Resolution-vs.-Accuracy Dilemma in Machine Learning Modeling of Electronic Excitation Spectra

TL;DR

This paper investigates the trade-off between resolution and accuracy in machine learning models for predicting electronic excitation spectra, introducing a new dataset and demonstrating improved spectral predictions with ML over traditional methods.

Contribution

The study presents a new large dataset of molecular spectra and shows that ML models trained on limited data can accurately predict high-resolution spectra, addressing the resolution-accuracy dilemma.

Findings

ML models outperform small basis set predictions in spectral accuracy

A new dataset with 12,880 molecules enables better spectral modeling

High-resolution spectra can be recovered with less than 10% training data

Abstract

In this study, we explore the potential of machine learning for modeling molecular electronic spectral intensities as a continuous function in a given wavelength range. Since presently available chemical space datasets provide excitation energies and corresponding oscillator strengths for only a few valence transitions, here, we present a new dataset -- \bigqm -- with 12,880 molecules containing up to 7 CONF atoms and report ground state and excited state properties. A publicly accessible web-based data-mining platform is presented to facilitate on-the-fly screening of several molecular properties including harmonic vibrational and electronic spectra. We present all singlet electronic transitions from the ground state calculated using the time-dependent density functional theory framework with the $\omega$B97XD exchange-correlation functional and a diffuse-function augmented basis set. The resulting spectra predominantly span the X-ray to deep-UV region (10--120 nm). To compare the target spectra with predictions based on small basis sets, we bin spectral intensities and show good agreement is obtained only at the expense of the resolution. Compared to this, machine learning models with latest structural representations trained directly using $<10 \%$ of the target data recover the spectra of the remaining molecules with better accuracies at a desirable $<1$ nm wavelength resolution.