Java .ulp()

The Math.ulp() method returns the unit of least precision of a given number. It calculates the distance between the given float or double value and the next largest float or double value in magnitude.

Syntax

Math.ulp(num);

The num parameter can be of type float or double and the return value will be of the same type.

Special cases include the following:

• If num is NaN, NaN will be returned.
• If num is positive or negative zero, then the return value will be Double.MIN_VALUE or Float.MIN_VALUE which is the lowest number a float or double can represent.
• If num is positive or negative infinity, then the return value will be positive infinity.
• If num is a number, then num will have the same return value as -num.
• If num is Double.MAX_VALUE or Float.MAX_VALUE, then the return value will be 2⁹⁷¹ for double and 2¹⁰⁴ for float.

Example 1

// Test.java
import java.lang.Math;

public class Test {
  public static void main(String args[]) {
    double num = 23.44;
    System.out.println(Math.ulp(num));
  }
}

Copy to clipboard
Copy to clipboard

This will output:

3.552713678800501E-15

Copy to clipboard
Copy to clipboard

Example 2

import java.lang.Math;

public class Test {
  public static void main(String args[]) {
    double num = 23.44;
    System.out.println(Math.ulp(num) + " = "  + Math.ulp(-num));
    System.out.println(Math.ulp(0/0.));
    System.out.println(Math.ulp(0.));
    System.out.println(Math.ulp(1./0));
  }
}

Copy to clipboard
Copy to clipboard

This will output:

3.552713678800501E-15 = 3.552713678800501E-15
NaN
4.9E-324
Infinity

Copy to clipboard
Copy to clipboard