weightsGTP: Function for determining the (monotone part of the) local...

View source: R/informSCI.R

weightsGTPR Documentation

Function for determining the (monotone part of the) local significance levels

Description

Function for determining the monotone part (eta.mu) of the local significance levels for the key equation of the informative SCI algorithm. The function creates dual graphs and rejects some of its hypotheses to obtain the local significance levels.

Usage

weightsGTP(mu, g, weights, alpha, q, mu_0)

Arguments

mu

A real-valued vector (-Inf is also allowed) of dimension m indicating which dual graph should be created and which null hypotheses should be rejected. mu[i]>mu_0[i] iff the corresponding hypothesis is rejected, 1\leq i\leq m.

g

A numeric square matrix of transition weights for the graphical test procedure.

weights

A numeric vector of dimension m of initial weights for the graphical test procedure.

alpha

Overall level of the graphical test procedure.

q

A numeric vector of dimension m of information weights.

mu_0

A numeric vector of dimension m of bounds of the null hypotheses.

Details

m = number of hypotheses.

The function is not suitable if for all 1\leq i\leq m it holds q[i]==0 and mu[i]>mu_0[i].

Value

Returns a numeric vector of dimension m (eta.mu) used for solving the key equation of the informSCI algorithm. It contains the local levels in mu divided by q^{max(mu-mu_0,0)} or divided by adapted information weights (only if q[i]>0).


informativeSCI documentation built on June 22, 2024, 10:01 a.m.