How to find repeated sequence , and the times of repeat, in array by Matlab?

I want to input an array e.g. 12341212356

and find all repeated sequences(and times)in that array like: 12 : repeat 3 times 123 : repest 2 times

How could I use Matlab to do this work efficiently?

thanks!

---

thanks for the comments! supplement explanation for my question:

1.The length of sequence need to large than 1 otherwise every data can be a pattern, that's not what I want.

2.Every sequence that appear more than two times will be record.

3.I want to find every patterns that agree with the rules mentioned above.

e.g. 121212 will be 12:trice, 121:twice, 212 twice, 1212:twice e.g. 123234345 will be 23:twice, 34:twic

ps: I don't really care about the output form right now. I think a matrix that record those information will be fine.

thanks

[Andrew Newell]

Are overlaps allowed? E.g., in 121212 is 1212 repeated once or twice?

[Daniel Shub]

You need to supply more information.

Do you want the repeats for sequences you specify or all sequences? What about sequences of length 1?

What doe you expect the output to look like?

Can you give a couple of examples (worked out by hand) of inputs and outputs.

Answer 1

[Matt Fig]

Here is another version which returns in strings and only returns those patterns that repeat and are of length >1:

    % This code works with numbers or vectors of digits.
    % M = round(rand(1,500)*3);  % A bigger vector, try this out!
    M = 123412123562356; % An example number to try...
    As = sprintf('%.0f',M);
    cnt = 1;
    L = length(As);

    for jj = 1:L-2
        for ii = jj+1:L
            I = strfind(As,As(jj:ii));
            if length(I)>1
                T{cnt} = sprintf('%s_%i',As(jj:ii),length(I));
                cnt = cnt + 1;
            end
        end
    end

    T = unique(T)

Note that this could easily be modified to return numeric results instead...

%

%

%

%

%

EDIT Update with a functional form.

I made this into a function which is more general. For instance, you will want to switch to using a vector if your length gets long enough because double precision limits how many digits you can hold. For example:

    M = 123412123562356778834239877;

Would be better represented as a vector of digits, as it cannot be stored exactly. The following works fast on vectors of numbers or characters, and scalar numbers like M above (subject to the discussed limits).

function T = find_patterns(M)
%FIND_PATTERNS finds all repeating patterns in a vector or number.
% T = FIND_PATTERNS(M) returns an n-by-2 cell array.  For each row of the
% cell array, column one has the pattern, and column two has the number of
% times the pattern occurs.
% Takes as input either a vector of digits or charcters, or a scalar number  
% and finds all repeating patterns of length greater one. 
% If the input argument is numeric, the output argument will be numeric,
% and if the input argument is a vector of characters, the output argument
% will be a vector of characters.
%
% Examples:
%
%  M = 123412123562356; % A scalar number.
%  T = find_patterns(M);
%  T{8,:}  % Show that [2 3 5 6] occurs 2 times.
%
%  M = 'c31a234121a23562c356'; % A character array.
%  T = find_patterns(M);
%
%  M = round(rand(1,300)*3); % A vector of digits.
%  T = find_patterns(M);
%  T{1,:}
%
%
% Author: Matt Fig

if ~isscalar(M) && ~isvector(M)
    error('Only scalar and vector arguments are allowed.');
end

flg = 0;

if ~ischar(M)
    M = sprintf('%.0f',M);
    flg = 1;
else
    M = M(:).'; % Make sure we have a row vector...
end

cnt = 1;
L = length(M);
T = {};

for jj = 1:L-2
    for ii = jj+1:L-1
        I = strfind(M,M(jj:ii));
        if length(I)>1
            T{cnt} = sprintf('%s %i',M(jj:ii),length(I));%#ok
            cnt = cnt + 1;
        end
    end
end

if ~isempty(T)
    T = regexp(unique(T),'\ ','split');
    T = cat(1,T{:});

      if flg
          T(:,1) = cellfun(@(x) str2num(x.').',T(:,1),'Un',0);%#ok
          T(:,2) = cellfun(@str2double,T(:,2),'Un',0);
      end

      [J,J] = sort(cellfun('length',T(:,1)));
      T = T(J,:);
end

[Xiaonan Xing]

Your function is just too beautiful and I have to log in to my MATLAB account to give you an upvote, also my very first upvote. Thanks for the code!

Answer 2

[Andrei Bobrov]

variant

A = str2num(num2str(12341212356)')
N = length(A);
out = cell(N-1,1);
for j = 1:N-1
    [a, ~, n] = unique(A(bsxfun(@plus,1:j,(0:N-j)')),'rows');
    out{j} = [a sum(bsxfun(@(x,y)x==y,1:size(a,1),n))'];
end

repeated combinations:

o2 = cellfun(@(x)x(x(:,end) > 1,:),out,'un',0)
out2 = o2(~cellfun('isempty',o2))

in each cell of the last column the number of combinations, the combination of the left hand side