Installation

pytket-custatevec requires Python 3.10+ and a Linux machine with an NVIDIA GPU (Compute Capability 7.0+).

Prerequisites

Ensure your machine has the NVIDIA Drivers installed and working (check with nvidia-smi).

You generally do not need to install the full CUDA Toolkit system package, as the Python packages below include the necessary runtime libraries.

1. Install Dependencies

Select your package manager to install the required libraries.

Conda (Recommended)Pip

We recommend Conda because it reliably manages the system-level CUDA libraries alongside Python packages.

# 1. Install NVIDIA libraries
conda install -c conda-forge cuquantum-python

# 2. Install the package
pip install pytket-custatevec

You can install everything via pip. This will automatically download the binary wheels for cuquantum and cupy.

pip install pytket-custatevec

Linux Requirements

If you see ImportError: libcustatevec.so not found, ensure you have a recent version of pip (pip install --upgrade pip) and a compatible Linux distribution (glibc 2.17+).

2. Verify Installation

You can check if the backend loads correctly by running this one-liner:

python -c "from pytket.extensions.custatevec import CuStateVecStateBackend; print('✅ Backend loaded successfully')"