Particle system algorithm and chaos propagation related to non-conservative McKean type stochastic differential equations (Q523374)

The present paper focuses on numerical aspects of a special forward probabilistic representation of stochastic differential equations (SDEs). The authors consider a regularized McKean-type nonlinear SDE of the form \[ Y_t=Y_0+\int_0^t \Phi(s,Y_s,u(s,Y_s))\,dW_s +\int_0^t g(s,Y_s,u(s,Y_s))\,ds \] \[ u(t,y)=E[K(y-Y_t)exp\{\int_0^t \Lambda(s,Y_s,u(s,Y_s))\,ds\}],\quad t\in t\in[0,T], \] where the solution is the couple process function \((Y,u)\). The novelty with respect to the classical McKean-type equation consists in the form of the second equation. In the present paper, the authors extend the McKean based numerical schemes to the case of non-conservative partial differential equations with \(\Lambda\neq0\). An interacting particle system associated with the considered equation is introduced and one simple McKean-type SDE is replaced with an unknown process \(Y\) with a system of \(N\) ordinary SDEs, whose solution consist of a system of particles replacing the law of the process \(Y\) by the empirical mean law. Convergence of the time discretized particle system is proved under the Lipschitz condition on the coefficients. Evaluation of the article approximation errors is discussed in details. Specifically, the authors concentrate on its convergence. Finally, numerical simulations illustrate the performance. Details of the implementation are concisely described, a pseudocode of the algorithm is presented.