lmmControl: Control options for 'smsn.lmm()', 'smn.lmm()' and...

In skewlmm: Scale Mixture of Skew-Normal Linear Mixed Models

Control options for smsn.lmm(), smn.lmm() and smn.clmm()

Description

The values supplied in the function call replace the defaults and a list with all possible arguments is returned. The returned list has class "lmmControl" and is used as the control argument to the smsn.lmm(), smn.lmm() and smn.clmm() functions.

Usage

lmmControl(tol = 1e-06, max.iter = 300, calc.se = TRUE, lb = NULL,
           lu = NULL, luDEC = 10,
           initialValues = list(beta = NULL, sigma2 = NULL, D = NULL,
                                lambda = NULL, phi = NULL, nu = NULL),
           quiet = FALSE, showCriterium = FALSE, algorithm = "DAAREM",
           parallelphi = NULL, parallelnu = NULL, ncores = NULL,
           control.daarem = list())

Arguments

tol	Tolerance for the convergence criterion. Default = 1e-6.
max.iter	Maximum number of iterations for the EM algorithm. Default = 200.
calc.se	A logical value indicating if standard errors should be calculated.
lb	Optional. Bottom limit for estimating nu.
lu	Optional. Upper limit for estimating nu.
luDEC	Optional. Upper limit for estimating the "damping" parameter for DEC covariance. If luDEC<=1, only attenuation of the exponential decay can be obtained.
initialValues	Optional. A named list containing initial parameter values, with at most the following elements: beta, sigma2, D, lambda, phi, nu.
quiet	A logical value indicating if the iteration message should be suppressed. Useful when calling the function in R Markdown.
showCriterium	A logical value indicating if the criterium should be shown at each iteration.
algorithm	Algorithm to be used for estimation, either "DAAREM" (default) or "EM". DAAREM is an acceleration method and usually converges with fewer iterations, but it may result in numerical errors (in this case, please use the "EM" option).
parallelphi	A logical value indicating if parallel optimization should be used in the numerical update of the parameters related to the within-subject dependence structure. Default is TRUE if the data contains more than 30 subjects, and FALSE otherwise. Meaningless if depStruct = "UNC".
parallelnu	A logical value indicating if parallel optimization should be used in the numerical update of nu. Meaningless if distr="norm" or distr="sn".
ncores	Number of cores to be used for the parallel optimization. Meaningless if parallelphi = FALSE and parallelnu = FALSE.
control.daarem	List of control for the daarem algorithm. See help(daarem, package = "daarem") for details. Meaningless if algorithm = "EM"

Author(s)

Fernanda L. Schumacher, Larissa A. Matos and Victor H. Lachos

References

Henderson, N.C. and Varadhan, R. (2019) Damped Anderson acceleration with restarts and monotonicity control for accelerating EM and EM-like algorithms, Journal of Computational and Graphical Statistics, Vol. 28(4), 834-846.

Schumacher, F. L., Lachos, V. H., and Matos, L. A. (2021). Scale mixture of skew-normal linear mixed models with within-subject serial dependence. Statistics in Medicine 40(7), 1790-1810.

See Also

smsn.lmm, smn.lmm, smn.clmm, update

Examples

lmmControl(algorithm = "EM")

fm1 = smn.lmm(nlme::Orthodont, formFixed=distance ~ age+Sex,
              groupVar="Subject", control=lmmControl(tol=1e-7))

skewlmm documentation built on July 9, 2023, 7:29 p.m.