cutensornet-cu13

What it is and what it does

cuTensorNet is NVIDIA's GPU-accelerated library for tensor network computations, part of the cuQuantum SDK. It provides both C and Python APIs to perform high-performance tensor operations on NVIDIA CUDA devices, targeting quantum computing and scientific computing applications. The library is designed to exploit GPU parallelism for tensor contractions and network manipulations that would be slow on CPU.

The package is a thin wrapper around compiled CUDA kernels and requires CUDA 13 and the cutensor-cu13 dependency to function. It is actively maintained by NVIDIA, with wheels pre-built for Linux x86_64 and aarch64 architectures. Users working with tensor networks in quantum simulation, physics, or machine learning may access it either directly via this package or through the higher-level cuQuantum Python interface.

Use it for:

• Accelerate tensor network contractions in quantum circuit simulations on GPU hardware.
• Perform high-performance tensor operations for physics simulations and scientific computing.
• Build quantum algorithm implementations that leverage GPU acceleration via NVIDIA CUDA.
• Integrate GPU-accelerated tensor computations into Python-based quantum computing workflows.
• Optimize tensor network algorithms for research in quantum information and condensed matter physics.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

cuTensorNet is a GPU-accelerated library for tensor network computations, providing high-performance C and Python APIs for quantum and scientific computing workloads on NVIDIA CUDA hardware.

Yes, if you have CUDA 13 hardware and need GPU-accelerated tensor network computations for quantum or scientific computing. The package is actively maintained, has no known vulnerabilities, and is backed by NVIDIA. However, the unclear proprietary license and lack of specified Python version support warrant review before production deployment. Not suitable if you lack NVIDIA GPU hardware or CUDA 13 support.

Install

cutensornet-cu13 on PyPI

pip install cutensornet-cu13

uv add cutensornet-cu13

poetry add cutensornet-cu13

Installing cutensornet-cu13

Before you install

Medium install friction: requires CUDA 13 and the cutensor-cu13 runtime dependency; wheels are pre-built for x86_64 and aarch64 Linux. Package is actively maintained with recent releases.

License in practice

Licensed under NVIDIA Proprietary Software with unclear license treatment—review NVIDIA's terms before use in commercial or redistributed projects.

Quickstart

pip install cutensornet-cu13

import cutensornet as ctn
# Access tensor network APIs via the imported module

Requires CUDA 13 and the cutensor-cu13 runtime dependency installed; GPU hardware and NVIDIA drivers must be present.

Verify before relying

• Python version support is unspecified in the fact sheet; confirm compatibility with your target Python version.
• Whether the package can be used standalone or requires cuQuantum Python for practical tensor network workflows.
• Specific tensor network operations and API surface available in version 2.13.0.

Package facts

Evidence: cutensornet_cu13-2.13.0-py3-none-manylinux2014_aarch64.whl; cutensornet_cu13-2.13.0-py3-none-manylinux2014_x86_64.whl

Keywords: cuda, nvidia, tensor network, high-performance computing, quantum computing

Tags

More Scientific/Engineering packages