C++ Precision behavior with double/long double

Question

I am just curious about something.

I have the following code:

if (d1 == 2.3) cout << "2.3 is my value\n"; if (d1 == 2.2999999999999998) cout << "2.2999999999999998 is my value\n";

VS2013 c++11 screenshots

enter image description here

And it goes into both ifs. I know the precision for double is with 15decimals, so I will have to use a more appropiated type for this kind of data.

Could anybody link me a detailed reference? And also a "way" to storage just "2.3" or a data with more precision than 2.2999999999999998? (long double throws me complie error

Error 1 error C2398: Element '1': conversion from 'long double' to 'const std::complex<double>::_Ty &' requires a narrowing conversion...

Thanks.

Edit: Added complex

enter image description here

Try reading over this question. Also, did you try complex<long double>? — jxh
– jxh, Commented Feb 7, 2014 at 10:05
Yes, I tried with complex<long double> and there is no compile error, but they both storage still 2.99999999999998 — blfuentes
– blfuentes, Commented Feb 7, 2014 at 10:10
"I know the precision for double is with 15decimals" - Uhm, well... It's a bit more complicated than that.. — BoBTFish
– BoBTFish, Commented Feb 7, 2014 at 10:11

molbdnilo · Accepted Answer · 2014-02-07 10:03:33Z

4

This is the nature of floating point - there is no way to represent 2.3 exactly as a floating point binary.

The "2.3" in your first test can't be represented exactly in binary, so it's stored as the closest possible double, which is 2.2999999999999998.
In other words, both ifs perform the same comparison.

answered Feb 7, 2014 at 10:03

molbdnilo

67.1k3 gold badges46 silver badges90 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

blfuentes Over a year ago

I know about the binary problem, so I was just thinking on how about some "string" storage conversion, or a bigger precision type, but long double throws me the error I posted.

molbdnilo Over a year ago

@blacai You'd need an infinite amount of storage in order to be able to represent all numbers exactly. You're always going to run into a limit somewhere. You can decide on what precision you're after, then implement that yourself, but it's probably not worth it.

mlatu · Accepted Answer · 2014-02-07 10:11:05Z

maybe reading this might help you: http://www.cplusplus.com/forum/beginner/34088/

As far as i can tell there seems to be no real reference except with the compiler you use. Try http://msdn.microsoft.com I suppose 2.2999999999999998 is beeing rounded to the closest double value which would be 2.3, or the other way around as the other answerer suggests. if you want long double literals, you need to add a L at the end of your literal:

//checks for equalty to a double if (d1 == 2.2999999999999998) cout << d1 << " == 2.2999999999999998 ?\n"; //checks for equalty to a long double if (d1 == 2.2999999999999998L) cout << d1 << " == 2.2999999999999998L ?\n";

btw: why dont you use

complex<long double>

?

Cplusplus forums and tutorials, please dont. Cplusplus is a source of errors, outdated practices, and missunderstandings
As I answered before, I also try with complex<long double> but it seems the same to me with '2.999999999999998'

Collectives™ on Stack Overflow

C++ Precision behavior with double/long double

2 Answers 2

2 Comments

3 Comments

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

2 Comments

3 Comments

Linked

Related