Done with Undergrad, Procedural Generation, Model Checking

17 05 2008

I’m finally done with undergrad.  It feels like 4 years just blew by in the blink of an eye.  And come fall, I’ll be working, but also applying to grad schools.  What a ride.

Incidentally, I did finish my thesis.  It’s here: Type-Safe Stack Inspection for Garbage Collector Implementation.  Every time I write that title now I wish I’d cut it off as “Type-Safe Stack Inspection.”  Oh well, too late now.  I meant to post about it right after I finished, but the final edits ran into finals period, which consumed the past two weeks of my life.

On to other projects.

Currently, Brendan, Rob and I are working on building a fairly ambitious mapping tool.  We decided to use procedural modeling implement a multiple-level-of-detail real-time procedural generator for, well, planets.  The idea is that eventually we’ll start by rendering a planet with continents and oceans on screen, and the user can zoom in – to nearly any level of detail.  Moving closer to the planet gradually adds greater and greater levels of detail to the planet, laying out rivers, mountains and valleys, then cities and roads, laying out city streets and buildings, and possibly even building internals.  We think we can do it quickly and without using unreasonable amounts of memory by taking advantage of the fact that procedural generation is deterministic if you reuse the same random number generator seed.  So we can cache seeds for areas along with constraints on edges for some areas, and use negligible amounts of memory for storing enough information that you can back out or cross the planet, have the old region freed from memory, return and see the same things.  We’re aiming pretty high, so we’ll see what happens.

I also picked up a copy of a book my advisor told me about, Principles of Model Checking.  It was just released earlier this month, and it’s great so far.  I read a couple of the original LTL and CTL model checking papers (LTL: Model Checking Using Automata Theory, CTL: Automatic Verification of Finite-State Concurrent Systems Using Temporal Logic Specifications) a couple years ago in a seminar, but I didn’t have enough background in CS to really have all the details quite click for me back then (those papers were actually the first time I saw automata in CS, before I took a theory of computation class).  This book is shaping up to be a good refresher for me (also filling in gaps from what I missed the first time I looked at this material) and seems like it would be a good introduction for those new to model checking as well.  Sadly there is not yet a place to look inside it online.