OctoInformBeta

OpenAI preprint, September 2026 · §3. Affine counting and integer patches

3.4: Deletion estimate: Tree-stopping and deletion-average bounds alongside adaptive and physical block selections keeping bad-prime fractions at most ξ\xi on explicit point shares

Deterministic prime blocks control minor divisors across adaptive stages.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Setting

Take a finite set S⊆ℤnS \subseteq \mathbb{Z}^n (n∈ℤ≥0n \in \mathbb{Z}_{\ge 0}) of bounded coordinate height together with an ideal and a local system of polynomial equations around each point. Requiring point coordinates to be congruent modulo primes pp restricts SS to smaller subsets on which low-degree polynomials vanish, driving an rr-stage adaptive transition (r∈ℤ≥0r \in \mathbb{Z}_{\ge 0}) that produces a sequence of Jacobian minor integers at each point z∈Sz \in S.

The problem is to select deterministic prime blocks B0,…,Br−1B_0, \dots, B_{r-1} from given prime pools so that, for a large share of points z∈Sz \in S and every block BiB_i, at most a prescribed fraction ξ>0\xi > 0 of the primes in BiB_i divide the product of those minor values along the adaptive path from zz. Accompanying this block-selection problem in an abstract vector-space setting and in the geometric setting are an averaging estimate for bad elements under subset deletion and a cardinality bound for sets fibered over a bounded-degree tree.

Combinatorial Definitions

For a set Σ\Sigma, a finite subset R0⊆ΣR_0 \subseteq \Sigma of root nodes, a map cc assigning a finite subset c(i,s)⊆Σc(i, s) \subseteq \Sigma to each i∈ℤ≥0i \in \mathbb{Z}_{\ge 0} and s∈Σs \in \Sigma, and an index i∈ℤ≥0i \in \mathbb{Z}_{\ge 0}, the ii-th tree level Li(R0,c)L_i(R_0, c) of R0R_0 and cc is the finite subset of Σ\Sigma defined recursively by (1): L0(R0,c)=R0,Li+1(R0,c)=⋃s∈Li(R0,c)c(i,s).L_0(R_0, c) = R_0, \qquad L_{i+1}(R_0, c) = \bigcup_{s \in L_i(R_0, c)} c(i, s).(1)✦

Given a finite set PP, an integer t∈ℤ≥0t \in \mathbb{Z}_{\ge 0}, and a map bad\mathrm{bad} assigning a finite set bad(S)\mathrm{bad}(S) to each finite set SS, the bad fraction β(P,t,bad)\beta(P, t, \mathrm{bad}) is the real number defined by (2): β(P,t,bad)=1t(|P|t)∑S⊆P|S|=t|S∩bad(S)|,\beta(P, t, \mathrm{bad}) = \frac{1}{t\binom{|P|}{t}} \sum_{\substack{S \subseteq P \\ |S| = t}} |S \cap \mathrm{bad}(S)|,(2)✦ with β(P,t,bad)=0\beta(P, t, \mathrm{bad}) = 0 when t=0t = 0 or t>|P|t > |P|.

Fix a field KK, a KK-vector space VV, a state set Σ\Sigma, a parameter set 𝒜\mathcal{A}, a map WW assigning a KK-subspace W(i,s,A)⊆VW(i, s, A) \subseteq V to each i∈ℤ≥0i \in \mathbb{Z}_{\ge 0}, s∈Σs \in \Sigma, and A∈𝒜A \in \mathcal{A}, and a transition map next\mathrm{next} assigning a state next(i,s,U)∈Σ\mathrm{next}(i, s, U) \in \Sigma to each i∈ℤ≥0i \in \mathbb{Z}_{\ge 0}, s∈Σs \in \Sigma, and KK-subspace U⊆VU \subseteq V. For m,i∈ℤ≥0m, i \in \mathbb{Z}_{\ge 0}, s∈Σs \in \Sigma, and A∈𝒜A \in \mathcal{A}, the adaptive state sequence 𝒮m(W,next,i,s,A)\mathcal{S}_m(W, \mathrm{next}, i, s, A) is the (m+1)(m+1)-tuple of states in Σ\Sigma defined recursively by 𝒮0(W,next,i,s,A)=(s)\mathcal{S}_0(W, \mathrm{next}, i, s, A) = (s) and concatenation (3): 𝒮m+1(W,next,i,s,A)=(s,𝒮m(W,next,i+1,next(i,s,W(i,s,A)),A)).\mathcal{S}_{m+1}(W, \mathrm{next}, i, s, A) = \bigl(s,\, \mathcal{S}_m\bigl(W, \mathrm{next}, i + 1, \mathrm{next}(i, s, W(i, s, A)), A\bigr)\bigr).(3)✦

