Information on reproducing:

approximately, one python file in the root of this folder corresponds to one type of experiment. sometimes different variations are available.
the tests.py file is an exemption (which was a bad design choice in hindsight): it contains smaller setups, each in one function.

contents of this folders are:
  - make-plots.xsh (using the xonsh shell language) generates all plots required for the master's thesis.

  - 'pickle': experimental data and intermediate results are cached in this folder using the python pickle library (not the best choice...). more or
    less every run (generating usually one plot) creates one folder in 'pickle'. runs from tests.py are stored in the subfolder 'pickle/ex'. the kwargs
    files store the kwargs used in the run (see EvalNode documentation for more). they can be read using some code in tests.py (without sensible interface)
    pickled data that isn't used in the plots for the thesis is stored in 'pickle-old'

  - 'figures': the exported figures

  - docs for the code are in 'code/docs-build'. they are not complete, but some parts are quite extensively documented.
  
  - 'src': the source code, cpp and python/cython parts.
  
  - the Makefile contains several useful targets:
      'make all': c++ library, cython bindings, linking
      'make docs'; compile documentation with sphinx and doxygen

known issues:
  - the EvalNode stuff is not perfect, but quite good i think (for alpha stage stuff)
  - you need to add the 'code/lib' folder to the $LD_LIBRARY_PATH environment variable (at least for some runs) because the cython library is dynamically linked
  - some directories for build output or for example generated figures are not created automatically, which leads to errors. just create them manually.

versions of the software stack I used for this:
  Archlinux OS (up to date as of 2022-06-01)
    for more on the software tools see github.com/pguenth/config (or whereever this may be in the future)

  Python 3.10.4
  g++ 12.1.0

  Python libs:
    Proplot 0.9.5.post301
    Matplotlib 3.5.2
    eigency 1.77
    cython 0.29.30
    numpy 1.22.3
    scipy 1.8.1
    astropy 5.0.1
    agnpy 0.1.8
    cysignals 1.11.2

  Python optional (but possibly code needs to be commented out):
    networkx 2.7.1 [for creating visualisations of the evaluation tree]

  C++ libs:
    Eigen 3.4.0
    boost 1.78.0

  for the docs:
    python-sphinx 5.0.1
    python-sphinx_rtd_theme 1.0.0
    doxygen 1.9.3

run stuff in the main directory of the source code and set
$PYTHONPATH=".:$PYTHONPATH" to be able to run for example
python runs/foo.py