Misha and Grisha are funny boys, so they like to use new underground. The underground has n stations connected with n - 1 routes so that each route connects two stations, and it is possible to reach every station from any other.

The boys decided to have fun and came up with a plan. Namely, in some day in the morning Misha will ride the underground from station s to station f by the shortest path, and will draw with aerosol an ugly text "Misha was here" on every station he will pass through (including s and f). After that on the same day at evening Grisha will ride from station t to station f by the shortest path and will count stations with Misha's text. After that at night the underground workers will wash the texts out, because the underground should be clean.

The boys have already chosen three stations a, b and c for each of several following days, one of them should be station s on that day, another should be station f, and the remaining should be station t. They became interested how they should choose these stations s, f, t so that the number Grisha will count is as large as possible. They asked you for help.

Input

The first line contains two integers n and q (2 ≤ n ≤ 105, 1 ≤ q ≤ 105) — the number of stations and the number of days.

The second line contains n - 1 integers p2, p3, ..., pn (1 ≤ pi ≤ n). The integer pi means that there is a route between stations pi and i. It is guaranteed that it's possible to reach every station from any other.

The next q lines contains three integers a, b and c each (1 ≤ a, b, c ≤ n) — the ids of stations chosen by boys for some day. Note that some of these ids could be same.

Output

Print q lines. In the i-th of these lines print the maximum possible number Grisha can get counting when the stations s, t and f are chosen optimally from the three stations on the i-th day.

Examples
Input
3 2
1 1
1 2 3
2 3 3
Output
2 3
Input
4 1
1 2 3
1 2 3
Output
2
Note

In the first example on the first day if s = 1, f = 2, t = 3, Misha would go on the route 1 2, and Grisha would go on the route 3 1 2. He would see the text at the stations 1 and 2. On the second day, if s = 3, f = 2, t = 3, both boys would go on the route 3 1 2. Grisha would see the text at 3 stations.

In the second examle if s = 1, f = 3, t = 2, Misha would go on the route 1 2 3, and Grisha would go on the route 2 3 and would see the text at both stations.


  题目大意 给定一个有n个点的树,有m个询问,每个询问给出3个点,你需要分别选出一个点作为s,f,t,如果将s到f的路径上的所有点权加1,计算t到f的路径上的点权和,问可能的点权和最大为多少。

  显然树链剖分。

  枚举6种可能,然后取个max。然后水完了。

  然而比赛最后30分钟发现这特么是道水题,早知道该先写D题(内心崩溃...)。然后最后30分钟内20分钟写了出来,然后gdb挂了。。

  然后在没gdb的情况下,今天上午把它改出来(不可思议),然后发现,不会线段树区间lazy标记如何赋初值,树链剖分数组开始滥用。

  另外线段树似乎容易被卡常,所以我就改用树状数组了(动态new点,能不TLE就怪了)。

  (PS:其实这道题有一个很傻逼的$O(nlogn)$的做法,根本没必要树剖,直接求lca就行了,可惜当时没有想出来。。)

