diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml index 3baf87d..08299ab 100644 --- a/.github/workflows/test.yml +++ b/.github/workflows/test.yml @@ -34,3 +34,4 @@ jobs: - run: node turbulence/table.mjs quick - run: node turbulence/edges.mjs quick - run: node turbulence/edges.mjs 256 + - run: node halos/table.mjs diff --git a/CORRECTIONS.md b/CORRECTIONS.md index ce676fb..7403695 100644 --- a/CORRECTIONS.md +++ b/CORRECTIONS.md @@ -187,3 +187,31 @@ Claims made during the exploration that did not survive it, kept so the record s See [cosmic-homogeneity](cosmic-homogeneity/). 22. **"The nesting stops, the way a coastline's roughness stops at the size of molecules."** Points the wrong way. A coastline is rough down to a smallest scale; galaxy clustering is clumpy up to a largest one. + +## 19 October 2026: halos within halos (issue #18) + +23. **"Halos sit inside halos with nearly the same shape from about an Earth mass up to galaxy clusters, some 20 + decades in mass (Wang et al. 2020)."** Two slips, the second partly contested. Wang et al. zoomed eight times, each + time into a smaller region of the previous level chosen to avoid any massive halo: their 20 decades are field + halos of many masses, not a chain of halos inside halos. Inside a Milky Way the nesting of a 100 GeV WIMP spans + 16 to 18 decades: at most 9 to 11 levels at the ratio the Aquarius subhalo law gives (masses of all the matter; + counting the dark matter alone, the 9 becomes 10), and 7 to 9 if sub-subhalos are 2 or 3 times fewer, as + Aquarius finds (it resolves four levels itself). And halos near the cut-off form with steeper ρ ∝ r^−1.5 + centres. Whether those survive inside later halos is contested: Wang et al. find NFW and Einasto profiles down to + an Earth mass, Delos and White suggest the steeper centres lie at radii those simulations do not resolve, and a + later study finds them buried in NFW-like profiles but not destroyed. See [halos](halos/). +24. **"Hu–Barkana–Gruzinov's quoted k_J ≈ 66.5 a^¼ Mpc⁻¹ and M_J ≈ 1.5×10⁷ M☉ for 10⁻²² eV are not mutually + consistent with ρ = ρ_m."** Neither number is Hu, Barkana and Gruzinov's (2000), who give the Jeans + wavenumber at equality, 9 m₂₂^½ Mpc⁻¹, and a Jeans length. The 66.5 form is Marsh's (2016) and uses the dark + matter density alone, 0.12 ρ_crit/h². 1.5×10⁷ M☉ is Hui et al.'s (2017, as quoted in search text), and a sphere + of diameter λ_J reproduces it. With the dark matter density the two agree (66.5 Mpc⁻¹ and 1.47×10⁷ M☉); with + all matter they are 69.5 Mpc⁻¹ and 1.54×10⁷ M☉. Both go as ρ^¼, 4.5% between the two densities: quoted to two + figures, 1.5×10⁷ hides the difference, while 66.5, quoted to three, does not. +25. **"About 18 decades for a 100 GeV WIMP and only 2 to 4.5 for 10⁻²² eV wave dark matter."** The WIMP figure + survives: 16 to 18 decades. The wave figure mixes definitions. Where the halo population turns down (half + mode, half power, Schive et al.'s M₀) it is 1.3 to 1.8 decades; 4.8 is from the Jeans mass today, which bounds + any halo from below but is not where halos become rare. +26. **"Excluded by Lyman-α and Milky Way satellite bounds (from about 2×10⁻²¹ to 3×10⁻¹⁹ eV)."** Out of date at + the top: Ursa Major III, if it is a galaxy, would raise the dwarf-galaxy bound from 3×10⁻¹⁹ to 8×10⁻¹⁸ eV (May, + Dalal and Kravtsov 2025). The issue's "at 2×10⁻²⁰ eV the suppression mass falls to about 10⁷ M☉" holds for + Schive et al.'s M₀ (1.4×10⁷ M☉); the half-mode and half-power masses are 3.4 and 4.6×10⁷ M☉. diff --git a/README.md b/README.md index 1931581..c7dd25c 100644 --- a/README.md +++ b/README.md @@ -40,6 +40,13 @@ with Claude (Anthropic), for [Foundation0](https://github.com/Foundation-0/f0). continues. The fractal edges of the Taurus clouds (dimension 1.36) agree in number with laboratory turbulent surfaces, but the two measure different things, and across clouds the dimension runs from about 1.2 to 1.6 ([turbulence](turbulence/)). +- **Halos nest inside halos, down to a floor the dark matter particle sets.** Dark matter clumps from the + bottom up, and a Milky Way's halo holds subhalos that hold their own. For a 100 GeV WIMP the nesting spans + 16 to 18 decades in mass, down to halos of 10⁻⁶ to 10⁻⁴ Suns; below that, the particle's coupling to the + radiation, and then its own motion, have smoothed everything out. For wave dark matter at the lightest masses + still allowed it spans only 3 to 8 decades, depending on which bound holds, cut off by quantum pressure. + The first halos form with steeper centres; whether those survive inside the halos built from them is still + argued ([halos](halos/)). Some repetition across scales has to be tuned, and some sets its own range. At a critical point, such as water's liquid-vapour critical point or the Curie temperature of a magnet whose spins point only up or down, fluctuations diff --git a/halos/README.md b/halos/README.md new file mode 100644 index 0000000..c1bd654 --- /dev/null +++ b/halos/README.md @@ -0,0 +1,288 @@ +# Halos within halos: how deep the nesting goes, and what stops it + +Status: **exploration, not peer reviewed, no new result claimed.** Issue +[#18](https://github.com/Foundation-0/fractals/issues/18). + +Dark matter clumps from the bottom up: the smallest halos form first, merge into larger ones, and many survive +inside them as subhalos, which hold sub-subhalos of their own. The nesting cannot go on forever. Below some mass +the dark matter particle itself has smoothed the universe out, and no halo forms. This study computes that floor +for a WIMP and for wave (fuzzy) dark matter from the linear power spectrum. It counts how many levels of nesting +fit between the floor and a Milky Way, and asks whether the halos stop looking alike before the floor is reached. + +## The answer + +| Dark matter | Smallest halos (M☉) | Decades below a 10¹² M☉ Milky Way | Levels at the Aquarius ratio (an upper bound) | +|---|---|---|---| +| 100 GeV WIMP (bino), kinetic decoupling at 40 to 10 MeV | 1.4×10⁻⁶ to 1.0×10⁻⁴ | 17.9 to 16.0 | 11 to 9 | +| WIMPs in general (Bringmann 2009; his 10⁻¹¹ to 10⁻³ counts the dark matter alone) | 1.2×10⁻¹¹ to 1.2×10⁻³ | 23 to 15 | 14 to 9 | +| Wave, 10⁻²² eV (the lensing notes' benchmark, excluded) | 1.6 to 5.4×10¹⁰ | 1.8 to 1.3 | 1 to 0 | +| Wave, 2 to 2.9×10⁻²¹ eV (Lyman-α; Milky Way satellites) | 1.8 to 9.9×10⁸ | 3.7 to 3.0 | 2 to 1 | +| Wave, 2×10⁻²⁰ eV (Lyman-α, Rogers and Peiris 2021) | 1.4 to 4.6×10⁷ | 4.9 to 4.3 | 3 to 2 | +| Wave, 3×10⁻¹⁹ eV (dwarf galaxies, Dalal and Kravtsov 2022) | 3.7×10⁵ to 1.2×10⁶ | 6.4 to 5.9 | 4 to 3 | +| Wave, 8×10⁻¹⁸ eV (if Ursa Major III is a galaxy, 2025) | 4.6×10³ to 1.6×10⁴ | 8.3 to 7.8 | 5 to 4 | + +Each wave mass is the lower bound of the probe named (as seen in search text; checked in review); each assumes the +wave is all of the dark matter. The ranges come from the definition of "smallest" (below) and, for the bino, from +the decoupling temperature. A level is a factor μ = 0.025 in mass: where the subhalo law fitted to the Aquarius +simulation expects the largest subhalo of a host, N(>μM_host) = 1. Using that ratio at every level is the +self-similar case, which Springel et al. (2008) call an upper limit: they find sub-subhalos 2 or 3 times fewer. +With that at every level below the first, the bino has 9 to 8 or 8 to 7 levels, Bringmann's range 11 to 7 or 10 to +7, and no wave row loses more than one. μ is measured against the host's M200; against M50, the mass inside the +radius where the law was fitted, it is 0.018, and counted down from M50 the levels drop by up to one. The bino's 9 +at 10 MeV is a knife edge between 9 and 10 (below). At 10× per level the bino has 15 to 17 levels; at 100×, 7 or 8. + +So the two kinds of dark matter stop nesting eleven decades or more apart at the Lyman-α bounds, and eight to ten +apart at the strongest bound proposed, 8×10⁻¹⁸ eV. Wave dark matter at the benchmark mass hardly nests at all: a +Milky Way's largest expected subhalo (2.5×10¹⁰ M☉) is already near the floor. + +## The WIMP floor + +Two effects erase the smallest scales of a WIMP's spectrum (Green, Hofmann and Schwarz 2004, 2005; Loeb and +Zaldarriaga 2005; Bertschinger 2006): + +- **Acoustic damping.** Until kinetic decoupling, at T_kd, the WIMPs scatter off the radiation and move with its + sound waves, which damp out everything inside the horizon. The matter inside the comoving horizon at T_kd is + M_ao = 1.03×10⁻⁴ M☉ at 10 MeV, falling roughly as T_kd⁻³. Loeb and Zaldarriaga give 10⁻⁴ (T_d/10 MeV)⁻³ M☉. + The expansion rate comes from the radiation then: photons, neutrinos and e± (g = 10.75), joined above about + 10 MeV by muons and pions (g = 12.6 at 30 MeV). +- **Free streaming.** After decoupling the WIMPs coast. A 100 GeV WIMP decoupling at 30 MeV has a speed spread + √(T_kd/m) and coasts s = 0.84 pc (comoving, through e± annihilation and on to today), so modes shorter than + k_fs = √2/s = 1.68×10⁶ Mpc⁻¹ are smeared out. Green, Hofmann and Schwarz give 1.70×10⁶ Mpc⁻¹, 1.4% higher; from + 10 to 40 MeV theirs runs from 5.4% higher to 1.1% lower, as the radiation's g changes. The floors below use + Green, Hofmann and Schwarz's k_fs and its T_kd scaling; the derived value is a check on it. + +The floor is the larger of the two (Bringmann 2009). For a bino decoupling at 10 to 40 MeV the acoustic mass is +larger than the free-streaming mass M(π/k_fs) throughout, and comparable to the half-mode mass at 30 to 40 MeV. +Across WIMP models Bringmann finds floors from about 10⁻¹¹ to nearly 10⁻³ M☉. He writes both masses with the dark +matter density ρ_χ, as Cornell, Profumo and Shepherd (2013) write the acoustic mass. Every mass here counts all the +matter, 1.19 times more (0.07 decades), so the answer table converts his range. In his convention the bino at 10 +MeV has 10.01 levels at μ instead of 9.97, so its count sits on the edge between 9 and 10. + +**"About 10⁻⁶ M☉" is a convention.** The same cut-off gives 1.05×10⁻⁶ M☉ for a sphere of radius π/k_fs = 1.85 pc, +which is how Green, Hofmann and Schwarz define their "about 10⁻⁶ M☉" (2004 and 2005, as #35's review read them; +9×10⁻⁷ for their 2004 model A). Their "comoving radius of about 1 pc" agrees with it only to a factor 2 (a sphere +of radius 1 pc holds 1.7×10⁻⁷ M☉). The half-mode mass, where the transfer function falls to 1/2, is 4.3×10⁻⁶; the +half-power mass, where it falls to 1/√2, is 1.2×10⁻⁵. That is a factor of 11, one decade of the answer. The +transfer function, T = [1 − (2/3)(k/k_fs)²] exp[−(k/k_fs)² − (k/k_d)²] (in Green, Hofmann and Schwarz's JCAP paper, +the equations labelled dfs and dk in its arXiv source), is stated for k < k_fs, and both lie inside that range +(0.625 and 0.45 k_fs). A plain Gaussian with the same k_fs would move the half-mode mass from 4.3 to 1.8×10⁻⁶ M☉. + +## The wave floor + +A wave of mass m cannot be squeezed below its Jeans length: quantum pressure beats gravity for wavenumbers above +k_J = (16πGρ)^¼ (m/ħ)^½. The cut-off this imprints on the spectrum (Hu, Barkana and Gruzinov 2000) falls to half +power at k_½ = 4.58 m₂₂^(4/9) Mpc⁻¹ (they quote 4.5), with m₂₂ = m/10⁻²² eV. + +**Which mass is "the smallest wave halo"** matters more here than for WIMPs. At 10⁻²² eV: + +| Definition | Mass (M☉) | +|---|---| +| Half power, T² = 1/2 | 5.4×10¹⁰ | +| Half mode, T = 1/2 | 3.9×10¹⁰ | +| Schive et al. (2016): the halo mass function turns down at M₀ = 1.6×10¹⁰ m₂₂^(−4/3) | 1.6×10¹⁰ | +| Jeans mass today, a sphere of diameter λ_J (Hui et al. 2017's sphere, here in all the matter) | 1.5×10⁷ | +| Jeans mass at matter-radiation equality, same sphere | 6.9×10⁹ | +| Jeans mass today, other spheres: radius λ_J, radius 1/k_J | 1.2×10⁸, 5.0×10⁵ | + +The first three mark where halos become rare, and the table above uses them. The Jeans mass today is a lower +bound on any halo, not the floor of the population, which Schive et al.'s fit already suppresses steeply below M₀, +three decades higher. The Jeans masses alone span a factor 10⁵ between definitions and epochs. They scale as +m^(−3/2), the population floor as m^(−4/3). At 2×10⁻²⁰ eV the population floor is 1.4 to 4.6×10⁷ M☉. + +Every row at a bound is an extrapolation. Hu, Barkana and Gruzinov's cut-off was computed near 10⁻²² eV, and Schive +et al. fitted M₀ to runs at 0.8 to 3.2×10⁻²² eV and flag it as uncertain below 10⁹ M☉ (as #35's review read them). +At 2×10⁻²⁰ eV the particle is about 60 times heavier than the heaviest fitted. There Rogers and Peiris quote a +half-mode mass of 7.2×10⁷ M☉, in a convention matched to warm dark matter (Drlica-Wagner et al. 2019); that is 2.1 +times the 3.4×10⁷ here. + +**The wave clock and the Jeans scale.