AEC BAC for ωB97X-D3/def2-TZVP and B97-D3/def2-mSVP #459
Conversation
|
@kspieks, could you make a twin RMG-Py PR to update the model chemistry table in the docs? |
|
@alongd Will do! Thanks for reminding me |
Note that 6 species failed during QM due to the frequency being negative for spcs112, spcs229, spcs303, spcs351, spcs407, spcs419 Note that 4 species failed to converge during QM optimization: spcs149, spcs161, spcs170, spcs397 I tried changing convergence criteria, but could not fix the issue. The remaining 411 species were used when fitting BACs.
Note that 17 species failed during QM due to the frequency being negative or optimization failing to converge: spcs68, spcs117, spcs126, spcs140, spcs177, spcs229, spcs223, spcs286, spcs339, spcs351, spcs149, spcs161, spcs170, spcs397, spcs195, spcs267, and spcs320 The remaining 404 species were used for fitting BACs.
kspieks
force-pushed
the
aec_bac_wb97xd3_b97d3
branch
from
4ffebb3 to
772cb4e
Compare
@xiaoruiDong Does this PR look ready to be merged? I’m unsure why the continuous integration of rmg-tests are failing, but it seems like that has failed for many recent PRs. I can add it as a hackathon topic for us to look into. Regarding conformers, my understanding was that the experimental geometry from CCCBDB was used as the initial guess for ωb97mv optimization (whose converged geometry was the initial guess for my calculations). It’s possible Colin did a conformer search with ARC for these species, but I’m unsure where the log files are so I can’t check for sure. Looks like other PRs (#412) for adding BACs also did not consider rotors (or the PR (#417) doesn't comment on it). |
|
@kspieks Thanks. This PR looks good. Can you also update the reference database yamls with D97-B3/def2-mSVP energies? Once they are updated, please rebase the PR, and I will merge it into the master branch. Thanks! |
Thanks for reviewing! I updated the reference database with b97d3/def2-msvp data :) |
|
FYI: all log files and Arkane outputs related to these jobs have been copied to my account on our new greencloud sever at |
This PR adds frequency scaling factors, AEC, and BAC for two levels of theory:
ωB97X-D3/def2-TZVPandB97-D3/def2-mSVP. All calculations were run with QChem 5.3Frequency scaling factors were computed using ARC.
AECs were determined by running sp calculations of the 16 small molecules from the reference database. The experimental geometries were used i.e. no optimization was performed.
BACs were determined by running opt + freq jobs in QChem using ARC. The reference database previously had stored the optimized geometries from
ωB97M-V/def2-TZVPD; these geometries were used as the initial guess. Rotors were not considered. A few molecules failed during QM by either not converging during geometry optimization or yielding a negative frequency despite being a stable species, not a TS; these molecules were omitted and the remaining 400+ molecules were used for BACs fitting.For
ωB97X-D3/def2-TZVP:1_wb97xd3_def2tzvp_errors.pdf
For
B97-D3/def2-mSVP:1_b97d3_def2msvp_errors.pdf