Multivariate taylor



Hello,


I don't understand the result (or maybe only the notation) of 
multivariate taylor expansion of a formal function. I would expect the 
taylor expansion of f(x,y) to order 1 to be

f(0,0) + x (df(x,y)/dx)(0,0) + y (df(x,y)/dy)(0,0)

but what I understand of maxima's answer is

f(0,0) + x (df(x,y)/dx)(0,y) + y (df(0,y)/dy)(0)

which is not the same in general (due to terms of higher order).


However, the taylor expansion of a "concrete" function seems ok.

I didn't succeed to plug a concrete function in a formal taylor expansion.

Could somebody explain what happens below ?

How to handle  ?%at(...)   ?



(%i1) build_info () ;   display2d:false$
Maxima version: 5.14.0
Maxima build date: 17:10 2/5/2008
host type: i686-pc-linux-gnu
lisp-implementation-type: CLISP
lisp-implementation-version: 2.41 (2006-10-13) (built 3374673657) 
(memory 3411216640)

(%o1)
(%i3) taylor(f(x,y),[x,y],[0,0],[1,1])$   define(g(x,y),%);
(%o4) g(x,y):=f(0,0)
         +((?%at('diff(f(x,y),x,1),x = 0))*x+(?%at('diff(f(0,y),y,1),y = 
0))*y)
(%i5) f(x,y):=cos(x+y)$     g(x,y);
(%o6) (?%at('diff(cos(y+x),x,1),x = 0))*x+(?%at('diff(cos(y),y,1),y = 
0))*y+1
(%i7) diff(cos(x+y),x,1)$   a:at(%,x=0)$    diff(cos(y),y,1)$ 
b:at(%,y=0)$   a*x+b*y+1;
(%o11) 1-x*sin(y)
(%i12) taylor(cos(x+y),[x,y],[0,0],[1,1]);
(%o12) +1


I expected (%o11) to be a taylor expansion, i.e. a polynomial in x and 
y, precisely the same as (%o12).


By the way, when  display2d is false then
(%i2) taylor(-x,[x,y],[0,0],[1,1]);
(%o2) +(-x)
the output is not simplified. Should it be so ?


Eric Reyssat