** In the Witness Map's terms, the wave clock m·Δx²/ħ at Δx = 1/k_J equals +1/(4√(πGρ)), gravity's own clock: the Jeans condition restated, an identity for any constants. At Δx = λ_J/2 = +π/k_J the clock is π² longer, so the factor is a convention. The often-quoted k_J = 66.5 a^¼ m₂₂^½ Mpc⁻¹ (Marsh +2016; the equation labelled axion_jeans in the arXiv source) uses the dark matter density alone (0.12 ρ_crit/h²); +all matter gives 69.5. Hui et al.'s M_J = 1.5×10⁷ m₂₂^(−3/2) M☉ (the equation labelled jeans) also uses the wave's +own density, with Ω = 0.27 and h = 0.7: there this model gives 1.505×10⁷. At Planck 2018 it gives 1.47×10⁷ in the +dark matter alone and 1.54×10⁷ in all the matter. + +**The smallest wave structure: the soliton.** At the centre of every wave halo the wave settles into its ground +state, the soliton (the Witness Map's `SOL`, which solves the Schrödinger–Poisson equations). Mass times size is +fixed: M r_c = 2.68 ħ²/(Gm²), with r_c the half-density radius, and the mass inside r_c times r_c is +0.633 ħ²/(Gm²). That is 5.4×10¹² (m/10⁻²³ eV)⁻² M☉ pc. Schive, Chiueh and Broadhurst's core mass-radius relation +(their Supplementary Information), M_c ≈ 5.5×10⁹ (m/10⁻²³ eV)⁻² (r_c/kpc)⁻¹ M☉, gives 5.5×10¹²; their fitted +profile, integrated to r_c, gives 5.3×10¹². A gravitational atom (the same wave around a point mass, see +[gravitational-atom](../gravitational-atom/)) has M r_½ = (ln 2/2) ħ²/(Gm²) = 0.347 ħ²/(Gm²) with the same +combination. The soliton is that atom with its own mass for a nucleus, spread through it, and 7.7 times larger. + +## How many levels + +Springel et al.'s Aquarius halo (2008) follows dN/dM = a₀ (M/m₀)^−1.9 above about 10⁵ M☉. Extended down to each +floor, the law gives the counts in [`table.mjs`](table.mjs). Above 10⁻⁶ M☉ it gives 1.05×10¹⁵ subhalos in Aq-A +(1.84×10¹² M☉) and 5.7×10¹⁴ in a 10¹² M☉ host ([`tests/halos.mjs`](../tests/halos.mjs)), tides ignored. Diemand, +Moore and Stadel (2005) find "over 10¹⁵" that survive: Aq-A's count clears that by 5%, and the lighter host falls +short by a factor 1.75. Their own M^−1.9 law, extrapolated the same way, gives about 5×10¹⁵. Like for like, the +Aquarius law gives 4.8 times fewer for Aq-A and 8.8 times fewer for the 10¹² M☉ host; they give no host mass. The +levels are a separate count. Each level's largest subhalo, where the law expects one, is μ = 0.025 of its host (1.6 +decades per level), an upper bound (above). Aquarius itself resolves "up to four generations" of subhalos within +subhalos, and four levels at Aq-A's own ratio (0.027) reach 9×10⁵ M☉, within its resolved range. With sub-subhalos +2 or 3 times fewer, four levels reach 9×10⁴ to 2×10⁴ M☉, at or below the law's fitted range. + +Two cautions. The extrapolation runs 9 to 11 decades below where the law was fitted for the bino (8 to 16 across +Bringmann's WIMP range, at most 1.3 for the wave masses at the bounds listed), and assumes that what forms +survives. Whether microhalos survive in the Galaxy is disputed. Zhao et al. (2007) argue that encounters with stars +turn the ones near the Sun into streams; Moore et al. (2005) and Goerdt et al. (2007) find that their dense centres +survive; van den Bosch et al. (2018) argue that much of the disruption seen in simulations is numerical. The second +caution is that μ sets the levels, not the decades: at 10× per level the bino has 15 to 17, at μ 9 to 11, and +7 to 9 with fewer sub-subhalos. + +## Why small halos form almost together, and whether they stop looking alike + +On small scales the CDM spectrum's effective slope approaches −3, where every mass would collapse at once. It gets +there only logarithmically: n_eff = −2.15 at 10¹² M☉, −2.84 at 10⁻⁶ M☉. So σ(M), the typical fluctuation, grows +only elevenfold over those eighteen decades (2.2 to 24), and without a cut-off halos from 10⁻¹² to 10⁻³ M☉, nine +decades, all collapse within 157 million years (z = 24 to 14). + +For the bino the first halos (1σ, half-mode mass) collapse at z ≈ 16, or 11 to 13 once the dark matter grows +alone: the gas never falls into wells this small, which the growth factor assumes it does after the drag epoch. + +Green, Hofmann and Schwarz's z = 60 ± 20 (JCAP 2005, section "The first structures") is a different quantity. It is +z_nl, the redshift at which typical fluctuations reach σ(R, z_nl) = 1, with a top-hat window, σ₈ = 0.9, n = 1 and +the baryons ignored; they put Nσ peaks at about 60N. The same definition here, at M_hm to M(π/k_fs), gives z_nl = +28 to 29 with Planck 2018, 39 with WMAP-era round numbers (h = 0.7, Ω_m = 0.3, n_s = 1, σ₈ = 0.9), and 44 to 45 +with those numbers and the baryons removed, inside their range. A factor of about 1.3 in 1 + z remains. At this +model's own threshold, 1.686, the WMAP-era numbers collapse 1σ and 2σ peaks at z = 22.5 and 46, and with Planck +2018 collapse at z = 60 takes a 3.5σ peak (reaching δ = 1 there, their threshold, takes 2.1σ). These use the growth +factor without the baryon lag, as Green, Hofmann and Schwarz ignore the baryons; with the lag, z = 60 takes 3.7 to +4.4σ. + +Halos near the cut-off form with a different shape: steeper ρ ∝ r^−1.5 centres, prompt cusps, not the r^−1 centres +of larger halos (Ishiyama 2014; Angulo et al. 2017; Delos and White 2023). Phase-space simulations free of +artificial fragmentation find them too (Ondaro-Mallea et al. 2024, as seen in search text). The inner slope is +still about −1.3 at fifty times the cut-off mass (Ishiyama 2014). Whether the shapes stay different inside the +halos built from them is contested. Delos and White find that the cusps survive, at the smallest radii the +simulations resolve; Angulo et al. find that major mergers flatten them. J. Wang et al. (2020) find NFW and Einasto +profiles, nearly universal over 20 decades of mass from an Earth mass to clusters, and Delos and White suggest the +cusps lie at radii those simulations do not resolve. Y. Wang, Bose, Frenk and Jenkins (2025, as seen in search +text) find the whole profile turning NFW-like while the cusp stays visible in the particles of the original peak: +"buried but not destroyed". J. Wang et al. zoomed eight times, each time into a smaller region of the previous +level chosen to avoid any massive halo: their 20 decades are field halos of many masses, not a chain of halos +inside halos. + +## The spectrum twenty decades down + +σ(M) and the collapse redshifts sit on Eisenstein and Hu's (1998) fit, extrapolated twenty decades in mass below +where it was made. As a check, `model.js` integrates the radiation-era perturbations itself (Ma and Bertschinger +1995). A mode entering the horizon kicks the dark matter logarithmically, δ_c → −A ln(B kτ), and Mészáros growth +then gives the transfer function's tail. With the neutrinos streaming freely, as in the Boltzmann code Eisenstein +and Hu fitted, their tail agrees with this one to 1.5% from 10³ to 10⁹ h/Mpc. The bino's modes, though, entered +the horizon before the neutrinos decoupled, at about 1 MeV, so the neutrinos were still a fluid. That tail is 10% +lower. It lowers σ at the bino half-mode mass from 23.0 to 22.3 and z_c from 16.3 to 15.8, and leaves the floors +where they are: they come from the cut-off, not from the spectrum's amplitude. + +## Run it + +``` +node halos/table.mjs # the tables above and more, under a second +npm test # all studies' checks, including tests/halos.mjs +``` + +`model.js` needs [`cosmic-homogeneity/homogeneity.js`](../cosmic-homogeneity/) for the transfer function and +growth factor. The tests and table also load [`witness-map/model.js`](../witness-map/) for its constants and its +soliton. + +## Checks + +`tests/halos.mjs` compares the model with published values, exact results and its own numerics: + +- the free-streaming wavenumber and the cut-off and collisional damping masses of Green, Hofmann and Schwarz; +- the acoustic mass against Loeb and Zaldarriaga (within ×1.5) and, in the dark matter alone, against Cornell, + Profumo and Shepherd, who count only that (within ×2; it is 1.24 times theirs); +- Hu, Barkana and Gruzinov's Jeans wavenumber at equality and half-power wavenumber, the 66.5 Mpc⁻¹ form with the + density it needs, and Hui et al.'s Jeans mass at their normalisation (within 1%); +- the Jeans wavenumber found by bisection against its closed form (a check of the numerics in `jeans()`); +- the soliton's M r_c against Schive, Chiueh and Broadhurst's core mass-radius relation; +- the Aquarius normalisation scaled from Aq-A to a 10¹² M☉ host against the quoted one (a guard on the transcribed + constants); +- Aq-A's count above 10⁻⁶ M☉ against the "over 10¹⁵" that Diemand, Moore and Stadel find survive (a 5% margin); +- the radiation-era kick against its closed form, and Eisenstein and Hu's tail against the derived one; +- σ(M) against a direct quadrature, and its convergence in the grid. + +Four comparisons are notes, not checks: the fluid-neutrino tail; Aquarius's four generations; Diemand, Moore and +Stadel's own extrapolated total, about 5×10¹⁵ above 10⁻⁶ M☉, which the Aquarius law falls short of by 4.8 to 8.8 +times; and the onset of nonlinearity for typical fluctuations against Green, Hofmann and Schwarz's z = 60 ± 20. + +## What the model leaves out + +- Linear theory and spherical collapse throughout. No halo mass function near the cut-off: simulations find + subhalos strongly suppressed below the free-streaming scale (Ishiyama and Ando 2020), and how the mass + function turns over against the free-streaming scale is studied by Schneider, Smith and Reed (2013). +- The acoustic mass is the matter (dark matter and baryons, the convention for every mass here) inside the horizon, + an order-of-magnitude scale; the dark matter alone, which Bringmann and Cornell et al. use, is 16% less. + Bringmann (2009) finds that working out T_kd properly puts a given model's cut-off typically about ten times + above earlier order-of-magnitude estimates (as seen in search text; checked in review); here T_kd is an input, so + that uncertainty enters through it. +- The thermal history is ideal gases in equilibrium, with the neutrinos decoupling at once at 1 MeV. It leaves out + kaons and heavier hadrons, which would add about 1% to g at 70 MeV and 7% at 100 MeV (#35's review's own + calculation agreed); it is used here only up to 50 MeV (0.1%). The particle masses in it are recalled. +- The spectrum has no running of the spectral index: a running of −0.01 / +0.01 moves the bino's σ by −29% / +49% + and z_c to 11 / 25. Massive neutrinos and the QCD transition's imprint on modes above about 10⁶ Mpc⁻¹ are + ignored. +- Whether the matter's or the dark matter's contrast must reach 1.686 when the gas stays smooth is left open, + hence 11 to 13 for the bino's first halos. +- The subhalo law is extrapolated over up to 16 decades (9 to 11 for the bino) with no tidal losses, and the + levels assume the host's μ at every level, an upper bound. +- The wave fits are of N-body simulations with wave initial conditions (Schive et al. 2016) or of the linear + spectrum. The first halo mass function from a fully self-consistent wave simulation (May and Springel 2023: + 7×10⁻²³ eV, z = 3, 68 haloes) agrees