E. Fairy
time limit per test

1.5 seconds

memory limit per test

256 megabytes

input

standard input

output

standard output

Once upon a time there lived a good fairy A. One day a fine young man B came to her and asked to predict his future. The fairy looked into her magic ball and said that soon the fine young man will meet the most beautiful princess ever and will marry her. Then she drew on a sheet of paper n points and joined some of them with segments, each of the segments starts in some point and ends in some other point. Having drawn that picture, she asked the young man to erase one of the segments from the sheet. Then she tries to colour each point red or blue so, that there is no segment having points of the same colour as its ends. If she manages to do so, the prediction will come true. B wants to meet the most beautiful princess, that's why he asks you to help him. Find all the segments that will help him to meet the princess.

Input

The first input line contains two integer numbers: n — amount of the drawn points and m — amount of the drawn segments (1 ≤ n ≤ 104, 0 ≤ m ≤ 104). The following m lines contain the descriptions of the segments. Each description contains two different space-separated integer numbers vu (1 ≤ v ≤ n, 1 ≤ u ≤ n) — indexes of the points, joined by this segment. No segment is met in the description twice.

Output

In the first line output number k — amount of the segments in the answer. In the second line output k space-separated numbers — indexes of these segments in ascending order. Each index should be output only once. Segments are numbered from 1 in the input order.

