[MMTK] problems with molecular dynamics

Konrad Hinsen hinsen@cnrs-orleans.fr
Fri, 8 Nov 2002 18:40:45 +0100

> Here is the gdb output (Same sigsegv for Steep & Conjugate):

Not much help, unfortunately, I'd need at least line numbers or function
names within the module MMTK_forcefield.

> Now the weird this is I did try this code:
> from MMTK.Minimization import SteepestDescentMinimizer
> from MMTK import *
> from MMTK.ForceFields import Amber94ForceField
> from MMTK.Proteins import Protein
> universe = InfiniteUniverse(Amber94ForceField())
> universe.protein = Protein('insulin')
> minimizer = SteepestDescentMinimizer(universe)
> minimizer(steps = 100)
> And it worked.

There is one difference: this version has a well-defined energy, your
previous one didn't (Amber94 cannot handle protein models without
hydrogens). In fact, when I run your previous script on my machine, it
doesn't crash, but it doesn't return a useful state either. And a
simple energy evaluation for your system contains NaN in a few places.

Therefore I wonder if those NaN-reducing operations could cause a
crash on your machine due to compiler optimizations or whatever.
Before looking at any details, could you just try to use a model with
hydrogens? You can construct positions for them using

> Can you tell me what versions of all things relevent to MMTK you're using
> Conrad? Cause I think versions are important, cause that solved my
> previous core dump.

Python 2.2.1
Numerical Python 21.3
Scientific Python 2.3.4
NetCDF 3.5.0
MMTK 2.2

Not the latest versions, I know, but I don't do updates so often on my
home machine.

Konrad Hinsen                            | E-Mail: hinsen@cnrs-orleans.fr
Centre de Biophysique Moleculaire (CNRS) | Tel.: +33-
Rue Charles Sadron                       | Fax:  +33-
45071 Orleans Cedex 2                    | Deutsch/Esperanto/English/
France                                   | Nederlands/Francais