剑指offer--面试题31：连续子数组的最大和



题目描述

HZ偶尔会拿些专业问题来忽悠那些非计算机专业的同学。今天测试组开完会后,他又发话了:在古老的一维模式识别中,常常需要计算连续子向量的最大和,当向量全为正数的时候,问题很好解决。但是,如果向量中包含负数,是否应该包含某个负数,并期望旁边的正数会弥补它呢？例如:{6,-3,-2,7,-15,1,2,2},连续子向量的最大和为8(从第0个开始,到第3个为止)。你会不会被他忽悠住？

python实现：

# -*- coding:utf-8 -*-
class Solution:
     
    #法1：动态规划
    def FindGreatestSumOfSubArray1(self, array):
        # write code here
        #dp[i]表示以array[i]为结尾的子数组(0, i)的最大和
        #如果dp[i-1]>0, dp[i] = dp[i-1]+array[i]
        #如果dp[i-1]<=0, dp[i] = array[i]
        n = len(array)
        if n==0:
            return 0
        dp = [0]*n
        #初始化
        dp[0] = array[0]
        for i in range(1, n):
            if dp[i-1]<=0:
                dp[i] = array[i]
            else:
                dp[i] = dp[i-1] + array[i]
        return max(dp)
     
    #法2：分治
    #参考：http://blog.csdn.net/a272846945/article/details/50822601
    def FindGreatestSumOfSubArray(self, array):
        n = len(array)
        if n==0:
            return 0
        return self.devideAndConquer(array, 0, n-1)
     
    def devideAndConquer(self, nums, low, high):
        if low==high:
            return nums[low]
        mid = (low+high)/2
        leftMaxSum = self.devideAndConquer(nums, low, mid)
        rightMaxSum = self.devideAndConquer(nums, mid+1, high)
        crossMaxSum = self.merge(nums, low, mid, high)
        return max(leftMaxSum, rightMaxSum, crossMaxSum)
     
    def merge(self, nums, low, mid, high):
        continousLeftSum = nums[mid]
        tmpLeftSum = nums[mid]
        for i in range(mid-1, low-1, -1):
            tmpLeftSum += nums[i]
            continousLeftSum = max(continousLeftSum, tmpLeftSum)
        continousRightSum = nums[mid+1]
        tmpRightSum = nums[mid+1]
        for i in range(mid+2, high+1):
            tmpRightSum += nums[i]
            continousRightSum = max(continousRightSum, tmpRightSum)
        return continousLeftSum + continousRightSum