Non-commutative polynomials of independent Gaussian random matrices. The real and symplectic cases. (Q1769078)

This paper shows the following extension of Voiculescu's random matrix model to the sets of independent Gaussian random matrices with real entries and random matrices ensembles: Let \(r,s\in \mathbb N_0\) with \(r+s\geq 1\), and for each \(n\in \mathbb N_0\), let \(X_1^{(n)},\ldots,X_{r+s}^{(n)}\) be stochastically independent random matrices defined on \((\Omega,{\mathcal F},P)\) such that \(X_1^{(n)},\ldots,X_{r}^{(n)}\in \text{GOE}(n,\frac{1}{n})\) and \(X_{r+1}^{(n)},\ldots,X_{r+s}^{(n)}\in \text{GOE}^*(n,\frac{1}{n})\). Furthermore, let \(({\mathcal A},\tau)\) be a \(C^*\)-probability space with \(\tau\) a faithful state on \(\mathcal A\), and let \(\{x_1,\ldots,x_{r+s}\}\) be a semicircular system in \(({\mathcal A},\tau)\). Then there is a P-null set \(N\subseteq\Omega\) such that for every \(\omega\in \Omega\setminus N\) and for any polynomial \(p\) in \(r+s\) non-commuting variables, \[ \lim_{n\rightarrow\infty}\| p(X_1^{(n)}(\omega),\ldots,X_{r+s}^{(n)}(\omega))\| =\| p(x_1,\ldots,x_{r+s})\| . \] The above limitation is also satisfied for the sets of independent Gaussian random matrices with symplectic entries \(\text{GSE}(n,\frac{1}{n}\)) and random matrices ensembles related to these \(\text{GSE}^*(n,\frac{1}{n})\).