Heavy-light Decompositions Problem Code: HLDOTSSubmit

All submissions for this problem are available.

Heavy-light decomposition of a tree is a powerful tool that often helps in the most difficult tree data structure problems.

Heavy-light decomposition is to be built on a rooted tree. In this problem, the node with the number 1 should be considered the root of a tree. Heavy light decomposition is a colouring of edges of the tree. Each edge is either heavy or light. For each non leaf node (node having degree greater than 1), from all the edges emanating from it into the subtree rooted at this vertex should have exactly one heavy edge.

The heavy-light decomposition is called correct, if you can reach any node from the root node by using no more that ⌊ log2 N ⌋ light edges, where N is the number of nodes in the tree.

Given a tree, calculate the number of its' correct heavy-light decompositions. As answer could be very large, please print it modulo 19101995.

Input

There is exactly one test case.

The first line of input consists of a single integer N, denoting the number of the nodes in the tree.

Each of the following N - 1 lines contains a pair of integers, denoting the numbers of the nodes that are connected with an edge. The nodes are enumerated by positive integers in the range [1; N].

Output

Output the number of correct heavy-light decompositions of the given tree. Since this number can be huge, please output it modulo 19101995.

Constraints

  • (Subtask 1): 1 ≤ N ≤ 20 - 21 point.
  • (Subtask 2): 1 ≤ N ≤ 1000 - 34 points.
  • (Subtask 3): 1 ≤ N ≤ 100000 - 45 points.

Example

Input:
7
1 2
3 1
3 4
3 5
2 6
2 7 Output:
8

Explanation

Example case 1. Input is a complete binary tree. It consists of 7 nodes, therefore you can't have more than ⌊log2 7⌋ = ⌊(2.80735492206)⌋ = 2 light edges on the path from the root node to any other one. But the tree's height is 2, so you can choose the decomposition in any way you like. All the decompositions will be correct ones. There are three nodes that has outgoing edges from them (in the direction opposite to the root's one), their numbers are 1, 2 and 3. Each of them has 2 outgoing edges from which you can colour exactly one of them heavy, so overall you'll have 2 * 2 * 2 = 8 options of creating the correct heavy-light decompositions.

 
 
题意:求从根到任意节点经过的轻链的个数不超过$log_2n$的轻重链划分的方案数。
一个树dp。
由于模数是一个合数,所以不能直接用快速幂求逆元,可以用改用前缀后缀和做。
f是前缀和,g是后缀和。
注意没有儿子(叶子)和只有一个儿子的特殊情况。
//Serene
#include<algorithm>
#include<iostream>
#include<cstring>
#include<cstdlib>
#include<cstdio>
#include<cmath>
using namespace std;
const long long maxn=1e5+10,mod=19101995;
long long n,f[maxn][20],g[maxn][20],ans[maxn][20],fa[maxn],sz,fr[2*maxn]; long long aa;char cc;
long long read() {
aa=0;cc=getchar();
while(cc<'0'||cc>'9') cc=getchar();
while(cc>='0'&&cc<='9') aa=aa*10+cc-'0',cc=getchar();
return aa;
} long long fir[maxn],nxt[2*maxn],to[2*maxn],e=0;
void add(long long x,long long y) {
to[++e]=y;nxt[e]=fir[x];fir[x]=e;
to[++e]=x;nxt[e]=fir[y];fir[y]=e;
} long long rs;
long long qp(long long x,long long k) {
rs=1;if(x<=1) return x;
while(k) {
if(k&1) (rs*=x)%=mod;
x=x*x%mod;k>>=1;
}
return rs;
} void dfs(long long pos) {
long long z,tt=0,tot;
for(long long y=fir[pos];y;y=nxt[y]) {
if((z=to[y])==fa[pos]) continue;
fa[z]=pos; dfs(z);
}
for(long long i=0;i<=sz;++i) f[0][i]=1ll;
for(long long y=fir[pos];y;fr[nxt[y]]=y,y=nxt[y]) {
if((z=to[y])==fa[pos]) continue; tt++;
for(long long i=0;i<=sz;++i) (f[tt][i]=f[tt-1][i]*ans[z][i])%=mod;
}
tot=tt;
for(long long i=0;i<=sz;++i) g[tt][i]=1ll;
fr[fir[pos]]=0;
for(long long y=fr[0];y;y=fr[y]) {
if((z=to[y])==fa[pos]) continue; tt--;
for(long long i=0;i<=sz;++i) (g[tt][i]=g[tt+1][i]*ans[z][i])%=mod;
}
if(tot>1) for(long long y=fir[pos];y;y=nxt[y]) {
if((z=to[y])==fa[pos]) continue; tt++;
for(long long i=1;i<=sz;++i) (ans[pos][i]+=f[tt-1][i-1]*g[tt][i-1]%mod*ans[z][i])%=mod;
}
else if(!tot)for(long long i=0;i<=sz;++i) ans[pos][i]=1;
else {
if(to[fr[0]]==fa[pos]) fr[0]=fr[fr[0]];
for(long long i=0;i<=sz;++i) ans[pos][i]=ans[to[fr[0]]][i];
}
} int main() {
n=read();long long x,y;
sz=(long long)((double)log(n)/(double)log(2)+1e-8);
for(long long i=1;i<n;++i) {
x=read();y=read();
add(x,y);
}
dfs(1);
printf("%lld",ans[1][sz]);
return 0;
}
/*
7
1 2
3 2
4 3
5 1
6 2
7 3 right answer:
6
*/

  

对拍的rand:

//Serene
#include<algorithm>
#include<iostream>
#include<cstring>
#include<cstdlib>
#include<cstdio>
#include<cmath>
#include<ctime>
using namespace std;
const int maxn=15;
int n; int main() {
srand((unsigned)time(NULL));
n=rand()%maxn+2;int x;
printf("%d\n1 2\n",n);
for(int i=3;i<=n;++i) {
x=rand()%(i-1)+1;
printf("%d %d\n",i,x);
}
return 0;
}

  

codechef Heavy-light Decompositions的更多相关文章

  1. Heavy Light Decomposition

    Note 1.DFS1 mark all the depth mark fathers mark the heavy/light children mark the size of each subt ...

