OctoInformBeta

OpenAI preprint, September 2026 · §8. Quartic surface Hilbert bound

Proposition 8.1: A point-count versus quadratic Hilbert lower bound dichotomy for two-dimensional prime ideals alongside two-pole and radial estimates

Weighted Hilbert bounds and alternatives for quartic box systems.

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 two-dimensional prime ideal I⊂ℂ[X1,…,Xn]I \subset \mathbb{C}[X_1,\dots,X_n] containing four quartic relations ai2bi=q−θia_i^2 b_i = q - \theta_i (1≤i≤41 \le i \le 4) for pairwise distinct complex numbers θ1,…,θ4\theta_1,\dots,\theta_4 and polynomials b1,…,b4,q∈ℂ[X1,…,Xn]b_1,\dots,b_4, q \in \mathbb{C}[X_1,\dots,X_n], where a1,…,a4a_1,\dots,a_4 are linear forms in four distinguished coordinates. Given positive coordinate weights, how fast must the dimension of the space of weighted polynomial functions modulo II grow as the weight bound TT increases, uniformly once the degree of II is bounded?

When integer points z∈ℤnz \in \mathbb{Z}^n on the zero locus of II satisfy a1(z)a2(z)a3(z)a4(z)≠0a_1(z)a_2(z)a_3(z)a_4(z) \ne 0, one asks whether that point set can exceed the sum of its maximum coordinate-fibre and complex-ray multiplicities without forcing a quadratic lower bound in TT on the weighted dimension. The same problem splits into two algebraic questions over an arbitrary field: what quadratic lower bound holds for two-pole and four-equation systems when the shifts are not all equal, and how do the values (a1(z),…,ad(z))(a_1(z),\dots,a_d(z)) behave on integer points when dd polynomials a1,…,ada_1,\dots,a_d collapse modulo II onto scalar multiples of one linear parameter?

Weighted Spaces and Affine Invariants

The monomial weight wω(α)w_\omega(\alpha) of a nonnegative integer exponent vector α∈ℕσ\alpha \in \mathbb{N}^\sigma with respect to a real weight function ω:σ→ℝ\omega : \sigma \to \mathbb{R} on a finite variable set σ\sigma is the real number defined by (1): wω(α)=∑i∈σω(i)αi.w_\omega(\alpha) = \sum_{i \in \sigma} \omega(i)\,\alpha_i.(1)✦ Given a finite variable set σ\sigma, a real weight function ω:σ→ℝ\omega : \sigma \to \mathbb{R}, and a real threshold TT, the weighted exponent bound βω,T∈ℕσ\beta_{\omega,T} \in \mathbb{N}^\sigma is the finitely supported exponent vector assigning to each i∈σi \in \sigma the nonnegative integer floor of T/ω(i)T/\omega(i). For a finite variable set σ\sigma, a real weight function ω:σ→ℝ\omega : \sigma \to \mathbb{R}, and a real threshold TT, the set of weighted monomials ℳω(T)⊂ℕσ\mathcal{M}_\omega(T) \subset \mathbb{N}^\sigma is the finite set of nonnegative integer exponent vectors on σ\sigma associated with ω\omega and TT.✦

The space of weighted functions ℱI,ω(T)\mathcal{F}_{I,\omega}(T) for an ideal I⊂K[σ]I \subset K[\sigma] over a field KK with finite variable set σ\sigma, real weight function ω:σ→ℝ\omega : \sigma \to \mathbb{R}, and real threshold TT is the KK-linear subspace of K[σ]/IK[\sigma]/I given by (2): ℱI,ω(T)=span⁡K{Xαmod⁡I:α∈ℳω(T)}.\mathcal{F}_{I,\omega}(T) = \operatorname{span}_K \{ X^\alpha \bmod I : \alpha \in \mathcal{M}_\omega(T) \}.(2)✦ For an ideal I⊂K[σ]I \subset K[\sigma] over a field KK with finite variable set σ\sigma, real weight function ω:σ→ℝ\omega : \sigma \to \mathbb{R}, and real threshold TT, the weighted Hilbert function HI,ω(T)H_{I,\omega}(T) is the KK-dimension dim⁡KℱI,ω(T)\dim_K \mathcal{F}_{I,\omega}(T) of the space of weighted functions (2).✦

