Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

Java: Check if two double values match on specific no of decimal places

Tags:

java

double

Comparing those two values shall result in a "true":

53.9173333333333  53.9173
like image 404
tzippy Avatar asked May 31 '10 15:05

tzippy


4 Answers

here is the simple example if you still need this :)

public static boolean areEqualByThreeDecimalPlaces(double a, double b) {

    a = a * 1000;

    b = b * 1000;

    int a1 = (int) a;

    int b1 = (int) b;

    if (a1 == b1) {
        System.out.println("it works");
        return true;
    }

    else
        System.out.println("it doesn't work");
    return false;
like image 88
Norby Avatar answered Nov 15 '22 12:11

Norby


If you want a = 1.00001 and b = 0.99999 be identified as equal:

return Math.abs(a - b) < 1e-4;

Otherwise, if you want a = 1.00010 and b = 1.00019 be identified as equal, and both a and b are positive and not huge:

return Math.floor(a * 10000) == Math.floor(b * 10000);
// compare by == is fine here because both sides are integral values.
// double can represent integral values below 2**53 exactly.

Otherwise, use the truncate method as shown in Are there any functions for truncating a double in java?:

BigDecimal aa = new BigDecimal(a);
BigDecimal bb = new BigDecimal(b);
aa = aa.setScale(4, BigDecimal.ROUND_DOWN);
bb = bb.setScale(4, BigDecimal.ROUND_DOWN);
return aa.equals(bb);
like image 34
kennytm Avatar answered Nov 15 '22 12:11

kennytm


Naively:

if(Math.abs(a-b) < 0.0001)

However, that's not going to work correctly for all values. It's actually impossible to get it to work as long as you're using double, because double is implemented as binary fractons and does not even have decimal places.

You'll have to convert your values to String or BigDecimal to make meaningful tests about their decimal places.

You may want to read the Floating-Point Guide to improve your understanding of how floating point values work.

like image 6
Michael Borgwardt Avatar answered Nov 15 '22 12:11

Michael Borgwardt


Apache commons has this: org.apache.commons.math3.util.Precision equals(double x, double y, double eps)

epsilon would be the distance you would allow. Looks like yours would be 1e-5?

The source-code of this method looks like it uses Math.abs as suggested in other answers.

like image 3
DaveEdelstein Avatar answered Nov 15 '22 12:11

DaveEdelstein