Basically, everything works if I set standardize=TRUE, the default. But I find that when standardize=FALSE I get the error message "Warning in sqrt(var.U) : NaNs produced" (followed by some related errors). I believe this is because the grid range is out of bounds because when I set the grid range manually using spy.range and spz.range the error goes away and the program produces a result.
It may be something pathological with this data set. It could be related to the fact that I am treating a binary treatment variable as Gaussian (forcing treatSense to work by adding a little jitter to the treatment indicator).
Quoting Joel, here:
standardizeBug.tar.gz