Exponential ergodicity for stochastic functional differential equations with Markovian switching (Q6150483)

The authors consider stochastic functional equations of type \[ d[X(t)-G(X_t,\Lambda(t))]=b(X_t,\Lambda(t))\,dt+\sigma(X_t,\Lambda(t))\,dW(t) \] or \[ dX(t)=b(X_t,\Lambda(t))\,dt+\sigma(X_t,\Lambda(t))\,dW(t)+\int_Uc(X_{t-},\Lambda(t),z)\widetilde N(dt,dz) \] in \(\mathbb R^m\) where \(X_t(\theta)=X(t+\theta)\) for \(\theta\in[-T,0]\), \(W\) is a finite-dimensional Wiener process, \(\widetilde N\) is a compensated Poisson random measure and \(\Lambda\) is a right-continuous, irreducible Markov switching taking values in a finite state space, possessing a unique invariant probability measure. It is proved that under suitable monotonicity and continuity assumptions on the coefficients \(b\), \(\sigma\), \(c\) and under several other assumptions on the parameters of the Markov process \(\Lambda\) and the Poisson random measure \(N\), the equation has a unique solution \(X\) such that the process \((X_t,\Lambda(t))\) is Markov and admits a unique invariant measure which is exponentially ergodic.