Over a commutative semiring KK, the integer-weighted polynomial space 𝒫w(T)⊂K[σ]\mathcal{P}_w(T) \subset K[\sigma] for an index set σ\sigma, a nonnegative integer weight function w:σ→ℕw : \sigma \to \mathbb{N}, and an integer bound T≥0T \ge 0 is the KK-submodule of polynomials supported on exponent vectors of ww-weighted degree at most TT.✦ The bounded-degree polynomial space VT(K[σ])⊂K[σ]V_T(K[\sigma]) \subset K[\sigma] for a field KK, variable set σ\sigma, and integer bound T≥0T \ge 0 is the KK-subspace of polynomials supported on exponent vectors of total degree at most TT.✦

For a field KK and an ideal I⊂K[X1,…,Xn]I \subset K[X_1,\dots,X_n] in n≥0n \ge 0 coordinates, the affine Hilbert polynomial PI∈ℚ[X]P_I \in \mathbb{Q}[X] is a rational univariate polynomial whose value PI(T)P_I(T) equals the KK-dimension of the image of VT(K[X1,…,Xn])V_T(K[X_1,\dots,X_n]) in K[X1,…,Xn]/IK[X_1,\dots,X_n]/I for all sufficiently large integers TT.✦ The affine dimension affdim⁡(I)\operatorname{affdim}(I) of an ideal I⊂K[X1,…,Xn]I \subset K[X_1,\dots,X_n] over a field KK with affine Hilbert polynomial PI∈ℚ[X]P_I \in \mathbb{Q}[X] is the nonnegative integer degree deg⁡PI\deg P_I of PIP_I (taken as 00 when PI=0P_I = 0).✦ For an ideal I⊂K[X1,…,Xn]I \subset K[X_1,\dots,X_n] over a field KK, the affine degree affdeg⁡(I)∈ℚ\operatorname{affdeg}(I) \in \mathbb{Q} of II is the rational number given in terms of affdim⁡(I)\operatorname{affdim}(I) and the coefficient coeff⁡affdim⁡(I)(PI)\operatorname{coeff}_{\operatorname{affdim}(I)}(P_I) of PIP_I by (3): affdeg⁡(I)=(affdim⁡(I))!⋅coeff⁡affdim⁡(I)(PI).\operatorname{affdeg}(I) = (\operatorname{affdim}(I))! \cdot \operatorname{coeff}_{\operatorname{affdim}(I)}(P_I).(3)✦

A point z=(z1,…,zn)∈ℤnz = (z_1,\dots,z_n) \in \mathbb{Z}^n on an ideal I⊂K[X1,…,Xn]I \subset K[X_1,\dots,X_n] over a field KK is an integer tuple such that every polynomial p∈Ip \in I vanishes at the image of (z1,…,zn)(z_1,\dots,z_n) in KnK^n; over K=ℂK = \mathbb{C}, this evaluation is written p(z)p(z).

Quartic Box Systems

For nonnegative integers nn and dd, a field KK, an index map x:{1,…,d}→{1,…,n}x : \{1,\dots,d\} \to \{1,\dots,n\}, and coefficients c=(c1,…,cd)∈Kdc = (c_1,\dots,c_d) \in K^d, the coordinate linear form ℓx,c∈K[X1,…,Xn]\ell_{x,c} \in K[X_1,\dots,X_n] is the polynomial defined by (4): ℓx,c=∑j=1dcjXx(j).\ell_{x,c} = \sum_{j=1}^d c_j X_{x(j)}.(4)✦

