CHARMM Development Project Forums User Discussion & Questions Molecular Dynamics velocity verlet (VVER) propagation run MD

Hi,

In order to restart from the recent point in case of crash, I am making the chunks of the my propagation run. If I am using LEAP (Leapfrog Verlet) integrator everything is fine. The time I switch to VVER (velocity verlet) integrator, the first chunk which is using the restart file from the equilibration run runs fine. However, the 2nd chunk which uses the restart file from the 1st chunk does not work and script ends there. On the command line I see the following error message.

At line 6678 of file dynio.f
Fortran runtime error: Range error during floating point read

The reason of moving from LEAP to VVER is that I need to go for parallel version of CHARMM and old integrator is not supported. When I try to run the same job using VVER on parallel CHARMM I get the following error message.

At line 6819 of file dynio.f
Fortran runtime error: Range error during floating point read

Except line numbers of file dynio.f, both the errors refer to same issue. Anyone has any idea why this error is appearing and not letting the job run.

Best

LEAPfrog is well supported in parallel; there is certainly no need to switch to VVER for that reason.

There is too little context to comment further.

I am using Constant Temperature (NOSE) (Nose-Hoover method) with VVER.

The code that I am using is

-----
SHAKE BONH

!----------------------Heating Phase of Molecular Dynamics---------------------------!
! T=300K

open write card unit 30 name heat.rst
open write file unit 31 name heat.dcd
dyna strt vver nstep 3000 timestep 0.002 -
iprfrq 3000 ihtfrq 50 ieqfrq 0 ihbfrq 0 inbfrq -1 -
firstt 0.0 finalt 300.0 teminc 10.0 -
iasors 1 iasvel 1 ichecw 0 -
twindh 20.0 twindl -20.0 -
iunrea -1 iunwri 30 iuncrd 31 iunvel -1 kunit -1 -
nprint 500 nsavc 500 nsavv 0

!----------------------Equilibration Phase of Molecular Dynamics---------------------!
! T=300K

open read card unit 30 name heat.rst
open write card unit 31 name equi.rst
open write file unit 32 name equi.dcd
dyna rest vver nstep 10000 time 0.002 -
iprfrq 10000 ihtfrq 0 ieqfrq 100 ihbfrq 0 inbfrq -1 -
firstt 50.0 finalt 300.0 -
iasors 0 iscvel 1 ichecw 1 -
twindh 20 twindl -20 -
iunrea 30 iunwri 31 iuncrd 32 iunvel -1 kunit -1 -
nprint 500 nsavc 500 nsavv 0

!-----------------------Molecular Dynamics for analysis------------------------------!
! NVT ensemble, T=300K


open read card unit 30 name equi.rst
open write card unit 31 name dyna1.rst
open write file unit 32 name dyna1.dcd
dyna rest vver nstep 2500 timestep 0.002 -
iprfrq 2500 ihtfrq 0 ieqfrq 0 ihbfrq 0 inbfrq -1 -
firstt 300.0 finalt 300.0 teminc 0.0 -
iasors 0 iscvel 1 ichecw 0 -
twindh 20.0 twindl -20.0 -
iunrea 30 iunwri 31 iuncrd 32 iunvel -1 kunit -1 -
nprint 500 nsavc 500 nsavv 0 -
NOSE RSTN TREF 300 QREF 1000 NCYC 10
!QREF tempbath coupling should be [10-1000]; Large QREF is approximates microcanonical

CLOSE UNIT 30
CLOSE UNIT 31
CLOSE UNIT 32

open read card unit 30 name dyna1.rst
open write card unit 31 name dyna2.rst
open write file unit 32 name dyna2.dcd
dyna rest vver nstep 2500 timestep 0.002 -
iprfrq 2500 ihtfrq 0 ieqfrq 0 ihbfrq 0 inbfrq -1 -
firstt 300.0 finalt 300.0 teminc 0.0 -
iasors 0 iscvel 1 ichecw 0 -
twindh 20.0 twindl -20.0 -
iunrea 30 iunwri 31 iuncrd 32 iunvel -1 kunit -1 -
nprint 500 nsavc 500 nsavv 0 -
NOSE RSTN TREF 300 QREF 1000 NCYC 10
!QREF tempbath coupling should be [10-1000]; Large QREF is approximates microcanonical

