Chains are independent sequences of sampling that explore the posterior distribution of model parameters. And yes, that is a Markov chain. The Markov Chain Monte Carlo (MCMC) algorithm constructs independent "Markov chains" to perform a Monte Carlo approximation for the estimated parameters (only to combine the samples from all chains at the end). In the case of `brms`, [[No U-Turn Sampler - NUTS]] is used as an adaptive form of Hamiltonian Monte Carlo (HMC) sampling. > [!info]- Last updated: October 31, 2024