Let n,D,A,Bn, D, A, B be nonnegative integers, UU a real number, ω:{1,…,n}→ℝ\omega : \{1,\dots,n\} \to \mathbb{R} a real weight function, and w:{1,…,n}→ℕw : \{1,\dots,n\} \to \mathbb{N} an integer weight function. A quartic box system MM with parameters (n,D,A,B,U,ω,w)(n, D, A, B, U, \omega, w) consists of a characteristic-zero field K0K_0, four ℚ\mathbb{Q}-linearly independent elements v1,…,v4∈K0v_1,\dots,v_4 \in K_0, four field homomorphisms σ1,…,σ4:K0→ℂ\sigma_1,\dots,\sigma_4 : K_0 \to \mathbb{C}, four pairwise distinct complex numbers θ1,…,θ4∈ℂ\theta_1,\dots,\theta_4 \in \mathbb{C}, an index map x:{1,…,4}→{1,…,n}x : \{1,\dots,4\} \to \{1,\dots,n\} with coordinate indices xi=x(i)x_i = x(i) satisfying ω(xi)=U\omega(x_i) = U for 1≤i≤41 \le i \le 4, and polynomials b1,…,b4,q∈ℂ[X1,…,Xn]b_1,\dots,b_4, q \in \mathbb{C}[X_1,\dots,X_n] such that, writing ℓi=ℓx,(σi(v1),…,σi(v4))\ell_i = \ell_{x, (\sigma_i(v_1),\dots,\sigma_i(v_4))} as in (4), the polynomials ℓ1,…,ℓ4,b1,…,b4\ell_1,\dots,\ell_4, b_1,\dots,b_4 generate ℂ[X1,…,Xn]\mathbb{C}[X_1,\dots,X_n] as a ℂ\mathbb{C}-algebra, deg⁡ℓi≤D\deg \ell_i \le D and deg⁡q≤D\deg q \le D for 1≤i≤41 \le i \le 4, ℓi∈𝒫w(A)\ell_i \in \mathcal{P}_w(A) for 1≤i≤41 \le i \le 4, and bi∈𝒫w(B)b_i \in \mathcal{P}_w(B) for 1≤i≤41 \le i \le 4.✦

The ii-th quartic box linear form ai∈ℂ[X1,…,Xn]a_i \in \mathbb{C}[X_1,\dots,X_n] (1≤i≤41 \le i \le 4) of a quartic box system MM is ℓx,(σi(v1),…,σi(v4))\ell_{x, (\sigma_i(v_1),\dots,\sigma_i(v_4))}✦, and the ii-th quartic box equation fi∈ℂ[X1,…,Xn]f_i \in \mathbb{C}[X_1,\dots,X_n] of MM is the polynomial given by (5): fi=ai2bi−q+θi.f_i = a_i^2 b_i - q + \theta_i.(5)✦ The quartic box ideal IM⊂ℂ[X1,…,Xn]I_M \subset \mathbb{C}[X_1,\dots,X_n] of MM is the ideal generated by f1,f2,f3,f4f_1, f_2, f_3, f_4 from (5).✦

The Claim

Theorem 1. Let n≥0n \ge 0 be an integer, KK a field, I⊂K[X1,…,Xn]I \subset K[X_1,\dots,X_n] an ideal, and ω:{1,…,n}→ℝ>0\omega : \{1,\dots,n\} \to \mathbb{R}_{>0} a positive real weight function. Suppose θ1,…,θ4∈K\theta_1,\dots,\theta_4 \in K are pairwise distinct, r1,…,r4∈Kr_1,\dots,r_4 \in K are not all equal, and elements t,q,b1,…,b4∈K[X1,…,Xn]/It, q, b_1,\dots,b_4 \in K[X_1,\dots,X_n]/I satisfy (t−ri)2bi=q−θi(t - r_i)^2 b_i = q - \theta_i for 1≤i≤41 \le i \le 4, with tt and qq algebraically independent over KK. If U,V>0U, V > 0 are real numbers such that t∈ℱI,ω(U)t \in \mathcal{F}_{I,\omega}(U) and b1,…,b4∈ℱI,ω(V)b_1,\dots,b_4 \in \mathcal{F}_{I,\omega}(V), then for every real number T≥0T \ge 0, V+4U2UV2T2−(V+4U)(U+V)UV2T≤HI,ω(T).\frac{V + 4U}{2 U V^2}\,T^2 - \frac{(V + 4U)(U + V)}{U V^2}\,T \le H_{I,\omega}(T).(6)✦

