MATLAB Answers

lower triangular semidefinite matrix

6 views (last 30 days)
how can i build a sparse matrix in this form for a given semidefinite matrix? i only need the sparse lower triangular of semidefinite matrix
for example a semidefinite matrix as Q0:
2 0 -1
Q0 = 0 0.2 0
-1 0 2
is written as:
i = [1 2 3 3 ];
j = [1 2 3 1 ];
c = [2 0.2 2 -1 ];
where i shows number of row
j shows number of column
and c is the elemnt that belong to (i,j) that is not zero and belong to lower triangular of semidefinite matrix

  2 Comments

Guillaume
Guillaume on 4 May 2020
I'm not sure I understand the problem, sparse(i, j, c) would create your sparse matrix.
Guillaume
Guillaume on 4 May 2020
Albert Stirling's comment mistakenly posted as an Answer moved here:
I need a simple algorithm that select the non-zero values in shown lower triangular part of this symmetric matrix and return it as below:
i = [1 2 3 3 ];
j = [1 2 3 1 ];
c = [2 0.2 2 -1 ];
i need the algorithm to be flexible so i can give it any desired symmetric matrix and have an answer like the one mentioned.

Sign in to comment.

Accepted Answer

Guillaume
Guillaume on 4 May 2020
[i, j, c] = find(tril(yourmatrix))
seems to be what you're after.

  1 Comment

Albert Stirling
Albert Stirling on 4 May 2020
thank you so much that was a quick way to handle this problam.
however i found a long one myself as below:
Q0=[-2 0 0.2;
0 -2 0;
0.2 0 -0.2]
[, c]=size(Q0);
for i=1:c
for j=1:3-i
Q0(i,i+j)=0;
end
end
Q1=sparse(Q0);
[i,j,s] = find(Q1);
f=[i,j,s]'

Sign in to comment.

More Answers (1)

John D'Errico
John D'Errico on 4 May 2020
Edited: John D'Errico on 4 May 2020
You just want the lower triangle elements. semi-definite-ness is irrelevant to what you asked for at the end.
A = sprand(5,5,.3);
>> full(A)
ans =
0.46421 0.83266 0 0.022104 0.18026
0.26627 0 0 0 0
0 0 0 0 0
0 0 0 0.92865 0
0 0 0.37763 0.42783 0
>> [Rind,Cind,val] = find(tril(A))
Rind =
1
2
5
4
5
Cind =
1
1
3
4
4
val =
0.46421
0.26627
0.37763
0.92865
0.42783
That is how you would EXTRACT the lower triangle into a set of row and column indices, and the non-zero elements in those positions. Which on the face of it, seems to be your question. However, my guess is that you really want to build the sparse matrix from those values? Or, perhaps you just don't understand that sparse matrices already exist in MATLAB and can be used as such? It is not at all clear what is the real problem.
Regardless, if you wanted to build the matrix as a sparse one, then just call sparse using those vectors.
Atril = sparse(Rind,Cind,val,5,5)
Atril =
(1,1) 0.46421
(2,1) 0.26627
(5,3) 0.37763
(4,4) 0.92865
(5,4) 0.42783
>> full(Atril)
ans =
0.46421 0 0 0 0
0.26627 0 0 0 0
0 0 0 0 0
0 0 0 0.92865 0
0 0 0.37763 0.42783 0

  1 Comment

Albert Stirling
Albert Stirling on 4 May 2020
thank you, your solution seems to be a good way to solve this problam. thank you again.

Sign in to comment.

Products


Release

R2019b

Translated by