Ergodicity for the stochastic quantization problems on the 2D-torus (Q529631)

This is a strongly worded paper placed at the crossroad of stochastic quantization and Euclidean field theory. The authors recall indeed that in 1981 ``\textit{G. Parisi} and \textit{Y. S. Wu} [Sci. Sin. 24, No. 4, 483--496 (1981; Zbl 1480.81051)] proposed a program for Euclidean quantum field theory of getting Gibbs states of classical statistical mechanics as limiting distributions of stochastic processes, especially those which are solutions to non-linear stochastic differential equations. Then, one can use the stochastic differential equations to study the properties of the Gibbs states.'' In particular, the \(\Phi_2^4\)-quantum field in Euclidean quantum field theory (the simplest non-trivial Euclidean quantum field) consists in the measure \[ \nu(d\phi)=c\, e^{-2\int:q(\phi):\,d\xi}\mu(d\phi) \] where \[ q(\phi)=\frac{a_1}{4}\,\phi^4-\frac{a_2}{2}\,\phi^2 \] \(a_1>0\), \(a_2\) is real and \(\mu\) is the Gaussian free field. The Wick powers \(:\phi^n:\) are discussed in the Section 2. This measure happens to be an invariant measure for the following \textit{SDE} on the torus \(\mathbb{T}^2\) \[ dX=(AX-a_1:\!X^3\!:+\,a_2X)\,dt\qquad\quad X(0)=x \] where (to be concise we neglect a specification of the definition domains) \(Ax=\Delta x-x\), and \(W\) is a Wiener process. Now ``the issue of the stochastic quantization of the \(\Phi_2^4\) model is to solve the [previous] equation and to prove that the invariant measure is the limit of the time marginals as \(t\to\infty\)'' (ergodicity). This problem has been tackled in several ways in the previous decades, also in the case of the torus, namely in finite volume. In particular, the authors already solved an equivalent problem (based on the so called \textit{shifted} form of the previous equation) and proved that the \(\Phi_2^4\) quantum field \(\nu\) is in fact an invariant measure for the process \(X_0\) solution of our equation. It is all too natural to ask, however, whether this \(\nu\) is the \textit{unique} invariant measure for \(X_0\), because in this event \(\nu\) would also be the limiting distribution of the stochastic processes \(X_0\), and ``this problem is main point in the stochastic field quantization as we mentioned above in the \(\Phi_2^4\) model on the torus.'' This question too has been previously analyzed, but the arguments used there can not be applied for the torus, so that the authors undertook to study it in this paper by using the techniques from \textit{SPDE}: they analyze indeed the shifted equation directly and obtain that \(\nu\) is the unique invariant measure of \(X_0\), as summarized in their main result stated in Theorem 1.1. There they also prove the weak convergence to \(\nu\) of the associated semigroup. The paper contains also several other results (in Theorems 1.4 and 1.5) giving a characterization of \(\nu\) in terms of its density under translation, and a uniqueness result for the \(L\)-symmetrizing measures, where \(L\) is the corresponding generator. The core of the paper is of course the long final Section 4 entirely devoted to the proof of the results, while the remainder (Sections 2 and 3) is consecrated to preliminaries, notations and to the necessary estimates.