[lammps-users] [[email protected]: Question about minimize command in LAMMPS]

First, please post Qs to the LAMMPS mailing list. Others may
then comment.

Minimize in LAMMPS means 0K. I don't know what it means
to minimize at a finite T, at least for MD.

To relax at a high T does not mean minimize, but to run dynamics
at a high T and let the system equlibrate. E.g. run a few 1000 steps
with fix nvt or some other thermostat.

To anneal, just means to change the temperature at some desired
rate. The various thermostats can all ramp their target T over the
course of a long run, by specifying an initial and final T. So you
can do an anneal that way.



I have a question confusing me for a long time. For crystal or equibrium state materials, it is easy to understand that the minimization means letting the system find a local/global energy minimization state without themal fluctuation (0K). But for an non-equilibrium system, like disordered glass, a lot of argument arise for the “inherent structure” when I read many literatures. The whole story like this:

The glass sturcutre was obtained by quenching the system from a liquid conf to the temperature well-below the glass transition Tg. PBC were used in three dimensions. With the as-quenched sample at a given T, they tried to get a more stable conf by using energy minimization method. They called the minimized structure “inherent structure” , which means it is the local minimum in the energy landscape at finite T. Note that even liquid, they also have “inherent structure”.

Do you think the “inherent structure” make sense, physically?

Any comments are highly appreciated.


I think quenching means reducing the T while running dynamics.
If they do an eng min, then they mean T = 0 and no dynamics, just
a min. You could alternate these to try and fully relax a glass.