Segmentation fault (core dumped) when reading data file

Dear LAMMPS developers and users,

I’m trying to build my system using multiple files read option in read_data command (LAMMPS, stable 12 Aug 2018). Firstly, system consisting of calcite and aqueous solution run with NVT ensemble and then another aqueous solution is added to the system;

package gpu 1 neigh yes newton off
suffix gpu

units real
boundary p p p
dimension 3

atom_style full
bond_style harmonic
angle_style harmonic
improper_style cvff
pair_style lj/cut/coul/long 13.0 14.0

kspace_style pppm 1.0e-4
read_data F.data

pair_coeff 3 3 0 0 #H,H
pair_coeff 3 6 0 1.594 #H,O
pair_coeff 3 11 0.000006968 4.497 #H,aO3
pair_coeff 6 9 0.116079 3.47 #aC2,O
pair_coeff 6 10 0.269896 2.76 #aCa1,O
pair_coeff 6 11 0.14603 3.12 #O,O3
pair_coeff 9 9 2.65 2.62 #Ca2,aC2
pair_coeff 9 11 6.266 1.196 #aC2,aO3
pair_coeff 10 11 0.2388 2.805 3.435 #Ca1,aO3
pair_coeff 11 11 0.00027276 4.744 #aO3,aO3

pair_modify mix arithmetic

neighbor 2 bin
neigh_modify every 1 delay 0 check yes

group Calcite type 9 10 11
group water type 3 6
group sulphate type 7 8

fix 1 Calcite setforce 0.0 0.0 0.0
minimize 1.0e-4 1.0e-6 10000 100000

velocity all create 348.0 12345678 dist uniform
velocity Calcite set 0 0 0 units box

fix 2 water shake 1e-4 500 0 b 1 a 1
fix 3 all nvt temp 348.0 348.0 100.0

thermo 10

thermo_style custom step temp press &
pe ke etotal enthalpy

timestep 1

dump 1 all xyz 10 dump1.xyz
dump_modify 1 sort id
reset_timestep 0
run 500

read_data I.data add append shift 0.0 0.0 60.0

unfix 3
fix 4 all nvt temp 348.0 348.0 100.0
dump 2 all xyz 10 dump2.xyz
dump_modify 2 sort id

run 500

when reading the second data file, I received the following error:

Reading data file …
orthogonal box = (-40.2788 -42.8653 -38.1737) to (41.4712 42.5147 110.506)
1 by 1 by 1 MPI processor grid
reading atoms …
6190 atoms
[System2:16456] *** Process received signal ***
[System2:16456] Signal: Segmentation fault (11)
[System2:16456] Signal code: Address not mapped (1)
[System2:16456] Failing at address: 0x561fe7248db0
[System2:16456] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x12890)[0x7f1573f1a890]
[System2:16456] [ 1] lmp_ss(+0x1c26bb)[0x5620c31a76bb]
[System2:16456] [ 2] lmp_ss(+0x964562)[0x5620c3949562]
[System2:16456] [ 3] lmp_ss(+0x969ee0)[0x5620c394eee0]
[System2:16456] [ 4] lmp_ss(+0x17531e)[0x5620c315a31e]
[System2:16456] [ 5] lmp_ss(+0x172821)[0x5620c3157821]
[System2:16456] [ 6] lmp_ss(+0x1732fa)[0x5620c31582fa]
[System2:16456] [ 7] lmp_ss(+0xbe368)[0x5620c30a3368]
[System2:16456] [ 8] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xe7)[0x7f1573b38b97]
[System2:16456] [ 9] lmp_ss(+0xc946a)[0x5620c30ae46a]
[System2:16456] *** End of error message ***

Segmentation fault (core dumped)

Due to limitation for size of attachment, I cannot send associating input and data files, however it will be OK to send anyone if want. It should be mentioned that the type of atoms, bond and angles are the same in both data files.

I would appreciate if anyone help me with this.

Thank you in advance

what happens if
- you do not enable the GPU package, i.e. run on the CPU only?
- you only try to load the second data file?
- you use the latest stable version (12 Dec 2018)?

axel.

Dear Axel, thanks for your comments. I follow two first comments with the LAMMPS, stable 12 Aug 2018 version.

  • I run the input file on CPU, for a few steps (2000). The job was successful and the system works well after adding the second data file. However when I try the same on GPU, Segmentation fault (core dumped) error was received. Due to large size of my system, rate of simulation on CPU is very slow and I prefer to run on GPU.

  • I try to load only the second data file on GPU. The system works well without any error in a few first steps (1000).

I cannot understand what is the problem with running on GPU.

what happens if

  • you do not enable the GPU package, i.e. run on the CPU only?
  • you only try to load the second data file?
  • you use the latest stable version (12 Dec 2018)?

axel.