From 3dbc81675a659f70e979efbf1b6e21d1bc106379 Mon Sep 17 00:00:00 2001 From: Dylan Roy <249851642+d6g8k5htny-coder@users.noreply.github.com> Date: Sat, 17 Oct 2026 19:03:38 -0500 Subject: [PATCH] Preserve reviewed smoothing, A4 sector and finite-L coefficient sources --- .../EVERYWHERE_MARK_EXTENSION.md | 58 ++ .../PROOF.md | 614 ++++++++++++++ .../README.md | 15 + .../REVIEW.md | 575 +++++++++++++ .../coauthor/DERIVATION.md | 232 ++++++ .../FOLLOWON_SOURCE_INDEX.json | 373 +++++++++ .../PROOF.md | 768 ++++++++++++++++++ .../README.md | 15 + .../REVIEW.md | 657 +++++++++++++++ .../FINITE_SMOOTH_INCIDENCE_RECONSTRUCTION.md | 83 ++ .../PROOF.md | 445 ++++++++++ .../README.md | 15 + .../REVIEW.md | 438 ++++++++++ .../FOCUSED_TENSOR_NORMALIZATION_REVIEW.md | 133 +++ 14 files changed, 4421 insertions(+) create mode 100644 frontiers/gaussian_a4_elder_sector_20261017/EVERYWHERE_MARK_EXTENSION.md create mode 100644 frontiers/gaussian_a4_elder_sector_20261017/PROOF.md create mode 100644 frontiers/gaussian_a4_elder_sector_20261017/README.md create mode 100644 frontiers/gaussian_a4_elder_sector_20261017/REVIEW.md create mode 100644 frontiers/gaussian_a4_elder_sector_20261017/coauthor/DERIVATION.md create mode 100644 frontiers/nongaussian_smoothing_admission_20261017/FOLLOWON_SOURCE_INDEX.json create mode 100644 frontiers/nongaussian_smoothing_admission_20261017/PROOF.md create mode 100644 frontiers/nongaussian_smoothing_admission_20261017/README.md create mode 100644 frontiers/nongaussian_smoothing_admission_20261017/REVIEW.md create mode 100644 frontiers/nongaussian_smoothing_admission_20261017/imports/FINITE_SMOOTH_INCIDENCE_RECONSTRUCTION.md create mode 100644 frontiers/original_finite_l_gaussian_binding_20261017/PROOF.md create mode 100644 frontiers/original_finite_l_gaussian_binding_20261017/README.md create mode 100644 frontiers/original_finite_l_gaussian_binding_20261017/REVIEW.md create mode 100644 frontiers/original_finite_l_gaussian_binding_20261017/focused-tensor-normalization/FOCUSED_TENSOR_NORMALIZATION_REVIEW.md diff --git a/frontiers/gaussian_a4_elder_sector_20261017/EVERYWHERE_MARK_EXTENSION.md b/frontiers/gaussian_a4_elder_sector_20261017/EVERYWHERE_MARK_EXTENSION.md new file mode 100644 index 000000000..7b60528b0 --- /dev/null +++ b/frontiers/gaussian_a4_elder_sector_20261017/EVERYWHERE_MARK_EXTENSION.md @@ -0,0 +1,58 @@ +# Exact observable extension at the full-jet Rice join + +This records an interpretation-sensitive but valid join in §§2,3,7,8 of the +frozen A4 proposal, SHA256 abed6f94da080a33458b78f1e60ae92ad90083323af41d241de05939d4377ff0. +It is reviewer reconstruction, not an author-source edit or a new hypothesis. + +Let S(x,f) be the local sector tests in §2, including criticality, the simple +negative Hessian with the specified hard eigenvalue, the unique Q5>0 frame, +the compact jet boxes, 0c} to a point with value greater than f(x), with empty +supremum -infinity. Countable geodesic-ball chains and dense endpoint tests +make this Borel. For Borel B subset (0,infinity), use the everywhere Borel mark + + w_B(x,f)=S(x,f) 1{D_f(x) is finite} + 1{f(x)-D_f(x) in B}. + +The original field F is Morse with distinct critical values almost surely. +On this event the sum of w_B over its critical points is exactly the actual +sector-bar count N_A4(B): classical elder H0 identifies D with the death, +and the deterministic local enclosure proves that every marked birth has +an older point and hence is not essential. The count defined to be zero on +the original null complement therefore has the same expectation as this +extended critical-root sum. This is an equality of observables almost surely +under the ORIGINAL field law, not a claim of pointwise equality on every +conditional fiber. + +Apply the one-point all-Borel Rice identity to w_B. Its canonical full-jet +kernel is f_z=omega+m(z). Since Hm(z)=0 and H omega=omega, its sector residual +cutoff is exactly 1{||omega||C7<=R}; the actual mark has not been replaced by an +independent auxiliary draw. At every target z_r in the sector, for EVERY +cutoff residual, §7 proves D_(f_zr)(x)=b-Delta(r,omega) by the boundary barrier +and outgoing ridge path. It follows directly that the conditional mark equals + + 1{||omega||C7<=R} 1{Delta(r,omega) in B} + +when the other displayed jet-sector tests are met. The coordinate changes then +give (25). No conditional global Morse indicator has been deleted. One was +never present in the Borel extension to which Rice was applied. + +At the limiting target z0, the contact is forced degenerate; no global Morse +statement about that law is needed either. The coefficient limit concerns the +explicit finite jet density, deterministic local lifetime and fixed residual +cutoff, not an attempted exceptional-target extension of a genericity theorem. + +The source explicitly supplies the everywhere intrinsic threshold (§7), the +original almost-sure actual-bar identification, and the all-Borel identity +(§3), and says that its conditional trapping/density argument does not assume +global Morse behavior. Thus this is a legitimate conventional inference from +its written constructions. A sentence spelling out w_B before (25) would aid +readers, but no required mathematical amendment or added hypothesis is needed. +The reviewed interpretation is precisely this one. Inserting a global Morse +indicator in the kernel and discarding it by unconditional genericity would +be invalid and is not accepted. diff --git a/frontiers/gaussian_a4_elder_sector_20261017/PROOF.md b/frontiers/gaussian_a4_elder_sector_20261017/PROOF.md new file mode 100644 index 000000000..86371a58b --- /dev/null +++ b/frontiers/gaussian_a4_elder_sector_20261017/PROOF.md @@ -0,0 +1,614 @@ +# Actual Gaussian A4 sectors with the global elder partner + +Conventional mathematical author proposal, 17 October 2026. Authors: the +OpenAI/Codex A4-sector author, Root01a11cbf (mechanism coauthor), and the +focused ridge/frame coauthor identified in the author result. This exact +proposal requires a separate nonauthor review. All participants are +source-exposed; no organizational, provider, blind, or human independence is +claimed. The mathematical result below is a restricted population theorem on +a fixed two-dimensional flat torus, not a whole-density correction theorem. + +## 1. Actual law and theorem + +Fix L>0, X=R^2/(L Z^2), its Euclidean metric and orientation, and V=L^2. +Let phi_j be real smooth periodic functions with dense real span in C-infinity(X), +let a_j>0 satisfy sum_j a_j ||phi_j||_Cq=0, +and let Z_j be the original independent standard real Gaussians. Use the +actual smooth random field + + F=sum_j a_j Z_j phi_j. (1) + +All assertions concern each fixed such law separately. One explicit example +uses k_n=2 pi n/L, w_n=exp(-|k_n|^2/2)/sum_m exp(-|k_m|^2/2), a set H containing +one member of each nonzero pair {n,-n}, and independent standard Gaussians: + + F(x)=sqrt(w_0) Z_0 + sum_{n in H} sqrt(2w_n) + [Z_n^c cos(k_n.x)+Z_n^s sin(k_n.x)]. (2) + +Its covariance is sum_n w_n cos(k_n.(x-y)). The positive exponentially +decreasing weights give every required smooth-norm summability, and Fourier +partial sums give dense smooth span. No finite-field truncation is performed. +Stationarity and isotropy are not needed in the theorem. + +The consumed smooth-Gaussian admission theorem G, exact SHA256 +8d992e194f0a898e563c4c3d05b8079cde292f9e87d7ff51030ad12ef5faf319, +proves smoothness, all smooth-norm moments, finite-jet ranks, literal +whole-field Gaussian kernels, all-Borel finite-incidence Rice identities, +and almost-sure Morse/distinct-critical-value behavior for (1). Its full +40231-byte proof and full 38035-byte accepted scoped review were read. Its +finite-incidence appendix was also read. Section 3 recalls exactly the +interfaces used here. Its fold conclusion is not an A4 premise. + +Here is the new assertion. There exist a nonempty compact rectangular +parameter box U around (6,22,24), with coordinates (alpha,beta,q), a finite +R>0, and r_0>0, such that the actual ordinary finite superlevel-H0 elder +bars whose birth maxima satisfy the Borel sector mark in Section 2 have +expected lifetime measure per area + + M_A4(B)=V^-1 E N_A4(B), B subset (0,infinity) Borel. + +On a fixed interval (0,ell_0), this measure has the particular finite Borel +density given by (26), and + + nu_A4(ell) ~ C_A4 ell^(4/5), + M_A4((0,t]) ~ (5/9) C_A4 t^(9/5), ell,t -> 0+, (3) + +where 00, -lambda>a, a in [-2,-1]. + +For a unit soft eigenvector u put v=Ju. Use raw symmetric derivative +coordinates, one entry F_{u^i v^(n-i)} for each i=0,...,n, at each order +n=0,...,5. In particular, order zero is F, order one has two entries, +order two has three, and so on. This full 5-jet is a vector of dimension +1+2+3+4+5+6=21, denoted J_{x,u}F. Its entries are actual derivatives, +not derivatives divided by factorials. In this frame F_uv=0. Put + + gamma=F_uuv, eta=F_uuuv, B=F_uvv, + T=F_uuu, + Q4=F_uuuu-3 gamma^2/a, + Q5=F_uuuuu-10 eta gamma/a+15 gamma^2 B/a^2. (4) + +These are precisely the third, fourth and fifth derivatives of the +transverse maximizing ridge through x, as proved in Section 5. Q4 is not +the raw fourth derivative. Let zeta be the ordered list of the other twelve +entries of orders 3,4,5: exclude T,F_uuuu,F_uuuuu and retain every other +entry once. Thus gamma,eta,B are entries of zeta. Fix their block + + zeta in [-1,1]^12, b=F(x) in [-1,1]. (5) + +Different fixed bounded boxes with nonempty interior would work. For +Q5!=0 exactly one of the two soft directions has Q5>0: reversal +(u,v)->(-u,-v) fixes even-order entries and negates odd-order entries, +so Q5 changes sign. Use this unique direction. Set + + r=-Q4/36, alpha=lambda/r^3, beta=T/r^2, q=Q5. (6) + +Require 00 give a unique Borel lift. Formula (7) is a continuous +linear operation at each parameter and depends smoothly on x,u. Norms, +jet entries, and the rational expressions (4) are Borel on a<=-1. +The Morse/distinct-height locus is Borel, and (1) belongs to it almost +surely by G. On that locus every maximum has a unique birth label. Count +its actual ordinary finite elder bar if and only if this mark holds. +Set the count to zero on the null complement. The deterministic proof +below also specifies its lifetime without using conditional global Morse +genericity. No essential class can carry this mark, because an explicitly +constructed point has larger height than the birth. + +## 3. Literal kernels and the one-point all-Borel Rice identity + +All twenty-one jet coordinates are independent continuous linear functionals +on smooth functions: a polynomial times a cutoff prescribes any 5-jet at x. +Dense smooth span in (1) therefore implies positive definite Gamma_{x,u}. +Indeed a row annihilating all coefficient columns would annihilate every +smooth function; a finite subset of columns already spans R^21. Compactness +of X times S^1 and smooth covariance give positive covariance floors and +uniform bounds for Gamma^-1 and the global smooth norms of c Gamma^-1. +The resulting Gaussian density p_{x,u}(z) is positive, finite and smooth +in all finite variables. There is no independence assumption between +jet orders or entries. + +For each literal jet value z define a whole-field conditional law by + + f_z=R_{x,u}+m_{x,u}(z), + R_{x,u}=H_{x,u}F, m_{x,u}(z)=c_{x,u}Gamma_{x,u}^-1 z. (8) + +The residual is a smooth centered Gaussian field independent of the full +jet; its own 5-jet at x is exactly zero. This formula specifies the kernel +at every z, including the degenerate limit below. It is not conditioning +on an event with positive probability. For clarity about Borel marks, G +derives its Rice identity from actual finite original coefficient minors. +Fix a smooth independent tail; solve the two critical equations for two +invertible original Gaussian coordinates; the incidence Jacobian is +|det Hess f| divided by the gradient-minor determinant. Ordinary finite +Gaussian fiber integration cancels that denominator. Equal-dimensional +smooth Sard, local inverse charts and ordinary change of variables give +the counting identity for every nonnegative Borel whole-field mark. +Integrate the smooth tails and sum a disjoint minor/chart partition. + +The literal fiber agrees with (8), not just almost everywhere in z: +the tail receives its Bayes density p_block(z-J tail)/p_full(z), and the +remaining finite block its affine Gaussian kernel. Completing the joint +Gaussian square gives the regression characteristic function at EVERY +target. Countably many field evaluations generate the smooth Borel +sigma-field, so the resulting whole-field laws agree. Thus arbitrary +Borel marks transfer at the zero-gradient and full-jet targets used here. +No continuity of the elder indicator on a null fiber is required. + +In particular the exact one-point identity, with height and higher-jet +disintegration, is + + E sum_{x: grad F(x)=0} w(x,F) + = integral_X integral p_fixed(z) |det H| + E[w(x,f_z)] d(non-gradient jet z) dx, (9) + +where the two gradient entries of z are set to zero and the remaining +nineteen entries have raw Lebesgue measure. This follows equally by +disintegrating the one-point identity at gradient zero using the same +literal block/tail calculation. Its total unmarked integral is finite +by G's compact covariance floors and conditional Hessian moments. + +The unconditional Morse/distinct-height statement is used only to identify +the original bar count almost surely. Under a literal degenerate limiting +jet the field need not be globally Morse. None of the global trapping or +density arguments assumes that it is. + +## 4. Eigenframe and reduced-jet changes, with exact multiplicities + +Let theta be the angle of u in a fixed oriented Euclidean frame. In raw +Hessian coordinates (H11,H12,H22), + + H=R_theta diag(-lambda,a) R_theta^T, + |d(H11,H12,H22)/d(lambda,a,theta)|=|a+lambda|. (10) + +For example at theta=0 the three derivative columns have diagonal entries +(-1,0,0), (0,0,1), and (0,-lambda-a,0); their determinant has absolute +value |a+lambda|. Conjugation by a plane rotation has determinant one +on the three raw symmetric entries, so the same formula holds at every +theta. This uses dH11 dH12 dH22, not an orthonormalized Frobenius volume. + +Angles in [0,pi) count each soft eigenline once. We instead use theta in +[0,2pi) together with Q5>0. Exactly one of each antipodal pair satisfies +this condition, so this also counts once. There is NO additional factor +1/2. In the full jet, the SO(2) action on each Sym^n(R^2) has determinant +one: its determinant is a continuous homomorphism from the compact connected +circle to positive real numbers, hence one. The gradient rotation also has +determinant one. In the combined change of variables the theta derivatives +of the higher entries form only a lower off-diagonal block; (10) is +therefore the only frame Jacobian. The Rice weight at the maximum is + + |det H|=lambda(-a). (11) + +Keep b,a,zeta fixed. Changing the pure fourth/fifth entries to (Q4,Q5) +has absolute determinant one. The correction to F_uuuu in (4) depends +only on a,zeta; the correction to F_uuuuu also depends only on a,zeta. +Consequently the inverse target for (6) is + + F=b, (F_u,F_v)=(0,0), + (F_uu,F_uv,F_vv)=(-alpha r^3,0,a), + T=beta r^2, + F_uuuu=-36r+3 gamma^2/a, + F_uuuuu=q+10 eta gamma/a-15 gamma^2 B/a^2, + remaining entries=zeta. (12) + +Denote this vector z_r and its r=0 value z_0. It is a smooth bounded +function of all parameters for 0<=r<=r_0 and a in [-2,-1]. The nonlinear +jet correction has not been replaced by a fictional independent Gaussian +law of Q4,Q5; the density is the full original joint p_{x,u}(z_r). + +Finally, + + |d(lambda,T,Q4)/d(alpha,beta,r)|=36 r^5. (13) + +Its columns have rows (r^3,0,3 alpha r^2), (0,r^2,2 beta r), +(0,0,-36), so the claim is exact. Combining (10)-(13), the density +before lifetime pushforward carries the precise factor + + 36 alpha(-a)|a+alpha r^3| r^8 dr. (14) + +It tends, after division by r^8, to 36 alpha a^2. There is one Rice +determinant and one spectral-coordinate Jacobian. No second Rice +determinant, independent saddle density, probability-normalized sphere +measure, or extra unordered-pair half is present. + +## 5. Actual ridge derivatives and uniform conditional bounds + +Let all compact parameters range as in (5), a in [-2,-1], and in a +sufficiently small closed rectangular U around (6,22,24). Its exact open +stability requirements are given in Section 6. From (8), all smooth-norm +moments of R_{x,u} are uniformly bounded over X times S^1. For instance +||R||_C7<=||F||_C7+C||F||_C5, with one finite constant. Choose R so large +that + + h(x,u):=P(||R_{x,u}||_C7<=R)>=1/2 for every x,u. (15) + +This follows directly from a uniform first-moment bound and Markov; no +small-ball or threshold-atom assertion is needed. The same residual is +used for EVERY r in (8), now with z=z_r. On the cutoff event the fields + + f_r=R_{x,u}+m_{x,u}(z_r) (16) + +and their r derivatives are bounded in C7 by one deterministic constant. +The residual distribution, the cutoff event, and its threshold do not +depend on r. All constants from now on are uniform over the compact +parameters and all residuals in this event, for this one fixed law. + +Use embedded Euclidean coordinates (s,t) centered at x in frame (u,v). +There is a fixed small coordinate rectangle on which f_vv<=-1/2. +This follows from f_vv(0)=a<=-1 and the uniform C3 bound. Since f_v(0)=0 +and f_uv(0)=0, the implicit equation f_v(s,h_r(s))=0 has a unique +smooth ridge on |s|<=rho, with + + h_r(0)=h'_r(0)=0, |h_r(s)|<=C s^2. (17) + +Existence with uniform bounds can also be seen without a parameterwise +implicit radius: f_v(s,0)=O(s^2), while f_v decreases in t with slope +at most -1/2; evaluate at t=+-C_1 s^2 and use the intermediate value +theorem. The implicit derivative formulas then give the uniform bounds +and smooth r dependence on a common rectangle. Shrink rho so that this +ridge stays inside it. Write g_r(s)=f_r(s,h_r(s)). Transverse strict +concavity implies f_r(s,t)<=g_r(s), and more quantitatively + + f_r(s,h_r(s)+z)<=g_r(s)-z^2/4 (18) + +whenever the displayed points stay in that rectangle. + +At s=0 the exact derivatives are + + g_r(0)=b, g'_r(0)=0, g''_r(0)=-alpha r^3, + g'''_r(0)=beta r^2, g''''_r(0)=-36r, g'''''_r(0)=q. (19) + +To verify the nonlinear entries rather than assume them, write +h_r(s)=c s^2+d s^3+O(s^4), c=-gamma/(2a). The coefficient of s^4 in +f(s,h(s)) beyond the pure fourth term is a c^2/2+gamma c/2 +=-gamma^2/(8a), hence the fourth correction is -3 gamma^2/a. +At order s^5 the d terms cancel because ac+gamma/2=0; what remains +is eta c/6+B c^2/2=-eta gamma/(12a)+B gamma^2/(8a^2). +Multiplication by 120 gives exactly Q5 in (4). No lambda-dependent +term is missing, because f_uv(0)=h'(0)=f_v(0)=0. + +The ridge is C6 in s under a C7 field bound and g'_r=f_u(s,h_r(s)), +so g_r is C7 with uniform bounds. Its parameter derivative satisfies +the envelope identity partial_r g_r=(partial_r f_r)(s,h_r(s)), because +f_v=0 on the ridge; this is C6 in s with uniform bounds. Taylor's +integral remainder therefore gives + + g_r(s)=b-alpha r^3 s^2/2+beta r^2 s^3/6 + -3r s^4/2+q s^5/120+E_r(s), + |E_r^(j)(s)|<=C|s|^(6-j), j=0,1,2, + |partial_r E_r(s)|<=C|s|^6. (20) + +The last inequality follows by Taylor-expanding partial_r g at zero; +its derivatives through order five are the derivatives of (19). The +same bounds for mixed derivatives needed below follow by differentiating +the integral remainder. In particular on every fixed bounded xi interval, + + r^-5(g_r(r xi)-b)=P_{alpha,beta,q}(xi)+O(r) in C2_xi, + |partial_r [r^-5(g_r(r xi)-b)]|<=C, (21) + +where + + P_{alpha,beta,q}(xi)=-alpha xi^2/2+beta xi^3/6 + -3 xi^4/2+q xi^5/120. + +The bounds in (21) hold also after one xi derivative of the r derivative, +as follows from E_s=O(s^5) and partial_r E_s=O(s^5). + +## 6. A nonempty regular unfolding box and all nearby critical points + +At (alpha,beta,q)=(6,22,24), + + P(xi)=xi^5/5-3xi^4/2+11xi^3/3-3xi^2, + P'(xi)=xi(xi-1)(xi-2)(xi-3). (22) + +The maxima at 0,2 have heights 0,-4/15; the minima at 1,3 have heights +-19/30,-9/10. In particular the height order is + + P(3)0 and let U be the closed rectangular +box |alpha-6|<=epsilon, |beta-22|<=epsilon, |q-24|<=epsilon. Require: +alpha,beta,q positive; the three nonzero roots xi_1,xi_2,xi_3 of P' +are simple and lie respectively in (3/4,5/4), (7/4,9/4), (11/4,13/4); +all adjacent gaps in the preceding strict height order exceed 1/8; +and 1/2g_r(s_2)>g_r(s_1)>g_r(s_3), + Delta(r):=b-g_r(s_3)=r^5 L_r, + L_r=L_0+O(r), |partial_r L_r|<=C. (23) + +All these estimates are uniform on the compact boxes and residual cutoff. +To justify the derivative bound, the scaled reduced functions in (21) +have bounded r derivative in C1, and their three roots have uniformly +nonzero derivative in xi. Their roots therefore have bounded r +derivatives. At a critical root differentiation of the critical value +has no root-motion term, so the same bound follows for L_r. + +There are exactly these four critical points in the whole fixed ridge +interval [-rho,rho], once r_0 is small. Here is the needed exclusion, +which local root perturbation alone would not prove. For s<0 the four +displayed terms of + + g'_r(s)=-alpha r^3 s+(beta/2)r^2s^2-6rs^3+(q/24)s^4+O(s^5) + +are positive. Shrink rho so the remainder is at most q_min |s|^4/96; +then g'_r(s)>0 on [-rho,0). For s>=C_2 r choose a fixed C_2>4 so large +that 6/C_2+alpha_max/C_2^30 up to rho. +On the compact scaled interval [0,C_2], (21), the simple roots of P', +and a positive derivative floor away from small disjoint root intervals +give exactly the three stated positive roots and the anchored root 0. +Near zero use the derivative floor of P' itself, P''(0)=-alpha<0. +There are no others. The transverse equation has exactly the one ridge, +so these are all critical points in the coordinate tube. + +The same Taylor expansion at fixed endpoints, after shrinking rho and +then r_0, yields uniform constants c_0>0 with + + g_r(-rho)<=b-c_0 rho^5, g_r(rho)>=b+c_0 rho^5. (24) + +At r=0 the leading quintic has these signs; its remainder is O(rho^6), +and the r-dependent lower terms tend uniformly to zero at fixed rho. +The critical signs are maximum at 0, saddle at s_1, maximum at s_2, +saddle at s_3, because the transverse Hessian is negative and the +reduced second derivatives alternate. Inertia follows from the Hessian +Schur complement g'' on the ridge. All four are nondegenerate for r>0. + +## 7. A physical enclosure and the actual global elder death + +Choose a fixed delta>0 small enough that the curved tube + + D_r={ (s,t): -rho<=s<=s_3(r), |t-h_r(s)|<=delta } + +is inside the embedded rectangle with transverse concavity. All choices +can be made uniformly: (17) keeps the ridge O(rho^2) from the axis; +one may first choose the transverse rectangle, then rho, then a fixed +delta, for example a sufficiently small positive multiple of rho^2. +It need not shrink with r. Decrease r_0 until + + Delta(r)c} to a point higher than b. + +Define the intrinsic elder threshold of a point at height b by the +supremum of rational cc} to a +point of value >b. Connectivity is Borel using countable chains of +small basis balls whose closure minima exceed c. The boundary argument +gives threshold <=d_r, and the ridge path gives threshold >=d_r. +Consequently the intrinsic threshold is EXACTLY d_r on every field +(16) under the residual cutoff, even if remote critical points are +degenerate or share heights. No unconditional-Morse statement has +been applied to a conditional field. + +For the original Gaussian field, which is Morse with distinct critical +values almost surely, classical H0 elder linear algebra identifies this +threshold with the unique ordinary finite bar death of M. It is the +actual index-one saddle at s_3. The saddle at s_1 is closer in physical +distance, since s_1/r tends to a number near one and s_3/r to one near +three; ridge offsets are O(r^2). It does not kill M. Its two incident +local components have maxima M and the lower maximum at s_2; at this +earlier level the lower maximum dies. This can also be seen directly +by enclosing the s_2 maximum between the gate s_1 and a point between +s_2,s_3 whose ridge height is below g(s_1), with the same transverse +side walls. The ridge reaches M at height at least g(s_1), giving its +death g(s_1). Thus the profile is an actual global selector counterexample +to assigning each maximum to its nearest saddle, not merely an algebraic +ordering of candidate roots. + +Only the birth at M is selected by the sector (6): for the other maximum +the third and fourth jet relations would have to be tested afresh, and +it is not counted a second time for M's bar. The unique birth labeling +and the unique Q5>0 soft orientation are the once-counting rules. + +## 8. Exact actual-bar measure and the density pushforward + +Let K be the product of the boxes for b,a,zeta and U, with Lebesgue +measure db da dzeta d alpha d beta dq. Use ordinary arc length dtheta +on S^1, of total mass 2 pi, and dx on X. Let mu_{x,u} be the actual +Gaussian residual law in (8), and write A_R={||R||_C7<=R}. The repeated +letter R in this event means the fixed cutoff constant on the right; +inside integrals the residual is denoted omega to avoid ambiguity. +The exact one-point identity (9), changes (10)-(14), and the deterministic +death identification give, for every Borel B subset (0,infinity), + + M_A4(B)=(36/V) integral_{X,S^1,K} integral_0^r0 + alpha(-a)|a+alpha r^3| r^8 p_{x,u}(z_r) + integral 1{||omega||_C7<=R} 1{Delta(r,omega) in B} + mu_{x,u}(d omega) dr dK dtheta dx. (25) + +This is a measure of actual ordinary bars. The inner lifetime is not +an artificial sampled normal-form amplitude; it is the proved global +death difference of the conditionally coupled actual Gaussian field. +The root and its value are jointly Borel (indeed continuous on the +bounded local fields) by the uniform simple-root construction. Every +operation in (25) therefore permits Tonelli for nonnegative marks. + +Decrease r_0 again. From (23), uniformly, + + Delta'(r)=r^4(5 L_r+r partial_r L_r)>0, + 00 uniformly in all marks and cutoff +residuals. Choose ell_0 below this positive lower bound. For each +00 uniformly, r_ell=O(ell^(1/5)). (27) + +Every other factor in the integrand of ell^(-4/5)nu_A4 converges uniformly +over marks and cutoff residuals to its r=0 value. For example the density +is uniformly smooth on the compact set of z_r, and L_r-L_0=O(r), +while the denominator tends uniformly to 5L_0. For varying x,u, integrals +may be written using the single original field F and omega=H_{x,u}F; +this supplies a joint Borel kernel and an ordinary fixed probability +space for dominated convergence. There is no assertion that a varying +norm-cutoff boundary has zero probability; its event is fixed in r. +All limiting factors except the indicator are independent of omega. +Thus the coefficient in (3) is exactly + + C_A4=(36/(5V)) integral_{X,S^1,K} + alpha a^2 L_0(alpha,beta,q)^(-9/5) + p_{x,u}(z_0(b,a,zeta,q)) h(x,u) + dK dtheta dx. (28) + +The notation z_0 retains its full twenty-one entries from (12), with +the gradient, soft Hessian, T and Q4 pins equal to zero in reduced +coordinates, and Q5=q. The raw fourth/fifth entries retain their mixed +jet corrections. No factorization of this joint Gaussian density is +asserted. The universal scalar 36/5 comes from the chosen definition +r=-Q4/36 and the fifth-order height map; it is not a universal numerical +coefficient for every differently normalized A4 sector. + +The integration domain is compact with positive finite Lebesgue measure. +The full joint Gaussian density has a positive lower bound on its compact +target set, alpha>0, |a|>=1, L_0 is bounded above and below, and h>=1/2. +It follows directly that 00+, (29) + +because all remaining errors are uniformly O(r_ell). This estimate is +only for this fixed marked sector. Integrating the density asymptotic +yields (5/9)C_A4 t^(9/5). Equation (25) also gives that power directly +from r^8 dr. Density power 4/5 and cumulative power 9/5 are distinct. + +## 9. Exclusions and why this is not a whole-density correction + +Every bound above is uniform over exactly the stated compact marks and +the fixed residual cutoff. The proof excludes, rather than declares +negligible, the following regions: + +* Fold/cusp discriminants: U stays a positive distance from multiple + roots, root collisions and critical-height ties. Other scaled unfoldings + can change the elder partner and can dominate a whole-population term. +* Vanishing fifth ridge derivative: q stays near 24, in particular away + from zero. No sixth-order contact admission or matching is supplied. +* A second soft Hessian direction: a lies in [-2,-1]. No small-hard- + eigenvalue tail bound or higher-corank singularity law is supplied. +* Large height, mixed jets and residual norms: all