Vladik was bored on his way home and decided to play the following game. He tookn cards and put them in a row in front of himself. Every card has a positive integer number not exceeding8 written on it. He decided to find the longest subsequence of cards which satisfies the following conditions:
- the number of occurrences of each number from 1 to8 in the subsequence doesn't differ by more then 1 from the number of occurrences of any other number. Formally, if there areck cards with numberk on them in the subsequence, than for all pairs of integers the condition|ci - cj| ≤ 1 must hold.
- if there is at least one card with number x on it in the subsequence, then all cards with numberx in this subsequence must form a continuous segment in it (but not necessarily a continuous segment in the original sequence). For example, the subsequence[1, 1, 2, 2] satisfies this condition while the subsequence[1, 2, 2, 1] doesn't. Note that [1, 1, 2, 2] doesn't satisfy the first condition.
Please help Vladik to find the length of the longest subsequence that satisfies both conditions.
The first line contains single integer n (1 ≤ n ≤ 1000) — the number of cards in Vladik's sequence.
The second line contains the sequence of n positive integers not exceeding8 — the description of Vladik's sequence.
Print single integer — the length of the longest subsequence of Vladik's sequence that satisfies both conditions.
3 1 1 1
1
8 8 7 6 5 4 3 2 1
8
24 1 8 1 2 8 2 3 8 3 4 8 4 5 8 5 6 8 6 7 8 7 8 8 8
17
In the first sample all the numbers written on the cards are equal, so you can't take more than one card, otherwise you'll violate the first condition.
题目大意:
给你一个长度为N的一个序列,需要我们找到一个最长的序列,保证每个元素(1~8)的个数之前的绝对值差小于等于1,而且对应选出来的序列中,需要保证相同的元素是连续的。
思路(补题真好玩系列0.0):思路来源于(http://www.cnblogs.com/Saurus/p/6183721.html);
1、首先我们很容易想到状压,我们肯定是状压8个数字的状态,其中1表示这个位子上的数已经选完了,0表示没有选完,那么我们在想dp数组的维度的时候,以及初算时间复杂度的时候,发现这个出现的次数一直是一个很难处理的点,那么我们首先对这个出现的次数进行优化:
①我们可以通过枚举来dp,每次枚举一个出现的次数,进行最大值的维护 ,我们不难想到:对应这个出现的次数越多,结果就可能越长,那么其具有单调性。
②对于可以枚举的具有单调性存在的变量,我们可以通过二分查找来进行优化。
2、那么我们每一次二分枚举出的当前值mid,表示本次二分枚举出来每个数将要出现的次数。
①对于当前值mid,每种数字出现的次数要么是mid次,要么是mid-1次。
②考虑dp,我们设定dp【i】【j】表示dp过程进行到第i位,对应j状态下的最多出现mid次的数字的个数。
③那么考虑状态转移方程:
如果我们直接暴力的话肯定时间复杂度是吃不消的,那么我们考虑调用一个vector<int >mp;mp【i】用于存储数字i出现的各个位子,我们按照升序排列存储。
那么接下来:
④维护好每次二分情况的最大值即可。
Ac代码:
#include<stdio.h>
#include<string.h>
#include<algorithm>
#include<vector>
using namespace std;
int a[100050];
int ans,n;
vector<int >mp[1050];
int cur[1050];
int dp[1050][1<<8];
int Slove(int mid)
{
int end=(1<<8);
for(int i=1;i<=8;i++)cur[i]=0;
memset(dp,-1,sizeof(dp));
dp[1][0]=0;
for(int i=1;i<=n;i++)
{
for(int j=0;j<end;j++)
{
if(dp[i][j]<0)continue;
for(int k=1;k<=8;k++)
{
if(j &(1<<(k-1)))continue;
if(mp[k].size()-cur[k]<mid-1)continue;
dp[mp[k][cur[k]+mid-2]][j|(1<<(k-1))]=max( dp[mp[k][cur[k]+mid-2]][j+(1<<(k-1))],dp[i][j]);
if(mp[k].size()-cur[k]<mid)continue;
dp[mp[k][cur[k]+mid-1]][j|(1<<(k-1))]=max( dp[mp[k][cur[k]+mid-1]][j+(1<<(k-1))],dp[i][j]+1);
}
}
cur[a[i]]++;
}
int sum=-0x3f3f3f3f;
for(int i=1;i<=n;i++)sum=max(sum,dp[i][(1<<8)-1]);
if(sum<=0)return 0;
ans=max(ans,sum*mid+(8-sum)*(mid-1));
return 1;
}
int main()
{
while(~scanf("%d",&n))
{
for(int i=1;i<=n;i++)scanf("%d",&a[i]);
for(int i=1;i<=n;i++)mp[a[i]].push_back(i);
int l=2,r=n;
ans=0;
while(r-l>=0)
{
int mid=(l+r)/2;
if(Slove(mid)==1)
{
l=mid+1;
}
else r=mid-1;
}
if(ans==0)
{
for(int i=1;i<=8;i++)if(mp[i].size()>0)ans++;
}
printf("%d\n",ans);
}
}