[ase-users] Units in ASE

[Gaël Donval]

Dear all,

I'm working on integrating VASP velocities in ASE.

VASP always outputs velocities in A/fs (one can give velocities as
base_vector/time_step but this is not the point here).

I had problems recalculating the kinetic energy and the temperature from
ASE using the appropriate functions: atoms.get_kinetic_energy() and
atoms.get_temperature(). Here is what I did:

1) Convert units to SI (I'm more comfortable with them):
        mass_si = mass * ase.units._amu # Da to kg
        v_si = v_vasp * 1e5 # A/fs to m/s

2) Calculate the kinetic energy in eV:
        ((mass_si*(v_si**2).sum(axis=1)/2.)*units.J).sum() # in eV

3) Calculate the temperature in Kelvins:
        (2*ekin/(3*len(atoms)))/units.kB
        
I could compare 2) and 3) with the results given in VASP. Basically, I
get almost the same values, but not exactly (VASP: 5.326 eV,
Calculations: 5.313 eV). I'm curious to get your opinion about this
difference. I know the mass and the velocity with at least 6
significative figures ... I acknowledge that these results are close
from each other but I expect at the very least 4 identical figures here.
Maybe the masses...


This is not all. When I use ase and put the velocities in as m/s, just
like above, the calculated energies and temperatures are completely off
(by a factor of 100, Ek=0.05434 u. --this is not even 0.05313...).

I guessed this was because of units. I can't change masses, but only
velocities. Since I have a factor of around 100, I just had to multiply
my velocities by 10. The problem is that I don't see how that would make
sensible units. Would anyone more familiar with atomic sets of units be
willing to shed light on this?

Thanks in advance,
Gael