QR decomposition

[varun]

I am trying to implement QR decomposition for Hermitian Matrix
But the conventional Gram Schmidt Process Fails please help!!

My code:
clc
clear all
close all

%A = [1 1 1; 1 1 0 ; 1 1 2];
A = [1+6j 1-4j 3-2j; 1+4j 8-1j 2+4j ; 3+2j 2-4j 2];
[qq rr] = qr(A);

MM=A;
[m n] = size(A);

r = zeros(n,n);mm=0;
QT = zeros(n,m);
Q = A;

for k=1:n
QT(k,:) = A(:,k)';
end
for k = 1:n
r(k,k) = sqrt(QT(k,:)*QT(k,:)');
QT(k,:) = QT(k,:)/r(k,k);
if(k<n)
for j=k+1:n
r(k,j)= (A(:,j)'*QT(k,:)')';
QT(j,:)= (QT(j,:)'-(r(k,j)*QT(k,:)'))';
end
end
end

[Matt J]

"varun" wrote in message <jhb6na$7sf$1...@newscl01ah.mathworks.com>...

> I am trying to implement QR decomposition for Hermitian Matrix
> But the conventional Gram Schmidt Process Fails please help!!

Why do you even need to do Gram Schmidt when the QR command gives you the decomposition straight away? Gram Schmidt is a numerically unstable algorithm, which is why people always use QR instead.

[Steven_Lord]

"varun " <signal...@gmail.com> wrote in message
news:jhb6na$7sf$1...@newscl01ah.mathworks.com...

> I am trying to implement QR decomposition for Hermitian Matrix
> But the conventional Gram Schmidt Process Fails please help!!

Define "fails" -- does your code throw an error? Does it return results
different from what you expect?

The more specific information you provide, the more likely it is that
someone will be able to help you.

*snip code*

--
Steve Lord
sl...@mathworks.com
To contact Technical Support use the Contact Us link on
http://www.mathworks.com

[Matt J]

"varun" wrote in message <jhb6na$7sf$1...@newscl01ah.mathworks.com>...

> I am trying to implement QR decomposition for Hermitian Matrix
> But the conventional Gram Schmidt Process Fails please help!!

In any case, it doesn't look like the process failed. When I run your code, I get
QT'=qq
r=rr
up to inconsequential sign differences.