No label defined (Q4493411)

Let \(X\) be a separable Banach space with dual \(X^*\), and \((\Omega,{\mathcal F},P)\) be a probability space. Let \((W_t)_{t\in [0,1]}\) be a Wiener process in \(X\) with Gaussian covariance operator \(R: X^*\to X\). Let \(Y\) be another Banach space. The definition of nonanticipating function \(\varphi: [0,1]\times \Omega\to L(X, Y)\) is given in order to define a generalized stochastic integral \(\xi= \int^1_0 \varphi dW_t\) of a random function \(\varphi\) with respect to the Wiener process. The following Itô formula is proved: Let \(\psi: [0,1]\times X\to Y\) be a continuous function with continuous Fréchet derivatives \(\psi_t': [0,1]\times X\to Y\), \(\psi_x': [0,1]\times X\to L(X, Y)\) and \(\psi''_{xx}: [0,1]\times X\to L(X, L(X, Y))\). Then \[ \psi(t, W_t)= \psi(0,0)+ \int^t_0 \psi_t'(s, W_s) ds+ \int^t_0 \psi_x'(s, W_s) dW_s+ {1\over 2} \int^t_0 \Biggl(\sum^\infty_{i=1} \psi''_{xx}(s, W_s) a_ia_i\Biggr) ds, \] where \(Rx^*= \sum^\infty_{k=1} \langle a_k,x^*\rangle a_k\) is the covariance operator of \(W_1\).