have the fixed bounds + in (5), Section 5 and U. Gaussian tails make ordinary moments finite, + but do not by themselves prove that the omitted sectors are smaller + than ell^(4/5) in a matched whole-density expansion. +* Other homology degrees, other dimensions or curved geometries, and + convergence of point processes or higher factorial moments are not + claimed here. The geometric proof explicitly uses a two-dimensional + ridge and a single hard transverse direction. + +The theorem is not vacuous after these restrictions: (15) gives at least +half of every literal residual law, the finite target density is uniformly +positive, the sector has positive finite-dimensional volume, and (28) is +strictly positive. It is an actual Gaussian higher-singularity population +inside a field whose leading generic-bar law is already known. It supplies +a concrete taxonomy entry and a falsification of nearest-saddle selection; +it does not substitute for matching the rest of the taxonomy. + +As a diagnostic only, a potential A_k with height r^(k+1), control weights +k-1,...,1 on derivatives 2,...,k, and a one-point Rice weight r^(k-1) +would have radial cumulative degree (k-1)(k+2)/2. The corresponding formal +cumulative prediction is k/2-1/(k+1), giving 2/3,5/4,9/5 for k=2,3,4. +This is not a general A_k theorem: observation changes, corank, selection, +and complements must be proved separately. For k=4 those ingredients +have been proved above only on this regular compact sector. + +## 10. Classical imports, source custody, and evidence boundary + +The conventional mathematical imports are finite-dimensional inverse and +implicit function theorems, Taylor's integral remainder, Euclidean plane +coordinate changes, elementary finite Gaussian regression and uniqueness, +Tonelli and dominated convergence, and ordinary finite-dimensional change +of variables. The one-point all-Borel identity and original Morse/distinct- +height facts are imported from the exact fully read G proof, which derives +them using finite original coefficient minors, smooth equal-dimensional +Sard, and ordinary smooth incidence area change of variables. The global +H0 interpretation uses the classical Morse level/handle and elder linear +algebra package already explicit in G. The threshold enclosure itself was +proved for every conditional cutoff field and does not import a stochastic +global-selector hypothesis. + +The earlier Math240 fifth-order ridge correction was inspected only in +its setting/notation and equation (0.1); this source is not imported as an +A4 theorem. The corrections (4) were independently reconstructed here. +The scout identifies actual older A3 results and prevents claiming them as +new. It is an ownership/source survey, not mathematical acceptance of this +proposal. Exact input copies, SHA256 identities and read extents are in +INPUT_MANIFEST.json. The focused contributor's complete derivation is an +author contribution, not independent review. New exact controls check the +polynomial heights/derivatives, the two Jacobians, nonlinear ridge +corrections, exponent ledger and refusal cases. Those controls are finite +algebra checks, not a random-field experiment or proof of asymptotics. + +No old control suite was rerun. No field realization, numerical Rice +integration, coefficient enclosure, sampler/grid qualification, Lean command, +kernel audit, source-branch edit, remote mutation or scientific promotion +was performed for this author unit. The source is a local conventional +proof proposal until the separate exact-source review. No provider-distinct +acceptance, blind50-model validation, Hk theorem, manuscript acceptance, +external uptake, or institutional adoption follows. The larger research +goal remains ACTIVE / UNMET. diff --git a/frontiers/gaussian_a4_elder_sector_20261017/README.md b/frontiers/gaussian_a4_elder_sector_20261017/README.md new file mode 100644 index 000000000..50fba3dde --- /dev/null +++ b/frontiers/gaussian_a4_elder_sector_20261017/README.md @@ -0,0 +1,15 @@ +# Actual Gaussian compact A4 elder sector + +This is a source-author publication proposal carrying the full unchanged [proof](PROOF.md) and [actual integrated nonauthor review](REVIEW.md). Historical proposal/PENDING wording stays unchanged beside the actual later scoped acceptance. This index is not a new proof, review, source adoption, owner decision or scientific status register. + +One positive fixed compact regular marked Gaussian A4 ordinaryH0 elder sector on a fixed oriented flat two-torus; density C_A4 ell^(4/5)+O(ell), cumulative~(5/9)C_A4 t^(9/5), as ell,t decrease to0. These are the stated small-argument results and must not be treated as exact finite-scale power laws. + +Not whole-bar leading law/next whole-density correction; no negligibility of omitted discriminants, Q5=0/second-soft/large-jet/residual sectors, other dimensions/curvature/Hk/PPP, numerical or science. The full original proof and review control the exact hypotheses and limitations. All remaining original numerical/scientific obligations retain their evidence-specific dispositions; this packet does not close all five groups or the113retained failures. + +Exact GAD literal kernels/all-Borel one-point incidence plus shared finite-incidence supplement; named classical local/global Morse and differential facts; coauthor source is authorship, mark extension is reviewed reconstruction. + +The exact [everywhere Borel mark extension](EVERYWHERE_MARK_EXTENSION.md) is accepted reviewer reconstruction. The [ridge/frame derivation](coauthor/DERIVATION.md) is exposed COAUTHOR source, not nonauthor review or independent standalone theorem acceptance. + +The shared [source/dependency/exclusion index](../nongaussian_smoothing_admission_20261017/FOLLOWON_SOURCE_INDEX.json) records exact original prose identities, canonical source/review publications, immutable existing Math475/GAD and Main331/HF references, neutral historical-seal hashes and scope boundaries. Already landed GAD proofs/reviews and existing HF sources are referenced, not duplicated or replaced. Original sources and old seals were not edited. + +No Lean source, package, manifest, workflow, runtime, compiler target or execution receipt is registered here. Full books/PDFs/PNGs, private filesystem/fileID metadata, programs, failed attempts and recovery archives remain local; this is not a public ZIP, hermetic proof closure or portable execution package. New routing metadata has source-author provenance. The completed author/review sources here are source-exposed OpenAI, organizational/provider-distinct/blind/human credit0. Supplementary Claude input on the separate finite-L binding (moment/He/normalization only) is PENDING at Math193/6243336796; it supplies no completed review, adoption or independence credit. Existing C261 workflow/source history at this base is separate; this candidate changes none of it. Scientific effect NONE. diff --git a/frontiers/gaussian_a4_elder_sector_20261017/REVIEW.md b/frontiers/gaussian_a4_elder_sector_20261017/REVIEW.md new file mode 100644 index 000000000..47f8584aa --- /dev/null +++ b/frontiers/gaussian_a4_elder_sector_20261017/REVIEW.md @@ -0,0 +1,575 @@ +# Nonauthor review of the actual Gaussian A4 elder sector + +**Verdict: ACCEPT_SCOPED. No must-fix finding or required mathematical source +amendment was found for the exact frozen proposal under the stated law, +sector restrictions, and classical imports.** This review makes explicit the +everywhere Borel observable extension needed to pass from the original +almost-sure bar count to equation (25); it does not infer conditional Morse +behavior from unconditional genericity. + +Reviewer: OpenAI/Codex `/root/reverse_localization_nonauthor_review_20261017`. +Date: 17 October 2026. Root, the A4-sector author, and its focused ridge/frame +contributor are mathematical coauthors. The present reviewer did not author +this A4 proof, its sector mechanism, or its focused derivation. The reviewer +previously reviewed the consumed Gaussian admission theorem G and authored +its finite smooth incidence reconstruction. That earlier work is disclosed +infrastructure exposure, not fresh independent review credit for this A4 +claim. The focused A4 derivation is exposed COAUTHOR evidence only. All +participants share OpenAI and are source/route exposed. Organizational, +provider-distinct, blind, and human independence credit are all zero. No +helper was spawned for this unit. + +## 1. Exact source, actual reads and disposition boundary + +The reviewed A4 source is the complete frozen `PROOF.md`, 32065 UTF-8 bytes, +614 lines, SHA256 +`abed6f94da080a33458b78f1e60ae92ad90083323af41d241de05939d4377ff0`. +The exact author freeze is 2651 bytes, SHA256 +`4ffbe56ef508c6e543c82e6319c2cde0aad603fcf49988920f6a3ee9bfc7ea8d`, +binding 15 files totaling 225875 bytes. The full proof was read only after +that identity was supplied as final. All 614 lines were read. There was no +verdict on a draft, and no author file has been edited during review. + +The precise consumed documents are: + +| Source | Bytes | SHA256 and actual extent | +| --- | ---: | --- | +| Gaussian admission G proof | 40231 | `8d992e194f0a898e563c4c3d05b8079cde292f9e87d7ff51030ad12ef5faf319`; full 739-line body read in the preceding task in this conversation, exact bytes reverified here | +| G source-bound scoped review | 38035 | `6fedc43ba34408f23f8934166b3898e2cd19175c57e61e37227c85f3ca5d4c70`; full body, previously authored by this reviewer; exposure disclosed | +| G finite smooth incidence appendix | 5032 | `b1b928ce32a6a39b464cb9f1c6dd82f61b95d2bb095459d9b02f301168bd7fb2`; full, previously authored and freshly read here | +| Higher-singularity scout report | 13771 | `584e035a80a1832e727ad9995f1880dee4713041bfe63c2570ece612ac75cf38`; full fresh read | +| Focused ridge/frame COAUTHOR derivation | 14969 | `d667ca3541fc7187abbc6757fb73f1c27784e4aefc2dc42f215bdee86d786946`; full fresh read | +| Math240 cusp-parity source | 51848 | `35b9c5943cf35919f2046ba194945ba759be654edf5d5ad77ba17b1f5c5377fd`; retained whole, lines 44–95 read for notation and equation (0.1), no full-source review claimed | + +The author result and input manifest, focused coauthor result and freeze, +and exact freeze-bound file identities were also inspected. Retained copies +are under `inputs/`. The scout reports previous A3 results and proposes A4 +powers provisionally; its report is not itself an A4 theorem. Math240 is +used only as a notation/formula reference. The new ridge formulas are actually +derived in the A4 proof and independently reconstructed below, so no Math240 +whole-density theorem, matching lemma, or inherited acceptance is required. +No old source suite or author control program was executed. + +Acceptance is for each fixed actual Gaussian series on the fixed oriented +flat two-torus X=R^2/(L Z^2), L>0, + + F=sum_j a_j Z_j phi_j, + Z_j original independent standard real Gaussians, + a_j>0, span_R(phi_j) dense in C-infinity(X), + sum_j a_j ||phi_j||_Cq=0. + +This includes the explicit positive exponentially weighted Fourier example +in equation (2). All law/geometry/sector constants are fixed-law constants. +Stationarity and rotational invariance are not required. The theorem is +broader than the one example but narrower than all abstract smooth Gaussian +laws with no proof of this representation. G supplies smoothness, moments, +named finite jet rank, exact Borel kernels/Rice, and original Morse/distinct +critical values for this explicit class. Its generic fold conclusion is not +being used as an A4 geometry or selection proof. + +The observable is an actual finite ordinary superlevel H0 elder bar marked +at its birth maximum. The mark retains hard eigenvalue a in [-2,-1], birth +height b in [-1,1], twelve other raw third-to-fifth derivative entries in +[-1,1]^12, a nonempty small compact box of (alpha,beta,q) near (6,22,24), +a positive fifth reduced derivative, 00, a in [-2,-1], and -lambda>a. In raw ambient coordinates the measure +is dH11 dH12 dH22, with one unweighted off-diagonal entry. The rotation +parametrization H=R_theta diag(-lambda,a)R_theta^T has absolute Jacobian +|a+lambda|. This follows by direct differentiation at theta=0 and transport +by determinant-one conjugation on raw symmetric coordinates. It is also +verified independently at nonzero rational rotations in the fresh controls. +The eigenvalue ordering fixes which eigenline is soft; no eigenvalue-swap +multiplicity remains. + +Each Hessian eigenline has two oriented lifts theta and theta+pi on the +full circle. Reversing both frame vectors negates every odd-order derivative +and fixes every even-order derivative. Thus Q5 changes sign, while Q4 does +not. The condition Q5>0 chooses exactly one lift. Consequently full ordinary +arc length on [0,2pi) with Q5>0 counts once. There is no remaining antipodal +half. In particular the half used to average an oriented cubic moment in +the previously reviewed fold formula cannot be carried into this different +one-point positive-Q5 chart. + +At each derivative order n the SO(2) action on raw Sym^n coordinates has +determinant one: it preserves a positive tensor norm, is connected to the +identity, and has positive determinant. The explicit binomial rotation +matrices through n=5 confirm this independently. The combined Jacobian is +block triangular: theta dependence of the higher rotated jets contributes +only lower off-diagonal blocks beneath the Hessian block. Gradient rotation +also has determinant one. Hence the full joint density is the actual +p_(x,u)(z_r), with the sole spectral Jacobian |a+lambda|. A Frobenius- +normalized off-diagonal coordinate would require its own density adjustment; +mixing that measure with the raw density would change the constant. + +These are coordinate changes of the actual full Gaussian jet integral. +They are not a Gaussian regression conditioned on a field-selected frame, +and do not require rotational invariance of the law. The resulting covariance +and density may depend on x and theta and retain all cross-correlations. + +## 4. Reduced jets and the radial mass + +In the birth eigenframe F_v=F_uv=0 and F_vv=a is bounded away from zero. +The unique transverse critical ridge t=h(s) has h(0)=h'(0)=0. Solving +F_v(s,h(s))=0 by coefficient comparison yields + + h(s)=-(gamma/(2a))s^2 + +[-eta/(6a)+B gamma/(2a^2)]s^3+O(s^4). + +Substitution into F gives + + g''(0)=-lambda, g'''(0)=T, + g''''(0)=F_uuuu-3 gamma^2/a, + g'''''(0)=F_uuuuu-10 eta gamma/a+15 gamma^2 B/a^2. + +The terms involving the cubic ridge coefficient cancel at order five because +the transverse critical equation holds. There is no missing lambda term: +the vanished mixed Hessian entry and first ridge slope preclude it. This +was independently checked by solving the formal ridge equation through order +five in fresh rational polynomial arithmetic across several hard/soft and +mixed-jet choices, rather than inserting the author's formulas into themselves. + +All twelve other higher raw entries are retained. For fixed a and those +entries, replacing the pure fourth and fifth derivatives by Q4,Q5 is a +triangular translation with determinant one. Thus the raw Gaussian target +at r=0 retains + + F_uuuu=3 gamma^2/a, + F_uuuuu=q+10 eta gamma/a-15 gamma^2 B/a^2. + +Setting these raw coordinates to (0,q), or imposing small raw fourth derivative +instead of small reduced fourth derivative, would describe a different sector +and give the wrong density. There is no independent Gaussian density for +the nonlinear Q4,Q5 coordinates. + +The exact radial definitions are lambda=alpha r^3, T=beta r^2, Q4=-36r. +Their absolute three-variable Jacobian is 36r^5. One-point Rice contributes +lambda(-a), and the spectral coordinates contribute |a+lambda|. Therefore +before lifetime pushforward the exact mass is + + 36 alpha(-a)|a+alpha r^3| r^8 dr. + +There is one Rice determinant, no candidate saddle density, and no second +root-counting determinant. Both hard factors survive: the leading weight +is 36 alpha a^2 r^8. The radial exponent is not deduced solely from a quintic +height scale; it is obtained from the actual observation-coordinate volume +and physical birth Hessian weight. + +## 5. Fixed residual cutoff and the uniform C7 argument + +G gives uniform smooth moments for omega=HF; explicitly +||omega||C7<=||F||C7+C||F||C5. Compact covariance coefficients bound the +right-hand moments uniformly in x,u. Choose finite R above twice the uniform +first expectation. Markov then gives h(x,u)=P(||omega||C7<=R)>=1/2. This +uses neither a small-ball theorem nor absence of a norm atom. + +For each x,u, the same omega is used at every r. The target z_r is a smooth +polynomial in r on the compact parameter boxes, with a uniformly bounded +r derivative. The mean map c Gamma^-1 is bounded into every required smooth +norm. Thus f_r=omega+m(z_r) and partial_r f_r have one uniform spatial C7 +bound on the fixed residual event. In the actual conditional field, + + H f_r=H omega+H m(z_r)=omega. + +Accordingly the cutoff in the original birth mark is precisely the cutoff +used in the integral. It is not an independent auxiliary variable, a +renormalized conditional field law, or a total-field norm cutoff that moves +with r. A positive-probability atom at the threshold would cause no difficulty, +because this event is fixed throughout each radial change of variables. + +A common embedded rectangle exists with F_vv<=-1/2 by its value a<=-1 and +the uniform C3 bound. The implicit ridge has uniform size and derivative +bounds without dividing by the soft eigenvalue. One can obtain existence +by the transverse intermediate value theorem at t=+-C s^2 and then use the +implicit derivative identities. All denominators involve the hard transverse +Hessian. This gives a uniform common local chart for every conditional +cutoff field. + +Under spatial C7 bounds the ridge is C6. The identity g_s=F_u(s,h(s)) gives +g itself C7 without a seventh derivative of h. The parameter envelope identity +partial_r g=(partial_r f)(s,h(s)) is spatial C6. Taylor's integral formula +therefore gives a spatial order-six remainder E_r(s), with E_s^j=O(s^(6-j)) +for j<=2 and partial_r E=O(s^6). The required mixed estimates follow by +applying the same Taylor formula to partial_r g; in particular +partial_r E_s=O(s^5). Consequently + + r^-5(g_r(r xi)-b)=P_(alpha,beta,q)(xi)+O(r) in C2, + +and its r derivative is uniformly bounded in C1 on every fixed bounded +xi interval. Differentiating r^-5 E_r(r xi) includes both the explicit r +remainder derivative and the spatial chain term; the powers cancel to a +bounded expression. There is no hidden C8 requirement. These bounds justify +radial critical-root motion and lifetime differentiation, rather than only +a pointwise Taylor asymptotic. + +## 6. Nonempty regular sector and exclusion of extra roots + +At (alpha,beta,q)=(6,22,24), + + P=xi^5/5-3xi^4/2+11xi^3/3-3xi^2, + P'=xi(xi-1)(xi-2)(xi-3). + +The critical heights are P(0)=0, P(1)=-19/30, P(2)=-4/15, +P(3)=-9/10. Their order is strictly P(3)=C2 r, the +quartic term dominates the two negative lower terms when C2 is large, and +again dominates the remainder when rho is small. In the bounded scaled +interval, the nonzero reduced derivative away from disjoint root neighborhoods +and the nonzero root derivatives inside those neighborhoods preclude extras. +At zero the fixed negative reduced second derivative anchors uniqueness. +Hence exactly four critical points occur in the entire coordinate tube. + +Their physical inertia follows from the Hessian Schur complement, with the +negative transverse direction fixed: maximum, index-one saddle, maximum, +index-one saddle. Their height ordering persists. Implicit critical roots +have bounded r derivatives because the scaled derivative has bounded parameter +derivative and a uniform nonzero xi derivative at each root. At a critical +value the root-motion term vanishes. Therefore + + Delta(r)=b-g_r(s3)=r^5 L_r, + L_r=L0+O(r), |partial_r L_r|<=C, + L0=-P_(alpha,beta,q)(xi3)>0 + +uniformly over every compact mark and every cutoff residual. At fixed endpoint +rho the positive leading quintic and the bounded sixth remainder give +g(-rho)b+c rho^5 once rho, then r0, are small. All +choices depend on the fixed law and cutoff, which is allowed. + +## 7. Global physical enclosure and the farther elder saddle + +The crucial deterministic construction works for every conditional cutoff +field, even with arbitrary remote degeneracies or equal critical heights. +Use a curved tube from s=-rho to s=s3, of fixed transverse half-width delta +around the ridge. Uniform ridge size permits one embedded rectangle and +fixed positive delta; r0 is then made smaller so Delta is below both the +left-face quintic margin and the transverse delta^2 margin. + +Transverse concavity gives f(s,h(s)+z)<=g(s)-z^2/4. The tube's left and side +faces are strictly below d=b-Delta, and its right face is at most d, attaining +it only on the ridge. Its entire interior has height at most b, because the +birth at 0 is the highest of its local extrema and the ridge has no additional +critical points. Thus every continuous path from the birth to ANY point of +height greater than b must leave this embedded tube and cross a level at most +d. No path elsewhere on the torus can avoid that boundary. + +Conversely the actual ridge from 0 to rho reaches a point higher than b. +Its lowest level is exactly d at s3, since the intervening first saddle s1 +has strictly higher height and there are no further local critical points. +For every cc} to a higher point. For every c>d the +boundary forbids such a connection. The intrinsic supremum death threshold +is therefore exactly d. Whether the strict superlevel set connects at c=d +is irrelevant; the supremum recovers the closed ordinary death level. + +This proves actual global elder selection once applied to the original +almost-sure Morse/distinct field. It does not assume remote branches are +older, appeal to unconditional genericity on a null conditioning event, or +pair a maximum to its nearest critical point. The birth at 0 survives the +near saddle s1 because the other local maximum at s2 is younger and dies +there. A smaller analogous enclosure around s2, with the ridge path to the +older birth 0, proves that earlier death. Birth 0 dies at the farther saddle +s3. The outgoing ridge point proves the selected birth is not essential. +The physical distances are r times roots near 1 and 3 plus O(r^2) transverse +offsets, so the closer saddle is indeed s1 for small r. + +The fresh exact ridge-graph control independently gives birth0->saddle3 and +birth2->saddle1. Adding a bypass vertex at a higher level changes birth0's +death. That refusal demonstrates why algebraic root ordering alone is not a +global proof. The source's embedded boundary rules out precisely such a +bypass. A symmetric full quintic box cannot have every boundary point below +the death level, because the positive arm must rise above birth; the actual +asymmetric tube ends at s3 and avoids that false barrier. + +Each ordinary bar has one birth maximum. The mark chooses one positive-Q5 +orientation at that maximum. A second nearby maximum is a different birth, +not a second label for the first bar. At the seed it also has positive local +reduced fourth derivative, whereas this sector requires Q4<0; the once-counting +argument fundamentally uses unique birth labels and orientation, not a claim +that the Gaussian field has only one marked maximum globally. + +## 8. The important original-count/full-jet-kernel join + +Section 2 defines the actual count to be zero off the original Morse/distinct +locus. Section 7 proves an intrinsic threshold for every cutoff conditional +field. These statements must not be joined by asserting that conditional +full-5jet fields inherit unconditional global Morse behavior. That inference +would be invalid. The valid join uses an everywhere Borel observable extension; +its full details are retained in `EVERYWHERE_MARK_EXTENSION.md`. + +Let S(x,f) be the local sector tests without any global Morse indicator, and +let D_f(x) be the everywhere Borel intrinsic threshold supplied in §7. Define +on all smooth fields + + w_B(x,f)=S(x,f) 1{D_f(x) finite} + 1{f(x)-D_f(x) in B}. + +The original Gaussian law is Morse with distinct heights almost surely. +On that event its critical-root sum of w_B is exactly N_A4(B), by the global +threshold proof and ordinary H0 elder identification. Consequently their +expectations agree even though the definitions may differ on the original +null complement. Apply the all-Borel Rice identity to w_B. At every literal +z_r its value on a cutoff field is precisely 1{Delta(r,omega) in B}, by the +EVERY-field enclosure proof. Also H f_z=omega, so the residual part of the +actual sector mark is exactly the fixed residual event in that kernel. +This gives (25) directly. No conditional Morse indicator was present in that +extension, so no such indicator was dropped. + +This is a standard almost-sure observable equality followed by a valid Borel +Rice application. The source explicitly supplies the threshold, the original +almost-sure identification, and the all-Borel identity, and states that its +conditional density/trapping proof does not assume global Morse. Thus its +intended inference is mathematically complete without a new hypothesis. +A one-sentence declaration of this w_B before (25) would make the join easier +to read; I do not require a source amendment for that explanatory sentence. +Acceptance is precisely for this valid reconstruction, not for a transfer +of unconditional genericity to every full-jet target. This interpretation +was independently identified by the reviewer and separately raised by Root; +Root is a coauthor and receives no nonauthor credit for that observation. + +## 9. Exact pushforward, density version and cutoff order + +After the one-point, spectral and reduced-jet changes, equation (25) is an +identity of actual expected bar measures for every Borel B. Its radial weight +is the exact r^8 expression above, multiplied by the original joint density +p_(x,u)(z_r) and the actual residual law. The root construction and the +smooth regression map make the local lifetime jointly Borel. Nonnegative +Tonelli is therefore legitimate before invoking any asymptotic estimate. + +The order of choices matters: fix the law and compact regular parameter box; +choose the finite residual cutoff using its uniform moment; obtain a common +hard-concave rectangle, rho, delta, and uniform Taylor constants; then shrink +r0 for the simple-eigenvalue chamber, root exclusion, height/trap margins and +lifetime monotonicity. From L_r=L0+O(r) and bounded partial_r L_r, + + Delta'(r)=r^4(5 L_r+r partial_r L_r)>0 + +with uniform positive upper/lower factors for 0=c r0^5>0 supplies ell0>0 below the endpoint for every +mark and residual. The source's informal `min Delta(r0,omega)` is used only +for this uniform positive lower bound; attainment on a C7 ball is neither +proved nor needed. An infimum would state the same necessary condition. + +Each 0L0, 5L_(r_ell)+r_ell L'_(r_ell)->5L0. + +The finite density p(z_r) is uniformly smooth on the compact target set, +including the nonlinear mixed-jet translations. The remaining coefficients +also converge uniformly. No Gaussian tail estimate beyond the fixed cutoff +is needed here. To integrate x,u with their varying residual laws, write +omega=H_(x,u)F on the single original probability space; joint Borelness +permits Tonelli and dominated convergence. The norm event is fixed in r, +so no zero-probability cutoff-boundary assumption is made. + +It follows that + + C_A4=(36/(5V)) integral_(X,S1,K) + alpha a^2 L0(alpha,beta,q)^(-9/5) + p_(x,u)(z0(b,a,zeta,q)) h(x,u) dK dtheta dx. + +The mark space has positive finite volume; alpha>0, |a|>=1, L0 is bounded +above and below; the full jet Gaussian density is positive with a compact +lower bound, and h>=1/2. Hence the coefficient is finite and strictly +positive. No independent-order factorization is available or used. The +36/5 scalar belongs to the specific normalization Q4=-36r; changing the +sector normalization changes its numerical expression together with the +domain and target. It is not a universal numerical constant for all A4 +normal forms. + +Uniform O(r_ell) errors give +nu_A4(ell)=C_A4 ell^(4/5)+O(ell) within this fixed sector. Integrating gives +the coefficient 5/9 and cumulative power 9/5. The same cumulative exponent +also follows from r^8 dr and a fifth-order height. The fresh exact tests +keep the L0^-9/5 factor and distinguish density from cumulative powers. +No all-population matching or differentiation of a cumulative estimate is +hidden in this passage. + +## 10. Cutoff complement and unresolved matching + +The following exclusions remain explicit, not negligible errors: + +* U stays away from fold/cusp discriminants, multiple roots and