compute_W_c_volB {thamesmix} | R Documentation |
Determine the function W, radius c and volume of B from Metodiev et al.(2025)
Description
Determine the function W, radius c and volume of B from Metodiev et al.(2025)
Usage
compute_W_c_volB(
params,
limit,
lps,
logpost,
sims,
iters,
ellipse,
type = "simple",
lps_unif = NULL,
max_iters = Inf
)
Arguments
params |
sample from the posterior (as a matrix) |
limit |
the limit placed on the lps |
lps |
values of the unnormalized log posterior density |
logpost |
function used to compute the lps |
sims |
n_simul x G x (u+1) array of parameters sampled from the posterior, where n_simul is the number of simulations from the posterior, G is the number of components, u is the number of mixture component parameters (parameter u+1 is the mixture weight) |
iters |
half the number of simulations from the posterior |
ellipse |
the ellipsoid E from Metodiev et al. (2025) |
type |
THAMES variant ("simple", "permutations", or "standard") |
lps_unif |
values of the unnormalized log posterior density, evaluated on a uniform sample on the posterior ellipsoid |
max_iters |
maximum number of shrinkage iterations |
Value
a named list with the following elements: perms: number of permutations that were evaluated graph: the overlap graph for G c_opt: the radius of the ellipsoid in Metodiev et al. (2025) scaling: list of fit determined via QDA (means and covariances) log_cor: used to compute the volume of B in Metodiev et al.(2025) param_test: Monte Carlo sample used to compute the volume of B co: the criterion of overlap for G ellipse: the ellipsoid, potentially shifted to the sample mode
References
Martin Metodiev, Nicholas J. Irons, Marie Perrot-Dockès, Pierre Latouche, Adrian E. Raftery. "Easily Computed Marginal Likelihoods for Multivariate Mixture Models Using the THAMES Estimator." arXiv preprint arXiv:2504.21812.