hchasestevens / monkeys Goto Github PK

View Code? Open in Web Editor NEW

120.0 5.0 14.0 181 KB

A strongly-typed genetic programming framework for Python

Python 100.00%

python ipynb genetic-programming framework metaprogramming optimization ai ant-colony-optimization

monkeys's Introduction

monkeys

"If an army of monkeys were strumming on typewriters they might write all the books in the British Museum."

monkeys is a framework designed to make genetic programming in Python accessible, quick, flexible, and fun.

Get started here.

What is genetic programming?

Genetic programming algorithms are a class of evolutionary algorithms in which solutions to a problem are represented as executable tree structures - programs. In order to use genetic programming in order to solve a problem, two things must be specified:

What form(s) a potential solution can take.
A method of scoring solutions based on their meritoriousness.

Given these, a genetic programming system can perform intelligent exploration and search through the space of potential solutions, narrowing in on those programs that best solve the problem as specified. Genetic programming has achieved human-competitive results in a wide swath of domains, including:

Satellite antenna design for NASA
The creation of novel quantum computing algorithms
Evolving game AI
The automatic repair of buggy code.

monkeys to the rescue!

"Ford, there's an infinite number of monkeys outside who want to talk to us about this script for Hamlet they've worked out."

monkeys makes getting started with genetic programming painless and fun. Install monkeys by running:

pip install monkeys

monkeys uses a variant of genetic programming called "strongly-typed genetic programming" in order to allow you to quickly and easily specify how your programs should be structured.

monkeys supports Python 2.7 and 3.x.

Examples

Tutorials:

Getting started with monkeys: An introduction to the monkeys framework, which outlines the type system.
Monkeys in abstract syntax trees: using monkeys to generate Python code directly.
Debugging with monkeys: A guide for discovering and diagnosing bugs in your codebase with monkeys.

Sample usages:

Linting by example: Generating linting rules by supplying positive/negative code examples.
Solving logic puzzles with monkeys: Solving logic puzzles using the monkeys diagnostic tool.
More to come!

Contacts

Name: H. Chase Stevens
Twitter: @hchasestevens

monkeys's People

Contributors

Stargazers

Watchers

Forkers

corpsecamper barleyj suchow echo66 glavin001 rsrawat06 rubiruchi rmatam jacob414 afcarl thetradingflow hugoamaro python-repository-hub jesseventura

monkeys's Issues

How to use multiple types?

Hi,

I tried the following code, attempting to use two types :
@params(bool,float,float)
@rtype(float)
def IfCond(c,a,b):
if c==True :
return a
return b

recevied the following error:

File "twoTypes.py", line 64, in IfCond
if c==True :

TypeError: 'bool' object is not callable

I wanted my function to receive a boolean value (produced by a function with a boolean rtype, for example) and return either a or b, depending upon that value

Could anyone please correct my code, or provide a working example with a boolean-returning function, used by a float-returning function?

Thanks in advance!
Ami

Error at "Getting started" notebook

When running type_graph(), I'm getting the following error:

TypeError: object of type 'map' has no len()

assertions_as_score must not set max score for complex functions

assertions_as_score cannot determine maximum score for functions in which Assert nodes are descendants of control structures (If, While, For). In these cases, simply leave max score unset.

AST example: unable to reproduce evolved my_abs function

Running Monkeys in abstract syntax trees in Jupyter (on Python 2.7), I was unable to reproduce the result in the original notebook.

from monkeys.tools.display import type_graph
type_graph()

The code above generated the exact same type graph as shown in the original notebook. So far so good.
However, when calling optimize

best_abs = optimize(score, iterations=25)

it never managed to output a useful abs function. Most of the time the output was either

def my_abs(x):
    return int(x)

def my_abs(x):
    return int(-x)

After dumping all the generated intermediary ASTs, I have noticed that the conditional if_else never once appeared in any of the generated trees. What might be the reason for this?

Another Error at "Getting started" notebook

When running

best_equation = optimize(score)

I'm getting the error:

TypeError: bad operand type for unary -: 'NoneType'

I can post the whole traceback if it helps.