Exports a single method, taking a (set of) rng[s], a problem adhering to the LogDensityProblems.jl interface, and optional keyword arguments:
adaptive_warmup_mcmc(
rng[s], problem;
n_draws=1000,
target_acceptance_rate=.8,
max_tree_depth=10,
progress=nothing, options...
)With progress=Term.ProgressBar, prints a progress bar which is prettier than most.
The result is a NamedTuple, with its posterior_position field containing the posterior draws.
Results should come in faster than with "standard" methods, and should often be better.