AIMA Python Code

This file gives an overview of the Python code for the algorithms in the textbook AI: A Modern Approach. The code is Copyright (c) 2002 by Peter Norvig and is offered free of charge for your use. As you may know, the textbook presents algorithms in pseudo-code format; as a supplement we provide this Python code as well as Lisp code. The intent is to implement all the algorithms in both languages, so that you can choose whichever language you prefer. As yet neither implementation is complete, but the Lisp version is closer.

Installation Instructions

Here is how to download the files and make them ready for use. You only need to do this once, and if you are taking a course, your instructor may have set this up for you.

Download the file aima-python.zip
Unzip it. Your browser may do this automatically, or you can give the command "unzip aima-python.zip" or drag the file to your zip program icon.
You must have Python (version 2.2 or later) installed. If you don't have it yet, you can download it.
Make sure the files are in your module search path. You do this either by always starting Python from the directory where you keep the files, or by editing the environment variable PYTHONPATH.

User's Guide

Once you have the files installed, you can use them in several ways.

Read the code. This can enhance your understanding of the algorithms, and clarify parts that were not spelled out in the book's pseudo-code.
Run the documented examples that are interspersed in the code by running "python docex.py *.py" from the command line. After a minute or so you should see printed "<Test: passed all n>", for some number n. If you want, you can also look at the log file docex-log.html for details on each test. From within Python, you can also import the individual module(s) that you want to test and then do "import docex" and "docex.Test([modules])".
Run the existing code on your own data. For the module(s) you want, do "import module" and then run the functions you want on the data you want.
Experiment with extending the code.

Code File Summary

The following table lists the code files (modules), the chapters in the book to which they refer, the number of lines of code in each file, and a brief description of each file. You can also browse through the directory norvig.com/python/ to see all the files, or look at the example output log and link from there to the files.

Chapter Module Lines Description
1-2 agents 485 Implement Agents and Environments (Chapters 1-2).
3-4 search 470 Search (Chapters 3-4)
5 csp 435 CSP (Constraint Satisfaction Problems) problems and solvers. (Chapter 5).
6 games 285 Games, or Adversarial Search. (Chapters 6)
7,9 logic 335 Representations and Inference for Logic (Chapters 7,9)
18-21 learning 579 Learn to estimate functions from examples. (Chapters 18-21)
22 nlp 155 A chart parser and some grammars. (Chapter 22)
23 text 137 Statistical Language Processing tools. Not all working yet. (Chapter 23)
docex 205 A framework for running unit test examples, written in docstrings.
py2html 126 Pretty-print Python code to colorized, hyperlinked html.
utils 604 Provide some widely useful utilities. Safe for "from utils import *".
3816

Chapter	Module	Lines	Description
1-2	agents	485	Implement Agents and Environments (Chapters 1-2).
3-4	search	470	Search (Chapters 3-4)
5	csp	435	CSP (Constraint Satisfaction Problems) problems and solvers. (Chapter 5).
6	games	285	Games, or Adversarial Search. (Chapters 6)
7,9	logic	335	Representations and Inference for Logic (Chapters 7,9)
18-21	learning	579	Learn to estimate functions from examples. (Chapters 18-21)
22	nlp	155	A chart parser and some grammars. (Chapter 22)
23	text	137	Statistical Language Processing tools. Not all working yet. (Chapter 23)
	docex	205	A framework for running unit test examples, written in docstrings.
	py2html	126	Pretty-print Python code to colorized, hyperlinked html.
	utils	604	Provide some widely useful utilities. Safe for "from utils import *".
		3816

Developer's Guide

If you'd like to contribute to this project, we could really use the help. Read the guidelines and then let me know what you'd like to contribute.

Peter Norvig