##########################
Pandas and xarray in pyCIF
##########################

.. role:: bash(code)
      :language: bash


pyCIF uses `pandas <http://pandas.pydata.org/>`__ and `xarray <http://xarray.pydata.org/en/stable/>`__
to store data in a convenient format.
Time series are stored as :bash:`pandas.DataFrame` and higher dimension data
sets as :bash:`xarray.Dataset` or :bash:`xarray.DataArray`.

Pandas is mostly used for its convenient and efficient way of handling dates and reindexing.
Xarray is only used as the default format to share data between functions in pyCIF.

.. note::

    We do not use all the potential of pandas and xarray for two main reasons:

        - to avoid compatibility issues between versions of the two modules
        - no adjoint is coded for the methods of the two modules, so they are replaced by custom equivalents

Intermediate data in pyCIF have a standardized format as detailed
:doc:`here </documentation/plugins/transforms/index>`.
It is possible to dump intermediate temporary data for debugging purposes
using the option :bash:`save_debug` in the observation operator: see details
:doc:`here </documentation/plugins/obsoperators/standard>`.