No label defined (Q4940292)

Let \((\Omega,{\mathcal F},{\mathcal F}_{t},P)\) be a probability space with filtration \({\mathcal F}_{t}\), \(t\in[0,T].\) The author considers the Itô stochastic equation \[ \xi_{\varepsilon}(t)=x+\int_0^{t}[b_{\varepsilon}(\xi_{\varepsilon}(s))+ g_{\varepsilon}(\xi_{\varepsilon}(s))]ds+ \int_0^{t}\sigma_{\varepsilon} (\xi_{\varepsilon}(s)) dw(s), \quad t\in[0,T], \tag{1} \] where \((w(t),{\mathcal F}_{t})\) is a standard Wiener process and \(\varepsilon\) is a small parameter. For any \(\varepsilon>0\) the functions \(b_{\varepsilon}(x)\), \(g_{\varepsilon}(x)\) and \(\sigma_{\varepsilon}(x)\) are measurable and such that there exists a unique weak solution of equation (1). There exist some constants \(0<\lambda<\Lambda\) such that \(|g_{\varepsilon}(x)|\leq\Lambda\), \(\lambda\leq \sigma_{\varepsilon}(x)\leq\Lambda.\) For any \(x\in R\) \(\bigl|\int_0^{x}{b_{\varepsilon}(y)\over \sigma_{\varepsilon}^2(y)} dy\bigr|\leq\Lambda.\) The author introduces functions \(b(x)\), \(g(x)\), \(\sigma(x)\) which are close to \(b_{\varepsilon}(x)\), \(g_{\varepsilon}(x)\), \(\sigma_{\varepsilon}(x)\) and considers the Itô stochastic equation \[ \xi(t)=x+\int_0^{t}B(\xi(s)) ds+ \int_0^{t}\sigma(\xi(s)) dw(s), \tag{2} \] where \(B(x)=b(x)+g(x).\) Let \(\mu^{\varepsilon}\) be a sequence of probability measures induced by the random processes \(\xi^{\varepsilon}(t)\) on \(C([0,T])\) and let \(\mu\) be a measure induced by the random process \(\xi(t)\) on \(C([0,T]).\) The author finds sufficient conditions for weak convergence of the measures \(\mu^{\varepsilon}\) to the measure \(\mu.\)