OctoInformBeta

OpenAI preprint, September 2026 · §6. The tangent obstruction (proof of Theorem 1.1)

§6: Tangent isomorphism and contradiction: Wedge coordinates give Tv≃ℝ3T_v \simeq \mathbb{R}^3 for unit v=(v0,v′)v = (v_0, v') with v0≠0v_0 \neq 0, ruling out normalized pencils via operator-norm bounds

Tangent obstruction proves the hyperbolicity cone KK is not spectrahedral.

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 the hyperbolicity cone KK in the space of triples (X,Z,y)(X, Z, y) consisting of two 4×44 \times 4 real symmetric matrices and a vector in ℝ3\mathbb{R}^3. The question is whether KK is spectrahedral, or whether a normalized matrix pencil can represent its Schur-complement inequality. Along a unit vector v=(v0,v′)∈ℝ4v = (v_0, v') \in \mathbb{R}^4 with v0≠0v_0 \neq 0, wedge coordinates map the orthogonal complement of vv to ℝ3\mathbb{R}^3, raising the question of whether this map is a linear isomorphism and whether a biquadratic form can match the squared Euclidean operator norm of a linear matrix family on that orthogonal complement.

The Hyperbolicity Cone and Normalized Pencils

For a nonnegative integer nn and a set RR, write Rn×nR^{n \times n} for the n×nn \times n square matrices over RR, and for nonnegative integers mm and nn, write ℝm×n\mathbb{R}^{m \times n} for the real m×nm \times n matrices. Denote by ℝn\mathbb{R}^n the nn-dimensional Euclidean space and by 𝒮n⊂ℝn×n\mathcal{S}^n \subset \mathbb{R}^{n \times n} the subspace of real symmetric n×nn \times n matrices, writing A≽0A \succeq 0 when A∈𝒮nA \in \mathcal{S}^n is positive semidefinite, A≻0A \succ 0 when it is positive definite, and InI_n for the n×nn \times n identity matrix. The ambient space is the 23-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 base point e=(I4,I4,0)e = (I_4, I_4, 0).

Over a commutative ring RR and for 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 matrix over RR defined with cyclic indices i,j∈{1,2,3}i, j \in \{1, 2, 3\} by (1): Q(y)ij={yi2+yi+12,i=j,−yiyj,i≠j.Q(y)_{ij} = \begin{cases} y_i^2 + y_{i+1}^2, & i = j, \\ -y_i y_j, & i \neq j. \end{cases}(1) Given a commutative ring RR and z,y∈R3z, y \in R^3, the Choi-Lam biquadratic form b(z,y)b(z, y) is the polynomial z⊤Q(y)zz^\top Q(y) z defined with cyclic indices i∈{1,2,3}i \in \{1, 2, 3\} by (2): b(z,y)=∑i=13zi2yi2−2(z1y1z2y2+z1y1z3y3+z2y2z3y3)+∑i=13zi2yi+12.b(z, y) = \sum_{i=1}^3 z_i^2 y_i^2 - 2(z_1 y_1 z_2 y_2 + z_1 y_1 z_3 y_3 + z_2 y_2 z_3 y_3) + \sum_{i=1}^3 z_i^2 y_{i+1}^2.(2)

Given a commutative ring RR and y∈R3y \in R^3, the matrix map Φy:R4×4→R4×4\Phi_y : R^{4 \times 4} \to R^{4 \times 4} is the linear map sending X=(ar⊤sT)X = \begin{pmatrix} a & r^\top \\ s & T \end{pmatrix} with a∈Ra \in R, r,s∈R3r, s \in R^3, and T∈R3×3T \in R^{3 \times 3} to (3): 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}.(3)✦

Over a commutative ring RR, the determinantal formula for X,Z∈R4×4X, Z \in R^{4 \times 4} and y∈R3y \in R^3 is (4): det⁡((det⁡X)Z−Φy(adj⁡X)).\det\bigl((\det X) Z - \Phi_y(\operatorname{adj} X)\bigr).(4)✦ The polynomial p(X,Z,y)p(X, Z, y) in the 2323 coordinates of 𝒮4×𝒮4×ℝ3\mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3 is the real polynomial defined by applying (4) to the symmetric 4×44 \times 4 matrices determined by 1010 upper-triangular indeterminates 1≤i≤j≤41 \le i \le j \le 4 for XX and 1010 for ZZ, together with 33 indeterminates for yy.