critical-height + ties. Approaching these boundaries can change the partner and population + scaling; no matching estimate is provided. +* q=Q5 stays near 24, away from zero. No sixth-order or more degenerate + contact is admitted. +* The hard eigenvalue stays in [-2,-1]. A second soft direction and hard- + curvature tails require another analysis. +* Height, twelve mixed/raw higher jets, and the actual residual norm retain + their fixed cutoffs. Ordinary Gaussian moments do not establish that the + omitted large-mark sectors are o(ell^(4/5)). +* There is no exhaustion over all compact sectors, dimensions, geometries or + higher homology; no factorial-moment or Poisson argument is present. + +The sector is nevertheless a genuine positive population, as the coefficient +proof shows. It coexists with G's all-bar generic fold intensity of order +ell^-1/3 and does not modify that leading law. The source's A_k weight formula +in §9 is explicitly diagnostic. Only k=4 in this fixed two-dimensional regular +compact sector is proved by this unit. No taxonomy-wide theorem is inferred. + +## 11. New controls, retained negative evidence and imports + +The new reviewer-authored `independent_a4_controls.py` ran once and passed +226 exact checks with nine negative/refusal families, exit 0 and empty stderr. +It imports no author or previous test program. Its methods are formal power +series elimination of F_y=0, raw binomial tensor rotation matrices at nonzero +rational angles, exact polynomial values, a finite graph elder algorithm, +and rational power/coefficient algebra. They check the two mixed ridge +corrections independently, raw Hessian/angular measures, nearby elder pairing, +and lifetime coefficient powers. The bypass graph and r-dependent toy cutoff +are deliberate refusal examples, not simulated Gaussian counterexamples. + +The nine negative families reject raw fourth in place of reduced fourth; +omitted mixed fifth-jet terms; mixing Frobenius and raw off-diagonal measure; +nearest-saddle pairing; remote-bypass exclusion from local order alone; +a full symmetric quintic boundary barrier; an extra antipodal half; +a dropped spectral hard factor; and replacing a shrinking residual mark +by fixed positive mass. Some assertions concern exact coordinate or toy +pushforward algebra; no claim is made that every finite control is itself +an actual Gaussian field or a proof of the analytic asymptotic. + +The author's 281 reported controls in eight refusal families were not rerun +or counted as independent review evidence. The focused ridge/frame document +is an author derivation, not a second reviewer. Its result retains an optional +failed `sympy` import (`ModuleNotFoundError`, no installation attempted). That +probe is an access/dependency event, not a mathematical failure, and is +preserved in its retained result. The present review's fresh control run had +no failure. No field was sampled and no numerical integral or approximate +eigenvalue/root computation was used. + +The conventional imports remain finite-dimensional implicit/inverse +function theorems, Taylor integral remainder, elementary Euclidean coordinate +changes, finite Gaussian regression/uniqueness, Tonelli, dominated convergence, +and ordinary change of variables. G explicitly supplies the finite-original- +coefficient smooth Sard/incidence proof, actual smooth moments/ranks and +original almost-sure Morse/distinct-height facts. Classical Morse local form, +critical-free flow/handle facts and elementary H0 elder algebra are retained +as the same named premises of the consumed G source. The threshold enclosure +itself is directly proved here and does not import a new global stochastic +selector theorem. + +No new external primary-book read or network lookup was needed for this +unit's independent finite reconstruction. The complete Math240 source is +retained for identity, while only its specified ridge-formula excerpt was +read. No full-source adoption of that old A3 chain is claimed, and the scout's +linked remote source findings are not independently promoted by this review. +This distinction prevents a source-pointer read or author lineage from +silently becoming a newly proved dependency. + +## 12. Final source-bound disposition + +The actual global enclosure, full-jet conditional law, one-point Borel +observable extension, coordinate multiplicities, uniform C7 radial argument, +unique inverse lifetime map, finite pointwise density representative and +strictly positive coefficient all survive independent reconstruction. +The exact frozen source therefore receives **ACCEPT_SCOPED**, with +`must_fix_findings=[]` and `required_amendments=[]`. + +The two explanatory details recorded above are not new mathematical premises: +the all-field intrinsic threshold supplies the valid Rice-mark extension, +and a uniform infimum bound suffices where the source writes `min`. Neither +requires source alteration for the theorem as stated. Acceptance does not +permit a different derivation based on conditional Morse transfer or an +attained minimum on an unproved compact residual ball. + +The final machine verdict and manifest bind the full review, all retained +inputs, new control program/result/receipt, and exact source readback. The +author freeze and all old sources remain unchanged. No source/native write, +field draw, Lean/kernel operation, archive rebuild, institutional status +change or scientific promotion occurred. Root must read the entire sealed +review before any source adoption or publication decision. Organizational, +blind, provider-distinct and human independence credit remain zero. The larger +project goal remains ACTIVE / UNMET; scientific effect is NONE. diff --git a/frontiers/gaussian_a4_elder_sector_20261017/coauthor/DERIVATION.md b/frontiers/gaussian_a4_elder_sector_20261017/coauthor/DERIVATION.md new file mode 100644 index 000000000..ae7c81553 --- /dev/null +++ b/frontiers/gaussian_a4_elder_sector_20261017/coauthor/DERIVATION.md @@ -0,0 +1,232 @@ +# A4 compact-sector eigenframe and ridge coauthor derivation + +This is a focused analytic **coauthor contribution**, not a nonauthor review or a claim that the complete Gaussian A4 sector theorem has been proved. The main author owns the global selection argument and final proof. No source publication, network mutation, Lean execution, field draw, or reservation is made here. + +## 1. Conventions and precise conclusions + +Work on a fixed flat oriented two-dimensional torus. Write an oriented orthonormal frame as `(u,v)`, with `v=Ju`, and use the raw mixed-derivative coordinates of every symmetric tensor. In particular the Hessian measure is + +`dH11 dH12 dH22`, + +with one, unweighted, off-diagonal coordinate. At a critical maximum let the soft eigenvalue be `s=-lambda`, the hard eigenvalue be `a`, and assume + +`a <= -a0 < 0`, `0 < lambda`, and `-lambda > a`. + +The conclusions established below are: + +1. The Hessian change of variables has absolute Jacobian `|a+lambda|`. +2. A full angular circle together with the restriction `Q5>0` counts each admissible oriented soft eigendirection exactly once. There is no further factor `1/2` in this convention. +3. The ridge corrections are + + `Q4 = Fuuuu - 3 gamma^2/a`, + + `Q5 = Fuuuuu - 10 eta gamma/a + 15 gamma^2 B/a^2`, + + where `gamma=Fuuv`, `eta=Fuuuv`, and `B=Fuvv`. + +4. Replacing the two pure derivatives `(Fuuuu,Fuuuuu)` by `(Q4,Q5)`, with all other jets fixed, has Jacobian one. +5. The sector substitution + + `(lambda,T3,Q4)=(alpha r^3,beta r^2,-36r)` + + has absolute Jacobian `36 r^5`. Together with one-point Rice and the Hessian parametrization, the exact finite-r algebraic weight is + + `36 alpha (-a) |a+alpha r^3| r^8`. + + Its leading coefficient is `36 alpha a^2 r^8`. +6. Uniform spatial `C7` bounds on the conditional fields and their parameter derivative suffice for the bounded-r-derivative remainder needed in the lifetime change of variables. No `C8` bound is required by this step. + +All density formulas below retain the actual full joint Gaussian jet density in the moving frame. Rotation invariance of the field and independence of derivative orders are not used. + +## 2. Hessian Jacobian with the raw off-diagonal coordinate + +Let `R_theta` have columns `u_theta=(cos theta,sin theta)` and `v_theta=(-sin theta,cos theta)`. Then + +`H=R_theta diag(-lambda,a) R_theta^T`. + +In the fixed ambient frame, + +`H11=-lambda cos^2 theta+a sin^2 theta`, + +`H12=(-lambda-a) sin theta cos theta`, + +`H22=-lambda sin^2 theta+a cos^2 theta`. + +At `theta=0`, the columns of the derivative with respect to `(lambda,a,theta)` are + +`(-1,0,0)`, `(0,0,1)`, and `(0,-lambda-a,0)`. + +The determinant therefore has absolute value `|-lambda-a|=|a+lambda|`. Conjugation by `R_theta` on symmetric matrices has determinant one in the coordinate volume `dH11 dH12 dH22`: it is a continuous representation of the rotation group; its determinant is positive, and it preserves the equivalent quadratic form `H11^2+2H12^2+H22^2`, so its determinant is exactly one. This transports the calculation at zero to every angle. + +Consequently, on the simple-eigenvalue chamber with the eigenvalues ordered as above, + +`dH11 dH12 dH22=|a+lambda| d lambda da d theta` + +when `theta` ranges over any half-open interval of length `pi`. This is a one-to-one parametrization except for the angular endpoint, a null set. Using an orthonormalized off-diagonal coordinate `sqrt(2)H12` would change the coordinate density and Jacobian together. That is a different convention and supplies no extra factor here. + +The gradient rotation has determinant one. For each order `k`, rotation on `Sym^k(R^2)` has determinant one in the raw mixed-derivative coordinates: it preserves a positive factorial-weighted quadratic form and is connected to the identity. Thus simultaneous conversion of all jets through order five into the moving frame has unit determinant at fixed `theta`. Although the moving-frame high-jet coordinates depend on `theta`, the combined Jacobian is block triangular once the Hessian coordinates have been parametrized. Its only nonunit factor is the Hessian factor just calculated. + +## 3. Why the full circle with Q5>0 has no extra half + +Changing `theta` to `theta+pi` changes both frame vectors to their negatives. A derivative tensor component of total order `k` gains the factor `(-1)^k`. Thus `a`, `lambda`, `Fuuuu`, and `eta` are unchanged; `gamma`, `B`, `T3=Fuuu`, and `Fuuuuu` change sign. The formulas in Section 4 give + +`Q4(theta+pi)=Q4(theta)`, `Q5(theta+pi)=-Q5(theta)`. + +On the sector `Q5` in a compact interval contained in `(0,infinity)`, each soft eigenline has exactly one representative on the full circle satisfying `Q5>0`. Therefore the parametrization + +`theta in [0,2pi)`, `Q5>0` + +is one-to-one on this chamber, up to the angular endpoint. Equivalently, start with the length-pi angular parametrization and split its `Q5>0` and `Q5<0` pieces. On the negative piece, send `theta` to `theta+pi` and negate all odd-order components. This is a unit-Jacobian transformation and sends the negative piece into the other half of the full-circle positive-Q5 chart. + +The same argument applies with additional sector restrictions, including `T3/r^2` in a positive interval: those restrictions are evaluated only in the uniquely selected positive-Q5 frame. There is no claim that their values are the same in both frames. There is no unordered-pair convention in this one-point birth count. + +The event `Q5=0` is outside the chosen compact sector. Under a nonsingular full jet density it is also a null hypersurface. It plays no role in this normalization. + +## 4. Exact nonlinear ridge derivatives + +Use Euclidean coordinates `(s,y)` along `(u,v)` centered at the critical point. Suppose + +`F_s=F_y=F_sy=0`, `F_yy=a != 0`. + +The implicit-function theorem gives the transverse critical ridge `y=h(s)`, defined near zero by `F_y(s,h(s))=0`, with `h(0)=h'(0)=0`. Set + +`c2=h''(0)`, `c3=h'''(0)`. + +Differentiating the defining equation, or comparing its Taylor coefficients, gives + +`a c2+gamma=0`, + +`a c3+3B c2+eta=0`. + +Hence + +`c2=-gamma/a`, `c3=-eta/a+3B gamma/a^2`. + +Let `g(s)=F(s,h(s))`. Up to order five, the terms affected by the ridge are precisely + +`a h(s)^2/2 + gamma s^2 h(s)/2 + B s h(s)^2/2 + eta s^3 h(s)/6`. + +Writing `h(s)=c2 s^2/2+c3 s^3/6+O(s^4)`, the order-four coefficient is + +`a c2^2/8+gamma c2/4=-gamma^2/(8a)`. + +After multiplication by `4!`, this gives + +`g''''(0)=Fuuuu-3gamma^2/a=Q4`. + +The order-five coefficient is + +`a c2 c3/12+gamma c3/12+B c2^2/8+eta c2/12`. + +The first two terms cancel because `a c2+gamma=0`. Multiplying the remaining terms by `5!` gives + +`g'''''(0)=Fuuuuu-10 eta gamma/a+15 gamma^2 B/a^2=Q5`. + +Also `g''(0)=-lambda` and `g'''(0)=T3`. None of these identities requires `lambda=0`. + +The local source `singularity-taxonomy-next-gap-20261017/math-elder-cusp-proof.md`, SHA-256 `35b9c5943cf35919f2046ba194945ba759be654edf5d5ad77ba17b1f5c5377fd`, lines 63-74, uses the same correction: its `120 Q1` is the present `Q5`. Only this formula and its stated conventions were consulted here; no claim of a new whole-source review is made. + +Fix `a`, `gamma`, `eta`, `B`, and every jet coordinate other than the two pure components. Then + +`(Fuuuu,Fuuuuu) -> (Q4,Q5)` + +has derivative matrix equal to the two-dimensional identity. It is an affine translation in these two coordinates; its Jacobian is exactly one. With the other coordinates retained as variables, the complete transformation remains triangular with unit determinant. The inverse formulas are + +`Fuuuu=Q4+3gamma^2/a`, + +`Fuuuuu=Q5+10 eta gamma/a-15 gamma^2 B/a^2`. + +They are smooth with uniformly bounded derivatives on a compact hard-eigenvalue interval bounded away from zero. A condition on the raw pure fourth derivative alone is not the condition `Q4=-36r`. + +## 5. Radial sector substitution and the coefficient ledger + +Let `omega` denote all moving-frame derivatives of orders three through five other than `T3`, `Fuuuu`, and `Fuuuuu`; there are twelve such raw mixed-derivative coordinates. In particular `gamma`, `eta`, and `B` are retained in `omega`. Retain `b=F(0)` and the hard eigenvalue `a`. Use + +`lambda=alpha r^3`, `T3=beta r^2`, `Q4=-36r`, `Q5=q`. + +The derivative of `(lambda,T3,Q4)` with respect to `(alpha,beta,r)` is + +`[[r^3,0,3alpha r^2],[0,r^2,2beta r],[0,0,-36]]`. + +Its absolute determinant is `36r^5`, with no factors involving alpha or beta. The one-point Rice determinant is `|det H|=lambda(-a)` on the maximum chamber. Multiplying it by the Hessian and radial Jacobians gives + +`lambda(-a)|a+lambda| * 36r^5 = 36alpha(-a)|a+alpha r^3|r^8`. + +For sufficiently small r, compactness and `a<=-a0` give `-lambda>a`; the absolute value can then be written `-a-alpha r^3`. The factor divided by `r^8` tends uniformly to `36alpha a^2`. + +To specify the density argument without hiding any Gaussian independence, order the complete moving-frame 5-jet as + +`Jtheta=(F,Fu,Fv,Fuu,Fuv,Fvv, all raw order-3, order-4, order-5 entries)`. + +Its dimension is 21. The target is + +`z_r=(b,0,0,-alpha r^3,0,a,T3=beta r^2, omega, Fuuuu=-36r+3gamma^2/a, Fuuuuu=q+10eta gamma/a-15gamma^2B/a^2)` + +with entries placed in their stated tensor order. Let `p_{x,theta}(z_r)` be its actual 21-dimensional density. At r=0 the limiting target retains the nonlinear mixed-jet terms: + +`Fuuuu=3gamma^2/a`, + +`Fuuuuu=q+10eta gamma/a-15gamma^2B/a^2`. + +These raw derivatives are generally nonzero. Replacing them by `(0,q)` in the Gaussian density would be an error. + +For fixed `(x,theta)`, write the canonical Gaussian conditioning as `F_r=m(z_r)+R`, with R independent of the jet and with a distribution not depending on r. If all components through order five are conditioned, the residual projection is unchanged by a moving-frame rotation, because that rotation is an invertible change of coordinates on the same finite jet span. The notation may still retain `(x,theta)` if convenient. + +Suppose the main proof supplies a deterministic marked lifetime + +`Delta_r=r^5 L_r`, + +on a compact parameter box and on a fixed residual cutoff, with `L_r -> L0>0` uniformly and `sup |partial_r L_r|0`. + +The lifetime density obtained from this radial coordinate therefore has leading coefficient + +`(36/(5 Vol X)) integral alpha a^2 L0^(-9/5) p_{x,theta}(z_0) P(residual cutoff at x,theta) dx dtheta db da domega dq dalpha dbeta`. + +The integration ranges are the main proof's fixed compact boxes, including the full angular circle and positive q interval. This displayed coefficient is conditional on the main proof establishing the deterministic lifetime and count-once mark. The coarea algebra supplies the power `ell^(4/5)` because `r^8/(partial_r Delta_r)` tends to `r^4/(5L0)` and `r^4=ell^(4/5)L_r^(-4/5)`. The corresponding cumulative exponent is `9/5`, with coefficient `(5/9)` times the density coefficient. This concerns the restricted compact sector, not the leading density of all bars. + +## 6. C7 control is enough for the radial derivative + +Here is the precise sufficient setup. Fix the compact jet parameters other than r, including a hard eigenvalue bounded away from zero. Suppose `r -> F_r` is continuously differentiable with values in spatial `C7`, and both `||F_r||C7` and `||partial_r F_r||C7` are uniformly bounded. The Gaussian regression above has this property on a fixed `C7` residual ball: the mean is linear in the finite targets, which are polynomial in r with smooth coefficients on the chosen parameter box. Compact invertibility of the jet covariance controls the mean and its r derivative. + +On a sufficiently small fixed rectangle, the transverse derivative remains uniformly negative. The implicit ridge `h(r,s)` has uniformly bounded spatial derivatives through order six. Its direct construction uses `F_y` in spatial `C6`. The ridge value `g_r(s)=F_r(s,h(r,s))` has bounded spatial `C7` norm because + +`partial_s g_r=(F_r)_s(s,h(r,s))`. + +The right side is spatial `C6`, so this identity gives the seventh spatial derivative without needing a seventh derivative of h. The parameter envelope identity is + +`partial_r g_r(s)=(partial_r F_r)(s,h(r,s))`. + +It is spatial `C6` with a uniform bound. These assertions follow by the implicit-function chain rule and induction; the denominator in each implicit derivative is the uniformly nonzero transverse Hessian. There is no division by the small soft curvature. + +Let `T_r(s)` be the exact degree-five Taylor polynomial of `g_r` at zero, whose coefficients are the ridge identities above. Write `R_r(s)=g_r(s)-T_r(s)`. Taylor's integral formula, also applied to `partial_r g_r`, yields uniformly + +`|partial_s^j R_r(s)| <= C |s|^(6-j)` for `0<=j<=3`, + +`|partial_s^j partial_r R_r(s)| <= C |s|^(6-j)` + +where the second bound is used for `0<=j<=2`. Differentiation of the Taylor coefficients is legitimate because of the established mixed regularity. On a fixed compact xi interval define + +`E_r(xi)=r^(-5) R_r(r xi)`. + +Then `||E_r||C2(xi)<=Cr`. Moreover + +`partial_r E_r=-5r^(-6)R_r(rxi)+r^(-5)(partial_r R_r(rxi)+xi partial_s R_r(rxi))`, + +and the preceding bounds, differentiated in xi up to order two, give `||partial_r E_r||C2(xi)<=C`. + +Consequently the scaled ridge converges in `C2`, and its r derivative stays bounded in `C2`. On a fixed compact polynomial sector with simple, uniformly separated roots, the implicit critical roots have bounded r derivatives. The lifetime between the selected birth and selected critical saddle then has `L_r=L0+O(r)` and bounded `partial_r L_r`. The main proof must establish which root gives the actual global elder death; these estimates alone do not establish that selection. + +## 7. Pitfalls and remaining owner work + +- Keep the raw Hessian measure and full-circle positive-Q5 convention together. Adding an antipodal half to that convention undercounts. +- Keep both hard-Hessian factors: one comes from Rice and the other from the eigenframe Jacobian. Their product tends to `a^2`, not `|a|`. +- Keep all mixed jets in the Gaussian density, including the nonlinear limiting values of the pure fourth and fifth derivatives. +- Use a fixed regression-residual cutoff to make the conditioning measure fixed in r. If changing that definition, recheck the lifetime pushforward and boundary contribution. +- The simple-root bounds and actual curved global trap remain the main author's responsibility. This document does not establish a full restricted-sector theorem by itself. +- No numerical onset, all-sector matching, full next correction, Poisson law, higher homology, Lean verification, scientific acceptance, or provider-distinct review follows from these calculations. + +This finite contribution ends with the two coordinate transformations and regularity ledger delivered. There is no continuing source or runtime reservation. diff --git a/frontiers/nongaussian_smoothing_admission_20261017/FOLLOWON_SOURCE_INDEX.json b/frontiers/nongaussian_smoothing_admission_20261017/FOLLOWON_SOURCE_INDEX.json new file mode 100644 index 000000000..fa21d3326 --- /dev/null +++ b/frontiers/nongaussian_smoothing_admission_20261017/FOLLOWON_SOURCE_INDEX.json @@ -0,0 +1,373 @@ +{ + "schema": "public-followon-source-proposal-index-v1", + "state": "PROPOSED_SOURCE_COMPOSITION_REQUIRES_DISTINCT_REVIEW_NOT_ADOPTED", + "preparer_role": "SOURCEAUTHOR_PREPARER; prior focused binding-review author, not new independent reviewer", + "organizational_independence_credit": 0, + "native_cut": { + "math_main": "2b43d27eebce765cf30090c03e74714c72431114", + "math_tree": "62f0920bdb94be6e19661641077b70e554010169", + "main_reference": "46835abc657ef90deeaf9d83a47f912274b3ccfd" + }, + "original_exact_prose": [ + { + "unit": "NG", + "path": "frontiers/nongaussian_smoothing_admission_20261017/PROOF.md", + "role": "full original accepted proof", + "bytes": 38656, + "sha256": "322195ff6693f17047d9ec60c52f8178f5f3a215c4f4cba42dd342ff59dbc116", + "git_blob_sha1": "9df1a2c4ac9ec668b90729cc31c9195685a4b7c7" + }, + { + "unit": "NG", + "path": "frontiers/nongaussian_smoothing_admission_20261017/REVIEW.md", + "role": "full actual nonauthor review", + "bytes": 38360, + "sha256": "2dfe06601aa07bf0fa54923461f8367ba3018d30f1d6f43ff37a64a81d3b8adb", + "git_blob_sha1": "4af49a9319f86cf0d24248deb0ca0fbba66be404" + }, + { + "unit": "NG", + "path": "frontiers/nongaussian_smoothing_admission_20261017/imports/FINITE_SMOOTH_INCIDENCE_RECONSTRUCTION.md", + "role": "earlier accepted GAD reviewer mathematical appendix; NEW exact dependency publication, not previously landed49", + "bytes": 5032, + "sha256": "b1b928ce32a6a39b464cb9f1c6dd82f61b95d2bb095459d9b02f301168bd7fb2", + "git_blob_sha1": "66a784fbd6ed16411d97afb5fda5da20641b0910" + }, + { + "unit": "A4", + "path": "frontiers/gaussian_a4_elder_sector_20261017/PROOF.md", + "role": "full original accepted restricted-sector proof", + "bytes": 32065, + "sha256": "abed6f94da080a33458b78f1e60ae92ad90083323af41d241de05939d4377ff0", + "git_blob_sha1": "86371a58bdb12d86158c79dcc0c91da01c7c3e5b" + }, + { + "unit": "A4", + "path": "frontiers/gaussian_a4_elder_sector_20261017/REVIEW.md", + "role": "full actual nonauthor review", + "bytes": 33042, + "sha256": "1f87a54d883e8611e72c9c396c2f0e77a5c872546e8ef9242b7437448ac27fea", + "git_blob_sha1": "47f8584aae2b27ad37cdb71a4eceb496c5049ed6" + }, + { + "unit": "A4", + "path": "frontiers/gaussian_a4_elder_sector_20261017/EVERYWHERE_MARK_EXTENSION.md", + "role": "accepted reviewer reconstruction of everywhere Borel observable join", + "bytes": 3442, + "sha256": "3f5586edf75aa92ce1a17bd44dcbfb3116a5918722d341c8272b3e5dc4276164", + "git_blob_sha1": "7b60528b0c8117aa4bd9bbfd07880bb9d086bc8d" + }, + { + "unit": "A4", + "path": "frontiers/gaussian_a4_elder_sector_20261017/coauthor/DERIVATION.md", + "role": "focused analytic COAUTHOR source, not independent review or standalone theorem acceptance", + "bytes": 14969, + "sha256": "d667ca3541fc7187abbc6757fb73f1c27784e4aefc2dc42f215bdee86d786946", + "git_blob_sha1": "ae7c81553eb877b0707d644aa007a0ca13f780b5" + }, + { + "unit": "BIND", + "path": "frontiers/original_finite_l_gaussian_binding_20261017/PROOF.md", + "role": "full original accepted finiteL ideal-law alignment proof", + "bytes": 22163, + "sha256": "ad9ab5afea28062dfc661d0d19b43f20e67cfd67cb6910d3cc6c3fb8b81398d8", + "git_blob_sha1": "e8bb07fe8fab94616f02bdf792c34e3b337e1cf6" + }, + { + "unit": "BIND", + "path": "frontiers/original_finite_l_gaussian_binding_20261017/REVIEW.md", + "role": "full actual integrated nonauthor review", + "bytes": 23795, + "sha256": "ea09be7dbe4baea3b141e75ce563b5adb8752bb42ff16fc65735ed44974ce096", + "git_blob_sha1": "95c0d25a1caa49f95c8bcd3a2d3f131d460da890" + }, + { + "unit": "BIND", + "path": "frontiers/original_finite_l_gaussian_binding_20261017/focused-tensor-normalization/FOCUSED_TENSOR_NORMALIZATION_REVIEW.md", + "role": "actual focused tensor-normalization review; preparer authored this earlier focused review, no new independence credit", + "bytes": 12199, + "sha256": "fe2207d69495c41f40cea47b5643f72208a0ef5f6a1fe75e4af5042e462eedfb", + "git_blob_sha1": "0d9baf6f872df93facff76ea85f86be8e89b43b4" + } + ], + "canonical_publications": { + "NG": [ + "https://github.com/d6g8k5htny-coder/main/issues/419#issuecomment-6242848584", + "https://github.com/d6g8k5htny-coder/main/issues/419#issuecomment-6242848784", + "https://github.com/d6g8k5htny-coder/main/issues/419#issuecomment-6242943453" + ], + "A4": [ + "https://github.com/d6g8k5htny-coder/main/issues/419#issuecomment-6242536232", + "https://github.com/d6g8k5htny-coder/main/issues/419#issuecomment-6242537990", + "https://github.com/d6g8k5htny-coder/main/issues/419#issuecomment-6242623075" + ], + "BIND": [ + "https://github.com/d6g8k5htny-coder/Math-/issues/193#issuecomment-6243294265", + "https://github.com/d6g8k5htny-coder/Math-/issues/193#issuecomment-6243298158", + "https://github.com/d6g8k5htny-coder/Math-/issues/193#issuecomment-6243304101" + ] + }, + "original_seals_reference_only": [ + { + "unit": "NG", + "role": "original author seal", + "original_artifact_basename": "AUTHOR_FREEZE_MANIFEST.json", + "bytes": 3062, + "sha256": "8afee5ec816f3c729f25b3e608e0e93ba58d51a8a4601fc5d4be23830052e96f", + "git_blob_sha1": "fbd5a2264ecd239e515ea52e7a2011f0deda878e", + "published_raw_metadata": false + }, + { + "unit": "NG", + "role": "original actual nonauthor review seal", + "original_artifact_basename": "NONAUTHOR_SMOOTHING_REVIEW_MANIFEST.json", + "bytes": 12780, + "sha256": "b860c65e46428109e6cda07e0a82080ed511626cdb2260b94c3d9e18bd01d16a", + "git_blob_sha1": "1c0dc2df1a982a64fba6c69e6a8033845b397514", + "published_raw_metadata": false + }, + { + "unit": "A4", + "role": "original author seal", + "original_artifact_basename": "AUTHOR_FREEZE_MANIFEST.json", + "bytes": 2651, + "sha256": "4ffbe56ef508c6e543c82e6319c2cde0aad603fcf49988920f6a3ee9bfc7ea8d", + "git_blob_sha1": "24b1d661b77e8956003025c1d77c5d121205a776", + "published_raw_metadata": false + }, + { + "unit": "A4", + "role": "original actual nonauthor review seal", + "original_artifact_basename": "NONAUTHOR_A4_REVIEW_MANIFEST.json", + "bytes": 10261, + "sha256": "693e466116b5e7b17b5543153829904ecf18ee634f20427f2c7f7f6bec41de66", + "git_blob_sha1": "d015ac94c48c170623b9ce8a4f91596228e0deaa", + "published_raw_metadata": false + }, + { + "unit": "BIND", + "role": "original author seal", + "original_artifact_basename": "AUTHOR_FREEZE_MANIFEST.json", + "bytes": 3308, + "sha256": "83361eeb0f74574464b77638005794ced810691309592879a2b5f3a3fb20f065", + "git_blob_sha1": "f6bf23e175062706cca7e93c1209777bab44d868", + "published_raw_metadata": false + }, + { + "unit": "BIND", + "role": "original actual nonauthor review seal", + "original_artifact_basename": "NONAUTHOR_BINDING_REVIEW_MANIFEST.json", + "bytes": 19171, + "sha256": "b2f744006cc37a4b91b2ea20965175029e08bf7b5b45c5402466874e4f8d6ec1", + "git_blob_sha1": "fde082634a894f34cc4c84b7a0d50a6bf67b8da3", + "published_raw_metadata": false + }, + { + "unit": "A4", + "role": "focused COAUTHOR seal", + "original_artifact_basename": "COAUTHOR_FREEZE_MANIFEST.json", + "bytes": 943, + "sha256": "21f134a3839eb83958510119228413374d0c93a7b71112bc121cfa31bd6d4858", + "git_blob_sha1": "7d852d1ef9ace652fa8c1358b6a70b01a751c621", + "published_raw_metadata": false + }, + { + "unit": "BIND", + "role": "focused nonauthor review seal", + "original_artifact_basename": "FOCUSED_TENSOR_NORMALIZATION_REVIEW_MANIFEST.json", + "bytes": 2823, + "sha256": "046c47ecfcc131ef8908a434708f4a849ea18e6f9733026432cc750a62dfafbc", + "git_blob_sha1": "e14a23426461e4eca6522193b536992b2fe481bb", + "published_raw_metadata": false + } + ], + "existing_dependencies_reference_only": [ + { + "unit": "GAD", + "repository": "d6g8k5htny-coder/Math-", + "commit": "2b43d27eebce765cf30090c03e74714c72431114", + "path": "frontiers/gaussian_short_bar_theory_20261017/smooth_gaussian_manifold_admission/PROOF.md", + "bytes": 