6–10 May 2024
ECT*
Europe/Rome timezone

Automated Active Space Selection Using Large Language Models

Not scheduled
30m
Talk Talks

Speaker

Romit Chakraborty (University of Chicago, University of California, Berkeley)

Description

The Multi-Reference Electronic Structure Theory involves choosing an active space with knowledge of the subspace's spatial and energetic information within the Hilbert space of a molecular electronic Hamiltonian that exhibits strong correlation. This process can be automated with the help of an AI assistant. This paper presents such an assistant that utilizes tools like the Approximate Pair Coefficient (APC) and the Atomic Valence Active Space (AVAS). Additionally, the assistant has a fine-tuned Large Language Model that can determine the active space required for a given molecular state with informed decisions.

Primary author

Romit Chakraborty (University of Chicago, University of California, Berkeley)

Presentation materials

There are no materials yet.