OctoInformBeta

OpenAI preprint, September 2026 · §3. Hyperbolicity and two cone slices

Proposition 3.2: The positive-definite first-matrix and zero-vector cases, positive-shift nonvanishing, an open subset of the cone, and an interior base point

Semidefinite inequalities characterize two slices of a hyperbolicity cone.

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 determinantal hyperbolicity cone on pairs of 4×44 \times 4 real symmetric matrices together with a vector in ℝ3\mathbb{R}^3. Two questions are which semidefinite matrix inequalities describe its slices when the first matrix is positive definite or when the vector vanishes, and whether its base point lies in the interior of the cone.

Definitions

Given a natural number nn and a set RR, the n×nn \times n square matrices over RR are the matrices with entries in RR whose rows and columns are indexed by {1,…,n}\{1, \dots, n\}.✦ For a natural number nn, the space 𝒮n\mathcal{S}^n of real symmetric matrices is the subspace of self-adjoint real n×nn \times n square matrices.✦ When A∈𝒮nA \in \mathcal{S}^n, write A≽0A \succeq 0 for positive semidefinite and A≻0A \succ 0 for positive definite. The ambient vector space is the 2323-dimensional product 𝒮4×𝒮4×ℝ3\mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3 of triples (X,Z,y)(X, Z, y) with X,Z∈𝒮4X, Z \in \mathcal{S}^4 and y∈ℝ3y \in \mathbb{R}^3. Inside 𝒮4×𝒮4×ℝ3\mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3, the base point ee is the triple (I4,I4,0)(I_4, I_4, 0), where I4I_4 is the 4×44 \times 4 identity matrix and 00 is the zero vector in ℝ3\mathbb{R}^3.✦

Over a commutative ring RR, for a triple y=(y1,y2,y3)∈R3y = (y_1, y_2, y_3) \in R^3, the quadratic matrix Q(y)Q(y) is the 3×33 \times 3 square matrix over RR with cyclic indices in {1,2,3}\{1, 2, 3\} defined by (1): Q(y)ij={yi2+yi+12if i=j,−yiyjif i≠j.Q(y)_{ij} = \begin{cases} y_i^2 + y_{i+1}^2 & \text{if } i = j, \\ -y_i y_j & \text{if } i \ne j. \end{cases}(1)✦ Given a commutative ring RR and y∈R3y \in R^3, the matrix map Φy\Phi_y sends each 4×44 \times 4 square matrix over RR in (1+3)×(1+3)(1 + 3) \times (1 + 3) block form (ar⊤sT)\begin{pmatrix} a & r^\top \\ s & T \end{pmatrix}, where a∈Ra \in R, r,s∈R3r, s \in R^3, and TT is a 3×33 \times 3 square matrix over RR, to the 4×44 \times 4 matrix over RR given by (2): Φy(ar⊤sT)=(tr⁡(Q(y)T)−r⊤Q(y)−Q(y)saQ(y)).\Phi_y \begin{pmatrix} a & r^\top \\ s & T \end{pmatrix} = \begin{pmatrix} \operatorname{tr}(Q(y) T) & -r^\top Q(y) \\ -Q(y) s & a Q(y) \end{pmatrix}.(2)✦ When y∈ℝ3y \in \mathbb{R}^3, the symmetric matrix map Φy:𝒮4→𝒮4\Phi_y : \mathcal{S}^4 \to \mathcal{S}^4 restricts (2) to 𝒮4\mathcal{S}^4, mapping each real symmetric matrix X∈𝒮4X \in \mathcal{S}^4 to a symmetric matrix Φy(X)∈𝒮4\Phi_y(X) \in \mathcal{S}^4.✦

For a commutative ring RR, the determinantal formula (3) is the scalar assigned to 4×44 \times 4 square matrices XX and ZZ over RR and a triple y∈R3y \in R^3: det⁡((det⁡X)Z−Φy(adj⁡X)).\det\!\bigl((\det X) Z - \Phi_y(\operatorname{adj} X)\bigr).(3) The hyperbolic polynomial pp is the real multivariate polynomial in the 2323 coordinates of 𝒮4×𝒮4×ℝ3\mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3 obtained by applying (3) to generic symmetric matrices X,Z∈𝒮4X, Z \in \mathcal{S}^4 and a vector y∈ℝ3y \in \mathbb{R}^3 of coordinate indeterminates.✦

At any point x∈𝒮4×𝒮4×ℝ3x \in \mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3, the univariate restriction of pp is the real univariate polynomial t↦p(te−x)t \mapsto p(t e - x).✦ The closed hyperbolicity cone K=Λ+(p,e)K = \Lambda_+(p, e) is the subset of points x∈𝒮4×𝒮4×ℝ3x \in \mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3 for which every complex root t∈ℂt \in \mathbb{C} of the univariate restriction t↦p(te−x)t \mapsto p(t e - x) is real and nonnegative.✦

