There is a need in the materials design and electronic-structure communities to reduce the cost and increase the accuracy of correlation energy calculations for real systems. One barrier to satisfying this need is the error associated with simulating real systems with a finite number of particles. We address these so-called finite-size errors by using machine learning to calculate correlation energies for large system sizes using coupled cluster data for smaller system sizes. In doing so, we reduce the cost associated with calculating high accuracy energies, thus making the simulation of real materials more affordable.

Machine learning for correcting finite-size errors in coupled cluster singles and doubles correlation energies

ACS Fall 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.