Energy rise

Dear Lammps Users,

Hello, everyone. I used the lammps to equilibrium small organic system. The system seems equilibrated after a long simulation period under npt ensemble. However, when I continued to run the simulation, the energy increase at a point and enter to equilibration again. Dose anyone know what happen with the system, or is this result reliable? Any help would be appreciated.

Kerwin

etot.jpg

Dear Lammps Users,

Hello, everyone. I used the lammps to equilibrium small organic system. The
system seems equilibrated after a long simulation period under npt ensemble.
However, when I continued to run the simulation, the energy increase at a
point and enter to equilibration again. Dose anyone know what happen with
the system, or is this result reliable? Any help would be appreciated.

nobody can know for certain from just looking at the evolution of the
energy. there are several possible explanations.
the first thing to check is whether there is a significant change in
geometry that coincides with the change of total energy.
or whether this change happens at the point in time when you restart.

axel.

Dear Kerwin,

Does this happen as soon as you restart the simulation?
If this is the case, such a sudden increase in the energy could come from a change in the the mass of a group of atoms, or in the parameters of the some interaction potential (or in those of a bond type).
Check that you are not changing anything important when restarting the simulation.

Valerio

It did happen at restart process. And the only change is time step. When I turned the time step form 1fs to 2fs, the energy of system rose. Is that the problem?

Dear Kerwin,

Does this happen as soon as you restart the simulation?
If this is the case, such a sudden increase in the energy could come from a change in the the mass of a group of atoms, or in the parameters of the some interaction potential (or in those of a bond type).
Check that you are not changing anything important when restarting the simulation.

Valerio

2018-05-04 0:09 GMT+02:00 Chuanxing Zhan via lammps-users <[email protected]>:

Dear Kerwin,

I would say that changing the timestep shouldn’t give that result, but why don’t you test it yourself?
Try to do the same without changing the timestep and see whether the problem persists.

Valerio

Thanks for your advice. I did the test. The energy doesn’t change if I keep the same time step, it will rise if I increase the time step. I also agree with your point, the energy result should not relate to time step. I don’t think I changed any other parameter in my scripts.

Kerwin

Dear Kerwin,

I would say that changing the timestep shouldn’t give that result, but why don’t you test it yourself?
Try to do the same without changing the timestep and see whether the problem persists.

Valerio

2018-05-04 18:14 GMT+02:00 Chuanxing Zhan <[email protected]…16…>:

Thanks for your advice. I did the test. The energy doesn't change if I keep
the same time step, it will rise if I increase the time step. I also agree
with your point, the energy result should not relate to time step. I don't
think I changed any other parameter in my scripts.

it is difficult to give any more specific advice without actually
having a simple test case.
are you certain, that your system is able to run an MD in a stable
fashion at a 2fs timestep?
in other words, does your system, when run without a thermostat,
conserve energy?

you mentioned, that you are looking at small organic compounds. for
those often 0.5fs or less is required unless you use constraints on
all bonds to hydrogen atoms.
the graph you produced would be consistent with a scenario where you
do not have constraints, and then your thermostat compensates for the
instability. in that case, your average temperature would be above the
desired target.

axel.