A simple approach to rotationally invariant machine learning of avector quantity

TL;DR

This paper introduces a simple, rotation-invariant machine learning method for vector and tensor properties using a three-step rotate-predict-rotate approach based on the molecular tensor of inertia, avoiding complex equivariant techniques.

Contribution

The authors propose a novel, straightforward RPR technique that guarantees rotational covariance in ML predictions of vector and tensor properties without auxiliary constructions.

Findings

The RPR method accurately predicts dipole moments along MD trajectories.

The approach is computationally efficient and easily extensible to tensors.

Validated on 1,2-dichloroethane MD data.

Abstract

Unlike with the energy, which is a scalar property, machine learning (ML) predictions of vector or tensor properties poses the additional challenge of achieving proper invariance (covariance) with respect to molecular rotation. If the properties cannot be obtained by differentiation, other appropriate methods should be applied to retain the covariance. There have been several approaches suggested to properly treat this issue. For nonadiabatic couplings and polarizabilities, for example, it was possible to construct virtual quantities from which the above tensorial properties are obtained by differentiation and thus guarantee the covariance. Here we propose a simpler alternative technique, which does not require construction of auxiliary properties or application of special equivariant ML techniques. We suggest a three-step approach, using the molecular tensor of inertia. In the first step, the molecule is rotated using the eigenvectors of this tensor to its principal axes. In the second step, the ML procedure predicts the vector property relative to this orientation, based on a training set where all vector properties were in this same coordinate system. As third step, it remains to transform the ML estimate of the vector property back to the original orientation. This rotate-predict-rotate (RPR) procedure should thus guarantee proper covariance of a vector property and is trivially extensible also to tensors such as polarizability. The PRP procedure has an advantage that the accurate models can be trained very fast for thousands of molecular configurations which might be beneficial where many trainings are required (e.g., in active learning). We have implemented the RPR technique, using the MLatom and Newton-X programs for ML and MD and performed its assessment on the dipole moment along MD trajectories of 1,2-dichloroethane.