As I have reduced the number of projects I am working on in my day job (down from ten to just… less than ten) I have compensated by raising the number of projects I am working on at home. These range from cleaning up the flowerbeds to updating and revising the Markov chain computations book. The term rewriter has, again, moved to the back of the queue.
I like to have a lot going on, but I also like to focus on one thing for an extended period. The problem is that I have a lot of ideas, and very seldom does anyone–including me–step in to help course correct.
I have dug in a bit and I am focused in my not-work time on two things: cleaning up our flower beds, which is a monumental task that no sane person would try without first resorting to fire, and revising the Markov chain computations text. The latter is my “fun” project, and I have needed to do this almost since the moment it was originally published.
The Markov chain project has required me to do two things:
- Learn Julia. I originally thought I would rewrite the algorithms in the book (they were originally coded for Scilab) in R, and then I figured Python would be a good choice. I have since settled on Julia for the algorithms in the text itself, with an appendix for the algorithms in Python. I have to say… I really, really like Julia. I am starting to see why everyone else does.
- Re-learn a lot of computational mathematics. I remembered I wanted to include a section on significant digits from the computations, and so I needed the condition number… but the structure of the book means I need to explain it, and it’s been a long time since I thought about all this. This part is both fun, and frustrating.
The neat thing about this effort is that I get to include stuff that was not in the original. There have been new results (better derivations of some of my prior results by other folk) and some old results that were never properly published (the model segmentation analysis method for which I received my first patent). Along the way I hope to implement a nice, open source TML parser to build the matrices for all this.
It’s a lot of fun, and working on this helps me avoid pulling grass our of our flowerbed. Which I really need to do right now…