I have a set of molecular dimers (~25 atoms) for which DFT-D3 interaction energies are calculated by Psi4 and Turbomole program. Now I want to compare these values against Force-Field methods like UFF/Amber/OPLS. I have known that in Gromacs or Amber suite of programs can do that but for these small systems, I think those will be overkill. Any suggestion of programs which can calculate single point energies of the small dimers?
$\begingroup$ $\endgroup$
15 
- 3$\begingroup$ What do you mean by overkill? Lammps might have this capability. See their documentation for details. Maybe their user forum might have some insight. $\endgroup$PBH– PBH2021-10-19 06:21:35 +00:00Commented Oct 19, 2021 at 6:21
- 2$\begingroup$ The ultimate simplicity would be to just write your own FF code. It would be short for dimers. No need for ewald or anything. Just sum the lennard-jones and simple coulomb interactions. The longest part of the program would be reading in atom type lennard-jones parameters and charges and storing them in some kind of structure for recall later. $\endgroup$Wesley– Wesley2021-10-19 15:53:54 +00:00Commented Oct 19, 2021 at 15:53
- 2$\begingroup$ Most MD codes also have the ability to perform a geometry optimization. I would suggest using that so that you can be sure the structure has the appropriate minimum with each force field. Otherwise you have to hack the MD code to only run for one step which I always find concerning because it's hard to know if it will move the geometry before printing the energies. Also, make sure you don't use periodic boundary conditions, otherwise the MD package will probably use them automatically and give incorrect results. $\endgroup$jheindel– jheindel2021-10-20 07:10:45 +00:00Commented Oct 20, 2021 at 7:10
- 2$\begingroup$ @PrasantaBandyopadhyay Right. You probably still want the single point energies at the minimum of the corresponding force field, even if you also want them at the MP2 geometries. The optimization will give you both at once because the first energy in the optimization routine is the single point energy at the MP2 geometry. $\endgroup$jheindel– jheindel2021-10-20 16:50:04 +00:00Commented Oct 20, 2021 at 16:50
- 1$\begingroup$ OpenMM should be able to do this pretty easily, if the FF you want is in a format that can be converted by github.com/openmm/openmmforcefields, and if you have a way to get the topology information (atom types) into OpenMM. Update the Q with an example configuration/topology information and I can see if I can the details to work. However, I second @jheindel's point that this isn't expected to be a good comparison. (This kind of single-point comparison is exactly what is done to test the conversion in that GH repository.) $\endgroup$dwhswenson– dwhswenson2021-11-10 22:53:28 +00:00Commented Nov 10, 2021 at 22:53
| Show 10 more comments