On Linear Row-Finite Systems of Stochastic Differential Equations (Q4328501)

Let \({\mathcal A}\) be a continuous linear operator on the non-normable topological vector space \(E\) and \(f\) a continuous \(E\)-valued function on \([0,\infty)\). Then it is known that, in general, neither the deterministic differential equation \[ \dot{X}(t) = {\mathcal A}X(t) + f(t), \qquad X(0)=X_0 \in E,\tag{1} \] nor the stochastic differential equation \[ d\dot{X}(t) = {\mathcal A}X(t) dt + f(t) dt + dW(t),\qquad X(0,\omega)=X_0(\omega) \in E,\tag{2} \] (\(W(t)\) a Wiener process) need to have a solution. However, it is also known that if \(E={\mathbb R}^{\infty}\) is the sequence space with the weak topology of coordinatewise convergence and \(W_t\) is a natural Wiener process on \({\mathbb R}^{\infty}\) (that is an iid sequence of standard Wiener processes \(W_n(t)\) with the property that \(\sum_{n=1}^{\infty}{k_nW_n(t)}\) is a standard Wiener process for any \(k \in l^{2}\) with \(\sum_{n=1}^{\infty}{k_n^2}=1\)), then both the deterministic and the stochastic differential equations are solvable. Using methods originally introduced for proving the solvability of (1), the author generalizes the result for the stochastic situation and shows that (2) still has a strong solution if the Wiener process is replaced by an adapted and continuous (i.e. predictable) or \(L^2\)-continuous stochastic process.