Check out my new paper
Introduction to Judea Pearl’s Do-Calculus (arXiv:1305.5506)
It’s a purely pedagogical paper with no new results. Its goal is to give a fairly self-contained introduction to Judea Pearl’s Do-Calculus, including proofs of his 3 rules.
As I have mentioned in previous posts, lately I’ve been obsessed with Judea Pearl’s causality theory. Eventually, I want to adapt it to quantum mechanics. I already have some nice ideas about how to do that, but this paper has the much more modest goal of introducing the reader to one of Pearl’s early, landmark papers about causality, R218-B, published in 1995.
This paper of mine is my way of dipping my toes into the strong river of knowledge that is Pearl’s causality theory. In order to understand Pearl’s work better, I forced myself to give a mini-lecture about it to a small group of my local friends. Then I turned those mad ravings into something that resembles the thoughts of a sane person and wrote that up.
While writing this paper, I kept on thinking that the nodes of a Bayesian network resemble cows, a whole herd of them enclosed in several corrals. From this perspective, the d-separation theorem began to make a lot of sense to me as a statement about bovine traffic between the corrals. That’s why I titled this blog post “do-calculus for ranchers” (also for cows, cowhands and rustlers). Okay, my paper mentions the cow analogy only very briefly. It’s mentioned in only one or two sentences in the whole 16 page paper. So you might still enjoy the paper, even if you aren’t too interested in my cow application.