Polynomial Ideals and Patches

Over a field KK in n∈ℤ≥0n \in \mathbb{Z}_{\ge 0} variables 𝒙=(xl)0≤l<n\mathbf{x} = (x_l)_{0 \le l < n}, for each T∈ℤ≥0T \in \mathbb{Z}_{\ge 0}, the space of polynomials of bounded degree VT(n,K)V_T(n, K) is the KK-subspace of K[𝒙]K[\mathbf{x}] consisting of polynomials of total degree at most TT.✦ For an ideal I⊆K[𝒙]I \subseteq K[\mathbf{x}] and the quotient map πI:K[𝒙]→K[𝒙]/I\pi_I : K[\mathbf{x}] \to K[\mathbf{x}]/I, the affine Hilbert polynomial PIP_I of II is the polynomial in ℚ[t]\mathbb{Q}[t] of degree at most nn such that dim⁡KπI(VT(n,K))=PI(T)\dim_K \pi_I(V_T(n, K)) = P_I(T) for all sufficiently large T∈ℤ≥0T \in \mathbb{Z}_{\ge 0}.✦ For an ideal I⊆K[𝒙]I \subseteq K[\mathbf{x}] with affine Hilbert polynomial PIP_I, the affine dimension dim⁡affI\dim_{\mathrm{aff}} I of II is the degree of PIP_I (with deg⁡0=0\deg 0 = 0)✦, and the affine degree deg⁡affI\deg_{\mathrm{aff}} I of II is the rational number defined by (4): deg⁡affI=(dim⁡affI)!coeffdim⁡affI(PI),\deg_{\mathrm{aff}} I = (\dim_{\mathrm{aff}} I)!\, \mathrm{coeff}_{\dim_{\mathrm{aff}} I}(P_I),(4)✦ where coeffk(PI)\mathrm{coeff}_k(P_I) denotes the coefficient of tkt^k in PIP_I.

An integer patch P=(F,(Ei)i∉F)P = (F, (E_i)_{i \notin F}) in nn variables consists of a subset F⊆{0,…,n−1}F \subseteq \{0, \dots, n-1\} of free coordinates and a defining polynomial Ei∈ℤ[𝒙]E_i \in \mathbb{Z}[\mathbf{x}] for each non-free coordinate i∈{0,…,n−1}\Fi \in \{0, \dots, n-1\} \setminus F.✦ The patch Jacobian minor ΔP\Delta_P of an integer patch P=(F,(Ei)i∉F)P = (F, (E_i)_{i \notin F}) in nn variables is the polynomial in ℤ[𝒙]\mathbb{Z}[\mathbf{x}] defined by the determinant (5): ΔP=det⁡(∂Ei∂xj)i,j∉F.\Delta_P = \det\left(\frac{\partial E_i}{\partial x_j}\right)_{i, j \notin F}.(5)✦ An integer patch P=(F,(Ei)i∉F)P = (F, (E_i)_{i \notin F}) in nn variables is a patch on an ideal I⊆K[𝒙]I \subseteq K[\mathbf{x}] if |F|=dim⁡affI|F| = \dim_{\mathrm{aff}} I, the image of every EiE_i (i∉Fi \notin F) under the coefficient homomorphism ℤ→K\mathbb{Z} \to K belongs to II, and the image of its patch Jacobian minor ΔP\Delta_P under ℤ→K\mathbb{Z} \to K does not belong to II.✦

