OctoInformBeta

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

Theorem 3.1: Upper bounds for integer point sets of bounded height via equation locks and weighted Hilbert functions, alongside a stopped forest power bound

Bounding integer points in weighted boxes on prime ideals.

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

Fix an integer n≥0n \ge 0 and consider a finite set S⊆ℤnS \subseteq \mathbb{Z}^n of integer points of bounded height in a weighted box, all lying on the zero locus of a complex prime ideal of bounded dimension and degree. The problem is to bound the cardinality |S||S| by a power of the scale parameter XX under a dichotomy at each intermediate dimension: every prime ideal locked on SS by equations of bounded degree either admits few points of SS or has a large weighted Hilbert function. A companion combinatorial estimate bounds a finite sum of products of branching counts that grow as powers of XX.

Kernel Geometries and Equation Locks

For an integer n≥0n \ge 0, an integral kernel geometry G=(S,H)G = (S, H) in nn variables consists of a finite set S⊆ℤnS \subseteq \mathbb{Z}^n and a real height H≥1H \ge 1 such that |zj|≤H|z_j| \le H for every point z=(z1,…,zn)∈Sz = (z_1,\dots,z_n) \in S and every 1≤j≤n1 \le j \le n.✦

For a field KK, an integer tuple z∈ℤnz \in \mathbb{Z}^n is a point on an ideal I⊆K[x1,…,xn]I \subseteq K[x_1,\dots,x_n] if every polynomial p∈Ip \in I vanishes at zz.✦

An equation lock for an ideal I⊆K[x1,…,xn]I \subseteq K[x_1,\dots,x_n] over a field KK, a finite set S⊆ℤnS \subseteq \mathbb{Z}^n, and a degree bound D∈ℤ≥0D \in \mathbb{Z}_{\ge 0} is the condition that a polynomial p∈K[x1,…,xn]p \in K[x_1,\dots,x_n] of total degree at most DD belongs to II if and only if pp vanishes at every point z∈Sz \in S on II.✦

Affine Dimension and Degree

Given a field KK, a variable set σ\sigma, and an integer T≥0T \ge 0, the space of polynomials of bounded degree K[𝒙σ]≤TK[\mathbf{x}_\sigma]_{\le T} is the KK-subspace of the polynomial ring K[𝒙σ]K[\mathbf{x}_\sigma] in variables indexed by σ\sigma consisting of polynomials supported on monomials of total degree at most TT; for nn variables x1,…,xnx_1,\dots,x_n, the subspace is written K[x1,…,xn]≤TK[x_1,\dots,x_n]_{\le T}.✦

For every field KK, integer n≥0n \ge 0, and ideal I⊆K[x1,…,xn]I \subseteq K[x_1,\dots,x_n], the Hilbert function is eventually polynomial: there exist a polynomial P∈ℚ[t]P \in \mathbb{Q}[t] of degree at most nn and an integer T0≥0T_0 \ge 0 such that for every integer T≥T0T \ge T_0, dim⁡KπI(K[x1,…,xn]≤T)=P(T),\dim_K \pi_I\bigl(K[x_1,\dots,x_n]_{\le T}\bigr) = P(T),(1)✦ where πI:K[x1,…,xn]→K[x1,…,xn]/I\pi_I : K[x_1,\dots,x_n] \to K[x_1,\dots,x_n]/I is the canonical KK-linear quotient map. The affine Hilbert polynomial PI∈ℚ[t]P_I \in \mathbb{Q}[t] of an ideal I⊆K[x1,…,xn]I \subseteq K[x_1,\dots,x_n] over a field KK is a polynomial realizing (1) for II.✦

