ckuhl/gp

genetic programming in python

Genetic Programming

Define the problem and expected outputs, and a solution will be evolved to solved the defined problem.

Quickstart

Call make setup to set up the virtual environment and then make debug to watch it start churning away at an example problem.

Project Info

• requirements.txt containts the requirements
• tests in ./tests, can be run using make test
• to run the program call make run or make debug
• documentation in ./docs, can be updated using make docs

Useful Tools

python -m cProfile -s tottime gp/__main__.py profile, sorted by total time in each function.

To profile a specific function:

import pprofile
profiler = pprofile.Profile()
with profiler:
    ...

and then start it from the command line as such: pprofile --threads 0 main.py

To do

• change gene.run into a generator
  • i.e. terminate as soon as sufficient output is generated
• make generation running async
  • each gene.run is not dependant on any other
• create a custom logging formatter to colourize "WARN", etc.
• add a few utility classes to making typing simpler
  • i.e. make it harder to mess up what we're returning