40231, + "sha256": "8d992e194f0a898e563c4c3d05b8079cde292f9e87d7ff51030ad12ef5faf319", + "git_blob_sha1": "d61fd11a826856baecb4cf4cd8a76ad5c25350dd", + "action": "EXACT_EXISTING_SOURCE_REFERENCE_NO_COPY" + }, + { + "unit": "GAD", + "repository": "d6g8k5htny-coder/Math-", + "commit": "2b43d27eebce765cf30090c03e74714c72431114", + "path": "frontiers/gaussian_short_bar_theory_20261017/smooth_gaussian_manifold_admission/REVIEW.md", + "bytes": 38035, + "sha256": "6fedc43ba34408f23f8934166b3898e2cd19175c57e61e37227c85f3ca5d4c70", + "git_blob_sha1": "554994ce28ea6b9e63450aa12301e5f220ec2e40", + "action": "EXACT_EXISTING_SOURCE_REFERENCE_NO_COPY" + }, + { + "unit": "HF", + "repository": "d6g8k5htny-coder/main", + "commit": "89657c7c48289378949dc59e7fbe6dc1a55d9c62", + "path": "experiments/universality/higher_homology_fold/PROOF.md", + "bytes": 46174, + "sha256": "87f43e6ed94c5c63c385ecf177f68212afa5db54b24062fa008f081f47a78999", + "git_blob_sha1": "8e18a7900e7d6308649239cc300a49a7c07c2f27", + "action": "EXACT_EXISTING_OPEN_PR331_SOURCE_REFERENCE_NO_COPY", + "import_scope": "NG uses deterministic sections1–7; old stationary-Gaussian analytic adapter is not inherited; D1 source/body amendment remains separate" + }, + { + "unit": "HF", + "repository": "d6g8k5htny-coder/main", + "commit": "89657c7c48289378949dc59e7fbe6dc1a55d9c62", + "path": "experiments/universality/higher_homology_fold/REVIEW.md", + "bytes": 5443, + "sha256": "4a24a4fe3e6bdd6e380ce1436f0b07eed587a03e01dc3ad6c59c59c4fbfba5fb", + "git_blob_sha1": "22fd94c885cf97c332a3f8bc5a7e6f805a5146c2", + "action": "EXACT_EXISTING_OPEN_PR331_SOURCE_REFERENCE_NO_COPY", + "import_scope": "NG uses deterministic sections1–7; old stationary-Gaussian analytic adapter is not inherited; D1 source/body amendment remains separate" + }, + { + "unit": "ORIGINAL_FINITE_L", + "canonical_public_comment": "https://github.com/d6g8k5htny-coder/Math-/issues/193#issuecomment-6242350213", + "original_artifact_basename": "independent-derivation.md", + "bytes": 24027, + "sha256": "f38c085ab5b15368e6989a1fd6298490f3b671f5a6c3a01bcdc81b516022bc33", + "git_blob_sha1": "d537bade335adb32050bd5b9470fec5a66c947e4", + "action": "REFERENCE_ORIGINAL_EXACT_PUBLIC_SOURCE_NO_DUPLICATE", + "scope": "Binding proof restates exact tensors/formula; original global proof passages are not retroactively accepted" + }, + { + "unit": "MATH240", + "canonical_public_pr": "https://github.com/d6g8k5htny-coder/Math-/pull/240", + "original_artifact_basename": "PROOF.md", + "bytes": 51848, + "sha256": "35b9c5943cf35919f2046ba194945ba759be654edf5d5ad77ba17b1f5c5377fd", + "git_blob_sha1": "16a1db0658918057ff741df888c13702763d9919", + "current_math_exact_paths": [ + "frontiers/elder_cusp_parity_20261002/PROOF.md" + ], + "action": "REFERENCE_NARROW_PRIOR_SOURCE_NO_COPY", + "scope": "A4 author used lines44–95/ridge notation/equation0.1, reconstructed independently; no whole Math240 theorem import" + } + ], + "accepted_scope_and_imports": { + "NG": { + "positive": "Fixed explicit Gaussian G plus independent Borel smooth W with all polynomial smooth-norm moments; H0 d>=1, and orientable d>=2 finite Hp per fixed field; asymptotic density−1/3/cumulative2/3 and iid whole-copy marked process within full review scope", + "limits": "No arbitrary nonGaussian law, uniform degenerating-noise/metric/dimension statement, within-field PPP/higher factorial theorem, numerical window, Lean or science", + "imports": "Existing exact GAD and finite smooth incidence supplement; exact HF deterministic sections1–7 only; named smooth/Morse/measure/Gaussian/classical imports retained" + }, + "A4": { + "positive": "One positive fixed compact regular marked Gaussian A4 ordinaryH0 elder sector on a fixed oriented flat two-torus; density C_A4 ell^(4/5)+O(ell), cumulative~(5/9)C_A4 t^(9/5), as ell,t decrease to0", + "limits": "Not whole-bar leading law/next whole-density correction; no negligibility of omitted discriminants, Q5=0/second-soft/large-jet/residual sectors, other dimensions/curvature/Hk/PPP, numerical or science", + "imports": "Exact GAD literal kernels/all-Borel one-point incidence plus shared finite-incidence supplement; named classical local/global Morse and differential facts; coauthor source is authorship, mark extension is reviewed reconstruction" + }, + "BIND": { + "positive": "Exact fixed finiteL original Gaussian law admitted through GAD; specified finite Borel ordinaryH0 density representative has C_alg ell^(−1/3) asymptotic and integrated(3/2)C_alg t^(2/3) per physical volume", + "limits": "Alternative ideal-law identification, not retrospective validation of every original argument; no M31/rate/onset/sampler/grid/count/confidence certificate, evaluated interval, all-five-gap closure or science", + "imports": "Exact GAD theorem and stated classical imports; original exact formula/tensors; numerical METHODV2 and P3/P7 advisory are not mathematical premises" + } + }, + "not_in_prior49": "The5032B GAD incidence appendix is newly published exactly once here, not falsely described as alreadylanded", + "local_only_exclusions": [ + "third-party PDFs/full books and rendered PNGs", + "raw old input/output/result/seal metadata containing private filesystem or fileID locators", + "programs,controls,failed attempts,raw streams,binaries,oleans,package/cache trees", + "prior complete recovery ZIPs; no public archive", + "old METHODV1/V2/proxy mathematics as proof premises; exact versions remain canonical history" + ], + "closure_limit": "Full new proofs and actual reviews plus named mathematical supplements are present. This is not physical closure of all historical manifests, transitive classical books or a portable runtime. Historical paths/PENDING labels remain literal and are interpreted by new metadata only.", + "relative_link_policy": "No frozen prose Markdown relative href needs repair; A4 extension colocated and binding focused subdirectory retained. Historical inline input filenames are routed by this index, not rewritten.", + "no_theorem_or_status_promotion": true, + "source_adoption": false, + "owner_adoption": false, + "scientific_effect": "NONE", + "supplementary_technical_input": { + "canonical_claim": "https://github.com/d6g8k5htny-coder/Math-/issues/193#issuecomment-6243336796", + "actual_actor": "Anthropic Claude01Cz", + "scope": "binding moment/regression/He/normalization only", + "status": "PENDING_NO_COMPLETED_RESULT_OR_ADOPTION_OR_INDEPENDENCE_CREDIT", + "root_observed_cut": "2026-10-17T23:49:56Z", + "claim_created_at": "2026-10-17T23:49:21Z", + "future_label_correction": "Body says23:52contendercheck, which was future at Root23:49cut and is not used as completed verification", + "prior_completed_binding_acceptance": "PRESERVED_UNCHANGED", + "local_preparation_waits_for_result": false + }, + "current_runtime_boundary": "Currentbase separately contains the landed C261 source-smooth workflow/history. This candidate adds no runtime/CI/Lean registration and makes no new claim about that separate execution.", + "existing_C261_workflow_scope": { + "path": ".github/workflows/source-smooth-realization.yml", + "commit": "2b43d27eebce765cf30090c03e74714c72431114", + "bytes": 35002, + "sha256": "313952a5a8f6f87e27584be0b8d4064edecb41a5f655d0bbf3a83d4e76b21515", + "git_blob_sha1": "ecc9d40d9f12c068975fd2574032536edb3fc085", + "observation": "FULL579literalworkflowlines and invoked audit/lakefile read; static source/coverage only, no execution", + "trigger": { + "event": "pull_request", + "types": [ + "opened" + ], + "paths": [ + "frontiers/source_smooth_realization_20261011/**", + ".github/workflows/source-smooth-realization.yml" + ], + "push_dispatch_schedule": false + }, + "candidate14_paths_match_filters": false, + "historical_guard": { + "deadline": "2026-10-16T02:50:00Z", + "expected_base": "319f9c45f66dbcce4a71b9c2f3e0fb62a666bfec", + "expected_head_ref": "codex/c261a-hosted-successor-20261016-root01a11cbf", + "original_attempt_only": 1, + "meaning": "Past fixed route plus external observer record; not renewed by landing or this prose packet" + }, + "runtime_config": { + "runner": "ubuntu-latest", + "job_minutes": 30, + "threads": 2, + "Lean": "4.34.1", + "Lake": "5.0.0-src+5045d00", + "locked_packages": 9, + "mathlib": "d13f23b723b8a846827a245b89c10fc7d3f11612", + "actions": { + "checkout": "3d3c42e5aac5ba805825da76410c181273ba90b1", + "lean-action": "f061402b660e0c34644504b324e830f2991d4865", + "upload-artifact": "043fb46d1a93c77aae656e7c1c64a875d1fc6a0a" + }, + "lean_action_inputs": "explicit source-package directory; auto-config/build/test/lint false; mathlib-cache true; github-cache false" + }, + "invoked_scope": { + "baseline_modules": [ + "GaussianSeriesFoundation", + "SpectralWeights", + "SourceFourierSeries", + "SourceC4", + "SourceGaussianLaw", + "SourceLattice", + "SourceKernel", + "SourceCovariance", + "SourceTorus", + "SourceTorusC4" + ], + "proof_modules": [ + "SourceSmooth", + "SourceSmoothCarrier" + ], + "control_modules": [ + "SourceSmoothControls" + ], + "standard_kernel_modules": [ + "SourceSmooth", + "SourceSmoothCarrier", + "SourceSmoothControls" + ], + "audit": "SourceSmoothAxiomAudit.lean: actual13module-origin declaration census,16requirednames and4private-carrier guards,standard axioms only", + "phase_caps_seconds": [ + 540, + 90, + 90, + 90, + 90 + ], + "kernel90scope": "three kernel calls combined90seconds" + }, + "custody": "always final identity/outcome capture and artifact upload;21packageleaves+workflow exact22-addition guard", + "coverage_limit": "This C261 source scope does not compile/audit NG,A4,binding prose or475elder modules. Dated475core63 receipt retains its scope; no blanket current-Math-only63 claim. No new runtime/admission claim." + } +} diff --git a/frontiers/nongaussian_smoothing_admission_20261017/PROOF.md b/frontiers/nongaussian_smoothing_admission_20261017/PROOF.md new file mode 100644 index 000000000..9df1a2c4a --- /dev/null +++ b/frontiers/nongaussian_smoothing_admission_20261017/PROOF.md @@ -0,0 +1,768 @@ +# Independent Gaussian smoothing admits nonGaussian fold universality + +Conventional mathematical author proposal, 17 October 2026. Authors: +OpenAI/Codex nongaussian-smoothing-admission author and Root01a11cbf +(mechanism coauthor). This is a new reconstruction, pending a separate +exact-source nonauthor review. The independently frozen A4 source is not +changed. All contributors are source-exposed OpenAI agents; organizational +independence is zero. No scientific promotion is asserted. + +## 1. Actual product law and conclusions + +Let X be a fixed nonempty connected compact smooth manifold without +boundary, with fixed smooth Riemannian metric g and dimension d>=1. +Write V=vol_g(X). Fix an actual centered Gaussian smooth random field + + G=sum_{j>=1} a_j Z_j phi_j, (1) + +where Z_j are the original independent standard real Gaussians, a_j>0, +the real span of the deterministic phi_j is dense in C-infinity(X), and + + sum_j a_j ||phi_j||_Cq < infinity for every integer q>=0. (2) + +Use increasing smooth norms from covariant derivatives or a finite +coordinate atlas. Fix independently a Borel C-infinity(X)-valued random +field W such that + + E ||W||_Cq^m < infinity for every integer q>=0 and every m>0. (3) + +W need not be centered, Gaussian, stationary, bounded, or have any +exponential moment. A deterministic smooth W is allowed. The actual +law under consideration is the independent product construction + + F=G+W. (4) + +The nonzero Gaussian component and independence are substantive +hypotheses. Constants are for this fixed decomposition, law, metric and +dimension. There is no assertion uniform as the Gaussian component +vanishes or its spectra/geometry degenerate. The theorem does not say +every smooth nonGaussian law has this decomposition. + +For ordinary persistence over any fixed coefficient field K, exclude +essential intervals. If d>=2 and X is orientable, let p=0,...,d-1. If +X is not assumed orientable, the assertion here is p=0 only. In d=1 +it is also p=0 only. For each of these degrees the expected finite +superlevel-Hp lifetime measure per volume has the explicitly specified +finite nonnegative Borel density representative nu_p in (12), and + + nu_p(ell) ~ C_p ell^(-1/3), + E N_p((0,t])/V ~ (3/2) C_p t^(2/3), ell,t ->0+, (5) + +with 0=2 the leading coefficients, and the limiting marked +measure below, are independent of the chosen field K: the local +topological incidence is +/-1. Global pairings and lower-order terms +can depend on K. No equality C_p=C_(d-1-p) is inferred. The centered +Gaussian transverse sign symmetry used in a prior special model is +unavailable for a general noncentered mixture. + +There is also an independent-whole-copy corollary, proved in Section 12. +For independent copies of this one fixed F, the rare short-bar processes, +with n t_n^(2/3)->lambda>0, converge to an explicitly marked Poisson +random measure with degree, scaled lifetime, birth height, cubic mark, +midpoint and direction marks. This is not a within-one-field spatial +Poisson theorem or a factorial-moment assertion. + +## 2. Exact consumed theorems and what is new here + +GAD is the smooth dense-support Gaussian admission proof, 40231 bytes, +739 lines, SHA256 +8d992e194f0a898e563c4c3d05b8079cde292f9e87d7ff51030ad12ef5faf319. +Its full proof, full 38035-byte accepted scoped review, and finite smooth +incidence appendix were read in this session. It proves for (1)-(2): +the actual smooth law and all norm moments; finite original coefficient +minors from dense smooth span; the actual full-field Gaussian conditional +version at every literal target; all-Borel finite-incidence Rice; off- +contact conditional genericity using finite residual coefficient ranks; +and the deterministic H0 elder-transfer geometry. Its hypotheses concern +G, not F. Applying its Gaussian conditional covariance formula directly +to F would be incorrect. + +HF is Main331, Higher homology: attaching obstructions and isolated-fold +unit incidence, exact 46174-byte/911-line proof SHA256 +87f43e6ed94c5c63c385ecf177f68212afa5db54b24062fa008f081f47a78999, +with exact 5443-byte review SHA256 +4a24a4fe3e6bdd6e380ce1436f0b07eed587a03e01dc3ad6c59c59c4fbfba5fb. +Both were read completely. HF Sections 1-7 prove a deterministic +isolated regular-fold unit incidence and Borel once-counting on a closed +oriented manifold: a same-field globally C4-convergent continuation with +a single cubic contact, nonsingular transverse Hessian, exact two pins, +and all other critical points Morse with distinct heights away from +the contact gives the exact finite Hp interval. Mixed transverse +inertia changes its degree. This is the topology imported here. + +HF's probabilistic Sections 8-11 admit only its named stationary Gaussian +torus law. They do not already prove (4), arbitrary geometry, or (3)'s +polynomial-only envelope. The source also retains a documentary README +audit qualification; this author does not modify that source or claim +an integration/metadata resolution. The new work here supplies the +actual product-law kernels, every analytic interface, manifold binding, +nonGaussian joint-jet coefficient, and bounded-cluster/iid-copy passage. + +The deterministic HF import uses the specific classical relative handle +boundary formula: one transverse attaching/belt intersection has +incidence +/-1 over every field. Its proof constructs the local product +flow, excludes remote connecting trajectories by a fixed-annulus positive +height cost, proves no intervening critical value, and handles C4 by +aligned smoothing and regular-level isotopy. Its arbitrary adjacent-cell +counterexamples remain exclusions when these hypotheses are absent. +No new handle theorem or whole classical-book proof is claimed here. + +## 3. Smooth product law, ranks and literal kernels + +By GAD, G is a Borel smooth random element and all its smooth norms have +all finite moments. Thus F in (4) is Borel smooth, with all finite norm +moments by Minkowski and (3). Independence lets us use the actual product +space P_W tensor P_G throughout. Nothing requires W's raw coefficient +representation, a density for W, or recovery of W from F. + +If L is any of the finite independent jet/observation maps used below, +the Gaussian covariance Gamma_L=Cov(LG) is positive. This follows from +dense-span bump-polynomial interpolation and a finite subset of the +original Gaussian coefficient columns. Write c_L=Cov(G,LG), + + R_L=G-c_L Gamma_L^-1 LG, + G_{L,y}=R_L+c_L Gamma_L^-1 y. (7) + +R_L is a smooth Gaussian field independent of LG. Its finite observation +is zero. Formula (7) is the actual Gaussian conditional law at every y, +not an assertion that the pin event has positive probability. It is +jointly Borel in sites, frames and targets in local frame charts. + +Let pi_L^G denote the positive Gaussian density of LG. Then LF has the +everywhere positive convolution density + + p_L^F(v)=E_W pi_L^G(v-LW). (8) + +It is finite and smooth in v; every derivative of a nondegenerate +Gaussian density is bounded, so differentiation under this mixture is +justified without any W exponential moment. Dependence on bounded +site/frame parameters is continuous by dominated convergence and the +proven covariance floors. The literal whole-field conditional version +of F at LF=v is + + Q^F_(L,v)(A)= [p_L^F(v)]^-1 + E_W[ pi_L^G(v-LW) + P_G{ W+R_L+c_L Gamma_L^-1(v-LW) in A } ]. (9) + +For each v this is a probability kernel; positivity of (8) makes its +denominator legitimate. Its measurability and normalization follow from +the explicit regression and Gaussian convolution. This is not a Gaussian +law in general. Its posterior distribution of W is weighted by the +displayed likelihood. Replacing it with the unweighted W law would give +the wrong conditional field. + +To verify its literal version, condition first on a fixed smooth w, +split G into a finite original coefficient block plus its independent +smooth tail, and use the block's invertible observation minor. Its tail +receives the Bayes density pi_block(v-Lw-Ltail)/pi_G(v-Lw). +The remaining finite block has its affine Gaussian conditional law. +Finite Gaussian completion of squares identifies this with (7) at +every target v-Lw. Integrating the original W law multiplies by +pi_G(v-Lw), giving (9). Equality of finite evaluation characteristic +functions, followed by countable evaluation generation of the smooth +Borel sigma-field as in GAD, identifies the whole-field law, including +all Borel marks. A continuous-mark extension or almost-every-target +argument has not been substituted for this calculation. + +Formula (9) names the literal whole-field version associated with the +fixed product decomposition (4). No uniqueness of all-target versions +across different decompositions is needed or asserted. The finite-jet +density is the intrinsic continuous convolution density of F, and all +counting identities use this explicitly matched version. + +On compact separated site/frame domains, all needed Gaussian covariance +floors and smooth regression coefficient bounds are uniform. W is a +translation of the field, not of those Gaussian minors, so their ranks +are unchanged. Full finite F-jet laws are mixtures of shifted full-rank +Gaussian laws; they have strictly positive densities on every full +independent jet coordinate space. + +## 4. COUNT_p and original genericity, conditioning on W first + +For each fixed smooth w, the actual field w+G has the same independent +finite Gaussian coefficient directions as G. Fix a smooth Gaussian tail. +At a root, solve the gradient equations for an invertible finite +Gaussian coefficient minor. The incidence projection from site and +remaining coefficients has absolute Jacobian |det Hess(w+G)| divided +by that minor determinant. Smooth finite-dimensional Sard proves Morse +behavior for almost every Gaussian block. For distinct sites the joint +first jets of G have full rank; literal height disintegration rules +out ties between critical heights. A countable separated-domain/minor +cover gives these statements for the whole field, for each fixed w. +Integrating W proves that F is Morse with distinct critical values +almost surely. Compactness makes its critical set finite. + +The same finite smooth incidence proof, with an arbitrary nonnegative +Borel whole-field mark, gives the two-point identity at fixed w. +Integrating W and using (8)-(9) gives the extended actual identity + + E sum_{M!=S, grad F(M)=grad F(S)=0} h(M,S,F) + = integral_{M!=S} p^F_(grad_M,grad_S)(0,0) + E_QF[ |det H_M det H_S| h(M,S,F) ] dvol(M)dvol(S). (10) + +The mark was Borel before tail integration; no continuity of a global +pairing is required. Each physical Hessian determinant appears once. +Expected total critical count is finite: the one-point Gaussian +gradient density is bounded uniformly, and its conditional Hessian +moment is bounded by C(1+||w||_C2)^(d), with constants enlarged if +necessary; (3) integrates this bound. Ties need only separated local +integrals and countable exhaustion, not a collision-uniform floor. + +For oriented X,d>=2, use HF Section 7's Borel selector sigma_p(F,M,S) +of the unique ordinary finite Hp interval with these exact ordered +endpoints, extended by zero off the Morse/distinct locus. Its local +constancy is for the continued critical labels in a small Morse +neighborhood, as proved using aligned regular-level isotopies. Each +critical point has at most one endpoint role. For H0 in the other +cases use GAD's intrinsic rational strict-superlevel connection +selector and standard H0 elder linear algebra. Both count each finite +interval once and exclude essential intervals. Thus + + N_p((0,t])=sum_{M!=S, gradients zero} + sigma_p(F,M,S) 1{00, M is a birth critical point, not generally a maximum. +An index predicate alone would count candidates rather than bars. + +Disintegrate both heights in (10) using the same literal kernels. +If O_MS=(F(M),grad F(M),F(S),grad F(S)), define for ell>0 + + nu_p(ell)=V^-1 integral_{M!=S} integral_R + p^F_O(b,0,b-ell,0) + E_QF[ |det H_M det H_S| sigma_p(F,M,S)] db dvol(M)dvol(S). (12) + +The redundant inertia conditions are included in sigma_p. This +nonnegative Borel representative is the actual expected finite-bar +density by Tonelli. Its finiteness is proved below. We have obtained +COUNT_p for the actual product law, not merely a conditional Gaussian +candidate count. + +## 5. Exact near observations and the same product-space coupling + +Choose a fixed short-geodesic radius r_0>0. At x,u in SX let +M=exp_x(-ru/2), S=exp_x(ru/2). Parallel-transport a perpendicular +orthonormal frame along the geodesic. The literal 2d+2 observation +rows, applied to any smooth scalar field f, are + + U_r f=((f(M)+f(S))/2, (f(S)-f(M))/r, + (f_u(S)-f_u(M))/r, + (6/r^2)[f_u(M)+f_u(S)-2(f(S)-f(M))/r], + ((f_vi(M)+f_vi(S))/2,(f_vi(S)-f_vi(M))/r)_i). + +Their target for the two exact critical heights b,b-kr^3 is + + v_r=(b-kr^3/2,-kr^2,0,12k,(0,0)_i), k>0, + v_0=(b,0,0,12k,(0,0)_i). (13) + +The r=0 extension is U_0=(f,f_u,f_uu,f_uuu,(f_vi,f_uvi)_i). +Integral finite-difference identities give ||U_r f||<=C||f||_C3 +and ||(U_r-U_0)f||<=Cr||f||_C4, uniformly on SX. The exact change +from the two first jets O_MS has absolute determinant +12 r^(-(d+3)), unchanged by the law of f. In particular + + p_O^F(b,0,b-kr^3,0)=12 r^(-(d+3)) p^F_(U_r)(v_r). (14) + +This is the specific height/observation binding used in (12). + +Use G's c_r=Cov(G,U_r G), Gamma_r=Cov(U_r G), and R_r from (7). +GAD's contact rank and compactness give common covariance floors +for 0<=r<=r_0 after decreasing r_0. Smooth covariance and the row +estimates imply convergence of c_r Gamma_r^-1 in each global smooth +norm, and R_r->R_0 globally C4 almost surely and in every finite +Lp norm. On the ORIGINAL product W,G define + + F_{r,w}=w+R_r+c_r Gamma_r^-1(v_r-U_r w). (15) + +For each fixed w, this is the exact full-field conditional law +given U_r F=v_r within that component. It has the forced pins +exactly and converges to F_{0,w} globally C4, using the same G. +For every q>=3, + + ||F_{r,w}||_Cq <= C_q(||G||_Cq+||w||_Cq+1+|b|+k), (16) + +with harmless equivalent norms and enlarged constants. The bound +includes the deterministic conditional mean, which can grow with w. +There is no Gaussian assertion about F or inverse posterior normalizer +in (16). All unnormalized expectations below use this fixed product +space and the weight pi_r^G(v_r-U_r w). + +## 6. Literal conditional off-contact genericity and contact isolation + +Fix x,u, b in R and k>0, and ANY smooth w, and condition G on +U_0 G=v_0-U_0 w. The Gaussian finite block whose joint rows are +U_0 and one or two remote first jets has full rank on each compact +domain separated from x and, for two sites, from its diagonal. +Surjectivity of (U_0,J_remote) implies that J_remote is surjective +on the kernel of U_0: lift (0,y). This rank is independent of the +target and of w. + +Under this literal target, the finite Gaussian block/tail description +is its positive Bayes-weighted smooth tail and an affine Gaussian on +the finite coefficient kernel. In an orthonormal basis of that kernel +the remaining coordinates are independent standard Gaussians. Adding +the deterministic smooth w and conditional mean does not change the +surjective residual gradient directions. Apply the finite incidence/ +Sard argument in those coordinates. For every fixed w and target, +all other critical points of F_{0,w} are Morse almost surely. Residual +first-jet density excludes critical height b; residual two-first-jet +density excludes ties among other critical heights. Countably many +separated domains give the assertions on X minus {x}. A different +finite block can be used on each domain; all describe the same kernel. + +The transverse A_0 of F_{0,w} has a shifted full-rank Gaussian density +on all independent symmetric entries. It is nonsingular almost surely, +and every inertia cone has strictly positive weighted determinant-square +mass at every fixed w,b,k,x,u. No centeredness of that matrix is assumed. +At the forced contact the gradient and axial Hessian column vanish, +the axial cubic derivative is 12k, and A_0 is nonsingular. The implicit +transverse critical ridge therefore has reduced derivative +6k s^2+o(s^2)>0 at sufficiently small nonzero s. The contact is isolated. +All other critical points lie in a compact complement, so they are +finite and have distinct heights with positive gaps from b. + +This proves the required conditional genericity separately from original +Morse genericity. It holds for each fixed w and contact target, with +probability one in the Gaussian residual. Fubini gives the same statement +for the original W,G product at each fixed target. There is no uncountable +simultaneous event over every w,target,frame, and none is needed. + +## 7. CONTACT_p and actual SELECT_p on the manifold + +Put j=d-p-1. For H0 this means j=d-1. In the parallel endpoint frames +define + + W_(p,r)=|det H_M det H_S| + 1{index H_M=j+1,index H_S=j}, + V_(p,r)=W_(p,r)/r^2. + +The exact gradient pins give the Hessian axial-column average zero. +As in GAD, the physical Hessian blocks and polynomial identity are + + H_i=[r alpha_i, r beta_i^T; r beta_i,A_i], + det H_i/r=alpha_i det A_i-r beta_i^T adj(A_i)beta_i, + alpha_M->-6k, alpha_S->6k, A_i->A_0. (17) + +The r^-1/2 axial congruence gives the inertia. For each fixed w and +almost every residual G, + + V_(p,r) -> V_(p,0) + =36 k^2 (det A_0)^2 1{index A_0=j}, + V_(p,r)<=C(1+||F_{r,w}||_C3)^(2d). (18) + +If A_0 is nonsingular with a different inertia the indicator eventually +fails; at a singular A_0 the determinant vanishes in the limit regardless +of indicator oscillations. The polynomial identity avoids any inverse +curvature moment. Every index cone has finite positive contact mass. + +For oriented X,d>=2, (15) and Section 6 meet HF's deterministic theorem. +To bind its physical chart to the manifold observations, use Fermi +coordinates (s,t) around the short geodesic: the axial coordinate is +unit speed and the transverse coordinate vectors on the axis are the +parallel frame. The two pins are literally (+-r/2,0). At the contact, +the zero gradient identifies coordinate and covariant Hessians, and +the axial third derivative is the geodesic derivative 12k. The fields +converge globally C4, hence in that fixed chart. These are HF's exact +geometric inputs on the stated closed oriented manifold. Its unit +incidence gives the actual finite Hp interval at the pinned endpoints, +and sigma_p=1 eventually whenever V_(p,0)>0. There is no invocation +of negative-definite H0 concavity for a mixed transverse block. + +For H0 without the orientability/d>=2 premise, GAD's deterministic +asymmetric elder enclosure applies on the negative transverse cone, +with the older outgoing arm. The side wall is absent in d=1. It gives +the same exact finite H0 pair and eventual selector. Both mechanisms +also give global eventual Morse/distinct behavior and the