In [6]: a = float('nan')
In [7]: a == a
Out[7]: False
Why?
Asked
Active
Viewed 1,703 times
-1
-
6Because `NaN != NaN` – styvane Jul 10 '17 at 22:13
-
2@cᴏʟᴅsᴘᴇᴇᴅ Not really. He asking about _why_ they behave a certain way, not _how_ they behave a certain way. – Christian Dean Jul 10 '17 at 22:14
-
@ChristianDean He'll need to visit https://math.stackexchange.com then – cs95 Jul 10 '17 at 22:14
-
2The correct test probably is `math.isnan(a)`. – Paulo Scardine Jul 10 '17 at 22:16
-
@cᴏʟᴅsᴘᴇᴇᴅ That, I'd tend to agree with you on. What he's asking about really isn't specific to any programming language. That's how most languages implement `NaN`. – Christian Dean Jul 10 '17 at 22:17
-
By definition, `NaN` fails *any* comparison. – Prune Jul 10 '17 at 22:31
1 Answers
8
There are two reasons, one practical and one historical. The historical reason is that this makes it trivial to test for a NaN. If the value isn't equal to itself, it's a NaN.
The practical reason is that NaN is used to indicate various different cases where the value is not defined, and not all undefined values are equal.
Should NaN/NaN equal 1? Think about how crazy that would be. Arguing that NaN should equal itself is only a bit less crazy.
David Schwartz
- 179,497
- 17
- 214
- 278
-
1Interesting. To your question of Should NaN/NaN equal 1? I think the answer I would give is: simply don't allow division by NaN since it doesn't represent a number, it's like dividing by a string or something – nichochar Jul 10 '17 at 22:16
-
@nichochar The way you typically disallow something is by having it return a `NaN`. What precisely are you suggesting? – David Schwartz Jul 10 '17 at 22:17
-
1
-
1Yes. This allows a `NaN` to 'bubble up' through a series of computations, so you only have to test the final result for `NaN`. – David Schwartz Jul 10 '17 at 22:19
-
I was curious if `math.isnan` would be defined simply as `return a != a` (it is not). – Paulo Scardine Jul 10 '17 at 22:28