Code

 /**
* Codeforces
* Problem#832D
* Accepted
* Time:1466ms
* Memory:14200k
*/
#include <iostream>
#include <cstdio>
#include <ctime>
#include <cmath>
#include <cctype>
#include <cstring>
#include <cstdlib>
#include <fstream>
#include <sstream>
#include <algorithm>
#include <map>
#include <set>
#include <stack>
#include <queue>
#include <vector>
#include <stack>
#ifndef WIN32
#define Auto "%lld"
#else
#define Auto "%I64d"
#endif
using namespace std;
typedef bool boolean;
const signed int inf = (signed)((1u << ) - );
const signed long long llf = (signed long long)((1ull << ) - );
const double eps = 1e-;
const int binary_limit = ;
#define smin(a, b) a = min(a, b)
#define smax(a, b) a = max(a, b)
#define max3(a, b, c) max(a, max(b, c))
#define min3(a, b, c) min(a, min(b, c))
template<typename T>
inline boolean readInteger(T& u){
char x;
int aFlag = ;
while(!isdigit((x = getchar())) && x != '-' && x != -);
if(x == -) {
ungetc(x, stdin);
return false;
}
if(x == '-'){
x = getchar();
aFlag = -;
}
for(u = x - ''; isdigit((x = getchar())); u = (u << ) + (u << ) + x - '');
ungetc(x, stdin);
u *= aFlag;
return true;
} #define LL int
#define lowbit(x) (x & (-x)) typedef class IndexedTree {
public:
LL* a;
int s;
IndexedTree():a(NULL), s() { }
IndexedTree(int n):s(n) {
a = new LL[(n + )];
memset(a, , sizeof(LL) * (n + ));
} inline void add(int idx, LL val) {
for(; idx <= s; idx += lowbit(idx))
a[idx] += val;
} inline LL getSum(int idx) {
LL rt = ;
for(; idx; idx -= lowbit(idx))
rt += a[idx];
return rt;
}
}IndexedTree; typedef class SegmentableIndexedTree {
public:
IndexedTree s; // sum
IndexedTree us; // unique sum
stack< pair<int, int> > bs; SegmentableIndexedTree() { }
SegmentableIndexedTree(int n) {
s = IndexedTree(n + );
us = IndexedTree(n + );
} inline void add(int l, int r, LL val) {
s.add(l, val), us.add(l, l * val);
s.add(r + , -val), us.add(r + , -(r + ) * val);
if(val > )
bs.push(pair<int, int>(l, r));
} inline LL getSum(int idx) {
return (idx + ) * s.getSum(idx) - us.getSum(idx);
} inline LL getSum(int l, int r) {
return getSum(r) - getSum(l - );
} inline void reset() {
pair<int, int> p;
while(!bs.empty()) {
p = bs.top();
bs.pop();
add(p.first, p.second, -);
}
}
}SegmentableIndexedTree; const int maxn = 1e5 + ;
int n, m;
vector<int> g[maxn]; inline void init() {
readInteger(n);
readInteger(m);
for(int i = , x; i < n; i++) {
readInteger(x);
g[x].push_back(i + );
}
} int cnt = ;
int visitID[maxn], size[maxn], fa[maxn];
int zson[maxn], dep[maxn], top[maxn];
SegmentableIndexedTree st; void dfs1(int node, int last) {
dep[node] = dep[last] + ;
size[node] = ;
fa[node] = last;
int maxs = , maxid = ;
for(int i = ; i < (signed)g[node].size(); i++) {
int& e = g[node][i];
if(e == last) continue;
dfs1(e, node);
size[node] += size[e];
if(size[e] > maxs) maxs = size[e], maxid = e;//if(siz[son[u]] < siz[v]) son[u] = v;
}
zson[node] = maxid;
} void dfs2(int node, int last, boolean iszson) {
