envelope() with fun=F3est about spatstat HOT 2 CLOSED

@baddstats is the real expert on this. Either it is a bug or it is related to the algorithm that calculates the estimate of F. At least in 2D the distance transform algorithm is used to estimate F in Fest and this imposes some requirements on the r-values that F is calculated for. See these lines of the documentation for details.

If something similar is true for F3est, that could be the explanation, but I haven't looked at the details of the code. To circumvent the problem you can set rmax = 0.1*128/127 and nrval = 129 and then simply pick out the first 128 r-values. When I do this with your example the first 127 values km and rs are identical with the values obtained with rmax=0.1. For cs there is a small difference between the two results. In all cases the modification gives a correct looking 128th value (and then the omitted 129th value looks strange instead).

from spatstat.

This is a bug in the C code for F3est. It has now been fixed, in spatstat 1.47-0.008.
Thank you for bring this to our attention.

from spatstat.