Efficient Markov chain Monte Carlo sampling for electrical impedance tomography
Journal Title: Computer Assisted Methods in Engineering and Science - Year 2014, Vol 21, Issue 3
Abstract
This paper studies electrical impedance tomography (EIT) using Bayesian inference [1]. The resulting posterior distribution is sampled by Markov chain Monte Carlo (MCMC) [2]. This paper studies a toy model of EIT as the one presented in [3], and focuses on efficient MCMC sampling for this model. First, this paper analyses the computation of forward map of EIT which is the bottleneck of each MCMC update. The forward map is computed by the finite element method [4]. Here its exact computation was conducted up to five times more efficient, by updating the Cholesky factor of the stiffness matrix [5]. Since the forward map computation takes up nearly all the CPU time in each MCMC update, the overall efficiency of MCMC algorithms can be improved almost to the same amount. The forward map can also be computed approximately by local linearisation, and this approximate computation is much more efficient than the exact one. Without loss of efficiency, this approximate computation is more accurate here, after a log transformation is introduced into the local linearisation process. Later on, this improvement of accuracy will play an important role when the approximate computation of forward map will be employed for devising efficient MCMC algorithms. Second, the paper presents two novel MCMC algorithms for sampling the posterior distribution in the toy model of EIT. The two algorithms are made within the 'multiple prior update' [6] and 'the delayed-acceptance Metropolis-Hastings' [7] schemes respectively. Both of them have MCMC proposals that are made of localized updates, so that the forward map computation in each MCMC update can be made efficient by updating the Cholesky factor of the stiffness matrix. Both algorithms' performances are compared to that of the standard single-site Metropolis [8], which is considered hard to surpass [3]. The algorithm of 'multiple prior update' is found to be six times more efficient, while 'the delayed-acceptance Metropolis-Hastings' with single-site update is at least twice more efficient.
Authors and Affiliations
Erfang Ma
Keywords
Related Articles
- EP ID EP73497
- DOI -
- Views 190
- Downloads 0
Application of inverse analysis in electromagnetic grinding of brown coal to obtain an optimal particle size distribution- a heuristic approach
This paper presents the research results of milling process optimization in the electromagnetic mill to obtain the predetermined particle size distribution of brown coal. Because of an important role of brown coal in Pol...
Identification of solidification process parameters. (Received in the final form November 12, 2010)
In the paper, the identification problems connected with estimation of cast iron and mould thermophysical parameters are discussed. The additional information necessary to solve the problem results from the knowledge of...
Bayesian neural networks and Gaussian processes in identification of concrete properties
This paper gives a concise overview of concrete properties prediction using advanced nonlinear regression approach and Bayesian inference. Feed-forward layered neural network (FLNN) with Markov chain Monte Carlo stochast...
A time-step insensitive recurrent approach to analyze non-stationary random responses. (Received in the final form October 28, 2008).
The recurrent approach constructed via the stochastic central difference (SCD) is a very fast method for analyzing non-stationary random responses. However, the computational results depend to a great extent upon the dis...
Simulation of in-plane vibrations of 2D structural solids with singularities using an efficient wave based prediction technique
This paper proposes the wave based method for the steady-state dynamic analysis of the in-plane behaviour of 2D structural solids. This novel prediction technique relaxes the frequency limitations of the commonly used fi...