Stéphane,
Thanks, this is great, the use of eye is the way to go to overcome the
problem of an earlier suggestion which discarded, for instance, A + 2
for not being what Horner expects; but A + 2*eye(A) is!
Thanks also for making me aware of the simplified syntax 2*eye() which
adapts its
Why just not writing Horner's algorithm directly ?
functionout=horner_mat(p,
X)a=coeff(p)out=zeros(X);fork=degree(p):-1:0out=out*X+a(k+1)*eye()endendfunction-->
a=[1 2;3 4]a = 1. 2.3. 4.--> horner_mat(1+%s-%s^2,a)ans =-5. -8. -12.
-17.--> eye() + a - a^2ans =-5. -8. -12. -17.S.
Le
Hi,
I've got some time this morning (-: so this one seems to work and should be a
bit more efficient:
// **
P = %s^2 + 2*%s + 3
C = coeff(P)
n = size(C, "c")
if n > 1 then
stringP = "X1 = x; "
for i = 3:n
stringP = stringP+"X"+string(i-1)+" = X"+string(i-2)+"*x; "
I'm aware that this is an inefficient way to compute the result
but unfortunately, the Horner form of polynomial does not work on matrices:
(A + 2)*A is not the same as A*A + 2*A for Scilab.
It is possible to write a better function manually,
by first calculating the successive powers before
Hello Frederico,
> De : Federico Miyara
> Envoyé : mardi 24 septembre 2019 00:25
>
> Is there some way of evaluating a polynomial on a square matrix in a
> matrix-wise (not component-wise) fashion?
You can transform your polynomial into a usual external function Then apply it
to a matrix.
e.g.
