How do I split an array into small array of fixed size with increments by 1?

For example, I have an array, say

aa = [2 5 6 10 2 4 2 1 -1 -12 0 3 7];

What I want to do is create arrays of length 4, such as

[2 5 6 10], [5 6 10 2], [6 10 2 4], [10 2 4 2], [2 4 2 1], [4 2 1 -1], ....

and group them into a matrix. I understand this can be done in a for loop, but is there a way to vectorize this process?

Answer 1

[Bruno Luong]

A=randi(10,1,10)

A = 1×10

3 10 1 4 8 8 9 3 10 6

k=4;
B = hankel(A(1:end-k+1),A(end-k+1:end))

B = 7×4

3 10 1 4 10 1 4 8 1 4 8 8 4 8 8 9 8 8 9 3 8 9 3 10 9 3 10 6

[DGM]

Good gravy. I thought "oh I could do this with toeplitz()" but figured it'd just be slightly more awkward to flip everything, so I didn't.

[Alekhya Hati]

I found another method as well later on that worked for me

aa = [2 5 6 10 2 4 2 1 -1 -12 0 3 7];
bb = buffer(aa, 4, 3, 'nodelay')

bb = 4×10

2 5 6 10 2 4 2 1 -1 -12 5 6 10 2 4 2 1 -1 -12 0 6 10 2 4 2 1 -1 -12 0 3 10 2 4 2 1 -1 -12 0 3 7

[Bruno Luong]

I never use this command buffer, so I learn something.

It requires the signal processing toolbox though.

Answer 2

[DGM]

I'm sure there are other ways, but here's one way.

% the input
A = [2 5 6 10 2 4 2 1 -1 -12 0 3 7];

% the parameter
blocklen = 4;

% generate a 2D array of indices
nrows = numel(A)-blocklen+1;
idx = (1:blocklen) + (0:nrows-1).';

% build the output by direct addressing
B = A(idx)

B = 10×4

2 5 6 10 5 6 10 2 6 10 2 4 10 2 4 2 2 4 2 1 4 2 1 -1 2 1 -1 -12 1 -1 -12 0 -1 -12 0 3 -12 0 3 7