Skip to content

Integrate CliqueMerging presolver and tune presolver settings#415

Merged
rapids-bot[bot] merged 40 commits intoNVIDIA:branch-25.10from
rg20:tune_presolve_settings
Oct 3, 2025
Merged

Integrate CliqueMerging presolver and tune presolver settings#415
rapids-bot[bot] merged 40 commits intoNVIDIA:branch-25.10from
rg20:tune_presolve_settings

Conversation

@rg20
Copy link
Copy Markdown
Contributor

@rg20 rg20 commented Sep 24, 2025 •
edited
Loading

Description

This PR integrates the CliqueMerging presolver from Papilo and customizes some presolver settings for cuOpt. This PR also provides --dual-postsolve option. This option allows users to disable dual post solve which makes presolver apply more reductions. By default this flag is set to false.

Issue

Closes #356

Checklist

  • [ x] I am familiar with the Contributing Guidelines.
  • Testing
    • [ x] New or existing tests cover these changes
    • Added tests
    • Created an issue to follow-up
    • NA
  • Documentation
    • The documentation is up to date with these changes
    • Added new documentation
    • [x ] NA

@rg20 rg20 requested review from a team as code owners September 24, 2025 22:07
@copy-pr-bot
Copy link
Copy Markdown

copy-pr-bot bot commented Sep 24, 2025

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

@rg20 rg20 added non-breaking Introduces a non-breaking change improvement Improves an existing functionality labels Sep 24, 2025
@rg20 rg20 added this to the 25.10 milestone Sep 24, 2025
@rg20 rg20 force-pushed the tune_presolve_settings branch from 5edeb2b to dc2937b Compare September 24, 2025 22:07
@rg20 rg20 requested review from aliceb-nv and removed request for akifcorduk and kaatish September 24, 2025 22:08
@aliceb-nv
Copy link
Copy Markdown
Contributor

aliceb-nv commented Sep 25, 2025

Thanks a lot for the super quick and great work Rajesh! :)
I'm seeing a consistent regression on comp07-2idx.mps compared to the previous commit that only appears when the CliqueMerging presolver is added, do you have an idea what could cause this (overconstraining?)
Previous commit finds near-optimal solutions (obj <15) in 10-20 seconds, while w/ cliquemerging it remains above 100

I think we can merge the PR as it's overall beneficial but we should probably track this with a GH issue

@aliceb-nv
Copy link
Copy Markdown
Contributor

aliceb-nv commented Sep 25, 2025

/ok to test dc2937b

@rg20
Copy link
Copy Markdown
Contributor Author

rg20 commented Sep 25, 2025

/ok to test 2d9f080

tmckayus
tmckayus approved these changes Sep 26, 2025
View reviewed changes
Copy link
Copy Markdown
Contributor

@tmckayus tmckayus left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

cmake lgtm. What's the strategy to remember to change this someday from a specifc hash? We might forget :)

@rg20
Copy link
Copy Markdown
Contributor Author

rg20 commented Sep 26, 2025

cmake lgtm. What's the strategy to remember to change this someday from a specifc hash? We might forget :)

This is the latest commit from development branch. Once things merge to main branch, we can switch to that. I can add some notes.

@rg20 rg20 changed the title Integrate CliqueMerging presolver Integrate CliqueMerging presolver and tune presolver settings Sep 26, 2025
@rg20
Copy link
Copy Markdown
Contributor Author

rg20 commented Sep 26, 2025

/ok to test d3ac1bc

@rgsl888prabhu
Copy link
Copy Markdown
Collaborator

rgsl888prabhu commented Sep 30, 2025

/ok to test 547ccf2

akifcorduk
akifcorduk reviewed Sep 30, 2025
View reviewed changes
@rgsl888prabhu
Copy link
Copy Markdown
Collaborator

rgsl888prabhu commented Sep 30, 2025

/ok to test f3d9bf5

@rg20
Copy link
Copy Markdown
Contributor Author

rg20 commented Sep 30, 2025

/ok to test 6d5b773

@rg20
Copy link
Copy Markdown
Contributor Author

rg20 commented Sep 30, 2025

/ok to test b339605

@rg20
Copy link
Copy Markdown
Contributor Author

rg20 commented Sep 30, 2025

/ok to test 9e13e09

@rg20
Copy link
Copy Markdown
Contributor Author

rg20 commented Sep 30, 2025

/ok to test 556d96a

@rg20
Copy link
Copy Markdown
Contributor Author

rg20 commented Oct 1, 2025

/ok to test 1c1bddd

@rg20
Copy link
Copy Markdown
Contributor Author

rg20 commented Oct 1, 2025

/ok to test 036117b

@rg20
Copy link
Copy Markdown
Contributor Author

rg20 commented Oct 1, 2025

/ok to test d0fc79f

@rg20
Copy link
Copy Markdown
Contributor Author

rg20 commented Oct 2, 2025

/ok to test b9c0dcb

@rg20
Copy link
Copy Markdown
Contributor Author

rg20 commented Oct 2, 2025

/ok to test 5fe2f3c

@rg20
Copy link
Copy Markdown
Contributor Author

rg20 commented Oct 3, 2025

/merge

@rapids-bot rapids-bot bot merged commit a57d894 into NVIDIA:branch-25.10 Oct 3, 2025
258 of 260 checks passed
jieyibi pushed a commit to yining043/cuopt that referenced this pull request Mar 26, 2026
@rg20
Integrate CliqueMerging presolver and tune presolver settings (NVIDIA…
7fcafc8 
…#415)

This PR integrates the CliqueMerging presolver from Papilo and customizes some presolver settings for cuOpt. This PR also provides `--dual-postsolve` option. This option allows users to disable dual post solve which makes presolver apply more reductions. By default this flag is set to false.  

## Issue

Closes NVIDIA#356

Authors:
  - Rajesh Gandham (https://github.com/rg20)
  - Ramakrishnap (https://github.com/rgsl888prabhu)

Approvers:
  - Alice Boucher (https://github.com/aliceb-nv)
  - Trevor McKay (https://github.com/tmckayus)
  - Ramakrishnap (https://github.com/rgsl888prabhu)

URL: NVIDIA#415
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@akifcorduk akifcorduk akifcorduk left review comments

@tmckayus tmckayus tmckayus approved these changes

@rgsl888prabhu rgsl888prabhu rgsl888prabhu approved these changes

@aliceb-nv aliceb-nv aliceb-nv approved these changes

Assignees

No one assigned

Labels

improvement Improves an existing functionality non-breaking Introduces a non-breaking change

Projects

None yet

Milestone

25.10

Development

Successfully merging this pull request may close these issues.

[FEA] Add a new --presolve option that supports dual-preserving presolve methods

6 participants

@rg20 @aliceb-nv @rgsl888prabhu @tmckayus @akifcorduk @anandhkb