CEGO (version 2.4.3)

repairConditionsCorrelationMatrix: Repair Conditions of a Correlation Matrix

Description

This function repairs correlation matrices, so that the following two properties are ensured: The correlations values should be between -1 and 1, and the diagonal values should be one.

Usage

repairConditionsCorrelationMatrix(mat)

Value

repaired correlation matrix

Arguments

mat

symmetric, PSD distance matrix. If your matrix is not CNSD, use correctionDefinite first. Or use correctionKernelMatrix.

References

Martin Zaefferer and Thomas Bartz-Beielstein. (2016). Efficient Global Optimization with Indefinite Kernels. Parallel Problem Solving from Nature-PPSN XIV. Accepted, in press. Springer.

See Also

correctionDefinite, correctionDistanceMatrix, correctionKernelMatrix, correctionCNSD, repairConditionsDistanceMatrix

Examples

x <- list(c(2,1,4,3),c(2,4,3,1),c(4,2,1,3),c(4,3,2,1),c(1,4,3,2))
D <- distanceMatrix(x,distancePermutationInsert)
K <- exp(-0.01*D)
K <- correctionDefinite(K,type="PSD")$mat
K
K <- repairConditionsCorrelationMatrix(K)