top[node] = (iszson) ? (top[last]) : (node);
visitID[node] = ++cnt;
if(zson[node] != ) dfs2(zson[node], node, true);
for(int i = ; i < (signed)g[node].size(); i++) {
int& e = g[node][i];
if(e == last || e == zson[node]) continue;
dfs2(e, node, false);
}
} void update(int a, int b){
// cout << st.query(st.root, 1, n, 1, n) << endl;
while(top[a] != top[b]){
int& d = (dep[top[a]] > dep[top[b]]) ? (a) : (b);
st.add(visitID[top[d]], visitID[d], );
d = fa[top[d]];
}
if(visitID[a] > visitID[b]) swap(a, b);
st.add(visitID[a], visitID[b], );
// cout << st.query(st.root, 1, n, 1, n) << endl;
} int query(int a, int b) {
int ret = ;
while(top[a] != top[b]){
int& d = (dep[top[a]] > dep[top[b]]) ? (a) : (b);
ret += st.getSum(visitID[top[d]], visitID[d]);
d = fa[top[d]];
}
if(visitID[a] > visitID[b]) swap(a, b);
ret += st.getSum(visitID[a], visitID[b]);
// cout << ret << endl;
// cout << "============================" << endl;
return ret;
} int query(int s, int t, int f) {
st.reset();
update(s, f);
return query(t, f);
} inline void solve() {
int a, b, c;
dfs1(, );
dfs2(, , false);
st = SegmentableIndexedTree(n);
// for(int i = 1; i <= n; i++)
// cout << zson[i] << " ";
while(m--) {
readInteger(a);
readInteger(b);
readInteger(c);
int res = query(a, b, c);
smax(res, query(b, a, c));
smax(res, query(b, c, a));
smax(res, query(b, a, c));
smax(res, query(c, a, b));
smax(res, query(c, b, a));
printf("%d\n", res);
}
} int main() {
init();
solve();
return ;
}
 /**
* Codeforces
* Problem#832D
* Accepted
* Time:1466ms
* Memory:14200k
*/
#include <iostream>
#include <cstdio>
#include <ctime>
#include <cmath>
#include <cctype>
#include <cstring>
#include <cstdlib>
#include <fstream>
#include <sstream>
#include <algorithm>
#include <map>
#include <set>
#include <stack>
#include <queue>
#include <vector>
#include <stack>
#ifndef WIN32
#define Auto "%lld"
#else
#define Auto "%I64d"
#endif
using namespace std;
typedef bool boolean;
const signed int inf = (signed)((1u << ) - );
const signed long long llf = (signed long long)((1ull << ) - );
const double eps = 1e-;
const int binary_limit = ;
#define smin(a, b) a = min(a, b)
#define smax(a, b) a = max(a, b)
#define max3(a, b, c) max(a, max(b, c))
#define min3(a, b, c) min(a, min(b, c))
template<typename T>
inline boolean readInteger(T& u){
char x;
int aFlag = ;
while(!isdigit((x = getchar())) && x != '-' && x != -);
if(x == -) {
ungetc(x, stdin);
return false;
}
if(x == '-'){
x = getchar();
aFlag = -;
}
for(u = x - ''; isdigit((x = getchar())); u = (u << ) + (u << ) + x - '');
ungetc(x, stdin);
u *= aFlag;
return true;
} #define LL int
#define lowbit(x) (x & (-x)) typedef class IndexedTree {
public:
LL* a;
int s;
IndexedTree():a(NULL), s() { }
IndexedTree(int n):s(n) {
a = new LL[(n + )];
memset(a, , sizeof(LL) * (n + ));
} inline void add(int idx, LL val) {
for(; idx <= s; idx += lowbit(idx))
a[idx] += val;
} inline LL getSum(int idx) {
LL rt = ;
