[gmx-users] PME error in gromacs-3.3.1

[Mark Abraham]

Yuhui Li wrote:
> Hi, all
> 
> I installed gromacs-3.3.1 recently and tested a system consisting only 
> thress water molecules. When I use coulombtype = Cut-off, it works well, 
> but if I use coulombtype = PME, there was a segmentation fault. I have 
> checked the maillist archives and found that some people had the same 
> problem in gromacs-3.3. From their instruction, the new pme.c file was 
> tried and   pme_order = 4 was set, but the problem is still there. Do 
> you have idea about this problem?

Actually, if you mean only "three" water molecules, the problem is 
likely to be that the fftw grid interpolation will fail if there is not 
at least one something per something (can't remember, check the archives 
if you care). Try doing a system of reasonable size.

Mark