Notes on permanental and subpermanental inequalities (Q1195335): Difference between revisions

Let \(J_ n\) denote the \(n\times n\) matrix whose entries are all \({1\over n}\). For any \(n\times n\) matrix \(A\) let \(\sigma_ k(A)\) denote the sum of all subpermanents of order \(k\) in \(A,k=1,2,\dots,n\). For convenience put \(\sigma_ 0(A)=1\). It had been known that if \(t\in\left[-{1\over n},1\right]-\{0\}\) and \(A\) is an \(n\times n\) stochastic matrix, then the maximum of \(\sigma_ k(tA+(1-t)J_ n)\) is achieved if and only if \(A\) is a permutation matrix when \(k>1\). The author shows that \({1\over n}\) can be replaced by \({1\over n-1}\). It is shown that if \(A\) is doubly stochastic then for each \(k\), \(\sigma_ k(tA+(1-t)J_ n)\leq t\sigma_ k(A)+(1-t)\sigma_ k(J_ n)\) for \(t\) in some interval \([0,t_ k]\). In particular, \(t_ 2=1\) and \(t_ 3={1\over 2}\). The number \(t_ k=1\) for all \(k\) if in addition \(A\) is normal with spectrum in the sector \(-\pi/2n\leq\theta\leq\pi/2n\) in the complex plane.