Keyboard shortcuts

Press ← or → to navigate between chapters

Press ? to show this help

Press Esc to hide this help

Sra: reciprocal-permanent Lorentzian assertion

Abstract

The reciprocal-permanent polynomial of a positive inverse M matrix need not be Lorentzian.

Definition 1.1 (InverseMMatrix).

Formalization. D5/S3/Combinatorics/Permanental/SraReciprocalPermanentLorentzianRefutation.InverseMMatrix (✓ std3).

Citation. Suvrit Sra (2026). A Koteljanskii inequality for permanents. DOI: 10.48550/arXiv.2609.39979. URL: https://arxiv.org/abs/2609.39979v1.

Commentary.

Source, p. 1: “A matrix A is an inverse M-matrix if A≥0 entrywise, A is invertible, and A^{-1} has nonpositive off-diagonal entries.” Invertibility is expressed by IsUnit of the determinant; indices are Fin n.

Definition 1.2 (reciprocalPolynomial).

Formalization. D5/S3/Combinatorics/Permanental/SraReciprocalPermanentLorentzianRefutation.reciprocalPolynomial (✓ std3).

Citation. Suvrit Sra (2026). A Koteljanskii inequality for permanents. DOI: 10.48550/arXiv.2609.39979. URL: https://arxiv.org/abs/2609.39979v1.

Commentary.

Source, p. 6, §4(d): “For an inverse M-matrix A, consider” the displayed sum f_A(x_0,x)=∑{S⊆[n]} per(A_S)⁻¹ x_0^{n−|S|} ∏{i∈S} x_i. The variable none is x₀ and some i is the paper’s variable xᵢ₊₁. PSW.principalPermanent uses Matrix.permanent on the subtype S; the empty permanent and product are one. Nat.sub in the exponent is truncated natural subtraction.

Definition 1.3 (positiveHomogeneous).

Formalization. D5/S3/Combinatorics/Permanental/SraReciprocalPermanentLorentzianRefutation.positiveHomogeneous (✓ std3).

Citation. Petter Brändén, June Huh (2020). Lorentzian polynomials. DOI: 10.4007/annals.2020.192.3.4. URL: https://arxiv.org/abs/1902.03719v8.

Commentary.

Brändén–Huh, p. 6, §2.1: “We write H⁽ᵈ⁾ₙ for the set of degree d homogeneous polynomials in ℝ[w₁,…,wₙ].” “We define a topology on H⁽ᵈ⁾ₙ using the Euclidean norm for the coefficients, and write P⁽ᵈ⁾ₙ ⊆ H⁽ᵈ⁾ₙ for the open subset of polynomials all of whose coefficients are positive.” The formula quantifies over all degree-d exponents, including exponents outside a polynomial’s support. The centered dot in the lambda denotes Lean’s anonymous unused index binder _.

Definition 1.4 (quadraticHessian).

Formalization. D5/S3/Combinatorics/Permanental/SraReciprocalPermanentLorentzianRefutation.quadraticHessian (✓ std3).

Citation. Petter Brändén, June Huh (2020). Lorentzian polynomials. DOI: 10.4007/annals.2020.192.3.4. URL: https://arxiv.org/abs/1902.03719v8.

Commentary.

Brändén–Huh, p. 6, §2.1: “The Hessian of f ∈ ℝ[w₁,…,wₙ] is the symmetric matrix” ℋ_f(w)=(∂ᵢ∂ⱼf)ᵢ,ⱼ, “where ∂ᵢ stands for the partial derivative ∂/∂wᵢ.” For a homogeneous quadratic these derivatives are constant, so their coefficients at exponent zero give that matrix.

Definition 1.5 (strictlyLorentzian).

Formalization. D5/S3/Combinatorics/Permanental/SraReciprocalPermanentLorentzianRefutation.strictlyLorentzian (✓ std3).

Citation. Petter Brändén, June Huh (2020). Lorentzian polynomials. DOI: 10.4007/annals.2020.192.3.4. URL: https://arxiv.org/abs/1902.03719v8.

Commentary.

Brändén–Huh, p. 7, Definition 2.1: “We set L̊⁰ₙ=P⁰ₙ, L̊¹ₙ=P¹ₙ, and L̊²ₙ={f∈P²ₙ | ℋ_f is nonsingular and has exactly one positive eigenvalue}.” “For d larger than 2, we define L̊ᵈₙ recursively by setting L̊ᵈₙ={f∈Pᵈₙ | ∂ᵢf∈L̊ᵈ⁻¹ₙ for all i∈[n]}.” The variable set here is Option(Fin n). The Hermitian eigenvalue index counts multiplicity; every eigenvalue is nonzero and exactly one index has a positive eigenvalue.

Definition 1.6 (Lorentzian).

Formalization. D5/S3/Combinatorics/Permanental/SraReciprocalPermanentLorentzianRefutation.Lorentzian (✓ std3).

Citation. Petter Brändén, June Huh (2020). Lorentzian polynomials. DOI: 10.4007/annals.2020.192.3.4. URL: https://arxiv.org/abs/1902.03719v8.

Commentary.

Brändén–Huh, p. 7, Definition 2.1: “The polynomials in L̊⁽ᵈ⁾ₙ are called strictly Lorentzian, and the limits of strictly Lorentzian polynomials are called Lorentzian.” The coefficient epsilon condition expresses closure in the finite-dimensional homogeneous degree-d space. The centered dot in the lambda denotes Lean’s anonymous unused index binder _.

Definition 1.7 (claim).

Formalization. D5/S3/Combinatorics/Permanental/SraReciprocalPermanentLorentzianRefutation.claim (✓ std3).

Citation. Suvrit Sra (2026). A Koteljanskii inequality for permanents. DOI: 10.48550/arXiv.2609.39979. URL: https://arxiv.org/abs/2609.39979v1.

Commentary.

Source, p. 6, §4(d): “Is f_A Lorentzian for every inverse M-matrix A?” The assertion quantifies over every n>0 and every real n×n inverse M matrix, with no symmetry hypothesis. Fin n index i is source index i+1.

Theorem 1.8 (result).

Proof. Machine-checked in Lean as D5/S3/Combinatorics/Permanental/SraReciprocalPermanentLorentzianRefutation.result (✓ std3). ∎

Resolves. Problems/sra2609-koteljanskii (refuted) by D5/S3/Combinatorics/Permanental/SraReciprocalPermanentLorentzianRefutation.result.

Source. Repository-derived.

Acknowledgement. Suvrit Sra (2026). A Koteljanskii inequality for permanents. DOI: 10.48550/arXiv.2609.39979. URL: https://arxiv.org/abs/2609.39979v1.

Commentary.

The 5×5 matrix of Sra’s Theorem 1.2 has positive entries and an inverse with nonpositive off-diagonal entries. Differentiating in its last three variables gives a quadratic whose Hessian is positive definite on span(e₀,e₁+e₂). This property persists for sufficiently close coefficient approximations, whereas every strictly Lorentzian quadratic has only one positive eigenvalue.

References