Skip to content

Patch bfgs upper bound #399

Open
MaxenceGollier wants to merge 2 commits intoJuliaSmoothOptimizers:mainfrom
MaxenceGollier:patch-bfgs-upper-bound
Open

Patch bfgs upper bound #399
MaxenceGollier wants to merge 2 commits intoJuliaSmoothOptimizers:mainfrom
MaxenceGollier:patch-bfgs-upper-bound

Conversation

@MaxenceGollier
Copy link
Contributor

Following these runs: https://github.com/JuliaSmoothOptimizers/RegularizedOptimization.jl/actions/runs/21675732671
There is a mistake with how we compute the upper bound. With this patch, tests complete in RegularizedOptimization.jl at least locally.

@dpo, if you agree, please merge.

Copy link
Member

@dpo dpo left a comment

Choose a reason for hiding this comment

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

Yes, that was a bug.

@MaxenceGollier
Copy link
Contributor Author

Can we merge @dpo ? Test aren't failing due to this modification.

Copilot AI review requested due to automatic review settings February 12, 2026 16:22
Copy link

Copilot AI left a comment

Choose a reason for hiding this comment

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

Pull request overview

This PR fixes the computation of the stored upper bound on the forward L-BFGS operator norm, aligning the update with the stated bound formula in LBFGSData and preventing underestimated bounds that can break downstream checks (as seen in RegularizedOptimization.jl runs).

Changes:

  • Update opnorm_upper_bound maintenance to add/subtract ‖bᵢ‖² (squared norms) instead of ‖bᵢ‖.
  • Keep the bound consistent with the documented inequality ‖Bₖ‖₂ ≤ ‖B₀‖₂ + ∑ᵢ ‖bᵢ‖₂².

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants