3 ms·
Break the Markov Chains of Oppression: Modeling Without MCMC
- kem 10y agoNice to see someone reapproaching the question of "why MCMC?" You don't really need it to do Bayesian inference, although it might seem that way nowadays. It brings up good arguments for using MCMC, although they're not as compelling to me as the author seems to assume they are. I wish the topic were discussed more often--it would be nice to see some followups.