complete +critical list: the local two roots and the finitely many old branches. + +Define the UNNORMALIZED contact/selected kernels + + Z_(p,r)=E_W[ pi_r^G(v_r-U_r W) E_G V_(p,r) ], + Y_(p,r)=E_W[ pi_r^G(v_r-U_r W) E_G(V_(p,r) sigma_p) ]. (19) + +For each fixed b,k the bound (16),(18), pi_r^G<=C, and all moments +of W,G supply an integrable product-space bound. The weight converges +pointwise, and on V_(p,0)>0 the selector is eventually one; on V_(p,0)=0 +the weighted determinant vanishes. Dominated convergence proves + + Z_(p,r)->Z_(p,0)>0, + Y_(p,r)->Z_(p,0), + E_W[pi_r^G E_G(V_(p,r)(1-sigma_p))]->0. (20) + +These are actual same-product-space limits, not limits under a +fictionally r-independent posterior of W. Also p^F_(U_r)(v_r)-> +p^F_(U_0)(v_0)>0 by bounded Gaussian densities and continuity. Division +by that FIXED-target positive normalizer gives CONTACT_p and SELECT_p +for the canonical F kernels (9), if normalized formulations are wanted. +No global lower bound for that normalizer has been used. + +## 8. ENVELOPE_p with only polynomial moments of W + +Here is the all-mark estimate which cannot be obtained by asserting +Gaussianity of F. Put B=1+|b|+k and N=1+||W||_Cq with q>=4 fixed. +The exact first/fourth target entries in (13) give uniformly + + |v_r|>=c_1 B-C, |U_r W|<=C N. + +Absorb the harmless constant in C_1 N. The Gaussian covariance floor +and upper bound give + + pi_r^G(v_r-U_r W) + <=C exp[-c(c_1 B-C_1 N)_+^2]. (21) + +From (16),(18) and the Gaussian norm moments, with m=2d, + + E_G V_(p,r)<=C(B+N)^m. + +Take epsilon=c_1/(2C_1)>0. On N<=epsilon B, the exponential in (21) +is at most exp(-c'B^2) and B+N<=C B. On N>epsilon B, +B+N<=C_epsilon N; discard the density exponential and use, for any +fixed M>0, + + E[N^m 1{N>epsilon B}] + <=(epsilon B)^(-M) E N^(m+M). + +The latter moment is finite by (3), even though E exp(cN^2) may be +infinite for every c>0. Thus for EVERY fixed M>0, + + 0<=12Y_(p,r)<=12Z_(p,r) + <=C_M [B^m exp(-c'B^2)+B^(-M)] + <=H_M(b,k):=C'_M(1+|b|+k)^(-M). (22) + +Constants are uniform in 04/3, for example M=6. Directly integrating b gives + + integral_R integral_0^infinity k^(-2/3) H_M(b,k) dk db1 for b; at large k, b +integration gives order (1+k)^(1-M), and the remaining power is +k^(1/3-M), integrable precisely when M>4/3. The sphere bundle has +finite volume. This proves ENVELOPE_p without an inverse posterior +normalizer, Gaussian tails for W, or an exponential W moment. + +The same argument works with untyped determinant weight, so it also +dominates candidate and bounded extra global marks. There is no +assertion of a uniform selector radius or rate over all W realizations. + +## 9. FAR_p, finite density versions, and the curved Jacobian + +On dist(M,S)>=r_0, G's joint two-first-jet covariance O_MS has a +positive floor on the compact domain. For 0(exp_x(-h/2),exp_x(h/2)) has natural volume Jacobian J(x,h), +smooth, positive, J(x,0)=1 and J(x,-h)=J(x,h). Its derivative at +h=0 is (a,b)->(a-b/2,a+b/2), of absolute determinant one. Therefore + + dvol(M)dvol(S)=J(x,ru) r^(d-1) dr dvol(x)d sigma_x(u), + J(x,ru)=1+O(r^2) (24) + +uniformly on the compact sphere bundle. Curved geometry has not been +silently treated as flat. The near fixed-ell substitution k=ell/r^3 +in (12), with (14),(18), gives + + ell^(1/3) nu_p,near(ell) + =1/(3V) integral_{SX,R,(0,infinity)} + 1{k>=ell/r_0^3} k^(-2/3) + J(x,(ell/k)^(1/3)u) 12Y_(p,(ell/k)^(1/3)) + db dk dvol(x)d sigma_x(u). (25) + +The radial ledger is (d-1)+3-(d+3)+2=1: spatial polar measure, +height change, literal observation density, and physical determinant +orders. No exponent was inferred merely from the cubic height gap. + +The envelope (22) makes (25) finite at each ell>0, and uniformly +bounded after multiplication by ell^(1/3), with local bounds on +positive-lifetime intervals. Together with (23), (12) is the finite +Borel density representative asserted in Section 1. Equations +(20),(22),(24) and dominated convergence give + + C_p=4/V integral_{SX,R,(0,infinity)} + k^(-2/3) Z_(p,0)(b,k,x,u) + db dk dvol(x)d sigma_x(u). (26) + +Finiteness follows from the envelope. Positivity follows because for +every smooth w and every fixed b,k>0, the Gaussian likelihood is +positive and the shifted full Gaussian transverse matrix has positive +determinant-square mass on the required open inertia cone. Integration +over W and any nonempty finite mark window is strictly positive. +The cumulative factor 3/2 follows by integration of ell^(-1/3). +No density continuity or O(1) near remainder is asserted. + +## 10. The full actual joint-jet invariant and geometric half + +Let D,V_u,T_u,A_u have the meanings in Section 1. The full joint +law of their independent coordinates, with optional birth height F, +is a mixture of shifted nondegenerate Gaussian densities. Use exactly +the literal mixture kernels (9) for all their pins and disintegrations. +In particular the value at D=0,V_u=0 is specified, not an arbitrary +version of a conditional expectation. Normalized moments at this +fixed pin are finite: the Gaussian likelihood is bounded, its residual +jet moment is polynomial in W's finite jet, (3) integrates it, and +the fixed pin density is strictly positive. + +At the contact, U_0 F=(b,0,0,12k,0,...) pins F=b,D=0,V_u=0,T_u=12k. +By (18), integrate the actual birth height b in (26), using Tonelli +and the same full joint convolution law. This yields + + C_p=144/V integral_SX integral_0^infinity k^(4/3) + p^F_(D,V_u,T_u)(0,0,12k) + E_F[(det A_u)^2 1{A_u has p positive eigenvalues} + | D=0,V_u=0,T_u=12k] + dk dvol(x)d sigma_x(u). (27) + +The change t=12k gives 144*12^(-7/3)=12^(-1/3), hence + + C_p=12^(-1/3)/V integral_SX p^F_(D,V_u)(0,0) + E_F[(T_u)_+^(4/3) (det A_u)^2 1{A_u has p positives} + | D=0,V_u=0] dvol(x)d sigma_x(u). (28) + +No covariance factorization, centered conditional matrix, or independent +odd/even jets have been assumed. This is the actual joint-jet invariant +of F. Its conditional distribution is generally nonGaussian. + +For the alternative absolute-value formula (6), replace u by -u. +The same field has T_-u=-T_u. The transverse subspace, determinant +square and inertia cone are unchanged. The coordinates of D,V_u +transform linearly with absolute determinant one and preserve their +zero pin. Their actual density at zero and conditional law transform +accordingly. Ordinary sphere area is antipodally invariant. Therefore +the integrals of T_+^(4/3) and (-T)_+^(4/3), with these weights, are +equal AFTER integration over the full sphere. Their sum is the +absolute-value integrand. Averaging gives precisely (6). + +This half is geometric averaging over opposite directions. It does +not require F or W to be centered, the law to be invariant under +F->-F, stationarity, reflection symmetry of the law, or any jet +independence. It is not a second division of ordered birth/death +pairs. Each actual interval was already counted once in (11). + +For oriented X the local coefficient is independent of K by HF's +unit incidence; (27)-(28) also show no K in its limit. But changing +A to -A is a DIFFERENT transformation from changing u to -u. +There is no general sign symmetry of the transverse conditional +mixture. Already a scalar A distributed as N(m,1), m>0, has +E[A^2 1{A>0}]>E[A^2 1{A<0}], since for a>0 its two density terms +satisfy exp(-(a-m)^2/2)>exp(-(a+m)^2/2). This rejects importing the +old centered-Gaussian cone argument into this proof. No unproved +degree-duality identity for (4) is asserted. + +## 11. Genuine nonGaussian examples and exact boundaries + +Let H be an admitted smooth Gaussian series independent of G, and +epsilon!=0. Then + + W(x)=epsilon[H(x)^4-E H(x)^4] (29) + +is an actual smooth Borel field independent of G satisfying (3). +The deterministic function E H(x)^4=3 Var(H(x))^2 is smooth by covariance +summability. Leibniz gives ||H^4||_Cq<=C_q||H||_Cq^4; all required +moments follow from H's norm moments and Holder. At any point with +sigma^2=Var H(x)>0, the centered Gaussian even moments give + + E W(x)^3=epsilon^3[10395-3*3*105+2*27] sigma^12 + =9504 epsilon^3 sigma^12 !=0. + +Independence and centering make the third cumulant of F(x) the same, +since G(x)'s third cumulant is zero and mixed centered terms vanish. +Consequently F is genuinely nonGaussian. Moreover for every c>0, +E exp(c |W(x)|)=infinity: for sufficiently large |H(x)| the absolute +value is bounded below by a positive multiple of |H(x)|^4, whose +positive quartic exponent dominates the Gaussian negative quadratic +density exponent. This admitted example has NO positive pointwise +absolute exponential moment. Such a moment cannot be a hidden +hypothesis of the envelope. The simpler centered-square example +epsilon[H^2-EH^2] is admitted too; its third cumulant is +8 epsilon^3 sigma^6, and it has no exponential-square moment. + +If W is an independent random spatial CONSTANT S with all moments, +then every field realization G+S has exactly G's lifetime list and +pairing in every degree: all critical heights shift by S and all +height differences and sublevel inclusions simply reparameterize. +Formula (6) agrees exactly, because all the derivative jets in that +formula are unchanged. The birth integral (27) removes the random +height convolution. This is an exact consistency control, not a +claim that nonconstant W leaves the coefficient unchanged. + +Independence cannot be omitted: W=-G has every smooth norm moment +and is smooth, but F=0 has no nondegenerate contact law or positive +finite-bar coefficient. This violates independence and so is outside +the theorem. Likewise the Gaussian component must satisfy the +stated actual dense-span ranks; merely writing a degenerate finite +Gaussian noise does not suffice. The constants and asymptotic window +need not remain controlled as that component approaches zero. + +The prior covariance-matched random-amplitude counterexamples still +forbid universalizing over every smooth nonGaussian field with a +chosen covariance. Actual independent Gaussian smoothing, its +nonvanishing finite ranks, product moments, conditional selection +and envelope are the mechanism proved here. Covariance alone does +not assert that decomposition or those interfaces. + +## 12. Bounded-cluster estimate and degree-marked iid-copy Poisson law + +For the oriented manifold case sum p=0,...,d-1 and put N_t^all equal +to the number of all ordinary finite bars of lifetime <=t. No finite +degree-d bar can be killed by a (d+1)-handle. In the H0-only case +use just N_t^all=N_(0,t), and the degree mark is the singleton 0. +Each original field is Morse with a finite critical/lifetime list +almost surely; all its positive lifetimes are then bounded away +from zero for that realization. This fact alone would not prove +a Poisson limit, so we establish the required weighted cluster bound. + +At fixed contact marks and fixed smooth w, for almost every Gaussian +residual with V_(p,0)>0, Section 7 gives the actual pinned bar and +the complete critical list for all small r. The two local critical +points' unique endpoint roles are consumed by this one bar. Every +other bar in the stated degree set therefore uses two OLD critical +branches. Their limiting heights are finitely many and pairwise +distinct, so every possible nonzero gap between them has a positive +lower bound. This remains true if their global pairing changes. +There are no additional critical branches near the forced contact. + +Fix 00, and set r=(ta/k)^(1/3). The pinned lifetime +is exactly ta<=Tt. Eventually it is the ONLY bar with lifetime <=Tt. +On V_(p,0)=0 the determinant weight tends to zero regardless of other +counts. Insert the bounded Borel mark 1{N_(Tt)^all>=2} into the +once-counted identity (11)-(12). After scaling ell=ta, the near +integrand divided by t^(2/3) is bounded by a finite constant times + + a^(-1/3) k^(-2/3) H_M(b,k), 0=2}]=o(t^(2/3)). (30) + +This is not a second factorial moment estimate. There is no extra +unbounded witness count, four-site rank or same-field mixing input. + +For near bars use the intrinsic midpoint x and ordered unit direction +u from birth M to death S, as in Section 5. Give each the marks + + (p,a=ell/t,b=F(M),k=ell/dist(M,S)^3,x,u). + +The unique short geodesic makes these jointly Borel. Bars outside +the near band may be omitted from this marked process; their expected +number on a<=T is O(t), so their probability of affecting any finite +window tends to zero also in the independent-copy limit below. +If desired they can receive a cemetery geometry mark with the same +vanishing effect. There is no need to choose a global geodesic at +cut-locus pairs. + +The first-moment contact passage with a bounded continuous test of +these marks, using (20),(22),(24), gives the limiting one-copy mean +measure after division by t^(2/3): + + d eta_p=4 a^(-1/3) k^(-2/3) Z_(p,0)(b,k,x,u) + da db dk dvol_g(x)d sigma_x(u), a,k>0. (31) + +It is finite on the COMPLETE mark windows 00 +with n t_n^(2/3)->lambda in (0,infinity), and superpose their marked +short-bar measures. For a nonnegative bounded continuous process test +f supported in a finite scaled-lifetime window, write Xi_t for the +single-copy measure and h=1-exp(-f). On a field with at most one +relevant point, + + 1-exp(-)=. + +In general their difference in absolute value is at most +N_(Tt)^all 1{N_(Tt)^all>=2}, plus the negligible far count if those +bars are omitted. Equations (30)-(31) give + + E[1-exp(-)] + =t^(2/3) sum_p integral (1-exp(-f)) d eta_p+o(t^(2/3)). + +Independence then yields the Laplace limit + + E exp(-) + -> exp[-lambda sum_p integral (1-exp(-f)) d eta_p]. (32) + +This is the Poisson random measure with intensity +lambda sum_p delta_p tensor eta_p on the marked space. The usual +Laplace-functional characterization on its locally compact separable +mark space gives convergence in law; (22) also controls the complete +finite-lifetime windows including a=0 and unbounded b,k. Limiting +degree populations are independent Poisson restrictions. The actual +degree populations within one finite field need not be independent. +The convergence is for independent whole copies, with fixed geometry +and fixed Gaussian component, not expanding volume, same-field spatial +Poisson, or factorial-moment convergence. + +## 13. Evidence, imports and remaining boundaries + +All five interfaces COUNT_p, CONTACT_p, SELECT_p, ENVELOPE_p and FAR_p +have been derived for (4), under the exact deterministic topology +imports stated in Section 2. The only probabilistic moments required +of W in the proof are polynomial moments of its actual global smooth +norms. Literal kernel normalization is used for definition and +fixed-target interpretation; all estimates and limiting passages +use unnormalized likelihoods on the original product space. + +Classical mathematical imports remaining are finite-dimensional +Gaussian regression/uniqueness, smooth finite-dimensional Sard and +smooth incidence counting via local inverses and ordinary change +of variables, Taylor/implicit-function calculus and short-geodesic +coordinates, the named Morse handle/relative-incidence and finite +persistence package in HF, Tonelli, Minkowski/Holder, dominated +convergence, and the elementary Laplace characterization of a Poisson +random measure. GAD and HF state and derive their precise uses; +no new infinite-dimensional transversality or Kac-Rice theorem is +silently imported. The full copies and read extents are recorded in +INPUT_MANIFEST.json. No new full primary-book reading is claimed. + +The new exact finite controls check Bayes weighting, shifted conditional +means, the target coercivity/tail split, exponent integrability, the +nonGaussian third cumulant, constant-shift invariance, cone-symmetry +refusal, antipodal averaging and radial/Poisson coefficient factors. +These finite algebra controls support the proof; they are not an +infinite-field sample, numerical integral, asymptotic experiment or +independent mathematical review. No old controls are rerun. + +This author unit does not qualify any numerical sampler, grid, empirical +50-model trial, coefficient enclosure, Lean theorem, same-field Poisson +or all-order factorial result. It does not remove higher-singularity, +degenerate-noise, or arbitrary-law boundaries. No source branch, Drive +record, external message or scientific-status register is changed. +The larger research objective remains ACTIVE / UNMET. A separate +exact-source nonauthor review is still required for this proposal. diff --git a/frontiers/nongaussian_smoothing_admission_20261017/README.md b/frontiers/nongaussian_smoothing_admission_20261017/README.md new file mode 100644 index 000000000..3838da251 --- /dev/null +++ b/frontiers/nongaussian_smoothing_admission_20261017/README.md @@ -0,0 +1,15 @@ +# Independent nonGaussian smoothing admission + +This is a source-author publication proposal carrying the full unchanged [proof](PROOF.md) and [actual integrated nonauthor review](REVIEW.md). Historical proposal/PENDING wording stays unchanged beside the actual later scoped acceptance. This index is not a new proof, review, source adoption, owner decision or scientific status register. + +Fixed explicit Gaussian G plus independent Borel smooth W with all polynomial smooth-norm moments; H0 d>=1, and orientable d>=2 finite Hp per fixed field; asymptotic density−1/3/cumulative2/3 and iid whole-copy marked process within full review scope. These are the stated small-argument results and must not be treated as exact finite-scale power laws. + +No arbitrary nonGaussian law, uniform degenerating-noise/metric/dimension statement, within-field PPP/higher factorial theorem, numerical window, Lean or science. The full original proof and review control the exact hypotheses and limitations. All remaining original numerical/scientific obligations retain their evidence-specific dispositions; this packet does not close all five groups or the113retained failures. + +Existing exact GAD and finite smooth incidence supplement; exact HF deterministic sections1–7 only; named smooth/Morse/measure/Gaussian/classical imports retained. + +The shared exact [finite smooth incidence appendix](imports/FINITE_SMOOTH_INCIDENCE_RECONSTRUCTION.md) is earlier GAD reviewer mathematics newly published here once; it was absent from the prior49 files. A4 and the finite-L binding reference this same appendix. + +The shared [source/dependency/exclusion index](../nongaussian_smoothing_admission_20261017/FOLLOWON_SOURCE_INDEX.json) records exact original prose identities, canonical source/review publications, immutable existing Math475/GAD and Main331/HF references, neutral historical-seal hashes and scope boundaries. Already landed GAD proofs/reviews and existing HF sources are referenced, not duplicated or replaced. Original sources and old seals were not edited. + +No Lean source, package, manifest, workflow, runtime, compiler target or execution receipt is registered here. Full books/PDFs/PNGs, private filesystem/fileID metadata, programs, failed attempts and recovery archives remain local; this is not a public ZIP, hermetic proof closure or portable execution package. New routing metadata has source-author provenance. The completed author/review sources here are source-exposed OpenAI, organizational/provider-distinct/blind/human credit0. Supplementary Claude input on the separate finite-L binding (moment/He/normalization only) is PENDING at Math193/6243336796; it supplies no completed review, adoption or independence credit. Existing C261 workflow/source history at this base is separate; this candidate changes none of it. Scientific effect NONE. diff --git a/frontiers/nongaussian_smoothing_admission_20261017/REVIEW.md b/frontiers/nongaussian_smoothing_admission_20261017/REVIEW.md new file mode 100644 index 000000000..4af49a931 --- /dev/null +++ b/frontiers/nongaussian_smoothing_admission_20261017/REVIEW.md @@ -0,0 +1,657 @@ +# Nonauthor review: independent Gaussian smoothing and finite-bar universality + +**Verdict: ACCEPT_SCOPED. No must-fix finding or required source amendment was +found for the exact frozen proposal, under its explicit product-law, +geometry/degree, fixed-coefficient-field, and classical-import hypotheses.** +The review accepts the actual density and independent-whole-copy marked +Poisson statements below. It does not accept a covariance-only nonGaussian +extension, Gaussian parity for a nonGaussian mixture, factorial moments, or +a spatial Poisson limit within one field. + +Reviewer: OpenAI/Codex `/root/reverse_localization_nonauthor_review_20261017`, +17 October 2026. Root and the distinct nonGaussian-smoothing author unit are +mechanism/proof coauthors. The earlier A4 focused contributor made no new +contribution to this unit and receives no authorship or review credit here. +The present reviewer did not author the new smoothing mechanism or proof. +The reviewer did previously review GAD and author its consumed source-bound +review and finite smooth incidence appendix. That exposure/self-authorship +is disclosed; it is not fresh independent validation of that earlier work. +All parties are source-exposed OpenAI agents. Organizational, provider-distinct, +blind and human independence credits are zero. No helper was used in this unit. + +## 1. Exact object, source reads and claim + +The exact new `PROOF.md` has 38656 UTF-8 bytes, 768 lines, SHA256 +`322195ff6693f17047d9ec60c52f8178f5f3a215c4f4cba42dd342ff59dbc116`. +The author freeze has 3062 bytes, SHA256 +`8afee5ec816f3c729f25b3e608e0e93ba58d51a8a4601fc5d4be23830052e96f`, +and binds 17 files totaling 246569 bytes, including five exact consumed inputs. +The reviewer read the entire 768-line frozen proof in untruncated ranges. +There was no final-source read or theorem verdict on an unstable draft. All +freeze-bound files and the freeze identity were freshly checked. Hashes bind +the object; they are not substitutes for the reconstruction in this report. + +The dependencies actually inspected are: + +| Source | Bytes | SHA256 and extent | +| --- | ---: | --- | +| GAD Gaussian admission proof | 40231 | `8d992e194f0a898e563c4c3d05b8079cde292f9e87d7ff51030ad12ef5faf319`; full 739-line proof read earlier in this conversation, exact bytes reverified here | +| GAD source-bound review | 38035 | `6fedc43ba34408f23f8934166b3898e2cd19175c57e61e37227c85f3ca5d4c70`; full prior own review, exposure disclosed | +| GAD finite smooth incidence appendix | 5032 | `b1b928ce32a6a39b464cb9f1c6dd82f61b95d2bb095459d9b02f301168bd7fb2`; full prior own appendix, freshly read in the preceding task | +| Main331 higher-homology proof HF | 46174 | `87f43e6ed94c5c63c385ecf177f68212afa5db54b24062fa008f081f47a78999`; fresh full 911-line read | +| Main331 source-bound review | 5443 | `4a24a4fe3e6bdd6e380ce1436f0b07eed587a03e01dc3ad6c59c59c4fbfba5fb`; fresh full read | +| Main331 seven native comments | 97253 | `c80ccd83a632831dd94ca50d155fbde013ed4aefb5c743f377f69f244203d971`; all seven complete bodies decoded and read, original JSON retained | + +The new author result, input manifest, freeze and preserved authoring-history +record were read. The author control programs were not rerun or treated as +independent review evidence. Retained local inputs and read extents are bound +by this review's manifest. The earlier independently frozen A4 proof is a +separate object and is not a mathematical premise of this theorem. + +The actual law is F=G+W on each fixed nonempty connected compact smooth +boundaryless Riemannian manifold X of dimension d>=1 and volume V>0. The +centered Gaussian field G is an actual series of the original independent +standard real Gaussian coefficients, with strictly positive deterministic +weights, dense real smooth span, and summability of every weighted smooth +norm. W is an independent Borel smooth random element with every finite +positive polynomial moment of every smooth norm. W may be deterministic, +noncentered, nonstationary, nonGaussian, atomic in its law, and without a +positive exponential moment. The decomposition and independence are genuine +hypotheses. No representation theorem for every smooth nonGaussian law, +uniformity as G vanishes, or uniformity over laws/metrics/dimensions is claimed. + +For H0 the result covers every such X, including d=1 and nonorientable X. +For the higher-degree assertion X must be orientable and d>=2; choose an +orientation and fix any coefficient field K. The admitted degrees are +p=0,...,d-1. The observables are actual finite ordinary superlevel Hp bars, +with essential intervals excluded. The exact Borel density version (12) +satisfies + + nu_p(ell)~C_p ell^(-1/3), + E N_p((0,t])/V~(3/2) C_p t^(2/3), + 0lambda, +the stated marked process converges to a Poisson random measure of intensity +lambda sum_p delta_p tensor eta_p. The raw spatial measure in eta_p is dvol, +not probability-normalized dvol/V. + +## 2. What Main331 supplies, and the documentary hold it does not resolve + +Main331 §§1–7 prove a deterministic unit-fold theorem and Borel finite-bar +selectors on a closed oriented manifold. The required continuation is one +physical family converging globally in C4, with one isolated cubic contact, +nonsingular transverse Hessian, exact two critical pins, and all other +critical points Morse with distinct heights away from the contact value. +Those hypotheses supply the complete critical list and no intervening event. +Its §§8–11 prove an analytic/probabilistic specialization only for its named +stationary torus Gaussian law SG. The new proof does not inherit that +probabilistic admission, its parity factorization, stationarity cancellation, +or its cone symmetry. It independently supplies all analytic F=G+W contracts. + +All seven native comments were read, not only a PASS label. The IDs are +6138705550, 6139276556, 6139465647, 6154502316, 6154541570, 6154559578 and +6154562912. Claude's delivered 6139276556 gives PASS_MATHEMATICS_SCOPED under +the stated imports and AMEND_DOCUMENTARY_SCOPED for D1: the repository review +and PR body disagree about a corrected-README metadata PASS. Later comments +plan an amendment, debate a line-number attribution, correct that route and +accept the corrected route. They do not establish that an actual D1 amendment +was made to Main331. The retained source still preserves that documentary +qualification. This review leaves D1 UNAMENDED and does not promote route +acceptance to completed source repair. + +That documentary mismatch does not supply a missing mathematical premise to +this new theorem. The consumed deterministic statement and construction are +available in the actual exact source, with their own hypotheses and named +classical imports. This review reconstructs those uses directly. It does not +certify an unobserved metadata audit, edit Main331, or adopt its unrelated +Gaussian probabilistic specialization as a theorem about W. + +## 3. Actual product law and canonical kernels at every target + +GAD gives an actual Borel C-infinity Gaussian field G with all finite smooth +norm moments. Adding the independent Borel W preserves smoothness, Borelness +and all polynomial smooth-norm moments. No density or coefficient realization +for W, or recovery of W from the observed F, is required. Work on the original +product probability space and then push forward to the field. + +For each named independent finite observation map L, dense smooth span and +bump-polynomial jet interpolation make Gamma_L=Cov(LG) strictly positive. +Some finite subset of the ORIGINAL independent Gaussian coefficient columns +already spans its target. The covariance field c_L is smooth and its regression +map is bounded into each needed smooth norm on compact separated parameter +domains. The Gaussian residual R_L=G-c_L Gamma_L^-1 LG is independent of LG, +with zero observations. Thus for every y the actual Gaussian conditional law +is R_L+c_L Gamma_L^-1 y. + +At fixed smooth w the field F has observation shift Lw. Integrating over the +original W law gives the actual everywhere positive continuous convolution +density + + p_L^F(v)=E_W pi_L^G(v-LW). + +Every target derivative is justified by bounded derivatives of the finite +Gaussian density. Continuity in compact site/frame parameters follows from +pointwise continuity for every smooth w and a uniform Gaussian density bound. +There is no need for exponential W moments or a nonsingular covariance of W. + +The literal full-field version is + + Q_(L,v)^F(A)=1/p_L^F(v) E_W[ + pi_L^G(v-LW) P_G{W+R_L+c_L Gamma_L^-1(v-LW) in A}]. + +This is a nonGaussian mixture in general. Its denominator is positive at +every fixed target and its numerator is a probability mixture with the stated +normalization. Its W posterior is likelihood-weighted. Using the unweighted +W law as a conditional posterior would be false. All later estimates retain +the numerator, so no lower bound for this normalizer over growing marks is +assumed. + +The all-target version is actually tied to coefficient