[gmx-users] Differences in version 3.0 and 3.3

[chris.neale at utoronto.ca]

Rgardng the Coul 1-4 interactions, are you only changing the  
executable in this comparison,
or are you also (perhaps unintentionally) changing the .itp files? If  
so, could the
difference be related to changes to the forcefield files as mentioned here:
http://www.gromacs.org/pipermail/gmx-users/2006-April/021184.html

Chris.

-- original message --

I was running an NPT molecular dynamics simulation on a system of 6  
protein strands.  I am using Gromacs version 3.3 and for some specific  
purposes, I was only using version 3.0.  Starting from the same input  
structure, forcefield (OPLS) and same run parameters, I noticed that  
the Coul 1-4 interactions are different in the two versions, even in  
the very first step in calculating the energy of the starting  
configuration.  All the other interactions are the same except for the  
Coul 1-4.  The charges are obviously the same, and the same fudgeQQ  
value was used.  Is there an inherent difference in how the two  
versions calculate the Coul 1-4 interactions?

The second difference I noticed involves the pressure coupling.  I was  
doing a semiisotropic Berendsen pressure coupling.  In version 3.3,  
the pressure scaling seems to work properly, as in, the pressure in  
the xy direction and in the z direction averages out to the pressure  
value I had set.  However in version 3.0, the scaling seems to work  
properly in the xy direction, but the pressure in the z direction does  
not go to the desired pressure.  I am using in both cases, a time  
constant of 0.5ps, a pressure of 1 bar, and a compressibility of  
4.5e-5 for both the xy and z directions.  I am interested in the  
surface tension, so if the normal pressure is off, then so is the  
surface tension.  Again is this difference in behavior known or  
expected? Or am I missing something simple and fundamental here?

Any advice or suggestions will be greatly appreciated.

Thank you.

Daniel