Mathematics > Numerical Analysis
[Submitted on 5 Aug 2016 (this version), latest version 14 Nov 2017 (v2)]
Title:Iterative importance sampling algorithms for parameter estimation problems
View PDFAbstract:In parameter estimation problems one approximates a posterior distribution over uncertain param- eters defined jointly by a prior distribution, a numerical model, and noisy data. Typically, Markov Chain Monte Carlo (MCMC) is used for the numerical solution of such problems. An alternative to MCMC is importance sampling, where one draws samples from a proposal distribution, and attaches weights to each sample to account for the fact that the proposal distribution is not the posterior distribution. Drawing samples and computing their weights can be done independently, which leads to massively parallel algorithms that exhibit near perfect scaling with the number of cores. These advantages however are conditional on ones ability to construct an appropriate proposal distribution. Here we present iterative algorithms that refine the mean and covariance matrix of Gaussian or multivariate t-proposal distributions. Such ideas have been discussed in the literature before. We revisit and revise these algorithms and investigate their applicability by test- ing their performance in two realistic test problems, one in subsurface flow, and one in combustion modeling. In our implementation, the algorithms leverage massively parallel computers, and we also discuss how to use coarse MCMC simulations or optimization in connection with Gaussian mixture models to initialize the iterations.
Submission history
From: Matthias Morzfeld [view email][v1] Fri, 5 Aug 2016 17:59:17 UTC (983 KB)
[v2] Tue, 14 Nov 2017 17:46:22 UTC (1,157 KB)
Current browse context:
math.NA
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.