Issues with Restart Command

Dear LAMMPS Users,

Hello All.

I am currently experiencing an issue with the restart command.

I am performing a uniaxial elongation of a simulation box that contains polymer chains.

The simulation box are being deformed via fix deform command where I specified an equation that describes the rate and displacement of the box length.

After carefully reading the run command in the documentation, I am still obtaining an error using the restart command. I used the start/stop keywords since my fix deform command changes over time. Therefore, I always output restart files after every segmented runs.

When I restart my simulation using one of the outputted restart files, I am obtaining the exact thermodynamic quantities at the start of the restart simulation.

However, after one time interval, I obtain incorrect instantaneous simulation box length and other thermodynamic parameters compared to the parameters that were to be obtained had the entire simulation was completed.

If I further increased the number of chains and atoms inside the simulation box, I would obtain FENE bond too long.

Moreover, I am using the same number of processors in both of my restart and in my original input file.

I was curious if anyone had experienced this issue.

I have displayed a portion of my original script:

units lj
atom_style full
neighbor 0.36 bin
neigh_modify delay 2
pair_style lj/cut 2.5
bond_style hybrid harmonic fene
special_bonds lj 0.0 1.0 1.0
read_data FJC_Single_HET

pair_coeff * * 1.0 1.0 2.5
bond_coeff 1*2 fene 30.0 1.5 1.0 1.0
bond_coeff 3 harmonic 1000.0 1.20
pair_modify shift yes
timestep 0.010
reset_timestep 0

fix 1 all langevin 1.0 1.0 2.0 542305
fix 2 all press/berendsen iso 0.0 0.0 10.0
fix 3 all nve

fix 4 all deform 1 x variable v_displace v_rate y volume z volume remap v

fix 5 all ave/time 4 25 100 c_p[1] c_p[2] c_p[3] c_p[4] c_p[5] c_p[6] file press.1

dump 1 all custom 100000 *_Monomer_Coordinates id type x y z ix iy iz

restart 1000000 *_Strain_Rate.restart

run 1000000 start 0 stop 10000000
run 1000000 start 0 stop 10000000
run 1000000 start 0 stop 10000000
run 1000000 start 0 stop 10000000
run 1000000 start 0 stop 10000000
run 1000000 start 0 stop 10000000
run 1000000 start 0 stop 10000000
run 1000000 start 0 stop 10000000
run 1000000 start 0 stop 10000000
run 1000000 start 0 stop 10000000

And a portion of my restart script if I were to use 3000000_Strain_Rate.restart

processors 2 2 1
units lj
atom_style full
read_restart 3000000_Strain_Rate.restart

neighbor 0.36 bin
neigh_modify delay 2
pair_style lj/cut 2.5
bond_style hybrid harmonic fene
special_bonds lj 0.0 1.0 1.0
pair_coeff * * 1.0 1.0 2.5
bond_coeff 1*2 fene 30.0 1.5 1.0 1.0
bond_coeff 3 harmonic 1000.0 1.20
pair_modify shift yes

timestep 0.010

fix 1 all langevin 1.0 1.0 2.0 542305
fix 3 all nve
thermo 1000
thermo_style custom step temp pe etotal epair pxx pyy pzz lx ly lz

fix 4 all deform 1 x variable v_displace v_rate y volume z volume remap v
compute p all pressure thermo_temp
fix 5 all ave/time 4 25 100 c_p[1] c_p[2] c_p[3] c_p[4] c_p[5] c_p[6] file press.1

dump 1 all custom 100000 *_Monomer_Coordinates id type x y z ix iy iz

restart 1000000 *_Strain_Rate.restart

run 1000000 start 0 stop 10000000
run 1000000 start 0 stop 10000000
run 1000000 start 0 stop 10000000
run 1000000 start 0 stop 10000000
run 1000000 start 0 stop 10000000
run 1000000 start 0 stop 10000000

I would greatly appreciate if I could request for any suggestions. Thank you.

Sincerely,

Masato Koizumi

Dear LAMMPS Users,

Hello All.

I am currently experiencing an issue with the restart command.

no, i don't think so. it looks to me, that you are using restart
correctly, but that the two inputs for the original and the restarted
runs are inconsistent. in fact, the first run is bogus, as you seem to
have two fixes active that modify the same simulation cell parameters
and that is a *major* mistake. doesn't LAMMPS print a warning about
it?

axel.

Dear Dr. Axel Kohlmeyer,

Hello Dr. Kohlmeyer. Thank you for your e-mail.

Actually, the portion of my original input file that specifies the fix commands is:

fix 1 all langevin 1.0 1.0 2.0 542305
fix 2 all press/berendsen iso 0.0 0.0 10.0
fix 3 all nve
thermo 1000
thermo_style custom step temp pe etotal epair pxx pyy pzz lx ly lz
run 100000
unfix 2

fix 4 all deform 1 x variable v_displace v_rate y volume z volume remap v

Therefore, press/berendsen is unfixed prior to the start of uniaxial deformation such that I am performing the stress strain simulation under NVT ensemble.

Once again, thank you so much for your time and effort.

Sincerely,

Masato Koizumi