OctoInformBeta

OpenAI preprint, September 2026 · §4. The adjugate scheme in characteristic two

Lemma 4.1: The affine determinant has a prescribed squarefree leading form over any field

Distinct slopes give separable 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

Consider a determinant representation in two variables with the identity matrix as the coefficient of the spectral variable and a diagonal matrix as the other coefficient. Let KK be a field, let II be a finite index set of cardinality nn, and let θ:I→K\theta\colon I\to K be a family of slopes. Write Dθ=diag⁡(θi)i∈ID_\theta=\operatorname{diag}(\theta_i)_{i\in I} and let InI_n denote the identity matrix.

For a scalar a∈Ka\in K, the binary form ℓa∈K[t,s]\ell_a\in K[t,s] is the linear polynomial (1): ℓa(s,t)=s+at.\ell_a(s,t)=s+at.(1)✦ For A∈Mat⁡I(K)A\in\operatorname{Mat}_I(K), the diagonal binary pencil SθS_\theta and the bivariate spectral pencil Mθ,AM_{\theta,A} are the polynomial matrices in (2): Sθ=diag⁡(ℓθi)i∈I,Mθ,A=Sθ+A=sIn+tDθ+A.S_\theta=\operatorname{diag}(\ell_{\theta_i})_{i\in I}, \qquad M_{\theta,A}=S_\theta+A=sI_n+tD_\theta+A.(2) The spectral curve equation Fθ,A∈K[t,s]F_{\theta,A}\in K[t,s] is the determinant (3): Fθ,A(s,t)=det⁡Mθ,A(s,t).F_{\theta,A}(s,t)=\det M_{\theta,A}(s,t).(3)✦ Write [Fθ,A]n[F_{\theta,A}]_n for its homogeneous component of total degree nn.

The Claim

Theorem 1. Let KK be a field, let II be a finite index set of cardinality nn, let θ:I→K\theta\colon I\to K, and let A∈Mat⁡I(K)A\in\operatorname{Mat}_I(K). Then deg⁡totFθ,A≤n,\deg_{\mathrm{tot}}F_{\theta,A}\leq n,(4)✦ [Fθ,A]n=∏i∈Iℓθi.[F_{\theta,A}]_n=\prod_{i\in I}\ell_{\theta_i}.(5)✦ If the slopes θi\theta_i are pairwise distinct, then Fθ,A(s,t)is separable as a polynomial in s over K(t),F_{\theta,A}(s,t) \quad\text{is separable as a polynomial in }s\text{ over }K(t),(6)✦ ∏i∈Iℓθiis squarefree in K[t,s].\prod_{i\in I}\ell_{\theta_i} \quad\text{is squarefree in }K[t,s].(7)✦

In the paper, the result reads: writing nn for its matrix size kk, for every A∈Mat⁡n(K)A\in\operatorname{Mat}_n(K) in its setting, with homogeneous pencil M̃θ,A=sIn+tDθ+uA\widetilde M_{\theta,A}=sI_n+tD_\theta+uA, the projective spectral curve CA=V(det⁡M̃θ,A)⊂ℙK2C_A=V(\det\widetilde M_{\theta,A})\subset\mathbb P^2_K is reduced, the projection π:CA→ℙK1\pi\colon C_A\to\mathbb P^1_K, [s:t:u]↦[t:u][s:t:u]\mapsto[t:u], is finite flat of degree nn and generically etale, every component meets u=0u=0, where M̃θ,A\widetilde M_{\theta,A} has rank n−1n-1, and π*𝒪ℙ1(1)≃𝒪CA(1)\pi^*\mathcal O_{\mathbb P^1}(1)\simeq\mathcal O_{C_A}(1), π*𝒪CA≃⨁j=0n−1𝒪ℙ1(−j)\pi_*\mathcal O_{C_A}\simeq\bigoplus_{j=0}^{n-1}\mathcal O_{\mathbb P^1}(-j), while π*𝒪CA(n−1)≃⨁j=0n−1𝒪ℙ1(n−1−j)\pi_*\mathcal O_{C_A}(n-1)\simeq \bigoplus_{j=0}^{n-1}\mathcal O_{\mathbb P^1}(n-1-j).

What Is Checked

The assertions (4) and (5) are stated for finite matrices over an arbitrary field with arbitrary slope families; (6) and (7) also assume that the slopes are pairwise distinct.