incidence. At fixed w, +choose a finite original Gaussian block with an invertible observation minor, +fix its independent smooth Gaussian tail, and change finite coefficients on +the exact fiber. The Gaussian tail receives its correct Bayes likelihood. +Completing the finite Gaussian square for each finite field evaluation vector +identifies its mean and covariance at EVERY target v-Lw. Multiply by +pi_L^G(v-Lw), integrate the original W law, and divide by p_L^F(v). This is the +same kernel displayed above at every v. Countably many values generate the +smooth Borel sigma-field through measurable derivative difference quotients, +countable seminorm suprema and a countable smooth dense set. Equality of the +finite evaluation laws therefore identifies the whole field, including all +nonlocal Borel barcode marks. No continuity of the selector on a null fiber, +almost-every-target extrapolation, or Gaussian regression applied to F's +covariance is used. + +Noninjectivity of the product decomposition causes no problem: the augmented +law is conditioned and then pushed forward. The source fixes that literal +version for its chosen decomposition; it need not prove uniqueness of every +null-target kernel across all possible decompositions of an identical field +law. Its finite-jet continuous density remains intrinsic to the law of F. + +## 4. COUNT_p: genericity, arbitrary Borel marks and coefficient fields + +For every fixed smooth w, the original G directions are still full-rank +finite Gaussian coefficient directions in w+G. Smooth equal-dimensional +incidence/Sard proves Morse genericity before invoking root counting. Pair +first-jet rank makes the two critical heights conditionally nondegenerate, +so ties have zero expected count on each separated compact pair domain. +Countable covers and annuli exhaust all distinct sites. Integrating W proves +original F is Morse with distinct heights almost surely. Compactness gives +finitely many critical points. One-point Rice also gives finite expected +critical count: its Gaussian observation density is bounded and the conditional +Hessian moment grows only polynomially in ||w||C2, which W integrates. + +The finite smooth incidence map has one absolute physical Hessian determinant +per root. For the two-site formula it has the product, divided by the selected +coefficient-minor determinant; finite Gaussian observation change of variables +cancels that denominator. The nonnegative whole-field mark is Borel before +integration over either the Gaussian tail or W. Local inverse charts, disjoint +Borel minor partitions, coefficient exhaustion, ordinary change of variables +and Tonelli yield the extended identity. Integrating W gives exactly the +convolution kernels above. There is no hidden extra determinant or global +continuity hypothesis on the mark. + +For p>0, HF §7 supplies the actual Borel barcode selector, not an H0 elder +rule renamed. In a neighborhood of a Morse/distinct field, all critical +branches/order persist by the implicit function theorem and a compact gradient +floor. Align their finitely many critical heights, then flow the finitely +many chosen regular-level boundaries along that aligned path. The sampled +filtration inclusions are simultaneously carried by ambient maps, so its +finite persistence module over this fixed K is unchanged. Pairings of labeled +branches are locally constant. A countable cover of the separable good locus +therefore gives a Borel graph of selected endpoint pairs. Each Morse handle +changes one rank and each critical point has at most one endpoint role. + +This construction works for each FIXED field K; it need not algorithmically +compute a boundary matrix or choose one Borel basis across every K. Different +Morse chambers may have different attaching maps and K-dependent global +pairings. The source does not conflate their local constancy with global +constancy. Essential intervals have no finite death and are omitted. For +H0 without the higher-degree geometric hypotheses, GAD's Borel intrinsic +strict-superlevel connection selector and H0 elder algebra suffice. + +The resulting once-count sum is the actual N_p. The index condition is already +forced by that selector, so the typed determinant used near the diagonal is +redundant on selected pairs, not an additional determinant weight. M in higher +degrees denotes a birth critical point and need not be a maximum. The exact +two-height disintegration gives (12) with the same specified kernels; Tonelli +proves it is an actual expected finite-bar measure. Finiteness of its chosen +density version is established by the later near/far bounds. + +## 5. Same product-space contact coupling and conditional genericity + +The literal U_r rows and target v_r are those in (13). Along a short geodesic +with parallel frames the divided differences are integral averages. Thus +|U_r f|<=C||f||C3 and |(U_r-U0)f|<=Cr||f||C4, with a common compact-frame +constant. The row change from the two first jets has determinant +12r^(-(d+3)). These are geometric linear identities and do not depend on the +law being Gaussian. They bind the actual nonGaussian density (12) to its +specified contact kernel. + +G's observation covariance has the uniform floor. On the ORIGINAL independent +W,G product use + + F_(r,w)=w+R_r+c_r Gamma_r^-1(v_r-U_r w). + +At fixed w this has exact critical/height pins and is the genuine Gaussian +component fiber of the product law. It converges to F_(0,w) globally in C4 +on the same G realization. Smooth covariance convergence and the averaged-row +bounds also give + + ||F_(r,w)||Cq<=C_q(||G||Cq+||w||Cq+1+|b|+k), q>=3. + +The conditional mean is allowed to grow with w. Treating it as a centered +Gaussian family with no w-dependent mean would be incorrect. The posterior +of W can depend on r; the proof never couples those normalized posteriors +as if they were identical. It couples the Gaussian components on the original +product and retains their likelihood weights. + +At the literal contact target, joint rank of (U0,remote first jets) in G +implies surjectivity of each remote map on ker U0 by lifting (0,y). On every +compact domain separated from the contact, finite original coefficient blocks +provide these ranks. Given a fixed smooth w and a fixed literal target, the +correct Bayes-weighted Gaussian tail remains smooth, and an orthonormal basis +of the finite coefficient kernel gives independent standard residual coordinates. +Adding w and the affine mean preserves those ranks. Finite residual Sard/Rice +then proves off-contact Morse behavior and excludes off-contact critical +height b and pairwise ties. Countably many separated domains may use different +blocks, all identified with the SAME canonical kernel. + +Every conclusion is for each fixed w,target with residual probability one. +The bad events are measurable via the countable compact incidence domains; +Fubini gives the needed product-space statement at each fixed target. This is +not an uncountable simultaneous genericity assertion. The transverse conditional +A is a shifted full-rank Gaussian matrix at each w,target, so is invertible +almost surely and has positive determinant-square mass on every open inertia +cone. Its mean need not vanish. + +The forced contact has zero gradient and axial Hessian column and axial cubic +12k>0. An implicit transverse ridge has reduced derivative 6ks^2+o(s^2), +positive away from zero nearby. This isolates the contact for every nonsingular +A, including mixed inertia. All other critical points then form a finite Morse +set in a compact complement, with distinct heights and positive gaps from b. +That is the complete conditional old critical list required by deterministic +selection. It was not inferred from unconditional Morse genericity. + +## 6. Physical Hessians, manifold chart and the unit interval + +Exact endpoint gradient pins give the physical block identity + + H_i=[r alpha_i,r beta_i^T;r beta_i,A_i], + det H_i/r=alpha_i det A_i-r beta_i^T adj(A_i)beta_i, + alpha_M->-6k, alpha_S->6k, A_i->A0. + +The identity is polynomial and valid for singular blocks. The axial congruence +diag(r^-1/2,I) gives the inertia. With j=d-p-1, the typed divided weight +converges to + + V_(p,0)=36k^2(det A0)^2 1{index A0=j}. + +This is exactly the cone of p positive transverse eigenvalues. If A0 is +nonsingular with another inertia, the type fails eventually. At a singular +limit the determinant vanishes despite any indicator oscillation. The bound +V_(p,r)<=C(1+||F_(r,w)||C3)^(2d) avoids inverse-curvature moments. + +HF's physical-chart hypotheses are genuinely met on the manifold. Fermi +coordinates around the fixed short geodesic put the pins at (+-r/2,0), with +unit-speed axial coordinate and the parallel transverse vectors on the axis. +At the contact, zero gradient identifies coordinate and covariant Hessians; +the axial third derivative is the geodesic third derivative. Global C4 +convergence implies convergence in that fixed chart. A global frame is not +required; finitely many local frame trivializations suffice. No flat-metric +approximation is used to invoke the topology. + +For orientable X,d>=2, the two endpoints have g=-F handle degrees p and p+1. +The exact splitting in HF reduces F locally to a cubic axial function plus +a quadratic form of fixed mixed transverse inertia. Its reduced derivative +is strictly convex, so the two pinned roots are the only local roots. The +product ascent field has one axial S-to-M trajectory: nonzero negative +transverse coordinates diverge at one time end and nonzero positive ones at +the other. The connection is transverse. Uniform comparison of this adapted +field with the physical gradient, plus a positive gradient floor on a fixed +annulus, makes any excursion cost a fixed positive height. The total kr^3 +height gap is eventually smaller. Hence no other global connecting trajectory +can cancel the local incidence. + +One transverse attaching/belt intersection gives relative boundary coefficient ++/-1, a unit over every K. With no intervening critical value, the finite +chain/triple-sequence argument forces the precise interval born at M and dying +at S. Earlier global attaching data do not change that pivot. HF's C4 bridge +uses close smooth approximation, aligned critical heights and regular-level +isotopy; its tolerance may depend on the shrinking gap. It transfers the +actual endpoints and finite filtration, not only their ordering. This is +why the reused unit-fold theorem applies to the C4 regression continuation +without a smooth unfolding parameter or Morse-Smale premise. + +For H0 on other manifolds/d=1 the negative transverse cone permits GAD's +asymmetric elder enclosure and older outgoing arm. The side face is absent +in d=1. No mixed-inertia concavity is asserted. Both mechanisms give the exact +selected pair and eventual global Morse/distinct behavior with the two new +roots plus finitely many old branches. On V_(p,0)>0 the actual sigma_p is +therefore eventually one; on V_(p,0)=0 its determinant-weighted defect vanishes. + +## 7. Unnormalized limits and the polynomial all-mark envelope + +The decisive kernels are + + Z_r=E_W[pi_r^G(v_r-U_r W) E_G V_r], + Y_r=E_W[pi_r^G(v_r-U_r W) E_G(V_r sigma_p)]. + +At fixed b,k, the Gaussian likelihood is uniformly bounded, the coupled +weight converges pointwise, and the norm bound is polynomial in ||W||,||G|| +and the fixed marks. All required product moments are finite. Dominated +convergence gives Z_r->Z0>0, Y_r->Z0 and weighted defect ->0. The actual +normalizer p_(Ur)^F(v_r) also converges to its positive fixed-target value. +Dividing there gives normalized CONTACT/SELECT if desired. There is no +uniform inverse-normalizer estimate over all marks, and none is needed. + +For all marks let B=1+|b|+k and N=1+||W||Cq, q>=4. The target's first and +fourth entries give |v_r|>=c1 B-C, while |U_r W|<=C N. The Gaussian covariance +floor/upper bound gives + + pi_r^G(v_r-U_r W)<=C exp[-c(c1 B-C1 N)_+^2], + E_G V_r<=C(B+N)^(2d). + +On N<=epsilon B, with epsilon=c1/(2C1), the integrand is bounded by a polynomial +in B times exp(-c'B^2). On N>epsilon B, discard that exponential and use + + E[N^(2d)1{N>epsilon B}] + <=(epsilon B)^(-M) E N^(2d+M). + +Thus for any fixed M>0 the UNNORMALIZED selected and candidate kernels obey + + 12Y_r<=12Z_r<=C_M(1+|b|+k)^(-M). + +This is the actual new envelope. Completing a square in a way that introduces +E exp(c||W||^2), or dividing by a small posterior density first, would destroy +the stated class and is not the proof given. + +The integrability threshold is correct and independent of dimension once the +needed W moment has been paid. For M>1, + + integral_R (1+|b|+k)^(-M) db=2/(M-1)*(1+k)^(1-M). + +After multiplying by k^-2/3, the large-k power is k^(1/3-M), integrable exactly +when M>4/3. The small-k power -2/3 is integrable. Choosing M=6 gives the common +all-mark envelope on finite-volume SX and the finite degree set. No all-mark +selector radius or deterministic convergence rate over W is required. The +same bound holds with untyped determinants and any additional bounded global +mark, which is crucial for the later cluster estimate. + +## 8. Far density, curved volume and the actual density representative + +On a separated pair domain, the Gaussian full first-jet covariance has a +compact positive floor. Fixed-w regression gives conditional C2/Hessian +moments polynomial in 1+|b|+||w||C2 for 0=ell/r0^3 and k^-2/3 times J*12Y_r. The polynomial envelope makes this finite +at every ell>0 and bounded after multiplication by ell^(1/3), with local +positive-lifetime bounds. The far term is also finite. This proves the +particular finite Borel representative claimed in (12); arbitrary +Radon–Nikodym null-set modifications need not have its pointwise asymptotic. + +At each fixed k>0 and b,x,u the contact limits apply. Integrating them by the +common all-mark envelope, then removing the O(1) far term after multiplication +by ell^(1/3), gives + + C_p=(4/V) integral k^-2/3 Z_(p,0)(b,k,x,u) db dk dvol d sigma. + +The likelihood and every shifted full Gaussian transverse cone mass are +strictly positive at each smooth w and fixed positive k. Their original W +average and any nonempty finite mark-window integral are positive. The same +envelope gives finiteness. Integrating ell^-1/3 gives the cumulative factor +3/2; no density continuity or O(1) near remainder is asserted. + +## 9. Full nonGaussian joint-jet coefficient and the two different symmetries + +The actual finite joint F-jet is a convolution of shifted nondegenerate +Gaussian densities. Its continuous density is positive everywhere. At the +fixed pin D=grad F=0, V_u=Hess F(u,.)=0, the normalized moments are finite: +the Gaussian likelihood is bounded, the residual jet moment is polynomial +in W's finite jet, and the fixed denominator is positive. This statement +uses the specified literal product kernel, not an arbitrary conditional +expectation version on a null event. + +At contact, U0 pins F=b,D=0,V=0,T=12k. Integrate b in the actual joint law +and use the limit V_(p,0)=36k^2(det A)^2 on its cone. The exact scalar in (27) +is 144, followed by t=12k, giving 144*12^(-7/3)=12^(-1/3). Hence + + C_p=12^(-1/3)/V integral_SX p_(D,V)^F(0,0) + E_F[(T_+)^(4/3)(det A)^2 1{A has p positives}|D=0,V=0] + dvol d sigma. + +This is a full nonGaussian joint moment. No Gaussian Schur-complement moment +of F's covariance, centered transverse matrix, or odd/even independence may +replace it. All W-induced correlations and shifts remain. + +For the same field, changing u to -u changes T to -T while leaving the +transverse Hessian and its inertia cone unchanged. The zero-pinned D,V +coordinate map has absolute determinant one. Sphere area is antipodally +invariant, so AFTER integration over directions the positive-part moment is +half the absolute moment. This gives exactly (6), including its scalar +1/(2*12^(1/3)*V). It does not use symmetry of F's law under F->-F, centeredness, +stationarity, or reflection invariance. It is not another division of ordered +endpoint pairs. In d=1 the empty determinant/cone equal one and ordinary S^0 +measure counts its two directions. + +Changing F to -F, in contrast, changes A to -A and exchanges the transverse +cones. Without that additional law symmetry, C_p=C_(d-1-p) is not inferred. +A shifted scalar Gaussian already has unequal determinant-square masses on +the positive and negative cones. This is a refusal of an invalid centered- +Gaussian inference, not a claim that one finite scalar example by itself is +a global manifold barcode counterexample. For every fixed K the local +incidence remains a unit, so the leading coefficient and eta_p do not depend +on K; finite-field global pairings and lower-order behavior still may. + +## 10. Genuine examples and exact decomposition boundaries + +For an independent admitted Gaussian H, W=epsilon(H^4-EH^4), epsilon!=0, +is a Borel smooth field with every polynomial smooth-norm moment. Leibniz +bounds its norm by a constant times ||H||Cq^4 plus a smooth deterministic +mean. Its pointwise third centered moment is + + (10395-3*3*105+2*27) epsilon^3 sigma^12 + =9504 epsilon^3 sigma^12 !=0. + +Independence and centering preserve this third cumulant after adding G, +so the actual F is nonGaussian. For every c>0, exp(c|W(x)|) has infinite +expectation: its positive quartic exponent eventually dominates the negative +quadratic Gaussian density exponent. This is an admitted example with no +positive pointwise absolute exponential moment, not merely no exponential- +square moment. The centered-square example has third cumulant 8 epsilon^3 +sigma^6 and also meets the polynomial assumptions. These calculations are +consistent with, and test the necessity of, the envelope's actual moment use. + +If W is an independent spatial constant S, each realization's finite +lifetimes and pairings are exactly those of G. The derivative jets in (6) +are unchanged, and the birth integral removes the height convolution, so +all C_p are unchanged. The full birth-height mark does shift/convolve; the +source does not claim its entire b-marked intensity is unchanged. This +separates lifetime invariance from marked-height invariance. + +Dropping independence permits W=-G, which has all norm moments but gives +F=0 and destroys rank and the positive coefficient. A degenerate Gaussian +component likewise need not supply the minors. The constants and asymptotic +window are not uniform as that component tends to zero. Covariance-matched +amplitude-mixture counterexamples remain boundaries of arbitrary-law claims; +this theorem admits the actual nonvanishing independent smoothing mechanism, +not every law sharing a covariance. + +## 11. Endpoint exhaustion and the bounded-cluster input + +A finite original lifetime list tending empty under scaling is not sufficient +to deduce Poisson. The proof supplies the missing global cluster estimate. +At a fixed contact-good product realization with V_(p,0)>0 the selected +pinned interval consumes both new critical points' unique endpoint roles. +Every other finite interval in the admitted degree set uses two old critical +branches. Their finitely many distinct limiting heights have a positive +minimum nonzero gap, regardless of how their old global pairing changes. +Thus for r=(ta/k)^(1/3), fixed 00, eventually the pinned ta interval +is the only interval of lifetime <=Tt. In the H0-only case the degree set is +just {0}; the proof does not claim all higher degrees on a nonorientable X. + +Insert the BOUNDED Borel mark 1{N_(Tt)^all>=2} into each once-counted endpoint +sum. The anchored sum itself accounts for N; there is no extra unbounded +witness count in the conditional integrand. On positive limiting weight the +mark vanishes eventually; on zero limiting weight the divided determinant +vanishes regardless of the mark. The fixed-target product-space bound +justifies the inner limit. The all-mark bound after ell=ta is proportional to + + a^-1/3 k^-2/3 H_M(b,k), 0=2}]=o(t^(2/3)). + +No second factorial estimate follows. The new independent finite control +exhibits epsilon=m^-2, P(N=1)=epsilon/2, P(N=m^2)=m^-5: its cluster expectation +divided by epsilon is 1/m->0, while its second factorial expectation divided +by epsilon is m-1/m->infinity. This explicitly distinguishes the accepted +bounded-cluster mechanism from an unsupported factorial claim. + +## 12. Raw marked mean, iid Laplace product and the endpoint a=0 + +The actual near-bar marks are degree, a=ell/t, birth height b, cubic k, +midpoint x and the ordered unit direction from birth to death. A unique short +geodesic defines them Borelly; no global cut-locus geodesic convention is +needed because far bars may be omitted. For independent copies their total +far contribution is at most n O(t_n)=O(t_n^(1/3))->0. Birth point and midpoint +have the same limit if a birth-position projection is desired, by their +vanishing displacement and the complete mark bound. + +The first-moment limit retains raw spatial volume: + + d eta_p=4 a^-1/3 k^-2/3 Z_(p,0)(b,k,x,u) + da db dk dvol(x)d sigma(u). + +Its mass on the COMPLETE window 0 and E[1-exp(-)] is zero +when there is at most one relevant bar and is bounded by the cluster count +otherwise. Thus it is o(t^(2/3)). The first-moment limit gives its leading +coefficient sum_p integral h d eta_p. Independent copies then give the +ordinary Laplace product, with n t_n^(2/3)->lambda and log-product remainder +vanishing, to yield intensity lambda sum_p delta_p tensor eta_p. + +This is a degree-marked Poisson random measure for iid WHOLE copies of the +fixed F. Disjoint limiting degree restrictions are independent Poisson +populations; the degree counts in one finite field need not be independent. +There is no spatial expanding-domain limit, one-field mixing theorem, +coalescing-site factorial estimate, or empirical field experiment hidden in +the proof. Those remain outside the accepted statement. + +## 13. Bounded primary check and exact import boundary + +The new reviewer downloaded and retained Milnor's primary book scan, +3572752 bytes, SHA256 +`658bfefbdfe7838b689eaddbd3a28b8d769ae51f7e47025fe7e5c9402b761af7`. +The bounded read covered §5's two-critical-point setup and Theorem 5.4, +and §7's setup, Lemma 7.2/Corollary 7.3 and the displayed corollary proof. +The relevant incidence pages and unique-point theorem were rendered and +visually inspected. Corollary 7.3 identifies the relative boundary matrix +with signed attaching/belt intersections; its setup imposes no simply-connected +or middle-index hypothesis. Theorem 5.4 corroborates the unique transverse +point criterion. The stronger algebraic cancellation theorem is not substituted. +This was not a full-book read, and no new full Laudenbach or Antony read is +claimed. [Milnor primary scan](https://webhomes.maths.ed.ac.uk/~v1ranick/surgery/hcobord.pdf) + +The main unit-fold reconstruction above is of the supplied complete HF source. +Its classical relative incidence premise remains explicit; the primary check +verifies the particular import rather than re-proving all Morse theory. The +bounded scan-read paragraph does not manufacture whole-book acceptance or +import unrelated h-cobordism dimension restrictions. + +Other retained classical premises are smooth local differential calculus, +finite-dimensional Sard and smooth incidence change of variables, Gaussian +regression and uniqueness, finite persistence decomposition and Morse level/ +handle/critical-free-flow facts, Tonelli, Minkowski/Holder, dominated convergence, +and the Laplace characterization of a Poisson random measure on the stated +locally compact separable mark space. GAD/HF derive their precise applications. +No general infinite-dimensional transversality theorem or arbitrary-mark +Kac–Rice formula is imported without the actual finite coefficient proof. + +## 14. Controls, failures retained and final disposition + +The fresh reviewer program `independent_smoothing_controls.py` ran once, +passing 884 exact finite assertions and seven negative/refusal families, +exit 0 and empty stderr. The controls use Gaussian-plus-two-atom W models at +literal zero targets, taking delta^2=2log(m) so Gaussian likelihood ratios +are exact rationals; polynomial moment identities; the all-mark tail threshold; +Gaussian-polynomial cumulants; inertia/sign bookkeeping; the cluster/factorial +counterexample; and raw spatial-volume normalization. They import no author or +old program. General all-target kernels, infinite-field convergence, topology +and limits are justified analytically above, not by the finite check count. + +The seven refusals distinguish an unweighted W posterior, the borderline +nonintegrable M=4/3 mark tail, polynomial versus exponential moments, +direction reversal versus transverse sign reversal, centered cone symmetry +versus shifted mixtures, bounded-cluster versus factorial control, and raw +versus probability-normalized spatial intensity. Finite scalar or count +examples are not misrepresented as full Gaussian manifold fields. + +The author reports 443 final checks and eight refusals. Its earlier passing +pre-clarification run is retained, not a second independent set of 443 tests. +Neither version was rerun here. Its preserved authoring history also records +one atomic prefreeze patch-context mismatch, which changed no file; matching +contexts were then used before freezing. That is an editing failure, not a +mathematical counterexample. This review does not erase it or count it as a +semantic proof failure. The present new control run had no failure. + +All five interfaces for the actual product law survive reconstruction. The +literal mixture kernel, nonnormalized product-space convergence, polynomial +all-mark envelope, actual manifold/chart and Hp selector, complete endpoint +exhaustion, correct joint coefficient and raw iid intensity are actually +proved rather than inherited from SG's stationary Gaussian case. The exact +source receives **ACCEPT_SCOPED**, with `must_fix_findings=[]` and +`required_amendments=[]`. + +Acceptance retains the actual independent decomposition, fixed positive dense +Gaussian component, all polynomial smooth-norm moments of W, geometry/degree +restrictions, fixed K and the explicit classical imports. It does not resolve +Main331 D1 or unrelated source/documentary/formal holds. It does not prove +all smooth nonGaussian laws, Gaussian parity, uniform degenerating-noise +constants, a numerical onset, higher singularity matching, within-field +Poisson or factorial results, sampler/grid qualification, Lean/kernel evidence, +blind50 validation, manuscript acceptance, external uptake or scientific +promotion. + +The sealed machine verdict and manifest bind this full public-safe report, +retained exact inputs, primary identity/extent, the new control evidence and +fresh final readback. No author or old source, independently frozen A4 artifact, +source branch, native page, private cloud record, or status register was +changed by this review. Root must read the entire report before any publication +or adoption decision. Organizational, provider-distinct, blind and human +independence credits remain zero. The full research goal remains ACTIVE / +UNMET; scientific effect is NONE. diff --git a/frontiers/nongaussian_smoothing_admission_20261017/imports/FINITE_SMOOTH_INCIDENCE_RECONSTRUCTION.md b/frontiers/nongaussian_smoothing_admission_20261017/imports/FINITE_SMOOTH_INCIDENCE_RECONSTRUCTION.md new file mode 100644 index 000000000..66a784fbd --- /dev/null +++ b/frontiers/nongaussian_smoothing_admission_20261017/imports/FINITE_SMOOTH_INCIDENCE_RECONSTRUCTION.md @@ -0,0 +1,83 @@ +# Independent finite smooth incidence reconstruction + +17 October 2026. Nonauthor reviewer `/root/reverse_localization_nonauthor_review_20261017`. +This is an analytic reconstruction of the finite case used by the frozen proposal, +not a new stochastic hypothesis, an implementation test, or a general nonsmooth +area/coarea claim. + +Let Phi:Omega subset R^n -> R^n be smooth. On a compact subdomain its derivative +is bounded and uniformly continuous. For a grid cube of side delta meeting the +critical set, choose a critical point x in that cube. Taylor's first-order +remainder is bounded by epsilon(delta)*delta throughout that cube, with +epsilon(delta)->0 uniformly. Since D Phi(x) has rank at most n-1, the image +of the cube lies in a rotated box whose n-1 long dimensions are at most C delta +and whose remaining dimension is at most C epsilon(delta) delta. Its outer +Lebesgue measure is at most C epsilon(delta) delta^n. There are O(delta^-n) +cubes in a fixed bounded subdomain. Thus the image of the critical set there +has outer measure at most C epsilon(delta), hence zero. Exhaust compact +subdomains. This proves exactly equal-dimensional smooth Sard as needed here; +no infinite-dimensional transversality is involved. The same argument works +in chart domains by