This is a package to approximate the response of radiation detectors to a gamma source in a heterogeneous environment. Read any of the papers I've written recently for details, because I doubt you got here by accident.
Gefry reads an input deck that describes the model geometry. This input deck is written in YAML, a simple markup format. It's pretty self explanatory and you shouldn't need to change stuff much.
All geometry is represented as polygons (curves have to be approximated). These polygons are described by lists of vertices. Each polygon also has an associated material, with the materials described in their own section. Each material is characterized by a number density and a microscopic cross section. If you want to assign macroscopic cross sections then set number density to 1.0. Materials can be reused. There is also an "interstitial material", which is the material of everything that is not one of the defined regions (i.e., the air).
NOTE the order of the regions is important! If you want to run a problem with varying cross sections, the order of the cross sections you provide must match the order of polygons in the input file!
This way to use this is read an input deck, which returns an object
that can be called to evaluate the counts seen by the detector. The arguments
you provide vary depending on what kind of problem you are running,
currently either Simple_Problem (fixed cross sections, you just provide
the source location and intensity) or Perturbable_XS_Problem (variable
cross sections, you provide source location and intensity, plus a list
of cross sections).
The biggest thing to know is that this version of the code is purely deterministic. It only evaluates the ray tracing model for the detector network and does not include any statistical effects or background. You add those on your own, however you desire.
I develop and use this code on Python 3. NumPy support for Python 2.7 is being dropped at the end of 2018, so Python 2.7 is dead and it's time to move on. The code should be compatible, but I don't really test it on Python 2 so you might encounter bugs. If you do let me know and I will try to fix them unless it's really hopeless.