Optimization of Monte Carlo weighted methods for part of variables

algorithm collision estimate importance sampling linear functionals modelling the free path Monte Carlo weighted method partial optimization of modelling second kind integral equation sub-Markovian kernel terminating Markov chain variance reduction

Mathematics Subject Classification ID

Computational methods in Markov chains (60J22) Monte Carlo methods (65C05) Numerical analysis or methods applied to Markov chains (65C40)

Review text

Estimates for the linear functionals \(\int \phi (x)h(x)dx\) of the solution \(\phi (x)\) to the second kind integral equation with a strictly subMarkovian kernel \(k(x,x')\) and a positive free term \(f(x)\) with integral 1 can be estimated directly by using \(\sum_n h(x_n)\) where \(x_n\) is the homogeneous terminating Markov chain with transition function \(k(x,x')\) and initial density \(f(x)\). In this paper the minimal problem of variance of the estimate of the functional is considered by imitating the idea of the importance sampling of the traditional variance reduction technique by using another auxillary Markov chain with appropriate selected transition kernel and initial density. This improved estimate is also called weight `collision estimate'. A practical efficient approximation algorithm is suggested.

Recommendations

Cited in

(8)

This page was built for publication: Optimization of Monte Carlo weighted methods for part of variables

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4463681)