The affine dimension dim⁡aff(I)∈ℤ≥0\dim_{\mathrm{aff}}(I) \in \mathbb{Z}_{\ge 0} of an ideal I⊆K[x1,…,xn]I \subseteq K[x_1,\dots,x_n] over a field KK is the nonnegative integer degree of its affine Hilbert polynomial PIP_I (taken to be 00 when PI=0P_I = 0).✦ The affine degree deg⁡aff(I)∈ℚ\deg_{\mathrm{aff}}(I) \in \mathbb{Q} of an ideal I⊆K[x1,…,xn]I \subseteq K[x_1,\dots,x_n] over a field KK is dim⁡aff(I)!cdim⁡aff(I)(PI)\dim_{\mathrm{aff}}(I)!\,c_{\dim_{\mathrm{aff}}(I)}(P_I), where cd(PI)c_d(P_I) is the coefficient of tdt^d in the affine Hilbert polynomial PIP_I.

Weighted Hilbert Functions

Given a finite variable set σ\sigma, a weight function ω:σ→ℝ\omega : \sigma \to \mathbb{R}, and an exponent vector α∈ℤ≥0σ\alpha \in \mathbb{Z}_{\ge 0}^\sigma, the monomial weight wω(α)∈ℝw_\omega(\alpha) \in \mathbb{R} is ∑j∈σω(j)α(j)\sum_{j \in \sigma} \omega(j)\alpha(j).✦ The weighted exponent bound βω,T∈ℤ≥0σ\beta_{\omega, T} \in \mathbb{Z}_{\ge 0}^\sigma for a weight function ω:σ→ℝ\omega : \sigma \to \mathbb{R} on a finite variable set σ\sigma and a real threshold TT is the exponent vector assigning to each j∈σj \in \sigma the nonnegative integer floor of T/ω(j)T/\omega(j).✦ The set of weighted monomials ℳ(ω,T)⊆ℤ≥0σ\mathcal{M}(\omega, T) \subseteq \mathbb{Z}_{\ge 0}^\sigma for a weight function ω:σ→ℝ\omega : \sigma \to \mathbb{R} on a finite variable set σ\sigma and a real threshold TT is a finite set of exponent vectors.

For an ideal I⊆K[𝒙σ]I \subseteq K[\mathbf{x}_\sigma] over a field KK, a weight function ω:σ→ℝ\omega : \sigma \to \mathbb{R} on a finite variable set σ\sigma, and a threshold T∈ℝT \in \mathbb{R}, the space of weighted functions WI(ω,T)W_I(\omega, T) is the real-weighted KK-subspace of K[𝒙σ]/IK[\mathbf{x}_\sigma]/I spanned over KK by the quotient classes modulo II of the monomials 𝒙α=∏j∈σxjα(j)\mathbf{x}^\alpha = \prod_{j \in \sigma} x_j^{\alpha(j)} as α\alpha ranges over ℳ(ω,T)\mathcal{M}(\omega, T).✦ The weighted Hilbert function HI(ω,T)∈ℤ≥0H_I(\omega, T) \in \mathbb{Z}_{\ge 0} of an ideal I⊆K[𝒙σ]I \subseteq K[\mathbf{x}_\sigma] over a field KK for a weight function ω:σ→ℝ\omega : \sigma \to \mathbb{R} on a finite variable set σ\sigma and a real threshold TT is the KK-dimension dim⁡KWI(ω,T)\dim_K W_I(\omega, T) of the space of weighted functions.✦

