Limiting models for equations with large random potential: a review (Q2348476)

The authors review recent results on convergence of solutions \(u^\varepsilon\) of elliptic and parabolic equations with large and highly oscillatory random potentials, as \(\varepsilon\downarrow 0\), to solutions \(u^*\) of homogenized deterministic or stochastic equations. A special attention is paid to the rate of convergence of the error \(u^\varepsilon-u^*\). More precisely, it is shown that \(u^\varepsilon\) converges to \(u^*\) (in some sense) and sometimes it is also possible to show that the rescaled fluctuations \(\varepsilon^{-\alpha}(u^\varepsilon-u^*)\) converge in distribution to a deterministic or a random limit \(\xi\) that is statistically characterized as a random variable or a solution of a deterministic or a stochastic equation. In the elliptic case, equations of the form \(-\nabla\cdot q(\varepsilon^{-1}x,\omega)\nabla u^\varepsilon=f\) and \(P(x,D)u^\varepsilon+q(\varepsilon^{-1}x,\omega) u^\varepsilon=f\), amended with some boundary conditions, are considered for a deterministic elliptic operator \(P\) and a random potential \(q\). In the parabolic case, equations of the form \(\partial_tu^\varepsilon+P(D)u^\varepsilon+a\varepsilon^{-\beta}q(\varepsilon^{-\gamma}t,\varepsilon^{-1}x)u^\varepsilon=0\) on \(\mathbb R^d\) are considered for \(a\in\{-1,i\}\). The random potential is assumed to satisfy a boundedness, stationarity, ergodicity and other conditions. The authors demonstrate three techniques of proofs: a combinatorial technique based on the Duhamel expansion (the diagrammatic expansion), the Feynman-Kac formula and the standard multi-scale method.