For N∈ℤ≥0N \in \mathbb{Z}_{\ge 0} and H∈ℝH \in \mathbb{R}, the integral kernel height bound ℋ(N,H)\mathcal{H}(N, H) is the real number defined by (6): ℋ(N,H)=(1+N2H2)N!(1+NH2)N.\mathcal{H}(N, H) = (1 + N^2 H^2)\, N!\, (1 + N H^2)^N.(6)✦ An integer patch P=(F,(Ei)i∉F)P = (F, (E_i)_{i \notin F}) in nn variables is bounded with degree bound D∈ℤ≥0D \in \mathbb{Z}_{\ge 0} and height parameter H∈ℝH \in \mathbb{R} if there exists d∈ℤ≥0d \in \mathbb{Z}_{\ge 0} with d≤Dd \le D such that every EiE_i (i∉Fi \notin F) has total degree at most dd and every integer coefficient cc of EiE_i satisfies |c|≤ℋ((d+nn),Hd)|c| \le \mathcal{H}\bigl(\binom{d+n}{n}, H^d\bigr).✦

Bounded Point Sets and Adaptive Paths

An integral kernel geometry G=(SG,HG)G = (S_G, H_G) in nn variables consists of a finite set SG⊆ℤnS_G \subseteq \mathbb{Z}^n and a real height HG≥1H_G \ge 1 such that |zl|≤HG|z_l| \le H_G for every z∈SGz \in S_G and every 0≤l<n0 \le l < n.✦ For an ideal I⊆K[𝒙]I \subseteq K[\mathbf{x}], an integer tuple z∈ℤnz \in \mathbb{Z}^n is a point on II if evaluating every p∈Ip \in I at the image of zz in KnK^n yields 00.✦

A patch piece over a field KK is a pair (I,P)(I, P) consisting of an ideal I⊆K[𝒙]I \subseteq K[\mathbf{x}] and an integer patch PP in nn variables; the patch piece (I,P)(I, P) covers a point z∈ℤnz \in \mathbb{Z}^n if zz is a point on II and ΔP(z)≠0\Delta_P(z) \ne 0 in ℤ\mathbb{Z}.✦ For h,e∈ℤ≥0h, e \in \mathbb{Z}_{\ge 0}, a prime ideal I⊆ℂ[𝒙]I \subseteq \mathbb{C}[\mathbf{x}] is a kernel parent ideal for (h,e)(h, e) if dim⁡affI≤h\dim_{\mathrm{aff}} I \le h and deg⁡affI≤e\deg_{\mathrm{aff}} I \le e.✦ Given an integral kernel geometry G=(SG,HG)G = (S_G, H_G), integers h,e∈ℤ≥0h, e \in \mathbb{Z}_{\ge 0}, and z∈ℤnz \in \mathbb{Z}^n, a complex patch piece (I,P)(I, P) is a physical state at zz for (G,h,e)(G, h, e) if II is a kernel parent ideal for (h,e)(h, e), PP is a patch on II bounded with degree bound ee and height parameter HGH_G, and (I,P)(I, P) covers zz.✦

