From c0dca77a6c2f60ef6663b087e94aac3250227d7c Mon Sep 17 00:00:00 2001 From: Claude Date: Mon, 19 Oct 2026 11:13:42 +0000 Subject: [PATCH 01/10] Orbits study: where Kepler's law holds, where rotation curves go flat, and what ends a Kepler region A study for #36, which asks whether a galaxy is a big solar system. - orbits/model.js: - Kepler's law against JPL's approximate elements; - the tidal (Jacobi) radius and the density ratio 4A/Omega; - S2's orbit; - a power-law star cusp around Sgr A*; - MWPotential2014 plus Sgr A*; - a Hill-problem integrator for a point-mass or disk host; - the planets' tilts to the invariable plane. - orbits/table.mjs prints every number in the README; quick mode skips the stability scans. - tests/orbits.mjs has 21 checks. - The README gains a bullet on solar systems and galaxies, and CI runs the quick table. Co-Authored-By: Claude Opus 5.5 Claude-Session: https://claude.ai/code/session_01C2bpiHyHSkC8DAL5CcyciK --- .github/workflows/test.yml | 1 + README.md | 6 + orbits/README.md | 239 ++++++++++++++++++++++++++++ orbits/model.js | 315 +++++++++++++++++++++++++++++++++++++ orbits/table.mjs | 112 +++++++++++++ tests/orbits.mjs | 95 +++++++++++ 6 files changed, 768 insertions(+) create mode 100644 orbits/README.md create mode 100644 orbits/model.js create mode 100644 orbits/table.mjs create mode 100644 tests/orbits.mjs diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml index 34ea1d4..c44831d 100644 --- a/.github/workflows/test.yml +++ b/.github/workflows/test.yml @@ -31,3 +31,4 @@ jobs: - run: node witness-map/table.mjs - run: node witness-map/table.mjs 1e-6 cryo - run: node notes/bh-lensing.mjs + - run: node orbits/table.mjs quick diff --git a/README.md b/README.md index 152e37a..079aead 100644 --- a/README.md +++ b/README.md @@ -18,6 +18,12 @@ with Claude (Anthropic), for [Foundation0](https://github.com/Foundation-0/f0). with gravity in place of electric attraction (the "gravitational atom"). It is still a wave cloud, not planets on orbits. The same equation, a wave held by a pull that weakens with distance, gives the same solution at any scale. The nesting lives in the equations, not the objects. +- **Solar systems are not little galaxies, except at a galaxy's centre.** Around one dominant mass, orbits follow + Kepler's law at every radius: the planets match it to 0.13%, the size of their own masses. Across the Milky Way's + disk the speed instead stays near 230 km/s, because the mass is spread out. A Kepler region ends where its host's + tide wins, at a density ratio rather than a size: 1.4 pc for the Sun in the Galaxy, a third of an au for Jupiter. + A planet keeps moons, and the Sun comets, out to about half that radius. The Galaxy has its own small solar system + around its central black hole, a parsec or two across ([orbits](orbits/)). - **Quantum paths are fractal, between two lengths.** A free particle's path, recorded by position measurements finer than ħ/p, has fractal dimension 2, against 1 for a classical trajectory (Abbott and Wise, 1981); below ħ/mc, in a single-particle model, the speed of light makes it smooth again. Its surroundings draw diff --git a/orbits/README.md b/orbits/README.md new file mode 100644 index 0000000..f4c0a51 --- /dev/null +++ b/orbits/README.md @@ -0,0 +1,239 @@ +# Orbits: is a galaxy a big solar system? + +Status: **exploration, not peer reviewed, no new result claimed.** Issue +[#36](https://github.com/Foundation-0/fractals/issues/36). + +The repository's README opens with "atoms look a little like solar systems, solar systems a little like galaxies". +The first half has a study, the Witness Map. This one takes the second half. A solar system and a spiral galaxy +are both flat, rotating disks, but they do not orbit alike. This study asks where each one's law of motion holds, +what ends it, and what, if anything, repeats from one scale to the next. + +## Short answer + +| Question | Answer | +|---|---| +| Is a galaxy a big solar system? | No. Around one dominant mass, orbital speed falls as Kepler's v ∝ r^−1/2. Across the Milky Way's disk the speed stays near 230 km/s. From 5 to 25 kpc the mass inside grows 3.7-fold (Eilers et al. 2019), where a Kepler system's would stay fixed. | +| How exact is Kepler's law in the Solar System? | For all eight planets n²a³ matches GM☉ to within 1.3×10⁻³, below the planets' total mass of 1.34×10⁻³ of the Sun's. P ∝ a^1.49986 over 0.39 to 30 au. The mass inside Neptune's orbit is 0.13% more than the mass inside Mercury's. | +| Does the Galaxy have a Kepler region? | Yes, around Sgr A*. The star S2 orbits it as a planet orbits the Sun, with general relativity's 12.1′ precession per orbit on top. The region ends 1 to 3 pc out, where the stars around the black hole begin to weigh as much as it does. | +| What ends a Kepler region? | The host's tide. A body keeps satellites only inside its tidal radius, r_J³ = Gm/(4A(A − B)). There, the body's mean density is 4A/Ω times the host's mean density inside the orbit: 3 for a planet around the Sun, 2.25 for the Sun in the Galaxy. For the Sun, r_J = 1.37 pc = 2.8×10⁵ au. | +| How much of that region is used? | In this study's three-body runs (circular starting orbits, all orientations), about half: 0.53 r_J around a planet and 0.45 r_J for the Sun, which is 1.3×10⁵ au. Measured edges are about 0.4 to 0.6 r_H for the giant planets' outer moons and 1 to 2×10⁵ au for the Oort cloud. | +| What does repeat? | Kepler regions nested inside one another, each ending at its host's tide at a similar fraction of its tidal radius. Also flat disks: the planets' orbits are tilted by 2° on average (6° for Mercury), and the Galaxy's thin disk is about 0.1 as thick as it is wide. | + +## Kepler's law in the Solar System + +Around one mass M, a circular orbit of radius r has speed √(GM/r), and P² = 4π²a³/G(M + m). Nothing in the law +sets a size, so it holds from the Sun's surface out to wherever another pull competes. + +JPL's approximate elements (Standish, Table 1, a fit to 1800–2050) give each planet's semi-major axis a and mean +motion n. Against the IAU 2015 nominal GM☉ (`node orbits/table.mjs`, section 1): + +| | Mercury | Venus | Earth–Moon | Mars | Jupiter | Saturn | Uranus | Neptune | +|---|---|---|---|---|---|---|---|---| +| n²a³/GM☉ − 1 | 4.8e−6 | 1.7e−5 | 1.1e−5 | 3.7e−5 | 9.0e−4 | 2.2e−4 | 1.03e−3 | 1.27e−3 | +| m/M☉ | 1.7e−7 | 2.4e−6 | 3.0e−6 | 3.2e−7 | 9.5e−4 | 2.9e−4 | 4.4e−5 | 5.2e−5 | + +Every excess is below 1.34×10⁻³, the eight planets' total mass over the Sun's. The excesses are of that size +because the planets pull on one another as well as on the Sun: + +- Jupiter's excess is close to its own mass. +- Uranus's and Neptune's are closer to the mass of the planets inside their orbits. +- Saturn's, 2.2×10⁻⁴, is below its own mass and far below the mass inside its orbit. + +A 250-year fit to perturbed orbits does not separate these effects, so the study claims only the bound. The fitted +exponent of P against a is 1.49986 over all eight planets and 1.49999 over the inner four. + +## Where a Kepler region ends + +A body of mass m orbiting in a host feels the host's tide. Its own pull wins only inside the tidal (Jacobi) +radius, r_J³ = Gm/(4Ω² − κ²) = Gm/(4A(A − B)), where A and B are the host's Oort constants at the orbit (Binney +and Tremaine 2008, sec. 8.3): + +- **Point-mass host** (A = 3Ω/4, B = −Ω/4). This is Hill's radius, a(m/3M)^(1/3). Jupiter's is 0.355 au. +- **Flat rotation curve** (A = −B = Ω/2). For the Sun, Bovy's (2017) A = 15.3 and B = −11.9 km s⁻¹ kpc⁻¹ give 1.37 pc, which is 2.83×10⁵ au. Moving A or B by its error gives 1.35 to 1.39 pc, and Correa-Otto and Calandra (2019) quote 2.8×10⁵ au. + +Written as densities, the edge needs no length scale. Inside r_J the body's mean density, 3m/4πr_J³, is 4A/Ω +times the host's mean density inside the orbit, 3Ω²/4πG (spherical host). That ratio is exactly 3 for a planet +and 2 for a flat rotation curve; with Bovy's values it is 2.25. A Kepler region therefore ends where the central +body stops being two or three times denser than its surroundings. That is a statement about a ratio, not a size. + +### How much of the region a body keeps + +Inside r_J not every orbit survives. The model follows circular orbits in Hill's problem: the body, its host's tide +expanded to first order, and the Coriolis force. Lengths are in r_J and times in 1/Ω: + +``` +x'' = 2y' + Γx − Γx/r³, y'' = −2x' − Γy/r³, z'' = −ν²z − Γz/r³ +``` + +Two numbers set the host: + +- **Γ = 4A/Ω**: 3 for a point-mass host, 2.25 for the Sun's place in the Galaxy. +- **The vertical tide, ν²/Ω²**: + - It is 1 for a point mass. + - For the Galaxy it is (4πGρ₀ + 2(A² − B²))/Ω² = 7.6, from Poisson's equation with a local mid-plane density ρ₀ = 0.10 M☉ pc⁻³ (McKee, Parravano and Hollenbach 2015: 0.097 ± 0.013). + - The Galaxy's disk squeezes the Sun's sphere vertically more than seven times as hard as the Sun squeezes a planet's. + +An orbit is followed for 20 turns of the host. For the Sun that is 4.5 billion years, about the Solar System's age; +for Jupiter it is 237 years. It is counted as lost once it passes 3 r_J. The radius by which half the orbits (8 +starting angles each) have escaped (`node orbits/table.mjs`, section 5): + +| Orbit | Point-mass host | The Sun in the Galaxy | +|---|---|---| +| Prograde, in the plane | 0.45 | 0.475 | +| Retrograde, in the plane | 0.75 | 0.925 | +| Polar, plane through the host's direction | 0.45 | 0.425 | +| Polar, plane across it | 0.475 | 0.39 | +| Facing every way (24 directions × 4 angles) | 0.53 | 0.45 | + +For a planet the model's 0.45 and 0.75 are within 0.05 of published values. Shen and Tremaine (2008) find that many +prograde and retrograde satellites survive to about 0.5 and 0.7 r_H over up to 10⁹ years. The +observed irregular moons reach about 0.4 and 0.6 r_H. Followed for 100 turns instead of 20, the all-directions +value moves from 0.53 to 0.52. + +For the Sun the half-way radius is 0.45 r_J, or 1.26×10⁵ au. It is 0.44 to 0.46 r_J for ρ₀ from 0.08 to 0.12. The +Oort cloud's outer edge is quoted at 1 to 2×10⁵ au (Dones et al. 2004, as cited), so the Sun keeps comets about as +far out, as a fraction of its tidal radius, as a giant planet keeps moons. The Galaxy's stronger vertical tide +loses polar orbits a little sooner and keeps in-plane retrograde ones a little longer. + +The model also stops an orbit that plunges within 10⁻³ r_J. That is about Jupiter's radius, and for the Sun it is +280 au. Up to 17% of the Sun's inner orbits do so within 20 turns. This is the tide driving comets inward (Heisler +and Tremaine 1986), and the study does not follow it further. + +## A Kepler region at the Galaxy's centre + +GRAVITY's 2020 elements for S2 are a = 125.058 mas at R₀ = 8246.7 pc (1031 au), P = 16.0455 yr and e = 0.884649. +They give: + +- **The mass.** Kepler's third law gives M = a³/P² = 4.2606×10⁶ M☉. Their fit is 4.261×10⁶. The agreement shows only that the elements are consistent, since they fitted the mass with the same law. +- **The precession.** General relativity predicts 6πGM/(c²a(1 − e²)) = 12.16′ per orbit. They quote "the 12.1′ precession" and measured 1.10 ± 0.19 times it. +- **Pericentre.** 119 au, about 1,400 Schwarzschild radii, at 7,740 km/s, which is 2.6% of c. They quote 120 au and 7,700 km/s. + +Around Sgr A* (4.297×10⁶ M☉, GRAVITY 2022) lies the nuclear star cluster: + +- **The stars inside 1 pc.** 0.5 to 1.5×10⁶ M☉ (Schödel, Merritt and Eckart 2009). +- **The density.** It falls as r^−1.13 inside a break near 3 pc (Schödel et al. 2018). + +With the mass growing as r^1.87, the stars inside r weigh as much as Sgr A* at 1.8 to 3.2 pc, and twice as much at +2.5 to 4.6 pc. The literature puts that "twice" radius, the usual influence radius, at 3 to 4 pc: Alexander 2005, +3 pc; Feldmeier-Krause et al. 2017, about 4 pc; Merritt 2013, 3.9 pc, as cited by GRAVITY 2024. + +The speed's log slope is −0.497 at 0.1 pc and −0.32 at 1 pc. It passes −1/4 at 1.3 pc and turns positive by 3 pc. +For comparison, Bovy's MWPotential2014 bulge has no nuclear cluster and was fitted on kiloparsec scales; with +Sgr A* added it puts the "twice" radius at 9.4 pc. + +So a galaxy does contain a small solar system, at its centre and about a parsec across. That its size matches +the Sun's own 1.4 pc tidal radius is a coincidence: +- Sgr A* is four million times the Sun's mass. +- Its surroundings are about a million times denser than the Sun's. + +## Across the disk: flat, not Kepler + +Eilers et al. (2019) fit v_c = 229.0 − 1.7(R − 8.122) km/s from 5 to 25 kpc, with 2–5% systematic uncertainty: + +- **Log slope.** d ln v/d ln R is −0.04 at 5 kpc, −0.06 at the Sun and −0.21 at 25 kpc. Kepler's would be −0.5. +- **Mass inside.** Read as spherical, v²R/G grows 3.7-fold between 5 and 25 kpc. +- **The stars alone.** The stars weigh 5 ± 1×10¹⁰ M☉ (Bland-Hawthorn and Gerhard 2016). Even 10¹¹ M☉ all inside 25 kpc would hold only 131 km/s there, against about 200 measured. That is the missing-mass problem in one line. + +The model's Galaxy is Bovy's MWPotential2014: a power-law bulge with a cutoff, a Miyamoto–Nagai disk and an NFW +halo. It is normalised to 220 km/s at 8 kpc, with Sgr A* added. Its curve has Eilers et al.'s shape to within 2% +from 5 to 25 kpc, but sits 4% lower. Its log slope stays within ±0.15 of zero from 3.6 to 49 kpc. + +Going outward, the model's speed falls by less than Kepler's law would give. The log slope is −0.21 at 100 kpc +and −0.27 at 200 kpc, because the NFW halo's mass still grows. Gaia DR3 analyses differ on whether the measured +curve turns down past about 19 kpc: +- Jiao et al. (2023) report "a Keplerian decline". +- Koop et al. (2024) find the curves robust only to about 15 kpc. + +## An acceleration scale, and why this study does not use it + +In other galaxies the shift from Kepler-like to flat tracks an acceleration, not a radius. The observed and +baryonic accelerations follow one relation, which bends at g† = 1.20 ± 0.02 (random) ± 0.24 (systematic) × 10⁻¹⁰ +m s⁻² (McGaugh, Lelli and Schombert 2016). + +- **Where the Sun's pull reaches it.** At 7,030 au, inside its tidal radius. +- **The Galaxy's own pull at the Sun.** v²/R₀ = 2.1×10⁻¹⁰ m s⁻², or 1.7 g†. + +Wide binary stars at 2 to 30 thousand au test gravity at these accelerations, and the results disagree: +- Chae (2023) reports a 1.43 ± 0.06 boost. +- Banik et al. (2024) prefer Newtonian dynamics at 19σ and exclude MOND at 16σ. +- Later analyses still split. + +The study reports the scale and leaves the question to the Witness Map and issue #3. + +## What does repeat: thin disks + +The planets' orbits lie within a few degrees of the invariable plane, the plane perpendicular to the Solar System's +total angular momentum (Souami and Souchay 2012). Computed from JPL's elements, the inclinations are: +- Mercury 6.3°, Venus 2.2°, Earth–Moon 1.6°, Mars 1.7°; +- Jupiter 0.33°, Saturn 0.93°, Uranus 1.0°, Neptune 0.73°. + +The rms of sin i is 0.045, or 0.023 without Mercury. The Galaxy's thin disk is 300 ± 50 pc thick over a scale +length of 2.6 ± 0.5 kpc, a ratio of 0.12 (Bland-Hawthorn and Gerhard 2016). Both are flat for the same reason: gas +that radiates its energy but keeps its angular momentum settles into a disk. What the disks orbit differs: one +dominant mass, or mass spread through the disk and halo. + +## Run it + +``` +node orbits/table.mjs # every number above (about 45 s; the stability scans take nearly all of it) +node orbits/table.mjs quick # sections 1-4 only, well under a second +node tests/orbits.mjs # the checks, about 1 s (also run by npm test) +``` + +## Checks + +`tests/orbits.mjs` has 21 checks. Each prints its source, the published or expected value, the model's value and +the tolerance. + +| What | Against | Kind | +|---|---|---| +| n²a³/GM☉ − 1 below the planets' mass fraction, for all eight; exponent 3/2 | JPL Table 1, IAU 2015, DE430 | published elements | +| 4A/Ω = 3 and 2; Jacobi radius = Hill radius for a point mass | closed forms | arithmetic, model-internal | +| The Sun's r_J = 2.8×10⁵ au | Correa-Otto and Calandra 2019 | search text | +| S2: Kepler mass, 12.1′ precession, 120 au, 7,700 km/s | GRAVITY 2020 | abstract and appendix, via a transcription | +| Stars weigh 2 M_BH at 3 to 4 pc | Alexander 2005; Feldmeier-Krause et al. 2017; Merritt 2013 | search text | +| Kepler slope −1/2 inside 0.1 pc | model | model | +| MWPotential2014: 220 km/s at 8 kpc; shape within 5% of Eilers et al.; flat (|slope| < 0.25) at 5 to 25 kpc | Eilers et al. 2019 | normalisation; abstract | +| Jupiter's and Saturn's tilts to the invariable plane | Souami and Souchay 2012 | search text | +| Planet: prograde kept at 0.4 r_H, lost by 0.6; retrograde kept at 0.6, lost by 0.9 | Shen and Tremaine 2008 (0.5, 0.7) | abstract | +| The Galaxy's ν²/Ω² | Poisson's equation | arithmetic | +| The Sun keeps every orientation at 0.3 r_J, loses most by 0.6 | Oort cloud edge 1 to 2×10⁵ au | model; as cited | + +## What the model leaves out + +- **The planets' pull on one another.** Kepler's law is checked only to the size of the planets' masses. +- **Hill's problem is the tide to first order, on a circular host orbit.** The study drops the Sun's vertical oscillation through the disk (a period of about 80 Myr with this model's ν) and its radial epicycles. It also leaves out passing stars and molecular clouds, which also strip the Oort cloud, and the planets' scattering of comets that plunge inward. +- **Circular starting orbits only.** Real moons and comets start on eccentric orbits. The edges are set at half the orbits escaping within 20 turns of the host, on a grid of 0.025 r_J (in the plane) or 0.1 r_J (all directions). +- **MWPotential2014 is one model among many.** Its halo sets the flat range, and it has no nuclear star cluster. The Galactic-centre numbers use the separate power-law cusp above, which is valid only inside its 3 pc break. +- **Spherical mass estimates.** v²R/G treats a flattened galaxy as spherical. The disk changes the numbers by tens of percent, not the 3.7-fold growth. + +## Sources + +Read here only through search results and abstracts, or as noted. Each entry says how it was read. + +- W. M. Folkner et al., *The Planetary and Lunar Ephemerides DE430 and DE431*, IPN Progress Report 42-196 (2014). GM values seen in copies of the file. +- E. M. Standish, *Keplerian Elements for Approximate Positions of the Major Planets*, JPL Solar System Dynamics, Table 1 (1800–2050). Seen in a copy of the file. +- A. Prša et al., *Nominal values for selected solar and planetary quantities: IAU 2015 Resolution B3*, Astron. J. 152, 41 (2016). https://doi.org/10.3847/0004-6256/152/2/41. Search text of the resolution. +- D. Souami and J. Souchay, *The solar system's invariable plane*, A&A 543, A133 (2012). https://doi.org/10.1051/0004-6361/201219011. Search text. +- J. Binney and S. Tremaine, *Galactic Dynamics*, 2nd ed., Princeton (2008), sec. 8.3. From memory. +- J. Bovy, *Galactic rotation in Gaia DR1*, MNRAS 468, L63 (2017). https://arxiv.org/abs/1610.07610. Abstract, search text. +- J. A. Correa-Otto and M. F. Calandra, MNRAS 490, 2495 (2019). https://doi.org/10.1093/mnras/stz2671. Search text. +- Y. Shen and S. Tremaine, *Stability of Distant Satellites of the Giant Planets in the Solar System*, AJ (2008). https://arxiv.org/abs/0806.2882. Abstract and introduction, search text. +- C. F. McKee, A. Parravano and D. J. Hollenbach, *Stars, gas, and dark matter in the solar neighborhood*, ApJ 814, 13 (2015). https://arxiv.org/abs/1509.05334. Search text, secondary. +- M. Heisler and S. Tremaine, *The influence of the Galactic tidal field on the Oort comet cloud*, Icarus 65, 13 (1986). Abstract, search text. +- L. Dones, P. R. Weissman, H. F. Levison and M. J. Duncan, *Oort cloud formation and dynamics*, in *Comets II*, 153 (2004). As cited in search text. +- GRAVITY Collaboration, *Detection of the Schwarzschild precession in the orbit of the star S2 near the Galactic centre massive black hole*, A&A 636, L5 (2020). https://doi.org/10.1051/0004-6361/202037813. Abstract, appendix B and Table E.1, read in a transcription. +- GRAVITY Collaboration, A&A 657, L12 (2022). https://doi.org/10.1051/0004-6361/202142465. Search text. +- R. Schödel, D. Merritt and A. Eckart, A&A 502, 91 (2009). https://doi.org/10.1051/0004-6361/200810922. Abstract, search text. +- R. Schödel et al., A&A 609, A27 (2018). Abstract, search text. +- T. Alexander, Phys. Rep. 419, 65 (2005). As cited by Feldmeier-Krause et al. 2017. +- A. Feldmeier-Krause et al., MNRAS 466, 4040 (2017). https://arxiv.org/abs/1701.01583. Search text. +- D. Merritt, *Dynamics and Evolution of Galactic Nuclei*, Princeton (2013). As cited by GRAVITY Collaboration 2024 (https://arxiv.org/abs/2409.12261). +- J. Bovy, *galpy: A Python Library for Galactic Dynamics*, ApJS 216, 29 (2015). https://doi.org/10.1088/0067-0049/216/2/29. MWPotential2014's parameters read in galpy's source. +- C. Eilers, D. W. Hogg, H.-W. Rix and M. K. Ness, ApJ 871, 120 (2019). https://doi.org/10.3847/1538-4357/aaf648. Abstract, search text. +- J. Bland-Hawthorn and O. Gerhard, *The Galaxy in Context*, ARA&A 54, 529 (2016). https://doi.org/10.1146/annurev-astro-081915-023441. As cited in search text. +- Y. Jiao et al., A&A 678, A208 (2023). https://doi.org/10.1051/0004-6361/202347513. Abstract, search text. +- H. Koop et al., A&A 692, A50 (2024). https://arxiv.org/abs/2405.19028. Abstract, search text. +- S. S. McGaugh, F. Lelli and J. M. Schombert, PRL 117, 201101 (2016). https://doi.org/10.1103/PhysRevLett.117.201101. Search text. +- K.-H. Chae, ApJ 952, 128 (2023). https://doi.org/10.3847/1538-4357/ace101. Search text. +- I. Banik et al., MNRAS 527, 4573 (2024). https://doi.org/10.1093/mnras/stad3393. Abstract, search text. diff --git a/orbits/model.js b/orbits/model.js new file mode 100644 index 0000000..78cb5b1 --- /dev/null +++ b/orbits/model.js @@ -0,0 +1,315 @@ +/* Orbits: is a galaxy a big solar system? Where Kepler's law holds, where rotation curves go flat, and what sets the + * boundary. Part of https://github.com/Foundation-0/fractals (issue #36). No dependencies. Loads in Node (require) and + * in a browser (window.ORB). + * + * Kepler's law. Around one dominant mass M, a circular orbit of radius r has speed v = sqrt(GM/r), so v falls as + * r^(-1/2) and the period P obeys P^2 = 4 pi^2 a^3 / (G(M + m)). The law holds at every radius from the central body to + * where something else's pull competes, so it repeats across scales with no size of its own. + * + * Where it stops. A body of mass m on a circular orbit of angular speed Omega in a host keeps its own satellites + * inside the tidal (Jacobi) radius, where its pull matches the host's tidal stretch less the centrifugal term + * (Binney & Tremaine 2008, sec. 8.3): + * r_J^3 = G m / (4 Omega^2 - kappa^2) = G m / (4 A (A - B)), Omega = A - B, kappa^2 = -4 B (A - B), + * with A and B the host's Oort constants at the orbit. For a point-mass host A = 3 Omega / 4 and B = -Omega / 4, and + * r_J is Hill's radius a (m / 3M)^(1/3). For a flat rotation curve A = -B = Omega / 2. Either way the mean density of + * m inside r_J, 3m / (4 pi r_J^3), equals 4A/Omega times the host's mean density inside the orbit, 3 Omega^2 / (4 pi G) + * (spherical host): 3 times for a point mass, 2 for a flat curve. So the edge is set by a density ratio, not a size. + * + * Flat curves. Where the enclosed mass grows as M( { const d = Math.PI / 180; return [Math.sin(i * d) * Math.sin(Om * d), -Math.sin(i * d) * Math.cos(Om * d), Math.cos(i * d)]; }; + // A planet's orbital inclination to the invariable plane [deg]. + function tiltToInvariable(p) { + const n = orbitNormal(p.i, p.Om), m = orbitNormal(INVARIABLE.i, INVARIABLE.Om); + return Math.acos(Math.min(1, n[0] * m[0] + n[1] * m[1] + n[2] * m[2])) * 180 / Math.PI; + } + const planetMassFraction = () => PLANETS.reduce((s, p) => s + 1 / p.sunOver, 0); + + // Kepler's third law for one planet: the GM its orbit implies, n^2 a^3, against GM_sun, with n the mean-longitude + // rate. A 250-year fit to a perturbed orbit, so the excess carries the other planets' pull as well as the planet's + // own mass. + function kepler(p) { + const n = p.Ldot * Math.PI / 180 / (36525 * C.day), a = p.a * C.au; + const GM = n * n * a * a * a; + return { name: p.name, n, P: 2 * Math.PI / n / C.yr, GM, excess: GM / C.GMsun - 1, own: 1 / p.sunOver }; + } + + // Least-squares slope of log P against log a: 3/2 for Kepler. + function keplerExponent(list = PLANETS) { + const xs = list.map(p => Math.log(p.a)), ys = list.map(p => Math.log(kepler(p).P)); + const mx = xs.reduce((s, x) => s + x, 0) / xs.length, my = ys.reduce((s, y) => s + y, 0) / ys.length; + let sxy = 0, sxx = 0; + xs.forEach((x, i) => { sxy += (x - mx) * (ys[i] - my); sxx += (x - mx) ** 2; }); + return sxy / sxx; + } + + // ---------- where a Kepler domain ends ---------- + // Jacobi radius [m] of a body of GM [m^3/s^2] in a host with Oort constants A, B [km/s/kpc]. + const jacobi = (GM, A, B) => Math.cbrt(GM / (4 * A * (A - B) * C.kmskpc ** 2)); + // Hill radius [same unit as a] for a point-mass host: the Jacobi radius with A = 3 Omega/4, B = -Omega/4. + const hill = (a, massRatio) => a * Math.cbrt(massRatio / 3); + // Oort constants of a point-mass host and of a flat rotation curve, at angular speed Omega. + const oortKepler = Omega => ({ A: 0.75 * Omega, B: -0.25 * Omega }); + const oortFlat = Omega => ({ A: 0.5 * Omega, B: -0.5 * Omega }); + // The body's mean density inside r_J over the host's mean density inside the orbit (spherical host): 4A/Omega. + const densityRatio = (A, B) => 4 * A / (A - B); + // Where a mass's own pull falls to an acceleration g: r = sqrt(GM/g) [m]. + const radiusAtAcceleration = (GM, g) => Math.sqrt(GM / g); + + // ---------- Sgr A* and S2 ---------- + // GRAVITY Collaboration 2020 (A&A 636, L5), Table E.1: S2's semi-major axis [mas], eccentricity, period [yr]; their + // fitted mass [M_sun] and distance [pc]. + const S2 = { aMas: 125.058, e: 0.884649, P: 16.0455, M: 4.261e6, R0: 8246.7, fSP: 1.10, fSPerr: 0.19 }; + function orbitOf({ aMas, e, P, R0 }, M) { + const aAU = aMas * 1e-3 * R0, Mk = aAU ** 3 / (P * P); // Kepler's third law in au, yr, M_sun (m << M) + const mass = M === undefined ? Mk : M, GM = mass * C.GMsun, a = aAU * C.au; + const precession = 6 * Math.PI * GM / (C.c * C.c * a * (1 - e * e)); // rad per orbit + const rp = a * (1 - e); + return { + aAU, keplerMass: Mk, precessionArcmin: precession * 180 / Math.PI * 60, + periAU: rp / C.au, periSpeed: Math.sqrt(GM * (1 + e) / rp) / 1e3, periOverRs: rp / (2 * GM / (C.c * C.c)), + }; + } + + // The stars around Sgr A*: inside the break near 3 pc the density falls as r^-gamma, gamma = 1.13 (Schodel et al. + // 2018), so their mass grows as r^(3 - gamma); 0.5 to 1.5 million solar masses lie inside 1 pc (Schodel, Merritt & + // Eckart 2009). Radii in pc. + const CUSP = { M1: 1.0e6, M1lo: 0.5e6, M1hi: 1.5e6, gamma: 1.13, rBreak: 3 }; + const cuspMass = (r, M1 = CUSP.M1, gamma = CUSP.gamma) => M1 * Math.pow(r, 3 - gamma); + // The radius where the stars inside weigh k times Sgr A* (k = 1, or k = 2 for the usual influence radius). + const cuspRadius = (k, Mbh, M1 = CUSP.M1, gamma = CUSP.gamma) => Math.pow(k * Mbh / M1, 1 / (3 - gamma)); + // dln v / dln r for Sgr A* plus the cusp: (dln M / dln r - 1) / 2. + const cuspSlope = (r, Mbh, M1 = CUSP.M1, gamma = CUSP.gamma) => { const m = cuspMass(r, M1, gamma); return ((3 - gamma) * m / (Mbh + m) - 1) / 2; }; + + // ---------- the Milky Way ---------- + function lgamma(z) { // Lanczos, g = 7; |error| < 1e-14 for z > 0 + const c = [0.99999999999980993, 676.5203681218851, -1259.1392167224028, 771.32342877765313, -176.61502916214059, + 12.507343278686905, -0.13857109526572012, 9.9843695780195716e-6, 1.5056327351493116e-7]; + if (z < 0.5) return Math.log(Math.PI / Math.sin(Math.PI * z)) - lgamma(1 - z); + z -= 1; let x = c[0]; + for (let i = 1; i < 9; i++) x += c[i] / (z + i); + const t = z + 7.5; + return 0.5 * Math.log(2 * Math.PI) + (z + 0.5) * Math.log(t) - t + Math.log(x); + } + function gammaP(s, x) { // regularised lower incomplete gamma: series below s + 1, continued fraction above + if (x <= 0) return 0; + const pre = Math.exp(-x + s * Math.log(x) - lgamma(s)); + if (x < s + 1) { + let sum = 1 / s, term = 1 / s; + for (let n = 1; n < 1000; n++) { term *= x / (s + n); sum += term; if (term < sum * 1e-17) break; } + return pre * sum; + } + let b = x + 1 - s, c = 1e300, d = 1 / b, h = d; // modified Lentz + for (let i = 1; i < 1000; i++) { + const an = -i * (i - s); b += 2; + d = an * d + b; if (Math.abs(d) < 1e-300) d = 1e-300; + c = b + an / c; if (Math.abs(c) < 1e-300) c = 1e-300; + d = 1 / d; const del = d * c; h *= del; + if (Math.abs(del - 1) < 1e-16) break; + } + return 1 - pre * h; + } + + // Bovy 2015 (ApJS 216, 29), sec. 3.5: MWPotential2014. + const MWP14 = { R0: 8, v0: 220, f: { bulge: 0.05, disk: 0.6, halo: 0.35 }, alpha: 1.8, rc: 1.9, a: 3, b: 0.28, rs: 16 }; + const SGRA = { M: 4.297e6 }; // GRAVITY Collaboration 2022 (A&A 657, L12) + + function galaxy(par = MWP14, { bh = SGRA.M } = {}) { + const shape = { + bulge: r => gammaP((3 - par.alpha) / 2, (r / par.rc) ** 2), // M( R * R / Math.pow(R * R + (par.a + par.b) ** 2, 1.5), // v^2 / (G M_disk), mid-plane + halo: r => Math.log(1 + r / par.rs) - (r / par.rs) / (1 + r / par.rs), // M( ({ + bh: C.G * M.bh / R, bulge: C.G * M.bulge * shape.bulge(R) / R, disk: C.G * M.disk * shape.disk(R), + halo: C.G * M.halo * shape.halo(R) / R, + }); + const v2 = R => { const p = parts(R); return p.bh + p.bulge + p.disk + p.halo; }; + const v = R => Math.sqrt(v2(R)); + const slope = R => { const h = 1e-4; return Math.log(v(R * (1 + h)) / v(R * (1 - h))) / Math.log((1 + h) / (1 - h)); }; + const bulgeMass = r => M.bulge * shape.bulge(r); + // the smallest radius above r0 where f crosses zero, by a log grid then bisection + const cross = (f, r0, r1) => { + let lo = r0, flo = f(lo); + for (let r = r0 * 1.02; r <= r1; r *= 1.02) { + const fr = f(r); + if (Math.sign(fr) !== Math.sign(flo)) { + let a = lo, b = r; + for (let i = 0; i < 100; i++) { const m = Math.sqrt(a * b); if (Math.sign(f(m)) === Math.sign(flo)) a = m; else b = m; } + return Math.sqrt(a * b); + } + lo = r; flo = fr; + } + return NaN; + }; + return { + par, M, parts, v, slope, bulgeMass, cross, + // where the stars inside r first weigh as much as Sgr A* (the influence radius), and where v's log slope is -1/4 + influence: () => cross(r => bulgeMass(r) - M.bh, 1e-6, 1), + keplerEdge: () => cross(r => slope(r) + 0.25, 1e-6, 1), + // the longest run of radii (kpc, on a grid 0.5% apart) where |dln v / dln R| < tol + flatRange: (tol = 0.15, rMin = 1e-3, rMax = 300) => { + let best = [NaN, NaN], start = NaN, last = NaN; + for (let r = rMin; r <= rMax * 1.0001; r *= 1.005) { + if (Math.abs(slope(r)) < tol) { if (Number.isNaN(start)) start = r; last = r; } + else if (!Number.isNaN(start)) { + if (Number.isNaN(best[0]) || last / start > best[1] / best[0]) best = [start, last]; + start = NaN; + } + } + if (!Number.isNaN(start) && (Number.isNaN(best[0]) || last / start > best[1] / best[0])) best = [start, last]; + return best; + }, + }; + } + + // ---------- Hill's problem: which orbits a body keeps inside its tidal radius ---------- + // A satellite of a body on a circular orbit in a host, in the frame rotating with the body (x away from the host's + // centre, y along the motion, z out of the plane), lengths in r_J and times in 1/Omega (Binney & Tremaine 2008, + // sec. 8.3; Heisler & Tremaine 1986 for the vertical tide): + // x'' = 2y' + Gam x - Gam x/r^3, y'' = -2x' - Gam y/r^3, z'' = -nu2 z - Gam z/r^3, + // with Gam = 4A/Omega (3 for a point-mass host, so these are Hill's equations) and nu2 = (vertical tidal frequency / + // Omega)^2: 1 for a point-mass host; for a disk, (4 pi G rho0 + 2(A^2 - B^2)) / Omega^2 by Poisson's equation, with + // rho0 the local mid-plane density. The tidal radius r_J is where the x-axis forces balance (x = 1). + const HOST_KEPLER = { name: 'point-mass host (a planet around the Sun)', Gam: 3, nu2: 1 }; + function hostDisk(A, B, rho0) { // A, B in km/s/kpc; rho0 in M_sun/pc^3 + const Omega = A - B, nuz = 4 * Math.PI * C.G * rho0 * 1e9 + 2 * (A * A - B * B); // (km/s/kpc)^2; 1 pc^-3 = 1e9 kpc^-3 + return { name: 'disk host (the Sun in the Galaxy)', Gam: 4 * A / Omega, nu2: nuz / (Omega * Omega), A, B, rho0 }; + } + + function hillDeriv(s, Gam, nu2, out) { + const x = s[0], y = s[1], z = s[2], vx = s[3], vy = s[4], vz = s[5]; + const r2 = x * x + y * y + z * z, k = Gam / (r2 * Math.sqrt(r2)); + out[0] = vx; out[1] = vy; out[2] = vz; + out[3] = 2 * vy + Gam * x - k * x; out[4] = -2 * vx - k * y; out[5] = -nu2 * z - k * z; + } + // Dormand-Prince 5(4) with step control (Hairer, Norsett & Wanner, Solving ODEs I, sec. II.5). + const DP = { + a: [[], [1 / 5], [3 / 40, 9 / 40], [44 / 45, -56 / 15, 32 / 9], [19372 / 6561, -25360 / 2187, 64448 / 6561, -212 / 729], + [9017 / 3168, -355 / 33, 46732 / 5247, 49 / 176, -5103 / 18656], [35 / 384, 0, 500 / 1113, 125 / 192, -2187 / 6784, 11 / 84]], + e: [71 / 57600, 0, -71 / 16695, 71 / 1920, -17253 / 339200, 22 / 525, -1 / 40], + }; + // Follows one orbit for `orbits` turns of the host (time 2 pi orbits). Fate: 'escapes' once r > rEsc, 'plunges' + // once r < rIn (into the inner system: the planet itself, or the Sun's planets), otherwise 'stays'. + function hillOrbit(host, s0, { orbits = 20, rEsc = 3, rIn = 1e-3, tol = 1e-10 } = {}) { + const T = 2 * Math.PI * orbits, k = Array.from({ length: 7 }, () => new Float64Array(6)); + const y = new Float64Array(6), s = Float64Array.from(s0); + let t = 0, h = 1e-3, steps = 0; + hillDeriv(s, host.Gam, host.nu2, k[0]); + while (t < T) { + if (t + h > T) h = T - t; + for (let j = 1; j < 7; j++) { + for (let i = 0; i < 6; i++) { let acc = s[i]; for (let m = 0; m < j; m++) acc += h * DP.a[j][m] * k[m][i]; y[i] = acc; } + hillDeriv(y, host.Gam, host.nu2, k[j]); + } + let err = 0; + for (let i = 0; i < 6; i++) { + let ei = 0; for (let m = 0; m < 7; m++) ei += DP.e[m] * k[m][i]; + err = Math.max(err, Math.abs(h * ei) / (tol * (1 + Math.abs(s[i])))); + } + if (err <= 1) { + t += h; steps++; + for (let i = 0; i < 6; i++) { s[i] = y[i]; k[0][i] = k[6][i]; } + const r = Math.hypot(s[0], s[1], s[2]); + if (r > rEsc) return { fate: 'escapes', t, steps }; + if (r < rIn) return { fate: 'plunges', t, steps }; + } + h *= Math.min(5, Math.max(0.2, 0.9 * Math.pow(err, -0.2))); + } + return { fate: 'stays', t, steps }; + } + // A circular orbit of radius a (in r_J) about the body, at the two-body speed, in the plane with unit normal n + // (rotating-frame axes; n = +z is prograde, -z retrograde), starting at angle phi. + function hillStart(host, a, n, phi) { + const [nx, ny, nz] = n; + let ux, uy, uz; + if (Math.abs(nz) < 0.9) { ux = -ny; uy = nx; uz = 0; } else { ux = 0; uy = -nz; uz = ny; } + const un = Math.hypot(ux, uy, uz); ux /= un; uy /= un; uz /= un; + const wx = ny * uz - nz * uy, wy = nz * ux - nx * uz, wz = nx * uy - ny * ux; + const c = Math.cos(phi), s = Math.sin(phi), v = Math.sqrt(host.Gam / a); + const x = a * (c * ux + s * wx), y = a * (c * uy + s * wy), z = a * (c * uz + s * wz); + // velocity in the rotating frame: the inertial circular velocity less Omega x r, Omega = +z + return [x, y, z, v * (-s * ux + c * wx) + y, v * (-s * uy + c * wy) - x, v * (-s * uz + c * wz)]; + } + // n directions spread evenly over the sphere (Fibonacci lattice). + function sphere(n) { + const out = [], g = Math.PI * (3 - Math.sqrt(5)); + for (let i = 0; i < n; i++) { const z = 1 - 2 * (i + 0.5) / n, r = Math.sqrt(1 - z * z); out.push([r * Math.cos(g * i), r * Math.sin(g * i), z]); } + return out; + } + // Fractions of orbits that stay, escape and plunge, over the given orbit normals and nPhase starting angles each. + function hillFates(host, a, normals, { nPhase = 4, ...opt } = {}) { + const n = { stays: 0, escapes: 0, plunges: 0 }; + for (const nv of normals) for (let p = 0; p < nPhase; p++) n[hillOrbit(host, hillStart(host, a, nv, 2 * Math.PI * (p + 0.5) / nPhase), opt).fate]++; + const tot = normals.length * nPhase; + return { stays: n.stays / tot, escapes: n.escapes / tot, plunges: n.plunges / tot }; + } + // Fates on a grid of radii (in r_J), and the radius where the fraction that escapes first reaches one half (linear + // interpolation between grid points; NaN if it never does). + function hillScan(host, normals, as, opt = {}) { + const rows = as.map(a => ({ a, ...hillFates(host, a, normals, opt) })); + let edge = NaN; + for (let i = 0; i < rows.length; i++) { + if (rows[i].escapes >= 0.5) { + edge = i === 0 ? rows[0].a : rows[i - 1].a + (rows[i].a - rows[i - 1].a) * (0.5 - rows[i - 1].escapes) / (rows[i].escapes - rows[i - 1].escapes); + break; + } + } + return { rows, edge }; + } + + // McGaugh, Lelli & Schombert 2016: the radial acceleration relation, g_obs = g_bar / (1 - exp(-sqrt(g_bar / g+))). + const GDAGGER = 1.20e-10; // m s^-2 + const rar = (gbar, gd = GDAGGER) => gbar / (1 - Math.exp(-Math.sqrt(gbar / gd))); + + const api = { + C, PLANETS, INVARIABLE, tiltToInvariable, planetMassFraction, kepler, keplerExponent, jacobi, hill, oortKepler, oortFlat, densityRatio, + radiusAtAcceleration, S2, orbitOf, CUSP, cuspMass, cuspRadius, cuspSlope, lgamma, gammaP, MWP14, SGRA, galaxy, GDAGGER, rar, + HOST_KEPLER, hostDisk, hillOrbit, hillStart, sphere, hillFates, hillScan, + }; + if (typeof module !== 'undefined' && module.exports) module.exports = api; else root.ORB = api; +})(typeof window !== 'undefined' ? window : globalThis); diff --git a/orbits/table.mjs b/orbits/table.mjs new file mode 100644 index 0000000..6768445 --- /dev/null +++ b/orbits/table.mjs @@ -0,0 +1,112 @@ +// Prints every number in orbits/README.md. Run: node orbits/table.mjs (about a minute; the stability scans of +// section 5 take nearly all of it), or node orbits/table.mjs quick (well under a second; section 5 left out). +import { createRequire } from 'node:module'; +const O = createRequire(import.meta.url)('./model.js'); +const { C } = O; +const e3 = x => x.toExponential(3), f = (x, d = 3) => x.toFixed(d); +const quick = process.argv[2] === 'quick'; + +console.log('1. Kepler in the Solar System (JPL approximate elements, Table 1, 1800-2050; GM_sun IAU 2015; masses DE430)\n'); +console.log(' planet a [au] P [yr] n^2 a^3/GM_sun - 1 own mass/Sun own + inner planets'); +let inner = 0; +for (const p of O.PLANETS) { + const k = O.kepler(p); + console.log(` ${p.name.padEnd(9)} ${f(p.a, 6).padStart(10)} ${f(k.P, 5).padStart(11)} ${e3(k.excess).padStart(10)} ${e3(k.own)} ${e3(k.own + inner)}`); + inner += k.own; +} +const mu = O.planetMassFraction(); +console.log(` all eight planets together: ${e3(mu)} of the Sun's mass; largest |excess| ${e3(Math.max(...O.PLANETS.map(p => Math.abs(O.kepler(p).excess))))}`); +console.log(` fitted exponent of P against a: ${f(O.keplerExponent(), 6)} over all eight (0.39 to 30 au), ${f(O.keplerExponent(O.PLANETS.slice(0, 4)), 6)} over the inner four; Kepler 1.5`); +console.log(` mass inside Neptune's orbit over mass inside Mercury's: 1 + ${e3(mu)}`); + +console.log('\n2. Where a Kepler domain ends: the tidal (Jacobi) radius, r_J^3 = G m / (4 A (A - B))\n'); +const OORT = { A: 15.3, B: -11.9, dA: 0.4, dB: 0.4 }; // Bovy 2017 +const rJ = O.jacobi(C.GMsun, OORT.A, OORT.B); +const rJs = [[OORT.A + OORT.dA, OORT.B], [OORT.A - OORT.dA, OORT.B], [OORT.A, OORT.B + OORT.dB], [OORT.A, OORT.B - OORT.dB]].map(([A, B]) => O.jacobi(C.GMsun, A, B) / C.pc); +console.log(` the Sun in the Galaxy (A = ${OORT.A}, B = ${OORT.B} km/s/kpc, Bovy 2017): r_J = ${f(rJ / C.pc)} pc = ${f(rJ / C.au / 1e5, 2)}e5 au` + + ` (moving A or B by its error: ${f(Math.min(...rJs))} to ${f(Math.max(...rJs))} pc)`); +console.log(` the Sun's mean density inside r_J over the Galaxy's inside the Sun's orbit: 4A/(A - B) = ${f(O.densityRatio(OORT.A, OORT.B), 2)}` + + ` (3 for a point-mass host, 2 for a flat rotation curve)`); +const rhoJ = 1 / (4 / 3 * Math.PI * (rJ / C.pc) ** 3); +console.log(` that mean density: ${f(rhoJ, 3)} M_sun/pc^3`); +for (const p of O.PLANETS) console.log(` Hill radius of ${p.name.padEnd(8)} a (m/3M)^(1/3) = ${f(O.hill(p.a, 1 / p.sunOver), 4)} au = ${f(O.hill(p.a, 1 / p.sunOver) / p.a, 4)} a`); +console.log(` the Sun's own pull falls to g+ = ${O.GDAGGER.toExponential(2)} m/s^2 at ${f(O.radiusAtAcceleration(C.GMsun, O.GDAGGER) / C.au, 0)} au` + + ` (${f(O.radiusAtAcceleration(C.GMsun, O.GDAGGER) / rJ, 3)} r_J)`); + +console.log('\n3. Sgr A*: a Kepler domain at the centre of the Galaxy (S2, GRAVITY Collaboration 2020)\n'); +const s2 = O.orbitOf(O.S2), s2M = O.orbitOf(O.S2, O.S2.M); +console.log(` a = ${O.S2.aMas} mas at R0 = ${O.S2.R0} pc: ${f(s2.aAU, 1)} au; P = ${O.S2.P} yr; e = ${O.S2.e}`); +console.log(` Kepler's third law: M = a^3/P^2 = ${f(s2.keplerMass / 1e6, 4)}e6 M_sun (their fit ${O.S2.M / 1e6}e6)`); +console.log(` Schwarzschild precession 6 pi GM / (c^2 a (1 - e^2)) = ${f(s2M.precessionArcmin, 2)} arcmin per orbit; measured f_SP = ${O.S2.fSP} +- ${O.S2.fSPerr} of it`); +console.log(` pericentre ${f(s2M.periAU, 0)} au = ${f(s2M.periOverRs, 0)} Schwarzschild radii, at ${f(s2M.periSpeed, 0)} km/s (${f(s2M.periSpeed * 1e3 / C.c * 100, 2)}% of c)`); +const MBH = O.SGRA.M, K = O.CUSP; +console.log(` Sgr A* today: ${MBH / 1e6}e6 M_sun (GRAVITY Collaboration 2022). The stars around it: ${K.M1lo / 1e6} to ${K.M1hi / 1e6}e6 M_sun inside 1 pc` + + ` (Schodel, Merritt & Eckart 2009), density ~ r^-${K.gamma} inside ~${K.rBreak} pc (Schodel et al. 2018)`); +for (const [lab, M1] of [['low', K.M1lo], ['middle', K.M1], ['high', K.M1hi]]) { + console.log(` ${lab.padEnd(6)} (${M1 / 1e6}e6 inside 1 pc): stars weigh as much as Sgr A* inside ${f(O.cuspRadius(1, MBH, M1), 2)} pc, twice as much inside ${f(O.cuspRadius(2, MBH, M1), 2)} pc`); +} +{ + let lo = 1e-3, hi = 30; + for (let i = 0; i < 100; i++) { const m = Math.sqrt(lo * hi); if (O.cuspSlope(m, MBH) < -0.25) lo = m; else hi = m; } + console.log(` dln v/dln r (middle): ${[0.01, 0.1, 0.5, 1, 2, 3].map(r => `${f(O.cuspSlope(r, MBH))} at ${r} pc`).join(', ')}; -1/4 at ${f(lo, 2)} pc`); +} +const g = O.galaxy(); +console.log(` for comparison, MWPotential2014's bulge (fit on kpc scales, no nuclear star cluster) plus Sgr A*: stars weigh as much as Sgr A* inside` + + ` ${f(g.influence() * 1e3, 2)} pc, twice as much inside ${f(g.cross(r => g.bulgeMass(r) - 2 * g.M.bh, 1e-6, 1) * 1e3, 2)} pc`); + +console.log('\n4. The Galaxy: v(R) and its log slope in the mid-plane (MWPotential2014, Bovy 2015, plus Sgr A*)\n'); +console.log(` component masses: bulge ${e3(g.M.bulge)}, disk ${e3(g.M.disk)} M_sun; halo 4 pi rho_s r_s^3 = ${e3(g.M.halo)} M_sun`); +console.log(' R [kpc] v [km/s] dln v/dln R fractions of v^2: BH, bulge, disk, halo'); +for (const R of [1e-5, 1e-4, 1e-3, 3e-3, 1e-2, 0.03, 0.1, 0.3, 1, 3, 5, 8, 15, 25, 50, 100, 200]) { + const p = g.parts(R), v2 = p.bh + p.bulge + p.disk + p.halo; + console.log(` ${String(R).padEnd(10)} ${f(g.v(R), 1).padStart(9)} ${f(g.slope(R)).padStart(8)} ${[p.bh, p.bulge, p.disk, p.halo].map(x => f(x / v2, 3)).join(', ')}`); +} +const flat = g.flatRange(0.15, 1, 300); +console.log(` |dln v/dln R| < 0.15 from ${f(flat[0], 2)} to ${f(flat[1], 1)} kpc`); +const EILERS = { v0: 229.0, dv: -1.7, R0: 8.122, Rmin: 5, Rmax: 25 }; // Eilers et al. 2019 +const ve = R => EILERS.v0 + EILERS.dv * (R - EILERS.R0); +console.log(` Eilers et al. 2019: v = ${EILERS.v0} ${EILERS.dv} (R - ${EILERS.R0}) km/s for ${EILERS.Rmin} to ${EILERS.Rmax} kpc;` + + ` dln v/dln R = ${f(EILERS.dv * EILERS.Rmin / ve(EILERS.Rmin))} at ${EILERS.Rmin}, ${f(EILERS.dv * EILERS.R0 / ve(EILERS.R0))} at R0, ${f(EILERS.dv * EILERS.Rmax / ve(EILERS.Rmax))} at ${EILERS.Rmax} kpc`); +const Mgrow = ve(EILERS.Rmax) ** 2 * EILERS.Rmax / (ve(EILERS.Rmin) ** 2 * EILERS.Rmin); +console.log(` mass inside 25 kpc over mass inside 5 kpc, v^2 R / G (spherical): ${f(Mgrow, 2)}; for Kepler it would be 1`); +for (const Mb of [5e10, 6e10, 1e11]) console.log(` ${e3(Mb)} M_sun all inside 25 kpc, as a point mass: v(25 kpc) = ${f(Math.sqrt(C.G * Mb / 25), 0)} km/s (measured ${f(ve(25), 0)})`); +console.log(' (the stars: 5 +- 1 x 10^10 M_sun, Bland-Hawthorn & Gerhard 2016; 10^11 is twice that)'); +const gR0 = (ve(EILERS.R0) * 1e3) ** 2 / (EILERS.R0 * C.kpc); +console.log(` the Galaxy's pull at the Sun: v^2/R0 = ${e3(gR0)} m/s^2 = ${f(gR0 / O.GDAGGER, 2)} g+`); + +console.log('\n4b. What does repeat: thin disks\n'); +console.log(` inclinations to the invariable plane (JPL Table 1 elements; plane from Souami & Souchay 2012): ` + + O.PLANETS.map(p => `${p.name} ${f(O.tiltToInvariable(p), 3)}`).join(', ') + ' deg'); +const tilts = O.PLANETS.map(p => O.tiltToInvariable(p) * Math.PI / 180); +console.log(` rms sin(i) over the eight: ${f(Math.sqrt(tilts.reduce((a, t) => a + Math.sin(t) ** 2, 0) / 8), 3)}; without Mercury: ${f(Math.sqrt(tilts.slice(1).reduce((a, t) => a + Math.sin(t) ** 2, 0) / 7), 3)}`); +console.log(` the Galaxy's thin disk: scale height 300 +- 50 pc over scale length 2.6 +- 0.5 kpc = ${f(0.3 / 2.6, 3)} (Bland-Hawthorn & Gerhard 2016)`); + +if (!quick) { + console.log('\n5. Which orbits a body keeps: circular orbits followed for 20 turns of the host (Hill\'s problem)\n'); + const RHO0 = 0.10; // M_sun/pc^3, local mid-plane density (McKee, Parravano & Hollenbach 2015: 0.097 +- 0.013) + const sun = O.hostDisk(OORT.A, OORT.B, RHO0); + console.log(` point-mass host: Gam = 4A/Omega = ${O.HOST_KEPLER.Gam}, (nu/Omega)^2 = ${O.HOST_KEPLER.nu2}`); + console.log(` the Sun in the Galaxy (A, B as above, rho0 = ${RHO0} M_sun/pc^3): Gam = ${f(sun.Gam, 3)}, (nu/Omega)^2 = ${f(sun.nu2, 2)}`); + console.log(` 20 turns of the host: ${f(20 * 2 * Math.PI / ((OORT.A - OORT.B) * C.kmskpc) / C.yr / 1e9, 2)} Gyr for the Sun; ${f(20 * O.kepler(O.PLANETS[4]).P, 0)} yr for Jupiter`); + const planes = { 'prograde, in the plane': [[0, 0, 1]], 'retrograde, in the plane': [[0, 0, -1]], 'polar, plane holds the host': [[0, 1, 0]], 'polar, plane across it': [[1, 0, 0]] }; + const as = []; for (let a = 0.3; a <= 1.2001; a += 0.025) as.push(a); + console.log(' radius [r_J] where half the orbits (8 starting angles) have escaped beyond 3 r_J:'); + console.log(' orbit point-mass host the Sun in the Galaxy'); + for (const [k, n] of Object.entries(planes)) { + const e = [O.HOST_KEPLER, sun].map(h => O.hillScan(h, n, as, { nPhase: 8 }).edge); + console.log(` ${k.padEnd(28)} ${f(e[0], 3).padStart(10)} ${f(e[1], 3).padStart(10)}`); + } + const dirs = O.sphere(24), asIso = []; for (let a = 0.2; a <= 1.2001; a += 0.1) asIso.push(a); + console.log('\n orbits facing every way (24 directions x 4 starting angles): fractions that stay, escape, and plunge within 1e-3 r_J'); + const iso = {}; + for (const [name, h] of [['point-mass host', O.HOST_KEPLER], ['the Sun in the Galaxy', sun]]) { + iso[name] = O.hillScan(h, dirs, asIso, { nPhase: 4 }); + console.log(` ${name}: half have escaped by ${f(iso[name].edge, 3)} r_J`); + for (const r of iso[name].rows) console.log(` a = ${f(r.a, 1)} r_J: stay ${f(r.stays, 2)}, escape ${f(r.escapes, 2)}, plunge ${f(r.plunges, 2)}`); + } + console.log(` for the Sun: ${f(iso['the Sun in the Galaxy'].edge * rJ / C.au / 1e5, 2)}e5 au`); + for (const rho of [0.08, 0.12]) { + const e = O.hillScan(O.hostDisk(OORT.A, OORT.B, rho), dirs, asIso, { nPhase: 4 }).edge; + console.log(` with rho0 = ${rho}: (nu/Omega)^2 = ${f(O.hostDisk(OORT.A, OORT.B, rho).nu2, 2)}, half escaped by ${f(e, 3)} r_J`); + } +} diff --git a/tests/orbits.mjs b/tests/orbits.mjs new file mode 100644 index 0000000..bcf069a --- /dev/null +++ b/tests/orbits.mjs @@ -0,0 +1,95 @@ +// Checks the orbits study (orbits/model.js) against published values and its own closed forms. Run: npm test. +// Each check prints the source, the published or expected value, the model's value and the tolerance. "Arithmetic" and +// "model-internal" mark checks of the code against a closed form or itself, not against nature. Sources were read in +// search results and abstracts only (orbits/README.md, Sources), except where a check says otherwise. +import { createRequire } from 'node:module'; +const O = createRequire(import.meta.url)('../orbits/model.js'); +const { C } = O; + +let fails = 0; +const fmt = x => (Math.abs(x) >= 1e-3 && Math.abs(x) < 1e6 ? x.toPrecision(5) : x.toExponential(3)); +function near(name, got, want, tol, src) { + const ok = Number.isFinite(got) && Math.abs(got - want) <= tol; + console.log(`${ok ? 