QuantumATK Forum
QuantumATK => General Questions and Answers => Topic started by: Dipankar Saha on May 23, 2016, 21:21
-
Hi,
1) What is the reason for checking _"remove the center-of-mass momentum"_ in the case of NPT simulation?
2) Moreover, why do you go for the NPT relaxation, prior to the "NVT Nose Hoover "..., where the actual target is to equilibrate the sample with a constant temp. ?? / What is chain length (By setting this..., are you diving the entire system in to a number of slabs..?) ??
Thanks and Regards_
Dipankar
-
1) What is the reason for checking _"remove the center-of-mass momentum"_ in the case of NPT simulation?
The center-of-mass sometimes may show some drift due to non-zero center-of-mass-forces (e.g. caused by random forces as in Langevin, or by inaccuracies in the force calculations, e.g. when using reciprocal methods to calculate long-ranged-interactions such as CoulombSPME, or also in DFT), which is often unwanted. This option will fix the center-of-mass of the system during the simulation and remove such effects, caused by non-zero forces on the center-of-mass.
Be aware that in NPT simulations this constraint does not remove the fluctuations of the center-of-mass due to changing cell sizes, but this will only be small oscillations.
2) Moreover, why do you go for the NPT relaxation, prior to the "NVT Nose Hoover "..., where the actual target is to equilibrate the sample with a constant temp. ??
The NPT simulations are used to relax the strain which has been introduced by generating the interface. So first, one would equilibrate the sample at constant pressure and temperature to find the equilibrium lattice constants, and then with these lattice constants one can equilibrate again at constant temperature to thermalize the system. The second step might not absolutely be necessary but it will facilitate a constant average temperature in the subsequent NVE-simulation.
What is chain length (By setting this..., are you diving the entire system in to a number of slabs..?) ??
In modern Nose-Hoover-like thermostats and Barostats (i.e. NVTNoseHoover and NPTMartynaTobiasKlein) one uses not one thermostat but a series of thermostats which subsequently act on each other, so one thermostat which couples to the velocities of the system, another thermostat which couples to the first thermostat and so on. This technique suppresses the oscillations in the temperature (and in the cell vectors), especially when small systems are simulated. The chain length-parameter specifies how many of these thermostats are used, but it should not affect the results of your simulations.
The default value of 3 should work fine in almost all situations, only if still you encounter large oscillations in the temperature, you can increase this value.
-
I understand your points..... / Thank you so much Julian... for all the details !! :)
________
I quick quest. _
what's the expression used for finding the temperature (T)...while deriving 'temp. gradient' ?
Is it_
T=(1/ 3NKB) * Summ.{i to N} [mi (vi)2] ??
If so... what will be the value of "N"...(if I go with the default value of chain length, i.e. 3 ) ? Is it... something like...the total no. of atoms of the entire system divided by 3 ??
Best_
Dipankar
-
what's the expression used for finding the temperature (T)...while deriving 'temp. gradient' ?
Is it_
T=(1/ 3NKB) * Summ.{i to N} [mi (vi)2] ??
Yes.
If so... what will be the value of "N"...(if I go with the default value of chain length, i.e. 3 ) ? Is it... something like...the total no. of atoms of the entire system divided by 3 ??
No, the chain length has nothing to do with the temperature calculation. It is more a technical thing, that you can essentially ignore, in particular when you analyze the results.
-
What's this "N" then? Is this the total no. of atoms (considering NPT /or, NVT ensemble) ??
-
N is the total number of atoms for the global temperature, or the number of atoms inside a spatial bin in the Temperature profile analysis.
-
Okay... / Thanks...!! :)