数组的最大子数和问题

数组的最大子数和问题就是

有一个数组a，求a中和最大的连续子序列。

下面是看了《编程之美》的五种解法，计算复杂度和空间复杂度递减

1 O(N^3)

def maxsubsum(a):
    l = len(a)
    max = min(a)
    for i in range(l):
        for j in range(i, l):
            sum = 0
            for k in range(i, j+1):
                sum += a[k]
            if sum > max:
                max = sum
    return max

这为最普遍的遍历求解

2 O(N^2)

def maxsubsum1(a):
    l= len(a)
    max = min(a)
    for i in range(l):
        sum = 0
        for j in range(i, l):
            sum += a[j]
            if sum > max:
                max = sum
    return max

 由于sum[k] = sum[k-1]+a[i]，所以可以将1中的k循环省去

3 O(N * logN)

def maxsubsum2(a):
    l = len(a)
    start = [None] * l
    all = [None] * l
    start[l - 1] = a[l - 1]
    all[l - 1] = a[ l - 1]
    l = l - 2
    while l >= 0:
        start[l] = max(a[l] + start[l + 1], a[l])
        all[l] = max(start[l], all[l+1])
        l -= 1
    return all[0]

 次方法使用动态规划方法，start[i]为以a[i]为开头的最大字数和，all[i]为从i开始的最大子数和

动态规划公式为  all[l] = max(a[l], a[l] + start[l+1], all[l+1])

4 O(N * logN) 空间复杂度为O(1)

def maxsubsum3(a):
    l = len(a)
    start = a[l - 1]
    all = a[l - 1]
    l = l - 2
    while l >= 0:
        start = max(start + a[l], a[l])
        all = max(all, start)
        l -= 1
    return all

 对3进行修改，去掉了数组start和all的使用

5 O(N * logN)空间复杂度O(1)

当a中全部为负数时，返回0

def maxsubsum4(a):
    l = len(a)
    start = a[l - 1]
    all = a[l - 1]
    l = l - 2
    while l >= 0:
        if start < 0:
            start = 0
        start += a[l]
        if start > all:
            all = start
        l -= 1
    return all