Regarding some problem in output in two different machine

LAMMPS LAMMPS Mailing List Mirror
1

Hi everyone,

I have a one small question, we have two cluster. i am submitting job same script file, same parameter ,same configuration file in two cluster. initially in both cluster giving same results but after 10000 step output data are changing. can any any one tell me why this is happening?

My output data in cluster1 after 50000 step is:

step atom temp press total energy Pe ke

50000 512 0.04134527 0.011288482 -1.7540739 -1.8159707 0.061896777

volume

1057.6832

Output data in cluster2 after 50000 steps is:

50000 512 0.043425209 0.00053326428 -1.7466686 -1.8116792 0.065010591 1058.3514

I am using same script file & same configuration every thing same.

Regards

Krishna

2

Round-off drift? See section 12.1 of the manual

which discusses this.

Steve

3

Thank you, But which one i will consider correct, because i am trying to find the transition point. In cluster1 transition point is 20000000 and cluster2 it is showing 7000000 step. There is to much large difference. This happen with only one temp. also i checked this two to three temp. I found its transition point in both cluster is approx same step. only i am worried about that temp .

4

If it’s round-off they are both equally “correct”. Read up about
MD trajectories - you are just exploring an energy surface.

From your description, it sounds like you would need to

many simulations and gather statistics on “events” occurring
to say something meaningful about transitions.

Steve

5

Hi,

First i want to sorry again that i am asking same question which is little bit not clear. I am finding transition point base on cumulative potential energy. If it is not round-off error then what exactly i will check? If any one want to see my script file i will attach. Can any one give me some suggestion?

Regards

krishan

6

Krishna,

Steve already told you what is probably going on. Like he said, you need to read up on how MD trajectories work and how to report statistically meaningful results. Popular books are Allen and Tildesley’s Computer Simulation of Liquids or Rapaport’s The Art of Molecular Dynamics Simulation.

Salomon