Page 1 of 1

OpenMM accelerated Tinker

Posted: Thu Feb 02, 2012 3:56 pm
by akila
We have installed OpenMM accelerated TINKER (tried both v5.05/OpenMM3 and v6/OpenMM4) on our GPU workstation. We are trying to use the AMOEBA polarizable force field for a system of water molecules.

When using OpenMM accelerated Tinker executable, dynamic_openmm with the input files from Tinker, waterball.xyz, waterball.key and water.prm, it reports that the Beeman integrator is not supported. Using the same input files with the TINKER executable without OpenMM, dynamic, the MD simulation runs fine. Any insights into this will be appreciated.

Akila Gothandaraman
University of Pittsburgh

Re: OpenMM accelerated Tinker

Posted: Thu Feb 02, 2012 4:31 pm
by joyku
Not all the AMOEBA forces, options, and integrators have been implemented in the OpenMM version. This includes the Beeman integrator. The OpenMM Users Guide (p. 113-115 in the Release 4.0 version) describes all the limitations.

You can request new features be added to OpenMM by going to the project page (http://simtk.org/home/openmm) and clicking on the link Advanced -> Features & Bugs.

Joy

Re: OpenMM accelerated Tinker

Posted: Thu Feb 02, 2012 4:53 pm
by friedrim
Hi Akila,

Try adding one of the following lines to the TINKER key file:

integrator VERLET

or

integrator STOCHASTIC

Mark

Re: OpenMM accelerated Tinker

Posted: Fri Feb 03, 2012 1:25 pm
by akila
Hello Joy and Mark,

Thanks for your comments.

I think Mark was suggesting to use

integrate verlet (and not integrator verlet). Thanks for the tip. This prevented the program from using the default setting (Beeman) in the source file, dynamic.f.

-Akila