A numerical method for SDEs with discontinuous drift (Q285276)

Existence and uniqueness of a solution is proved for the stochastic differential equation (SDE) \(dX_t=\mu(X_t)\,dt+ \sigma(X_t)\,dW_t\), \(X_t= x\), where \(W\) is a standard Brownian motion, \(\mu\) is piecewise Lipschitz, and \(\sigma\) is globally Lipschitz and bounded away from zero. The proof relies on transforming the SDE into one with Lipschitz coefficients. As this transform is explicit and constructive, a numerical scheme for approximating the solution of the SDE is obtained in which the solution of the transformed SDE is approximated using the Euler-Marayama method and then inverse transformed. The scheme is proved to converge with order \({1\over 2}\). Examples are presented which illustrate the method and show its accuracy.