Description

In the deep universe, there is a beautiful planet named as CS on which scientists have decided to build Immense Colossal Particle Collider (i.e. ICPC) to find the ultimate theory of the universe. The ICPC is made up with several fragments, and each fragment
has a series of energy level. Any continuous sub-series of energy level corresponds to one type of microscopic particle and can accelerate it with a remarkable effect. Scientists have found that the observation of the certain type of particle is remarkable
enough if its corresponding energy level sub-series appears in more than one half fragments. Another thing, the reverse of one specific sub-series of energy level corresponds to the antiparticle of the particle corresponded by its original sub-series. As we
all know, when a particle meets its antiparticle, DUANG DUANG, a very remarkable phenomenon can be observed by scientists. For simplicity, scientists have declared that it is not remarkable enough until the total count of the appearance in the different fragments
of the original sub-series and its reverse is more than one half the number of fragments. Lastly, both in the first and the second condition, the longer the sub-series is, the more remarkable observation can be get.

Well, so long a paragraph, science is really complicated. Now, questions come: given a set of fragments with a series of energy level, find the sub-series which can get the most remarkable observation.

Input

There are several cases. Every case comes a line with a positive integer N (N <= 10) first of all, followed by N lines each of which contains a nonempty series of capital letters representing energy levels. All series have a length not more than 1000.

Output

For every case, output the wanted sub-series. If there are more than one, output them in the alphabetical order, each in one line. If there is none, output NONE. Note that whenever one sub-series and its reverse appear simultaneously with the satisfied condition,
it is available to output only the less one in alphabetical order of them two even if any of them two appears more than one half N times.

