From 3ab12c6542c88554ad0d42e0e19aff2461dc3c0d Mon Sep 17 00:00:00 2001 From: Dylan Roy <249851642+d6g8k5htny-coder@users.noreply.github.com> Date: Thu, 8 Oct 2026 17:29:38 -0700 Subject: [PATCH] Prove stationary quartic selected H0 law with explicit sampling weights --- experiments/universality/README.md | 15 + .../universality/quartic_cusp_field/PROOF.md | 751 ++++++++++++++++++ .../universality/quartic_cusp_field/REVIEW.md | 92 +++ 3 files changed, 858 insertions(+) create mode 100644 experiments/universality/quartic_cusp_field/PROOF.md create mode 100644 experiments/universality/quartic_cusp_field/REVIEW.md diff --git a/experiments/universality/README.md b/experiments/universality/README.md index edb79ca74..9c95539f8 100644 --- a/experiments/universality/README.md +++ b/experiments/universality/README.md @@ -163,6 +163,21 @@ AMEND, corrected preparation reviews and completed incorporated-source review. This exact declared mixture admits no arbitrary non-Gaussian law, numerical sampler, blind experiment or same-field spatial process. +The [stationary quartic field construction](quartic_cusp_field/PROOF.md) +gives a second actual singularity mechanism. A smooth global torus family +has exactly one finite H0 bar, proved by a complete critical list and +superlevel enclosure, with exact lifetime C*lambda^2. Haar translation +makes its full field law stationary. A parameter density proportional to +lambda^(gamma-1) gives exact lifetime density power gamma/2-1 on its +specified finite interval, and an independent-whole-copy marked Poisson +limit. Flat parameter density realizes -1/2, while gamma=4/3 realizes +-1/3 at the same quartic germ. Contact sampling weight therefore belongs +in the exponent taxonomy. Its [review](quartic_cusp_field/REVIEW.md) +preserves the exact field/topology/probability proof and input exposure. +This restricted parameter path is C1 but not C2 at zero and its two-site +observation law is singular; it supplies no generic two-parameter cusp, +Gaussian jet-law admission, numerical coupling or spatial process. + These conventional analytic results discharge selected-bar counting and the leading near/far intensity composition for the stated ideal finite laws and admitted stationary Gaussian spectral/carrier laws. Further diff --git a/experiments/universality/quartic_cusp_field/PROOF.md b/experiments/universality/quartic_cusp_field/PROOF.md new file mode 100644 index 000000000..92c8effc4 --- /dev/null +++ b/experiments/universality/quartic_cusp_field/PROOF.md @@ -0,0 +1,751 @@ +# Actual stationary quartic contact: sampling-dependent selected H0 law + +Incorporated on 8 October 2026 by OpenAI/Codex root, task +01a11cbf-068d-7102-861e-75814e715c98, under exact GitHub307 amendment +6071510105 and the existing README claim. Mathematical author: +OpenAI/Codex universality_review. Initial detached base is published +amplitude draft334 head `6bd133d591e43e75a9292463bb4bfde7a34984dc`. +No main landing or model-catalogue admission is implied. + +Sections1-11 preserve the original32,334-byte/695-line preparation, +SHA256 `24e6f4149c0a2df2f8df1dbfa8d66da3e01a7d12bee4b803bac093cfea36bf3f`, +exactly: 26,625 bytes/602 lines, SHA256 +`9f1426a21b40b524ad4b25c81368489b0aa7a0c29f79659eb556bb4d1579f6d0`. +Both final_contract_review and benchmark_formal_audit have separate fresh +FULL original695-line mathematical PASS verdicts. Root fully read the +actual original1-350/351-695. These new wrapper/context/README bytes await +their own incorporated-source and metadata reviews; original verdicts are +not rebound. The [review record](REVIEW.md) retains source identities. + +The self-contained construction gives an actual globally selected finite +H0 bar in a stationary smooth field, with quartic contact and exact +lifetime C*lambda^2. Sampling weight near zero determines its density +power. The path is restricted and only C1 in its parameter at zero; +no generic full two-parameter cusp or Gaussian contact law is admitted. +All performers are prior source/route-exposed OpenAI/Codex AI agents, +organizational independence0. No numerical, formal, blind or scientific +promotion follows from these conventional analytic results. + +## 1. Exact claim and conventions + +Fix an integer d>=2, a side length L>0, and the flat torus + + X=(R/LZ)^d, V=L^d, omega=2 pi/L. + +Homology is ordinary superlevel persistent homology over an arbitrary +fixed field K. A finite H0 interval has a maximum birth value B and a +merging-saddle death value D, with lifetime ell=B-D>0. The component with +the larger maximum value survives a merger. One H0 class is essential +because X is connected. Essential intervals are excluded from all counts. + +For any + + 00, (1) + +we construct a deterministic family F_lambda, 0<=lambda<=lambda_0, +of C-infinity functions on X and a stationary random field Z. Its random +parameters are a uniform translation and an