Description

CIA headquarter collects data from across the country through its classified network. They have been using optical fibres long before it's been deployed on any civilian projects. However they are still under a lot pressure recently because the data are growing rapidly. As a result they are considering upgrading the network with new technologies that provide a few times wider bandwidth. In the experiemental stage, they would like to upgrade one segment of their original network in order to see how it performs. And as a CIA intern it's your responsibility to investigate which segment could actually help increase the total bandwidth the headquarter receives, suppose that all the cities have infinite data to send and the routing algorithm is optimized. As they have prepared the data for you in a few minutes, you are told that they need the result immediately. Well, practically immediately.

Input

Input contains multiple test cases. First line of each test case contains three integers n, m and l, they represent the number of cities, the number of relay stations and the number of segments. Cities will be referred to as integers from 1 to n, while relay stations use integers from n+1 to n+m. You can saves assume that n + m <= 100, l <= 1000 (all of them are positive). The headquarter is identified by the integer 0.

The next l lines hold a segment on each line in the form of a b c, where a is the source node and b is the target node, while c is its bandwidth. They are all integers where a and b are valid identifiers (from 0 to n+m). c is positive. For some reason the data links are all directional.

The input is terminated by a test case with n = 0. You can safely assume that your calculation can be housed within 32-bit integers.

Output

For each test print the segment id's that meets the criteria. The result is printed in a single line and sorted in ascending order, with a single space as the separator. If none of the segment meets the criteria, just print an empty line. The segment id is 1 based not 0 based.

题目大意:有n个发射点,m个中间点,和一个接受点0,中间由L条有向光纤连接,每个光纤有一定的带宽,问扩大哪些光纤(只能扩大一条)的带宽可以扩大发射点的总带宽。

思路:从源点S到n个发射点连一条容量为无穷大的边,再连接L条光纤,容量为该光纤的带宽,以0为汇点T。题目转化成扩大哪些边的容量可以增加最大流。先在图上求最大流,在残量网络中,把S能到的点标记为1,把能到T的点标记为2,那么当一条边的出发点为1、结束点为2,这条边就是关键割边,也就是所求的边之一。如果有一条边的其中一个点未被标记,那么扩大这条边的容量,也就不过是从源点到达的点多了一个,或者到达汇点的点多了一个,但是在这个残量网络中依旧没有从S到T的增广路。如果两个点都未被标记……那更加不可以啦……

PS:出发点为2结束点为1的边,增加容量也不会有增广路出现,曾因为这个WA了一次>_<

BFS+ISAP(30MS):