Given r∈ℤ≥0r \in \mathbb{Z}_{\ge 0}, a tuple B=(B0,…,Br−1)B = (B_0, \dots, B_{r-1}) of finite subsets of ℤ≥0\mathbb{Z}_{\ge 0}, and i∈ℤ≥0i \in \mathbb{Z}_{\ge 0}, the cumulative block primes Q≤i(B)Q_{\le i}(B) of BB up to stage ii is the subset of ℤ≥0\mathbb{Z}_{\ge 0} defined by Q≤i(B)=⋃0≤j≤i,j<rBjQ_{\le i}(B) = \bigcup_{0 \le j \le i,\; j < r} B_j.✦ For a finite set S⊆ℤnS \subseteq \mathbb{Z}^n, an ideal I⊆K[𝒙]I \subseteq K[\mathbf{x}], a point z∈ℤnz \in \mathbb{Z}^n, and a finite set Q⊆ℤ≥0Q \subseteq \mathbb{Z}_{\ge 0}, the evaluation cell 𝒞(S,I,z,Q)\mathcal{C}(S, I, z, Q) of S,I,zS, I, z, and QQ is the subset of SS defined by (7): 𝒞(S,I,z,Q)={w∈S:w is a point on I and q∣(wl−zl) for all q∈Q,0≤l<n}.\mathcal{C}(S, I, z, Q) = \{w \in S : w \text{ is a point on } I \text{ and } q \mid (w_l - z_l) \text{ for all } q \in Q,\; 0 \le l < n\}.(7)✦ For S⊆ℤnS \subseteq \mathbb{Z}^n and D∈ℤ≥0D \in \mathbb{Z}_{\ge 0}, the evaluation space ED(S)E_D(S) of SS is the ℂ\mathbb{C}-subspace of VD(n,ℂ)V_D(n, \mathbb{C}) consisting of polynomials vanishing on SS✦, and for d,E∈ℤ≥0d, E \in \mathbb{Z}_{\ge 0}, the staged evaluation space Vd,E(S)V_{d, E}(S) of SS is the ℂ\mathbb{C}-subspace of VE(n,ℂ)V_E(n, \mathbb{C}) defined by (8): Vd,E(S)={p∈EE(S):deg⁡p≤d}.V_{d, E}(S) = \{p \in E_E(S) : \deg p \le d\}.(8)✦

Specializing adaptive sequences to an integral kernel geometry G=(SG,HG)G = (S_G, H_G), a stage-degree map d:ℤ≥0→ℤ≥0d : \mathbb{Z}_{\ge 0} \to \mathbb{Z}_{\ge 0}, integers r,A,h,e,E∈ℤ≥0r, A, h, e, E \in \mathbb{Z}_{\ge 0}, a point z∈ℤnz \in \mathbb{Z}^n, an index i∈ℤ≥0i \in \mathbb{Z}_{\ge 0}, a complex patch piece s=(I,P)s = (I, P), and an rr-tuple of finite blocks B=(B0,…,Br−1)B = (B_0, \dots, B_{r-1}), the physical kernel subspace WG,d,E,zphys(i,s,B)W^{\mathrm{phys}}_{G, d, E, z}(i, s, B) is the ℂ\mathbb{C}-subspace of VE(n,ℂ)V_E(n, \mathbb{C}) defined by Vd(i),E(𝒞(SG,I,z,Q≤i(B)))V_{d(i), E}(\mathcal{C}(S_G, I, z, Q_{\le i}(B))).✦ At stage ii, the physical transition nextG,d,A,h,e,zphys(i,s,W)\mathrm{next}^{\mathrm{phys}}_{G, d, A, h, e, z}(i, s, W) sends ss and a subspace W⊆VE(n,ℂ)W \subseteq V_E(n, \mathbb{C}) to a covering child patch piece for zz chosen from a finite list of child patch pieces associated to GG, did_i, AA, hh, a recursively determined stage-ii degree budget, II, and WW. Given also m∈ℤ≥0m \in \mathbb{Z}_{\ge 0} and a map start\mathrm{start} assigning a complex patch piece start(z)\mathrm{start}(z) to each z∈SGz \in S_G, the physical path product Πmphys(G,d,A,h,e,E,start,B,z)\Pi^{\mathrm{phys}}_m(G, d, A, h, e, E, \mathrm{start}, B, z) is the integer defined by (9): Πmphys(G,d,A,h,e,E,start,B,z)=∏k=0mΔPk(z),\Pi^{\mathrm{phys}}_m(G, d, A, h, e, E, \mathrm{start}, B, z) = \prod_{k=0}^m \Delta_{P_k}(z),(9)✦ where ((I0,P0),…,(Im,Pm))=𝒮m(WG,d,E,zphys,nextG,d,A,h,e,zphys,0,start(z),B)((I_0, P_0), \dots, (I_m, P_m)) = \mathcal{S}_m\bigl(W^{\mathrm{phys}}_{G, d, E, z}, \mathrm{next}^{\mathrm{phys}}_{G, d, A, h, e, z}, 0, \mathrm{start}(z), B\bigr). For ξ∈ℝ\xi \in \mathbb{R}, a point z∈SGz \in S_G is physically regular for (B,ξ)(B, \xi) with mm adaptive steps (relative to G,d,A,h,e,E,startG, d, A, h, e, E, \mathrm{start}) if (10) holds for every 0≤j<r0 \le j < r: 1|Bj||{q∈Bj:q∣Πmphys(G,d,A,h,e,E,start,B,z)}|≤ξ,\frac{1}{|B_j|}\bigl|\bigl\{q \in B_j : q \mid \Pi^{\mathrm{phys}}_m(G, d, A, h, e, E, \mathrm{start}, B, z)\bigr\}\bigr| \le \xi,(10)✦ reading the fraction as 00 when Bj=∅B_j = \emptyset.

