Le 20/12/2012 13:35, Volker Braun a écrit :
It is common practice to encode missing/invalid data points as NaN, see for example the NaN toolbox for Matlab/Octave (http://pub.ist.ac.at/~schloegl/matlab/NaN/). Having arbitrary base grids would be nice, of course, but certainly comes at a significant performance cost. Ideally there would be a simple implementation with square grids and Nans that propagate through the surface mesh generation (no if/else branches) for missing numbers, and a more complicated plot that can deal with arbitrary grids.
Using some special value as meaning invalid is what NaN is about, in fact. :-)
Still, the fact that an invalid point could be given by something like (Nan, 2) in a plane or (0,1,Nan) in 3d-space is probably wrong: after all, those are *valid* tuples... which happen to contain invalid coordinates. There should be a NaT (not-a-tuple) for this situation. I can't help but think about the haskell language's Maybe monad here.
Snark on #sagemath -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To post to this group, send email to sage-devel@googlegroups.com. To unsubscribe from this group, send email to sage-devel+unsubscr...@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel?hl=en.
