Empirical Hamiltonian Monte Carlo

Source: R/ehmc.R

Perform Empirical Hamiltonian Monte Carlo

ehmc_helper(
  log_posterior,
  gradient,
  n_steps,
  init_parameters,
  iters,
  warmup_iters = iters/2,
  logging = TRUE
)

Arguments

log_posterior

a function for the unnormalised log-posterior from parameters -> log-likelihood
gradient

the gradient of the log_posterior
n_steps

the initial number of leapfrog steps
init_parameters

a named vector containing the initial parameters
iters

the number of iterations

Value

a matrix of iterations from the HMC algorithm representing draws from the posterior distribution