linear search alpha

Dear lammps users,

Recently, when checking the minimize algorithm, I found a very puzzling problem. I increased the model cross-sectional area from 10x5 to 10x6, then the stopping criterion changed from “force tolerance” to “linesearch alpha is zero”.

From the manual I learn that “linesearch alpha is zero” is often caused by the discontinuity of energy around the potential cutoff, but this change of model size also lead to alpha=0. Could you please give me some hint on the possible reason? For convenience, the input script is attached.

Best wishes,

in-check (971 Bytes)

As far as I understand minimizes, an alpha of zero means that it could not find a change in the potential energy along the forces or that it is already at the minimum (i.e.,the energy increases in both directions along the forces). Does the minimized system look good? If so I would not worry too much. You could also try a smoothed potential if it is applicable for you (like lj/smooth/linear).

Steve or Axel might be able to provide more insight.