Examples
input

Copy
4 4
1 2
1 3
2 4
3 4
output
4
1 2 3 4
input

Copy
4 5
1 2
2 3
3 4
4 1
1 3
output
1
5
题目大意:给定一个无向图,问删哪些边之后这个图变成二分图。求的是一个边的集合,实际上只删一条边.
分析:挺好的一道题.
   一个图是二分图的充要条件是不存在奇环.将一条奇环上的边删去就能破坏掉这个奇环,如果要破坏掉所有的奇环,那么删的边就必须是所有奇环的交集.
   仅仅只是删掉交集这么简单吗?如果一条边同时在偶环和奇环上,删掉这条边后偶环和奇环就会重新组合成一个奇环.那么删的这条边就必须满足两个条件:1.在所有奇环的交集中. 2.不在任何偶环上.
   那么找环就好了.天真的我以为直接dfs+栈维护一下就好了.这道题的环是会重叠的,这种做法行不通......换一种做法,每个点记录第一条连向这个点的边的编号(其实记录的就是树边),那么可以把边转换为点,在点上对树边进行操作,非树边需要特判一下.
   每次找到一条非树边,这条边连接的两个点是u,v,如果构成了一个奇环,就在维护奇环线段树中把u,v这条链+1,否则在维护偶环线段树中把u,v这条链+1.怎么提取这条链?树链剖分!
   最后是一些细节:如果没有奇环,所有的边都满足条件;如果奇环只有1个,那么那个奇环的非树边要考虑进来;维护的是一个图而不是树,在dfs时只考虑树边!
#include <stack>
#include <cstdio>
#include <cstring>
#include <iostream>
#include <algorithm> using namespace std; const int maxn = ; int n,m,head[maxn],to[maxn],nextt[maxn],id[maxn],tot = ,vis[maxn],cnta,ans,cntb,anss[maxn];
int bianhao[maxn],h[maxn],son[maxn],top[maxn],pos[maxn],sizee[maxn],cnt,fa[maxn],flag;
int L[maxn << ],R[maxn << ],sum1[maxn << ],sum2[maxn << ],tag1[maxn << ],tag2[maxn << ]; struct node
{
int x,y;
}e[maxn]; void add(int x,int y)
{
to[tot] = y;
nextt[tot] = head[x];
head[x] = tot++;
} void dfs1(int u,int faa)
{
h[u] = h[faa] + ;
sizee[u] = ;
fa[u] = faa;
for (int i = head[u];i;i = nextt[i])
{
int v = to[i];
if (v == faa || h[v])
continue;
bianhao[v] = (i / ) + (i % );
vis[(i / ) + (i % )] = ;
dfs1(v,u);
sizee[u] += sizee[v];
if (sizee[v] > sizee[son[u]])
son[u] = v;
}
} void dfs2(int u,int topp)
{
pos[u] = ++cnt;
id[cnt] = u;
top[u] = topp;
if (son[u])
dfs2(son[u],topp);
for (int i = head[u];i;i = nextt[i])
{
int v = to[i];
if (v == fa[u] || v == son[u] || fa[v] != u)
continue;
dfs2(v,v);
}
} void build(int o,int l,int r)
{
L[o] = l;
R[o] = r;
if (l == r)
return;
int mid = (l + r) >> ;
build(o * ,l,mid);
build(o * + ,mid + ,r);
} void pushup(int o)
{
sum1[o] = sum1[o * ] + sum1[o * + ];
sum2[o] = sum2[o * ] + sum2[o * + ];
} void pushdown(int o)
{
if (tag1[o])
{
tag1[o * ] += tag1[o];
tag1[o * + ] += tag1[o];
sum1[o * ] += tag1[o] * (R[o * ] - L[o * ] + );
sum1[o * + ] += tag1[o] * (R[o * + ] - L[o * + ] + );
tag1[o] = ;
}
if (tag2[o])
{
tag2[o * ] += tag2[o];
tag2[o * + ] += tag2[o];
sum2[o * ] += tag2[o] * (R[o * ] - L[o * ] + );
sum2[o * + ] += tag2[o] * (R[o * + ] - L[o * + ] + );
tag2[o] = ;
}
} void update1(int o,int l,int r,int x,int y)
{
if(x <= l && r <= y)
{
sum1[o] += (r - l + );
tag1[o]++;
return;
}
pushdown(o);
int mid = (l + r) >> ;
if (x <= mid)
update1(o * ,l,mid,x,y);
if (y > mid)
update1(o * + ,mid + ,r,x,y);
} void update2(int o,int l,int r,int x,int y)
{
if(x <= l && r <= y)
{
sum2[o] += (r - l + );
tag2[o]++;
return;
}
pushdown(o);
int mid = (l + r) >> ;
if (x <= mid)
update2(o * ,l,mid,x,y);
if (y > mid)
update2(o * + ,mid + ,r,x,y);
} void change(int x,int y,int tagg)
{
if (h[x] < h[y])
swap(x,y);
while (top[x] != top[y])
{
if (h[top[x]] < h[top[y]])
swap(x,y);
int t = top[x];
if (tagg == )
update1(,,n,pos[t],pos[x]);
else
update2(,,n,pos[t],pos[x]);
x = fa[t];
}
if (x == y)
return;
if (h[x] < h[y])
swap(x,y);
if (tagg == )
update1(,,n,pos[y] + ,pos[x]); //为什么要+1?因为实际维护的是边.
else
update2(,,n,pos[y] + ,pos[x]);
} int query1(int o,int l,int r,int v)
{
if (l == r)
return sum1[o];
pushdown(o);
int mid = (l + r) >> ;
if (v <= mid)
return query1(o * ,l,mid,v);
else
return query1(o * + ,mid + ,r,v);
} int query2(int o,int l,int r,int v)
{
if (l == r)
return sum2[o];
pushdown(o);
int mid = (l + r) >> ;
if (v <= mid)
return query2(o * ,l,mid,v);
else
return query2(o * + ,mid + ,r,v);
} int main()
{
scanf("%d%d",&n,&m);
for (int i = ; i <= m; i++)
{
int x,y;
scanf("%d%d",&x,&y);
e[i].x = x;
e[i].y = y;
add(x,y);
add(y,x);
}
for (int i = ; i <= n; i++)
if (!h[i])
dfs1(i,),dfs2(i,i);
build(,,n);
for (int i = ; i <= m; i++)
{
if (!vis[i])
{
int x = e[i].x,y = e[i].y;
if (abs(h[x] - h[y]) % == ) //偶环
{
cntb++;
change(x,y,-);
}
else
{
cnta++;
flag = i;
change(x,y,);
}
}
}
if (cnta == )
{
for (int i = ; i <= m; i++)
anss[++ans] = i;
}
else
{
if (cnta == )
anss[++ans] = flag;
for (int i = ; i <= n; i++)
{
if (query1(,,n,pos[i]) == cnta && query2(,,n,pos[i]) == )
anss[++ans] = bianhao[i];
}
sort(anss + ,anss + + ans);
}
printf("%d\n",ans);
for(int i = ; i <= ans; i++)
printf("%d ",anss[i]);
printf("\n"); return ;
}
    

Codeforces 19.E Fairy的更多相关文章

  1. codeforces 19 D. Points(线段树+set二分)

    题目链接:http://codeforces.com/contest/19/problem/D 题意:给出3种操作:1)添加点(x,y),2)删除点(x,y),3)查询离(x,y)最近的右上方的点. ...