for(; idx; idx -= lowbit(idx))
rt += a[idx];
return rt;
}
}IndexedTree; typedef class SegmentableIndexedTree {
public:
IndexedTree s; // sum
IndexedTree us; // unique sum
stack< pair<int, int> > bs; SegmentableIndexedTree() { }
SegmentableIndexedTree(int n) {
s = IndexedTree(n + );
us = IndexedTree(n + );
} inline void add(int l, int r, LL val) {
s.add(l, val), us.add(l, l * val);
s.add(r + , -val), us.add(r + , -(r + ) * val);
if(val > )
bs.push(pair<int, int>(l, r));
} inline LL getSum(int idx) {
return (idx + ) * s.getSum(idx) - us.getSum(idx);
} inline LL getSum(int l, int r) {
return getSum(r) - getSum(l - );
} inline void reset() {
pair<int, int> p;
while(!bs.empty()) {
p = bs.top();
bs.pop();
add(p.first, p.second, -);
}
}
}SegmentableIndexedTree; const int maxn = 1e5 + ;
int n, m;
vector<int> g[maxn]; inline void init() {
readInteger(n);
readInteger(m);
for(int i = , x; i < n; i++) {
readInteger(x);
g[x].push_back(i + );
}
} int cnt = ;
int visitID[maxn], size[maxn], fa[maxn];
int zson[maxn], dep[maxn], top[maxn];
SegmentableIndexedTree st; void dfs1(int node, int last) {
dep[node] = dep[last] + ;
size[node] = ;
fa[node] = last;
int maxs = , maxid = ;
for(int i = ; i < (signed)g[node].size(); i++) {
int& e = g[node][i];
if(e == last) continue;
dfs1(e, node);
size[node] += size[e];
if(size[e] > maxs) maxs = size[e], maxid = e;//if(siz[son[u]] < siz[v]) son[u] = v;
}
zson[node] = maxid;
} void dfs2(int node, int last, boolean iszson) {
top[node] = (iszson) ? (top[last]) : (node);
visitID[node] = ++cnt;
if(zson[node] != ) dfs2(zson[node], node, true);
for(int i = ; i < (signed)g[node].size(); i++) {
int& e = g[node][i];
if(e == last || e == zson[node]) continue;
dfs2(e, node, false);
}
} void update(int a, int b){
// cout << st.query(st.root, 1, n, 1, n) << endl;
while(top[a] != top[b]){
int& d = (dep[top[a]] > dep[top[b]]) ? (a) : (b);
st.add(visitID[top[d]], visitID[d], );
d = fa[top[d]];
}
if(visitID[a] > visitID[b]) swap(a, b);
st.add(visitID[a], visitID[b], );
// cout << st.query(st.root, 1, n, 1, n) << endl;
} int query(int a, int b) {
int ret = ;
while(top[a] != top[b]){
int& d = (dep[top[a]] > dep[top[b]]) ? (a) : (b);
ret += st.getSum(visitID[top[d]], visitID[d]);
d = fa[top[d]];
}
if(visitID[a] > visitID[b]) swap(a, b);
ret += st.getSum(visitID[a], visitID[b]);
// cout << ret << endl;
// cout << "============================" << endl;
return ret;
} int query(int s, int t, int f) {
st.reset();
update(s, f);
return query(t, f);
} inline void solve() {
int a, b, c;
dfs1(, );
dfs2(, , false);
st = SegmentableIndexedTree(n);
// for(int i = 1; i <= n; i++)
// cout << zson[i] << " ";
while(m--) {
readInteger(a);
readInteger(b);
readInteger(c);
int res = query(a, b, c);
smax(res, query(b, a, c));
smax(res, query(b, c, a));
smax(res, query(b, a, c));
smax(res, query(c, a, b));
smax(res, query(c, b, a));
printf("%d\n", res);
}
} int main() {
init();
solve();
return ;
}