 #include <cstdio>
#include <cstring>
#include <algorithm>
#include <queue>
using namespace std; const int MAXN = ;
const int MAXE = ;
const int INF = 0x3fff3fff; struct SAP {
int head[MAXN], dis[MAXN], pre[MAXN], cur[MAXN], gap[MAXN];
int to[MAXE], next[MAXE], flow[MAXE];
int n, st, ed, ecnt; void init() {
memset(head, , sizeof(head));
ecnt = ;
} void add_edge(int u, int v, int c) {
to[ecnt] = v; flow[ecnt] = c; next[ecnt] = head[u]; head[u] = ecnt++;
to[ecnt] = u; flow[ecnt] = ; next[ecnt] = head[v]; head[v] = ecnt++;
//printf("%d->%d flow = %d\n", u, v, c);
} void bfs() {
memset(dis, 0x3f, sizeof(dis));
queue<int> que; que.push(ed);
dis[ed] = ;
while(!que.empty()) {
int u = que.front(); que.pop();
++gap[dis[u]];
for(int p = head[u]; p; p = next[p]) {
int &v = to[p];
if(flow[p ^ ] && dis[v] > n) {
dis[v] = dis[u] + ;
que.push(v);
}
}
}
} int Max_flow(int ss, int tt, int nn) {
st = ss; ed = tt; n = nn;
int ans = , minFlow = INF, u;
for(int i = ; i <= n; ++i) {
cur[i] = head[i];
gap[i] = ;
}
u = pre[st] = st;
bfs();
while(dis[st] < n) {
bool flag = false;
for(int &p = cur[u]; p; p = next[p]) {
int &v = to[p];
if(flow[p] && dis[u] == dis[v] + ) {
flag = true;
minFlow = min(minFlow, flow[p]);
pre[v] = u;
u = v;
if(u == ed) {
ans += minFlow;
while(u != st) {
u = pre[u];
flow[cur[u]] -= minFlow;
flow[cur[u] ^ ] += minFlow;
}
minFlow = INF;
}
break;
}
}
if(flag) continue;
int minDis = n - ;
for(int p = head[u]; p; p = next[p]) {
int &v = to[p];
if(flow[p] && minDis > dis[v]) {
minDis = dis[v];
cur[u] = p;
}
}
if(--gap[dis[u]] == ) break;
gap[dis[u] = minDis + ]++;
u = pre[u];
}
return ans;
} int mark[MAXN]; void make_cut() {
memset(mark, , sizeof(mark));
queue<int> que;
que.push(st); mark[st] = ;
while(!que.empty()) {
int u = que.front(); que.pop();
for(int p = head[u]; p; p = next[p]) {
int &v = to[p];
if(flow[p] && !mark[v]) {
mark[v] = ;
que.push(v);
}
}
}
que.push(ed); mark[ed] = ;
while(!que.empty()) {
int u = que.front(); que.pop();
for(int p = head[u]; p; p = next[p]) {
int &v = to[p];
if(flow[p ^ ] && !mark[v]) {
mark[v] = ;
que.push(v);
}
}
}
}
} G; int from[MAXE], to[MAXE];
int n, m, L, c; int main() {
while(scanf("%d%d%d", &n, &m, &L) != EOF) {
if(n == ) break;
int ss = n + m + , tt = ;
G.init();
for(int i = ; i <= n; ++i) G.add_edge(ss, i, INF);
for(int i = ; i <= L; ++i) {
scanf("%d%d%d", &from[i], &to[i], &c);
G.add_edge(from[i], to[i], c);
}
//printf("%d\n", G.Max_flow(ss, tt, ss));
G.Max_flow(ss, tt, ss);
G.make_cut();
bool flag = false;
for(int i = ; i <= L; ++i) {
int &u = from[i], &v = to[i];
if(G.mark[u] == && G.mark[v] == ) {
if(flag) printf(" ");
printf("%d", i);
flag = true;
}
}
puts("");
}
}

ZOJ 2532 Internship(最大流找关键割边)的更多相关文章

  1. ZOJ 2532 Internship 求隔边

    Internship Time Limit: 5 Seconds      Memory Limit: 32768 KB CIA headquarter collects data from acro ...

  2. zoj 2532 Internship【最小割】

    就是求哪些边在最大流上满流,也就是找割边.把0作为t点,s向所有的1~n连流量为inf的边,其他的边按照流量连.跑一遍最大流,从s顺着有残余流量的正向边dfs打标记fr,从t顺着正向边有残余流量的反向 ...

  3. ZOJ 2532 Internship

    Internship Time Limit: 5000ms Memory Limit: 32768KB This problem will be judged on ZJU. Original ID: ...

  4. B - Internship (网络流关键割边)

    题目链接:https://cn.vjudge.net/contest/281961#problem/B 题目大意:给你n个城市,中间有一些中转站,然后给你终点,再给你l条轨道以及流量,问你增加哪几条轨 ...

  5. poj 3204(最小割--关键割边)

    Ikki's Story I - Road Reconstruction Time Limit: 2000MS   Memory Limit: 131072K Total Submissions: 7 ...