independent lambda with +density gamma lambda^(gamma-1)/lambda_0^gamma on (0,lambda_0). +The positive constant lambda_0 is chosen by explicit deterministic +inequalities in Section 5. Optional deterministic centering and variance +normalization are included in Z, and never change the pairing. + +There are constants C>0 and ell_0=C lambda_0^2 such that each sample of Z +has exactly one finite H0 bar, and + + ell=C lambda^2, + P(ell<=t)=(t/ell_0)^(gamma/2), 0 0, n t_n^beta -> Lambda in (0,infinity), (3) + +their finite-H0 points with lifetime mark ell/t_n and birth-location +mark x converge to a Poisson random measure with intensity + + Lambda beta ell_0^(-beta) a^(beta-1) da dx/V. (4) + +The exact direction from the birth maximum to its death saddle is the +positive first coordinate direction, so that mark can be added as a +Dirac mass at e_1. Section 10 proves the Laplace functional directly. +This is an independent-copies process, not a spatial expanding-domain +limit or a Poisson statement inside one field. + +## 2. An explicit Morse seed and exact maximum coordinates + +Let w_i=3^(-i), i=1,...,d, and define + + h(y)=sum_i w_i cos(omega y_i), b=sum_i w_i. (5) + +The critical points have y_i in {0,L/2}. At each one the Hessian is +diagonal with nonzero entries -omega^2 w_i cos(omega y_i). +There are exactly 2^d critical points, all nondegenerate, and a unique +maximum at y=0. Their values are distinct. Indeed, if two sign patterns +differ first at index i, their value difference has leading magnitude +2w_i, while the absolute sum of all later differences is at most +2 sum_(j>i) w_j < w_i. In particular every other critical value is at +most b-Delta, where + + Delta=2 w_d>0. + +Use representatives |y_i|0 with R^2b-R^2} is exactly the chart ball D_R={|z|[0,1] +such that + + c(v)=v/4 for 0<=v<=r_1^2, + c(v)=1 for v>=R^2/2, + c(v)>0 for v>0. (7) + +Such a function exists with c'>=0. To see this without a matching +assumption, take a smooth nonnegative derivative d_0 equal to 1/4 on +[0,r_1^2], tapering to zero strictly before R^2/2. Its integral I_0 is +less than 1, since R^2<1. Add a nonnegative smooth bump, supported in +(r_1^2,R^2/2), with integral 1-I_0, and integrate their sum from 0. +The resulting c has exactly the properties (7). + +Replace h on D_R by + + F_0=b-s^2 c(v)-|t|^2, v=s^2+|t|^2, (8) + +and retain h elsewhere. On the collar v>=R^2/2 the replacement equals +h exactly. Thus it defines a global C-infinity function, including all +derivatives at the chart boundary. In the chart, + + (F_0)_s=-2s[c(v)+s^2 c'(v)], + (F_0)_(t_i)=-2t_i[1+s^2 c'(v)]. (9) + +The second equations force t=0 at any critical point. For s!=0 the +bracket in the first equation is strictly positive. Consequently the +origin is the only critical point in D_R; all other critical points +and their values remain those of h. + +Near the origin, (8) is exactly + + F_0=b-s^4/4-(1+s^2/4)|t|^2. + +The smooth coordinate change u=t sqrt(1+s^2/4) gives + + F_0=b-s^4/4-|u|^2. (10) + +Its Hessian has corank one, with a negative definite transverse block; +its third axial derivative vanishes and its fourth axial derivative is +-6. In this source an A3 quartic contact means precisely this real +function-germ normal form, up to the displayed nonsingular coordinate +changes and a constant. We use no classification theorem to infer it. +The usual A_k function-germ notation is independently identifiable in +Arnold--Gusein-Zade--Varchenko's primary chapter preview, which gives +x^(k+1)+y^2 as its A_k series. Our signs specify the real maximum-type +variant. [Actual primary preview, pp. 242--243](https://page-one.springer.com/pdf/preview/10.1007/978-1-4612-5154-5_15). +This is function-germ terminology, not an identification of +wave-front or caustic names in a different convention. + +## 4. The restricted unfolding and its exact three critical values + +Take a smooth radial cutoff chi(z)=psi(|z|^2), with + + 0<=chi<=1, + chi=1 on |z|<=r_1/2, + chi=0 on |z|>=r_1. + +Extend it by zero outside D_R and define, for lambda>=0, + + F_lambda=F_0+chi(z)[lambda s^2/2+eta lambda^(3/2)s]. (11) + +Each field is globally C-infinity in the spatial variable. On the core +|z|<=r_1/2 the formula is exact: + + F_lambda=b+Psi_lambda(s)-(1+s^2/4)|t|^2, + Psi_lambda(s)=-s^4/4+lambda s^2/2+eta lambda^(3/2)s. + +For lambda>0 put s=sqrt(lambda) z and + + Phi_eta(z)=-z^4/4+z^2/2+eta z. + +At a core critical point t=0, and the axial equation is + + -z^3+z+eta=0. (12) + +For (1) it has exactly three simple real roots z_-Phi_eta(z_-)>Phi_eta(z_0). (14) + +For the second inequality, Phi' is negative on (z_-,z_0), so its +integral there is strictly negative. For the comparison of the two +maxima, extend their simple root branches to eta=0. The maximum-value +difference is zero at eta=0 and has derivative z_+(eta)-z_-(eta)>0, +since the derivative of Phi at a root vanishes. This proves