Extract unknown number of vectors from matrix

Hey there,

I want to use the function sub2ind on the column vectors of a matrix M, let's say of size . The function sub2ind asks for: ind = sub2ind(sz,I1,I2,...,In), i.e. I would need n vectors of size as input. If I'd know the (small) number of vectors, of course I could go for I1 = M(:,1), I2 = M(:,2), ...

The problem: it doesn't run on the matrix M directly and I don't know the number of columns n beforehand. num2cell would create separated cells with the n vectors of size , but sub2ind also does not run on cells.

I couldn't find a solution for the problem; how do I get efficiently n separate vectors from a matrix with unknown size n? Is that possible? If not, how can I solve the described problem with sub2ind?

Thanks a lot for your help!

Answer 1

[Stephen23]

" num2cell would create separated cells with the n vectors of size, but sub2ind also does not run on cells."

True, but you can easily use a cell array to create a comma-separated list, e.g.:

spl = num2cell(M,1);
idx = sub2ind(sz,spl{:});

https://www.mathworks.com/help/matlab/matlab_prog/comma-separated-lists.html

https://www.mathworks.com/matlabcentral/answers/320713-how-to-operate-on-comma-separated-lists

"how do I get efficiently n separate vectors from a matrix with unknown size n? Is that possible?"

Yes, use num2cell to create a cell array of vectors.

"how can I solve the described problem with sub2ind?"

Use a comma-separated list generated from that cell array.

[math4]

Thanks for getting back to me! No your code isn't doing the right thing - I don't want every entry to be comma-separated, but the two vectors contained in separate cell entries.

However, I just found a mathworks extension of sub2ind to matrices, great!

[math4]

If anyone else should ever need it, here we go: https://www.mathworks.com/matlabcentral/fileexchange/36458-augmented-version-of-sub2ind?s_tid=answers_rc2-1_p4_MLT

Just be carefule about the transposed input dimension!

[Stephen23]

"No your code isn't doing the right thing"

Really? Did you even try it? It gives exactly the same output as that (badly-named) third party code you linked to. I downloaded that FEX function, renamed it to sub2idx, and compared its output aginst the (very standard MATLAB) code that I gave you in my answer. Lets have a look:

>> M = randi(9,3,5);  % 3x5 random indices
>> S = 9*ones(1,5);   % size
>> C = num2cell(M,1); % split
>> X = sub2ind(S,C{:})
X =
   20421
    8960
   52607
>> Y = sub2idx(S,M.') % FEX 36458
Y =
   20421    8960   52607

So far I am struggling to see why you would write that my "code isn't doing the right thing", given that it returns exactly the same values as what your linked-to FEX code does, just without any third-party functions required.

[math4]

Yes, of course I tried, but just quickly and my sincere apologies - I mismatched dimensions, since I had it transposed for the other code. Thanks for your solution, that's great.