R/make.control.flexrsurv.R

In flexrsurv: Flexible Relative Survival Analysis

control.flexrsurv.additive <- function (epsilon = 1e-08, maxit = 25, trace = FALSE, ...){
	# controls for flexrsurv additive models are glm.controls
	return(glm.control(epsilon = epsilon, maxit = maxit, trace = trace))
}



control.flexrsurv.multiplicative <- function (epsilon = 1e-8, maxit = 100, trace = FALSE, epsilon.glm = 1e-1, maxit.glm=30, ...){
	# controls for flexrsurv multiplicative models
	# epsilon : positive convergence tolerance of the iterative alternative algorithm;
	#           the iterations converge when |loglik - loglik_old}/(|loglik| + 0.1) < epsilon.
	# maxit  : integer giving the maximal number of iteration of the iterative alternative algorithm.
	# trace : logical indicating if output should be produced for each iteration.  
	# epsilon.glm,  maxit.glm  : contols passed to the inner glm call of the itertive alternative algorithm
	
	if (!is.numeric(epsilon) || epsilon <= 0)
		stop("'epsilon' must be > 0", call.=FALSE)
	if (!is.numeric(epsilon.glm) || epsilon.glm <= 0) 
		stop("'epsilon.glm' must be > 0", call.=FALSE)
	if (!is.numeric(maxit.glm) || maxit.glm <= 0) 
		stop("maximum number of iterations 'maxit.glm' must be > 0", call.=FALSE)
	return(list(epsilon = epsilon, maxit = maxit, trace = trace, epsilon.glm = epsilon.glm, maxit.glm = maxit.glm))
}