ERROR: Temperature compute degrees of freedom < 0

Hi,

I am using granular package in conjunction with the fix rigid for simulation of non-spherical particles. I am simulating simple shear flow with the following (simplified) input script (initial data file attached)

atom_style sphere
boundary p p p
newton off
comm_modify mode single vel yes

fix addMolecule all property/atom mol ghost yes
read_data init.dat fix addMolecule NULL Molecules #Initial particle cluster attached

set atom * mass 0.5
set atom * diameter 1.0

pair_style gran/hooke/history 10000 NULL 100 NULL 0.5 1
pair_coeff * *

timestep 0.001

fix 1 all rigid molecule

change_box all triclinic

compute biasTemp all temp/profile 1 0 0 y 10
compute sphTemp all temp/sphere bias biasTemp
thermo_style custom c_sphTemp

fix shearRun all deform 1 xy erate 0.05 remap v

run 100

the script works fine with older version of lammps (9Dec2014) and results seem reasonable, however I recently updated to 7Dec2015 to rerun some simulations and I get the following error:

ERROR: Temperature compute degrees of freedom < 0 (…/compute_temp_profile.cpp:248)

The problem seems to stem from a new line testing for (dof<0) and there seems to be slight changes to the way dof_compute() works (additional term d*Nbins subtracted)

I am just wondering if the developer can shed some lights on this issue.

Many thanks

sina

init.dat (313 KB)

Hi,

I am using granular package in conjunction with the fix rigid for simulation
of non-spherical particles. I am simulating simple shear flow with the
following (simplified) input script (initial data file attached)

  atom_style sphere
  boundary p p p
  newton off
  comm_modify mode single vel yes

  fix addMolecule all property/atom mol ghost yes
  read_data init.dat fix addMolecule NULL Molecules #Initial particle
cluster attached

  set atom * mass 0.5
  set atom * diameter 1.0

  pair_style gran/hooke/history 10000 NULL 100 NULL 0.5 1
  pair_coeff * *

  timestep 0.001

  fix 1 all rigid molecule

  change_box all triclinic

  compute biasTemp all temp/profile 1 0 0 y 10
  compute sphTemp all temp/sphere bias biasTemp
  thermo_style custom c_sphTemp

  fix shearRun all deform 1 xy erate 0.05 remap v

  run 100

the script works fine with older version of lammps (9Dec2014) and results
seem reasonable, however I recently updated to 7Dec2015 to rerun some
simulations and I get the following error:

ERROR: Temperature compute degrees of freedom < 0
(../compute_temp_profile.cpp:248)

The problem seems to stem from a new line testing for (dof<0) and there
seems to be slight changes to the way dof_compute() works (additional term
d*Nbins subtracted)

I am just wondering if the developer can shed some lights on this issue.

compute temp/profile doesn't account for the additional degrees of
freedom of extended particles. the error exit is likely the
consequence of improved handling of degrees of freedom elsewhere.

axel.