Particle MCMC for Bayesian Microwave Control

4th International Workshop on New Computational Methods for Inverse Problems

物理学;计算机科学

Minvielle, P.^1 ; Todeschini, A.^2 ; Caron, F.^3 ; Del Moral, P.^4

CEA-CESTA, Le Barp

33114, France^1

INRIA Bordeaux Sud-Ouest, cours de la Liberation, 351, Talence Cedex

33405, France^2

University of Oxford, 1 South Parks Road, Oxford, United Kingdom^3

UNSW, High Street, Kensington Sidney, Australia^4

关键词: Bayesian inference;    Electromagnetic scattering;    Higher-dimensional problems;    Metropolis Hastings;    Microwave control;    Multiple components;    Property estimation;    Rao-Blackwellised Sequential Monte Carlo;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/542/1/012007/pdf
DOI  :  10.1088/1742-6596/542/1/012007

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

We consider the problem of local radioelectric property estimation from global electromagnetic scattering measurements. This challenging ill-posed high dimensional inverse problem can be explored by intensive computations of a parallel Maxwell solver on a petaflopic supercomputer. Then, it is shown how Bayesian inference can be perfomed with a Particle Marginal Metropolis-Hastings (PMMH) approach, which includes a Rao-Blackwellised Sequential Monte Carlo algorithm with interacting Kalman filters. Material properties, including a multiple components "Debye relaxation"/"Lorenzian resonant" material model, are estimated; it is illustrated on synthetic data. Eventually, we propose different ways to deal with higher dimensional problems, from parallelization to the original introduction of efficient sequential data assimilation techniques, widely used in weather forecasting, oceanography, geophysics, etc.