CLOSE UNIT 30
CLOSE UNIT 31
CLOSE UNIT 32

open read card unit 30 name dyna2.rst
open write card unit 31 name dyna3.rst
open write file unit 32 name dyna3.dcd
dyna rest vver nstep 2500 timestep 0.002 -
iprfrq 2500 ihtfrq 0 ieqfrq 0 ihbfrq 0 inbfrq -1 -
firstt 300.0 finalt 300.0 teminc 0.0 -
iasors 0 iscvel 1 ichecw 0 -
twindh 20.0 twindl -20.0 -
iunrea 30 iunwri 31 iuncrd 32 iunvel -1 kunit -1 -
nprint 500 nsavc 500 nsavv 0 -
NOSE RSTN TREF 300 QREF 1000 NCYC 10
!QREF tempbath coupling should be [10-1000]; Large QREF is approximates microcanonical

CLOSE UNIT 30
CLOSE UNIT 31
CLOSE UNIT 32

open read card unit 30 name dyna3.rst
open write card unit 31 name dyna4.rst
open write file unit 32 name dyna4.dcd
dyna rest vver nstep 2500 timestep 0.002 -
iprfrq 2500 ihtfrq 0 ieqfrq 0 ihbfrq 0 inbfrq -1 -
firstt 300.0 finalt 300.0 teminc 0.0 -
iasors 0 iscvel 1 ichecw 0 -
twindh 20.0 twindl -20.0 -
iunrea 30 iunwri 31 iuncrd 32 iunvel -1 kunit -1 -
nprint 500 nsavc 500 nsavv 0 -
NOSE RSTN TREF 300 QREF 1000 NCYC 10
!QREF tempbath coupling should be [10-1000]; Large QREF is approximates microcanonical

CLOSE UNIT 30
CLOSE UNIT 31
CLOSE UNIT 32

!------------------------------------------------------------------------------------!

-----

The error message is

At line 6819 of file dynio.f
Fortran runtime error: Range error during floating point read

Check the output from the end of the previous run; for some reason CHARMM writes out a bad restart file for an energy change tolerance failure.

I highly recommend SHAKE BONH PARAM

I suggest starting the heating at 150 or 200 K; there's no need to start from zero K. You should also set IEQFRQ for the heating run.

These are very, very short simulations.

I took all the suggestions into consideration and still the same error message.

How one can see the the bad restart file format.

To track the cause of problem, I made number of simulations with different conditions and figured out that this error message occurs when I run dynamics either on single or parallel machines using Nose-Hoover dynamics with velocity verlet integrator. VVER alone without Nose-Hoover runs perfectly fine.

The restart file generated at the end of first chunk of propagation phase is not being read by the second chunk of propagation phase. which is is very difficult for me to understand.

Best

I'm not very familiar with VVER/NOSE; I do know it's fairly old code, from a developer I believe is no longer active in the CHARMM community.

You'd have to check the code in source/dynamc for the format details. I think the format may be the same, but with some nonsense data written.

The NOSE-HOOVER PARAMETES entry of restart file with VERL/NOSE (single processor)



The NOSE-HOOVER PARAMETES entry of restart file with VVER/NOSE (single processor)



The difference as anyone can notice is that every entry except line no. 1 in case of VERL/NOSE and line no. 11 in case of VVER/NOSE is zero. Is this some bug in the code which makes the simulation not being able to read the NOSE generated restart to be read properly in the subsequent chunks of the propagation run.

A bug can be a possibility; what CHARMM version is this?

c35b1