Cutting (dividing) a matrix into a specific number of rows.

Hi, I have an exemplary matrix A(8,3) which contains 2 smaller matrices A1(4,3) and A2(4,3). I want to convert A into matrix B(12,2).

A=[11 12 13;21 22 23;31 32 33;41 42 43;51 52 53;61 62 63;71 72 73;81 82 83]

A = 8x3

11 12 13 21 22 23 31 32 33 41 42 43 51 52 53 61 62 63 71 72 73 81 82 83

B=[11 51;21 61;31 71;41 81;12 52;22 62;32 72;42 82;13 53;23 63;33 73;43 83]

B = 12x2

11 51 21 61 31 71 41 81 12 52 22 62 32 72 42 82 13 53 23 63

Are two for loops enough, or do I need more? How many to iterate to in each loop? I am trying to figure this out in many ways and still without success. Can I ask for help or at least a hint?

Answer 1

[Stephen23]

A = [11,12,13;21,22,23;31,32,33;41,42,43;51,52,53;61,62,63;71,72,73;81,82,83];
B = reshape(permute(reshape(A,[],2,3),[1,3,2]),[],2)

B = 12x2

11 51 21 61 31 71 41 81 12 52 22 62 32 72 42 82 13 53 23 63

[Radoslaw Puchalski]

OK, thanks. Unfortunately this example with 2 matrices A1 and A2 is just a minimalistic one. In reality, I will have thousands of such smaller matrices that I need to rearrange. The number of columns in A can also be larger. Will it work then, too? Why are there as many as 3 parameters in permute?

[Stephen23]

"In reality, I will have thousands of such smaller matrices that I need to rearrange. The number of columns in A can also be larger. Will it work then, too?"

Yes, when you know and specify the correct number of columns, etc, and all submatrices have the same sizes, etc.

Nothing I have shown you is limited to two submatrices. That is rather the point of using this general approach.

"Why are there as many as 3 parameters in permute?"

PERMUTE supports exactly two inputs, so presumably you are actually asking about its 2nd argument, the dimension order. I specified three dimensions because that is how many dimensions the intermediate array has.

[Radoslaw Puchalski]

It works great. Thank you so much.

Answer 2

[Alexander]

Just another approach. Simple but it might be helpful. I assume in the first half of your matrix are x or t, in the second half y data (measured or simulated channel 1). So the code below is only for one channel, but I think it's easy to modify if more channels are needed.

clear;

%{

% Your matrix:

A = [11,12,13

21,22,23

31,32,33

41,42,43

51,52,53

61,62,63

71,72,73

81,82,83]

%}

% Just to construct flexible test matrices remove until the "A" if you trust the code below. If you don't

% trust the

Rows = 9; Cols = 4; nn = 1;

for (ii = 1:Rows)

for (jj = 1:Cols)

A(ii,jj) = nn;

nn = nn +1;

end

%nn = nn +1;

end

A

% Here begins the code you migth need.

[Rows,Cols]=size(A);

if(mod(Rows,2)); fprintf(2,'No. of rows must be even\n'); return; end

ll = numel(A)/2;

B = []; C = [];

for(ii = Rows/2:Rows/2:Rows-1)

for(jj = 1:Cols)

B = [B;A(1:ii,jj)];

C = [C;A(1+ii:ii*2,jj)];

end

end

% The matrix you want to have:

B = [B, C]