Codeforces Round #425 (Div. 2) Problem D Misha, Grisha and Underground (Codeforces 832D) - 树链剖分 - 树状数组的更多相关文章

  1. 【 Codeforces Round #425 (Div. 2) D】Misha, Grisha and Underground

    [Link]:http://codeforces.com/contest/832/problem/D [Description] 给你一棵树; 然后给你3个点 让你把这3个点和点s,t,f对应; 然后 ...

  2. Codeforces Round #425 (Div. 2) Problem C Strange Radiation (Codeforces 832C) - 二分答案 - 数论

    n people are standing on a coordinate axis in points with positive integer coordinates strictly less ...

  3. Codeforces Round #425 (Div. 2) Problem B Petya and Exam (Codeforces 832B) - 暴力

    It's hard times now. Today Petya needs to score 100 points on Informatics exam. The tasks seem easy ...

  4. Codeforces Round #425 (Div. 2) Problem A Sasha and Sticks (Codeforces 832A)

    It's one more school day now. Sasha doesn't like classes and is always bored at them. So, each day h ...

  5. Codeforces Round #425 (Div. 2) Misha, Grisha and Underground(LCA)

    Misha, Grisha and Underground time limit per test 2 seconds memory limit per test 256 megabytes inpu ...

  6. Codeforces Round #716 (Div. 2), problem: (B) AND 0, Sum Big位运算思维

    & -- 位运算之一,有0则0 原题链接 Problem - 1514B - Codeforces 题目 Example input 2 2 2 100000 20 output 4 2267 ...

  7. Codeforces Round #425 (Div. 2))——A题&&B题&&D题

    A. Sasha and Sticks 题目链接:http://codeforces.com/contest/832/problem/A 题目意思:n个棍,双方每次取k个,取得多次数的人获胜,Sash ...

  8. Codeforces Round #425 (Div. 2) - D

    题目链接:http://codeforces.com/contest/832/problem/D 题意:给定一棵n个点的树,然后给你q个询问,每个询问为三元组(a,b,c),问你从这三个点中选取一个作 ...

  9. Codeforces Round #425 (Div. 2) D.Misha, Grisha and Underground

    我奇特的脑回路的做法就是 树链剖分 + 树状数组 树状数组是那种 区间修改,区间求和,还有回溯的 当我看到别人写的是lca,直接讨论时,感觉自己的智商收到了碾压... #include<cmat ...

随机推荐

  1. selenium元素单击不稳定解决方法

    selenium自动化测试过程中,经常会发现某一元素单击,很不稳定,有时候执行了点击没有反映. 以下总结两种解决方法:都是通过js注入的方式去点击. 1.F12查一看,要点击的按钮,或连接,有没有on ...

  2. java基础(一) -语法

    基本语法 编写Java程序时,应注意以下几点: 大小写敏感:Java是大小写敏感的,这就意味着标识符Hello与hello是不同的. 类名:对于所有的类来说,类名的首字母应该大写.如果类名由若干单词组 ...

  3. Lua class

    local _class = {} function class(super) local class_type = {} class_type.ctor = false class_type.sup ...

  4. MYSQL5.6.X 非在线安装版(解压版)安装过程

    一.卸载以前旧版本(本人5.5版本) 1.关闭MySQL服务 以管理员身份运行cmd,执行以下命令: net stop mysql 或者右键我的电脑,在管理——服务——停止MySQL 2.卸载MySQ ...

  5. java中的锁之Lock接口与Condition接口

    一.Lock源码. 1.是一个接口.一共有6个方法. 2.方法详细如下: (1)当前线程尝试获取锁.结果分两种情况,一是成功获取到锁,则返回:二是获取锁失败,则一直等待.不响应中断请求. (2)当前线 ...

  6. Python并发编程之线程池/进程池--concurrent.futures模块

    一.关于concurrent.futures模块 Python标准库为我们提供了threading和multiprocessing模块编写相应的多线程/多进程代码,但是当项目达到一定的规模,频繁创建/ ...

  7. QWidget 的 close 与 Qt::WA_DeleteOnClose

    [1]close 与 Qt::WA_DeleteOnClose简介 1.1 Qt源码 /*! Closes this widget. Returns \c true if the widget was ...

  8. python 内置函数enumerate()

    enumerate() 函数用于将一个可遍历的数据对象(如列表.元组或字符串)组合为一个索引序列,同时列出数据和数据下标,一般用在 for 循环当中.在python 3中返回一个生成器,代码如下: a ...

  9. 机器学习-LR推导及与SVM的区别

    之前整理过一篇关于逻辑回归的帖子,但是只是简单介绍了一下了LR的基本思想,面试的时候基本用不上,那么这篇帖子就深入理解一下LR的一些知识,希望能够对面试有一定的帮助. 1.逻辑斯谛分布 介绍逻辑斯谛回 ...

  10. linux常用命令:touch 命令

    linux的touch命令不常用,一般在使用make的时候可能会用到,用来修改文件时间戳,或者新建一个不存在的文件. 1.命令格式: touch [选项]... 文件... 2.命令参数: -a    ...