cuEST

cuEST Examples

This directory contains standalone examples for the cuEST library, covering the C API, Python API, and the cuEST SCF Python library.

For full API documentation, see https://docs.nvidia.com/cuda/cuest/.

Prerequisites

• CUDA Toolkit 12 or later
• Python 3.10 or later
• An NVIDIA GPU (Ampere or later recommended)

Quick Start

1. Install Python packages

From this directory:

pip install ".[cu12]"  # or ".[cu13]" for CUDA 13

This installs:

• helpers — shared utilities used by the Python API examples
• cuest_scf — the cuEST SCF Python library
• nvidia-cuest-cu12 (or nvidia-cuest-cu13)

2. Run the examples

C API examples — see c_examples/README.md for build instructions, then:

cd c_examples
./run_all.sh

Python API examples:

cd python_examples
./run_all.sh

cuEST SCF examples and tests:

cd cuest_scf_examples
./run_all.sh

Third-Party Notices

See THIRD_PARTY_NOTICES.txt for licenses and attributions for bundled third-party data and software dependencies.

Input Data

Shared geometry and basis set files used by all examples are in data/:

data/
  geometry/
    h2o.xyz               # Water molecule geometry
    ch2i2.xyz             # Diiodomethane geometry (ECP examples)
  basis_set/
    def2-svp.gbs          # def2-SVP basis set
    def2-universal-jkfit.gbs  # def2-universal-JKFIT auxiliary basis set
    def2-svp-ecp.gbs      # def2-SVP ECP basis set