4 ms·
Stan uses Markov chain Monte Carlo for inference. MCMC algorithms can do very well at approximating the high-dimensional integrals required to calculate the pos
by ppcsf 10y ago
Stan uses Markov chain Monte Carlo for inference. MCMC algorithms can do very well at approximating the high-dimensional integrals required to calculate the posterior; estimating 1000-dimensional distributions with MCMC methods is not unusual.
- glial 10y agoNot unusual, but the sampling can take a long time and scales poorly with the amount of data you have.