Exponential mixing for finite-dimensional approximations of the Schrödinger equation with multiplicative noise (Q843090)

The author considers the problem \[ i\frac{dz}{dt}=Az+\beta (t)Bz+\varepsilon F(z), z(0)=z_0, \tag{1} \] where \(A\) and \(B\) are Hermitian matrices, \(F:\mathbb{C}^n \rightarrow \mathbb{C}^n\) is a real-analytic function such that the scalar product \(\langle F(z),z\rangle\) is real \(\forall z\in \mathbb{C}^n\) and \(\varepsilon\in \mathbb{R}\) is small constant, \(\beta (t)\) is a random process \(\beta (t)=\sum_{k=0}^{\infty} I_k (t)\eta_k (t-k), t\geq 0\), where \(I_k\) is the indicator function of the interval \([k,K+1)\) and \(\eta_k \) are independent identically distributed random variables in \(L_2([0,1],\mathbb{R})\). The restriction of the solution of (1) at integer times forms a Markov chain. The aim of the paper is the study of ergodicity of this chain. It is shown that the system has a unique stationary measure \(\mu\) on the unit sphere \(S\subset \mathbb{C}^n\) and \(\mu\) is absolutely continuous with respect to the Riemannian volume on \(S\). Moreover, for any initial condition in \(S\), the solution converges exponentially fast to the measure \(\mu\) in the variational norm.