On stochastic differential equations in a configuration space (Q2761602)

Let NEWLINE\[NEWLINEdx_k(t)= \sum_{i\neq k} a\bigl(x_k(t)- x_i(t)\bigr)d t+ \sigma dw_k(t), \quad k=1,2,\dots, \tag{1}NEWLINE\]NEWLINE where \(a(x)= -U_x(x)\), \(U:R^d\to R\) is a smooth function for \(|x|>0\) and \(\sigma>0\) is a constant, \(w_k (t) \), \(k=1,2,\dots\), is a sequence of independent Wiener processes in \(R^d\). System (1) describes the evolution of systems of pairwise interacting particles with the pairwise potential \(U(x)\) which is perturbed by Wiener noises. Let \(\Gamma\) denote the space of locally finite counting measures \(\gamma\) on the Borel \(\sigma\)-algebra of the space \(R^d\) with topology generated by the weak convergence of measures. For \(\Gamma\)-valued function \(\gamma_t\) the system (1) is rewritten in the form NEWLINE\[NEWLINE\langle\varphi, \gamma\rangle= \sum_k\varphi \bigl( x_k(t)\bigr), \quad\varphi\in C_f, \tag{2}NEWLINE\]NEWLINE where \(C_f\) denotes the set of continuous functions \(\varphi:R^d\to R\) with bounded supports and using Itô's formula NEWLINE\[NEWLINEd\langle \varphi,\gamma_t \rangle=\bigl\langle (\varphi',a), \gamma_t\times\gamma_t \bigr\rangle dt+{\sigma^2\over 2} \langle\Delta \varphi,\gamma_t \rangle+\sum_k \sigma\biggl( \varphi'\bigl(x_k(t) \bigr),dw_k(t)\biggr),\quad \varphi\in C_f^{ (2)}, \tag{3}NEWLINE\]NEWLINE where \(\Delta\varphi (x)=\text{Tr} \varphi''(x)\) and \(C_f^{(2)}\) is the set of \(\varphi\in C_f\) for which \(\varphi'(x)\) and \(\varphi''(x)\) are continuous bounded functions. A \(\Gamma\)-valued stochastic process \(\gamma_t (\omega)\) is called a weak solution to system (3) if, for all \(\varphi\in C_f^{ (2)}\), the stochastic process NEWLINE\[NEWLINE\mu_\varphi (\omega,t)= \langle\varphi, \gamma_t \rangle -\int^t_0 \left[\bigl \langle( \varphi',a), \gamma_s\times \gamma_s\bigr \rangle +{\sigma^2 \over 2}\langle\Delta\varphi,\gamma_s\rangle \right]dsNEWLINE\]NEWLINE is a martingale and the square characteristic of the martingale is NEWLINE\[NEWLINE\langle\mu_\varphi, \mu_\varphi \rangle= \sigma^2\int^t_0 \bigl\langle (\varphi', \varphi'), \gamma_s\bigr\rangle ds.NEWLINE\]NEWLINE If \(\gamma_t(\omega)\) is a weak solution to system (3) and NEWLINE\[NEWLINE\bigl\langle \varphi, \gamma_t(\omega) \bigr\rangle= \sum_k\varphi \bigl(x_k(t) \bigr), \quad \varphi \in C_f,NEWLINE\]NEWLINE then the sequence \(\{x_k(t)\), \(k\in {\mathcal N}\}\) is a weak solution to system (1). In the present paper it is proved that under some assumptions of the function \(U(x)\) there exists a unique local weak solution of system (1).