Virtual Screening of Chemical Space based on Quantum Annealing

TL;DR

This paper demonstrates that quantum annealing can efficiently sample and identify important features in chemical space, significantly reducing the search space for new materials and potentially accelerating material discovery.

Contribution

The study introduces a novel approach using quantum annealing as a sampler to extract feature importance, reducing chemical space by over 99%.

Findings

Chemical space can be reduced to less than 1% using feature importance.

Quantum annealer effectively samples data faster than classical methods.

Accelerates material research by narrowing down candidate molecules.

Abstract

For searching a new chemical material which satisfies the target characteristic value, for example emission wavelength, many cut and trial of experiments/calculations are required since the chemical space is astronomically large (organic molecules generates >10^60 candidates). Extracting feature importance is a method to reduce the chemical space, and limiting the search space to those features leads to shorter development time. Quantum computer can generate sampling data faster than classical computers, and this property is utilized to extract feature importance. In this paper, quantum annealer was used as a sampler to make data for extracting feature importance of material properties. By screening the chemical space with feature importance, it was found that the chemical space can be reduced to less than 1 percent. This result suggests that the acceleration of material research can be achievable.