'PASS' : 'FAIL'} ${name}\n model ${fmt(got)} expected ${fmt(want)} +- ${fmt(tol)} ${src}`); + if (!ok) fails++; +} +function check(name, ok, detail, src) { + console.log(`${ok ? 'PASS' : 'FAIL'} ${name}\n ${detail} ${src}`); + if (!ok) fails++; +} + +// 1. Kepler's third law across the planets. +const JPL = 'JPL approximate elements, Table 1 (1800-2050); GM_sun IAU 2015 B3; masses DE430'; +const mu = O.planetMassFraction(), ex = O.PLANETS.map(p => O.kepler(p).excess), big = Math.max(...ex.map(Math.abs)); +check('n^2 a^3 = GM_sun for every planet, to within the planets\' total mass over the Sun\'s', ex.every(x => Math.abs(x) <= mu), + `largest |n^2 a^3 / GM_sun - 1| = ${fmt(big)} (${O.PLANETS[ex.map(Math.abs).indexOf(big)].name}); sum of m/M_sun = ${fmt(mu)}`, JPL); +near('Kepler\'s exponent: P ~ a^p over 0.39 to 30 au', O.keplerExponent(), 1.5, 1.5 * mu, `${JPL}; tolerance 3/2 times that mass fraction`); + +// 2. Where a Kepler domain ends. +const kk = O.oortKepler(1), kf = O.oortFlat(1); +near('Point-mass host: 4A/Omega = 3', O.densityRatio(kk.A, kk.B), 3, 1e-12, 'arithmetic'); +near('Flat rotation curve: 4A/Omega = 2', O.densityRatio(kf.A, kf.B), 2, 1e-12, 'arithmetic'); +{ + const a = O.PLANETS[4].a * C.au, q = 1 / O.PLANETS[4].sunOver, o = O.oortKepler(Math.sqrt(C.GMsun / a ** 3) / C.kmskpc); + near('Jacobi radius with point-mass Oort constants / Hill radius a (m/3M)^(1/3), for Jupiter', O.jacobi(C.GMsun * q, o.A, o.B) / O.hill(a, q), 1, 1e-12, 'model-internal'); +} +near('The Sun\'s tidal radius in the Galaxy [1e5 au], from Bovy 2017\'s A = 15.3, B = -11.9 km/s/kpc', O.jacobi(C.GMsun, 15.3, -11.9) / C.au / 1e5, 2.8, 0.05, + 'Correa-Otto & Calandra 2019 (MNRAS 490, 2495): "r_J ~ 2.8 x 10^5 au"; tolerance the rounding of 2.8'); + +// 3. Sgr A*: S2's orbit and the stars around it. +const G20 = 'GRAVITY Collaboration 2020 (A&A 636, L5)'; +const s2 = O.orbitOf(O.S2), s2M = O.orbitOf(O.S2, O.S2.M); +near('S2: Kepler\'s third law returns GRAVITY\'s fitted mass [1e6 M_sun]', s2.keplerMass / 1e6, O.S2.M / 1e6, 0.012, + `${G20}, Table E.1; consistency with their own fit (tolerance its 0.012 error), not a measurement`); +near('S2: Schwarzschild precession per orbit [arcmin]', s2M.precessionArcmin, 12.1, 0.1, `${G20}, appendix B: "the 12.1' precession"`); +near('S2: pericentre distance [au]', s2M.periAU, 120, 5, `${G20}, abstract: "pericentre at 120 AU", rounded to 10 au`); +near('S2: speed at pericentre [km/s]', s2M.periSpeed, 7700, 50, `${G20}, abstract: "7700 km s^-1", rounded to 100 km/s`); +{ + const K = O.CUSP, M = O.SGRA.M, r2 = [K.M1hi, K.M1, K.M1lo].map(m => O.cuspRadius(2, M, m)); + check('The stars inside r weigh twice Sgr A* at 3 to 4 pc in the literature; the cusp model brackets it', + r2[0] <= 3 && r2[2] >= 4, + `model ${fmt(r2[0])} to ${fmt(r2[2])} pc (middle ${fmt(r2[1])}) for 0.5-1.5e6 M_sun inside 1 pc and density ~ r^-1.13`, + 'Alexander 2005 (3 pc), Feldmeier-Krause et al. 2017 (about 4 pc), Merritt 2013 (3.9 pc); inputs Schodel et al. 2009, 2018 (search text)'); + near('Kepler inside 0.1 pc of Sgr A*: dln v/dln r', O.cuspSlope(0.1, M), -0.5, 0.01, 'model (Sgr A* plus the cusp)'); +} + +// 4. The Galaxy. +const g = O.galaxy(), g0 = O.galaxy(O.MWP14, { bh: 0 }); +near('MWPotential2014: v_c(8 kpc) without Sgr A* [km/s]', g0.v(8), 220, 1e-9, 'model-internal (the normalisation; parameters as in galpy\'s source)'); +{ + const Rs = [5, 8.122, 15, 25], ve = R => 229.0 - 1.7 * (R - 8.122); + const rel = Rs.map(R => (g.v(R) / g.v(8.122)) / (ve(R) / ve(8.122)) - 1); + check('The shape of MWPotential2014\'s curve, v(R)/v(R0), within 5% of Eilers et al.\'s from 5 to 25 kpc', rel.every(r => Math.abs(r) < 0.05), + `ratio to Eilers - 1 = ${rel.map(r => r.toFixed(3)).join(', ')} at ${Rs.join(', ')} kpc; the level differs by ${(g.v(8.122) / 229 - 1).toFixed(3)} (220 against 229 km/s)`, + 'Eilers et al. 2019 (ApJ 871, 120), linear fit; tolerance their 2-5% systematic uncertainty'); + const sl = Rs.map(R => g.slope(R)), se = Rs.map(R => -1.7 * R / ve(R)); + check('Flat, not Kepler, from 5 to 25 kpc: |dln v/dln R| < 0.25 in the model and in Eilers et al.\'s fit', + sl.every(s => Math.abs(s) < 0.25) && se.every(s => Math.abs(s) < 0.25), + `model ${sl.map(s => s.toFixed(3)).join(', ')}; Eilers ${se.map(s => s.toFixed(3)).join(', ')}; Kepler -0.5`, 'Eilers et al. 2019; model'); +} + +// 5. Thin disks: the planets' tilts to the invariable plane. +const SS = 'Souami & Souchay 2012 (A&A 543, A133), as quoted in search text; tolerance covers epoch and element differences'; +near('Jupiter\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O.PLANETS[4]), 0.3219, 0.01, SS); +near('Saturn\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O.PLANETS[5]), 0.9254, 0.01, SS); + +// 6. Hill's problem: which orbits a body keeps (20 turns of the host, 8 starting angles each). +{ + const K = O.HOST_KEPLER, fates = (h, a, n) => O.hillFates(h, a, [n], { nPhase: 8 }).escapes; + const pro = [fates(K, 0.4, [0, 0, 1]), fates(K, 0.6, [0, 0, 1])], ret = [fates(K, 0.6, [0, 0, -1]), fates(K, 0.9, [0, 0, -1])]; + check('Point-mass host: prograde circular orbits kept at 0.4 r_H, lost by 0.6 (published limit about 0.5)', pro[0] === 0 && pro[1] >= 0.5, + `fraction escaped ${pro[0].toFixed(2)} at 0.4 r_H, ${pro[1].toFixed(2)} at 0.6 r_H`, + 'Shen & Tremaine 2008 (about 0.5 r_H), Hamilton & Krivov 1997 (0.53 r_H): search text'); + check('Point-mass host: retrograde orbits kept at 0.6 r_H, lost by 0.9 (published limit about 0.7)', ret[0] === 0 && ret[1] >= 0.5, + `fraction escaped ${ret[0].toFixed(2)} at 0.6 r_H, ${ret[1].toFixed(2)} at 0.9 r_H`, + 'Shen & Tremaine 2008 (about 0.7 r_H), Hamilton & Krivov 1997 (0.69 r_H): search text'); + const sun = O.hostDisk(15.3, -11.9, 0.10); + near('The Sun in the Galaxy: (vertical tidal frequency / Omega)^2', sun.nu2, (4 * Math.PI * C.G * 0.1e9 + 2 * (15.3 ** 2 - 11.9 ** 2)) / 27.2 ** 2, 1e-12, 'arithmetic (Poisson\'s equation)'); + const dirs = O.sphere(12), iso = [0.3, 0.6].map(a => O.hillFates(sun, a, dirs, { nPhase: 2 }).escapes); + check('The Sun keeps orbits facing every way at 0.3 r_J (0.85e5 au) and loses most by 0.6 r_J (1.7e5 au)', iso[0] === 0 && iso[1] >= 0.5, + `fraction escaped ${iso[0].toFixed(2)} at 0.3 r_J, ${iso[1].toFixed(2)} at 0.6 r_J (12 directions x 2 angles)`, + 'model; the Oort cloud\'s outer edge is quoted at 1 to 2 x 10^5 au (Dones et al. 2004, as cited)'); +} + +console.log(fails ? `\n${fails} check(s) failed` : '\nall orbits checks pass'); +process.exit(fails ? 1 : 0); From d6708536a88067fa32a65114b02b1be6b3450767 Mon Sep 17 00:00:00 2001 From: Claude Date: Mon, 19 Oct 2026 11:59:08 +0000 Subject: [PATCH 02/10] Orbits study: fixes from the pre-PR review These come from three reviewers with refuters (physics, claims against output, tests and code). - Blocking: the vertical tide cannot act on orbits that start in the plane. In-plane retrograde orbits last to 0.925 r_J because Gam = 4A/Omega is 2.25, not 3. The table gains a column with (nu/Omega)^2 = 1, which shows this. - Blocking: the Kepler-exponent check's tolerance is now the bound the excess check implies, 4.1e-4, and is labelled model-internal. - Kepler's law: - n = Ldot is stated as the angular rate (Ldot - wdot would give Neptune 4.2e-3). - The planets' total mass is an observed size, not a bound. - A ring-averaged secular estimate is added (ringExcess) and printed. - The great inequality is named as a long-period term the fit carries. - Density ratio: the comparison is to the host's mean density inside the orbit (spherical), not the local density. 4A/Omega = 2(1 - dln v/dln R). - Sgr A*: - The short answer gives the slope -1/4 radius, 1.0 to 1.8 pc, and the equal-mass radius separately. - Cusp radii past the 3 pc break are flagged as lower limits. - The 3.9 pc is credited to GRAVITY 2024, using Merritt's definition, and Feldmeier-Krause's black-hole mass is noted. - The Kepler region is "about 1.3 pc in radius", and it is the Milky Way's, not every galaxy's. - Oort cloud: its edge is attributed to Correa-Otto & Calandra (near 1e5 au) and Dones et al. (as far as 2e5 au), which is 0.35 to 0.7 r_J. A new check compares the model's edge with that range. - Tests: - A Jacobi-integral conservation check is added. - A retrograde check that distinguishes the hosts is added. - nu2 is computed in SI, independently of C.G. - 4A/Omega is taken from the model's own curve. - A hillScan edge is compared with Shen & Tremaine. - The unverified Hamilton & Krivov values are dropped. - S2's pericentre values are labelled as from the introduction. - The 0.3 r_J check prints its plunges. - Table: - A long mode prints the 100-turn runs. - New printed numbers: the vertical period, the cusp density ratio, the Eilers comparison numbers and the bulge's flat run. - Neptune's inside-mass is fixed. - The A, B errors now include the systematic. - Code: dead keplerEdge and rar are removed; flatRange defaults to outside the bulge; hillOrbit returns its final state. - Wording: the scale height is written as an axis ratio, and the Eilers speeds and the root bullet are made precise. Co-Authored-By: Claude Opus 5.5 Claude-Session: https://claude.ai/code/session_01C2bpiHyHSkC8DAL5CcyciK --- README.md | 10 +-- orbits/README.md | 198 +++++++++++++++++++++++++++-------------------- orbits/model.js | 63 ++++++++++----- orbits/table.mjs | 72 +++++++++++------ tests/orbits.mjs | 95 +++++++++++++++-------- 5 files changed, 274 insertions(+), 164 deletions(-) diff --git a/README.md b/README.md index 079aead..fb22373 100644 --- a/README.md +++ b/README.md @@ -19,11 +19,11 @@ with Claude (Anthropic), for [Foundation0](https://github.com/Foundation-0/f0). planets on orbits. The same equation, a wave held by a pull that weakens with distance, gives the same solution at any scale. The nesting lives in the equations, not the objects. - **Solar systems are not little galaxies, except at a galaxy's centre.** Around one dominant mass, orbits follow - Kepler's law at every radius: the planets match it to 0.13%, the size of their own masses. Across the Milky Way's - disk the speed instead stays near 230 km/s, because the mass is spread out. A Kepler region ends where its host's - tide wins, at a density ratio rather than a size: 1.4 pc for the Sun in the Galaxy, a third of an au for Jupiter. - A planet keeps moons, and the Sun comets, out to about half that radius. The Galaxy has its own small solar system - around its central black hole, a parsec or two across ([orbits](orbits/)). + Kepler's law at every radius. The planets match it to 0.13%, about what they weigh together against the Sun. Across + the Milky Way's disk the speed instead stays between 200 and 235 km/s, because the mass is spread out. A Kepler + region ends where its host's tide wins, at a density ratio rather than a size: 1.4 pc for the Sun in the Galaxy, + a third of an au for Jupiter. A planet keeps moons, and the Sun comets, out to about half that radius. The Milky + Way has its own small solar system around its central black hole, about a parsec in radius ([orbits](orbits/)). - **Quantum paths are fractal, between two lengths.** A free particle's path, recorded by position measurements finer than ħ/p, has fractal dimension 2, against 1 for a classical trajectory (Abbott and Wise, 1981); below ħ/mc, in a single-particle model, the speed of light makes it smooth again. Its surroundings draw diff --git a/orbits/README.md b/orbits/README.md index f4c0a51..df5cb5e 100644 --- a/orbits/README.md +++ b/orbits/README.md @@ -12,35 +12,42 @@ what ends it, and what, if anything, repeats from one scale to the next. | Question | Answer | |---|---| -| Is a galaxy a big solar system? | No. Around one dominant mass, orbital speed falls as Kepler's v ∝ r^−1/2. Across the Milky Way's disk the speed stays near 230 km/s. From 5 to 25 kpc the mass inside grows 3.7-fold (Eilers et al. 2019), where a Kepler system's would stay fixed. | -| How exact is Kepler's law in the Solar System? | For all eight planets n²a³ matches GM☉ to within 1.3×10⁻³, below the planets' total mass of 1.34×10⁻³ of the Sun's. P ∝ a^1.49986 over 0.39 to 30 au. The mass inside Neptune's orbit is 0.13% more than the mass inside Mercury's. | -| Does the Galaxy have a Kepler region? | Yes, around Sgr A*. The star S2 orbits it as a planet orbits the Sun, with general relativity's 12.1′ precession per orbit on top. The region ends 1 to 3 pc out, where the stars around the black hole begin to weigh as much as it does. | -| What ends a Kepler region? | The host's tide. A body keeps satellites only inside its tidal radius, r_J³ = Gm/(4A(A − B)). There, the body's mean density is 4A/Ω times the host's mean density inside the orbit: 3 for a planet around the Sun, 2.25 for the Sun in the Galaxy. For the Sun, r_J = 1.37 pc = 2.8×10⁵ au. | -| How much of that region is used? | In this study's three-body runs (circular starting orbits, all orientations), about half: 0.53 r_J around a planet and 0.45 r_J for the Sun, which is 1.3×10⁵ au. Measured edges are about 0.4 to 0.6 r_H for the giant planets' outer moons and 1 to 2×10⁵ au for the Oort cloud. | -| What does repeat? | Kepler regions nested inside one another, each ending at its host's tide at a similar fraction of its tidal radius. Also flat disks: the planets' orbits are tilted by 2° on average (6° for Mercury), and the Galaxy's thin disk is about 0.1 as thick as it is wide. | +| Is a galaxy a big solar system? | No. Around one dominant mass, orbital speed falls as Kepler's v ∝ r^−1/2. Across the Milky Way's disk the speed falls only from 234 to 200 km/s between 5 and 25 kpc. Read as spherical, the mass inside grows 3.7-fold over that range, where a Kepler system's would stay fixed. Both figures come from Eilers et al.'s (2019) fit. | +| How exact is Kepler's law in the Solar System? | In JPL's 1800–2050 fit, n²a³ matches GM☉ to within 1.3×10⁻³ for all eight planets. That is about the planets' total mass, 1.34×10⁻³ of the Sun's: the planets pull on one another. P ∝ a^1.49986 over 0.39 to 30 au. | +| Does the Galaxy have a Kepler region? | Yes, around Sgr A*. The star S2 orbits it as a planet orbits the Sun, plus general relativity's 12.1′ precession per orbit. The region ends about 1.3 pc out, where the speed's log slope passes −1/4 (1.0 to 1.8 pc across the cluster-mass range). The stars inside weigh as much as the black hole by 1.8 to 3.2 pc. | +| What ends a Kepler region? | For a body orbiting in a host (a planet, the Sun), the host's tide. Satellites are held only inside its tidal radius, r_J³ = Gm/(4A(A − B)). Inside it, the body's mean density is 4A/Ω times the host's mean density inside the orbit (host taken as spherical): 3 for a planet around the Sun, 2.25 for the Sun in the Galaxy. For the Sun, r_J = 1.37 pc = 2.8×10⁵ au. At the Galaxy's centre, where Sgr A* sits still, the region ends instead where the surrounding stars' mass approaches the black hole's. | +| How much of that region is used? | About half, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.53 r_J around a planet and by 0.45 r_J (1.3×10⁵ au) for the Sun. Shen and Tremaine (2008) put stable prograde and retrograde moons out to about 0.5 and 0.7 r_H; the observed outer moons reach about 0.4 and 0.6 r_H. The Oort cloud's outer edge is placed between 10⁵ and 2×10⁵ au, which is 0.35 to 0.7 r_J. | +| What does repeat? | Kepler regions nested inside one another, each ending at its host's tide at a similar fraction of its tidal radius: moons in a planet's, planets and comets in the Sun's. Also flat disks: the planets' orbits are tilted by 2° on average (6° for Mercury), and the Galaxy's thin disk is about 0.1 as thick as it is wide. | ## Kepler's law in the Solar System Around one mass M, a circular orbit of radius r has speed √(GM/r), and P² = 4π²a³/G(M + m). Nothing in the law sets a size, so it holds from the Sun's surface out to wherever another pull competes. -JPL's approximate elements (Standish, Table 1, a fit to 1800–2050) give each planet's semi-major axis a and mean -motion n. Against the IAU 2015 nominal GM☉ (`node orbits/table.mjs`, section 1): +JPL's approximate elements (Standish, Table 1, a fit to 1800–2050) give each planet's semi-major axis a and its +mean-longitude rate, n = L̇: +- This is the orbit's angular rate, which obeys n²a³ = GM( 1 kpc), its log slope stays within ±0.15 of zero from 3.6 to 49 kpc. The r^−1.8 bulge is about as flat, from 7 to 180 pc, but that is a dense bulge, not a single mass. -Going outward, the model's speed falls by less than Kepler's law would give. The log slope is −0.21 at 100 kpc -and −0.27 at 200 kpc, because the NFW halo's mass still grows. Gaia DR3 analyses differ on whether the measured -curve turns down past about 19 kpc: -- Jiao et al. (2023) report "a Keplerian decline". -- Koop et al. (2024) find the curves robust only to about 15 kpc. +Going outward, the model's speed falls by less than Kepler's law would give: +- **The model.** The log slope is −0.21 at 100 kpc and −0.27 at 200 kpc, because the NFW halo's mass still grows. +- **The data.** Gaia DR3 analyses differ on whether the measured curve turns down past about 19 kpc. Jiao et al. (2023) report "a Keplerian decline"; Koop et al. (2024) find the curves robust only to about 15 kpc. ## An acceleration scale, and why this study does not use it @@ -167,45 +186,51 @@ total angular momentum (Souami and Souchay 2012). Computed from JPL's elements, - Mercury 6.3°, Venus 2.2°, Earth–Moon 1.6°, Mars 1.7°; - Jupiter 0.33°, Saturn 0.93°, Uranus 1.0°, Neptune 0.73°. -The rms of sin i is 0.045, or 0.023 without Mercury. The Galaxy's thin disk is 300 ± 50 pc thick over a scale -length of 2.6 ± 0.5 kpc, a ratio of 0.12 (Bland-Hawthorn and Gerhard 2016). Both are flat for the same reason: gas -that radiates its energy but keeps its angular momentum settles into a disk. What the disks orbit differs: one -dominant mass, or mass spread through the disk and halo. +The rms of sin i is 0.045, or 0.023 without Mercury. The Galaxy's thin disk has a scale height of 300 ± 50 pc over +a scale length of 2.6 ± 0.5 kpc (Bland-Hawthorn and Gerhard 2016). Their ratio, 0.12, is the axis ratio of its +density contours. + +Both are flat for the same reason: gas that radiates its energy but keeps its angular momentum settles into a disk. +What the disks orbit differs: one dominant mass, or mass spread through the disk and halo. ## Run it ``` -node orbits/table.mjs # every number above (about 45 s; the stability scans take nearly all of it) +node orbits/table.mjs # every number above except the 100-turn runs (about 50 s; section 5 takes nearly all of it) node orbits/table.mjs quick # sections 1-4 only, well under a second -node tests/orbits.mjs # the checks, about 1 s (also run by npm test) +node orbits/table.mjs long # adds the 100-turn runs, about 2 minutes in all +node tests/orbits.mjs # the checks, about 2 s (also run by npm test) ``` ## Checks -`tests/orbits.mjs` has 21 checks. Each prints its source, the published or expected value, the model's value and +`tests/orbits.mjs` has 23 checks. Each prints its source, the published or expected value, the model's value and the tolerance. | What | Against | Kind | |---|---|---| -| n²a³/GM☉ − 1 below the planets' mass fraction, for all eight; exponent 3/2 | JPL Table 1, IAU 2015, DE430 | published elements | -| 4A/Ω = 3 and 2; Jacobi radius = Hill radius for a point mass | closed forms | arithmetic, model-internal | +| In the 1800–2050 fit, n²a³/GM☉ − 1 below the planets' mass fraction for all eight | JPL Table 1, IAU 2015, DE430 | published elements; an observed size, not a bound | +| Exponent 3/2, to the most the previous check allows | the same | model-internal, independent of GM☉ | +| 4A/Ω = 3 from the model's own curve near Sgr A*; Jacobi radius = Hill radius for a point mass | the model | model-internal | | The Sun's r_J = 2.8×10⁵ au | Correa-Otto and Calandra 2019 | search text | -| S2: Kepler mass, 12.1′ precession, 120 au, 7,700 km/s | GRAVITY 2020 | abstract and appendix, via a transcription | -| Stars weigh 2 M_BH at 3 to 4 pc | Alexander 2005; Feldmeier-Krause et al. 2017; Merritt 2013 | search text | -| Kepler slope −1/2 inside 0.1 pc | model | model | -| MWPotential2014: 220 km/s at 8 kpc; shape within 5% of Eilers et al.; flat (|slope| < 0.25) at 5 to 25 kpc | Eilers et al. 2019 | normalisation; abstract | +| S2: Kepler mass, 12.1′ precession, 120 au, 7,700 km/s | GRAVITY 2020 | Table E.1, appendix B and introduction, via a transcription | +| Stars weigh 2 M_BH at 3 to 4 pc | Alexander 2005; Feldmeier-Krause et al. 2017; GRAVITY 2024 | search text | +| Kepler slope −1/2 inside 0.1 pc | — | model | +| MWPotential2014: the 220 km/s normalisation; shape within 5% of Eilers et al.; flat (abs. slope < 0.25) at 5 to 25 kpc | Eilers et al. 2019 | model-internal; abstract | | Jupiter's and Saturn's tilts to the invariable plane | Souami and Souchay 2012 | search text | -| Planet: prograde kept at 0.4 r_H, lost by 0.6; retrograde kept at 0.6, lost by 0.9 | Shen and Tremaine 2008 (0.5, 0.7) | abstract | -| The Galaxy's ν²/Ω² | Poisson's equation | arithmetic | -| The Sun keeps every orientation at 0.3 r_J, loses most by 0.6 | Oort cloud edge 1 to 2×10⁵ au | model; as cited | +| Planet: half the prograde orbits escaped by 0.5 ± 0.1 r_H; retrograde kept at 0.6, lost by 0.9 | Shen and Tremaine 2008 (0.5, 0.7) | abstract, search text | +| Retrograde in the plane at 0.875 r_J: lost around a point mass, kept by the Sun | — | model | +| The Jacobi integral conserved over 20 turns; the Galaxy's ν²/Ω² in SI | equations of motion; Poisson's equation | model-internal; arithmetic | +| The Sun: none escape from 0.3 r_J, most by 0.6 | — | model | +| The Sun's all-directions edge within the Oort cloud's 1 to 2×10⁵ au | Correa-Otto and Calandra 2019; Dones et al. 2004 | search text; as cited | ## What the model leaves out -- **The planets' pull on one another.** Kepler's law is checked only to the size of the planets' masses. -- **Hill's problem is the tide to first order, on a circular host orbit.** The study drops the Sun's vertical oscillation through the disk (a period of about 80 Myr with this model's ν) and its radial epicycles. It also leaves out passing stars and molecular clouds, which also strip the Oort cloud, and the planets' scattering of comets that plunge inward. +- **The planets' long-period pull on one another.** Kepler's law is checked only to the size of the planets' masses. +- **Hill's problem is the tide to first order, on a circular host orbit.** The study drops the Sun's vertical oscillation through the disk (82 Myr with this model's ν) and its radial epicycles. It also leaves out passing stars and molecular clouds, which also strip the Oort cloud, and the planets' scattering of comets that plunge inward. - **Circular starting orbits only.** Real moons and comets start on eccentric orbits. The edges are set at half the orbits escaping within 20 turns of the host, on a grid of 0.025 r_J (in the plane) or 0.1 r_J (all directions). - **MWPotential2014 is one model among many.** Its halo sets the flat range, and it has no nuclear star cluster. The Galactic-centre numbers use the separate power-law cusp above, which is valid only inside its 3 pc break. -- **Spherical mass estimates.** v²R/G treats a flattened galaxy as spherical. The disk changes the numbers by tens of percent, not the 3.7-fold growth. +- **Spherical mass estimates.** v²R/G and 3Ω²/4πG treat a flattened galaxy as spherical. For a disk the true enclosed mass and mean density differ somewhat, which shifts the 3.7-fold growth and the 2.25, but not the contrast with Kepler's fixed mass. ## Sources @@ -217,18 +242,19 @@ Read here only through search results and abstracts, or as noted. Each entry say - D. Souami and J. Souchay, *The solar system's invariable plane*, A&A 543, A133 (2012). https://doi.org/10.1051/0004-6361/201219011. Search text. - J. Binney and S. Tremaine, *Galactic Dynamics*, 2nd ed., Princeton (2008), sec. 8.3. From memory. - J. Bovy, *Galactic rotation in Gaia DR1*, MNRAS 468, L63 (2017). https://arxiv.org/abs/1610.07610. Abstract, search text. -- J. A. Correa-Otto and M. F. Calandra, MNRAS 490, 2495 (2019). https://doi.org/10.1093/mnras/stz2671. Search text. +- J. A. Correa-Otto and M. F. Calandra, MNRAS 490, 2495 (2019). https://doi.org/10.1093/mnras/stz2671. Search text: "r_J ~ 2.8 × 10^5 au", and a structure near 10⁵ au marking an outer stability boundary. - Y. Shen and S. Tremaine, *Stability of Distant Satellites of the Giant Planets in the Solar System*, AJ (2008). https://arxiv.org/abs/0806.2882. Abstract and introduction, search text. - C. F. McKee, A. Parravano and D. J. Hollenbach, *Stars, gas, and dark matter in the solar neighborhood*, ApJ 814, 13 (2015). https://arxiv.org/abs/1509.05334. Search text, secondary. - M. Heisler and S. Tremaine, *The influence of the Galactic tidal field on the Oort comet cloud*, Icarus 65, 13 (1986). Abstract, search text. -- L. Dones, P. R. Weissman, H. F. Levison and M. J. Duncan, *Oort cloud formation and dynamics*, in *Comets II*, 153 (2004). As cited in search text. -- GRAVITY Collaboration, *Detection of the Schwarzschild precession in the orbit of the star S2 near the Galactic centre massive black hole*, A&A 636, L5 (2020). https://doi.org/10.1051/0004-6361/202037813. Abstract, appendix B and Table E.1, read in a transcription. +- L. Dones, P. R. Weissman, H. F. Levison and M. J. Duncan, *Oort cloud formation and dynamics*, in *Comets II*, 153 (2004). As cited in search text, for an outer edge "as far as 2 × 10^5 AU". +- GRAVITY Collaboration, *Detection of the Schwarzschild precession in the orbit of the star S2 near the Galactic centre massive black hole*, A&A 636, L5 (2020). https://doi.org/10.1051/0004-6361/202037813. Abstract, introduction, appendix B and Table E.1, read in a transcription. - GRAVITY Collaboration, A&A 657, L12 (2022). https://doi.org/10.1051/0004-6361/202142465. Search text. +- GRAVITY Collaboration (2024). https://arxiv.org/abs/2409.12261. Search text, for the 3.9 pc influence radius in their models. - R. Schödel, D. Merritt and A. Eckart, A&A 502, 91 (2009). https://doi.org/10.1051/0004-6361/200810922. Abstract, search text. - R. Schödel et al., A&A 609, A27 (2018). Abstract, search text. - T. Alexander, Phys. Rep. 419, 65 (2005). As cited by Feldmeier-Krause et al. 2017. -- A. Feldmeier-Krause et al., MNRAS 466, 4040 (2017). https://arxiv.org/abs/1701.01583. Search text. -- D. Merritt, *Dynamics and Evolution of Galactic Nuclei*, Princeton (2013). As cited by GRAVITY Collaboration 2024 (https://arxiv.org/abs/2409.12261). +- A. Feldmeier-Krause et al., *Triaxial orbit-based modelling of the Milky Way Nuclear Star Cluster*, MNRAS 466, 4040 (2017). https://arxiv.org/abs/1701.01583. Search text. +- D. Merritt, *Dynamics and Evolution of Galactic Nuclei*, Princeton (2013). Definition only, as cited by GRAVITY Collaboration 2024. - J. Bovy, *galpy: A Python Library for Galactic Dynamics*, ApJS 216, 29 (2015). https://doi.org/10.1088/0067-0049/216/2/29. MWPotential2014's parameters read in galpy's source. - C. Eilers, D. W. Hogg, H.-W. Rix and M. K. Ness, ApJ 871, 120 (2019). https://doi.org/10.3847/1538-4357/aaf648. Abstract, search text. - J. Bland-Hawthorn and O. Gerhard, *The Galaxy in Context*, ARA&A 54, 529 (2016). https://doi.org/10.1146/annurev-astro-081915-023441. As cited in search text. diff --git a/orbits/model.js b/orbits/model.js index 78cb5b1..4aa6d35 100644 --- a/orbits/model.js +++ b/orbits/model.js @@ -13,7 +13,9 @@ * with A and B the host's Oort constants at the orbit. For a point-mass host A = 3 Omega / 4 and B = -Omega / 4, and * r_J is Hill's radius a (m / 3M)^(1/3). For a flat rotation curve A = -B = Omega / 2. Either way the mean density of * m inside r_J, 3m / (4 pi r_J^3), equals 4A/Omega times the host's mean density inside the orbit, 3 Omega^2 / (4 pi G) - * (spherical host): 3 times for a point mass, 2 for a flat curve. So the edge is set by a density ratio, not a size. + * (spherical host): 3 times for a point mass, 2 for a flat curve, and in general 2(1 - dln v/dln R). So the edge is set + * by a ratio of mean densities, not a size. The host's density here is its mean inside the orbit, not its local + * density. * * Flat curves. Where the enclosed mass grows as M( PLANETS.reduce((s, p) => s + 1 / p.sunOver, 0); // Kepler's third law for one planet: the GM its orbit implies, n^2 a^3, against GM_sun, with n the mean-longitude - // rate. A 250-year fit to a perturbed orbit, so the excess carries the other planets' pull as well as the planet's - // own mass. + // rate Ldot: the orbit's angular rate, which obeys n^2 a^3 = G M( 1e-16 * a; i++) { const an = (a + b) / 2; b = Math.sqrt(a * b); a = an; } + return Math.PI / (2 * a); + } + const ringPotential = (r, b) => -(2 / Math.PI) * ellipticK(2 * Math.sqrt(r * b) / (r + b)) / (r + b); // per unit G m + function ringExcess(p) { + let ex = 1 / p.sunOver; + for (const q of PLANETS) { + if (q === p) continue; + const h = 1e-6 * p.a, pull = (ringPotential(p.a + h, q.a) - ringPotential(p.a - h, q.a)) / (2 * h); // inward > 0 + ex += p.a * p.a * pull / q.sunOver; + } + return ex; + } + // Least-squares slope of log P against log a: 3/2 for Kepler. function keplerExponent(list = PLANETS) { const xs = list.map(p => Math.log(p.a)), ys = list.map(p => Math.log(kepler(p).P)); @@ -88,7 +112,8 @@ // Oort constants of a point-mass host and of a flat rotation curve, at angular speed Omega. const oortKepler = Omega => ({ A: 0.75 * Omega, B: -0.25 * Omega }); const oortFlat = Omega => ({ A: 0.5 * Omega, B: -0.5 * Omega }); - // The body's mean density inside r_J over the host's mean density inside the orbit (spherical host): 4A/Omega. + // The body's mean density inside r_J over the host's mean density inside the orbit (spherical host): 4A/Omega, + // which is 2(1 - dln v/dln R) for a rotation curve v(R). const densityRatio = (A, B) => 4 * A / (A - B); // Where a mass's own pull falls to an acceleration g: r = sqrt(GM/g) [m]. const radiusAtAcceleration = (GM, g) => Math.sqrt(GM / g); @@ -113,7 +138,9 @@ // Eckart 2009). Radii in pc. const CUSP = { M1: 1.0e6, M1lo: 0.5e6, M1hi: 1.5e6, gamma: 1.13, rBreak: 3 }; const cuspMass = (r, M1 = CUSP.M1, gamma = CUSP.gamma) => M1 * Math.pow(r, 3 - gamma); - // The radius where the stars inside weigh k times Sgr A* (k = 1, or k = 2 for the usual influence radius). + // The radius where the stars inside weigh k times Sgr A* (k = 1, the equal-mass radius; k = 2, the usual influence + // radius). Radii beyond rBreak extrapolate the power law past the break, where the density falls faster, so they are + // lower limits. const cuspRadius = (k, Mbh, M1 = CUSP.M1, gamma = CUSP.gamma) => Math.pow(k * Mbh / M1, 1 / (3 - gamma)); // dln v / dln r for Sgr A* plus the cusp: (dln M / dln r - 1) / 2. const cuspSlope = (r, Mbh, M1 = CUSP.M1, gamma = CUSP.gamma) => { const m = cuspMass(r, M1, gamma); return ((3 - gamma) * m / (Mbh + m) - 1) / 2; }; @@ -188,11 +215,11 @@ }; return { par, M, parts, v, slope, bulgeMass, cross, - // where the stars inside r first weigh as much as Sgr A* (the influence radius), and where v's log slope is -1/4 - influence: () => cross(r => bulgeMass(r) - M.bh, 1e-6, 1), - keplerEdge: () => cross(r => slope(r) + 0.25, 1e-6, 1), - // the longest run of radii (kpc, on a grid 0.5% apart) where |dln v / dln R| < tol - flatRange: (tol = 0.15, rMin = 1e-3, rMax = 300) => { + // where the bulge's stars inside r first weigh k times Sgr A* (k = 1, equal mass; k = 2, the influence radius) + influence: (k = 1) => cross(r => bulgeMass(r) - k * M.bh, 1e-6, 1), + // the longest run of radii (kpc, on a grid 0.5% apart) where |dln v / dln R| < tol; by default outside the bulge + // (R > 1 kpc), since the r^-1.8 bulge is also flat to +-0.15 from about 7 to 180 pc + flatRange: (tol = 0.15, rMin = 1, rMax = 300) => { let best = [NaN, NaN], start = NaN, last = NaN; for (let r = rMin; r <= rMax * 1.0001; r *= 1.005) { if (Math.abs(slope(r)) < tol) { if (Number.isNaN(start)) start = r; last = r; } @@ -234,7 +261,8 @@ e: [71 / 57600, 0, -71 / 16695, 71 / 1920, -17253 / 339200, 22 / 525, -1 / 40], }; // Follows one orbit for `orbits` turns of the host (time 2 pi orbits). Fate: 'escapes' once r > rEsc, 'plunges' - // once r < rIn (into the inner system: the planet itself, or the Sun's planets), otherwise 'stays'. + // once r < rIn (1e-3 r_J is about a giant planet's own radius; for the Sun it is 280 au, outside the planets), + // otherwise 'stays'. Returns the final state s too. function hillOrbit(host, s0, { orbits = 20, rEsc = 3, rIn = 1e-3, tol = 1e-10 } = {}) { const T = 2 * Math.PI * orbits, k = Array.from({ length: 7 }, () => new Float64Array(6)); const y = new Float64Array(6), s = Float64Array.from(s0); @@ -255,12 +283,12 @@ t += h; steps++; for (let i = 0; i < 6; i++) { s[i] = y[i]; k[0][i] = k[6][i]; } const r = Math.hypot(s[0], s[1], s[2]); - if (r > rEsc) return { fate: 'escapes', t, steps }; - if (r < rIn) return { fate: 'plunges', t, steps }; + if (r > rEsc) return { fate: 'escapes', t, steps, s: Array.from(s) }; + if (r < rIn) return { fate: 'plunges', t, steps, s: Array.from(s) }; } h *= Math.min(5, Math.max(0.2, 0.9 * Math.pow(err, -0.2))); } - return { fate: 'stays', t, steps }; + return { fate: 'stays', t, steps, s: Array.from(s) }; } // A circular orbit of radius a (in r_J) about the body, at the two-body speed, in the plane with unit normal n // (rotating-frame axes; n = +z is prograde, -z retrograde), starting at angle phi. @@ -302,13 +330,12 @@ return { rows, edge }; } - // McGaugh, Lelli & Schombert 2016: the radial acceleration relation, g_obs = g_bar / (1 - exp(-sqrt(g_bar / g+))). + // McGaugh, Lelli & Schombert 2016: the acceleration where the radial acceleration relation bends. const GDAGGER = 1.20e-10; // m s^-2 - const rar = (gbar, gd = GDAGGER) => gbar / (1 - Math.exp(-Math.sqrt(gbar / gd))); const api = { - C, PLANETS, INVARIABLE, tiltToInvariable, planetMassFraction, kepler, keplerExponent, jacobi, hill, oortKepler, oortFlat, densityRatio, - radiusAtAcceleration, S2, orbitOf, CUSP, cuspMass, cuspRadius, cuspSlope, lgamma, gammaP, MWP14, SGRA, galaxy, GDAGGER, rar, + C, PLANETS, INVARIABLE, tiltToInvariable, planetMassFraction, kepler, ellipticK, ringExcess, keplerExponent, jacobi, hill, oortKepler, oortFlat, densityRatio, + radiusAtAcceleration, S2, orbitOf, CUSP, cuspMass, cuspRadius, cuspSlope, lgamma, gammaP, MWP14, SGRA, galaxy, GDAGGER, HOST_KEPLER, hostDisk, hillOrbit, hillStart, sphere, hillFates, hillScan, }; if (typeof module !== 'undefined' && module.exports) module.exports = api; else root.ORB = api; diff --git a/orbits/table.mjs b/orbits/table.mjs index 6768445..9e6eaca 100644 --- a/orbits/table.mjs +++ b/orbits/table.mjs @@ -1,34 +1,42 @@ -// Prints every number in orbits/README.md. Run: node orbits/table.mjs (about a minute; the stability scans of -// section 5 take nearly all of it), or node orbits/table.mjs quick (well under a second; section 5 left out). +// Prints every number in orbits/README.md. Run: node orbits/table.mjs (about 50 s; the stability scans of section 5 +// take nearly all of it), node orbits/table.mjs quick (well under a second; section 5 left out), or +// node orbits/table.mjs long (adds the all-directions scans followed for 100 turns of the host; about 2 minutes in all). import { createRequire } from 'node:module'; const O = createRequire(import.meta.url)('./model.js'); const { C } = O; const e3 = x => x.toExponential(3), f = (x, d = 3) => x.toFixed(d); -const quick = process.argv[2] === 'quick'; +const quick = process.argv[2] === 'quick', long = process.argv[2] === 'long'; console.log('1. Kepler in the Solar System (JPL approximate elements, Table 1, 1800-2050; GM_sun IAU 2015; masses DE430)\n'); -console.log(' planet a [au] P [yr] n^2 a^3/GM_sun - 1 own mass/Sun own + inner planets'); +console.log(' n = the mean-longitude rate Ldot (the angular rate); excess = n^2 a^3/GM_sun - 1'); +console.log(' planet a [au] P [yr] excess own mass/Sun own + inner planets other planets as rings'); let inner = 0; for (const p of O.PLANETS) { const k = O.kepler(p); - console.log(` ${p.name.padEnd(9)} ${f(p.a, 6).padStart(10)} ${f(k.P, 5).padStart(11)} ${e3(k.excess).padStart(10)} ${e3(k.own)} ${e3(k.own + inner)}`); + console.log(` ${p.name.padEnd(9)} ${f(p.a, 6).padStart(10)} ${f(k.P, 5).padStart(11)} ${e3(k.excess).padStart(10)} ${e3(k.own)} ${e3(k.own + inner)} ${e3(O.ringExcess(p)).padStart(10)}`); inner += k.own; } +{ + const nep = O.PLANETS[7], nA = (nep.Ldot + 0.32241464) * Math.PI / 180 / (36525 * C.day); + console.log(` with the mean-anomaly rate Ldot - wdot instead (Table 1's wdot = -0.32241464 deg/century), Neptune's excess would be ${e3(nA * nA * (nep.a * C.au) ** 3 / C.GMsun - 1)}`); +} const mu = O.planetMassFraction(); console.log(` all eight planets together: ${e3(mu)} of the Sun's mass; largest |excess| ${e3(Math.max(...O.PLANETS.map(p => Math.abs(O.kepler(p).excess))))}`); console.log(` fitted exponent of P against a: ${f(O.keplerExponent(), 6)} over all eight (0.39 to 30 au), ${f(O.keplerExponent(O.PLANETS.slice(0, 4)), 6)} over the inner four; Kepler 1.5`); -console.log(` mass inside Neptune's orbit over mass inside Mercury's: 1 + ${e3(mu)}`); +console.log(` mass inside Neptune's orbit over mass inside Mercury's: 1 + ${e3(mu - 1 / O.PLANETS[7].sunOver)}`); console.log('\n2. Where a Kepler domain ends: the tidal (Jacobi) radius, r_J^3 = G m / (4 A (A - B))\n'); const OORT = { A: 15.3, B: -11.9, dA: 0.4, dB: 0.4 }; // Bovy 2017 const rJ = O.jacobi(C.GMsun, OORT.A, OORT.B); const rJs = [[OORT.A + OORT.dA, OORT.B], [OORT.A - OORT.dA, OORT.B], [OORT.A, OORT.B + OORT.dB], [OORT.A, OORT.B - OORT.dB]].map(([A, B]) => O.jacobi(C.GMsun, A, B) / C.pc); console.log(` the Sun in the Galaxy (A = ${OORT.A}, B = ${OORT.B} km/s/kpc, Bovy 2017): r_J = ${f(rJ / C.pc)} pc = ${f(rJ / C.au / 1e5, 2)}e5 au` + - ` (moving A or B by its error: ${f(Math.min(...rJs))} to ${f(Math.max(...rJs))} pc)`); + ` (moving A or B by its statistical error: ${f(Math.min(...rJs))} to ${f(Math.max(...rJs))} pc; adding A's 0.5 systematic: ` + + `${f(O.jacobi(C.GMsun, OORT.A + 0.9, OORT.B) / C.pc)} to ${f(O.jacobi(C.GMsun, OORT.A - 0.9, OORT.B) / C.pc)} pc)`); console.log(` the Sun's mean density inside r_J over the Galaxy's inside the Sun's orbit: 4A/(A - B) = ${f(O.densityRatio(OORT.A, OORT.B), 2)}` + - ` (3 for a point-mass host, 2 for a flat rotation curve)`); + ` (3 for a point-mass host, 2 for a flat rotation curve; 2(1 - dln v/dln R) with dln v/dln R = -(A + B)/(A - B) = ${f(-(OORT.A + OORT.B) / (OORT.A - OORT.B))})`); const rhoJ = 1 / (4 / 3 * Math.PI * (rJ / C.pc) ** 3); -console.log(` that mean density: ${f(rhoJ, 3)} M_sun/pc^3`); +console.log(` that mean density: ${f(rhoJ, 3)} M_sun/pc^3; the Galaxy's spherical mean inside the Sun's orbit, 3 Omega^2/(4 pi G): ` + + `${f(3 * ((OORT.A - OORT.B) * 1e-3) ** 2 / (4 * Math.PI * C.G * 1e9 * 1e-6), 3)} M_sun/pc^3; the local mid-plane density is about 0.10`); for (const p of O.PLANETS) console.log(` Hill radius of ${p.name.padEnd(8)} a (m/3M)^(1/3) = ${f(O.hill(p.a, 1 / p.sunOver), 4)} au = ${f(O.hill(p.a, 1 / p.sunOver) / p.a, 4)} a`); console.log(` the Sun's own pull falls to g+ = ${O.GDAGGER.toExponential(2)} m/s^2 at ${f(O.radiusAtAcceleration(C.GMsun, O.GDAGGER) / C.au, 0)} au` + ` (${f(O.radiusAtAcceleration(C.GMsun, O.GDAGGER) / rJ, 3)} r_J)`); @@ -42,17 +50,18 @@ console.log(` pericentre ${f(s2M.periAU, 0)} au = ${f(s2M.periOverRs, 0)} Schw const MBH = O.SGRA.M, K = O.CUSP; console.log(` Sgr A* today: ${MBH / 1e6}e6 M_sun (GRAVITY Collaboration 2022). The stars around it: ${K.M1lo / 1e6} to ${K.M1hi / 1e6}e6 M_sun inside 1 pc` + ` (Schodel, Merritt & Eckart 2009), density ~ r^-${K.gamma} inside ~${K.rBreak} pc (Schodel et al. 2018)`); +const quarter = M1 => { let lo = 1e-3, hi = 30; for (let i = 0; i < 100; i++) { const m = Math.sqrt(lo * hi); if (O.cuspSlope(m, MBH, M1) < -0.25) lo = m; else hi = m; } return lo; }; for (const [lab, M1] of [['low', K.M1lo], ['middle', K.M1], ['high', K.M1hi]]) { - console.log(` ${lab.padEnd(6)} (${M1 / 1e6}e6 inside 1 pc): stars weigh as much as Sgr A* inside ${f(O.cuspRadius(1, MBH, M1), 2)} pc, twice as much inside ${f(O.cuspRadius(2, MBH, M1), 2)} pc`); -} -{ - let lo = 1e-3, hi = 30; - for (let i = 0; i < 100; i++) { const m = Math.sqrt(lo * hi); if (O.cuspSlope(m, MBH) < -0.25) lo = m; else hi = m; } - console.log(` dln v/dln r (middle): ${[0.01, 0.1, 0.5, 1, 2, 3].map(r => `${f(O.cuspSlope(r, MBH))} at ${r} pc`).join(', ')}; -1/4 at ${f(lo, 2)} pc`); + console.log(` ${lab.padEnd(6)} (${M1 / 1e6}e6 inside 1 pc): dln v/dln r passes -1/4 at ${f(quarter(M1), 2)} pc; stars weigh as much as Sgr A* inside` + + ` ${f(O.cuspRadius(1, MBH, M1), 2)} pc, twice as much inside ${f(O.cuspRadius(2, MBH, M1), 2)} pc`); } +console.log(` (radii past the ${K.rBreak} pc break extrapolate the power law; the density falls faster there, so they are lower limits)`); +console.log(` dln v/dln r (middle): ${[0.01, 0.1, 0.5, 1, 2, 3].map(r => `${f(O.cuspSlope(r, MBH))} at ${r} pc`).join(', ')}`); +console.log(` the stars inside 1 pc (middle): ${f(O.cuspMass(1) / MBH, 2)} of Sgr A*'s mass; their density at 1 pc, (3 - gamma) M1/(4 pi): ` + + `${e3((3 - K.gamma) * K.M1 / (4 * Math.PI))} M_sun/pc^3, ${e3((3 - K.gamma) * K.M1 / (4 * Math.PI) / 0.10)} times the 0.10 near the Sun; Sgr A* is ${e3(MBH)} Suns`); const g = O.galaxy(); console.log(` for comparison, MWPotential2014's bulge (fit on kpc scales, no nuclear star cluster) plus Sgr A*: stars weigh as much as Sgr A* inside` + - ` ${f(g.influence() * 1e3, 2)} pc, twice as much inside ${f(g.cross(r => g.bulgeMass(r) - 2 * g.M.bh, 1e-6, 1) * 1e3, 2)} pc`); + ` ${f(g.influence(1) * 1e3, 2)} pc, twice as much inside ${f(g.influence(2) * 1e3, 2)} pc`); console.log('\n4. The Galaxy: v(R) and its log slope in the mid-plane (MWPotential2014, Bovy 2015, plus Sgr A*)\n'); console.log(` component masses: bulge ${e3(g.M.bulge)}, disk ${e3(g.M.disk)} M_sun; halo 4 pi rho_s r_s^3 = ${e3(g.M.halo)} M_sun`); @@ -61,12 +70,18 @@ for (const R of [1e-5, 1e-4, 1e-3, 3e-3, 1e-2, 0.03, 0.1, 0.3, 1, 3, 5, 8, 15, 2 const p = g.parts(R), v2 = p.bh + p.bulge + p.disk + p.halo; console.log(` ${String(R).padEnd(10)} ${f(g.v(R), 1).padStart(9)} ${f(g.slope(R)).padStart(8)} ${[p.bh, p.bulge, p.disk, p.halo].map(x => f(x / v2, 3)).join(', ')}`); } -const flat = g.flatRange(0.15, 1, 300); -console.log(` |dln v/dln R| < 0.15 from ${f(flat[0], 2)} to ${f(flat[1], 1)} kpc`); +const flat = g.flatRange(0.15), flatBulge = g.flatRange(0.15, 1e-3, 1); +console.log(` |dln v/dln R| < 0.15 outside the bulge (R > 1 kpc) from ${f(flat[0], 2)} to ${f(flat[1], 1)} kpc; the r^-1.8 bulge is also that flat from` + + ` ${f(flatBulge[0] * 1e3, 0)} to ${f(flatBulge[1] * 1e3, 0)} pc`); const EILERS = { v0: 229.0, dv: -1.7, R0: 8.122, Rmin: 5, Rmax: 25 }; // Eilers et al. 2019 const ve = R => EILERS.v0 + EILERS.dv * (R - EILERS.R0); console.log(` Eilers et al. 2019: v = ${EILERS.v0} ${EILERS.dv} (R - ${EILERS.R0}) km/s for ${EILERS.Rmin} to ${EILERS.Rmax} kpc;` + ` dln v/dln R = ${f(EILERS.dv * EILERS.Rmin / ve(EILERS.Rmin))} at ${EILERS.Rmin}, ${f(EILERS.dv * EILERS.R0 / ve(EILERS.R0))} at R0, ${f(EILERS.dv * EILERS.Rmax / ve(EILERS.Rmax))} at ${EILERS.Rmax} kpc`); +{ + const Rs = [5, 8.122, 15, 25]; + console.log(` Eilers et al.'s v at ${Rs.join(', ')} kpc: ${Rs.map(R => f(ve(R), 1)).join(', ')} km/s; the model's: ${Rs.map(R => f(g.v(R), 1)).join(', ')}`); + console.log(` shape, (v/v(R0)) / (v_Eilers/v_Eilers(R0)) - 1: ${Rs.map(R => f(g.v(R) / g.v(8.122) / (ve(R) / ve(8.122)) - 1)).join(', ')}; level at R0: ${f(g.v(8.122) / 229 - 1)}`); +} const Mgrow = ve(EILERS.Rmax) ** 2 * EILERS.Rmax / (ve(EILERS.Rmin) ** 2 * EILERS.Rmin); console.log(` mass inside 25 kpc over mass inside 5 kpc, v^2 R / G (spherical): ${f(Mgrow, 2)}; for Kepler it would be 1`); for (const Mb of [5e10, 6e10, 1e11]) console.log(` ${e3(Mb)} M_sun all inside 25 kpc, as a point mass: v(25 kpc) = ${f(Math.sqrt(C.G * Mb / 25), 0)} km/s (measured ${f(ve(25), 0)})`); @@ -79,7 +94,8 @@ console.log(` inclinations to the invariable plane (JPL Table 1 elements; plan O.PLANETS.map(p => `${p.name} ${f(O.tiltToInvariable(p), 3)}`).join(', ') + ' deg'); const tilts = O.PLANETS.map(p => O.tiltToInvariable(p) * Math.PI / 180); console.log(` rms sin(i) over the eight: ${f(Math.sqrt(tilts.reduce((a, t) => a + Math.sin(t) ** 2, 0) / 8), 3)}; without Mercury: ${f(Math.sqrt(tilts.slice(1).reduce((a, t) => a + Math.sin(t) ** 2, 0) / 7), 3)}`); -console.log(` the Galaxy's thin disk: scale height 300 +- 50 pc over scale length 2.6 +- 0.5 kpc = ${f(0.3 / 2.6, 3)} (Bland-Hawthorn & Gerhard 2016)`); +console.log(` the Galaxy's thin disk: scale height 300 +- 50 pc over scale length 2.6 +- 0.5 kpc = ${f(0.3 / 2.6, 3)}, the axis ratio of its density` + + ` contours (Bland-Hawthorn & Gerhard 2016)`); if (!quick) { console.log('\n5. Which orbits a body keeps: circular orbits followed for 20 turns of the host (Hill\'s problem)\n'); @@ -88,14 +104,17 @@ if (!quick) { console.log(` point-mass host: Gam = 4A/Omega = ${O.HOST_KEPLER.Gam}, (nu/Omega)^2 = ${O.HOST_KEPLER.nu2}`); console.log(` the Sun in the Galaxy (A, B as above, rho0 = ${RHO0} M_sun/pc^3): Gam = ${f(sun.Gam, 3)}, (nu/Omega)^2 = ${f(sun.nu2, 2)}`); console.log(` 20 turns of the host: ${f(20 * 2 * Math.PI / ((OORT.A - OORT.B) * C.kmskpc) / C.yr / 1e9, 2)} Gyr for the Sun; ${f(20 * O.kepler(O.PLANETS[4]).P, 0)} yr for Jupiter`); + console.log(` the Sun's vertical oscillation through the disk with this nu: ${f(2 * Math.PI / (Math.sqrt(sun.nu2) * (OORT.A - OORT.B) * C.kmskpc) / C.yr / 1e6, 0)} Myr`); const planes = { 'prograde, in the plane': [[0, 0, 1]], 'retrograde, in the plane': [[0, 0, -1]], 'polar, plane holds the host': [[0, 1, 0]], 'polar, plane across it': [[1, 0, 0]] }; const as = []; for (let a = 0.3; a <= 1.2001; a += 0.025) as.push(a); console.log(' radius [r_J] where half the orbits (8 starting angles) have escaped beyond 3 r_J:'); - console.log(' orbit point-mass host the Sun in the Galaxy'); + const sunNu1 = { ...sun, nu2: 1, name: 'the Sun in the Galaxy with (nu/Omega)^2 = 1' }; + console.log(' orbit point-mass host the Sun in the Galaxy the same with (nu/Omega)^2 = 1'); for (const [k, n] of Object.entries(planes)) { - const e = [O.HOST_KEPLER, sun].map(h => O.hillScan(h, n, as, { nPhase: 8 }).edge); - console.log(` ${k.padEnd(28)} ${f(e[0], 3).padStart(10)} ${f(e[1], 3).padStart(10)}`); + const e = [O.HOST_KEPLER, sun, sunNu1].map(h => O.hillScan(h, n, as, { nPhase: 8 }).edge); + console.log(` ${k.padEnd(28)} ${f(e[0], 3).padStart(10)} ${f(e[1], 3).padStart(10)} ${f(e[2], 3).padStart(10)}`); } + console.log(' (an orbit that starts in the plane stays in it, so the vertical tide cannot act on the first two rows)'); const dirs = O.sphere(24), asIso = []; for (let a = 0.2; a <= 1.2001; a += 0.1) asIso.push(a); console.log('\n orbits facing every way (24 directions x 4 starting angles): fractions that stay, escape, and plunge within 1e-3 r_J'); const iso = {}; @@ -109,4 +128,13 @@ if (!quick) { const e = O.hillScan(O.hostDisk(OORT.A, OORT.B, rho), dirs, asIso, { nPhase: 4 }).edge; console.log(` with rho0 = ${rho}: (nu/Omega)^2 = ${f(O.hostDisk(OORT.A, OORT.B, rho).nu2, 2)}, half escaped by ${f(e, 3)} r_J`); } + const oort = [1e5, 2e5].map(x => x * C.au / rJ); + console.log(` the Oort cloud's outer edge, near 1e5 au and as far as 2e5 au: ${f(oort[0], 2)} to ${f(oort[1], 2)} r_J`); + if (long) { + console.log('\n the same, followed for 100 turns of the host:'); + for (const [name, h] of [['point-mass host', O.HOST_KEPLER], ['the Sun in the Galaxy', sun]]) { + const r = O.hillScan(h, dirs, asIso, { nPhase: 4, orbits: 100 }); + console.log(` ${name}: half have escaped by ${f(r.edge, 3)} r_J; plunged by then, at a = 0.2 to 0.4 r_J: ${r.rows.slice(0, 3).map(x => f(x.plunges, 2)).join(', ')}`); + } + } } diff --git a/tests/orbits.mjs b/tests/orbits.mjs index bcf069a..fd331dc 100644 --- a/tests/orbits.mjs +++ b/tests/orbits.mjs @@ -1,7 +1,8 @@ // Checks the orbits study (orbits/model.js) against published values and its own closed forms. Run: npm test. // Each check prints the source, the published or expected value, the model's value and the tolerance. "Arithmetic" and -// "model-internal" mark checks of the code against a closed form or itself, not against nature. Sources were read in -// search results and abstracts only (orbits/README.md, Sources), except where a check says otherwise. +// "model-internal" mark checks of the code against a closed form or itself, not against nature; "model" marks a model +// result with no published value to compare. Sources were read in search results and abstracts only +// (orbits/README.md, Sources), except where a check says otherwise. import { createRequire } from 'node:module'; const O = createRequire(import.meta.url)('../orbits/model.js'); const { C } = O; @@ -19,49 +20,64 @@ function check(name, ok, detail, src) { } // 1. Kepler's third law across the planets. -const JPL = 'JPL approximate elements, Table 1 (1800-2050); GM_sun IAU 2015 B3; masses DE430'; +const JPL = 'JPL approximate elements, Table 1 (1800-2050; a copy of the file); GM_sun IAU 2015 B3; masses DE430'; const mu = O.planetMassFraction(), ex = O.PLANETS.map(p => O.kepler(p).excess), big = Math.max(...ex.map(Math.abs)); -check('n^2 a^3 = GM_sun for every planet, to within the planets\' total mass over the Sun\'s', ex.every(x => Math.abs(x) <= mu), - `largest |n^2 a^3 / GM_sun - 1| = ${fmt(big)} (${O.PLANETS[ex.map(Math.abs).indexOf(big)].name}); sum of m/M_sun = ${fmt(mu)}`, JPL); -near('Kepler\'s exponent: P ~ a^p over 0.39 to 30 au', O.keplerExponent(), 1.5, 1.5 * mu, `${JPL}; tolerance 3/2 times that mass fraction`); +check('In this 1800-2050 fit, n^2 a^3 matches GM_sun for every planet to within the planets\' total mass over the Sun\'s', + ex.every(x => Math.abs(x) <= mu), + `largest |n^2 a^3 / GM_sun - 1| = ${fmt(big)} (${O.PLANETS[ex.map(Math.abs).indexOf(big)].name}); sum of m/M_sun = ${fmt(mu)}` + + ' (an observed size, not a physical bound: smeared into rings, the planets\' pull alone gives up to 1.5e-3)', JPL); +{ + const xs = O.PLANETS.map(p => Math.log(p.a)), mx = xs.reduce((s, x) => s + x, 0) / xs.length; + const tolP = 0.5 * mu * xs.reduce((s, x) => s + Math.abs(x - mx), 0) / xs.reduce((s, x) => s + (x - mx) ** 2, 0); + near('Kepler\'s exponent: P ~ a^p over 0.39 to 30 au', O.keplerExponent(), 1.5, tolP, + `${JPL}; tolerance 0.5 mu sum|dln a| / sum(dln a)^2, the most the first check allows: model-internal, implied by it but independent of GM_sun`); +} -// 2. Where a Kepler domain ends. -const kk = O.oortKepler(1), kf = O.oortFlat(1); -near('Point-mass host: 4A/Omega = 3', O.densityRatio(kk.A, kk.B), 3, 1e-12, 'arithmetic'); -near('Flat rotation curve: 4A/Omega = 2', O.densityRatio(kf.A, kf.B), 2, 1e-12, 'arithmetic'); +// 2. Where a Kepler region ends. { + const g = O.galaxy(), oortAt = R => { // Oort constants from the model's own curve: A = (v/R - dv/dR)/2, B = -(v/R + dv/dR)/2 + const v = g.v(R), dv = v / R * g.slope(R); + return { A: (v / R - dv) / 2, B: -(v / R + dv) / 2 }; + }; + const k = oortAt(1e-5); + near('4A/Omega from the model Galaxy\'s curve 1e-5 kpc from Sgr A*, where a point mass rules: Hill\'s 3', O.densityRatio(k.A, k.B), 3, 0.005, + 'model-internal (densityRatio against the curve\'s own Oort constants)'); const a = O.PLANETS[4].a * C.au, q = 1 / O.PLANETS[4].sunOver, o = O.oortKepler(Math.sqrt(C.GMsun / a ** 3) / C.kmskpc); near('Jacobi radius with point-mass Oort constants / Hill radius a (m/3M)^(1/3), for Jupiter', O.jacobi(C.GMsun * q, o.A, o.B) / O.hill(a, q), 1, 1e-12, 'model-internal'); } near('The Sun\'s tidal radius in the Galaxy [1e5 au], from Bovy 2017\'s A = 15.3, B = -11.9 km/s/kpc', O.jacobi(C.GMsun, 15.3, -11.9) / C.au / 1e5, 2.8, 0.05, - 'Correa-Otto & Calandra 2019 (MNRAS 490, 2495): "r_J ~ 2.8 x 10^5 au"; tolerance the rounding of 2.8'); + 'Correa-Otto & Calandra 2019 (MNRAS 490, 2495): "r_J ~ 2.8 x 10^5 au", search text; tolerance the rounding of 2.8'); // 3. Sgr A*: S2's orbit and the stars around it. -const G20 = 'GRAVITY Collaboration 2020 (A&A 636, L5)'; +const G20 = 'GRAVITY Collaboration 2020 (A&A 636, L5), read in a transcription'; const s2 = O.orbitOf(O.S2), s2M = O.orbitOf(O.S2, O.S2.M); near('S2: Kepler\'s third law returns GRAVITY\'s fitted mass [1e6 M_sun]', s2.keplerMass / 1e6, O.S2.M / 1e6, 0.012, `${G20}, Table E.1; consistency with their own fit (tolerance its 0.012 error), not a measurement`); -near('S2: Schwarzschild precession per orbit [arcmin]', s2M.precessionArcmin, 12.1, 0.1, `${G20}, appendix B: "the 12.1' precession"`); -near('S2: pericentre distance [au]', s2M.periAU, 120, 5, `${G20}, abstract: "pericentre at 120 AU", rounded to 10 au`); -near('S2: speed at pericentre [km/s]', s2M.periSpeed, 7700, 50, `${G20}, abstract: "7700 km s^-1", rounded to 100 km/s`); +near('S2: Schwarzschild precession per orbit [arcmin]', s2M.precessionArcmin, 12.1, 0.1, + `${G20}, appendix B: "the 12.1' precession"; tolerance one unit in the last digit (the formula with their Table E.1 M, a, e gives 12.16')`); +near('S2: pericentre distance [au]', s2M.periAU, 120, 5, `${G20}, introduction: "pericentre at 120 AU", rounded to 10 au`); +near('S2: speed at pericentre [km/s]', s2M.periSpeed, 7700, 50, `${G20}, introduction: "7700 km s^-1", rounded to 100 km/s`); { const K = O.CUSP, M = O.SGRA.M, r2 = [K.M1hi, K.M1, K.M1lo].map(m => O.cuspRadius(2, M, m)); check('The stars inside r weigh twice Sgr A* at 3 to 4 pc in the literature; the cusp model brackets it', r2[0] <= 3 && r2[2] >= 4, - `model ${fmt(r2[0])} to ${fmt(r2[2])} pc (middle ${fmt(r2[1])}) for 0.5-1.5e6 M_sun inside 1 pc and density ~ r^-1.13`, - 'Alexander 2005 (3 pc), Feldmeier-Krause et al. 2017 (about 4 pc), Merritt 2013 (3.9 pc); inputs Schodel et al. 2009, 2018 (search text)'); + `model ${fmt(r2[0])} to ${fmt(r2[2])} pc (middle ${fmt(r2[1])}) for 0.5-1.5e6 M_sun inside 1 pc and density ~ r^-1.13; radii past the 3 pc ` + + 'break extrapolate the cusp, where the density falls faster, so they are lower limits and the upper side holds a fortiori', + 'Alexander 2005 (3 pc, as cited), Feldmeier-Krause et al. 2017 (about 4 pc, with their M_BH of 3.0e6), GRAVITY Collaboration 2024 ' + + '(3.9 pc in their models; definition from Merritt 2013); inputs Schodel et al. 2009, 2018; all search text'); near('Kepler inside 0.1 pc of Sgr A*: dln v/dln r', O.cuspSlope(0.1, M), -0.5, 0.01, 'model (Sgr A* plus the cusp)'); } // 4. The Galaxy. const g = O.galaxy(), g0 = O.galaxy(O.MWP14, { bh: 0 }); -near('MWPotential2014: v_c(8 kpc) without Sgr A* [km/s]', g0.v(8), 220, 1e-9, 'model-internal (the normalisation; parameters as in galpy\'s source)'); +near('MWPotential2014 normalisation identity: v_c(8 kpc) without Sgr A* [km/s] (G and the shapes cancel)', g0.v(8), 220, 1e-9, + 'model-internal (parameters as in galpy\'s source)'); { const Rs = [5, 8.122, 15, 25], ve = R => 229.0 - 1.7 * (R - 8.122); const rel = Rs.map(R => (g.v(R) / g.v(8.122)) / (ve(R) / ve(8.122)) - 1); check('The shape of MWPotential2014\'s curve, v(R)/v(R0), within 5% of Eilers et al.\'s from 5 to 25 kpc', rel.every(r => Math.abs(r) < 0.05), `ratio to Eilers - 1 = ${rel.map(r => r.toFixed(3)).join(', ')} at ${Rs.join(', ')} kpc; the level differs by ${(g.v(8.122) / 229 - 1).toFixed(3)} (220 against 229 km/s)`, - 'Eilers et al. 2019 (ApJ 871, 120), linear fit; tolerance their 2-5% systematic uncertainty'); + 'Eilers et al. 2019 (ApJ 871, 120), linear fit, abstract; tolerance their 2-5% systematic uncertainty'); const sl = Rs.map(R => g.slope(R)), se = Rs.map(R => -1.7 * R / ve(R)); check('Flat, not Kepler, from 5 to 25 kpc: |dln v/dln R| < 0.25 in the model and in Eilers et al.\'s fit', sl.every(s => Math.abs(s) < 0.25) && se.every(s => Math.abs(s) < 0.25), @@ -73,22 +89,35 @@ const SS = 'Souami & Souchay 2012 (A&A 543, A133), as quoted in search text; tol near('Jupiter\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O.PLANETS[4]), 0.3219, 0.01, SS); near('Saturn\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O.PLANETS[5]), 0.9254, 0.01, SS); -// 6. Hill's problem: which orbits a body keeps (20 turns of the host, 8 starting angles each). +// 6. Hill's problem: which orbits a body keeps (20 turns of the host, 8 starting angles each unless stated). { - const K = O.HOST_KEPLER, fates = (h, a, n) => O.hillFates(h, a, [n], { nPhase: 8 }).escapes; - const pro = [fates(K, 0.4, [0, 0, 1]), fates(K, 0.6, [0, 0, 1])], ret = [fates(K, 0.6, [0, 0, -1]), fates(K, 0.9, [0, 0, -1])]; - check('Point-mass host: prograde circular orbits kept at 0.4 r_H, lost by 0.6 (published limit about 0.5)', pro[0] === 0 && pro[1] >= 0.5, - `fraction escaped ${pro[0].toFixed(2)} at 0.4 r_H, ${pro[1].toFixed(2)} at 0.6 r_H`, - 'Shen & Tremaine 2008 (about 0.5 r_H), Hamilton & Krivov 1997 (0.53 r_H): search text'); + const K = O.HOST_KEPLER, sun = O.hostDisk(15.3, -11.9, 0.10), ST = 'Shen & Tremaine 2008 (arXiv 0806.2882), abstract, search text'; + const edgePro = O.hillScan(K, [[0, 0, 1]], [0.4, 0.45, 0.5, 0.55, 0.6], { nPhase: 8 }).edge; + near('Point-mass host: radius where half the prograde circular orbits have escaped [r_H]', edgePro, 0.5, 0.1, + `${ST}: "about 0.5 r_H"; tolerance covers their different criterion and integration time`); + const esc = (h, a, n) => O.hillFates(h, a, [n], { nPhase: 8 }).escapes, ret = [esc(K, 0.6, [0, 0, -1]), esc(K, 0.9, [0, 0, -1])]; check('Point-mass host: retrograde orbits kept at 0.6 r_H, lost by 0.9 (published limit about 0.7)', ret[0] === 0 && ret[1] >= 0.5, - `fraction escaped ${ret[0].toFixed(2)} at 0.6 r_H, ${ret[1].toFixed(2)} at 0.9 r_H`, - 'Shen & Tremaine 2008 (about 0.7 r_H), Hamilton & Krivov 1997 (0.69 r_H): search text'); - const sun = O.hostDisk(15.3, -11.9, 0.10); - near('The Sun in the Galaxy: (vertical tidal frequency / Omega)^2', sun.nu2, (4 * Math.PI * C.G * 0.1e9 + 2 * (15.3 ** 2 - 11.9 ** 2)) / 27.2 ** 2, 1e-12, 'arithmetic (Poisson\'s equation)'); - const dirs = O.sphere(12), iso = [0.3, 0.6].map(a => O.hillFates(sun, a, dirs, { nPhase: 2 }).escapes); - check('The Sun keeps orbits facing every way at 0.3 r_J (0.85e5 au) and loses most by 0.6 r_J (1.7e5 au)', iso[0] === 0 && iso[1] >= 0.5, - `fraction escaped ${iso[0].toFixed(2)} at 0.3 r_J, ${iso[1].toFixed(2)} at 0.6 r_J (12 directions x 2 angles)`, - 'model; the Oort cloud\'s outer edge is quoted at 1 to 2 x 10^5 au (Dones et al. 2004, as cited)'); + `fraction escaped ${ret[0].toFixed(2)} at 0.6 r_H, ${ret[1].toFixed(2)} at 0.9 r_H`, `${ST}: "about 0.7 r_H"`); + const r875 = [esc(K, 0.875, [0, 0, -1]), esc(sun, 0.875, [0, 0, -1])]; + check('Retrograde in the plane at 0.875 r_J: lost around a point mass, kept by the Sun (its flatter curve, Gam = 2.25, not its vertical tide)', + r875[0] >= 0.75 && r875[1] <= 0.25, `fraction escaped ${r875[0].toFixed(2)} (point mass), ${r875[1].toFixed(2)} (the Sun)`, 'model'); + { + const s0 = O.hillStart(sun, 0.3, [0.6, 0, 0.8], 0.7), end = O.hillOrbit(sun, s0, { orbits: 20 }); + const EJ = ([x, y, z, vx, vy, vz]) => (vx * vx + vy * vy + vz * vz) / 2 - sun.Gam * x * x / 2 + sun.nu2 * z * z / 2 - sun.Gam / Math.hypot(x, y, z); + near('The Sun in the Galaxy: Jacobi integral |v|^2/2 - Gam x^2/2 + nu2 z^2/2 - Gam/r over 20 turns (a = 0.3, tilted orbit)', + EJ(end.s) - EJ(s0), 0, 1e-6, `model-internal (fate ${end.fate}; a conserved quantity of the equations of motion)`); + } + const Om = 27.2 * C.kmskpc, want = (4 * Math.PI * C.GMsun * 0.10 / C.pc ** 3 + 2 * (15.3 ** 2 - 11.9 ** 2) * C.kmskpc ** 2) / Om ** 2; + near('The Sun in the Galaxy: (vertical tidal frequency / Omega)^2', sun.nu2, want, 1e-12 * want, 'arithmetic in SI (Poisson\'s equation), independent of C.G'); + const dirs = O.sphere(12), fz = [0.3, 0.6].map(a => O.hillFates(sun, a, dirs, { nPhase: 2 })); + check('The Sun: no orbit, facing any way, escapes from 0.3 r_J (0.85e5 au); most escape by 0.6 r_J (1.7e5 au)', + fz[0].escapes === 0 && fz[1].escapes >= 0.5, + fz.map((r, i) => `at ${[0.3, 0.6][i]} r_J: stay ${r.stays.toFixed(2)}, escape ${r.escapes.toFixed(2)}, plunge ${r.plunges.toFixed(2)}`).join('; ') + + ' (12 directions x 2 angles)', 'model'); + const rJ = O.jacobi(C.GMsun, 15.3, -11.9), edge = O.hillScan(sun, dirs, [0.3, 0.4, 0.5, 0.6], { nPhase: 2 }).edge * rJ / C.au; + check('The Sun keeps orbits facing every way out to the Oort cloud\'s quoted outer edge, 1e5 to 2e5 au', edge >= 1e5 && edge <= 2e5, + `half have escaped by ${fmt(edge)} au (12 directions x 2 angles, grid 0.1 r_J)`, + 'near 1e5 au: Correa-Otto & Calandra 2019, search text; as far as 2e5 au: Dones et al. 2004, as cited in search text'); } console.log(fails ? `\n${fails} check(s) failed` : '\nall orbits checks pass'); From 34593f212ba0f775f5fddfed576c1442687ce0c0 Mon Sep 17 00:00:00 2001 From: Claude Date: Mon, 19 Oct 2026 12:14:14 +0000 Subject: [PATCH 03/10] Orbits study: fixes from the verification of d670853 - n^2 a is the mean inward pull at a. It equals G M( Claude-Session: https://claude.ai/code/session_01C2bpiHyHSkC8DAL5CcyciK --- README.md | 2 +- orbits/README.md | 41 +++++++++++++++++++++-------------------- orbits/model.js | 15 ++++++++------- orbits/table.mjs | 8 ++++++-- tests/orbits.mjs | 7 +++++-- 5 files changed, 41 insertions(+), 32 deletions(-) diff --git a/README.md b/README.md index fb22373..1fa58f5 100644 --- a/README.md +++ b/README.md @@ -18,7 +18,7 @@ with Claude (Anthropic), for [Foundation0](https://github.com/Foundation-0/f0). with gravity in place of electric attraction (the "gravitational atom"). It is still a wave cloud, not planets on orbits. The same equation, a wave held by a pull that weakens with distance, gives the same solution at any scale. The nesting lives in the equations, not the objects. -- **Solar systems are not little galaxies, except at a galaxy's centre.** Around one dominant mass, orbits follow +- **Solar systems are not little galaxies, except at the Milky Way's centre.** Around one dominant mass, orbits follow Kepler's law at every radius. The planets match it to 0.13%, about what they weigh together against the Sun. Across the Milky Way's disk the speed instead stays between 200 and 235 km/s, because the mass is spread out. A Kepler region ends where its host's tide wins, at a density ratio rather than a size: 1.4 pc for the Sun in the Galaxy, diff --git a/orbits/README.md b/orbits/README.md index df5cb5e..7290b29 100644 --- a/orbits/README.md +++ b/orbits/README.md @@ -14,10 +14,10 @@ what ends it, and what, if anything, repeats from one scale to the next. |---|---| | Is a galaxy a big solar system? | No. Around one dominant mass, orbital speed falls as Kepler's v ∝ r^−1/2. Across the Milky Way's disk the speed falls only from 234 to 200 km/s between 5 and 25 kpc. Read as spherical, the mass inside grows 3.7-fold over that range, where a Kepler system's would stay fixed. Both figures come from Eilers et al.'s (2019) fit. | | How exact is Kepler's law in the Solar System? | In JPL's 1800–2050 fit, n²a³ matches GM☉ to within 1.3×10⁻³ for all eight planets. That is about the planets' total mass, 1.34×10⁻³ of the Sun's: the planets pull on one another. P ∝ a^1.49986 over 0.39 to 30 au. | -| Does the Galaxy have a Kepler region? | Yes, around Sgr A*. The star S2 orbits it as a planet orbits the Sun, plus general relativity's 12.1′ precession per orbit. The region ends about 1.3 pc out, where the speed's log slope passes −1/4 (1.0 to 1.8 pc across the cluster-mass range). The stars inside weigh as much as the black hole by 1.8 to 3.2 pc. | +| Does the Galaxy have a Kepler region? | Yes, around Sgr A*. The star S2 orbits it as a planet orbits the Sun, plus general relativity's precession: 12.2′ per orbit from their elements (they quote 12.1′). The region ends about 1.3 pc out, where the speed's log slope passes −1/4 (1.0 to 1.8 pc across the cluster-mass range). The stars inside weigh as much as the black hole by 1.8 to 3.2 pc. | | What ends a Kepler region? | For a body orbiting in a host (a planet, the Sun), the host's tide. Satellites are held only inside its tidal radius, r_J³ = Gm/(4A(A − B)). Inside it, the body's mean density is 4A/Ω times the host's mean density inside the orbit (host taken as spherical): 3 for a planet around the Sun, 2.25 for the Sun in the Galaxy. For the Sun, r_J = 1.37 pc = 2.8×10⁵ au. At the Galaxy's centre, where Sgr A* sits still, the region ends instead where the surrounding stars' mass approaches the black hole's. | -| How much of that region is used? | About half, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.53 r_J around a planet and by 0.45 r_J (1.3×10⁵ au) for the Sun. Shen and Tremaine (2008) put stable prograde and retrograde moons out to about 0.5 and 0.7 r_H; the observed outer moons reach about 0.4 and 0.6 r_H. The Oort cloud's outer edge is placed between 10⁵ and 2×10⁵ au, which is 0.35 to 0.7 r_J. | -| What does repeat? | Kepler regions nested inside one another, each ending at its host's tide at a similar fraction of its tidal radius: moons in a planet's, planets and comets in the Sun's. Also flat disks: the planets' orbits are tilted by 2° on average (6° for Mercury), and the Galaxy's thin disk is about 0.1 as thick as it is wide. | +| How much of that region is used? | About half, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.53 r_J around a planet and by 0.45 r_J (1.3×10⁵ au) for the Sun. Shen and Tremaine (2008) put stable prograde and retrograde moons out to about 0.5 and 0.7 r_H; the observed irregular moons lie out to about 0.6 r_H, the retrograde ones further out than the prograde. The Oort cloud's outer edge is placed between 10⁵ and 2×10⁵ au, which is 0.35 to 0.7 r_J. | +| What does repeat? | Kepler regions nested inside one another, each ending at its host's tide at a similar fraction of its tidal radius: moons in a planet's, planets and comets in the Sun's. Also flat disks: the planets' orbits are tilted by about 2° on average (6° for Mercury), and the Galaxy's thin disk is about 0.1 as thick as it is wide. | ## Kepler's law in the Solar System @@ -26,24 +26,24 @@ sets a size, so it holds from the Sun's surface out to wherever another pull com JPL's approximate elements (Standish, Table 1, a fit to 1800–2050) give each planet's semi-major axis a and its mean-longitude rate, n = L̇: -- This is the orbit's angular rate, which obeys n²a³ = GM( PLANETS.reduce((s, p) => s + 1 / p.sunOver, 0); // Kepler's third law for one planet: the GM its orbit implies, n^2 a^3, against GM_sun, with n the mean-longitude - // rate Ldot: the orbit's angular rate, which obeys n^2 a^3 = G M( -(2 / Math.PI) * ellipticK(2 * Math.sqrt(r * b) / (r + b)) / (r + b); // per unit G m - function ringExcess(p) { - let ex = 1 / p.sunOver; + function ringExcess(p, { own = true } = {}) { + let ex = own ? 1 / p.sunOver : 0; for (const q of PLANETS) { if (q === p) continue; const h = 1e-6 * p.a, pull = (ringPotential(p.a + h, q.a) - ringPotential(p.a - h, q.a)) / (2 * h); // inward > 0 diff --git a/orbits/table.mjs b/orbits/table.mjs index 9e6eaca..3e4a90c 100644 --- a/orbits/table.mjs +++ b/orbits/table.mjs @@ -9,7 +9,7 @@ const quick = process.argv[2] === 'quick', long = process.argv[2] === 'long'; console.log('1. Kepler in the Solar System (JPL approximate elements, Table 1, 1800-2050; GM_sun IAU 2015; masses DE430)\n'); console.log(' n = the mean-longitude rate Ldot (the angular rate); excess = n^2 a^3/GM_sun - 1'); -console.log(' planet a [au] P [yr] excess own mass/Sun own + inner planets other planets as rings'); +console.log(' planet a [au] P [yr] excess own mass/Sun own + inner planets own + other planets as rings'); let inner = 0; for (const p of O.PLANETS) { const k = O.kepler(p); @@ -24,6 +24,8 @@ const mu = O.planetMassFraction(); console.log(` all eight planets together: ${e3(mu)} of the Sun's mass; largest |excess| ${e3(Math.max(...O.PLANETS.map(p => Math.abs(O.kepler(p).excess))))}`); console.log(` fitted exponent of P against a: ${f(O.keplerExponent(), 6)} over all eight (0.39 to 30 au), ${f(O.keplerExponent(O.PLANETS.slice(0, 4)), 6)} over the inner four; Kepler 1.5`); console.log(` mass inside Neptune's orbit over mass inside Mercury's: 1 + ${e3(mu - 1 / O.PLANETS[7].sunOver)}`); +console.log(` the other planets' rings alone: ${O.PLANETS.map(p => `${p.name} ${e3(O.ringExcess(p, { own: false }))}`).join(', ')}`); +console.log(` Saturn's ring estimate minus its fit: ${e3(O.ringExcess(O.PLANETS[5]) - O.kepler(O.PLANETS[5]).excess)}`); console.log('\n2. Where a Kepler domain ends: the tidal (Jacobi) radius, r_J^3 = G m / (4 A (A - B))\n'); const OORT = { A: 15.3, B: -11.9, dA: 0.4, dB: 0.4 }; // Bovy 2017 @@ -93,6 +95,7 @@ console.log('\n4b. What does repeat: thin disks\n'); console.log(` inclinations to the invariable plane (JPL Table 1 elements; plane from Souami & Souchay 2012): ` + O.PLANETS.map(p => `${p.name} ${f(O.tiltToInvariable(p), 3)}`).join(', ') + ' deg'); const tilts = O.PLANETS.map(p => O.tiltToInvariable(p) * Math.PI / 180); +console.log(` mean inclination over the eight: ${f(O.PLANETS.reduce((a, p) => a + O.tiltToInvariable(p), 0) / 8, 2)} deg`); console.log(` rms sin(i) over the eight: ${f(Math.sqrt(tilts.reduce((a, t) => a + Math.sin(t) ** 2, 0) / 8), 3)}; without Mercury: ${f(Math.sqrt(tilts.slice(1).reduce((a, t) => a + Math.sin(t) ** 2, 0) / 7), 3)}`); console.log(` the Galaxy's thin disk: scale height 300 +- 50 pc over scale length 2.6 +- 0.5 kpc = ${f(0.3 / 2.6, 3)}, the axis ratio of its density` + ` contours (Bland-Hawthorn & Gerhard 2016)`); @@ -104,6 +107,7 @@ if (!quick) { console.log(` point-mass host: Gam = 4A/Omega = ${O.HOST_KEPLER.Gam}, (nu/Omega)^2 = ${O.HOST_KEPLER.nu2}`); console.log(` the Sun in the Galaxy (A, B as above, rho0 = ${RHO0} M_sun/pc^3): Gam = ${f(sun.Gam, 3)}, (nu/Omega)^2 = ${f(sun.nu2, 2)}`); console.log(` 20 turns of the host: ${f(20 * 2 * Math.PI / ((OORT.A - OORT.B) * C.kmskpc) / C.yr / 1e9, 2)} Gyr for the Sun; ${f(20 * O.kepler(O.PLANETS[4]).P, 0)} yr for Jupiter`); + console.log(` an orbit counts as plunged inside 1e-3 r_J: ${f(1e-3 * rJ / C.au, 0)} au for the Sun, ${f(1e-3 * O.hill(O.PLANETS[4].a, 1 / O.PLANETS[4].sunOver) * C.au / 7.1492e7, 2)} Jupiter radii (71,492 km)`); console.log(` the Sun's vertical oscillation through the disk with this nu: ${f(2 * Math.PI / (Math.sqrt(sun.nu2) * (OORT.A - OORT.B) * C.kmskpc) / C.yr / 1e6, 0)} Myr`); const planes = { 'prograde, in the plane': [[0, 0, 1]], 'retrograde, in the plane': [[0, 0, -1]], 'polar, plane holds the host': [[0, 1, 0]], 'polar, plane across it': [[1, 0, 0]] }; const as = []; for (let a = 0.3; a <= 1.2001; a += 0.025) as.push(a); @@ -131,7 +135,7 @@ if (!quick) { const oort = [1e5, 2e5].map(x => x * C.au / rJ); console.log(` the Oort cloud's outer edge, near 1e5 au and as far as 2e5 au: ${f(oort[0], 2)} to ${f(oort[1], 2)} r_J`); if (long) { - console.log('\n the same, followed for 100 turns of the host:'); + console.log(`\n the same, followed for 100 turns of the host (${f(100 * 2 * Math.PI / ((OORT.A - OORT.B) * C.kmskpc) / C.yr / 1e9, 1)} Gyr for the Sun):`); for (const [name, h] of [['point-mass host', O.HOST_KEPLER], ['the Sun in the Galaxy', sun]]) { const r = O.hillScan(h, dirs, asIso, { nPhase: 4, orbits: 100 }); console.log(` ${name}: half have escaped by ${f(r.edge, 3)} r_J; plunged by then, at a = 0.2 to 0.4 r_J: ${r.rows.slice(0, 3).map(x => f(x.plunges, 2)).join(', ')}`); diff --git a/tests/orbits.mjs b/tests/orbits.mjs index fd331dc..9ed4a84 100644 --- a/tests/orbits.mjs +++ b/tests/orbits.mjs @@ -72,6 +72,8 @@ near('S2: speed at pericentre [km/s]', s2M.periSpeed, 7700, 50, `${G20}, introdu const g = O.galaxy(), g0 = O.galaxy(O.MWP14, { bh: 0 }); near('MWPotential2014 normalisation identity: v_c(8 kpc) without Sgr A* [km/s] (G and the shapes cancel)', g0.v(8), 220, 1e-9, 'model-internal (parameters as in galpy\'s source)'); +near('MWPotential2014: halo mass inside 245 kpc [1e12 M_sun]', g0.parts(245).halo * 245 / C.G / 1e12, 0.8, 0.05, + 'Bovy 2015, Table 1: virial mass 0.8e12 M_sun with r_vir about 245 kpc, as secondary papers quote it (search text); tolerance the rounding of 0.8'); { const Rs = [5, 8.122, 15, 25], ve = R => 229.0 - 1.7 * (R - 8.122); const rel = Rs.map(R => (g.v(R) / g.v(8.122)) / (ve(R) / ve(8.122)) - 1); @@ -97,10 +99,11 @@ near('Saturn\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O `${ST}: "about 0.5 r_H"; tolerance covers their different criterion and integration time`); const esc = (h, a, n) => O.hillFates(h, a, [n], { nPhase: 8 }).escapes, ret = [esc(K, 0.6, [0, 0, -1]), esc(K, 0.9, [0, 0, -1])]; check('Point-mass host: retrograde orbits kept at 0.6 r_H, lost by 0.9 (published limit about 0.7)', ret[0] === 0 && ret[1] >= 0.5, - `fraction escaped ${ret[0].toFixed(2)} at 0.6 r_H, ${ret[1].toFixed(2)} at 0.9 r_H`, `${ST}: "about 0.7 r_H"`); + `fraction escaped ${ret[0].toFixed(2)} at 0.6 r_H (kept: must be 0), ${ret[1].toFixed(2)} at 0.9 r_H (lost: must be >= 0.5)`, `${ST}: "about 0.7 r_H"`); const r875 = [esc(K, 0.875, [0, 0, -1]), esc(sun, 0.875, [0, 0, -1])]; check('Retrograde in the plane at 0.875 r_J: lost around a point mass, kept by the Sun (its flatter curve, Gam = 2.25, not its vertical tide)', - r875[0] >= 0.75 && r875[1] <= 0.25, `fraction escaped ${r875[0].toFixed(2)} (point mass), ${r875[1].toFixed(2)} (the Sun)`, 'model'); + r875[0] >= 0.75 && r875[1] <= 0.25, + `fraction escaped ${r875[0].toFixed(2)} (point mass; lost: must be >= 0.75), ${r875[1].toFixed(2)} (the Sun; kept: must be <= 0.25)`, 'model'); { const s0 = O.hillStart(sun, 0.3, [0.6, 0, 0.8], 0.7), end = O.hillOrbit(sun, s0, { orbits: 20 }); const EJ = ([x, y, z, vx, vy, vz]) => (vx * vx + vy * vy + vz * vz) / 2 - sun.Gam * x * x / 2 + sun.nu2 * z * z / 2 - sun.Gam / Math.hypot(x, y, z); From 2db95f6c892ccae373f83f68c3bf559feaaded52 Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 27 Oct 2026 00:55:51 +0000 Subject: [PATCH 04/10] Orbits study: fixes from zero-aska's review of 34593f2 Blocking: 1. Polar orbits are lost by plunging (Lidov-Kozai), not only by escaping. The table and README now give escaped and lost (escaped or plunged) edges. Polar orbits are lost by 0.14-0.15 r_H around a planet and by 0.06 r_J for the Sun within 20 turns; a check covers the Sun's polar plunges. 2. The in-plane edges are compared with Nesvorny et al. 2003 as Shen & Tremaine quote it (0.4 and 0.7 r_H for near-circular, near-ecliptic orbits), with S&T's coplanar ~r_H stated. The edges are now 0.46 and 0.775. The prograde check is two-sided, and the retrograde check uses the range 0.7 to 1 r_H. Starting angles span [0, pi) because Hill's equations are symmetric under inversion, so every angle is a different orbit. 3. Oort cloud: Correa-Otto & Calandra's a_crit is 6.5e4 au, set by passing stars, and half of their comets are bound at about 2e4 au. The quoted outer limits are now 0.23 to 0.71 r_J, and the check's lower bound is 6.5e4 au. 4. Feldmeier-Krause et al.: 4.0 pc with M_BH = 4e6 and Merritt's (2004) definition. 5. The "twice" radius at 3 pc is credited to Schodel et al. 2018 (7.8e6 inside 3 pc). Alexander's r_h ~ 3 pc is Gm/sigma^2, with 2m reached by about 4 pc. 6. The unsourced 0.5 systematic on A is removed. 7. Jupiter's ring term is the net of all the other planets' rings (outer -3.5e-5, inner +0.6e-5), now printed. Non-blocking, also applied: - All-directions scans use 48 directions on a 0.05 grid: 0.52 and 0.435 r_J. - The 100-turn values, and the plunge-radius, rho0 and Gam sensitivities, are printed. - Model: hillOrbit throws on a non-finite step or a step cap. Gauss's-theorem sphere mass for MWPotential2014 (v^2 R/G overstates the mass inside 8 kpc by 12%; the true 5-25 kpc growth is 4.03). Closed-form cusp slope radii. - The text now gives: - the -1/4 threshold's dependence and the zero crossings at 1.9-3.4 pc; - the gamma range 1.1-1.4; - Bovy's warning on A and B, with Gam = 2.12 from Eilers et al.'s slope; - the like-for-like disk ratio 0.037; - wording for the Eilers, Koop, Chae and Banik results. - New checks: - the inner planets' excess, which pins GM_sun and the century; - ellipticK, and gammaP against erf; - the cusp mass against Schodel et al. 2018; - the closed-form slope radius; - the disk sphere mass; - the Jacobi-integral check now requires the orbit to stay. - table.mjs exits 1 on NaN or Infinity; the CI job gets a timeout. - Sources: corrected citations, DOIs and provenance labels, as read in #37's review. - Root README: the numbers match the study, and the run block lists the orbits table. Co-Authored-By: Claude Opus 5.5 Claude-Session: https://claude.ai/code/session_01C2bpiHyHSkC8DAL5CcyciK --- .github/workflows/test.yml | 1 + README.md | 11 +- orbits/README.md | 259 +++++++++++++++++++++---------------- orbits/model.js | 72 ++++++++--- orbits/table.mjs | 95 +++++++++----- tests/orbits.mjs | 95 ++++++++++---- 6 files changed, 340 insertions(+), 193 deletions(-) diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml index c44831d..37ba1c7 100644 --- a/.github/workflows/test.yml +++ b/.github/workflows/test.yml @@ -17,6 +17,7 @@ concurrency: jobs: test: runs-on: ubuntu-latest + timeout-minutes: 20 strategy: matrix: node-version: [18, 20, 22] diff --git a/README.md b/README.md index 1fa58f5..86cd5c0 100644 --- a/README.md +++ b/README.md @@ -20,10 +20,11 @@ with Claude (Anthropic), for [Foundation0](https://github.com/Foundation-0/f0). solution at any scale. The nesting lives in the equations, not the objects. - **Solar systems are not little galaxies, except at the Milky Way's centre.** Around one dominant mass, orbits follow Kepler's law at every radius. The planets match it to 0.13%, about what they weigh together against the Sun. Across - the Milky Way's disk the speed instead stays between 200 and 235 km/s, because the mass is spread out. A Kepler + the Milky Way's disk the speed instead stays between 200 and 234 km/s, because the mass is spread out. A Kepler region ends where its host's tide wins, at a density ratio rather than a size: 1.4 pc for the Sun in the Galaxy, - a third of an au for Jupiter. A planet keeps moons, and the Sun comets, out to about half that radius. The Milky - Way has its own small solar system around its central black hole, about a parsec in radius ([orbits](orbits/)). + 0.36 au for Jupiter. In a three-body model a planet keeps moons, and the Sun comets, out to about half that + radius, though passing stars cut the Oort cloud shorter. The Milky Way has its own small solar system around its + central black hole, about 1.3 pc in radius ([orbits](orbits/)). - **Quantum paths are fractal, between two lengths.** A free particle's path, recorded by position measurements finer than ħ/p, has fractal dimension 2, against 1 for a classical trajectory (Abbott and Wise, 1981); below ħ/mc, in a single-particle model, the speed of light makes it smooth again. Its surroundings draw @@ -207,13 +208,15 @@ node witness-map/table.mjs # every object at a 250 nm gap, then th node witness-map/table.mjs 1e-6 cryo # 1 µm gap, everything in a 4 K vacuum node notes/bh-lensing.mjs # every number in the black hole note node critical/table.mjs # the critical-point study's numbers (about 65 s) +node orbits/table.mjs quick # the orbits study's numbers, without its stability scans ``` `witness-map/model.js` is the whole model, including `gravAtom` for the black hole note. It loads in Node (`require`) and in a browser (`window.WM`). `npm test` also checks `gravitational-atom/atom.js` (second study, below). The third study's checks, `tests/critical.mjs`, add about 65 s of seeded Monte Carlo runs, most of it percolation with nine seeds. CI runs every line above except the `critical/table.mjs` line: it repeats the tests' -seeded runs and prints the results without checking them. +seeded runs and prints the results without checking them. The orbits study's full table (`node orbits/table.mjs`, +a few minutes) is not run by CI either. ## Checks diff --git a/orbits/README.md b/orbits/README.md index 7290b29..b2bf146 100644 --- a/orbits/README.md +++ b/orbits/README.md @@ -13,19 +13,19 @@ what ends it, and what, if anything, repeats from one scale to the next. | Question | Answer | |---|---| | Is a galaxy a big solar system? | No. Around one dominant mass, orbital speed falls as Kepler's v ∝ r^−1/2. Across the Milky Way's disk the speed falls only from 234 to 200 km/s between 5 and 25 kpc. Read as spherical, the mass inside grows 3.7-fold over that range, where a Kepler system's would stay fixed. Both figures come from Eilers et al.'s (2019) fit. | -| How exact is Kepler's law in the Solar System? | In JPL's 1800–2050 fit, n²a³ matches GM☉ to within 1.3×10⁻³ for all eight planets. That is about the planets' total mass, 1.34×10⁻³ of the Sun's: the planets pull on one another. P ∝ a^1.49986 over 0.39 to 30 au. | -| Does the Galaxy have a Kepler region? | Yes, around Sgr A*. The star S2 orbits it as a planet orbits the Sun, plus general relativity's precession: 12.2′ per orbit from their elements (they quote 12.1′). The region ends about 1.3 pc out, where the speed's log slope passes −1/4 (1.0 to 1.8 pc across the cluster-mass range). The stars inside weigh as much as the black hole by 1.8 to 3.2 pc. | -| What ends a Kepler region? | For a body orbiting in a host (a planet, the Sun), the host's tide. Satellites are held only inside its tidal radius, r_J³ = Gm/(4A(A − B)). Inside it, the body's mean density is 4A/Ω times the host's mean density inside the orbit (host taken as spherical): 3 for a planet around the Sun, 2.25 for the Sun in the Galaxy. For the Sun, r_J = 1.37 pc = 2.8×10⁵ au. At the Galaxy's centre, where Sgr A* sits still, the region ends instead where the surrounding stars' mass approaches the black hole's. | -| How much of that region is used? | About half, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.53 r_J around a planet and by 0.45 r_J (1.3×10⁵ au) for the Sun. Shen and Tremaine (2008) put stable prograde and retrograde moons out to about 0.5 and 0.7 r_H; the observed irregular moons lie out to about 0.6 r_H, the retrograde ones further out than the prograde. The Oort cloud's outer edge is placed between 10⁵ and 2×10⁵ au, which is 0.35 to 0.7 r_J. | -| What does repeat? | Kepler regions nested inside one another, each ending at its host's tide at a similar fraction of its tidal radius: moons in a planet's, planets and comets in the Sun's. Also flat disks: the planets' orbits are tilted by about 2° on average (6° for Mercury), and the Galaxy's thin disk is about 0.1 as thick as it is wide. | +| How exact is Kepler's law in the Solar System? | In JPL's 1800–2050 fit, n²a³ matches GM☉ to within 1.3×10⁻³ for all eight planets. That is about the planets' total mass, 1.34×10⁻³ of the Sun's. Jupiter's excess is mostly its own mass in the two-body law; the outer planets' is mostly the others' pull. P ∝ a^1.49986 over 0.39 to 30 au. | +| Does the Galaxy have a Kepler region? | Yes, around Sgr A*. The star S2 orbits it as a planet orbits the Sun, plus general relativity's precession: 12.2′ per orbit from their elements (they quote 12.1′). The speed's log slope passes −1/4 about 1.3 pc out (1.0 to 1.8 pc across the cluster-mass range; 0.7 to 1.9 pc for thresholds from −0.4 to −0.1). The stars inside weigh as much as the black hole by 1.8 to 3.2 pc. | +| What ends a Kepler region? | For a body orbiting in a host (a planet, the Sun), the host's tide. Satellites are held inside its tidal radius, r_J³ = Gm/(4A(A − B)), apart from quasi-satellite orbits beyond about 2 r_H that Shen and Tremaine (2008) find around Jupiter, Uranus and Neptune. Inside r_J, the body's mean density is 4A/Ω times the host's mean density inside the orbit (host taken as spherical): 3 for a planet around the Sun, 2.25 for the Sun in the Galaxy. For the Sun, r_J = 1.37 pc = 2.8×10⁵ au. At the Galaxy's centre, where Sgr A* sits still, the region ends instead where the surrounding stars' mass approaches the black hole's. | +| How much of that region is used? | About half, for orbits that do not plunge, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.52 r_J around a planet and by 0.435 r_J (1.23×10⁵ au) for the Sun. Polar orbits are lost much sooner, to Lidov–Kozai cycles: half within 0.14 to 0.15 r_H around a planet and 0.06 r_J for the Sun in 20 turns of the host. For near-circular moons near the ecliptic, Nesvorný et al. (2003) put the limits at about 0.4 and 0.7 r_H (prograde, retrograde), and the model gives 0.46 and 0.775. The Oort cloud's outer limit is placed between 6.5×10⁴ au (Correa-Otto and Calandra 2019, where passing stars set it) and 2×10⁵ au, which is 0.23 to 0.71 r_J. | +| What does repeat? | Kepler regions nested inside one another: moons in a planet's, planets and comets in the Sun's. Each is bounded by its host's tide at a similar fraction of the tidal radius, though for the Oort cloud passing stars cut it shorter. Also flat disks: the planets' orbits are tilted by about 2° on average (6° for Mercury), and the Galaxy's thin disk is 0.04 as thick as the Sun's orbit is wide. | ## Kepler's law in the Solar System Around one mass M, a circular orbit of radius r has speed √(GM/r), and P² = 4π²a³/G(M + m). Nothing in the law sets a size, so it holds from the Sun's surface out to wherever another pull competes. -JPL's approximate elements (Standish, Table 1, a fit to 1800–2050) give each planet's semi-major axis a and its -mean-longitude rate, n = L̇: +JPL's approximate elements (Standish and Williams, Table 1, a fit to 1800–2050) give each planet's semi-major axis a +and its mean-longitude rate, n = L̇: - This is the orbit's angular rate. For a near-circular orbit, n²a is the mean inward pull at a. That equals GM( 1 kpc), its log slope stays within ±0.15 of zero from 3.6 to 49 kpc. The r^−1.8 bulge is about as flat, from 7 to 180 pc, but that is a dense bulge, not a single mass. +- **The spherical reading.** Gauss's theorem gives the true mass inside a sphere (the disk by quadrature). v²R/G overstates it by 12% at 8 kpc: 9.00 against 8.00×10¹⁰ M☉. The true growth from 5 to 25 kpc is 4.03-fold rather than 3.65. +- **Γ.** Bovy (2017) warns that A and B cannot be read directly as the circular velocity and its slope, since his C and K are about a fifth of A. Γ = 2(1 − d ln v/d ln R) is 2.12 from Eilers et al.'s slope and 2.21 from MWPotential2014's, against 2.25 from A and B. r_J hardly moves: 1.366 pc from Eilers et al.'s curve. Going outward, the model's speed falls by less than Kepler's law would give: - **The model.** The log slope is −0.21 at 100 kpc and −0.27 at 200 kpc, because the NFW halo's mass still grows. -- **The data.** Gaia DR3 analyses differ on whether the measured curve turns down past about 19 kpc. Jiao et al. (2023) report "a Keplerian decline"; Koop et al. (2024) find the curves robust only to about 15 kpc. +- **The data.** Gaia DR3 analyses differ on whether the measured curve turns down past about 19 kpc. + - Jiao et al. (2023) report "a Keplerian decline". + - Koop et al. (2024) find Jeans-equation curves robust with their statistics only to about 15 kpc. In simulations, such curves are off by about 10%. ## An acceleration scale, and why this study does not use it @@ -174,9 +195,9 @@ m s⁻² (McGaugh, Lelli and Schombert 2016). - **The Galaxy's own pull at the Sun.** v²/R₀ = 2.1×10⁻¹⁰ m s⁻², or 1.7 g†. Wide binary stars at 2 to 30 thousand au test gravity at these accelerations, and the results disagree: -- Chae (2023) reports a 1.43 ± 0.06 boost. -- Banik et al. (2024) prefer Newtonian dynamics at 19σ and exclude MOND at 16σ. -- Later analyses still split. +- **Chae (2023).** Finds observed over predicted acceleration of 1.43 ± 0.06 at g_N = 10^−10.15 m s⁻². +- **Banik et al. (2024).** Prefer the best Newtonian model over the best MOND one at 19σ; their fit to α_grav excludes MOND at 16σ. +- **Later analyses** still split. The study reports the scale and leaves the question to the Witness Map and issue #3. @@ -187,80 +208,98 @@ total angular momentum (Souami and Souchay 2012). Computed from JPL's elements, - Mercury 6.3°, Venus 2.2°, Earth–Moon 1.6°, Mars 1.7°; - Jupiter 0.33°, Saturn 0.93°, Uranus 1.0°, Neptune 0.73°. -The rms of sin i is 0.045, or 0.023 without Mercury. The Galaxy's thin disk has a scale height of 300 ± 50 pc over -a scale length of 2.6 ± 0.5 kpc (Bland-Hawthorn and Gerhard 2016). Their ratio, 0.12, is the axis ratio of its -density contours. +The rms of sin i is 0.045, or 0.023 without Mercury. + +The Galaxy's thin disk has an exponential scale height of 300 ± 50 pc over a scale length of 2.6 ± 0.5 kpc +(Bland-Hawthorn and Gerhard 2016). Their ratio, 0.12, is the axis ratio of its density contours. The like-for-like +comparison with the planets, height over orbital radius, is 300 pc over the Sun's 8.2 kpc: 0.037, against the +planets' 0.045. Both are flat for the same reason: gas that radiates its energy but keeps its angular momentum settles into a disk. -What the disks orbit differs: one dominant mass, or mass spread through the disk and halo. +The Galaxy's present thickness is also set by the stars' later heating. What the disks orbit differs: one dominant +mass, or mass spread through the disk and halo. ## Run it ``` -node orbits/table.mjs # every number above except the 100-turn runs (about 50 s; section 5 takes nearly all of it) -node orbits/table.mjs quick # sections 1-4 only, well under a second -node orbits/table.mjs long # adds the 100-turn runs, about 2 minutes in all -node tests/orbits.mjs # the checks, about 2 s (also run by npm test) +node orbits/table.mjs # every computed number above except the 100-turn runs (about 3 minutes; section 5 takes nearly all of it) +node orbits/table.mjs quick # sections 1-4 only, well under a second (run by CI) +node orbits/table.mjs long # adds the 100-turn runs, about 5 minutes in all +node tests/orbits.mjs # the checks, about 3 s (also run by npm test) ``` +The table exits with an error if any printed value is not a finite number. + ## Checks -`tests/orbits.mjs` has 24 checks. Each prints its source, the published or expected value, the model's value and -the tolerance. +`tests/orbits.mjs` has 35 checks. Each prints its source, the published or expected value, the model's value and +the tolerance. "#37's review" marks values a reviewer read in the paper's arXiv source. | What | Against | Kind | |---|---|---| -| In the 1800–2050 fit, n²a³/GM☉ − 1 below the planets' mass fraction for all eight | JPL Table 1, IAU 2015, DE430 | published elements; an observed size, not a bound | -| Exponent 3/2, to the most the previous check allows | the same | model-internal, independent of GM☉ | +| In the 1800–2050 fit, n²a³/GM☉ − 1 below the planets' mass fraction for all eight, and below 5×10⁻⁵ for the inner four | JPL Table 1, IAU 2015, DE430 | published elements; an observed size, not a bound | +| Exponent 3/2, to the most the first check allows | the same | model-internal, independent of GM☉ | | 4A/Ω = 3 from the model's own curve near Sgr A*; Jacobi radius = Hill radius for a point mass | the model | model-internal | -| The Sun's r_J = 2.8×10⁵ au | Correa-Otto and Calandra 2019 | search text | -| S2: Kepler mass, 12.1′ precession, 120 au, 7,700 km/s | GRAVITY 2020 | Table E.1, appendix B and introduction, via a transcription | -| Stars weigh 2 M_BH at 3 to 4 pc | Alexander 2005; Feldmeier-Krause et al. 2017; GRAVITY 2024 | search text | -| Kepler slope −1/2 inside 0.1 pc | — | model | -| MWPotential2014: the 220 km/s normalisation; halo mass 0.8×10¹² inside 245 kpc; shape within 5% of Eilers et al.; flat (abs. slope < 0.25) at 5 to 25 kpc | Bovy 2015 Table 1, as secondary papers quote it; Eilers et al. 2019 | model-internal; search text; abstract | -| Jupiter's and Saturn's tilts to the invariable plane | Souami and Souchay 2012 | search text | -| Planet: half the prograde orbits escaped by 0.5 ± 0.1 r_H; retrograde kept at 0.6, lost by 0.9 | Shen and Tremaine 2008 (0.5, 0.7) | abstract, search text | -| Retrograde in the plane at 0.875 r_J: lost around a point mass, kept by the Sun | — | model | -| The Jacobi integral conserved over 20 turns; the Galaxy's ν²/Ω² in SI | equations of motion; Poisson's equation | model-internal; arithmetic | +| The Sun's r_J = 2.8×10⁵ au | Correa-Otto and Calandra 2019 | #37's review; on Jiang and Tremaine's nearby constants | +| S2: Kepler mass, 12.1′ precession, 120 au, 7,700 km/s | GRAVITY 2020 | Table E.1, appendix B, sect. 1.2 | +| Stars weigh 2 M_BH at 3 to 4 pc; the cusp's mass inside 3 pc and 0.01 pc | Schödel et al. 2018; Alexander 2005; Feldmeier-Krause et al. 2017; GRAVITY 2024 | #37's review | +| Kepler slope −1/2 inside 0.1 pc; the −1/4 radius in closed form against bisection | — | model; model-internal | +| ellipticK(1/√2); gammaP(½, x) = erf √x at three points | closed forms | arithmetic | +| MWPotential2014: the 220 km/s normalisation; halo mass 0.8×10¹² inside 245 kpc; the disk's sphere mass tending to its total; shape within 5% of Eilers et al.; flat (abs. slope < 0.25) at 5 to 25 kpc | Bovy 2015 Table 1; Eilers et al. 2019 | model-internal; #37's review | +| Jupiter's and Saturn's tilts to the invariable plane | Souami and Souchay 2012 | search text, not confirmed in the paper | +| Planet: prograde mostly kept at 0.4 r_H and lost by 0.6; prograde edge 0.4 ± 0.1; retrograde edge between 0.7 and 1 r_H | Nesvorný et al. 2003 and Shen and Tremaine 2008, as Shen and Tremaine give them | model; #37's review | +| Retrograde in the plane at 0.85 r_J: mostly lost around a point mass, kept by the Sun | — | model | +| The Jacobi integral kept over 20 turns on an orbit that stays; the Galaxy's ν²/Ω² in SI | equations of motion; Poisson's equation | model-internal; arithmetic | +| The Sun's polar orbits at 0.3 r_J lost by plunging, not escape | Lidov–Kozai | model | | The Sun: none escape from 0.3 r_J, most by 0.6 | — | model | -| The Sun's all-directions edge within the Oort cloud's 1 to 2×10⁵ au | Correa-Otto and Calandra 2019; Dones et al. 2004 | search text; as cited | +| The Sun's all-directions edge within the Oort cloud's quoted 6.5×10⁴ to 2×10⁵ au | Correa-Otto and Calandra 2019; Dones et al. 2004 | #37's review; as cited | ## What the model leaves out - **The planets' long-period pull on one another.