Stochastic blockmodeling for learning the structure of optimization problems

Ilias Mitrai, Wentao Tang, Prodromos Daoutidis

Research output: Contribution to journalArticlepeer-review

13 Scopus citations

Abstract

Decomposition-based solution algorithms for optimization problems depend on the underlying latent block structure of the problem. Methods for detecting this structure are currently lacking. In this article, we propose stochastic blockmodeling (SBM) as a systematic framework for learning the underlying block structure in generic optimization problems. SBM is a generative graph model in which nodes belong to some blocks and the interconnections among the nodes are stochastically dependent on their block affiliations. Hence, through parametric statistical inference, the interconnection patterns underlying optimization problems can be estimated. For benchmark optimization problems, we show that SBM can reveal the underlying block structure and that the estimated blocks can be used as the basis for decomposition-based solution algorithms which can reach an optimum or bound estimates in reduced computational time. Finally, we present a general software platform for automated block structure detection and decomposition-based solution following distributed and hierarchical optimization approaches.

Original languageEnglish (US)
Article numbere17415
JournalAIChE Journal
Volume68
Issue number6
DOIs
StatePublished - Jun 2022
Externally publishedYes

Bibliographical note

Funding Information:
Financial support from NSF-CBET (award number 1926303) is acknowledged.

Funding Information:
Directorate for Engineering, Grant/Award Number: 1926303; National Science Foundation, Grant/Award Number: 1926303; NSF‐CBET Funding information

Funding Information:
Financial support from NSF‐CBET (award number 1926303) is acknowledged.

Publisher Copyright:
© 2021 American Institute of Chemical Engineers.

Keywords

  • Benders decomposition
  • Lagrangean decomposition
  • network decomposition
  • stochastic blockmodeling

Fingerprint

Dive into the research topics of 'Stochastic blockmodeling for learning the structure of optimization problems'. Together they form a unique fingerprint.

Cite this