Theorem 2. Let σ\sigma be a finite variable set, KK a field, I⊂K[σ]I \subset K[\sigma] an ideal, and ω:σ→ℝ>0\omega : \sigma \to \mathbb{R}_{>0} a positive real weight function. Suppose r0,r1∈Kr_0, r_1 \in K are distinct, θ0,θ1∈K\theta_0, \theta_1 \in K are scalars, and elements t,q,b0,b1,u0∈K[σ]/It, q, b_0, b_1, u_0 \in K[\sigma]/I satisfy (t−r0)2b0=q−θ0(t - r_0)^2 b_0 = q - \theta_0, (t−r1)2b1=q−θ1(t - r_1)^2 b_1 = q - \theta_1, and (t−r0)2u0=1(t - r_0)^2 u_0 = 1, with tt and qq algebraically independent over KK. If U,V>0U, V > 0 are real numbers such that t∈ℱI,ω(U)t \in \mathcal{F}_{I,\omega}(U) and b0,b1,u0∈ℱI,ω(V)b_0, b_1, u_0 \in \mathcal{F}_{I,\omega}(V), then for every real number T≥0T \ge 0, V+4U2UV2T2−(V+4U)(U+V)UV2T≤HI,ω(T).\frac{V + 4U}{2 U V^2}\,T^2 - \frac{(V + 4U)(U + V)}{U V^2}\,T \le H_{I,\omega}(T).(7)✦

Theorem 3. Let n,d≥0n, d \ge 0 be integers, I⊂ℂ[X1,…,Xn]I \subset \mathbb{C}[X_1,\dots,X_n] an ideal, a1,…,ad∈ℂ[X1,…,Xn]a_1,\dots,a_d \in \mathbb{C}[X_1,\dots,X_n] polynomials, c1,…,cd,r∈ℂc_1,\dots,c_d, r \in \mathbb{C}, and t∈ℂ[X1,…,Xn]/It \in \mathbb{C}[X_1,\dots,X_n]/I such that the residue class aimod⁡Ia_i \bmod I equals ci(t−r)c_i(t - r) in ℂ[X1,…,Xn]/I\mathbb{C}[X_1,\dots,X_n]/I for 1≤i≤d1 \le i \le d. Given an index s∈{1,…,d}s \in \{1,\dots,d\} and a point z0∈ℤnz_0 \in \mathbb{Z}^n on II with as(z0)≠0a_s(z_0) \ne 0, for every point z∈ℤnz \in \mathbb{Z}^n on II there exists a complex number u∈ℂu \in \mathbb{C} such that ai(z)=uai(z0)(1≤i≤d).a_i(z) = u\,a_i(z_0) \qquad (1 \le i \le d).(8)✦

