How to perform augemented DOE in R I have run a series of experiments for a continuous relationship between a response (y) and a single covariate (x). The model for this relationship is probably somewhat logarithmic, so y = Beta log(x). I need to choose the design points for the next round. These points should be conditional on previous design points and their results to obtain a global best value for a design criterion, like the D criterion.
In JMP it is possible to do this with Augemented Designs. Is there a similar function in R or a way to achieve it with a sequence of calculations (also fine)? I don't see this directly in AlgDesign, which I use often, and it is not clear to me if the new DOE.base package can do this.
 A: This is supported in AlgDesign. Check the help pages for optFederov: It has two optional arguments, augment and rows which do what you ask for. 
Below is a short example:
library(AlgDesign)

M1 <-as.matrix( gen.factorial(2, nVars=4, varNames=LETTERS[1:4]))
M2 <- as.matrix(gen.mixture(11,4))
colnames(M2) <- letters[1:4]

car_prod <- function(M1, M2) { # Cartesian product of two matrices
    ind  <- expand.grid(1:NROW(M1), 1:NROW(M2))
    rbind(cbind(M1[ind[,1],], M2[ind[,2],]))
}

M <- car_prod(M1,M2)

set.seed(7* 11* 13)
# Note we need to remove the intercept from the formula because of the mixture vars
des <- optFederov(~ .-1, data=M, nTrials=16, maxIteration=1000, nRepeats=100)

# Now we want to augment this design:

des_aug <- optFederov(~ .-1, data=M, nTrials=32, augment=TRUE, rows=des$rows, maxIteration=1000, nRepeats=100)

But look at the solution! All the mixture variables has one mixture component at 1.0, which cannot be good ... so for practical use much more thought is needed, but this example shows the mechanics of use of R
