compute group/group -serial and parallel

LAMMPS LAMMPS Mailing List Mirror
1

Hi users,

I am trying to calculate the potential energy of each molecule with respect to rest of the system using

compute group/group using rerun command.

I ran serial and parallel, after 600 steps the values are diverging.

Can anyone advice me using compute group/group in parallel.

input script

group mol29 molecule 29

group rest29 subtract all mol29
compute pe29 mol29 group/group rest29 kspace yes

group mol30 molecule 30
group rest30 subtract all mol30
compute pe30 mol30 group/group rest30 kspace yes

thermo_style custom pe ecoul elong evdwl c_pe29 c_pe30
thermo 1
thermo_modify flush yes line multi

rerun …/…/…/dump.npt skip 2 dump x y z

1 proc

---------------- Step 0 ----- CPU = 0.0000 (sec) ----------------
PotEng = -8392.4635 E_coul = 323.5779 E_long = -547.3164
E_vdwl = -53345.9066 pe29 = -71.6026 pe30 = -60.5491
WARNING: Inconsistent image flags (…/domain.cpp:600)

---------------- Step 200 ----- CPU = 11.2114 (sec) ----------------
PotEng = -8426.3724 E_coul = 316.4418 E_long = -547.0831
E_vdwl = -53390.2288 pe29 = -71.9553 pe30 = -64.7873
WARNING: Inconsistent image flags (…/domain.cpp:600)
---------------- Step 400 ----- CPU = 16.6296 (sec) ----------------
PotEng = -8531.6209 E_coul = 319.1370 E_long = -546.9825
E_vdwl = -53546.8306 pe29 = -74.8128 pe30 = -64.6230
WARNING: Inconsistent image flags (…/domain.cpp:600)
---------------- Step 600 ----- CPU = 22.0560 (sec) ----------------
PotEng = -8520.1535 E_coul = 316.5312 E_long = -546.9687
E_vdwl = -53499.4717 pe29 = -73.7694 pe30 = -65.5856
WARNING: Inconsistent image flags (…/domain.cpp:600)
---------------- Step 800 ----- CPU = 27.4513 (sec) ----------------
PotEng = -8314.1948 E_coul = 313.8077 E_long = -547.0886
E_vdwl = -53435.4951 pe29 = -75.5358 pe30 = -65.1085
WARNING: Inconsistent image flags (…/domain.cpp:600)
---------------- Step 1000 ----- CPU = 32.8729 (sec) ----------------
PotEng = -8337.4114 E_coul = 320.6469 E_long = -547.0152
E_vdwl = -53408.4821 pe29 = -75.3971 pe30 = -63.1014
WARNING: Inconsistent image flags (…/domain.cpp:600)

8 proc

---------------- Step 0 ----- CPU = 0.0000 (sec) ----------------
PotEng = -8392.4635 E_coul = 323.5779 E_long = -547.3164
E_vdwl = -53345.9066 pe29 = -71.6026 pe30 = -60.5491
WARNING: Inconsistent image flags (…/domain.cpp:600)
---------------- Step 200 ----- CPU = 1.5362 (sec) ----------------
PotEng = -8426.3724 E_coul = 316.4418 E_long = -547.0831
E_vdwl = -53390.2288 pe29 = -71.9553 pe30 = -64.7873
WARNING: Inconsistent image flags (…/domain.cpp:600)
---------------- Step 400 ----- CPU = 2.2872 (sec) ----------------
PotEng = -8531.6209 E_coul = 319.1370 E_long = -546.9825
E_vdwl = -53546.8306 pe29 = -74.8128 pe30 = -64.6230
WARNING: Inconsistent image flags (…/domain.cpp:600)
---------------- Step 600 ----- CPU = 3.0379 (sec) ----------------
PotEng = -8520.1535 E_coul = 316.5312 E_long = -546.9687
E_vdwl = -53499.4717 pe29 = -73.7694 pe30 = -65.5856
WARNING: Inconsistent image flags (…/domain.cpp:600)
---------------- Step 800 ----- CPU = 3.7885 (sec) ----------------
PotEng = -8314.1948 E_coul = 313.8077 E_long = -547.0886
E_vdwl = -53435.4951 pe29 = -81.3596 pe30 = -70.9322
WARNING: Inconsistent image flags (…/domain.cpp:600)
---------------- Step 1000 ----- CPU = 4.5453 (sec) ----------------
PotEng = -8337.4114 E_coul = 320.6469 E_long = -547.0152
E_vdwl = -53408.4821 pe29 = -82.3076 pe30 = -70.0119

Thanks

Karteek Kumar

2

Hi users,

I am trying to calculate the potential energy of each molecule with respect
to rest of the system using

compute group/group using rerun command.

I ran serial and parallel, after 600 steps the values are diverging.

Can anyone advice me using compute group/group in parallel.

it is *very* unlikely that compute group/group is broken and drawing
that conclusion from your observation is unfounded. it is much more
likely that there is something wrong with how you feed your trajectory
data to LAMMPS or that you are using an outdated version with a known
bug. have you looked here?
http://lammps.sandia.gov/bug.html ? have you visualized the trajectory
around the jump in the energy? does it coincide with an atom/molecule
jumping across a periodic boundary? doesn't the warning about
inconsistent image flags bother you?

in addition have have to make the "usual complaints":
- you don't specify the version of LAMMPS you are using?
- you don't provide a way to reproduce the behavior

after being reminded of this quite a few times already, you should know better.

axel.

3

Thanks Axel,

I have updated lammps to latest version ( 30sep2013) - which overcomes inconsistent image flags.

But still difference in energies with 1proc and 8 proc continues after 600 step.

I visualized the movie, but couldn’t find any atoms to overlap each other.

I am doing npt simulation in original traj. so while using rerun, simulation box is being replaced by dump box ( using box yes, which is default anyway).

I will mail you the respective files for reproducing the error.

Thanks a lot

Karteek Kumar