--- title: Cattani's Theorem in Codimension-Two Algebras url: https://www.emergentmind.com/topics/cattani-s-theorem type: topic --- # Cattani's Theorem in Codimension-Two Algebras Searching arXiv for Cattani's theorem and related codimension-two/Hodge-Riemann papers. Cattani’s theorem, in the form treated for graded Artinian Gorenstein algebras, asserts that on a convex cone the ordinary Hodge–Riemann relations are equivalent to the mixed Hodge–Riemann relations. In the codimension-two setting, where the algebra is defined by a homogeneous polynomial in two variables, this statement admits an explicit reformulation in terms of higher Hessians and total positivity of Toeplitz matrices. A 2025 arXiv paper gives a new proof in that special case, replacing the original variation-of-Hodge-structure machinery by an explicitly algebraic, matrix-theoretic, and topological argument [2507.15043]. ## 1. Statement and scope In the setting of the 2025 treatment, one starts from a graded Artinian Gorenstein algebra \(A_F\) associated with a real homogeneous polynomial \(F\). Cattani’s theorem is stated as follows: if \(U\) is a convex cone, then \(A_F\) satisfies ordinary \(\mathrm{HRR}_{\lfloor d/2\rfloor}(U)\) if and only if it satisfies mixed \(\mathrm{HRR}_{\lfloor d/2\rfloor}(U)\) [2507.15043]. The theorem compares two levels of Lefschetz-type positivity. Ordinary Hodge–Riemann relations test positivity using powers of a single linear form \(\ell\). Mixed Hodge–Riemann relations allow products of possibly distinct linear forms \(\ell_0,\ell_1,\dots\). Mixed HRR imply ordinary HRR by specializing all linear forms to be equal; the content of Cattani’s theorem is the converse under the convex cone hypothesis. Two caveats are part of the statement as presented in the codimension-two proof. First, the convex cone condition on \(U\) is essential. Second, in general one cannot replace top-degree HRR by HRR only up to some smaller \(i<\lfloor d/2\rfloor\). These restrictions delimit the exact range in which the equivalence is valid. ## 2. Algebraic framework in codimension two The codimension-two case is the case of graded Artinian Gorenstein algebras coming from a homogeneous polynomial in two variables. The paper works with \[ F\in Q_d=\mathbb{R}[X,Y]_d, \] and its associated algebra \[ A_F=\frac{R}{\operatorname{Ann}(F)},\qquad R=\mathbb{R}[x,y], \] where \(R\) acts on \(Q=\mathbb{R}[X,Y]\) by differential operators, \[ x\circ F=\frac{\partial F}{\partial X},\qquad y\circ F=\frac{\partial F}{\partial Y}. \] The annihilator is \[ \operatorname{Ann}(F)=\{f\in R\mid f\circ F\equiv 0\}. \] The algebra is graded, \[ A_F=\bigoplus_{i=0}^d (A_F)_i, \] with socle degree \(d\), and it carries perfect pairings \[ (-,-)_i:A_i\times A_{d-i}\to \mathbb{R},\qquad (\alpha,\beta)_i=(\alpha\beta)\circ F. \] The paper defines the Sperner number \[ s(F)=\max\{h_i\mid h_i=\dim(A_i)\}, \] and records that in codimension two the Hilbert function has the standard symmetric form \[ H(A_F)=(1,2,\dots,s-1,s,\dots,s,s-1,\dots,2,1). \] This codimension-two regime is structurally special. The coefficient data of \(F\) is a single sequence, the graded pieces have dimension at most \(i+1\), and the relevant Lefschetz forms can be written explicitly in a monomial basis. These features are what make the later Hessian and Toeplitz reformulations possible in a concrete form. ## 3. Ordinary and mixed Hodge–Riemann relations For a linear form \(\ell\in (A_F)_1\), the \(i\)-th primitive space is \[ P_{i,\ell}=\ker\{\times \ell^{d-2i+1}:A_i\to