LAMMPs with colvars Segmentation Fault when using gyration colvar.

Hi LAMMPs users,

I’ve been using LAMMPs for a short while, and recently tried to run an instance with the colvars package and specifically the gyration colvar. Whenever I try running either a harmonic instance or a metadynamics instance with a gyration colvar as a parameter, it gives a segmentation fault. (I will provide the debug.txt as an attachment.) I am running Mac OS X 10.10.5, and I have tried this on two different machines, one with Openmpi 1.10, and another with Openmpi 2.0. I’m using the most up to date version of the colvars package and have tried with two different builds of LAMMPs, July 30, 2016 and May 14, 2016.

To get the error, you can modify the peptide.colvars file in the examples provided by the LAMMPs install. I’ve defined the gyration colvar as such:

colvar {
name three
lowerBoundary 0
upperBoundary 10
gyration {
atoms {
atomNumbers 82 83 84
}
}
}

Then add the colvar “three” to the metadynamics instance.

The same error occurs when trying to define an inertia or inertiaZ colvar.

Thank you,

  • Ross

debug.txt (13.4 KB)

Hi LAMMPs users,

I've been using LAMMPs for a short while, and recently tried to run an
instance with the colvars package and specifically the gyration colvar.
Whenever I try running either a harmonic instance or a metadynamics instance
with a gyration colvar as a parameter, it gives a segmentation fault. (I
will provide the debug.txt as an attachment.) I am running Mac OS X 10.10.5,
and I have tried this on two different machines, one with Openmpi 1.10, and
another with Openmpi 2.0. I'm using the most up to date version of the
colvars package and have tried with two different builds of LAMMPs, July 30,
2016 and May 14, 2016.

steve just released a patch today with the latest colvars updates and bugfixes.
please give that a try.

thanks,
      axel.

Thank you for reporting the seg fault. Unfortunately, I couldn’t reproduce it.

Your execution stack indicates that the error occurred in the gradients function.

Unless the error is fixed, would you be able to get the line position where the segfault occurred?

Thanks

Giacomo