[Pw_forum] Contributing to CP & CG bug

Paul Tangney tangney at civet.berkeley.edu
Sat Nov 26 18:54:29 CET 2005


Hi Nicola,

I would be very happy to contribute to the
development of the code....in fact I have a
number of simple features implemented in my own version
of espresso that others might find useful (e.g. the
ability to perform s.c.f calculations in PWSCF in
the presence of an arbitrary potential V(r) ) .

I hate asking for features, when I could just
implement them myself if I was a member of the
"developers club". There are actually quite a few people
here in Berkeley that are using espresso (I persuaded
many people to try it, and most people are sold on it once
they've tried it), are developing methodology, and
who could give back to the Espresso community if there
was a means of doing so.

Espresso is definitely the best written and most trustworthy
code that I have encountered (and I think I've tried everything
that is freely available) so an obvious danger with extending
the list of developers is that it turns into a sprawling 
incomprehensible mess....like another freely available
multi-featured code that I tried.

----------------

What bug did you find in CG ?
I have been trying to use CG for a watery system with an
organic solute on an Opteron cluster with Myrinet.
It is *terribly* slow it takes about half an hour for one
time step on 16 processors.
This is a system with 205 atoms (550 or so electrons).
I haven't had a chance to delve into the code to figure
out the reason yet, but it might just be that I'm using crap
libraries or am using crap parameters. If the bug could
explain it, though, I'd appreciate the fix. On the other
hand, if you have parameters that you have found to work well
for water, that would be useful too as I could eliminate
my parameters as a source of the problem.

Last question : how are wavefunctions extrapolated between
time steps in the CG implementation ? Is there a way to
specify the order of extrapolation ?

Sorry for the long email.

Thanks,

Paul




--
ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
Dr. Paul Tangney
Theory of Nanostructured Materials Facility
The Molecular Foundry
Lawrence Berkeley National Lab.         E-mail: PTTangney at lbl.gov
1 Cyclotron Road, Bldg 66                  Phone: (510) 642-2635
Berkeley, CA 94720                         Fax :  (510) 643-9345
ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo




More information about the Pw_forum mailing list