A_{d-i+1}\}. \] The ordinary Lefschetz map in degree \(i\) is \[ \times \ell^{d-2i}: A_i\to A_{d-i}, \] and the ordinary \(i\)-th Lefschetz form is \[ (\alpha,\beta)_i^\ell = (-1)^i(\ell^{d-2i}\alpha\beta)\circ F. \] Ordinary \(\mathrm{HRR}_i\) requires that for every \(0\le j\le i\), the form \((-, -)_j^\ell\) is positive definite on \(P_{j,\ell}\), concretely \[ (\alpha,\alpha)_j^\ell = (-1)^j(\ell^{d-2j}\alpha^2)\circ F >0 \qquad \forall\, 0\neq \alpha\in P_{j,\ell}. \] For a tuple \(\mathcal L=(\ell_0,\ell_1,\dots,\ell_d)\), the mixed primitive space is \[ P_{i,\mathcal L} = \ker\{\times \ell_0\ell_1\cdots \ell_{d-2i}:A_i\to A_{d-i+1}\}, \] and the mixed Lefschetz form is \[ (\alpha,\beta)_i^{\mathcal L} = (-1)^i(\ell_1\cdots \ell_{d-2i}\alpha\beta)\circ F. \] Mixed \(\mathrm{HRR}_i\) requires that for every \(0\le j\le i\), the form \((-, -)_j^{\mathcal L}\) is nondegenerate on \(A_j\) and positive definite on \(P_{j,\mathcal L}\), equivalently \[ (\alpha,\alpha)_j^{\mathcal L} = (-1)^j(\ell_1\cdots \ell_{d-2j}\alpha^2)\circ F>0 \qquad \forall\, 0\neq \alpha\in P_{j,\mathcal L}. \] The paper works especially with two cones: \[ U=\{ax+by\mid a,b>0\}, \qquad \overline U=\{ax+by\mid a,b\ge 0,\ (a,b)\neq (0,0)\}. \] The distinction between \(U\) and \(\overline U\) controls a parallel distinction between total nonnegativity and total positivity. That dichotomy is not cosmetic: it is built into the theorem’s reformulation and into the deformation argument used in the proof. ## 4. Hessians, Toeplitz matrices, and the codimension-two reformulation The codimension-two proof translates ordinary HRR into positivity of higher Hessians and mixed HRR into total positivity properties of Toeplitz matrices [2507.15043]. If \[ F=\sum_{k=0}^d \binom{d}{k} c_k X^kY^{d-k}, \] then the normalized Toeplitz matrix is \[ \phi_d^i(F)= \begin{pmatrix} c_i & c_{i+1} & \cdots & c_d\\ \vdots & \ddots & \ddots & \vdots\\ c_0 & c_1 & \cdots & c_{d-i} \end{pmatrix}, \] an \((i+1)\times(d-i+1)\) Toeplitz matrix with \[ \operatorname{rank}(\phi_d^i(F))=\min\{i+1,s(F)\}=\dim((A_F)_i). \] The \(i\)-th higher Hessian matrix is \[ \operatorname{Hess}_i(F)= \left( x^{p+q}y^{2i-p-q}\circ F \right)_{0\le p,q\le i} = \left( \frac{\partial^{2i}F}{\partial X^{p+q}\partial Y^{2i-p-q} \right)_{0\le p,q\le i}. \] For \(0\le i\le s-1\), this matrix represents the Lefschetz map \[ \times \ell^{d-2i}: (A_F)_i\to (A_F)_{d-i}, \qquad \ell=Xx+Yy, \] up to a nonzero scalar multiple. The structural equivalences used in the proof are these. Ordinary \(\mathrm{HRR}_i(\overline U)\) is equivalent to positivity of the higher Hessians on \(\mathbb R_{\ge 0}^2\setminus\{0\}\), and ordinary \(\mathrm{HRR}_i(U)\) is equivalent to positivity on \(\mathbb R_{>0}^2\). Mixed \(\mathrm{HRR}_i(\overline U)\) is equivalent to \(\phi_d^i(F)\) being totally positive, while mixed \(\mathrm{HRR}_i(U)\) is equivalent to \(\phi_d^i(F)\) being totally non-negative. The paper packages the resulting codimension-two form of Cattani’s theorem as a Hessian criterion. For \(F\in Q_d\) of Sperner number \(s\): - \(\phi_d^{\,s-1}(F)\) is totally positive if and only if \[ H_i^F(X,Y)>0 \quad \forall (X,Y)\in \mathbb R_{\ge 0}^2\setminus\{(0,0)\}, \quad 0\le i\le s-1. \] - \(\phi_d^{\,s-1}(F)\) is totally non-negative if and only if \[ H_i^F(X,Y)>0 \quad \forall (X,Y)\in \mathbb R_{>0}^2, \quad 0\le i\le s-1. \] An equivalent maximal-size Toeplitz version is also given, using \(\phi_d^{\lfloor d/2\rfloor}(F)\). In this sense, Cattani’s theorem in two variables becomes a theorem that converts positivity of a finite family of higher Hessians into total positivity or total nonnegativity of a Toeplitz matrix. ## 5. Proof architecture in two variables The 2025 proof is organized as a sequence of translations and deformation arguments rather than by variation of Hodge structure [2507.15043]. Its central technical bridge is a Plücker expansion of the Hessian: \[ H_i^F(X,Y)= \left(\frac{d!