Séminaire Lotharingien de Combinatoire, 89B.29 (2023), 12 pp.

Mark Skandera and Daniel Soskin

Barrett-Johnson Inequalities for Totally Nonnegative Matrices

Abstract. Given a matrix A, let AI,J denote the submatrix of A determined by rows I and columns J. The Barrett-Johnson Inequalities relate sums of products of principal minors of positive semidefinite (PSD) matrices, when orders of the minors are given by integer partitions λ = (λ1, ..., λr, μ = (μ1, ..., μs of n. Specifically, we have
          ∑                                      ∑
λ1!⋅⋅⋅λr!       det(AI1,I1)⋅⋅⋅det(AIr,Ir) ≥ μ1!⋅⋅⋅μs!       det(AJ1,J1)⋅⋅⋅det(AJs,Js),
        (I1,...,Ir)                                (J1,...,Js)
for all PSD n × n matrices A, where sums are over ordered set partitions of {1,...,n} satisfying |Ik| = λk, |Jk| = μk, if and only if λ is majorized by μ. We show that these inequalities hold for totally nonnegative matrices as well.


Received: November 15, 2022. Accepted: February 20, 2023. Final version: April 1, 2023.

The following versions are available: