logLik_from_vector_param: Compute the log-likelihood from parameter vector

In courtiol/SileR: Survival Analysis of Elephants Using an Extension of the William Siler's Framework

Description

This function is called internally by other functions. It is a wrapper to the function logLik_from_design_matrix which has inputs consistent to the requirements of optimisation routines.

Usage

logLik_from_vector_param(param_values, param_str, data, indices = NULL,
  scale = 1, param_matrix = NULL, empty_param_matrix_full = NULL,
  empty_design_matrix = NULL)

Arguments

param_values	The vector of values for the parameters
param_str	The named vector indicating the parameter structure, that is, the type of subcategory for each meta parameter
data	The dataset to be used
indices	The list of indices computed by compute_indices
scale	The scaling parameter (useful if the optimisation function can only minimise)
param_matrix	The matrix storing parameter values
empty_param_matrix_full	The empty expanded or complete parameter matrix
empty_design_matrix	The empty design matrix

Details

If the argument param_matrix is given, the parameter matrix will be updated and not build from scratch. This will save some computation time.

Value

The scaled log-likelihood.

See Also

logLik_from_design_matrix

Examples

logLik_from_vector_param(param_values = c(0.1, 0.200001), param_str = c("w1" = "s"),
                         data = SurvEles_small)
pmat <- build_param_matrix(param_values = c(0.1, 0.2), param_str = c("w1" = "s"))
logLik_from_vector_param(param_values = c(0.1, 0.200001, rep(0, 14)), param_str = c("w1" = "s"),
                         data = SurvEles_small, param_matrix = pmat)

courtiol/SileR documentation built on May 16, 2020, 8:10 p.m.