}{(d-2i)!}\right)^{i+1} \sum_{J\in \binom{[d-i+1]}{i+1}} N'_J\,\Delta_J(\phi_d^i(F))\,X^{|\lambda(J)|}Y^{D_i-|\lambda(J)|}, \] where \[ D_i=(i+1)(d-2i). \] This writes each Hessian polynomial as a positive linear combination of monomials whose coefficients are maximal minors of the Toeplitz matrix, weighted by positive integers \(N'_J\). That expansion has two immediate consequences. If the relevant Toeplitz minors are positive or nonnegative, Hessian positivity follows directly. Conversely, Hessian positivity forces nonvanishing of specific minors, especially corner minors, because they occur as extreme coefficients. The proof supplements this with nonvanishing of maximal contiguous minors, obtained via an argument using HRR for derivatives \(x^r\circ F\). A topological step then identifies an open set \(\mathcal O_s\) of rank-\(s\) Toeplitz matrices characterized by nonvanishing corner minors and contiguous minors, and proves \[ \mathcal T_s^{>0}=\mathcal T_s^{\ge 0}\cap \mathcal O_s. \] In particular, \(\mathcal T_s^{>0}\) is a union of connected components of \(\mathcal O_s\). This allows one to use deformation. The shift operator \[ S_t(F)=F(X+tY,Y) \] preserves ordinary HRR on \(\overline U\) for \(t\ge 0\), and for \(t\gg 0\) an elementary perturbation argument yields mixed HRR on a suitable cone, hence total positivity of the corresponding Toeplitz matrix. Since the path remains inside \(\mathcal O_s\), the initial matrix must lie in the same connected component. For the open-cone statement, the proof uses the double shift \[ R_t(F)=F(X+tY,tX+Y). \] For \(00 \quad \forall (X,Y)\in \mathbb R_{\ge 0}^2\setminus\{0\}, \quad 0\le i\le s-1, \] then all coefficients of each \(H_i^F(X,Y)\) are strictly positive. If positivity holds only on \(\mathbb R_{>0}^2\), then all coefficients are nonnegative. The paper highlights this because positivity of a homogeneous polynomial on the positive orthant does not usually imply coefficientwise positivity; here it does because the Hessians arise from Toeplitz/Plücker data with total positivity constraints. A frequent source of confusion is the similarity between **Cattani’s theorem** and **Catlin’s theorem**. They are unrelated. Catlin’s theorem is a graph-coloring result: if \(G\) has maximum degree \(d\ge 3\) and does not contain \(K_{d+1}\), then \(G\) has a \(d\)-coloring in which one color class has size \(\alpha(G)\), strengthening Brooks’ theorem [1402.6298]. Cattani’s theorem, by contrast, is a theorem about ordinary and mixed Hodge–Riemann relations for graded Artinian Gorenstein algebras. The similarity is purely nominal. In its codimension-two form, Cattani’s theorem shows that mixed Hodge–Riemann phenomena can be reconstructed from explicitly computable algebraic data attached to a bivariate form. This suggests a broader perspective in which Hodge-theoretic positivity, higher Hessians, and total positivity belong to a single structural framework, even though the explicit Toeplitz and Hessian technology used here is particular to two variables. Source: https://www.emergentmind.com/topics/cattani-s-theorem