Santiago Cambronero,David Campos,C. A. Fonseca-Mora,Darío Mena
Abstract:Using the theory of stochastic integration developed recently by the authors, in this article is proved an Itô formula for Hilbert space-valued Itô processes defined with respect to a cylindrical-martingale valued measure. This formula is applied to prove an integration by parts formula for Itô processes, to prove a Burkholder inequality for the stochastic integral, and finally to prove an Itô formula for Hilbert space-valued Itô processes defined with respect to a Lévy process.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is to propose an Itô formula for Hilbert - space - valued Itô processes, which are defined with respect to cylindrical martingale - valued measures. Specifically, using the stochastic integral theory they recently developed, the authors prove such an Itô formula and apply it in the following aspects:
1. **Prove an integration - by - parts formula**: for two stochastic integral processes defined by cylindrical martingale - valued measures.
2. **Prove a Burkholder inequality**: for stochastic integrals of cylindrical martingale - valued measures with continuous paths.
3. **Prove an Itô formula**: for Hilbert - space - valued Itô processes driven by Lévy processes.
### Background and Motivation
The Itô formula is one of the most important tools in stochastic analysis and is widely used in theoretical and practical problems, such as solving stochastic differential equations. For the real - valued case and the Wiener process, the Itô formula was originally established by Kiyoshi Itô in [13]. Subsequently, Curtain and Falb [9] extended it to Hilbert - space - valued Itô processes driven by the Wiener process. Kunita [15] further extended it to Hilbert - space - valued Itô processes driven by square - integrable martingales. Métivier [18] extended it to Hilbert - space - valued semimartingales. In recent years, some studies have also established Itô formulas for other types of noise or in certain classes of Banach spaces [4, 6, 10, 11, 12, 19, 20].
### Main Contributions
The main contribution of this paper is to propose an Itô formula for Hilbert - space - valued Itô processes driven by cylindrical martingale - valued measures. Specifically, let \(H\), \(G\), and \(K\) be separable Hilbert spaces, \(A: \Omega\times\mathbb{R}^+\to\mathbb{R}\) be an adapted càdlàg process of locally finite variation, \(\mathcal{A}\) be a ring of subsets of the topological space \(U\), and \((M(t, A):t\geq0, A\in\mathcal{A})\) be a cylindrical martingale - valued measure. Let \(X\) be a \(G\)-valued Itô process:
\[dX_t=\Psi(t)dA(t)+\int_U\Phi(t, u)M(dt, du)\]
where the coefficients \(\Psi:\Omega\times\mathbb{R}^+\to G\) and \(\Phi:\Omega\times\mathbb{R}^+\times U\to HS(H, G)\) satisfy appropriate integrability conditions. If \(f:[0, T]\times G\to K\) is a function of class \(C^{1, 2}\), the authors prove the following Itô formula:
\[f(t, X_t)=f(0,\xi)+\int_0^t f_t(s, X_{s -})ds+\int_0^t\int_U f_x(s, X_{s -})\Phi(s, u)M(ds, du)\]
\[+\int_0^t f_x(s, X_{s -})\psi(s)dA(s)+\frac{1}{2}\int_0^t\int_U\text{Tr}(\Phi(s, u)Q^{1 / 2}_M c(f_{xx}(s, X_{s -}))\langle\langle M_c\rangle\rangle(ds, du)\]
\[+\sum_{0 < s\leq t}[f(s, X_s)-f(s, X_{s -})-f_x(s, X_{s -})(\Delta X_s)]\]
### Applications
To demonstrate the practicality of the proposed Itô formula, the authors introduce three applications in Section 5:
1. **Integration - by - parts formula**: for two stochastic integral processes defined by cylindrical martingale - valued measures.
2. **Burkholder inequality**: for cylindrical martingales with continuous paths.