At each point x∈𝒮4×𝒮4×ℝ3x \in \mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3, the univariate restriction of pp along ee is the real polynomial t↦p(te−x)t \mapsto p(t e - x).✦ The closed hyperbolicity cone K=Λ+(p,e)K = \Lambda_+(p, e) of pp at ee is the subset of 𝒮4×𝒮4×ℝ3\mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3 defined by (5): K={x∈𝒮4×𝒮4×ℝ3|every t∈ℂ with p(te−x)=0 has Im⁡(t)=0 and Re⁡(t)≥0}.K = \bigl\{ x \in \mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3 \;\big|\; \text{every } t \in \mathbb{C} \text{ with } p(t e - x) = 0 \text{ has } \operatorname{Im}(t) = 0 \text{ and } \operatorname{Re}(t) \ge 0 \bigr\}.(5)

A normalized pencil (D,E,B)(D, E, B) consists of positive integers a,c≥1a, c \ge 1 together with real linear maps D:𝒮4→𝒮aD : \mathcal{S}^4 \to \mathcal{S}^a, E:𝒮4→𝒮cE : \mathcal{S}^4 \to \mathcal{S}^c, and B:ℝ3→ℝa×cB : \mathbb{R}^3 \to \mathbb{R}^{a \times c} such that DD and EE are positive (X≽0X \succeq 0 implies D(X)≽0D(X) \succeq 0 and E(X)≽0E(X) \succeq 0 for all X∈𝒮4X \in \mathcal{S}^4) and unital (D(I4)=IaD(I_4) = I_a and E(I4)=IcE(I_4) = I_c), and such that for all X,Z∈𝒮4X, Z \in \mathcal{S}^4 with X≻0X \succ 0 and all y∈ℝ3y \in \mathbb{R}^3, the Schur-complement equivalence (6) holds: Z−Φy(X−1)≽0⇔E(Z)−B(y)⊤D(X)−1B(y)≽0.Z - \Phi_y(X^{-1}) \succeq 0 \iff E(Z) - B(y)^\top D(X)^{-1} B(y) \succeq 0.(6)

Projections, Operator Norm, and Tangent Coordinates

For a nonnegative integer nn and a square matrix A∈ℝn×nA \in \mathbb{R}^{n \times n}, the kernel projection proj⁡ker⁡A∈ℝn×n\operatorname{proj}_{\ker A} \in \mathbb{R}^{n \times n} is the matrix of the orthogonal projection of ℝn\mathbb{R}^n onto ker⁡A\ker A.✦ For a vector v∈ℝnv \in \mathbb{R}^n, the outer product vv⊤∈𝒮nv v^\top \in \mathcal{S}^n is the symmetric matrix with (i,j)(i, j)-entry vivjv_i v_j. For a normalized pencil (D,E,B)(D, E, B) with dimensions a,c≥1a, c \ge 1 and v∈ℝ4v \in \mathbb{R}^4, the left kernel projection P(v)P(v) is the matrix in ℝa×a\mathbb{R}^{a \times a} defined by (7): P(v)=proj⁡ker⁡D(I4−vv⊤).P(v) = \operatorname{proj}_{\ker D(I_4 - v v^\top)}.(7)✦

The Euclidean operator norm ∥A∥op\|A\|_{\mathrm{op}} of a real matrix A∈ℝm×nA \in \mathbb{R}^{m \times n} (for nonnegative integers mm and nn) is the operator norm of the linear map ℝn→ℝm\mathbb{R}^n \to \mathbb{R}^m determined by AA.✦

Write ⟨⋅,⋅⟩\langle \cdot, \cdot \rangle and ∥⋅∥\|\cdot\| for the Euclidean inner product and norm on ℝ4\mathbb{R}^4. At v∈ℝ4v \in \mathbb{R}^4, the tangent space Tv⊂ℝ4T_v \subset \mathbb{R}^4 is the linear subspace of vectors orthogonal to vv, given by (8): Tv={h∈ℝ4∣⟨v,h⟩=0}.T_v = \{ h \in \mathbb{R}^4 \mid \langle v, h \rangle = 0 \}.(8)

For vectors u=(u0,u′)u = (u_0, u') and v=(v0,v′)v = (v_0, v') in ℝ4\mathbb{R}^4 with u0,v0∈ℝu_0, v_0 \in \mathbb{R} and u′,v′∈ℝ3u', v' \in \mathbb{R}^3, the wedge coordinates of vv and uu form the vector v0u′−u0v′∈ℝ3v_0 u' - u_0 v' \in \mathbb{R}^3.✦ For any v=(v0,v′)∈ℝ4v = (v_0, v') \in \mathbb{R}^4, the tangent coordinate map is the real linear map from TvT_v to ℝ3\mathbb{R}^3 defined by (9): Tv→ℝ3,h=(h0,h′)↦v0h′−h0v′.T_v \longrightarrow \mathbb{R}^3, \qquad h = (h_0, h') \longmapsto v_0 h' - h_0 v'.(9)✦