  2. 树链剖分I 原理

    树链剖分(Heavy Light Decomposition, HLD)是一种将对[树上两点间的路径]上[边或点]的[修改与查询]转化到[序列]上来处理的方法. 目的:将树的边或点转化到一个线性结构( ...

  3. ACM/ICPC 之 拓扑排序-反向(POJ3687)

    难点依旧是题意....需要反向构图+去重+看题 POJ3687-Labeling Balls 题意:1-N编号的球,输出满足给定约束的按原编号排列的重量序列,如果有多组答案,则输出编号最小的Ball重 ...

  4. poj1013

    题目大意:假造的银币 Sally Jones有一些游客给的银币,但是只有11枚是真正的银币(有一枚是假的),从颜色和大小是无法区分真比还是假币的,但是它的重量和真币是不同的,Sally Jones它是 ...

  5. Java8 in action(1) 通过行为参数化传递代码--lambda代替策略模式

    [TOC] 猪脚:以下内容参考<Java 8 in Action> 需求 果农需要筛选苹果,可能想要绿色的,也可能想要红色的,可能想要大苹果(>150g),也可能需要红的大苹果.基于 ...

  6. TCP与UDP区别小结

    TCP(Transmission Control Protocol):传输控制协议 UDP(User Datagram Protocol):用户数据报协议       主要从连接性(Connectiv ...

  7. POJ1013 称硬币

    题目链接:http://poj.org/problem?id=1013 题目大意 有12枚硬币.其中有11枚真币和1枚假币.假币和真币重量不同,但不知道假币比真币轻还是重.现在,用一架天平称了这些币三 ...

  8. 神奇的树上启发式合并 (dsu on tree)

    参考资料 https://www.cnblogs.com/zhoushuyu/p/9069164.html https://www.cnblogs.com/candy99/p/dsuontree.ht ...

  9. HDU 5111 Alexandra and Two Trees 树链剖分 + 主席树

    题意: 给出两棵树,每棵树的节点都有一个权值. 同一棵树上的节点的权值互不相同,不同树上节点的权值可以相同. 要求回答如下询问: \(u_1 \, v_1 \, u_2 \, v_2\):询问第一棵树 ...

随机推荐

  1. Jeecms之JSP访问action类

    因为Jeecms采用spring容器管理类,在web容器加载的时候类都已经实例化好了.我们可以通过在JSP中访问spring上下文的方式来调用action业务类例: ApplicationContex ...

  2. Python基础-列表、元组、字典、字符串(精简解析),全网最齐全。

    一.列表 =====================================================1.列表的定义及格式: 列表是个有序的,可修改的,元素用逗号隔开,用中括号包围的序列 ...

  3. 关于hive表同步类型问题

    今天华为做实施的时候发现kylin做刷cube的时候发现源表数据类型不适合刷到kylin提供查询接口.问了下同事发现一个比较简单的解决办法. 源表是String类型,做hive视图可以做个hive表提 ...

  4. A20地址线科普【转载】

    1981 年8 月,IBM 公司最初推出的个人计算机IBM PC 使用的CPU 是Intel 8088.在该微机中地址线只有20 根(A0 – A19).在当时内存RAM 只有几百KB 或不到1MB ...

  5. c/c++ explicit用法

    试想一个类,它的构造函数只有一个参数, class test { test( int a ) {} 4 } 这时我想用一个int b = 10初始化它,这当然没问题,但如果你用一个字符例如'c'初始化 ...

  6. spring源码学习之AOP(一)

    继续源码学习,看了spring中基础的容器和AOP感觉自己也没有什么长进,哈哈,我也不知道到底有用没有,这可能是培养自己的一种精神吧,不管那么多,继续学习!AOP中 AOP中几个重要的概念:(1)Ad ...

  7. 【笔记】LR11中关联设置

    LR中关联建议都手动进行,自动不好用,也容易出错. 在LR中我们什么要做关联:1.关联解决的是动态数据的参数化.2.关联的数据一定是服务器响应的数据.3.服务器响应过来的数据在后面的服务还要使用. 手 ...

  8. How to class-dump iPad apps?

    http://stackoverflow.com/questions/4776593/how-to-class-dump-ipad-apps The issue here is that the bi ...

  9. 数据库完整性 ch.5

    数据库的完整性 是指 数据的正确性(correctness) 和 相容性 (compat-ability) 5.1 实体完整性 定义 对单属性码的说明有两种方法,一种是定义为表约束条件,一种是定义为列 ...

  10. Java 8最快的垃圾收集器是什么?

    OpenJDK 8 有多种 GC(Garbage Collector)算法,如 Parallel GC.CMS 和 G1.哪一个才是最快的呢?如果在 Java 9 中将 Java 8 默认的 GC 从 ...