one of my favorites was the transitions from directed graphs interpreted as state machines to regular expressions. I did some translation back and forth in some undergrad theory classes, but never got to deal with doing things like traversal problems while the data is in regexp form. From a pure theory standpoint it offers an interesting way to use the solution of a known set of problems (graph coverage, etc) to be effective in a new space (operating on regexps). I wish I'd explored them further.

G'luck

Trinary