[gmx-users] thermodynamic integration yields different values in gromacs 3.3.1 and later versions

[Justin A. Lemkul]

Schneck at uni-heidelberg.de wrote:
> Dear Gromacs users,
> 
> I am doing thermodynamic integration in order to compute the change in free
> energy upon increasing the electric charges of a water molecule from zero
> (lambda=0) to its actual partial charges (lambda=1). I get significantly
> different results of <dVpot/dlambda> with Gromacs 3.3.1 on one hand and 
> 3.3.3
> (and later versions) on the other hand. I use linear scaling of the charges
> with lambda.
> 
> There is a difference for all lambda values but it's most prominent near
> lambda=0, where the values are different by an order of magnitude. The
> difference is independent of the trajectory, i.e., when I use the 
> trajectory
> generated with 3.3.1 and rerun it with 3.3.3, I get the same results as 
> if I
> had generated the trajectory with 3.3.3 directly.
> 
> Is this a known phenomenon?
> Can anyone tell me what is the difference between versions 3.3.1 and later
> versions in this context?
> Are there any "hidden" corrections that have been added between 3.3.1 
> and 3.3.3?
> 

Are you using PME?  There is a bug in the 3.3.1 free energy code with PME.  It 
is a known issue, but unfortunately it looks like the mailing list search is 
down so I can't locate the thread where this is discussed.

-Justin

> Thanks
> 
> Emanuel
> 

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================