The Claim

Theorem 1. (1) Suppose KK is a field, VV is a finite-dimensional KK-vector space, Σ\Sigma is a state set, SS is a finite point set, r,T,m∈ℤ≥0r, T, m \in \mathbb{Z}_{\ge 0}, P0,…,Pr−1P_0, \dots, P_{r-1} are finite subsets of ℤ≥0\mathbb{Z}_{\ge 0}, and ℓ0,…,ℓr−1∈ℤ≥0\ell_0, \dots, \ell_{r-1} \in \mathbb{Z}_{\ge 0}. For each z∈Sz \in S, let WzW_z assign a KK-subspace Wz(j,s,A)⊆VW_z(j, s, A) \subseteq V to each j∈ℤ≥0j \in \mathbb{Z}_{\ge 0}, s∈Σs \in \Sigma, and rr-tuple A=(A0,…,Ar−1)A = (A_0, \dots, A_{r-1}) of finite subsets of ℤ≥0\mathbb{Z}_{\ge 0}, let nextz\mathrm{next}_z assign a state nextz(j,s,U)∈Σ\mathrm{next}_z(j, s, U) \in \Sigma to each j∈ℤ≥0j \in \mathbb{Z}_{\ge 0}, s∈Σs \in \Sigma, and KK-subspace U⊆VU \subseteq V, let start(z)∈Σ\mathrm{start}(z) \in \Sigma, and let minorz:Σ→ℤ\mathrm{minor}_z : \Sigma \to \mathbb{Z}. Let X,ζ,Λ,η,ξ∈ℝX, \zeta, \Lambda, \eta, \xi \in \mathbb{R} satisfy X>1X > 1, ζ>0\zeta > 0, ξ>0\xi > 0, T>0T > 0, and ℓi+T<|Pi|\ell_i + T < |P_i| for all 0≤i<r0 \le i < r. Write (t0(z,A),…,tm(z,A))=𝒮m(Wz,nextz,0,start(z),A)(t_0(z, A), \dots, t_m(z, A)) = \mathcal{S}_m(W_z, \mathrm{next}_z, 0, \mathrm{start}(z), A), and assume that: for all z∈Sz \in S, j∈ℤ≥0j \in \mathbb{Z}_{\ge 0}, s∈Σs \in \Sigma, rr-tuple BB, 0≤i<r0 \le i < r, and finite sets A′⊆A″⊆ℤ≥0A' \subseteq A'' \subseteq \mathbb{Z}_{\ge 0}, replacing the ii-th entry of BB by A′A' yields a subspace of Wz(j,s,B)W_z(j, s, B) with ii-th entry A″A''; every p∈Pip \in P_i (0≤i<r0 \le i < r) is prime with p≥Xζp \ge X^\zeta; for every z∈Sz \in S and every A=(A0,…,Ar−1)A = (A_0, \dots, A_{r-1}) with Ai⊆PiA_i \subseteq P_i, each 0≤k≤m0 \le k \le m satisfies minorz(tk(z,A))≠0\mathrm{minor}_z(t_k(z, A)) \ne 0 and |minorz(tk(z,A))|≤XΛ|\mathrm{minor}_z(t_k(z, A))| \le X^\Lambda; and (dim⁡KV+1)m/T+⌊Λ(m+1)/ζ⌋+/(|Pi|−(ℓi+T))≤η(\dim_K V + 1)^m / T + \lfloor \Lambda(m + 1)/\zeta \rfloor_+ / (|P_i| - (\ell_i + T)) \le \eta for all 0≤i<r0 \le i < r, where ⌊x⌋+=max⁡(0,⌊x⌋)\lfloor x \rfloor_+ = \max(0, \lfloor x \rfloor). Then there exists an rr-tuple A=(A0,…,Ar−1)A = (A_0, \dots, A_{r-1}) with Ai⊆PiA_i \subseteq P_i and ℓi<|Ai|≤ℓi+T\ell_i < |A_i| \le \ell_i + T for all 0≤i<r0 \le i < r satisfying (11): |{z∈S:∀0≤i<r,1|Ai||{p∈Ai:p∣∏k=0mminorz(tk(z,A))}|≤ξ}|≥(1−rηξ)|S|.\left|\left\{z \in S : \forall\, 0 \le i < r,\; \frac{1}{|A_i|}\left|\left\{p \in A_i : p \mid \prod_{k=0}^m \mathrm{minor}_z(t_k(z, A))\right\}\right| \le \xi\right\}\right| \ge \left(1 - \frac{r\eta}{\xi}\right)|S|.(11)✦

