Getting started
===============

Requirements
------------

``ps_eor`` supports Python 3.10 and later. Its functionality is split between
the core package and optional dependencies for ML-GPR inference.

Core installation
-----------------

Install the data handling, flagging, power-spectrum, spherical-analysis,
simulation, and command-line functionality with:

.. code-block:: console

   $ pip install ps-eor

This also installs the ``pstool`` command. The installation can be checked
with:

.. code-block:: console

   $ python -c "import ps_eor; print(ps_eor.__version__)"
   $ pstool --help

ML-GPR installation
-------------------

Install the Gaussian-process backend and its standard inference methods,
including MCMC, MAP, Dynesty nested sampling, and Pyro NUTS, with:

.. code-block:: console

   $ pip install "ps-eor[ml-gpr]"

The high-level ML-GPR interface can then be imported from the package:

.. code-block:: console

   $ python -c "from ps_eor.ml_gpr import MLGPRForegroundFitter"

UltraNest is a separate optional dependency:

.. code-block:: console

   $ pip install "ps-eor[ml-gpr,ultranest]"

Installing from source
----------------------

For an editable development installation:

.. code-block:: console

   $ git clone https://gitlab.com/flomertens/ps_eor.git
   $ cd ps_eor
   $ pip install -e ".[ml-gpr,dev]"

Documentation sections
----------------------

The :doc:`topic-based user guide <user_guide/index>` covers data handling,
power spectra, foreground separation, ML-GPR, simulations, and spherical
analysis. The :doc:`command-line documentation <cli/index>` describes every
``pstool`` command. Exact classes, functions, parameters, and return values
are listed in the :doc:`API reference <api/index>`.