jonlym
commented
4 years ago This is an interesting question. I would advise to always be careful when eliminating frequencies as symmetry may result in duplicate frequencies that are physical (e.g. methane). That being said, the example you provided is legitimate.
You can modify your code to check if the molecular is linear/non-linear (perhaps by using pmutt.statmech.rot.get_geometry_from_atoms) and if the number of frequencies does not match your array, filter them further using a function like the following:
def get_unique_float(in_data, tol=1.e-6):
"""Returns unique values. Solution from https://stackoverflow.com/a/5429049/6865095
Parameters
----------
in_data : numpy.ndarray
Array with duplicate values
tol : float, optional
Tolerance to use to determine if values are unique. Default value
is 1.e-6
Returns
-------
out_data : numpy.ndarray
Array without duplicate values
"""
b = in_data.copy()
b.sort()
d = np.append(True, np.diff(b))
out_data = b[d>tol]
return out_dataWhen I fed the function the frequencies you posted, it returned the following:
[ 63.163151 64.352654 4300.073749]Let me know if you think that solves your problem.
MCVifpen
Hello,
I need some help with the set_vib_wavenumbers_from_outcar method to parse VASP frequencies from an OUTCAR file to a dictionary. I need a way to get automatically the right frequencies for a thermochemistry calculation.
VASP yields 3N frequencies, though for thermochemistry calculations you only need 3N-6 or 3N-5 eigenvalues, i.e.: it computes redundant values from translations and rotations. For example, if I apply the set_vib_wavenumbers_from_outcar method to a H2 frequency calculation (3N-5 = 1 expected frequency) I get
{'vib_wavenumbers': [4300.073749, 64.352654, 63.163151, 4300.073749, 64.352654, 63.163151]}
If I apply what seems to me a sensible cutoff to remove unwanted frequencies, such as set_vib_wavenumbers_from_outcar("OUTCAR",,min_frequency_cutoff=100) I still get {'vib_wavenumbers': [4300.073749, 4300.073749]} that is , two frequencies instead of one.
What would you to get one single frequency and in a robust way that you can apply to any molecule ?
Thank you for your help.
Kind regards, Manuel