BAGEL interface for XMS and MS-CASPT2 in FSSH and LZSH #153
Conversation
The interface for BAGEL and ABIN to perform FSSH and LZSH simulations. It is set to do XMS-CASPT2 by default but input can be modified for MS-CASPT2 and CASSCF. It was tested on cyclopropanone molecule and benchmarked to MOLPRO. Credits to Tomas Jira and Jiri Suchan.
Quick fix of the previous commit based on review by Dan Hollas to make the code working. Renaming files correctly so that ABIN can call BAGEL interface. The rest of the review will be processed later.
|
@JanosJiri you can ignore the build errors on this PR, they are not real issues with the code. |
This commit contains partial improvements of BAGEL interface. It is not finished and working, but serves to share progress with Michal and Honza.
Codecov Report
Additional details and impacted files@@ Coverage Diff @@
## master #153 +/- ##
=======================================
Coverage 92.37% 92.37%
=======================================
Files 43 43
Lines 6084 6084
Branches 719 741 +22
=======================================
Hits 5620 5620
Misses 452 452
Partials 12 12
Flags with carried forward coverage won't be shown. Click here to find out more. |
Input is now in separate files so that r. files need not to be modified. Gradient calculation detects convergence failure and restarts with new thresholds if necessary.
|
@JanosJiri is this ready for a new round of review or not yet? |
|
Yes, it's ready for review @danielhollas. |
There was a problem hiding this comment.
This is looking great now @JanosJiri! Some final small things, and then I'll test it locally with our BAGEL.
Suggestion by Dan Hollas were incorporated all except the nproc variable. We don't have mpi ready here so I could not test it. These versions are working at our krypton cluster.
|
Thanks @JanosJiri 👏 I'll do some testing locally (hopefully next week) and then we should be good to go! |
This interface is a simple modification of the BAGEL-LZ interface. It was tested on cyclobutanone MD with GLE thermostat.
|
@JanosJiri I've cleaned up the interfaces in BAGEL-SH and tested XMS-CASPT2, MS-CASPT2 and CASSCF. I've been able to reduce a lot of duplication by moving stuff into functions in bagel.inp. If this looks good to you I'll apply the same thing to LZ and ground state BAGEL. |
shellcheck -a -S info -x bagel.inp r.bagel-sh
|
Okay, I think I am done. 😅 I've cleaned up all the three interfaces and tested them on Hamilton. The scripts now also pass the @JanosJiri It would be great if you could do last tests on Prague clusters and then we're good to go. When testing, you can increase the |
The interface for BAGEL and ABIN to perform FSSH and LZSH simulations. It is set to do XMS-CASPT2 by default but input can be modified for MS-CASPT2 and CASSCF. It was tested on cyclopropanone molecule and benchmarked to MOLPRO. Credits to Tomas Jira and Jiri Suchan.