possible bug in "unique" function

Dear all,

Could you, guys, help me please. I have a matrix

 X =

    2.2000    3.3000
    2.2000    3.3000
    4.0000    5.0000
    3.0000    5.0000

So I would like to find matrix A with unique rows in it.

So I use A=unique(X,'rows'). I get:

 A =

    2.2000    3.3000
    3.0000    5.0000
    4.0000    5.0000

Which is great. Then I go to my larger example. The size is 675041x2. I run the same line A=unique(X,'rows') as above. However, it does not give me unique numbers it has a lot of rows that are repeated. Grrrr. Moreover, I expect the first row in X to be equal to the first row in A. But in my case, it is not. Any suggestions please. The only thing I can think of is the truncation of floats. For example, the first element in my original X matrix is equal to:

X(1,:)

ans =

     1.0e+06 *

      1.2123    0.9458

Not sure at all

Answer 1

[Walter Roberson]

format long g
X(1,:)  - A(1,:)

Answer 2

[the cyclist]

I suggest that you type "format long" to see more of the digits that MATLAB is carrying along in the calculation. I expect you will see differences in the less-significant digits.

Such comparisons between floating-point numbers can be tricky. Perhaps you could do some rounding before using the unique() function.

Also, unique() sorts the result, which is why the ordering is "wrong" from your point of view. However, the unique() function also outputs indices that allow you to see exactly how the array was sorted. "doc unique" for details.

[Peter Perkins]

"[C,IA,IC] = unique(A,'stable') returns the values of C in the same order that they appear in A, while [C,IA,IC] = unique(A,'sorted') returns the values of C in sorted order."

Answer 3

[Stephen23]

There is no bug. You need to learn about the practical limits of floating point numbers:

http://www.mathworks.com/help/matlab/matlab_prog/floating-point-numbers.html

http://www.mathworks.com/matlabcentral/answers/69-why-does-1-2-3-1-3-not-equal-zero

http://matlab.wikia.com/wiki/FAQ#Why_is_0.3_-_0.2_-_0.1_.28or_similar.29_not_equal_to_zero.3F

http://www.mathworks.com/matlabcentral/answers/57444-faq-why-is-0-3-0-2-0-1-not-equal-to-zero

http://www.mathworks.com/company/newsletters/news_notes/pdf/Fall96Cleve.pdf

http://www.mathworks.com/matlabcentral/answers/41536-why-is-2-24-10-22-4-not-equal-to-0

https://www.mathworks.com/matlabcentral/newsreader/view_thread/66553

http://www.mathworks.com/company/newsletters/news_notes/pdf/Fall96Cleve.pdf

http://blogs.mathworks.com/loren/2006/08/23/a-glimpse-into-floating-point-accuracy/

http://www.mathworks.com/help/fixedpoint/ug/floating-point-numbers.html

And some external links on this topic:

http://www.lahey.com/float.htm

http://matlabgeeks.com/tips-tutorials/floating-point-comparisons-in-matlab/

Answer 4

[Steven Lord]

Since this question was originally asked, in release R2015a, we introduced the UNIQUETOL function. You can call UNIQUETOL with a small tolerance to consolidate elements or rows that are "almost" the same when generating a list of unique elements or rows.