Infrastructure for bespoke LJ parameter assignment based on QM calculations | Sophie Kantonen | Concept is to reduce number of adjustable parameters by not having LJ types but instead to parameterize a mapping from QM to sigma & epsilon for each atom in the molecule. (https://bit.ly/2VnnIsE). See also D Coles’s methods. JW – Maybe could look at implementing this in the toolkit. Touch base with engineering team later in May. DLM – Performance of this method? SK – MEasured using HVap, seems comparable to GAFF, with far fewer parameters. CB – I like the idea. As you put more charge on an atom, the radius goes up, the density goes up. So we’d expect a correlation between charge and radius or beta. MKG –Having both sigma and epsilon depend on the single variable beta, for a given element, already makes sigma and epsilon covary more than one might wish. If partial charge also were to covary with both of these, that might reduce the flexibility of the nonbonded interaction model to the point that it would be hard to match experimental condensed phase data. SK– The algorithm that gives beta_i (the exponential decay coefficient of atom i central to SDLJ) also gives Ni, the number of electrons in the atom’s Slater orbital. It may be interesting to look at whether N_i correlates with beta_i. If so, that would support Chris’s idea. If not, it may on the other hand, provide another variable that could be used to break the correlation between sigma and epsilon for each element. CB–The variability of the parent element's size (r*) and its well depth (epsilon) are distinguished in the physics underlying Sophie's mapping. Knowing that empirical dispersion in condensed phases is intrinsically non-local and many-body, this may affect epsilon more empirically than r*. Finally, any successful mapping scheme may be used to "bin" like mappings to come up with superior SMARTS-based LJ parameters. CB – This is reminiscent of Simon’s property calculation work, which implicitly considers all sorts of intermolecular interactions CB – It’d be interesting to see whether, emerging from the map potential, which is based on electronic structure, a correspondence for the purely empirical LJ that come out of the purely-empirical fit. This way, map potentials could guide parameter types. MKG – The challenge is that, when you have a lot of LJ types, it’s hard to be confident that they’ve been optimized enough. This is my big motivation for using less overall parameters CB – If, out of the mapping, bins appear, then this is a good guide to separating parameter types.
|