Java的NaN

看JDK源码时看到Math下边有个函数：

public static float max(float a, float b) {
    if (a != a)
        return a;   // a is NaN
    if ((a == 0.0f) &&
        (b == 0.0f) &&
        (Float.floatToRawIntBits(a) == negativeZeroFloatBits)) {
        // Raw conversion ok since NaN can't map to -0.0.
        return b;
    }
    return (a >= b) ? a : b;
}

注意  if (a != a) return a; // a is NaN

a竟然可能不等于a，这是什么逻辑？查了一下，NaN表示非数值，它不与任何数值相等，包括它自己。注意这一点跟

POSITIVE_INFINITY、NEGATIVE_INFINITY不一样，写个demo验证一下：

public static void main(String[] args) {
    System.out.println(Float.POSITIVE_INFINITY * 0);
    System.out.println(Float.NEGATIVE_INFINITY * 0);
    System.out.println((Float.POSITIVE_INFINITY * 0) == Float.POSITIVE_INFINITY);
    System.out.println((Float.NEGATIVE_INFINITY * 0) == Float.NEGATIVE_INFINITY);
    System.out.println(Float.POSITIVE_INFINITY / 0);
    System.out.println(Float.NEGATIVE_INFINITY / 0);
    System.out.println((Float.POSITIVE_INFINITY / 0) == Float.POSITIVE_INFINITY);
    System.out.println((Float.NEGATIVE_INFINITY / 0) == Float.NEGATIVE_INFINITY);
    float a = Float.POSITIVE_INFINITY;
    System.out.println(a == a);
    float b = Float.NaN;
    System.out.println(b == b);
    System.out.println(Double.isNaN(b));
}

输出：

NaN
NaN
false
false
Infinity
-Infinity
true
true
true
false
true