New Sigma Files (averaging = Mullins)

dinisAbranches commented 2 years ago

Sigma files (stored in UD/sigma1) obtained using the generate_all_profiles.py script with the options num_profiles = 1 and averaging='Mullins' (line 40, f = functools.partial(write_one, num_profiles = 1, averaging='Mullins')). These sigma files are different from the files inside UD/sigma for unknown reasons. They are needed for the validation of code that uses the Mullins averaging procedure to obtain sigma profiles, so perhaps it would be useful to include them in the repo?

ianhbell commented 2 years ago

I forwarded to my my colleagues for review. Hopefully they respond quickly. In the meantime, can I please ask you to add names to the header for the sigma files?

erikmickoleit commented 2 years ago

Hey there, I just checked the sigma profiles and the to_sigma.py script. The profiles in the UD/sigma folder seem to be very off. I do not know why. I use another sigma profile generator and it gives the exact same sigma profiles as with the generate_all_profiles.py and to_sigma.py scripts. I also compared the results to the original VT database by Mullins et al. The sigma profiles there deviate slightly from the sigma profiles here. The deviations are due to the following modifications:

double precision for Pi, the averaging radius r_av and square roots has been used here, which was not the case in the original sigma profile generator by Mullins et al.
sigma(i) values of each surface segment from the cosmo files were recalculated by calculating sigma(i)=charge(i) / area(i). We did that to get more accurate numbers as the cosmo file always just gives 7 digits for the sigma(i) values of each surface segment
Also we used the conversion factor from bohr radius to angstrom of 0.52917721067 instead of 0.529177249 originally used by Mullins et al. That includes the recalculation of r_av = sqrt(a_av/Pi)*0.52917721067 with a_av = 7.5 a.u.^2.

The sigma profiles in UD/sigma should be replaced by the profiles generated with the generate_all_profiles.py script. Does that answer the question?

dinisAbranches commented 2 years ago

Dear Erik,

Thank you for your reply. Yes, that answers my question. Would you mind replacing the current UD/sigma folder with the sigma profiles calculated using the generate_all_profiles.py and to_sigma.py scripts? I believe you can simply accept this pull request, as the changes proposed are just a new sigma profile folder calculated with those scripts. Of course, you may want to double check these files or regenerate them... Thanks for the help!

ianhbell commented 2 years ago

I think it might be better to close this PR rather than to have two folders of one-sigma profiles. Then we could regenerate, and add a README file.

@erikmickoleit want me to help you do this?

ianhbell commented 2 years ago

@dinisAbranches We have updated all the UD/sigma profiles with Mullins averaging with https://github.com/usnistgov/COSMOSAC/commit/32d03ebae2dce4e8863e5a73b10f7bcc88954b1f

usnistgov / COSMOSAC

New Sigma Files (averaging = Mullins) #18