Stochastic Taylor Expansions for the Expectation of Functionals of Diffusion Processes

From MaRDI portal



Abstract: Stochastic Taylor expansions of the expectation of functionals applied to diffusion processes which are solutions of stochastic differential equation systems are introduced. Taylor formulas w.r.t. increments of the time are presented for both, It^o and Stratonovich stochastic differential equation systems with multi-dimensional Wiener processes. Due to the very complex formulas arising for higher order expansions, an advantageous graphical representation by coloured trees is developed. The convergence of truncated formulas is analyzed and estimates for the truncation error are calculated. Finally, the stochastic Taylor formulas based on coloured trees turn out to be a generalization of the deterministic Taylor formulas using plain trees as recommended by Butcher for the solutions of ordinary differential equations.


The author considers systems of stochastic ordinary differential equations (SODEs) both in Itô and Stratonovich formulation, driven by a multi-dimensional Wiener process. In order to develop and analyse weak approximation methods, one needs stochastic Taylor expansions of \({\mathbf E}f(X_t)\), the expectation of a functional \(f\) of the solution \(X_t\) to the SODEs in both formulations. In this well-written article a representation of \({\mathbf E}f(X_t)\) in terms of truncated Taylor expansions and remainder terms is proved. Further, estimates of the remainder terms are given. The representations and the proofs are based on multi-coloured stochastic rooted tree theory. This generalizes the rooted tree theory well-known in the deterministic analysis of Runge-Kutta methods. In particular, the Taylor expansions derived here coincide with those derived in the deterministic setting if the diffusion term vanishes.




Cited in
(40)








This page was built for publication: Stochastic Taylor Expansions for the Expectation of Functionals of Diffusion Processes

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3158191)