Using BigDecimal instead of float

suggest change

Due to way that the float type is represented in computer memory, results of operations using this type can be inaccurate - some values are stored as approximations. Good examples of this are monetary calculations. If high precision is necessary, other types should be used. e.g. Java 7 provides BigDecimal.

import java.math.BigDecimal;

public class FloatTest {

public static void main(String[] args) {
    float accountBalance = 10000.00f;
    System.out.println("Operations using float:");
    System.out.println("1000 operations for 1.99");
    for(int i = 0; i<1000; i++){
        accountBalance -= 1.99f;
    }
    System.out.println(String.format("Account balance after float operations: %f", accountBalance));
    
    BigDecimal accountBalanceTwo = new BigDecimal("10000.00");
    System.out.println("Operations using BigDecimal:");
    System.out.println("1000 operations for 1.99");
    BigDecimal operation = new BigDecimal("1.99");
    for(int i = 0; i<1000; i++){
        accountBalanceTwo = accountBalanceTwo.subtract(operation);
    }
    System.out.println(String.format("Account balance after BigDecimal operations: %f", accountBalanceTwo));
}

Output of this program is:

Operations using float:
1000 operations for 1.99
Account balance after float operations: 8009,765625
Operations using BigDecimal:
1000 operations for 1.99
Account balance after BigDecimal operations: 8010,000000

For a starting balance of 10000.00, after 1000 operations for 1.99, we expect the balance to be 8010.00. Using the float type gives us an answer around 8009.77, which is unacceptably imprecise in the case of monetary calculations. Using BigDecimal gives us the proper result.

Found a mistake? Have a question or improvement idea? Let me know.

Table Of Contents