GWLP: Function for fast calculation of GWLP

View source: R/fastlengths.R

GWLPR Documentation

Function for fast calculation of GWLP

Description

Calculates GWLP using the formulae from Xu and Wu (2001)

Usage

GWLP(design, ...)
## S3 method for class 'design'
GWLP(design, kmax=design.info(design)$nfactors, 
     attrib.out=FALSE, with.blocks = FALSE, digits = NULL, ...)
## Default S3 method:
GWLP(design, kmax=ncol(design), attrib.out=FALSE, digits = NULL, ...)

Arguments

design

a design, not necessarily of class design; class design properties are exploited by using only factor columns (or factor and block columns, if with.blocks is TRUE)

kmax

the maximum word length requested

attrib.out

the detail added to the output (see Value section)

with.blocks

if TRUE, the block column contributes to the GWLP, otherwise it does not

digits

the number of decimals to round to; NULL prevents rounding

...

further arguments to generic GWLP; not used in the methods

Details

Function GWLP is much faster but also more inaccurate than the function lengths, which calculates numbers of words for lengths 2 to 5 only. Note, however, that function lengths can be faster for designs with very many rows.
If a design factor contains only some of the intended levels, design must be a data frame, and the factor must be an R factor with the complete set of levels specified, in order to make function GWLP aware of the missing levels.

Value

The GWLP methods output a named vector with the numbers of generalized words of lengths zero to kmax. If attrib.out is TRUE, this vector comes with the attributes B and levels.info, the latter documenting the level situation of the design, the former the distance distribution B (Xu and Wu 2001).

Author(s)

Hongquan Xu, Ulrike Groemping

References

Xu, H.-Q. and Wu, C.F.J. (2001). Generalized minimum aberration for asymmetrical fractional factorial designs. Annals of Statistics 29, 1066–1077.

See Also

See Also lengths

Examples

GWLP(L18)
GWLP(L18, attrib.out=TRUE)

DoE.base documentation built on Nov. 15, 2023, 1:06 a.m.