Sample Input

3
ABC
ABD
BCD
3
AAA
BBB
CCC
2
ABC
DBA

Sample Output

AB
BC
NONE
AB

HINT

Source



题意:
要求全部正向或者反向出如今超过k/2个串中的子串

思路:
还是和曾经一样二分答案。使用二进制来标记状态

#include <iostream>
#include <stdio.h>
#include <string.h>
#include <stack>
#include <queue>
#include <map>
#include <set>
#include <vector>
#include <math.h>
#include <bitset>
#include <algorithm>
#include <climits>
using namespace std; #define LS 2*i
#define RS 2*i+1
#define UP(i,x,y) for(i=x;i<=y;i++)
#define DOWN(i,x,y) for(i=x;i>=y;i--)
#define MEM(a,x) memset(a,x,sizeof(a))
#define W(a) while(a)
#define gcd(a,b) __gcd(a,b)
#define LL long long
#define N 1000005
#define MOD 1000000007
#define INF 0x3f3f3f3f
#define EXP 1e-8
int wa[N],wb[N],wsf[N],wv[N],sa[N];
int rank1[N],height[N],s[N],a[N];
//sa:字典序中排第i位的起始位置在str中第sa[i]
//rank:就是str第i个位置的后缀是在字典序排第几
//height:字典序排i和i-1的后缀的最长公共前缀
int cmp(int *r,int a,int b,int k)
{
return r[a]==r[b]&&r[a+k]==r[b+k];
}
void getsa(int *r,int *sa,int n,int m)//n要包括末尾加入的0
{
int i,j,p,*x=wa,*y=wb,*t;
for(i=0; i<m; i++) wsf[i]=0;
for(i=0; i<n; i++) wsf[x[i]=r[i]]++;
for(i=1; i<m; i++) wsf[i]+=wsf[i-1];
for(i=n-1; i>=0; i--) sa[--wsf[x[i]]]=i;
p=1;
j=1;
for(; p<n; j*=2,m=p)
{
for(p=0,i=n-j; i<n; i++) y[p++]=i;
for(i=0; i<n; i++) if(sa[i]>=j) y[p++]=sa[i]-j;
for(i=0; i<n; i++) wv[i]=x[y[i]];
for(i=0; i<m; i++) wsf[i]=0;
for(i=0; i<n; i++) wsf[wv[i]]++;
for(i=1; i<m; i++) wsf[i]+=wsf[i-1];
for(i=n-1; i>=0; i--) sa[--wsf[wv[i]]]=y[i];
t=x;
x=y;
y=t;
x[sa[0]]=0;
for(p=1,i=1; i<n; i++)
x[sa[i]]=cmp(y,sa[i-1],sa[i],j)? p-1:p++;
}
}
void getheight(int *r,int n)//n不保存最后的0
{
int i,j,k=0;
for(i=1; i<=n; i++) rank1[sa[i]]=i;
for(i=0; i<n; i++)
{
if(k)
k--;
else
k=0;
j=sa[rank1[i]-1];
while(r[i+k]==r[j+k])
k++;
height[rank1[i]]=k;
}
} char str[N];
int id[N];
map<string,int> mat,ans;
map<string,int>::iterator it; int check(int x)//统计该状态包括几个串
{
int i,cnt = 0;
for(i = 1; i<=10; i++)
if((1<<i)&x)
cnt++;
return cnt;
} int main()
{
int n,i,j,k,len;
while(~scanf("%d",&k))
{
MEM(id,0);
n = 0;
int p = 200;
for(i = 1; i<=k; i++)
{
scanf("%s",str);
len = strlen(str);
for(j = 0; j<len; j++)
{
id[n] = i;
s[n++] = str[j];
}
s[n++] = p++;
for(j = len-1; j>=0; j--)
s[n++] = str[j];
s[n++] = p++;
}
if(k == 1)
{
printf("%s\n",str);
continue;
}
getsa(s,sa,n,p);
getheight(s,n);
int l = 1,r = 1000;
ans.clear();
while(l<=r)
{
int mid = (l+r)/2;
i = 0;
mat.clear();
while(i<n)
{
if(height[i]>=mid)
{
int tem = 1<<id[sa[i-1]];
len = 2000;
while(height[i]>=mid && i<n)//二进制记录串
{
tem |= (1<<id[sa[i]]);
len = min(len,height[i]);
i++;
}
if(tem!=1)
{
char s1[1005],s2[1005];
for(j = len-1; j>=0; j--)
{
s1[len-1-j] = s[sa[i-1]+j];
s2[j] = s[sa[i-1]+j];
}
s1[len] = s2[len] = '\0';
if(mat.find(string(s1)) != mat.end())
mat[string(s1)] |= tem;
else
mat[string(s2)] = tem;
}
}
i++;
}
int flag = 0;
for(it = mat.begin(); it!=mat.end(); it++)
{
if(check(it->second) >= k/2+1)
{
if(flag==0)
{
ans.clear();
flag = 1;
}
ans.insert(*it);
}
}
if(flag==0) r = mid-1;
else l = mid+1;
}
if(ans.size()==0)
printf("NONE\n");
else
{
for(it = ans.begin(); it!=ans.end(); it++)
{
printf("%s\n",it->first.c_str());
}
}
} return 0;
}

CSU1608: Particle Collider(后缀数组)的更多相关文章

  1. 后缀数组的倍增算法(Prefix Doubling)

    后缀数组的倍增算法(Prefix Doubling) 文本内容除特殊注明外,均在知识共享署名-非商业性使用-相同方式共享 3.0协议下提供,附加条款亦可能应用. 最近在自学习BWT算法(Burrows ...

  2. BZOJ 4199: [Noi2015]品酒大会 [后缀数组 带权并查集]

    4199: [Noi2015]品酒大会 UOJ:http://uoj.ac/problem/131 一年一度的“幻影阁夏日品酒大会”隆重开幕了.大会包含品尝和趣味挑战两个环节,分别向优胜者颁发“首席品 ...

  3. BZOJ 1692: [Usaco2007 Dec]队列变换 [后缀数组 贪心]

    1692: [Usaco2007 Dec]队列变换 Time Limit: 5 Sec  Memory Limit: 64 MBSubmit: 1383  Solved: 582[Submit][St ...

  4. POJ3693 Maximum repetition substring [后缀数组 ST表]

    Maximum repetition substring Time Limit: 1000MS   Memory Limit: 65536K Total Submissions: 9458   Acc ...

