torchani#

The TorchANI neural network potential library main namespace

Most of the functions and classes of the library are accessible from their specific modules. For convenience, some useful classes are accessible also from here. These are:

Functions

single_point

Calculate properties for a batch of molecules

Modules

neighbors

Modular neighborlists to improve scaling for large systems

aev

Atomic Environment Vectors (AEVs) are TorchANI's name for local atomic features, calculated from the local chemical environment of each atom.

nn

Classes that represent atomic (and groups of element-specific) neural networks

grad

Wrapper functions that make use of torch.autograd

models

Provides access to all published ANI models.

potentials

Callable objects that can be attached to TorchANI models.

cutoffs

Collection of Cutoff functions

datasets

Functions and classes for creating batched and map-like datasets.

transforms

Composable functions that modify to properties when training or batching a dataset.

io

Convenience functions for reading and writing molecules in ".xyz" format

electro

Utilities for working with charged systems

arch

Construction of ANI-style models and definition of their architecture

constants

Atomic constants, and Density Functional constants

utils

Misc utilities used throughout the code

units

Unit conversion factors used in TorchANI

sae

Simple module that calculates a "1-body atomic potential"

sae_estimation

Functions to calculate self atomic energies (SAEs) via linear regression.

cli

Contains the TorchANI CLI entrypoints.

paths

Locations used by TorchANI to cache various resources

ase

Calculator subclass, used for interfacing with the ASE library

annotations

Type aliases; common type annotations used throughout the code

neurochem

This module is part of the Legacy API of TorchANI 2 and should not be used in new code.

legacy_data

This module is part of the Legacy API of TorchANI 2 and should not be used in new code.