Abstract:Modeling non-empirical and highly flexible interatomic potential energy surfaces (PES) using machine learning (ML) approaches is becoming popular in molecular and materials research. Training an ML-PES is typically performed in two stages: feature extraction and structure-property relationship modeling. The feature extraction stage transforms atomic positions into a symmetry-invariant mathematical representation. This representation can be fine-tuned by adjusting on a set of so-called "hyper-parameters" (HPs). Subsequently, an ML algorithm such as neural networks or Gaussian process regression (GPR) is used to model the structure-PES relationship based on another set of HPs. Choosing optimal values for the two sets of HPs is critical to ensure the high quality of the resulting ML-PES model. In this paper, we explore HP optimization strategies tailored for ML-PES generation using a custom-coded parallel particle swarm optimizer (available freely at <a class="link-external link-https" href="https://github.com/suresh0807/PPSO.git" rel="external noopener nofollow">this https URL</a>). We employ the smooth overlap of atomic positions (SOAP) descriptor in combination with GPR-based Gaussian approximation potentials (GAP) and optimize HPs for four distinct systems: a toy C dimer, amorphous carbon, $\alpha$-Fe, and small organic molecules (QM9 dataset). We propose a two-step optimization strategy in which the HPs related to the feature extraction stage are optimized first, followed by the optimization of the HPs in the training stage. This strategy is computationally more efficient than optimizing all HPs at the same time by means of significantly reducing the number of ML models needed to be trained to obtain the optimal HPs. This approach can be trivially extended to other combinations of descriptor and ML algorithm and brings us another step closer to fully automated ML-PES generation.

Hyperparameter Optimization for Atomic Cluster Expansion Potentials

Cartesian atomic cluster expansion for machine learning interatomic potentials

Particle Swarm Based Hyper-Parameter Optimization for Machine Learned Interatomic Potentials

Optimal design of experiments in the context of machine-learning inter-atomic potentials: improving the efficiency and transferability of kernel based methods

ACEpotentials.jl: A Julia Implementation of the Atomic Cluster Expansion

On the Atomic Cluster Expansion: interatomic potentials and beyond

Adaptive-precision potentials for large-scale atomistic simulations

Bayesian optimization of atomic structures with prior probabilities from universal interatomic potentials

Atomic cluster expansion potential for large scale simulations of hydrocarbons under shock compression

Improve the Performance of Machine-Learning Potentials by Optimizing Descriptors

Introduction to machine learning potentials for atomistic simulations

Atomic Cluster Expansion for a General-Purpose Interatomic Potential of Magnesium

De novo exploration and self-guided learning of potential-energy surfaces

Atomic cluster expansion for quantum-accurate large-scale simulations of carbon

Atomic Cluster Expansion without Self-Interaction

Machine Learning Interatomic Potentials and Long-Range Physics

Active learning strategies for atomic cluster expansion models

Constructing accurate and efficient general-purpose atomistic machine learning model with transferable accuracy for quantum chemistry

Machine-learning interatomic potentials for materials science

Including physics-informed atomization constraints in neural networks for reactive chemistry