  6. POJ3204 Ikki's Story - Road Reconstruction 网络流图的关键割边

    题目大意:一个有源有汇的城市,问最少增加城市中的多少道路可以增加源到汇上各个路径上可容纳的总车流量增加. 网络流关键割边集合指如果该边的容量增加,整个网络流图中的任意从原点到汇点的路径的流量便可增加. ...

  7. hdu3313 最大流找关键点,或者最短路找关键点.

    题意:      给你一个有向图,然后给你起点和终点,问你从起点到终点有多少个关键点,如果当前的这个点删除了就无法从起点到终点,那么这个点就是一个关键点.. 思路:      (1)有两种做法,我用的 ...

  8. hdu 4975 A simple Gaussian elimination problem 最大流+找环

    原题链接 http://acm.hdu.edu.cn/showproblem.php?pid=4975 这是一道很裸的最大流,将每个点(i,j)看作是从Ri向Cj的一条容量为9的边,从源点除法连接每个 ...

  9. HDU-3416 Marriage Match IV 最短路+最大流 找各最短路的所有边

    题目链接:https://cn.vjudge.net/problem/HDU-3416 题意 给一个图,求AB间最短路的条数(每一条最短路没有重边.可有重复节点) 思路 首先把全部最短路的边找出来,再 ...

随机推荐

  1. Windows10 IIS安装php manager和IIS URL Rewrite 2.0组件的方法

    Windows10中自带的Server:Microsoft-IIS///8.5/10上安装.微软脑子秀逗,跳过了9,以为能解决版本识别的问题,没想到弄成10,还是出现了版本识别的问题,真是自己打自己的 ...

  2. 模板——最小生成树kruskal算法+并查集数据结构

    并查集:找祖先并更新,注意路径压缩,不然会时间复杂度巨大导致出错/超时 合并:(我的祖先是的你的祖先的父亲) 找父亲:(初始化祖先是自己的,自己就是祖先) 查询:(我们是不是同一祖先) 路径压缩:(每 ...

  3. 哈希查找解决地址冲突的两种最常见方法(线性探测再散列,链地址法)C++实现

    #include<iostream>#include<iomanip>using namespace std; typedef struct Node{ int data; s ...

  4. 分布式版本控制系统之Git

    Git Git 是目前世界上最先进的分布式版本控制系统(没有之一) 作用 源代码管理 为什么要进行源代码管理? 方便多人协同开发 方便版本控制 Git的诞生 作者是 Linux 之父:Linus Be ...

  5. Jquery中菜单的展开和折叠

    jquery内容 <script> $(function () { $("dl dt").click(function () { $(this).siblings(). ...

  6. DXP常用的设置及快捷键

    原文地址:http://www.cnblogs.com/NickQ/p/8799240.html 测试环境:Altium Designer Summer 16 一.快捷键 1.原理图和PCB通用快捷键 ...

  7. python学习笔记:第20天 多继承、MRO C3算法

    目录 一.多继承 二.旧式类的MRO 三.新式类的MRO 四.super 一.多继承 之前已经学习过了继承,当出现了x是⼀种y的的时候,就可以使⽤继承关系,即"is-a" 关系.在 ...

  8. Python学习手册之捕获组和特殊匹配字符串

    在上一篇文章中,我们介绍了 Python 的字符类和对元字符进行了深入讲解,现在我们介绍 Python 的捕获组和特殊匹配字符串.查看上一篇文章请点击:https://www.cnblogs.com/ ...

  9. 成都Uber优步司机奖励政策(3月4日)

    滴快车单单2.5倍,注册地址:http://www.udache.com/ 如何注册Uber司机(全国版最新最详细注册流程)/月入2万/不用抢单:http://www.cnblogs.com/mfry ...

  10. Android Stadio配置了gralde的本地路径,但是windos 命令行还是会下载gradle

    如下图: 已经在stadio 里面设置了gradle 的路径,但是在cmd 命令行里面不会去用这个路径. 解决方案:需要在环境变量里面设置一个gradle home GRADLE_USER_HOME ...