PhBaseWorkChain: better handler for convergence
#986
Conversation
This file contains 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
Plain lowering down the alpha mix could not work always. The `nmix_ph` is already an improvement, leading to the usage of more iterations into the mixing. The rate of the alpha mix change is also increased, and below a certain threshold a multi mix approach is tried. Afterwards, the handler finishes the strategies.
mbercx
requested changes
Nov 24, 2023
|
|
||
| return ProcessHandlerReport(True) | ||
|
|
||
| # now try playing with the mixing parameters |
There was a problem hiding this comment.
In my runs I had immediately set alpha_mix to 0.4 in the overrides. I didn't gather any statistics, but I did get a noticable improvement in convergence. Something to consider?
There was a problem hiding this comment.
One additional note on the line below this one (which was already there ^^): The user can also set alpha_mix instead of alpha_mix(1) and this would work fine. If we only look for alpha_mix(1) and then set it, I think we override the alpha_mix specified by the user.
There was a problem hiding this comment.
I think the user have to specify alpha_mix(i) = X, and not simply alpha_mix = X
There was a problem hiding this comment.
I've always just used alpha_mix:
mabercx@uan04:/pfs/lustrep3/scratch/project_465000106/mabercx/88/a0/a0e0-3391-4e6b-b4da-fc98886eaa4a> grep alpha_mix aiida.in
alpha_mix = 4.0000000000d-01
mabercx@uan04:/pfs/lustrep3/scratch/project_465000106/mabercx/88/a0/a0e0-3391-4e6b-b4da-fc98886eaa4a> grep alpha_mix aiida.out | head -n 5
thresh= 1.000E-02 alpha_mix = 0.400 |ddv_scf|^2 = 1.410E-08
thresh= 1.187E-05 alpha_mix = 0.400 |ddv_scf|^2 = 3.525E-09
thresh= 5.937E-06 alpha_mix = 0.400 |ddv_scf|^2 = 7.637E-11
thresh= 8.739E-07 alpha_mix = 0.400 |ddv_scf|^2 = 2.221E-13
thresh= 4.713E-08 alpha_mix = 0.400 |ddv_scf|^2 = 3.612E-14
mbercx
requested changes
Nov 28, 2023
…esso into handler/ph/convergence
Add `alpha_mix` and `ndim_ph`.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Plain lowering down the alpha mix could not work always. The
nmix_phis already an improvement, leading to the usage of more iterations into the mixing. The rate of the alpha mix change is also increased, and below a certain threshold a multi mix approach is tried. Afterwards, the handler finishes the strategies.