Matrix transformation (sorting)
5 views (last 30 days)
Show older comments
I have a sparse matrix containing either 0 or 1, for example :
A = [0; 1; 0];
I would like to find a transformation to sort this matrix, i.e. find another matrix such that:
T*A = [1 ; 0 ; 0]
All the 1 should come first.
In this example, T should be :
T = [0 1 0; 0 0 0; 0 0 0];
I can not use builtin functions such as "sort" because I need to apply this transformations to other matrices.
i.e. once I have calculated T corresponding to the A matrix, I need to apply the same transformation to other (not particularly sorted) matrices.
How can I write an algorithm for more complicated examples ? Is there a builtin function that does this already ?
Best regards,
Peter
0 Comments
Accepted Answer
the cyclist
on 4 Feb 2020
Edited: the cyclist
on 4 Feb 2020
i = find(sort(A,'descend'));
j = find(A);
s = numel(A);
T = zeros(s,s);
T(sub2ind([s s],i,j)) = 1;
I hope it's clear what is going on here.
Note that in this case, the pseudoinverse of T is equal to the transpose of T. The transpose is actually the easier way to understand what is happening. If there is a 1 in element (x,y) of T, then there has to be a 1 in element (y,x) of the inverse transformation. (You can think of T and pinv(T) as just "sending" 1's back and forth to the appropriate positions.)
More Answers (1)
the cyclist
on 3 Feb 2020
Edited: the cyclist
on 3 Feb 2020
T = sort(A,'descend')/A
Note that I am only using sort here to define the result you need. The transformation matrix you get as a result doesn't rely on any actual sorting algorithm, and can be applied you your other matrices directly.
See Also
Categories
Find more on Shifting and Sorting Matrices in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!