Question regarding MD: ENCUT & SMASS

Message

simha · #1 Post by **simha** » Tue Aug 18, 2009 12:27 pm

Dear All

I would like to do (NVT) MD on a system containing H2O and a TM oxide. I have the following questions:

1) Will it be correct to do MD with default ENCUT for the equilibration run and then change the ENCUT to higher value for the production run? or one has to use higher ENCUT from staring?

2) In the electronic convergence its by default going to E-04. While running MD, if the NELM exceeds (does not reach E-04) are we forced to stop and do the MD again or is it acceptable ? lets say in the initial iterations.

2) A similar type of MD was done in this paper using VASP: J. Carrasco et.al, PRL 100 , 016101 (2008). The authors have used nose mass parameter = 24.3 E-27 a.u. But when I do MD with SMASS=1 , the nose mass parameter in OUTCAR is 0.540E-27a.u.

a) Is there a formula wherein I can check the SMASS value for 24.3 E-27 a.u.?

b) by using SMASS =1, In general can one get the dynamics right as compared to a higher SMASS value?

c) Is a higher value of SMASS more good in retaining a canonical ensemble?

I have read previous post and realize that there is no as such thumb rule for SMASS in NVT (if I am not wrong). But still I will be thankful and glad if some one share his/her experience in choosing SMASS.

Regards
Narasimham

#2 Post by **admin** » Mon Oct 12, 2009 2:59 pm

1) in general, you need not use a high ENCUT from the start. However, as H2O has rather short bond lengths, I would first of all test (in a small system) how large ENCUT has to be chosen to give reliable results (mind the strongly varying wavefunction oin the adsorbate molecule!). With the standard PPs for O and PREC=medium you should be on the safe side.
2) if a ionic step is not converged resonably, the forces are not reliable. I would stop the MD if this happened, because it may take some computing time to 'amend' this step. If electronic convergence is bad at some stage please check for the possible reason. H2O/TMO systems converge very well usually. Please also increase NELM in order to prevent that a MD step is not converged well before proceeding (default: NELM=60)
3) SMASS should be calculated from the characteristic phonon freqequencies of the system in order to avoid decoupling of the system from the Nose thermostat. If you are not sure about them. pelase use SMASS=0

VASP001 · #3 Post by **VASP001** » Tue Oct 27, 2009 4:45 pm

[quote author=medium you should be on the safe side.
2) if a ionic step is not converged resonably, the forces are not reliable. I would stop the MD if this happened, because it may take some computing time to 'amend' this step. If electronic convergence is bad at some stage please check for the possible reason. H2O/TMO systems converge very well usually. Please also increase NELM in order to prevent that a MD step is not converged well before proceeding (default: NELM=60)
3) SMASS should be calculated from the characteristic phonon freqequencies of the system in order to avoid decoupling of the system from the Nose thermostat. If you are not sure about them. pelase use SMASS=0 [/quote]
If I have caculate the frequency of the phonon for the system ,how can I make a correlation between the SMASS and the frequency?