Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Statistical Inversion of Electromagnetic Logging DataBeyond the Random-Walk: A Hybrid Monte Carlo Sampling

Statistical Inversion of Electromagnetic Logging Data: Beyond the Random-Walk: A Hybrid Monte... [Metropolis-Hastings algorithm realizes MCMC via random-walk sampling and recovers PPD through discrete samples. However, the slow convergence rate makes the sampling efficiency a big obstacle for any real-time data processing workflow. On the contrary, many deterministic optimizations follow a gradient update and have relatively fast searching speed compared with random move. One attractive realization is to combine two schemes, where people introduce the gradient as a moving force and combine it with a statistical sampling process. The fusion brings in the topic of this chapter, a hybrid scheme of MCMC, or named hybrid Monte Carlo sampling. In the following sections, we will introduce the concepts of hybrid Monte Carlo, Hamiltonian dynamics, and their mix, Hamiltonian Monte Carlo sampling method. We also use a group of examples to demonstrate the effect and advantage when applying HMC on the interpretation of directional EM LWD data.] http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png

Statistical Inversion of Electromagnetic Logging DataBeyond the Random-Walk: A Hybrid Monte Carlo Sampling

Download PDF
10 pages

Loading next page...
 
/lp/springer-journals/statistical-inversion-of-electromagnetic-logging-data-beyond-the-Fm0AoZDf10

References (6)

Publisher
Springer International Publishing
Copyright
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2021
ISBN
978-3-030-57096-5
Pages
31 –41
DOI
10.1007/978-3-030-57097-2_3
Publisher site
See Chapter on Publisher Site

Abstract

[Metropolis-Hastings algorithm realizes MCMC via random-walk sampling and recovers PPD through discrete samples. However, the slow convergence rate makes the sampling efficiency a big obstacle for any real-time data processing workflow. On the contrary, many deterministic optimizations follow a gradient update and have relatively fast searching speed compared with random move. One attractive realization is to combine two schemes, where people introduce the gradient as a moving force and combine it with a statistical sampling process. The fusion brings in the topic of this chapter, a hybrid scheme of MCMC, or named hybrid Monte Carlo sampling. In the following sections, we will introduce the concepts of hybrid Monte Carlo, Hamiltonian dynamics, and their mix, Hamiltonian Monte Carlo sampling method. We also use a group of examples to demonstrate the effect and advantage when applying HMC on the interpretation of directional EM LWD data.]

Published: Aug 28, 2020

There are no references for this article.