[lammps-users] Lammps minimization

Dear LAMMPS USERS

I wanted to ask a question about the minimizer in Lammps. Is it possible that the minimizer gives a status of energy tolerance even if it is in a local minimum not a global minimum? And if that is possible and happens, is there is another way to fix that other than changing the starting configuration ? For example using a damped dynamics method ?

Thank you

Dear LAMMPS USERS

I wanted to ask a question about the minimizer in Lammps. Is it possible that the minimizer gives a status of energy tolerance even if it is in a local minimum not a global minimum?

a minimizer algorithm cannot not know whether a minimum is a local or global minimum.

And if that is possible and happens, is there is another way to fix that other than changing the starting configuration ?

the only reliable way to identify a global minimum would be to sample the entire phase space, however that is usually of such a high dimensionality, that it is practically impossible.

For example using a damped dynamics method ?

both, the quickmin and fire style minimizer styles are damped dynamics methods. However, all of the above applies to them just as well as to the other minimization algorthims.

Thank you for your assistance Dr. Axel

I’m doing some tests for my work and I’m testing a potential with the case of two single atoms in the simulation box paired with an LJ potential. I do the minimization to check the minimum energy which would become epsilon (theoretically it should be). When using the cg minimization the energy does not reduce to epsilon but when using the fire algorithm it does. So I was trying to understand why would that happen.