Stochastic curvature flows: Asymptotic derivation, level set formulation and numerical experiments (Q5946037)

A stochastic Ginzburg-Landau equation \[ u^\varepsilon_{t} -\Delta u^\varepsilon + \varepsilon^{-2}F^\prime(u^\varepsilon) + \sqrt{2}\varepsilon^{(d-2)/2}\beta^{-1/2}J(r,t) = 0,\tag{1} \] where \((r,t)\in\mathbb R^{d}\times\mathbb R_{+}\), \(J\) is a white noise on \(\mathbb R^{d}\times\mathbb R_{+}\) and \(F\) is a symmetric double-well potential, is studied in the paper in connection with investigation of effects of microscopic random fluctuations on macroscopic models of phase transitions. Suppose that the solution \(u^\varepsilon\) to (1) has a formal asymptotic expansion \[ u^\varepsilon(r,t) = p(z,r, \varepsilon^{-2}t,t,\varepsilon^2t) + \varepsilon q(z,r, \varepsilon^{-2}t,t,\varepsilon^2t) + O(\varepsilon^2), \] where \(z= \varepsilon^{-1}\varphi(r,t,\varepsilon^2 t)\) and \(\varGamma _{t} \equiv \{r\in\mathbb R^{d}; \varphi(r,t) = 0\}\) describes an evolving surface of codimension one in \(\mathbb R^{d}\) (the interface). In the first part of the paper it is shown, in a non-rigorous way, that \(\varphi\) satisfies a stochastic equation \[ \varphi_{t} - \mu\sigma\text{tr}\left(\left( I - {\nabla\varphi\otimes\nabla\varphi \over |\nabla\varphi |^2}\right)\nabla^2\varphi \right) + \sqrt{2\mu}\beta^{-1/2}|\nabla \varphi|{\mathcal J}(\varrho, t) = 0, \] for \((\varrho,t)\in \varGamma_{t}\times \mathbb R_{+}\), \({\mathcal J}\) being again a white noise; which means that \(\varGamma_{t}\) evolves as stochastic mean curvature flow. In the second part of the paper, numerical approximations to (1) are discussed and results of numerical experiments are presented.