Potential Energy at Finite Temprature

LAMMPS LAMMPS Mailing List Mirror
1

Dear All,

I want to find the potential energy of a metal/ceramic material at finite temperature, 300K for example, such that two fixed regions (metal at the top and the ceramic a the bottom with the thickness of 1nm for each) are embedding the material. Moreover, the boundary condition perpendicular to the interface is shrink-wrapped and periodic in others. I have some questions as follows.

  1. Using MD with NVT ensemble with 100 ps and timestep of 1 fs, I want to equilibrate the material at the given temperature, then quench it to 0.001 K within 1 ps using NVT again, and finally minimizing the potential energy of the system by minimization. By the way, I assumed this annealing and quenching scheme is meaningful; please let me know if it is not. I’m afraid that I may lose the effort made by MD to reach an equilibrium state due to minimization, and the system may fall into a trap during the minimization.

  2. Using MD with NVT ensemble with 100 ps and timestep of 1 fs, I want to equilibrate the material at the given temperature and then calculate the average potential energy using fix/ave for the last 50 ps.

Could anyone please tell me which scheme is more physical meaningful and especially feasible in lammps to be used? I would greatly appreciate it if someone can tell me about another scheme that I am not aware of it.

Bests,

Reza.

2

You should take a text book on statistical mechanics and gather essential background information of what you are about to do. Asking on this mailing list is no substitute for it. Not having the necessary insight is likely to result in you wasting your time and computing useless data.

two core points:

  1. a minimization will try to bring the system to the next (best?) potential energy minimum that the minimization algorithm will find. for trivial systems, that will be the global minimum, for all others some local minimum will be found (often dependent on the specific settings and minimizer algorithm and the ruggedness of the potential hypersurface). so doing finite temperature MD followed by minimization can be used to map out accessible local minima in the the potential energy landscape, but will not provide the information you are looking for.

  2. the distribution of states will follow the statistical mechanical ensemble that you are studying. this will also allow you to determine the distribution of potential energy. i haven’t looked at this in quite a while, so i am not certain whether a simply averaged potential energy will be meaningful in this context. a statistical mechanics textbook should provide an answer or at least show a path how to derive the answer.

axel.