Is there an efficient way to check for an approximate floating point equality in C++?

Asked Sep 22 '11 at 02:16

Active Sep 22 '11 at 02:16

Viewed 82 times

Possible Duplicate:
Most effective way for float and double comparison

I have a function that calculates distance and then that distance is then later compared to the square root of 2 to see if two items are adjacent on a grid. In Java this works fine, apparently the Math.sqrt(Math.pow(x1-x2,2)....) produces the same as Math.sqrt(2) but in C++ cmath's sqrt(pow(x1-x2,2)...) is not equal.

I cout<<ed the doubles and to the precision shown (I will also output higher precision just to see) they looked equal but didn't evaluate as equal. On the other hand, if I test for values within +- 0.01 of sqrt(2) then this test for "approximate equality" works. Is this what I should do or is something else the matter? Is there some better way of doing this test if this is what I must do?

Thanks.

edited May 23 '17 at 12:03

Community

asked Sep 22 '11 at 02:16

Anonymous

3,226
2
34
49

There are about a billion questions about floating-point comparisons here on SO... – Oliver Charlesworth Sep 22 '11 at 02:21
Sorry, but also note that it actually prints as absolutely equal to 16 places... – Anonymous Sep 22 '11 at 02:22

Is there an efficient way to check for an approximate floating point equality in C++?

0 Answers0