[gmx-developers] pull code generalization feedback request
hess at kth.se
Tue Oct 22 10:20:12 CEST 2013
Such options would be nice, but this would lead to replication of a
large part of the bonded code in the pull code (and a crazy amount of
mdp options). If you are use umbrella potentials, all this can probably
be achieved with virtual sites and the normal bonded code.
As for parallelization, with domain decomposition this should still work
up to moderate parallelization when you use mdrun -rdd. In 5.0 we will
have shifted LJ in the verlet cut-off scheme and then you can run pure
OpenMP parallelization, where everything will work, or hybrid MPI+OpenMP
so you can use far fewer domains.
On 10/21/2013 06:44 PM, XAvier Periole wrote:
> The dimensions we use to define the relative orientation of two objects are defined using three anchors on each object (A,B,C) and (a,b,c). It is easier to place them on a equilateral triangle:
> - the distance between the centres of masses (anchors A and a) of the two objects: d(A,a),
> - the second and third ones are the bond angles between the vector defined by the centres of masses (A and a) and the second anchor point (B and b): <Aab and aAB.
> - the forth is the dihedral angle around the Aa vector: [BAab
> - the last two are rotations around the objects themselves: [Abcd and [aBCD
> Each of these dimensions are independent and may vary independently, that is actually the all point of it :)
> I hope this is more clear if not ask again :))
> On Oct 21, 2013, at 5:37 PM, "Shirts, Michael (mrs5pt)" <mrs5pt at eservices.virginia.edu> wrote:
>> Hi, Xavier-
>>> I realise that this is a bit late but it would be really great to be able to
>>> include multi-dimensional umbrella sampling with the option to define multiple
>>> bonds, angles and dihedrals simultaneously.
>> Can you define "multi-dimensional" in this context? If the bond, angles,
>> and dihedrals are defined simultaneously, then isn't there just one
>> dimension? What is the second dimension?
>> It's possible some of this can be done now, actually, depending on what you
>> mean. But if you can be more clear as to exactly what the dimensions you
>> want to simulate now, and how you code the restraints in currently, then it
>> will be very straightforward to include such functionality in what is
>> already planned for 5.0.
>> Michael Shirts
>> Assistant Professor
>> Department of Chemical Engineering
>> University of Virginia
>> michael.shirts at virginia.edu
>> gmx-developers mailing list
>> gmx-developers at gromacs.org
>> Please don't post (un)subscribe requests to the list. Use the
>> www interface or send it to gmx-developers-request at gromacs.org.
More information about the gromacs.org_gmx-developers