The extremal permanent conjecture for unitarily invariant norms

From papers

Let nn be a positive integer, let S(n,k)S(n,k) denote the class of matrices under consideration, let G(n,k)G(n,k) be the distinguished matrix, and let T{\mathcal T} be the class of unitarily invariant matrix norms. For TTT\in {\mathcal T}, define

G~(n,k)=G(n,k)T(G(n,k)).\tilde{G}(n,k)=\frac{G(n,k)}{T(G(n,k))}.

For k{2,,n}k\in\{2,\dots,n\} and MS(n,k)M\in S(n,k) satisfying T(M)=1T(M)=1, the permanent of MM is at most the permanent of G~(n,k)\tilde{G}(n,k).

Extremal permanent conjecture. For all MS(n,k)M\in S(n,k) such that T(M)=1T(M)=1,

Per(M)Per(G~(n,k)).\operatorname{Per}(M)\leq \operatorname{Per}(\tilde{G}(n,k)).

The bound is asserted to be tight because G~(n,k)S(n,k)\tilde{G}(n,k)\in S(n,k) attains it. The surrounding text gives a theorem for this extremal bound and, for the Frobenius normalization, identifies the global upper bound as 1/321/32, attained at G~(4,2)\tilde{G}(4,2); the parser provides no evidence that this conjecture has been resolved independently.

Progress summary

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Sources & referencesView supporting material

Primary source

Papri Dey, “Polynomials with Lorentzian Signature, and Computing Permanents via Hyperbolic Programming”, arXiv:2206.02759 (2025).

Solutions 0

No solutions have been posted yet.