Hi,
i’m running a calculation using ReaxFF. I initially want to minimize the structure and then do
an NVT run.
The NVT run works; the minimization run does not: it runs one single step and then gives
the following error message:
miguel,
---
_pmii_daemon(SIGCHLD): PE 1 exit signal Floating point exception
[NID 15833]Apid 283071: initiated application termination
Application 283071 exit codes: 136
---
No other info. is given, so it's difficult to find out what is happening.
you could compile with debug info and run the input from within a debugger.
or alternatively set up the signal handling to cause a core dump on floating
point exceptions and then inspect the coredump in a debugger.
However, in the "bugs fixes"
tabs of lammps' web-site, i notice there's one entry on April 3 2009
reporting a fix for the energy
minimizer. I have a Feb 2009 version of lammps so i wonder whether this
April 3 bug could
be causing the minimization to crash. Can this bug produce the message
above?
floating point exceptions can be caused by many things. e.g. incorrect inputs,
unphysical starting coordinates.
the easiest way to prove if the issues are related would be to compile the
updated version with all patches included and see if that makes a difference.
for difficult to minimize cases, a better step to begin with would be to run
MD with annealing and use fix nve/limit.
cheers,
axel.
Don't know. Please try the latest version and see.
Steve