【UVALive - 5131】Chips Challenge(上下界循环费用流)
Description
A prominent microprocessor company has enlisted your help to lay out some interchangeable components
(widgets) on some of their computer chips. Each chip’s design is an N × N square of slots. One
slot can hold a single component, and you are to try to fit in as many widgets as possible.
Modern processor designs are complex, of course. You unfortunately have several restrictions:
• Some of the slots are disabled.
• Some of the slots are already occupied by other components and cannot be used for widgets.
• There are sibling memory buses connected to the horizontal and vertical edges of the chip and
their bandwidth loads need to match. As such, there must be exactly as many components in
the first row as in the first column, exactly as many in the second row as in the second column,
and so on. Component counts include both the components already specified on the chip and the
added widgets.
• Similarly, the power supply is connected at the end of each row and column. To avoid hot spots,
any given row or column must have no more than A/B of the total components on the chip for
a given A and B.
A specification for a chip is N lines of N characters, where ‘.’ indicates an open slot, ‘/’ indicates
a disabled slot, and ‘C’ indicates a slot already occupied by a component. For example:
CC/..
././/
..C.C
/.C..
/./C/
If no more than 3/10 of the components may be in any one row or column, the maximum number of
widgets that can be added to this 5 × 5 chip is 7. A possible arrangement is below, where ‘W’ indicates
a widget added in an open slot.
CC/W.
W/W//
W.C.C
/.CWW
/W/C/erbInput
The input consists of several test cases. Each case starts with a line containing three integers: The size
of the chip N (1 ≤ N ≤ 40), and A and B (1 ≤ B ≤ 1000, 0 ≤ A ≤ B) as described above. Each of the
following N lines contains N characters describing the slots, one of ‘.’, ‘/’ or ‘C’, as described above.
The last test case is followed by a line containing three zeros.Output
For each test case, display a single line beginning with the case number. If there is a solution, display
the maximum number of widgets that can be added to the chip. Display ‘impossible’ if there is no
solution.
Follow the format of the sample output.Sample Input
2 1 1
/.
//
2 50 100
/.
C/
2 100 100
./
C.
5 3 10
CC/..
././/
..C.C
/.C..
/./C/
5 2 10
CC/..
././/
..C.C
/.C..
/./C/
0 0 0Sample Output
Case 1: 0
Case 2: 1
Case 3: impossible
Case 4: 7
Case 5: impossible
【题意】
有一个n*n的矩阵(n<=40),每个位置有三种情况,/表示不能用,C表示这个位置有一个芯片,'.'表示这个位置可以放芯片,要求第i行的芯片总数等于第i列的芯片总数,每行或的每列的芯片总数不能超过总芯片数的A/B。
【分析】
没有行列约束的时候直接行列建边。对于A/B的约束其实可以枚举行列的最大值然后加限制跑最大流(因为行很少)。不过有行等于列的约束,于是再建一条列到行的反向边,上限定为你枚举的约束,做循环流。因为循环流要加边而且判满流,所以要加一个费用才能计算,于是就是上下界循环费用流了。
#include<cstdio>
#include<cstdlib>
#include<cstring>
#include<iostream>
#include<algorithm>
#include<queue>
using namespace std;
