RIPM #135
Open
RIPM #135
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR is not finished, please do not merge.
A first implementation of RIPM and RIPMDH.
The values of the objective per iteration (or objective evaluation) are
objdec = solver_out.solver_specific[:Fhist] + solver_out.solver_specific[:Hhist].The benchmark plots are not working yet, because I would need to add to all solvers (TR, R2, TRDH) a list to indicate which iterations of the algorithm correspond to a gradient evaluation (not done in this PR) (because the gradient is not necessarly computed at every iteration).
To get the values of the objective per gradient evaluation, I could do$x_k$ and $x_{k+1}$ , or functions with constant gradients.
objdec = unique!(solver_out.solver_specific[:Fhist] + solver_out.solver_specific[:Hhist]), but I'm not sure that this woud be 100% accurate for small variations between@dpo