CF1088F Ehab and a weird weight formula 题解

文章讲述了如何在给定树结构和点权的情况下,通过优化策略重构树,使得除了一个点权最小的节点外,其他节点连接权值较小的节点,同时最小化总代价。关键策略包括考虑边的代价计算,以及利用点的2^k级祖先来预处理。

题目来源

Description

给定一棵有 n n n 个点的树,点有点权 { a i } \{a_i\} {ai}。且除了唯一的一个点权最小的点,其余的点都与一个权值比它小的点相邻。

现在需要用这 n n n 个点重新构建一棵树,使得代价最小。一个点 u u u 的代价为 deg ⁡ u × a u \deg_u\times a_u degu×au deg ⁡ u \deg_u degu 表示其度数;而一条边 ( u , v ) (u,v) (u,v) 的代价为 ⌈ log ⁡ 2 dis ( u , v ) ⌉ × min ⁡ { a u , a v } \lceil\log_2\text{dis}(u,v)\rceil\times\min\{a_u,a_v\} log2dis(u,v)⌉×min{au,av}

  • 2 ≤ n ≤ 5 × 1 0 5 2\le n\le5\times10^5 2n5×105

Solution

首先,对于点的代价可以很容易的转化到边上。即一条边 ( u , v ) (u,v) (u,v) 的代价为 ⌈ log ⁡ 2 dis ( u , v ) ⌉ × min ⁡ { a u , a v } + a u + a v \lceil\log_2\text{dis}(u,v)\rceil\times\min\{a_u,a_v\}+a_u+a_v log2dis(u,v)⌉×min{au,av}+au+av

其次,由于除了唯一的一个点权最小的点,其余的点都与一个权值比它小的点相邻,则我们可以将这棵树的根节点设为点权最小的点,且每个节点的父亲节点的点权都小于它。

那么,对于一个点,其向自己的祖先连边一定更优于向自己的后代或兄弟连边的(若向兄弟连边,不如向 lca \text{lca} lca 连边,因为距离更短且点权更小;后代则显然)。

同时,若每个点都向自己的祖先连边也就满足了连边完仍为树的条件。

接着,由于 ⌈ log ⁡ 2 dis ( u , v ) ⌉ \lceil\log_2\text{dis}(u,v)\rceil log2dis(u,v)⌉ 的值域只有 O ( log ⁡ n ) O(\log n) O(logn),可直接枚举 k = ⌈ log ⁡ 2 dis ( u , v ) ⌉ k=\lceil\log_2\text{dis}(u,v)\rceil k=log2dis(u,v)⌉

而对于相同的 k k k,显然 a v a_v av 越小答案越优,而 u u u 2 k 2^k 2k 级祖先即为最优的答案。因此我们只需要预处理出每个点的 2 k 2^k 2k 级祖先,并每次枚举 k k k 取最小值,最后相加即可。

Code

#include <bits/stdc++.h>
using namespace std;
int n,a[500005],Min=1e9+1,root,Fa[500005][20],head[1000005],ver[1000005],nxt[1000005],tot;
long long ans;
void add(int u,int v){
    ver[++tot]=v;
    nxt[tot]=head[u],head[u]=tot;
}
void dfs(int u,int fa){
    Fa[u][0]=fa;
    for (int i=1;i<=19;i++) Fa[u][i]=Fa[Fa[u][i-1]][i-1];
    for (int i=head[u];i;i=nxt[i]){
        int v=ver[i];
        if (v==fa) continue;
        dfs(v,u);
    }
}
int main(){
    scanf("%d",&n);
    for (int i=1;i<=n;i++){
        scanf("%d",&a[i]);
        if (a[i]<Min) Min=a[i],root=i;
    }
    for (int i=1,u,v;i<n;i++) scanf("%d%d",&u,&v),add(u,v),add(v,u);
    dfs(root,root);
    for (int i=1;i<=n;i++){
        if (i==root) continue;
        long long Minn=1e18;
        for (int j=0;j<=19;j++){
            int v=Fa[i][j];
            Minn=min(Minn,1ll*(j+1)*a[v]+a[i]);
        }
        ans+=Minn;
    }
    printf("%lld\n",ans);
    return 0;
}
This time Baby Ehab will only cut and not stick. He starts with a piece of paper with an array a a of length n n written on it, and then he does the following: he picks a range ( l , r ) (l,r) and cuts the subsegment a l , a l + 1 , … , a r a l ​ ,a l+1 ​ ,…,a r ​ out, removing the rest of the array. he then cuts this range into multiple subranges. to add a number theory spice to it, he requires that the elements of every subrange must have their product equal to their least common multiple (LCM). Formally, he partitions the elements of a l , a l + 1 , … , a r a l ​ ,a l+1 ​ ,…,a r ​ into contiguous subarrays such that the product of every subarray is equal to its LCM. Now, for q q independent ranges ( l , r ) (l,r), tell Baby Ehab the minimum number of subarrays he needs. Input The first line contains 2 2 integers n n and q q ( 1 ≤ n , q ≤ 10 5 1≤n,q≤10 5 ) — the length of the array a a and the number of queries. The next line contains n n integers a 1 a 1 ​ , a 2 a 2 ​ , … …, a n a n ​ ( 1 ≤ a i ≤ 10 5 1≤a i ​ ≤10 5 ) — the elements of the array a a. Each of the next q q lines contains 2 2 integers l l and r r ( 1 ≤ l ≤ r ≤ n 1≤l≤r≤n) — the endpoints of this query's interval. Output For each query, print its answer on a new line. Examples Inputcopy Outputcopy 6 3 2 3 10 7 5 14 1 6 2 4 3 5 3 1 2 Note The first query asks about the whole array. You can partition it into [ 2 ] [2], [ 3 , 10 , 7 ] [3,10,7], and [ 5 , 14 ] [5,14]. The first subrange has product and LCM equal to 2 2. The second has product and LCM equal to 210 210. And the third has product and LCM equal to 70 70. Another possible partitioning is [ 2 , 3 ] [2,3], [ 10 , 7 ] [10,7], and [ 5 , 14 ] [5,14]. The second query asks about the range ( 2 , 4 ) (2,4). Its product is equal to its LCM, so you don't need to partition it further. The last query asks about the range ( 3 , 5 ) (3,5). You can partition it into [ 10 , 7 ] [10,7] and [ 5 ] [5]. 解法:条件就是两两互质,预处理质因⼦集,找出每个数下⼀个不互质的位置。 i 进⼀步的,可以找出每个 作为左端点,可⾏的右端点最⼤是多⼤(区间 min)。 2j 因此每次划分极⻓就可以了,令 f(i, j) 然后枚举答案的⼆进制下每⼀位即可。 。 求C++代码.
06-10
评论
成就一亿技术人!
拼手气红包6.0元
还能输入1000个字符
 
红包 添加红包
表情包 插入表情
 条评论被折叠 查看
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值