Title:Strongly Asymptotically Optimal Schemes for the Strong Approximation of Non-Lipschitzian Stochastic Differential Equations with respect to the Supremum Error

Abstract:Our subject of study is strong approximation of systems of stochastic differential equations (SDEs) with respect to the supremum error criterion, and we seek approximations that perform strongly asymptotically optimal. In this context, we focus on two principal sequences of classes of approximations, namely, the classes of approximations that are based only on the evaluation of the initial value and on at most finitely many sequential evaluations of the driving Brownian motion on average and the classes of approximations that are based only on the evaluation of the initial value and on at most finitely many evaluations of the driving Brownian motion at equidistant sites. On the one hand, for SDEs with globally Lipschitz continuous coefficients, Müller-Gronbach [Ann. Appl. Probab. 12 (2002), no. 2, 664-690] showed that specific Euler-Maruyama schemes corresponding to adaptive and to equidistant time discretizations perform strongly asymptotically optimal in these classes. On the other hand, for SDEs with super-linearly growing coefficients, the main theorem of Hutzenthaler et al. [Proc. R. Soc. Lond. Ser. A Math. Phys. Eng. Sci. 467 (2011), no. 2130, 1563-1576] implies that the errors of these particular approximations tend to infinity as the numbers of discretization sites tend to infinity. In the present article, we generalize the results of the first mentioned reference such that SDEs of the latter type are incorporated. More precisely, we show under rather mild assumptions on the underlying SDE, notably Khasminskii-type and monotonicity conditions, that specific tamed Euler schemes corresponding to adaptive and to equidistant time discretizations perform strongly asymptotically optimal in the two aforementioned sequences of classes. To illustrate our findings, we numerically analyze the SDE relating to the Heston-$3/2$-model originating from mathematical finance.