  2. Codeforces Round #404 (Div. 2) C. Anton and Fairy Tale 二分

    C. Anton and Fairy Tale 题目连接: http://codeforces.com/contest/785/problem/C Description Anton likes to ...

  3. Codeforces Gym100735 I.Yet another A + B-Java大数 (KTU Programming Camp (Day 1) Lithuania, Birˇstonas, August 19, 2015)

    I.Yet another A + B You are given three numbers. Is there a way to replace variables A, B and C with ...

  4. Codeforces Gym100735 G.LCS Revised (KTU Programming Camp (Day 1) Lithuania, Birˇstonas, August 19, 2015)

    G.LCS Revised   The longest common subsequence is a well known DP problem: given two strings A and B ...

  5. Codeforces Gym100735 E.Restore (KTU Programming Camp (Day 1) Lithuania, Birˇstonas, August 19, 2015)

    E - Restore Given a matrix A of size N * N. The rows are numbered from 0 to N-1, the columns are num ...

  6. Educational Codeforces Round 19 A, B, C, E(xjb)

    题目链接:http://codeforces.com/contest/797 A题 题意:给出两个数n, k,问能不能将n分解成k个因子相乘的形式,不能输出-1,能则输出其因子: 思路:将n质因分解, ...

  7. 【19.77%】【codeforces 570D】Tree Requests

    time limit per test2 seconds memory limit per test256 megabytes inputstandard input outputstandard o ...

  8. 【19.46%】【codeforces 551B】ZgukistringZ

    time limit per test2 seconds memory limit per test256 megabytes inputstandard input outputstandard o ...

  9. 【codeforces 785C】Anton and Fairy Tale

    [题目链接]:http://codeforces.com/contest/785/problem/C [题意] 容量为n的谷仓,每一天都会有m个谷子入仓(满了就视为m);第i天 会有i只鸟叼走i个谷子 ...

随机推荐

  1. [network]RIP协议

    水平分割:一种避免路由环路的出现和加快路由汇聚的技术. 原理:路由器从某个接口接收到的更新信息不允许再从这个接口发送回去. 优点:1. 阻止路由环路产生:2. 减少路由器更新信息占用的链路带宽资源. ...

  2. Nginx内容缓存

    本节介绍如何启用和配置从代理服务器接收的响应的缓存.主要涉及以下内容 - 缓存介绍 启用响应缓存 涉及缓存的NGINX进程 指定要缓存的请求 限制或绕过缓存 从缓存中清除内容 配置缓存清除 发送清除命 ...

  3. JavaScript学习笔记(二)——函数和数组

    第二章 函数简介 1 第一个函数示例 <script language="JavaScript" type="text/JavaScript"> f ...

  4. 基于C#的机器学习--颜色混合-自组织映射和弹性神经网络

    自组织映射和弹性神经网络 自组织映射(SOM),或者你们可能听说过的Kohonen映射,是自组织神经网络的基本类型之一.自组织的能力提供了对以前不可见的输入数据的适应性.它被理论化为最自然的学习方式之 ...

  5. Paper Reading - Mind’s Eye: A Recurrent Visual Representation for Image Caption Generation ( CVPR 2015 )

    Link of the Paper: https://ieeexplore.ieee.org/document/7298856/ A Correlative Paper: Learning a Rec ...

  6. AutoCAD2015激活码和密钥

    CAD2015序列号和密钥 序列号:    666-69696969     产品密钥:     001G1 序列号:    356-72378422     产品密钥:     206G1 安装说明 ...

  7. 第二章:Internet地址结构

    引言 本章主要介绍了: 如何为Internet中的设备分配IP地址. 有助于理由可扩展性的地址结构分配方式. 特殊用途的地址. 表示IP地址 IPv4地址 长32位,采用点分四组或点分十进制来表示. ...

  8. Objective-C Json转Model(利用Runtime特性)

    封装initWithNSDictionary:方法 该方法接收NSDictionary对象, 返回PersonModel对象. #pragma mark - 使用runtime将JSON转成Model ...

  9. 团队作业4——第一次项目冲刺(Alpha版本)第二次

    一.会议内容 各人进行下一步工作 发现沟通流程问题并解决 二.各人工作 成员 计划任务 遇见难题 贡献比 塗家瑜(组长) 后端逻辑处理 无 1 张新磊 数据库搭建 无 1 姚燕彬 测试计划编写 无 1 ...

  10. 【final】140字互评②

    按照产品发布顺序 nice!----约吧 我们的团队展示相对于上次的手足无措,有了一定进步.但是整体还是不那么流畅总结起来的缺点是: 1.发布时,摄像头不清晰 且抖动 我们没有把摄像头固定,并且为了让 ...