Sort a matrix according to the index of a vector

I have following labels for the rows and columns of a square matrix:

labels = {'dog', 'car', 'fish'}'

The corresponding matrix that expresses the semantic similarity between the terms is

D = [[0 0.8 0.2];
     [0.8 0 0.7];
     [0.2 0.7 0]];

I want to sort matrix and labels after following index

idx = [1 2 1];

Sorting the labels is easy:

[values,isort]=sort(idx);
labels_sort = labels(isort)

labels_sort =

3×1 cell array

    {'dog' }
    {'fish'}
    {'car' }

But how can i sort the matrix D accordingly?

[KL]

You have created the very same question 3 times. You haven't replied to the comments on your previous posts but just deleted them entirely. Please do not spam the forum, if the answer below is not what you are looking for, provide more information (such as the expected result) and ask follow-up questions.

[zeu]

You are absolutly right. It was my first time a postet a question here. Next time I ll think more before!

Answer 1

[KL]

if you want to sort the rows of D according to isort, try

 newD = D(isort,:)

newD =

           0    0.8000    0.2000
      0.2000    0.7000         0
      0.8000         0    0.7000

[zeu]

Thank you, but your answer is not what I am looking for. My solution should be this:

newD = [[0 0.2 0.8];
        [0.2 0 0.7];
        [0.8 0.7 0]];

So from the original matrix:

           Dog Car Fish
      Dog   DD  DC   DF
      Car   CD  CC   CF
      Fish  FD  FC   FF

I want to go to that matrix:

          Dog Fish Car
    Dog   DD   DF   DC
    Fish  FD   FF   FC
    Car   CD   CF   CC

In the end I want to do it for a much bigger matrix. So just swichting columnd and row 2 with 3 is not enough.

[KL]

Try this,

 newD(isort,:) = D(:,isort)

 newD =

           0    0.2000    0.8000
      0.2000         0    0.7000
      0.8000    0.7000         0

[zeu]

Thank you, this indeed works for that index but as soon as I use an index with repeating elements it stops working:

X = [[0 8 2 4 5 4];
     [8 0 7 3 5 8];
     [2 7 0 9 4 9];
     [4 3 9 0 2 3];
     [5 5 4 2 0 1];
     [4 8 9 3 1 0]];

 idx = [1 3 3 1 4 2];

[values,isort]=sort(idx);
newX(isort,:) = X(:,isort);

The result is:

 newX =

     0     4     4     8     2     5
     4     0     3     3     9     2
     5     2     1     5     4     0
     8     3     8     0     7     5
     4     3     0     8     9     1
     2     9     9     7     0     4

[Image Analyst]

Attach your huge matrix in a .mat file.

[zeu]

Thank you very much! Attached you find the matrix and the index.

[KL]

I still do not understand what your index array mean. But if you want to sort the upper(along 2nd dim) and lower triangular(along 1st dim) matrices separately within the matrix, then try

 newX = sort(tril(X))+sort(triu(X),2)

 newX = 

     0     2     4     4     5     8
     2     0     3     5     7     8
     4     3     0     4     9     9
     4     5     4     0     2     3
     5     7     9     2     0     1
     8     8     9     3     1     0

[zeu]

Your answer is unfortunatley not what I m looking for. The indexing array means that the words with the same index belong to the same category. E.g. "dog" and "fish" belong to category 1 (animals), thats why they have the index 1."Car" belongs to category 2 (man-made objects) and has index 2. In the end there will be around 6 different categories and a few hundred terms to sort after them.

[zeu]

Is it possible to solve it by uniting the matrix and the labels into a table and then sorting the table after the labels? Or maybe by using a name-value pair?