On pathwise uniqueness for stochastic heat equations with non-Lipschitz coefficients (Q858987)

From MaRDI portal
scientific article
Language Label Description Also known as
English
On pathwise uniqueness for stochastic heat equations with non-Lipschitz coefficients
scientific article

    Statements

    On pathwise uniqueness for stochastic heat equations with non-Lipschitz coefficients (English)
    0 references
    0 references
    0 references
    0 references
    12 January 2007
    0 references
    This paper treats the stochastic heat equations of the form \[ \frac{\partial}{\partial t} u(t,x) = \frac{1}{2} \Delta u(t,x) dt + \sigma ( u(t,x)) \dot{W}(x,t), \tag{1} \] where \(\Delta\) denotes the Laplacian and \(W\) is Gaussian martingale measure on \({\mathbb R}_+ \times {\mathbb R}^d\) in the sense of \textit{J. B. Walsh} [in: École d'été de probabilités de Saint-Flour XIV-1984. Lect. Notes Math. 1180, 265--437 (1986; Zbl 0608.60060)]. Note that \(W\) is defined on a filtered probability space \(( \Omega, {\mathcal F}, {\mathcal F}_t, {\mathbb P})\) and \(W_t(\varphi)\) \(=\) \(\int_0^t \int_{{\mathbb R}^d} \varphi(s,x) W(dx, ds)\) is an \({\mathcal F}_t\)-martingale for \(\varphi\) \(\in\) \(C_c^{\infty}( {\mathbb R}_+ \times {\mathbb R}^d)\) the space of compactly supported, infinitely differentiable functions on \({\mathbb R}_+ \times {\mathbb R}^d\). Here \(u\) is a random function on \({\mathbb R}_+ \times {\mathbb R}^d\), and the coefficient \(\sigma\) is a real-valued continuous function on \({\mathbb R}\), satisfying the global growth condition: for all \(u \in {\mathbb R}\), there exists a constant \(c_0\) such that \[ | \sigma(u) | \leqslant c_0 ( 1 + | u | ). \tag{2} \] The authors prove the existence and pathwise uniqueness of (1) especially in the case of non-Lipschitz coefficients and singular spatial noise correlations. A stochastic process \(u :\) \(\Omega \times {\mathbb R}_+ \times {\mathbb R}^d\) \(\to {\mathbb R}\), which is jointly measurable and \({\mathcal F}_t\)-adapted, is said to be a solution to the stochastic heat equation (1) in the variation of constants sense with respect to the martingale measure \(W\), defined on \(\Omega\), and initial condition \(u_0\), if for each \(t \geq 0\), a.s. for almost all \(x \in {\mathbb R}^d\), \[ u(t,x) = \int_{{\mathbb R}^d} p_t(x-y) u_0(y) \,dy + \int_0^t \int_{{\mathbb R}^d} p_{t-s}(x-y) \sigma( u(s,y)) W(dy, ds), \tag{3} \] where \(p\) denotes the \(d\)-dimensional heat kernel. For norm \(\| f \|_{\lambda, \infty}\) \(=\) \(\sup_{x \in {\mathbb R}^d} | f(x) |\) \(\times\) \(e^{ - \lambda | x |}\), set \(C_{\text{tem}}\) \(=\) \(\{ f \in C({\mathbb R}^d)\); \(\| f \|_{\lambda, \infty} < \infty\) for any \(\lambda > 0 \}\), endowed with the topology induced by the norms \(\| \cdot \|_{\lambda, \infty}\) for \(\lambda > 0\). It is said that pathwise uniqueness holds for solutions of (3) in \(C({\mathbb R}_+, C_{\text{tem}})\) if, for every \(u_0 \in\) \(C_{\text{tem}}\), any two solutions to (3) with sample paths a.s. in \(C({\mathbb R}_+, C_{\text{tem}})\) must be equal with probability 1. Let the function \(k = k(x,y)\) : \({\mathbb R}^{2d} \to {\mathbb R}\) be the correlation kernel of \(W\). For \(u_0 \in C_{\text{tem}}\), under the assumption that \[ | k(x,y) | \leqslant c_1 \{ | x - y |^{-\alpha} + 1 \} \tag{4} \] holds for all \(x, y \in {\mathbb R}^d\) and some \(\alpha \in ( 0, 2 \wedge d)\), there exists a stochastically weak solution to (3) with sample paths a.s. in \(C( {\mathbb R}_+, C_{\text{tem}})\). Here is the main result, which is valid for any spatial dimension \(d\). Theorem Assume that, for some \(\alpha \in (0,1)\), \(\sigma :\) \({\mathbb R} \to {\mathbb R}\) is Hölder continuous of index \(\gamma\) for some \(\gamma \in ( (1 + \alpha)/2, 1]\), and (4) holds. Then pathwise uniqueness holds for solutions of (3) in \(C(\mathbb R_+, C_{\text{tem}})\). For other related works, see e.g. \textit{L. Mytnik} [Ann. Probab. 26, No. 3, 968--984 (1998; Zbl 0935.60045)] and \textit{A. Sturm} [Electron. J. Probab. 8, Paper No. 6 (2003; Zbl 1064.60199)].
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    colored noise
    0 references
    stochastic partial differential equation
    0 references
    0 references
    0 references