Java run-off error(on float and double)

We can't store real 3.9 in Base 2.

double a = 3.9;
boolean b = false;

if(a == 3.9){
   b = true;
}

System.out.println(b);

输出为：

false

(3.9)Base 10 = (1     1 .    1       1       1    00   1      1......)Base 2

      2^1 2^0 2^-1 2^-2 2^-3 ... 2^-6 2^-7 ...

    1/2    1/4   1/8      1/64 1/128

≈ 3 +0.8984375...

∴ We can not store real 3.9 in Base 2.

Example 1:

public class FloatingPointError {

	public static void main(String[] args) {
		double x = 3.9;
		double y = 3.8; 
		double difference = x - y;
		System.out.println(difference);
		if (difference == 0.1) {
			System.out.println("I can do simple math.");
		} else {
			System.out.println("I am having trouble with easy problems.");
		}
	}
}

输出为：

0.10000000000000009
I am having trouble with easy problems.

我们可以通过“epsilon”（极小的正数）来检测是否x = 3.9。

Example 2:

public class FloatingPointDoneRight {

	public static void main(String[] args) {
		double EPSILON = 0.00000000001;
		double x = 3.9;
		double y = 3.8;
		double difference = x - y;
		if (Math.abs(difference - 0.1) < EPSILON) {
			System.out.println("I can do simple math.");
		} else {
			System.out.println("I am having trouble with easy problems.");
		}
	}

}

输出为：

I can do simple math.