I have tested a simple example using Polymer Modeler.

I found that in the lammps input generated by this module, the minimization parts are divided into 3 steps.

The epsilon and sigma values for L-J interactions are reduced to 50% of original values,

then they return to the original value within 2 steps.

Are there any special reasons to minimize systems with 3 steps?

This procedure using 3 steps minimization could be efficient but I think we should not change the sigma values which could affect the geometry strongly.

The second question is about the total energy.

During the minimization step with reduced LJ parameters, the total energy seems to be negative which looks okay, but after the parameters are returned to the original values and MD steps using the buckingham potential, the total energy is positive.

I understand the absolute value of total energy does not mean so much. But I still have a doubt.

0 Like 0 Dislike

Alejandro Strachan@ onThanks for the comments.

We use the three step minimization with scaled L-J parameters to avoid large forces due to bad contacts (nearby atoms) during the initial steps after the Monte Carlo build. This is explained in detail in our paper:

https://arxiv.org/abs/1503.03894

You are also correct that the total energy does not mean much. Most of the covalent terms arr designed to be positive. Only relative energies are important.

Reply Report abuse

Please login to answer the question.