create_ns_mcmc: Create an MCMC object to estimate the marginal likelihood...

In beautier: 'BEAUti' from R

Create an MCMC object to estimate the marginal likelihood using Nested Sampling.

Description

This will result in a BEAST run that estimates the marginal likelihood until convergence is achieved. In this context, chain_length is only an upper bound to the length of that run.

Usage

create_ns_mcmc(
  chain_length = 1e+07,
  store_every = -1,
  pre_burnin = 0,
  n_init_attempts = 3,
  particle_count = 1,
  sub_chain_length = 5000,
  epsilon = "1e-12",
  tracelog = create_tracelog(),
  screenlog = create_screenlog(),
  treelog = create_treelog()
)

Arguments

chain_length	upper bound to the length of the MCMC chain
store_every	number of states the MCMC will process before the posterior's state will be saved to file. Use -1 or NA to use the default frequency.
pre_burnin	number of burn in samples taken before entering the main loop
n_init_attempts	number of initialization attempts before failing
particle_count	number of particles
sub_chain_length	sub-chain length
epsilon	epsilon
tracelog	a tracelog, as created by create_tracelog
screenlog	a screenlog, as created by create_screenlog
treelog	a treelog, as created by create_treelog

Value

an MCMC object

Author(s)

Richèl J.C. Bilderbeek

References

* [1] Patricio Maturana Russel, Brendon J Brewer, Steffen Klaere, Remco R Bouckaert; Model Selection and Parameter Inference in Phylogenetics Using Nested Sampling, Systematic Biology, 2018, syy050, https://doi.org/10.1093/sysbio/syy050

See Also

Use create_mcmc to create a regular MCMC. Use create_test_ns_mcmc to create an NS MCMC for testing, with, among others, a short MCMC chain length. Use check_ns_mcmc to check that an NS MCMC object is valid.

Examples

if (is_on_ci()) {

  mcmc <- create_ns_mcmc(
    chain_length = 1e7,
    store_every = 1000,
    particle_count = 1,
    sub_chain_length = 1000,
    epsilon = 1e-12
  )

  beast2_input_file <- get_beautier_tempfilename()
  create_beast2_input_file(
    get_fasta_filename(),
    beast2_input_file,
    mcmc = mcmc
  )
  file.remove(beast2_input_file)

  remove_beautier_folder()
}

beautier documentation built on Nov. 2, 2023, 5:08 p.m.