Help building a matrix-calculus package



On 2/23/12, Stavros Macrakis <macrakis at alum.mit.edu> wrote:

> Robert's general approach works, but the rule itself has a bug.

Yikes, yeah, that's pretty embarrassing. Let me try to recover.

matchdeclare (aa, nctimesp);
nctimesp (e) := not atom(e) and op(e) = ".";
tellsimp (aa^^-1, map (lambda ([e], e^^-1), reverse (aa)));

(a . b . c)^^-1;
 => c^^(-1) . b^^(-1) . a^^(-1)

That's better ....

Robert Dodier