[gmx-developers] Brainstorming ideas for minimally painless incorporation of an MC barostat (later MC moves in general)

Berk Hess hess at kth.se
Thu Jul 17 15:49:49 CEST 2014

On 07/14/2014 04:05 PM, Shirts, Michael R. (mrs5pt) wrote:
> Hi, all-
> Especially for Berk, but other ideas are likely to be very useful as well.
> This is also filed as http://redmine.gromacs.org/issues/1562, which will
> likely be a better place to hold the conversation moving forward (but I
> wanted to publicize it first).
> As we eliminate the iterative constraint pressure control method, it will
> be important to introduce a replacement in-theory-exact barostat for
> systems with constraints, and as discussed, and MC barostat is the best
> replacement.
> For an MC barostat, this becomes specifically:
> Every nstpcouple steps:
>    1. Compute the energy (force may not be needed - will have to think
> about combinations with MD)
>    2. Modify the coordinates (optionally the velocities, depending on the
> flavor of MC barostat)
>       a. For a MC barostat, no communication between nodes is required, as
> atoms are guaranteed stay on the same node -- everything only changes by
> scaling
This is not really true, since when constraints are present, these need 
to be applied after scaling, which leads to (minor) coordinate changes 
and requires communication. But we can probably do this without changing 
the decomposition.
>       b. For general MC moves (will likely be best to eventually make it
> general) communication between notes might be required.
>    3. Reevaluate the energy
>    4. Accept/reject based on energy difference (may require some other
> Jacobian terms). KEY PART: 'reject' reverts the system to state before the
> proposal
Reverting is easy, a simple mechanism is present in the energy 
minimization code.
>    5. Continue
> What will need to change in the way the forces are calculated for this to
> be possible? What other changes?
I don't think any changes are needed for the forces.
The constraining and calculating virial contributions are probably the 
most tricky parts.


> ~~~~~~~~~~~~
> Michael Shirts
> Assistant Professor
> Department of Chemical Engineering
> University of Virginia
> michael.shirts at virginia.edu
> (434)-243-1821

More information about the gromacs.org_gmx-developers mailing list