Java - Math.round Method

Math.round. Our numbers are 10.1 and 10.5. We want to round them to 10 and 11. With Math.round, part of java.lang.Math, we can do this with a single method call.

Math

For performance, Math.round seems fairly efficient. We benchmark it against a common cache collection (HashMap) just to ensure it is fast.

First example, double array. This program introduces an array of 4 doubles. These all must be rounded up or down. We use a for-each loop to call Math.round on each double.

Result The Math.round method changes 10.1 to 10, and 10.5 to 11. So it rounds down and up depending on the number's fractional value.

import java.lang.Math; public class Program { public static void main(String[] args) { double[] values = { 10.1, 10.5, 10.9, 10 }; // Round some double values. for (double v : values) { // This Math.round receives a double and returns along. long result = Math.round(v); System.out.println(v + ": " + result); } } }

10.1: 10 10.5: 11 10.9: 11 10.0: 10

Floats. With a float, we have a 4-byte floating-point value (half the size of a double). The Math.round method receives floats and returns ints.

import java.lang.Math; public class Program { public static void main(String[] args) { float value1 = (float) 10.1; float value2 = (float) 10.5; // Use Math.round on floats. int result1 = Math.round(value1); int result2 = Math.round(value2); // Display results. System.out.println(value1 + ": " + result1); System.out.println(value2 + ": " + result2); } }

10.1: 10 10.5: 11

Benchmark, HashMap. Here is an experiment. I wanted to see if a HashMap can retrieve a value faster than a call to Math.round. I found that Math.round is faster.

HashMap

Version 1 This version of the code repeatedly calls Math.round to get a round number from a double.

Version 2 Here we store the result of Math.round in a HashMap, and then repeatedly call get() to access the cached result.

Result It appears that Math.round is faster than a HashMap lookup. This defeats the optimization.

Tip Lookup tables often improve program performance, but only if the lookup time is less than the original computation time.

import java.util.HashMap; public class Program { public static void main(String[] args) { // Place a Double key in this HashMap. HashMap<Double, Long> roundCache = new HashMap<>(); double value = 1000.5; roundCache.put(value, Math.round(value)); long t1 = System.currentTimeMillis(); // Version 1: use round. for (int i = 0; i < 1000000; i++) { for (int y = 0; y < 200; y++) { // Use Math.round method directly. long result = Math.round(value); if (result != 1001) { return; } } } long t2 = System.currentTimeMillis(); // Version 2: use get() to look up a value. for (int i = 0; i < 1000000; i++) { for (int y = 0; y < 200; y++) { // Use get. long result = roundCache.get(value); if (result != 1001) { return; } } } long t3 = System.currentTimeMillis(); // ... Times. System.out.println(t2 - t1); System.out.println(t3 - t2); } }

7 ms: Math.round calls 358 ms: HashMap get calls

A summary. Math.round is a fast way to round up or down based on the fractional value of a number (a double or float). For double arguments, it returns long. And for floats, it returns int.