Skip to content

Conversation

nicoladicicco
Copy link

@nicoladicicco nicoladicicco commented Jul 19, 2024

This draft PR implements coordinate search for continuous variables in place of the current random search.
The objective function is assumed to be $C^1$ w.r.t. the continuous variables. Gradients are (momentarily) estimated via a simple finite difference method.
The algorithm uses the Armijo line search criterion to determine the step size.

The following features should be implemented before marking for review.

  • Implement a basic line search algorithm for coordinate descent on continuous variables
  • Integrate the coordinate descent algorithm with the management of the Tabu list
  • Add a test for a continuous problem (e.g., chemical equilibrium from ConstraintModels.jl) and check for convergence. The test should fail on the current version of the package

@Azzaare
Copy link
Member

Azzaare commented Jul 20, 2024

Regarding ConstraintModels, I still need to update the package!

Thank you for the nice work on this PR @nicoladicicco

@Azzaare Azzaare added enhancement New feature or request new feature labels Jul 20, 2024
@Azzaare Azzaare added this to the v1.0 milestone Jul 20, 2024
@Azzaare Azzaare changed the base branch from main to dev April 10, 2025 00:45
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request
Projects
Status: Todo
Development

Successfully merging this pull request may close these issues.

2 participants