aima-python: Structure of the Project

Python code for the book Artificial Intelligence: A Modern Approach. When complete, this project will cover all the major topics in the book, for each topic, such as logic, we will have the following Python 3.5 files:

• logic.py: Implementations of all the pseudocode algorithms in the book.
• logic_test.py: A lightweight test suite, using assert statements, designed for use with py.test.
• logic.ipynb: A Jupyter notebook, with examples of usage. Does a from logic import * to get the code.

Style Guide

There are a few style rules that are unique to this project:

• The first rule is that the code should correspond directly to the pseudocode in the book. When possible this will be almost one-to-one, just allowing for the syntactic differences between Python and pseudocode, and for different library functions.
• Don't make a function more complicated than the pseudocode in the book, even if the complication would add a nice feature, or give an efficiency gain. Instead, remain faithful to the pseudocode, and if you must, add a new function (not in the book) with the added feature.
• I use functional programming (functions with no side effects) in many cases, but not exclusively (sometimes classes and/or functions with side effects are used). Let the book's pseudocode be the guide.

Beyond the above rules, we use Pep 8, with a few minor exceptions:

• I'm not too worried about an occasional line longer than 79 characters.

• You don't need two spaces after a sentence-ending period.

• Strunk and White is not a good guide for English.

• I prefer more concise docstrings; I don't follow Pep 257.

• Not all constants have to be UPPERCASE.

• Pep 484 type annotations are allowed but not required. If your parameter name is already suggestive of the name of a type, such as url below, then i don't think the type annotation is useful. Return type annotations, such as -> None below, can be very useful.

    def retry(url: Url) -> None: 
  

Language Choice

Are we right to concentrate on Java and Python versions of the code? I think so; both languages are popular; Java is fast enough for our purposes, and has reasonable type declarations (but can be verbose); Python is popular and has a very direct mapping to the pseudocode in the book (ut lacks type declarations and can be solw). The TIOBE Index says the top five most popular languages are:

    Java, C, C++, C#, Python

So it might be reasonable to also support C++/C# at some point in the future. It might also be reasonable to support a language that combines the terse readability of Python with the type safety and speed of Java; perhaps Go or Julia. And finally, Javascript is the language of the browser; it would be nice to have code that runs in the browser, in Javascript or a variant such as Typescript.

What languages are instructors recommending for their AI class?
To get an approximate idea, I gave the query norvig russell "Modern Approach" along with the names of various languages and looked at the estimated counts of results on various dates. However, I don't have much confidence in these figures...

Language	2004	2005	2007	2010	2016
none	8,080	20,100	75,200	150,000	132,000
java	1,990	4,930	44,200	37,000	50,000
c++	875	1,820	35,300	105,000	35,000
lisp	844	974	30,100	19,000	14,000
prolog	789	2,010	23,200	17,000	16,000
python	785	1,240	18,400	11,000	12,000