Floating point error while running lammps simulation

[magnamad]

Hi ,

I am getting the following floating point error while running a lammps simulation from deepmd kit in my local pc. It is not even integrating one time step. I am facing the following error. Kindly , let me know where the issue lies.Any help would be greatly appreciated.

Thanks ,

Maddy

Generated 0 of 3 mixed pair_coeff terms from geometric mixing rule
Neighbor list info ...
  update: every = 1 steps, delay = 0 steps, check = yes
  max neighbors/atom: 2000, page size: 100000
  master list distance cutoff = 8
  ghost atom cutoff = 8
  binsize = 4, bins = 5 5 5
  1 neighbor lists, perpetual/occasional/extra = 1 0 0
  (1) pair deepmd, perpetual
      attributes: full, newton on
      pair build: full/bin/atomonly
      stencil: full/bin/3d
      bin: standard
Setting up Verlet run ...
  Unit style    : metal
  Current step  : 0
  Time step     : 0.0005
Per MPI rank memory allocation (min/avg/max) = 3.965 | 3.966 | 3.966 Mbytes
   Step          Temp          PotEng         KinEng         TotEng         Press          Volume    
         0   2500          -268.46512      19.712172     -248.75295      16094.353      8000         
/home/madhavan/Softwares/deepmd-kit/bin/lmp: line 11: 652385 Floating point exception(core dumped) /home/madhavan/Softwares/deepmd-kit/bin/_lmp "$@"

Data file for lammps

LAMMPS data file. CGCMM style. atom_style atomic generated by VMD/TopoTools v1.7 on Sat Oct 14 12:37:53 IST 2023
 62 atoms
 0 bonds
 0 angles
 0 dihedrals
 0 impropers
 3 atom types
 0 bond types
 0 angle types
 0 dihedral types
 0 improper types
 0.000000 20.00000  xlo xhi
 0.000000 20.00000  ylo yhi
 0.000000 20.00000  zlo zhi

# Pair Coeffs
#
# 1  Mo
# 2  O
# 3  S

 Masses

 1 2  # A
 2 4  # B
 3  8 # C

 Atoms # atomic

1 1 10.542255 8.453936 12.541114 # A
2 1 7.540054 9.377467 11.202197 # A
3 1 10.355869 11.167134 10.475779 # A
4 2 11.140261 9.915177 11.583379 # B
5 2 10.843721 8.629598 14.146713 # B
6 2 11.295051 7.075884 12.057318 # B
7 2 8.728238 8.380307 12.203994 # B
8 2 6.352739 10.012576 12.143702 # B
.....

Input script for lammps

``

plugin load libdeepmd_lmp.so

variable        NSTEPS          equal 10000000
variable        THERMO_FREQ     equal 1000
variable        DUMP_FREQ       equal 500
variable        TEMP            equal 2500.000000
variable        TAU_T           equal 0.100000

units           metal   # time - ps
boundary        p p p
atom_style      atomic

neighbor        2.0 bin

read_data       data.data
pair_style      deepmd graph-compress.pb
pair_coeff      * *

timestep        0.0005    #  0.5 fs
fix             1 all nvt temp ${TEMP} ${TEMP} ${TAU_T}
velocity        all create ${TEMP} 424242
thermo_style    custom step temp pe ke etotal press vol
thermo          ${THERMO_FREQ}
dump            1 all custom ${DUMP_FREQ} MoS2.lammpstrj id type x y z
dump_modify     1 sort id element A B C
run             ${NSTEPS}

DeepMD version 2.2.5 installed using offline package.

I used the following command to run the lammps simulation
export OMP_NUM_THREADS=2
mpirun -np 12 lmp -in input.lammps

[njzjz]

The bug is fixed in #2923.

[magnamad]

Oh great . Sorry but I am new to this. So , will the bug fixed version be available as an offline package soon (as it is easier to install for me ) , or I should install from source only.

Thanks again ,
Maddy

[njzjz]

The offline package is only provided for the stable version. For the development version (when merged), refer to https://docs.deepmodeling.com/projects/deepmd/en/master/install/easy-install-dev.html.

The error happens only when the number of local atoms in a processor is zero, so a workaround is to reduce the number of processors or increase the number of atoms.

[davy12344]

Hi developer,

Does this link correspond to the version without fault "Floating point error while running lammps simulation"? Can I just follow the description to install directly?
By the way, does this error just occur in the built-in mode lammps?

Best regards;

[njzjz]

It didn't occur in LAMMPS. The latest DeePMD-kit version has fixed it.