simtk.org

Hello,

I was trying to perform a RPMD simulation where I have applied constraints to some of the atoms in the simulation.

When I try to get a context, I get the following error:

Exception: RPMDIntegrator cannot be used with Systems that include constraints

Would it be possible to include constraints within the RPMD integrator?

Cheers,
Kevin

In principle yes, but the theory would need to be revised to properly account for the constraints. A while back I spoke to Tom Markland (who was our collaborator in developing this feature) about it, and he said deriving the theory for constraints was on his to-do list, but he hadn't done it yet. I don't know if he or someone else has done it since then.

Peter

Hi Peter,

Thank you for the quick reply!

There has been work done using constraints on the beads in the following papers: J. Chem. Phys. 115, 10758 (2001) and J. Chem. Phys. 123, 154504 (2005) but the theory had not been derived as you suggested.

Two recent papers have been published that justify the use of constraints on the beads to simulate quantum rigid bodies: J. Chem. Phys. 138, 184101 (2013) and J. Chem. Phys. 139, 184115 (2013).

The application of RATTLE/SHAKE correctly calculates the structural properties but requires a different energy estimator, but this can be calculated from the bead position and potential energy from the RPMD integrator.

Please let me know if you think this would be a reasonable feature to add and if you have any questions!
Kevin

It is definitely not that this is not do-able (we have been doing PIMD/RPMD with constraints for the last decade and it has been done before that). The main thing is that it was not a priority in the initial implementation. One could add it to the feature requests for the RPMD/PIMD integrator.

Thank you for your reply!

I have created a feature request as you suggested.

Additionally, if you need any help testing the implementation of the constraints, I would be more than happy to help!