Split matrix into N Equal Parts by rows

Hi,

I have an Nx10 matrix. How can I split this into three equally sized matrices (by number of rows) ? Is this something the reshape command can handle?

Thank you!

IP

Answer 1

[Matt J]

If N is a multiple of 3, you can download blkReshape,

https://www.mathworks.com/matlabcentral/fileexchange/115340-block-transposition-permutation-and-reshaping-of-arrays

N=12; M=10;

A=reshape(1:N*M,N,M),

B=blkReshape(A,[N/3,M],1,1,[]),

A =

1 13 25 37 49 61 73 85 97 109

2 14 26 38 50 62 74 86 98 110

3 15 27 39 51 63 75 87 99 111

4 16 28 40 52 64 76 88 100 112

5 17 29 41 53 65 77 89 101 113

6 18 30 42 54 66 78 90 102 114

7 19 31 43 55 67 79 91 103 115

8 20 32 44 56 68 80 92 104 116

9 21 33 45 57 69 81 93 105 117

10 22 34 46 58 70 82 94 106 118

11 23 35 47 59 71 83 95 107 119

12 24 36 48 60 72 84 96 108 120

B(:,:,1) =

1 13 25 37 49 61 73 85 97 109

2 14 26 38 50 62 74 86 98 110

3 15 27 39 51 63 75 87 99 111

4 16 28 40 52 64 76 88 100 112

B(:,:,2) =

5 17 29 41 53 65 77 89 101 113

6 18 30 42 54 66 78 90 102 114

7 19 31 43 55 67 79 91 103 115

8 20 32 44 56 68 80 92 104 116

B(:,:,3) =

9 21 33 45 57 69 81 93 105 117

10 22 34 46 58 70 82 94 106 118

11 23 35 47 59 71 83 95 107 119

12 24 36 48 60 72 84 96 108 120

[Inna Pelloso]

Wow! This is a seriously useful function! I wonder why there isn't a built in function like that!

Is there any way to add the last rows that are greater that the last multiple (e.g. the last two rows if I have 302 rows) to the first/middle/last block? This is a possible additional feature for this function.

[Matt J]

You can pad with additional rows, e.g.,

N=ceil(N/3)*3;
A(end+1:N,:)=nan;

Answer 2

[Walter Roberson]

pieces = 3;
N = size(A, 1);
part_sizes = floor(N/pieces) * [1 1];
part_sizes(end+1) = N - sum(part_sizes);
C = mat2cell(A, part_sizes, size(A,2));

C is now a cell array in which the rows are equally divided as possible, with the last block being shorter if necessary.

[Gorkem]

Thanks for the code. I was having trouble with the last block being too small compared to other blocks so I distributed residuals of the last block through the equally spaced blocks. It's not the best but here is the version I updated in this way:

pieces = 100;
N = size(A, 1);
part_sizes = floor(N/pieces) * [ones(1:pieces)];
Res = N - sum(part_sizes);
part_sizes(1:Res) = part_sizes(1:Res) + 1;
C = mat2cell(A, part_sizes, size(A,2));

[AlexDiz]

it's [ones(1, pieces)] instead of [ones(1:pieces)]

but it works perfectly !

thanks

Answer 3

[Matt J]

You can also use mat2tiles,

https://www.mathworks.com/matlabcentral/fileexchange/35085-mat2tiles-divide-array-into-equal-sized-sub-arrays

which doesn't require even multiples of the block size. However, it is never advisable to use cell arrays when you don't have to.

A=rand(14,6);
B=mat2tiles(A,[4,inf])  %group every 4 rows

B = 4x1 cell array
    {4x6 double}
    {4x6 double}
    {4x6 double}
    {2x6 double}