The Weyl calculus and a Cayley-Hamilton theorem for pairs of selfadjoint matrices (Q5926394)

Following H.~Weyl one can define a Weyl calculus \(W_{\mathbf A}\) for a pair of selfadjoint matrices \({\mathbf A}=(A_1,A_2)\) as a mean of construction which associates a matrix \(W_{\mathbf A}(f)\) to each smooth function \(f\) defined on~\({\mathbb R}^2\). The association \(f\mapsto W_{\mathbf A}(f)\) is linear but typically not multiplicative. Recently it has been shown [see e.g., \textit{G.~Greiner} and \textit{W. J.~Ricker}, Stud. Math. 112, No. 2, 109-125 (1995; Zbl 0817.47004), \textit{W. J.~Ricker}, Integral Equations Oper. Theory 22, No. 3, 333-338 (1995; Zbl 0847.47009)] that certain algebraic, geometric and topological properties of \(W_{\mathbf A}\) and/or the support \(\text{supp} (W_{\mathbf A})\) have strong implications for the relationship between \(A_1\) and~\(A_2\). The purpose of this paper is to construct a further equivalence of a distinctly different (algebraic) nature. Namely, if \(c_{\mathbf A}\) denotes the joint characteristic polynomial of the pair \({\mathbf A}\); i.e., the function \(\lambda \mapsto \det [(A_1-\lambda_1 I)^2+ (A_2-\lambda _2I)^2]\) for \(\lambda \in {\mathbb R}^2\), then \(A_1A_2=A_2A_1\) if and only if \(W_{\mathbf A}\) vanishes on the single polynomial function~\(c_{\mathbf A}\). That is the ``vector analogue'' \(W_{\mathbf A}(c_{\mathbf A})=0\) of the Cayley-Hamilton theorem is satisfied if and only if \(A_1\) and \(A_2\) commute.