Space mapping based derivative-free optimization framework for supply chain optimization

Atharv Bhosekar, Marianthi Ierapetritou

Research output: Chapter in Book/Report/Conference proceedingChapter

Abstract

Derivative-free optimization (DFO) addresses the problem of optimizing over simulations where a closed form of the objective function is not available. Developments in the theory of DFO algorithms have made them useful for many practical applications. However, most of the existing DFO algorithms do not show satisfactory performance on high-dimensional problems. One contributor to this difficulty is the accuracy of the surrogate models used to guide the search for the optimum. Space mapping approach exploits a simplified simulation, which is a physical surrogate of the original problem at hand. As the simplified simulation is computationally efficient, a larger number of samples can be collected for guiding the search. This work aims to assess the potential of space mapping for derivative-free optimization, understand the difficulties and display its performance on a supply chain simulation optimization problem for identifying optimal inventory allocation. Computational results illustrate the potential of this approach.

Original languageEnglish (US)
Title of host publicationComputer Aided Chemical Engineering
PublisherElsevier B.V.
Pages985-990
Number of pages6
DOIs
StatePublished - Jan 1 2018

Publication series

NameComputer Aided Chemical Engineering
Volume44
ISSN (Print)1570-7946

All Science Journal Classification (ASJC) codes

  • Chemical Engineering(all)
  • Computer Science Applications

Keywords

  • derivative-free optimization
  • multi-fidelity analysis
  • physical surrogate models
  • space mapping
  • supply chain simulation

Fingerprint

Dive into the research topics of 'Space mapping based derivative-free optimization framework for supply chain optimization'. Together they form a unique fingerprint.

Cite this