Initialization of molecular dynamics run

Total energy, geometry optimization, DFT+U, spin....

Moderator: bguster

Locked
natalie
Posts: 16
Joined: Sun Jan 17, 2010 10:24 pm

Initialization of molecular dynamics run

Post by natalie » Sat Jul 27, 2019 11:31 pm

This is more of a suggestion than a real problem in running molecular dynamics simulations with abinit. We found that with ionmov 12 (isokinetic) and 8 (Nose), abinit
uses mdtemp to set the temperature. We wanted to try a NVE calculation with ionmov 6 and were surprised that mdtemp did not set the temperature. We are able to
to control the temperature by initializing the atom velocities with a small program we wrote and adding that information to the input file with the "vel" key word.
In retrospect, this is totally logical, but it might be helpful to users to add a few sentences on the webpage describing the use of ionmov 6 (and related values of ionmov
such as 1 or 24). Thanks, Natalie

gregory
Posts: 5
Joined: Fri Jan 30, 2015 4:13 pm

Re: Initialization of molecular dynamics run

Post by gregory » Tue Aug 20, 2019 3:28 pm

Dear Natalie
It seems indeed that mdtemp cannot be used to initialize the velocities in a MD run with some values of ionmov.
Alternatively, you can use the PIMD options of ABINIT to do this job. For example, you can use the following input:

imgmov 9 #do not use ionmov
mdtemp 600 300 #at the first place, put the value you want for the initial temperature (here 600 K)
nimage 1
ntimimage 2
... and then your other variables (you can set all convergence parameters -- ecut, k-points, nstep -- to a minimal value to go quickly after the first time step).

Then, run this input a few seconds and stop ABINIT: after the first time step, ABINIT prints the initial velocities, that you can copy-paste in your true input containing ionmov 6 (with the variable "vel").

Using these PIMD options, the velocities are generated correctly, i.e. :
- randomly drawn in a maxwellian distribution
- rescaled so as to have exactly the desired initial temperature
- the velocity of the center of mass is removed

Best,
Gregory

Locked