Thread Subject: how to vectorize this

Hi
A very simple question:
I want to replace the diagonals of a matrix.
for example,
a=[1 1 1
   1 1 1
   1 1 1]
I need to change 1 in the diagonals to 2, and the result
a=[2 1 1
   1 2 1
   1 1 2]
I can write this code in a For loop.
I'm sure there is a much better way to do the same job,
hmmm, writing it in a vectorized form. But I have not been
able to figure one out.

Can somebody help me, thanks!!!

Hi,

try this :

a=ones(3);
a(1:size(a,1)+1:end)=2

or

a=diag([1 1 1])+1

Jérôme

In article <fkqeqa$ckp$1@fred.mathworks.com>,
xiao chen <yuanyuan5220985@gmail.com> wrote:
>I want to replace the diagonals of a matrix.

%the generalized solution for matrices that might not be square
numdiag = min([size(a,1),size(a,2)]);
diagidx = (0:numdiag-1)*size(a,1)+(1:numdiag));
a(diagidx) = a(diagidx) + 1;
--
  "I will speculate that [...] applications [...] could actually see a
  performance boost for most users by going dual-core [...] because it
  is running the adware and spyware that [...] are otherwise slowing
  down the single CPU that user has today" -- Herb Sutter

roberson@ibd.nrc-cnrc.gc.ca (Walter Roberson) wrote in message

> %the generalized solution for matrices that might not be
square
> numdiag = min([size(a,1),size(a,2)]);
> diagidx = (0:numdiag-1)*size(a,1)+(1:numdiag));
> a(diagidx) = a(diagidx) + 1;
> --

And if you prefer readability, diagidx can be obained by
sub2ind() as well:

numdiag = min(size(a));
diagidx = sub2ind(size(a),1:numdiag,1:numdiag);%<-row,column
a(diagidx) = a(diagidx) + 1;

Bruno

roberson@ibd.nrc-cnrc.gc.ca (Walter Roberson) wrote in
message <fkqkau$97i$1@canopus.cc.umanitoba.ca>...
> In article <fkqeqa$ckp$1@fred.mathworks.com>,
> xiao chen <yuanyuan5220985@gmail.com> wrote:
> >I want to replace the diagonals of a matrix.
>
> %the generalized solution for matrices that might not be
square
> numdiag = min([size(a,1),size(a,2)]);
> diagidx = (0:numdiag-1)*size(a,1)+(1:numdiag));
> a(diagidx) = a(diagidx) + 1;
> --
> "I will speculate that [...] applications [...] could
actually see a
> performance boost for most users by going dual-core
[...] because it
> is running the adware and spyware that [...] are
otherwise slowing
> down the single CPU that user has today" --
Herb Sutter

pos = sub2ind(size(a),1:min(size(a)),1:min(size(a)));
a(pos) = 2;

Anh Huy Phan
RIKEN - BSI

"xiao chen"
<SNIP diag-filler evergreen...

one of the many other solutions

     m=zeros(4);
     m(eye(size(m))==1)=pi; % -or- a vec...
     disp(m);

us

assume M is a square matrix of size mXm, and d a vector of
size 1Xm, the following will replace the main diagonal with
the vector.
M=M-diag(diag(M))+diag(d)

In your example,
M=ones(3,3),
and d = 2*ones(1,3)

I don't know how slow it is due to the various diag
function calls.