计算幂的一种高效方法

计算x^n次方时，我们首先想到的是递归，时间复杂度为O(n)

 

public double pow(double x, int n)
{
	if(n==0)
		return 1.0;
	if(n<0)
		return 1.0/pow(x,-n);
	return x*pow(x,n-1);
}

 

 

从改善时间复杂度的角度考虑，可以利用pow(x,n) = pow(x,n/2)*pow(x,n-n/2)，时间复杂度变为O(logn)

 

double pow(double x, int n)
{
	if(n==0)
		return 1.0;
	if(n<0)
		return 1.0/pow(x,-n);
	double half = pow(x,n>>1);
	if(n%2==0)
		return half*half;
	else
		return half*half*x;
}

但是上述代码都存在一个bug：当n为int类型的最小值Integer.MIN_VALUE(-2^32)时， -n并不是Integer.MAX_VALUE(2^32-1),减1才是。

 

Consider the binary representation of n. For example, if it is "10001011", thenx^n = x^(1+2+8+128) = x^1 * x^2 * x^8 * x^128. Thus, we don't want to loop n times to calculate x^n. To speed up, we loop through each bit, if the i-th bit is 1, then we add x^(1 << i) to the result. Since (1 << i) is a power of 2, x^(1<<(i+1)) = square(x^(1<<i)). The loop executes for a maximum of log(n) times.

 

 public double pow(double x, int n) {
        if(n == 0)
            return 1.0;
        if(n < 0)
        {
            if(n == Integer.MIN_VALUE)
                return 1.0/((pow(x,Integer.MAX_VALUE)*x));
            else
                return 1.0/pow(x, -n);
        }
        double result = 1.0;
        for(;n>0;x*=x,n>>=1)
        {
            if((n&1) > 0)
                result *= x;
        }
        return result;