MAPLE Publications
Papers describing the MAPLE framework and its applications.
MAPLE: A General Framework for Automated Molecular Modeling Across Machine Learning Potentials
Presents MAPLE as a comprehensive framework for automated molecular modeling that works seamlessly across different machine learning potentials, providing a unified interface for diverse computational chemistry applications.
ML Potential References
Key publications for the machine learning potentials supported by MAPLE.
ANI-2x: A Transferable Neural Network Potential
Devereux et al. Extending the applicability of the ANI deep learning potential to include organic molecules containing sulfur, fluorine, and chlorine. J. Chem. Theory Comput., 2020.
ANI-1: An Extensible Neural Network Potential
Smith et al. ANI-1: an extensible neural network potential with DFT accuracy at force field computational cost. Chem. Sci., 2017.
ANI-1ccx: Approaching Coupled Cluster Accuracy
Smith et al. Approaching coupled cluster accuracy with a general-purpose neural network potential through transfer learning. Nat. Commun., 2019.
AIMNet2: A Neural Network Interatomic Potential
Anstine and Isayev. AIMNet2: a neural network interatomic potential to meet your neutral and charged molecular modeling needs. 2024.
MACE-OFF: Transferable Organic Force Fields
Kovacs et al. MACE-OFF23: transferable machine learning force fields for organic molecules. 2023.
UMA: Universal Machine-learning Architecture
Meta FAIR. A foundation model for chemistry covering broad chemical space with a single unified interatomic potential.
EGRET: Equivariant Graph Network
An equivariant graph neural network potential for reactive and non-reactive molecular simulations.