** Kepler's law is checked only to the size of the planets' masses. -- **Hill's problem is the tide to first order, on a circular host orbit.** The study drops the Sun's vertical oscillation through the disk (82 Myr with this model's ν) and its radial epicycles. It also leaves out passing stars and molecular clouds, which also strip the Oort cloud, and the planets' scattering of comets that plunge inward. -- **Circular starting orbits only.** Real moons and comets start on eccentric orbits. The edges are set at half the orbits escaping within 20 turns of the host, on a grid of 0.025 r_J (in the plane) or 0.1 r_J (all directions). +- **Hill's problem is the tide to first order, on a circular host orbit.** The study drops the Sun's vertical oscillation through the disk (82 Myr with this model's ν) and its radial epicycles. It also leaves out: + - passing stars and molecular clouds, which set the Oort cloud's actual edge; + - the planets' scattering of comets that plunge inward; + - for moons, the planet's oblateness and its other moons, which damp Lidov–Kozai cycles close in. +- **Circular starting orbits only.** Real moons and comets start on eccentric orbits. + - **The edges' definition.** They are set at half the orbits escaping, or lost, within 20 turns of the host, on a grid of 0.025 r_J (single planes) or 0.05 r_J (all directions). + - **Plunge radius.** It is 10⁻³ r_J. Raising it to 10⁻² moves the Sun's all-directions edge from 0.448 to 0.469 (24 directions). + - **Longer runs** move the edges inward a little (see the 100-turn values above). - **MWPotential2014 is one model among many.** Its halo sets the flat range, and it has no nuclear star cluster. The checks pin its halo; the disk's and bulge's shape parameters are taken from galpy's source and not otherwise tested. The Galactic-centre numbers use the separate power-law cusp above, which is valid only inside its 3 pc break. -- **Spherical mass estimates.** v²R/G and 3Ω²/4πG treat a flattened galaxy as spherical. For a disk the true enclosed mass and mean density differ somewhat, which shifts the 3.7-fold growth and the 2.25, but not the contrast with Kepler's fixed mass. +- **Spherical mass estimates.** v²R/G and 3Ω²/4πG treat a flattened galaxy as spherical. In MWPotential2014 this overstates the mass inside 8 kpc by 12%. It shifts the 3.7-fold growth to 4.0 and the Sun's 2.25 to 2.56, but not the contrast with Kepler's fixed mass. ## Sources -Read here only through search results and abstracts, or as noted. Each entry says how it was read. - -- W. M. Folkner et al., *The Planetary and Lunar Ephemerides DE430 and DE431*, IPN Progress Report 42-196 (2014). GM values seen in copies of the file. -- E. M. Standish, *Keplerian Elements for Approximate Positions of the Major Planets*, JPL Solar System Dynamics, Table 1 (1800–2050). Seen in a copy of the file. -- A. Prša et al., *Nominal values for selected solar and planetary quantities: IAU 2015 Resolution B3*, Astron. J. 152, 41 (2016). https://doi.org/10.3847/0004-6256/152/2/41. Search text of the resolution. -- D. Souami and J. Souchay, *The solar system's invariable plane*, A&A 543, A133 (2012). https://doi.org/10.1051/0004-6361/201219011. Search text. -- J. Binney and S. Tremaine, *Galactic Dynamics*, 2nd ed., Princeton (2008), sec. 8.3. From memory. -- J. Bovy, *Galactic rotation in Gaia DR1*, MNRAS 468, L63 (2017). https://arxiv.org/abs/1610.07610. Abstract, search text; journal volume and page recalled, not confirmed. -- J. A. Correa-Otto and M. F. Calandra, MNRAS 490, 2495 (2019). https://doi.org/10.1093/mnras/stz2671. Search text: "r_J ~ 2.8 × 10^5 au", and a structure near 10⁵ au marking an outer stability boundary. -- Y. Shen and S. Tremaine, *Stability of Distant Satellites of the Giant Planets in the Solar System*, AJ (2008). https://arxiv.org/abs/0806.2882. Abstract and paper text, search text. -- C. F. McKee, A. Parravano and D. J. Hollenbach, *Stars, gas, and dark matter in the solar neighborhood*, ApJ 814, 13 (2015). https://arxiv.org/abs/1509.05334. Search text, secondary. -- M. Heisler and S. Tremaine, *The influence of the Galactic tidal field on the Oort comet cloud*, Icarus 65, 13 (1986). Abstract, search text. +Each entry says how it was read. "#37's review" means a reviewer checked the values used here in the arXiv source, +or in the primary file, during #37's review. + +- W. M. Folkner et al., *The Planetary and Lunar Ephemerides DE430 and DE431*, IPN Progress Report 42-196 (2014), Table 8. #37's review. +- E. M. Standish and J. G. Williams, *Keplerian Elements for Approximate Positions of the Major Planets*, JPL Solar System Dynamics, Table 1 (1800–2050). Checked against the SSD page in #37's review. +- A. Prša et al., *Nominal values for selected solar and planetary quantities: IAU 2015 Resolution B3*, Astron. J. 152, 41 (2016). https://doi.org/10.3847/0004-6256/152/2/41. #37's review. +- D. Souami and J. Souchay, *The solar system's invariable plane*, A&A 543, A133 (2012). https://doi.org/10.1051/0004-6361/201219011. Search text; not reached in #37's review. The plane's orientation is confirmed through Li, Xia and Zhou 2019; the per-planet tilts are not. +- Y.-F. Jiang and S. Tremaine, *The evolution of wide binary stars*, MNRAS 401, 977 (2010; journal reference recalled). https://arxiv.org/abs/0907.2952. The tidal-radius formula and constants, as #37's review quotes them. +- J. Binney and S. Tremaine, *Galactic Dynamics*, 2nd ed., Princeton (2008). The formula, as Jiang and Tremaine use it; section not confirmed. +- J. Bovy, *Galactic rotation in Gaia DR1*, MNRAS 468, L63 (2017). https://doi.org/10.1093/mnrasl/slx027. #37's review. +- J. A. Correa-Otto and M. F. Calandra, *Stability in the most external region of the Oort Cloud: evolution of the ejected comets*, MNRAS 490, 2495 (2019). https://doi.org/10.1093/mnras/stz2671. #37's review (sect. 4.3). +- Y. Shen and S. Tremaine, *Stability of Distant Satellites of the Giant Planets in the Solar System*, AJ 136, 2453 (2008). https://doi.org/10.1088/0004-6256/136/6/2453. #37's review. +- D. Nesvorný, J. L. A. Alvarellos, L. Dones and H. F. Levison, *Orbital and collisional evolution of the irregular satellites*, AJ 126, 398 (2003). As Shen and Tremaine quote it; reference recalled. +- M. L. Lidov, Planet. Space Sci. 9, 719 (1962), and Y. Kozai, AJ 67, 591 (1962). Recalled. +- C. F. McKee, A. Parravano and D. J. Hollenbach, *Stars, gas, and dark matter in the solar neighborhood*, ApJ 814, 13 (2015). https://arxiv.org/abs/1509.05334. #37's review. +- J. Heisler and S. Tremaine, *The influence of the Galactic tidal field on the Oort comet cloud*, Icarus 65, 13 (1986). Search text; not reached in #37's review. - L. Dones, P. R. Weissman, H. F. Levison and M. J. Duncan, *Oort cloud formation and dynamics*, in *Comets II*, 153 (2004). As cited in search text, for an outer edge "as far as 2 × 10^5 AU". -- GRAVITY Collaboration, *Detection of the Schwarzschild precession in the orbit of the star S2 near the Galactic centre massive black hole*, A&A 636, L5 (2020). https://doi.org/10.1051/0004-6361/202037813. Abstract, introduction, appendix B and Table E.1, read in a transcription. -- GRAVITY Collaboration, A&A 657, L12 (2022). https://doi.org/10.1051/0004-6361/202142465. Search text. -- GRAVITY Collaboration (2024). https://arxiv.org/abs/2409.12261. Search text, for the 3.9 pc influence radius in their models. -- R. Schödel, D. Merritt and A. Eckart, A&A 502, 91 (2009). https://doi.org/10.1051/0004-6361/200810922. Abstract, search text. -- R. Schödel et al., A&A 609, A27 (2018). Abstract, search text. -- T. Alexander, Phys. Rep. 419, 65 (2005). As cited by Feldmeier-Krause et al. 2017. -- A. Feldmeier-Krause et al., *Triaxial orbit-based modelling of the Milky Way Nuclear Star Cluster*, MNRAS 466, 4040 (2017). https://arxiv.org/abs/1701.01583. Search text. -- D. Merritt, *Dynamics and Evolution of Galactic Nuclei*, Princeton (2013). Definition only, as cited by GRAVITY Collaboration 2024. -- J. Bovy, *galpy: A Python Library for Galactic Dynamics*, ApJS 216, 29 (2015). https://doi.org/10.1088/0067-0049/216/2/29. MWPotential2014's parameters read in galpy's source. -- A.-C. Eilers, D. W. Hogg, H.-W. Rix and M. K. Ness, ApJ 871, 120 (2019). https://doi.org/10.3847/1538-4357/aaf648. Abstract, search text. -- J. Bland-Hawthorn and O. Gerhard, *The Galaxy in Context*, ARA&A 54, 529 (2016). https://doi.org/10.1146/annurev-astro-081915-023441. As cited in search text. -- Y. Jiao et al., A&A 678, A208 (2023). https://doi.org/10.1051/0004-6361/202347513. Abstract, search text. -- H. Koop et al., A&A 692, A50 (2024). https://arxiv.org/abs/2405.19028. Abstract, search text. -- S. S. McGaugh, F. Lelli and J. M. Schombert, PRL 117, 201101 (2016). https://doi.org/10.1103/PhysRevLett.117.201101. Search text. -- K.-H. Chae, ApJ 952, 128 (2023). https://doi.org/10.3847/1538-4357/ace101. Search text. -- I. Banik et al., MNRAS 527, 4573 (2024). https://doi.org/10.1093/mnras/stad3393. Abstract, search text. +- GRAVITY Collaboration, *Detection of the Schwarzschild precession in the orbit of the star S2 near the Galactic centre massive black hole*, A&A 636, L5 (2020). https://doi.org/10.1051/0004-6361/202037813. #37's review (sect. 1.2, appendix B, Table E.1). +- GRAVITY Collaboration, A&A 657, L12 (2022). https://doi.org/10.1051/0004-6361/202142465. #37's review. +- GRAVITY Collaboration, A&A 692, A242 (2024). https://doi.org/10.1051/0004-6361/202452274. #37's review (sect. 4.3). +- R. Schödel, D. Merritt and A. Eckart, A&A 502, 91 (2009). https://doi.org/10.1051/0004-6361/200810922. #37's review. +- R. Schödel et al., A&A 609, A27 (2018). https://doi.org/10.1051/0004-6361/201730452. #37's review (sect. 4.5). +- T. Alexander, *Stellar processes near the massive black hole in the Galactic center*, Phys. Rep. 419, 65 (2005). https://doi.org/10.1016/j.physrep.2005.08.002. #37's review (sect. 3.1.2). +- A. Feldmeier-Krause et al., *Triaxial orbit-based modelling of the Milky Way Nuclear Star Cluster*, MNRAS 466, 4040 (2017). https://doi.org/10.1093/mnras/stw3377. #37's review (sect. 3.3.2). +- D. Merritt, *Dynamics and Evolution of Galactic Nuclei*, Princeton (2013), and D. Merritt, in *Coevolution of Black Holes and Galaxies*, ed. L. C. Ho (2004; recalled). Definitions only, as GRAVITY 2024 and Feldmeier-Krause et al. cite them. +- J. Bovy, *galpy: A Python Library for Galactic Dynamics*, ApJS 216, 29 (2015). https://doi.org/10.1088/0067-0049/216/2/29. Table 1 and galpy's source, #37's review. +- A.-C. Eilers, D. W. Hogg, H.-W. Rix and M. K. Ness, ApJ 871, 120 (2019). https://doi.org/10.3847/1538-4357/aaf648. #37's review (eq. 7). +- J. Bland-Hawthorn and O. Gerhard, *The Galaxy in Context*, ARA&A 54, 529 (2016). https://doi.org/10.1146/annurev-astro-081915-023441. #37's review. +- Y. Jiao et al., A&A 678, A208 (2023). https://doi.org/10.1051/0004-6361/202347513. #37's review. +- O. Koop et al., A&A 692, A50 (2024). https://arxiv.org/abs/2405.19028. #37's review. +- S. S. McGaugh, F. Lelli and J. M. Schombert, PRL 117, 201101 (2016). https://doi.org/10.1103/PhysRevLett.117.201101. #37's review. +- K.-H. Chae, ApJ 952, 128 (2023). https://doi.org/10.3847/1538-4357/ace101. #37's review. +- I. Banik et al., MNRAS 527, 4573 (2024). https://doi.org/10.1093/mnras/stad3393. #37's review. diff --git a/orbits/model.js b/orbits/model.js index e981a26..8a3de42 100644 --- a/orbits/model.js +++ b/orbits/model.js @@ -17,8 +17,8 @@ * by a ratio of mean densities, not a size. The host's density here is its mean inside the orbit, not its local * density. * - * Flat curves. Where the enclosed mass grows as M( -(2 / Math.PI) * ellipticK(2 * Math.sqrt(r * b) / (r + b)) / (r + b); // per unit G m - function ringExcess(p, { own = true } = {}) { + function ringExcess(p, { own = true, which = () => true } = {}) { let ex = own ? 1 / p.sunOver : 0; for (const q of PLANETS) { - if (q === p) continue; + if (q === p || !which(q)) continue; const h = 1e-6 * p.a, pull = (ringPotential(p.a + h, q.a) - ringPotential(p.a - h, q.a)) / (2 * h); // inward > 0 ex += p.a * p.a * pull / q.sunOver; } @@ -135,7 +136,8 @@ } // The stars around Sgr A*: inside the break near 3 pc the density falls as r^-gamma, gamma = 1.13 (Schodel et al. - // 2018), so their mass grows as r^(3 - gamma); 0.5 to 1.5 million solar masses lie inside 1 pc (Schodel, Merritt & + // 2018: the light-density slope of the innermost 1-2 pc, which mass follows at constant M/L; their robust range is + // 1.1-1.4), so their mass grows as r^(3 - gamma); 0.5 to 1.5 million solar masses lie inside 1 pc (Schodel, Merritt & // Eckart 2009). Radii in pc. const CUSP = { M1: 1.0e6, M1lo: 0.5e6, M1hi: 1.5e6, gamma: 1.13, rBreak: 3 }; const cuspMass = (r, M1 = CUSP.M1, gamma = CUSP.gamma) => M1 * Math.pow(r, 3 - gamma); @@ -145,6 +147,10 @@ const cuspRadius = (k, Mbh, M1 = CUSP.M1, gamma = CUSP.gamma) => Math.pow(k * Mbh / M1, 1 / (3 - gamma)); // dln v / dln r for Sgr A* plus the cusp: (dln M / dln r - 1) / 2. const cuspSlope = (r, Mbh, M1 = CUSP.M1, gamma = CUSP.gamma) => { const m = cuspMass(r, M1, gamma); return ((3 - gamma) * m / (Mbh + m) - 1) / 2; }; + // Where that slope equals s (between -1/2 and (2 - gamma)/2), in closed form: the stars' mass is then + // Mbh (2s + 1) / ((3 - gamma) - (2s + 1)). + const cuspSlopeRadius = (s, Mbh, M1 = CUSP.M1, gamma = CUSP.gamma) => + Math.pow(Mbh * (2 * s + 1) / ((3 - gamma) - (2 * s + 1)) / M1, 1 / (3 - gamma)); // ---------- the Milky Way ---------- function lgamma(z) { // Lanczos, g = 7; |error| < 1e-14 for z > 0 @@ -200,6 +206,21 @@ const v = R => Math.sqrt(v2(R)); const slope = R => { const h = 1e-4; return Math.log(v(R * (1 + h)) / v(R * (1 - h))) / Math.log((1 + h) / (1 - h)); }; const bulgeMass = r => M.bulge * shape.bulge(r); + // The disk's mass inside a sphere of radius r, by Gauss's theorem: (r^2 / G) times the mean over the sphere of the + // outward radial pull, here r^2/2 times the integral over mu = z/r of dPhi/dr for the Miyamoto-Nagai potential + // -G M / sqrt(R^2 + (a + sqrt(z^2 + b^2))^2) (Simpson's rule in mu on [0, 1], doubled by symmetry). + const diskSphereMass = r => { + const n = 4000, a = par.a, b = par.b; + const f = mu => { + const z = r * mu, R = r * Math.sqrt(Math.max(0, 1 - mu * mu)), zeta = Math.sqrt(z * z + b * b), D = Math.hypot(R, a + zeta); + return (R * R + (a + zeta) * z * z / zeta) / (r * D * D * D); + }; + let sum = f(0) + f(1); + for (let i = 1; i < n; i++) sum += (i % 2 ? 4 : 2) * f(i / n); + return M.disk * r * r * sum / (3 * n); + }; + // The total mass inside a sphere of radius r [M_sun]. + const sphereMass = r => M.bh + bulgeMass(r) + diskSphereMass(r) + M.halo * shape.halo(r); // the smallest radius above r0 where f crosses zero, by a log grid then bisection const cross = (f, r0, r1) => { let lo = r0, flo = f(lo); @@ -215,7 +236,7 @@ return NaN; }; return { - par, M, parts, v, slope, bulgeMass, cross, + par, M, parts, v, slope, bulgeMass, diskSphereMass, sphereMass, cross, // where the bulge's stars inside r first weigh k times Sgr A* (k = 1, equal mass; k = 2, the influence radius) influence: (k = 1) => cross(r => bulgeMass(r) - k * M.bh, 1e-6, 1), // the longest run of radii (kpc, on a grid 0.5% apart) where |dln v / dln R| < tol; by default outside the bulge @@ -262,9 +283,9 @@ e: [71 / 57600, 0, -71 / 16695, 71 / 1920, -17253 / 339200, 22 / 525, -1 / 40], }; // Follows one orbit for `orbits` turns of the host (time 2 pi orbits). Fate: 'escapes' once r > rEsc, 'plunges' - // once r < rIn (1e-3 r_J is about a giant planet's own radius; for the Sun it is 280 au, outside the planets), - // otherwise 'stays'. Returns the final state s too. - function hillOrbit(host, s0, { orbits = 20, rEsc = 3, rIn = 1e-3, tol = 1e-10 } = {}) { + // once r < rIn (1e-3 r_J is 0.74 of Jupiter's radius; for the Sun it is 283 au, outside the planets), otherwise + // 'stays'. Returns the final state s too. Throws on a non-finite step or after maxSteps, rather than looping. + function hillOrbit(host, s0, { orbits = 20, rEsc = 3, rIn = 1e-3, tol = 1e-10, maxSteps = 5e6 } = {}) { const T = 2 * Math.PI * orbits, k = Array.from({ length: 7 }, () => new Float64Array(6)); const y = new Float64Array(6), s = Float64Array.from(s0); let t = 0, h = 1e-3, steps = 0; @@ -280,6 +301,8 @@ let ei = 0; for (let m = 0; m < 7; m++) ei += DP.e[m] * k[m][i]; err = Math.max(err, Math.abs(h * ei) / (tol * (1 + Math.abs(s[i])))); } + if (!Number.isFinite(err) || !(h > 0)) throw new Error(`hillOrbit: non-finite step at t = ${t}`); + if (steps > maxSteps) throw new Error(`hillOrbit: more than ${maxSteps} steps`); if (err <= 1) { t += h; steps++; for (let i = 0; i < 6; i++) { s[i] = y[i]; k[0][i] = k[6][i]; } @@ -311,24 +334,31 @@ return out; } // Fractions of orbits that stay, escape and plunge, over the given orbit normals and nPhase starting angles each. + // Hill's equations are symmetric under r -> -r, v -> -v, and a start at phi + pi is the inverted start at phi, so the + // starting angles are spread over [0, pi) only: each one is a different orbit. function hillFates(host, a, normals, { nPhase = 4, ...opt } = {}) { const n = { stays: 0, escapes: 0, plunges: 0 }; - for (const nv of normals) for (let p = 0; p < nPhase; p++) n[hillOrbit(host, hillStart(host, a, nv, 2 * Math.PI * (p + 0.5) / nPhase), opt).fate]++; + for (const nv of normals) for (let p = 0; p < nPhase; p++) n[hillOrbit(host, hillStart(host, a, nv, Math.PI * (p + 0.5) / nPhase), opt).fate]++; const tot = normals.length * nPhase; return { stays: n.stays / tot, escapes: n.escapes / tot, plunges: n.plunges / tot }; } - // Fates on a grid of radii (in r_J), and the radius where the fraction that escapes first reaches one half (linear - // interpolation between grid points; NaN if it never does). - function hillScan(host, normals, as, opt = {}) { - const rows = as.map(a => ({ a, ...hillFates(host, a, normals, opt) })); - let edge = NaN; + // Fates on a grid of radii (in r_J), and the radius where a fraction first reaches one half (linear interpolation + // between grid points; NaN if it never does): edge for the orbits that escaped, edgeLost for those escaped or plunged. + // A half reached already at the first grid point returns that point, so start the grid inside the edge. + function halfPoint(rows, frac) { for (let i = 0; i < rows.length; i++) { - if (rows[i].escapes >= 0.5) { - edge = i === 0 ? rows[0].a : rows[i - 1].a + (rows[i].a - rows[i - 1].a) * (0.5 - rows[i - 1].escapes) / (rows[i].escapes - rows[i - 1].escapes); - break; + const f = frac(rows[i]); + if (f >= 0.5) { + if (i === 0) return rows[0].a; + const g = frac(rows[i - 1]); + return rows[i - 1].a + (rows[i].a - rows[i - 1].a) * (0.5 - g) / (f - g); } } - return { rows, edge }; + return NaN; + } + function hillScan(host, normals, as, opt = {}) { + const rows = as.map(a => ({ a, ...hillFates(host, a, normals, opt) })); + return { rows, edge: halfPoint(rows, r => r.escapes), edgeLost: halfPoint(rows, r => r.escapes + r.plunges) }; } // McGaugh, Lelli & Schombert 2016: the acceleration where the radial acceleration relation bends. @@ -336,7 +366,7 @@ const api = { C, PLANETS, INVARIABLE, tiltToInvariable, planetMassFraction, kepler, ellipticK, ringExcess, keplerExponent, jacobi, hill, oortKepler, oortFlat, densityRatio, - radiusAtAcceleration, S2, orbitOf, CUSP, cuspMass, cuspRadius, cuspSlope, lgamma, gammaP, MWP14, SGRA, galaxy, GDAGGER, + radiusAtAcceleration, S2, orbitOf, CUSP, cuspMass, cuspRadius, cuspSlope, cuspSlopeRadius, lgamma, gammaP, MWP14, SGRA, galaxy, GDAGGER, HOST_KEPLER, hostDisk, hillOrbit, hillStart, sphere, hillFates, hillScan, }; if (typeof module !== 'undefined' && module.exports) module.exports = api; else root.ORB = api; diff --git a/orbits/table.mjs b/orbits/table.mjs index 3e4a90c..0d0c23f 100644 --- a/orbits/table.mjs +++ b/orbits/table.mjs @@ -1,11 +1,16 @@ -// Prints every number in orbits/README.md. Run: node orbits/table.mjs (about 50 s; the stability scans of section 5 -// take nearly all of it), node orbits/table.mjs quick (well under a second; section 5 left out), or -// node orbits/table.mjs long (adds the all-directions scans followed for 100 turns of the host; about 2 minutes in all). +// Prints every computed number in orbits/README.md. Run: node orbits/table.mjs (a few minutes; the stability scans of +// section 5 take nearly all of it), node orbits/table.mjs quick (well under a second; section 5 left out), or +// node orbits/table.mjs long (adds the all-directions scans followed for 100 turns of the host). +// Exits with code 1 if any printed line holds NaN or Infinity. import { createRequire } from 'node:module'; const O = createRequire(import.meta.url)('./model.js'); const { C } = O; const e3 = x => x.toExponential(3), f = (x, d = 3) => x.toFixed(d); const quick = process.argv[2] === 'quick', long = process.argv[2] === 'long'; +const print = console.log; +let bad = false; +console.log = (...a) => { if (/NaN|Infinity/.test(a.join(' '))) bad = true; print(...a); }; +process.on('exit', () => { if (bad) { print('non-finite value printed'); process.exitCode = 1; } }); console.log('1. Kepler in the Solar System (JPL approximate elements, Table 1, 1800-2050; GM_sun IAU 2015; masses DE430)\n'); console.log(' n = the mean-longitude rate Ldot (the angular rate); excess = n^2 a^3/GM_sun - 1'); @@ -22,9 +27,15 @@ for (const p of O.PLANETS) { } const mu = O.planetMassFraction(); console.log(` all eight planets together: ${e3(mu)} of the Sun's mass; largest |excess| ${e3(Math.max(...O.PLANETS.map(p => Math.abs(O.kepler(p).excess))))}`); +console.log(` largest |excess| of the inner four: ${e3(Math.max(...O.PLANETS.slice(0, 4).map(p => Math.abs(O.kepler(p).excess))))}`); console.log(` fitted exponent of P against a: ${f(O.keplerExponent(), 6)} over all eight (0.39 to 30 au), ${f(O.keplerExponent(O.PLANETS.slice(0, 4)), 6)} over the inner four; Kepler 1.5`); console.log(` mass inside Neptune's orbit over mass inside Mercury's: 1 + ${e3(mu - 1 / O.PLANETS[7].sunOver)}`); console.log(` the other planets' rings alone: ${O.PLANETS.map(p => `${p.name} ${e3(O.ringExcess(p, { own: false }))}`).join(', ')}`); +{ + const J = O.PLANETS[4]; + console.log(` Jupiter's ring term from the planets inside its orbit: ${e3(O.ringExcess(J, { own: false, which: q => q.a < J.a }))};` + + ` outside: ${e3(O.ringExcess(J, { own: false, which: q => q.a > J.a }))}`); +} console.log(` Saturn's ring estimate minus its fit: ${e3(O.ringExcess(O.PLANETS[5]) - O.kepler(O.PLANETS[5]).excess)}`); console.log('\n2. Where a Kepler domain ends: the tidal (Jacobi) radius, r_J^3 = G m / (4 A (A - B))\n'); @@ -32,8 +43,9 @@ const OORT = { A: 15.3, B: -11.9, dA: 0.4, dB: 0.4 }; // Bovy 2017 const rJ = O.jacobi(C.GMsun, OORT.A, OORT.B); const rJs = [[OORT.A + OORT.dA, OORT.B], [OORT.A - OORT.dA, OORT.B], [OORT.A, OORT.B + OORT.dB], [OORT.A, OORT.B - OORT.dB]].map(([A, B]) => O.jacobi(C.GMsun, A, B) / C.pc); console.log(` the Sun in the Galaxy (A = ${OORT.A}, B = ${OORT.B} km/s/kpc, Bovy 2017): r_J = ${f(rJ / C.pc)} pc = ${f(rJ / C.au / 1e5, 2)}e5 au` + - ` (moving A or B by its statistical error: ${f(Math.min(...rJs))} to ${f(Math.max(...rJs))} pc; adding A's 0.5 systematic: ` + - `${f(O.jacobi(C.GMsun, OORT.A + 0.9, OORT.B) / C.pc)} to ${f(O.jacobi(C.GMsun, OORT.A - 0.9, OORT.B) / C.pc)} pc)`); + ` (moving A or B by its error: ${f(Math.min(...rJs))} to ${f(Math.max(...rJs))} pc)`); +console.log(` with Jiang & Tremaine's A = 14.8 and Omega = 29.6 km/s/kpc (B = A - Omega), as Correa-Otto & Calandra use: ` + + `${f(O.jacobi(C.GMsun, 14.8, 14.8 - 29.6) / C.au / 1e5, 2)}e5 au`); console.log(` the Sun's mean density inside r_J over the Galaxy's inside the Sun's orbit: 4A/(A - B) = ${f(O.densityRatio(OORT.A, OORT.B), 2)}` + ` (3 for a point-mass host, 2 for a flat rotation curve; 2(1 - dln v/dln R) with dln v/dln R = -(A + B)/(A - B) = ${f(-(OORT.A + OORT.B) / (OORT.A - OORT.B))})`); const rhoJ = 1 / (4 / 3 * Math.PI * (rJ / C.pc) ** 3); @@ -51,14 +63,17 @@ console.log(` Schwarzschild precession 6 pi GM / (c^2 a (1 - e^2)) = ${f(s2M.p console.log(` pericentre ${f(s2M.periAU, 0)} au = ${f(s2M.periOverRs, 0)} Schwarzschild radii, at ${f(s2M.periSpeed, 0)} km/s (${f(s2M.periSpeed * 1e3 / C.c * 100, 2)}% of c)`); const MBH = O.SGRA.M, K = O.CUSP; console.log(` Sgr A* today: ${MBH / 1e6}e6 M_sun (GRAVITY Collaboration 2022). The stars around it: ${K.M1lo / 1e6} to ${K.M1hi / 1e6}e6 M_sun inside 1 pc` + - ` (Schodel, Merritt & Eckart 2009), density ~ r^-${K.gamma} inside ~${K.rBreak} pc (Schodel et al. 2018)`); -const quarter = M1 => { let lo = 1e-3, hi = 30; for (let i = 0; i < 100; i++) { const m = Math.sqrt(lo * hi); if (O.cuspSlope(m, MBH, M1) < -0.25) lo = m; else hi = m; } return lo; }; + ` (Schodel, Merritt & Eckart 2009), density ~ r^-${K.gamma} inside ~${K.rBreak} pc (Schodel et al. 2018); mass ~ r^${f(3 - K.gamma, 2)}`); for (const [lab, M1] of [['low', K.M1lo], ['middle', K.M1], ['high', K.M1hi]]) { - console.log(` ${lab.padEnd(6)} (${M1 / 1e6}e6 inside 1 pc): dln v/dln r passes -1/4 at ${f(quarter(M1), 2)} pc; stars weigh as much as Sgr A* inside` + - ` ${f(O.cuspRadius(1, MBH, M1), 2)} pc, twice as much inside ${f(O.cuspRadius(2, MBH, M1), 2)} pc`); + console.log(` ${lab.padEnd(6)} (${M1 / 1e6}e6 inside 1 pc): dln v/dln r passes -1/4 at ${f(O.cuspSlopeRadius(-0.25, MBH, M1), 2)} pc and 0 at` + + ` ${f(O.cuspSlopeRadius(0, MBH, M1), 2)} pc; stars weigh as much as Sgr A* inside ${f(O.cuspRadius(1, MBH, M1), 2)} pc, twice as much inside` + + ` ${f(O.cuspRadius(2, MBH, M1), 2)} pc`); } console.log(` (radii past the ${K.rBreak} pc break extrapolate the power law; the density falls faster there, so they are lower limits)`); +console.log(` middle: where dln v/dln r = -0.4, -0.3, -0.25, -0.2, -0.1: ${[-0.4, -0.3, -0.25, -0.2, -0.1].map(s => f(O.cuspSlopeRadius(s, MBH), 2)).join(', ')} pc`); console.log(` dln v/dln r (middle): ${[0.01, 0.1, 0.5, 1, 2, 3].map(r => `${f(O.cuspSlope(r, MBH))} at ${r} pc`).join(', ')}`); +console.log(` the stars' mass (middle) inside 3 pc: ${e3(O.cuspMass(3))} M_sun; inside 0.01 pc: ${f(O.cuspMass(0.01), 0)} M_sun`); +console.log(` with gamma = 1.4 instead (the top of Schodel et al.'s 1.1-1.4), the middle "twice" radius: ${f(O.cuspRadius(2, MBH, K.M1, 1.4), 2)} pc`); console.log(` the stars inside 1 pc (middle): ${f(O.cuspMass(1) / MBH, 2)} of Sgr A*'s mass; their density at 1 pc, (3 - gamma) M1/(4 pi): ` + `${e3((3 - K.gamma) * K.M1 / (4 * Math.PI))} M_sun/pc^3, ${e3((3 - K.gamma) * K.M1 / (4 * Math.PI) / 0.10)} times the 0.10 near the Sun; Sgr A* is ${e3(MBH)} Suns`); const g = O.galaxy(); @@ -86,6 +101,16 @@ console.log(` Eilers et al. 2019: v = ${EILERS.v0} ${EILERS.dv} (R - ${EILERS. } const Mgrow = ve(EILERS.Rmax) ** 2 * EILERS.Rmax / (ve(EILERS.Rmin) ** 2 * EILERS.Rmin); console.log(` mass inside 25 kpc over mass inside 5 kpc, v^2 R / G (spherical): ${f(Mgrow, 2)}; for Kepler it would be 1`); +{ + const vRG = R => g.v(R) ** 2 * R / C.G; + console.log(` MWPotential2014: mass inside a sphere of 8 kpc ${e3(g.sphereMass(8))} M_sun, against v^2 R/G = ${e3(vRG(8))} (${f(vRG(8) / g.sphereMass(8) - 1)} over)`); + console.log(` MWPotential2014: mass inside 25 kpc over 5 kpc, true ${f(g.sphereMass(25) / g.sphereMass(5), 2)}, v^2 R/G ${f(vRG(25) / vRG(5), 2)}`); + const Rsun = 8.122, rhoMean = g.sphereMass(Rsun) / (4 / 3 * Math.PI * (Rsun * 1e3) ** 3); + console.log(` the Sun's mean density inside r_J over MWPotential2014's true mean density inside ${Rsun} kpc: ${f(rhoJ / rhoMean, 2)}`); + const gamE = 2 * (1 - EILERS.dv * EILERS.R0 / ve(EILERS.R0)), gamM = 2 * (1 - g.slope(Rsun)); + console.log(` Gam = 4A/Omega = 2(1 - dln v/dln R) from Eilers et al.'s slope at R0: ${f(gamE, 2)}; from MWPotential2014's: ${f(gamM, 2)}; from Bovy's A, B: 2.25`); + console.log(` r_J from Eilers et al.'s curve at R0 (Omega = v/R0, A = Gam Omega/4): ${f(O.jacobi(C.GMsun, gamE * ve(Rsun) / Rsun / 4, gamE * ve(Rsun) / Rsun / 4 - ve(Rsun) / Rsun) / C.pc)} pc`); +} for (const Mb of [5e10, 6e10, 1e11]) console.log(` ${e3(Mb)} M_sun all inside 25 kpc, as a point mass: v(25 kpc) = ${f(Math.sqrt(C.G * Mb / 25), 0)} km/s (measured ${f(ve(25), 0)})`); console.log(' (the stars: 5 +- 1 x 10^10 M_sun, Bland-Hawthorn & Gerhard 2016; 10^11 is twice that)'); const gR0 = (ve(EILERS.R0) * 1e3) ** 2 / (EILERS.R0 * C.kpc); @@ -98,47 +123,57 @@ const tilts = O.PLANETS.map(p => O.tiltToInvariable(p) * Math.PI / 180); console.log(` mean inclination over the eight: ${f(O.PLANETS.reduce((a, p) => a + O.tiltToInvariable(p), 0) / 8, 2)} deg`); console.log(` rms sin(i) over the eight: ${f(Math.sqrt(tilts.reduce((a, t) => a + Math.sin(t) ** 2, 0) / 8), 3)}; without Mercury: ${f(Math.sqrt(tilts.slice(1).reduce((a, t) => a + Math.sin(t) ** 2, 0) / 7), 3)}`); console.log(` the Galaxy's thin disk: scale height 300 +- 50 pc over scale length 2.6 +- 0.5 kpc = ${f(0.3 / 2.6, 3)}, the axis ratio of its density` + - ` contours (Bland-Hawthorn & Gerhard 2016)`); + ` contours; over the Sun's radius 8.2 kpc = ${f(0.3 / 8.2, 3)}, the like-for-like ratio (Bland-Hawthorn & Gerhard 2016)`); if (!quick) { console.log('\n5. Which orbits a body keeps: circular orbits followed for 20 turns of the host (Hill\'s problem)\n'); const RHO0 = 0.10; // M_sun/pc^3, local mid-plane density (McKee, Parravano & Hollenbach 2015: 0.097 +- 0.013) - const sun = O.hostDisk(OORT.A, OORT.B, RHO0); + const sun = O.hostDisk(OORT.A, OORT.B, RHO0), turnGyr = n => n * 2 * Math.PI / ((OORT.A - OORT.B) * C.kmskpc) / C.yr / 1e9; console.log(` point-mass host: Gam = 4A/Omega = ${O.HOST_KEPLER.Gam}, (nu/Omega)^2 = ${O.HOST_KEPLER.nu2}`); console.log(` the Sun in the Galaxy (A, B as above, rho0 = ${RHO0} M_sun/pc^3): Gam = ${f(sun.Gam, 3)}, (nu/Omega)^2 = ${f(sun.nu2, 2)}`); - console.log(` 20 turns of the host: ${f(20 * 2 * Math.PI / ((OORT.A - OORT.B) * C.kmskpc) / C.yr / 1e9, 2)} Gyr for the Sun; ${f(20 * O.kepler(O.PLANETS[4]).P, 0)} yr for Jupiter`); + console.log(` 20 turns of the host: ${f(turnGyr(20), 2)} Gyr for the Sun; ${f(20 * O.kepler(O.PLANETS[4]).P, 0)} yr for Jupiter`); console.log(` an orbit counts as plunged inside 1e-3 r_J: ${f(1e-3 * rJ / C.au, 0)} au for the Sun, ${f(1e-3 * O.hill(O.PLANETS[4].a, 1 / O.PLANETS[4].sunOver) * C.au / 7.1492e7, 2)} Jupiter radii (71,492 km)`); console.log(` the Sun's vertical oscillation through the disk with this nu: ${f(2 * Math.PI / (Math.sqrt(sun.nu2) * (OORT.A - OORT.B) * C.kmskpc) / C.yr / 1e6, 0)} Myr`); const planes = { 'prograde, in the plane': [[0, 0, 1]], 'retrograde, in the plane': [[0, 0, -1]], 'polar, plane holds the host': [[0, 1, 0]], 'polar, plane across it': [[1, 0, 0]] }; - const as = []; for (let a = 0.3; a <= 1.2001; a += 0.025) as.push(a); - console.log(' radius [r_J] where half the orbits (8 starting angles) have escaped beyond 3 r_J:'); + const grid = (a0, a1, da) => { const out = []; for (let a = a0; a <= a1 + 1e-9; a += da) out.push(+a.toFixed(4)); return out; }; + const big = grid(0.3, 1.2, 0.025), all = grid(0.025, 1.2, 0.025); const sunNu1 = { ...sun, nu2: 1, name: 'the Sun in the Galaxy with (nu/Omega)^2 = 1' }; - console.log(' orbit point-mass host the Sun in the Galaxy the same with (nu/Omega)^2 = 1'); + console.log(' radius [r_J] where half the orbits (8 starting angles in [0, pi)) have escaped beyond 3 r_J / are lost (escaped, or plunged within 1e-3 r_J):'); + console.log(' orbit point-mass host the Sun in the Galaxy the same with (nu/Omega)^2 = 1'); for (const [k, n] of Object.entries(planes)) { - const e = [O.HOST_KEPLER, sun, sunNu1].map(h => O.hillScan(h, n, as, { nPhase: 8 }).edge); - console.log(` ${k.padEnd(28)} ${f(e[0], 3).padStart(10)} ${f(e[1], 3).padStart(10)} ${f(e[2], 3).padStart(10)}`); + const polar = k.startsWith('polar'); + const e = [O.HOST_KEPLER, sun, sunNu1].map(h => O.hillScan(h, n, polar ? all : big, { nPhase: 8 })); + console.log(` ${k.padEnd(28)} ${e.map(r => `${f(r.edge, 3)} / ${f(r.edgeLost, 3)}`.padStart(13)).join(' ')}`); + } + console.log(' (an orbit that starts in the plane stays in it, so the vertical tide cannot act on the first two rows, and none plunge)'); + console.log(' (polar orbits are lost to Lidov-Kozai cycles, whose eccentricity reaches 1 at 90 deg; below the "lost" radius the cycle takes longer than 20 turns)'); + { + const gamE = 2 * (1 - EILERS.dv * EILERS.R0 / ve(EILERS.R0)), sunE = { ...sun, Gam: gamE }; + console.log(` retrograde in the plane with Gam = ${f(gamE, 2)} (Eilers et al.'s slope): escaped by ${f(O.hillScan(sunE, [[0, 0, -1]], big, { nPhase: 8 }).edge, 3)} r_J`); } - console.log(' (an orbit that starts in the plane stays in it, so the vertical tide cannot act on the first two rows)'); - const dirs = O.sphere(24), asIso = []; for (let a = 0.2; a <= 1.2001; a += 0.1) asIso.push(a); - console.log('\n orbits facing every way (24 directions x 4 starting angles): fractions that stay, escape, and plunge within 1e-3 r_J'); + const dirs = O.sphere(48), asIso = grid(0.2, 1.0, 0.05); + console.log('\n orbits facing every way (48 directions x 4 starting angles, grid 0.05 r_J): fractions that stay, escape, and plunge within 1e-3 r_J'); const iso = {}; for (const [name, h] of [['point-mass host', O.HOST_KEPLER], ['the Sun in the Galaxy', sun]]) { iso[name] = O.hillScan(h, dirs, asIso, { nPhase: 4 }); - console.log(` ${name}: half have escaped by ${f(iso[name].edge, 3)} r_J`); - for (const r of iso[name].rows) console.log(` a = ${f(r.a, 1)} r_J: stay ${f(r.stays, 2)}, escape ${f(r.escapes, 2)}, plunge ${f(r.plunges, 2)}`); + console.log(` ${name}: half have escaped by ${f(iso[name].edge, 3)} r_J; half are lost (escaped or plunged) by ${f(iso[name].edgeLost, 3)} r_J`); + for (const r of iso[name].rows) console.log(` a = ${f(r.a, 2)} r_J: stay ${f(r.stays, 2)}, escape ${f(r.escapes, 2)}, plunge ${f(r.plunges, 2)}`); } - console.log(` for the Sun: ${f(iso['the Sun in the Galaxy'].edge * rJ / C.au / 1e5, 2)}e5 au`); + console.log(` for the Sun: half escaped by ${f(iso['the Sun in the Galaxy'].edge * rJ / C.au / 1e5, 2)}e5 au`); + const dirs24 = O.sphere(24); for (const rho of [0.08, 0.12]) { - const e = O.hillScan(O.hostDisk(OORT.A, OORT.B, rho), dirs, asIso, { nPhase: 4 }).edge; - console.log(` with rho0 = ${rho}: (nu/Omega)^2 = ${f(O.hostDisk(OORT.A, OORT.B, rho).nu2, 2)}, half escaped by ${f(e, 3)} r_J`); + const e = O.hillScan(O.hostDisk(OORT.A, OORT.B, rho), dirs24, asIso, { nPhase: 4 }).edge; + console.log(` with rho0 = ${rho} (24 directions): (nu/Omega)^2 = ${f(O.hostDisk(OORT.A, OORT.B, rho).nu2, 2)}, half escaped by ${f(e, 3)} r_J`); } - const oort = [1e5, 2e5].map(x => x * C.au / rJ); - console.log(` the Oort cloud's outer edge, near 1e5 au and as far as 2e5 au: ${f(oort[0], 2)} to ${f(oort[1], 2)} r_J`); + console.log(` the Sun at rho0 = 0.10 with 24 directions, for comparison: half escaped by ${f(O.hillScan(sun, dirs24, asIso, { nPhase: 4 }).edge, 3)} r_J`); + console.log(` with plunges counted inside 1e-2 r_J instead (24 directions): half escaped by ${f(O.hillScan(sun, dirs24, asIso, { nPhase: 4, rIn: 1e-2 }).edge, 3)} r_J`); + const oort = [6.5e4, 1e5, 2e5].map(x => x * C.au / rJ); + console.log(` the Oort cloud's outer limit: 6.5e4 au (Correa-Otto & Calandra's a_crit), 1e5 au, 2e5 au (Dones et al. 2004) = ${oort.map(x => f(x, 2)).join(', ')} r_J`); if (long) { - console.log(`\n the same, followed for 100 turns of the host (${f(100 * 2 * Math.PI / ((OORT.A - OORT.B) * C.kmskpc) / C.yr / 1e9, 1)} Gyr for the Sun):`); + console.log(`\n the same (24 directions x 4 angles), followed for 100 turns of the host (${f(turnGyr(100), 1)} Gyr for the Sun):`); for (const [name, h] of [['point-mass host', O.HOST_KEPLER], ['the Sun in the Galaxy', sun]]) { - const r = O.hillScan(h, dirs, asIso, { nPhase: 4, orbits: 100 }); - console.log(` ${name}: half have escaped by ${f(r.edge, 3)} r_J; plunged by then, at a = 0.2 to 0.4 r_J: ${r.rows.slice(0, 3).map(x => f(x.plunges, 2)).join(', ')}`); + const r20 = O.hillScan(h, dirs24, asIso, { nPhase: 4 }), r = O.hillScan(h, dirs24, asIso, { nPhase: 4, orbits: 100 }); + console.log(` ${name}: half escaped by ${f(r.edge, 3)} r_J (${f(r20.edge, 3)} over 20 turns); half lost by ${f(r.edgeLost, 3)} (${f(r20.edgeLost, 3)})`); } } } diff --git a/tests/orbits.mjs b/tests/orbits.mjs index 9ed4a84..8cf5503 100644 --- a/tests/orbits.mjs +++ b/tests/orbits.mjs @@ -2,7 +2,8 @@ // Each check prints the source, the published or expected value, the model's value and the tolerance. "Arithmetic" and // "model-internal" mark checks of the code against a closed form or itself, not against nature; "model" marks a model // result with no published value to compare. Sources were read in search results and abstracts only -// (orbits/README.md, Sources), except where a check says otherwise. +// (orbits/README.md, Sources), except where a check says otherwise; "#37's review" marks values a reviewer read in the +// paper's arXiv source. import { createRequire } from 'node:module'; const O = createRequire(import.meta.url)('../orbits/model.js'); const { C } = O; @@ -20,12 +21,17 @@ function check(name, ok, detail, src) { } // 1. Kepler's third law across the planets. -const JPL = 'JPL approximate elements, Table 1 (1800-2050; a copy of the file); GM_sun IAU 2015 B3; masses DE430'; +const JPL = 'JPL approximate elements, Table 1 (1800-2050; checked against the SSD page in #37\'s review); GM_sun IAU 2015 B3; masses DE430'; const mu = O.planetMassFraction(), ex = O.PLANETS.map(p => O.kepler(p).excess), big = Math.max(...ex.map(Math.abs)); check('In this 1800-2050 fit, n^2 a^3 matches GM_sun for every planet to within the planets\' total mass over the Sun\'s', ex.every(x => Math.abs(x) <= mu), `largest |n^2 a^3 / GM_sun - 1| = ${fmt(big)} (${O.PLANETS[ex.map(Math.abs).indexOf(big)].name}); sum of m/M_sun = ${fmt(mu)}` + - ' (an observed size, not a physical bound: smeared into rings, the planets\' pull alone gives up to 1.5e-3)', JPL); + ' (an observed size, not a physical bound: smeared into rings, the other planets\' pull alone gives up to 1.36e-3, for Uranus)', JPL); +{ + const inner = Math.max(...ex.slice(0, 4).map(Math.abs)); + near('The inner four planets: largest |n^2 a^3 / GM_sun - 1|', inner, 0, 5e-5, + `${JPL}; a regression bound, which an error above 5e-5 in GM_sun or in the length of the century would break`); +} { const xs = O.PLANETS.map(p => Math.log(p.a)), mx = xs.reduce((s, x) => s + x, 0) / xs.length; const tolP = 0.5 * mu * xs.reduce((s, x) => s + Math.abs(x - mx), 0) / xs.reduce((s, x) => s + (x - mx) ** 2, 0); @@ -46,26 +52,41 @@ check('In this 1800-2050 fit, n^2 a^3 matches GM_sun for every planet to within near('Jacobi radius with point-mass Oort constants / Hill radius a (m/3M)^(1/3), for Jupiter', O.jacobi(C.GMsun * q, o.A, o.B) / O.hill(a, q), 1, 1e-12, 'model-internal'); } near('The Sun\'s tidal radius in the Galaxy [1e5 au], from Bovy 2017\'s A = 15.3, B = -11.9 km/s/kpc', O.jacobi(C.GMsun, 15.3, -11.9) / C.au / 1e5, 2.8, 0.05, - 'Correa-Otto & Calandra 2019 (MNRAS 490, 2495): "r_J ~ 2.8 x 10^5 au", search text; tolerance the rounding of 2.8'); + 'Correa-Otto & Calandra 2019 (MNRAS 490, 2495): "r_J ~ 2.8 x 10^5 au" (#37\'s review), from Jiang & Tremaine\'s constants (2.78e5 au), so this confirms ' + + 'the formula on nearby inputs; tolerance the rounding of 2.8'); // 3. Sgr A*: S2's orbit and the stars around it. -const G20 = 'GRAVITY Collaboration 2020 (A&A 636, L5), read in a transcription'; +const G20 = 'GRAVITY Collaboration 2020 (A&A 636, L5), read in a transcription and in #37\'s review'; const s2 = O.orbitOf(O.S2), s2M = O.orbitOf(O.S2, O.S2.M); near('S2: Kepler\'s third law returns GRAVITY\'s fitted mass [1e6 M_sun]', s2.keplerMass / 1e6, O.S2.M / 1e6, 0.012, `${G20}, Table E.1; consistency with their own fit (tolerance its 0.012 error), not a measurement`); near('S2: Schwarzschild precession per orbit [arcmin]', s2M.precessionArcmin, 12.1, 0.1, `${G20}, appendix B: "the 12.1' precession"; tolerance one unit in the last digit (the formula with their Table E.1 M, a, e gives 12.16')`); -near('S2: pericentre distance [au]', s2M.periAU, 120, 5, `${G20}, introduction: "pericentre at 120 AU", rounded to 10 au`); -near('S2: speed at pericentre [km/s]', s2M.periSpeed, 7700, 50, `${G20}, introduction: "7700 km s^-1", rounded to 100 km/s`); +near('S2: pericentre distance [au]', s2M.periAU, 120, 5, `${G20}, sect. 1.2: "pericentre at 120 AU", rounded to 10 au`); +near('S2: speed at pericentre [km/s]', s2M.periSpeed, 7700, 50, `${G20}, sect. 1.2: "7700 km/s", rounded to 100 km/s`); { const K = O.CUSP, M = O.SGRA.M, r2 = [K.M1hi, K.M1, K.M1lo].map(m => O.cuspRadius(2, M, m)); check('The stars inside r weigh twice Sgr A* at 3 to 4 pc in the literature; the cusp model brackets it', r2[0] <= 3 && r2[2] >= 4, `model ${fmt(r2[0])} to ${fmt(r2[2])} pc (middle ${fmt(r2[1])}) for 0.5-1.5e6 M_sun inside 1 pc and density ~ r^-1.13; radii past the 3 pc ` + 'break extrapolate the cusp, where the density falls faster, so they are lower limits and the upper side holds a fortiori', - 'Alexander 2005 (3 pc, as cited), Feldmeier-Krause et al. 2017 (about 4 pc, with their M_BH of 3.0e6), GRAVITY Collaboration 2024 ' + - '(3.9 pc in their models; definition from Merritt 2013); inputs Schodel et al. 2009, 2018; all search text'); + 'Schodel et al. 2018 sect. 4.5 (7.8e6 M_sun inside 3 pc, "roughly twice" Sgr A*), Alexander 2005 (2m reached by <~4 pc), Feldmeier-Krause et al. ' + + '2017 sect. 3.3.2 (4.0 +2.2/-1.1 pc with M_BH = 4e6 and Merritt\'s 2004 definition), GRAVITY Collaboration 2024 sect. 4.3 (3.9 pc in their cusp ' + + 'simulation); all #37\'s review; inputs Schodel et al. 2009, 2018'); + near('The cusp model\'s stars inside 3 pc [1e6 M_sun] (middle mass inside 1 pc)', O.cuspMass(3) / 1e6, 7.8, 0.6, + 'Schodel et al. 2018 sect. 4.5: "7.8+-0.6 x 10^6 M_sun" within 3 pc (#37\'s review)'); + near('The cusp model\'s stars inside 0.01 pc [M_sun]', O.cuspMass(0.01), 180, 30, 'Schodel et al. 2018: "180+-30" within 0.01 pc (#37\'s review)'); near('Kepler inside 0.1 pc of Sgr A*: dln v/dln r', O.cuspSlope(0.1, M), -0.5, 0.01, 'model (Sgr A* plus the cusp)'); + let lo = 1e-3, hi = 30; + for (let i = 0; i < 200; i++) { const m = Math.sqrt(lo * hi); if (O.cuspSlope(m, M) < -0.25) lo = m; else hi = m; } + near('Where dln v/dln r = -1/4: the closed form against bisection [pc]', O.cuspSlopeRadius(-0.25, M), lo, 1e-9, 'model-internal'); +} + +// Special functions the model uses: the complete elliptic integral and the incomplete gamma function. +near('ellipticK(1/sqrt 2) [the lemniscate value Gamma(1/4)^2 / (4 sqrt pi)]', O.ellipticK(Math.SQRT1_2), 1.8540746773013719, 1e-14, 'arithmetic'); +for (const x of [0.25, 1, 4]) { + const erf = { 0.25: 0.5204998778130465, 1: 0.8427007929497149, 4: 0.9953222650189527 }[x]; // erf(sqrt x) + near(`gammaP(1/2, ${x}) = erf(sqrt ${x})`, O.gammaP(0.5, x), erf, 1e-13, `arithmetic (${x < 1.5 ? 'series' : 'continued fraction'})`); } // 4. The Galaxy. @@ -73,13 +94,16 @@ const g = O.galaxy(), g0 = O.galaxy(O.MWP14, { bh: 0 }); near('MWPotential2014 normalisation identity: v_c(8 kpc) without Sgr A* [km/s] (G and the shapes cancel)', g0.v(8), 220, 1e-9, 'model-internal (parameters as in galpy\'s source)'); near('MWPotential2014: halo mass inside 245 kpc [1e12 M_sun]', g0.parts(245).halo * 245 / C.G / 1e12, 0.8, 0.05, - 'Bovy 2015, Table 1: virial mass 0.8e12 M_sun with r_vir about 245 kpc, as secondary papers quote it (search text); tolerance the rounding of 0.8'); + 'Bovy 2015, Table 1: M_vir = 0.8e12 M_sun with r_vir = 245 kpc (#37\'s review); tolerance the rounding of 0.8'); +near('The disk\'s mass inside a sphere of 1e4 kpc / its total (Gauss\'s theorem by quadrature)', g.diskSphereMass(1e4) / g.M.disk, 1, 1e-3, + 'model-internal (the Miyamoto-Nagai mass outside 1e4 kpc is about 3e-4 of the total)'); { const Rs = [5, 8.122, 15, 25], ve = R => 229.0 - 1.7 * (R - 8.122); const rel = Rs.map(R => (g.v(R) / g.v(8.122)) / (ve(R) / ve(8.122)) - 1); check('The shape of MWPotential2014\'s curve, v(R)/v(R0), within 5% of Eilers et al.