Réponse acceptée

Bruno Luong
Bruno Luong le 22 Juil 2020
Modifié(e) : Bruno Luong le 22 Juil 2020

0 votes

Is it good now?
n = 12;
m = ceil(n / 3);
n = m*3;
A = spdiags([1 -4 1]+zeros(3,1), -1:1, 3, 3);
c = repmat({A},1,m);
A = blkdiag(c{:});
A = A + spdiags([1 1]+zeros(n,1), [-3 3], n, n);
full(A)

7 commentaires

mathru
mathru le 22 Juil 2020
In the above mentioned matrix, upper and lower 2nd diagonal elements are not all 1. These diagonal pattern is: 1 1 0 1 1 0 1 1
Bruno Luong
Bruno Luong le 22 Juil 2020
Oops I miss read it. Now it's fixed
mathru
mathru le 22 Juil 2020
Modifié(e) : mathru le 22 Juil 2020
Please see the upper diagonal elements. This diagonal elements are not all 1.
Bruno Luong
Bruno Luong le 22 Juil 2020
Is it good now?
mathru
mathru le 22 Juil 2020
yes. thanks.
mathru
mathru le 27 Juil 2020
How can I write this matrix using for loop?
Sindar
Sindar le 27 Juil 2020
Modifié(e) : Sindar le 27 Juil 2020
Do you want to use the matrix multiple times in a loop or construct the matrix itself in a for loop instead of the above method?
  • If the first (e.g., solving for different U vectors), it's best to generate the matrix outside the loop, then call or copy it inside:
...
A = A + spdiags([1 1]+zeros(n,1), [-3 3], n, n);
for ind=1:5
% if you use the same matrix each loop
U = rand(1,9);
x{ind} = A*U;
% if you need to adjust the matrix each loop
B = A;
B(5) = rand(1);
y{ind} = B*U;
end
  • If the second, why? Matlab is designed to efficiently and cleanly handle matrices without loops. But, since the reason may be "it's the assignment", you need to think about the pattern of the indices. To start off:
n = 9;
% start with an empty matrix
A = zeros(n);
% loop through the rows (or columns)
for ind=1:n
% the diagonal (col = row) is always -4
A(ind,ind) = -4;
% except in certain cases (rows 3 6 9)
if mod(ind,3) ~= 0
% the column after the diagonal has a 1
A(ind,ind+1) = 1;
end
...
end
% this gets you to
A
-4 1 0 0 0 0 0 0 0
0 -4 1 0 0 0 0 0 0
0 0 -4 0 0 0 0 0 0
0 0 0 -4 1 0 0 0 0
0 0 0 0 -4 1 0 0 0
0 0 0 0 0 -4 0 0 0
0 0 0 0 0 0 -4 1 0
0 0 0 0 0 0 0 -4 1
0 0 0 0 0 0 0 0 -4
p.s. sorry, I also missed the gaps for my earlier answer. Depending on the situation, an alternate strategy is to make the simple matrix with consistent diagonals, then adjust the outlier elements, something like this:
n = 9;
% simple matrix
A = full(spdiags([1 1 -4 1 1]+zeros(n,1), [-3 -1:1 3], n, n));
% set to zero the elements in row-3/col-4, row-6/col-7, etc.
A(sub2ind([n n],[3 6 4 7],[4 7 3 6])) = 0

Connectez-vous pour commenter.

Plus de réponses (1)

Bruno Luong
Bruno Luong le 27 Juil 2020
Modifié(e) : Bruno Luong le 27 Juil 2020

0 votes

m = 4;
n = 3*m;
T = diag(ones(1,n-3),3);
A = T + T';
B = ones(3)-5*eye(3);
A = A + kron(eye(m),B)

1 commentaire

Bruno Luong
Bruno Luong le 27 Juil 2020
Modifié(e) : Bruno Luong le 27 Juil 2020
If you insist on for loop
m = 4;
n = 3*m;
T = diag(ones(1,n-3),3);
A = T + T';
B = ones(3)-5*eye(3);
for k=1:m
i = (k-1)*3+(1:3);
A(i,i) = B;
end
A

Connectez-vous pour commenter.

Catégories

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by