scientific article; zbMATH DE number 1409863

Random matrices (algebraic aspects) (15B52) Limit theorems for vector-valued random variables (infinite-dimensional case) (60B12) Functional limit theorems; invariance principles (60F17)

Review text

The author considers a sequence of random matrices \(\{ X_{n}\mid n=\overline{1,\infty}\}\) of order \(2\times 2\) with terms from the field of complex numbers \(C\) such that the \(X_{n}\) are independent and have the Bernoulli distribution: for every \(n\geq 1\) \[ P\{ X_{n}=A\}=p, \quad P\{ X_{n}=B\}=q\quad p+q=1, \quad 0<p<1, \tag{1} \] \[ A=\begin{pmatrix} a&b \\ 0&c \end{pmatrix},\quad a>0,\quad b\geq 0, \quad c>0, \tag{2} \] \[ B=\begin{pmatrix} d&0\\ f&g\end{pmatrix}, \quad d>0, \quad f\geq 0, \quad g>0. \tag{3} \] This paper is devoted to the problem of asymptotic behavior of the products \[ Y_{n}=X_{n}X_{n-1} \ldots X_2 X_1,\quad n=\overline{1,\infty}. \] The asymptotic behaviour of the sequence \(Y_{n}\) is characterized by the Lyapunov exponent \(\kappa=\lim_{n\to\infty}n^{-1} \ln\|Y_{n}\|=\lim_{n\to\infty}n^{-1}E \ln\|Y_{n}\|\), where \(\|\cdot\|\) is the usual operator norm in \(L(C^2).\) The author presents an expression for \(\kappa\) in explicit form as some functional from \(P\), \(A\) and \(B.\) The main result of this paper is the following assertion: Let matrices \(A\) and \(B\) in the Bernoulli scheme (1) belong to space \(L(R^2)\) and be of structure (2) and (3), respectively. Then the exponential growth exponent \(\kappa\) of the semigroup \(Y_{n}\) is of the form \(\kappa=q\ln g+p\ln c+qE\ln\Psi_{\infty}\), where \(\Psi_{\infty}\) is the infinite continuous fraction \[ \Psi_{\infty}= 1+\alpha_1+\beta_1- {{\alpha_1}\over\displaystyle {1+\alpha_2+\beta_2}-{\strut{\alpha_2}\over\displaystyle {1+\alpha_3+\beta_3-\ldots} {\strut{\alpha_{n}}\over {{1+\alpha_3+\beta_3-\ldots}}}}}, \] \(\alpha_{n}=dg^{-1}a^{\Theta_{n}}c^{-\Theta_{n}}\), \(\beta_{n}=fg^{-1}b(\Theta_{n})c^{-\Theta_{n}}\), where \(\{\Theta_{n}\mid n=1,2,\ldots,\infty\}\) is the totality of the mutually independent geometrically distributed random variables with the mean \(E\Theta_{n}=q^{-1}p\).

Recommendations

Cited in

(37)

This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4940274)