Strong convergence of an explicit numerical method for SDEs with nonglobally Lipschitz continuous coefficients (Q453249)

The tamed Euler method, a modification of the explicit Euler method, is introduced and is proved (unlike the explicit Euler method) to generate approximations that converge strongly with order \({1\over 2}\) to the solution of the stochastic differential equation (SDE) NEWLINE\[NEWLINEdX_t= \mu(X_t)\,dt+ \sigma(X_t)\,dW_t,\quad X_0= \xi,NEWLINE\]NEWLINE where \(\mu\) is globally one-sided Lipschitz continuous with a continuous derivative that grows at most polynomially, \(\sigma\) is globally Lipschitz continuous, and \(W_t\) is an \(m\)-dimensional Brownian motion. For three examples results are given that demonstrate that the tamed Euler method requires much less computer running time to provide approximations with error less than 0.001 than does the implicit Euler method.