Implementation of the Xiao Xu TA algorithm (experimental, for comparison with MDLEs only)

XiaoXuMDLE(
  oa,
  ell,
  noptim.oa = 1,
  nseq = 2000,
  nrounds = 50,
  nsteps = 3000,
  dmethod = "manhattan",
  p = 50
)

createF(Dc, Dp, s, ell, nseq = 2000)

optimize(
  Dc,
  s,
  ell,
  Fhat,
  nrounds = 50,
  nsteps = 3000,
  dmethod = "manhattan",
  p = 50
)

Arguments

oa: matrix or data.frame that contains an ingoing symmetric OA. Levels must be denoted as 0 to s-1 or as 1 to s.
ell: the multiplier for each number of levels
noptim.oa: integer: number of optimization rounds applied to initial oa itself before starting expansion
nseq: tuning parameters for TA algorithm
nrounds: tuning parameters for TA algorithm
nsteps: tuning parameters for TA algorithm
dmethod: distance method for phi_p, "manhattan" (default) or "euclidean"
p: p for phi_p (the larger, the closer to maximin distance)
Dc: matrix
Dp: matrix
s: original number of levels
Fhat: distribution function (created with createF)

Value

XiaoXuMDLE returns a matrix with attribute phi_p.

createF returns a distribution function.

optimize returns a matrix with attribute phi_p.

Details

The ingoing oa is optimized by function phi_optimize, using noptim.rounds=noptim.oa; this yields the matrix Dp for use in the internal functions DcFromDp and createF.
Function XiaoXuMDLE returns the value that is produced by applying the internal function optimize to the resulting Dc and F.

References

For full detail, see SOAs-package.

Xiao and Xu (2018)

Examples

## create 8-level columns from 4-level columns
XiaoXuMDLE(DoE.base::L16.4.5, 2, nrounds = 5, nsteps=50)
#>    A B C D E
#> 1  2 4 0 3 5
#> 2  3 6 7 4 0
#> 3  3 1 2 0 6
#> 4  2 2 4 7 2
#> 5  7 5 7 1 3
#> 6  6 7 1 6 6
#> 7  6 1 5 3 1
#> 8  7 2 2 5 5
#> 9  0 5 3 6 0
#> 10 0 7 5 1 4
#> 11 1 0 0 4 2
#> 12 1 3 6 2 7
#> 13 4 4 4 5 7
#> 14 4 6 3 2 3
#> 15 5 0 6 7 4
#> 16 5 3 1 0 1
#> attr(,"origin")
#> [1] "Kuhfeld collection"
#> attr(,"phi_p")
#> [1] 0.115849