Energy conserved in MD simulation using NVE ensemble

[gniding]

Dear Sir,

I am working on the MD simulation using NVE ensemble. In the INCAR, I set SMASS=-3, and I plotted the total energy, potential energy and kinetic energy. I found that the total energy is not equal to the summation of potential energy and kinetic energy. However, when I set SMASS=-3, MDALGO=1 and ANDERSEN_PROB=0.0, the total energy(potential energy + kinetic energy) is conserved. So, could anybody kindly tell me why?

Thanks,
chao

[martin.schlipf]

Could you elaborate what exactly you observe: Do you not understand the output of Vasp or do you get different results when you different Incar settings to get an NVE ensemble?

In the former case, I would like to point you to our MD section in the manual which is quite comprehensive.

If it is the latter case, could you please check if you observe the same for the tutorial example? If it is specific to your system, we would need to know about the particular setup. If you find the same trend for the tutorial, we have to figure out whether a fix is required.

[gniding]

Dear Sir

Thanks for your quick response. I got different results when I used the above two INCAR setting. I am running the MD simulation with NVE ensemble of the graphene sliding on the three layers Cu surface with the bottom Cu layer fixed. Given the initial velocity to graphene, then plot the total energy, the result was shown in the following link. https://drive.google.com/file/d/1JElUG0 ... sp=sharing

For SMASS = -3, the total free energy was not conserved. However, for SMASS=-3, MDALGO=1, ANDERSEN_PROB=0, the total free energy was conserved.

Thanks,
Chao

[martin.schlipf]

Did you try to reproduce the effect in the tutorial system?

If it only appears in your system, I would need to see the POSCAR, INCAR, KPOINTS, and POTCAR file.

Also your figure is not set to public, I cannot access it.

[gniding]

Thanks Sir. I will check it out.

One more thing that in VASP MD simulation, I will consider the single graphene layer on Cu surface, I will give an initial velocity to the graphene, Cu surface be rigid. During the MD simulation, can I keep the velocity as constant?

Thanks

[martin.schlipf]

You can constrain your MD run (see here and here). However, if I understood correctly what you want to do (constraining the graphene to have a certain velocity), I don't think that is possible. But I also do not understand why you would want to do such a thing. In which experiment would you have control over the velocity of individual atoms?