The spectral radius of large random matrices (Q1083753): Difference between revisions

The main result gives the following theorem: Let \(M_ n=\{m_{ij}(n)\}_{1\leq i,j\leq n}\) be a sequence of \(n\times n\) random matrices with \(m_{ij}(n)\), \(1\leq i,j\leq n\), independent and identically distributed random variables for each n. Assume, for each n, \(Em_{11}(n)=0\), \(Em^ 2_{11}(n)=\sigma^ 2\) and \(E| m_{11}(n)|^ p\leq p^{\alpha p}\), for all \(p\geq 2\), some \(\alpha\). Then \[ \limsup_{n\to \infty}\rho_ n\leq \sigma \quad a.s., \] where \(\rho_ n=\) spectral radius of \(M_ n/\sqrt{n}=\max \{| \lambda |:\lambda\) eigenvalue of \(M_ n/n\}\).