[gmx-developers] Wrong Ouput from g_density - PBC bug?

[Juan M. Vanegas]

Dear gmx-developers,

There is a problem with the output from g_density (4.5.5). If I take a 
200 ns trajectory of a membrane simulation in water I get different 
results depending on how much of the trajectory I analyze. If I use the 
whole 200 ns, I get very different results than if I use only 100 ns or 
a 50 ns section. The main difference is that the water density is lowest 
in the 200 ns analysis and highest in the 50 ns. It seems that the 
problem has to do with molecules that diffuse out of the box, which 
explains why you only see a difference in the water density and not on 
the membrane. If I comment out the following line in 
src/tools/gmx_density.c (line 281):

gmx_rmpbc(gpbc,natoms,box,x0);

Then the problem goes away and the density of water is the same no 
matter how much of the trajectory I analyze. Cheers,

-- 
Juan M. Vanegas
Postdoctoral Researcher
Laboratori de Càlcul Numèric (LaCàN)
Departament de Matemàtica Aplicada III
Universitat Politècnica de Catalunya
Parc K2M-004, Jordi Girona 1-3
Barcelona 08034, Spain