Finding the indices of duplicate values in one array

Given one array A=[ 1 1 2 3 5 6 7].

I need help to known the indices where there are duplicate values.

Thanks

Answer 1

[Stephan Koehler]

A = [1 2 3 2 5 3]
[v, w] = unique( A, 'stable' );
duplicate_indices = setdiff( 1:numel(A), w )

this should work too, and is elegant

[Jun W]

How about finding how many times are those elements repeated?

[Image Analyst]

Use histcounts and look for bins with more than 2 counts.

A = [1 2 3 2 5 3]
[counts, edges] = histcounts(A)

A =

1 2 3 2 5 3

counts =

1 2 2 0 1

edges =

Columns 1 through 5

0.5 1.5 2.5 3.5 4.5

Column 6

5.5

You can see that the bins for 2 and 3 both have 2 counts so there are multiples of 2 and 3 in A.

Note: This will find any repeats, and they don't have to be consecutive. If you want to look for consecutive repeats, call the diff() function and look for zeros.

Answer 2

[Image Analyst]

Here's one way:

A = [-2 0 1 1 2 3 5 6 6 6 7 11 40]
% Elements 3, 4, 8, 9, and 10 are repeats.
% Assume A is integers and get edges
edges = min(A) : max(A)
[counts, values] = histcounts(A, edges)
repeatedElements = values(counts >= 2)
% Assume they're integers
% Print them out and collect indexes of repeated elements into an array.
indexes = [];
for k = 1 : length(repeatedElements)
  indexes = [indexes, find(A == repeatedElements(k))];
end
indexes % Report to the command window.

You get [3,4,8,9,10] as you should.

[Image Analyst]

Arthur, with your new array A = [29892, 29051, 29051], my code still works. It returns 2 and 3. Can you tell me why you're still trying to use Adam's code even after I told you it doesn't work but mine does?

A = [29892, 29051, 29051]; 
% Elements 2 and 3 are repeats.
% Assume A is integers and get edges
edges = min(A) : max(A);
[counts, values] = histcounts(A, edges);
repeatedElements = values(counts >= 2)
% Assume they're integers
% Print them out and collect indexes of repeated elements into a cell array.
indexes = [];
for k = 1 : length(repeatedElements)
  indexes = [indexes, find(A == repeatedElements(k))];
end
indexes % Report to the command window.

[Arthur Souza]

because I don't have 'histcounts' function. Where do I find it?

[Image Analyst]

Then you have a version older than R2014b. What version do you have? This should work in old versions:

A = [29892, 29051, 29051]; 
% Elements 2 and 3 are repeats.
% Assume A is integers and get edges
edges = min(A) : max(A);
[counts, values] = histc(A, edges);
repeatedElements = edges(counts >= 2)
% Assume they're integers
% Print them out and collect indexes of repeated elements into an array.
indexes = [];
for k = 1 : length(repeatedElements)
  indexes = [indexes, find(A == repeatedElements(k))];
end
indexes % Report to the command window.

[Arthur Souza]

I have the 2013a version. and... IT WORKED! Thank you so much Image Analyst! I think my problem is solved now! Have a nice weekend!

[Tyann Hardyn]

You save my life (indirectly) again, Mr Image Analyst. Thank you so much. You helped someone else, then your help will be a good answer for the others, like me, lol.

[Roger]

I know this is an old thread but if anyone is still out there... I see a flaw in this method. Change A to include a duplicate of 40: A = [-2 0 1 1 2 3 5 6 6 6 7 11 40 40]

You still get [3,4,8,9,10] as a result. This method doesn't pick up the duplicate of 40. The reason is because counts is one index short of edges and it misses that last duplicate. Seems to be inherent in how histcounts works. Just need to be aware of that.

[Steven Lord]

The last bin includes both the left and right edges, while the earlier bins include only the left edges. This is stated in the description of the edges input on the histcounts documentation page: "Bin edges, specified as a vector. The first vector element specifies the leading edge of the first bin. The last element specifies the trailing edge of the last bin. The trailing edge is only included for the last bin."

