No label defined (Q3973414)

[For the entire collection see Zbl 0725.00015.] The Dirichlet problems for a class of singularly perturbed systems of PDE are considered. We use the stochastic representation for solutions of such systems that based on the exit distributions for the Markov process \((X^ \varepsilon_ t,\nu^ \varepsilon_ t)\) on the phase space \(\mathbb{R}^ r\times\{1,\ldots,n\}\), where \(X^ \varepsilon_ t\) satisfies SDE: \[ dX^ \varepsilon_ t = b_{\nu^ \varepsilon_ t}(X^ \varepsilon _ t)dt+\varepsilon\sigma_{ \nu^ \varepsilon_ t}(X^ \varepsilon_ t)dW_ t,\quad \varepsilon>0, \] \(W_ t\) is a Wiener process in \(\mathbb{R}^ r\) and the transition probabilities for \(\nu^ \varepsilon_ t\) have the form: \[ P_{x,k}\{\nu^ \varepsilon_ t=i\}=td_{ki}(x)+o(t),\quad t\downarrow 0. \] The conditions when these solutions converge to the generalized solution of the corresponding degenerate system are found. Some estimates of this convergence are obtained.