\'s from 5 to 25 kpc', rel.every(r => Math.abs(r) < 0.05), `ratio to Eilers - 1 = ${rel.map(r => r.toFixed(3)).join(', ')} at ${Rs.join(', ')} kpc; the level differs by ${(g.v(8.122) / 229 - 1).toFixed(3)} (220 against 229 km/s)`, - 'Eilers et al. 2019 (ApJ 871, 120), linear fit, abstract; tolerance their 2-5% systematic uncertainty'); + 'Eilers et al. 2019 (ApJ 871, 120), eq. 7 (#37\'s review); tolerance their 2-5% systematic uncertainty out to about 20 kpc (the slope\'s 0.46 ' + + 'km/s/kpc systematic gives 0.6% at 5 kpc and 3.9% at 25 kpc)'); const sl = Rs.map(R => g.slope(R)), se = Rs.map(R => -1.7 * R / ve(R)); check('Flat, not Kepler, from 5 to 25 kpc: |dln v/dln R| < 0.25 in the model and in Eilers et al.\'s fit', sl.every(s => Math.abs(s) < 0.25) && se.every(s => Math.abs(s) < 0.25), @@ -87,28 +111,43 @@ near('MWPotential2014: halo mass inside 245 kpc [1e12 M_sun]', g0.parts(245).hal } // 5. Thin disks: the planets' tilts to the invariable plane. -const SS = 'Souami & Souchay 2012 (A&A 543, A133), as quoted in search text; tolerance covers epoch and element differences'; +const SS = 'Souami & Souchay 2012 (A&A 543, A133), as quoted in search text, not confirmed in the paper; tolerance covers epoch and element differences'; near('Jupiter\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O.PLANETS[4]), 0.3219, 0.01, SS); near('Saturn\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O.PLANETS[5]), 0.9254, 0.01, SS); // 6. Hill's problem: which orbits a body keeps (20 turns of the host, 8 starting angles each unless stated). { - const K = O.HOST_KEPLER, sun = O.hostDisk(15.3, -11.9, 0.10), ST = 'Shen & Tremaine 2008 (arXiv 0806.2882), abstract, search text'; - const edgePro = O.hillScan(K, [[0, 0, 1]], [0.4, 0.45, 0.5, 0.55, 0.6], { nPhase: 8 }).edge; - near('Point-mass host: radius where half the prograde circular orbits have escaped [r_H]', edgePro, 0.5, 0.1, - `${ST}: "about 0.5 r_H"; tolerance covers their different criterion and integration time`); - const esc = (h, a, n) => O.hillFates(h, a, [n], { nPhase: 8 }).escapes, ret = [esc(K, 0.6, [0, 0, -1]), esc(K, 0.9, [0, 0, -1])]; - check('Point-mass host: retrograde orbits kept at 0.6 r_H, lost by 0.9 (published limit about 0.7)', ret[0] === 0 && ret[1] >= 0.5, - `fraction escaped ${ret[0].toFixed(2)} at 0.6 r_H (kept: must be 0), ${ret[1].toFixed(2)} at 0.9 r_H (lost: must be >= 0.5)`, `${ST}: "about 0.7 r_H"`); - const r875 = [esc(K, 0.875, [0, 0, -1]), esc(sun, 0.875, [0, 0, -1])]; - check('Retrograde in the plane at 0.875 r_J: lost around a point mass, kept by the Sun (its flatter curve, Gam = 2.25, not its vertical tide)', - r875[0] >= 0.75 && r875[1] <= 0.25, - `fraction escaped ${r875[0].toFixed(2)} (point mass; lost: must be >= 0.75), ${r875[1].toFixed(2)} (the Sun; kept: must be <= 0.25)`, 'model'); + const K = O.HOST_KEPLER, sun = O.hostDisk(15.3, -11.9, 0.10); + const NV = 'Nesvorny et al. 2003, as Shen & Tremaine 2008 (AJ 136, 2453) quote them: semi-major axes of "~0.7 r_H for retrograde satellites and ' + + '~0.4 r_H for prograde ones", for nearly circular orbits near the ecliptic (#37\'s review)'; + const esc = (h, a, n) => O.hillFates(h, a, [n], { nPhase: 8 }).escapes, pro = [esc(K, 0.4, [0, 0, 1]), esc(K, 0.6, [0, 0, 1])]; + check('Point-mass host: prograde circular orbits mostly kept at 0.4 r_H, mostly lost by 0.6', pro[0] < 0.5 && pro[1] >= 0.5, + `fraction escaped ${pro[0].toFixed(2)} at 0.4 r_H (must be < 0.5), ${pro[1].toFixed(2)} at 0.6 r_H (must be >= 0.5)`, 'model'); + const edgePro = O.hillScan(K, [[0, 0, 1]], [0.3, 0.35, 0.4, 0.45, 0.5, 0.55, 0.6, 0.65, 0.7], { nPhase: 8 }).edge; + near('Point-mass host: radius where half the prograde circular orbits in the plane have escaped [r_H]', edgePro, 0.4, 0.1, + `${NV}; tolerance covers their different criterion and integration time`); + const gRet = []; for (let a = 0.6; a <= 1.0001; a += 0.025) gRet.push(+a.toFixed(3)); + const edgeRet = O.hillScan(K, [[0, 0, -1]], gRet, { nPhase: 8 }).edge; + check('Point-mass host: the retrograde in-plane edge lies between 0.7 r_H (near-circular, near-ecliptic orbits) and r_H (some coplanar ones)', + edgeRet >= 0.7 && edgeRet <= 1, `half escaped by ${fmt(edgeRet)} r_H (must be 0.7 to 1)`, + `${NV}; Shen & Tremaine 2008 abstract: "some coplanar retrograde satellites ... can survive out to ~r_H" (#37's review)`); + const r85 = [esc(K, 0.85, [0, 0, -1]), esc(sun, 0.85, [0, 0, -1])]; + check('Retrograde in the plane at 0.85 r_J: mostly lost around a point mass, kept by the Sun (its flatter curve, Gam = 2.25, not its vertical tide)', + r85[0] >= 0.5 && r85[1] <= 0.25, + `fraction escaped ${r85[0].toFixed(2)} (point mass; must be >= 0.5), ${r85[1].toFixed(2)} (the Sun; must be <= 0.25)`, 'model'); { const s0 = O.hillStart(sun, 0.3, [0.6, 0, 0.8], 0.7), end = O.hillOrbit(sun, s0, { orbits: 20 }); const EJ = ([x, y, z, vx, vy, vz]) => (vx * vx + vy * vy + vz * vz) / 2 - sun.Gam * x * x / 2 + sun.nu2 * z * z / 2 - sun.Gam / Math.hypot(x, y, z); - near('The Sun in the Galaxy: Jacobi integral |v|^2/2 - Gam x^2/2 + nu2 z^2/2 - Gam/r over 20 turns (a = 0.3, tilted orbit)', - EJ(end.s) - EJ(s0), 0, 1e-6, `model-internal (fate ${end.fate}; a conserved quantity of the equations of motion)`); + const dE = EJ(end.s) - EJ(s0); + check('The Sun in the Galaxy: Jacobi integral |v|^2/2 - Gam x^2/2 + nu2 z^2/2 - Gam/r kept over 20 turns (a = 0.3, tilted orbit, which stays)', + end.fate === 'stays' && Math.abs(dE) <= 1e-6, `change ${fmt(dE)} (must be within 1e-6); fate ${end.fate} (must be stays)`, + 'model-internal (a conserved quantity of the equations of motion)'); + } + { + const pol = O.hillFates(sun, 0.3, [[1, 0, 0]], { nPhase: 8 }); + check('The Sun: polar circular orbits at 0.3 r_J are lost within 20 turns, by plunging (Lidov-Kozai), not by escaping', + pol.plunges >= 0.5 && pol.escapes === 0, `plunge ${pol.plunges.toFixed(2)} (must be >= 0.5), escape ${pol.escapes.toFixed(2)} (must be 0)`, + 'model (Lidov 1962, Kozai 1962: the eccentricity reaches 1 at 90 deg)'); } const Om = 27.2 * C.kmskpc, want = (4 * Math.PI * C.GMsun * 0.10 / C.pc ** 3 + 2 * (15.3 ** 2 - 11.9 ** 2) * C.kmskpc ** 2) / Om ** 2; near('The Sun in the Galaxy: (vertical tidal frequency / Omega)^2', sun.nu2, want, 1e-12 * want, 'arithmetic in SI (Poisson\'s equation), independent of C.G'); @@ -118,9 +157,9 @@ near('Saturn\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O fz.map((r, i) => `at ${[0.3, 0.6][i]} r_J: stay ${r.stays.toFixed(2)}, escape ${r.escapes.toFixed(2)}, plunge ${r.plunges.toFixed(2)}`).join('; ') + ' (12 directions x 2 angles)', 'model'); const rJ = O.jacobi(C.GMsun, 15.3, -11.9), edge = O.hillScan(sun, dirs, [0.3, 0.4, 0.5, 0.6], { nPhase: 2 }).edge * rJ / C.au; - check('The Sun keeps orbits facing every way out to the Oort cloud\'s quoted outer edge, 1e5 to 2e5 au', edge >= 1e5 && edge <= 2e5, - `half have escaped by ${fmt(edge)} au (12 directions x 2 angles, grid 0.1 r_J)`, - 'near 1e5 au: Correa-Otto & Calandra 2019, search text; as far as 2e5 au: Dones et al. 2004, as cited in search text'); + check('The Sun\'s all-directions escape edge lies within the Oort cloud\'s quoted outer limits, 6.5e4 to 2e5 au', edge >= 6.5e4 && edge <= 2e5, + `half have escaped by ${fmt(edge)} au (12 directions x 2 angles, grid 0.1 r_J; the tide alone, no passing stars)`, + 'Correa-Otto & Calandra 2019 sect. 4.3: a_crit ~ 6.5e4 au, set by passing stars (#37\'s review); Dones et al. 2004: as far as 2e5 au, as cited in search text'); } console.log(fails ? `\n${fails} check(s) failed` : '\nall orbits checks pass'); From 624ece2e67e0b63406d813faf08a2943eb279a06 Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 27 Oct 2026 01:24:46 +0000 Subject: [PATCH 05/10] Orbits study: fixes from the verification of 2db95f6 - Saturn's fitted excess is below its own mass, so only Uranus's and Neptune's are mostly the other planets' pull. - In-plane orbits plunge only outside their escape edge; the table prints where each first plunges. - Polar "lost" is the first radius where half are lost. The table prints the lost-fraction profile: around a point mass it falls back below half from about 0.35 r_H. - The 75 deg results are printed: none plunge for a planet, nor inside 0.3 r_J for the Sun; 25-38% of the Sun's plunge at 0.35-0.4. So is the inclination range of plunging directions: within 11 deg of polar for a planet, 29 deg for the Sun. - The ratio is 2.53 (2.25 times the 12% overstatement); 2.56 is against the true mean inside 8.122 kpc. - Checks: Kuzmin and Plummer limits of the disk sphere mass; the edgeLost path; the Oort grid starting at 0.2 r_J, so its 6.5e4 au bound can fail; the inner-planet bound's asymmetric window stated. - The chaos of single-plane edges is stated (nu2 +- 1e-4 moves 0.394 to 0.412) and printed. - Wording: thin-disk height over orbital radius; the run-length bullet; the 24-direction comparison for rIn; "every computed number except the values credited to #37's review". - Sources: Standish & Williams (1992), Li, Xia & Zhou 2019, and Binney & Tremaine's section left unconfirmed in model.js too. Co-Authored-By: Claude Opus 5.5 Claude-Session: https://claude.ai/code/session_01C2bpiHyHSkC8DAL5CcyciK --- orbits/README.md | 39 ++++++++++++++++++++++----------------- orbits/model.js | 6 +++--- orbits/table.mjs | 30 +++++++++++++++++++++++++++--- tests/orbits.mjs | 14 +++++++++++--- 4 files changed, 63 insertions(+), 26 deletions(-) diff --git a/orbits/README.md b/orbits/README.md index b2bf146..e8aa849 100644 --- a/orbits/README.md +++ b/orbits/README.md @@ -13,11 +13,11 @@ what ends it, and what, if anything, repeats from one scale to the next. | Question | Answer | |---|---| | Is a galaxy a big solar system? | No. Around one dominant mass, orbital speed falls as Kepler's v ∝ r^−1/2. Across the Milky Way's disk the speed falls only from 234 to 200 km/s between 5 and 25 kpc. Read as spherical, the mass inside grows 3.7-fold over that range, where a Kepler system's would stay fixed. Both figures come from Eilers et al.'s (2019) fit. | -| How exact is Kepler's law in the Solar System? | In JPL's 1800–2050 fit, n²a³ matches GM☉ to within 1.3×10⁻³ for all eight planets. That is about the planets' total mass, 1.34×10⁻³ of the Sun's. Jupiter's excess is mostly its own mass in the two-body law; the outer planets' is mostly the others' pull. P ∝ a^1.49986 over 0.39 to 30 au. | +| How exact is Kepler's law in the Solar System? | In JPL's 1800–2050 fit, n²a³ matches GM☉ to within 1.3×10⁻³ for all eight planets. That is about the planets' total mass, 1.34×10⁻³ of the Sun's. Jupiter's excess is mostly its own mass in the two-body law; Uranus's and Neptune's are mostly the other planets' pull, and Saturn's fitted 2.2×10⁻⁴ is below its own 2.9×10⁻⁴ (see below). P ∝ a^1.49986 over 0.39 to 30 au. | | Does the Galaxy have a Kepler region? | Yes, around Sgr A*. The star S2 orbits it as a planet orbits the Sun, plus general relativity's precession: 12.2′ per orbit from their elements (they quote 12.1′). The speed's log slope passes −1/4 about 1.3 pc out (1.0 to 1.8 pc across the cluster-mass range; 0.7 to 1.9 pc for thresholds from −0.4 to −0.1). The stars inside weigh as much as the black hole by 1.8 to 3.2 pc. | | What ends a Kepler region? | For a body orbiting in a host (a planet, the Sun), the host's tide. Satellites are held inside its tidal radius, r_J³ = Gm/(4A(A − B)), apart from quasi-satellite orbits beyond about 2 r_H that Shen and Tremaine (2008) find around Jupiter, Uranus and Neptune. Inside r_J, the body's mean density is 4A/Ω times the host's mean density inside the orbit (host taken as spherical): 3 for a planet around the Sun, 2.25 for the Sun in the Galaxy. For the Sun, r_J = 1.37 pc = 2.8×10⁵ au. At the Galaxy's centre, where Sgr A* sits still, the region ends instead where the surrounding stars' mass approaches the black hole's. | -| How much of that region is used? | About half, for orbits that do not plunge, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.52 r_J around a planet and by 0.435 r_J (1.23×10⁵ au) for the Sun. Polar orbits are lost much sooner, to Lidov–Kozai cycles: half within 0.14 to 0.15 r_H around a planet and 0.06 r_J for the Sun in 20 turns of the host. For near-circular moons near the ecliptic, Nesvorný et al. (2003) put the limits at about 0.4 and 0.7 r_H (prograde, retrograde), and the model gives 0.46 and 0.775. The Oort cloud's outer limit is placed between 6.5×10⁴ au (Correa-Otto and Calandra 2019, where passing stars set it) and 2×10⁵ au, which is 0.23 to 0.71 r_J. | -| What does repeat? | Kepler regions nested inside one another: moons in a planet's, planets and comets in the Sun's. Each is bounded by its host's tide at a similar fraction of the tidal radius, though for the Oort cloud passing stars cut it shorter. Also flat disks: the planets' orbits are tilted by about 2° on average (6° for Mercury), and the Galaxy's thin disk is 0.04 as thick as the Sun's orbit is wide. | +| How much of that region is used? | About half, for orbits that do not plunge, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.52 r_J around a planet and by 0.435 r_J (1.23×10⁵ au) for the Sun. Polar orbits are lost much sooner, to Lidov–Kozai cycles: half are lost by 0.14 to 0.15 r_H around a planet (up to about 0.3 r_H; from 0.35 r_H on, half or more survive the run) and by 0.06 r_J for the Sun, in 20 turns of the host. For near-circular moons near the ecliptic, Nesvorný et al. (2003) put the limits at about 0.4 and 0.7 r_H (prograde, retrograde), and the model gives 0.46 and 0.775. The Oort cloud's outer limit is placed between 6.5×10⁴ au (Correa-Otto and Calandra 2019, where passing stars set it) and 2×10⁵ au, which is 0.23 to 0.71 r_J. | +| What does repeat? | Kepler regions nested inside one another: moons in a planet's, planets and comets in the Sun's. Each is bounded by its host's tide at a similar fraction of the tidal radius, though for the Oort cloud passing stars cut it shorter. Also flat disks: the planets' orbits are tilted by about 2° on average (6° for Mercury), and the Galaxy's thin disk has a scale height 0.04 of the Sun's orbital radius. | ## Kepler's law in the Solar System @@ -63,7 +63,7 @@ and Tremaine 2010 write it, from Binney and Tremaine 2008): Written as densities, the edge needs no length scale. Inside r_J the body's mean density, 3m/4πr_J³, is 4A/Ω times the host's mean density inside the orbit, 3Ω²/4πG (spherical host). - **The ratio.** In terms of the rotation curve it is 2(1 − d ln v/d ln R): exactly 3 for a planet and 2 for a flat curve. -- **The Sun.** Bovy's values, taking the Galaxy as spherical, give 2.25. Against MWPotential2014's true mean density inside the Sun's orbit it would be 2.56 (see Across the disk). +- **The Sun.** Bovy's values, taking the Galaxy as spherical, give 2.25. Correcting for the 12% by which the spherical reading overstates MWPotential2014's mass inside 8 kpc gives 2.53 (2.56 against its true mean density inside 8.122 kpc; see Across the disk). So a Kepler region ends where the body's mean density falls to two or three times the host's mean density inside the body's orbit. That comparison density is a mean, not the local density. For the Sun it is 0.041 M☉ pc⁻³. The @@ -104,24 +104,27 @@ The radius by which half the orbits have escaped, and by which half are lost (`n | Polar, plane across it: escaped / lost | 0.485 / 0.15 | 0.39 / 0.06 | 0.50 / 0.16 | | Facing every way: escaped / lost | 0.52 / 0.515 | 0.435 / 0.42 | | -In-plane orbits never plunge, so for them escaped and lost are the same. +In-plane orbits plunge only outside their escape edge (first at 0.55 r_H, prograde around a point mass, and at +0.875 and 1.0 r_J retrograde), so for them the escaped and lost radii agree. For polar rows, "lost" is the first +radius where half are lost. **Polar orbits are lost by plunging.** - **The mechanism.** A circular orbit at 90° to the host's plane goes through Lidov–Kozai cycles (Lidov 1962, Kozai 1962), whose eccentricity reaches 1. So it dives into the body long before it can escape. -- **When.** Within 20 turns that happens from 0.14 to 0.15 r_H around a planet and from 0.06 r_J for the Sun. Further in, the cycle takes longer than 20 turns; it is not absent. -- **Other directions.** Tilted to 75° instead, none plunge between 0.1 and 0.4, around either host (#37's review). Facing every way, up to a fifth of the Sun's orbits plunge within 20 turns, and 8–11% of a planet's between 0.2 and 0.4 r_H, all from near-polar directions. For comets this is the tide driving them inward (Heisler and Tremaine 1986). +- **When.** Within 20 turns half are lost from 0.14 to 0.15 r_H around a planet and from 0.06 r_J for the Sun. Further in, the cycle takes longer than 20 turns; it is not absent. +- **Not everywhere outside.** Around a point mass the lost fraction peaks at 0.2 r_H (all 8 orbits) and falls back to 0.25–0.5 at 0.35–0.4 r_H, where most polar orbits survive the run; so does the Sun's with ν²/Ω² = 1. With the Galaxy's vertical tide the Sun loses 63–100% of its polar orbits at every radius from 0.075 to 0.45 r_J. +- **Other directions.** Tilted to 75° instead, none plunge from 0.1 to 0.4 r_H around a planet, nor inside 0.3 r_J for the Sun; at 0.35–0.4 r_J, 25–38% of the Sun's do. Facing every way, up to a fifth of the Sun's orbits plunge within 20 turns, from directions within 29° of polar, and 8–11% of a planet's between 0.2 and 0.4 r_H, from within 11° of polar. For comets this is the tide driving them inward (Heisler and Tremaine 1986). **For a planet.** - **Against the published limits.** The model's in-plane edges are 0.46 (prograde) and 0.775 (retrograde), against Nesvorný et al.'s (2003) semi-major axes of about 0.4 and 0.7 r_H for nearly circular orbits near the ecliptic, as Shen and Tremaine (2008) quote them. -- **Over all inclinations.** Shen and Tremaine find that many prograde and retrograde satellites survive to about 0.5 and 0.7 r_H, and some coplanar retrograde ones to about r_H. The observed irregular moons lie at about 0.05 to 0.6 r_H, the retrograde ones further out (about 0.6 against 0.4 r_H). -- **Run length.** These positions are not semi-major axes, and the runs are short: 20 turns is 237 years for Jupiter. +- **Over all inclinations.** Shen and Tremaine find that many prograde and retrograde satellites survive to about 0.5 and 0.7 r_H, and some coplanar retrograde ones to about r_H. The observed irregular moons lie at about 0.05 to 0.6 r_H, the retrograde ones further out (about 0.6 against 0.4 r_H). Their extents are instantaneous distances, not semi-major axes. +- **Run length.** The runs are short: 20 turns is 237 years for Jupiter. - Followed for 100 turns, the all-directions value moves from 0.53 to 0.51 (24 directions, `node orbits/table.mjs long`). - In a 48-orbit sample run to 3000 turns, #37's review found the point-mass edge drifting from 0.50 to 0.475. **For the Sun.** - **The edge.** Half of the orbits facing every way have escaped by 0.435 r_J, or 1.23×10⁵ au. - For ρ₀ from 0.08 to 0.12 it moves from 0.454 to 0.429 (on 24 directions, which give 0.448 at 0.10). - - Counting plunges inside 10⁻² r_J instead of 10⁻³ moves it to 0.469. + - Counting plunges inside 10⁻² r_J instead of 10⁻³ moves the 24-direction value from 0.448 to 0.469. - Followed for 100 turns (23 billion years, a convergence test only) it is 0.42, and 0.38 counting plunges, against 0.45 and 0.43 over 20 turns (24 directions). - **The Oort cloud.** Correa-Otto and Calandra (2019, sect. 4.3) find fewer than 5% of comets still bound beyond 6.5×10⁴ au and none at 10⁵ au. They define a_crit ≈ 6.5×10⁴ au as the cloud's outer limit; half are bound at about 2×10⁴ au. They name passing stars, not the tide, as what sets that limit. Dones et al. (2004, as cited) place the edge as far as 2×10⁵ au. So the quoted outer limits run from 0.23 to 0.71 r_J. The model's 0.435 r_J lies past Correa-Otto and Calandra's: the tide alone would keep comets further out than passing stars allow. - **The vertical tide.** It loses polar orbits sooner, by escape at 0.43 and 0.39 r_J against 0.46 and 0.50 with ν²/Ω² = 1 at the same Γ, and by plunging at 0.06 against 0.16. It never acts on an orbit that starts in the plane, because such an orbit stays there. @@ -222,7 +225,7 @@ mass, or mass spread through the disk and halo. ## Run it ``` -node orbits/table.mjs # every computed number above except the 100-turn runs (about 3 minutes; section 5 takes nearly all of it) +node orbits/table.mjs # every computed number above except the 100-turn runs and the values credited to #37's review (about 3 minutes) node orbits/table.mjs quick # sections 1-4 only, well under a second (run by CI) node orbits/table.mjs long # adds the 100-turn runs, about 5 minutes in all node tests/orbits.mjs # the checks, about 3 s (also run by npm test) @@ -232,7 +235,7 @@ The table exits with an error if any printed value is not a finite number. ## Checks -`tests/orbits.mjs` has 35 checks. Each prints its source, the published or expected value, the model's value and +`tests/orbits.mjs` has 38 checks. Each prints its source, the published or expected value, the model's value and the tolerance. "#37's review" marks values a reviewer read in the paper's arXiv source. | What | Against | Kind | @@ -245,12 +248,12 @@ the tolerance. "#37's review" marks values a reviewer read in the paper's arXiv | Stars weigh 2 M_BH at 3 to 4 pc; the cusp's mass inside 3 pc and 0.01 pc | Schödel et al. 2018; Alexander 2005; Feldmeier-Krause et al. 2017; GRAVITY 2024 | #37's review | | Kepler slope −1/2 inside 0.1 pc; the −1/4 radius in closed form against bisection | — | model; model-internal | | ellipticK(1/√2); gammaP(½, x) = erf √x at three points | closed forms | arithmetic | -| MWPotential2014: the 220 km/s normalisation; halo mass 0.8×10¹² inside 245 kpc; the disk's sphere mass tending to its total; shape within 5% of Eilers et al.; flat (abs. slope < 0.25) at 5 to 25 kpc | Bovy 2015 Table 1; Eilers et al. 2019 | model-internal; #37's review | +| MWPotential2014: the 220 km/s normalisation; halo mass 0.8×10¹² inside 245 kpc; the disk's sphere mass tending to its total, and its Kuzmin and Plummer limits; shape within 5% of Eilers et al.; flat (abs. slope < 0.25) at 5 to 25 kpc | Bovy 2015 Table 1; closed forms; Eilers et al. 2019 | model-internal; #37's review; arithmetic | | Jupiter's and Saturn's tilts to the invariable plane | Souami and Souchay 2012 | search text, not confirmed in the paper | | Planet: prograde mostly kept at 0.4 r_H and lost by 0.6; prograde edge 0.4 ± 0.1; retrograde edge between 0.7 and 1 r_H | Nesvorný et al. 2003 and Shen and Tremaine 2008, as Shen and Tremaine give them | model; #37's review | | Retrograde in the plane at 0.85 r_J: mostly lost around a point mass, kept by the Sun | — | model | | The Jacobi integral kept over 20 turns on an orbit that stays; the Galaxy's ν²/Ω² in SI | equations of motion; Poisson's equation | model-internal; arithmetic | -| The Sun's polar orbits at 0.3 r_J lost by plunging, not escape | Lidov–Kozai | model | +| The Sun's polar orbits at 0.3 r_J lost by plunging, not escape; a point mass's polar orbits lost between 0.1 and 0.2 r_H before any escapes | Lidov–Kozai | model | | The Sun: none escape from 0.3 r_J, most by 0.6 | — | model | | The Sun's all-directions edge within the Oort cloud's quoted 6.5×10⁴ to 2×10⁵ au | Correa-Otto and Calandra 2019; Dones et al. 2004 | #37's review; as cited | @@ -264,9 +267,10 @@ the tolerance. "#37's review" marks values a reviewer read in the paper's arXiv - **Circular starting orbits only.** Real moons and comets start on eccentric orbits. - **The edges' definition.** They are set at half the orbits escaping, or lost, within 20 turns of the host, on a grid of 0.025 r_J (single planes) or 0.05 r_J (all directions). - **Plunge radius.** It is 10⁻³ r_J. Raising it to 10⁻² moves the Sun's all-directions edge from 0.448 to 0.469 (24 directions). + - **Chaos.** Single-plane edges are chaotic at about the 0.02 level: changing ν²/Ω² by 10⁻⁴ moves the Sun's polar-across escape edge from 0.394 to 0.412. - **Longer runs** move the edges inward a little (see the 100-turn values above). -- **MWPotential2014 is one model among many.** Its halo sets the flat range, and it has no nuclear star cluster. The checks pin its halo; the disk's and bulge's shape parameters are taken from galpy's source and not otherwise tested. The Galactic-centre numbers use the separate power-law cusp above, which is valid only inside its 3 pc break. -- **Spherical mass estimates.** v²R/G and 3Ω²/4πG treat a flattened galaxy as spherical. In MWPotential2014 this overstates the mass inside 8 kpc by 12%. It shifts the 3.7-fold growth to 4.0 and the Sun's 2.25 to 2.56, but not the contrast with Kepler's fixed mass. +- **MWPotential2014 is one model among many.** Its halo sets the flat range, and it has no nuclear star cluster. The checks pin its halo; the disk's and bulge's shape parameters are taken from galpy's source and not otherwise tested, since Table 1's derived masses were not read for this study. The Galactic-centre numbers use the separate power-law cusp above, which is valid only inside its 3 pc break. +- **Spherical mass estimates.** v²R/G and 3Ω²/4πG treat a flattened galaxy as spherical. In MWPotential2014 this overstates the mass inside 8 kpc by 12%. It shifts the 3.7-fold growth to 4.0 and the Sun's 2.25 to 2.53, but not the contrast with Kepler's fixed mass. ## Sources @@ -274,9 +278,10 @@ Each entry says how it was read. "#37's review" means a reviewer checked the val or in the primary file, during #37's review. - W. M. Folkner et al., *The Planetary and Lunar Ephemerides DE430 and DE431*, IPN Progress Report 42-196 (2014), Table 8. #37's review. -- E. M. Standish and J. G. Williams, *Keplerian Elements for Approximate Positions of the Major Planets*, JPL Solar System Dynamics, Table 1 (1800–2050). Checked against the SSD page in #37's review. +- E. M. Standish and J. G. Williams, *Keplerian Elements for Approximate Positions of the Major Planets*, JPL Solar System Dynamics (1992), Table 1 (1800–2050). Checked against the SSD page in #37's review. - A. Prša et al., *Nominal values for selected solar and planetary quantities: IAU 2015 Resolution B3*, Astron. J. 152, 41 (2016). https://doi.org/10.3847/0004-6256/152/2/41. #37's review. - D. Souami and J. Souchay, *The solar system's invariable plane*, A&A 543, A133 (2012). https://doi.org/10.1051/0004-6361/201219011. Search text; not reached in #37's review. The plane's orientation is confirmed through Li, Xia and Zhou 2019; the per-planet tilts are not. +- Li, Xia and Zhou, *Calibration of the angular momenta of the minor planets in the solar system* (2019). https://arxiv.org/abs/1909.11293. For the invariable plane's orientation, as #37's review read it; journal reference not confirmed. - Y.-F. Jiang and S. Tremaine, *The evolution of wide binary stars*, MNRAS 401, 977 (2010; journal reference recalled). https://arxiv.org/abs/0907.2952. The tidal-radius formula and constants, as #37's review quotes them. - J. Binney and S. Tremaine, *Galactic Dynamics*, 2nd ed., Princeton (2008). The formula, as Jiang and Tremaine use it; section not confirmed. - J. Bovy, *Galactic rotation in Gaia DR1*, MNRAS 468, L63 (2017). https://doi.org/10.1093/mnrasl/slx027. #37's review. diff --git a/orbits/model.js b/orbits/model.js index 8a3de42..42d3313 100644 --- a/orbits/model.js +++ b/orbits/model.js @@ -8,7 +8,7 @@ * * Where it stops. A body of mass m on a circular orbit of angular speed Omega in a host keeps its own satellites * inside the tidal (Jacobi) radius, where its pull matches the host's tidal stretch less the centrifugal term - * (Binney & Tremaine 2008, sec. 8.3): + * (Binney & Tremaine 2008, as Jiang & Tremaine 2010 give it; section not confirmed): * r_J^3 = G m / (4 Omega^2 - kappa^2) = G m / (4 A (A - B)), Omega = A - B, kappa^2 = -4 B (A - B), * with A and B the host's Oort constants at the orbit. For a point-mass host A = 3 Omega / 4 and B = -Omega / 4, and * r_J is Hill's radius a (m / 3M)^(1/3). For a flat rotation curve A = -B = Omega / 2. Either way the mean density of @@ -39,7 +39,7 @@ C.kmskpc = 1e3 / C.kpc; // 1 km/s/kpc in s^-1 // ---------- the Solar System ---------- - // JPL "Keplerian Elements for Approximate Positions of the Major Planets" (Standish), Table 1 (1800 AD - 2050 AD), + // JPL "Keplerian Elements for Approximate Positions of the Major Planets" (Standish & Williams), Table 1 (1800 AD - 2050 AD), // heliocentric, mean ecliptic and equinox of J2000: a [au], inclination i and node Om [deg], mean-longitude rate // Ldot [deg/century]. Sun/planet mass ratios from the DE430 GM values (Folkner et al. 2014); "EM Bary" is the // Earth-Moon barycentre. @@ -259,7 +259,7 @@ // ---------- Hill's problem: which orbits a body keeps inside its tidal radius ---------- // A satellite of a body on a circular orbit in a host, in the frame rotating with the body (x away from the host's // centre, y along the motion, z out of the plane), lengths in r_J and times in 1/Omega (Binney & Tremaine 2008, - // sec. 8.3; Heisler & Tremaine 1986 for the vertical tide): + // as Jiang & Tremaine 2010 give it; Heisler & Tremaine 1986 for the vertical tide): // x'' = 2y' + Gam x - Gam x/r^3, y'' = -2x' - Gam y/r^3, z'' = -nu2 z - Gam z/r^3, // with Gam = 4A/Omega (3 for a point-mass host, so these are Hill's equations) and nu2 = (vertical tidal frequency / // Omega)^2: 1 for a point-mass host; for a disk, (4 pi G rho0 + 2(A^2 - B^2)) / Omega^2 by Poisson's equation, with diff --git a/orbits/table.mjs b/orbits/table.mjs index 0d0c23f..61fdc8b 100644 --- a/orbits/table.mjs +++ b/orbits/table.mjs @@ -1,4 +1,4 @@ -// Prints every computed number in orbits/README.md. Run: node orbits/table.mjs (a few minutes; the stability scans of +// Prints every computed number in orbits/README.md, except the values the README credits to #37's review. Run: node orbits/table.mjs (a few minutes; the stability scans of // section 5 take nearly all of it), node orbits/table.mjs quick (well under a second; section 5 left out), or // node orbits/table.mjs long (adds the all-directions scans followed for 100 turns of the host). // Exits with code 1 if any printed line holds NaN or Infinity. @@ -106,7 +106,8 @@ console.log(` mass inside 25 kpc over mass inside 5 kpc, v^2 R / G (spherical) console.log(` MWPotential2014: mass inside a sphere of 8 kpc ${e3(g.sphereMass(8))} M_sun, against v^2 R/G = ${e3(vRG(8))} (${f(vRG(8) / g.sphereMass(8) - 1)} over)`); console.log(` MWPotential2014: mass inside 25 kpc over 5 kpc, true ${f(g.sphereMass(25) / g.sphereMass(5), 2)}, v^2 R/G ${f(vRG(25) / vRG(5), 2)}`); const Rsun = 8.122, rhoMean = g.sphereMass(Rsun) / (4 / 3 * Math.PI * (Rsun * 1e3) ** 3); - console.log(` the Sun's mean density inside r_J over MWPotential2014's true mean density inside ${Rsun} kpc: ${f(rhoJ / rhoMean, 2)}`); + console.log(` the Sun's mean density inside r_J over MWPotential2014's true mean density inside ${Rsun} kpc: ${f(rhoJ / rhoMean, 2)};` + + ` 2.25 times the 8 kpc overstatement alone: ${f(O.densityRatio(OORT.A, OORT.B) * vRG(8) / g.sphereMass(8), 2)}`); const gamE = 2 * (1 - EILERS.dv * EILERS.R0 / ve(EILERS.R0)), gamM = 2 * (1 - g.slope(Rsun)); console.log(` Gam = 4A/Omega = 2(1 - dln v/dln R) from Eilers et al.'s slope at R0: ${f(gamE, 2)}; from MWPotential2014's: ${f(gamM, 2)}; from Bovy's A, B: 2.25`); console.log(` r_J from Eilers et al.'s curve at R0 (Omega = v/R0, A = Gam Omega/4): ${f(O.jacobi(C.GMsun, gamE * ve(Rsun) / Rsun / 4, gamE * ve(Rsun) / Rsun / 4 - ve(Rsun) / Rsun) / C.pc)} pc`); @@ -140,12 +141,31 @@ if (!quick) { const sunNu1 = { ...sun, nu2: 1, name: 'the Sun in the Galaxy with (nu/Omega)^2 = 1' }; console.log(' radius [r_J] where half the orbits (8 starting angles in [0, pi)) have escaped beyond 3 r_J / are lost (escaped, or plunged within 1e-3 r_J):'); console.log(' orbit point-mass host the Sun in the Galaxy the same with (nu/Omega)^2 = 1'); + const profiles = []; for (const [k, n] of Object.entries(planes)) { const polar = k.startsWith('polar'); const e = [O.HOST_KEPLER, sun, sunNu1].map(h => O.hillScan(h, n, polar ? all : big, { nPhase: 8 })); console.log(` ${k.padEnd(28)} ${e.map(r => `${f(r.edge, 3)} / ${f(r.edgeLost, 3)}`.padStart(13)).join(' ')}`); + const firstPlunge = e.map(r => { const row = r.rows.find(x => x.plunges > 0); return row ? row.a : NaN; }); + if (!polar) profiles.push(`${k}: first plunge at ${firstPlunge.map(x => (Number.isNaN(x) ? 'none' : f(x, 3))).join(', ')} r_J`); + else profiles.push(`${k}: lost fraction at a = 0.1, 0.15, 0.2, 0.3, 0.35, 0.4, 0.45: ` + + e.map(r => [0.1, 0.15, 0.2, 0.3, 0.35, 0.4, 0.45].map(a => { const x = r.rows.find(y => Math.abs(y.a - a) < 1e-9); return f(x.escapes + x.plunges, 2); }).join(' ')).join(' | ')); + } + console.log(' (an orbit that starts in the plane stays in it, so the vertical tide cannot act on the first two rows; their plunges, if any, begin only'); + console.log(' outside the escape edge, so the two radii agree. For polar rows "lost" is the first radius where half are lost.)'); + for (const p of profiles) console.log(` ${p} [point mass | Sun | Sun, nu2 = 1]`); + { + const across = [[1, 0, 0]], bigOnly = big; + const at = nu2 => O.hillScan({ ...sun, nu2 }, across, bigOnly, { nPhase: 8 }).edge; + console.log(` the Sun's polar-across escape edge at (nu/Omega)^2 = ${f(sun.nu2, 4)}, -1e-4, +1e-4: ${f(at(sun.nu2), 3)}, ${f(at(sun.nu2 - 1e-4), 3)}, ${f(at(sun.nu2 + 1e-4), 3)}`); + } + { + const s75 = Math.sin(75 * Math.PI / 180), c75 = Math.cos(75 * Math.PI / 180), as75 = grid(0.1, 0.4, 0.05); + for (const [name, h] of [['point mass', O.HOST_KEPLER], ['the Sun', sun]]) { + const row = [[s75, 0, c75], [0, s75, c75]].map(n => as75.map(a => f(O.hillFates(h, a, [n], { nPhase: 8 }).plunges, 2)).join(' ')); + console.log(` tilted 75 deg, ${name}: plunged fraction at a = ${as75.join(', ')}: node along y: ${row[0]}; node along x: ${row[1]}`); + } } - console.log(' (an orbit that starts in the plane stays in it, so the vertical tide cannot act on the first two rows, and none plunge)'); console.log(' (polar orbits are lost to Lidov-Kozai cycles, whose eccentricity reaches 1 at 90 deg; below the "lost" radius the cycle takes longer than 20 turns)'); { const gamE = 2 * (1 - EILERS.dv * EILERS.R0 / ve(EILERS.R0)), sunE = { ...sun, Gam: gamE }; @@ -155,6 +175,10 @@ if (!quick) { console.log('\n orbits facing every way (48 directions x 4 starting angles, grid 0.05 r_J): fractions that stay, escape, and plunge within 1e-3 r_J'); const iso = {}; for (const [name, h] of [['point-mass host', O.HOST_KEPLER], ['the Sun in the Galaxy', sun]]) { + // the directions that plunge at 0.2 to 0.4 r_J, as degrees from polar (from the orbit's normal against the host's z axis) + let far = 0; + for (const nv of dirs) for (const a of [0.2, 0.3, 0.4]) if (O.hillFates(h, a, [nv], { nPhase: 4 }).plunges > 0) far = Math.max(far, Math.abs(90 - Math.acos(nv[2]) * 180 / Math.PI)); + console.log(` ${name}: plunges at 0.2-0.4 r_J come from directions within ${f(far, 0)} deg of polar`); iso[name] = O.hillScan(h, dirs, asIso, { nPhase: 4 }); console.log(` ${name}: half have escaped by ${f(iso[name].edge, 3)} r_J; half are lost (escaped or plunged) by ${f(iso[name].edgeLost, 3)} r_J`); for (const r of iso[name].rows) console.log(` a = ${f(r.a, 2)} r_J: stay ${f(r.stays, 2)}, escape ${f(r.escapes, 2)}, plunge ${f(r.plunges, 2)}`); diff --git a/tests/orbits.mjs b/tests/orbits.mjs index 8cf5503..24a2c77 100644 --- a/tests/orbits.mjs +++ b/tests/orbits.mjs @@ -30,7 +30,7 @@ check('In this 1800-2050 fit, n^2 a^3 matches GM_sun for every planet to within { const inner = Math.max(...ex.slice(0, 4).map(Math.abs)); near('The inner four planets: largest |n^2 a^3 / GM_sun - 1|', inner, 0, 5e-5, - `${JPL}; a regression bound, which an error above 5e-5 in GM_sun or in the length of the century would break`); + `${JPL}; a regression bound, which a GM_sun error beyond -1.3e-5 or +5.5e-5, or a century-length error beyond -0.6e-5 or +2.7e-5, would break`); } { const xs = O.PLANETS.map(p => Math.log(p.a)), mx = xs.reduce((s, x) => s + x, 0) / xs.length; @@ -97,6 +97,10 @@ near('MWPotential2014: halo mass inside 245 kpc [1e12 M_sun]', g0.parts(245).hal 'Bovy 2015, Table 1: M_vir = 0.8e12 M_sun with r_vir = 245 kpc (#37\'s review); tolerance the rounding of 0.8'); near('The disk\'s mass inside a sphere of 1e4 kpc / its total (Gauss\'s theorem by quadrature)', g.diskSphereMass(1e4) / g.M.disk, 1, 1e-3, 'model-internal (the Miyamoto-Nagai mass outside 1e4 kpc is about 3e-4 of the total)'); +near('Disk sphere mass, Kuzmin limit (b -> 0): fraction inside 8 kpc = 1 - a / sqrt(r^2 + a^2)', ((G) => G.diskSphereMass(8) / G.M.disk)(O.galaxy({ ...O.MWP14, b: 1e-9 })), + 1 - 3 / Math.hypot(8, 3), 1e-8, 'arithmetic (closed form)'); +near('Disk sphere mass, Plummer limit (a = 0): fraction inside 0.3 kpc = r^3 / (r^2 + b^2)^(3/2)', ((G) => G.diskSphereMass(0.3) / G.M.disk)(O.galaxy({ ...O.MWP14, a: 0 })), + 0.3 ** 3 / (0.09 + 0.28 ** 2) ** 1.5, 1e-8, 'arithmetic (closed form)'); { const Rs = [5, 8.122, 15, 25], ve = R => 229.0 - 1.7 * (R - 8.122); const rel = Rs.map(R => (g.v(R) / g.v(8.122)) / (ve(R) / ve(8.122)) - 1); @@ -148,6 +152,10 @@ near('Saturn\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O check('The Sun: polar circular orbits at 0.3 r_J are lost within 20 turns, by plunging (Lidov-Kozai), not by escaping', pol.plunges >= 0.5 && pol.escapes === 0, `plunge ${pol.plunges.toFixed(2)} (must be >= 0.5), escape ${pol.escapes.toFixed(2)} (must be 0)`, 'model (Lidov 1962, Kozai 1962: the eccentricity reaches 1 at 90 deg)'); + const L = O.hillScan(K, [[1, 0, 0]], [0.1, 0.2], { nPhase: 8 }); + check('Point-mass host: polar orbits (plane across the host) are lost by plunging between 0.1 and 0.2 r_H, before any escapes', + Number.isNaN(L.edge) && L.edgeLost > 0.1 && L.edgeLost < 0.2, `half lost by ${fmt(L.edgeLost)} r_H (must be 0.1 to 0.2); half escaped: ${L.edge} (must be none)`, + 'model (the edgeLost path)'); } const Om = 27.2 * C.kmskpc, want = (4 * Math.PI * C.GMsun * 0.10 / C.pc ** 3 + 2 * (15.3 ** 2 - 11.9 ** 2) * C.kmskpc ** 2) / Om ** 2; near('The Sun in the Galaxy: (vertical tidal frequency / Omega)^2', sun.nu2, want, 1e-12 * want, 'arithmetic in SI (Poisson\'s equation), independent of C.G'); @@ -156,9 +164,9 @@ near('Saturn\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O fz[0].escapes === 0 && fz[1].escapes >= 0.5, fz.map((r, i) => `at ${[0.3, 0.6][i]} r_J: stay ${r.stays.toFixed(2)}, escape ${r.escapes.toFixed(2)}, plunge ${r.plunges.toFixed(2)}`).join('; ') + ' (12 directions x 2 angles)', 'model'); - const rJ = O.jacobi(C.GMsun, 15.3, -11.9), edge = O.hillScan(sun, dirs, [0.3, 0.4, 0.5, 0.6], { nPhase: 2 }).edge * rJ / C.au; + const rJ = O.jacobi(C.GMsun, 15.3, -11.9), edge = O.hillScan(sun, dirs, [0.2, 0.3, 0.4, 0.5, 0.6], { nPhase: 2 }).edge * rJ / C.au; check('The Sun\'s all-directions escape edge lies within the Oort cloud\'s quoted outer limits, 6.5e4 to 2e5 au', edge >= 6.5e4 && edge <= 2e5, - `half have escaped by ${fmt(edge)} au (12 directions x 2 angles, grid 0.1 r_J; the tide alone, no passing stars)`, + `half have escaped by ${fmt(edge)} au (12 directions x 2 angles, grid 0.1 r_J from 0.2 r_J = 5.7e4 au; the tide alone, no passing stars)`, 'Correa-Otto & Calandra 2019 sect. 4.3: a_crit ~ 6.5e4 au, set by passing stars (#37\'s review); Dones et al. 2004: as far as 2e5 au, as cited in search text'); } From 99ec2bc6e93ef41b56e866060b2d8e3d6ef7e2d7 Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 27 Oct 2026 08:00:10 +0000 Subject: [PATCH 06/10] Orbits study: fixes from #37's re-review of 624ece2 Blocking: README:19 said polar orbits around a planet are mostly kept "from 0.35 r_H on". The table's own profile loses more than half at 0.35; the row now says the lost fraction dips below half near 0.4 r_H, and the polar bullet gives the profile at the printed radii. Sampling and numerics: - Starting angles span the quarter turn [0, pi/2): besides r -> -r, Hill's equations are symmetric under time reversal with y -> -y, which pairs the start at pi - phi with the start at phi for every single-plane scan. hillFates takes quarter = 1 for [pi/2, pi); the table prints the all-directions fractions from that quarter, which agree to 0.03. Section 5 was re-run and every number in the README updated. - 64-angle in-plane edges, the 64-angle lost edge and first plunge, the Gam = 2.12 edge (about 1.0 r_J), tilt-from-polar plunge profiles, 75 deg plunges, both edges under the plunge radius, and the nu^2 sensitivity are printed; the chaos caveat now separates polar chaos from in-plane sampling error. - hillOrbit's step cap scales with the number of turns (2.5e5 per turn) and it throws on non-finite settings. A missing first plunge prints "none" without hiding a NaN radius. Checks (43): Bovy 2015 Table 1's bulge and disk masses, M(<60 kpc), slope at R0 and local dark-matter density (haloDensity added); the point-mass polar check requires no escapes at all; the prograde grid starts at 0.25; the Oort label states the check's range. Wording and sources: Alexander's r_h as he estimates it; Correa-Otto and Calandra's limit set by stellar passages acting with the tide; Shen and Tremaine's extents placed after their 0.5/0.7/r_H sentence; the SSD page's "not intended to represent any sort of mean"; molecular clouds "also strip it"; h_z/R0 against rms sin i as "both a few percent"; Koop et al.'s 15% / 10%; the averaged tide explains the near-polar core only; Table 1's derived values now checked. Li, Xia and Zhou's and Jiang and Tremaine's journal references, Bland-Hawthorn and Gerhard's subtitle, Merritt 2004's page, GRAVITY 2022 and 2024 titles, the SSD page title, and provenance for Nesvorny et al., Lidov, Kozai and Dones et al. Runtimes remeasured: about 7 minutes full, 10 long. Co-Authored-By: Claude Opus 5.5 Claude-Session: https://claude.ai/code/session_01C2bpiHyHSkC8DAL5CcyciK --- README.md | 2 +- orbits/README.md | 104 ++++++++++++++++++++++++++--------------------- orbits/model.js | 22 ++++++---- orbits/table.mjs | 48 ++++++++++++++++------ tests/orbits.mjs | 16 ++++++-- 5 files changed, 123 insertions(+), 69 deletions(-) diff --git a/README.md b/README.md index 86cd5c0..31e0153 100644 --- a/README.md +++ b/README.md @@ -216,7 +216,7 @@ node orbits/table.mjs quick # the orbits study's numbers, without i below). The third study's checks, `tests/critical.mjs`, add about 65 s of seeded Monte Carlo runs, most of it percolation with nine seeds. CI runs every line above except the `critical/table.mjs` line: it repeats the tests' seeded runs and prints the results without checking them. The orbits study's full table (`node orbits/table.mjs`, -a few minutes) is not run by CI either. +about 7 minutes) is not run by CI either. ## Checks diff --git a/orbits/README.md b/orbits/README.md index e8aa849..9528fb0 100644 --- a/orbits/README.md +++ b/orbits/README.md @@ -16,7 +16,7 @@ what ends it, and what, if anything, repeats from one scale to the next. | How exact is Kepler's law in the Solar System? | In JPL's 1800–2050 fit, n²a³ matches GM☉ to within 1.3×10⁻³ for all eight planets. That is about the planets' total mass, 1.34×10⁻³ of the Sun's. Jupiter's excess is mostly its own mass in the two-body law; Uranus's and Neptune's are mostly the other planets' pull, and Saturn's fitted 2.2×10⁻⁴ is below its own 2.9×10⁻⁴ (see below). P ∝ a^1.49986 over 0.39 to 30 au. | | Does the Galaxy have a Kepler region? | Yes, around Sgr A*. The star S2 orbits it as a planet orbits the Sun, plus general relativity's precession: 12.2′ per orbit from their elements (they quote 12.1′). The speed's log slope passes −1/4 about 1.3 pc out (1.0 to 1.8 pc across the cluster-mass range; 0.7 to 1.9 pc for thresholds from −0.4 to −0.1). The stars inside weigh as much as the black hole by 1.8 to 3.2 pc. | | What ends a Kepler region? | For a body orbiting in a host (a planet, the Sun), the host's tide. Satellites are held inside its tidal radius, r_J³ = Gm/(4A(A − B)), apart from quasi-satellite orbits beyond about 2 r_H that Shen and Tremaine (2008) find around Jupiter, Uranus and Neptune. Inside r_J, the body's mean density is 4A/Ω times the host's mean density inside the orbit (host taken as spherical): 3 for a planet around the Sun, 2.25 for the Sun in the Galaxy. For the Sun, r_J = 1.37 pc = 2.8×10⁵ au. At the Galaxy's centre, where Sgr A* sits still, the region ends instead where the surrounding stars' mass approaches the black hole's. | -| How much of that region is used? | About half, for orbits that do not plunge, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.52 r_J around a planet and by 0.435 r_J (1.23×10⁵ au) for the Sun. Polar orbits are lost much sooner, to Lidov–Kozai cycles: half are lost by 0.14 to 0.15 r_H around a planet (up to about 0.3 r_H; from 0.35 r_H on, half or more survive the run) and by 0.06 r_J for the Sun, in 20 turns of the host. For near-circular moons near the ecliptic, Nesvorný et al. (2003) put the limits at about 0.4 and 0.7 r_H (prograde, retrograde), and the model gives 0.46 and 0.775. The Oort cloud's outer limit is placed between 6.5×10⁴ au (Correa-Otto and Calandra 2019, where passing stars set it) and 2×10⁵ au, which is 0.23 to 0.71 r_J. | +| How much of that region is used? | About half, for orbits that do not plunge, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.52 r_J around a planet and by 0.433 r_J (1.23×10⁵ au) for the Sun. Polar orbits are lost much sooner, to Lidov–Kozai cycles: half are lost by 0.14 r_H around a planet (the lost fraction dips below half near 0.4 r_H) and by 0.06 r_J for the Sun, in 20 turns of the host. For near-circular moons near the ecliptic, Nesvorný et al. (2003) put the limits at about 0.4 and 0.7 r_H (prograde, retrograde), and the model gives 0.46 and 0.78. The Oort cloud's outer limit is placed between 6.5×10⁴ au (Correa-Otto and Calandra 2019, where the frequency of stellar passages, acting with the tide, sets it) and 2×10⁵ au, which is 0.23 to 0.71 r_J. | | What does repeat? | Kepler regions nested inside one another: moons in a planet's, planets and comets in the Sun's. Each is bounded by its host's tide at a similar fraction of the tidal radius, though for the Oort cloud passing stars cut it shorter. Also flat disks: the planets' orbits are tilted by about 2° on average (6° for Mercury), and the Galaxy's thin disk has a scale height 0.04 of the Sun's orbital radius. | ## Kepler's law in the Solar System @@ -43,7 +43,7 @@ own m/M☉ and adds the pull of each other planet, smeared into a ring along its - **Jupiter.