Published on:

12 October 2023

Primary Category:

Quantum Physics

Paper Authors:

Sonaldeep Halder,

Anish Dey,

Chinmay Shrikhande,

Rahul Maitra

•

Machine learning is used to expand an initial approximate wavefunction into dominant configurations

•

Incorporates high-rank excitations indirectly via 'scatterer' operators

•

Results in low-depth quantum circuits tailored to each molecule

•

Significantly enhances feasibility of quantum simulations

•

Easily integrates with error mitigation techniques

Machine learning helps construct compact quantum circuits for chemistry

This paper develops a new protocol combining machine learning and quantum chemistry principles to construct compact, customizable quantum circuits to represent molecular wavefunctions. The approach uses a restricted Boltzmann machine to expand an initial approximate wavefunction into the dominant configurations. It incorporates high-rank excitations indirectly via 'scatterer' operators, resulting in very low-depth circuits tailored to each molecule.

Machine learning reveals key insights on approximating the random phase approximation

Machine learning improves molecular force fields

Training Classical Neural Networks via Quantum Approximation

Neural networks model quantum ground states

Machine learning of electronic excitations

Quantum wave function collapse for procedural content

No comments yet, be the first to start the conversation...

Sign up to comment on this paper