Error in colvars module in parsing accumulatedWork when it is -nan

LAMMPS LAMMPS Mailing List Mirror
1

Hi,

I recently added colvarsRestartFrequency 100 to my colvars configuration and I got this error whenever my accumulatedWork is -nan in the colvars.state restart file.

colvars: Error: in parsing "accumulatedWork”.
ERROR on proc 0: Fatal error in the collective variables module. (../colvarproxy_lammps.cpp:315)

In my colvars.state restart file:

restraint {
  configuration {
    step 2000000
    name harmonic1
    centers 3.00000000000000e+00
    centers_raw 3.00000000000000e+00
    accumulatedWork -nan
  }
}

Without the colvarsRestartFrequency tag, I didn’t seem to get this error.

I was able to solve this problem by doing reading the problematic restart file, omit fix colvars and write_restart. So, I’m posting this question to see if there is a better way to solve this problem?

Best,
Tonnam Balankura
Penn State University

2

Hi,

I recently added colvarsRestartFrequency 100 to my colvars configuration
and I got this error whenever my accumulatedWork is -nan in the
colvars.state restart file.

colvars: Error: in parsing "accumulatedWork”.
ERROR on proc 0: Fatal error in the collective variables module.
(../colvarproxy_lammps.cpp:315)

In my colvars.state restart file:

restraint {
  configuration {
    step 2000000
    name harmonic1
    centers 3.00000000000000e+00
    centers_raw 3.00000000000000e+00
    accumulatedWork -nan
  }
}

Without the colvarsRestartFrequency tag, I didn’t seem to get this error.

I was able to solve this problem by doing reading the problematic restart
file, omit fix colvars and write_restart. So, I’m posting this question to
see if there is a better way to solve this problem?

​the real problem is, that there *should* not be a -NaN.

you will have to figure out, why this happens; i.e. whether this ​is a
(fixed?) bug in colvars or an indication of a problem with your simulation.

axel.

3

I’m not aware of a bug in the harmonic restraint, can you look at the values that the collective variable takes over time?

4

Thanks Axel and Giacomo for your replies. Looking at my lammpstrj file, the simulation seem to be fine.

My out.colvars.traj is:

step ZPULL x0_ZPULL W_harmonic1

1000000 3.86582832134200e+00 3.00000000000000e+00 0.00000000000000e+00
1000100 3.89272402639424e+00 3.00000000000000e+00 -nan
1000200 4.02273383910695e+00 3.00000000000000e+00 -nan
1000300 3.90598924756570e+00 3.00000000000000e+00 -nan
1000400 3.81480499193611e+00 3.00000000000000e+00 -nan
1000500 3.75193022228514e+00 3.00000000000000e+00 -nan

In other windows without such problem, my out.colvars.traj is typically:

step ZPULL x0_ZPULL W_harmonic1

1028000 1.54967464283919e+01 1.57600000000000e+01 0.00000000000000e+00
1028100 1.55348447711368e+01 1.57600000000000e+01 0.00000000000000e+00
1028200 1.56061368474879e+01 1.57600000000000e+01 0.00000000000000e+00
1028300 1.57186100264062e+01 1.57600000000000e+01 0.00000000000000e+00
1028400 1.57976618800188e+01 1.57600000000000e+01 0.00000000000000e+00
1028500 1.59025738774480e+01 1.57600000000000e+01 0.00000000000000e+00

My colvars configuration is:

colvarsTrajFrequency 100
colvarsRestartFrequency 100

colvar {
name ZPULL
width 1.0
UpperBoundary 46.0
LowerBoundary 0.0
distanceZ {
main { atomNumbersRange 5017-5106 }
ref {
atomNumbersRange 1-418
}
axis (0.0,0.0,-1.0)
}
}

harmonic {
colvars ZPULL
forceConstant 1.7
centers 15.76
targetCenters 15.76
targetNumSteps 1000000
outputCenters yes
outputAccumulatedWork yes
}

Best,
Tonnam

5

I think it might be an error on my computation node? After I restart it, the -nan have not shown up yet. The cluster I’m using is relatively new and may not have been properly setup yet.

Best,
Tonnam

6

If the error cannot be reproduced, I would not worry about it. If it surfaces again, try increasing the thermo frequency and colvarsTrajFrequency to get an idea of when things do go wrong.

Giacomo