** It gives 9.3×10⁻⁴: its own 9.5×10⁻⁴, less 2.9×10⁻⁵ net from the other planets' rings (−3.5×10⁻⁵ from those outside its orbit, +0.6×10⁻⁵ from those inside). The fit gives 9.0×10⁻⁴. - **Saturn, Uranus and Neptune.** It gives 1.4 to 1.5×10⁻³, above the planets' total mass. Mass inside an orbit pulls harder spread into a ring than gathered at the centre. -- **What a fit this short keeps.** A 250-year fit also carries long-period terms, such as the roughly 900-year great inequality of Jupiter and Saturn near their 5:2 resonance (recalled, not modelled here). Saturn's fitted 2.2×10⁻⁴ sits 1.3×10⁻³ below the ring estimate. Whether Table 1's outer-planet elements are secular means, which would settle this, is not checked. +- **What a fit this short keeps.** A 250-year fit also carries long-period terms, such as the roughly 900-year great inequality of Jupiter and Saturn near their 5:2 resonance (recalled, not modelled here). Saturn's fitted 2.2×10⁻⁴ sits 1.3×10⁻³ below the ring estimate. Table 1's elements are not secular means, which is consistent with this: the SSD page says such elements "are not intended to represent any sort of mean; they are simply the result of being adjusted for a best fit". - **The inner planets.** The ring estimate is below 6×10⁻⁶ and the fits are 5×10⁻⁶ to 4×10⁻⁵, so other terms dominate there too. So the study claims the size of the departures from Kepler's law, about the planets' total mass, and no more. The @@ -91,44 +91,55 @@ for Jupiter it is 237 years. - **Escaped:** once it passes 3 r_J. - **Plunged:** once it comes within 10⁻³ r_J, which is 0.74 of Jupiter's radius or, for the Sun, 283 au, outside the planets. - **Lost:** either. -- **Starting angles.** Hill's equations are symmetric under r → −r, v → −v, so they are spread over half a turn and each is a different orbit. +- **Starting angles.** Hill's equations are symmetric under r → −r, v → −v, so a start half a turn on is the inverted orbit, with the same fate. They are also symmetric under time reversal with y → −y. For every single-plane row here, that makes the start at π − φ the mirror image of the start at φ with time running backward, and the two fates nearly always match. So the starting angles are spread over a quarter turn, where no two are related by either symmetry. In the all-directions sample the mirror is a different direction. There, starting angles in the second quarter turn give the same fractions to within about 0.03 (section 5). The radius by which half the orbits have escaped, and by which half are lost (`node orbits/table.mjs`, section 5; 8 starting angles; grid 0.025 r_J; all directions on 48 directions × 4 angles and a 0.05 grid): | Orbit | Point-mass host | The Sun in the Galaxy | The same with ν²/Ω² = 1 | |---|---|---|---| -| Prograde, in the plane | 0.46 | 0.475 | 0.475 | -| Retrograde, in the plane | 0.775 | 0.95 | 0.95 | -| Polar, plane through the host's direction: escaped / lost | 0.46 / 0.14 | 0.43 / 0.06 | 0.46 / 0.16 | -| Polar, plane across it: escaped / lost | 0.485 / 0.15 | 0.39 / 0.06 | 0.50 / 0.16 | -| Facing every way: escaped / lost | 0.52 / 0.515 | 0.435 / 0.42 | | +| Prograde, in the plane | 0.46 | 0.49 | 0.49 | +| Retrograde, in the plane | 0.775 | 0.925 | 0.925 | +| Polar, plane through the host's direction: escaped / lost | 0.47 / 0.14 | 0.40 / 0.06 | 0.475 / 0.17 | +| Polar, plane across it: escaped / lost | 0.49 / 0.14 | 0.39 / 0.06 | 0.49 / 0.16 | +| Facing every way: escaped / lost | 0.52 / 0.52 | 0.433 / 0.42 | | -In-plane orbits plunge only outside their escape edge (first at 0.55 r_H, prograde around a point mass, and at -0.875 and 1.0 r_J retrograde), so for them the escaped and lost radii agree. For polar rows, "lost" is the first -radius where half are lost. +**The table's resolution.** With 64 starting angles instead of 8, the in-plane edges are 0.46 and 0.79 around a +point mass and 0.95 for the Sun's retrograde orbits. So the 8-angle values carry sampling errors of up to about 0.03. + +**In-plane orbits.** These plunge only rarely: with 8 angles, first at 0.55 r_J prograde and at 0.85 and 1.1 r_J +retrograde, all outside their escape edge. With 64, the point mass's first retrograde plunge is at 0.775, just inside +its 0.788 edge, and its lost radius is 0.783. So for in-plane orbits the escaped and lost radii agree to about 0.005. + +**Polar rows.** "Lost" is the first radius where half are lost. **Polar orbits are lost by plunging.** - **The mechanism.** A circular orbit at 90° to the host's plane goes through Lidov–Kozai cycles (Lidov 1962, Kozai 1962), whose eccentricity reaches 1. So it dives into the body long before it can escape. -- **When.** Within 20 turns half are lost from 0.14 to 0.15 r_H around a planet and from 0.06 r_J for the Sun. Further in, the cycle takes longer than 20 turns; it is not absent. -- **Not everywhere outside.** Around a point mass the lost fraction peaks at 0.2 r_H (all 8 orbits) and falls back to 0.25–0.5 at 0.35–0.4 r_H, where most polar orbits survive the run; so does the Sun's with ν²/Ω² = 1. With the Galaxy's vertical tide the Sun loses 63–100% of its polar orbits at every radius from 0.075 to 0.45 r_J. -- **Other directions.** Tilted to 75° instead, none plunge from 0.1 to 0.4 r_H around a planet, nor inside 0.3 r_J for the Sun; at 0.35–0.4 r_J, 25–38% of the Sun's do. Facing every way, up to a fifth of the Sun's orbits plunge within 20 turns, from directions within 29° of polar, and 8–11% of a planet's between 0.2 and 0.4 r_H, from within 11° of polar. For comets this is the tide driving them inward (Heisler and Tremaine 1986). +- **When.** Within 20 turns half are lost from 0.14 r_H around a planet and from 0.06 r_J for the Sun. Further in, the cycle takes longer than 20 turns; it is not absent. +- **Not everywhere outside.** Around a point mass the lost fraction is 75–100% at 0.15–0.2 r_H and dips below half near 0.4 r_H (0–25% at 0.4), where most polar orbits survive the run. With ν²/Ω² = 1 the Sun's lost fraction also dips below half, from 0.35 r_J in the plane across the host's direction and at 0.45 r_J in the plane through it. With the Galaxy's vertical tide the Sun loses half or more of its polar orbits at every radius printed, 0.1 to 0.45 r_J. +- **Other directions.** For a planet, plunges reach to about 8°, 10° and 14° from polar at 0.2, 0.3 and 0.4 r_H. Tilted to 75° (15° from polar), none plunge from 0.1 to 0.4 r_H. For the Sun they reach to about 8° from polar at 0.2 r_J and 24° at 0.3 r_J. At 75°, none plunge inside 0.3 r_J, and up to 63% do at 0.3–0.4 r_J. +- **All directions.** Facing every way, up to a fifth of the Sun's orbits plunge within 20 turns, and 8–12% of a planet's between 0.2 and 0.4 r_H. +- **What the averaged tide explains.** For comets, the tide driving them inward (Heisler and Tremaine 1986) accounts for the near-polar core. The Sun's wider cone near its escape edge, and its plunges at 75°, are not secular: #37's re-review found that the averaged tide gives a cone only 3.6 to 5.1° wide at 0.2 to 0.4 r_J, and no plunges at 75°. These plunges depend steeply on the plunge radius: counted inside 10⁻² r_J, the largest plunged fraction over all directions rises from 0.24 to 0.41 (24 directions). **For a planet.** -- **Against the published limits.** The model's in-plane edges are 0.46 (prograde) and 0.775 (retrograde), against Nesvorný et al.'s (2003) semi-major axes of about 0.4 and 0.7 r_H for nearly circular orbits near the ecliptic, as Shen and Tremaine (2008) quote them. -- **Over all inclinations.** Shen and Tremaine find that many prograde and retrograde satellites survive to about 0.5 and 0.7 r_H, and some coplanar retrograde ones to about r_H. The observed irregular moons lie at about 0.05 to 0.6 r_H, the retrograde ones further out (about 0.6 against 0.4 r_H). Their extents are instantaneous distances, not semi-major axes. +- **Against the published limits.** The model's in-plane edges are 0.46 (prograde) and 0.775 to 0.79 (retrograde), against Nesvorný et al.'s (2003) semi-major axes of about 0.4 and 0.7 r_H for nearly circular orbits near the ecliptic, as Shen and Tremaine (2008) quote them. +- **Over all inclinations.** Shen and Tremaine find that many prograde and retrograde satellites survive to about 0.5 and 0.7 r_H, and some coplanar retrograde ones to about r_H; their extents are instantaneous distances, not semi-major axes. The observed irregular moons lie at about 0.05 to 0.6 r_H, the retrograde ones further out (about 0.6 against 0.4 r_H). - **Run length.** The runs are short: 20 turns is 237 years for Jupiter. - - Followed for 100 turns, the all-directions value moves from 0.53 to 0.51 (24 directions, `node orbits/table.mjs long`). + - Followed for 100 turns, the all-directions value moves from 0.52 to 0.51 (24 directions, `node orbits/table.mjs long`). - In a 48-orbit sample run to 3000 turns, #37's review found the point-mass edge drifting from 0.50 to 0.475. **For the Sun.** -- **The edge.** Half of the orbits facing every way have escaped by 0.435 r_J, or 1.23×10⁵ au. - - For ρ₀ from 0.08 to 0.12 it moves from 0.454 to 0.429 (on 24 directions, which give 0.448 at 0.10). - - Counting plunges inside 10⁻² r_J instead of 10⁻³ moves the 24-direction value from 0.448 to 0.469. - - Followed for 100 turns (23 billion years, a convergence test only) it is 0.42, and 0.38 counting plunges, against 0.45 and 0.43 over 20 turns (24 directions). -- **The Oort cloud.** Correa-Otto and Calandra (2019, sect. 4.3) find fewer than 5% of comets still bound beyond 6.5×10⁴ au and none at 10⁵ au. They define a_crit ≈ 6.5×10⁴ au as the cloud's outer limit; half are bound at about 2×10⁴ au. They name passing stars, not the tide, as what sets that limit. Dones et al. (2004, as cited) place the edge as far as 2×10⁵ au. So the quoted outer limits run from 0.23 to 0.71 r_J. The model's 0.435 r_J lies past Correa-Otto and Calandra's: the tide alone would keep comets further out than passing stars allow. -- **The vertical tide.** It loses polar orbits sooner, by escape at 0.43 and 0.39 r_J against 0.46 and 0.50 with ν²/Ω² = 1 at the same Γ, and by plunging at 0.06 against 0.16. It never acts on an orbit that starts in the plane, because such an orbit stays there. -- **In-plane retrograde orbits.** These last to 0.95 r_J rather than 0.775 because Γ is 2.25 rather than 3: the flatter rotation curve weakens the in-plane tide. With Γ = 2.12 from Eilers et al.'s slope instead (see Across the disk) the edge is 0.975. +- **The edge.** Half of the orbits facing every way have escaped by 0.433 r_J, or 1.23×10⁵ au. + - For ρ₀ from 0.08 to 0.12 it moves from 0.446 to 0.431 (on 24 directions, which give 0.441 at 0.10). + - Counting plunges inside 10⁻² r_J instead of 10⁻³ moves the 24-direction escaped edge out, from 0.441 to 0.458, but the lost edge in, from 0.410 to 0.395. + - Followed for 100 turns (23 billion years, a convergence test only) it is 0.42, and 0.38 counting plunges, against 0.44 and 0.41 over 20 turns (24 directions). +- **The Oort cloud.** Correa-Otto and Calandra (2019, sect. 4.3) follow comets by their initial semi-major axis. They find fewer than 5% still bound beyond 6.5×10⁴ au and none at 10⁵ au. + - They define a_crit ≈ 6.5×10⁴ au as the cloud's outer limit. Half are bound at about 2×10⁴ au, a mean over three samples. + - They describe a joint ("synergy") effect of the Galactic potential and stellar passages, and name the frequency of stellar passages as the main modulator of that limit. + - Dones et al. (2004, as cited) place the edge as far as 2×10⁵ au. So the quoted outer limits run from 0.23 to 0.71 r_J. + - The model's 0.433 r_J lies past Correa-Otto and Calandra's: the tide alone would keep comets further out than the tide and passing stars together allow. +- **The vertical tide.** It loses polar orbits sooner: by escape at 0.40 and 0.39 r_J, against 0.475 and 0.49 with ν²/Ω² = 1 at the same Γ, and by plunging at 0.06 against 0.16 to 0.17. It never acts on an orbit that starts in the plane, because such an orbit stays there. +- **In-plane retrograde orbits.** These last to 0.925 r_J (0.95 with 64 angles) rather than 0.775 because Γ is 2.25 rather than 3: the flatter rotation curve weakens the in-plane tide. With Γ = 2.12 from Eilers et al.'s slope instead (see Across the disk), the edge is about 1.0 r_J: 1.00 with 8 angles and 1.03 with 64. ## A Kepler region at the Galaxy's centre @@ -151,7 +162,7 @@ those values are lower limits. With γ = 1.4 the middle "twice" radius is 3.8 pc The literature puts the "twice" radius, the usual influence radius, at 3 to 4 pc: - Schödel et al. 2018, sect. 4.5: 7.8×10⁶ M☉ within 3 pc, "roughly twice the mass of Sgr A*". -- Alexander 2005: r_h ~ 3 pc from Gm/σ², with twice the black hole's mass reached by about 4 pc. +- Alexander 2005: r_h ∼ 3 pc, estimated from the stars' mass reaching m by ≲2 pc and 2m by ≲4 pc (he calls Gm/σ² customary but not well defined). - Feldmeier-Krause et al. 2017: about 4 pc (4.0 +2.2/−1.1), with a black-hole mass of 4×10⁶ M☉ and Merritt's (2004) definition. - GRAVITY Collaboration 2024: 3.9 pc in their cusp simulation, with Merritt's (2013) definition. @@ -186,7 +197,7 @@ Going outward, the model's speed falls by less than Kepler's law would give: - **The model.** The log slope is −0.21 at 100 kpc and −0.27 at 200 kpc, because the NFW halo's mass still grows. - **The data.** Gaia DR3 analyses differ on whether the measured curve turns down past about 19 kpc. - Jiao et al. (2023) report "a Keplerian decline". - - Koop et al. (2024) find Jeans-equation curves robust with their statistics only to about 15 kpc. In simulations, such curves are off by about 10%. + - Koop et al. (2024) find Jeans-equation curves robust with their statistics only to about 15 kpc. In simulations they find such curves off by as much as 15%, but likely of order 10%. ## An acceleration scale, and why this study does not use it @@ -214,9 +225,9 @@ total angular momentum (Souami and Souchay 2012). Computed from JPL's elements, The rms of sin i is 0.045, or 0.023 without Mercury. The Galaxy's thin disk has an exponential scale height of 300 ± 50 pc over a scale length of 2.6 ± 0.5 kpc -(Bland-Hawthorn and Gerhard 2016). Their ratio, 0.12, is the axis ratio of its density contours. The like-for-like -comparison with the planets, height over orbital radius, is 300 pc over the Sun's 8.2 kpc: 0.037, against the -planets' 0.045. +(Bland-Hawthorn and Gerhard 2016). Their ratio, 0.12, is the axis ratio of its density contours. Height over +orbital radius is 300 pc over the Sun's 8.2 kpc, 0.037, against the planets' rms sin i of 0.045: both a few percent. +The two measures differ in convention, an exponential scale height against an rms, by about √2. Both are flat for the same reason: gas that radiates its energy but keeps its angular momentum settles into a disk. The Galaxy's present thickness is also set by the stars' later heating. What the disks orbit differs: one dominant @@ -225,9 +236,9 @@ mass, or mass spread through the disk and halo. ## Run it ``` -node orbits/table.mjs # every computed number above except the 100-turn runs and the values credited to #37's review (about 3 minutes) +node orbits/table.mjs # every computed number above except the 100-turn runs and the values credited to #37's review (about 7 minutes) node orbits/table.mjs quick # sections 1-4 only, well under a second (run by CI) -node orbits/table.mjs long # adds the 100-turn runs, about 5 minutes in all +node orbits/table.mjs long # adds the 100-turn runs, about 10 minutes in all node tests/orbits.mjs # the checks, about 3 s (also run by npm test) ``` @@ -235,7 +246,7 @@ The table exits with an error if any printed value is not a finite number. ## Checks -`tests/orbits.mjs` has 38 checks. Each prints its source, the published or expected value, the model's value and +`tests/orbits.mjs` has 43 checks. Each prints its source, the published or expected value, the model's value and the tolerance. "#37's review" marks values a reviewer read in the paper's arXiv source. | What | Against | Kind | @@ -249,6 +260,7 @@ the tolerance. "#37's review" marks values a reviewer read in the paper's arXiv | Kepler slope −1/2 inside 0.1 pc; the −1/4 radius in closed form against bisection | — | model; model-internal | | ellipticK(1/√2); gammaP(½, x) = erf √x at three points | closed forms | arithmetic | | MWPotential2014: the 220 km/s normalisation; halo mass 0.8×10¹² inside 245 kpc; the disk's sphere mass tending to its total, and its Kuzmin and Plummer limits; shape within 5% of Eilers et al.; flat (abs. slope < 0.25) at 5 to 25 kpc | Bovy 2015 Table 1; closed forms; Eilers et al. 2019 | model-internal; #37's review; arithmetic | +| MWPotential2014's derived values: bulge mass 0.5 and disk mass 6.8×10¹⁰ M☉, v²R/G at 60 kpc 4.1×10¹¹ M☉, log slope −0.10 at 8 kpc, dark-matter density 0.008 M☉ pc⁻³ at 8 kpc | Bovy 2015 Table 1 | #37's re-review (arXiv source); at Table 1's rounding | | Jupiter's and Saturn's tilts to the invariable plane | Souami and Souchay 2012 | search text, not confirmed in the paper | | Planet: prograde mostly kept at 0.4 r_H and lost by 0.6; prograde edge 0.4 ± 0.1; retrograde edge between 0.7 and 1 r_H | Nesvorný et al. 2003 and Shen and Tremaine 2008, as Shen and Tremaine give them | model; #37's review | | Retrograde in the plane at 0.85 r_J: mostly lost around a point mass, kept by the Sun | — | model | @@ -261,15 +273,15 @@ the tolerance. "#37's review" marks values a reviewer read in the paper's arXiv - **The planets' long-period pull on one another.** Kepler's law is checked only to the size of the planets' masses. - **Hill's problem is the tide to first order, on a circular host orbit.** The study drops the Sun's vertical oscillation through the disk (82 Myr with this model's ν) and its radial epicycles. It also leaves out: - - passing stars and molecular clouds, which set the Oort cloud's actual edge; + - passing stars, which with the tide set the Oort cloud's actual edge, and molecular clouds, which also strip it; - the planets' scattering of comets that plunge inward; - for moons, the planet's oblateness and its other moons, which damp Lidov–Kozai cycles close in. - **Circular starting orbits only.** Real moons and comets start on eccentric orbits. - **The edges' definition.** They are set at half the orbits escaping, or lost, within 20 turns of the host, on a grid of 0.025 r_J (single planes) or 0.05 r_J (all directions). - - **Plunge radius.** It is 10⁻³ r_J. Raising it to 10⁻² moves the Sun's all-directions edge from 0.448 to 0.469 (24 directions). - - **Chaos.** Single-plane edges are chaotic at about the 0.02 level: changing ν²/Ω² by 10⁻⁴ moves the Sun's polar-across escape edge from 0.394 to 0.412. + - **Plunge radius.** It is 10⁻³ r_J. Raising it to 10⁻² moves the Sun's all-directions escaped edge out, from 0.441 to 0.458, and its lost edge in, from 0.410 to 0.395 (24 directions). + - **Chaos and sampling.** The polar edges are chaotic: changing ν²/Ω² by 10⁻⁴ moves the Sun's polar-across escape edge from 0.388 to 0.392. The in-plane edges are not, but with 8 starting angles they carry sampling errors of up to about 0.03 (see the 64-angle values above). - **Longer runs** move the edges inward a little (see the 100-turn values above). -- **MWPotential2014 is one model among many.** Its halo sets the flat range, and it has no nuclear star cluster. The checks pin its halo; the disk's and bulge's shape parameters are taken from galpy's source and not otherwise tested, since Table 1's derived masses were not read for this study. The Galactic-centre numbers use the separate power-law cusp above, which is valid only inside its 3 pc break. +- **MWPotential2014 is one model among many.** Its halo sets the flat range, and it has no nuclear star cluster. The checks pin its halo, its Plummer limit pins the disk's b, and Table 1's derived values (bulge and disk masses, the mass inside 60 kpc, the slope at R₀ and the local dark-matter density) are checked at Table 1's rounding. The other shape parameters are taken from galpy's source and tested only through those totals. The Galactic-centre numbers use the separate power-law cusp above, which is valid only inside its 3 pc break. - **Spherical mass estimates.** v²R/G and 3Ω²/4πG treat a flattened galaxy as spherical. In MWPotential2014 this overstates the mass inside 8 kpc by 12%. It shifts the 3.7-fold growth to 4.0 and the Sun's 2.25 to 2.53, but not the contrast with Kepler's fixed mass. ## Sources @@ -278,31 +290,31 @@ Each entry says how it was read. "#37's review" means a reviewer checked the val or in the primary file, during #37's review. - W. M. Folkner et al., *The Planetary and Lunar Ephemerides DE430 and DE431*, IPN Progress Report 42-196 (2014), Table 8. #37's review. -- E. M. Standish and J. G. Williams, *Keplerian Elements for Approximate Positions of the Major Planets*, JPL Solar System Dynamics (1992), Table 1 (1800–2050). Checked against the SSD page in #37's review. +- E. M. Standish and J. G. Williams, *Approximate Positions of the Planets*, JPL Solar System Dynamics (credited there to Standish and Williams 1992), Table 1 (1800–2050). Checked against the SSD page in #37's review. - A. Prša et al., *Nominal values for selected solar and planetary quantities: IAU 2015 Resolution B3*, Astron. J. 152, 41 (2016). https://doi.org/10.3847/0004-6256/152/2/41. #37's review. - D. Souami and J. Souchay, *The solar system's invariable plane*, A&A 543, A133 (2012). https://doi.org/10.1051/0004-6361/201219011. Search text; not reached in #37's review. The plane's orientation is confirmed through Li, Xia and Zhou 2019; the per-planet tilts are not. -- Li, Xia and Zhou, *Calibration of the angular momenta of the minor planets in the solar system* (2019). https://arxiv.org/abs/1909.11293. For the invariable plane's orientation, as #37's review read it; journal reference not confirmed. -- Y.-F. Jiang and S. Tremaine, *The evolution of wide binary stars*, MNRAS 401, 977 (2010; journal reference recalled). https://arxiv.org/abs/0907.2952. The tidal-radius formula and constants, as #37's review quotes them. +- J. Li, Z. J. Xia and L. Zhou, *Calibration of the angular momenta of the minor planets in the solar system*, A&A 630, A68 (2019). https://doi.org/10.1051/0004-6361/201834196. For the invariable plane's orientation, as #37's review read it. +- Y.-F. Jiang and S. Tremaine, *The evolution of wide binary stars*, MNRAS 401, 977 (2010). https://doi.org/10.1111/j.1365-2966.2009.15744.x. The tidal-radius formula and constants, as #37's review quotes them. - J. Binney and S. Tremaine, *Galactic Dynamics*, 2nd ed., Princeton (2008). The formula, as Jiang and Tremaine use it; section not confirmed. - J. Bovy, *Galactic rotation in Gaia DR1*, MNRAS 468, L63 (2017). https://doi.org/10.1093/mnrasl/slx027. #37's review. - J. A. Correa-Otto and M. F. Calandra, *Stability in the most external region of the Oort Cloud: evolution of the ejected comets*, MNRAS 490, 2495 (2019). https://doi.org/10.1093/mnras/stz2671. #37's review (sect. 4.3). - Y. Shen and S. Tremaine, *Stability of Distant Satellites of the Giant Planets in the Solar System*, AJ 136, 2453 (2008). https://doi.org/10.1088/0004-6256/136/6/2453. #37's review. -- D. Nesvorný, J. L. A. Alvarellos, L. Dones and H. F. Levison, *Orbital and collisional evolution of the irregular satellites*, AJ 126, 398 (2003). As Shen and Tremaine quote it; reference recalled. -- M. L. Lidov, Planet. Space Sci. 9, 719 (1962), and Y. Kozai, AJ 67, 591 (1962). Recalled. +- D. Nesvorný, J. L. A. Alvarellos, L. Dones and H. F. Levison, *Orbital and collisional evolution of the irregular satellites*, AJ 126, 398 (2003). As Shen and Tremaine quote it; the reference matches Crossref (#37's re-review), the text was not read. +- M. L. Lidov, Planet. Space Sci. 9, 719 (1962), and Y. Kozai, AJ 67, 591 (1962). The references match Crossref (#37's re-review); the texts were not read. - C. F. McKee, A. Parravano and D. J. Hollenbach, *Stars, gas, and dark matter in the solar neighborhood*, ApJ 814, 13 (2015). https://arxiv.org/abs/1509.05334. #37's review. - J. Heisler and S. Tremaine, *The influence of the Galactic tidal field on the Oort comet cloud*, Icarus 65, 13 (1986). Search text; not reached in #37's review. -- L. Dones, P. R. Weissman, H. F. Levison and M. J. Duncan, *Oort cloud formation and dynamics*, in *Comets II*, 153 (2004). As cited in search text, for an outer edge "as far as 2 × 10^5 AU". +- L. Dones, P. R. Weissman, H. F. Levison and M. J. Duncan, *Oort cloud formation and dynamics*, in *Comets II*, 153–174 (2004). The reference matches Crossref (#37's re-review); the text was not read. The edge "as far as 2 × 10^5 AU" is as cited in search text. - GRAVITY Collaboration, *Detection of the Schwarzschild precession in the orbit of the star S2 near the Galactic centre massive black hole*, A&A 636, L5 (2020). https://doi.org/10.1051/0004-6361/202037813. #37's review (sect. 1.2, appendix B, Table E.1). -- GRAVITY Collaboration, A&A 657, L12 (2022). https://doi.org/10.1051/0004-6361/202142465. #37's review. -- GRAVITY Collaboration, A&A 692, A242 (2024). https://doi.org/10.1051/0004-6361/202452274. #37's review (sect. 4.3). +- GRAVITY Collaboration, *Mass distribution in the Galactic Center based on interferometric astrometry of multiple stellar orbits*, A&A 657, L12 (2022). https://doi.org/10.1051/0004-6361/202142465. #37's review. +- GRAVITY Collaboration, *Improving constraints on the extended mass distribution in the Galactic Center with stellar orbits*, A&A 692, A242 (2024). https://doi.org/10.1051/0004-6361/202452274. #37's review (sect. 4.3). - R. Schödel, D. Merritt and A. Eckart, A&A 502, 91 (2009). https://doi.org/10.1051/0004-6361/200810922. #37's review. - R. Schödel et al., A&A 609, A27 (2018). https://doi.org/10.1051/0004-6361/201730452. #37's review (sect. 4.5). - T. Alexander, *Stellar processes near the massive black hole in the Galactic center*, Phys. Rep. 419, 65 (2005). https://doi.org/10.1016/j.physrep.2005.08.002. #37's review (sect. 3.1.2). - A. Feldmeier-Krause et al., *Triaxial orbit-based modelling of the Milky Way Nuclear Star Cluster*, MNRAS 466, 4040 (2017). https://doi.org/10.1093/mnras/stw3377. #37's review (sect. 3.3.2). -- D. Merritt, *Dynamics and Evolution of Galactic Nuclei*, Princeton (2013), and D. Merritt, in *Coevolution of Black Holes and Galaxies*, ed. L. C. Ho (2004; recalled). Definitions only, as GRAVITY 2024 and Feldmeier-Krause et al. cite them. +- D. Merritt, *Dynamics and Evolution of Galactic Nuclei*, Princeton (2013), and D. Merritt, in *Coevolution of Black Holes and Galaxies*, ed. L. C. Ho, p. 263 (2004). Definitions only, as GRAVITY 2024 and Feldmeier-Krause et al. cite them. - J. Bovy, *galpy: A Python Library for Galactic Dynamics*, ApJS 216, 29 (2015). https://doi.org/10.1088/0067-0049/216/2/29. Table 1 and galpy's source, #37's review. - A.-C. Eilers, D. W. Hogg, H.-W. Rix and M. K. Ness, ApJ 871, 120 (2019). https://doi.org/10.3847/1538-4357/aaf648. #37's review (eq. 7). -- J. Bland-Hawthorn and O. Gerhard, *The Galaxy in Context*, ARA&A 54, 529 (2016). https://doi.org/10.1146/annurev-astro-081915-023441. #37's review. +- J. Bland-Hawthorn and O. Gerhard, *The Galaxy in Context: Structural, Kinematic, and Integrated Properties*, ARA&A 54, 529 (2016). https://doi.org/10.1146/annurev-astro-081915-023441. #37's review. - Y. Jiao et al., A&A 678, A208 (2023). https://doi.org/10.1051/0004-6361/202347513. #37's review. - O. Koop et al., A&A 692, A50 (2024). https://arxiv.org/abs/2405.19028. #37's review. - S. S. McGaugh, F. Lelli and J. M. Schombert, PRL 117, 201101 (2016). https://doi.org/10.1103/PhysRevLett.117.201101. #37's review. diff --git a/orbits/model.js b/orbits/model.js index 42d3313..5e0bf5c 100644 --- a/orbits/model.js +++ b/orbits/model.js @@ -221,6 +221,8 @@ }; // The total mass inside a sphere of radius r [M_sun]. const sphereMass = r => M.bh + bulgeMass(r) + diskSphereMass(r) + M.halo * shape.halo(r); + // The halo's (NFW) density at radius r [M_sun/pc^3]. + const haloDensity = r => M.halo / (4 * Math.PI * par.rs ** 3) / ((r / par.rs) * (1 + r / par.rs) ** 2) / 1e9; // the smallest radius above r0 where f crosses zero, by a log grid then bisection const cross = (f, r0, r1) => { let lo = r0, flo = f(lo); @@ -236,7 +238,7 @@ return NaN; }; return { - par, M, parts, v, slope, bulgeMass, diskSphereMass, sphereMass, cross, + par, M, parts, v, slope, bulgeMass, diskSphereMass, sphereMass, haloDensity, cross, // where the bulge's stars inside r first weigh k times Sgr A* (k = 1, equal mass; k = 2, the influence radius) influence: (k = 1) => cross(r => bulgeMass(r) - k * M.bh, 1e-6, 1), // the longest run of radii (kpc, on a grid 0.5% apart) where |dln v / dln R| < tol; by default outside the bulge @@ -284,8 +286,10 @@ }; // Follows one orbit for `orbits` turns of the host (time 2 pi orbits). Fate: 'escapes' once r > rEsc, 'plunges' // once r < rIn (1e-3 r_J is 0.74 of Jupiter's radius; for the Sun it is 283 au, outside the planets), otherwise - // 'stays'. Returns the final state s too. Throws on a non-finite step or after maxSteps, rather than looping. - function hillOrbit(host, s0, { orbits = 20, rEsc = 3, rIn = 1e-3, tol = 1e-10, maxSteps = 5e6 } = {}) { + // 'stays'. Returns the final state s too. Throws on non-finite settings, on a non-finite step, or after maxSteps + // (by default 2.5e5 per turn of the host; the hardest orbit in the tables needs about 1.5e4), rather than looping. + function hillOrbit(host, s0, { orbits = 20, rEsc = 3, rIn = 1e-3, tol = 1e-10, maxSteps = 2.5e5 * orbits } = {}) { + if (![orbits, rEsc, rIn, tol, maxSteps, host.Gam, host.nu2, ...s0].every(Number.isFinite)) throw new Error('hillOrbit: non-finite input'); const T = 2 * Math.PI * orbits, k = Array.from({ length: 7 }, () => new Float64Array(6)); const y = new Float64Array(6), s = Float64Array.from(s0); let t = 0, h = 1e-3, steps = 0; @@ -334,11 +338,15 @@ return out; } // Fractions of orbits that stay, escape and plunge, over the given orbit normals and nPhase starting angles each. - // Hill's equations are symmetric under r -> -r, v -> -v, and a start at phi + pi is the inverted start at phi, so the - // starting angles are spread over [0, pi) only: each one is a different orbit. - function hillFates(host, a, normals, { nPhase = 4, ...opt } = {}) { + // Hill's equations are symmetric under r -> -r, v -> -v: a start at phi + pi is the inverted start at phi, with the + // same fate. They are also symmetric under time reversal with y -> -y. For a normal in the x-z or y-z plane, as in + // every single-plane scan here, that makes a start at pi - phi the mirror of the start at phi run backward, whose + // fate nearly always matches. So the starting angles are spread over the quarter turn [0, pi/2) (quarter = 0), where + // no two are related by either symmetry; quarter = 1 takes [pi/2, pi) instead. For other normals the mirror is + // another normal, and [0, pi/2) samples half of each normal's distinct starts. + function hillFates(host, a, normals, { nPhase = 4, quarter = 0, ...opt } = {}) { const n = { stays: 0, escapes: 0, plunges: 0 }; - for (const nv of normals) for (let p = 0; p < nPhase; p++) n[hillOrbit(host, hillStart(host, a, nv, Math.PI * (p + 0.5) / nPhase), opt).fate]++; + for (const nv of normals) for (let p = 0; p < nPhase; p++) n[hillOrbit(host, hillStart(host, a, nv, Math.PI / 2 * (quarter + (p + 0.5) / nPhase)), opt).fate]++; const tot = normals.length * nPhase; return { stays: n.stays / tot, escapes: n.escapes / tot, plunges: n.plunges / tot }; } diff --git a/orbits/table.mjs b/orbits/table.mjs index 61fdc8b..759835b 100644 --- a/orbits/table.mjs +++ b/orbits/table.mjs @@ -44,7 +44,7 @@ const rJ = O.jacobi(C.GMsun, OORT.A, OORT.B); const rJs = [[OORT.A + OORT.dA, OORT.B], [OORT.A - OORT.dA, OORT.B], [OORT.A, OORT.B + OORT.dB], [OORT.A, OORT.B - OORT.dB]].map(([A, B]) => O.jacobi(C.GMsun, A, B) / C.pc); console.log(` the Sun in the Galaxy (A = ${OORT.A}, B = ${OORT.B} km/s/kpc, Bovy 2017): r_J = ${f(rJ / C.pc)} pc = ${f(rJ / C.au / 1e5, 2)}e5 au` + ` (moving A or B by its error: ${f(Math.min(...rJs))} to ${f(Math.max(...rJs))} pc)`); -console.log(` with Jiang & Tremaine's A = 14.8 and Omega = 29.6 km/s/kpc (B = A - Omega), as Correa-Otto & Calandra use: ` + +console.log(` with A = 14.8 and Omega = 29.6 km/s/kpc (B = A - Omega; Jiang & Tremaine's constants as Correa-Otto & Calandra round them): ` + `${f(O.jacobi(C.GMsun, 14.8, 14.8 - 29.6) / C.au / 1e5, 2)}e5 au`); console.log(` the Sun's mean density inside r_J over the Galaxy's inside the Sun's orbit: 4A/(A - B) = ${f(O.densityRatio(OORT.A, OORT.B), 2)}` + ` (3 for a point-mass host, 2 for a flat rotation curve; 2(1 - dln v/dln R) with dln v/dln R = -(A + B)/(A - B) = ${f(-(OORT.A + OORT.B) / (OORT.A - OORT.B))})`); @@ -139,20 +139,20 @@ if (!quick) { const grid = (a0, a1, da) => { const out = []; for (let a = a0; a <= a1 + 1e-9; a += da) out.push(+a.toFixed(4)); return out; }; const big = grid(0.3, 1.2, 0.025), all = grid(0.025, 1.2, 0.025); const sunNu1 = { ...sun, nu2: 1, name: 'the Sun in the Galaxy with (nu/Omega)^2 = 1' }; - console.log(' radius [r_J] where half the orbits (8 starting angles in [0, pi)) have escaped beyond 3 r_J / are lost (escaped, or plunged within 1e-3 r_J):'); + console.log(' radius [r_J] where half the orbits (8 starting angles in [0, pi/2)) have escaped beyond 3 r_J / are lost (escaped, or plunged within 1e-3 r_J):'); console.log(' orbit point-mass host the Sun in the Galaxy the same with (nu/Omega)^2 = 1'); const profiles = []; for (const [k, n] of Object.entries(planes)) { const polar = k.startsWith('polar'); const e = [O.HOST_KEPLER, sun, sunNu1].map(h => O.hillScan(h, n, polar ? all : big, { nPhase: 8 })); console.log(` ${k.padEnd(28)} ${e.map(r => `${f(r.edge, 3)} / ${f(r.edgeLost, 3)}`.padStart(13)).join(' ')}`); - const firstPlunge = e.map(r => { const row = r.rows.find(x => x.plunges > 0); return row ? row.a : NaN; }); - if (!polar) profiles.push(`${k}: first plunge at ${firstPlunge.map(x => (Number.isNaN(x) ? 'none' : f(x, 3))).join(', ')} r_J`); + const firstPlunge = e.map(r => r.rows.find(x => x.plunges > 0)); // undefined if none plunge, so a NaN radius still prints as NaN + if (!polar) profiles.push(`${k}: first plunge at ${firstPlunge.map(row => (row ? f(row.a, 3) : 'none')).join(', ')} r_J`); else profiles.push(`${k}: lost fraction at a = 0.1, 0.15, 0.2, 0.3, 0.35, 0.4, 0.45: ` + e.map(r => [0.1, 0.15, 0.2, 0.3, 0.35, 0.4, 0.45].map(a => { const x = r.rows.find(y => Math.abs(y.a - a) < 1e-9); return f(x.escapes + x.plunges, 2); }).join(' ')).join(' | ')); } - console.log(' (an orbit that starts in the plane stays in it, so the vertical tide cannot act on the first two rows; their plunges, if any, begin only'); - console.log(' outside the escape edge, so the two radii agree. For polar rows "lost" is the first radius where half are lost.)'); + console.log(' (an orbit that starts in the plane stays in it, so the vertical tide cannot act on the first two rows; with these 8 angles their'); + console.log(' plunges begin only outside the escape edge, so the two radii agree. For polar rows "lost" is the first radius where half are lost.)'); for (const p of profiles) console.log(` ${p} [point mass | Sun | Sun, nu2 = 1]`); { const across = [[1, 0, 0]], bigOnly = big; @@ -170,27 +170,51 @@ if (!quick) { { const gamE = 2 * (1 - EILERS.dv * EILERS.R0 / ve(EILERS.R0)), sunE = { ...sun, Gam: gamE }; console.log(` retrograde in the plane with Gam = ${f(gamE, 2)} (Eilers et al.'s slope): escaped by ${f(O.hillScan(sunE, [[0, 0, -1]], big, { nPhase: 8 }).edge, 3)} r_J`); + const s64 = (h, n) => O.hillScan(h, [n], grid(0.3, 1.3, 0.025), { nPhase: 64 }); + const re = s64(O.HOST_KEPLER, [0, 0, -1]), first = re.rows.find(r => r.plunges > 0); + console.log(` in the plane with 64 starting angles: point mass prograde ${f(s64(O.HOST_KEPLER, [0, 0, 1]).edge, 3)}, retrograde ${f(re.edge, 3)}` + + ` (lost ${f(re.edgeLost, 3)}, first plunge at ${first ? f(first.a, 3) : 'none'});`); + console.log(` the Sun retrograde ${f(s64(sun, [0, 0, -1]).edge, 3)}, and with Gam = ${f(gamE, 2)} ${f(s64(sunE, [0, 0, -1]).edge, 3)} r_J`); + } + { + // how far from polar plunges reach: orbit normals tilted by delta from the host's plane (i = 90 +- delta), 4 node + // azimuths x 8 starting angles each, at a few radii inside the escape edge + const fromPolar = (h, a, deltas) => deltas.map(d => { + const i = (90 - d) * Math.PI / 180, ns = []; + for (const s of [1, -1]) for (const om of [0, 0.5, 1, 1.5]) ns.push([Math.sin(i) * Math.cos(om * Math.PI), Math.sin(i) * Math.sin(om * Math.PI), s * Math.cos(i)]); + return O.hillFates(h, a, ns, { nPhase: 8 }).plunges; + }); + const dK = grid(0, 20, 2), dS = grid(0, 40, 4); + for (const a of [0.2, 0.3, 0.4]) console.log(` point mass, a = ${a} r_H: plunged fraction at ${dK[0]}-${dK[dK.length - 1]} deg from polar (step 2): ${fromPolar(O.HOST_KEPLER, a, dK).map(x => f(x, 2)).join(' ')}`); + for (const a of [0.2, 0.3]) console.log(` the Sun, a = ${a} r_J: plunged fraction at ${dS[0]}-${dS[dS.length - 1]} deg from polar (step 4): ${fromPolar(sun, a, dS).map(x => f(x, 2)).join(' ')}`); } const dirs = O.sphere(48), asIso = grid(0.2, 1.0, 0.05); console.log('\n orbits facing every way (48 directions x 4 starting angles, grid 0.05 r_J): fractions that stay, escape, and plunge within 1e-3 r_J'); const iso = {}; for (const [name, h] of [['point-mass host', O.HOST_KEPLER], ['the Sun in the Galaxy', sun]]) { - // the directions that plunge at 0.2 to 0.4 r_J, as degrees from polar (from the orbit's normal against the host's z axis) - let far = 0; - for (const nv of dirs) for (const a of [0.2, 0.3, 0.4]) if (O.hillFates(h, a, [nv], { nPhase: 4 }).plunges > 0) far = Math.max(far, Math.abs(90 - Math.acos(nv[2]) * 180 / Math.PI)); - console.log(` ${name}: plunges at 0.2-0.4 r_J come from directions within ${f(far, 0)} deg of polar`); iso[name] = O.hillScan(h, dirs, asIso, { nPhase: 4 }); console.log(` ${name}: half have escaped by ${f(iso[name].edge, 3)} r_J; half are lost (escaped or plunged) by ${f(iso[name].edgeLost, 3)} r_J`); for (const r of iso[name].rows) console.log(` a = ${f(r.a, 2)} r_J: stay ${f(r.stays, 2)}, escape ${f(r.escapes, 2)}, plunge ${f(r.plunges, 2)}`); } console.log(` for the Sun: half escaped by ${f(iso['the Sun in the Galaxy'].edge * rJ / C.au / 1e5, 2)}e5 au`); + { // the mirror pairs no starts within one of these directions, so check the second quarter turn gives the same fractions + const q = [['point mass', O.HOST_KEPLER, 0.5, iso['point-mass host']], ['the Sun', sun, 0.45, iso['the Sun in the Galaxy']]].map(([k, h, a, s]) => { + const r = s.rows.find(x => Math.abs(x.a - a) < 1e-9), r2 = O.hillFates(h, a, dirs, { nPhase: 4, quarter: 1 }); + return `${k} at ${a}: escaped ${f(r.escapes, 3)} -> ${f(r2.escapes, 3)}, lost ${f(r.escapes + r.plunges, 3)} -> ${f(r2.escapes + r2.plunges, 3)}`; + }); + console.log(` starting angles in [pi/2, pi) instead of [0, pi/2): ${q.join('; ')}`); + } const dirs24 = O.sphere(24); for (const rho of [0.08, 0.12]) { const e = O.hillScan(O.hostDisk(OORT.A, OORT.B, rho), dirs24, asIso, { nPhase: 4 }).edge; console.log(` with rho0 = ${rho} (24 directions): (nu/Omega)^2 = ${f(O.hostDisk(OORT.A, OORT.B, rho).nu2, 2)}, half escaped by ${f(e, 3)} r_J`); } - console.log(` the Sun at rho0 = 0.10 with 24 directions, for comparison: half escaped by ${f(O.hillScan(sun, dirs24, asIso, { nPhase: 4 }).edge, 3)} r_J`); - console.log(` with plunges counted inside 1e-2 r_J instead (24 directions): half escaped by ${f(O.hillScan(sun, dirs24, asIso, { nPhase: 4, rIn: 1e-2 }).edge, 3)} r_J`); + { + const base = O.hillScan(sun, dirs24, asIso, { nPhase: 4 }), r2 = O.hillScan(sun, dirs24, asIso, { nPhase: 4, rIn: 1e-2 }); + console.log(` the Sun at rho0 = 0.10 with 24 directions, for comparison: half escaped by ${f(base.edge, 3)}, half lost by ${f(base.edgeLost, 3)} r_J`); + console.log(` with plunges counted inside 1e-2 r_J instead (24 directions): half escaped by ${f(r2.edge, 3)}, half lost by ${f(r2.edgeLost, 3)} r_J;` + + ` largest plunged fraction ${f(Math.max(...r2.rows.map(x => x.plunges)), 2)} (against ${f(Math.max(...base.rows.map(x => x.plunges)), 2)})`); + } const oort = [6.5e4, 1e5, 2e5].map(x => x * C.au / rJ); console.log(` the Oort cloud's outer limit: 6.5e4 au (Correa-Otto & Calandra's a_crit), 1e5 au, 2e5 au (Dones et al. 2004) = ${oort.map(x => f(x, 2)).join(', ')} r_J`); if (long) { diff --git a/tests/orbits.mjs b/tests/orbits.mjs index 24a2c77..6c730c7 100644 --- a/tests/orbits.mjs +++ b/tests/orbits.mjs @@ -93,6 +93,14 @@ for (const x of [0.25, 1, 4]) { const g = O.galaxy(), g0 = O.galaxy(O.MWP14, { bh: 0 }); near('MWPotential2014 normalisation identity: v_c(8 kpc) without Sgr A* [km/s] (G and the shapes cancel)', g0.v(8), 220, 1e-9, 'model-internal (parameters as in galpy\'s source)'); +{ + const T1 = 'Bovy 2015, Table 1 (arXiv source, mwpotential2014.txt), as #37\'s re-review read it; tolerance the rounding of the printed value'; + near('MWPotential2014: bulge mass [1e10 M_sun]', g0.M.bulge / 1e10, 0.5, 0.05, T1); + near('MWPotential2014: disk mass [1e10 M_sun]', g0.M.disk / 1e10, 6.8, 0.05, T1); + near('MWPotential2014: mass inside 60 kpc, as v^2 R/G [1e11 M_sun]', g0.v(60) ** 2 * 60 / C.G / 1e11, 4.1, 0.05, `${T1} (the true sphere mass is 4.05)`); + near('MWPotential2014: dln v_c/dln R at R0 = 8 kpc', g0.slope(8), -0.10, 0.005, T1); + near('MWPotential2014: dark-matter density at R0 [M_sun/pc^3]', g0.haloDensity(8), 0.008, 0.0005, T1); +} near('MWPotential2014: halo mass inside 245 kpc [1e12 M_sun]', g0.parts(245).halo * 245 / C.G / 1e12, 0.8, 0.05, 'Bovy 2015, Table 1: M_vir = 0.8e12 M_sun with r_vir = 245 kpc (#37\'s review); tolerance the rounding of 0.8'); near('The disk\'s mass inside a sphere of 1e4 kpc / its total (Gauss\'s theorem by quadrature)', g.diskSphereMass(1e4) / g.M.disk, 1, 1e-3, @@ -127,7 +135,7 @@ near('Saturn\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O const esc = (h, a, n) => O.hillFates(h, a, [n], { nPhase: 8 }).escapes, pro = [esc(K, 0.4, [0, 0, 1]), esc(K, 0.6, [0, 0, 1])]; check('Point-mass host: prograde circular orbits mostly kept at 0.4 r_H, mostly lost by 0.6', pro[0] < 0.5 && pro[1] >= 0.5, `fraction escaped ${pro[0].toFixed(2)} at 0.4 r_H (must be < 0.5), ${pro[1].toFixed(2)} at 0.6 r_H (must be >= 0.5)`, 'model'); - const edgePro = O.hillScan(K, [[0, 0, 1]], [0.3, 0.35, 0.4, 0.45, 0.5, 0.55, 0.6, 0.65, 0.7], { nPhase: 8 }).edge; + const edgePro = O.hillScan(K, [[0, 0, 1]], [0.25, 0.3, 0.35, 0.4, 0.45, 0.5, 0.55, 0.6, 0.65, 0.7], { nPhase: 8 }).edge; near('Point-mass host: radius where half the prograde circular orbits in the plane have escaped [r_H]', edgePro, 0.4, 0.1, `${NV}; tolerance covers their different criterion and integration time`); const gRet = []; for (let a = 0.6; a <= 1.0001; a += 0.025) gRet.push(+a.toFixed(3)); @@ -154,7 +162,8 @@ near('Saturn\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O 'model (Lidov 1962, Kozai 1962: the eccentricity reaches 1 at 90 deg)'); const L = O.hillScan(K, [[1, 0, 0]], [0.1, 0.2], { nPhase: 8 }); check('Point-mass host: polar orbits (plane across the host) are lost by plunging between 0.1 and 0.2 r_H, before any escapes', - Number.isNaN(L.edge) && L.edgeLost > 0.1 && L.edgeLost < 0.2, `half lost by ${fmt(L.edgeLost)} r_H (must be 0.1 to 0.2); half escaped: ${L.edge} (must be none)`, + L.rows.every(r => r.escapes === 0) && L.edgeLost > 0.1 && L.edgeLost < 0.2, + `half lost by ${fmt(L.edgeLost)} r_H (must be 0.1 to 0.2); escaped fractions ${L.rows.map(r => r.escapes.toFixed(2)).join(', ')} (must all be 0)`, 'model (the edgeLost path)'); } const Om = 27.2 * C.kmskpc, want = (4 * Math.PI * C.GMsun * 0.10 / C.pc ** 3 + 2 * (15.3 ** 2 - 11.9 ** 2) * C.kmskpc ** 2) / Om ** 2; @@ -167,7 +176,8 @@ near('Saturn\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O const rJ = O.jacobi(C.GMsun, 15.3, -11.9), edge = O.hillScan(sun, dirs, [0.2, 0.3, 0.4, 0.5, 0.6], { nPhase: 2 }).edge * rJ / C.au; check('The Sun\'s all-directions escape edge lies within the Oort cloud\'s quoted outer limits, 6.5e4 to 2e5 au', edge >= 6.5e4 && edge <= 2e5, `half have escaped by ${fmt(edge)} au (12 directions x 2 angles, grid 0.1 r_J from 0.2 r_J = 5.7e4 au; the tide alone, no passing stars)`, - 'Correa-Otto & Calandra 2019 sect. 4.3: a_crit ~ 6.5e4 au, set by passing stars (#37\'s review); Dones et al. 2004: as far as 2e5 au, as cited in search text'); + 'Correa-Otto & Calandra 2019 sect. 4.3: a_crit ~ 6.5e4 au, set by the frequency of stellar passages acting with the tide (#37\'s review); ' + + 'Dones et al. 2004: as far as 2e5 au, as cited in search text; a wide window, which the edge passes for Gam from 0.6 to 1.5 times its value'); } console.log(fails ? `\n${fails} check(s) failed` : '\nall orbits checks pass'); From 80438b769ef79cf2f493bdcae8e80b414dc3fdef Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 27 Oct 2026 10:17:10 +0000 Subject: [PATCH 07/10] Orbits study: fixes from the verification of 99ec2bc Sampling (blocking): the quarter turn [0, pi/2) assumed the start at pi - phi duplicates the start at phi. It is that start's time-reversed mirror, whose fate matches nearly always in the plane but differs for 7-17% of polar pairs, so polar edges depended on which quarter was sampled (the Sun's polar-across edge moved 0.39 -> 0.43). Starting angles now span the half turn [0, pi), a quarter step off the grid pi p/n, so no inverted or mirrored pair falls in one sample. Section 5 re-run; the table prints the polar rows with 16 angles too (they move by at most 0.01) and the README carries the new numbers. Chaos (blocking): "the in-plane edges are not [chaotic]" was wrong: changing Gam by 1e-4 moves the Sun's retrograde edge 0.925 -> 0.95. The table prints that line; the caveat now says single orbits near any edge are chaotic. Also: - 64-angle lost edges and first plunges printed for every in-plane scan (with Gam = 2.12 the lost radius is 1.01 against 1.03 escaped). - The tilt-from-polar scan used node azimuths om and om + pi, which are the same orbits under z -> -z; it now uses 0 and 90 deg with 16 angles. - The symmetry text names z -> -z and the inversion for y-z planes. - Thin disks: an exponential scale height and rms sin i differ by a factor of about 2, not sqrt2; as rms z/r the disk is 0.052 against the planets' 0.032, printed; table.mjs no longer says "like-for-like". - The averaged-tide cone: 5.1 deg at 0.2 r_J to 3.6 deg at 0.4 r_J, and the Sun's whole cone beyond it, not only near the escape edge, is non-secular. - The polar check accepts half lost exactly at 0.2; the Oort check's label states its measured insensitivity to Gam; "#37's re-review" is defined; the Kuzmin limit pins a; the bulge's slope and cut-off are untested; Alexander's m is the black hole's mass; "Galactic center" as A&A prints it; the step-cap comment's figure; runtimes measured on an idle machine (about 10 minutes full, 13 long, 6 s for the checks). Co-Authored-By: Claude Opus 5.5 Claude-Session: https://claude.ai/code/session_01C2bpiHyHSkC8DAL5CcyciK --- README.md | 2 +- orbits/README.md | 69 ++++++++++++++++++++++++++---------------------- orbits/model.js | 18 +++++++------ orbits/table.mjs | 44 +++++++++++++++--------------- tests/orbits.mjs | 6 ++--- 5 files changed, 74 insertions(+), 65 deletions(-) diff --git a/README.md b/README.md index 31e0153..609b234 100644 --- a/README.md +++ b/README.md @@ -216,7 +216,7 @@ node orbits/table.mjs quick # the orbits study's numbers, without i below). The third study's checks, `tests/critical.mjs`, add about 65 s of seeded Monte Carlo runs, most of it percolation with nine seeds. CI runs every line above except the `critical/table.mjs` line: it repeats the tests' seeded runs and prints the results without checking them. The orbits study's full table (`node orbits/table.mjs`, -about 7 minutes) is not run by CI either. +about 10 minutes) is not run by CI either. ## Checks diff --git a/orbits/README.md b/orbits/README.md index 9528fb0..2ce1d58 100644 --- a/orbits/README.md +++ b/orbits/README.md @@ -16,7 +16,7 @@ what ends it, and what, if anything, repeats from one scale to the next. | How exact is Kepler's law in the Solar System? | In JPL's 1800–2050 fit, n²a³ matches GM☉ to within 1.3×10⁻³ for all eight planets. That is about the planets' total mass, 1.34×10⁻³ of the Sun's. Jupiter's excess is mostly its own mass in the two-body law; Uranus's and Neptune's are mostly the other planets' pull, and Saturn's fitted 2.2×10⁻⁴ is below its own 2.9×10⁻⁴ (see below). P ∝ a^1.49986 over 0.39 to 30 au. | | Does the Galaxy have a Kepler region? | Yes, around Sgr A*. The star S2 orbits it as a planet orbits the Sun, plus general relativity's precession: 12.2′ per orbit from their elements (they quote 12.1′). The speed's log slope passes −1/4 about 1.3 pc out (1.0 to 1.8 pc across the cluster-mass range; 0.7 to 1.9 pc for thresholds from −0.4 to −0.1). The stars inside weigh as much as the black hole by 1.8 to 3.2 pc. | | What ends a Kepler region? | For a body orbiting in a host (a planet, the Sun), the host's tide. Satellites are held inside its tidal radius, r_J³ = Gm/(4A(A − B)), apart from quasi-satellite orbits beyond about 2 r_H that Shen and Tremaine (2008) find around Jupiter, Uranus and Neptune. Inside r_J, the body's mean density is 4A/Ω times the host's mean density inside the orbit (host taken as spherical): 3 for a planet around the Sun, 2.25 for the Sun in the Galaxy. For the Sun, r_J = 1.37 pc = 2.8×10⁵ au. At the Galaxy's centre, where Sgr A* sits still, the region ends instead where the surrounding stars' mass approaches the black hole's. | -| How much of that region is used? | About half, for orbits that do not plunge, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.52 r_J around a planet and by 0.433 r_J (1.23×10⁵ au) for the Sun. Polar orbits are lost much sooner, to Lidov–Kozai cycles: half are lost by 0.14 r_H around a planet (the lost fraction dips below half near 0.4 r_H) and by 0.06 r_J for the Sun, in 20 turns of the host. For near-circular moons near the ecliptic, Nesvorný et al. (2003) put the limits at about 0.4 and 0.7 r_H (prograde, retrograde), and the model gives 0.46 and 0.78. The Oort cloud's outer limit is placed between 6.5×10⁴ au (Correa-Otto and Calandra 2019, where the frequency of stellar passages, acting with the tide, sets it) and 2×10⁵ au, which is 0.23 to 0.71 r_J. | +| How much of that region is used? | About half, for orbits that do not plunge, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.52 r_J around a planet and by 0.436 r_J (1.23×10⁵ au) for the Sun. Polar orbits are lost much sooner, to Lidov–Kozai cycles: half are lost by 0.14 r_H around a planet (half or more out to 0.35 r_H, fewer than half at 0.45 r_H) and by 0.06 r_J for the Sun, in 20 turns of the host. For near-circular moons near the ecliptic, Nesvorný et al. (2003) put the limits at about 0.4 and 0.7 r_H (prograde, retrograde), and the model gives 0.46 and 0.78. The Oort cloud's outer limit is placed between 6.5×10⁴ au (Correa-Otto and Calandra 2019, where the frequency of stellar passages, acting with the tide, sets it) and 2×10⁵ au, which is 0.23 to 0.71 r_J. | | What does repeat? | Kepler regions nested inside one another: moons in a planet's, planets and comets in the Sun's. Each is bounded by its host's tide at a similar fraction of the tidal radius, though for the Oort cloud passing stars cut it shorter. Also flat disks: the planets' orbits are tilted by about 2° on average (6° for Mercury), and the Galaxy's thin disk has a scale height 0.04 of the Sun's orbital radius. | ## Kepler's law in the Solar System @@ -91,7 +91,7 @@ for Jupiter it is 237 years. - **Escaped:** once it passes 3 r_J. - **Plunged:** once it comes within 10⁻³ r_J, which is 0.74 of Jupiter's radius or, for the Sun, 283 au, outside the planets. - **Lost:** either. -- **Starting angles.