broadly with Schive et al.'s fit, but has too few haloes to test M₀ precisely. + +## Sources + +The authors read published values in abstracts and search results ("as seen in search text" in the checks and +code), not in the papers' full texts, except where marked recalled. The independent review of #35 then read the +primary sources in their arXiv LaTeX and confirmed those values ("checked in review"; Eisenstein and Hu's fitting +range was not checked). Some values come from that reading alone ("as #35's review read it"): Green, Hofmann and +Schwarz's transfer-function prefactor (the JCAP equations labelled dfs and dk), cut-off mass definition, damping +mass and z_nl; Bringmann's and Cornell et al.'s use of the dark matter density; Diemand, Moore and Stadel's +extrapolated 5×10¹⁵; Springel et al.'s sub-subhalo counts; Wang et al.'s method; Hui et al.'s (equation label +jeans) and Marsh's (axion_jeans) normalisations; Schive, Chiueh and Broadhurst's core mass-radius relation; the +runs Schive et al. fitted M₀ to (MFFit); May and Springel's result; and Rogers and Peiris's half-mode mass. Aq-A's +M50 and Schive, Chiueh and Broadhurst's fitted profile (in `table.mjs`) are recalled. + +- A. M. Green, S. Hofmann and D. J. Schwarz, *The power spectrum of SUSY-CDM on sub-galactic scales*, MNRAS 353, L23 (2004). https://arxiv.org/abs/astro-ph/0309621 +- A. M. Green, S. Hofmann and D. J. Schwarz, *The first WIMPy halos*, JCAP 0508, 003 (2005). https://arxiv.org/abs/astro-ph/0503387 +- A. M. Green, S. Hofmann and D. J. Schwarz, *Small scale WIMP physics*, AIP Conf. Proc. 805, 431 (2006). https://arxiv.org/abs/astro-ph/0508553 +- A. Loeb and M. Zaldarriaga, *The small-scale power spectrum of cold dark matter*, Phys. Rev. D 71, 103520 (2005). https://arxiv.org/abs/astro-ph/0504112 +- E. Bertschinger, *Effects of cold dark matter decoupling and pair annihilation on cosmological perturbations*, Phys. Rev. D 74, 063509 (2006). https://arxiv.org/abs/astro-ph/0607319 +- T. Bringmann, *Particle models and the small-scale structure of dark matter*, New J. Phys. 11, 105027 (2009). https://arxiv.org/abs/0903.0189 +- J. M. Cornell, S. Profumo and W. Shepherd, *Kinetic decoupling and small-scale structure in effective theories of dark matter*, Phys. Rev. D 88, 015027 (2013). https://arxiv.org/abs/1305.4676 +- J. Diemand, B. Moore and J. Stadel, *Earth-mass dark-matter haloes as the first structures in the early Universe*, Nature 433, 389 (2005). https://arxiv.org/abs/astro-ph/0501589 +- V. Springel et al., *The Aquarius Project: the subhaloes of galactic haloes*, MNRAS 391, 1685 (2008). https://arxiv.org/abs/0809.0898 +- D. Erkal et al., *The number and size of subhalo-induced gaps in stellar streams*, MNRAS 463, 102 (2016), for the Aquarius fit's parameters. https://arxiv.org/abs/1606.04946 +- J. Wang et al., *Universal structure of dark matter haloes over a mass range of 20 orders of magnitude*, Nature 585, 39 (2020). https://arxiv.org/abs/1911.09720 +- T. Ishiyama, *Hierarchical formation of dark matter halos and the free streaming scale*, ApJ 788, 27 (2014). https://arxiv.org/abs/1404.1650 +- T. Ishiyama and S. Ando, *The abundance and structure of subhaloes near the free streaming scale and their impact on indirect dark matter searches*, MNRAS 492, 3662 (2020). https://arxiv.org/abs/1907.03642 +- R. E. Angulo, O. Hahn, A. D. Ludlow and S. Bonoli, *Earth-mass haloes and the emergence of NFW density profiles*, MNRAS 471, 4687 (2017). https://arxiv.org/abs/1604.03131 +- M. S. Delos and S. D. M. White, *Inner cusps of the first dark matter haloes: formation and survival in a cosmological context*, MNRAS 518, 3509 (2023). https://arxiv.org/abs/2207.05082 +- M. S. Delos and S. D. M. White, *Prompt cusps and the dark matter annihilation signal*, JCAP 10, 008 (2023). https://arxiv.org/abs/2209.11237 +- L. Ondaro-Mallea et al., *Phase-space simulations of prompt cusps: simulating the formation of the first haloes without artificial fragmentation*, MNRAS 527, 10802 (2024). https://arxiv.org/abs/2309.05707 +- Y. Wang, S. Bose, C. S. Frenk and A. Jenkins, *Buried but not destroyed: the evolution from prompt cusps to NFW haloes* (2025). https://arxiv.org/abs/2512.13804 +- H. Zhao, D. Hooper, G. W. Angus, J. E. Taylor and J. Silk, *Tidal disruption of the first dark microhalos*, ApJ 654, 697 (2007). https://arxiv.org/abs/astro-ph/0508215 +- B. Moore, J. Diemand, J. Stadel and T. Quinn, *On the survival and disruption of Earth mass CDM micro-haloes* (2005). https://arxiv.org/abs/astro-ph/0502213 +- T. Goerdt, O. Y. Gnedin, B. Moore, J. Diemand and J. Stadel, *The survival and disruption of cold dark matter microhaloes: implications for direct and indirect detection experiments*, MNRAS 375, 191 (2007). https://arxiv.org/abs/astro-ph/0608495 +- F. C. van den Bosch, G. Ogiya, O. Hahn and A. Burkert, *Disruption of dark matter substructure: fact or fiction?*, MNRAS 474, 3043 (2018). https://arxiv.org/abs/1711.05276 +- A. Schneider, R. E. Smith, A. V. Macciò and B. Moore, *Non-linear evolution of cosmological structures in warm dark matter models*, MNRAS 424, 684 (2012), for the half-mode mass. https://arxiv.org/abs/1112.0330 +- A. Schneider, R. E. Smith and D. Reed, *Halo mass function and the free streaming scale*, MNRAS 433, 1573 (2013). https://arxiv.org/abs/1303.0839 +- W. Hu, R. Barkana and A. Gruzinov, *Fuzzy cold dark matter: the wave properties of ultralight particles*, Phys. Rev. Lett. 85, 1158 (2000). https://arxiv.org/abs/astro-ph/0003365 +- L. Hui, J. P. Ostriker, S. Tremaine and E. Witten, *Ultralight scalars as cosmological dark matter*, Phys. Rev. D 95, 043541 (2017). https://arxiv.org/abs/1610.08297; its Jeans mass as quoted by F. Toguz, D. Kawata, G. Seabroke and J. I. Read, MNRAS 511, 1757 (2022). https://arxiv.org/abs/2106.02526 +- D. J. E. Marsh, *Axion cosmology*, Phys. Rep. 643, 1 (2016). https://arxiv.org/abs/1510.07633; the 66.5 form as quoted by M. Kulkarni and J. P. Ostriker, *What is the halo mass function in a fuzzy dark matter cosmology?*, MNRAS 510, 1425 (2022). https://arxiv.org/abs/2011.02116 +- H.-Y. Schive, T. Chiueh and T. Broadhurst, *Cosmic structure as the quantum interference of a coherent dark wave*, Nature Phys. 10, 496 (2014), and its Supplementary Information. https://arxiv.org/abs/1406.6586; the 5.5×10¹² first seen restated by D. Wardana, K. Hayashi, M. Chiba and E. G. M. Ferreira, *Fuzzy dark matter and the impact of core–halo diversity on its particle mass constraints* (2026), which credits it to Phys. Rev. Lett. 113, 261302. https://arxiv.org/abs/2603.07175 +- H.-Y. Schive, T. Chiueh, T. Broadhurst and K.-W. Huang, *Contrasting galaxy formation from quantum wave dark matter, ψDM, with ΛCDM, using Planck and Hubble data*, ApJ 818, 89 (2016). https://arxiv.org/abs/1508.04621 +- S. May and V. Springel, *The halo mass function and filaments in full cosmological simulations with fuzzy dark matter*, MNRAS 524, 4256 (2023). https://arxiv.org/abs/2209.14886 +- V. Iršič, M. Viel, M. G. Haehnelt, J. S. Bolton and G. D. Becker, *First constraints on fuzzy dark matter from Lyman-α forest data and hydrodynamical simulations*, Phys. Rev. Lett. 119, 031302 (2017). https://arxiv.org/abs/1703.04683 +- E. Armengaud, N. Palanque-Delabrouille, C. Yèche, D. J. E. Marsh and J. Baur, *Constraining the mass of light bosonic dark matter using SDSS Lyman-α forest*, MNRAS 471, 4606 (2017). https://arxiv.org/abs/1703.09126 +- K. K. Rogers and H. V. Peiris, *Strong bound on canonical ultralight axion dark matter from the Lyman-alpha forest*, Phys. Rev. Lett. 126, 071302 (2021). https://arxiv.org/abs/2007.12705 +- A. Drlica-Wagner et al., *Probing the fundamental nature of dark matter with the Large Synoptic Survey Telescope* (2019), for the half-mode convention Rogers and Peiris use. https://arxiv.org/abs/1902.01055 +- E. O. Nadler et al. (DES), *Constraints on dark matter properties from observations of Milky Way satellite galaxies*, Phys. Rev. Lett. 126, 091101 (2021). https://arxiv.org/abs/2008.00022 +- N. Dalal and A. Kravtsov, *Excluding fuzzy dark matter with sizes and stellar kinematics of ultrafaint dwarf galaxies*, Phys. Rev. D 106, 063517 (2022). https://arxiv.org/abs/2203.05750 +- S. May, N. Dalal and A. Kravtsov, *Updated bounds on ultra-light dark matter from the tiniest galaxies* (2025). https://arxiv.org/abs/2509.02781 +- C.-P. Ma and E. Bertschinger, *Cosmological perturbation theory in the synchronous and conformal Newtonian gauges*, ApJ 455, 7 (1995). https://arxiv.org/abs/astro-ph/9506072 +- D. J. Eisenstein and W. Hu, *Baryonic features in the matter transfer function*, ApJ 496, 605 (1998). https://arxiv.org/abs/astro-ph/9709112 +- Planck Collaboration, *Planck 2018 results. VI. Cosmological parameters*, A&A 641, A6 (2020). https://arxiv.org/abs/1807.06209 diff --git a/halos/model.js b/halos/model.js new file mode 100644 index 0000000..0f5375e --- /dev/null +++ b/halos/model.js @@ -0,0 +1,425 @@ +/* Halos within halos: where the nesting of dark matter halos stops, and what sets the smallest halo. Its only + * dependency is the repository's cosmic-homogeneity model, which supplies the Eisenstein & Hu (1998) no-wiggle + * transfer function and the linear growth factor (in a browser, load cosmic-homogeneity/homogeneity.js first: it + * sets window.HOMOGENEITY). + * + * 1. Linear matter power today: P(k) = A k^ns Tnw(k)^2 Tcut(k)^2, with A set by sigma8 (Planck 2018) at Tcut = 1. + * Wavenumbers in h/Mpc on the grid (as in homogeneity.js); the cut-off formulas are in Mpc^-1, as in their papers. + * Masses in Msun (not h^-1 Msun): M = (4 pi/3) rho_m R^3, rho_m today's mean matter density (dark matter and + * baryons, for every mass in this study). + * sigma^2(R) = int Delta^2(k) Tcut^2 W^2(kR) dln k, Delta^2 = k^3 P/(2 pi^2), W the top-hat 3 j1(x)/x. + * Effective slope: sigma ~ M^-alpha, n_eff = 6 alpha - 3. Near n_eff = -3 all masses collapse at nearly once. + * Half mode: Tcut(k_hm) = 1/2 in amplitude, M_hm = (4 pi/3) rho_m (pi/k_hm)^3, a sphere whose radius is half the + * wavelength 2 pi/k_hm (Schneider et al. 2012). Half power, Tcut^2 = 1/2 (Hu, Barkana & Gruzinov 2000): {power: true}. + * Collapse: sigma(M) D(z_c) = 1.686, D = 1 today (or another threshold: Green, Hofmann & Schwarz's z_nl is where + * sigma = 1). Optional factors: the radiation-era growing mode 1 + 2 a_eq/(3a) (Meszaros), and the dark matter + * growing alone on scales the gas never falls into (integrated; see growth()). + * 2. Cut-offs set by the particle. + * WIMP, free streaming (Green, Hofmann & Schwarz 2005): T = [1 - (2/3)(k/k_fs)^2] exp[-(k/k_fs)^2 - (k/k_d)^2], + * in their JCAP paper (the equations labelled dfs and dk in its arXiv source), stated as valid for k < k_fs (the + * half mode is at 0.625 k_fs), with + * k_fs = 1.70e6 Mpc^-1 (m/100 GeV)^1/2 (T_kd/30 MeV)^1/2 / [1 + ln(T_kd/30 MeV)/19.2], + * k_d = 3.76e7 Mpc^-1 (m/100 GeV)^1/2 (T_kd/30 MeV)^1/2. + * The scalings were first read in search text and the prefactor recalled; #35's review read all three in the + * papers' arXiv source. + * Also derived here: after kinetic decoupling a WIMP of one-dimensional speed spread sigma_v = sqrt(T_kd/m) coasts + * a comoving distance s; a Maxwellian smears a mode as exp(-k^2 s^2/2), so k_fs = sqrt(2)/s. + * WIMP, acoustic damping (Loeb & Zaldarriaga 2005; Bertschinger 2006): until kinetic decoupling the WIMPs move with + * the radiation, whose sound waves erase everything inside the horizon. M_ao = (4 pi/3) rho_m tau_kd^3, the mass + * inside the comoving horizon tau_kd = 1/(a H) at T_kd. The floor is the larger of the two (Bringmann 2009, + * Sec. 4, who gives 1e-11 to nearly 1e-3 Msun across WIMP models; as seen in search text, checked in review). + * Bringmann, and Cornell et al. for M_ao, count the dark matter alone, rho_chi = f_c rho_m (as #35's review read + * them): their floors are f_c (0.84 for Planck 2018) times the ones here. + * The