Lattice approximation to the dynamical \(\Phi_{3}^{4}\) model (Q1747755)

This paper deals with the lattice approximations to the dynamical \(\Phi_3^4\) model by paracontrolled distributions. For \(n\geq 1\), let \(\Lambda^{N}=\{-N,-(N-1),\ldots,N\}^3\). Set \(\varepsilon=2(2N+1)^{-1}\). Every point \(k\in\Lambda^ N\) can be identified with \(x=\varepsilon k\in \Lambda_{\varepsilon}=\{x=(x^1,x^2,x^3)\in\varepsilon {\mathbb Z}^3:-1<x^{i}<1, i=1,2,3\}\). Let \(W_{N}(x)=2^{-3/2}\sum_{| k|_{\infty}\leq N}\beta_{k}e^{i\pi k\cdot x}\) for \(x\in\Lambda_{\varepsilon}\) be a cylindrical Wiener process on \(L^2(\Lambda_{\varepsilon})\). Here, \(\{\beta_{k}\}\) is a family of complex-valued Brownian motions with \(\bar \beta_{-k}(t)=\beta_{k}(t)\), \(E[\beta_{k_1}(t_1)\beta_{k_2}(t_2)]=1_{\{k_1+k_2=0\}}t_1\wedge t_2\) and \(| k|_{\infty}=\max(| k^1|, | k^2|, | k^3|)\). For \(Y:\;\Lambda_{\varepsilon}\to {\mathbb R}\) it is defined \(\text{Ext}\, Y(x)=2^{-3}\sum_{k\in\{-N,\ldots,N\}^3}\sum_{y\in \Lambda_{\varepsilon}}\varepsilon^3 e^{i\pi k\cdot(x-y)}Y(y)\), and for \(x\in\Lambda_{\varepsilon}\) it is defined \(\Delta_{\varepsilon}f(x)=\varepsilon^{-2}\sum_{y\in\Lambda_{\varepsilon},y\sim x}(f(y)-f(x))\), and the nearest neighbor relation \(x\sim y\) is to be understood with periodic boundary conditions on \(\Lambda_{\varepsilon}\). The main result of this paper is the following theorem. Let \(\kappa\in(1/2,2/3)\) and \(\Phi_0\in C^{-\kappa}\), where \(C^{\alpha}, \alpha\in{\mathbb R}\) is the Hölder-Besov space. Let \((\Phi,\tau)\) be the unique (maximal in time) solution to equation \(d\Phi=(\Delta\Phi-\Phi^3)dt+dW(t)\), \(\Phi(0)=\Phi_0\), where \(W\) is a cylindrical Wiener process, and let for \(\varepsilon\in (0,1)\) the function \(\Phi^{\varepsilon}\) be the unique solution to equation \(d\Phi^{\varepsilon}(t,x)=(\Delta_{\varepsilon}\Phi^{\varepsilon}(t,x)-(\Phi^{\varepsilon})^3(t,x)+(3C_0^{\varepsilon}- 9C_1^{\varepsilon})\Phi^{\varepsilon}(t,x))dt+d W_{N}(t,x)\), \(\Phi_{\varepsilon}(0)=\Phi^{\varepsilon}_0\), where \(C_0^{\varepsilon},C_1^{\varepsilon}\) are some constants defined in the paper. If \(\Phi^{\varepsilon}_0-\Phi_0\to 0\) in \(C^{-\kappa}\), then there exists a sequance of random times \(\tau_{L}\) such that \(\lim_{L\to\infty}\tau_{L}=\tau\) and \(\sup_{t\in[0,\tau_{L}]}\|\text{Ext}\,\Phi^{\varepsilon}-\Phi\|_{-\kappa}\to0\) in probability, as \(\varepsilon\to0\).