Concatenate logical/numerical arrays element wise

I have five 2D arrays (912384x224) with either a logical 1 or 0 (essentially five logical masks). The five arrays are stored together in a single cell.

I wish to combine the logical arrays such that I have a new cell that combines the 1s and 0s of each array.

For example, if

H{1}(1,1)=1;

H{2}(1,1)=0;

H{3}(1,1)=1;

H{4}(1,1)=1;

H{5}(1,1)=0;

then the new cell array should be:

combinedH{1}(1,1)=[1,0,1,1,0] (or just 10110)

If I do the regular cat function, the arrays do not merge element-wise. Converting the values to string and then back to double after concatenation is memory consuming.

I can obtain the desired result through the below code, but it is also incredibly time-consuming.

for m=1:912384

for n=1:224

logic{m,n}=[Hmask{1}(m,n),Hmask{2}(m,n),Hmask{3}(m,n),Hmask{4}(m,n),Hmask{5}(m,n)];

end

end

Any help would be appreciated. I have tried many examples to concatenate without any success.

Regards

[James Tursa]

The main question you should ask yourself is how will this be used downstream in your code? That will determine what the best way to concatenate is. The easiest, of course, is just to concatenate them in the 3rd dimension into a 3D logical array. Another way is to combine them into bits of a 2D int8 matrix. Or perhaps into decimal numbers of a 2D int16 matrix such that the decimal conversion for display shows the 10110 etc. patterns. Bottom line is there are multiple ways to combine them, but you need to tell us how they will be used downstream in your code so we can reasonably advise you on the best way to do it. Is it just for display purposes? Will you still need to get at the original "bits"? Or ...?

[Aiman Raza]

Thanks for your reply. This was what I was looking for. I will use the concatenated matrix to create a binary sequence, which I will multiply further with a data matrix (double) of the same size.

Answer 1

[James Tursa]

Here is one way:

% Generate sample data
H{1} = rand(2,3)<0.5;
H{2} = rand(2,3)<0.5;
H{3} = rand(2,3)<0.5;
Hc = cat(3,H{:}) % combined

Hc = 2×3×3 logical array
Hc(:,:,1) =

   1   0   0
   0   0   1


Hc(:,:,2) =

   1   1   0
   1   0   1


Hc(:,:,3) =

   0   1   0
   1   1   0

% Extract a vector of the original data
squeeze(Hc(1,1,:))

ans = 3×1 logical array
   1
   1
   0

squeeze(Hc(1,2,:))

ans = 3×1 logical array
   0
   1
   1

Or you could permute the vectors into the 1st dimension. E.g.,

Hp = permute(Hc,[3,1,2]);

Then extraction is on the 1st dimension and doesn't need squeezing:

Hp(:,1,1)

ans = 3×1 logical array
   1
   1
   0

Hp(:,1,2)

ans = 3×1 logical array
   0
   1
   1

And yet another way if you want row vector output is to permute your vectors into the 2nd dimension. E.g.,

H2 = permute(Hc,[4,3,1,2]);

Then extraction is on the first two dimensions:

H2(:,:,1,1)

ans = 1×3 logical array
   1   1   0

H2(:,:,2,1)

ans = 1×3 logical array
   0   1   1

[Aiman Raza]

Thanks a lot for your response!