[lammps-users] Serius problem in combining compute reduce/region and compute stress/atom in current version


I have got a serius problem when I combine compute reduce/region and compute stress/atom using the latest version. The error massage is just “Segmentation fault.”

The following is the script I used, a modification of “in. melt”.

3d Lennard-Jones melt

units lj
atom_style atomic

lattice fcc 0.8442
region box block 0 50 0 50 0 10
create_box 1 box
create_atoms 1 box
mass 1 1.0

velocity all create 3.0 87287

pair_style lj/cut 2.5
pair_coeff 1 1 1.0 1.0 2.5

neighbor 0.3 bin
neigh_modify every 20 delay 0 check no

region ct block 10 25 10 25 2 5 units box
compute stress all stress/atom virial
compute str_ct all reduce/region ct sum c_stress[1] c_stress[2] c_stress[3]
variable P_o equal -(c_str_ct[1]+c_str_ct[2]+c_str_ct[3])

fix 1 all nve

dump id all atom 100 dump.melt

thermo 50
thermo_style custom step temp press v_P_o
run 100

  1. When I use compute reduce instead of reduce/region, it works fine.
  2. It worked with older version of Lammps e.g. April, 2010.
  3. It always happen whether I compile serial version(lmp_serial) or parallel version(lmp_linux).

Please check this whether this also happens in your machine.

Thanks for your attention.

Sangrak Kim

This was a bug. There was some left-over code
in compute reduce/region from when it and compute reduce
used to process global values, which they no longer do.

I’ll post a patch later today.


2011/1/2 김상락 <srkim@…581…>

Hi Steve
I also encountered this problem when I use “compute reduce” command. Can you please also post a patch for it? Thanks.


2011/1/3 Steve Plimpton <sjplimp@…43…4…>

It’s the 3Jan11 patch.


2011/1/3 yang lu <luyang0309@…24…>