-------- Original Message -------- Subject: tpsRegr update to version 1.38 Date: Sun, 30 Oct 2011 15:11:07 -0400 From: F. James Rohlf <[email protected]> Reply-To: [email protected] Organization: Stony Brook University To: morphmet <[email protected]> I have just uploaded version 1.38 of tpsRegr to the Stony Brook Morphometrics server. It fixed a problem in computing Wilks lambda when doing permutation tests on data with a large number of landmarks. Previously I had taken a shortcut to try to speed up the computations when performing permutation tests. That was not sufficiently accurate when there are many landmarks and hence determinants need to computed for large sums of squares matrices. Permutation tests will now be slower but more accurate (a not unusual trade-off). Thanks to Karel Kleisner for providing a dataset that demonstrated the problem (there were 72 landmarks). ---------------------- F. James Rohlf, John S. Toll Professor Dept. Ecology and Evolution, Stony Brook University, NY 11794-5245 PPlease consider the environment before printing this email
