On Thu, 3 Mar 2005, IOhannes m zmoelnig wrote:
Winfried Ritsch wrote: jfyi: the major inconsistancy with zexy's [matrix~] is, that when multiplying a vector of signals with a matrix the actual operation that takes place is "x^ = a^ * A", while in literature normally we find things like "y^ = B * b^", which is only the same if we assume x^=y^' and A=B'.
GridFlow's @inner was originally doing something like the latter, actually R = A*transpose(B), of which the latter is a special case (because of the commutativity of Real Dot Product).
At a certain point in 2002, I changed it so that @inner is R=A*B, and @inner2 is R=A*transpose(B). In the new system (2005) there is #inner and #transpose.
The other differences are that matrix~ is for signals only, and handles only two dimensions.
anyone has an idea how to find an elegant solution to this problem (while still maintaining an intuitive object) ?
just write your matrices pretransposed.
_____________________________________________________________________ Mathieu Bouchard -=- Montréal QC Canada -=- http://artengine.ca/matju