[gmx-users] Re: g_rama/xrama problem in version 3.1.4.

[istvan at kolossvary.hu]

Hi, David, Erik, 

Erik Lindahl wrote: 

> Hi David, 
> 
>> I suspect it's a force field problem, that is you used OPLS, whereas I'm
>> afraid to admit that part of this code expects the topology to have
>> gromos-like dihedrals. If this is the case you coud remake a topology
>> using the ffgmx2 force field and use that for this analysis only.
> 

I did this and it works. I suppose if I change the ACE/NAC atom names for 
the methyl carbon from CH3 to CA in the OPLS-AA .rtp and .hdb files that 
would be OK, too. I'll let you know. 

In the meantime, two quick questions. What is nrexcl in the topology file? 
There is a reference to this in the manual saying that, e.g., benzene should 
have a nrexcl=5 in order to exclude all intramolecular contributions to make 
autocorrelation functions smoother. With my ACE-ALA-NAC nrexcl is set to 3 
by default. What does it mean? 

My other question is -fee in g_energy. It says G = -kT * ln< e^(E/kT)>. What 
exactly does < e^(E/kT)> mean? If I sum up ALL of the energy levels of a 
system that way, it gives the partition function and that is OK. However, if 
I do sampling (Metropolis MC or MD), I cannot guarantee that I visit all of 
the energy levels. All I can say is that I can keep detailed balance and 
sample low-energy states more frequently (exponentially more frequently) 
than high-energy states. Therefore, the exponential character is already 
there in the sampling. My concern is that averaging "exponentially" 
according to < e^(E/kT)> while sampling "exponentially" via MD would somehow 
"double-exponentiate" the average. I am probably missing something. Could 
you clarify this. 

Many thanks, 

  Istvan 


> _______________________________________________
> gmx-users mailing list
> gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the www 
> interface or send it to gmx-users-request at gromacs.org.