Solving Wordbrain

Robert Stone presented this talk as a practice run for using this presentation for a conference. He first described both depth-first search and breadth-first search. This ensured that everyone was familiar with the fundamentals for the solution techniques we will be exploring.

Then, he described the game WordBrain and gave examples to clarify how the game is played. The main points of the game are a set of letters, a given number of words to find, and the number of letters in each of the target words. There are no time limits. Hints can be requested. Unfortnately, we find that there may be more than one set of words that meet the criteria supplied, but only one answer.

Robert showed how he used a standard search technique to generate all possible word combinations from the gameboard. These potential solutions are filtered to generate only actual solutions. The resulting solutions are formatted and displayed to simplify testing the potential answers by entering them into the program.

He showed the code and explored many of the design decisions that he had made. Since the code had run slower than he expected, he had turned to Devel::NYTProf to determine the bottlenecks in the code. These measurements lead to potential optimizations, which he discussed.

The meeting quickly turned into a discussion of his coding decisions with different suggestions on speed improvements. We discussed different approaches to limit the depth of recursion and ways to more quickly recognize dead-ends. The group was quite engaged in suggestions to help with performance.

Robert has provided his materials for others to study, including

We had 11 people attending this month. As always, we'd like to thank HostGator, LLC for providing the meeting space and food for the group.