Answers are grounded in this paper and cite their sources.
Let M be a compact set of real square matrices with joint spectral radius one, and let $g(n)$ be the largest norm of a product of n matrices from M. Varney and Morris asked whether g is always weakly increasing and weakly upper regularly varying; this is Problem MF-11 of the collection Open Problems in Numerical Linear Algebra. We show that both properties can fail, even for pairs of matrices. We construct an explicit pair of real 20×20 matrices, depending on a Liouville number θ_, for which the ratios $g(N/2)/g(N)$ and $g(2N)/g(N)$ are unbounded along the lengths N=16ᵖ for which pθ_ is extremely close to an integer. Both properties also fail for a pair of invertible matrices and for a pair of 22×22 matrices with n≤ g(n)≤ Cn(log n)². On the other hand, both properties hold for every finite set of 3×3 matrices, so that the smallest dimension of a counterexample lies between $4$ and $20$, and g is weakly increasing for block triangular families with two product bounded diagonal blocks, one of which contains an elliptic product or is irreducible with a product of spectral radius one. The proofs reduce products to a seven-dimensional commutative nilpotent algebra, in which a quadratic term cancels exactly at certain lengths.
No takes yet. Share an insight, caveat, or question.
Pedro Martins (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: