Malliavin's calculus and stochastic integral representations of functional of diffusion processes†

From MaRDI portal
Publication:3330239





Let F be a Fréchet differentiable function on the space C[0,1] with continuous paths, with Fréchet derivative \(dF(b)(u)=\int^{1}_{0}u(s)\lambda^ F(ds;b)\) for b and u in C[0,1]. If \(B=\{B(t)| 0\leq t\leq 1\}\) is a Brownian motion and \(\beta_ t=\sigma \{B(s)| o\leq s\leq t\}\) and \(EF(B)=0\), Clark's formula states that \(F(b)=\int^{1}_{0}E\{\lambda^ F((s,1],\cdot)| \beta_ s\}dB(s)\). This paper extends Clark's formula to functionals F which are weakly H-differentiable in the sense of Shigekawa. For such F, \[ E\{F(b)\int^{1}_{0}<\alpha(s),dB(s)>\}=E\{\int^{1}_{0}<DF(b)'(s),\;alpha(s)>ds\} \] for all bounded, \(\beta_ t\)-adapted processes \(\alpha\) (s), where DF(b)(s), \(o\leq s\leq 1\), denotes the weak H-derivative of F. Clark's formula is a simple consequence of this result. A simple proof of this result is given using Stroock's development of the Malliavin calculus. Malliavin calculus techniques are also used to derive \textit{U. G. Haussmann's} [SIAM J. Control Optimization 16, 252-269 (1978; Zbl 0375.60070)] stochastic integral representation of a functional F(y) of the diffusion \(dy=m(t,y)dt+\sigma(t,y)dB\). In doing this, it is shown that y(t,B) is weakly H-differentiable if m and \(\sigma\) have bounded, continuous first derivatives in y.




Cited in
(70)








This page was built for publication: Malliavin's calculus and stochastic integral representations of functional of diffusion processes†

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3330239)