[lammps-users] read_restart Error: charge/regulation fix

LAMMPS LAMMPS Mailing List Mirror
1

Hi all,

I am using LAMMPS version 29 September 2021 to simulate multiple coarse-grained polymer chains in a cubic box with PBCs in constant NVE + Langevin thermostat. The atom style is full and forces are computed using lj/cut/coulomb/debye and harmonic springs for covalent bonds.

In addition to this, I am using charge/regulation fixes for acid/base protonation/deprotonation.

The simnulations are running nicely except when I am restarting them using the read_restart command. I have tried using simple restart and mpiio options and both of them are giving different errors; the input script is attached. I have checked the remap option while read_restart but couldn’t get any help.

With MPIIO:

Reading restart file …

restart file = 29 Sep 2021, LAMMPS = 29 Sep 2021

restoring atom style full from restart

orthogonal box = (-250.00000 -250.00000 -250.00000) to (250.00000 250.00000 250.00000)

2 by 3 by 4 MPI processor grid

restoring pair style lj/cut/coul/debye from restart

restoring bond style harmonic from restart

ERROR on proc 0: Unexpected end of file while reading file ‘rst.mpiio.5000000’ (src/modify.cpp: 1540)

Without MPIIO:

Reading restart file …

restart file = 29 Sep 2021, LAMMPS = 29 Sep 2021

restoring atom style full from restart

orthogonal box = (-250.00000 -250.00000 -250.00000) to (250.00000 250.00000 250.00000)

2 by 3 by 4 MPI processor grid

restoring pair style lj/cut/coul/debye from restart

restoring bond style harmonic from restart

libc++abi: terminating with uncaught exception of type std::bad_alloc: std::bad_alloc

libc++abi: terminating with uncaught exception of type std::bad_alloc: std::bad_alloc

[] *** Process received signal ***

[] Signal: Aborted (6)

[] Signal code: (-6)

[] [ 0] libc++abi: terminating with uncaught exception of type std::bad_alloc: std::bad_alloc

The command for running simulations is given below:

mpirun -np $NCPUS lmp_mpi -log my.log -in in.poly

Here, $NCPUS is supplied from the submission script. Each MPI process is using 1 OpenMP thread.

Without charge/regulation without and with MPIIO, simulations are restarting without any error.

Any help is appreciated.

Sincerely,

Piyush

in.poly (1.71 KB)

2

Thanks for your report.

From looking at the source code of fix charge/regulation, it appears that it does not support restarting, but had by accident a flag for triggering a restart set.
However, only changing that does not resolve problems restarting runs, so some time debugging effort is needed.
I will report here if a solution was found.

3

please try recompiling with the following patch added.
I recommend against using the mpiio restart format, too.
The change should apply to both variants and resolve either issue, though.

commit f7d049ac2dee42dcd077fe07fec1d0e21e8cc733 (HEAD -> maintenance-2021-09-29, akohlmey/maintenance-2021-09-29)
gpg: Signature made Wed 24 Nov 2021 03:36:16 PM EST
gpg: using RSA key EEA103764C6C633EDC8AC428D9B44E93BF0C375A
gpg: Good signature from "Axel Kohlmeyer <[[email protected]](mailto:[email protected])>" [ultimate]
Author: Axel Kohlmeyer <[[email protected]](mailto:[email protected])>

generate atom tags for newly created atoms, if tags are enabled. triclinic support.

diff --git a/src/MC/fix_charge_regulation.cpp b/src/MC/fix_charge_regulation.cpp
index 2497b3d976..626a2e389d 100644
--- a/src/MC/fix_charge_regulation.cpp
+++ b/src/MC/fix_charge_regulation.cpp
@@ -75,7 +75,7 @@ FixChargeRegulation::FixChargeRegulation(LAMMPS *lmp, int narg, char **arg) :
size_vector = 8;
global_freq = 1;
extvector = 0;
- restart_global = 1;
+ restart_global = 0;
time_depend = 1;
cr_nmax = 0;
overlap_flag = 0;
@@ -985,9 +985,17 @@ int FixChargeRegulation::insert_particle(int ptype, double charge, double rd, do
modify->create_attribute(m);

}
+ atom->natoms++;
atom->nghost = 0;
+ if (atom->tag_enable) {
+ if (atom->tag_enable) {
+ atom->tag_extend();
+ if (atom->map_style != Atom::MAP_NONE) atom->map_init();
+ }
+ }
+ if (triclinic) domain->x2lamda(atom->nlocal);
comm->borders();
- atom->natoms++;
+ if (triclinic) domain->lamda2x(atom->nlocal+atom->nghost);
return m;
}
4

Hi Axel,

Thank you so much for the quick response and modifying the source code. I have checked this new patch. Unfortunately, I am still encountering error shown below:

read_restart rst.mpiio.* remap
Reading restart file …
ERROR: Incomplete or corrupted LAMMPS restart file (src/read_restart.cpp:1160)
Last command: read_restart rst.mpiio.* remap

Best regards,
Piyush

5

don’t use mpiio.

6

also, you cannot read only restarts.
both, reading and writing restarts had problems with fix charge/regulation

7

that was supposed to old restarts.

8

Hi Axel,

It is working now without mpiio (“old restarts”), despite a warning shown below.