What thermodynamic conditions data gathering run considers

LAMMPS LAMMPS General Discussion
1

hello Everyone
I am working on calculating the MSD values accurately, after running the NPT and NVT for 1 ns I did the data gathering run for 1fs so I was able to see alot of diffusion jumps by plotting the MSD. I want to know what thermodynamics conditions data gathering run considers because I see a continuous growth in the MSD as I increase the duration of data gathering run.

2

This question makes not much sense to me because
a) compute msd like any other compute is executed whenever its data is retrieved for processing/output for the first time on a given timestep and it is computed relatively to the positions at the time that the compute was created, and
b) shouldn’t the MSD increase linearly in the long time limit?

1 Like
3

LAMMPS does not know or consider anything beyond what you input. LAMMPS does not know about temperature, pressure, diffusion, phase transitions, or any other number of quantities its users keep asking it for. LAMMPS only does what it is told to do, since it is a computer program, but if LAMMPS does “know” something, it simply knows how to integrate equations of motion for particles beginning from an initial configuration, and it also “knows” that users sometimes ask it to calculate some strange combination of functions of the particles’ positions and velocities.

One of those strange functions is the MSD, or mean square displacement, and all LAMMPS knows is how to measure how far each particle has moved from its initial position and then average that over some user-specified group of particles. You can verify for yourself that compute_msd.cpp contains little beyond that. LAMMPS does not know what shape this function should have over time, and even if it did, LAMMPS does not know how those shapes correspond to temperature, phase state, viscosity, or non-Newtonian time scale effects. Knowing such things is your job, not LAMMPS’s.

Scientists solve scientific problems. LAMMPS simply solves partial differential equations. :slight_smile:

1 Like
4

Hello Dr. Akohlmey

  1. I guess question was not complete, while making some tests by using pure SiC without any vacancy when I activated the msd compute, I still observed an unphysical increase in the msd vs time graph. even analyzing the movie i didn’t see any diffusion or interstitials, so I was wondering what actually happened between MSD run and msd_computation that increased the MSD, I guess you have answered this.
  2. I agree with the current method of MSD calculation in LAMMPS total MSD should increase, if there is actually diffusion.
5

You need to check if your system is accumulating a center of mass momentum and thus drifting as a whole. You can define variables that look at the ‘vcm’ for each direction and output them with thermo_style custom.

1 Like
6

Dear Dr. Akohlmey
thanks for comment this seems reasonable to me. i can check it out