averaging structure gives very high temperature

LAMMPS LAMMPS Mailing List Mirror
1

Hi,
      I am getting strange energy and temperature value for my runs. I am running my metal-U at 800K under NPT followed by NVT for 50 ps. I am dumping my structure at 2 ps intervals. After equilibrating energy and temperature, building average final structure by taking final 100 or 1000 ps steps' structures. Here I am getting weird results:

1. If I take the average structure and run NVT at same temperature (800K), it gives very high energy and temperature
output for last 2 steps averaged structure (first step is 800K because I set velocity to 800K)
           0 800 -11398.427 0 -10700.526 511646.31
        1 4438.0344 -14901.775 0 -11030.137 488895.39
        2 11154.093 -21109.81 0 -11379.24 423065.87
        3 15812.259 -25070.487 0 -11276.241 366431.75
        4 18046.79 -26931.656 0 -11188.056 337600.32
        5 19019.659 -27722.54 0 -11130.233 323727.49

2. if I take only last step structure, it gives slightly different numbers : -35884.8 is the cohesive energy at 800K
        0 800 -35882.458 0 -35184.557 164.93185
        1 799.95928 -35882.423 0 -35184.557 161.86385
        2 799.91102 -35882.38 0 -35184.557 159.0894
        3 799.85506 -35882.332 0 -35184.557 156.62886
        4 799.79118 -35882.276 0 -35184.557 154.5017
        5 799.71923 -35882.213 0 -35184.557 152.7292

I am unable to understand, what is the reason behind this high value of energy for averaged structure. Has anyone else noticed similar behavior?

Thanks
Priyank

2

Your first run shows a temp increase from 800 to 4400
in a single timestep. Obviously this is bad dynamics.
No thermostat will correct for that.

Steve

3

Hi,
     I am getting strange energy and temperature value for my runs. I
am running my metal-U at 800K under NPT followed by NVT for 50 ps. I am
dumping my structure at 2 ps intervals. After equilibrating energy and
temperature, building average final structure by taking final 100 or
1000 ps steps' structures. Here I am getting weird results:

1. If I take the average structure and run NVT at same temperature
(800K), it gives very high energy and temperature

Why do you expect, that an averaged structure over a long period is of
low potential energy?

Axel

4

No, I don't expect it to be of low energy but atleast of the same order. If you see the temperature it is going from 800 to 4000 in 1 step and then 20k in 5 steps. That is troubling me.

5

No, I don't expect it to be of low energy but atleast of the same order. If
you see the temperature it is going from 800 to 4000 in 1 step and then 20k
in 5 steps. That is troubling me.

just consider an atom that moves around an area of
high potential energy in a circle. its average position
will be in the center where there is very high potential energy.
this is a very simplistic example, but shows how an
average position can be very much misrepresenting
the potential hypersurface that atoms in your system
are experiencing.

cheers,
      axel.