#define Maxn 10100
#define Maxm 1001000
#define INF 0xfffffff int map[][];
int first[Maxn],dis[Maxn],pre[Maxn],flow[Maxn];
char s[];
bool inq[Maxn]; int st,ed,sum,h;
int n,A,B; struct node
{
int x,y,f,c,o,next;
}t[Maxm];int len; int mymax(int x,int y) {return x>y?x:y;}
int mymin(int x,int y) {return x<y?x:y;} void ins(int x,int y,int f,int c)
{
if(f==) return;
if(x==st) sum+=f;
t[++len].x=x;t[len].y=y;t[len].f=f;t[len].c=c;
t[len].next=first[x];first[x]=len;t[len].o=len+;
t[++len].x=y;t[len].y=x;t[len].f=;t[len].c=-c;
t[len].next=first[y];first[y]=len;t[len].o=len-;
} void make_edge(int x,int y,int k1,int k2,int c)
{
ins(st,y,k2,);
ins(x,ed,k2,c);
ins(y,x,k2-k1,-c);
} queue<int > q;
bool bfs(int f1,int f2)
{
while(!q.empty()) q.pop();
memset(dis,-,sizeof(dis));
memset(inq,,sizeof(inq));
memset(pre,-,sizeof(pre));
inq[f1]=;q.push(f1);flow[f1]=INF;pre[f1]=;dis[st]=;
while(!q.empty())
{
int x=q.front();q.pop();
for(int i=first[x];i;i=t[i].next) if(t[i].f>)
{
int y=t[i].y;
if(dis[y]<dis[x]+t[i].c)
{
dis[y]=dis[x]+t[i].c;
if(!inq[y]) {q.push(y);inq[y]=;}
pre[y]=i;
flow[y]=mymin(flow[x],t[i].f);
}
}
inq[x]=;
}
if(pre[f2]==-) return ;
return flow[f2];
} int ffind(int x,int y)
{
int a,sc=,now;h=;
while(a=bfs(st,ed))
{
now=y;sc+=a*dis[y];
h+=a;
while(now!=x)
{
t[pre[now]].f-=a;
t[t[pre[now]].o].f+=a;
now=t[pre[now]].x;
}
}
return sc;
} int get_ans(int x)
{
st=*n+,ed=st+;
len=;sum=;
memset(first,,sizeof(first));
for(int i=;i<=n;i++)
for(int j=;j<=n;j++)
{
if(map[i][j]==) make_edge(i,j+n,,,);
else if(map[i][j]==) ins(i,j+n,,);
}
for(int i=;i<=n;i++) make_edge(i+n,i,,x,);
int a=ffind(st,ed);
if(h!=sum) return -;
if(x*B<=a*A) return a;
return -;
} int main()
{
int kase=;
while()
{
scanf("%d%d%d",&n,&A,&B);
if(n==&&A==&&B==) break;
bool ok=;
int sc=;
for(int i=;i<=n;i++)
{
scanf("%s",s);
for(int j=;j<n;j++)
{
if(s[j]=='/') map[i][j+]=;
else if(s[j]=='.') map[i][j+]=;
else map[i][j+]=,sc++;
}
}
for(int i=;i<=n;i++)
{
int s1=,s2=;
for(int j=;j<=n;j++)
{
if(map[i][j]==) s1++;
if(map[j][i]==) s2++;
}
if(s1!=s2||s1*B>sc*A||s2*B>sc*A) {ok=;break;}
}
printf("Case %d: ",++kase);
int maxx=-;
if(ok) maxx=sc;
for(int i=;i<=n;i++)
{
maxx=mymax(maxx,get_ans(i)); }
if(maxx!=-) maxx-=sc;
if(maxx==-) printf("impossible\n");
else printf("%d\n",maxx);
}
return ;
}
[LA5131]
2016-06-04 13:27:29
【UVALive - 5131】Chips Challenge(上下界循环费用流)的更多相关文章
- 【BZOJ-4213】贪吃蛇 有上下界的费用流
4213: 贪吃蛇 Time Limit: 5 Sec Memory Limit: 64 MBSubmit: 58 Solved: 24[Submit][Status][Discuss] Desc ...
- BZOJ2055 80人环游世界 网络流 费用流 有源汇有上下界的费用流
https://darkbzoj.cf/problem/2055 https://blog.csdn.net/Clove_unique/article/details/54864211 ←对有上下界费 ...
- BZOJ 2055: 80人环游世界(有上下界的费用流)
题面 Time Limit: 10 Sec Memory Limit: 64 MB Submit: 693 Solved: 434 [Submit][Status][Discuss] Descript ...
- 【上下界网络流 费用流】bzoj2055: 80人环游世界
EK费用流居然写错了…… Description 想必大家都看过成龙大哥的<80天环游世界>,里面的紧张刺激的打斗场面一定给你留下了深刻的印象.现在就有这么 一个80人的团 ...