compact exhaustion. + +At every regular point the inverse-function theorem supplies a neighborhood +on which Phi is a diffeomorphism. A countable such cover exists. Partition the +regular set into disjoint Borel pieces subordinate to that cover. On each +piece, ordinary change of variables, first for indicators and then simple +functions and monotone limits, gives for nonnegative Borel h + + integral h(z) |det D Phi(z)| dz + = integral sum_{z in Phi^-1(y)} h(z) dy. + +Sum the pieces by Tonelli. The critical source set contributes zero on the +left; all of its target values form a null set by the preceding argument, +so it contributes zero on the right. Truncate h first if necessary to avoid +any ambiguous infinite-times-zero notation. This is the full Borel weighted +equal-dimensional area formula for the particular smooth maps used in the +proposal. General locally Lipschitz area theory is a permitted explicit +import, but is stronger than necessary for these incidence maps. + +For a fixed smooth Gaussian-tail realization and a finite standard coefficient +block split into (a,eta), invert a gradient minor B(p). The critical-point +incidence is parameterized by + + (p,eta) -> (a(p,eta),eta), + a=-B(p)^-1(grad T+D(p)eta). + +Implicit differentiation at a zero gives a_p=-B^-1 H in physical orthonormal +coordinates. Therefore the incidence Jacobian is |det H|/|det B|. For two +sites the Jacobian is |det H_p det H_q|/|det B_pair|. The coefficient density +and linear observation-fiber Jacobian cancel the minor denominator to give +the actual Gaussian density of the observations. This retains precisely one +Hessian determinant at each root. With coordinate rather than orthonormal +rows, the corresponding volume factors cancel in the ordinary chart +change of variables. Smooth local frames suffice. + +The mark h can include every nonnegative Borel function of the sites and the +entire smooth field, including a nonlocal elder bar indicator. There is no +continuity assumption on that indicator. Fixing a smooth tail makes its +field dependence a Borel function of the remaining finite coefficients. +Tonelli integrates tails only after the finite identity has been proved. +The possible infinitude of total counts is harmless for the extended identity; +local finiteness follows later from the actual covariance and moment bounds. +The same construction under a contact target uses an orthonormal basis of the +finite contact kernel and its remaining standard Gaussian coordinates. + +This reduction still requires identifying the finite-coefficient fiber at its +literal observation target with the canonical whole-field conditional kernel. +That is a separate Gaussian calculation, not a consequence of area formula +or support. It is reconstructed in the main review and independently by the +focused dense-support/Rice reviewer. In particular the unweighted original +tail must not replace its target-dependent Bayes-weighted distribution. + +Primary-source check: Abhishek Shivkumar's UT differential-topology notes for +Daniel Allcock's Spring 2023 course state smooth Sard as Theorem 1.5.1 on +one-based PDF page 14. The reviewer read that exact statement and its +surrounding definition of manifold null sets, including a rendered page. +This is a bounded theorem-statement read, not a read of the full 60-page +notes or a claim that the notes prove every import. The exact retained PDF +is 636727 bytes, SHA256 +f1018cc3f838da11b3b98a5c39abc6498c0edb0eb615a3ab1f408b68e485e807. +Source: https://web.ma.utexas.edu/users/abhishek/files/notes/Differential%20Topology.pdf +The independent finite argument above and IFT/change of variables suffice for +the area formula used here; no abstract-only area/coarea citation is treated +as a proof of an all-Borel conditional Rice identity. diff --git a/frontiers/original_finite_l_gaussian_binding_20261017/PROOF.md b/frontiers/original_finite_l_gaussian_binding_20261017/PROOF.md new file mode 100644 index 000000000..e8bb07fe8 --- /dev/null +++ b/frontiers/original_finite_l_gaussian_binding_20261017/PROOF.md @@ -0,0 +1,445 @@ +# The original finite-L expression is the ideal Gaussian H0 coefficient + +New conventional source-alignment author proposal, 17 October 2026. +Authors: OpenAI/Codex reconstruction author +`/root/a4_gaussian_elder_sector_author_20261017` and Root01a11cbf +(mechanism coauthor). Actual native pickup: +[Math193/6243100632](https://github.com/d6g8k5htny-coder/Math-/issues/193#issuecomment-6243100632). +The pickup was read before mathematical source writes. This exact new +proposal requires separate nonauthor review. All authors are source-exposed +and share OpenAI; organizational/provider/blind/human independence is zero. +Scientific effect is NONE. + +## 1. Exact question, positive result and restricted meaning + +Fix an integer d>=2 and a finite real L>0, and use physical coordinates +on X=R^d/(L Z^d) with its flat metric and physical volume L^d. ORIGINAL +is the exact24027-byte/629-line independent-derivation.md, SHA256 +`f38c085ab5b15368e6989a1fd6298490f3b671f5a6c3a01bcdc81b516022bc33`, +public in +[Math193/6242350213](https://github.com/d6g8k5htny-coder/Math-/issues/193#issuecomment-6242350213). +Its expression(1), denoted C_alg(d,L), is + + C_alg = Gamma(7/6)/[24^(1/3)sqrt(pi)(2pi)^d a^(d/2)] + integral_S^(d-1) F_L(e) V_L(e)^(2/3) + /sqrt(det B_L(e)) d omega(e). (1) + +The original moments, conditioned Hessian law and full sphere AREA measure +in(1) are retained exactly, as defined below. This is not a fitted amplitude, +an infinite-L limit, a truncated-spectrum law, a height-pinned covariance +or an empirical coefficient. + +**Source-alignment theorem, conditional on the explicit already proved GAD +theorem and its stated classical imports.** The EXACT fixed finite-L +centered Gaussian smooth law specified in ORIGINAL satisfies GAD's actual +series hypotheses. Its expected finite ordinary superlevel-H0 bar measure +per physical volume has GAD's finite Borel density representative nu_GAD +with + + nu_GAD(ell) ~ C_alg(d,L) ell^(-1/3), + L^(-d) E N_bar((0,t]) ~ (3/2) C_alg(d,L) t^(2/3), + 00+. (2) + +All essential intervals are excluded. The density statement concerns the +specified representative, not every modification of it on a null set. +The integrated assertion has no version ambiguity. +The mathematical identification holds for any fixed real L>0; a later +numerical evaluation separately needs the method's effective representation +of that particular L. Such input access is not inferred from this theorem. + +This is an ALTERNATIVE source-bound proof of the ideal-law asymptotic +identification. It does not retrospectively certify each argument or +claim in ORIGINAL Sections7-9, erase their historical qualifications, +or assert that a numerical driver/grid/sample implements this law. +It supplies no effective positive-lifetime window, convergence rate, +M31 remainder, sampler/count-transfer/confidence certificate, enclosure +digits or diagnosis of the113retained failures. It does not close all +five scientific input groups. Frozen original and method sources remain +unchanged; the existing numerical coordinator owns their numerical work. + +## 2. Exact theorem imports and current method version + +GAD is Actual fold admission for smooth Gaussian fields with dense smooth +support, exact40231bytes/739lines, SHA256 +`8d992e194f0a898e563c4c3d05b8079cde292f9e87d7ff51030ad12ef5faf319`. +Its full accepted scoped review is38035bytes/641lines, SHA256 +`6fedc43ba34408f23f8934166b3898e2cd19175c57e61e37227c85f3ca5d4c70`; +its finite-incidence appendix is5032bytes/83lines, SHA256 +`b1b928ce32a6a39b464cb9f1c6dd82f61b95d2bb095459d9b02f301168bd7fb2`. +All three full actual sources were read earlier in this same author session, +not inferred from a summary; their bytes were freshly reverified for this +task and the relevant coefficient/target passages reread. ORIGINAL was +also read completely and its actual tensor/regression passages reread. +No additional whole-book or primary-paper read is claimed here. + +GAD admits EACH actual centered series sum_j a_j Z_j phi_j on a fixed +compact connected boundaryless Riemannian manifold when all a_j>0, +the real smooth span of phi_j is dense in C-infinity, and +sum_j a_j||phi_j||_Cq is finite for every integer q>=0. It proves the +same-law literal Gaussian kernels and all-Borel finite-incidence count, +conditional off-contact genericity, actual global elder selection, +contact determinant limit, all-mark domination and far estimate. Its +equations(19),(25),(27),(28) supply the actual finite H0 density and +joint-jet coefficient. Those are conventional theorem inputs, under +GAD's explicit finite differential-topology/measure/Morse imports. +They are not compiler, experiment, coefficient-enclosure or scientific +evidence. No new stochastic selection premise is silently added here. + +The numerical method is now the EXACT corrected METHOD V2 at +[Math193/6243041893](https://github.com/d6g8k5htny-coder/Math-/issues/193#issuecomment-6243041893), +32593bytes/414lines, SHA256 +`8bdea4b42e9efd2d846cd1aa73691e8194fa26cd937ac02b79991334f88204f6`. +It was read completely. The actual V1 finding +[6242995585](https://github.com/d6g8k5htny-coder/Math-/issues/193#issuecomment-6242995585) +and the full literal V1-to-V2 diff were read. V1's31426-byte/d947 source +is HISTORICAL, AMEND_REQUIRED for its all-real spectral cutoff domain. +V2 restricts the spectral cutoff to positive integers and chooses +integer Gaussian radius/rational boxes; the remaining edits disclose +version/claim/history/review assistance. All33displayed mathematical +blocks are byte-identical between V1 and V2. In particular V2 changes +no ideal field law, M1 coefficient, M4 height-unconditioned Hessian law +or density-version qualification. This binding uses ORIGINAL and GAD +directly; numerical-method acceptance is not a mathematical premise +for(2). The new peer perturbation advisory is likewise not a premise. + +## 3. Construct the exact full-spectrum Gaussian smooth field + +For m in Z^d put + + p_m=2pi m/L, + Z_L=sum_m exp(-|p_m|^2/2), + w_m=exp(-|p_m|^2/2)/Z_L. (3) + +All sums are over the ENTIRE lattice. Z_L is finite and strictly +positive and every w_m>0. Factorization of nonnegative sums gives + + Z_L=z_L^d, + z_L=sum_{k in Z} exp(-(2pi k/L)^2/2). (4) + +Thus the spectral probability vector with masses w_m has exactly +ORIGINAL's independent symmetric coordinates Q_i of Section2. The +notation p_m here separates wave numbers from probability masses; +it does not change ORIGINAL's q_m=2pi m/L convention. + +Choose a set H containing one member of every nonzero pair {m,-m}. +Let Z_0 and (X_m,Y_m)_{m in H} be mutually independent standard real +Gaussians. Define the ACTUAL field + + F_L(x)=sqrt(w_0)Z_0 + +sum_{m in H} sqrt(2w_m) + [X_m cos(p_m.x)+Y_m sin(p_m.x)]. (5) + +This retains the random spatially constant zero mode. There is no +additional L^(-d/2) basis normalization or samplewise mean removal. +The deterministic smooth functions in(5) are1 and the literal sine/ +cosine functions, with strictly positive coefficients sqrt(w_0) and +sqrt(2w_m). Each fixed Cq norm is bounded by C_q(1+|p_m|)^q. + +For every integer q>=0, + + sum_{m in H} sqrt(2w_m)(1+|p_m|)^q < infinity. (6) + +For a direct proof, sqrt(w_m) is a fixed positive multiple of +exp(-|2pi m/L|^2/4). Group lattice points by integer shells in |m|. +The number in the k-th shell is at most C_d(1+k)^d. Its polynomial +factor times exp(-c_L k^2) is summable, since for all sufficiently +large k any prescribed polynomial is at most exp(c_L k^2/2), and +sum exp(-c_L k^2/2) is bounded by a geometric tail. The finitely +many smaller k do not matter. Constants depend on this fixed L,d. + +The real span is dense in C-infinity(X). For a smooth periodic f, +repeated integration by parts makes its Fourier coefficients decay +faster than any polynomial in |m|. For each q choose enough integrations +so the derivative Fourier coefficients are absolutely summable. +Finite symmetric partial sums converge in Cq, and the same sequence +converges in every Cq. Their real form is a combination of the1, +sine and cosine functions in(5). This proves the required smooth +density, not merely L2 completeness. + +One can see smooth convergence of(5) directly as well. For each q, +the expectation of the sum of Cq bounds with |X_m|,|Y_m| is finite +by(6), hence the sum is finite almost surely. Countably intersect +over q. The derivative series then converge absolutely and uniformly, +giving an actual smooth Borel field. Minkowski gives all finite +positive moments of its smooth norms. This is the same actual series +construction admitted by GAD, with no truncation used as a source law. + +## 4. Equality of Gaussian Borel field laws, not just covariance matching + +Finite collections of(5)'s values and derivatives are centered jointly +Gaussian: their finite sums are Gaussian, converge in L2 by(6), and +their characteristic functions converge to the corresponding Gaussian +characteristic function. Independent coefficients give, exactly, + + Cov(F_L(x),F_L(y)) + =w_0+sum_{m in H}2w_m cos(p_m.(x-y)) + =sum_{m in Z^d}w_m cos(p_m.(x-y)) + =E cos(Q.(x-y))=K_L(x-y). (7) + +This is ORIGINAL's literal full-spectrum covariance, with pointwise +variance sum_m w_m=1. The spatial mean of(5) is sqrt(w_0)Z_0, not +zero for each sample. Any driver mean removal or variance rescaling +would require its own law comparison; none is presumed here. + +The ideal source in ORIGINAL is centered Gaussian, so its finite +evaluation laws are determined by(7). Equality of covariance is being +used WITH joint Gaussianity, not as an assertion about arbitrary +nonGaussian laws sharing K_L. A countable dense evaluation family +generates the smooth Borel sigma-field: derivatives are limits of +value finite differences on countable chart grids, Cq seminorms are +countable suprema, and rational trigonometric polynomials give a +countable dense family of norm-ball centers. Hence the Gaussian +finite-dimensional identity determines the same law as a Borel +C-infinity random element. Equivalently(5) constructs that actual +ideal smooth version. Translation invariance of(7) also gives +stationarity of the full Borel field law by the same argument. + +GAD's hypotheses are now proved for this exact law. It supplies a +finite Borel density nu_GAD for the actual finite ordinary H0 bars, +with its finite positive coefficient C_GAD and exponent -1/3. Because +the global bar count is a Borel functional as established in GAD, +the smooth-law identity transfers that observable, not just its jets. +It remains to identify C_GAD with the particular original expression. + +## 5. Actual derivative covariance and the odd regression + +Use the physical orthonormal coordinates and put D=grad F_L, +H=Hess F_L, and T_e=partial_e^3 F_L at one point. Differentiating +(7) termwise is justified by(6). Let a=mu_2,b=mu_4,c=mu_6 and +kappa_4=b-3a^2,kappa_6=c-15ab+30a^3 as in ORIGINAL(4). The product +spectral law gives + + Cov(D)=aI_d, + Cov(H_ij,H_kl) + =a^2(delta_ij delta_kl+delta_ik delta_jl+delta_il delta_jk) + +kappa_4 sum_s delta_is delta_js delta_ks delta_ls. (8) + +These identities follow by grouping indices into even multiplicities: +four equal indices have moment b, two distinct equal pairs have a^2, +and any odd coordinate multiplicity has zero moment. This yields +the three pairings plus the fourth-cumulant correction exactly. + +All covariance entries between D or T_e and H vanish by odd spectral +parity. These variables are jointly centered Gaussian, so the ENTIRE +block (D,T_e) is independent of H, including its transverse and axial +entries. Independence here is proved for this exact stationary +Gaussian law; it is not inherited from GAD's general nonstationary +coefficient theorem. + +Differentiating in one first- and one third-derivative argument gives + + Cov(T_e,D_i)=-v_i(e), + v_i(e)=E[Q_i(Q.e)^3]=3a^2 e_i+kappa_4 e_i^3. (9) + +The minus sign comes from the three derivatives in the second field +argument. Expansion by coordinate multiplicities gives + + E(Q.e)^6=15a^3+15a kappa_4 sum_i e_i^4 + +kappa_6 sum_i e_i^6. + +Therefore the literal Gaussian law of T_e conditioned on D=0 is +centered with variance + + V(e)=E(Q.e)^6-a^(-1)|v(e)|^2 + =6a^3+9a kappa_4 sum_i e_i^4 + +(kappa_6-kappa_4^2/a)sum_i e_i^6, (10) + +because |v|^2=9a^4+6a^2 kappa_4 sum e_i^4+kappa_4^2 sum e_i^6. +This is EXACTLY ORIGINAL(8) and V2(M3). In a coordinate direction +it is c-b^2/a. Omitting the gradient regression changes this object. +Positivity follows from GAD's finite independent jet rank, or directly +because a cubic polynomial (Q.e)^3 cannot agree with a linear one +on the full lattice support. No finite spectral carrier replaces +that support. Also a>0 by the nonzero spectral masses. + +## 6. The exact He=0 transverse Gaussian law + +For a unit e, contraction of(8) gives + + B(e)=Cov(He)=a^2(I_d+2ee^T) + +kappa_4 diag(e_i^2). (11) + +It is positive definite by the full Hessian/jet rank. This is +ORIGINAL(6), not a rotationally averaged approximation. Let U have +orthonormal columns spanning e-perp. Contracting(8) once with U +and e gives + + R_ab,i=Cov((U^THU)_ab,(He)_i) + =a^2 delta_ab e_i+kappa_4 U_ia U_ib e_i. (12) + +In the isotropic pairing terms the other two contractions contain +U^Te=0 and vanish. Contracting all four entries with U gives the +unconditioned transverse covariance + + C_ab,cd=a^2(delta_ab delta_cd+delta_ac delta_bd+delta_ad delta_bc) + +kappa_4 sum_i U_ia U_ib U_ic U_id. + +Consequently the actual canonical centered Gaussian law of U^THU +conditioned on He=0 has covariance + + Gamma_e(ab,cd)=C_ab,cd + -sum_ij R_ab,i (B(e)^(-1))_ij R_cd,j. (13) + +This reproduces ORIGINAL equation(11) entry for entry and the +conditioned object V2(M4). Independent free entries ab,cd are listed +once; no orthonormalized off-diagonal factor is introduced. The +conditional covariance is nondegenerate by the enlarged local jet +rank, so its negative cone has positive determinant-square mass. +Define + + F(e)=E[(det A_e)^2 1{A_e<0}], A_e=U^THU | He=0. (14) + +This is the ORIGINAL F_L(e), including its raw Gaussian integral(11a). +Changing U conjugates A by an orthogonal matrix and preserves its +determinant square/inertia, so no global continuous frame is needed. + +There is NO condition F_L(x)=0 in(13)-(14). The scalar height is +correlated with Hessian entries; introducing such a condition would +generally alter this law. GAD's coefficient already integrated out +the birth height. Independence of (D,T_e) from H implies, under the +literal Gaussian pins D=0,He=0, that T_e|D=0 and A_e|He=0 are +independent. This is valid at the exact zero targets by elementary +full-rank Gaussian regression, not merely an almost-every-target +disintegration. Both factors are the actual ones in(10),(14). + +## 7. Bind the actual density representative and pair coordinates + +For M!=S let O=(F(M),grad F(M),F(S),grad F(S)) and use GAD's +canonical full-field Gaussian regression Q_O. Its actual density is + + nu_GAD(ell)=L^(-d) integral_(M!=S) integral_R + p_O(b,0,b-ell,0) + E_QO[ |det H_M det H_S| sigma_eld(M,S,F)] db dM dS. (15) + +Here sigma_eld is GAD's proved Borel once-counting selector for finite +ordinary H0 elder bars. This is its equation(19), now applied to the +law whose identity was proved in Sections3-4. GAD proves all-Borel +Rice, finite conditional kernels, global elder selection, integrable +near dominance and bounded far density. Thus(15) is a particular +finite Borel density of the genuine expected lifetime measure. +It is not the unselected maximum/saddle gap measure. + +To connect ORIGINAL's height-integrated coordinates literally, set +L_MS=(grad F(M),grad F(S),Delta=F(M)-F(S)). The invertible change +O -> (b=F(M),L_MS) has absolute determinant one. Integrating b in +the joint Gaussian density and canonical kernel gives exactly + + p_L(0,0,ell) Q_L(0,0,ell) + =integral_R p_O(b,0,b-ell,0) Q_O(b,0,b-ell,0) db (16) + +as a whole-field measure identity. For finite evaluation tests this +is ordinary Gaussian marginal/conditional integration at every +target; equality extends to whole smooth Borel laws by the countable +evaluation argument above. Therefore(15) also uses precisely the +gradient/gap pin object intended by ORIGINAL, with a proved global +mark. No extra common-height density remains. + +For short M=x-re/2,S=x+re/2 define ORIGINAL's + + P_r=(D_M+D_S)/2, D_r=(D_S-D_M)/r, + W_r=(12/r^3)[Delta+r e.P_r]. + +The gradient transform has determinant r^(-d); the last row adds +12/r^3. Hence its exact density Jacobian is12r^(-(d+3)), matching +ORIGINAL(18) and GAD's observation normalization after height +integration. On the zero-gradient target W_r=12ell/r^3. This is a +finite-coordinate identity, not an admission of every original +limiting argument. GAD's distinct all-mark proof supplies the +asymptotic for(15) under the now-verified law hypotheses. + +## 8. Exact coefficient equality, with every normalization + +Write mathcalV=L^d to avoid confusing volume with V(e). GAD's +equation(28), specialized to the ordinary H0 cone, is + + C_GAD=1/[2*12^(1/3)*mathcalV] integral_X integral_S^(d-1) + p_(D,He)(0,0) + E[|T_e|^(4/3)(det A_e)^2 1{A_e<0}|D=0,He=0] d omega dx. (17) + +GAD may express the Hessian column in frame(e,U); ORIGINAL expresses +He in the fixed physical frame. Their coordinate transformation is +orthogonal, of absolute determinant one, and preserves the zero +target. Thus the zero densities and matrix determinant below are +exactly the same; no frame-volume factor occurs. + +Stationarity of the actual full law makes the jet integrand in(17) +independent of x. The x integral is mathcalV and cancels its inverse +once. Odd/even independence and(10)-(14) give + + p_(D,He)(0,0)=(2pi)^(-d) a^(-d/2)(det B(e))^(-1/2), + E[|T_e|^(4/3)(det A_e)^2 1{A_e<0}|D=0,He=0] + =E[|N(0,V(e))|^(4/3)] F(e). (18) + +For completeness, integrating the centered scalar Gaussian density +on both half-axes and substituting y=t^2/(2V) gives + + E|N(0,V)|^(4/3) + =2^(2/3) Gamma(7/6) V^(2/3)/sqrt(pi). (19) + +The numerical algebra is exact: + + 2^(2/3)/(2*12^(1/3))=24^(-1/3). + +Substitute(18)-(19) into(17). The result is EXACTLY(1): + + C_GAD=C_alg(d,L). (20) + +One can check the same normalization before evaluating Gamma: +GAD(27) becomes12^(-1/3)integral p_D(0)p_He(0)F(e) +E[(T_e)_+^(4/3)|D=0]d omega, identical to ORIGINAL(22). +Its one-sided Gaussian moment is half(19), as ORIGINAL(23) states. +The full sphere already counts the ordered birth/death direction +once. GAD's antipodal half in(17) and the absolute moment in(19) +are one equivalent representation of that same count, not grounds +for another half. The scalar d2 negative-cone half inside F(e) +has a separate distributional origin and remains unchanged. + +Every e-dependence is retained. Cubic lattice symmetry has not been +replaced by continuous rotational invariance. The finite L is fixed +through(3)-(4) in a,b,c,B,V and Gamma_e; no limiting substitutions +a=1,b=3,c=15 were made. No extra L^(-d) is left after the explicit +volume cancellation, and no height density or off-diagonal Gaussian +measure normalization is missing. GAD gives0 actual ordinary finite H0 bar density +from GAD -> original derivative/conditional laws -> exact coefficient +equality(20). Combining GAD's conclusion with(20) proves(2), within +the stated conventional theorem imports. The integrated coefficient +factor3/2 follows by integrating ell^(-1/3), not by differentiating +a cumulative approximation. Any other density representative agrees +almost everywhere, but changing it on a null sequence can destroy +pointwise asymptotics; such a representative is not asserted in(2). + +This supplies a new alternative proof of qualitative ideal-law +identification. It does not declare ORIGINAL's own uniform-tube, +Sobolev-tightness, marked approximation or measure-sandwich passages +individually verified. Their original historical verification +qualifications remain in their frozen bytes. Similarly V2's statement +that its ENCLOSURE METHOD alone did not identify the barcode coefficient +remains accurate as a statement of that unit's evidence. If this new +binding is separately accepted, a later explicit source-bound record +may add the alternative theorem; the old record must not be rewritten +as if the method itself proved it. + +No finite-ell modulus, onset scale or M31 error functions are obtained. +A certified interval for the algebraic expression, if actually executed +and reviewed, could then also enclose the ideal-law limiting coefficient +identified by(20); it would still say nothing quantitative about a +particular positive lifetime, finite sampler/grid or empirical fit +without their separate error/transfer/confidence evidence. No interval +has been computed by this unit. The five missing scientific input +groups,113retained failures, stopped TA/TRIG routes, distinct-model +blind50 requirement, and broader scientific acceptance are not closed. + +No source formula, method, numerical driver, status register, external +message, branch or Drive object was changed. The only executions here +were read-only source retrieval/text comparison and inert byte custody. +No compiler, Lean/control suite, field draw, numerical coefficient +evaluation, experiment or old rerun was launched. All exact inputs and +read extents are in INPUT_MANIFEST.json; author result and output seal +identify this new proposal. Independent nonauthor review of these exact +bytes remains required. The full research goal remains ACTIVE / UNMET. diff --git a/frontiers/original_finite_l_gaussian_binding_20261017/README.md b/frontiers/original_finite_l_gaussian_binding_20261017/README.md new file mode 100644 index 000000000..ecc6300eb --- /dev/null +++ b/frontiers/original_finite_l_gaussian_binding_20261017/README.md @@ -0,0 +1,15 @@ +# Original fixed finite-L ideal Gaussian binding + +This is a source-author publication proposal carrying the full unchanged [proof](PROOF.md) and [actual integrated nonauthor review](REVIEW.md). Historical proposal/PENDING wording stays unchanged beside the actual later scoped acceptance. This index is not a new proof, review, source adoption, owner decision or scientific status register. + +Exact fixed finiteL original Gaussian law admitted through GAD; specified finite Borel ordinaryH0 density representative has C_alg ell^(−1/3) asymptotic and integrated(3/2)C_alg t^(2/3) per physical volume. These are the stated small-argument results and must not be treated as exact finite-scale power laws. + +Alternative ideal-law identification, not retrospective validation of every original argument; no M31/rate/onset/sampler/grid/count/confidence certificate, evaluated interval, all-five-gap closure or science. The full original proof and review control the exact hypotheses and limitations. All remaining original numerical/scientific obligations retain their evidence-specific dispositions; this packet does not close all five groups or the113retained failures. + +Exact GAD theorem and stated classical imports; original exact formula/tensors; numerical METHODV2 and P3/P7 advisory are not mathematical premises. + +The exact [focused tensor review](focused-tensor-normalization/FOCUSED_TENSOR_NORMALIZATION_REVIEW.md) is part of the actual integrated review. Its author is this later packet preparer; no new independent review credit is created. + +The shared [source/dependency/exclusion index](../nongaussian_smoothing_admission_20261017/FOLLOWON_SOURCE_INDEX.json) records exact original prose identities, canonical source/review publications, immutable existing Math475/GAD and Main331/HF references, neutral historical-seal hashes and scope boundaries. Already landed GAD proofs/reviews and existing HF sources are referenced, not duplicated or replaced. Original sources and old seals were not edited. + +No Lean source, package, manifest, workflow, runtime, compiler target or execution receipt is registered here. Full books/PDFs/PNGs, private filesystem/fileID metadata, programs, failed attempts and recovery archives remain local; this is not a public ZIP, hermetic proof closure or portable execution package. New routing metadata has source-author provenance. The completed author/review sources here are source-exposed OpenAI, organizational/provider-distinct/blind/human credit0. Supplementary Claude input on the separate finite-L binding (moment/He/normalization only) is PENDING at Math193/6243336796; it supplies no completed review, adoption or independence credit. Existing C261 workflow/source history at this base is separate; this candidate changes none of it. Scientific effect NONE. diff --git a/frontiers/original_finite_l_gaussian_binding_20261017/REVIEW.md b/frontiers/original_finite_l_gaussian_binding_20261017/REVIEW.md new file mode 100644 index 000000000..95c0d25a1 --- /dev/null +++ b/frontiers/original_finite_l_gaussian_binding_20261017/REVIEW.md @@ -0,0 +1,438 @@ +# Nonauthor mathematical review: original finite-L coefficient and GAD + +**Verdict: ACCEPT_SCOPED.