Diffusion processes on graphs: Stochastic differential equations, large deviation principle (Q1974437)

The authors study diffusion processes on a connected graph consisting of vertices \(O_{k}\), \(k=1,2,\ldots,M\), and edges \(I_{i}\), \(i=1,2,\ldots,N\). On the edges the definition of such process is easy and one should prescribe the behaviour of the process after it reaches the vertices [see also \textit{M. I. Freidlin} and \textit{A. D. Wentzell}, Ann. Probab. 21, No. 4, 2215-2245 (1993; Zbl 0795.60042)]. This is done using sets of nonegative constants \(\alpha_{k}\) and \(\alpha_{ki}\) for each \(O_{k}\) and each \(I_{i}\) having \(O_{k}\) as one of the end points. In the present paper the authors develop stochastic analysis for such processes and give a useful description for them. The main point is to derive Itô's differential rule for the processes. Then large deviation principles for these processes and their local times at the vertices are studied and applications are given [see also \textit{T.-S. Chiang} and \textit{S.-J. Sheu}, Stochastic Anal. Appl. 15, No. 1, 31-50 (1997; Zbl 0873.60013) or \textit{A. P. Korostelev} and \textit{S. L. Leonov}, Theory Probab. Appl. 37, No. 3, 543-550 (1992); translation from Teor. Veroyatn. Primen. 37, No. 3, 570-576 (1992; Zbl 0765.60078)].