(2) Given n,A,h,e,m,E∈ℤ≥0n, A, h, e, m, E \in \mathbb{Z}_{\ge 0} and d:ℤ≥0→ℤ≥0d : \mathbb{Z}_{\ge 0} \to \mathbb{Z}_{\ge 0}, there exist C∈ℝC \in \mathbb{R} with C≥1C \ge 1 and b∈ℤ≥0b \in \mathbb{Z}_{\ge 0} with the following property. Let G=(SG,HG)G = (S_G, H_G) be an integral kernel geometry in nn variables, let start\mathrm{start} assign a physical state start(z)\mathrm{start}(z) at zz for (G,h,e)(G, h, e) to each z∈SGz \in S_G, let r,T∈ℤ≥0r, T \in \mathbb{Z}_{\ge 0}, let P0,…,Pr−1P_0, \dots, P_{r-1} be finite sets of primes, let ℓ0,…,ℓr−1∈ℤ≥0\ell_0, \dots, \ell_{r-1} \in \mathbb{Z}_{\ge 0}, and let X,ζ,w,η,ξ∈ℝX, \zeta, w, \eta, \xi \in \mathbb{R}. Assume X>1X > 1, X≥CX \ge C, w≥0w \ge 0, HG≤XwH_G \le X^w, ζ>0\zeta > 0, ξ>0\xi > 0, T>0T > 0, p≥Xζp \ge X^\zeta for all 0≤i<r0 \le i < r and p∈Pip \in P_i, ℓi+T<|Pi|\ell_i + T < |P_i| for all 0≤i<r0 \le i < r, and ((E+nn)+1)m/T+⌊(wb+1)(m+1)/ζ⌋/(|Pi|−(ℓi+T))≤η\bigl(\binom{E+n}{n} + 1\bigr)^m / T + \lfloor (wb + 1)(m + 1)/\zeta \rfloor / (|P_i| - (\ell_i + T)) \le \eta for all 0≤i<r0 \le i < r. Then there exists B=(B0,…,Br−1)B = (B_0, \dots, B_{r-1}) with Bi⊆PiB_i \subseteq P_i and ℓi<|Bi|≤ℓi+T\ell_i < |B_i| \le \ell_i + T for all 0≤i<r0 \le i < r satisfying (12): |{z∈SG:z is physically regular for (B,ξ) with m adaptive steps}|≥(1−rηξ)|SG|.\bigl|\{z \in S_G : z \text{ is physically regular for } (B, \xi) \text{ with } m \text{ adaptive steps}\}\bigr| \ge \left(1 - \frac{r\eta}{\xi}\right)|S_G|.(12)✦