** Hill's equations are symmetric under r → −r, v → −v, so a start half a turn on is the inverted orbit, with the same fate. They are also symmetric under time reversal with y → −y. For every single-plane row here, that makes the start at π − φ the mirror image of the start at φ with time running backward, and the two fates nearly always match. So the starting angles are spread over a quarter turn, where no two are related by either symmetry. In the all-directions sample the mirror is a different direction. There, starting angles in the second quarter turn give the same fractions to within about 0.03 (section 5). +- **Starting angles.** Hill's equations are symmetric under r → −r, v → −v, so a start half a turn on is the inverted orbit, with the same fate; the starting angles therefore span a half turn. The equations are also symmetric under z → −z and under time reversal with y → −y. For every single-plane row here, these make the start at π − φ the mirror image of the start at φ with time running backward (inverted, for a plane through the host's direction). In the plane the two fates nearly always match; for polar orbits they often differ, so neither is a duplicate. The angles sit a quarter step off the grid πp/8, so that no such pair falls in one sample. The radius by which half the orbits have escaped, and by which half are lost (`node orbits/table.mjs`, section 5; 8 starting angles; grid 0.025 r_J; all directions on 48 directions × 4 angles and a 0.05 grid): @@ -100,45 +100,48 @@ The radius by which half the orbits have escaped, and by which half are lost (`n |---|---|---|---| | Prograde, in the plane | 0.46 | 0.49 | 0.49 | | Retrograde, in the plane | 0.775 | 0.925 | 0.925 | -| Polar, plane through the host's direction: escaped / lost | 0.47 / 0.14 | 0.40 / 0.06 | 0.475 / 0.17 | -| Polar, plane across it: escaped / lost | 0.49 / 0.14 | 0.39 / 0.06 | 0.49 / 0.16 | -| Facing every way: escaped / lost | 0.52 / 0.52 | 0.433 / 0.42 | | +| Polar, plane through the host's direction: escaped / lost | 0.47 / 0.14 | 0.41 / 0.06 | 0.475 / 0.16 | +| Polar, plane across it: escaped / lost | 0.475 / 0.145 | 0.40 / 0.06 | 0.49 / 0.17 | +| Facing every way: escaped / lost | 0.519 / 0.515 | 0.436 / 0.421 | | **The table's resolution.** With 64 starting angles instead of 8, the in-plane edges are 0.46 and 0.79 around a -point mass and 0.95 for the Sun's retrograde orbits. So the 8-angle values carry sampling errors of up to about 0.03. +point mass, 0.49 and 0.95 for the Sun, and 1.03 for the Sun's retrograde orbits with Γ = 2.12. So the in-plane +8-angle values carry sampling errors of up to about 0.03. With 16 angles the polar edges move by at most 0.01. -**In-plane orbits.** These plunge only rarely: with 8 angles, first at 0.55 r_J prograde and at 0.85 and 1.1 r_J -retrograde, all outside their escape edge. With 64, the point mass's first retrograde plunge is at 0.775, just inside -its 0.788 edge, and its lost radius is 0.783. So for in-plane orbits the escaped and lost radii agree to about 0.005. +**In-plane orbits.** These plunge only rarely: with 8 angles, first at 0.55 to 0.575 r_J prograde and at 1.1 r_J +retrograde for the Sun, all outside their escape edge, and not at all retrograde around a point mass (to 1.2 r_J). +With 64 angles the escaped and lost radii agree to about 0.005 (0.788 against 0.783 retrograde around a point mass), +except with Γ = 2.12, where the lost radius is 1.01 against 1.03 escaped; there the first plunge, at 0.975, falls +inside the escape edge. **Polar rows.** "Lost" is the first radius where half are lost. **Polar orbits are lost by plunging.** - **The mechanism.** A circular orbit at 90° to the host's plane goes through Lidov–Kozai cycles (Lidov 1962, Kozai 1962), whose eccentricity reaches 1. So it dives into the body long before it can escape. - **When.** Within 20 turns half are lost from 0.14 r_H around a planet and from 0.06 r_J for the Sun. Further in, the cycle takes longer than 20 turns; it is not absent. -- **Not everywhere outside.** Around a point mass the lost fraction is 75–100% at 0.15–0.2 r_H and dips below half near 0.4 r_H (0–25% at 0.4), where most polar orbits survive the run. With ν²/Ω² = 1 the Sun's lost fraction also dips below half, from 0.35 r_J in the plane across the host's direction and at 0.45 r_J in the plane through it. With the Galaxy's vertical tide the Sun loses half or more of its polar orbits at every radius printed, 0.1 to 0.45 r_J. -- **Other directions.** For a planet, plunges reach to about 8°, 10° and 14° from polar at 0.2, 0.3 and 0.4 r_H. Tilted to 75° (15° from polar), none plunge from 0.1 to 0.4 r_H. For the Sun they reach to about 8° from polar at 0.2 r_J and 24° at 0.3 r_J. At 75°, none plunge inside 0.3 r_J, and up to 63% do at 0.3–0.4 r_J. -- **All directions.** Facing every way, up to a fifth of the Sun's orbits plunge within 20 turns, and 8–12% of a planet's between 0.2 and 0.4 r_H. -- **What the averaged tide explains.** For comets, the tide driving them inward (Heisler and Tremaine 1986) accounts for the near-polar core. The Sun's wider cone near its escape edge, and its plunges at 75°, are not secular: #37's re-review found that the averaged tide gives a cone only 3.6 to 5.1° wide at 0.2 to 0.4 r_J, and no plunges at 75°. These plunges depend steeply on the plunge radius: counted inside 10⁻² r_J, the largest plunged fraction over all directions rises from 0.24 to 0.41 (24 directions). +- **Not everywhere outside.** Around a point mass the lost fraction is 63–100% at 0.15–0.2 r_H, half or more out to 0.35 r_H, and below half at 0.45 r_H (13–38%), where most polar orbits survive the run. With ν²/Ω² = 1 the Sun's lost fraction falls to half or less from 0.35 r_J. With the Galaxy's vertical tide the Sun loses half or more of its polar orbits at every radius printed, 0.1 to 0.45 r_J. +- **Other directions.** For a planet, plunges reach to about 8°, 10° and 14° from polar at 0.2, 0.3 and 0.4 r_H. Tilted to 75° (15° from polar), none plunge from 0.1 to 0.4 r_H. For the Sun they reach to about 8° from polar at 0.2 r_J and 24° at 0.3 r_J. At 75°, none plunge out to 0.3 r_J, and up to 63% do at 0.35–0.4 r_J. +- **All directions.** Facing every way, up to 23% of the Sun's orbits plunge within 20 turns, and 9–11% of a planet's between 0.2 and 0.4 r_H. +- **What the averaged tide explains.** For comets, the tide driving them inward (Heisler and Tremaine 1986) accounts for the near-polar core: at 0.2 r_J all the Sun's orbits within 4° of polar plunge. The rest of its cone (to 8° at 0.2 r_J and 24° at 0.3 r_J) and its plunges at 75° are not secular: #37's re-review found that the averaged tide limits plunges to within 5.1° of polar at 0.2 r_J and 3.6° at 0.4 r_J, and gives none at 75°. These plunges depend steeply on the plunge radius: counted inside 10⁻² r_J, the largest plunged fraction over all directions rises from 0.27 to 0.43 (24 directions). **For a planet.** - **Against the published limits.** The model's in-plane edges are 0.46 (prograde) and 0.775 to 0.79 (retrograde), against Nesvorný et al.'s (2003) semi-major axes of about 0.4 and 0.7 r_H for nearly circular orbits near the ecliptic, as Shen and Tremaine (2008) quote them. - **Over all inclinations.** Shen and Tremaine find that many prograde and retrograde satellites survive to about 0.5 and 0.7 r_H, and some coplanar retrograde ones to about r_H; their extents are instantaneous distances, not semi-major axes. The observed irregular moons lie at about 0.05 to 0.6 r_H, the retrograde ones further out (about 0.6 against 0.4 r_H). - **Run length.** The runs are short: 20 turns is 237 years for Jupiter. - - Followed for 100 turns, the all-directions value moves from 0.52 to 0.51 (24 directions, `node orbits/table.mjs long`). + - Followed for 100 turns, the all-directions value moves from 0.52 to 0.50 (24 directions, `node orbits/table.mjs long`). - In a 48-orbit sample run to 3000 turns, #37's review found the point-mass edge drifting from 0.50 to 0.475. **For the Sun.** -- **The edge.** Half of the orbits facing every way have escaped by 0.433 r_J, or 1.23×10⁵ au. - - For ρ₀ from 0.08 to 0.12 it moves from 0.446 to 0.431 (on 24 directions, which give 0.441 at 0.10). - - Counting plunges inside 10⁻² r_J instead of 10⁻³ moves the 24-direction escaped edge out, from 0.441 to 0.458, but the lost edge in, from 0.410 to 0.395. - - Followed for 100 turns (23 billion years, a convergence test only) it is 0.42, and 0.38 counting plunges, against 0.44 and 0.41 over 20 turns (24 directions). +- **The edge.** Half of the orbits facing every way have escaped by 0.436 r_J, or 1.23×10⁵ au. + - For ρ₀ from 0.08 to 0.12 it moves from 0.446 to 0.427 (on 24 directions, which give 0.437 at 0.10). + - Counting plunges inside 10⁻² r_J instead of 10⁻³ moves the 24-direction escaped edge out, from 0.437 to 0.460, but the lost edge in, from 0.411 to 0.394. + - Followed for 100 turns (23 billion years, a convergence test only) it is 0.41, and 0.38 counting plunges, against 0.44 and 0.41 over 20 turns (24 directions). - **The Oort cloud.** Correa-Otto and Calandra (2019, sect. 4.3) follow comets by their initial semi-major axis. They find fewer than 5% still bound beyond 6.5×10⁴ au and none at 10⁵ au. - They define a_crit ≈ 6.5×10⁴ au as the cloud's outer limit. Half are bound at about 2×10⁴ au, a mean over three samples. - They describe a joint ("synergy") effect of the Galactic potential and stellar passages, and name the frequency of stellar passages as the main modulator of that limit. - Dones et al. (2004, as cited) place the edge as far as 2×10⁵ au. So the quoted outer limits run from 0.23 to 0.71 r_J. - - The model's 0.433 r_J lies past Correa-Otto and Calandra's: the tide alone would keep comets further out than the tide and passing stars together allow. -- **The vertical tide.** It loses polar orbits sooner: by escape at 0.40 and 0.39 r_J, against 0.475 and 0.49 with ν²/Ω² = 1 at the same Γ, and by plunging at 0.06 against 0.16 to 0.17. It never acts on an orbit that starts in the plane, because such an orbit stays there. + - The model's 0.436 r_J lies past Correa-Otto and Calandra's: the tide alone would keep comets further out than the tide and passing stars together allow. +- **The vertical tide.** It loses polar orbits sooner: by escape at 0.41 and 0.40 r_J, against 0.475 and 0.49 with ν²/Ω² = 1 at the same Γ, and by plunging at 0.06 against 0.16 to 0.17. It never acts on an orbit that starts in the plane, because such an orbit stays there. - **In-plane retrograde orbits.** These last to 0.925 r_J (0.95 with 64 angles) rather than 0.775 because Γ is 2.25 rather than 3: the flatter rotation curve weakens the in-plane tide. With Γ = 2.12 from Eilers et al.'s slope instead (see Across the disk), the edge is about 1.0 r_J: 1.00 with 8 angles and 1.03 with 64. ## A Kepler region at the Galaxy's centre @@ -162,7 +165,7 @@ those values are lower limits. With γ = 1.4 the middle "twice" radius is 3.8 pc The literature puts the "twice" radius, the usual influence radius, at 3 to 4 pc: - Schödel et al. 2018, sect. 4.5: 7.8×10⁶ M☉ within 3 pc, "roughly twice the mass of Sgr A*". -- Alexander 2005: r_h ∼ 3 pc, estimated from the stars' mass reaching m by ≲2 pc and 2m by ≲4 pc (he calls Gm/σ² customary but not well defined). +- Alexander 2005: r_h ∼ 3 pc, estimated from the stars' mass reaching the black hole's by ≲2 pc and twice it by ≲4 pc (he calls Gm/σ² customary but not well defined). - Feldmeier-Krause et al. 2017: about 4 pc (4.0 +2.2/−1.1), with a black-hole mass of 4×10⁶ M☉ and Merritt's (2004) definition. - GRAVITY Collaboration 2024: 3.9 pc in their cusp simulation, with Merritt's (2013) definition. @@ -227,7 +230,9 @@ The rms of sin i is 0.045, or 0.023 without Mercury. The Galaxy's thin disk has an exponential scale height of 300 ± 50 pc over a scale length of 2.6 ± 0.5 kpc (Bland-Hawthorn and Gerhard 2016). Their ratio, 0.12, is the axis ratio of its density contours. Height over orbital radius is 300 pc over the Sun's 8.2 kpc, 0.037, against the planets' rms sin i of 0.045: both a few percent. -The two measures differ in convention, an exponential scale height against an rms, by about √2. +The two measures differ by a factor of about 2 in convention. An exponential scale height h has rms z = √2 h, and an +orbit tilted by i has rms z/r = sin i/√2. Compared as rms height over radius, the thin disk is 0.052 and the planets +0.032. Both are flat for the same reason: gas that radiates its energy but keeps its angular momentum settles into a disk. The Galaxy's present thickness is also set by the stars' later heating. What the disks orbit differs: one dominant @@ -236,10 +241,10 @@ mass, or mass spread through the disk and halo. ## Run it ``` -node orbits/table.mjs # every computed number above except the 100-turn runs and the values credited to #37's review (about 7 minutes) +node orbits/table.mjs # every computed number above except the 100-turn runs and the values credited to #37's review (about 10 minutes) node orbits/table.mjs quick # sections 1-4 only, well under a second (run by CI) -node orbits/table.mjs long # adds the 100-turn runs, about 10 minutes in all -node tests/orbits.mjs # the checks, about 3 s (also run by npm test) +node orbits/table.mjs long # adds the 100-turn runs, about 13 minutes in all +node tests/orbits.mjs # the checks, about 6 s (also run by npm test) ``` The table exits with an error if any printed value is not a finite number. @@ -247,7 +252,8 @@ The table exits with an error if any printed value is not a finite number. ## Checks `tests/orbits.mjs` has 43 checks. Each prints its source, the published or expected value, the model's value and -the tolerance. "#37's review" marks values a reviewer read in the paper's arXiv source. +the tolerance. "#37's review" marks values a reviewer read in the paper's arXiv source; "#37's re-review" marks +what the second review checked, in an arXiv source or against Crossref, as each label says. | What | Against | Kind | |---|---|---| @@ -278,16 +284,17 @@ the tolerance. "#37's review" marks values a reviewer read in the paper's arXiv - for moons, the planet's oblateness and its other moons, which damp Lidov–Kozai cycles close in. - **Circular starting orbits only.** Real moons and comets start on eccentric orbits. - **The edges' definition.** They are set at half the orbits escaping, or lost, within 20 turns of the host, on a grid of 0.025 r_J (single planes) or 0.05 r_J (all directions). - - **Plunge radius.** It is 10⁻³ r_J. Raising it to 10⁻² moves the Sun's all-directions escaped edge out, from 0.441 to 0.458, and its lost edge in, from 0.410 to 0.395 (24 directions). - - **Chaos and sampling.** The polar edges are chaotic: changing ν²/Ω² by 10⁻⁴ moves the Sun's polar-across escape edge from 0.388 to 0.392. The in-plane edges are not, but with 8 starting angles they carry sampling errors of up to about 0.03 (see the 64-angle values above). + - **Plunge radius.** It is 10⁻³ r_J. Raising it to 10⁻² moves the Sun's all-directions escaped edge out, from 0.437 to 0.460, and its lost edge in, from 0.411 to 0.394 (24 directions). + - **Chaos and sampling.** Single orbits near an edge are chaotic: changing Γ by 10⁻⁴ either way moves the Sun's in-plane retrograde edge from 0.925 to 0.95, as one orbit of eight changes its fate. (Changing ν²/Ω² by 10⁻⁴ leaves the Sun's polar-across edge at 0.40 in this sample.) With 8 starting angles the in-plane edges carry sampling errors of up to about 0.03 and the polar edges up to about 0.01 (see the 64- and 16-angle values above). - **Longer runs** move the edges inward a little (see the 100-turn values above). -- **MWPotential2014 is one model among many.** Its halo sets the flat range, and it has no nuclear star cluster. The checks pin its halo, its Plummer limit pins the disk's b, and Table 1's derived values (bulge and disk masses, the mass inside 60 kpc, the slope at R₀ and the local dark-matter density) are checked at Table 1's rounding. The other shape parameters are taken from galpy's source and tested only through those totals. The Galactic-centre numbers use the separate power-law cusp above, which is valid only inside its 3 pc break. +- **MWPotential2014 is one model among many.** Its halo sets the flat range, and it has no nuclear star cluster. The checks pin its halo, its Kuzmin and Plummer limits pin the disk's a and b to galpy's values, and Table 1's derived values (bulge and disk masses, the mass inside 60 kpc, the slope at R₀ and the local dark-matter density) are checked at Table 1's rounding. The bulge's slope and cut-off are taken from galpy's source and not tested: its mass follows from its 5% share of the force at R₀. The Galactic-centre numbers use the separate power-law cusp above, which is valid only inside its 3 pc break. - **Spherical mass estimates.** v²R/G and 3Ω²/4πG treat a flattened galaxy as spherical. In MWPotential2014 this overstates the mass inside 8 kpc by 12%. It shifts the 3.7-fold growth to 4.0 and the Sun's 2.25 to 2.53, but not the contrast with Kepler's fixed mass. ## Sources Each entry says how it was read. "#37's review" means a reviewer checked the values used here in the arXiv source, -or in the primary file, during #37's review. +or in the primary file, during #37's review. "#37's re-review" marks what the second review checked: values in an +arXiv source, or a reference matched against Crossref where the entry says so. - W. M. Folkner et al., *The Planetary and Lunar Ephemerides DE430 and DE431*, IPN Progress Report 42-196 (2014), Table 8. #37's review. - E. M. Standish and J. G. Williams, *Approximate Positions of the Planets*, JPL Solar System Dynamics (credited there to Standish and Williams 1992), Table 1 (1800–2050). Checked against the SSD page in #37's review. @@ -306,7 +313,7 @@ or in the primary file, during #37's review. - L. Dones, P. R. Weissman, H. F. Levison and M. J. Duncan, *Oort cloud formation and dynamics*, in *Comets II*, 153–174 (2004). The reference matches Crossref (#37's re-review); the text was not read. The edge "as far as 2 × 10^5 AU" is as cited in search text. - GRAVITY Collaboration, *Detection of the Schwarzschild precession in the orbit of the star S2 near the Galactic centre massive black hole*, A&A 636, L5 (2020). https://doi.org/10.1051/0004-6361/202037813. #37's review (sect. 1.2, appendix B, Table E.1). - GRAVITY Collaboration, *Mass distribution in the Galactic Center based on interferometric astrometry of multiple stellar orbits*, A&A 657, L12 (2022). https://doi.org/10.1051/0004-6361/202142465. #37's review. -- GRAVITY Collaboration, *Improving constraints on the extended mass distribution in the Galactic Center with stellar orbits*, A&A 692, A242 (2024). https://doi.org/10.1051/0004-6361/202452274. #37's review (sect. 4.3). +- GRAVITY Collaboration, *Improving constraints on the extended mass distribution in the Galactic center with stellar orbits*, A&A 692, A242 (2024). https://doi.org/10.1051/0004-6361/202452274. #37's review (sect. 4.3). - R. Schödel, D. Merritt and A. Eckart, A&A 502, 91 (2009). https://doi.org/10.1051/0004-6361/200810922. #37's review. - R. Schödel et al., A&A 609, A27 (2018). https://doi.org/10.1051/0004-6361/201730452. #37's review (sect. 4.5). - T. Alexander, *Stellar processes near the massive black hole in the Galactic center*, Phys. Rep. 419, 65 (2005). https://doi.org/10.1016/j.physrep.2005.08.002. #37's review (sect. 3.1.2). diff --git a/orbits/model.js b/orbits/model.js index 5e0bf5c..c4efc05 100644 --- a/orbits/model.js +++ b/orbits/model.js @@ -287,7 +287,8 @@ // Follows one orbit for `orbits` turns of the host (time 2 pi orbits). Fate: 'escapes' once r > rEsc, 'plunges' // once r < rIn (1e-3 r_J is 0.74 of Jupiter's radius; for the Sun it is 283 au, outside the planets), otherwise // 'stays'. Returns the final state s too. Throws on non-finite settings, on a non-finite step, or after maxSteps - // (by default 2.5e5 per turn of the host; the hardest orbit in the tables needs about 1.5e4), rather than looping. + // (by default 2.5e5 per turn of the host; the hardest orbit in the tables, a polar orbit at 0.025 r_J that stays, + // needs about 7e4 per turn), rather than looping. function hillOrbit(host, s0, { orbits = 20, rEsc = 3, rIn = 1e-3, tol = 1e-10, maxSteps = 2.5e5 * orbits } = {}) { if (![orbits, rEsc, rIn, tol, maxSteps, host.Gam, host.nu2, ...s0].every(Number.isFinite)) throw new Error('hillOrbit: non-finite input'); const T = 2 * Math.PI * orbits, k = Array.from({ length: 7 }, () => new Float64Array(6)); @@ -339,14 +340,15 @@ } // Fractions of orbits that stay, escape and plunge, over the given orbit normals and nPhase starting angles each. // Hill's equations are symmetric under r -> -r, v -> -v: a start at phi + pi is the inverted start at phi, with the - // same fate. They are also symmetric under time reversal with y -> -y. For a normal in the x-z or y-z plane, as in - // every single-plane scan here, that makes a start at pi - phi the mirror of the start at phi run backward, whose - // fate nearly always matches. So the starting angles are spread over the quarter turn [0, pi/2) (quarter = 0), where - // no two are related by either symmetry; quarter = 1 takes [pi/2, pi) instead. For other normals the mirror is - // another normal, and [0, pi/2) samples half of each normal's distinct starts. - function hillFates(host, a, normals, { nPhase = 4, quarter = 0, ...opt } = {}) { + // same fate. So the starting angles span the half turn [0, pi). The equations are also symmetric under z -> -z, and + // under time reversal with y -> -y. For a normal in the x-z plane (or, with z -> -z as well and an inversion, the + // y-z plane), as in every single-plane scan here, the start at pi - phi is then the mirror of the start at phi run + // backward. Its fate matches nearly always in the plane but often not for polar orbits, so it is no duplicate there; + // still, the angles phi = pi (p + offset)/nPhase are offset by a quarter step so that no such pair falls in one + // sample. offset = 0.75 takes the mirrored set pi - phi instead. + function hillFates(host, a, normals, { nPhase = 4, offset = 0.25, ...opt } = {}) { const n = { stays: 0, escapes: 0, plunges: 0 }; - for (const nv of normals) for (let p = 0; p < nPhase; p++) n[hillOrbit(host, hillStart(host, a, nv, Math.PI / 2 * (quarter + (p + 0.5) / nPhase)), opt).fate]++; + for (const nv of normals) for (let p = 0; p < nPhase; p++) n[hillOrbit(host, hillStart(host, a, nv, Math.PI * (p + offset) / nPhase), opt).fate]++; const tot = normals.length * nPhase; return { stays: n.stays / tot, escapes: n.escapes / tot, plunges: n.plunges / tot }; } diff --git a/orbits/table.mjs b/orbits/table.mjs index 759835b..6ab8ed8 100644 --- a/orbits/table.mjs +++ b/orbits/table.mjs @@ -1,4 +1,4 @@ -// Prints every computed number in orbits/README.md, except the values the README credits to #37's review. Run: node orbits/table.mjs (a few minutes; the stability scans of +// Prints every computed number in orbits/README.md, except the values the README credits to #37's review. Run: node orbits/table.mjs (about 10 minutes; the stability scans of // section 5 take nearly all of it), node orbits/table.mjs quick (well under a second; section 5 left out), or // node orbits/table.mjs long (adds the all-directions scans followed for 100 turns of the host). // Exits with code 1 if any printed line holds NaN or Infinity. @@ -124,7 +124,9 @@ const tilts = O.PLANETS.map(p => O.tiltToInvariable(p) * Math.PI / 180); console.log(` mean inclination over the eight: ${f(O.PLANETS.reduce((a, p) => a + O.tiltToInvariable(p), 0) / 8, 2)} deg`); console.log(` rms sin(i) over the eight: ${f(Math.sqrt(tilts.reduce((a, t) => a + Math.sin(t) ** 2, 0) / 8), 3)}; without Mercury: ${f(Math.sqrt(tilts.slice(1).reduce((a, t) => a + Math.sin(t) ** 2, 0) / 7), 3)}`); console.log(` the Galaxy's thin disk: scale height 300 +- 50 pc over scale length 2.6 +- 0.5 kpc = ${f(0.3 / 2.6, 3)}, the axis ratio of its density` + - ` contours; over the Sun's radius 8.2 kpc = ${f(0.3 / 8.2, 3)}, the like-for-like ratio (Bland-Hawthorn & Gerhard 2016)`); + ` contours; over the Sun's radius 8.2 kpc = ${f(0.3 / 8.2, 3)} (Bland-Hawthorn & Gerhard 2016)`); +console.log(` as rms height over radius: the thin disk ${f(Math.SQRT2 * 0.3 / 8.2, 3)} (an exponential profile has rms z = sqrt2 h), the planets` + + ` ${f(Math.sqrt(tilts.reduce((a, t) => a + Math.sin(t) ** 2, 0) / 8) / Math.SQRT2, 3)} (an orbit tilted by i has rms z/r = sin i/sqrt2)`); if (!quick) { console.log('\n5. Which orbits a body keeps: circular orbits followed for 20 turns of the host (Hill\'s problem)\n'); @@ -139,7 +141,8 @@ if (!quick) { const grid = (a0, a1, da) => { const out = []; for (let a = a0; a <= a1 + 1e-9; a += da) out.push(+a.toFixed(4)); return out; }; const big = grid(0.3, 1.2, 0.025), all = grid(0.025, 1.2, 0.025); const sunNu1 = { ...sun, nu2: 1, name: 'the Sun in the Galaxy with (nu/Omega)^2 = 1' }; - console.log(' radius [r_J] where half the orbits (8 starting angles in [0, pi/2)) have escaped beyond 3 r_J / are lost (escaped, or plunged within 1e-3 r_J):'); + console.log(' radius [r_J] where half the orbits (8 starting angles over [0, pi), a quarter step off pi p/8) have escaped beyond 3 r_J / are lost'); + console.log(' (escaped, or plunged within 1e-3 r_J):'); console.log(' orbit point-mass host the Sun in the Galaxy the same with (nu/Omega)^2 = 1'); const profiles = []; for (const [k, n] of Object.entries(planes)) { @@ -154,10 +157,13 @@ if (!quick) { console.log(' (an orbit that starts in the plane stays in it, so the vertical tide cannot act on the first two rows; with these 8 angles their'); console.log(' plunges begin only outside the escape edge, so the two radii agree. For polar rows "lost" is the first radius where half are lost.)'); for (const p of profiles) console.log(` ${p} [point mass | Sun | Sun, nu2 = 1]`); + console.log(' polar rows with 16 starting angles, escaped / lost [point mass | Sun | Sun, nu2 = 1]:'); + for (const [k, n] of Object.entries(planes).filter(([k]) => k.startsWith('polar'))) + console.log(` ${k.padEnd(28)}${[O.HOST_KEPLER, sun, sunNu1].map(h => { const r = O.hillScan(h, n, all, { nPhase: 16 }); return `${f(r.edge, 3)} / ${f(r.edgeLost, 3)}`.padStart(13); }).join(' ')}`); { - const across = [[1, 0, 0]], bigOnly = big; - const at = nu2 => O.hillScan({ ...sun, nu2 }, across, bigOnly, { nPhase: 8 }).edge; - console.log(` the Sun's polar-across escape edge at (nu/Omega)^2 = ${f(sun.nu2, 4)}, -1e-4, +1e-4: ${f(at(sun.nu2), 3)}, ${f(at(sun.nu2 - 1e-4), 3)}, ${f(at(sun.nu2 + 1e-4), 3)}`); + const at = h => O.hillScan(h, [[1, 0, 0]], big, { nPhase: 8 }).edge, atR = h => O.hillScan(h, [[0, 0, -1]], big, { nPhase: 8 }).edge; + console.log(` the Sun's polar-across escape edge at (nu/Omega)^2 = ${f(sun.nu2, 4)}, -1e-4, +1e-4: ${[0, -1e-4, 1e-4].map(d => f(at({ ...sun, nu2: sun.nu2 + d }), 3)).join(', ')}`); + console.log(` the Sun's retrograde in-plane escape edge at Gam = ${f(sun.Gam, 4)}, -1e-4, +1e-4: ${[0, -1e-4, 1e-4].map(d => f(atR({ ...sun, Gam: sun.Gam + d }), 3)).join(', ')}`); } { const s75 = Math.sin(75 * Math.PI / 180), c75 = Math.cos(75 * Math.PI / 180), as75 = grid(0.1, 0.4, 0.05); @@ -170,19 +176,20 @@ if (!quick) { { const gamE = 2 * (1 - EILERS.dv * EILERS.R0 / ve(EILERS.R0)), sunE = { ...sun, Gam: gamE }; console.log(` retrograde in the plane with Gam = ${f(gamE, 2)} (Eilers et al.'s slope): escaped by ${f(O.hillScan(sunE, [[0, 0, -1]], big, { nPhase: 8 }).edge, 3)} r_J`); - const s64 = (h, n) => O.hillScan(h, [n], grid(0.3, 1.3, 0.025), { nPhase: 64 }); - const re = s64(O.HOST_KEPLER, [0, 0, -1]), first = re.rows.find(r => r.plunges > 0); - console.log(` in the plane with 64 starting angles: point mass prograde ${f(s64(O.HOST_KEPLER, [0, 0, 1]).edge, 3)}, retrograde ${f(re.edge, 3)}` + - ` (lost ${f(re.edgeLost, 3)}, first plunge at ${first ? f(first.a, 3) : 'none'});`); - console.log(` the Sun retrograde ${f(s64(sun, [0, 0, -1]).edge, 3)}, and with Gam = ${f(gamE, 2)} ${f(s64(sunE, [0, 0, -1]).edge, 3)} r_J`); + const s64 = (h, n) => { const r = O.hillScan(h, [n], grid(0.3, 1.3, 0.025), { nPhase: 64 }), first = r.rows.find(x => x.plunges > 0); + return `${f(r.edge, 3)} (lost ${f(r.edgeLost, 3)}, first plunge at ${first ? f(first.a, 3) : 'none'})`; }; + console.log(' in the plane with 64 starting angles, escaped (lost, first plunge) [r_J]:'); + console.log(` point mass prograde ${s64(O.HOST_KEPLER, [0, 0, 1])}, retrograde ${s64(O.HOST_KEPLER, [0, 0, -1])}`); + console.log(` the Sun prograde ${s64(sun, [0, 0, 1])}, retrograde ${s64(sun, [0, 0, -1])}; retrograde with Gam = ${f(gamE, 2)} ${s64(sunE, [0, 0, -1])}`); } { - // how far from polar plunges reach: orbit normals tilted by delta from the host's plane (i = 90 +- delta), 4 node - // azimuths x 8 starting angles each, at a few radii inside the escape edge + // how far from polar plunges reach: orbit normals tilted by delta from polar (i = 90 +- delta), with the line of + // nodes along x or along y, x 16 starting angles each, at a few radii inside the escape edge (a node azimuth om + pi + // is the mirror image of om under z -> -z, so it would repeat the same orbits) const fromPolar = (h, a, deltas) => deltas.map(d => { const i = (90 - d) * Math.PI / 180, ns = []; - for (const s of [1, -1]) for (const om of [0, 0.5, 1, 1.5]) ns.push([Math.sin(i) * Math.cos(om * Math.PI), Math.sin(i) * Math.sin(om * Math.PI), s * Math.cos(i)]); - return O.hillFates(h, a, ns, { nPhase: 8 }).plunges; + for (const s of [1, -1]) for (const om of [0, 0.5]) ns.push([Math.sin(i) * Math.cos(om * Math.PI), Math.sin(i) * Math.sin(om * Math.PI), s * Math.cos(i)]); + return O.hillFates(h, a, ns, { nPhase: 16 }).plunges; }); const dK = grid(0, 20, 2), dS = grid(0, 40, 4); for (const a of [0.2, 0.3, 0.4]) console.log(` point mass, a = ${a} r_H: plunged fraction at ${dK[0]}-${dK[dK.length - 1]} deg from polar (step 2): ${fromPolar(O.HOST_KEPLER, a, dK).map(x => f(x, 2)).join(' ')}`); @@ -197,13 +204,6 @@ if (!quick) { for (const r of iso[name].rows) console.log(` a = ${f(r.a, 2)} r_J: stay ${f(r.stays, 2)}, escape ${f(r.escapes, 2)}, plunge ${f(r.plunges, 2)}`); } console.log(` for the Sun: half escaped by ${f(iso['the Sun in the Galaxy'].edge * rJ / C.au / 1e5, 2)}e5 au`); - { // the mirror pairs no starts within one of these directions, so check the second quarter turn gives the same fractions - const q = [['point mass', O.HOST_KEPLER, 0.5, iso['point-mass host']], ['the Sun', sun, 0.45, iso['the Sun in the Galaxy']]].map(([k, h, a, s]) => { - const r = s.rows.find(x => Math.abs(x.a - a) < 1e-9), r2 = O.hillFates(h, a, dirs, { nPhase: 4, quarter: 1 }); - return `${k} at ${a}: escaped ${f(r.escapes, 3)} -> ${f(r2.escapes, 3)}, lost ${f(r.escapes + r.plunges, 3)} -> ${f(r2.escapes + r2.plunges, 3)}`; - }); - console.log(` starting angles in [pi/2, pi) instead of [0, pi/2): ${q.join('; ')}`); - } const dirs24 = O.sphere(24); for (const rho of [0.08, 0.12]) { const e = O.hillScan(O.hostDisk(OORT.A, OORT.B, rho), dirs24, asIso, { nPhase: 4 }).edge; diff --git a/tests/orbits.mjs b/tests/orbits.mjs index 6c730c7..37cc4f8 100644 --- a/tests/orbits.mjs +++ b/tests/orbits.mjs @@ -162,8 +162,8 @@ near('Saturn\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O 'model (Lidov 1962, Kozai 1962: the eccentricity reaches 1 at 90 deg)'); const L = O.hillScan(K, [[1, 0, 0]], [0.1, 0.2], { nPhase: 8 }); check('Point-mass host: polar orbits (plane across the host) are lost by plunging between 0.1 and 0.2 r_H, before any escapes', - L.rows.every(r => r.escapes === 0) && L.edgeLost > 0.1 && L.edgeLost < 0.2, - `half lost by ${fmt(L.edgeLost)} r_H (must be 0.1 to 0.2); escaped fractions ${L.rows.map(r => r.escapes.toFixed(2)).join(', ')} (must all be 0)`, + L.rows.every(r => r.escapes === 0) && L.edgeLost > 0.1 && L.edgeLost <= 0.2, + `half lost by ${fmt(L.edgeLost)} r_H (must be above 0.1 and at most 0.2); escaped fractions ${L.rows.map(r => r.escapes.toFixed(2)).join(', ')} (must all be 0)`, 'model (the edgeLost path)'); } const Om = 27.2 * C.kmskpc, want = (4 * Math.PI * C.GMsun * 0.10 / C.pc ** 3 + 2 * (15.3 ** 2 - 11.9 ** 2) * C.kmskpc ** 2) / Om ** 2; @@ -177,7 +177,7 @@ near('Saturn\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O check('The Sun\'s all-directions escape edge lies within the Oort cloud\'s quoted outer limits, 6.5e4 to 2e5 au', edge >= 6.5e4 && edge <= 2e5, `half have escaped by ${fmt(edge)} au (12 directions x 2 angles, grid 0.1 r_J from 0.2 r_J = 5.7e4 au; the tide alone, no passing stars)`, 'Correa-Otto & Calandra 2019 sect. 4.3: a_crit ~ 6.5e4 au, set by the frequency of stellar passages acting with the tide (#37\'s review); ' + - 'Dones et al. 2004: as far as 2e5 au, as cited in search text; a wide window, which the edge passes for Gam from 0.6 to 1.5 times its value'); + 'Dones et al. 2004: as far as 2e5 au, as cited in search text; a wide window: at fixed r_J the edge stays within 1.24-1.29e5 au for Gam from 0.5 to 2 times its value, so this checks the scale, not the tide\'s details'); } console.log(fails ? `\n${fails} check(s) failed` : '\nall orbits checks pass'); From 9ec913e288bedfafe5646078966777937014308f Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 27 Oct 2026 11:03:50 +0000 Subject: [PATCH 08/10] Orbits study: polar profiles from 32 angles, and the verification of 80438b7 - The polar lost-fraction wording rested on 8-angle values at a few printed radii; the unprinted grid points swing by about 0.25, and at 0.325 r_H across the host the converged fraction is about a third. table.mjs now prints 32-angle polar profiles from 0.1 to 0.5 r_J on the 0.025 grid, with their edges, and the README describes those: half or more lost from 0.15 to 0.3 r_H around a point mass, then swinging between about a fifth and two-thirds out to 0.45 r_H; half or more at every radius for the Sun with its vertical tide. - The polar sampling error is stated from the mirrored set of 8 angles (escape edges up to 0.025), printed, as well as against 32 angles. - Printed now: the mirror-pair fate differences around a point mass (0, 16 and 32 of 64 in the plane and the two polar planes) and the orbits escaped at 0.925 r_J behind the Gam +- 1e-4 line (4, 3, 3 of 8). - The averaged-tide sentence separates the secular cone (to 5.1 deg at 0.2 r_J) from the rest; the Gam = 2.12 first-plunge clause is dropped; the Oort label gives the dense-grid range, 1.20 to 1.33e5 au; "#37's re-review" covers its calculations, and the Run-it line and the table's header credit it. Co-Authored-By: Claude Opus 5.5 Claude-Session: https://claude.ai/code/session_01C2bpiHyHSkC8DAL5CcyciK --- orbits/README.md | 25 +++++++++++++------------ orbits/table.mjs | 37 ++++++++++++++++++++++++++++--------- tests/orbits.mjs | 2 +- 3 files changed, 42 insertions(+), 22 deletions(-) diff --git a/orbits/README.md b/orbits/README.md index 2ce1d58..b0df4bc 100644 --- a/orbits/README.md +++ b/orbits/README.md @@ -16,7 +16,7 @@ what ends it, and what, if anything, repeats from one scale to the next. | How exact is Kepler's law in the Solar System? | In JPL's 1800–2050 fit, n²a³ matches GM☉ to within 1.3×10⁻³ for all eight planets. That is about the planets' total mass, 1.34×10⁻³ of the Sun's. Jupiter's excess is mostly its own mass in the two-body law; Uranus's and Neptune's are mostly the other planets' pull, and Saturn's fitted 2.2×10⁻⁴ is below its own 2.9×10⁻⁴ (see below). P ∝ a^1.49986 over 0.39 to 30 au. | | Does the Galaxy have a Kepler region? | Yes, around Sgr A*. The star S2 orbits it as a planet orbits the Sun, plus general relativity's precession: 12.2′ per orbit from their elements (they quote 12.1′). The speed's log slope passes −1/4 about 1.3 pc out (1.0 to 1.8 pc across the cluster-mass range; 0.7 to 1.9 pc for thresholds from −0.4 to −0.1). The stars inside weigh as much as the black hole by 1.8 to 3.2 pc. | | What ends a Kepler region? | For a body orbiting in a host (a planet, the Sun), the host's tide. Satellites are held inside its tidal radius, r_J³ = Gm/(4A(A − B)), apart from quasi-satellite orbits beyond about 2 r_H that Shen and Tremaine (2008) find around Jupiter, Uranus and Neptune. Inside r_J, the body's mean density is 4A/Ω times the host's mean density inside the orbit (host taken as spherical): 3 for a planet around the Sun, 2.25 for the Sun in the Galaxy. For the Sun, r_J = 1.37 pc = 2.8×10⁵ au. At the Galaxy's centre, where Sgr A* sits still, the region ends instead where the surrounding stars' mass approaches the black hole's. | -| How much of that region is used? | About half, for orbits that do not plunge, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.52 r_J around a planet and by 0.436 r_J (1.23×10⁵ au) for the Sun. Polar orbits are lost much sooner, to Lidov–Kozai cycles: half are lost by 0.14 r_H around a planet (half or more out to 0.35 r_H, fewer than half at 0.45 r_H) and by 0.06 r_J for the Sun, in 20 turns of the host. For near-circular moons near the ecliptic, Nesvorný et al. (2003) put the limits at about 0.4 and 0.7 r_H (prograde, retrograde), and the model gives 0.46 and 0.78. The Oort cloud's outer limit is placed between 6.5×10⁴ au (Correa-Otto and Calandra 2019, where the frequency of stellar passages, acting with the tide, sets it) and 2×10⁵ au, which is 0.23 to 0.71 r_J. | +| How much of that region is used? | About half, for orbits that do not plunge, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.52 r_J around a planet and by 0.436 r_J (1.23×10⁵ au) for the Sun. Polar orbits are lost much sooner, to Lidov–Kozai cycles: half are lost by 0.14 r_H around a planet (half or more out to 0.3 r_H; further out the fraction swings around a half, down to about a fifth near 0.4 r_H) and by 0.06 r_J for the Sun, in 20 turns of the host. For near-circular moons near the ecliptic, Nesvorný et al. (2003) put the limits at about 0.4 and 0.7 r_H (prograde, retrograde), and the model gives 0.46 and 0.78. The Oort cloud's outer limit is placed between 6.5×10⁴ au (Correa-Otto and Calandra 2019, where the frequency of stellar passages, acting with the tide, sets it) and 2×10⁵ au, which is 0.23 to 0.71 r_J. | | What does repeat? | Kepler regions nested inside one another: moons in a planet's, planets and comets in the Sun's. Each is bounded by its host's tide at a similar fraction of the tidal radius, though for the Oort cloud passing stars cut it shorter. Also flat disks: the planets' orbits are tilted by about 2° on average (6° for Mercury), and the Galaxy's thin disk has a scale height 0.04 of the Sun's orbital radius. | ## Kepler's law in the Solar System @@ -91,7 +91,7 @@ for Jupiter it is 237 years. - **Escaped:** once it passes 3 r_J. - **Plunged:** once it comes within 10⁻³ r_J, which is 0.74 of Jupiter's radius or, for the Sun, 283 au, outside the planets. - **Lost:** either. -- **Starting angles.** Hill's equations are symmetric under r → −r, v → −v, so a start half a turn on is the inverted orbit, with the same fate; the starting angles therefore span a half turn. The equations are also symmetric under z → −z and under time reversal with y → −y. For every single-plane row here, these make the start at π − φ the mirror image of the start at φ with time running backward (inverted, for a plane through the host's direction). In the plane the two fates nearly always match; for polar orbits they often differ, so neither is a duplicate. The angles sit a quarter step off the grid πp/8, so that no such pair falls in one sample. +- **Starting angles.** Hill's equations are symmetric under r → −r, v → −v, so a start half a turn on is the inverted orbit, with the same fate; the starting angles therefore span a half turn. The equations are also symmetric under z → −z and under time reversal with y → −y. For every single-plane row here, these make the start at π − φ the mirror image of the start at φ with time running backward (inverted, for a plane through the host's direction). In the plane the two fates nearly always match; for polar orbits they often differ (around a point mass at 0.15–0.5 r_H, none of 64 pairs differ in the plane, 16 and 32 of 64 in the two polar planes), so neither is a duplicate. The angles sit a quarter step off the grid πp/8, so that no such pair falls in one sample. The radius by which half the orbits have escaped, and by which half are lost (`node orbits/table.mjs`, section 5; 8 starting angles; grid 0.025 r_J; all directions on 48 directions × 4 angles and a 0.05 grid): @@ -106,23 +106,24 @@ The radius by which half the orbits have escaped, and by which half are lost (`n **The table's resolution.