the first +inequality for the whole interval (1). Define + + kappa_eta=Phi_eta(z_-)-Phi_eta(z_0)>0. (15) + +The larger maximum has Phi_eta(z_+)>0; this also follows by comparing +its value with Phi_eta(1)=1/4+eta. We do not require the smaller maximum +to lie above b. For eta sufficiently close to the endpoint of (1) its +value can lie below b, which does not affect (14) or the argument. + +## 5. Deterministic parameter bounds and the complete critical list + +All constants in this section depend only on fixed d,L,eta,R,c,chi. +They are chosen before lambda is randomized. Let + + A={r_1/2<=|z|<=r_1}, + g_* = min_A |grad_z F_0|>0. + +Positivity follows from (9) and compactness; this annulus avoids the +only critical point of F_0. Define finite constants + + C_0=sup |chi s^2/2|+eta sup |chi s|, + C_1=sup |grad_z(chi s^2/2)| + +eta sup |grad_z(chi s)|, + C_P=1+max_i |Phi_eta(z_i)|, + delta=r_1^4/64. + +For 0<=lambda<=1, the perturbation in (11) has sup norm at most C_0 +lambda and z-gradient norm at most C_1 lambda on its support. Choose +lambda_0>0 so small that + + lambda_0<1, + C_1 lambda_0r_1, the perturbation vanishes and (9) excludes all +new roots. Its support and boundary are covered by these regions. +All old critical points lie outside D_R and remain unchanged. +The complete critical list consequently contains + + (2^d-1) old critical points plus M_-,S,M_+, + hence 2^d+2 points in total. + +Every one is nondegenerate at positive lambda. The three new critical +values are distinct by (14), and lie in (b-Delta/4,b+Delta/4) by (16). +All old values are at most b-Delta and are mutually distinct. Thus +F_lambda is Morse with distinct critical values. In particular it has +exactly two maxima: the seed's only maximum was replaced by M_- and +M_+. The right maximum M_+ is the unique global maximum. + +There is also a uniform old-height gap. The set of unchanged old +critical values is finite and distinct, so its distinct pairwise +differences have a positive minimum when that set has at least two +elements. Its distance from each of the new values is at least +3Delta/4. There is no vanishing old-critical-value collision in this +family; the only closing critical-height gaps are among the three +explicit local points. + +## 6. A fixed high-level barrier and an actual H0 elder bar + +For any chart point with |z|>=r_1/2, monotonicity gives + + c(|z|^2)>=c(r_1^2/4)=r_1^2/16. + +Since c<=1, its loss in (8) is at least + + s^2 c(|z|^2)+|t|^2 + >= (r_1^2/16)|z|^2 >= delta. + +Outside D_R the loss is at least R^2>delta. Combining this with (16) +gives the global barrier + + F_lambda <= b-3delta/4 outside {|z|b-delta/4. (17) + +Every superlevel set at a height between F_lambda(S) and F_lambda(M_-) +is therefore contained strictly inside the exact core. A path outside +that core cannot connect its two components at any such height. + +The topology inside is explicit. At height q in this interval, its +axial section is determined by + + Psi_lambda(s)>=q-b, + +and the transverse fiber at an allowed s is the closed ball + + |t|^2 <= [b+Psi_lambda(s)-q]/(1+s^2/4). (18) + +There are exactly two axial intervals, one containing M_- and one +containing M_+, because Psi_lambda is a negative-leading quartic with +the three simple critical points already listed and q is strictly +between its middle minimum and its lower maximum. The intervals and +their fibers lie within the core: its boundary is below q by (17). +One can also see this by following the axial quartic from its local +maxima to either boundary; it has no further turning point there. + +The homotopy (s,t)->(s,(1-u)t), 0<=u<=1, remains in the core and +increases F_lambda. It retracts the whole superlevel set to these two +axial intervals. Each interval is connected, its fibers are connected, +and no path between intervals exists because every point projects to +an allowed s. Thus there are exactly two superlevel components. + +At q=F_lambda(S), those intervals first meet at s=sqrt(lambda)z_0; +the fiber there is the single point t=0. Just below this level the +axial section is one interval and its superlevel set is connected. +This proves an actual merger in the complete torus superlevel set. +It is not only a local saddle candidate or a presumed desired pair. + +At its birth M_- forms a new component, while M_+ has the strictly +larger birth value by (14). The elder rule therefore pairs M_- with +S and preserves the component born at M_+. Its exact finite lifetime +before fixed normalization is + + F_lambda(M_-)-F_lambda(S)=kappa_eta lambda^2. (19) + +There are no other finite H0 intervals. A component of a nonempty +superlevel set contains a maximum: maximize F_lambda on that +component, or use the elementary critical-free gradient flow between +critical levels. Each maximum gives one H0 birth and there are only +two maxima. The explicit merger accounts for the only finite