scientific article; zbMATH DE number 4135179

From MaRDI portal
Publication:3468422





[For the entire collection see Zbl 0668.00026.] Let \(\Lambda \subset R^ 2\) be a finite open rectangle and \(S'\) denote \({\mathcal D}'(\Lambda)\) the space of distributions on \(\Lambda\) and let \({\mathcal S}'\) denote the space of tempered distributions on \(\Lambda\). Let \(C_ i=(-\Delta +I)^{-1}\), \(i=1,2\) with Dirichlet (resp. free) boundary conditions on \(\Lambda\). \(C_ i\), \(i=1,2\) are covariance operators and for C a covariance operator let C(\(\cdot.\cdot)\) denote its integral kernel, \(C^{\alpha}\) its \(\alpha^{th}\) operator power and let \(\mu_ C\) denote the centered Gaussian measure with variance operator C. Consider the following \(S'\)-valued stochastic differential equation, \[ d\phi (t)=-C_ 1^{-\epsilon}\phi (t)dt+dw(t),\quad \phi (0)=\phi \in S',\quad 0<\epsilon <1, \] where W(t) is a Wiener process with covariance \(C_ 1^{1-\epsilon}\). It is not difficult to prove that this equation has a unique solution and has a path continuous version as an \(H^{-\alpha}\)-valued process on (0,\(\infty)\). Moreover \(\phi\) (\(\cdot)\) is ergodic and has \(\mu_{C_ 1}\) as its unique invariant measure. The same claims can be made with \(C_ 1\) replaced by \(C_ 2\). This procedure of creating a stochastic differential equation with unique invariant measure a desired invariant measure is termed stochastic quantization.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3468422)