How to find rows with multiple entries that are equal?

I have a NxM matrix and I want to find the rows that have M-1 duplicate entries. For example:

A=[1 1 2 4;
   2 2 2 1;
   2 9 3 0;
   3 0 3 3;
   4 3 2 2];
fun(A);
ans = [2 2 2 1;
       3 0 3 3]

How could I do this, preferably elegantly in a few lines? I am trying to avoid loops because I anticipate working with large matrices eventually.

Answer 1

[Kirby Fears]

In two lines:

sA=sort(A,2);
A(sum(sA(:,2:end)==sA(:,1:end-1),2)==size(sA,2)-2,:)
  ans =
     2     2     2     1
     3     0     3     3

Here's another method that doesn't use sort(). It's slightly faster for large matrices on my machine:

rowIdx=(sum(repmat(A(:,1),1,size(A,2))~=A,2)==1) | ...
    (sum(repmat(A(:,2),1,size(A,2))~=A,2)==1);
A=A(rowIdx,:);

You might want to break that down into more lines of code for readability.

Hope this helps.

Answer 2

[the cyclist]

A(sum(diff([nan(size(A,1),1),sort(A,2)],1,2)==0,2)==(size(A,2)-2),:)

Answer 3

[Andrei Bobrov]

out = A(any(hist(A')==3),:);

[Colton]

Thank you. Needs the correction below though.

M=size(A,2);
out= A(~any(hist(A')==(M),:);