[gmx-developers] Flat-bottom potentials

[Mark Abraham]

On 5/03/2012 7:24 PM, Jochen Hub wrote:
> OK, since there are obviously no preferences for how to implement 
> flat-bottom potentials, I'll just use new function types and submit 
> the implementation to Gerrit (and hope that it is acceptable).

Looked fine to me. You will need to allow for a parameter for the force 
constant and for the radius, of course. There may be some efficiency to 
be had by reusing code that is used for flat-bottomed distance restraints.

Mark

>
> Best,
> Jochen
>
>
> Am 3/2/12 12:23 PM, schrieb Jochen Hub:
>> Hi developers,
>>
>> I would (finally) like to implement some flat-bottom quadratic
>> potentials into Gromacs. They are useful for many applications, but so
>> far I usually implemented them as more or less dirty hacks, which of
>> course never made it into the main branch. Berk, David and me talked
>> about that topic already years ago, and now I would like to continue
>> this debate.
>>
>> I think the flat-bottom quadratic potentials should allow:
>>
>> 1) different geometries:
>> - spherical (e.g. to avoid evaporation from a spherical droplet)
>> - cylindrical in x-y plane, but free in z (needed to compute PMFs for
>> membrane channels, to keep the solute aligned with the channel pore)
>> - 1D, e.g. to keep a layer of some thickness stable
>>
>> 2) different reference positions. This is important if you compute PMFs
>> for mutimeric membrane channels, where you want to have a cylindrical
>> flat-bottom quadratic potential at each monomer.
>>
>> I can think of two ways to implement them:
>> a) not preferred: into the pull code. AFAIK, that would not allow
>> different reference positions at the moment (tell me if I am wrong).
>>
>> b) preferred: New function types (ft) for the [ position_restraints ]
>> section in the topology. So ft=1 would be the normal posres, and ft 2,
>> 3, 4 5, 6 could be for spherical, cylindrical in the xy plane, and 1D in
>> X, Y, and Z, respectively. So ft>1 would ask for one additional
>> parameter r which gives the radius/half-width of the flat bottom. Only
>> one force constant would be required for the flat-bottom potential.
>>
>> Would you agree with that? Please let me know.
>>
>> Cheers,
>> Jochen
>>
>>
>>
>