Hello everyone,
I want to know simple explanation of difference between warm up and samples and what happens when we do discard_adapt = false. Below you can see plots for two different conditions with and without discard_adapt. Now what I expected that in first plot I should see 250 iterations because i’m not discarding adaptive samples. Is there something wrong in my understanding?
alg = NUTS(200, 0.65)
chain_first = sample(model, alg, MCMCThreads(), 50, 1, discard_adapt = false, save_state = false)
alg = NUTS(200, 0.65)
chain_first = sample(model, alg, MCMCThreads(), 50, 1, discard_adapt = true, save_state = false)