Large deviations and the Malliavin calculus (Q792536)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Large deviations and the Malliavin calculus |
scientific article |
Statements
Large deviations and the Malliavin calculus (English)
0 references
1984
0 references
This book uses the so-called Malliavin calculus and large deviation techniques to study the asymptotics as \(t\downarrow \downarrow 0\) of the conditional probabilities of bridges associated with certain hypoelliptic diffusions. In particular, stochatic differential equations of the form \(dx=\sum^{m}_{i=1}X_ i(x)\quad dw^ i;\quad x(0)=x_ 0\) are studied, where \(X_ 1,...,X_ m\) are smooth vector fields as on a manifold M. Let be \(x_ s=\psi_ s(w)\). Associated to the latter differential equation is the second order operator \(L=1/2\sum^{m}_{i=1}X^ 2_ i\). The law of \(\psi_ 1(w)\) is given by the kernel \(e^ L(x_ 0,.)\). Under certain standard assumptions of Hörmander, one considers the asymptotics of the law \(p_ t(x)\) of \(\psi_ 1(\sqrt{t}dw)\) as \(t\downarrow \downarrow 0.\) In the elliptic case, rather complete results are obtained. In case ''hypoelliptic'' is substituted for ''elliptic'', some basic machinery is developed and some conjectures are formulated. The special case of the Hessenberg group is considered in some detail. The book assumes a fair amount of sophistication on the part of the reader - some background in probability, Riemannian geometry, and the basic language of differential equations is advised. However the author is fairly kind to his readers. He works out special cases, writes out most details, and gives copious references. The author is able to keep his book fairly brief and yet provide an introduction to an active area of research.
0 references
diffusion processes
0 references
stochastic analysis on manifolds
0 references
Gaussian processes
0 references
Malliavin calculus
0 references
large deviation techniques
0 references
hypoelliptic diffusions
0 references