** The exact frozen source establishes +`C_GAD = C_alg(d,L)` for the same original fixed finite-L centered Gaussian +smooth law, conditional on the explicitly consumed GAD theorem and its +stated classical imports. Its construction verifies the actual Gaussian +series hypotheses, its density statement concerns the genuine finite +ordinary H0 bar measure and specified representative, and its tensor, +regression and normalization reductions recover the original expression. +No must-fix finding or required source amendment was found in this binding. + +This is a new alternative qualitative identification proof. It is not a +retrospective acceptance of the original's individual G1–G4 arguments, +an effective lifetime window, an executed coefficient enclosure, a sampler +or count-transfer certificate, a diagnosis of retained failures, or a closure +of all five scientific input groups. + +Reviewer: OpenAI/Codex `/root/a4_local_archive_consumer_20261017`, +17 October 2026. Root is mechanism coauthor and +`/root/a4_gaussian_elder_sector_author_20261017` is reconstruction author. +The present reviewer did not author this binding, its mechanism, or the +GAD theorem. Earlier work here was local archive consumption and a distinct +source-composition review; that exposure is disclosed and supplies no new +mathematical validation of those earlier objects. The present unit actually +reads and reconstructs the new mathematical claim. All participants are +source-exposed OpenAI/Codex; organizational, provider-distinct, blind and +human independence credits are zero. + +## 1. Exact source and actual reading + +The reviewed binding `PROOF.md` is 22163 UTF-8 bytes, 445 lines, SHA256 +`ad9ab5afea28062dfc661d0d19b43f20e67cfd67cb6910d3cc6c3fb8b81398d8`. +Its original author seal is 3308 bytes, SHA256 +`83361eeb0f74574464b77638005794ced810691309592879a2b5f3a3fb20f065`, +binding 19 files totaling 400004 bytes. All 19 rows and the seal were +verified before review; no final verdict was issued on an unstable source. + +The following complete exact documents were read for this unit: + +| Input | Bytes / lines | SHA256 | +| --- | --- | --- | +| New binding proof | 22163 / 445 | `ad9ab5afea28062dfc661d0d19b43f20e67cfd67cb6910d3cc6c3fb8b81398d8` | +| ORIGINAL independent derivation | 24027 / 629 | `f38c085ab5b15368e6989a1fd6298490f3b671f5a6c3a01bcdc81b516022bc33` | +| GAD proof | 40231 / 739 | `8d992e194f0a898e563c4c3d05b8079cde292f9e87d7ff51030ad12ef5faf319` | +| Actual source-bound GAD review | 38035 / 641 | `6fedc43ba34408f23f8934166b3898e2cd19175c57e61e37227c85f3ca5d4c70` | +| Finite smooth incidence appendix | 5032 / 83 | `b1b928ce32a6a39b464cb9f1c6dd82f61b95d2bb095459d9b02f301168bd7fb2` | +| Current METHOD V2 | 32593 / 414 | `8bdea4b42e9efd2d846cd1aa73691e8194fa26cd937ac02b79991334f88204f6` | + +The original pickup, V1 adverse domain finding, complete literal V1-to-V2 +diff, author result/input manifest and version reconciliation were also read. +Historical V1 bytes and displayed mathematical blocks were compared; no +complete rereading of every V1 prose line is claimed. Exact retained copies +and read extents accompany this report. No original source, method or seal +was edited. No new primary-paper or classical-book reading is claimed. + +The source question is precisely the [original finite-L expression](https://github.com/d6g8k5htny-coder/Math-/issues/193#issuecomment-6242350213), +under the [binding pickup](https://github.com/d6g8k5htny-coder/Math-/issues/193#issuecomment-6243100632). +The [current V2 method](https://github.com/d6g8k5htny-coder/Math-/issues/193#issuecomment-6243041893) +is context for which expression is being evaluated, not a mathematical +premise that supplies the barcode identification. The retained V1 finding +at [6242995585](https://github.com/d6g8k5htny-coder/Math-/issues/193#issuecomment-6242995585) +remains adverse history. + +## 2. Actual Fourier law, including the zero mode + +Fix an integer d>=2 and any fixed finite real L>0. Physical coordinates +give X=R^d/(L Z^d), flat metric and volume L^d. With h=2pi/L, set + + p_m=h m, Z_L=sum_(m in Z^d) exp(-|p_m|^2/2), + w_m=exp(-|p_m|^2/2)/Z_L. + +Nonnegative product summation gives Z_L=z_L^d. Hence the corresponding +spectral vector Q has exactly the original independent symmetric coordinate +law. All weights are positive, including w_0, and sum to one. + +Choose one representative of each nonzero pair {m,-m}. The proposed real +field is + + sqrt(w_0) Z_0 + sum_H sqrt(2w_m) + [X_m cos(p_m.x)+Y_m sin(p_m.x)], + +with all displayed scalar coefficients independent N(0,1). A single pair +contributes 2w_m cos(p_m.(x-y)) to covariance. Summing all pairs and the +constant contributes exactly sum_(Z^d) w_m cos(p_m.(x-y)), the original +K_L(x-y). Point variance is one. The random spatial mean is sqrt(w_0)Z_0; +it is not removed. There is no unaccounted L^(-d/2) basis normalization. +Thus the real representation does not halve or double the nonzero spectrum. + +For fixed q, the deterministic sine/cosine Cq norms are bounded by a fixed +multiple of (1+|m|)^q. Their positive amplitudes decay as exp(-c_L|m|^2). +Polynomial shell counts therefore prove the required weighted norm sum is +finite. The constant mode is harmless. For each q the expectation of the +absolute Gaussian norm-bound sum is finite, so that sum is finite almost +surely. Countably intersect over q: every derivative series converges +absolutely uniformly. This constructs an actual Borel smooth field. +Minkowski supplies all finite p>=1 norm moments; smaller positive moments +follow from those. No finite cutoff field is used as the source law. + +For smooth periodic f, repeated integration by parts makes its Fourier +coefficients decay faster than every polynomial. Choose sufficiently many +integrations for each derivative order. Symmetric finite Fourier partial +sums then converge in every Cq norm. The real trigonometric functions above +have dense real span in C-infinity, not just in L2. Relabeling this countable +list supplies exactly the positive-amplitude independent series admitted +by GAD. The construction works for every fixed L>0 without claiming any +common constants or useful window as L varies. + +## 3. Equality of full smooth Gaussian laws and the actual bar observable + +Finite evaluation and derivative vectors of the series are jointly Gaussian: +their finite prefixes converge in L2, and their Gaussian characteristic +functions converge to the corresponding centered Gaussian characteristic +function. Their covariance is the original covariance, termwise +differentiation being justified by the all-norm summability. + +Covariance determines these laws because both objects here are centered +jointly Gaussian. This is not a covariance-only assertion about an arbitrary +nonGaussian field. To pass from finite evaluations to the smooth random +element, use countable coordinate grids. Derivatives are measurable limits +of difference quotients of values; derivative seminorms are countable +suprema on dense grids; rational trigonometric combinations supply countable +ball centers. Thus countably many values generate the smooth Borel +sigma-field. Equality of their joint laws gives equality of the actual +Borel C-infinity field laws. The stationary covariance also makes the +whole smooth law stationary by the same argument. + +The finite ordinary H0 bar measure is GAD's proved Borel functional, with +the essential class excluded. Equality of field laws therefore transfers +that global observable, not merely its local jets. Applying the admitted +GAD theorem gives its actual finite Borel density representative and +finite positive coefficient. There is no need to adopt the original's +unverified marked-sandwich arguments as premises. + +The theorem concerns a specified density version. Another version can +differ on a null sequence and fail a pointwise asymptotic. The integrated +expected finite-bar statement is version-independent. The source keeps +this distinction and does not claim that every historical symbol rho has +already been shown pointwise equal to the selected GAD representative. + +## 4. Physical jet tensor and the odd regression + +Write D=grad F, H=Hess F and T_e=partial_e^3 F. Set +a=mu_2, b=mu_4, c=mu_6, kappa_4=b-3a^2 and +kappa_6=c-15ab+30a^3. Product symmetry gives + + Cov(D)=a I, + Cov(H_ij,H_kl)=a^2(delta_ij delta_kl+delta_ik delta_jl + +delta_il delta_jk) + +kappa_4 sum_s delta_is delta_js delta_ks delta_ls. + +Four equal spectral indices give b; two distinct equal pairs give a^2; +an odd coordinate multiplicity gives zero. This reconstructs the tensor +without assuming rotation invariance or positivity of kappa_4. + +Odd/even spectral parity makes the covariance of the entire (D,T_e) block +with the entire H block zero. Since the complete jet is jointly Gaussian, +the two blocks are independent. This is separately proved for the original +stationary Gaussian law; it is not a simplification that GAD allows for +every general manifold Gaussian law. + +The physical derivative signs give + + Cov(T_e,D_i)=-v_i, + v_i=3a^2 e_i+kappa_4 e_i^3. + +The minus sign follows by differentiating once in the first field argument +and three times in the second. Expanding the directional sixth moment and +subtracting the actual gradient projection gives + + E(Q.e)^6=15a^3+15a kappa_4 sum e_i^4+kappa_6 sum e_i^6, + V(e)=E(Q.e)^6-|v|^2/a + =6a^3+9a kappa_4 sum e_i^4 + +(kappa_6-kappa_4^2/a)sum e_i^6. + +Thus T_e given D=0 is centered N(0,V(e)), exactly ORIGINAL (8) and V2 M3. +In a coordinate direction V=c-b^2/a; the unregressed sixth moment c is a +different object. Positivity follows from the admitted full finite-jet rank. +It also has a direct spectral proof: if the residual cubic had zero +variance, its cubic polynomial would agree with a linear polynomial on +every lattice point of positive mass. Restrict to an axis where e_i is +nonzero and vary the integer coordinate; a nonzero cubic cannot equal a +linear polynomial for all those integers. Likewise a>0. No cutoff-support +argument is substituted for the original full support. + +## 5. He=0, raw symmetric coordinates, and the forbidden height pin + +Contracting the fourth tensor with e twice gives + + B(e)=Cov(He)=a^2(I+2ee^T)+kappa_4 diag(e_i^2). + +For orthonormal transverse columns U, let A=U^T H U. Direct contraction gives + + Cov(A_ab,(He)_i)=R_ab,i=a^2 delta_ab e_i+kappa_4 U_ia U_ib e_i, + Cov(A_ab,A_cd)=a^2(delta_ab delta_cd+delta_ac delta_bd + +delta_ad delta_bc) + +kappa_4 sum_i U_ia U_ib U_ic U_id. + +The other isotropic cross contractions vanish because U^T e=0. Conditioning +on He=0 subtracts R B^(-1) R^T, exactly the original equation (11). +The indices a<=b label each free symmetric entry once. No Frobenius +orthonormalization or extra off-diagonal density factor has appeared. +Consequently the original raw Gaussian integral (11a) is exactly + + F_L(e)=E[(det A)^2 1{A<0} | He=0]. + +The joint map H -> (He,U^T H U) is an isomorphism on symmetric matrices +after choosing the frame (e,U). Full Hessian covariance rank therefore +makes B and the transverse Schur complement positive definite. This proves +nondegeneracy of the raw covariance and positive determinant-square mass +on the open negative cone. Conjugating U by an orthogonal transverse basis +change preserves that expectation; no global continuous frame is needed. + +This covariance is **not** conditioned on f=0. The source has already +integrated height. A concrete refusal is d=2, e=e_1: unconditioned-height +variance is b-a^4/b, whereas adding f=0 gives b-a^2. Their difference is +a^2(b-a^2)/b>0. Here Cov(f,H)=-a I, so the extra height pin genuinely +changes the law. In the isotropic parameter check a=1,b=3 this is 8/3 +versus 2, not an interchangeable convention. This check is algebraic, not +a newly sampled field or evaluated original coefficient. + +Since H is independent of (D,T_e), conditioning on D=0 and He=0 retains +independence between T_e|D=0 and A|He=0. Full-rank finite Gaussian regression +specifies this at the literal zero targets. It is not merely an +almost-every-target independence assertion. + +## 6. The null-target whole-field marginalization + +For fixed distinct sites M,S, GAD supplies a nondegenerate full observation +O=(f(M),D_M,f(S),D_S) and its canonical whole-field Gaussian kernel. +Let B=f(M) and Y=(D_M,D_S,Delta=f(M)-f(S)). The linear map O -> (B,Y) +has absolute determinant one, and both covariance matrices are nondegenerate +by the admitted finite-jet rank. + +For every literal y, finite Gaussian regression gives a Gaussian conditional +law for B given Y=y with positive variance. For any finite field evaluation +vector E, decompose its conditional mean/covariance first under (B,Y) and +then integrate that Gaussian conditional B law. The mean is the direct +regression on Y, and the covariance is + + Cov(E | B,Y) + A_B Var(B | Y) A_B^T = Cov(E | Y), + +where A_B is the B coefficient in the regression on (B,Y). These are Schur +identities, valid at every y. Their characteristic functions therefore +identify the whole Gaussian measure after the countable-evaluation Borel +argument. Hence, as measures on smooth fields, + + p_Y(y) Q_Y(y)=integral_R p_(B,Y)(b,y) Q_(B,Y)(b,y) db. + +This proves the binding's (16) at the specific zero-gradient/gap target. +It is stronger than invoking an unspecified regular conditional probability +identity almost everywhere. It applies to every nonnegative Borel mark by +equality of measures and monotone limits. In particular the Hessian product +times GAD's global elder mark may be inserted without assuming that mark is +continuous on a null fiber. The determinant is present once at each endpoint. + +The actual GAD density thus equals the height-integrated gradient/gap +kernel expression for the same bar measure. Its elder selector already +includes the maximum and index-(d-1) conditions; writing the untyped absolute +determinant product multiplied by that selector introduces no missing type +weight. The formula is not an unselected maximum/saddle gap density. + +For the original literal rows, + + P=(D_M+D_S)/2, D_r=(D_S-D_M)/r, + W=(12/r^3)(Delta+r e.P), + +the gradient map has determinant r^(-d), and the Delta coefficient in W is +12/r^3. The exact density Jacobian is therefore 12 r^(-(d+3)). The extra +W gradient entries do not change this block-triangular determinant. At zero +gradients W=12ell/r^3. + +There is also an exact row-level check against GAD: its secant row is +q=(f(S)-f(M))/r, and its cubic row is +(6/r^2)(D_uM+D_uS-2q). That is precisely W. The original axial P obeys +P_u=q+(r^2/12)W; all transverse average/difference rows agree. After a +permutation this remaining change is triangular with diagonal one. Thus no +unnoticed scale factor is lost when moving between the two observation +conventions, before or after integrating the common height. + +## 7. Coefficient equality and every scalar factor + +GAD (28), already for the actual selected finite H0 measure, is + + C_GAD=1/(2*12^(1/3)*L^d) integral_X integral_S^(d-1) + p_(D,He)(0,0) + E[|T_e|^(4/3)(det A)^2 1{A<0} | D=0,He=0] d omega dx. + +GAD's column components may use the frame (e,U), whereas the original uses +fixed physical coordinates. The change is orthogonal, has absolute +determinant one, and preserves the zero pin. The two zero densities agree. +Stationarity makes this integrand independent of x. The x integral is L^d +and cancels the inverse physical volume once, leaving no extra L^(-d). + +The separately proved odd/even independence and literal regression imply + + p_(D,He)(0,0)=(2pi)^(-d) a^(-d/2) (det B(e))^(-1/2), + E[|T_e|^(4/3)(det A)^2 1{A<0} | pins] + =E|N(0,V(e))|^(4/3) F_L(e). + +Integrating the scalar Gaussian over both half-axes and substituting +t^2/(2V) gives + + E|N(0,V)|^(4/3)=2^(2/3) Gamma(7/6) V^(2/3)/sqrt(pi). + +Combining with GAD's scalar uses exactly +2^(2/3)/(2*12^(1/3))=24^(-1/3). Thus + + C_GAD=Gamma(7/6)/(24^(1/3)*sqrt(pi)*(2pi)^d*a^(d/2)) + integral_S^(d-1) F_L(e)V(e)^(2/3)/sqrt(det B(e)) d omega, + +which is precisely ORIGINAL (1), and V2 M1. Every finite-L directional +dependence remains; lattice symmetry has not been replaced by continuous +rotation invariance or by a=1,b=3,c=15. + +Equivalently GAD (27) uses the positive-part moment, whose value is half +the absolute moment. That yields ORIGINAL (22)–(23) directly. Full sphere +AREA measure counts the role-directed maximum-to-saddle displacement once. +GAD's antipodal half is not permission for another unordered-pair half. +The d=2 scalar negative-cone half inside F_L(e) is a different distributional +factor and is already part of that expectation. No probability-normalized +sphere measure is used. Physical derivative units yield +length^(3d) * length^(-4(d-1)) * length^(-4)=length^(-d), the expected +per-volume coefficient units. + +GAD gives finite strict positivity for its admitted law. The exact equality +therefore also gives 00. Executable interval evaluation +needs certified effective access to its particular positive L. Nothing +here manufactures that numerical input, a verified implementation, outward +rounding evidence, finite precision/PCG Gaussian sampling, barcode-count +transfer, confidence intervals or a useful positive-lifetime error bound. +Even a future certified interval for C_alg would then enclose the ideal-law +limiting coefficient only. It would not supply M31's error functions or +connect an empirical finite-lifetime fit to that limit. + +## 9. Independent diagnostics and focused review join + +The main reviewer wrote a new exact Fraction matrix/polynomial program. +It checks the all-target finite Gaussian mean/covariance chain, the literal +gradient/gap and common-height determinants, physical midpoint Jacobian, +cubic row on monomials through degree seven, fourth/sixth moment contractions, +the regressed variance, height-pin refusal, scalar half factors and units. +The captured final run passed 58 unique checks in five refusal families, +exit 0 with empty stderr. It evaluates no original Fourier spectral sum, +Gaussian field, coefficient integral, quadrature or asymptotic limit. +Finite rational example moments diagnose polynomial identities only. + +An initial direct run also passed those 58 checks. Its exact program and +receipt remain under attempts. Two unused construction snippets were removed +before the captured run; no check or source conclusion changed. These runs +are not counted as 116 independent checks. The analytic reconstruction +above, rather than the finite diagnostic count, establishes the general +claims. + +The separately assigned focused reviewer is the actual nonauthor +`/root/c261a_successor_preparation_review_20261016`. Its complete +12199-byte hand tensor/normalization report, SHA256 +`fe2207d69495c41f40cea47b5643f72208a0ef5f6a1fe75e4af5042e462eedfb`, +returns ACCEPT_SCOPED with no must-fix for new Sections 5–6 and 8. I read +that complete report, machine verdict, seal and read extent, compared its +hand contractions with the actual new source and my own reconstruction, +and freshly matched all five input and four output seal rows. Its exact +machine is 2004 bytes / SHA256 +`e4a971fbe7cfa3b57b46a693f4a0b11ed9b2c771d32c0fc4fc5541dc66e25c8f`; +its manifest is 2823 bytes / SHA256 +`046c47ecfcc131ef8908a434708f4a849ea18e6f9733026432cc750a62dfafbc`. + +It read the entire new 445-line proof, original lines 1–386, GAD lines +1–125 and 540–739, and specified GAD-review coefficient/final-disposition +ranges. It independently derived the physical derivative sign rule, +fourth/sixth tensors, B and R contractions, raw He-only Schur covariance, +gradient regression, Gaussian 4/3 moment and full normalization ledger. +Its prior source preparation, method V1/V2 delta exposure and related +formal-interface review exposure are disclosed. It performed hand algebra +and inert identity work only; no mathematical diagnostic program, old +control, Lean, field, quadrature or coefficient digits. Organizational, +provider-distinct, blind and human independence credits remain zero. +This focused result is an actual limited review input. It does not issue +a whole-proof verdict or substitute for the law/Borel/kernel/observable +joins reconstructed in this integrated report. + +One combined local read printed the correct full focused report but then +returned exit 1 because two guessed generic machine/manifest filenames did +not exist. Listing the directory and reading their actual names resolved +that lookup. The attempt is retained in the join readback; it is neither +a source mismatch nor a mathematical proof failure. + +## 10. Final scope, retained imports and release + +Acceptance is conditional on the exact consumed GAD theorem with its explicit +classical smooth/Morse/measure imports. The finite incidence appendix explains +the needed finite all-Borel counting identity from equal-dimensional Sard, +inverse charts and ordinary change of variables. The new binding adds the +literal full Fourier realization, dense/summable admission, complete smooth +Gaussian law identity, correct null-target marginalization and exact original +tensor/coefficient reduction. It does not rely on a generic weighted Rice +citation alone or invent a new unproved global selector premise. + +The original uniform-tube, conditional Sobolev-tightness and marked-measure +sandwich arguments are not individually certified here. No effective ell +window, M31 remainder, numerical coefficient enclosure, sampler/grid/count +confidence result, diagnosis of the 113 retained failures, all-five-input +closure, distinct-model blind50 validation, Lean/kernel verification, +manuscript acceptance, external uptake or scientific promotion follows. +Unrelated numerical owners and held TA/TRIG routes remain outside this unit. + +Only the new review directory was written. No author/source/method/seal, +native page, branch, Drive object, numerical driver or status register was +modified; no old control suite, compiler, Lean, field draw, coefficient +quadrature or numerical experiment was run. The final machine verdict, +source readback and input/output seal bind the exact reviewed source and +all newly retained review evidence. Root must read this complete report +before publication or adoption. This finite review creates no continuing +source, numerical, integration or runtime reservation. Scientific effect +is NONE; the broad research goal remains ACTIVE / UNMET. diff --git a/frontiers/original_finite_l_gaussian_binding_20261017/focused-tensor-normalization/FOCUSED_TENSOR_NORMALIZATION_REVIEW.md b/frontiers/original_finite_l_gaussian_binding_20261017/focused-tensor-normalization/FOCUSED_TENSOR_NORMALIZATION_REVIEW.md new file mode 100644 index 000000000..0d9baf6f8 --- /dev/null +++ b/frontiers/original_finite_l_gaussian_binding_20261017/focused-tensor-normalization/FOCUSED_TENSOR_NORMALIZATION_REVIEW.md @@ -0,0 +1,133 @@ +# Focused nonauthor tensor and normalization review + +**Verdict: ACCEPT_SCOPED for the tensor, regressions and coefficient normalization in §§5–6 and8 of the exact new binding proof. No must-fix finding in this focused scope.** The hand derivation below verifies that, once the exact ideal Gaussian law has been admitted to the stated GAD theorem, GAD(27)/(28) specializes to the ORIGINAL algebraic coefficient with the displayed constants. This is input to the integrated review, not an independent acceptance of the entire new source-law/Borel-observable binding or a numerical qualification. + +The exact new PROOF.md is22163bytes/445lines, SHA-256 **ad9ab5afea28062dfc661d0d19b43f20e67cfd67cb6910d3cc6c3fb8b81398d8**. I read it completely. The original source is24027bytes/SHAf38c085ab5b15368e6989a1fd6298490f3b671f5a6c3a01bcdc81b516022bc33. GAD is40231bytes/SHA8d992e194f0a898e563c4c3d05b8079cde292f9e87d7ff51030ad12ef5faf319, with its exact38035-byte accepted scoped review. READ_EXTENT.json gives actual bounded rereads; no new full GAD/classical-source read is invented. + +Actual reviewer: OpenAI/Codex /root/c261a_successor_preparation_review_20261016, nonauthor of the new binding proof, GAD and ORIGINAL. I have prior source preparation/custody exposure to GAD, exact finite-L method V1/V2 delta exposure and related asymmetric/formal-interface review exposure. Those are disclosed source contacts, not blind evidence. Shared provider/principal, organizational independence0, provider-distinct0, blind0, human0. I did not author or repair this proposal. The validation here is independent hand algebra/index contraction/Gaussian integration; only inert byte/hash work was executed. + +## 1. Physical derivative tensor, including its signs + +Write K(x-y)=E cos(Q·(x-y)), where the physical wave numbers are h m, h=2pi/L, and Q has the original full product lattice distribution. Termwise differentiation is justified by the supplied summable Fourier moments. For multiindices alpha,beta of even total order2r, + + Cov(partial^alpha F,partial^beta F) + =(-1)^(|beta|+r) E Q^(alpha+beta); + +odd total order gives zero. This is just the y-derivative sign together with the derivative of cosine at zero. It gives, in particular, + + Cov(D_i,D_j)=a delta_ij, + Cov(F,H_ij)=-a delta_ij, + Cov(D_i,H_jk)=Cov(T_ijk,H_lm)=0, + Cov(D_i,T_jkl)=-M_ijkl, + Cov(H_ij,H_kl)=M_ijkl, + Cov(T_ijk,T_lmn)=M_ijklmn. + +Here a=E Q_i², b=E Q_i⁴, c=E Q_i⁶. For the fourth tensor, coordinate parity leaves only one fourfold index or two equal pairs. Thus + + M_ijkl=a²(delta_ij delta_kl+delta_ik delta_jl+delta_il delta_jk) + +kappa4 sum_s delta_is delta_js delta_ks delta_ls, + kappa4=b-3a². + +This gives b when all indices agree and a² for each two-distinct-pair case. It is the exact ORIGINAL(5), not an isotropic replacement. For completeness the sixth tensor is the sum of a³ over its15pairings, a*kappa4 over its15partitions into a fourfold block and a pair, and kappa6 on the all-equal index block. Here kappa6=c-15ab+30a³. The cumulants are those of the spectral coordinate Q, not an assumption that Q is Gaussian. + +All these field jets are jointly Gaussian. Therefore the zero cross-covariance of the entire odd block(D,T_e) with H proves actual block independence. It does not prove independence between T_e and D; that correlation is explicitly retained. The nonzero Cov(F,H) also shows why common-height pinning cannot silently be added. + +## 2. Independent contractions: B, R and the raw Schur complement + +Put Y=He, so Y_i=sum_j H_ij e_j. Contracting the three pairings of M_i j k l against e_j e_l gives respectively a²delta_ik, a²e_i e_k and a²e_i e_k. The fourth-cumulant contraction is kappa4 delta_ik e_i². Hence + + B(e)=Cov(Y)=a²(I+2ee^T)+kappa4 diag(e_i²), + +exactly ORIGINAL(6) and new(11). + +Let A^0_ab=sum_jk U_ja H_jk U_kb, where U^T U=I and U^T e=0. In Cov(A^0_ab,Y_i), the trace pairing gives a²delta_ab e_i. The other two pairings contain U_a·e or U_b·e and vanish. The fourth-cumulant term leaves + + R_ab,i=a²delta_ab e_i+kappa4 U_ia U_ib e_i. + +Contracting all four tensor slots against U gives + + C_ab,cd=a²(delta_ab delta_cd+delta_ac delta_bd+delta_ad delta_bc) + +kappa4 sum_i U_ia U_ib U_ic U_id. + +Gaussian conditioning at the literal zero target therefore gives + + Gamma_e(ab,cd)=C_ab,cd-R_ab B(e)^(-1) R_cd^T. + +This is precisely ORIGINAL(11) and new(13), including correlated diagonal/off-diagonal entries. The free indices a<=b are coordinate positions, not statistically independent Gaussian entries. No diagonalization or independence simplification of Gamma is permitted. In particular the free-entry Gaussian density is the ordinary N-dimensional density with N=(d-1)d/2; its Lebesgue element lists each A_ab once. There is no sqrt(2) Jacobian to insert into that original measure. + +For positivity, a nonzero symmetric S gives a nonzero quadratic polynomial Q^T S Q. Full lattice support cannot annihilate that polynomial, so the full Hessian covariance is positive definite. The linear coordinate map H ->(He,U^THU) is invertible on Sym_d. Its jointly positive covariance implies B>0 and Gamma_e>0 by the Schur complement. The negative cone is nonempty open and the nondegenerate centered Gaussian density is positive there; determinant-square weight is positive off its zero hypersurface and has finite Gaussian moments. Thus F(e)>0 and finite. This argument does not replace the field by a finite spectral carrier. + +## 3. Odd regression, independently expanded + +Expanding E[Q_i(Q·e)^3] by coordinate multiplicity gives + + v_i=b e_i³+3a² e_i sum_(j!=i)e_j² + =3a²e_i+kappa4 e_i³, + Cov(T_e,D_i)=-v_i. + +Write S4=sum_i e_i⁴ and S6=sum_i e_i⁶. The sixth directional moment, obtained either from the full sixth tensor or from its one/two/three-coordinate multiplicities, is + + E(Q·e)^6=15a³+15a*kappa4 S4+kappa6 S6. + +Indeed the direct multiplicity expansion is c sum e_i⁶+15ab sum_(i!=j)e_i⁴e_j²+90a³ sum_(i