[gmx-developers] Re: [gmx-revision] pair types for free-energy decoupling

[Berk Hess]

Michael Shirts wrote:

>Hi, all-
>
>I've been thinking about the free energy coupling changes.  First, I
>had a question that Berk posted to the gmx-revision list
>
>On 2/3/06, Berk Hess <hessb at mpip-mainz.mpg.de> wrote:
>  
>
>>Added new pair type 2, which is identical to the standard type 1,
>>except that it only the A-state charges and LJ parameters.
>>    
>>
>
>Could you restate this?  There appears to be a typo here, and I'm not
>sure what the differences are with pair 3.
>  
>
A typo indeed.
It should be:
except that it only USES the ....

>  
>
>>Added new pair type 3, which use the non-scaled charges
>>and the normal non-bonded parameters, also only from the A-state.
>>    
>>
>
>So, as I know understand it, if one were doing free energy decoupling,
>which would be done with pair type 3, the atoms of the ligand would
>always have the unperturbed interactions.  But if PME or Ewald is
>being computed, won't the ligand intramolecular interactions also be
>calculated using the scaled charges, leading to double-counding? 
>Won't this need to be subtracted off with a separate Ewald
>calculation?
>  
>
The idea is that for the solute to be decoupled one would replace
the in its [ pairs ] section type 1 by type 2
and exclude all intramolecular non-bonded interactions and replace
add them again explicitly in a [ pairs ] section as type 3.

The standard 3.3 mdrun already subtracts the PME mesh contribution
to the force, energy and dH/dl for exclusions, so there is no double 
counting.

BTW I tested it and it works.

Berk.