Theorem 4. Let n,A,B,E≥0n, A, B, E \ge 0 and D>0D > 0 be integers, let U,V,L>0U, V, L > 0 be real numbers satisfying A=LUA = L U and B=LVB = L V, let ω:{1,…,n}→ℝ>0\omega : \{1,\dots,n\} \to \mathbb{R}_{>0} and w:{1,…,n}→ℤ≥1w : \{1,\dots,n\} \to \mathbb{Z}_{\ge 1} satisfy w(i)=Lω(i)w(i) = L \omega(i) for 1≤i≤n1 \le i \le n, and let MM be a quartic box system with parameters (n,D,A,B,U,ω,w)(n, D, A, B, U, \omega, w). Then there exists a real constant H≥0H \ge 0 such that for every finite set S⊂ℤnS \subset \mathbb{Z}^n, all nonnegative integers FF and RR, and every prime ideal I⊂ℂ[X1,…,Xn]I \subset \mathbb{C}[X_1,\dots,X_n] containing IMI_M such that every z∈Sz \in S is a point on II with ai(z)≠0a_i(z) \ne 0 for 1≤i≤41 \le i \le 4, every coordinate fibre {z∈S:zxi=ci for 1≤i≤4}\{z \in S : z_{x_i} = c_i \text{ for } 1 \le i \le 4\} (c=(c1,…,c4)∈ℤ4c = (c_1,\dots,c_4) \in \mathbb{Z}^4) has at most FF points, for every z0∈Sz_0 \in S the complex ray {z∈S:ai(z)=λai(z0) for 1≤i≤4 for some λ∈ℂ}\{z \in S : a_i(z) = \lambda\,a_i(z_0) \text{ for } 1 \le i \le 4 \text{ for some } \lambda \in \mathbb{C}\} has at most RR points, no aia_i vanishes modulo II, the residue class of qq modulo II is transcendental over ℂ\mathbb{C}, affdim⁡(I)=2\operatorname{affdim}(I) = 2, and affdeg⁡(I)≤E\operatorname{affdeg}(I) \le E, one has |S|≤F+Rormin⁡(min⁡(2A,B)(2A+B)L22A2B2,V+4U2UV2)T2−HT≤HI,ω(T)for all T≥1.\begin{gathered} |S| \le F + R \qquad\text{or}\\ \min\!\left(\frac{\min(2A, B)(2A + B)L^2}{2 A^2 B^2},\, \frac{V + 4U}{2 U V^2}\right) T^2 - H T \le H_{I,\omega}(T) \quad\text{for all } T \ge 1. \end{gathered}(9)✦

In the paper, the result reads: for a geometrically irreducible surface YY defined over ℚ\mathbb{Q} in the arithmetic coordinates, meeting a1a2a3a4≠0a_1 a_2 a_3 a_4 \ne 0 and satisfying ai2bi=q−θia_i^2 b_i = q - \theta_i (1≤i≤41 \le i \le 4) with qq (written nn in the paper) nonconstant and algebraic over ℚ(a1,a2,a3,a4)\mathbb{Q}(a_1, a_2, a_3, a_4), with block weights U,V>0U, V > 0 and D0=min⁡(3U,2V,U+V)D_0 = \min(3U, 2V, U+V), uniformly for bounded deg⁡Y\deg Y and fixed weights one has HY(T)≥(T2/2)min⁡((2U+V)D0/(U2V2),3/U2)−O(T)H_Y(T) \ge (T^2 / 2)\min((2U + V) D_0 / (U^2 V^2), 3 / U^2) - O(T). In (9), the surface YY, its degree deg⁡Y\deg Y, and HY(T)H_Y(T) are replaced by a prime ideal I⊂ℂ[X1,…,Xn]I \subset \mathbb{C}[X_1,\dots,X_n] of affine dimension 22 and affine degree at most EE containing the ideal of a quartic box system MM (with qq transcendental modulo II), the finite-set alternative |S|≤F+R|S| \le F + R is included, and the constants D0D_0 and 3/U23/U^2 are replaced by min⁡(2U,V)\min(2U, V) (since A=LUA = LU and B=LVB = LV) and by (V+4U)/(2UV2)(V + 4U)/(2 U V^2).

