Limit theorem for multidimensional renewal equation (Q2132094): Difference between revisions

This paper focuses on the extension to matrices of well-known results on limit theorems for renewal processes. The renewal equation in matrix form is written as \[X^{\varepsilon}(t)=A^{\varepsilon}(t)+\int_0^tF^{\varepsilon}(du)X^{\varepsilon}(t-u),\] where \(t>0\), \(\varepsilon >0\), \(X^\varepsilon\), \(A^\varepsilon\) are stochastic matrix-valued stochastic processes and \(F^\varepsilon\) is a matrix-valued positive measure given at the outset. The main result of the paper is the weak convergence of measures \(F^\varepsilon\) towards a matrix-valued measure \(F\) which is block-diagonal decomposable. The authors consider \(F^\varepsilon\) decomposed in blocks as \[F^\varepsilon=F+g_1(\varepsilon)B+g_2(\varepsilon)B^2+\ldots+g_n(\varepsilon)B^n+\] So, the result naturally depends on the convergence to zero of the coefficients \(g_1(\varepsilon),\ldots, g_n(\varepsilon)\) as \(\varepsilon\downarrow 0\). The authors use standard Laplace transform methods together with elementary matrix decomposition.