For a natural number nn, the symmetric matrix inverse X−1X^{-1} of a symmetric matrix X∈𝒮nX \in \mathcal{S}^n is the matrix in 𝒮n\mathcal{S}^n that is its n×nn \times n matrix inverse.✦ Given x=(X,Z,y)∈𝒮4×𝒮4×ℝ3x = (X, Z, y) \in \mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3, the ambient residual ℛ(x)\mathcal{R}(x) is the symmetric matrix in 𝒮4\mathcal{S}^4 defined by (4): ℛ(x)=Z−Φy(X−1).\mathcal{R}(x) = Z - \Phi_y(X^{-1}).(4) When X,Z∈𝒮4X, Z \in \mathcal{S}^4, y∈ℝ3y \in \mathbb{R}^3, and t∈ℝt \in \mathbb{R}, the shifted cone residual ℛt(X,Z,y)\mathcal{R}_t(X, Z, y) is the symmetric matrix in 𝒮4\mathcal{S}^4 given by (5): ℛt(X,Z,y)=Z+tI4−Φy((X+tI4)−1).\mathcal{R}_t(X, Z, y) = Z + t I_4 - \Phi_y\bigl((X + t I_4)^{-1}\bigr).(5)

The Claim

Theorem 1. Let X,Z∈𝒮4X, Z \in \mathcal{S}^4 and y∈ℝ3y \in \mathbb{R}^3. If X≻0X \succ 0, then membership in KK satisfies (6): (X,Z,y)∈K⇔Z−Φy(X−1)≽0.(X, Z, y) \in K \iff Z - \Phi_y(X^{-1}) \succeq 0.(6)✦ Given X,Z∈𝒮4X, Z \in \mathcal{S}^4, membership on the zero-vector slice satisfies (7): (X,Z,0)∈K⇔X≽0 and Z≽0.(X, Z, 0) \in K \iff X \succeq 0 \text{ and } Z \succeq 0.(7)✦ Denoting by int⁡(K)\operatorname{int}(K) the interior of KK in 𝒮4×𝒮4×ℝ3\mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3, the base point ee satisfies (8): e∈int⁡(K).e \in \operatorname{int}(K).(8)✦ For any x∈𝒮4×𝒮4×ℝ3x \in \mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3, membership in KK is characterized by (9): x∈K⇔p(x+te)≠0for all real t>0.x \in K \iff p(x + t e) \ne 0 \quad \text{for all real } t > 0.(9)✦ Whenever X,Z∈𝒮4X, Z \in \mathcal{S}^4 and y∈ℝ3y \in \mathbb{R}^3 satisfy X≻0X \succ 0 and X−1−(X+tI4)−1≽0X^{-1} - (X + t I_4)^{-1} \succeq 0 for every real t≥0t \ge 0, the equivalence (10) holds: (X,Z,y)∈K⇔ℛ0(X,Z,y)≽0.(X, Z, y) \in K \iff \mathcal{R}_0(X, Z, y) \succeq 0.(10)✦ Write U={x=(X,Z,y)∈𝒮4×𝒮4×ℝ3:X≻0 and ℛ(x)≻0}U = \{x = (X, Z, y) \in \mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3 : X \succ 0 \text{ and } \mathcal{R}(x) \succ 0\} for the strict positive slice; then (11) holds: U is open in 𝒮4×𝒮4×ℝ3,U \text{ is open in } \mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3,(11)✦ Moreover, (12) holds: U⊆K.U \subseteq K.(12)✦

In the paper, the result reads: for K=Λ+(p,e)K = \Lambda_+(p, e), X,Z∈𝒮4X, Z \in \mathcal{S}^4, and y∈ℝ3y \in \mathbb{R}^3, if X≻0X \succ 0 then (X,Z,y)∈K(X, Z, y) \in K if and only if Z≽Φy(X−1)Z \succeq \Phi_y(X^{-1}), and (X,Z,0)∈K(X, Z, 0) \in K if and only if X≽0X \succeq 0 and Z≽0Z \succeq 0; moreover, ee lies in the interior of KK in the full ambient space.

What Is Checked

The claim is stated over the real symmetric matrix spaces 𝒮n\mathcal{S}^n, the 2323-dimensional ambient space 𝒮4×𝒮4×ℝ3\mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3 with base point e=(I4,I4,0)e = (I_4, I_4, 0), the determinantal polynomial pp constructed from Q(y)Q(y) and Φy\Phi_y, the closed hyperbolicity cone K=Λ+(p,e)K = \Lambda_+(p, e) defined by the univariate restriction, the symmetric matrix inverse with the residuals ℛ(x)\mathcal{R}(x) and ℛt(X,Z,y)\mathcal{R}_t(X, Z, y), and the slice conditions X≻0X \succ 0, X−1−(X+tI4)−1≽0X^{-1} - (X + t I_4)^{-1} \succeq 0 for t≥0t \ge 0, y=0y = 0, and X≻0X \succ 0 with ℛ(x)≻0\mathcal{R}(x) \succ 0.