Well... this current behavior contradicts https://www.jsoftware.com/help/dictionary/d031.htm and https://code.jsoftware.com/wiki/Vocabulary/underdot
So I think it's unintended behavior. That said, https://code.jsoftware.com/wiki/Vocabulary/percentdot also says (for %. monadic): "If the columns of y are dependent, domain error is signaled" so there's a tension here between %. behavior and _. issues which presumably lead into the current situation. Thanks, -- Raul On Thu, Feb 24, 2022 at 4:06 PM Clifford Reiter <reit...@lafayette.edu> wrote: > > In my mind the inverse of a non invertible matrix is a domain error. > > But I have become aware of otherwise > > > JVERSION > > Engine: j902/j64avx2/windows > > Release-a: commercial/2020-12-05T13:36:01 > > Library: 9.02.08 > > Qt IDE: 1.9.2s/5.12.10(5.12.10) > > Platform: Win 64 > > Installer: J902 install > > InstallPath: c:/program files/j902 > > Contact: www.jsoftware.com > > (Also 9.03 MacOs) > > Likely due to avx changes in 9.01 > > %. i. 4 4 > > |domain error > > | %.i.4 4 > > %. 3 3$1 > > |NaN error > > %. i. 3 3 > > _. _. _. > > _. _. _. > > _. _. _. > > Run (;%.)a=:(,+/)?2 3$4 > > a few times and see all three types of results. I find the array of > indeterminants disquieting but I am not sure it is worth doing anything > about. > > Thoughts? > > Back to grading... > > Best, Cliff > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
