On convergence of approximations of Itô-Volterra equations (Q4357560)

The author considers the equation \[ \xi(t)=h(t)+\int_0^ta(t,s,\xi) d\alpha(s) +\int_0^tb(t,s,\xi) d\mu(s),\quad t\geq 0,\tag{1} \] for \({\mathbb{C}}\)-valued processes \(\xi.\) Here \(h\) is a \({\mathbb{C}}\)-valued process, \(\alpha\) is a finite-valued locally bounded variation continuous process, \(\mu\) is finite-valued local martingale and \(a,b\) are matrix functions. Let \(\delta:=\{0=s_0<s_1<s_2<\cdots,\;s_j\to\infty\};\;|\delta|=\sup_{j\geq 1} s_j\) and \[ \lambda_{\delta}(s)=s_{j-1},\;\;s\in[s_{j-1},s_j[,\quad j\in{\mathbb{N}}; \] \[ \nu_{\delta}(x,u)=x(0),\;\;u\in[0,s_1[; \] \ \[ \nu_{\delta}(x,u)={s_{j+1}-u\over s_{j+1}-s_j}x(s_{j-1})+{u-s_j\over s_{j+1}-s_j}x(s_j),\;\;u\in[s_j,s_{j+1}[,\quad j\in{\mathbb{N}}. \] The approximation of equation (1) is defined as follows \[ \xi_{\delta}(t)=h(t) +\int_0^ta(t,\lambda_{\delta}(s),\nu_{\delta}(\xi_{\delta})) d\alpha(s) +\int_0^tb(t,\lambda_{\delta}(s),\nu_{\delta}(\xi_{\delta})) d\mu(s),\quad t\geq 0. \] Under appropriate conditions it is proved that \(\xi_{\delta}\) converges in law to \(\xi\) as \(|\delta|\to 0.\) Some generalization of the Burkholder-Gundy inequality is also given. For the entire collection see [Zbl 0870.00040].