wspDesign {DiceDesign} | R Documentation |
The WSP (Wooton, Sergent, Phan-Tan-Luu) algorithm is an iterative algorithm based on suppression of some experiments from an initial design in each step. WSP leads to a space filling design
wspDesign(design, dmin, init = "center")
design |
a matrix (or a data.frame) corresponding to the design of experiments |
dmin |
a minimum bound for mindist value of the final design |
init |
defines the initialization point (input coordinates) of the algorithm: "center" (default value) takes the central point of the domain "random" takes a random point inside the domain |
WSP enables to create a design D which is such that mindist(D)>dmin. However, it cannot assess the number of experiments. Similarly to straussDesign function, WSP is a powerful algorithm to construct space filling designs in high dimension
A list containing:
InitialDesign |
the starting design |
dmin |
minimum bound for mindist value of the final design |
design |
the matrix of the final design |
ResidualDesign |
the matrix of the residual design (points of InitialDesign not in design) |
G. Damblin & B. Iooss
J. Santiago, M. Claeys-Bruno, M.Sergent (2012). Construction of space filling designs using WSP algorithm for high dimensional spaces, Chenometrics and Intelligent Laboratory Systems, 113:26-31.
dimension <- 2
n <- 100
X <- matrix(runif(n*dimension), n, dimension)
m <- wspDesign(X, 0.1)
plot(m$design)
xDRDN(m, letter = "T", dgts = 2, range = c(-10, 10))