Approximating commuting operators (Q5935368): Difference between revisions

Let \(\mathcal{C}(m,n)\) be the set of \(m\)-tuples of commuting \(n\times n\) complex matrices, and let \(\mathcal{G}(m,n)\) be the subset consisting of \(m\)-tuples of generic commuting matrices (that is, matrices which have distinct eigenvalues). There has been an extensive study of the approximation problem: when is the topological closure of \(\mathcal{G}(m,n)\) equal to \(\mathcal{C}(m,n)\)? [see, for example, \textit{M. Gerstenhaber}, Ann. Math. (2) 73, 324-348 (1961; Zbl 0168.28201)]. It was shown that the answer is always ``yes'' if \(n\leq 3\) or \(m\leq 2\); and it is ``no'' if \(n\geq 4\) and \(m\geq 4\). \textit{R. M. Guralnick} [Linear Multilinear Algebra 31, No. 1-4, 71-75 (1992; Zbl 0754.15011)] proved that the answer is ``no'' if \(m=3\) and \(n\geq 32\) leaving the unsettled cases where \(m=3\) and \(4\leq n\leq 31\). The authors of the present paper settle some of these cases and show that when \(m=3\) the answer is ``no'' if \(n\geq 30\) and ``yes'' if \(n\leq 5\). They also solve the approximation problem for \(3\)-tuples satisfying some special conditions. The proofs are largely computational.