Supported Python and NumPy versions are determined according to the NEP 29 deprecation policy.

Using conda/pip#

Conda install (preferred):

conda install -c conda-forge stumpy

PyPI install with pip:

python -m pip install stumpy

From source#

To install stumpy from source, first clone the source repository:

git clone
cd stumpy

Next, you’ll need to install the necessary dependencies. For maximum performance (or if you are installing stumpy for the Apple M1 ARM-based chip), it is recommended that you install all dependencies using conda:

conda install -c conda-forge -y numpy scipy numba

Alternatively, but with lower performance, you can also install these dependencies using the requirements.txt file in the root of this repository:

python -m pip install -r requirements.txt

Once the dependencies are installed (stay inside of the stumpy directory), execute:

python -m pip install .