Stationary directed polymers and energy solutions of the Burgers equation (Q2196537)

The authors prove an approximation in distribution in the space \(C([0,T],\mathcal S')\) of the unique energy solution to the stochastic Burgers equation \[ \partial_t u = \frac 12\partial_x^2 u + c\partial_x u - \frac 12 \partial_x(u^2) + \partial_x\mathcal W, \] where \(c\in\mathbb R\) is some constant and \(\mathcal W\) is space-time white noise. The approximation consists of a sequence of processes \(\mathcal X_t^n\), \(t\in [0,T]\), of the following form \[ \mathcal X_t^n(\phi) = \sum_j (u^n(tn,j) - \rho_n)\phi\left(\tfrac{j-nt-a_n\sqrt n}{\sqrt n}\right),\quad\phi\in C_c^\infty. \] Here, \((a_n)_n\) is some increasing sequence such that \(a_n\uparrow\infty\) and \(a_n n^{-\frac 12}\to 0\), \(\rho_n = \mathbb E u^n\) with \(u^n(t,j) := u_{\beta,\theta}(t,j)\) with \(\beta=n^{-1/4}\) and \(\theta = 1 + 1/(2\sqrt n)\). The \(u_{\beta,\theta}(t,j)\) are stochastic processes of the form \[ u_{\beta,\theta}(t,j) = \log Z_{\beta,\theta}(t,j) - \log Z_{\beta,\theta}(t,j-1). \] with \begin{align*} Z_{\beta,\theta}(t,n) &= \idotsint\limits_{-\infty < s_0 < s_1 < \dots < s_{n-1}< t} \exp\big[ \beta B^{(0)}(s_0) + \theta s_0 \\ &\qquad\mbox+\beta\big\{B^{(1)}(s_0,s_1)+\dots+B^{(n)}(s_{n-1},t)\big\}\big] ds_0 \,ds_1\dots ds_{n-1}. \end{align*} By \(B^{(k)}(u,v)\) the increment \(B^{(k)}(u)-B^{(k)}(v)\) is meant and \(B^{(0)}\) is a Brownian motion which is independent of the standard Brownian motions \(B^{(j)}(t)\), \(j=1,2,\dots, n\). Note that all Brownian motions are two-sided in time.