“Dividing and Conquering” and “Caching” in Molecular Modeling

Molecular modeling is widely utilized in subjects including but not limited to physics, chemistry, biology, materials science and engineering. Impressive progress has been made development of theories, algorithms software packages. To divide conquer, cache intermediate results have long standing principles algorithms. Not surprisingly, most important methodological advancements more than half century molecular are various implementations these two fundamental principles. In the mainstream classical computational science, tremendous efforts invested on lines algorithm development. The first coarse graining, which represent multiple basic particles higher resolution as a single larger softer particle lower counterpart, with resulting force fields partial transferability at expense some information loss. second enhanced sampling, realizes “dividing conquering” and/or “caching” configurational space focus either reaction coordinates collective variables metadynamics related algorithms, or transition matrix state discretization Markov models. For this line spatial maintained transferable. Deep learning realize efficient accurate ways along algorithmic research. We proposed demonstrated local free energy landscape approach, new framework for science. This based third class that facilitates through partially transferable distributions clusters degrees freedom. Differences, connections potential interactions among three directions discussed, hope stimulate elegant, reliable formulations complex systems.