[gmx-developers] [Fwd: coulomb SR in 3.3cvs]

Michel Cuendet michel.cuendet at epfl.ch
Mon Jul 18 20:13:19 CEST 2005

Hi David,

>Is this from the same tpr? How can the density be different?
This is from the same tpr, but it's an average over a short md run 
(about 12 ps), for each versions. Like always, minor fluctuations are 
expected, right? But the Coulomb energy gap is more than that...

>Did you check this:
Yes I did. Which point are you referring to exactly ?

If it's this one :
Berk Hess  :  17 Aug 2004   :   PME energy accuracy in single precision
It can't be, because I did the tests with a double precision version of 
3.2.1 (single precision wouldn't work at all on itanium). On an x86 cpu, 
the single precision version 3.2.1 gives an even bigger potential energy 
gap w/r to 3.3 (almost double). This additional gap is probably due to 
the above bugfix. Indeed, going to double precision version 3.2.1, still 
on x86, brings back the potential energy gap w/r to 3.3 very close to 
the gap observed on itanium.

Could also be this one :
Erik Lindahl   :  27 Dec 2004   :   New solvent detection
But then something has gone wrong...

Am I the only one outthere who has observed this energy gap ?


Michel Cuendet, Ph.D. student
Laboratory of Computational Biochemistry and Chemistry
Swiss Federal Institute of Technology in Lausanne (EPFL)
CH-1015 Lausanne						
Switzerland                            Phone : +41 21 693 0324
lcbcpc21.epfl.ch/~mitch	                       +41  1 633 4194

More information about the gromacs.org_gmx-developers mailing list