A mixed random walk on nonnegative matrices: A law of large numbers (Q1900335)

Let \(Y_n\) be a sequence of i.i.d. Bernoulli random variables, and \((X_n)\) a sequence of i.i.d. nonnegative matrices. The so-called mixed random walks \((S_n)\) on the semigroup of nonnegative matrices is obtained by multiplying the \(X_n\) left or right according to \(Y_n\), i.e., \(S_n=S_{n-1} X_n\) if \(Y_n=0\) and \(S_n=X_nS_{n-1}\) if \(Y_n=1\). Let \(\mu\) be the distribution of \(X_1\) with support \(S\). The point of the paper is that \(S\) is not required to be compact. Under the condition of tightness of \((\mu^n)\) it is shown that a unique invariant probability measure for \((S_n)\) exists and a strong law of large numbers is proved.