OctoInformBeta

OpenAI preprint, September 2026 · §6. From curve defects to matrix codimension

Lemma 6.3: For distinct diagonal coefficients in size k+1≥2k+1\geq 2, the row quotient has dimension (k+12)\binom{k+1}{2}

Row ideals and determinants recover spectral pencils.

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

An affine spectral pencil attaches polynomial data to a constant matrix: its determinant and the column sums of its adjugate. The row ideal records the latter data, and its quotient measures their colength.

Let KK be a field and kk a nonnegative integer. Write n=k+1n=k+1 for the matrix size, corresponding to the paper’s parameter kk. For θ∈Kn\theta\in K^n, put Dθ=diag⁡(θ1,…,θn)D_\theta=\operatorname{diag}(\theta_1,\ldots,\theta_n). The diagonal pencil BA(t)B_A(t) associated to A∈Mat⁡n(K)A\in\operatorname{Mat}_n(K) and θ\theta is the polynomial matrix BA(t)=tDθ+A∈Mat⁡n(K[t]).B_A(t)=tD_\theta+A\in\operatorname{Mat}_n(K[t]).(1)✦

For an r×rr\times r matrix MM over a commutative ring RR, its spectral pencil PM(u)P_M(u) is the polynomial matrix defined in (2). For BA(t)B_A(t) from (1), write PA(t,s)P_A(t,s) for the resulting matrix over K[t][s]K[t][s]: PM(u)=uIr+M,PA(t,s)=PBA(t)(s)=sIn+tDθ+A.P_M(u)=uI_r+M, \qquad P_A(t,s)=P_{B_A(t)}(s)=sI_n+tD_\theta+A.(2)✦

The adjugate column sum ρi(M;u)∈R[u]\rho_i(M;u)\in R[u] is the sum of the entries in column ii of the adjugate of the spectral pencil (2): ρi(M;u)=∑j=1radj⁡(PM(u))ji.\rho_i(M;u)=\sum_{j=1}^r\operatorname{adj}(P_M(u))_{ji}.(3)✦ The adjugate row ideal I(M)⊆R[u]I(M)\subseteq R[u] is the ideal generated by the column sums (3): I(M)=(ρ1(M;u),…,ρr(M;u)).I(M)=\bigl(\rho_1(M;u),\ldots,\rho_r(M;u)\bigr).(4)✦ The adjugate row ideal IAI_A of the affine pencil PA(t,s)P_A(t,s) is the ideal of K[t][s]K[t][s] obtained by specializing (4) as in (5). The row quotient QAQ_A of the affine pencil PA(t,s)P_A(t,s) is the quotient of the polynomial algebra K[t][s]K[t][s] by its row ideal IAI_A, and FAF_A is the determinant polynomial of PA(t,s)P_A(t,s): IA=I(BA(t))⊆K[t][s],QA=K[t][s]/IA,FA(t,s)=det⁡PA(t,s).I_A=I(B_A(t))\subseteq K[t][s], \qquad Q_A=K[t][s]/I_A, \qquad F_A(t,s)=\det P_A(t,s).(5)

The Claim

Theorem 1. Suppose k>0k>0 and the diagonal coefficients θ1,…,θn\theta_1,\ldots,\theta_n are pairwise distinct. For every A∈Mat⁡n(K)A\in\operatorname{Mat}_n(K), QA is finite-dimensional over KQ_A\text{ is finite-dimensional over }K(6)✦ and dim⁡KQA=(k+12).\dim_K Q_A=\binom{k+1}{2}.(7)✦ For matrices A,B∈Mat⁡n(K)A,B\in\operatorname{Mat}_n(K) with IA=IBI_A=I_B and FA=FBF_A=F_B, A=B.A=B.(8)✦

In the paper, the result reads: for every matrix AA in the fixed matrix fiber over the curve CC, the subscheme ZA′⊆CZ'_A\subseteq C is finite of length N=n(n−1)/2N=n(n-1)/2, and two matrices A,BA,B in that fiber with ZA′=ZB′Z'_A=Z'_B as subschemes of CC are equal.

What Is Checked

The assertions (6), (7), and (8) are stated over matrices of size k+1k+1 over a field KK, with k>0k>0 and pairwise distinct diagonal coefficients, and use the row ideals in K[t][s]K[t][s] and the KK-vector space quotients of (5).