.. ## Copyright (c) Lawrence Livermore National Security, LLC and other
.. ## Axom Project Contributors. See top-level LICENSE and COPYRIGHT
.. ## files for dates and other details.
.. ##
.. ## SPDX-License-Identifier: (BSD-3-Clause)

.. _output-format-label:

=============
HDF5 and JSON
=============

Sina's C++ and Python Code supports loading both JSON and HDF5 files as Documents,
and saving Documents to either file type.

Both options offer differing strengths and weaknesses when compared to each other,
the following data on the size and speed has been compiled to give a better idea
of which file type fits your requirements.

------------------------------
Why You Should Use JSON
------------------------------

* JSON is more universally parsable
* JSON code is human readable allowing you more, easier flexibility with searching for relevant data.  This is most applicable with smaller files since large amounts of data or curve sets quickly become more efficient to navigate digitally
* JSON outperforms HDF5 in speed and size efficiency when dealing with non-curve set data and outperforms at smaller curve set sizes (just before :math:`10^2` Curve Sets for Size and around :math:`10^3.25` for Speed)

------------------------------
Why You Should Use HDF5
------------------------------

* HDF5 offers better size and speed efficiency when dealing with larger files/curve sets and only outperforms more dramatically as size increases
* Hierarchical structure leads to being 2.5x more size efficient and 5x faster at our largest tested files

------------------------------
JSON vs HDF5 Size
------------------------------

.. image:: ../imgs/JSON_vs_HDF5_size.png
    :alt: JSON Size / HDF5 Size when dealing with 10 - :math:`10^4` curves with 10 - :math:`10^4` items


------------------------------
JSON vs HDF5 Speed
------------------------------

.. image:: ../imgs/JSON_vs_HDF5_speed.png
    :alt: JSON Size / HDF5 Size when dealing with 10 - :math:`10^4` curves with 10 - :math:`10^4` items