What Is Checked

The main alternative (9) is stated over nonnegative integers n,A,B,En, A, B, E, a positive integer DD, positive real parameters U,V,LU, V, L with A=LUA = LU and B=LVB = LV, positive weights ω\omega and ww satisfying w(i)=Lω(i)w(i) = L\omega(i), a quartic box system MM, a finite set S⊂ℤnS \subset \mathbb{Z}^n of points on II where a1,…,a4a_1,\dots,a_4 are nonzero and coordinate fibres and complex rays have sizes at most FF and RR, and a prime ideal I⊂ℂ[X1,…,Xn]I \subset \mathbb{C}[X_1,\dots,X_n] containing IMI_M on which a1,…,a4a_1,\dots,a_4 are nonzero modulo II, qq is transcendental over ℂ\mathbb{C} modulo II, affdim⁡(I)=2\operatorname{affdim}(I) = 2, and affdeg⁡(I)≤E\operatorname{affdeg}(I) \le E. The affine-pole bound (6) is stated over an ideal I⊂K[X1,…,Xn]I \subset K[X_1,\dots,X_n] over an arbitrary field KK, positive weights ω\omega, pairwise distinct θ1,…,θ4\theta_1,\dots,\theta_4, not all equal r1,…,r4r_1,\dots,r_4, algebraically independent t,q∈K[X1,…,Xn]/It, q \in K[X_1,\dots,X_n]/I, and b1,…,b4∈K[X1,…,Xn]/Ib_1,\dots,b_4 \in K[X_1,\dots,X_n]/I satisfying (t−ri)2bi=q−θi(t - r_i)^2 b_i = q - \theta_i with t∈ℱI,ω(U)t \in \mathcal{F}_{I,\omega}(U) and b1,…,b4∈ℱI,ω(V)b_1,\dots,b_4 \in \mathcal{F}_{I,\omega}(V) for U,V>0U, V > 0; while the two-pole bound (7) is stated over a finite variable set σ\sigma, an arbitrary field KK, an ideal I⊂K[σ]I \subset K[\sigma], positive weights ω\omega, distinct r0,r1∈Kr_0, r_1 \in K, θ0,θ1∈K\theta_0, \theta_1 \in K, algebraically independent t,q∈K[σ]/It, q \in K[\sigma]/I, and b0,b1,u0∈K[σ]/Ib_0, b_1, u_0 \in K[\sigma]/I satisfying (t−r0)2b0=q−θ0(t - r_0)^2 b_0 = q - \theta_0, (t−r1)2b1=q−θ1(t - r_1)^2 b_1 = q - \theta_1, and (t−r0)2u0=1(t - r_0)^2 u_0 = 1 with t∈ℱI,ω(U)t \in \mathcal{F}_{I,\omega}(U) and b0,b1,u0∈ℱI,ω(V)b_0, b_1, u_0 \in \mathcal{F}_{I,\omega}(V) for U,V>0U, V > 0. The radial lemma (8) is stated over an ideal I⊂ℂ[X1,…,Xn]I \subset \mathbb{C}[X_1,\dots,X_n], polynomials a1,…,ad∈ℂ[X1,…,Xn]a_1,\dots,a_d \in \mathbb{C}[X_1,\dots,X_n] with aimod⁡I=ci(t−r)a_i \bmod I = c_i(t - r), a point z0∈ℤnz_0 \in \mathbb{Z}^n on II satisfying as(z0)≠0a_s(z_0) \ne 0, and points z∈ℤnz \in \mathbb{Z}^n on II.

On the bounded-degree spaces VT(K[X1,…,Xn])V_T(K[X_1,\dots,X_n]) underlying the affine invariants, the library proves that for every ideal I⊂K[X1,…,Xn]I \subset K[X_1,\dots,X_n] over a field KK, there exist a rational polynomial P∈ℚ[X]P \in \mathbb{Q}[X] of degree at most nn and a nonnegative integer T0T_0 such that the KK-dimension of the image of VT(K[X1,…,Xn])V_T(K[X_1,\dots,X_n]) in K[X1,…,Xn]/IK[X_1,\dots,X_n]/I equals P(T)P(T) for every integer T≥T0T \ge T_0.✦