Using the SantaFe Ant Trail Problem as a data souce while I work through Sutton's Reinforcement Learning book.

SantaFe Ant Trail is a well know test problem for genetic algorithms https://en.wikipedia.org/wiki/Santa_Fe_Trail_problem

This is an attempt to first solve it using a reinforcement neural network in Python http://karpathy.github.io/2016/05/31/rl/ This is in Python and Numpy to make it easier to understand code and use as a toy problem

Then made a second pass on the problem with Python and Theano

Making a another pass his time with TensorFlow using this tutorial https://medium.com/emergent-future/simple-reinforcement-learning-with-tensorflow-part-0-q-learning-with-tables-and-neural-networks-d195264329d0#.72rnep32m

Mostly this is an exercise to get familiar with various deep learning libraries.