M_step: M-step in the EAM algorithm described in KMS19.

In depCensoring: Statistical Methods for Survival Data with Dependent Censoring

M-step in the EAM algorithm described in KMS19.

Description

This function performs the maximization step in the EAM algorithm. More specifically, it maximizes the expected improvement. ToDo: implement sample space contractions (see comment made in documentation of draw.sv.init).

Usage

M_step(
  dir,
  evaluations,
  theta.hash,
  fit.krige,
  test.fun,
  c,
  par.space,
  hyperparams,
  verbose
)

Arguments

dir	Direction to search in. dir = 1 corresponds to finding the upper bound of the confidence interval. dir = -1 corresponds to finding the lower bound.
evaluations	Matrix containing each point that was already evaluated, alongside the corresponding test statistic and critical value, as its rows.
theta.hash	Tentative best value of theta. Obtained from the E-step.
fit.krige	Kriging model obtained from the A-step.
test.fun	The test function to be inverted in order to obtain the identified set.
c	Projection vector.
par.space	Bounds of the parameter space.
hyperparams	Parameters used in obtaining initial values for the maximization algorithm. If NULL, default values are used. Default is hyperparams = NULL.
verbose	Verbosity parameter.

depCensoring documentation built on April 4, 2025, 1:52 a.m.