BugFix: use str(freq_level) to get frequency scale factor from dict

[oscarwumit]

Motivation or Problem

The keys of data.freq_dict are strings, but freq_level is an object. This mismatch results in KeyError, and subsequently gave a wrong frequency scale factor of 1 for model chemistries exist in the database.

This bug also caused ARC to crash as it incorrectly informs ARC about the frequency scale factor. Related to ReactionMechanismGenerator/ARC#398

Execute the following in a Jupiter notebook will reveal the bug:

from arkane.encorr.corr import assign_frequency_scale_factor
from arkane.modelchem import LevelOfTheory
import arkane.encorr.data as data

lot = LevelOfTheory(method='b3lyp',basis='6311+g(3df,2p)',software='gaussian')
assign_frequency_scale_factor(lot)

Output: WARNING:root:No frequency scaling factor found for LevelOfTheory(method='b3lyp',basis='6311+g(3df,2p)',software='gaussian'). Assuming a value of unity. This will affect the partition function and all quantities derived from it (thermo quantities and rate coefficients).

1

Description of Changes

Convert freq_level to string before using it as a key.

Testing

After the change, repeat the above example code in the Jupiter notebook gave the correct frequency scale factor of 0.967.

[cgrambow]

The keys in data.freq_dict are not strings. They get converted to level of theory objects when loading arkane.encorr.data. I ran your example and it works fine for me using the level of theory object.

[oscarwumit]

Thanks, @cgrambow. When I did the test, I forgot to make clean and make. Because RMGobject has been modified, it is important to recompile the code. After recompiling, the code behaves normally.