FastMmLHD: Fast Maximin Distance LHD

In LHD: Latin Hypercube Designs (LHDs)

Description

FastMmLHD returns a n by k maximin distance LHD matrix generated by the construction method of Wang, L., Xiao, Q., and Xu, H. (2018)

Usage

FastMmLHD(n, k, method = "manhattan", t1 = 10)

Arguments

n	A positive integer, which stands for the number of rows (or run size).
k	A positive integer, which stands for the number of columns (or factor size).
method	A distance measure method. The default setting is "manhattan", and it could be one of the following: "euclidean", "maximum", "manhattan", "canberra", "binary" or "minkowski". Any unambiguous substring can be given.
t1	A tunning parameter, which determines how many repeats will be implemented to search for the optimal design. The default is set to be 10.

Value

If all inputs are logical, then the output will be a n by k maximin distance LHD under under the maximin L_1 distance criterion..

References

Wang, L., Xiao, Q., and Xu, H. (2018) Optimal maximin $L_1$-distance Latin hypercube designs based on good lattice point designs. The Annals of Statistics, 46(6B), 3741-3766.

Examples

#n by n design when 2n+1 is prime
try=FastMmLHD(8,8)
try
phi_p(try)   #calculate the phi_p of "try".

#n by n design when n+1 is prime
try2=FastMmLHD(12,12)
try2
phi_p(try2)   #calculate the phi_p of "try2".

#n by n-1 design when n is prime
try3=FastMmLHD(7,6)
try3
phi_p(try3)   #calculate the phi_p of "try3".

#General cases
try4=FastMmLHD(24,8)
try4
phi_p(try4)   #calculate the phi_p of "try4".

LHD documentation built on Aug. 1, 2021, 1:06 a.m.