# I tried NVT ensemble to release the initial stress of a model but it did not work well

Dear all,

I tried to use NVE ensemble to release the initial stress of a model. NPT ensemble cannot be used because the boundary is not periodic. I tried NVT ensemble but it did not work well. How should I do?

Thanks a lot.

Liu Bo

Dear all,

I tried to use NVE ensemble to release the initial stress of a model. NPT
ensemble cannot be used because the boundary is not periodic. I tried NVT
ensemble but it did not work well. How should I do?

who can tell from this extremely vague description?

axel.

What Stress are u talking about…?? Why is your system in a ""Initial Stress “” State?? What happened when u used NVE ensamble??? What did NVT ensamble give you as an o/p???

Sagar

Rule #1 of computer simulation: You have to do the thinking and then
make the computer work for you.
The opposite doesn't work.
Let's say your system is a finite size body embedded in a simulation
box. Your body is initially stressed (that's OK, could happen when
switching force fields, or taking data from ab-initio papers, etc).
Why do you think pure NVE didn't work? Think a bit... In NVE you
conserve energy and if you want to do a stress relaxation the initial
potential energy of your system has to be dissipated. There is no
dissipation mechanism in NVE. You'll only get conversion from
potential into kinetic energy forever.

1: Add to the NVE ensemble a damping force
2: Just employ an energy minimization technique to relax the initial
stress in your finite size body.

If you are working at a finite temperature and you have a finite size
body you can still apply the procedure above at T=0 and then crank up
the temperature making sure that you produce is done in a way that no
energy is added to the translational/rotational degrees of freedom
(assuming you want your system at rest).

More permanent solution to your problem:
Your systematic posts show that your background is very weak in many
areas. The library is a good place to start...

Carlos

Thanks. I tried to get the stress-strain curve of deforming a model. The initial pressure of the model should be released to zeros. I used the NPT ensemble as follows to accomplish this.

fix 1 all npt temp 300 300 1 iso 0 0 1 drag 1

I changed the Pdamp from 1 to 1000. It did not work. As you know NPT ensemble can only be used for periodic model. If I build a finite sized model, I should use NVT or NVE. I tried them but they did not work too.

Thanks and best regards,

Liu Bo

Thanks. Your comments are precisely on the points. I am trying you suggestions.

Best regards,

Liu Bo

Using Pdamp either 1 or 1000 isnt good. Pdamp=1 means your pressure is fluctuating wildly, and pdamp=1000 means your system is taking a very long time to equilibrate. Try using Pdamp=100.0 or 10.0

Also, have u tried without using the Drag keyword.?? Because Drag keyword interferes with the normal dynamics of NPT, so that u may not get what u expect. First try without drag. Too much a high value of drag is totally undesirable.(for the same reason)

U may also try using a continous chain of thermostats,as given in the documentation.

Sagar

Thank very much for helping. Your suggestions are very useful. I will try them.

Best regards,

Liu Bo