bug with newton off?

Hi,

For atomstyle template (polymer) with Fene bonds I get different outcomes for newton on and off. It crashes for newton off. Possible bug?

Thanks and regards,

Hi,

For atomstyle template (polymer) with Fene bonds I get different outcomes
for newton on and off. It crashes for newton off. Possible bug?

Crash output? LAMMPS version? OS? Compiler and settings? example script?

what about #3 and #9 of http://lammps.sandia.gov/guidelines.html ?

Apologies for witholding details.

I used LAMMPS: Jul 30, 2016 version (serial) on Ubuntu compiled with debug setting (-g -O). I am also attaching all required files for run herewith (tar.gz file).

To summarize the issue,

With “newton off” it crashed with following error where as with “newton on” it didn’t

tail -5 newton_off:
WARNING: FENE bond too long: 702 164 165 1.54581 (…/bond_fene.cpp:91)
WARNING: FENE bond too long: 703 61 62 2.24083 (…/bond_fene.cpp:91)
WARNING: FENE bond too long: 703 106 107 2.15495 (…/bond_fene.cpp:91)
WARNING: FENE bond too long: 703 108 109 3.00592 (…/bond_fene.cpp:91)
ERROR on proc 0: Bad FENE bond (…/bond_fene.cpp:92)

Input script used:

#temperature in reduced units

variable temp equal 3.0

dimension 2

units lj

molecule poly1 poly1.molecule

atom_style template poly1

boundary p p p

#neighbor settings

neighbor 2.0 bin

neigh_modify every 1 delay 0 check yes

newton off

#add additional particles to box

region box block -32 32 -32 32 -2.0 2.0

region myCylinder cylinder z 0 0 30 -2.0 2.0

create_box 1 box bond/types 1 angle/types 1 extra/bond/per/atom 2 extra/angle/per/atom 1 extra/special/per/atom 6

read_data poly.input add merge

mass * 1

group mobile type 1

pair_style lj/cut 1.12246204830937

pair_coeff * * 1.0 1.0 1.12246204830937

pair_modify shift yes

#FENE type bond

bond_style fene

bond_coeff 1 30.0 1.5 1.0 1.0

#bond_style harmonic

#bond_coeff 1 40.0 0.96

special_bonds fene

angle_style harmonic

angle_coeff 1 100 180

#remove overlap between polymer and obstacles

minimize 1e-10 1e-10 100 100

reset_timestep 0

Thanks so much,
Abhijeet

fene_newton_crash.tar.gz (66 KB)

Perhaps it could be generic issue with template style.

If I understand correctly, for newton off case number of bonds associated with each polymer bead (except the end ones) should be equal to 2. However, I notice it to be registered as 1.

Update.

I think I found out what’s going on.

I set newton command before command molecule and it worked without issues.

Best,
Abhijeet