mcmc_sampling: MCMC Sampling
In mdkarcher/phylodyn: Statistical Tools for Phylodynamics

MCMC Sampling

mcmc_sampling(dataset, alg, nsamp, nburnin = 0, nsubsamp = 1,
  ngrid = 100, nugget = "1,1", prec_alpha = 0.01, prec_beta = 0.01,
  TrjL = NULL, Nleap = NULL, szkappa = NULL, rand_leap = NULL,
  f_init = rep(1, ngrid - 1), kappa = 1, covariates = NULL,
  power_covariates = NULL, betas = rep(0, 2 + length(covariates)),
  pow_betas = rep(0, length(power_covariates)), samp_alg = "none",
  kappa_alg = "gibbs", beta_vars = rep(100, length(betas)),
  pow_beta_vars = rep(100, length(pow_betas)), printevery = 100,
  first_elem_prec = 0.01)

`dataset`	`phylo` object or list containing vectors of coalescent times `coal_times`, sampling times `samp_times`, and number sampled per sampling time `n_sampled`.
`alg`	string selecting which MCMC sampler to use. Options are "HMC", "splitHMC", "MALA", "aMALA", and "ESS".
`nsamp`	integer number of MCMC steps to compute.
`nburnin`	integer number of MCMC steps to discard as burn-in.
`nsubsamp`	integer after burn-in, how often to record a step to the output.
`ngrid`	integer number of grid point in the latent field.
`nugget`	string selecting which "nugget" adjustment to apply to the precision matrix to make it full-rank. Options are '1,1' for an adjustment to the first element, 'diag' for an adjustment to the entire main diagonal, or 'none' which may result in a non-full-rank precision matrix.
`prec_alpha, prec_beta`	numeric hyperparameters for the prior on precision.
`TrjL`	numeric tuning parameter.
`Nleap`	integer tuning parameter.
`szkappa`	numeric tuning parameter.
`rand_leap`	logical tuning parameter.
`f_init`	numeric vector starting log effective population size values.
`kappa`	numeric starting kappa.
`covariates`	list of functions representing covariate trajectories that (may) influence sampling frequency.
`power_covariates`	list of functions representing interaction covariate trajectories that (may) influence sampling frequency.
`betas`	numeric vector of (starting) values for the coefficients of the log-linear sampling time model.
`pow_betas`	numeric vector of (starting) values for the interaction coefficients of the log-linear sampling time model.
`samp_alg`	string selecting sampling algorithm for sampling time intensity coefficients. One of "none" (default), "fixed", "MH", and "ESS".
`kappa_alg`	selects sampling algorithm for kappa. One of "gibbs" (default) or "whiten".
`beta_vars`	numeric vector prior variances of the beta hyperparameters.
`pow_beta_vars`	numeric vector prior variances of the pow_beta hyperparameters.
`printevery`	integer how many MCMC steps between writing output to the console.
`first_elem_prec`	numeric the precision of the first element.