death; +the unique global maximum gives the one essential interval. Later +lower levels cannot create new H0 births without another maximum. + +If one additionally tracks ordinary higher homology, the standard +one-handle/one-endpoint rule for a smooth Morse function over K gives +a useful bounded consequence. The three local critical points have +already spent their endpoint roles on the essential H0 birth and +this finite H0 birth/death. Any other finite interval has two old +critical values and hence a lifetime uniformly bounded below by the +old-height gap in Section 5. This uses that named Morse-handle fact, +not an asserted Hk elder generalization. The H0 theorem (2)--(4) does +not need this additional consequence. + +## 7. A genuine stationary random-field law + +Let Y be Haar-uniform on X and independent of a variable lambda with +density + + p_lambda(v)=gamma v^(gamma-1)/lambda_0^gamma, + 0=0, + + sup_(0<=lambda<=lambda_0) ||F_lambda||_(C^q(X))0: conditional on any positive lambda, +the nonconstant continuous function F_lambda evaluated at a uniform +point has strictly positive variance. The law of total variance +then gives sigma^2>0. Centering and multiplying by this positive +constant preserve all critical locations, inertias and elder pairs. +Stationarity, smoothness and norm-moment bounds persist. + +Set C=kappa_eta/sigma for (21), or set sigma=1, mu=0 when using G +without normalization. The parameters gamma and the deterministic +construction are fixed before sigma is computed. Changing gamma +may change sigma and the covariance. We make no claim that the +different gamma laws are covariance matched. + +The parameter family is smooth for lambda>0 and converges globally +in every spatial C^q norm to F_0 as lambda decreases to zero. As a +family at lambda=0 it is C^1 but not C^2, since the nonzero coefficient +eta lambda^(3/2) has that regularity. In the standard two coefficient +controls (u,v) of the quartic potential, our path is + + (u,v)=(lambda,eta lambda^(3/2)). (22) + +It stays inside the three-root cusp region and approaches its tip +along a prescribed path. It is neither a generic full two-parameter +random unfolding nor a smooth transverse one-parameter crossing of +that tip. The restriction is part of the actual law, not hidden in +a universal-class claim. + +## 8. Exact lifetime density, location and physical distance + +The unique finite-H0 lifetime of Z is C lambda^2 by (19),(21). +The change of variables lambda=sqrt(ell/C) in (20) proves (2): + + p_ell(ell)=gamma/(2 lambda_0^gamma C^(gamma/2)) + ell^(gamma/2-1) + =beta ell_0^(-beta) ell^(beta-1), + 0z_-. Its length is + + r_lambda=(2/omega)[ + arcsin(sqrt(lambda)z_0/sqrt(2w_1)) + -arcsin(sqrt(lambda)z_-/sqrt(2w_1))] + =D sqrt(lambda)+O(lambda^(3/2)), + D=2(z_0-z_-)/(omega sqrt(2w_1))>0. (24) + +All parameters in this expansion are fixed. Thus the actual physical +height gap has quartic order in endpoint distance, + + ell=C D^(-4) r_lambda^4 [1+O(lambda)]. (25) + +The exact law is quadratic in lambda; (25) is a distance asymptotic +and is not promoted to an exact Euclidean-radius identity. +The regular-fold cubic mark ell/r_lambda^3 tends to zero like +(C/D^3)sqrt(lambda). These samples approach a different contact +boundary from fixed positive-cubic-mark windows in the Gaussian +regular-fold theorem. + +The global mean-zero birth height is + + B_Z=b_0+(Phi_eta(z_-)/kappa_eta) ell, + b_0=(b-mu)/sigma. (26) + +In particular unscaled physical birth heights collapse to b_0 as +lifetimes decrease to zero. If desired the rescaled mark +(B_Z-b_0)/t is exactly (Phi_eta(z_-)/kappa_eta)(ell/t). +The centered lifetime still has no dependence on mu. + +## 9. What this proves about the order taxonomy + +For the identical limiting germ (10) and the identical fixed path +shape eta, changing only gamma in (20) gives + +| Sampling near lambda=0 | Exact small-lifetime density power | +| --- | --- | +| gamma=1, flat lambda density | ell^(-1/2) | +| gamma=4/3 | ell^(-1/3) | +| gamma=2 | ell^0 | +| gamma>2 | ell^(gamma/2-1), vanishing at zero | +| any gamma>0 | ell^(gamma/2-1) | + +The constants in (23) may change with the fixed variance normalization, +but these powers do not. The associated cumulative power is gamma/2. +This is an actual stationary-field construction with an actual selected +bar, rather than a formal replacement of the cubic order by four in +a Gaussian Kac--Rice formula. + +The underlying probability order can also be stated in physical radius. +By (24), the induced small-r parameter mass is proportional to +r^(2gamma-1) dr; together with ell proportional to r^4 this gives +the same density exponent gamma/2-1. Uniform translation supplies +uniform contact location, not the ordinary off-diagonal polar contact +measure of a nondegenerate two-site jet law. There is no reason to +substitute the Gaussian observation/determinant orders into this