Abstract:We develop a strategy that integrates machine learning and first-principles calculations to achieve technical accurate predictions of infrared spectra. Specifically, the methodology allows to predict infrared spectra for complex systems at finite temperatures. The method's effectiveness is demonstrated in challenging scenarios, such as the analysis of water and the organic-inorganic halide perovskite MAPbI$_{3}$, where our results consistently align with experimental data. A distinctive feature of the methodology is the incorporation of derivative learning, which proves indispensable for obtaining accurate polarization data in bulk materials and facilitates the training of a machine learning surrogate model of the polarization adapted to rotational and translational symmetries. We achieve polarisation prediction accuracies of about 1 % by training only on the predicted Born effective charges.

What problem does this paper attempt to address?

The paper aims to address the issue of accurately predicting the infrared spectra of complex materials at finite temperatures. To solve this problem, the authors developed a method that combines machine learning with ab initio calculations. The features of this method include: 1. **Construction of Machine Learning Force Field (MLFF)**: Generating high-quality molecular dynamics trajectories through machine learning to simulate the dynamic behavior of materials. 2. **Tensorial Learning**: Specifically learning polarization data in bulk materials, achieved by learning the Born effective charge tensor, which helps build a machine learning surrogate model that can adapt to rotational and translational symmetries. 3. **Prediction Accuracy**: By training on Born effective charges, a polarization prediction accuracy of approximately 1% can be achieved. 4. **Addressing Polarization Uncertainty in Bulk Materials**: Determining polarization in bulk materials through derivative learning, using the Born effective charge tensor as derivative information for polarization, thus resolving the issue of non-uniqueness in bulk material polarization. 5. **Validation of Method Effectiveness**: Validating the method's effectiveness in challenging scenarios (such as water and organic-inorganic halide perovskite MAPbI3), with results showing consistency with experimental data. 6. **Theoretical Framework**: The paper also details the theoretical framework used, including energy expressions, Green-Kubo relations, etc., and discusses the specific implementation of molecular dynamics simulations. In summary, this research proposes an innovative method that can effectively predict the infrared spectra of complex systems at finite temperatures, which is of significant importance for fields such as catalysis, clinical analysis, and materials science research.

Derivative learning of tensorial quantities -- Predicting finite temperature infrared spectra from first principles

Machine Learning Molecular Dynamics for the Simulation of Infrared Spectra

Efficient Composite Infrared Spectroscopy: Combining the Doubly-Harmonic Approximation with Machine Learning Potentials

Tensorial properties via the neuroevolution potential framework: Fast simulation of infrared and Raman spectra

Neural Network Approach for Predicting Infrared Spectra from 3D Molecular Structure

First-principles Predictions of Temperature-Dependent Infrared Dielectric Function of Polar Materials by Including Four-Phonon Scattering and Phonon Frequency Shift

Deep learning of optical spectra of semiconductors and insulators

Performance metrics for tensorial learning: prediction of Li4Ti5O12 nuclear magnetic resonance observables at experimental accuracy

Boosting the Modeling of InfraRed and Raman Spectra of Bulk Phase Chromophores with Machine Learning

Infrared Spectra at Coupled Cluster Accuracy from Neural Network Representations

Deep learning of spectra: Predicting the dielectric function of semiconductors

Machine-learning prediction of infrared spectra of interstellar polycyclic aromatic hydrocarbons

First-Principles Spectroscopy of Aqueous Interfaces using Machine-Learned Electronic and Quantum Nuclear Effects

Infrared Spectra Prediction for Functional Group Region Utilizing a Machine Learning Approach with Structural Neighboring Mechanism

Finite temperature dielectric properties of KTaO$_3$ from first principles and machine learning: Phonon spectra, Barrett law, strain engineering and electrostriction

Machine Learning Assisted Prediction of Dielectric Temperature Spectrum of Ferroelectrics

Infrared spectra of neutral polycyclic aromatic hydrocarbons based on machine learning potential energy surface and dipole mapping

Dielectric Tensor Prediction for Inorganic Materials Using Latent Information from Preferred Potential

Phase transitions of hybrid perovskites simulated by machine-learning force fields trained on-the-fly with Bayesian inference

Augmenting zero-Kelvin quantum mechanics with machine learning for the prediction of chemical reactions at high temperatures