Stage 4: Iterate to convergence
The details of the solvent electrostatic field that we created in
Stage 2 depend on the details of the solute that we
laid down in Stage 1. Now that the
solute properties have changed from our initial guess, we should make sure that
the field itself is consistent with the new charge distribution. Using the
new library file created at the end of Stage 3,
regenerate each of the topologies for the solvated systems in from the end of
Stage 1 (to save time in pmemd computations, it's
reasonable to take the coordinates from the previously equilibrated water boxes
as a starting point). Proceed in Stage 2 to run
mdgx to perform dynamics and collect the solvent electrostatic
field, launch quantum calculations, collect the grids, and run the fitting in
Stage 3 again. This single iteration should produce
a pretty well converged answer as to the IPolQ charges for the molecule, but if
the initial guess was really bad (let's say you said each of the solute charges
was zero at the beginning of Stage 1), a third iteration might be in order.
Copy the latest charges into the
Glycerol.prepi file and it is
ready to use in any subsequent simulations.
Return to the introduction.