law. + +Consequently a quartic singularity fixes the local relation between +the parameter, endpoints and height gap, but does not by itself fix +an observed lifetime density exponent. Contact sampling weight is a +necessary part of a taxonomy. The flat-parameter example realizes a +different exponent from the admitted regular Gaussian fold law; the +gamma=4/3 example also shows that the exponent -1/3 alone cannot +identify the singularity class. + +No claim is made about a generic two-parameter cusp sampled with a +specified two-dimensional density, about all A3 laws, or about an +Arnold class assigning one covariance-independent persistence power. +Those would require their own actual contact law and global selection +proof, just as the regular-fold interface theorem requires them. + +## 10. An actual iid-copy marked process, with exact cluster avoidance + +For a single sample let Xi_t be its point measure with lifetime +ell/t and birth location x. For every finite T, on {ell<=Tt} it +has exactly one point, and otherwise it has none on (0,T]xX. +As t decreases to zero the measure on this window is eventually +empty for each sample, since its one lifetime is strictly positive. +Its expected count is exactly (Tt/ell_0)^beta when Tt=0 +be a bounded continuous compactly supported test function on +(0,infinity)xX, with lifetime support in (0,T]. For t0, including at the lifetime boundary zero. The same calculation +also applies to bounded continuous tests that extend to zero. + +There is no inference of a Poisson limit from first moments alone. +Single-field multiple-witness avoidance is exact here: + + N_(H0,Tt) is in {0,1}, + E[N_(H0,Tt); N_(H0,Tt)>=2]=0. (28) + +The process is a rare-event limit of independent entire field copies. +It does not assert spatial independence, a factorial limit within a +copy, a growing-domain regime, or convergence of unscaled barcodes. +Adding degree and direction marks puts a Dirac mass at degree zero +and e_1 in (4). Adding the unscaled physical birth height puts a +Dirac mass at b_0, by (26); it does not retain a nontrivial birth law. + +One may retain the explicit rescaled birth height in (26), or the +separation mark r_lambda/t^(1/4). The latter converges, at lifetime +mark a, to D(a/C)^(1/4) by (24). These give deterministic mark +pushforwards of (4), not extra probabilistic independence assertions. +If the higher-degree old-gap consequence in Section 6 is used, all +other finite bars are absent from (0,Tt] for sufficiently small t, +so the identical limit holds for the entire degree-marked finite +barcode. The main theorem is the H0 assertion and needs no such +higher-degree extension. + +## 11. Admission boundaries, measurable selectors and law deficits + +The chosen finite-H0 pair is a Borel function of (lambda,Y): its +locations, values and lifetime are explicit continuous expressions +for positive lambda. Equivalently it is the actual elder pair on +the open Morse/distinct field locus, proved in Section 6. The count +in (28) is its Borel lifetime indicator. Thus there is no unresolved +root-count or global-selector measurability premise in this family. + +Every sample has good topology and uniformly bounded spatial +derivatives, but the law fails the full joint-density contract of +the Gaussian or regular contact-counting adapters. To make this +deficit precise, fix two distinct sites p,q and consider their +values and gradients, a vector in R^(2d+2). For positive lambda this +vector is a smooth function of (lambda,Y), which has only d+1 +real parameters. Cover the parameter space by countably many +compact coordinate charts with lambda bounded away from zero. +The map on each is Lipschitz. A Lipschitz image of a bounded subset +of R^(d+1) has zero (2d+2)-dimensional Lebesgue measure: subdividing +into cubes of side epsilon covers the image by O(epsilon^(-(d+1))) +balls of radius O(epsilon), with total (2d+2)-volume tending to zero. +Taking the countable union proves that this observation law is +singular, and hence has no full Lebesgue joint density. + +This is a law-density deficit, not a claim that its covariance matrix +must be singular. A nonlinear distribution with a small-dimensional +support can have a nonsingular covariance; covariance rank alone +does not decide this issue. Similarly we do not assert any particular +positive or zero Fourier block for the constructed covariance. + +The translated smooth templates give a real stationary, non-Gaussian +law with rapidly controlled spatial derivatives. Non-Gaussianity also +has a direct scalar witness: G(0) has bounded support and positive +variance, whereas a nonconstant Gaussian variable has unbounded +support. Its Fourier phases +are correlated through Y and its deformation through lambda. No +independent Gaussian coefficient carrier, arbitrary jet-rank property, +or generic full two-parameter contact density is