(DP动态规划) Leetcode 312. Burst Balloons
第一次做动态规划的题目
题目
Given n balloons, indexed from 0
to n-1
. Each balloon is painted with a number on it represented by array nums. You are asked to burst all the balloons. If the you burst balloon i you will get nums[left] * nums[i] * nums[right]
coins. Here left and right are adjacent indices of i
. After the burst, the left and right then becomes adjacent.
Find the maximum coins you can collect by bursting the balloons wisely.
Note:
- You may imagine
nums[-1] = nums[n] = 1
. They are not real therefore you can not burst them. -
0 ≤ n ≤ 500
,0 ≤ nums[i] ≤ 100
Example:
Input: [3,1,5,8]
Output: 167
Explanation: nums = [3,1,5,8] --> [3,5,8] --> [3,8] --> [8] --> []
coins = 3*1*5 + 3*5*8 + 1*3*8 + 1*8*1 = 167
题目解析与分析
题意
大概就是:有一排气球,每个标了数字,然后戳一个气球得到的奖励是左气球×该气球×右气球
,当然如果左或右没有气球就乘1。问该以什么顺序戳气球得到的奖励最大
最笨的思路
枚举法:将所有的情况列举一遍。当有n
个气球的时候,第一步我们有n
种选择,第二步我们又有n-1
个选择......显然全枚举的算法复杂度为,效率不敢恭维,因此这里就不实现——因为不可能过。
参阅
进一步想法
我们需要去考虑上面枚举法做了什么重复的计算。我们可以想到,给定一组气球,它所能获得的最大的奖励应该和前面已经被戳的气球无关——被戳过的气球只是在求和的时候累积上了而已。
对于给定k
k
(从1开始)的所有情况都记录在内存上,k+1
就可以基于k
进行计算,那么我们总共需要进行的计算就是
这种算法优于, 但仍然坏于; 我们需要更优的算法
分治的想法
我们考虑用分治去思考这一道题。先是正常地考虑分治,我戳爆某个气球,可不可以把剩下的气球分成两堆呢?这是否可行的前提是两堆会不会互相干扰:答案是肯定的,在戳爆某个气球以后,左堆的最右气球会需要右堆的最左气球来进行计算。
这时候我们需要反向的思维:我们正向地想戳气球的过程,当然会导致两堆相互影响。那如果我们考虑的是在这一堆里最后戳爆的那个气球呢?假如A,B,C,D,E中我最后戳C, 那左堆(A,B)在戳的过程中显然会以C为右边界,而右堆(D,E)以C为左边界——这就实现了分治,两个子问题是相互不干扰的。
具体的算法如下:
func maxCoins(nums []int) int {
record := []int{1}
record = append(record, nums...)
record = append(record, 1)
fmt.Println(record)
return DC(0, len(record)-1, record)
}
// DC: Calculate the max coins got from (leftB, rightB), boundary exclueded
func DC(leftB, rightB int, record []int) int{
// Go through every possibility of last balloon burst within the boundaries
if leftB + 1 == rightB {
return 0
}
max := 0
for i := leftB + 1; i < rightB; i++ {
temp := record[leftB] * record[i] * record[rightB] + DC(leftB, i, record) + DC(i, rightB, record)
if temp > max {
max = temp
}
}
return max
}
很可惜这个算法超时了
动态规划-更进一步的优化
但是这并不是结束。我们很容易可以发现这里面有大量重复的运算。这时我们不难想到借用动态规划的思想来解决这个问题——将重复的结果保存到内存当中
func maxCoins(nums []int) int {
record := []int{1}
record = append(record, nums...)
record = append(record, 1)
fmt.Println(record)
mem := make([][]int, len(record))
for i := range(mem){
mem[i] = make([]int, len(record))
}
return DC(0, len(record)-1, record, mem)
}
// DC: Calculate the max coins got from (leftB, rightB), boundary exclueded
func DC(leftB, rightB int, record []int, mem [][]int) int{
// Go through every possibility of last balloon burst within the boundaries
if leftB + 1 == rightB {
return 0
}
if mem[leftB][rightB] != 0 {
return mem[leftB][rightB]
}
max := 0
for i := leftB + 1; i < rightB; i++ {
temp := record[leftB] * record[i] * record[rightB] + DC(leftB, i, record, mem) + DC(i, rightB, record, mem)
if temp > max {
max = temp
mem[leftB][rightB] = temp
}
}
return max
}
很遗憾,这份代码也只打败了20%多的代码。
纯粹的动态规划
上面的局部动态规划这么低的KO率怎能够让我们满足呢?通过上面的代码编写,我们很显然发现可以将递归的过程变成循环。
简单来说,就是先将left + 1 == right
的值先全部赋为0(由于Go语言的机制默认初始化为0就不用额外操作了);而后left + 2 == right
的值可以根据left + 1 == right
的值计算出来;再然后逐渐递增直到算出(0, length+2)
即为结果(由于按照题目要求两端都加了1作为墙壁,所以是len+2)。
代码如下:
func maxCoins(nums []int) int {
record := []int{1}
record = append(record, nums...)
record = append(record, 1)
fmt.Println(record)
mem := make([][]int, len(record))
for i := range(mem){
mem[i] = make([]int, len(record))
}
// mem 全部初始化为 0
for stretch := 2; stretch < len(record); stretch++ {
for left := 0; left + stretch < len(record); left++ {
max := 0
right := left + stretch
for i := left + 1; i < left + stretch; i++ {
temp := record[left] * record[i] * record[right] + mem[left][i] + mem[i][right]
if temp > max {
max = temp
mem[left][right] = temp
}
}
}
}
return mem[0][len(record)-1]
}
Runtime: 8 ms, faster than 57.14% of Go online submissions for Burst Balloons.
这样的结果就差强人意了
还能再优化吗
还能再优化。对代码结构可以做进一步的优化
func maxCoins4(nums []int) int {
record := append([]int{1}, nums...)
record = append(record, 1)
fmt.Println(record)
mem := make([][]int, len(record))
for i := range(mem){
mem[i] = make([]int, len(record))
}
// mem 全部初始化为 0
lens := len(record)
for stretch := 2; stretch < lens; stretch++ {
for left := 0; left + stretch < lens; left++ {
max := 0
right := left + stretch
for i := left + 1; i < right; i++ {
temp := record[left] * record[i] * record[right] + mem[left][i] + mem[i][right]
if temp > max {
max = temp
mem[left][right] = temp
}
}
}
}
return mem[0][lens-1]
}
Runtime: 4 ms, faster than 100.00% of Go online submissions for Burst Balloons.
令人满足。