Adaptive Multiple Importance Sampling - Statistics > ComputationReport as inadecuate

Adaptive Multiple Importance Sampling - Statistics > Computation - Download this document for free, or read online. Document in PDF available to download.

Abstract: The Adaptive Multiple Importance Sampling AMIS algorithm is aimed at anoptimal recycling of past simulations in an iterated importance samplingscheme. The difference with earlier adaptive importance samplingimplementations like Population Monte Carlo is that the importance weights ofall simulated values, past as well as present, are recomputed at eachiteration, following the technique of the deterministic multiple mixtureestimator of Owen and Zhou 2000. Although the convergence properties of thealgorithm cannot be fully investigated, we demonstrate through a challengingbanana shape target distribution and a population genetics example that theimprovement brought by this technique is substantial.

Author: Jean-Marie Cornuet CBGP, INRA, Montpellier, Jean-Michel Marin I3M, Montpellier, Antonietta Mira University of Lugano, Christian P


Related documents