So add on a number that's greater than the maximum element in your data. Inf is a good choice.

A = [-2 0 1 1 2 3 5 6 6 6 7 11 40 40]

A = 1×14

-2 0 1 1 2 3 5 6 6 6 7 11 40 40

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

edges = [min(A):max(A) Inf];
[counts1, edges1] = histcounts(A, edges);
repeatedElements = edges1(counts1 >= 2)

repeatedElements = 1×3

1 6 40

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

Or you could use non-equally spaced bins containing the unique elements from your data.

[counts2, edges2] = histcounts(A, [unique(A) Inf])

counts2 = 1×10

1 1 2 1 1 1 3 1 1 2

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

edges2 = 1×11

-2 0 1 2 3 5 6 7 11 40 Inf

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

repeatedElements = edges2(counts2 >= 2)

repeatedElements = 1×3

1 6 40

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

If your data spans a wide range this can reduce the number of bins histcounts uses.

whos counts1 edges1 counts2 edges2

  Name         Size            Bytes  Class     Attributes

  counts1      1x43              344  double              
  counts2      1x10               80  double              
  edges1       1x44              352  double              
  edges2       1x11               88  double              

The unique approach uses 10 bins, the non-unique approach uses 43. This is a fairly small difference for your sample A, but the impact is much larger if you have a distant outlier.

B = [A 5000]; % 5000 is far from the rest of the elements in A
edges = [min(B):max(B) Inf];
[counts1, edges1] = histcounts(B, edges);
[counts2, edges2] = histcounts(B, [unique(B) Inf]);
whos counts1 edges1 counts2 edges2

  Name         Size              Bytes  Class     Attributes

  counts1      1x5003            40024  double              
  counts2      1x11                 88  double              
  edges1       1x5004            40032  double              
  edges2       1x12                 96  double              

[Walter Roberson]

@Steven Lord

Using edges = [min(B):max(B) Inf]; assumes that the input data is integer.

Answer 3

[Adam]

[~, uniqueIdx] = unique( A );
duplicateLocations = ismember( A, find( A( setdiff( 1:numel(A), uniqueIdx ) ) ) );

then

find( duplicateLocations )

will give you the indices if you want them rather than a logical vector.

There are probably neater methods though.

If you want only the duplicates after the first then simply

setdiff( 1:numel(A), uniqueIdx )

should do the job.

[Arthur Souza]

I tried this:

[U,I]=unique(A(:,1)); repeated=setdiff(1:size(A,1),I)

but MATLAB returns me this -> Error using unique Too many input arguments. Error in setdiff>setdiffR2012a (line 505) c = unique(c,order); Error in setdiff (line 84) [varargout{1:nlhs}] = setdiffR2012a(varargin{:});

Can somebody help me? I'm new here

[Image Analyst]

Adam, this doesn't work in general:

A=[ 1 1 2 3 5 6 6 7]
[~, uniqueIdx] = unique( A );
duplicateLocations = ismember( A, find( A( setdiff( 1:numel(A), uniqueIdx ) ) ) )
indexes = find(duplicateLocations)

It gets the wrong indexes for the repeated 6's:

duplicateLocations =
  1×8 logical array
   1   1   1   0   0   0   0   0
indexes =
     1     2     3

It should give [1,2,6,7].

Arthur, your code worked for me for the A given. If it doesn't work for you, give us your A.

[Arthur Souza]

My A is an arbitrary vector, like this one you used here. I would like to know why this error occurs and try to fix it. I have to find these indexes to use them on another vector. Don't know why, but the A you showed here didn't work for me =/

[Arthur Souza]

I am trying with an A like this: A = [29892, 29051, 29051]; But it still doesn't wokr for me.

[Jan]

"Doesn't work" is a weak description of the problem. Please post the error message or explain the difference between the results and your expectations.

