Floating point comparison without any tolerance

Question

This is what I got to know about Floating point equals operation:

To actually compare floating-point numbers for equality, it is generally desirable to compare them within some tiny range of allowable differences; this range is often re‐garded as a tolerance or as epsilon.

Ie,

/** Compare two doubles, using default  epsilon */
public static boolean equals(double a, double b) {
    return equals(a, b, EPSILON);
}

Is this the only method or is there any other way to compare the exact value floating point numbers without any tolerance.

Why is it that you want to exactly compare two floating point numbers? — u8sand, Feb 27 '16 at 13:04
Yes, of course you can. The point of the advice you are quoting is that it is rarely a good idea to do so. See http://stackoverflow.com/questions/588004/is-floating-point-math-broken — Raedwald, Feb 27 '16 at 13:21

Clashsoft · Answer 1 · 2017-07-07T14:55:55.317

3

The == operator compares two double values for exact equality.

Example:

1.0 == 2.0 - 1.0 // true
0.1 + 0.2 == 0.3 // false
0.25 == 0.125 + 0.125 // true
1.0 / 3.0 == 2.0 / 6.0 // true

edited Jul 07 '17 at 14:55

answered Feb 27 '16 at 13:14

Clashsoft

11,553
5
40
79

score 2 · Accepted Answer · edited May 23 '17 at 12:15

Please note that there is no such thing as exactness in floating points due to the fundamental limitation of the binary code. You cannot express decimal in binary exactly, instead halves, quarters (and other ratios with denominator being a power of two) are used to express values that are 'close enough' within some acceptable tolerance. You cannot remove this tolerance and compare exact values, because otherwise you cannot express decimals in binary.

Useful link.

UPDATE: Just found on the stackoverflow this great response, so that I don't have to write it all over. Read this and maybe you will consider using BigDecimal for the task at hand.

Andriy Kryvtsun · Answer 3 · 2016-02-27T13:27:03.927

1

Use static Double#compare(double d1, double d2) method for internal double representation comparison if you really need it.

edited Feb 27 '16 at 13:27

answered Feb 27 '16 at 13:08

Andriy Kryvtsun

3,220
3
27
41

score 1 · Answer 4 · edited May 23 '17 at 11:50

1

Advice to just use ==, or to use a tolerance (epsilon) are wrong. They will give unexpected results if the float values are NaN (not a number). You should instead use floatToIntBits Why use Float.floatToIntBits() in Java float comparisons?

edited May 23 '17 at 11:50

Community

1
1

answered Feb 27 '16 at 14:06

Raedwald

46,613
43
151
237

Floating point comparison without any tolerance

4 Answers4