Yet another Push-based genetic programming system in Clojure.
Propeller is a component of several active research projects and it is subject to change as a part of these activities.
See the git commit comments for guidance with respect to recent changes.
Some documentation is available at https://lspector.github.io/propeller/, but this should be evaluated in the context of the commit messages and current source code.
If you are working in a Clojure IDE with an integrated REPL, the first
thing you may want to do is to open src/propeller/session.cljc and
evaluate the namespace declaration and the commented-out expressions
therein. These demonstrate core components of Propeller including
complete genetic programming runs. When conducting complete genetic
programming runs this way (using gp/gp), depending on your IDE you
may need to explicitly open and load the problem file before evaluating
the calls to require and gp/gp.
To run Propeller from the command line, on a genetic programming problem
that is defined within this project, you will probably want to use either
the Clojure CLI tools or
leiningen. In the examples below, the leiningen
and CLI commands are identical except that the former begin with
lein run -m, while the latter begin with clj -M -m.
To start a run use clj -M -m <namespace> or
lein run -m <namespace>, replacing <namespace>
with the actual namespace that you will find at the top of the problem file.
For example, you can run the integer-regression genetic programming problem with:
clj -M -m propeller.problems.integer-regression
or
lein run -m propeller.problems.integer-regression
Additional command-line arguments may be provided to override the default key/value pairs specified in the problem file, for example:
clj -M -m propeller.problems.integer-regression :population-size 100
or
lein run -m propeller.problems.integer-regression :population-size 100
On Unix operating systems, including MacOS, you can use something
like the following to send output both to the terminal
and to a text file (called outfile in this example):
clj -M -m propeller.problems.integer-regression | tee outfile
or
lein run -m propeller.problems.integer-regression | tee outfile
If you want to provide command line arguments that include
characters that may be interpreted by your command line shell
before they get to Clojure, then enclose those in double
quotes, like in this example that provides a non-default
value for the :variation argument, which is a clojure map
containing curly brackets that may confuse your shell:
clj -M -m propeller.problems.integer-regression :variation "{:umad 1.0}"
or
lein run -m propeller.problems.integer-regression :variation "{:umad 1.0}"
For many genetic operator hyperparameters, collections may be provided in place of single values. When this is done, a random element of the collection will be chosen (with each being equally likely) each time the operator is used. When specied at the command line, these collections will also have to be quoted, for example with :umad-rate "[0.01 0.05 0.1]" to mean that UMAD rates of 0.01, 0.05, and 0.1 can be used.
By default, Propeller will conduct many processes concurrently on multiple
cores using threads. If you want to disable this behavior (for example, during
debugging) then provide the argument :single-thread-mode with the value true.
Threads are not available in Javascript, so no processes are run concurrnetly
when Propeller is run in Clojurescript.
Run in development:
yarn
(mkdir -p target && cp assets/index.html target/)
yarn shadow-cljs watch appshadow-cljs will be installed in node_modules/ when you run yarn.
:dev-http specifies that target/ will be served at http://localhost:8080 .
After page is loaded, you may also start a REPL connected to browser with:
yarn shadow-cljs cljs-repl appOnce the REPL is loaded, load the core namespace with:
(ns propeller.core)
Calling (-main) will run the default genetic programming problem.
Propeller is an implementation of the Push programming language and the PushGP genetic programming system in Clojure, based on Tom Helmuth's little PushGP implementation propel.
For more information on Push and PushGP see http://pushlanguage.org.