Dupuis, Paul; Williams, Ruth J. Lyapunov functions for semimartingale reflecting Brownian motions. (English) Zbl 0808.60068 Ann. Probab. 22, No. 2, 680-702 (1994). Consider a nonnegative orthant \(S\) in \(R^ d\) in which a Brownian motion with constant drift \(r^ 0\) and covariance \(\Delta\) evolves. When it hits the \(i\)-th face of \(S\), it is reflected with direction \(r^ i\). Such a “semimartingale reflecting Brownian motion” (SRBM) associated with \((S,r^ 0, r^ 1, \dots, r^ d, \Delta )\) models certain problems in queueing networks under heavy traffic. The authors show that a sufficient condition for the SRBM to be positive recurrent is that all solutions of a related deterministic Skorokhod problem are attracted to the origin, i.e. the paths stay arbitrarily near zero after a time \(T\). The method of proving ergodicity uses the construction of a smooth Lyapunov function for the SRBM. Reviewer: M.Kohlmann (Bonn) Cited in 4 ReviewsCited in 61 Documents MSC: 60J60 Diffusion processes 60J65 Brownian motion 60K25 Queueing theory (aspects of probability theory) 34D20 Stability of solutions to ordinary differential equations Keywords:semimartingale reflecting Brownian motion; recurrene; Skorokhod problem; Lyapunov function PDFBibTeX XMLCite \textit{P. Dupuis} and \textit{R. J. Williams}, Ann. Probab. 22, No. 2, 680--702 (1994; Zbl 0808.60068) Full Text: DOI