How to install unxt

How to install unxt#

unxt is on PyPI and supports the Python versions listed on its PyPI page.

pip install unxt

uv add unxt

pip install git+https://github.com/GalacticDynamics/unxt.git

git clone https://github.com/GalacticDynamics/unxt.git
cd unxt
pip install -e .  # editable mode

Installing unxt brings in JAX, quax, quaxed and astropy, which is the unit backend.

Install an add-on package#

The unxts.* packages are separate distributions. Install each through the matching extra rather than by name, so that it and unxt are resolved together as a compatible set:

Extra

Package

Adds

unxt[parametric]

unxts.parametric

ParametricQuantity — dimension in the type, runtime dimension checking

unxt[linalg]

unxts.linalg

QuantityMatrix — per-element units for Jacobians and metrics

unxt[interop-gala]

unxts.interop.gala

Unit-system conversion with gala

unxt[interop-mpl]

unxts.interop.matplotlib

Plotting quantities with matplotlib

unxt[interop-xarray]

unxts.interop.xarray

A .unxt accessor for DataArray and Dataset

 
pip install "unxt[parametric]"

Two packages have no extra of their own — install them by name:

Package

Adds

unxts.hypothesis

hypothesis strategies for property-based testing

unxts.api

The abstract dispatch API. Already a dependency of unxt; install it directly only if you are writing against the API without unxt

 
pip install unxts.hypothesis

Everything at once#

unxt[workspace] installs every package above, unxts.hypothesis and unxts.api included. unxt[all] is workspace plus backend-astropy, which pins the astropy unit backend explicitly:

pip install "unxt[all]"

Each package has its own documentation under Packages in the sidebar.

Check it worked#

>>> import unxt as u
>>> u.Q(1.0, "m")
Quantity(Array(1., dtype=float32...), unit='m')

If that prints a quantity, you are ready for Your first calculation with units.

See also#