(3) Given n,E,A,h,e∈ℤ≥0n, E, A, h, e \in \mathbb{Z}_{\ge 0}, r∈ℤ>0r \in \mathbb{Z}_{>0}, d:ℤ≥0→ℤ≥0d : \mathbb{Z}_{\ge 0} \to \mathbb{Z}_{\ge 0}, reals a0,…,ar−1≥0a_0, \dots, a_{r-1} \ge 0, and reals γ>0\gamma > 0, ξ>0\xi > 0, w≥0w \ge 0, there exist ζ>0\zeta > 0, integers U0,…,Ur−1∈ℤ≥0U_0, \dots, U_{r-1} \in \mathbb{Z}_{\ge 0} with ζUi≤ai+2γ\zeta U_i \le a_i + 2\gamma for all 0≤i<r0 \le i < r, and real constants K≥1K \ge 1 and X0X_0 such that for every real X>X0X > X_0, every integral kernel geometry G=(SG,HG)G = (S_G, H_G) in nn variables with HG≤XwH_G \le X^w, and every map start\mathrm{start} assigning a physical state start(z)\mathrm{start}(z) at zz for (G,h,e)(G, h, e) to each z∈SGz \in S_G, there exist pairwise disjoint finite sets of primes B=(B0,…,Br−1)B = (B_0, \dots, B_{r-1}) satisfying |Bi|≤Ui|B_i| \le U_i and ai+γ≤ζ|Bi|a_i + \gamma \le \zeta |B_i| for all 0≤i<r0 \le i < r, Xζ≤p≤KXζX^\zeta \le p \le K X^\zeta for every 0≤i<r0 \le i < r and p∈Bip \in B_i, and (13): |{z∈SG:z is physically regular for (B,ξ) with r adaptive steps}|≥|SG|2.\bigl|\{z \in S_G : z \text{ is physically regular for } (B, \xi) \text{ with } r \text{ adaptive steps}\}\bigr| \ge \frac{|S_G|}{2}.(13)✦

(4) Let PP be a finite set, let ℓ,T∈ℤ≥0\ell, T \in \mathbb{Z}_{\ge 0} satisfy T>0T > 0 and ℓ+T<|P|\ell + T < |P|, let RR assign a nonnegative integer R(S)R(S) to each finite set SS, let bad\mathrm{bad} assign a finite set bad(S)\mathrm{bad}(S) to each finite set SS, and let M,D∈ℤ≥0M, D \in \mathbb{Z}_{\ge 0}. If for every S⊆PS \subseteq P one has R(S)≤MR(S) \le M, |P∩bad(S)|≤D|P \cap \mathrm{bad}(S)| \le D, and for every q∈Sq \in S both R(S)≤R(S\{q})R(S) \le R(S \setminus \{q\}) and whenever R(S)=R(S\{q})R(S) = R(S \setminus \{q\}) one has bad(S)=bad(S\{q})\mathrm{bad}(S) = \mathrm{bad}(S \setminus \{q\}), then (14) holds: 1T∑j=0T−1β(P,ℓ+j+1,bad)≤MT+D|P|−(ℓ+T).\frac{1}{T}\sum_{j=0}^{T-1} \beta(P, \ell + j + 1, \mathrm{bad}) \le \frac{M}{T} + \frac{D}{|P| - (\ell + T)}.(14)✦