expansion before and after decoupling comes from ideal gases of photons, neutrinos, e+-, mu+- and pions + * (see thermal()), valid from below 1 MeV to about 80 MeV: kaons and heavier hadrons are left out (they would + * add about 1% to g at 70 MeV and 7% at 100 MeV, as #35's review's own calculation agreed). + * Wave dark matter (Hu, Barkana & Gruzinov 2000): T = cos(x^3)/(1 + x^8), x = 1.61 m22^(1/18) k/k_Jeq, + * k_Jeq = 9 m22^(1/2) Mpc^-1, m22 = m/1e-22 eV. Schive et al. (2016) fit the halo mass function's turn-down at + * M0 = 1.6e10 m22^(-4/3) Msun (the equation labelled MFFit in their arXiv source, fitted to runs at m22 = 0.8, 1.6 and 3.2, so every bound mass is + * an extrapolation). Both first seen as quoted by later papers in search text; #35's review read them in the + * papers' arXiv source. + * 3. The small-scale tail of the transfer function, as a check on extrapolating Eisenstein & Hu twenty decades in + * mass: a mode that enters the horizon in the radiation era leaves the dark matter with a logarithmic kick, + * delta_c -> -A ln(B k tau), after which it grows as in Meszaros's solution. Radiation-era perturbations in the + * conformal Newtonian gauge (Ma & Bertschinger 1995): photons a perfect fluid, neutrinos either a second perfect + * fluid (still coupled when the mode enters) or a truncated Boltzmann hierarchy (free streaming), the dark matter + * a test fluid. Matching onto Meszaros's growing mode gives, for k >> k_eq, + * T(k) -> (15/16) A (k_eq/k)^2 ln(4 sqrt2 e^-3 B k/k_eq), A in units of the adiabatic constant C + * (Ma & Bertschinger's; B is a pure number), with the large-scale matter-era potential (6/5) C. + * 4. The Jeans scale of a wave of mass m in density rho: omega^2 = (hbar k^2/2m)^2 - 4 pi G rho, so + * k_J = (16 pi G rho)^(1/4) (m/hbar)^(1/2), and the Witness Map's wave clock across 1/k_J, + * m (1/k_J)^2/hbar = 1/(4 sqrt(pi G rho)), is gravity's own clock. + * 5. Nesting: the decades of mass from a floor up to a host, and the subhalo counts of Springel et al. (2008), + * dN/dM = a0 (M/m0)^-1.9, extrapolated down to the floor. The levels apply the host's ratio at every level, the + * self-similar case Springel et al. call an upper limit; they find sub-subhalos 2 or 3 times fewer. + */ +(function (root) { + 'use strict'; + const HOM = typeof module !== 'undefined' && module.exports ? require('../cosmic-homogeneity/homogeneity.js') : root.HOMOGENEITY; + + // SI. CODATA 2018 G and hbar; IAU 2015 nominal solar mass; the parsec from the IAU 2012 astronomical unit. + const C = { G: 6.6743e-11, hbar: 1.054571817e-34, c: 2.99792458e8, e: 1.602176634e-19, kB: 1.380649e-23, + Msun: 1.98841e30, Mpc: 3.0856775814913673e22, yr: 3.15576e7 }; + C.eV = C.e / (C.c * C.c); // kilograms per eV/c^2 + C.pc = C.Mpc / 1e6; C.kpc = C.Mpc / 1e3; + const PLANCK18 = HOM.PLANCK18; + // Round numbers of the WMAP era, for comparing with papers of 2004-2005. sigma8 = 0.9 and n_s = 1 are Green, Hofmann & + // Schwarz's stated inputs (as #35's review read them); h, Omega_m and Omega_b are assumed. + const WMAP_ERA = { h: 0.7, Om: 0.3, Ob: 0.045, ns: 1, sigma8: 0.9, Tcmb: 2.725 }; + const DELTA_C = 1.686; // spherical collapse threshold (Einstein-de Sitter) + const MW = 1e12; // Milky Way halo mass, Msun + // Neutrinos' share of the radiation, 3.046 species: rho_nu/rho_gamma = 3.046 (7/8)(4/11)^(4/3). + const R_NU = (() => { const r = 3.046 * 7 / 8 * Math.pow(4 / 11, 4 / 3); return r / (1 + r); })(); + + // Background: densities, equality (EH 1998 eqs. 2-3), drag epoch, cosmic time. + function background(par = PLANCK18) { + const H100 = 1e5 / C.Mpc; // 100 km/s/Mpc in s^-1 + const rhoCrit = 3 * H100 * H100 / (8 * Math.PI * C.G); // kg m^-3 per h^2 + const rhoMsi = rhoCrit * par.h * par.h * par.Om; // kg m^-3 + const rhoM = rhoMsi * Math.pow(C.Mpc, 3) / C.Msun; // Msun Mpc^-3 + const tf = HOM.transferEH(par), zeq = tf.zeq, aeq = 1 / (1 + zeq); + const keq = 7.46e-2 * par.Om * par.h * par.h * Math.pow(par.Tcmb / 2.7, -2); // Mpc^-1, = a_eq H(a_eq) + const Or = par.Om * aeq, OL = 1 - par.Om - Or, H0 = par.h / 2997.92458; // H0 in Mpc^-1 + const fc = 1 - par.Ob / par.Om; + const hubbleGyr = 977.792 / (100 * par.h); // 1/H0 in Gyr + const E = a => Math.sqrt(Or / (a * a * a * a) + par.Om / (a * a * a) + OL); + function age(z) { // int_0^a da/(a H) by Simpson in ln a from a = 1e-12 + const a1 = 1 / (1 + z), l0 = Math.log(1e-12), l1 = Math.log(a1), n = 400, h = (l1 - l0) / n; + let s = 0; for (let i = 0; i <= n; i++) { const a = Math.exp(l0 + i * h), w = i === 0 || i === n ? 1 : i % 2 ? 4 : 2; s += w / E(a); } + return s * h / 3 * hubbleGyr; + } + return { par, rhoCrit, rhoMsi, rhoM, zeq, aeq, keq, Or, OL, H0, fc, zd: tf.zd, age, hubbleGyr, state: thermal(par, rhoMsi) }; + } + + // ---- The thermal history around kinetic decoupling: the scale factor and the expansion rate at photon temperature T + // (MeV), from ideal gases in equilibrium: photons (2), three neutrinos (6, fermions), e+- (4), mu+- (4), pi+- (2) and + // pi0 (1). The neutrinos decouple at T_NU = 1 MeV (taken as instantaneous); below it the photon-e+- plasma keeps its + // own entropy, so e+- annihilation heats the photons and not the neutrinos. Not valid above about 80 MeV, where the + // hadrons left out (kaons and heavier) start to matter. + const T_NU = 1, NU = 7 / 8 * 6, ME = 0.51099895, MMU = 105.6583755, MPI = 139.57039, MPI0 = 134.9768; // MeV + // One species of dof degrees of freedom and mass x T: g = rho/((pi^2/30) T^4) and gs = s/((2 pi^2/45) T^3), with + // rho = dof/(2 pi^2) int E p^2 f dp and P = dof/(6 pi^2) int p^4/E f dp; Simpson's rule over u = p/T up to 50. + function species(dof, x, fermi) { + if (x > 60) return { g: 0, gs: 0 }; + const n = 400, du = 50 / n; let r = 0, p = 0; + for (let i = 1; i <= n; i++) { // the integrands vanish at u = 0 + const u = i * du, e = Math.sqrt(u * u + x * x), f = 1 / (Math.exp(e) + (fermi ? 1 : -1)), w = i === n ? 1 : i % 2 ? 4 : 2; + r += w * u * u * e * f; p += w * u * u * u * u / e * f; + } + r *= du / 3 * dof / (2 * Math.PI * Math.PI); p *= du / 3 * dof / (6 * Math.PI * Math.PI); + return { g: r * 30 / (Math.PI * Math.PI), gs: (r + p) * 45 / (2 * Math.PI * Math.PI) }; + } + function dof(T) { // the photon-e+- plasma, and the muons and pions + const e = species(4, ME / T, true), mu = species(4, MMU / T, true), pc = species(2, MPI / T, false), p0 = species(1, MPI0 / T, false); + return { g: 2 + e.g, gs: 2 + e.gs, gH: mu.g + pc.g + p0.g, gsH: mu.gs + pc.gs + p0.gs }; + } + function thermal(par, rhoMsi) { + const T0 = par.Tcmb * C.kB / C.e / 1e6, dNu = dof(T_NU); + const aNu = T0 * Math.cbrt(2 / dNu.gs) / T_NU; // photon entropy today, g_s = 2 + const gsNu = dNu.gs + NU + dNu.gsH; + const U = (g, T) => Math.PI * Math.PI / 30 * g * Math.pow(T * 1e6 * C.e, 4) / Math.pow(C.hbar * C.c, 3); // J m^-3 + return function state(T) { + if (T > 80) throw new RangeError('thermal history: kaons and heavier hadrons, left out, add over 2% to g above about 80 MeV'); + const d = dof(T); let a, u, g = null; + if (T >= T_NU) { g = d.g + NU + d.gH; a = aNu * T_NU / T * Math.cbrt(gsNu / (d.gs + NU + d.gsH)); u = U(g, T); } + else { a = T0 * Math.cbrt(2 / d.gs) / T; u = U(d.g, T) + U(NU, T_NU * aNu / a); } + return { a, g, u, H: Math.sqrt(8 * Math.PI * C.G / 3 * (u / (C.c * C.c) + rhoMsi / (a * a * a))) }; + }; + } + + // Mass of a sphere of radius pi/k (half a wavelength), k in Mpc^-1. + const massOfK = (bg, k) => 4 * Math.PI / 3 * bg.rhoM * Math.pow(Math.PI / k, 3); + const radiusOfMass = (bg, M) => Math.cbrt(3 * M / (4 * Math.PI * bg.rhoM)); // Mpc + + // ---- WIMPs. m in GeV, T_kd in MeV. + // Green, Hofmann & Schwarz 2005 (JCAP 0508, 003), scalings as given in their proceedings, astro-ph/0508553. + function wimpGHS({ m = 100, Tkd = 30 } = {}) { + const r = Math.sqrt(m / 100 * Tkd / 30); + const kfs = 1.70e6 * r / (1 + Math.log(Tkd / 30) / 19.2), kd = 3.76e7 * r; + return { name: `WIMP ${m} GeV, T_kd ${Tkd} MeV`, short: `WIMP ${m}/${Tkd}`, kind: 'wimp', m, Tkd, kfs, kd, T: k => wimpT(k, kfs, kd) }; + } + function wimpT(k, kfs, kd) { const x = k / kfs, y = k / kd; return (1 - 2 / 3 * x * x) * Math.exp(-x * x - y * y); } + // The universe at kinetic decoupling: scale factor, expansion rate (s^-1), the radiation's g, and the comoving + // horizon tau = 1/(a H) in Mpc. + function decoupling(bg, Tkd) { const st = bg.state(Tkd); return { ...st, tau: C.c / (st.a * st.H) / C.Mpc }; } + // Free streaming from first principles. After decoupling at a_kd the speed spread falls as sigma_v a_kd/a, so the + // comoving distance coasted is s = sigma_v a_kd c int da/(a^3 H) = sigma_v a_kd c int dln a/(a^2 H): numerically + // through the thermal history down to 1 keV, where e+- annihilation is over, then in closed form for radiation and + // matter up to today, (F(y) - F(Om/Or))/(H0 sqrt(Or)), F(y) = 2 ln[(sqrt(1+y) + 1)/sqrt(y)], y = a Om/Or (Lambda + // ignored: it would lower s by 0.02%). And k_fs = sqrt(2)/s. + function wimpFreeStream(bg, { m = 100, Tkd = 30 } = {}) { + const sig = Math.sqrt(Tkd * 1e-3 / m), n = 400, l0 = Math.log(Tkd), l1 = Math.log(1e-3); // sigma_v in units of c + let prev = bg.state(Tkd), I = 0; + const akd = prev.a; + for (let i = 1; i <= n; i++) { // trapezoid in ln a + const st = bg.state(Math.exp(l0 + (l1 - l0) * i / n)); + I += Math.log(st.a / prev.a) * (1 / (st.a * st.a * st.H) + 1 / (prev.a * prev.a * prev.H)) / 2; prev = st; + } + const Or = prev.u / (C.c * C.c) * Math.pow(prev.a, 4) / (bg.rhoCrit * bg.par.h * bg.par.h), F = y => 2 * Math.log((Math.sqrt(1 + y) + 1) / Math.sqrt(y)); + const s = sig * akd * (C.c * I / C.Mpc + (F(prev.a * bg.par.Om / Or) - F(bg.par.Om / Or)) / (bg.H0 * Math.sqrt(Or))); // Mpc + const kfs = Math.SQRT2 / s, kd = 3.76e7 * Math.sqrt(m / 100 * Tkd / 30); + return { name: `WIMP ${m} GeV, T_kd ${Tkd} MeV (free streaming derived)`, kind: 'wimp', m, Tkd, kfs, kd, s, akd, T: k => wimpT(k, kfs, kd) }; + } + // Acoustic damping: the matter inside the horizon at kinetic decoupling (the dark matter alone is f_c of it). + const acousticMass = (bg, Tkd) => 4 * Math.PI / 3 * bg.rhoM * Math.pow(decoupling(bg, Tkd).tau, 3); + // The WIMP floor under two conventions: Bringmann's max[M(pi/k_fs), M_ao], and the half-mode mass with the same rule. + // Both in all the matter, as every mass here; Bringmann writes his with the dark matter density, f_c times these. + function wimpFloor(bg, { m = 100, Tkd = 30 } = {}) { + const w = wimpGHS({ m, Tkd }), Mfs = massOfK(bg, w.kfs), Mhm = massOfK(bg, halfMode(w)), Mao = acousticMass(bg, Tkd); + return { cut: w, Mfs, Mhm, Mao, bringmann: Math.max(Mfs, Mao), halfMode: Math.max(Mhm, Mao) }; + } + + // ---- Wave dark matter. Hu, Barkana & Gruzinov 2000 (PRL 85, 1158). + function fdmHBG(m22) { + const kJeq = 9 * Math.sqrt(m22), c = 1.61 * Math.pow(m22, 1 / 18) / kJeq; + return { name: `wave DM ${(m22 * 1e-22).toExponential(1)} eV`, short: `wave ${(m22 * 1e-22).toExponential(0)}`, kind: 'fdm', m22, kJeq, + T: k => { const x = c * k, x3 = x * x * x; return Math.cos(x3) / (1 + x3 * x3 * x * x); } }; + } + const schiveM0 = m22 => 1.6e10 * Math.pow(m22, -4 / 3); // Schive et al. 2016 (ApJ 818, 89), Msun + // Lower bounds on the mass (eV), each assuming the wave is all of the dark matter; values and confidence levels + // as seen in search text (abstracts and search summaries), and as #35's review read them in the papers. + const BOUNDS = [ + { m: 2.0e-21, cl: '2 sigma', src: 'Irsic et al. 2017, Lyman-alpha (conservative thermal history)' }, + { m: 2.9e-21, cl: '95%', src: 'Armengaud et al. 2017, Lyman-alpha; Nadler et al. 2021, Milky Way satellites' }, + { m: 2.0e-20, cl: '95%', src: 'Rogers & Peiris 2021, Lyman-alpha' }, + { m: 3e-19, cl: '99%', src: 'Dalal & Kravtsov 2022, sizes and stellar speeds of ultrafaint dwarfs' }, + { m: 8e-18, cl: '95%', src: 'May, Dalal & Kravtsov 2025, Ursa Major III / UNIONS 1, if it is a galaxy' }, + ]; + + const CDM = { name: 'CDM (no