Constraints: Formulating Constraints

In kkmann/adoptr: Adaptive Optimal Two-Stage Designs in R

Formulating Constraints

Description

Conceptually, constraints work very similar to scores (any score can be put in a constraint). Currently, constraints of the form 'score <=/>= x', 'x <=/>= score' and 'score <=/>= score' are admissible.

Usage

## S4 method for signature 'Constraint,TwoStageDesign'
evaluate(s, design, optimization = FALSE, ...)

## S4 method for signature 'ConditionalScore,numeric'
e1 <= e2

## S4 method for signature 'ConditionalScore,numeric'
e1 >= e2

## S4 method for signature 'numeric,ConditionalScore'
e1 <= e2

## S4 method for signature 'numeric,ConditionalScore'
e1 >= e2

## S4 method for signature 'ConditionalScore,ConditionalScore'
e1 <= e2

## S4 method for signature 'ConditionalScore,ConditionalScore'
e1 >= e2

## S4 method for signature 'UnconditionalScore,numeric'
e1 <= e2

## S4 method for signature 'UnconditionalScore,numeric'
e1 >= e2

## S4 method for signature 'numeric,UnconditionalScore'
e1 <= e2

## S4 method for signature 'numeric,UnconditionalScore'
e1 >= e2

## S4 method for signature 'UnconditionalScore,UnconditionalScore'
e1 <= e2

## S4 method for signature 'UnconditionalScore,UnconditionalScore'
e1 >= e2

Arguments

s	Score object
design	object
optimization	logical, if TRUE uses a relaxation to real parameters of the underlying design; used for smooth optimization.
...	further optional arguments
e1	left hand side (score or numeric)
e2	right hand side (score or numeric)

See Also

minimize

Examples

design <- OneStageDesign(50, 1.96)

cp     <- ConditionalPower(Normal(), PointMassPrior(0.4, 1))
pow    <- Power(Normal(), PointMassPrior(0.4, 1))

# unconditional power constraint
constraint1 <- pow >= 0.8
evaluate(constraint1, design)

# conditional power constraint
constraint2 <- cp  >= 0.7
evaluate(constraint2, design, .5)
constraint3 <- 0.7 <= cp # same as constraint2
evaluate(constraint3, design, .5)

kkmann/adoptr documentation built on Feb. 3, 2024, 6:55 p.m.