Function that gives list of algorithm variables used by
borrTest.
borrControl(nAlphaGrid=10000,nThetaGrid=1000, maxIter=0, digits=4, orderFunc=NULL)
number used for defining grid for searching over the (0,1) space for significance levels.
Used in borrOrderingAlphaGrid.
number of evenly spaced grid elements for searching over the (0,1) space for theta.
Used in both borrOrderingAlphaGrid and and borrOrderingByRR.
maximum number of searches over the alpha space. Used in borrOrderingAlphaGrid.
number of digits for rounding alpha star values. Used in borrOrderingByRR.
character vector to determine function to do the borr ordering. NULL checks for precalculated values then does borrOrderingAlphaGrid. 'AlphaGrid' uses borrOrderingAlphaGrid
and 'ByRR' uses borrOrderingByRR.
In borrOrderingAlphaGrid
we create a grid for searching over the significance level space, for the first calculation (zeroth iteration) we use alpha.seq where
alpha.seq <-
sort(unique(c(
10 ^ seq(log10(minAlpha), 0, length = nAlphaGrid / 2),
seq(minAlpha, 10 ^ 0, length = nAlphaGrid / 2)
))), where minAlpha is the one-sided p-value at the point (x1=n1, x2=0)
given by minAlpha<- dbinom(0,nT,nC/(nC+nT))*dbinom(nC,nC,nC/(nC+nT)).
If there are ties and maxIter is greater than 0, then replace each tied value with an equally spaced grid (with nAlphaGrid elements) between the adjacent non-tied values.
If the lowest value in the grid, minAlpha, is tied, then set minAlpha<-minAlpha/10
at the beginning of the iteration.
Repeat this process up to maxIter times.
See borrOrderingInternal for more details of algorithms.