= numerics =

/python experiments in numerics + plotting/

== What is numerics? ==

Over the years and at various companies, I have found the need to
perform quick pipelines of calculations on mathematical data (usually
numbers). The `numerics`
[python package](https://pypi.python.org/pypi/numerics)
is an assemblege of tools and methods needed to scratch this issue.
It is neither scoped nor complete.

`numerics` largely deals in the currency of Comma Separated Value (CSV)
files.

== Dependencies ==

The backbone of `numerics` are its dependencies, which I have either
used + liked or just used:

- [NumPy](http://www.numpy.org/)
- [pandas](http://pandas.pydata.org/) : python data analysis
- [matplotlib](https://matplotlib.org/)
- [bokeh](http://bokeh.pydata.org/en/latest/) : because I couldn't
  decide which visualization library I *actually* wanted

These dependencies are (mostly) listed in `numerics`'s
[setup.py](http://k0s.org/hg/numerics/file/tip/setup.py)


== Installation ==

Because `numerics` is largely a bunch of experiments, it requires
several packages.  Some of them *may* be challenging to install.
Here's how, as best I know.

I have found it necessary to `pip install bokeh` before
the rest of numerics...otherwise a version conflict re `pytz` may result.

Then one may do `python setup.py develop` in the `numerics` directory

=== OSX ===

For [matplotlib](https://matplotlib.org/):

http://superuser.com/questions/242190/how-to-install-matplotlib-on-os-x

You may also have to manually refresh the font cache:

    cd ~/.matplotlib/
    fc-list  # Should take a couple minutes.

(From: http://stackoverflow.com/questions/17490444/import-matplotlib-pyplot-hangs )


== Testing ==

Use `pytest` + `tox` per usual

== Libraries ==

The following libraries, listed by language, may be worth investigating.


=== JavaScript ===

- http://metricsgraphicsjs.org/
- https://github.com/mikedewar/d3py : python -> d3 bridge


== Links ==

- https://scipy.org/

----

Jeff Hammel

http://k0s.org/hg/numerics