Skip to content

A consensus-based optimization methods for saddle point problems (CBO-SP)

License

Notifications You must be signed in to change notification settings

KonstantinRiedl/CBOSaddlePoints

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

CBO for Saddle Point Problems (CBO-SP)

Implementation of a consensus-based optimization method for saddle point problems (CBO-SP).

CBO-SP is a novel multi-particle metaheuristic derivative-free optimization method capable of provably finding global Nash equilibria. Following the idea of swarm intelligence, the method employs a group of interacting particles, which perform a minimization over one variable and a maximization over the other.

Version 1.0

Date 23.12.2022


R e f e r e n c e s

Consensus-Based Optimization for Saddle Point Problems

https://arxiv.org/abs/2212.12334

by

  • Hui   H u a n g   (University of Graz),
  • Jinniao   Q i u   (University of Calgary),
  • Konstantin   R i e d l   (Technical University of Munich, Munich Center for Machine Learning)

D e s c r i p t i o n

MATLAB implementation of a consensus-based optimization method for saddle point problems.

For the reader's convenience we describe the folder structure in what follows:

BenchmarkFunctions

  • objective_function.m: objective function generator
  • ObjectiveFunctionPlot.m: plotting routine for objective function

EnergyBasedCBOAnalysis

  • analyses: convergence and parameter analyses of CBO-SP
    • CBOSPNumericalExample.m: testing script
  • CBOSP: code of CBO-SP optimizer
    • compute_consensus.m: computation of consensus point
    • CBOSP_update: one CBO-SP step
    • CBOSP.m: CBO-SP optimizer
  • visualizations: visualization of the CBO-SP dynamics
    • CBOSPIllustrative.m: Illustration of the CBO-SP at work

C i t a t i o n s

@article{CBOSPQiuHuangRiedl,
      title = {Consensus-Based Optimization for Saddle Point Problems},
     author = {Jinniao Qiu and Hui Huang and Konstantin Riedl},
       year = {2022},
    journal = {arXiv preprint arXiv:2212.12334},
}