[Arthur Souza]

This is the error message -> Error using unique Too many input arguments. Error in setdiff>setdiffR2012a (line 505) c = unique(c,order); Error in setdiff (line 84) [varargout{1:nlhs}] = setdiffR2012a(varargin{:});

[Arthur Souza]

My problem is the same as the topic of this forum: Finding the indices of duplicate values in one array. I use the same solution that has been put here, but only this error message is returned to me

[CompViscount]

Commenting here as it's led me to overall the best answer here, it just has a mistake. The "find" in the 2nd line changes the values into indices before passing to ismember, which just makes the output nonsense. I removed that. Using the same numbers as image analyst above:

A=[ 1 1 2 3 5 6 6 7]

A = 1×8

1 1 2 3 5 6 6 7

[~, uniqueIdx] = unique(A);
dupeIdx = ismember( A, A( setdiff( 1:numel(A), uniqueIdx ) ) );
dupes = A(dupeIdx)

dupes = 1×4

1 1 6 6

dupeLoc = find(dupeIdx)

dupeLoc = 1×4

1 2 6 7

[Gabor]

This works, thanks

Answer 4

[Jan]

function Ind = IndexOfMultiples(A)
T   = true(size(A));
off = false;
A   = A(:);
for iA = 1:numel(A)
  if T(iA)          % if not switched already
    d = (A(iA) == A);
    if sum(d) > 1   % More than 1 occurrence found
       T(d) = off;  % switch all occurrences
    end
  end
end
Ind = find(~T);
end

If the input has more than 45 elements, this is faster:

function T = isMultiple(A)
% T = isMultiple(A)
% INPUT:  A: Numerical or CHAR array of any dimensions.
% OUTPUT: T: TRUE if element occurs multiple times anywhere in the array.
%
% Tested: Matlab 2009a, 2015b(32/64), 2016b, 2018b, Win7/10
% Author: Jan, Heidelberg, (C) 2021
% License: CC BY-SA 3.0, see: creativecommons.org/licenses/by-sa/3.0/

T        = false(size(A));
[S, idx] = sort(A(:).');
m        = [false, diff(S) == 0];
if any(m)        % Any equal elements found:
   m(strfind(m, [false, true])) = true;
   T(idx) = m;   % Resort to original order
end
end

[GONZALEZ DE COSSIO ECHEVERRIA Francisco Jose]

Any way to speed up this code? Thanks

[Muhammad Imran]

Thanks Jan

Answer 5

[MRINAL BHAUMIK]

A=[ 1 1 2 3 5 6 7 6]
B = A'./A
B = B-diag(diag(B))
[pos1 pos2]=find(B==1)

o/p
pos1 =

     2
     1
     8
     6

Answer 6

[Anamika]

In MATLAB, you can find the indices of duplicate values in an array using the `find` function along with the `unique` function. Here's how you can do it:

A = [1 1 2 3 5 6 7];

% Finding the unique elements in the array

unique_elements = unique(A);

% Initializing an empty array to store the indices of duplicate values

duplicate_indices = [];

% Iterating through each unique element

for i = 1:numel(unique_elements)

% Finding the indices of occurrences of the current unique element

indices = find(A == unique_elements(i));

% If there are more than one occurrence, add the indices to the duplicate_indices array

if numel(indices) > 1

duplicate_indices = [duplicate_indices indices];

end

end

% Displaying the indices of duplicate values

disp(duplicate_indices);

Running this code will give you the indices of the duplicate values in the array A. In this case, the output will be: 1 2

This means that the duplicate values are located at indices 1 and 2 in the array A.

Answer 7

[Eduardo Gonzalez Rodriguez]

Here is my solution to find repeated values and their counts

function [dup, counts] = duplicates(A)
    [dup,~,n] = unique(A, 'rows', 'stable');
    counts = accumarray(n, 1, [], @sum);
    dup(counts==1) = [];
    counts(counts==1) = [];

Answer 8

[Piotr]

Hello,

here is my attempt to solve it. I faced similar problem but in my case I wanted to have the result in two column representation. Each row contains indices of repeated values.

A = [ 1 1 2 3 5 6 7 6];
nk = nchoosek(1:length(A),2);
nk(diff(A(nk),[],2)~=0,:) = [];
disp(nk)

Cheers, Piotr