Normalized stochastic integrals in topological vector spaces

From MaRDI portal





Let \(E\) be a quasicomplete locally convex topological vector space with weakly separable dual space \(E'\), i.e. \(E\) is a locally convex topological vector space so that all its bounded closed subsets are complete. Let \((Q_s)_{s\geq 0}\) be a predictable family of symmetric nonnegative linear forms from \(E'\) into \(E\) and \(\lambda_s\) be a predictable increasing process. By a locally square integrable cylindrical martingale in \(E\) (with covariance operator function \(Q_s\) and quadratic variation \(\int^t_0 Q_s d\lambda_s)\), the authors understand a family of real-valued locally square integrable martingales \(M_t(y)\), \(y\in E'\), such that \[ \bigl \langle M(y),M(y')\bigr\rangle_t=\int^t_0\langle Q_sy,y'\rangle_{E,E'} d\lambda_s. \tag{*} \] The stochastic integral (*) is constructed in Section 2. In Section 3 the authors prove that the stable space of a locally square integrable continuous cylindrical martingale \(\{M_t(y'),\;y'\in E'\}\) coincides with the set of normalized integrals \[ {\mathcal L}^1(M)= \left\{ {\mathcal R}(f): E\left[\left( \int^1_0 | f_s |^2_{H_s} d\lambda_s \right)^{1/2} \right]< \infty \right\}. \] In Section 4 the authors discuss various particular cases of the normalized integral. These include Hilbert-valued stochastic integrals, stochastic integrals with respect to orthogonal martingale measures, stochastic integrals with respect to stochastic flows, etc. In Section 5 the authors apply the same ideas for integration of vector-valued functions with respect to martingale measures. For the entire collection see [Zbl 0893.00035].











This page was built for publication: Normalized stochastic integrals in topological vector spaces

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4213419)