Closed maxscheurer closed 3 years ago
The only reason this PR is not moving forward is that I've not been able to come up with a user-friendly solution for integrating this feature properly.
The following feature scenarios are possible:
excitation_energy
if a PE ground state is present. Uncorrected excitation energies (i.e., the "actual" eigenvalues") are in excitation_energy_uncorrected
excitation_energy_uncorrected
values already contain the "linear response" contribution, so the ptLR does not make sense anymore. The ptSS still makes sense. The eigenvalues of this modified ADC matrix however still correspond to excitation_energy_uncorrected
. At the moment, both ptLR and ptSS corrections are always computed, also for the "postSCF scheme", which makes little sense.The same problematic will hold for other solvent models (PCM for example). So the following questions arise:
run_adc
and via "matrix construction".Just wanted to leave this here for discussion, because I have not made any progress alone so far...
This pull request introduces 1 alert when merging de5bbb93d9a286aca75d8a2a40e409859c43d26e into 041cdcb16ec60d1655db8d562ef2f3b53038c120 - view on LGTM.com
new alerts:
I hate clang-format
, why is it changing this one file anyways? I've added the .clang-format
file we used to have in adccore
, but it does not seem to care. I cannot reproduce this locally, however 😠
Different clang-format versions? Sometimes this happens then.
Different clang-format versions? Sometimes this happens then.
I've tried both v11 and v9 on CI, both giving the same diff
... I can make the change such that the CI runs without failure.
I would not get too much held up with this. either make the change or disable the relevant line from formatting or sth like that.
I think we have quite the functionality we want now! If you got a couple of minutes in the next days, @mfherbst, it would be nice if you could give it a brief look. I'm now working on refactoring the user-facing changes in run_adc
and some convenience wrappers, and then we can do refactoring etc. 🚀
Everything seems to be operational again 🎉
I've refactored the stuff we discussed 😄 The interface currently has a bit more flexibility, but it feels good from a user perspective imho. Let me know what you think @mfherbst 🧐
Docs now have a section on performing PE calculations 🚀 It's all coming together... 😄
@maxscheurer Noted it. Not sure when I get to it. Have a deadline next week.
Sure, I didn't want to be too annoying, sorry 😄 Then I'll start working on sth else and we move this forward when you have more time.
Don't worry ... I just wanted you to know.
@maxscheurer Unless you want to rebase I would squash-merge once tests pass.
Squash-merge is fine 😄 🚀
I guess a release of v0.15.9
would be adequate?
Preliminary PR, just to stash some work...
run_adc
and wrappers