[Pw_forum] Fwd: BOMD / FPMD / CP

Paolo Giannozzi giannozz at nest.sns.it
Fri Oct 28 20:21:48 CEST 2005


Paul Tangney <tangney at civet.berkeley.edu> asked me to post 
the following message:

----------  Forwarded Message  ----------

Subject: BOMD / FPMD / CP
Date: Friday 28 October 2005 19:06
From: Paul Tangney <tangney at civet.berkeley.edu>
To: giannozz at nest.sns.it

Hello,

I am using Espresso on an Opteron cluster
with LAM/MPI and Portland compilers.
My system is an Indole molecule surrounded
by 63 waters.


I have a few questions:

(1) Is FPMD still being maintained ?
Some of the features that used to be
in FPMD a few years ago are no longer working
and I had to delve into the code to 'reactivate'
them. e.g. velocity rescaling.
Also, some of the default parameters are different
from what the documentation says they are.

(2) I'm doing Born-Oppenheimer dynamics
with DIIS. It works great for a while (although
I'm sure the parameters I'm using are far from optimal)
and then, after a few hundred steps, the performance
drops by a factor of about 100000. The simulation basically stops.
Has anybody else experienced this ?
Is there a memory leak ?

(3) Does anybody know which code (CP/FPMD) is faster for
Born-Oppenheimer MD using norm-conserving PPs and which
minimization method works best for water ?  Is any effort
being put into making BOMD fast in these codes or is the
focus only on CPMD ?

Thanks for your help.


Paul

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



More information about the Pw_forum mailing list