cut-off)', short: 'CDM no cut', kind: 'cdm', T: () => 1 }; + // Sensitivity of the WIMP form: the same k_fs with a plain Gaussian, exp(-(k/k_fs)^2 - (k/k_d)^2). + const gaussianOf = w => ({ ...w, name: w.name + ', Gaussian only', short: w.short + ' G', T: k => Math.exp(-Math.pow(k / w.kfs, 2) - Math.pow(k / w.kd, 2)) }); + + // First k (Mpc^-1) where T falls to 1/2 (half mode), or T^2 does ({power: true}). Scans upward, then bisects; + // Infinity if it never does. + function halfMode(cut, { power = false } = {}) { + const f = k => { const t = cut.T(k); return (power ? t * t : t) - 0.5; }; + let k = 1e-4; while (f(k) > 0 && k < 1e14) k *= 1.05; + if (f(k) > 0) return Infinity; + let lo = k / 1.05, hi = k; + for (let i = 0; i < 80; i++) { const mid = Math.sqrt(lo * hi); if (f(mid) > 0) lo = mid; else hi = mid; } + return hi; + } + + // ---- The linear spectrum on a log-k grid (h/Mpc), and sigma(M) for any cut-off. + const tophat2 = x => { + if (x < 1e-3) return 1 - x * x / 5; + if (x > 10) return 4.5 * (1 + x * x) / Math.pow(x, 6); // phase average of W^2: the grid cannot follow the wiggles + const w = 3 * (Math.sin(x) - x * Math.cos(x)) / (x * x * x); return w * w; + }; + // opts: perDecade (grid density), kmax (h/Mpc), running (dn_s/dln k about 0.05 Mpc^-1), factor (a function of k in + // Mpc^-1 multiplying the transfer function at small scales, applied after the sigma8 normalisation). + function model(par = PLANCK18, opts = {}) { + const bg = background(par), tf = HOM.transferEH(par), gr = HOM.growth(par); + const perDec = opts.perDecade || 100, k0 = 1e-5, k1 = opts.kmax || 1e10; + const NK = Math.ceil(Math.log10(k1 / k0) * perDec), dl = Math.log(k1 / k0) / NK; + const alphaS = opts.running || 0, kp = 0.05 / par.h; + const K = new Float64Array(NK), D2 = new Float64Array(NK); + const shape = k => Math.pow(k, 3 + par.ns) * Math.pow(k / kp, 0.5 * alphaS * Math.log(k / kp)) * Math.pow(tf.Tnw(k), 2) / (2 * Math.PI * Math.PI); + for (let i = 0; i < NK; i++) { K[i] = k0 * Math.exp((i + 0.5) * dl); D2[i] = shape(K[i]); } + let s8 = 0; for (let i = 0; i < NK; i++) s8 += D2[i] * tophat2(8 * K[i]) * dl; + const A = par.sigma8 * par.sigma8 / s8; + for (let i = 0; i < NK; i++) { D2[i] *= A; if (opts.factor) D2[i] *= Math.pow(opts.factor(K[i] * par.h), 2); } + const cache = new Map(); + function weights(cut) { + if (cache.has(cut)) return cache.get(cut); + const w = new Float64Array(NK); for (let i = 0; i < NK; i++) { const t = cut.T(K[i] * par.h); w[i] = D2[i] * t * t * dl; } + cache.set(cut, w); return w; + } + const Rh = M => radiusOfMass(bg, M) * par.h; // h^-1 Mpc + function sigma(M, cut = CDM) { + const w = weights(cut), R = Rh(M); let s = 0; + for (let i = 0; i < NK; i++) s += w[i] * tophat2(K[i] * R); return Math.sqrt(s); + } + function slope(M, cut = CDM) { + const e = 0.05, a = -Math.log(sigma(M * Math.exp(e), cut) / sigma(M * Math.exp(-e), cut)) / (2 * e); + return { alpha: a, neff: 6 * a - 3 }; + } + // Peak of Delta^2 T^2 (Mpc^-1): where the power per log k is largest once the cut-off is applied. + function peak(cut) { const w = weights(cut); let j = 0; for (let i = 1; i < NK; i++) if (w[i] > w[j]) j = i; return K[j] * par.h; } + // Growth table: ln D against ln(1+z), from homogeneity.js (flat LCDM, no radiation). + const NZ = 300, lz1 = Math.log(1 + 3e4), LZ = new Float64Array(NZ + 1), LD = new Float64Array(NZ + 1); + for (let i = 0; i <= NZ; i++) { LZ[i] = lz1 * i / NZ; LD[i] = Math.log(gr.D(Math.exp(LZ[i]) - 1)); } + // The dark matter growing alone. Far below the gas's Jeans mass the baryons never fall in, but the growth factor + // (and Eisenstein & Hu's transfer function) assume that after the drag epoch they do. Both histories, in ln a with + // radiation, matter and Lambda, delta'' + (2 + dln H/dln a) delta' = (3/2)(Omega_c(a) delta_c + Omega_b(a) delta_b), + // the baryons smooth throughout, or released from rest at the drag epoch. Started at a_eq/100 on the radiation-era + // logarithmic mode, delta_c = 14 + ln(a/a_i) (a mode that entered at about 1e6 Mpc^-1; from 8 to 20 in place of + // 14 the collapse redshifts do not change at the 0.01 level). + let lag = null; + function lagTable() { + const fc = bg.fc, fb = 1 - fc, Om = par.Om, Or = bg.Or, OL = 1 - Om - Or; + const rhs = (N, y) => { + const a = Math.exp(N), m = Om / (a * a * a), r = Or / (a * a * a * a), E2 = r + m + OL; + const fr = 2 + (-4 * r - 3 * m) / (2 * E2), src = 1.5 * m / E2 * (fc * y[0] + fb * y[2]); + return [y[1], src - fr * y[1], y[3], src - fr * y[3]]; + }; + const rk4 = (N, y, h) => { + const k1 = rhs(N, y), k2 = rhs(N + h / 2, y.map((v, i) => v + h / 2 * k1[i])), k3 = rhs(N + h / 2, y.map((v, i) => v + h / 2 * k2[i])); + const k4 = rhs(N + h, y.map((v, i) => v + h * k3[i])); return y.map((v, i) => v + h / 6 * (k1[i] + 2 * k2[i] + 2 * k3[i] + k4[i])); + }; + const N0 = Math.log(bg.aeq / 100), Nd = -Math.log(1 + bg.zd), n1 = Math.ceil((Nd - N0) / 0.005), n2 = Math.ceil(-Nd / 0.005); + let y = [14, 1, 0, 0], N = N0; + for (let i = 0; i < n1; i++) { y = rk4(N, y, (Nd - N0) / n1); y[2] = y[3] = 0; N += (Nd - N0) / n1; } // smooth before drag + let ySmooth = y.slice(), yIn = y.slice(); + const tab = [[Nd, 1]]; + const smooth = (N, v) => { const d = rhs(N, v); d[2] = d[3] = 0; return d; }; + for (let i = 0; i < n2; i++) { + const h = -Nd / n2; + yIn = rk4(N, yIn, h); + const k1 = smooth(N, ySmooth), k2 = smooth(N + h / 2, ySmooth.map((v, j) => v + h / 2 * k1[j])), k3 = smooth(N + h / 2, ySmooth.map((v, j) => v + h / 2 * k2[j])); + const k4 = smooth(N + h, ySmooth.map((v, j) => v + h * k3[j])); + ySmooth = ySmooth.map((v, j) => v + h / 6 * (k1[j] + 2 * k2[j] + 2 * k3[j] + k4[j])); N += h; + tab.push([N, fc * ySmooth[0] / (fc * yIn[0] + fb * yIn[2])]); + } + return tab; + } + function growth(z, { radiation = false, baryonLag = false } = {}) { + const u = Math.log(1 + z) / lz1 * NZ, i = Math.min(NZ - 1, Math.floor(u)), t = u - i; + let D = Math.exp(LD[i] + t * (LD[i + 1] - LD[i])); + const a = 1 / (1 + z); + if (radiation) D *= 1 + 2 * bg.aeq / (3 * a); + // The lag on the matter contrast f_c delta_c (baryonLag true), or on the dark matter contrast itself ('cdm', + // which also divides by f_c): two readings of which contrast reaches 1.686. + if (baryonLag && z < bg.zd) { + if (!lag) lag = lagTable(); + const N = -Math.log(1 + z), j = Math.min(lag.length - 2, Math.floor((N - lag[0][0]) / (lag[1][0] - lag[0][0]))); + const w = (N - lag[j][0]) / (lag[j + 1][0] - lag[j][0]); + D *= lag[j][1] + w * (lag[j + 1][1] - lag[j][1]); + } + if (baryonLag === 'cdm') D /= bg.fc; // before drag too, the baryons are smooth + return D; + } + // Redshift at which a peak of height nu sigma reaches 1.686 (or o.delta); -1 if it has not yet. + function zc(M, cut = CDM, o = {}) { + const s = (o.nu ?? 1) * sigma(M, cut, o), dc = o.delta ?? DELTA_C; if (s * growth(0, o) < dc) return -1; + let lo = 0, hi = lz1; + for (let i = 0; i < 60; i++) { const mid = (lo + hi) / 2; if (s * growth(Math.exp(mid) - 1, o) > dc) lo = mid; else hi = mid; } + return Math.exp(lo) - 1; + } + const fac = k => (opts.factor ? Math.pow(opts.factor(k * par.h), 2) : 1); + return { par, bg, tf, A, K, D2, dl, NK, sigma, slope, peak, growth, zc, Rh, + Delta2: k => A * shape(k) * fac(k), P: k => A * shape(k) * fac(k) * 2 * Math.PI * Math.PI / (k * k * k) }; + } + + // ---- Radiation-era kick (section 3 above). Units k = 1, x = k tau, adiabatic constant C = 1. Rnu is the neutrinos' + // share of the radiation; free: true lets them stream through a hierarchy of lmax multipoles, free: false keeps them + // a perfect fluid (as photons are). The potential follows the 0i Einstein equation up to x = 1 and the algebraic + // Poisson constraint after, so it cannot drift. Returns {A, B} of delta_c -> -A ln(B x), fitted between xmax/4 and + // xmax, and the radiation-era potential psi0 = 20/(15 + 4 Rnu) (Ma & Bertschinger eq. 96). + // For a perfect fluid the asymptote is exact. The potential is phi = psi = 3 psi0 (sin y - y cos y)/y^3, y = x/sqrt3, + // and delta_c' = 3 phi' - theta_c, theta_c' = psi - theta_c/x integrate once to + // x delta_c' = 3 x phi' - int_0^x x' psi dx' = 3 x phi' - 9 psi0 (1 - sin y/y). + // With int_0^Y (1 - sin y/y) dy/y = ln Y + gamma_E - 1 + O(1/Y) and delta_c(0) = -3 psi0/2, phi(0) = psi0, + // delta_c -> -9 psi0 ln(x e^(gamma_E - 1/2)/sqrt3): A = 9 psi0, B = e^(gamma_E - 1/2)/sqrt3. + // This integration reaches it to 1e-3. + function radiationEra({ Rnu = R_NU, free = true, lmax = 60, xmax = 800 } = {}) { + if (!(xmax >= 4)) throw new RangeError('radiationEra: xmax must be at least 4'); + const fn = Rnu, fg = 1 - Rnu, L = free ? lmax : 1; + // state: delta_g, theta_g, delta_nu, theta_nu, F_2..F_L (neutrino multipoles, F_2 = 2 sigma), delta_c, theta_c, phi + const iF = 4, iDC = 4 + L - 1, iTC = iDC + 1, iPHI = iDC + 2, N = iDC + 3; + const psi0 = 20 / (15 + 4 * (free ? Rnu : 0)), phi0 = (1 + 0.4 * (free ? Rnu : 0)) * psi0; + let x = 1e-3; const s = new Float64Array(N); + s[0] = s[2] = -2 * psi0; s[1] = s[3] = 0.5 * x * psi0; + if (free) s[iF] = 2 * x * x / 15 * psi0; + s[iDC] = -1.5 * psi0; s[iTC] = 0.5 * x * psi0; s[iPHI] = phi0; + function deriv(x, s, d) { + const phi = x < 1 ? s[iPHI] : -1.5 / (x * x) * (fg * s[0] + fn * s[2] + 4 / x * (fg * s[1] + fn * s[3])); + const sig = free ? s[iF] / 2 : 0, psi = phi - 6 * fn * sig / (x * x); + const phip = -psi / x + 2 / (x * x) * (fg * s[1] + fn * s[3]); + d[0] = -4 / 3 * s[1] + 4 * phip; d[1] = s[0] / 4 + psi; + d[2] = -4 / 3 * s[3] + 4 * phip; d[3] = s[2] / 4 - sig + psi; + if (free) { + const F = l => (l === 1 ? 4 / 3 * s[3] : s[iF + l - 2]); + d[iF] = 8 / 15 * s[3] - 3 / 5 * (L >= 3 ? s[iF + 1] : 0); + for (let l = 3; l <= L; l++) { + const Fp1 = l < L ? s[iF + l - 1] : (2 * L + 1) / x * F(L) - F(L - 1); // Ma & Bertschinger's truncation + d[iF + l - 2] = (l * F(l - 1) - (l + 1) * Fp1) / (2 * l + 1); + } + } + d[iDC] = -s[iTC] + 3 * phip; d[iTC] = -s[iTC] / x + psi; d[iPHI] = phip; + } + const k1 = new Float64Array(N), k2 = new Float64Array(N), k3 = new Float64Array(N), k4 = new Float64Array(N), t = new Float64Array(N); + const step = h => { + deriv(x, s, k1); for (let i = 0; i < N; i++) t[i] = s[i] + h / 2 * k1[i]; + deriv(x + h / 2, t, k2); for (let i = 0; i < N; i++) t[i] = s[i] + h / 2 * k2[i]; + deriv(x + h / 2, t, k3); for (let i = 0; i < N; i++) t[i] = s[i] + h * k3[i]; + deriv(x + h, t, k4); for (let i = 0; i < N; i++) s[i] += h / 6 * (k1[i] + 2 * k2[i] + 2 * k3[i] + k4[i]); + x += h; + }; + const x1 = xmax / 4; let d1 = null; + for (const stop of [1, x1, xmax]) { // land exactly on x = 1 (the switch) and on x1 + while (x < stop - 1e-12) step(Math.min(0.02, 0.3 * x / (2 * L + 1), 0.1 * x, stop - x)); + if (stop === x1) d1 = s[iDC]; + } + const A = -(s[iDC] - d1) / Math.log(xmax / x1), B = Math.exp(-s[iDC] / A) / xmax; + return { A, B, psi0 }; + } + // The tail it implies, T(k) for k in Mpc^-1, written as coef (k_eq/k)^2 ln(c k/k_eq). + function smallScaleTail(bg, { A, B }) { + const coef = 15 / 16 * A, c = 4 * Math.SQRT2 * Math.exp(-3) * B; + return { coef, c, T: k => coef * Math.pow(bg.keq / k, 2) * Math.log(c * k / bg.keq) }; + } + + // ---- Jeans scale of a wave of mass mEV (eV) in density rho (kg m^-3), physical, SI. K: the constants to use + // (G, hbar, and eV in kilograms), e.g. the Witness Map's own. + function jeans(mEV, rho, K = C) { + const m = mEV * K.eV, closed = Math.pow(16 * Math.PI * K.G * rho, 0.25) * Math.sqrt(m / K.hbar); + // Root of omega^2(k) = (hbar k^2/2m)^2 - 4 pi G rho by bisection, independent of the closed form. + const w2 = k => Math.pow(K.hbar * k * k / (2 * m), 2) - 4 * Math.PI * K.G * rho; + let lo = closed / 10, hi = closed * 10; + for (let i = 0; i < 200; i++) { const mid = Math.sqrt(lo * hi); if (w2(mid) < 0) lo = mid; else hi = mid; } + const kJ = Math.sqrt(lo * hi); + return { kJ, closed, tWave: m / (K.hbar * kJ * kJ), tGrav: 1 / (4 * Math.sqrt(Math.PI * K.G * rho)), + tFF: Math.sqrt(3 * Math.PI / (32 * K.G * rho)), lambdaJ: 2 * Math.PI / kJ }; + } + // Comoving Jeans wavenumber (Mpc^-1) at scale factor a, in the mean matter density rho_m0 a^-3 (or a given density + // today, rho0 in kg m^-3), so k_J ~ a^(1/4). + function kJcomoving(bg, m22, a, { rho0 = bg.rhoMsi } = {}) { + return a * jeans(m22 * 1e-22, rho0 / (a * a * a)).kJ * C.Mpc; + } + // Jeans masses (Msun) under the definitions in use, in the comoving density rhoM (Msun Mpc^-3; mean matter by + // default); lambda_J = 2 pi/k_J. halfWave, a sphere of diameter lambda_J, is Hui, Ostriker, Tremaine & Witten's + // (2017) definition, M_J = (4/3) pi rho (lambda_J/2)^3 in the wave's own density (the equation labelled jeans in their arXiv source, as #35's review + // read it in the arXiv source; first seen quoted in search text). + function jeansMasses(bg, kJ, rhoM = bg.rhoM) { + const lam = 2 * Math.PI / kJ, V = r => 4 * Math.PI / 3 * r * r * r * rhoM; + return { halfWave: V(lam / 2), fullWave: V(lam), inverseK: V(1 / kJ), cube: rhoM * lam * lam * lam }; + } + // The soliton (a halo's core) from the Schroedinger-Poisson ground state in the Witness Map's units (its SOL: half- + // density radius xc, mass Ic inside it and I in all, in units hbar = m = 4 pi G rho_c = 1). Masses times the + // half-density radius r_c, in units of hbar^2/(G m^2): the same combination that sets a gravitational atom's size. + function solitonRelation(SOL) { + return { coreRc: SOL.Ic * SOL.xc, totalRc: SOL.I * SOL.xc, coreShare: SOL.Ic / SOL.I }; + } + const hbar2Gm2 = (mEV, K = C) => K.hbar * K.hbar / (K.G * Math.pow(mEV * K.eV, 2)); // kg m + + // ---- Subhalos. Springel et al. 2008 (Aquarius, MNRAS 391, 1685) fit dN/dM = a0 (M/m0)^-n for Aq-A-1 within + // r50 = 433 kpc (host M200 = 1.84e12 Msun): a0 = 3.26e-5 Msun^-1, m0 = 2.52e7 Msun, n = 1.9; and the same law scaled + // to a 1e12 Msun host within 353 kpc, a0 = 1.77e-5 (both as quoted by Erkal et al. 2016, arXiv:1606.04946; #35's + // review found Aq-A's in Springel et al. directly). m50, the mass inside r50: Aq-A's 2.52e12 Msun (their Table 1, + // recalled; it gives the mu of 0.0195 against M50 that #35's review found), scaled with the host. + const AQUARIUS = { a0: 3.26e-5, m0: 2.52e7, n: 1.9, host: 1.84e12, r: 433, m50: 2.52e12 }; + const AQUARIUS_MW = { a0: 1.77e-5, m0: 2.52e7, n: 1.9, host: 1e12, r: 353, m50: 2.52e12 / 1.84 }; + const subhaloCount = (Mmin, Mmax = Infinity, s = AQUARIUS_MW) => + s.a0 * s.m0 / (s.n - 1) * (Math.pow(Mmin / s.m0, 1 - s.n) - (isFinite(Mmax) ? Math.pow(Mmax / s.m0, 1 - s.n) : 0)); + const subhaloMass = (Mmin, Mmax, s = AQUARIUS_MW) => + s.a0 * s.m0 * s.m0 / (2 - s.n) * (Math.pow(Mmax / s.m0, 2 - s.n) - Math.pow(Mmin / s.m0, 2 - s.n)); + // The largest subhalo to expect, N(>M) = 1, as a fraction mu of its host's M200 (of its M50 with {m50: true}). + const muAquarius = (s = AQUARIUS_MW, { m50 = false } = {}) => s.m0 * Math.pow(s.a0 * s.m0 / (s.n - 1), 1 / (s.n - 1)) / (m50 ? s.m50 : s.host); + // Decades of mass between a floor and a host, and levels of nesting if each level is a factor mu below its host: + // levels (whole), exact (the fraction dec/log10(1/mu), for each mu). suppressed: the levels at mu = muAquarius() if, + // below the first, each level's subhalos are a factor f = 2 or 3 fewer than the self-similar count (Springel et al.'s + // sub-subhalos). N(>M) ~ M^-(n-1), so each of those levels is mu f^(-1/(n-1)) of its host. + const nesting = (floor, host = MW, mus = [0.1, 0.01, muAquarius()]) => { + const dec = Math.log10(host / floor), d1 = -Math.log10(muAquarius()); + const suppressed = [2, 3].map(f => (dec < d1 ? 0 : 1 + Math.floor((dec - d1) / (d1 + Math.log10(f) / (AQUARIUS_MW.n - 1))))); + return { decades: dec, levels: mus.map(mu => Math.max(0, Math.floor(dec / -Math.log10(mu)))), exact: mus.map(mu => dec / -Math.log10(mu)), suppressed }; + }; + + const api = { C, PLANCK18, WMAP_ERA, DELTA_C, MW, R_NU, HOM, background, massOfK, radiusOfMass, + wimpGHS, wimpT, decoupling, wimpFreeStream, acousticMass, wimpFloor, fdmHBG, schiveM0, BOUNDS, CDM, gaussianOf, halfMode, + model, radiationEra, smallScaleTail, jeans, kJcomoving, jeansMasses, solitonRelation, hbar2Gm2, + AQUARIUS, AQUARIUS_MW, subhaloCount, subhaloMass, muAquarius, nesting }; + if (typeof module !== 'undefined' && module.exports) module.exports = api; else root.HALOS = api; +})(typeof window !== 'undefined' ? window : globalThis); diff --git a/halos/table.mjs b/halos/table.mjs new file mode 100644 index 0000000..b889299 --- /dev/null +++ b/halos/table.mjs @@ -0,0 +1,169 @@ +// Prints how deep the nesting of dark matter halos goes and what stops it: the small-scale spectrum, the floors a WIMP +// and wave dark matter set, sigma(M) and when each mass collapses, the levels of nesting below a Milky Way, and the +// Jeans scale and soliton in the Witness Map's terms. Usage: node halos/table.mjs (under a second) +import { createRequire } from 'node:module'; +const require = createRequire(import.meta.url); +const H = require('./model.js'); +const WM = require('../witness-map/model.js'); +const e2 = x => (x === 0 ? '0' : Math.abs(x) >= 1e-2 && Math.abs(x) < 1e4 ? x.toPrecision(3) : x.toExponential(2)); +const pad = (s, n) => String(s).padStart(n); + +const P18 = H.PLANCK18, md = H.model(P18, { perDecade: 160 }), bg = md.bg, h = P18.h; +const hm = c => { const k = H.halfMode(c), k2 = H.halfMode(c, { power: true }); return { k, M: H.massOfK(bg, k), k2, M2: H.massOfK(bg, k2) }; }; + +// ---- 1. The spectrum +const tailFree = H.smallScaleTail(bg, H.radiationEra({ lmax: 120, xmax: 1600 })), tailFluid = H.smallScaleTail(bg, H.radiationEra({ free: false })); +const noB = H.HOM.transferEH({ ...P18, Ob: 1e-7 }); +console.log('1. Linear matter power today: Planck 2018, Eisenstein & Hu no-wiggle transfer function, sigma8 = 0.8111'); +console.log(` rho_m = ${e2(bg.rhoM)} Msun/Mpc^3, z_eq = ${bg.zeq.toFixed(0)}, k_eq = ${bg.keq.toFixed(5)} Mpc^-1\n`); +console.log(' k [h/Mpc] P [(Mpc/h)^3] Delta^2 dlnDelta^2/dlnk T_nw T_nw / T_nw(no baryons) EH (no baryons) / derived tail:'); +console.log(' free neutrinos fluid neutrinos'); +for (const k of [1e-3, 1e-2, 1e-1, 1, 1e1, 1e2, 1e3, 1e4, 1e5, 1e6, 1e7, 1e8, 1e9]) { + const sl = Math.log(md.Delta2(k * 1.01) / md.Delta2(k / 1.01)) / Math.log(1.01 * 1.01); + console.log(` ${pad(k.toExponential(0), 7)} ${pad(e2(md.P(k)), 11)} ${pad(e2(md.Delta2(k)), 9)} ${pad(sl.toFixed(3), 9)} ${pad(md.tf.Tnw(k).toExponential(2), 12)} ` + + `${pad((md.tf.Tnw(k) / noB.Tnw(k)).toFixed(3), 12)}` + (k >= 1e2 ? ` ${pad((noB.Tnw(k) / tailFree.T(k * h)).toFixed(3), 23)} ${pad((noB.Tnw(k) / tailFluid.T(k * h)).toFixed(3), 16)}` : '')); +} +console.log(` Derived tails T -> coef (k_eq/k)^2 ln(c k/k_eq): free-streaming neutrinos ${tailFree.coef.toFixed(2)}, c = ${tailFree.c.toFixed(4)};` + + ` neutrinos a fluid ${tailFluid.coef.toFixed(2)}, c = ${tailFluid.c.toFixed(4)}`); +const kNu = 1 / H.decoupling(bg, 1).tau; // horizon wavenumber when the neutrinos decouple, at about 1 MeV +console.log(` The neutrinos decouple at about 1 MeV, when the horizon is k = ${e2(kNu)} Mpc^-1. Modes entering well before that see them`); +console.log(` as a fluid: from about ${e2(10 * kNu)} Mpc^-1 up the fluid tail applies, and Eisenstein & Hu overstate the transfer function`); +console.log(' there, at the bino cut-off included, by about 10%.'); + +// ---- 2. The floors +const bino = H.wimpGHS({ m: 100, Tkd: 30 }), Mb = hm(bino).M; +const derived = H.wimpFreeStream(bg, { m: 100, Tkd: 30 }); +const ghsOver = Tkd => H.wimpGHS({ m: 100, Tkd }).kfs / H.wimpFreeStream(bg, { m: 100, Tkd }).kfs - 1; +console.log('\n2. Floors. Masses of spheres: radius pi/k (half a wavelength) for a wavenumber k; M_hm where T = 1/2, M_1/2 where T^2 = 1/2 [Msun]'); +console.log(` WIMP 100 GeV, T_kd 30 MeV: k_fs = ${e2(bino.kfs)} Mpc^-1 (derived from the coasting distance ${(derived.s * 1e6).toFixed(3)} pc: k_fs = sqrt2/s = ${e2(derived.kfs)}), k_d = ${e2(bino.kd)}`); +console.log(` M(pi/k_fs) = ${e2(H.massOfK(bg, bino.kfs))}, M_hm = ${e2(Mb)}, M_1/2 = ${e2(hm(bino).M2)}, M(pi/k_d) = ${e2(H.massOfK(bg, bino.kd))};` + + ` Gaussian form, same k_fs: M_hm = ${e2(hm(H.gaussianOf(bino)).M)}`); +console.log(` Green, Hofmann & Schwarz's k_fs against the derived one: ${[10, 30, 40].map(T => `${(100 * ghsOver(T)).toFixed(1)}% at ${T} MeV`).join(', ')}`); +console.log(` pi/k_fs = ${(Math.PI / bino.kfs * 1e6).toFixed(2)} pc comoving (a sphere of radius 1 pc holds ${e2(4 * Math.PI / 3 * bg.rhoM * 1e-18)} Msun)`); +const dk = H.decoupling(bg, 30); +console.log(` at 30 MeV the radiation has g = ${dk.g.toFixed(2)} (muons and pions add ${(dk.g - 10.75).toFixed(2)} to photons, e+- and neutrinos);` + + ` horizon ${(dk.tau * 1e6).toFixed(2)} pc comoving: acoustic mass M_ao = ${e2(H.acousticMass(bg, 30))}`); +const TK = [3, 10, 20, 30, 50]; +console.log('\n WIMP floor, max[M(pi/k_fs), M_ao] (Bringmann 2009) / the same with M_hm, by mass and kinetic decoupling temperature:'); +console.log(' ' + 'm [GeV] \\ T_kd [MeV]'.padEnd(21) + TK.map(t => pad(t, 20)).join('')); +for (const m of [10, 100, 1000]) console.log(' ' + pad(m, 6).padEnd(21) + TK.map(Tkd => { const f = H.wimpFloor(bg, { m, Tkd }); return pad(`${e2(f.bringmann)} / ${e2(f.halfMode)}`, 20); }).join('')); +console.log(' ' + 'acoustic mass alone'.padEnd(21) + TK.map(Tkd => pad(e2(H.acousticMass(bg, Tkd)), 20)).join('')); +console.log(' Bringmann (2009): across WIMP models the floor runs from about 1e-11 to nearly 1e-3 Msun (T_kd from several MeV to a few GeV).'); +console.log(` These masses are of all the matter. Bringmann, and Cornell et al. for M_ao, count the dark matter alone, f_c = ${bg.fc.toFixed(3)} of it:` + + ` in their convention each floor is ${(-Math.log10(bg.fc)).toFixed(3)} decades lower.`); + +const MASSES = [1e-22, 2e-21, 2.9e-21, 2e-20, 3e-19, 8e-18]; +const wave = MASSES.map(m => { const c = H.fdmHBG(m / 1e-22), r = hm(c), J = H.jeansMasses(bg, H.kJcomoving(bg, m / 1e-22, 1)); + return { m, c, k2: r.k2, M2: r.M2, Mhm: r.M, M0: H.schiveM0(m / 1e-22), MJ: J.halfWave }; }); +console.log('\n Wave dark matter: the halo population turns down near M_1/2, M_hm and Schive\'s M0; nothing collapses below the Jeans mass.'); +console.log(' m [eV] k_1/2 [Mpc^-1] M_1/2 M_hm M0 (Schive) Jeans today bound at this mass (each assumes all the dark matter)'); +for (const w of wave) { + const b = H.BOUNDS.find(x => x.m === w.m); + console.log(` ${w.m.toExponential(1)} ${pad(e2(w.k2), 12)} ${pad(e2(w.M2), 11)} ${pad(e2(w.Mhm), 9)} ${pad(e2(w.M0), 11)} ${pad(e2(w.MJ), 12)} ${b ? `${b.cl}: ${b.src}` : 'benchmark of the lensing notes; excluded by all of the bounds'}`); +} +const mFit = 3.2e-22, rp = wave.find(w => w.m === 2e-20); // Schive et al. fitted M0 at m22 = 0.8, 1.6 and 3.2 +console.log(` Hu et al.'s cut-off was computed near 1e-22 eV and Schive et al. fitted M0 at m22 = 0.8 to 3.2, so every bound row is an extrapolation:`); +console.log(` ${(MASSES[1] / mFit).toFixed(1)} to ${e2(MASSES[5] / mFit)} times the heaviest particle fitted (${(rp.m / mFit).toFixed(1)} at 2e-20 eV). Rogers & Peiris quote a half-mode mass of`); +console.log(` 7.2e7 Msun at their bound, in a convention matched to warm dark matter (Drlica-Wagner et al. 2019): ${(7.2e7 / rp.Mhm).toFixed(1)} times M_hm here.`); + +// ---- 3. sigma(M) and collapse +const masses = [1e-12, 1e-9, 1e-6, 1e-3, 1, 1e3, 1e6, 1e9, 1e12, 1e15]; +const models = [H.CDM, bino, wave[0].c, wave[3].c]; +console.log('\n3. sigma(M) today (top-hat), with n_eff = 6 alpha - 3 for sigma ~ M^-alpha in brackets: at n_eff = -3 every mass collapses at once'); +console.log(' M [Msun] ' + models.map(c => c.short.padStart(17)).join('')); +for (const M of masses) console.log(` ${pad(M.toExponential(0), 7)} ` + models.map(c => pad(`${md.sigma(M, c).toFixed(2)} (${md.slope(M, c).neff.toFixed(2)})`, 17)).join('')); +const tMyr = z => bg.age(z) * 1e3; +console.log('\n Collapse redshift of a 1 sigma peak, sigma(M) D(z_c) = 1.686, and the cosmic age then [Myr]'); +console.log(' M [Msun] ' + models.map(c => c.short.padStart(17)).join('')); +for (const M of masses) console.log(` ${pad(M.toExponential(0), 7)} ` + models.map(c => { const z = md.zc(M, c); return pad(z < 0 ? 