On Wong-Zakai approximation of stochastic differential equations (Q791231)

This paper is devoted to the proof of the following theorem: Let X(t) be the solution of the stochastic differential equation (SDE) \(dX(t)=F(X(t))dZ(t)\), where Z is a continuous vector martingale and F a smooth matrix-valued mapping. Let \(X_ n\) be the solution of the ODE which is obtained by replacing Z by \(Z_ n\), where \(Z_ n\) is a piecewise monotonic approximation of Z. Let denote \(DF=(F^ 1\!_ j\partial F_ k/\partial x^ 1;j,k=1,...,p)\) and \(\ll Z,Z\gg =(<Z^ j,Z^ k>)\). We establish the convergence of \(X_ n\) to the solution of the ''corrected'' SDE \(dX(t)=F(X(t))dZ(t)+(1/2)DF(X(t))d\ll Z,Z\gg_ t\) uniformly in probability on compact intervals. This result also yields an improvement of a theorem of \textit{S. I. Marcus} [Stochastics 4, 223-245 (1981; Zbl 0456.60064)] concerning SDE driven by discontinuous semimartingales.