A Gentle Introduction to Markov Chain Monte Carlo (2012)

theclevermachine.wordpress.com

105 points·phonebucket··4 comments

Not sure how "gentle" that was. Maybe it's just my own knowledge base, but there seemed to be a lot of mathematical symbol gobbledygook. If that's your thing, I get it, but if you are fluent in arcane math symbolism, you probably don't need a gentle introduction. And if those types of formal equations look to you like they are an alien language, they don't bring clarity.

In the same vein of "speaking the language of your audience" I'd think that Python would be a more readable and commonly understood language for examples.

I'm not sure where you are coming from. I recognize the sentiment, but I don't feel it for this one. The mathematical notation used here is commonly taught in high schools.

The basics of Probability Theory, and Integrals are prerequisite to the introduction. p(x|y) means the chance that x happens if y happened - that notation is more common in the US I believe.

There even is accompanying code. MatLab is admittedly less common than python, but the code looks nearly the same as it would in NumPy.

This might seem intimidating, but is conceptually rather easy.

> This might seem intimidating, but is conceptually rather easy.

Yes, that is exactly the problem. This article makes its subject unnecessarily confusing.

What are the best ways, if any, to test for the convergence (or pseudo-convergence) of the Markov chain to the equilibrium distribution?

Reply on news.ycombinator.com