Optimizing a matrix of values

3 views (last 30 days)
Allan Paolo on 14 Dec 2023
Answered: Matt J on 14 Dec 2023
Hi, I need help in optimizing a matrix of values. Assuming a 3x3 matrix, I would have this matrix:
I needed the values of A to F be optimized. What I would usually do is to just take [A B C D E F] as an input value and call it a day. To be fair, this works decently.
My main issue is that my matrix usually changes size, ie. it becomes this:
I will have to change the input to [A B] this time. My question, is, is there a more elegant way of doing these things: Is there a way for me to simply plug in the two matrices without reshaping them and forcing 0 on the diagonal?

Matt J on 14 Dec 2023
You could define the whole matrix as unknown, but with diagonal elements bound to zero:
LB=-inf(n);
LB(1:n+1:end)=0;
UB=-LB;
M=optimvar('M',[n,n],'Lower',LB,'Upper',UB);

madhan ravi on 14 Dec 2023
a = zeros(3);
[m, n] = size(a);
a(~diag(1 : m)) = 1 : m * n - m;
a.'
madhan ravi on 14 Dec 2023
Edited: madhan ravi on 14 Dec 2023
a = zeros(2);
[m, n] = size(a);
a(~diag(1 : m)) = m * n - m : -1 : 1
%a(~diag(1 : m)) = flip(your_input)
madhan ravi on 14 Dec 2023
where 1 : m * n - m is your input