repairConditionsDistanceMatrix: Repair Conditions of a Distance Matrix
Description
This function repairs distance matrices, so that the following two properties are ensured:
The distance values should be non-zero and the diagonal should be zero.
Other properties (conditionally negative semi-definitene (CNSD), symmetric) are
assumed to be given.
Usage
repairConditionsDistanceMatrix(mat)
Value
repaired distance matrix
Arguments
mat
symmetric, CNSD distance matrix. If your matrix is not CNSD, use correctionCNSD first. Or use correctionDistanceMatrix.
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.
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)
D <- correctionCNSD(D)
D
D <- repairConditionsDistanceMatrix(D)
D