.. highlight:: shell

============
Installation
============

Requirements
============

.. list-table::
    :widths: 20 25 55
    :header-rows: 1

    * - Module
      - Version
      - 
    * - Python_
      - >= 3.9
      - Other versions of Python (>3.6) might be supported. Python version 3.10 and above are now supported.
    * - Numpy_
      - >= 1.21.0
      - Older versions of numpy may be compatible.
    * - Scipy_
      - >= 1.8.0
      - 
    * - `MKL`_
      - >= 2022.0.1 
      - Intel® Math Kernel Library (installed with ``mkl-service``) provides the BLAS libraries. It is recommended that ``mkl-service`` is installed through conda
    * - Aesara_
      - >= 2.4.0
      - Latest version of Aesara can be downloaded from `<https://github.com/aesara-devs/aesara>`_

.. _Python: https://www.python.org/
.. _Numpy: https://numpy.org/
.. _Scipy: https://scipy.org/
.. _MKL: https://www.intel.com/content/www/us/en/develop/documentation/get-started-with-mkl-for-dpcpp/top.html
.. _Aesara: https://aesara.readthedocs.io/en/latest/index.html

.. _conda install:

Installing through Conda (miniconda)
====================================

Download and install `Miniconda <https://docs.conda.io/en/latest/miniconda.html>`_.

Run the following command to install Conda and the required Conda dependencies:

You will need `Miniconda <https://docs.conda.io/en/latest/miniconda.html>`_ (Full Anaconda works fine, but **miniconda** is recommmended for a minimal installation). Ensure that Conda is using at least **Python 3.9**.

Once Conda is installed, install the required dependencies from conda via:

**Windows**

.. code-block:: console

    $ conda install mkl-service conda-forge::cxx-compiler conda-forge::m2w64-toolchain -y

**Mac OS X**

.. code-block:: console

    $ conda install mkl-service Clang

**Linux**

.. code-block:: console

    $ conda install mkl-service conda-forge::cxx-compiler


Stable release installation
---------------------------

To install the package, run this command in your terminal to download and install the latest branch of ``PyCMTensor`` from `PyPI <https://pypi.org/project/pycmtensor/>`_  after you have installed the :ref:`Conda dependencies <conda install>`:

.. code-block:: console

    pip install pycmtensor -U

This is the preferred method to install ``PyCMTensor``, as it will always install the most recent stable release.

**Optional**: Alternatively, if you want the development version from the Github repository:

.. code-block:: console

    pip install git+https://github.com/mwong009/pycmtensor.git@develop -U

----

Development
===========

The sources for PyCMTensor can be downloaded from the `Github repo`_.

.. code-block:: console

    git clone git://github.com/mwong009/pycmtensor

To set up ``PyCMTensor`` in a local development environment, you need to set up a virtual environment and install the project requirements. Follow the instructions to install Conda (miniconda), then start a new virtual environment with the provided ``environment_<your OS>.yml`` file.

For example in windows:

.. code-block:: console

    conda env create -f environment_windows.yml

Next, activate the virtual environment and install poetry via ``pip``.

.. code-block:: console

    conda activate pycmtensor-dev
    pip install poetry

Lastly, install the project and development dependencies

.. code-block:: console

    poetry install -E dev

.. _Github repo: https://github.com/mwong009/pycmtensor