Limit theorems for a sequence of nonlinear reaction-diffusion systems (Q2366181): Difference between revisions

The author follows Kotelenez in comparing the stochastic model \(X^ N\) with \(\psi^ N\), a spatially discretized version of the deterministic model. Let \(\psi(0)\in C^ 1([0,1])\), \(M\) be the unique (in distribution) Gaussian martingale defined on some probability space \((\Omega,\overline F,\overline F_ t,P)\) such that for any \(\beta>3/2\), \(M\in C([0,\infty);H_{-\beta})\) a.s. Let \(\alpha>1/2\) and let \(V_ 0\) be \(H_{-\alpha}\) valued and \(\overline F_ 0\) measurable with \(E\| V_ 0\|^ 2_{-\alpha}<\infty\). For \(k\geq\max(\alpha,1)\) and \(\psi(0)\in C^ k([0,1])\) consider the stochastic partial differential equation \[ dV(t)=(\Delta+R'(\psi(t)))V(t)dt+dM(t),\;V(0)=V_ 0. \] The author proves the following main result: Fix \(\alpha>1/2\) and assume (i) \(N/l\to 0\) as \(N\to\infty\), (ii) \((Nl)^{1/4}\| X^ N(0)-\psi^ N(0)\|_ 0\to 0\) in probability, (iii) \((Nl)^{1/2}(X^ N(0)- \psi^ N(0))\to V_ 0\) in distribution on \(H_{-\alpha}\), (iv) \(\psi(0)\in C^ k([0,1])\) for \(k\geq\max(4,\alpha)\). Then it holds: (a) For \(\gamma\geq\alpha\) and any sequence \(l_ N\) satisfying (i)--(iii), \((Nl)^{1/2}(X^ N-\psi^ N)\to V\) in distribution on \(D([0,\infty);H_{-\gamma})\) with the Skorokhod topology. (b) For a particular sequence \(l_ N\) satisfying (i)--(iii), (a) holds with \(\psi\) in place of \(\psi^ N\).