Description of work
Updated the DOS calculation so that results are a bit more clear for new users.
Removal of the negative frequency part of the DOS.
Setting of the DOS to arb units. Since MDANSE DOS is actually the FT of a weighted VACF some actual units could be specified like previously. However A.U. are less confusing for new users since you would expect either A.U. or states per unit energy not the previous units of nm2/ps.
Set the default weighting scheme to atomic_weight so that the results are now proportional to the actual density of states.
Changed the DOS so that the sum of atomic density of states equal to the total.
Fixes
Closes #408, Closes #353
To test
Run a density of states calculation on the protos branch and then on the dos-weight-update branch the total DOS should be the same except the new result should only include positive frequency results and the DOS units should now be au.
Water diffusion constant DOS(0) test
Since we know that DOS(0) with equal weighting is proportional to the diffusion constant (D) and d/dt MSD = 6D. I ran the MSD with the long DLPOLY H2O trajectory runs and got the following plot.
Doing a simple two point gradient at the endpoints I get a approximate diffusion of 0.633738624175695 / (6 * 50) = 0.00211246208058565 nm^2/ps. Running DOS (equal weight) on the same trajectory I get.
DOS(0) = 0.0007601503675207263 and pi * DOS(0) = 0.002388082810226695 (The pi factor occurs because of the 1/2pi factor used in MDANSE for the foward FT). Which is similar to the MSD result above.
Description of work Updated the DOS calculation so that results are a bit more clear for new users.
Fixes Closes #408, Closes #353
To test Run a density of states calculation on the protos branch and then on the dos-weight-update branch the total DOS should be the same except the new result should only include positive frequency results and the DOS units should now be au.
Water diffusion constant DOS(0) test Since we know that DOS(0) with equal weighting is proportional to the diffusion constant (D) and d/dt MSD = 6D. I ran the MSD with the long DLPOLY H2O trajectory runs and got the following plot.
Doing a simple two point gradient at the endpoints I get a approximate diffusion of 0.633738624175695 / (6 * 50) = 0.00211246208058565 nm^2/ps. Running DOS (equal weight) on the same trajectory I get.
DOS(0) = 0.0007601503675207263 and pi * DOS(0) = 0.002388082810226695 (The pi factor occurs because of the 1/2pi factor used in MDANSE for the foward FT). Which is similar to the MSD result above.
Total DOS now a sum of atomic