problem during minimization of waterbox

Hello,

I am an undergraduate student learning about computational materials science at my university. I was introduced to LAMMPS and it was suggested that I try to run a waterbox simulation in order to learn the basics of LAMMPS. However, I am running into trouble during the minimization of the waterbox. My potential energy values decrease and approach zero, but then get very large and negative as shown below.

Setting up cg style minimization …

Unit style: real

Memory usage per processor = 9.9399 Mbytes

Step PotEng TotEng Press Temp

0 70292483 70292483 3.012731e+08 0

100 3569344.3 3569344.3 11499218 0

200 392467.63 392467.63 972098.97 0

300 72140.856 72140.856 135231.38 0

400 20213.842 20213.842 23171.154 0

500 9675.8074 9675.8074 1687.113 0

600 5908.8743 5908.8743 -360.53153 0

700 3438.2235 3438.2235 -80.822008 0

800 533.55157 533.55157 8.2978363 0

821 -1.6787387e+16 -1.6787387e+16 -3.4633089e+15 0

Loop time of 84.5311 on 4 procs for 821 steps with 4500 atoms

Minimization stats:

Stopping criterion = quadratic factors are zero

Energy initial, next-to-last, final =

70292482.5548 -1.67873873717e+16 -1.67873873717e+16

Force two-norm initial, final = 1.44413e+09 3.34124e+30

Force max component initial, final = 8.43602e+08 2.36261e+30

Final line search alpha, max atom move = 2.12658e-46 5.0243e-16

Iterations, force evaluations = 821 1273

I have tried a few things to troubleshoot, including: expanding the size of my box, adjusting the stopping criteria for the minimize command, generating a new waterbox. Do you have any suggestions on what might be causing this? I will attach my input files to this email.

Thanks,
Jeremy

water_input.in (1.87 KB)

out.4369 (4.54 KB)

packmol_waterbox.dat (307 KB)

Hello,

I am an undergraduate student learning about computational materials
science at my university. I was introduced to LAMMPS and it was suggested
that I try to run a waterbox simulation in order to learn the basics of
LAMMPS. However, I am running into trouble during the minimization of the
waterbox. My potential energy values decrease and approach zero, but then
get very large and negative as shown below.

Setting up cg style minimization ...

  Unit style: real

Memory usage per processor = 9.9399 Mbytes

Step PotEng TotEng Press Temp

       0 70292483 70292483 3.012731e+08 0

     100 3569344.3 3569344.3 11499218 0

     200 392467.63 392467.63 972098.97 0

     300 72140.856 72140.856 135231.38 0

     400 20213.842 20213.842 23171.154 0

     500 9675.8074 9675.8074 1687.113 0

     600 5908.8743 5908.8743 -360.53153 0

     700 3438.2235 3438.2235 -80.822008 0

     800 533.55157 533.55157 8.2978363 0

     821 -1.6787387e+16 -1.6787387e+16 -3.4633089e+15 0

Loop time of 84.5311 on 4 procs for 821 steps with 4500 atoms

Minimization stats:

  Stopping criterion = quadratic factors are zero

  Energy initial, next-to-last, final =

         70292482.5548 -1.67873873717e+16 -1.67873873717e+16

  Force two-norm initial, final = 1.44413e+09 3.34124e+30

  Force max component initial, final = 8.43602e+08 2.36261e+30

  Final line search alpha, max atom move = 2.12658e-46 5.0243e-16

  Iterations, force evaluations = 821 1273

I have tried a few things to troubleshoot, including: expanding the size
of my box, adjusting the stopping criteria for the minimize command,
generating a new waterbox. Do you have any suggestions on what might be
causing this? I will attach my input files to this email.

​this looks like bad potential parameters or bad settings (e.g. units)
related​ to them. this is something you should first and foremost discuss
with the person supervising and advising you.

axel.

FYI,

just got around to have a closer look at your input.
the major issue are the force constants for bonds and angles. since you have a rigid water model, you have to crank them up for the minimization (since there is no shake during minimization). otherwise, your molecules deform too much and you get a “coulomb catastrophe” due to hydrogen atoms being dragged on top of oxygens.

axel.