Matrix Multiplikation Speed Up

Hello, I would like to speed up this code, any sugestions

Rm, Zm and Cm are 1700x1700

f   = linspace(1e3,5e6,750);    % Hier gibt man den Frequenzbereich an
w   = 2*pi*f';
parfor k=1:length(f)
        Zm       = Rm+w(k)*1i*Lm;
        Gama     = Kk*(Zm\Kkt);
        YYn      = w(k)*1i*Cm + Gama;
        Zn       = inv(YYn);
        Amp(k)   = abs(1-Zn(jj,1)/Zn(1,1));
 end

I do not have a GPU

[Joss Knight]

What is Kkt? What is jj?

[Matthias Schneider]

Sorry I forgot to mention what Kk, KKt and jj are. Kk ist an transformations Matrix, and Kkt = Kk', jj is the index of an outer for loop.

a = 1700
Kk=eye(a);          
for i=1:(a-1)          
    Kk(i+1,i)=-1;
end

this little pice of code describes how Kk is built, but the Kk is built only once and is the used over and over again.

[madhan ravi]

a?

[Joss Knight]

Sorry, nothing occurs to me. You take the trouble to invert YYn but then only use a single row of it, so maybe you could focus on that piece of algebra.