Dear all,
I have two questions regarding simulation of a polymeric system.
-
When I try to equilibrate the polymer box with two different force fields, I find that in one case Potential Energy is positive, while in the other case it is negative. My understanding was that for solids, potential energy is typically negative as it describes the negative of the energy required to separate the molecules involved. So am I correct in assuming that the parameters for the force field in the first case are incorrect (where pe is +ve) or is this possible?
-
I have a small box of polymeric molecules but I plan to simulate a larger system by replicating. As mentioned in one of the latest mails, should I simply heat treat the box, equilibrate and then replicate it, or do I need to perform equilibration run again after the replication?
Thank you
Ganesh
Dear all,
I have two questions regarding simulation of a polymeric system.
1. When I try to equilibrate the polymer box with two different force
fields, I find that in one case Potential Energy is positive, while in the
other case it is negative. My understanding was that for solids, potential
energy is typically negative as it describes the negative of the energy
required to separate the molecules involved. So am I correct in assuming
that the parameters for the force field in the first case are incorrect
(where pe is +ve) or is this possible?
since you are using empirical potentials, the absolute energy
of your system depends on the functional form of the potential.
since the forces are all that you are interested in, and since those
are the gradient of the potential, any constant total energy term
will vanish when computing forces and thus have no impact
on the system. it is still possible, that you are making an error,
but it cannot be safely inferred from the absolute value of the
potential energy.
2. I have a small box of polymeric molecules but I plan to simulate a larger
system by replicating. As mentioned in one of the latest mails, should I
simply heat treat the box, equilibrate and then replicate it, or do I need
to perform equilibration run again after the replication?
it is a good idea to equilibrate for a bit after the replication, but
you don't have to equilibrate as much, if your system is already
well equilibrated before replication. due to replication, you impose
an unwanted periodicity onto your system which you want to get
rid of. also, larger cells allow more equilibration "pathways"
and you have to make sure that your smaller system is not trapped
in some meta-stable state.
cheers,
axel.
Hi Bala,
You should equilibrate the system after replicating it because the
additional degrees of freedom that appear after immediately
replication are not thermalized. This is especially important if you
want to do some equilibrium calculation (e.g. diffusion, thermal
conductivity, viscousity, etc) in which the physics is dominated by
extended ('long wavelength') degrees of freedom.
Zhun-Yong
I have tried to equilibrate the same system using two different approaches and I face some issues. In the first case I replicate a small box of molecules to the desired size and equilibrate. This works fine. In the second case I equilibrate the small box, and follow that up with a replication and further equilibration, but apparently get the error about lost atoms during the replication process. I tried using different equilibration schemes and sets of processors, but this seems to occur during all attempts. Periodic boundaries are used in all the cases. Is there a specific technique to equilibrate+replicate+equilibrate?
It’s hard to tell why you’re seeing the lost atoms error since it could be due to many different issues — this is a common problem users see when their system is in a non-physical configuration or has an incorrectly defined force field.
But since your first approach works fine, you can probably safely use that. The only drawback is that it might be more computationally expensive to do just replicate+equilibrate since the equilibration is on the large system and will likely need to be longer. But if computational cost isn’t a huge issue in this case, and you can ensure that your system has fully equilibrated, I’d recommend simply using this approach and dropping the other approach.
There’s no magic formula or standard procedure. Use what verifiably works.
Paul