rpl_loglike: Fisher transformation applied within function

In andreamrau/rpl: Randomized pairwise likelihood method for complex statistical inferences

Fisher transformation applied within function

Description

par is a vector of mean parameters followed by Fisher-transformed correlation parameters

Usage

rpl_loglike(
  par,
  mydata,
  pi = 1,
  offsets = NULL,
  block_indices = NULL,
  seed = NULL,
  count_pairs = FALSE,
  parallel = FALSE,
  BPPARAM = ifelse(parallel, bpparam(), NULL)
)

Arguments

par	Vector providing parameter values (means followed by correlations)
mydata	Data matrix
pi	Bernoulli sampling parameter (0 < pi <= 1)
offsets	If needed, a matrix of offsets (optional)
block_indices	List of vectors providing indices for each block for blockwise exchangeable correlation matrices
seed	Seed to use for random number generation
count_pairs	If TRUE, return the number of observations retained for each pair of variables used in the randomized pairwise likelihood
parallel	If TRUE, some computations are parallelized with the BiocParallel package
BPPARAM	If parallel=TRUE, the registry of back-ends for use in parallelized calculations

Value

Negative log-likelihood for user-provided parameter vector, calculated with the randomized pairwise likelihood approach. If count_pairs = TRUE, counts of the number of observations used for each pair are also returned.

References

Mazo, G., Karlis, D., Rau, A. (2021) A randomized pairwise likelihood method for complex statistical inferences. In revision. hal-03126621.

andreamrau/rpl documentation built on April 26, 2023, 3:57 p.m.