** With 64 starting angles instead of 8, the in-plane edges are 0.46 and 0.79 around a point mass, 0.49 and 0.95 for the Sun, and 1.03 for the Sun's retrograde orbits with Γ = 2.12. So the in-plane -8-angle values carry sampling errors of up to about 0.03. With 16 angles the polar edges move by at most 0.01. +8-angle values carry sampling errors of up to about 0.03. For the polar rows, 32 angles move the escape edges by at +most 0.011 and the lost edges by at most 0.004, while the mirrored set of 8 angles moves the escape edges by up to +0.025. **In-plane orbits.** These plunge only rarely: with 8 angles, first at 0.55 to 0.575 r_J prograde and at 1.1 r_J retrograde for the Sun, all outside their escape edge, and not at all retrograde around a point mass (to 1.2 r_J). With 64 angles the escaped and lost radii agree to about 0.005 (0.788 against 0.783 retrograde around a point mass), -except with Γ = 2.12, where the lost radius is 1.01 against 1.03 escaped; there the first plunge, at 0.975, falls -inside the escape edge. +except with Γ = 2.12, where the lost radius is 1.01 against 1.03 escaped. **Polar rows.** "Lost" is the first radius where half are lost. **Polar orbits are lost by plunging.** - **The mechanism.** A circular orbit at 90° to the host's plane goes through Lidov–Kozai cycles (Lidov 1962, Kozai 1962), whose eccentricity reaches 1. So it dives into the body long before it can escape. - **When.** Within 20 turns half are lost from 0.14 r_H around a planet and from 0.06 r_J for the Sun. Further in, the cycle takes longer than 20 turns; it is not absent. -- **Not everywhere outside.** Around a point mass the lost fraction is 63–100% at 0.15–0.2 r_H, half or more out to 0.35 r_H, and below half at 0.45 r_H (13–38%), where most polar orbits survive the run. With ν²/Ω² = 1 the Sun's lost fraction falls to half or less from 0.35 r_J. With the Galaxy's vertical tide the Sun loses half or more of its polar orbits at every radius printed, 0.1 to 0.45 r_J. +- **Not everywhere outside.** With 32 starting angles, the lost fraction around a point mass is half or more from 0.15 to 0.3 r_H. Between 0.325 and 0.45 r_H it swings between about a fifth and two-thirds, and at 0.4 r_H, 19–25% are lost and most polar orbits survive the run; past the escape edge it rises again as orbits escape. With ν²/Ω² = 1 the Sun's lost fraction is 19–47% from 0.325 to 0.45 r_J. With the Galaxy's vertical tide the Sun loses half or more of its polar orbits at every radius from 0.1 to 0.5 r_J. - **Other directions.** For a planet, plunges reach to about 8°, 10° and 14° from polar at 0.2, 0.3 and 0.4 r_H. Tilted to 75° (15° from polar), none plunge from 0.1 to 0.4 r_H. For the Sun they reach to about 8° from polar at 0.2 r_J and 24° at 0.3 r_J. At 75°, none plunge out to 0.3 r_J, and up to 63% do at 0.35–0.4 r_J. - **All directions.** Facing every way, up to 23% of the Sun's orbits plunge within 20 turns, and 9–11% of a planet's between 0.2 and 0.4 r_H. -- **What the averaged tide explains.** For comets, the tide driving them inward (Heisler and Tremaine 1986) accounts for the near-polar core: at 0.2 r_J all the Sun's orbits within 4° of polar plunge. The rest of its cone (to 8° at 0.2 r_J and 24° at 0.3 r_J) and its plunges at 75° are not secular: #37's re-review found that the averaged tide limits plunges to within 5.1° of polar at 0.2 r_J and 3.6° at 0.4 r_J, and gives none at 75°. These plunges depend steeply on the plunge radius: counted inside 10⁻² r_J, the largest plunged fraction over all directions rises from 0.27 to 0.43 (24 directions). +- **What the averaged tide explains.** For comets, the tide driving them inward (Heisler and Tremaine 1986) accounts for the near-polar core: at 0.2 r_J all the Sun's orbits within 4° of polar plunge. The rest of its cone, beyond 5.1° at 0.2 r_J (out to 8°) and out to 24° at 0.3 r_J, and its plunges at 75° are not secular: #37's re-review found that the averaged tide limits plunges to within 5.1° of polar at 0.2 r_J and 3.6° at 0.4 r_J, and gives none at 75°. These plunges depend steeply on the plunge radius: counted inside 10⁻² r_J, the largest plunged fraction over all directions rises from 0.27 to 0.43 (24 directions). **For a planet.** - **Against the published limits.** The model's in-plane edges are 0.46 (prograde) and 0.775 to 0.79 (retrograde), against Nesvorný et al.'s (2003) semi-major axes of about 0.4 and 0.7 r_H for nearly circular orbits near the ecliptic, as Shen and Tremaine (2008) quote them. @@ -241,7 +242,7 @@ mass, or mass spread through the disk and halo. ## Run it ``` -node orbits/table.mjs # every computed number above except the 100-turn runs and the values credited to #37's review (about 10 minutes) +node orbits/table.mjs # every computed number above except the 100-turn runs and the values credited to #37's review or re-review (about 10 minutes) node orbits/table.mjs quick # sections 1-4 only, well under a second (run by CI) node orbits/table.mjs long # adds the 100-turn runs, about 13 minutes in all node tests/orbits.mjs # the checks, about 6 s (also run by npm test) @@ -253,7 +254,7 @@ The table exits with an error if any printed value is not a finite number. `tests/orbits.mjs` has 43 checks. Each prints its source, the published or expected value, the model's value and the tolerance. "#37's review" marks values a reviewer read in the paper's arXiv source; "#37's re-review" marks -what the second review checked, in an arXiv source or against Crossref, as each label says. +what the second review checked or calculated, in an arXiv source or against Crossref, as each label says. | What | Against | Kind | |---|---|---| @@ -285,7 +286,7 @@ what the second review checked, in an arXiv source or against Crossref, as each - **Circular starting orbits only.** Real moons and comets start on eccentric orbits. - **The edges' definition.** They are set at half the orbits escaping, or lost, within 20 turns of the host, on a grid of 0.025 r_J (single planes) or 0.05 r_J (all directions). - **Plunge radius.** It is 10⁻³ r_J. Raising it to 10⁻² moves the Sun's all-directions escaped edge out, from 0.437 to 0.460, and its lost edge in, from 0.411 to 0.394 (24 directions). - - **Chaos and sampling.** Single orbits near an edge are chaotic: changing Γ by 10⁻⁴ either way moves the Sun's in-plane retrograde edge from 0.925 to 0.95, as one orbit of eight changes its fate. (Changing ν²/Ω² by 10⁻⁴ leaves the Sun's polar-across edge at 0.40 in this sample.) With 8 starting angles the in-plane edges carry sampling errors of up to about 0.03 and the polar edges up to about 0.01 (see the 64- and 16-angle values above). + - **Chaos and sampling.** Single orbits near an edge are chaotic: changing Γ by 10⁻⁴ either way moves the Sun's in-plane retrograde edge from 0.925 to 0.95, as one orbit of eight changes its fate. (Changing ν²/Ω² by 10⁻⁴ leaves the Sun's polar-across edge at 0.40 in this sample.) With 8 starting angles the in-plane edges carry sampling errors of up to about 0.03 and the polar escape edges up to about 0.025 (see the table's resolution above). - **Longer runs** move the edges inward a little (see the 100-turn values above). - **MWPotential2014 is one model among many.** Its halo sets the flat range, and it has no nuclear star cluster. The checks pin its halo, its Kuzmin and Plummer limits pin the disk's a and b to galpy's values, and Table 1's derived values (bulge and disk masses, the mass inside 60 kpc, the slope at R₀ and the local dark-matter density) are checked at Table 1's rounding. The bulge's slope and cut-off are taken from galpy's source and not tested: its mass follows from its 5% share of the force at R₀. The Galactic-centre numbers use the separate power-law cusp above, which is valid only inside its 3 pc break. - **Spherical mass estimates.** v²R/G and 3Ω²/4πG treat a flattened galaxy as spherical. In MWPotential2014 this overstates the mass inside 8 kpc by 12%. It shifts the 3.7-fold growth to 4.0 and the Sun's 2.25 to 2.53, but not the contrast with Kepler's fixed mass. @@ -293,8 +294,8 @@ what the second review checked, in an arXiv source or against Crossref, as each ## Sources Each entry says how it was read. "#37's review" means a reviewer checked the values used here in the arXiv source, -or in the primary file, during #37's review. "#37's re-review" marks what the second review checked: values in an -arXiv source, or a reference matched against Crossref where the entry says so. +or in the primary file, during #37's review. "#37's re-review" marks what the second review checked or calculated: +values in an arXiv source, a reference matched against Crossref where the entry says so, or its averaged-tide cone. - W. M. Folkner et al., *The Planetary and Lunar Ephemerides DE430 and DE431*, IPN Progress Report 42-196 (2014), Table 8. #37's review. - E. M. Standish and J. G. Williams, *Approximate Positions of the Planets*, JPL Solar System Dynamics (credited there to Standish and Williams 1992), Table 1 (1800–2050). Checked against the SSD page in #37's review. diff --git a/orbits/table.mjs b/orbits/table.mjs index 6ab8ed8..aa116b3 100644 --- a/orbits/table.mjs +++ b/orbits/table.mjs @@ -1,4 +1,4 @@ -// Prints every computed number in orbits/README.md, except the values the README credits to #37's review. Run: node orbits/table.mjs (about 10 minutes; the stability scans of +// Prints every computed number in orbits/README.md, except the values the README credits to #37's review or re-review. Run: node orbits/table.mjs (about 10 minutes; the stability scans of // section 5 take nearly all of it), node orbits/table.mjs quick (well under a second; section 5 left out), or // node orbits/table.mjs long (adds the all-directions scans followed for 100 turns of the host). // Exits with code 1 if any printed line holds NaN or Infinity. @@ -151,19 +151,38 @@ if (!quick) { console.log(` ${k.padEnd(28)} ${e.map(r => `${f(r.edge, 3)} / ${f(r.edgeLost, 3)}`.padStart(13)).join(' ')}`); const firstPlunge = e.map(r => r.rows.find(x => x.plunges > 0)); // undefined if none plunge, so a NaN radius still prints as NaN if (!polar) profiles.push(`${k}: first plunge at ${firstPlunge.map(row => (row ? f(row.a, 3) : 'none')).join(', ')} r_J`); - else profiles.push(`${k}: lost fraction at a = 0.1, 0.15, 0.2, 0.3, 0.35, 0.4, 0.45: ` + - e.map(r => [0.1, 0.15, 0.2, 0.3, 0.35, 0.4, 0.45].map(a => { const x = r.rows.find(y => Math.abs(y.a - a) < 1e-9); return f(x.escapes + x.plunges, 2); }).join(' ')).join(' | ')); } console.log(' (an orbit that starts in the plane stays in it, so the vertical tide cannot act on the first two rows; with these 8 angles their'); console.log(' plunges begin only outside the escape edge, so the two radii agree. For polar rows "lost" is the first radius where half are lost.)'); for (const p of profiles) console.log(` ${p} [point mass | Sun | Sun, nu2 = 1]`); - console.log(' polar rows with 16 starting angles, escaped / lost [point mass | Sun | Sun, nu2 = 1]:'); - for (const [k, n] of Object.entries(planes).filter(([k]) => k.startsWith('polar'))) - console.log(` ${k.padEnd(28)}${[O.HOST_KEPLER, sun, sunNu1].map(h => { const r = O.hillScan(h, n, all, { nPhase: 16 }); return `${f(r.edge, 3)} / ${f(r.edgeLost, 3)}`.padStart(13); }).join(' ')}`); { - const at = h => O.hillScan(h, [[1, 0, 0]], big, { nPhase: 8 }).edge, atR = h => O.hillScan(h, [[0, 0, -1]], big, { nPhase: 8 }).edge; + const polars = Object.entries(planes).filter(([k]) => k.startsWith('polar')), hosts = [['point mass', O.HOST_KEPLER], ['the Sun', sun], ['the Sun, nu2 = 1', sunNu1]]; + console.log(' polar rows with the mirrored set of 8 angles (pi - phi, a quarter step the other way), escaped / lost [point mass | Sun | Sun, nu2 = 1]:'); + for (const [k, n] of polars) + console.log(` ${k.padEnd(28)}${hosts.map(([, h]) => { const r = O.hillScan(h, n, all, { nPhase: 8, offset: 0.75 }); return `${f(r.edge, 3)} / ${f(r.edgeLost, 3)}`.padStart(13); }).join(' ')}`); + const as32 = grid(0.1, 0.5, 0.025); + console.log(` polar rows with 32 angles: the fraction lost (escaped or plunged) at a = ${as32[0]} to ${as32[as32.length - 1]} r_J in steps of 0.025, and the edges:`); + for (const [k, n] of polars) for (const [name, h] of hosts) { + const r = O.hillScan(h, n, as32, { nPhase: 32 }), lost = r.edgeLost === as32[0] ? `below ${as32[0]}` : f(r.edgeLost, 3); + console.log(` ${k}, ${name}: ${r.rows.map(x => f(x.escapes + x.plunges, 2)).join(' ')}; escaped by ${f(r.edge, 3)}, lost by ${lost}`); + } + } + { + const at = h => O.hillScan(h, [[1, 0, 0]], big, { nPhase: 8 }).edge, atR = h => O.hillScan(h, [[0, 0, -1]], big, { nPhase: 8 }); console.log(` the Sun's polar-across escape edge at (nu/Omega)^2 = ${f(sun.nu2, 4)}, -1e-4, +1e-4: ${[0, -1e-4, 1e-4].map(d => f(at({ ...sun, nu2: sun.nu2 + d }), 3)).join(', ')}`); - console.log(` the Sun's retrograde in-plane escape edge at Gam = ${f(sun.Gam, 4)}, -1e-4, +1e-4: ${[0, -1e-4, 1e-4].map(d => f(atR({ ...sun, Gam: sun.Gam + d }), 3)).join(', ')}`); + const rG = [0, -1e-4, 1e-4].map(d => atR({ ...sun, Gam: sun.Gam + d })), at925 = r => r.rows.find(x => Math.abs(x.a - 0.925) < 1e-9).escapes * 8; + console.log(` the Sun's retrograde in-plane escape edge at Gam = ${f(sun.Gam, 4)}, -1e-4, +1e-4: ${rG.map(r => f(r.edge, 3)).join(', ')}` + + ` (orbits of 8 escaped at 0.925 r_J: ${rG.map(at925).join(', ')})`); + } + { // a start and its time-reversed mirror (pi - phi): how often their fates differ, around a point mass + const as = grid(0.15, 0.5, 0.05), diff = n => { let d = 0, t = 0; + for (const a of as) for (let p = 0; p < 8; p++) { + const A = O.hillOrbit(O.HOST_KEPLER, O.hillStart(O.HOST_KEPLER, a, n, Math.PI * (p + 0.25) / 8)).fate; + const B = O.hillOrbit(O.HOST_KEPLER, O.hillStart(O.HOST_KEPLER, a, n, Math.PI - Math.PI * (p + 0.25) / 8)).fate; + t++; if (A !== B) d++; + } + return `${d} of ${t}`; }; + console.log(` point mass, a = 0.15 to 0.5 r_H: mirrored pairs whose fates differ: retrograde in the plane ${diff([0, 0, -1])}, polar through the host ${diff([0, 1, 0])}, polar across ${diff([1, 0, 0])}`); } { const s75 = Math.sin(75 * Math.PI / 180), c75 = Math.cos(75 * Math.PI / 180), as75 = grid(0.1, 0.4, 0.05); @@ -185,7 +204,7 @@ if (!quick) { { // how far from polar plunges reach: orbit normals tilted by delta from polar (i = 90 +- delta), with the line of // nodes along x or along y, x 16 starting angles each, at a few radii inside the escape edge (a node azimuth om + pi - // is the mirror image of om under z -> -z, so it would repeat the same orbits) + // is the mirror image of om under z -> -z, so it would repeat the same orbits; at 0 deg the two senses coincide) const fromPolar = (h, a, deltas) => deltas.map(d => { const i = (90 - d) * Math.PI / 180, ns = []; for (const s of [1, -1]) for (const om of [0, 0.5]) ns.push([Math.sin(i) * Math.cos(om * Math.PI), Math.sin(i) * Math.sin(om * Math.PI), s * Math.cos(i)]); diff --git a/tests/orbits.mjs b/tests/orbits.mjs index 37cc4f8..4c0a3e5 100644 --- a/tests/orbits.mjs +++ b/tests/orbits.mjs @@ -177,7 +177,7 @@ near('Saturn\'s inclination to the invariable plane [deg]', O.tiltToInvariable(O check('The Sun\'s all-directions escape edge lies within the Oort cloud\'s quoted outer limits, 6.5e4 to 2e5 au', edge >= 6.5e4 && edge <= 2e5, `half have escaped by ${fmt(edge)} au (12 directions x 2 angles, grid 0.1 r_J from 0.2 r_J = 5.7e4 au; the tide alone, no passing stars)`, 'Correa-Otto & Calandra 2019 sect. 4.3: a_crit ~ 6.5e4 au, set by the frequency of stellar passages acting with the tide (#37\'s review); ' + - 'Dones et al. 2004: as far as 2e5 au, as cited in search text; a wide window: at fixed r_J the edge stays within 1.24-1.29e5 au for Gam from 0.5 to 2 times its value, so this checks the scale, not the tide\'s details'); + 'Dones et al. 2004: as far as 2e5 au, as cited in search text; a wide window: at fixed r_J the edge stays between about 1.20 and 1.33e5 au for Gam from 0.5 to 2 times its value (in steps of 0.025), so this checks the scale, not the tide\'s details'); } console.log(fails ? `\n${fails} check(s) failed` : '\nall orbits checks pass'); From 905eb6eb97ab6dca03dfa459cd8d6ea2b59278af Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 27 Oct 2026 11:41:20 +0000 Subject: [PATCH 09/10] Orbits study: 75 deg rows from 32 angles, and the check of 9ec913e - The 75 deg rows used 8 angles; with 32 the Sun's prograde orbits plunge from 0.3 r_J (up to 16%) and up to 44% at 0.4 r_J, not 63%. The table prints 32-angle 75 deg rows, labelled as the prograde side; the tilt-from-polar cone counts both senses. - Row 19 and the 'When' bullet: half lost by 0.14 to 0.15 r_H; beyond 0.3 r_H the 32-angle fraction runs between about a fifth and two-thirds. - 32-angle fractions are rounded and given +-0.1 sampling noise; the lost-edge bound names the edges the 32-angle grid measures; the mirror-pair count in the plane is for retrograde orbits. Co-Authored-By: Claude Opus 5.5 Claude-Session: https://claude.ai/code/session_01C2bpiHyHSkC8DAL5CcyciK --- README.md | 2 +- orbits/README.md | 18 +++++++++--------- orbits/table.mjs | 6 +++--- 3 files changed, 13 insertions(+), 13 deletions(-) diff --git a/README.md b/README.md index 609b234..8696a99 100644 --- a/README.md +++ b/README.md @@ -216,7 +216,7 @@ node orbits/table.mjs quick # the orbits study's numbers, without i below). The third study's checks, `tests/critical.mjs`, add about 65 s of seeded Monte Carlo runs, most of it percolation with nine seeds. CI runs every line above except the `critical/table.mjs` line: it repeats the tests' seeded runs and prints the results without checking them. The orbits study's full table (`node orbits/table.mjs`, -about 10 minutes) is not run by CI either. +about 11 minutes) is not run by CI either. ## Checks diff --git a/orbits/README.md b/orbits/README.md index b0df4bc..938c1c7 100644 --- a/orbits/README.md +++ b/orbits/README.md @@ -16,7 +16,7 @@ what ends it, and what, if anything, repeats from one scale to the next. | How exact is Kepler's law in the Solar System? | In JPL's 1800–2050 fit, n²a³ matches GM☉ to within 1.3×10⁻³ for all eight planets. That is about the planets' total mass, 1.34×10⁻³ of the Sun's. Jupiter's excess is mostly its own mass in the two-body law; Uranus's and Neptune's are mostly the other planets' pull, and Saturn's fitted 2.2×10⁻⁴ is below its own 2.9×10⁻⁴ (see below). P ∝ a^1.49986 over 0.39 to 30 au. | | Does the Galaxy have a Kepler region? | Yes, around Sgr A*. The star S2 orbits it as a planet orbits the Sun, plus general relativity's precession: 12.2′ per orbit from their elements (they quote 12.1′). The speed's log slope passes −1/4 about 1.3 pc out (1.0 to 1.8 pc across the cluster-mass range; 0.7 to 1.9 pc for thresholds from −0.4 to −0.1). The stars inside weigh as much as the black hole by 1.8 to 3.2 pc. | | What ends a Kepler region? | For a body orbiting in a host (a planet, the Sun), the host's tide. Satellites are held inside its tidal radius, r_J³ = Gm/(4A(A − B)), apart from quasi-satellite orbits beyond about 2 r_H that Shen and Tremaine (2008) find around Jupiter, Uranus and Neptune. Inside r_J, the body's mean density is 4A/Ω times the host's mean density inside the orbit (host taken as spherical): 3 for a planet around the Sun, 2.25 for the Sun in the Galaxy. For the Sun, r_J = 1.37 pc = 2.8×10⁵ au. At the Galaxy's centre, where Sgr A* sits still, the region ends instead where the surrounding stars' mass approaches the black hole's. | -| How much of that region is used? | About half, for orbits that do not plunge, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.52 r_J around a planet and by 0.436 r_J (1.23×10⁵ au) for the Sun. Polar orbits are lost much sooner, to Lidov–Kozai cycles: half are lost by 0.14 r_H around a planet (half or more out to 0.3 r_H; further out the fraction swings around a half, down to about a fifth near 0.4 r_H) and by 0.06 r_J for the Sun, in 20 turns of the host. For near-circular moons near the ecliptic, Nesvorný et al. (2003) put the limits at about 0.4 and 0.7 r_H (prograde, retrograde), and the model gives 0.46 and 0.78. The Oort cloud's outer limit is placed between 6.5×10⁴ au (Correa-Otto and Calandra 2019, where the frequency of stellar passages, acting with the tide, sets it) and 2×10⁵ au, which is 0.23 to 0.71 r_J. | +| How much of that region is used? | About half, for orbits that do not plunge, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.52 r_J around a planet and by 0.436 r_J (1.23×10⁵ au) for the Sun. Polar orbits are lost much sooner, to Lidov–Kozai cycles: half are lost by 0.14 to 0.15 r_H around a planet (half or more out to 0.3 r_H; further out between about a fifth and two-thirds) and by 0.06 r_J for the Sun, in 20 turns of the host. For near-circular moons near the ecliptic, Nesvorný et al. (2003) put the limits at about 0.4 and 0.7 r_H (prograde, retrograde), and the model gives 0.46 and 0.78. The Oort cloud's outer limit is placed between 6.5×10⁴ au (Correa-Otto and Calandra 2019, where the frequency of stellar passages, acting with the tide, sets it) and 2×10⁵ au, which is 0.23 to 0.71 r_J. | | What does repeat? | Kepler regions nested inside one another: moons in a planet's, planets and comets in the Sun's. Each is bounded by its host's tide at a similar fraction of the tidal radius, though for the Oort cloud passing stars cut it shorter. Also flat disks: the planets' orbits are tilted by about 2° on average (6° for Mercury), and the Galaxy's thin disk has a scale height 0.04 of the Sun's orbital radius. | ## Kepler's law in the Solar System @@ -91,7 +91,7 @@ for Jupiter it is 237 years. - **Escaped:** once it passes 3 r_J. - **Plunged:** once it comes within 10⁻³ r_J, which is 0.74 of Jupiter's radius or, for the Sun, 283 au, outside the planets. - **Lost:** either. -- **Starting angles.** Hill's equations are symmetric under r → −r, v → −v, so a start half a turn on is the inverted orbit, with the same fate; the starting angles therefore span a half turn. The equations are also symmetric under z → −z and under time reversal with y → −y. For every single-plane row here, these make the start at π − φ the mirror image of the start at φ with time running backward (inverted, for a plane through the host's direction). In the plane the two fates nearly always match; for polar orbits they often differ (around a point mass at 0.15–0.5 r_H, none of 64 pairs differ in the plane, 16 and 32 of 64 in the two polar planes), so neither is a duplicate. The angles sit a quarter step off the grid πp/8, so that no such pair falls in one sample. +- **Starting angles.** Hill's equations are symmetric under r → −r, v → −v, so a start half a turn on is the inverted orbit, with the same fate; the starting angles therefore span a half turn. The equations are also symmetric under z → −z and under time reversal with y → −y. For every single-plane row here, these make the start at π − φ the mirror image of the start at φ with time running backward (inverted, for a plane through the host's direction). In the plane the two fates nearly always match; for polar orbits they often differ (around a point mass at 0.15–0.5 r_H, none of 64 retrograde pairs differ in the plane, 16 and 32 of 64 in the two polar planes), so neither is a duplicate. The angles sit a quarter step off the grid πp/8, so that no such pair falls in one sample. The radius by which half the orbits have escaped, and by which half are lost (`node orbits/table.mjs`, section 5; 8 starting angles; grid 0.025 r_J; all directions on 48 directions × 4 angles and a 0.05 grid): @@ -107,8 +107,8 @@ The radius by which half the orbits have escaped, and by which half are lost (`n **The table's resolution.** With 64 starting angles instead of 8, the in-plane edges are 0.46 and 0.79 around a point mass, 0.49 and 0.95 for the Sun, and 1.03 for the Sun's retrograde orbits with Γ = 2.12. So the in-plane 8-angle values carry sampling errors of up to about 0.03. For the polar rows, 32 angles move the escape edges by at -most 0.011 and the lost edges by at most 0.004, while the mirrored set of 8 angles moves the escape edges by up to -0.025. +most 0.011 and the lost edges around a point mass and with ν²/Ω² = 1 by at most 0.004 (the Sun's 0.06 lies below the +32-angle grid), while the mirrored set of 8 angles moves the escape edges by up to 0.025. **In-plane orbits.** These plunge only rarely: with 8 angles, first at 0.55 to 0.575 r_J prograde and at 1.1 r_J retrograde for the Sun, all outside their escape edge, and not at all retrograde around a point mass (to 1.2 r_J). @@ -119,9 +119,9 @@ except with Γ = 2.12, where the lost radius is 1.01 against 1.03 escaped. **Polar orbits are lost by plunging.** - **The mechanism.** A circular orbit at 90° to the host's plane goes through Lidov–Kozai cycles (Lidov 1962, Kozai 1962), whose eccentricity reaches 1. So it dives into the body long before it can escape. -- **When.** Within 20 turns half are lost from 0.14 r_H around a planet and from 0.06 r_J for the Sun. Further in, the cycle takes longer than 20 turns; it is not absent. -- **Not everywhere outside.** With 32 starting angles, the lost fraction around a point mass is half or more from 0.15 to 0.3 r_H. Between 0.325 and 0.45 r_H it swings between about a fifth and two-thirds, and at 0.4 r_H, 19–25% are lost and most polar orbits survive the run; past the escape edge it rises again as orbits escape. With ν²/Ω² = 1 the Sun's lost fraction is 19–47% from 0.325 to 0.45 r_J. With the Galaxy's vertical tide the Sun loses half or more of its polar orbits at every radius from 0.1 to 0.5 r_J. -- **Other directions.** For a planet, plunges reach to about 8°, 10° and 14° from polar at 0.2, 0.3 and 0.4 r_H. Tilted to 75° (15° from polar), none plunge from 0.1 to 0.4 r_H. For the Sun they reach to about 8° from polar at 0.2 r_J and 24° at 0.3 r_J. At 75°, none plunge out to 0.3 r_J, and up to 63% do at 0.35–0.4 r_J. +- **When.** Within 20 turns half are lost from 0.14 to 0.15 r_H around a planet and from 0.06 r_J for the Sun. Further in, the cycle takes longer than 20 turns; it is not absent. +- **Not everywhere outside.** With 32 starting angles, the lost fraction around a point mass is half or more from 0.15 to 0.3 r_H. Between 0.325 and 0.45 r_H it swings between about a fifth and two-thirds, and at 0.4 r_H only about a fifth to a quarter are lost, so most polar orbits survive the run; past the escape edge it rises again as orbits escape. With ν²/Ω² = 1 the Sun's lost fraction is about a fifth to a half from 0.325 to 0.45 r_J. With the Galaxy's vertical tide the Sun loses half or more of its polar orbits at every radius from 0.1 to 0.5 r_J (exactly half at 0.35 r_J across the host). These 32-angle fractions carry sampling noise of about ±0.1. +- **Other directions.** For a planet, plunges reach to about 8°, 10° and 14° from polar at 0.2, 0.3 and 0.4 r_H, counting both senses of tilt. On the prograde side, tilted to 75° (15° from polar; 32 angles), none plunge from 0.1 to 0.4 r_H. For the Sun plunges reach to about 8° from polar at 0.2 r_J and 24° at 0.3 r_J. At 75° on the prograde side, none plunge out to 0.25 r_J, up to 16% do at 0.3 r_J and up to 44% at 0.4 r_J. - **All directions.** Facing every way, up to 23% of the Sun's orbits plunge within 20 turns, and 9–11% of a planet's between 0.2 and 0.4 r_H. - **What the averaged tide explains.** For comets, the tide driving them inward (Heisler and Tremaine 1986) accounts for the near-polar core: at 0.2 r_J all the Sun's orbits within 4° of polar plunge. The rest of its cone, beyond 5.1° at 0.2 r_J (out to 8°) and out to 24° at 0.3 r_J, and its plunges at 75° are not secular: #37's re-review found that the averaged tide limits plunges to within 5.1° of polar at 0.2 r_J and 3.6° at 0.4 r_J, and gives none at 75°. These plunges depend steeply on the plunge radius: counted inside 10⁻² r_J, the largest plunged fraction over all directions rises from 0.27 to 0.43 (24 directions). @@ -242,9 +242,9 @@ mass, or mass spread through the disk and halo. ## Run it ``` -node orbits/table.mjs # every computed number above except the 100-turn runs and the values credited to #37's review or re-review (about 10 minutes) +node orbits/table.mjs # every computed number above except the 100-turn runs and the values credited to #37's review or re-review (about 11 minutes) node orbits/table.mjs quick # sections 1-4 only, well under a second (run by CI) -node orbits/table.mjs long # adds the 100-turn runs, about 13 minutes in all +node orbits/table.mjs long # adds the 100-turn runs, about 14 minutes in all node tests/orbits.mjs # the checks, about 6 s (also run by npm test) ``` diff --git a/orbits/table.mjs b/orbits/table.mjs index aa116b3..cae06c1 100644 --- a/orbits/table.mjs +++ b/orbits/table.mjs @@ -1,4 +1,4 @@ -// Prints every computed number in orbits/README.md, except the values the README credits to #37's review or re-review. Run: node orbits/table.mjs (about 10 minutes; the stability scans of +// Prints every computed number in orbits/README.md, except the values the README credits to #37's review or re-review. Run: node orbits/table.mjs (about 11 minutes; the stability scans of // section 5 take nearly all of it), node orbits/table.mjs quick (well under a second; section 5 left out), or // node orbits/table.mjs long (adds the all-directions scans followed for 100 turns of the host). // Exits with code 1 if any printed line holds NaN or Infinity. @@ -187,8 +187,8 @@ if (!quick) { { const s75 = Math.sin(75 * Math.PI / 180), c75 = Math.cos(75 * Math.PI / 180), as75 = grid(0.1, 0.4, 0.05); for (const [name, h] of [['point mass', O.HOST_KEPLER], ['the Sun', sun]]) { - const row = [[s75, 0, c75], [0, s75, c75]].map(n => as75.map(a => f(O.hillFates(h, a, [n], { nPhase: 8 }).plunges, 2)).join(' ')); - console.log(` tilted 75 deg, ${name}: plunged fraction at a = ${as75.join(', ')}: node along y: ${row[0]}; node along x: ${row[1]}`); + const row = [[s75, 0, c75], [0, s75, c75]].map(n => as75.map(a => f(O.hillFates(h, a, [n], { nPhase: 32 }).plunges, 2)).join(' ')); + console.log(` tilted 75 deg (prograde side), 32 angles, ${name}: plunged fraction at a = ${as75.join(', ')}: node along y: ${row[0]}; node along x: ${row[1]}`); } } console.log(' (polar orbits are lost to Lidov-Kozai cycles, whose eccentricity reaches 1 at 90 deg; below the "lost" radius the cycle takes longer than 20 turns)'); From 4345024c38e0d0784c9495e3d67759c7ff315468 Mon Sep 17 00:00:00 2001 From: Claude Date: Wed, 28 Oct 2026 00:03:14 +0000 Subject: [PATCH 10/10] Orbits study: non-blocking points from the check of 905eb6e - Plunge cones are given as brackets on the scan's steps (8-10, 10-12 and 14-16 deg for a planet; 8-12 and 24-28 deg for the Sun), not as the last step with plunges. - 32-angle sampling noise: about +-0.1, and up to 0.2; at 0.4 r_H a fifth to a quarter are lost here, about a quarter with 512 angles. - Row 19 bounds its outer range at 0.45 r_H; the polar mirror-pair counts name their planes. - Shen and Tremaine's title as Crossref gives it. - The bulge-mass check pins the bulge's force share only to about 10%; the step cap would fire only below about 0.012 r_J. No change to the program's output. Co-Authored-By: Claude Opus 5.5 Claude-Session: https://claude.ai/code/session_01C2bpiHyHSkC8DAL5CcyciK --- orbits/README.md | 14 +++++++------- orbits/model.js | 3 ++- 2 files changed, 9 insertions(+), 8 deletions(-) diff --git a/orbits/README.md b/orbits/README.md index 938c1c7..7ff4a84 100644 --- a/orbits/README.md +++ b/orbits/README.md @@ -16,7 +16,7 @@ what ends it, and what, if anything, repeats from one scale to the next. | How exact is Kepler's law in the Solar System? | In JPL's 1800–2050 fit, n²a³ matches GM☉ to within 1.3×10⁻³ for all eight planets. That is about the planets' total mass, 1.34×10⁻³ of the Sun's. Jupiter's excess is mostly its own mass in the two-body law; Uranus's and Neptune's are mostly the other planets' pull, and Saturn's fitted 2.2×10⁻⁴ is below its own 2.9×10⁻⁴ (see below). P ∝ a^1.49986 over 0.39 to 30 au. | | Does the Galaxy have a Kepler region? | Yes, around Sgr A*. The star S2 orbits it as a planet orbits the Sun, plus general relativity's precession: 12.2′ per orbit from their elements (they quote 12.1′). The speed's log slope passes −1/4 about 1.3 pc out (1.0 to 1.8 pc across the cluster-mass range; 0.7 to 1.9 pc for thresholds from −0.4 to −0.1). The stars inside weigh as much as the black hole by 1.8 to 3.2 pc. | | What ends a Kepler region? | For a body orbiting in a host (a planet, the Sun), the host's tide. Satellites are held inside its tidal radius, r_J³ = Gm/(4A(A − B)), apart from quasi-satellite orbits beyond about 2 r_H that Shen and Tremaine (2008) find around Jupiter, Uranus and Neptune. Inside r_J, the body's mean density is 4A/Ω times the host's mean density inside the orbit (host taken as spherical): 3 for a planet around the Sun, 2.25 for the Sun in the Galaxy. For the Sun, r_J = 1.37 pc = 2.8×10⁵ au. At the Galaxy's centre, where Sgr A* sits still, the region ends instead where the surrounding stars' mass approaches the black hole's. | -| How much of that region is used? | About half, for orbits that do not plunge, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.52 r_J around a planet and by 0.436 r_J (1.23×10⁵ au) for the Sun. Polar orbits are lost much sooner, to Lidov–Kozai cycles: half are lost by 0.14 to 0.15 r_H around a planet (half or more out to 0.3 r_H; further out between about a fifth and two-thirds) and by 0.06 r_J for the Sun, in 20 turns of the host. For near-circular moons near the ecliptic, Nesvorný et al. (2003) put the limits at about 0.4 and 0.7 r_H (prograde, retrograde), and the model gives 0.46 and 0.78. The Oort cloud's outer limit is placed between 6.5×10⁴ au (Correa-Otto and Calandra 2019, where the frequency of stellar passages, acting with the tide, sets it) and 2×10⁵ au, which is 0.23 to 0.71 r_J. | +| How much of that region is used? | About half, for orbits that do not plunge, in this study's three-body runs. Of circular orbits facing every way, half have escaped by 0.52 r_J around a planet and by 0.436 r_J (1.23×10⁵ au) for the Sun. Polar orbits are lost much sooner, to Lidov–Kozai cycles: half are lost by 0.14 to 0.15 r_H around a planet (half or more out to 0.3 r_H; further out, to 0.45 r_H, between about a fifth and two-thirds) and by 0.06 r_J for the Sun, in 20 turns of the host. For near-circular moons near the ecliptic, Nesvorný et al. (2003) put the limits at about 0.4 and 0.7 r_H (prograde, retrograde), and the model gives 0.46 and 0.78. The Oort cloud's outer limit is placed between 6.5×10⁴ au (Correa-Otto and Calandra 2019, where the frequency of stellar passages, acting with the tide, sets it) and 2×10⁵ au, which is 0.23 to 0.71 r_J. | | What does repeat? | Kepler regions nested inside one another: moons in a planet's, planets and comets in the Sun's. Each is bounded by its host's tide at a similar fraction of the tidal radius, though for the Oort cloud passing stars cut it shorter. Also flat disks: the planets' orbits are tilted by about 2° on average (6° for Mercury), and the Galaxy's thin disk has a scale height 0.04 of the Sun's orbital radius. | ## Kepler's law in the Solar System @@ -91,7 +91,7 @@ for Jupiter it is 237 years. - **Escaped:** once it passes 3 r_J. - **Plunged:** once it comes within 10⁻³ r_J, which is 0.74 of Jupiter's radius or, for the Sun, 283 au, outside the planets. - **Lost:** either. -- **Starting angles.** Hill's equations are symmetric under r → −r, v → −v, so a start half a turn on is the inverted orbit, with the same fate; the starting angles therefore span a half turn. The equations are also symmetric under z → −z and under time reversal with y → −y. For every single-plane row here, these make the start at π − φ the mirror image of the start at φ with time running backward (inverted, for a plane through the host's direction). In the plane the two fates nearly always match; for polar orbits they often differ (around a point mass at 0.15–0.5 r_H, none of 64 retrograde pairs differ in the plane, 16 and 32 of 64 in the two polar planes), so neither is a duplicate. The angles sit a quarter step off the grid πp/8, so that no such pair falls in one sample. +- **Starting angles.** Hill's equations are symmetric under r → −r, v → −v, so a start half a turn on is the inverted orbit, with the same fate; the starting angles therefore span a half turn. The equations are also symmetric under z → −z and under time reversal with y → −y. For every single-plane row here, these make the start at π − φ the mirror image of the start at φ with time running backward (inverted, for a plane through the host's direction). In the plane the two fates nearly always match; for polar orbits they often differ (around a point mass at 0.15–0.5 r_H, none of 64 retrograde pairs differ in the plane, 16 of 64 in the polar plane through the host's direction and 32 of 64 across it), so neither is a duplicate. The angles sit a quarter step off the grid πp/8, so that no such pair falls in one sample. The radius by which half the orbits have escaped, and by which half are lost (`node orbits/table.mjs`, section 5; 8 starting angles; grid 0.025 r_J; all directions on 48 directions × 4 angles and a 0.05 grid): @@ -120,10 +120,10 @@ except with Γ = 2.12, where the lost radius is 1.01 against 1.03 escaped. **Polar orbits are lost by plunging.** - **The mechanism.** A circular orbit at 90° to the host's plane goes through Lidov–Kozai cycles (Lidov 1962, Kozai 1962), whose eccentricity reaches 1. So it dives into the body long before it can escape. - **When.** Within 20 turns half are lost from 0.14 to 0.15 r_H around a planet and from 0.06 r_J for the Sun. Further in, the cycle takes longer than 20 turns; it is not absent. -- **Not everywhere outside.** With 32 starting angles, the lost fraction around a point mass is half or more from 0.15 to 0.3 r_H. Between 0.325 and 0.45 r_H it swings between about a fifth and two-thirds, and at 0.4 r_H only about a fifth to a quarter are lost, so most polar orbits survive the run; past the escape edge it rises again as orbits escape. With ν²/Ω² = 1 the Sun's lost fraction is about a fifth to a half from 0.325 to 0.45 r_J. With the Galaxy's vertical tide the Sun loses half or more of its polar orbits at every radius from 0.1 to 0.5 r_J (exactly half at 0.35 r_J across the host). These 32-angle fractions carry sampling noise of about ±0.1. -- **Other directions.** For a planet, plunges reach to about 8°, 10° and 14° from polar at 0.2, 0.3 and 0.4 r_H, counting both senses of tilt. On the prograde side, tilted to 75° (15° from polar; 32 angles), none plunge from 0.1 to 0.4 r_H. For the Sun plunges reach to about 8° from polar at 0.2 r_J and 24° at 0.3 r_J. At 75° on the prograde side, none plunge out to 0.25 r_J, up to 16% do at 0.3 r_J and up to 44% at 0.4 r_J. +- **Not everywhere outside.** With 32 starting angles, the lost fraction around a point mass is half or more from 0.15 to 0.3 r_H. Between 0.325 and 0.45 r_H it swings between about a fifth and two-thirds, and at 0.4 r_H a fifth to a quarter are lost (about a quarter with 512 angles, #37's re-review), so most polar orbits survive the run; past the escape edge it rises again as orbits escape. With ν²/Ω² = 1 the Sun's lost fraction is about a fifth to a half from 0.325 to 0.45 r_J. With the Galaxy's vertical tide the Sun loses half or more of its polar orbits at every radius from 0.1 to 0.5 r_J (exactly half at 0.35 r_J across the host). These 32-angle fractions carry sampling noise of about ±0.1, and up to 0.2. +- **Other directions.** For a planet, plunges reach 8–10°, 10–12° and 14–16° from polar at 0.2, 0.3 and 0.4 r_H (between the last tilt with plunges and the first without, on 2° steps), counting both senses of tilt. On the prograde side, tilted to 75° (15° from polar; 32 angles), none plunge from 0.1 to 0.4 r_H. For the Sun plunges reach 8–12° from polar at 0.2 r_J and 24–28° at 0.3 r_J (4° steps). At 75° on the prograde side, none plunge out to 0.25 r_J, up to 16% do at 0.3 r_J and up to 44% at 0.4 r_J. - **All directions.** Facing every way, up to 23% of the Sun's orbits plunge within 20 turns, and 9–11% of a planet's between 0.2 and 0.4 r_H. -- **What the averaged tide explains.** For comets, the tide driving them inward (Heisler and Tremaine 1986) accounts for the near-polar core: at 0.2 r_J all the Sun's orbits within 4° of polar plunge. The rest of its cone, beyond 5.1° at 0.2 r_J (out to 8°) and out to 24° at 0.3 r_J, and its plunges at 75° are not secular: #37's re-review found that the averaged tide limits plunges to within 5.1° of polar at 0.2 r_J and 3.6° at 0.4 r_J, and gives none at 75°. These plunges depend steeply on the plunge radius: counted inside 10⁻² r_J, the largest plunged fraction over all directions rises from 0.27 to 0.43 (24 directions). +- **What the averaged tide explains.** For comets, the tide driving them inward (Heisler and Tremaine 1986) accounts for the near-polar core: at 0.2 r_J all the Sun's orbits within 4° of polar plunge. The rest of its cone, beyond 5.1° at 0.2 r_J (out to 8–12°) and out to 24–28° at 0.3 r_J, and its plunges at 75° are not secular: #37's re-review found that the averaged tide limits plunges to within 5.1° of polar at 0.2 r_J and 3.6° at 0.4 r_J, and gives none at 75°. These plunges depend steeply on the plunge radius: counted inside 10⁻² r_J, the largest plunged fraction over all directions rises from 0.27 to 0.43 (24 directions). **For a planet.** - **Against the published limits.** The model's in-plane edges are 0.46 (prograde) and 0.775 to 0.79 (retrograde), against Nesvorný et al.'s (2003) semi-major axes of about 0.4 and 0.7 r_H for nearly circular orbits near the ecliptic, as Shen and Tremaine (2008) quote them. @@ -288,7 +288,7 @@ what the second review checked or calculated, in an arXiv source or against Cros - **Plunge radius.** It is 10⁻³ r_J. Raising it to 10⁻² moves the Sun's all-directions escaped edge out, from 0.437 to 0.460, and its lost edge in, from 0.411 to 0.394 (24 directions). - **Chaos and sampling.** Single orbits near an edge are chaotic: changing Γ by 10⁻⁴ either way moves the Sun's in-plane retrograde edge from 0.925 to 0.95, as one orbit of eight changes its fate. (Changing ν²/Ω² by 10⁻⁴ leaves the Sun's polar-across edge at 0.40 in this sample.) With 8 starting angles the in-plane edges carry sampling errors of up to about 0.03 and the polar escape edges up to about 0.025 (see the table's resolution above). - **Longer runs** move the edges inward a little (see the 100-turn values above). -- **MWPotential2014 is one model among many.** Its halo sets the flat range, and it has no nuclear star cluster. The checks pin its halo, its Kuzmin and Plummer limits pin the disk's a and b to galpy's values, and Table 1's derived values (bulge and disk masses, the mass inside 60 kpc, the slope at R₀ and the local dark-matter density) are checked at Table 1's rounding. The bulge's slope and cut-off are taken from galpy's source and not tested: its mass follows from its 5% share of the force at R₀. The Galactic-centre numbers use the separate power-law cusp above, which is valid only inside its 3 pc break. +- **MWPotential2014 is one model among many.** Its halo sets the flat range, and it has no nuclear star cluster. The checks pin its halo, its Kuzmin and Plummer limits pin the disk's a and b to galpy's values, and Table 1's derived values (bulge and disk masses, the mass inside 60 kpc, the slope at R₀ and the local dark-matter density) are checked at Table 1's rounding. The bulge's slope and cut-off are taken from galpy's source and not tested: its mass follows from its 5% share of the force at R₀, and the bulge-mass check, at Table 1's rounding, pins that share only to about 10%. The Galactic-centre numbers use the separate power-law cusp above, which is valid only inside its 3 pc break. - **Spherical mass estimates.** v²R/G and 3Ω²/4πG treat a flattened galaxy as spherical. In MWPotential2014 this overstates the mass inside 8 kpc by 12%. It shifts the 3.7-fold growth to 4.0 and the Sun's 2.25 to 2.53, but not the contrast with Kepler's fixed mass. ## Sources @@ -306,7 +306,7 @@ values in an arXiv source, a reference matched against Crossref where the entry - J. Binney and S. Tremaine, *Galactic Dynamics*, 2nd ed., Princeton (2008). The formula, as Jiang and Tremaine use it; section not confirmed. - J. Bovy, *Galactic rotation in Gaia DR1*, MNRAS 468, L63 (2017). https://doi.org/10.1093/mnrasl/slx027. #37's review. - J. A. Correa-Otto and M. F. Calandra, *Stability in the most external region of the Oort Cloud: evolution of the ejected comets*, MNRAS 490, 2495 (2019). https://doi.org/10.1093/mnras/stz2671. #37's review (sect. 4.3). -- Y. Shen and S. Tremaine, *Stability of Distant Satellites of the Giant Planets in the Solar System*, AJ 136, 2453 (2008). https://doi.org/10.1088/0004-6256/136/6/2453. #37's review. +- Y. Shen and S. Tremaine, *Stability of the Distant Satellites of the Giant Planets in the Solar System*, AJ 136, 2453 (2008). https://doi.org/10.1088/0004-6256/136/6/2453. #37's review; title as Crossref gives it (#37's re-review). - D. Nesvorný, J. L. A. Alvarellos, L. Dones and H. F. Levison, *Orbital and collisional evolution of the irregular satellites*, AJ 126, 398 (2003). As Shen and Tremaine quote it; the reference matches Crossref (#37's re-review), the text was not read. - M. L. Lidov, Planet. Space Sci. 9, 719 (1962), and Y. Kozai, AJ 67, 591 (1962). The references match Crossref (#37's re-review); the texts were not read. - C. F. McKee, A. Parravano and D. J. Hollenbach, *Stars, gas, and dark matter in the solar neighborhood*, ApJ 814, 13 (2015). https://arxiv.org/abs/1509.05334. #37's review. diff --git a/orbits/model.js b/orbits/model.js index c4efc05..12e28c8 100644 --- a/orbits/model.js +++ b/orbits/model.js @@ -288,7 +288,8 @@ // once r < rIn (1e-3 r_J is 0.74 of Jupiter's radius; for the Sun it is 283 au, outside the planets), otherwise // 'stays'. Returns the final state s too. Throws on non-finite settings, on a non-finite step, or after maxSteps // (by default 2.5e5 per turn of the host; the hardest orbit in the tables, a polar orbit at 0.025 r_J that stays, - // needs about 7e4 per turn), rather than looping. + // needs about 7e4 per turn; steps per turn grow as a^-1.5, so the cap would fire only below about 0.012 r_J, under + // the smallest radius scanned here), rather than looping. function hillOrbit(host, s0, { orbits = 20, rEsc = 3, rIn = 1e-3, tol = 1e-10, maxSteps = 2.5e5 * orbits } = {}) { if (![orbits, rEsc, rIn, tol, maxSteps, host.Gam, host.nu2, ...s0].every(Number.isFinite)) throw new Error('hillOrbit: non-finite input'); const T = 2 * Math.PI * orbits, k = Array.from({ length: 7 }, () => new Float64Array(6));