Gaussian unconverge trsh #758
Conversation
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #758 +/- ##
==========================================
+ Coverage 73.82% 73.96% +0.13%
==========================================
Files 101 101
Lines 27623 27776 +153
Branches 5795 5816 +21
==========================================
+ Hits 20393 20544 +151
+ Misses 5774 5773 -1
- Partials 1456 1459 +3
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
calvinp0
force-pushed
the
gaussian_unconverge_trsh
branch
2 times, most recently
from
6d424ba
to
e07ec32
Compare
| @@ -878,6 +887,8 @@ def trsh_ess_job(label: str, | |||
|
|
|||
| # Check if SCF is in the keyword | |||
| ess_trsh_methods, trsh_keyword, couldnt_trsh = trsh_keyword_scf(job_status, ess_trsh_methods, trsh_keyword, couldnt_trsh) | |||
| if 'scf=(maxcycle=512)' in ess_trsh_methods: | |||
There was a problem hiding this comment.
It was just a number that was suggested. There is no rhyme or reason why that value specifically. As I mentioned in the description, this troubleshoot addition is a last resort brute force.
There was a problem hiding this comment.
I have removed the scf cycle increase as it would have consequences on certain aspects if this is being done automatedly.
calvinp0
force-pushed
the
gaussian_unconverge_trsh
branch
from
e07ec32
to
a6a6bd1
Compare
calvinp0
force-pushed
the
gaussian_unconverge_trsh
branch
5 times, most recently
from
0687958
to
d6e9440
Compare
calvinp0
force-pushed
the
gaussian_unconverge_trsh
branch
from
d6e9440
to
b25f8b6
Compare
There are times when we do not get convergence of the optimisation of a molecule and we only receive an output of
l9999.exe. Currently, we do not make any drastic changes then to the input file for Gaussian for troubleshooting and may end up determining we cannot converge the molecule and therefore give up.However, further up the file we may find the number of steps is a limit (the example output file shows 27 steps is the max for opt). Therefore, this aims to troubleshoot it by increasing the cycles for opt. Removed increase of scf maxcycle as DM/QC (512) have their own default values compared to default algorithm (128). If non-convergence is still not occuring, then checking if the scf is shaking manually should be done. https://wongzit.github.io/method-to-solve-the-scf-not-converged/
Edit: I have now included more methods for dealing with other errors, such as L502 -
Inaccurate quadrature in CalDSuI have attempted to provide comments against each trsh_keyword to explain what they do