A stochastic simplex approximate gradient (StoSAG) for optimization under uncertainty

Rahul Fonseca, B Chen, Jan Dirk Jansen, Albert C. Reynolds

Research output: Contribution to journalArticleScientificpeer-review

94 Citations (Scopus)
57 Downloads (Pure)

Abstract

We consider a technique to estimate an approximate gradient using an ensemble of randomly chosen control vectors, known as Ensemble Optimization (EnOpt) in the oil and gas reservoir simulation community. In particular, we address how to obtain accurate approximate gradients when the underlying numerical mod- els contain uncertain parameters because of geological uncertainties. In that case, ‘robust optimization’ is performed by optimizing the expected value of the objective function over an ensemble of geological mod- els. In earlier publications, based on the pioneering work of Chen et al. (2009), it has been suggested that a straightforward one-to-one combination of random control vectors and random geological models is capa- ble of generating sufficiently accurate approximate gradients. However, this form of EnOpt does not always yield satisfactory results. In a recent article, Fonseca et al. (2015) formulate a modified EnOpt algorithm, referred to here as a Stochastic Simplex Approximate Gradient (StoSAG; in earlier publications referred to as ‘modified robust EnOpt’) and show, via computational experiments, that StoSAG generally yields significantly better gradient approximations than the standard EnOpt algorithm. Here, we provide theoreti- cal arguments to show why StoSAG is superior to EnOpt
Original languageEnglish
Pages (from-to)1756-1776
Number of pages21
JournalInternational Journal for Numerical Methods in Engineering
Volume109
Issue number13
DOIs
Publication statusPublished - 2016

Keywords

  • Approximate gradient
  • Stochastic gradient
  • Ensemble optimization
  • Robust optimization
  • StoSAG

Fingerprint

Dive into the research topics of 'A stochastic simplex approximate gradient (StoSAG) for optimization under uncertainty'. Together they form a unique fingerprint.

Cite this