supplied. It is not +admitted by covariance positivity plus sample smoothness alone. + +The apparent abundance of nearby critical points is also explicit: +three local points, not the two branches of a regular cubic fold. +The two maxima and one saddle have different roles. Choosing the +wrong maximum--saddle candidate would give the essential older +component or a different height difference; the actual finite bar +uses the lower maximum in (14). Its enclosure and elder comparison +are why this source can make an actual selected statement. + +The construction is not covariance matched to the spectral Gaussian +family, and it is not a random scalar-amplitude mixture of that law. +It supplies a different local contact order in an actual spatial +field while keeping its restricted path and sampling dependence +visible. The previously prepared amplitude example isolates a +different failure mechanism, namely a global scale changing the +effective contact probability even at fixed covariance. + +## 12. Operational context, source review and reading boundary + +These current repository references were freshly hashed by root. They +are context, not Gaussian assumptions imported into this non-Gaussian +law. Sections1-11 directly prove the principal H0 field/count/process +result. The optional higher-degree old-gap consequence consumes only the +explicitly named ordinary Morse one-handle/one-endpoint fact. + +| Context | Current repository source | Bytes / lines | SHA256 | +| --- | --- | --- | --- | +| ST | [regular-fold structural theorem](../fold_structural_universality/PROOF.md) | 43927 / 804 | 3d952a906553d578876456539a8f0dcd64aedd45099e8394cbe9db99866ec531 | +| SG | [spectral Gaussian admission](../spectral_gaussian_admission/PROOF.md) | 42478 / 826 | 6c0f34cb318d337b3b8c2c306f3b965f5e5a9d2fe38d5e001b8044de74b7f225 | +| HK | [higher-homology fold](../higher_homology_fold/PROOF.md) | 46174 / 911 | 87f43e6ed94c5c63c385ecf177f68212afa5db54b24062fa008f081f47a78999 | +| AMP | [exact amplitude crossover](../spectral_amplitude_crossover/PROOF.md) | 37461 / 778 | aeea6b55a1876c357d9beeeb5a96ee54a5b17b53c6138e2993c454f1117f65d5 | + +The author originally read the externalSG825/43e818 andHK884/de1cad +preparations and both amplitude720/5b185 and corrected730/9e8f sources; +the following historical ledger retains those exact identities and +actual reading extents. Root checked original/current SG Sections1-11 +at36656B/4cad43d3 and HK Sections1-12 at37220B/c770d4f7. No new whole +currentSG/HK author reading is inferred. The amplitude result is context; +no amplitude theorem is a premise of this quartic construction. + +Fresh exact original-source audits are separately bound. Final read all +695 actual lines in1-240/241-480/481-695, and benchmark read all695 in +1-232/233-464/465-695; both returned mathematicalPASS with no must-fix. +Both freshly reproduced the five original context hashes and additionally +reread ST691-737 and HK388-442. Their prior whole-source exposure is +separate; no five new whole dependency audits are claimed. Final supplied +the earlier fold-chart route and amplitude wrapper, benchmark CG/ET and +catalogue code, root the prefile flattening/unfolding idea, and the author +all related spectral/Hk/amplitude routes. Organizational independence0. + +The author's primaryAGV exposure is bounded to its two-page nomenclature +preview, not classification. Root and final each attempted that preview +once and received inaccessible/InternalError; neither obtained fresh +primary-content verification or retried. Benchmark made no new primary +fetch. The actual real normal form is explicitly derived in Section3; +no unseen classification theorem is used. No rawPDF/fullbook custody, +compiler, numerical field, sampler, seed or experiment is credited. + +The author's original source ledger below is preserved as historical +provenance, including its original request for exact-file review. Its +filesystem paths identify original artifacts, while current operational +links are above. The original's completed two reviews are not a new full +review of this incorporated wrapper. New incorporated-source review and +README metadata remain PENDING at this construction snapshot. + +## 13. Preserved original author context and primary ledger + +All paths below are under +`/Users/dylanroy/Documents/Codex/2026-10-08/turn-the-below-into-a-reality/`. +Their hashes were freshly reproduced before saving this source. +The author had previously read the full actual sources listed here; +the current task does not rebadge those prior reads as new independent +reviews, nor infer any current publication or scientific status. + +* `fold-universality-worktree/experiments/universality/fold_structural_universality/PROOF.md`, + SHA256 `3d952a906553d578876456539a8f0dcd64aedd45099e8394cbe9db99866ec531`. + Prior full source read; Section 10, actual lines 