  5. POJ1743 Musical Theme [后缀数组]

    Musical Theme Time Limit: 1000MS   Memory Limit: 30000K Total Submissions: 27539   Accepted: 9290 De ...

  6. 后缀数组(suffix array)详解

    写在前面 在字符串处理当中,后缀树和后缀数组都是非常有力的工具. 其中后缀树大家了解得比较多,关于后缀数组则很少见于国内的资料. 其实后缀数组是后缀树的一个非常精巧的替代品,它比后缀树容易编程实现, ...

  7. 【UOJ #35】后缀排序 后缀数组模板

    http://uoj.ac/problem/35 以前做后缀数组的题直接粘模板...现在重新写一下模板 注意用来基数排序的数组一定要开到N. #include<cstdio> #inclu ...

  8. 【BZOJ-2119】股市的预测 后缀数组

    2119: 股市的预测 Time Limit: 10 Sec  Memory Limit: 259 MBSubmit: 334  Solved: 154[Submit][Status][Discuss ...

  9. 【BZOJ-4698】Sandy的卡片 后缀数组

    4698: Sdoi2008 Sandy的卡片 Time Limit: 10 Sec  Memory Limit: 128 MBSubmit: 140  Solved: 55[Submit][Stat ...

随机推荐

  1. c++类模板初探

    #include <iostream> #include <string> using namespace std; // 你提交的代码将嵌入到这里 ; template &l ...

  2. TP5防sql注入、防xss攻击

    框架默认没有设置任何过滤规则 可以配置文件中设置全局的过滤规则 config.php 配置选项 default_filter 添加以下代码即可 // 默认全局过滤方法 用逗号分隔多个 'default ...

  3. mariadb数据库的链接查询和表格设计

    链接查询 练习准备: --创建学生表 create table students ( id int unsigned not null auto_increment primary key, name ...

  4. Edward Frenkel关于几何化朗兰兹纲领的采访

    本文来自:菲尔兹奖座谈会 博客 Edward Frenkel教授的主要研究方向是数学与量子物理中的对称.他现在在做的许多问题都与朗兰兹纲领有关.他现在是加州大学伯克利分校的数学教授. 在今年的菲尔兹奖 ...

  5. C#串口调试工具 (WPF/MVVM结构完整示例版)

    前文 由于经常用到串口调试, 尽管有现成的软件, 因为前端时间涉及一个二次开发, 就因为一个RtsEnable设置, 折腾半天,  网上各种版本的也很多, 功能扩展的很开也多.所以现在自己做了一个够用 ...

  6. hadoop学习;datajoin;chain签名;combine()

    hadoop有种简化机制来管理job和control的非线性作业之间的依赖.job对象时mapreduce的表现形式.job对象的实例化可通过传递一个jobconf对象到作业的构造函数中来实现. x. ...

  7. xcode 及 MAC 经常使用快捷键

    郝萌主倾心贡献,尊重作者的劳动成果,请勿转载. 假设文章对您有所帮助,欢迎给作者捐赠,支持郝萌主,捐赠数额任意.重在心意^_^ 我要捐赠: 点击捐赠 Cocos2d-X源代码下载:点我传送 黑苹果键盘 ...

  8. HDU 4965 Fast Matrix Calculation(矩阵高速幂)

    HDU 4965 Fast Matrix Calculation 题目链接 矩阵相乘为AxBxAxB...乘nn次.能够变成Ax(BxAxBxA...)xB,中间乘n n - 1次,这样中间的矩阵一个 ...

  9. 删除heroku上的数据库记录

    部署本地项目到heroku上.在线上插入数据到数据库,本地代码再次更新到heroku,线上的数据记录还存在单是图片丢失.问题还没有解决: 本地代码和heroku代码怎样同步? heroku使用的pg和 ...

  10. nyoj--1011--So Easy[II](数学几何水题)

    So Easy[II] 时间限制:1000 ms  |  内存限制:65535 KB 难度:2 描述 这是一道基础的计算几何问题(其实这不提示大家也都看的出).问题描述如下: 给你一个N边形.且N边形 ...