DMA, Local force constants, Marcus dimension #265
Conversation
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
| "max_cycles": 1, | ||
| }, | ||
| ) | ||
| opt = opt_result.opt |
Check notice
Code scanning / CodeQL
Unused local variable Note test
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #265 +/- ##
===========================================
- Coverage 61.58% 27.80% -33.78%
===========================================
Files 282 319 +37
Lines 29472 73521 +44049
===========================================
+ Hits 18149 20440 +2291
- Misses 11323 53081 +41758
Flags with carried forward coverage won't be shown. Click here to find out more.
|
eljost
force-pushed
the
dev
branch
2 times, most recently
from
01215f3
to
494de1c
Compare
| import warnings | ||
|
|
||
| try: | ||
| import ase |
Check notice
Code scanning / CodeQL
Unused import Note
There was a problem hiding this comment.
CodeQL found more than 10 potential problems in the proposed changes. Check the Files changed tab for more details.
eljost
force-pushed
the
dev
branch
3 times, most recently
from
9c63d47
to
20d0884
Compare
eljost
force-pushed
the
dev
branch
3 times, most recently
from
ec5fdb7
to
39f3968
Compare
There was a problem hiding this comment.
CodeQL found more than 20 potential problems in the proposed changes. Check the Files changed tab for more details.
eljost
force-pushed
the
dev
branch
2 times, most recently
from
d950481
to
50a7f5c
Compare
mod: also dump marcus_dim to xyz file at the end
fix: handle failed calculations in Marcus dim scans.
pysisyphus goes full quantum now!
using scans along Normal modes and using a sum-over-states partition function w/ eigenvalues determined via the Numerov method
now results are also dumped while the scan runs fix: relaxed_scan.trj contained unrelaxed geometries fix: revive 'pysisplot --scan'. this command plots files that match the pattern "*relaxed_scan.dat"
Now, the H2O2 tunnel splitting from the paper can be reproduced.
using Gaussian Process Regression w/ sklearn and optuna for hyperparameter optimization. See "tests/test_hindered_rotor/test_gpr_torsion.py" for a simple example. The general idea is to calculate the torsional potential energy curve on-the-fly at points, predicted by an acquisition function, based on a Gaussian process regression. When the potential energy curve is in place, nuclear wavefunctions and associated eigenvalues are evaluated using a periodic 1d-Numerov approach. The required mass is determined according to Pitzer/East/Radom (see partfuncs.inertmom).
add: higher level single point calcs in hindered_rotor add: ability to use splined potentials for testing
when the symlink already existed
as the reference image was created without it. I added a flag to disable graphivz, to allow a meaningful comparison
restricted trust_radius and added hessian_recalculation
to hopefully make them avaiable for nix
as scipy shifted the whole fragment's centroid into the origin, which was bad. Actually, the fragment must be shifted in a way, so the anchor atom is in the origin.
mod: set overachieve factor 3 in hindered_rotor opts
now convergence of partition function and predicted standard deviation are checked.
instead of np.int64
add: ability to scan transition states in hindered_rotor with appropriate opt_cls key add: hessian information is stored w/ correct prefix fix: calculation of force constant using periodic finite differences used one point too much
deprecatedandtests_staging