A stochastic log-Laplace equation. (Q1889784)

This paper treats a branching interacting particle system in a random environment. More precisely, the particles move in \({\mathbb R}\) with critical binary branching. Between branchings, the position \(\eta_t^i\) of the \(i\)th particle at time \(t\) is governed by \[ \,d \eta_t^i = b(\eta_t^i) \,dt + c(\eta_t^i) \,d W(t) + e(\eta_t^i) \,d B_i(t), \quad i=1,2,\dots, \tag{1} \] where \(W, B_i\) are independent Brownian motions, and \(b, c, e\) are real-valued bounded functions with bounded first and second derivatives. Further it is assumed that \(e\) is bounded away from 0 and \(c\) has third continuous and bounded derivative. Let \({\mathcal M}_F({\mathbb R})\) denote the set of all finite Borel measures on \({\mathbb R}\). \textit{G. Skoulakis} and \textit{R. J. Adler} [Ann. Appl. Probab. 11, No. 2, 488--543 (2001; Zbl 1018.60052)] established that a superprocess \(X_t\) in a random environment can be obtained as the high density limit of the above-mentioned particle system, and its characterization by the martingale problem is also given. The aim of this paper is to derive properties of \(X_t\) by employing the corresponding backward stochastic log-Laplace equation (LLE). Indeed, the author studies the nonlinear stochastic partial differential equation (SPDE) (i.e. stochastic LLE) of the type: \[ y_{s,t}(x)= f(x) + \int_s^t \{ b(x) \partial_x y_{r,t}(x) + a(x) \partial_x^2 y_{r,t}(x) - y_{r,t}(x)^2 \} \,dr + \int_s^t c(x) \partial_x y_{r,t}(x) \, \widehat {d}W_r, \tag{2} \] with \(a(x) = \frac{1}{2}( e(x)^2 + c(x)^2)\), where \(f\) is a nonnegative bounded function with compact support and the last integral is the backward Itô integral, and he proves that the conditional log-Laplace functional \({\mathbb E}^W \exp \{ - \langle X_t, f \rangle \}\) of the superprocess \(X_t\) is given by \(\exp \{ - \langle \mu, y_{0,t} \rangle \}\), where \(\mu\) is the initial measure \(X_0\), \(y_{0,t}\) is a solution of (2), and \({\mathbb E}^W\) is the expectation by the conditional probability measure \({\mathbb P}^W\) given \(W\). The author establishes the existence and uniqueness of solutions to (2) by smoothing out the nonlinear term and making use of the particle system representation developed by \textit{T. G. Kurtz} and \textit{J. Xiong} [Stochastic Processes Appl. {83}, No. 1, 103--126 (1999; Zbl 0996.60071)]. Moreover, the Wong-Zakai approximation for the equation is also derived. As an application, the author gives a direct proof of the moment formulas of Skoulakis and Adler [loc. cit.]. As for other related models, see e.g. \textit{H. Wang} [Stochastic Anal. Appl. {16}, No. 4, 753--786 (1998; Zbl 0913.60091)] and \textit{D. A. Dawson, Z. Li} and \textit{H. Wang} [Electron. J. Probab. { 6}, Paper No. 25 (2001; Zbl 1008.60093)]; as to other related works for SPDEs with similar setting, see \textit{P. Kotelenez} [Stochastics Stochastics Rep. {41}, No. 3, 177--199 (1992; Zbl 0766.60078) and Probab. Theory Relat. Fields { 93}, No. 1, 1--19 (1992; Zbl 0767.60053)].