[lammps-users] deformed box after npt run with pressure control style "xyz"

LAMMPS LAMMPS Mailing List Mirror
1

Thanks Steve.

I’ve tried two things:

1. As you told me, I increase the precision and found that the dilation factor is different in xyz even if it appears at 10^-7. That for the z-axis is smaller. The ratio of Lx:Ly:Lz changes from 0.386 to 0.391.

As follows:

  0 2.0232614 2.9235608 5.5379974 62108.72 -5.5149004 3.0347186 28.839654 28.839654 74.674529 5.6099543 5.7951015 5.2089364 -0.057484868 0.14193013 -0.017237353

BOX DILATE: 0 1.0000010618 28.8397

BOX DILATE: 1 1.0000010618 28.8397

BOX DILATE: 2 1.00000085774 74.6745

BOX DILATE: 0 1.0000010618 28.8397

BOX DILATE: 1 1.0000010618 28.8397

BOX DILATE: 2 1.00000085774 74.6746

...

15620 2.651821 3.8031009 2.6409022 93210.505 -1.9305568 3.977504 33.089003 33.089003 85.132912 2.6616096 2.6479815 2.6131157 -0.012724146 0.070555751 -0.061006088

BOX DILATE: 0 1.0000037263 33.089

BOX DILATE: 1 1.0000037263 33.089

BOX DILATE: 2 1.00000352223 85.1329

BOX DILATE: 0 1.0000037263 33.0891

BOX DILATE: 1 1.0000037263 33.0891

BOX DILATE: 2 1.00000352223 85.1332

...

29999 2.6016707 3.7902786 3.302639 86438.794 -2.0548799 3.9022829 32.330657 32.330657 82.695075 3.3702441 3.3133025 3.2243705 -0.0099711106 0.030020318 -0.028955017

BOX DILATE: 0 0.999995338825 32.3307

BOX DILATE: 1 0.999995338825 32.3307

BOX DILATE: 2 0.999995134759 82.6951

BOX DILATE: 0 0.999995338825 32.3305

BOX DILATE: 1 0.999995338825 32.3305

BOX DILATE: 2 0.999995134759 82.6947

2. The para-pure Gay-Berne bulk system has been run with the same p-style xyz which was built by delete all the LJ walls in the initial structure. It is a big surprise that the ratio of Lx:Ly:Lz changes from 0.3862 0.3862 1.0000 to 0.3910 0.3910 1.0000 after 30000 step npt xyz runnings. But the ratio keeps well as a constant for the pure GB bulk after the same simulation.

So maybe the error is caused by the precision. It just happens in some case.

2

ok - this helps. Please post your input script and any associated
data file and I'll take a look.

Steve

3

I ran your files. Your restart file has leftover info from a previous
run where you used a fix npt with the same name, but presumably
different params - was it a fix npt aniso? This is causing the new
fix npt to reset with those params (LAMMPS prints a message about that).
In the previous run, the dilation factors were not equal (aniso?) so
those get inherited in the new run, and then things don't proceed as
you expect.

If you simply replace
fix npt GB npt/asphere 2.99 2.99 0.1 xyz 4.4 4.4 100 dilate partial
with
fix nptnew GB npt/asphere 2.99 2.99 0.1 xyz 4.4 4.4 100 dilate partial

then the IDs are different and this won't happen. I.e. the dilation
factors that get printed out as the run starts up are identical in all
3 dims - so I don't think the aspect ratio will change.

Steve