Paul A. Razafimandimby

arXiv2018

Numerical approximation of stochastic evolution equations: Convergence in scale of Hilbert spaces

The present paper is devoted to the numerical approximation of an abstract stochastic nonlinear evolution equation in a separable Hilbert space {$\mathrm{H}$}. Examples of equations which fall into our framework include the GOY and Sabra shell models and { a class of nonlinear heat equations.} The space-time numerical scheme is defined in terms of a Galerkin approximation in space and a { semi-implicit Euler--Maruyama scheme in time}. {We prove the convergence in probability of our scheme by means of an estimate of the error on a localized set of arbitrary large probability.} Our error estimate is shown to hold in a more regular space $\mathrm{V}_β\subset \mathrm{H}$ with $β\in [0,\frac14)$ and { that the explicit rate of convergence of our scheme depends on this parameter $β$. }

math.NA↗

arXiv2016

Ergodicity of stochastic shell models driven by pure jump noise

In the present paper we study a stochastic evolution equation for shell (SABRA \& GOY) models with pure jump \levy noise $L=\sum_{k=1}^\infty l_k(t)e_k$ on a Hilbert space $\h$. Here $\{l_k, k\in \mathbb{N}\}$ is a family of independent and identically distributed (i.i.d.) real-valued pure jump \levy processes and $\{e_k, k\in \mathbb{N}\}$ is an orthonormal basis of $\h$. We mainly prove that the stochastic system has a unique invariant measure. For this aim we show that if the \levy measure of each component $l_k(t)$ of $L$ satisfies a certain order and a non-{degeneracy} condition and is absolutely continuous with respect to the Lebesgue measure, then the Markov semigroup associated {with} the unique solution of the system has the strong Feller property. If, furthermore, each $l_k(t)$ satisfies a small deviation property, then 0 is accessible for the dynamics independently of the initial condition. Examples of noises satisfying our conditions are a family of i.i.d tempered \levy noises $\{l_k, k\in \mathbb{N}\}$ and $\{l_k=W_k\circ G_k + G_k, k\in \mathbb{N} \}$ where $\{G_k, k \in \mathbb{N}\}$ (resp., $\{W_k, k\in \mathbb{N}\}$) is a sequence of i.i.d subordinator Gamma (resp., real-valued Wiener) processes with \levy density $f_G(z)=(\vartheta z)^{-1} e^{-\frac z\vartheta} \mathds{1}_{z>0}$. The proof of the strong Feller property relies on the truncation of the nonlinearity and the use of a gradient estimate for the Galerkin system of the truncated equation. The gradient estimate is a consequence of a Bismut-Elworthy-Li (BEL) type formula that we prove in the Appendix A of the paper.

math.PR↗