The Claim

Theorem 1. Fix integers n,r,D,e0≥0n, r, D, e_0 \ge 0 with r>0r > 0, real weights ω=(ω1,…,ωn)\omega = (\omega_1,\dots,\omega_n) with ωj>0\omega_j > 0 for 1≤j≤n1 \le j \le n, real parameters Cbox≥1C_{\mathrm{box}} \ge 1, w≥0w \ge 0, and ε>0\varepsilon > 0, real tuples a0,…,ar−1≥0a_0,\dots,a_{r-1} \ge 0 and δ0,…,δr−1>0\delta_0,\dots,\delta_{r-1} > 0 satisfying δi≤∑j=0iaj\delta_i \le \sum_{j=0}^i a_j for all 0≤i<r0 \le i < r, and real sequences c(e)≥0c(e) \ge 0 and Kex(e)≥1K_{\mathrm{ex}}(e) \ge 1 for integers e≥0e \ge 0. Then there exist constants C>0C > 0 and X0∈ℝX_0 \in \mathbb{R} with the following property. Let X>X0X > X_0, let G=(S,H)G = (S, H) be an integral kernel geometry in nn variables satisfying H≤XwH \le X^w and |zj|≤CboxXωj|z_j| \le C_{\mathrm{box}} X^{\omega_j} for all z=(z1,…,zn)∈Sz = (z_1,\dots,z_n) \in S and 1≤j≤n1 \le j \le n, and let I⊆ℂ[x1,…,xn]I \subseteq \mathbb{C}[x_1,\dots,x_n] be a prime ideal with dim⁡aff(I)≤r\dim_{\mathrm{aff}}(I) \le r and deg⁡aff(I)≤e0\deg_{\mathrm{aff}}(I) \le e_0 such that every z∈Sz \in S is a point on II. Suppose that for every 0≤i<r0 \le i < r, every integer e≥0e \ge 0, and every prime ideal J⊆ℂ[x1,…,xn]J \subseteq \mathbb{C}[x_1,\dots,x_n] containing II that admits an equation lock on SS up to degree DD, satisfies deg⁡aff(J)≤e\deg_{\mathrm{aff}}(J) \le e and dim⁡aff(J)=r−i\dim_{\mathrm{aff}}(J) = r - i, and has at least one point z∈Sz \in S on it, either |{z∈S:z is a point on J}|≤Kex(e)Xε/4|\{z \in S : z \text{ is a point on } J\}| \le K_{\mathrm{ex}}(e) X^{\varepsilon/4} or HJ(ω,T)≥Tr−i/((r−i)!δir−i)−c(e)Tr−i−1H_J(\omega, T) \ge T^{r-i} / ((r-i)!\,\delta_i^{r-i}) - c(e) T^{r-i-1} for every T≥1T \ge 1. Then |S|≤CX∑i=0r−1(r−i)ai+ε.|S| \le C\,X^{\sum_{i=0}^{r-1} (r-i)a_i + \varepsilon}.(2)✦

In the paper, the result reads: for fixed 1≤r≤n1 \le r \le n, e0e_0, CboxC_{\mathrm{box}}, weights ωj>0\omega_j > 0 for 1≤j≤n1 \le j \le n, and δr−h>0\delta_{r-h} > 0 for 1≤h≤r1 \le h \le r, if X≥2X \ge 2 and S⊆ℤnS \subseteq \mathbb{Z}^n satisfies |zj|≤CboxXωj|z_j| \le C_{\mathrm{box}} X^{\omega_j} for all z∈Sz \in S and 1≤j≤n1 \le j \le n and lies on an algebraic set ZZ of degree at most e0e_0 and dimension at most rr such that uniformly every ℚ\mathbb{Q}-irreducible subvariety A⊆ZA \subseteq Z of degree at most ee and dimension h>0h > 0 meeting SS satisfies either |S∩A|≪e,αXα|S \cap A| \ll_{e,\alpha} X^\alpha for every α>0\alpha > 0 or HA(ω,T)≥Th/(h!δr−hh)−c(e)Th−1H_A(\omega, T) \ge T^h / (h!\,\delta_{r-h}^h) - c(e) T^{h-1} for T≥1T \ge 1, then with Eh=max⁡h≤s≤rδr−sE_h = \max_{h \le s \le r} \delta_{r-s} one has |S|≪εXE1+…+Er+ε|S| \ll_\varepsilon X^{E_1 + \dots + E_r + \varepsilon} for every ε>0\varepsilon > 0, with no coefficient bound on the equations initially defining ZZ (where nn, CboxC_{\mathrm{box}}, δr−h\delta_{r-h}, c(e)c(e), and HA(ω,T)H_A(\omega, T) stand for the paper’s NN, CC, δh\delta_h, CeC_e, and HA(T)H_A(T)).

