Ab Initio versions are incredibly efficient. They strip away the aimless meandering of standard classes. You aren't spending six weeks learning colors; you are learning sentence structures, thematic vocabulary, and survival skills immediately.
ML potentials are getting shockingly good. But they depend on training data — and that data comes from the expensive, “ab initio version” codes. When the ab initio version changes (e.g., higher accuracy functional, core-valence correlation), the ML model’s ceiling moves too. ab initio versions
In the scientific community, "ab initio" (Latin for "from the beginning") refers to calculations based solely on fundamental physical constants and the laws of quantum mechanics. Review - SciELO Ab Initio versions are incredibly efficient
Why “Ab Initio” Versions Still Matter in an AI-Driven World ML potentials are getting shockingly good
Here’s a draft for an interesting post about ab initio versions — tailored for a computational chemistry, materials science, or ML/physics audience.