I have several multiple cell arrays with different sizes, and I want to find the intersection of all the arrays at the same time

[~,ia,ib] = intersect(C{1}(:,2),C{2}(:,2));

B = [C{1}(ia,2:3),C{2}(ib,3)];

Using the above code, I can find the intersection of 2 cell arrays. However, I have about 200 cell arrays with different sizes, how can I find the intersection for all the arrays

[dpb]

"... I can find the intersection of 2 cell arrays. However, I have about 200 cell arrays..."

Now you know why you shouldn't create multiple sequentially-numbered variables (or any other naming scheme); you've just created a nightmare to do anything with them effectively.

Redesign and use a single cell array to hold the lot and then you can, at least, iterate over it.

Better would be to show the use case that created all these and then the end result needed; probably somebody here can give far more efficient way to go at the end objective.

[Bruno Luong]

" why you shouldn't create multiple sequentially-numbered variables"

What numbered variables are you refered to? OP has cell C without number that stores various sets.

[dpb]

I guess I misinterpreted the plural in "I have about 200 cell arrays..." as 200 variables, not a single cell array of size 200.

Answer 1

[Bruno Luong]

This is what you want

https://fr.mathworks.com/matlabcentral/fileexchange/66614-mintersect-varargin

[Rukundo Wellen]

Dear Bruno, I have tried using that function but I still cannot get the answer

[Bruno Luong]

I cannot guess what exactly you have tried and why exactly you still cannot get the answer.

% Generate some dummy data (10 sets that intersect at least in 1:5)
C = cell(1,10);
for k=1:length(C)
    rset =  [1:5 ceil(100*rand(1,randi(5)))];
    rset = rset(randperm(end));
    C{k} = rset;
end
C{:}

ans = 1×10

54 4 1 75 9 2 5 41 68 3

ans = 1×7

3 2 49 5 76 1 4

ans = 1×8

91 17 3 2 4 1 5 95

ans = 1×10

38 3 79 92 2 62 66 1 5 4

ans = 1×6

71 5 3 1 2 4

ans = 1×10

76 4 1 3 5 2 40 57 75 48

ans = 1×9

1 3 89 39 90 4 64 5 2

ans = 1×9

36 47 1 3 14 73 4 5 2

ans = 1×9

3 5 45 86 5 4 59 1 2

ans = 1×10

27 1 5 3 5 2 4 47 63 2

mintersect(C{:})

ans = 1×5

1 2 3 4 5

function [S,varargout] = mintersect(varargin)
% [S, iA, iB, iC, ...] = mintersect(A, B, C, ...)
% Returns the data S common to numerical vectors A, B, C..., with no
% repetitions. Output S is in sorted order.
% Return in iA, iB, ... index vectors such that S = A(iA) = B(iB) = ...
%
% Syntax: [...] = mintersect(A, B, C, ..., 'rows')
% A, B, are arrays and must have the same number of column n. 
% MINTERSECT considers each row of input arguments as an entities
% for comparison. S is array of n-columns, each row appears at least once
% in each input array.
%
% See also: intersect, munion
% Author: Bruno Luong <[e-mail]>
s           = varargin(:);
rowflag     = ischar(s{end}) && strcmpi(s(end),'rows');
if rowflag
    s(end) = [];
end
nsets       = size(s,1);
m           = cellfun('size',s,1);
n           = cellfun('size',s,2);
isallrowv   = all(m==1);
if isallrowv
    m       = n;
    n       = 1;
else
    if any(diff(n))
        error('mintersect: input arrays must have the same number of columns');
    end
    n       = n(1);
end
cm          = [0; cumsum(m)];
A           = zeros(cm(end),n+1);
for k = 1:nsets
    r           = cm(k)+1:cm(k+1);
    A(r,1:n)    = s{k};
    A(r,end)    = k;
end
[v,K]       = uniquerow(A);
[u,I,J]     = uniquerow(v(:,1:end-1));
tf          = accumarray(J,1)==nsets;
S           = u(tf,:);
if isallrowv
    S = S.';
end
if nargout > 1
    nout  = nargout-1;
    if isempty(S)
        out         = cell(1,nout);
        [out{:}]    = deal([]);
    else
        i           = cumsum(accumarray(1+cumsum(m),-m)+1);
        iK          = i(K);
        a           = bsxfun(@plus, I(tf), (0:nsets-1));
        out         = num2cell(reshape(iK(a),size(a)),1);
    end
    varargout = out;
end
end % mintersect
%%
function [u,I,J] = uniquerow(a)
% perform [u,I,J] = unique(a,'rows') but without overhead
if size(a,2) == 1
    [b,K] = sort(a,'ascend');
else
    %[b,K] = sortrows(a,'ascend');
    [b,K] = sortrows(a); % R2014b does not recorgnize 'ascend' option
end
tf = [true; any(diff(b,1,1),2)];
u = b(tf,:);
%if nargout >= 2
I = K(tf);
if nargout >= 3
    J = cumsum(tf);
    J(K) = J;
end
%end
end % uniquerow

[Image Analyst]

@Rukundo Wellen, Show the hidden comments to see @Bruno Luong's answer where he had to make up some data because you keep forgetting to attach your own data. Is it surprising that the code didn't work for you when you didn't give us your data? Attach your actual cell arrays in a .mat file. Make it easy for people to help you not hard!

If you have any more questions, then attach your data and code to read it in with the paperclip icon after you read this:

TUTORIAL: How to ask a question (on Answers) and get a fast answer