In (2), the algebraic set and ℚ\mathbb{Q}-irreducible subvarieties are replaced by prime ideals I⊆J⊆ℂ[x1,…,xn]I \subseteq J \subseteq \mathbb{C}[x_1,\dots,x_n] with an equation lock on SS up to degree DD for an integral kernel geometry of height at most XwX^w, the exceptional exponent is fixed at ε/4\varepsilon/4, the maxima max⁡h≤s≤rδr−s\max_{h \le s \le r} \delta_{r-s} are replaced by prefix-dominant increments aia_i, and the bound holds for X>X0X > X_0.

Theorem 2. Let r,L≥0r, L \ge 0 be integers, let bi≥0b_i \ge 0 be integers and Ci≥1C_i \ge 1, κi≥0\kappa_i \ge 0 be real numbers for 0≤i<r0 \le i < r, and let X≥1X \ge 1 be a real number satisfying bi≤CiXκib_i \le C_i X^{\kappa_i} for all 0≤i<r0 \le i < r. Then ∑i=0rL∏j=0i−1bj≤(r+1)L(∏j=0r−1Cj)X∑j=0r−1κj.\sum_{i=0}^{r} L \prod_{j=0}^{i-1} b_j \le (r+1)L \left(\prod_{j=0}^{r-1} C_j\right) X^{\sum_{j=0}^{r-1} \kappa_j}.(3)✦

What Is Checked

The affine counting bound (2) is stated over an integral kernel geometry G=(S,H)G = (S, H) in nn variables satisfying H≤XwH \le X^w and |zj|≤CboxXωj|z_j| \le C_{\mathrm{box}} X^{\omega_j} for all z∈Sz \in S and 1≤j≤n1 \le j \le n, prime ideals I⊆J⊆ℂ[x1,…,xn]I \subseteq J \subseteq \mathbb{C}[x_1,\dots,x_n] (with every point of SS lying on II, dim⁡aff(I)≤r\dim_{\mathrm{aff}}(I) \le r, deg⁡aff(I)≤e0\deg_{\mathrm{aff}}(I) \le e_0, and each JJ having an equation lock on SS at degree DD, dim⁡aff(J)=r−i\dim_{\mathrm{aff}}(J) = r - i, deg⁡aff(J)≤e\deg_{\mathrm{aff}}(J) \le e, and at least one point of SS on it), the weighted Hilbert function HJ(ω,T)H_J(\omega, T) defined as the ℂ\mathbb{C}-dimension of the space of weighted functions WJ(ω,T)W_J(\omega, T) generated by the weighted monomials ℳ(ω,T)\mathcal{M}(\omega, T), and parameters r>0r > 0, ωj>0\omega_j > 0, Cbox≥1C_{\mathrm{box}} \ge 1, w≥0w \ge 0, ε>0\varepsilon > 0, ai≥0a_i \ge 0, δi>0\delta_i > 0 with δi≤∑j=0iaj\delta_i \le \sum_{j=0}^i a_j, c(e)≥0c(e) \ge 0, and Kex(e)≥1K_{\mathrm{ex}}(e) \ge 1 under the dichotomy on point counts and weighted Hilbert functions at each intermediate dimension r−ir - i. The stopped forest power bound (3) is stated over integers r,L≥0r, L \ge 0 and bi≥0b_i \ge 0, real numbers Ci≥1C_i \ge 1 and κi≥0\kappa_i \ge 0 for 0≤i<r0 \le i < r, and X≥1X \ge 1 satisfying bi≤CiXκib_i \le C_i X^{\kappa_i}. Beyond (2) and (3), the eventual polynomiality (1) of the Hilbert function holds for every field KK, integer n≥0n \ge 0, and ideal I⊆K[x1,…,xn]I \subseteq K[x_1,\dots,x_n] on the spaces K[x1,…,xn]≤TK[x_1,\dots,x_n]_{\le T} of polynomials of bounded degree.