- Codeforces Gym 101190 NEERC 16 .D Delight for a Cat (上下界的费用流)
ls是一个特别堕落的小朋友,对于n个连续的小时,他将要么睡觉要么打隔膜,一个小时内他不能既睡觉也打隔膜 ,因此一个小时内他只能选择睡觉或者打隔膜,当然他也必须选择睡觉或打隔膜,对于每一个小时,他选择睡 ...
- UvaL-7670 上下界可行费用流
#include <iostream> #include <cstdio> #include <cstring> #include <queue> #d ...
- BZOJ 3876 支线剧情 有源汇有上下界最小费用可行流
题意: 给定一张拓扑图,每条边有边权,每次只能从第一个点出发沿着拓扑图走一条路径,求遍历所有边所需要的最小边权和 分析: 这道题乍一看,可能会想到什么最小链覆盖之类的,但是仔细一想,会发现不行,一是因 ...
- BZOJ 2055 80人环游世界 有上下界最小费用可行流
题意: 现在有这么一个m人的团伙,也想来一次环游世界. 他们打算兵分多路,游遍每一个国家. 因为他们主要分布在东方,所以他们只朝西方进军.设从东方到西方的每一个国家的编号依次为1...N.假若第 ...
- zoj3229 Shoot the Bullet(有源汇有上下界的最大流)
题意: 一个屌丝给m个女神拍照,计划拍照n天,每一天屌丝给给定的C个女神拍照,每天拍照数不能超过D张,而且给每个女神i拍照有数量限制[Li,Ri],对于每个女神n天的拍照总和不能少于Gi,如果有解求屌 ...
随机推荐
- myeclipse设置技巧
如何设置jsp的默认打开为MyEclipse JSP Editor? windows -> General -> Editor - > File Associations 选择 *. ...
- cookie 和 HttpSession
保存会话数据的两种技术 Cookie Cookie 是客户端技术,程序把每个用户的数据以cookie的形式写给用户的浏览器.当用户使用浏览器再去访问服务器中的web资源时,就会带着各自的数据去.web ...
- oracle合并查询
1). Union 该操作符用于取得两个结果集的并集.当使用该操作符时,会自动去掉结果集中重复行. 2).union all 该操作符与union相似,但是它不会取消重复行,而且不会排序. 3). I ...
- Android(java)学习笔记180:Android MediaPlayer 播放prepareAsync called in state 8解决办法
使用android MediaPlayer播放音频文件时,有时会出现prepareasync called in state 8错误. 以下方法可以避免这个异常出现. 第一种方法: private ...
- 禁止输出重定向(>)覆盖已存在文件(防止误操作)
在输出重定向中,>表示重定向并覆盖已有文件内容,为了防止误操作,覆盖重要的内容,可以使用如下命令: set -C 这样输出重定向到一个已有文件就会提示: cannot overwrite exi ...
- LCA问题
基本概念 LCA:树上的最近公共祖先,对于有根树T的两个结点u.v,最近公共祖先LCA(T,u,v)表示一个结点x,满足x是u.v的祖先且x的深度尽可能大. RMQ:区间最小值查询问题.对于长度为n的 ...
- quartz.net 基于数据库的简单实现
前面简单学习了通过XML配置或者内存指定的方式实现调度任务.但此用法实战用途较小,企业上多需要分布式集群的方式.quart团队也考虑到了这点,于是有了我们今天要学习的.基于数据库实现分布式. Name ...
- JavaScript的垃圾回收
来自MDN,Memory Management 简介 在底层语言中,比如C,有专门的内存管理机制,比如malloc() 和 free().而Javascript是有垃圾回收(garbage colle ...
- JS获取页面上所有input
for (var i = 0; i < document.getElementsByTagName("input").length; i++) { if (document. ...
- jQuery UI 日期控件--datepicker
在web开发中,日期的输入经常会遇到.我们会用的解决方法有: 1.自己写css和js,对日期进行控制:----有点浪费精力和时间: 2.用easyui插件中的日期插件来实现: 3.用juqery-ui ...