441. Arranging Coins(可用二分搜索)

题目地址:https://leetcode.com/problems/arranging-coins/description/

You have a total of n coins that you want to form in a staircase shape, where every k-th row must have exactly k coins.

Given n, find the total number of full staircase rows that can be formed.

n is a non-negative integer and fits within the range of a 32-bit signed integer.

Example 1:

n = 5

The coins can form the following rows:
¤
¤ ¤
¤ ¤

Because the 3rd row is incomplete, we return 2.

Example 2:

n = 8

The coins can form the following rows:
¤
¤ ¤
¤ ¤ ¤
¤ ¤

Because the 4th row is incomplete, we return 3.

// 两种方法

方法一:

class Solution {
    public int arrangeCoins(int n) {
        long lo = 1, hi = n; // lo和hi定义为long纯属为了方便mid赋值,int范围够了
        long mid;
        if (n == 0) return 0;
        while (hi - lo > 1) {
            mid = (lo + hi) >>> 1; // 这里mid要定义long,虽然这里不会因为加法溢出而导致mid出问题,但是下面的mid*(1+mid)>>>1这个表达式默认int,可能出现乘法溢出,远远超出int范围,二进制多出的高位部分已经被截断了,会导致结果错误。
            long sum = mid * (1 + mid) >>> 1; // sum一定要long因为int取值到2147483647有一步会得到的1297036691877396480
            if (sum > n) {
                hi = mid;
            } else {
                lo = mid;
            }
        }
        return (int)lo;
    }
}


如果觉得mid=(lo+hi)>>>1不好理解,那就换成mid=lo + ((hi - lo) >>1)吧,因为lo在int范围,hi在int范围,hi-lo也在int范围,而lo+(hi-lo)>>1是小于hi的,所以也是int范围。关于为什么>>>1更好可以看我的另一篇博客:https://blog.csdn.net/qq_34115899/article/details/79859973

但是>>>1只能解决加法溢出的问题,几乎是解决不了乘法溢出的问题(除非有类似乘以2再>>>1的巧合,高位数据是被截断的,没有保存),解决办法是选用更大的数据类型来处理乘法溢出问题。

方法二:

很容易想到直接用数学方法

X*(1+X)/2≤n

X+X≤  2n

这里就要用配方法

4X+ 4X≤ 4*2*n

(2X+1)- 1 ≤ 8n

2X+1 ≤

X ≤ 

写成代码:

class Solution {
    public int arrangeCoins(int n) {
        double t = 8.0 * n + 1; // 不能写成8*n+1,这个表达式是默认int,可能超出int范围导致结果错误,所以换成8.0,表达式就成为double型
        return (int)((Math.sqrt(t) - 1) / 2);
    }
}

========================================Talk is cheap, show me the code=======================================

posted @ 2018-05-06 09:54  绿叶萌飞  阅读(125)  评论(0编辑  收藏  举报