(5) Let SS be a finite set, let R0⊆ΣR_0 \subseteq \Sigma be a finite root set, let cc assign a finite subset c(i,s)⊆Σc(i, s) \subseteq \Sigma to each i∈ℤ≥0i \in \mathbb{Z}_{\ge 0} and s∈Σs \in \Sigma, let (Bi)i≥0(B_i)_{i \ge 0} be a sequence of integers with Bi≥1B_i \ge 1 for all i∈ℤ≥0i \in \mathbb{Z}_{\ge 0}, let r,L∈ℤ≥0r, L \in \mathbb{Z}_{\ge 0}, and let depth:S→ℤ≥0\mathrm{depth} : S \to \mathbb{Z}_{\ge 0} and node:S→Σ\mathrm{node} : S \to \Sigma. If |c(i,s)|≤Bi|c(i, s)| \le B_i for all i∈ℤ≥0i \in \mathbb{Z}_{\ge 0} and s∈Li(R0,c)s \in L_i(R_0, c), depth(z)≤r\mathrm{depth}(z) \le r and node(z)∈Ldepth(z)(R0,c)\mathrm{node}(z) \in L_{\mathrm{depth}(z)}(R_0, c) for all z∈Sz \in S, and |{z∈S:depth(z)=i and node(z)=s}|≤L|\{z \in S : \mathrm{depth}(z) = i \text{ and } \mathrm{node}(z) = s\}| \le L for all 0≤i≤r0 \le i \le r and s∈Li(R0,c)s \in L_i(R_0, c), then (15) holds: |S|≤(r+1)|R0|(∏j=0r−1Bj)L.|S| \le (r + 1)\,|R_0|\,\left(\prod_{j=0}^{r-1} B_j\right) L.(15)✦

In the paper, the result reads: fixing r∈ℤ>0r \in \mathbb{Z}_{>0}, an index 0≤i<r0 \le i < r, a point z∈Sz \in S, all blocks BjB_j for j≠ij \ne i, and a prime pool PiP_i with |Pi|→∞|P_i| \to \infty consisting of primes in [Xζ,CrXζ][X^\zeta, C_r X^\zeta] (for a constant CrC_r depending on rr) disjoint from those blocks, if tt is chosen uniformly among TT consecutive integers (where TT stands for the paper’s T0T_0) and BiB_i is chosen uniformly among tt-element subsets of PiP_i, then the expected fraction of primes in BiB_i dividing at least one full-path integer Δ0(z),…,Δr(z)\Delta_0(z), \dots, \Delta_r(z) is at most M/T+o(1)M/T + o(1) uniformly in zz and the remaining blocks (where MM stands for the paper’s B0B_0). Rather than bounding a single-block expectation over full-path integers Δ0(z),…,Δr(z)\Delta_0(z), \dots, \Delta_r(z) as |Pi|→∞|P_i| \to \infty, the library states explicit finite-parameter inequalities in (14) and (15) and selects deterministic blocks simultaneously across all rr stages in (11) for abstract adaptive minor products and in (12) and (13) for physical patch states.

What Is Checked

The bound (15) is stated over a finite set SS with depth and node maps into the tree levels of a finite root set R0⊆ΣR_0 \subseteq \Sigma and bounded-size child maps c(i,⋅)c(i, \cdot) (Bi≥1B_i \ge 1) with uniform fiber bound LL, while (14) is stated over a finite set PP, ℓ+T<|P|\ell + T < |P| (T>0T > 0), a bounded monotone rank function RR, and a map bad\mathrm{bad} invariant under rank-preserving single-element deletions and meeting PP in at most DD elements. Across adaptive stages, (11) is stated over a finite point set SS, a finite-dimensional KK-vector space VV, prime pools PiP_i with primes at least XζX^\zeta, monotone subspace families WzW_z, transition maps nextz\mathrm{next}_z, nonzero bounded integer minor maps minorz\mathrm{minor}_z, and the dimension and pool-size inequality for η\eta, while (12) and (13) are stated over integral kernel geometries G=(SG,HG)G = (S_G, H_G) in nn variables of height HG≤XwH_G \le X^w, assignments of physical states at every z∈SGz \in S_G, and physical regularity with respect to staged evaluation spaces on prime-congruence cells and physical transition maps.

Beyond the main claims, the library proves that for every polynomial ideal I⊆K[𝒙]I \subseteq K[\mathbf{x}] in nn variables over a field KK, the Hilbert function is eventually polynomial: there exist a rational-coefficient polynomial PP of degree at most nn and an integer cutoff T0≥0T_0 \ge 0 such that dim⁡KπI(VT(n,K))=P(T)\dim_K \pi_I(V_T(n, K)) = P(T) for all T≥T0T \ge T_0.✦