scientific article; zbMATH DE number 3844992

From MaRDI portal
Publication:3314928





The vector integral equation \[ \phi(x)=\int_{X}K(x,y)\phi(y)dy+H(x),\phi,H\in E^ m,\quad K\sim m\times m, \] is solved by the Monte Carlo method. The sequence of points \(X_ 0=X\), \(X_ 1,...,X_ N\) with transitional density function \(P_{n-1,n}=P(X_ 0=X,X_ 1,...,X_{n-1})\) of the transition \(X_{n- 1}\to X_ n\) is considered. The problem of optimization of algorithms is solved for the given integral equation. It consist in a definition of densities \(P_{n-1,n}\), such that the variance of the estimator \({\hat \phi}\)(x) has a minimum. Algorithmes for densities \(P_{n-1,n}\) are presented. It is shown, that the optimal chain is in general not a Markov chain. In the class of Markov densities algorithms for \(P_{n-1,n}\) also are presented. A numerical example is given, realizing the method of optimization.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3314928)