designCriterion: Function of criterion value computing

In PingYangChen/DiscrimOD: DiscrimOD: Finding Optimal Discrimination Designs by Hybridizing Particle Swarm and L-BFGS Algorithms

Function of criterion value computing

Description

Computes the discrimination design criterion value

Computes the discrimination design criterion value

Usage

designCriterion(
  DESIGN1,
  MODEL_INFO,
  DISTANCE,
  dsLower,
  dsUpper,
  crit_type = "pair_fixed_true",
  MaxMinStdVals = NULL,
  PSO_INFO = NULL,
  LBFGS_INFO = NULL,
  environment,
  ...
)

designCriterion(
  DESIGN1,
  MODEL_INFO,
  DISTANCE,
  dsLower,
  dsUpper,
  crit_type = "pair_fixed_true",
  MaxMinStdVals = NULL,
  PSO_INFO = NULL,
  LBFGS_INFO = NULL,
  environment,
  ...
)

Arguments

DESIGN1	matrix. The approximate design.
MODEL_INFO	list of information of competing models. For details, run emptyModelList() and see the instruction below.
DISTANCE	function. The R/C++ function of distance measure. For T-optimal design, the function is the squared difference of two models means.
dsLower	vector. The finite lower bounds of the design space. Its length should be equal to the dimension of design space.
dsUpper	vector. The finite upper bounds of the design space. Its length should be equal to the dimension of design space.
crit_type	string. The name of the case of the discrimination design problem. The default is 'pair_fixed_true'.
MaxMinStdVals	vector. The values of demoninators in the design efficiency calculation for finding max-min discrimination design.
PSO_INFO	list. PSO and BFGS options.

Value

An List.

• cri_val the design criterion value of the resulting design in the PSO result or the given design.

• theta2 a matrix of the parameters of each input model that result the design criterion value.

An List.

• cri_val the design criterion value of the resulting design in the PSO result or the given design.

• theta2 a matrix of the parameters of each input model that result the design criterion value.

PingYangChen/DiscrimOD documentation built on June 10, 2025, 11:32 p.m.