Why are the return values of these doubles -1.#IND?

Question

I have :

double score = cvMatchContourTrees( CT1, CT2, CV_CONTOUR_TREES_MATCH_I1, 0.0 );
        cout<<score<<endl;

There are values returned as -1.#IND. Other than that, the positive values are normal, like 1.34543.

Why does this happen? How do I solve it?

`#IND` is [NaN](http://en.wikipedia.org/wiki/NaN). You might want to read [this](http://www.johndcook.com/IEEE_exceptions_in_cpp.html). — Frédéric Hamidi, Sep 19 '11 at 19:27
`cvMatchContourTrees` and OpenCV are complete red herrings in this question. — Lightness Races in Orbit, Sep 19 '11 at 19:58
@Tomalak, how can you be sure? Is the question "What does -1.#IND mean?" or is it "How did cvMatchContourTrees return that value?" — Mark Ransom, Sep 19 '11 at 20:40
See: http://blogs.msdn.com/b/oldnewthing/archive/2013/02/21/10395734.aspx and http://blogs.msdn.com/b/oldnewthing/archive/2013/02/28/10397976.aspx — Michael Burr, May 07 '14 at 20:49

score 15 · Answer 1 · edited Jun 20 '20 at 09:12

15

As Frederic says, it's the result of a 'Not a Number' being formatted by an application built with visual studio on windows. John D Cook has an excellent reference:

Windows displays a NaN as -1.#IND ("IND" for "indeterminate") while Linux displays nan.

...

In short, if you get 1.#INF or inf, look for overflow or division by zero. If you get 1.#IND or nan, look for illegal operations.

Watch out for truncations if you do any sort of formatting with your string; I've encountered related issues when handling these sorts of errors myself.

edited Jun 20 '20 at 09:12

Community

1
1

answered Sep 19 '11 at 20:41

Jon Cage

36,366
38
137
215

I'm not sure about `0/0` yielding `NaN`. I think it's like any other division by zero in that you'll get a floating-point exception from the OS. – Lightness Races in Orbit Sep 19 '11 at 20:45
@Tomalak: With Visual Studio, 0 / 0 throws an exception. 0.0 / 0.0 either returns NaN or throws an exception, you control that by a flag. The default is that it returns NaN. – Johan Råde Sep 19 '11 at 20:49
@user763305: Yes, exactly. And that's not what the original quote said. My point is, beware the perils of Wikipedia... – Lightness Races in Orbit Sep 19 '11 at 20:50
Maybe.. ...or maybe it's in a library call and that library silently masks the exception but still spits out NaN as per the OPs question... ? – Jon Cage Sep 19 '11 at 20:51

score 5 · Answer 2 · answered Sep 19 '11 at 19:57

5

std::cout << (0/0.f);
// Output: -1.#IND

It's NaN.

answered Sep 19 '11 at 19:57

Lightness Races in Orbit

378,754
76
643
1,055

score 3 · Answer 3 · answered Jun 18 '12 at 15:01

3

In my experience -1.#IND comes from imaginary numbers. So, doing cout << sqrt(-1.); should output -1.#IND

answered Jun 18 '12 at 15:01

viraj

1,784
4
35
52

Why are the return values of these doubles -1.#IND?

3 Answers3

Linked

Related