* Cleaner code layout (no mixing of guards and if-statements, etc.)

* Caching the size of the calculated independent node set rather than
  finding the length each time.