I did not download anything from GitHub.
Here is the output of ls from bash
jgough.node4:/scratch1/amber14/AmberTools/src/parmed> ls
chemistry compat24.py compat24.pyc Makefile parmed.Doxyfile parmed.py
ParmedTools setup.sh xparmed.py
jgough.node4:/scratch1/amber14/AmberTools/src/parmed> ls *
compat24.py compat24.pyc Makefile parmed.Doxyfile parmed.py setup.sh
xparmed.py
chemistry:
amber exceptions.py exceptions.pyc __init__.py __init__.pyc
molecule.py periodic_table.py periodic_table.pyc system.py tinker
ParmedTools:
add1264.py argumentlist.pyc changeljpair.pyc coarsegrain.py gui
logos.pyc ParmedActions.pyc parmlist.pyc
addljtype.py arraytools.py changeradii.py coarsegrain.pyc
__init__.py mod_molecules.py parmed_cmd.py simulations
addljtype.pyc arraytools.pyc changeradii.pyc exceptions.py
__init__.pyc mod_molecules.pyc parmed_cmd.pyc
argumentlist.py changeljpair.py checkvalidity.py exceptions.pyc
logos.py ParmedActions.py parmlist.py
jgough.node4:/scratch1/amber14/AmberTools/src/parmed>
I believe that something IS off when the output is parsed by sander. Our
calculations (on the energy) are WAY off. That's how I came across the
errors. The data from one node was correct and the output of the other was
off.
On Fri, Mar 13, 2015 at 3:21 PM, Jason Swails <jason.swails.gmail.com>
wrote:
> On Fri, 2015-03-13 at 14:31 -0400, Jonathan Gough wrote:
> > It's ParmEd.... That is where the issue is...
> >
> > I did "clean installs" on 2 nearly identical machines. One worked, one
> > didn't work. how can I trouble shoot that?
>
> Did you by chance download the Github version of ParmEd on one of those
> machines?
>
> The thing about ParmEd is that it's Python, so there is no chance of
> there being compiler weirdness causing two machines to yield different
> results from the same code. If one passes tests and the other doesn't,
> then the codes are different.
>
> I suspect you replaced ParmEd on one of the machines with the version
> from Github, which gives prmtops that differ according to "diff", but
> not according to "sander".
>
> The ParmEd Github repo has a test/ directory with unit tests that is not
> present in the AmberTools distribution. So if
> $AMBERHOME/AmberTools/src/parmed has a "test" directory, this would
> explain it (and like I mentioned, both versions of ParmEd give
> equivalent topologies, they just look different to dacdif).
>
> HTH,
> Jason
>
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Mar 13 2015 - 12:30:03 PDT