691--737, was freshly + reread for its conditional order taxonomy. It explicitly leaves a + higher-order field theorem unproved. The present construction supplies + one such actual restricted law, not a general admission of its class. +* `spectral-gaussian-universality-preparation.md`, + 42,038 bytes / 825 lines, SHA256 + `43e8180947b559e53f7afcf95ba61136f3fffcbe05a19e7a5953018fda27e294`. + Author's prior full actual-byte read and prior same-team technical + reviews are historical context. None of its Gaussian COUNT, + CONTACT, ENVELOPE, SELECT, FAR or coefficient conclusions are + imported into this non-Gaussian law. +* `higher-homology-fold-obstruction-preparation.md`, + 44,258 bytes / 884 lines, SHA256 + `de1cad971e113f097a8e29ba82fb295d93b23f57aeda5b265e6f3c88cef47f2e`. + Prior full author read; Section 7, actual lines 388--442, was freshly + reread for the ordinary Morse-handle endpoint rule used only in the + optional old-higher-bar-gap consequence. Its independent review state + is not a premise for the self-contained H0 enclosure proof here. +* `spectral-amplitude-crossover-preparation.md`, + 33,421 bytes / 720 lines, SHA256 + `5b1859080a137aed9eeeba20e1b6f0c5a61387e5eb6fa20bb1b8704860fb73d2`, + and its separate corrected preparation, 34,110 bytes / 730 lines, + `spectral-amplitude-crossover-corrected-preparation.md`, SHA256 + `9e8f10162e9fa21f308454df9463459cdd67ef39a2d8103a78a397b4ab82dd0d`. + Both had full saved-byte author reads. The original's two full reviews + found a missing t_n->0 hypothesis in its critical process normalization; + its AMEND history remains intact. The corrected source is a successor, + not a rewrite of that frozen source or its review record. No amplitude + theorem is consumed in this proof. In particular (3) includes the + small-threshold hypothesis explicitly. + +For the A_k nomenclature, the author freshly read the official publisher +book and chapter pages and the actual two-page primary chapter preview +of Arnold, Gusein-Zade and Varchenko, *Singularities of Differentiable +Maps*, Volume I, Section 15, printed pages 242--243. The preview states +the A_k series of function germs. It does not expose the whole chapter, +and no full-book or full real-classification read is claimed. The exact +real quartic normal form (10) is proved directly, so no unseen +classification result is needed. Primary links: + +* [AGV, Lists of singularities](https://link.springer.com/chapter/10.1007/978-1-4612-5154-5_15). +* [Actual chapter preview, pp. 242--243](https://page-one.springer.com/pdf/preview/10.1007/978-1-4612-5154-5_15). + +The optional all-degree old-gap statement consumes only the classical +fact that crossing a nondegenerate Morse level attaches one handle, +whose field-coefficient relative homology is one-dimensional in its +handle degree. In the prior higher-homology task the author actually +read the relevant bounded primary portions of Milnor's *Lectures on +the h-Cobordism Theorem*: Section 5's handle setup and Theorem 5.4, +and Section 7's relative-homology/intersection material, including +Corollary 7.3. The current proof uses no cancellation theorem, generic +gradient-trajectory assumption, or stronger Hk selected claim from +that book. [Primary Milnor scan](https://webhomes.maths.ed.ac.uk/~v1ranick/surgery/hcobord.pdf). + +The actual dependencies of (2)--(4) are elementary smooth cutoffs, +the explicit coordinate and derivative calculations, compact gradient +and height bounds, the connected-component computation (18), Haar +translation invariance, a scalar change of variables, and the iid +Laplace product. No Gaussian regression, numerical draw, compiler, +external message, repository source edit or status change was used. +The proposed stationary A3 family and its exact selected H0 law are +new analytic preparation, awaiting independent exact-file review. diff --git a/experiments/universality/quartic_cusp_field/REVIEW.md b/experiments/universality/quartic_cusp_field/REVIEW.md new file mode 100644 index 000000000..346f67067 --- /dev/null +++ b/experiments/universality/quartic_cusp_field/REVIEW.md @@ -0,0 +1,92 @@ +# Quartic field: construction and source-bound review + +Construction snapshot,8October2026. Mathematical author OpenAI/Codex +universality_review; wrapper/integrator root task01a11cbf-068d-7102-861e-75814e715c98. +Exact claim6071510105; detached initialbase amplitude334native +6bd133d591e43e75a9292463bb4bfde7a34984dc. No mainintegration implied. + +Current [proof](PROOF.md): 35931 UTF8 bytes/751lines, +SHA256 33ba8362069346058ba33bbd8c1bd1feed62b749a96b6cbbc8836ddcbc6e4f2e. +Original695 source:32334B/SHA25624e6f4149c0a2df2f8df1dbfa8d66da3e01a7d12bee4b803bac093cfea36bf3f. +Preserved Sections1-11:26625B/602lines, +SHA2569f1426a21b40b524ad4b25c81368489b0aa7a0c29f79659eb556bb4d1579f6d0. +Originalauthor saved-byte readFULL; root originalFULL1-350/351-695. +Two separate