LAMMPS WWW Site - LAMMPS Documentation - LAMMPS Mailing List Archives
Re: [lammps-users] linesearch alpha is zero - LAMMPS17
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lammps-users] linesearch alpha is zero - LAMMPS17

From: Axel Kohlmeyer <akohlmey@...24...>
Date: Tue, 13 Jun 2017 13:30:26 -0400

On Tue, Jun 13, 2017 at 12:53 PM, Alexandra Davila <davila@...6925...4...> wrote:
Hi lammps-users,

I am trying to minimize by using the force criterion a system of (only) 65 Au atoms with the reax potential.

This is my input file:

# REAX potential for CHO system

units real

boundary p p f
atom_style charge

group freeze id 1:32
group unfrozen subtract all freeze

pair_style reax/c NULL
pair_coeff * * ffield.reax.AuO Au
fix 1 all  qeq/reax 10 0.0 10.0 1e-6 reax/c

neighbor 2 bin
neigh_modify every 1 delay 0 check yes

thermo 1
thermo_style custom step temp pe

fix 2 freeze setforce 0.0 0.0 0.0

dump 1 all custom 1 out.trj id element type x y z  fx fy fz q

min_style cg
min_modify dmax 0.01
minimize 0.0 1.0e-10 10000 100000


In the output file I found:

 Minimization stats:
 Stopping criterion = linesearch alpha is zero

The forces don’t fulfill the criteria, why? I tried another potential (eam) and it worked!

I really would like to know what I am doing wrong. I have seen that many other people had a similar problem, but their solution didn’t help me.

​there is nothing really wrong. these kinds of things can happen with certain minimization algorithms. the error message effectively means, that the minimization algorithm cannot figure out in which direction to move. minimization algorithms are not guaranteed to always drop into the global minimum.
for some systems, and some force fields, it can be difficult to find a minimum. sometimes also, the force field can contribute to this by being corrugated with "ripples" and not as smooth as others. since you are immobilizing some atoms, you're making the situation even more complex.

here are some suggestions that can help:
- switch the minimization algorithm and then go back
- or run some MD steps and then minimize
- (slightly) randomize your coordinates and minimize again
- or -as you noted- switch the force field and then go back

all of these will do about the same thing: they will modify your positions in some more-or-less systematic way and may help to have your system "jump" out of where it is "trapped" on the potential hypersurface.




I am (very) new using lammps.
A. Dávila
AG. Pehlke
Institut für Theoretische Physik und Astrophysik
Christian-Albrechts-Universität zu Kiel
Leibnizstr. 15
24118 Kiel

Check out the vibrant tech community on one of the world's most
engaging tech sites,!
lammps-users mailing list

Dr. Axel Kohlmeyer  akohlmey@...12...24...
College of Science & Technology, Temple University, Philadelphia PA, USA
International Centre for Theoretical Physics, Trieste. Italy.