'not yet' : `${z.toFixed(1)} (${tMyr(z).toFixed(0)})`, 17); }).join('')); +console.log(' Below a cut-off sigma stops growing, so the flat redshifts there are not halos: nothing smaller collapses on its own.'); +const chain = [1e-12, 1e-9, 1e-6, 1e-3].map(M => ({ M, z: md.zc(M) })); +console.log(` CDM without a cut-off: ${chain.map(c => `${e2(c.M)} at z ${c.z.toFixed(1)}`).join(', ')}: nine decades within ${(tMyr(chain[3].z) - tMyr(chain[0].z)).toFixed(0)} Myr`); +const zb = o => md.zc(Mb, bino, o).toFixed(1), nu60 = (dc, o) => (dc / (md.sigma(Mb, bino) * md.growth(60, o))).toFixed(1); +console.log(`\n Bino halos at M_hm = ${e2(Mb)} Msun, 1 sigma: z_c = ${zb()}; with the radiation-era growing mode ${zb({ radiation: true })};` + + ` with the dark matter growing alone (the gas never falls in) ${zb({ baryonLag: true })} to ${zb({ baryonLag: 'cdm' })}; 2 sigma ${zb({ nu: 2 })}, 3 sigma ${zb({ nu: 3 })}.`); +console.log(` To collapse (1.686) at z = 60 takes a ${nu60(H.DELTA_C)} sigma peak, ${nu60(H.DELTA_C, { baryonLag: 'cdm' })} to ${nu60(H.DELTA_C, { baryonLag: true })} with the dark matter growing alone;` + + ` to reach delta = 1 there (Green, Hofmann & Schwarz's threshold), ${nu60(1)} sigma.`); +const runs = [-0.01, 0.01].map(a => H.model(P18, { perDecade: 160, running: a })); +console.log(` Running dn_s/dlnk = -0.01 / +0.01 (sigma8 fixed): sigma(M_hm) ${runs[0].sigma(Mb, bino).toFixed(1)} / ${runs[1].sigma(Mb, bino).toFixed(1)} (${md.sigma(Mb, bino).toFixed(1)} without), z_c ${runs[0].zc(Mb, bino).toFixed(1)} / ${runs[1].zc(Mb, bino).toFixed(1)}`); +// Neutrinos still coupled at horizon entry: scale T by the fluid/free tail ratio, ramped in ln k over a decade above k_nu. +const ramp = k => { const t = Math.min(1, Math.max(0, Math.log(k / kNu) / Math.LN10)); return 1 - t + t * tailFluid.T(k) / tailFree.T(k); }; +const nuF = H.model(P18, { perDecade: 160, factor: ramp }); +console.log(` Neutrinos coupled at horizon entry (the fluid tail from ${e2(10 * kNu)} Mpc^-1 up): sigma(M_hm) ${nuF.sigma(Mb, bino).toFixed(1)}, z_c ${nuF.zc(Mb, bino).toFixed(1)}`); +const wm = H.model(H.WMAP_ERA, { perDecade: 160 }), MbW = H.massOfK(wm.bg, H.halfMode(bino)); +console.log(` WMAP-era round numbers (h 0.7, Om 0.3, n_s 1, sigma8 0.9): sigma(M_hm) ${wm.sigma(MbW, bino).toFixed(1)}, z_c 1 sigma ${wm.zc(MbW, bino).toFixed(1)}, 2 sigma ${wm.zc(MbW, bino, { nu: 2 }).toFixed(1)}`); +// Green, Hofmann & Schwarz's z = 60 +- 20 (JCAP 2005, "The first structures") is z_nl, where typical fluctuations reach +// sigma(R, z_nl) = 1: top hat, sigma8 = 0.9, n = 1, baryons ignored. The same here, at M_hm and M(pi/k_fs). +const ghsLike = H.model({ ...H.WMAP_ERA, Ob: 1e-7 }, { perDecade: 160 }); +const znl = m => [H.halfMode(bino), bino.kfs].map(k => m.zc(H.massOfK(m.bg, k), bino, { delta: 1 })); +const zn = [md, wm, ghsLike].map(znl); +console.log(` Onset of nonlinearity for typical fluctuations, sigma(M, z_nl) = 1, at M_hm / M(pi/k_fs): Planck 2018 ${zn[0].map(z => z.toFixed(1)).join(' / ')};` + + ` WMAP-era ${zn[1].map(z => z.toFixed(1)).join(' / ')};`); +console.log(` WMAP-era without baryons ${zn[2].map(z => z.toFixed(1)).join(' / ')}. Green, Hofmann & Schwarz 2005: z_nl = 60 +- 20, (1 + 60)/(1 + z_nl) = ` + + `${zn[2].map(z => (61 / (1 + z)).toFixed(2)).join(' / ')}`); +console.log(` The bino's power per log k, Delta^2 T^2, peaks at k = ${e2(md.peak(bino))} Mpc^-1, a factor ${(hm(bino).k / md.peak(bino)).toFixed(1)} below k_hm: just before the cut-off.`); + +// ---- 4. Nesting +const muAq = H.muAquarius(); +console.log(`\n4. Nesting below a 1e12 Msun Milky Way: decades from the floor, and levels if each level is 10x, 100x or mu = ${muAq.toFixed(4)} below its host`); +console.log(' (mu: where the Aquarius law expects the largest subhalo, N(>mu M_host) = 1). Subhalos above the floor, dN/dM = a0 (M/m0)^-1.9,'); +console.log(' m0 = 2.52e7 Msun: a0 = 1.77e-5/Msun for a 1e12 Msun host within 353 kpc, and 3.26e-5 for Aq-A (1.84e12, within 433 kpc);'); +console.log(' f_sub: the mass in subhalos from the floor to 1e10 Msun over 1e12 Msun. All extrapolated below about 1e5 Msun, tides ignored.'); +console.log(' Levels at mu (exact fraction in brackets) are self-similar, an upper bound (Springel et al.); 2x/3x: sub-subhalos 2 or 3 times fewer.\n'); +const floors = [ + ['WIMP range, lightest (Bringmann; all matter)', 1e-11 / bg.fc], + ...[[40, 'bringmann'], [30, 'bringmann'], [30, 'halfMode'], [10, 'bringmann']].map(([Tkd, key]) => + [`Bino, T_kd ${Tkd} MeV, ${key === 'bringmann' ? 'max(M(pi/k_fs), M_ao)' : 'max(M_hm, M_ao)'}`, H.wimpFloor(bg, { m: 100, Tkd })[key]]), + ['WIMP range, heaviest (Bringmann; all matter)', 1e-3 / bg.fc], + ...wave.flatMap(w => [[`wave ${w.m.toExponential(1)} eV, M_hm`, w.Mhm], [`wave ${w.m.toExponential(1)} eV, Schive M0`, w.M0]]), +]; +console.log(' floor M_floor decades levels 10x 100x mu (exact) 2x/3x N(>floor), 1e12 Aq-A f_sub'); +for (const [name, M] of floors) { + const n = H.nesting(M); + console.log(` ${name.padEnd(45)} ${pad(e2(M), 9)} ${pad(n.decades.toFixed(1), 6)} ${pad(n.levels[0], 10)} ${pad(n.levels[1], 4)} ${pad(`${n.levels[2]} (${n.exact[2].toFixed(2)})`, 11)}` + + ` ${pad(n.suppressed.join('/'), 7)} ${pad(e2(H.subhaloCount(M)), 15)} ${pad(e2(H.subhaloCount(M, Infinity, H.AQUARIUS)), 9)} ${pad(M < 1e10 ? e2(H.subhaloMass(M, 1e10) / 1e12) : '-', 7)}`); +} +const mu50 = H.muAquarius(H.AQUARIUS_MW, { m50: true }), lev50 = M => H.nesting(M, H.AQUARIUS_MW.m50, [mu50]).levels[0], f10 = H.wimpFloor(bg, { m: 100, Tkd: 10 }).bringmann; +console.log(`\n mu is against the host's M200. Against M50, the mass inside r50 where the law was fitted, it is ${mu50.toFixed(4)} (Aq-A: ` + + `${H.muAquarius(H.AQUARIUS, { m50: true }).toFixed(4)}),\n and the levels at mu, counted down from M50, are ${lev50(floors[0][1])} to ${lev50(floors[5][1])} for Bringmann's range and ` + + `${lev50(floors[1][1])} to ${lev50(f10)} for the bino, against ${H.nesting(floors[0][1]).levels[2]} to ${H.nesting(floors[5][1]).levels[2]} and ${H.nesting(floors[1][1]).levels[2]} to ${H.nesting(f10).levels[2]}.`); +console.log(` In the dark matter density alone (Bringmann's convention) the bino at T_kd 10 MeV has ${H.nesting(bg.fc * f10).exact[2].toFixed(2)} levels at mu,` + + ` not ${H.nesting(f10).exact[2].toFixed(2)}: a knife edge between 9 and 10.`); +console.log('\n Wave dark matter, decades from the population floor (M_1/2, M_hm, M0) and, for comparison, from the Jeans mass today:'); +for (const w of wave) { + const ns = [w.M2, w.Mhm, w.M0].map(M => H.nesting(M)), d = ns.map(n => n.decades); + const span = lev => (Math.min(...lev) === Math.max(...lev) ? `${Math.min(...lev)}` : `${Math.min(...lev)}-${Math.max(...lev)}`); + console.log(` ${w.m.toExponential(1)} eV: ${Math.min(...d).toFixed(1)}-${Math.max(...d).toFixed(1)} decades, ${span(ns.map(n => n.levels[2]))} levels at mu ` + + `(${span(ns.map(n => n.suppressed[0]))} / ${span(ns.map(n => n.suppressed[1]))} with sub-subhalos 2x / 3x fewer); Jeans mass today ${H.nesting(w.MJ).decades.toFixed(1)} decades`); +} + +// ---- 5. The Jeans scale and the soliton +console.log('\n5. The Jeans scale in the Witness Map\'s terms (its constants): the wave clock m dx^2/hbar at dx = 1/k_J against 1/(4 sqrt(pi G rho))'); +for (const [lab, rho] of [['mean matter today', bg.rhoMsi], ['mean matter at equality', bg.rhoMsi / Math.pow(bg.aeq, 3)], ['the Sun\'s orbit, 0.4 GeV/cm^3', 0.4 * WM.GEV_CM3]]) { + const J = H.jeans(1e-22, rho, WM.C); + console.log(` ${lab.padEnd(30)} 1e-22 eV: lambda_J = ${e2(J.lambdaJ / WM.C.kpc)} kpc; wave clock ${e2(J.tWave / WM.C.yr)} yr = gravity's ${e2(J.tGrav / WM.C.yr)} yr; free fall ${e2(J.tFF / WM.C.yr)} yr`); +} +console.log(' At dx = lambda_J/2 = pi/k_J the clock is pi^2 = 9.87 times longer: the factor is a choice of convention.'); +const kJ0 = H.kJcomoving(bg, 1, 1), kJeq = H.kJcomoving(bg, 1, bg.aeq), kJdm = H.kJcomoving(bg, 1, 1, { rho0: 0.12 * bg.rhoCrit }); +console.log(` Comoving k_J at 1e-22 eV: today ${kJ0.toFixed(1)} Mpc^-1 (all matter) or ${kJdm.toFixed(1)} (0.12 rho_crit/h^2, the dark matter alone: the quoted 66.5);` + + ` at equality ${kJeq.toFixed(2)}`); +{ const toM = rho => rho * Math.pow(H.C.Mpc, 3) / H.C.Msun, MJ = rho => H.jeansMasses(bg, H.kJcomoving(bg, 1, 1, { rho0: rho }), toM(rho)).halfWave; + console.log(` Jeans mass today, sphere of diameter lambda_J, in the dark matter density alone: ${e2(MJ(0.12 * bg.rhoCrit))} Msun` + + ' (k_J and M_J both go as rho^(1/4));'); + console.log(` at Hui et al.'s normalisation, Omega_FDM = 0.27 and h = 0.7: ${(MJ(0.27 * 0.7 * 0.7 * bg.rhoCrit) / 1e7).toFixed(3)}e7 Msun (their 1.5e7)`); } +const J0 = H.jeansMasses(bg, kJ0), Jeq = H.jeansMasses(bg, kJeq); +console.log(` ${'Jeans mass at 1e-22 eV in all the matter [Msun]'.padEnd(50)} ${'today'.padStart(9)} at equality`); +for (const [key, lab] of [['fullWave', 'sphere of radius lambda_J'], ['cube', 'cube of side lambda_J'], ['halfWave', 'sphere of diameter lambda_J (Hui et al.\'s sphere)'], ['inverseK', 'sphere of radius 1/k_J']]) + console.log(` ${lab.padEnd(50)} ${pad(e2(J0[key]), 9)} ${pad(e2(Jeq[key]), 9)}`); +console.log(` Spread: ${e2(Jeq.fullWave / J0.inverseK)} between the largest and smallest; M_J ~ m^-3/2 while M_hm and M0 go as m^-4/3`); +const s = H.solitonRelation(WM.SOL), u22 = H.hbar2Gm2(1e-22, WM.C) / (WM.C.Msun * WM.C.kpc); +console.log(`\n Soliton (the Schroedinger-Poisson ground state, solved in witness-map/model.js), r_c its half-density radius:`); +console.log(` M( (x === 0 ? '0' : x.toExponential(2)); +function within(name, got, want, factor, src) { + const ok = got > 0 && got / want <= factor && want / got <= factor; + console.log(`${ok ? 'PASS' : 'FAIL'} ${name}\n model ${fmt(got)} expected ${fmt(want)} (within x${factor}) ${src}`); + if (!ok) fails++; +} +function near(name, got, want, rel, src) { + const ok = Math.abs(got / want - 1) <= rel; + console.log(`${ok ? 'PASS' : 'FAIL'} ${name}\n model ${got.toPrecision(5)} expected ${want.toPrecision(5)} (within ${rel * 100}%) ${src}`); + if (!ok) fails++; +} +function check(name, ok, detail, src) { + console.log(`${ok ? 