I'm very interested in this class of approximations on consumer hardware from a different perspective: Simulating (simple) biological systems and organic molecules. Background: I'm building out the rust structural bio OSS ecosystem, and a GUI/3D CAD-style visualizer. It has molecular dynamics integrated tightly, and is designed to "just work" without fuss, setup, or errors.
The problem: It is reliant on atom-centered partial charge, and pre-calculated parameters for its Newtonian forces. These are available for a set of 30k or so organic molecules, and most protein, lipid, and nucleic acid configurations. The problems: Making it work for arbitrary systems, and doing better than these specifically-tuned models. I want the general case, and it to work in a no-fuss way.
I am now looking into the TWA, and am interested in this class of approximation in general. I have the UI and traditional MD system mostly ready; getting ready to tackle this problem. This is a much simpler setup than the ones physicists are interested, and that the article covers, since we can focus on stable covalent-bonded systems of atoms.
replies(1):