ClojureGP

ClojureGP is a genetic programming framework written in Clojure. Genetic programming is an evolutionary technique that automatically generates programs (in this case Clojure functions) by iteratively evolving a population of candidate solutions toward some goal. ClojureGP aims to be highly configurable while remaining concise and easy to pick up.

A GP run is represented as a lazy sequence of evolutionary generations, which fits naturally with Clojure's sequence model. This means you can work with a run using ordinary sequence operations: map statistics over it as each generation is produced, call last to get the final result, or otherwise consume it however you like.

Quick start

The simplest way to see it in action is to load one of the included examples from the repository root:

$ clojure -M -i examples/reg_exp.clj --repl

Then call run from the REPL:

user=> (reg-exp/run)

This runs a symbolic regression experiment (evolving a Clojure expression to fit a target equation) and prints per-generation fitness statistics along with a summary at the end. A step-by-step walkthrough of this example is in docs/reg_example.md.

Documentation

The docs/ directory contains:

• overview.md: how ClojureGP works: the lazy sequence model, individual representation, and the configuration approach
• reg_example.md: a walkthrough of the regression example covering all the basics, a good place to start
• config_reference.md: full reference for all configuration keys

The examples/ directory has five experiments of varying complexity, from a bare skeleton in minimal.clj up to the Santa Fe trail ant problem in ant_exp.clj.

Using ClojureGP as a library

ClojureGP is available as a git dependency for the Clojure CLI. Add the following to your deps.edn (see the Clojure guide on git libraries):

io.github.svdm/ClojureGP {:git/tag "v2.0.0" :git/sha "f867199"}

ClojureGP only requires Clojure itself. Two optional extras are available via aliases in deps.edn:

• :unc-math: example of using a third party RNG using Uncommons Maths
• :graph: example of live fitness plotting using XChart

License

Copyright (c) Stefan A. van der Meer. All rights reserved. Distributed under the Eclipse Public License 1.0, which can be found in epl-v10.html at the root of this distribution.