'PASS' : 'FAIL'} ${name}\n ${detail} ${src}`); + if (!ok) fails++; +} +const note = (name, detail, src) => console.log(`NOTE ${name}\n ${detail} ${src}`); + +const P18 = H.PLANCK18, md = H.model(P18, { perDecade: 60 }), bg = md.bg, h = P18.h; +const bino = H.wimpGHS({ m: 100, Tkd: 30 }), Mhm = H.massOfK(bg, H.halfMode(bino)); +const GHS = 'Green, Hofmann & Schwarz, JCAP 0508, 003 (2005), arXiv:astro-ph/0503387', GHS04 = 'Green, Hofmann & Schwarz, MNRAS 353, L23 (2004)'; + +// 1. The spectrum sigma(M) and the collapse redshifts sit on: Eisenstein & Hu's no-wiggle transfer function extrapolated +// to 1e10 h/Mpc, twenty decades in mass below where it was fitted. Two checks of the integration, then the +// extrapolation itself. +{ // sigma(1 Mpc/h)/sigma8 by a direct quadrature with the exact top-hat window (the grid phase-averages W^2 at kR > 10). + const n = 400000, l0 = Math.log(1e-5), l1 = Math.log(1e3), dl = (l1 - l0) / n, W2 = x => { const w = 3 * (Math.sin(x) - x * Math.cos(x)) / (x * x * x); return w * w; }; + let s1 = 0, s8 = 0; + for (let i = 0; i < n; i++) { const k = Math.exp(l0 + (i + 0.5) * dl), d = Math.pow(k, 3 + P18.ns) * Math.pow(md.tf.Tnw(k), 2); s1 += d * W2(k); s8 += d * W2(8 * k); } + const Mr = 4 * Math.PI / 3 * bg.rhoM * Math.pow(1 / h, 3); + near('sigma(R = 1 Mpc/h)/sigma8: the 60-per-decade grid against a direct quadrature with the exact window', md.sigma(Mr) / P18.sigma8, Math.sqrt(s1 / s8), 1e-3, + 'numerical (400,000 points in ln k)'); + const fine = H.model(P18, { perDecade: 400, kmax: 1e12 }); + near('sigma at the bino half-mode mass: 60 points per decade to 1e10 h/Mpc against 400 to 1e12', md.sigma(Mhm, bino), fine.sigma(Mhm, bino), 2e-5, 'numerical convergence'); +} +// A mode that enters the horizon in the radiation era kicks the dark matter logarithmically. With the neutrinos +// a perfect fluid the asymptote is known exactly; with them streaming freely it is not, and Eisenstein & Hu fitted +// a Boltzmann code that let them stream. Modes entering before neutrino decoupling see the fluid case instead. +const fluid = H.radiationEra({ free: false }), freeNu = H.radiationEra({ lmax: 60, xmax: 800 }); +const tailFluid = H.smallScaleTail(bg, fluid), tailFree = H.smallScaleTail(bg, freeNu), noB = H.HOM.transferEH({ ...P18, Ob: 1e-7 }); +check('Radiation era, neutrinos a perfect fluid: delta_c -> -A ln(B k tau) with A = 9 psi0 and B = e^(gamma_E - 1/2)/sqrt3 (to 1e-3)', + Math.abs(fluid.A / fluid.psi0 / 9 - 1) < 1e-3 && Math.abs(fluid.B / (Math.exp(0.5772156649 - 0.5) / Math.sqrt(3)) - 1) < 1e-3, + `A/psi0 = ${(fluid.A / fluid.psi0).toFixed(5)}, B = ${fluid.B.toFixed(5)} (exact 9 and ${(Math.exp(0.5772156649 - 0.5) / Math.sqrt(3)).toFixed(5)})`, + 'closed form, derived in model.js above radiationEra(); Ma & Bertschinger, ApJ 455, 7 (1995), for the equations'); +const ratio = (k, tail) => noB.Tnw(k) / tail.T(k * h); +check("Eisenstein & Hu's tail, baryons removed, within 3% of the free-streaming-neutrino tail from 1e3 to 1e9 h/Mpc", + [1e3, 1e6, 1e9].every(k => Math.abs(ratio(k, tailFree) - 1) < 0.03), + `EH/derived ${[1e3, 1e6, 1e9].map(k => `${ratio(k, tailFree).toFixed(3)} at ${fmt(k)}`).join(', ')}; derived tail ${tailFree.coef.toFixed(2)} (k_eq/k)^2 ln(${tailFree.c.toFixed(4)} k/k_eq)`, + 'Eisenstein & Hu, ApJ 496, 605 (1998), eqs. 26-31, fitted to a Boltzmann code with free-streaming neutrinos'); +note('With the neutrinos still a fluid when the mode enters, the tail is lower than Eisenstein & Hu', + `EH/fluid ${[1e5, 1e7, 1e9].map(k => `${ratio(k, tailFluid).toFixed(3)} at ${fmt(k)}`).join(', ')}: bino-scale modes entered before the neutrinos decoupled, so EH overstates them by about 10%`, + 'model only: no published value'); + +// 2. The WIMP floor (100 GeV bino, kinetic decoupling at 30 MeV unless stated). +{ + const derived = H.wimpFreeStream(bg, { m: 100, Tkd: 30 }); + near('Free-streaming wavenumber derived from the coasting distance after kinetic decoupling, through the thermal history (Mpc^-1)', derived.kfs, 1.70e6, 0.05, + 'Green, Hofmann & Schwarz, astro-ph/0508553 and JCAP 0508, 003 (2005): k_fs = 1.70e6 Mpc^-1, as seen in search text; checked in review, with the same sqrt2 ' + + '(this checks k_fs = sqrt2/s through this model\'s thermal history)'); +} +within('Bino cut-off mass, a sphere of radius pi/k_fs (Msun)', H.massOfK(bg, bino.kfs), 1e-6, 2, GHS + ': "about 1e-6 Msun", as seen in the abstract; ' + + 'checked in review, which found the sphere of radius pi/k_fs to be their stated definition (' + GHS04 + ', gives 9e-7 for its model A)'); +within('Collisional damping mass, a sphere of radius pi/k_d (Msun)', H.massOfK(bg, bino.kd), 1e-10, 2, + GHS + ': "M_d ~ 1e-10 Msun" for a sphere of radius pi/k_d; ' + GHS04 + ': 9e-11 at T_kd = 33 MeV; both as #35\'s review read them in the arXiv source'); +{ // Cornell et al. write M_ao with the dark matter density rho_chi, f_c of the matter density used here. + const lz = H.acousticMass(bg, 10), m30 = H.acousticMass(bg, 30), g30 = H.decoupling(bg, 30).g, cps = 3.4e-6 * Math.pow(30 * Math.pow(g30, 0.25) / 50, -3); + const chi30 = bg.fc * m30; + check('Acoustic mass, the mass inside the horizon at kinetic decoupling, against two published scalings', + lz / 1e-4 < 1.5 && lz / 1e-4 > 1 / 1.5 && chi30 / cps < 2 && chi30 / cps > 0.5, + `${fmt(lz)} Msun of matter at 10 MeV against Loeb & Zaldarriaga's 1e-4 (T/10 MeV)^-3 (within x1.5); at 30 MeV, the dark matter alone, as Cornell et al. count it, ` + + `${fmt(chi30)} (${(chi30 / cps).toFixed(2)} times theirs; all matter ${fmt(m30)}, ${(m30 / cps).toFixed(2)} times) against their 3.4e-6 (T g^(1/4)/50 MeV)^-3 = ${fmt(cps)} ` + + `with this model's g = ${g30.toFixed(2)} (within x2)`, + 'Loeb & Zaldarriaga, PRD 71, 103520 (2005); Cornell, Profumo & Shepherd, PRD 88, 015027 (2013), eq. 14, with rho_chi; both as seen in search text, checked in review'); +} + +// 3. The wave floor. Hu, Barkana & Gruzinov's Jeans scale and cut-off, and the conventions behind the numbers quoted. +const kJeq = H.kJcomoving(bg, 1, bg.aeq); +near('Comoving Jeans wavenumber at matter-radiation equality, 1e-22 eV, mean matter density (Mpc^-1)', kJeq, 9, 0.05, + 'Hu, Barkana & Gruzinov, PRL 85, 1158 (2000): k_J,eq = 9 m22^(1/2) Mpc^-1, as seen in search text; checked in review'); +{ + const rhoDM = 0.12 * bg.rhoCrit, k0 = H.kJcomoving(bg, 1, 1, { rho0: rhoDM }), kM = H.kJcomoving(bg, 1, 1); + near('Jeans wavenumber today, 1e-22 eV, in a density of 0.12 rho_crit/h^2 (the dark matter alone) (Mpc^-1)', k0, 66.5, 0.005, + `Marsh, Phys. Rep. 643, 1 (2016), the equation labelled axion_jeans in its arXiv source: k_J = 66.5 a^(1/4) (Omega_a h^2/0.12)^(1/4) m22^(1/2), as quoted by Kulkarni & Ostriker (arXiv:2011.02116), ` + + `seen in search text; checked in review, Omega_a the axion density; all matter gives ${kM.toFixed(1)}`); +} +near("Hu, Barkana & Gruzinov's half-power wavenumber, T^2 = 1/2, at 1e-22 eV (Mpc^-1)", H.halfMode(H.fdmHBG(1), { power: true }), 4.5, 0.03, + 'Hu, Barkana & Gruzinov (2000): 4.5 m22^(4/9) Mpc^-1, as seen in search text; checked in review'); +{ // Hui et al. normalise to the wave's own density, Omega_FDM = 0.27 with h = 0.7. + const toM = rho => rho * Math.pow(H.C.Mpc, 3) / H.C.Msun, MJ = rho => H.jeansMasses(bg, H.kJcomoving(bg, 1, 1, { rho0: rho }), toM(rho)).halfWave; + const Oc = P18.Om - P18.Ob, hui = 1.5e7 * Math.pow(Oc / 0.27, 0.25) * Math.sqrt(h / 0.7); + near('Jeans mass today, 1e-22 eV, a sphere of diameter lambda_J in Omega_FDM = 0.27, h = 0.7 (Msun): a lower bound on any halo, not the floor of the halo population', + MJ(0.27 * 0.7 * 0.7 * bg.rhoCrit), 1.5e7, 0.01, 'Hui, Ostriker, Tremaine & Witten, PRD 95, 043541 (2017), the equation labelled jeans in its arXiv source: M_J = (4/3) pi rho (lambda_J/2)^3 = ' + + '1.5e7 (Omega_FDM/0.27)^(1/4) (H0/70)^(1/2) m22^(-3/2) Msun, as quoted in arXiv:2106.02526, seen in search text; checked in review. The 1% is tighter ' + + 'than the quoted rounding (about 3%), to separate the wave\'s own density from all the matter. ' + + `At this model's cosmology: dark matter alone ${fmt(MJ(Oc * h * h * bg.rhoCrit))}, their formula ${fmt(hui)}; all matter ${fmt(MJ(bg.rhoMsi))}`); +} +{ // The Witness Map's wave clock m dx^2/hbar across dx = 1/k_J is gravity's clock, with the Witness Map's constants. + let worst = 0; + for (const rho of [bg.rhoMsi, bg.rhoMsi / Math.pow(bg.aeq, 3), 0.4 * WM.GEV_CM3]) for (const m22 of [1, 20, 200]) { + const J = H.jeans(m22 * 1e-22, rho, WM.C); + worst = Math.max(worst, Math.abs(J.tWave / J.tGrav - 1), Math.abs(J.kJ / J.closed - 1)); + } + check('The wave clock at 1/k_J equals 1/(4 sqrt(pi G rho)), k_J found as the root of omega^2(k) = 0', worst < 1e-12, + `worst relative deviation ${worst.toExponential(1)} over 3 densities and 3 masses (tolerance 1e-12)`, + 'algebra: an identity for any constants, so this checks only the bisection root against the closed form'); +} +{ // The smallest wave structure: the soliton, the Schroedinger-Poisson ground state the Witness Map already solves. + const s = H.solitonRelation(WM.SOL), unit = H.hbar2Gm2(1e-23, WM.C) / (WM.C.Msun * WM.C.kpc / 1e3); + near('Soliton: mass inside the half-density radius times that radius, m = 1e-23 eV (Msun pc)', s.coreRc * unit, 5.5e12, 0.05, + 'Schive, Chiueh & Broadhurst, Nature Phys. 10, 496 (2014), Supplementary Information: the core mass-radius relation M_c ~ 5.5e9 m23^-2 (r_c/kpc)^-1 Msun, ' + + 'as #35\'s review read it (first seen restated in arXiv:2603.07175, which credits it to PRL 113, 261302; that paper does not contain it); ' + + `here ${s.coreRc.toFixed(3)} hbar^2/(G m^2); their fitted profile, integrated to r_c, gives 5.3e12 (halos/table.mjs)`); +} + +// 4. Counting the nesting: Springel et al.'s Aquarius subhalo law extended to the floor. +near('Aquarius normalisation scaled from the 1.84e12 Msun Aq-A to a 1e12 Msun host', H.AQUARIUS.a0 * 1e12 / H.AQUARIUS.host, H.AQUARIUS_MW.a0, 0.01, + 'Erkal et al., MNRAS 463, 102 (2016), quoting Springel et al., MNRAS 391, 1685 (2008), seen in search text; Aq-A\'s checked in review (a guard on the transcribed pair)'); +{ // Diemand, Moore & Stadel: "over 10^15" survive; their own m^-1.9 law, extrapolated, gives about 5e15 above 1e-6 Msun. + const nMW = H.subhaloCount(1e-6), nAq = H.subhaloCount(1e-6, Infinity, H.AQUARIUS), DMS = 'Diemand, Moore & Stadel, Nature 433, 389 (2005)'; + check('Earth-mass halos in Aq-A: the Aquarius law down to 1e-6 Msun, tides ignored, holds at least the "over 10^15" that survive', + nAq >= 1e15, `Aq-A (1.84e12 Msun, within 433 kpc) ${fmt(nAq)}, ${((nAq / 1e15 - 1) * 100).toFixed(0)}% above 1e15`, + DMS + ': "over 10^15" survive in the Galactic halo, as seen in the arXiv abstract; checked in review'); + note('Earth-mass halos, like for like: the Aquarius law against Diemand, Moore & Stadel\'s own m^-1.9 law, both without tides', + `above 1e-6 Msun: Aq-A ${fmt(nAq)}, ${(5e15 / nAq).toFixed(1)} times fewer than 5e15; a 1e12 Msun host (within 353 kpc) ${fmt(nMW)}, ${(5e15 / nMW).toFixed(1)} times fewer ` + + `(their host mass is not given); that host is also ${(1e15 / nMW).toFixed(2)} times short of the 1e15 that survive`, DMS + ', body text after Fig. 3: N(>1e-6 Msun) ~ 5e15, as #35\'s review read it in the arXiv source'); +} +{ // Where the levels come from: each level's largest subhalo, where the Aquarius law expects one, N(>M) = 1. + const mu = H.muAquarius(), muAq = H.muAquarius(H.AQUARIUS); + note('Levels Aquarius itself resolves', `the largest expected subhalo is mu = ${mu.toFixed(4)} of a 1e12 Msun host's M200 (${(-Math.log10(mu)).toFixed(2)} decades per level), ` + + `${muAq.toFixed(4)} of Aq-A's; four levels at that ratio below Aq-A's 1.84e12 Msun reach ${fmt(H.AQUARIUS.host * Math.pow(muAq, 4))} Msun, or ${[3, 2].map(f => fmt(H.AQUARIUS.host * muAq * Math.pow(muAq * Math.pow(f, -1 / (H.AQUARIUS.n - 1)), 3))).join(' to ')} with sub-subhalos 3 or 2 times fewer; published: "up to four generations" of subhalos within subhalos, ` + + 'and sub-subhalos "a factor of 2 or 3 below" the self-similar count, which "provides an upper limit"', + 'Springel et al. (2008), abstract, as seen in search text; Sec. 4, as #35\'s review read it'); +} + +// 5. When the bino's fluctuations go nonlinear. Green, Hofmann & Schwarz's z = 60 +- 20 is z_nl, where typical +// fluctuations reach sigma(R, z_nl) = 1 (top hat, sigma8 = 0.9, n = 1, baryons ignored), not this model's 1.686. +// A comparison, not a check: the rest of their calculation is not this one, and a factor of about 1.3 in 1 + z +// remains. +{ + const nl = m => [H.halfMode(bino), bino.kfs].map(k => m.zc(H.massOfK(m.bg, k), bino, { delta: 1 }).toFixed(1)).join('-'); + const wm = H.model(H.WMAP_ERA, { perDecade: 60 }), ghsLike = H.model({ ...H.WMAP_ERA, Ob: 1e-7 }, { perDecade: 60 }), z0 = ghsLike.zc(H.massOfK(ghsLike.bg, bino.kfs), bino, { delta: 1 }); + note('Bino: onset of nonlinearity for typical fluctuations, sigma(M, z_nl) = 1 at M_hm to M(pi/k_fs): published z_nl = 60 +- 20', + `Planck 2018 ${nl(md)}; WMAP-era round numbers (h = 0.7, Omega_m = 0.3, n_s = 1, sigma8 = 0.9) ${nl(wm)}; the same without baryons ${nl(ghsLike)}, ` + + `inside their range, (1 + 60)/(1 + z_nl) = ${(61 / (1 + z0)).toFixed(2)}; this model's first halos (1 sigma, 1.686, M_hm) collapse at z = ${md.zc(Mhm, bino).toFixed(1)}`, + GHS + ', section "The first structures", as #35\'s review read it in the arXiv source'); +} + +console.log(fails ? `\n${fails} check(s) failed` : '\nall checks pass'); +process.exit(fails ? 1 : 0);