Implementation Notes for CSC 3400

This document is a running commentary on the implementation of the Python code for Russell and Norvig's AIMA text to a PyCharm environment, for teaching CSC 3400 at Belmont University. It is intended primarily for students in the course, but may also be useful for instructors in similar courses.

My background: I have taught AI with a variety of textbooks over 25 years, and started using the 2nd edition of AIMA back when it was scheme-based. I switched to Java eight years ago, because we teach Java in the programming I/II sequence at Belmont. I am switching to Python this year, using PyCharm-EDU as the development platform.

Tools

• PyCharm installation was uneventful: Google, download, install. So far, PyCharm Edu seems adequate for this course.
• I used git to clone the AIMA code from https://github.com/aimacode/aima-python.git
• I launched PyCharm, opened the aima-python directory as a new project (as described in https://www.jetbrains.com/help/pycharm/2016.1/importing-project-from-existing-source-code.html).
• I had been using python 2.7, but aima-python requires python 3. I duly installed it and changed the project interpreter to python 3.5.
• I found that code and data are separate, so I used git to clone the AIMA data from https://github.com/aimacode/aima-data.git, then linked the aima-data directory into the aima-code tree.

First steps

• AIMA has a tests folder in aima-code, and PyCharm allowed me to right-click the folder and run all 102 tests, which passed. Whew!
• Better yet, I set a break point in the best_first_graph_search method of search.py, right-clicked test-search.py file in the tests folder, and launched the debugger; PyCharm sped to the breakpoint, where I could watch Arad give way to Zerind, then Timisoara, etc. Sweet!

Agents

• The Agent, Environment, Percept and Actuator classes are all crammed into agents.py