User Guide#
Welcome to the ALCHEMI Toolkit-Ops user guide. Start with installation below, then choose a component to learn its features, make your first call, and find guidance for your workload.
Quick Start#
The quickest way to install ALCHEMI Toolkit-Ops:
$ pip install nvalchemi-toolkit-ops
To install ALCHEMI Toolkit-Ops with a deep-learning backend:
$ pip install 'nvalchemi-toolkit-ops[torch]'
$ pip install 'nvalchemi-toolkit-ops[jax]'
Tip
Running on NVIDIA DGX Spark? The Blackwell GPU requires CUDA 13 wheels for PyTorch. See the CUDA 13 installation notes before proceeding.
About#
Core Components#
Neighbor Lists — find nearby atom pairs and use their distances, vectors, or pair-function outputs.
Electrostatics — calculate Coulomb interactions with direct, Ewald, PME, or DSF methods.
Dispersion Corrections — add two-body DFT-D3(BJ) dispersion with environment-dependent coefficients.
Dynamics — integrate molecular dynamics trajectories and relax structures with geometry optimization.
Segment Operations — reduce and transform data grouped by system or segment.