rgp (version 0.4-1)

paretoFrontKneeIndex: Find the knee of a two dimensional pareto front

Description

Given a matrix m of two rows and n columns, representing solutions of a two-dimensional optimization problem, returns the column index of the point with minimum euclidean distance to the utopia point. The utopia point is the point consisting of the row minima of m. NA or NaN values of m are ommited.

Usage

paretoFrontKneeIndex(m, normalize = TRUE)

Arguments

m

A matrix of two rows and n columns, representing the solutions of a two-dimensional optimization problem.

normalize

Whether to normalize both objectives to the interval of [0, 1], defaults to TRUE.

Value

The knee point index, i.e. the column index in m of the point of minimum euclidean distance to the utopia point.

Examples

m1 <- matrix(runif(200), ncol = 100)
plot(t(m1))
points(t(m1[,emoa::nds_rank(m1) == 1]), col = "red", pch = 16)
pKnee <- m1[, paretoFrontKneeIndex(m1)]
points(t(pKnee), col = "green4", pch = 16)