The Claim

Theorem 1. Let v=(v0,v′)∈ℝ4v = (v_0, v') \in \mathbb{R}^4 be a unit vector with v0≠0v_0 \neq 0. Then there exists a real linear isomorphism J:Tv→∼ℝ3J : T_v \xrightarrow{\sim} \mathbb{R}^3 satisfying (10): J(h)=v0h′−h0v′(h=(h0,h′)∈Tv).J(h) = v_0 h' - h_0 v' \qquad \bigl(h = (h_0, h') \in T_v\bigr).(10)✦

Theorem 2. Let mm and nn be nonnegative integers, let A∈ℝm×nA \in \mathbb{R}^{m \times n} be a real matrix, and let 1≤i≤m1 \le i \le m and 1≤j≤n1 \le j \le n. Then the entries of AA satisfy (11): Aij2≤∥A∥op2.A_{ij}^2 \le \|A\|_{\mathrm{op}}^2.(11)✦

Theorem 3. Let (D,E,B)(D, E, B) be a normalized pencil with dimensions a,c≥1a, c \ge 1, let v=(v0,v′)∈ℝ4v = (v_0, v') \in \mathbb{R}^4 be a unit vector with v0≠0v_0 \neq 0, and let T:ℝ4→ℝc×cT : \mathbb{R}^4 \to \mathbb{R}^{c \times c} be a real linear map. Then there exist h=(h0,h′)∈Tvh = (h_0, h') \in T_v and y∈ℝ3y \in \mathbb{R}^3 satisfying (12): b(v0h′−h0v′,y)≠∥P(v)B(y)T(h)∥op2.b(v_0 h' - h_0 v', y) \neq \|P(v) B(y) T(h)\|_{\mathrm{op}}^2.(12)✦

Theorem 4. No normalized pencil exists (13): {(D,E,B)∣(D,E,B) is a normalized pencil}=∅.\{ (D, E, B) \mid (D, E, B) \text{ is a normalized pencil} \} = \emptyset.(13)✦

Theorem 5. The hyperbolicity cone KK is not spectrahedral: for every positive integer NN and every real linear map L:𝒮4×𝒮4×ℝ3→𝒮NL : \mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3 \to \mathcal{S}^N, (14): K≠{x∈𝒮4×𝒮4×ℝ3∣L(x)≽0}.K \neq \{ x \in \mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3 \mid L(x) \succeq 0 \}.(14)✦

In the paper, the result reads: for v=(v0,v′)∈ℝ4v = (v_0, v') \in \mathbb{R}^4 with v0≠0v_0 \neq 0, J:Tv→ℝ3J : T_v \to \mathbb{R}^3 (with TvT_v standing for v⟂v^\perp) is a linear isomorphism, so F(z,y)=P(v)B(y)dRv(J−1z)F(z, y) = P(v) B(y)\, dR_v(J^{-1} z) (where R(u)=proj⁡ker⁡E(I4−uu⊤)R(u) = \operatorname{proj}_{\ker E(I_4 - u u^\top)} and dRvdR_v is its differential at vv) has bilinear entries ℓij(z,y)\ell_{ij}(z, y) satisfying ℓij(z,y)2≤∥F(z,y)∥op2=b(z,y)\ell_{ij}(z, y)^2 \le \|F(z, y)\|_{\mathrm{op}}^2 = b(z, y), and Lemma 2.1 forces F=0F = 0 and hence b=0b = 0, contradicting b(e1,e1)=1b(e_1, e_1) = 1 for the first standard basis vector e1∈ℝ3e_1 \in \mathbb{R}^3.

What Is Checked

The tangent coordinate map (9) and theorems (10) through (14) are stated over the tangent space TvT_v and wedge coordinates in ℝ4\mathbb{R}^4 (assuming ∥v∥=1\|v\| = 1 and v0≠0v_0 \neq 0 in (10) and (12)), the Euclidean operator norm ∥⋅∥op\|\cdot\|_{\mathrm{op}} on real m×nm \times n matrices in (11), normalized pencils (D,E,B)(D, E, B) with left kernel projection P(v)P(v), Choi-Lam form b(z,y)b(z, y), and real linear maps T:ℝ4→ℝc×cT : \mathbb{R}^4 \to \mathbb{R}^{c \times c} in (12) and (13), and the hyperbolicity cone K⊂𝒮4×𝒮4×ℝ3K \subset \mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3 of pp at e=(I4,I4,0)e = (I_4, I_4, 0) together with real linear maps L:𝒮4×𝒮4×ℝ3→𝒮NL : \mathcal{S}^4 \times \mathcal{S}^4 \times \mathbb{R}^3 \to \mathcal{S}^N for positive integers NN in (14).