freshFULL original mathematicalPASS reviews: final_contract +actual1-240/241-480/481-695 and benchmark1-232/233-464/465-695, +no must-fix. New current fullsource review, root current read and +new [README](../README.md) metadata review:PENDING construction snapshot. + +An explicit weightedcosine seed, exact sinechart, smooth matched surgery, +protected annulus and global highlevel barrier produce exactly one finite +H0 bar, lowermaximum to middlesaddle, lifetimeC*lambda². Haartranslation +makes the field stationary. The exactparameter density gives wholefield +lifetime density beta*ell0^(-beta)*ell^(beta-1), beta=gamma/2, on(0,ell0), +and its exactonepoint iid-copy Laplace law. All oldcritical branches and +the otheressentialmaximum are explicitly controlled. Flatgamma1 gives +power-1/2; gamma4/3 gives-1/3 atthesamequarticgerm. Probabilityweight is +partofthe taxonomy. The path is restrictedC1-not-C2; two-sitepin law is +singular. No Gaussian or genericfullA3 admission, numericalsampler or +samefieldspatiallimit. OrdinaryfixedcoefficientfieldH0; optionaloldHk +bar-gap fact is separately declared and notneeded forthe mainclaim. + +All performerssameproviderOpenAI/Codex, source/routeexposed, org0. +Fivecontext hashes and boundedST/HK rereads do notbecome fivewholeinput +reviews. AuthorAGVpreviewboundednomenclature only; root/finalfetches +inaccessible, no newroot/finalprimarycontent; benchmark no newfetch. +No unseenclassification import, rawPDF/fullbookcredit, compiler, unit +tests, fielddraws, sampler, blindseeds orscientificpromotion. Source +proof/diff/core/link checks and lateractualpublication/gates willbe +recorded separately. Full landmarkgoal ACTIVE. + +## Completed incorporated-source review, 9 October 2026 UTC + +The preceding 39 lines are the unchanged construction snapshot. Its +PENDING states describe that earlier stage. Final_contract_review has +now performed a fresh FULL mathematical read of the actual saved +751-line proof in untruncated ranges 1–250, 251–500 and 501–751, +with mathematical PASS and no must-fix. This verdict binds the current +35,931-byte proof and SHA256 33ba8362069346058ba33bbd8c1bd1feed62b749a96b6cbbc8836ddcbc6e4f2e. +It is separate from both original 695-line source verdicts. + +The reader repeated attacks on the exact surgery, root exhaustion, +annulus exclusion, global enclosure, two-component retraction and actual +lower-maximum elder pairing. The normalization, density and volume +factors, physical distance and birth marks, and exact one-point iid +Laplace law passed. The principal H0 result is self-contained; the +optional other-degree observation remains bounded by its declared +one-handle endpoint import. No Gaussian or generic cusp classification, +covariance matching, numerical sampler, spatial process, formal proof or +scientific acceptance is credited. + +The same reader fully read the actual construction companion at +2,425 bytes / 39 lines, SHA256 5fb816b2135df21d984f4db134310df176bec87697f45956c5145daecf700600, +and README at 17,387 bytes / 261 lines, SHA256 9070e18f4c5c8c76e96ad8a48c93bf73e5b07fbde65e75e7c201274236803bce. +Construction metadata PASS covers the complete original-to-current proof +diff, unchanged 26,625-byte Sections 1–11 and author-ledger text, four +current context hashes, actual original/current SG and HK core maps, +and the complete 1,000-byte README insertion. The 16,387-byte amplitude +README baseline is retained exactly. All 37 direct local links resolve; +navigation reports 29 pages, 167 local links, six fragments and zero +problems. External sources were not crawled. These fresh hash/core checks +are not four new whole-input reviews. + +Root separately fully read the actual 751 lines in ranges 1–380 and +381–751, the 39-line companion and 261-line README, and the complete +695-to-751 diff. Root reproduced the four current context hashes, +quartic Sections 1–11, original/current SG and HK cores and exact +README baseline restoration, then ran the declared navigation check. +All matched. The author-ledger text matches verbatim: root's extraction +starts after the heading newline; the reader's extraction includes it. +No discrepancy in original text is inferred from those byte boundaries. + +Root preserved this companion's exact construction bytes in the ordinary +external quartic-cusp-review-construction.md and the unchanged README +in quartic-cusp-readme-construction.md. This new append still requires +its own bounded metadata review. The mathematical proof and README +remain unchanged; a later receipt will bind the completed append review +and actual native publication without a self-hash cycle. + +All performers are source-exposed same-provider OpenAI/Codex AI agents, +organizational independence zero. No fresh primary fetch, compiler, test +suite, field draw, sampler or blind seed was used for this incorporation. +Full landmark goal ACTIVE.