結果

問題 No.2163 LCA Sum Query
ユーザー 👑 p-adicp-adic
提出日時 2022-12-17 10:34:17
言語 C++17
(gcc 12.3.0 + boost 1.83.0)
結果
WA  
実行時間 -
コード長 18,668 bytes
コンパイル時間 4,669 ms
コンパイル使用メモリ 268,100 KB
実行使用メモリ 40,204 KB
最終ジャッジ日時 2024-11-16 19:10:26
合計ジャッジ時間 169,023 ms
ジャッジサーバーID
(参考情報)
judge1 / judge2
このコードへのチャレンジ
(要ログイン)

テストケース

テストケース表示
入力 結果 実行時間
実行使用メモリ
testcase_00 AC 4 ms
32,608 KB
testcase_01 WA -
testcase_02 WA -
testcase_03 WA -
testcase_04 WA -
testcase_05 WA -
testcase_06 WA -
testcase_07 WA -
testcase_08 WA -
testcase_09 WA -
testcase_10 WA -
testcase_11 WA -
testcase_12 WA -
testcase_13 TLE -
testcase_14 TLE -
testcase_15 WA -
testcase_16 WA -
testcase_17 WA -
testcase_18 WA -
testcase_19 WA -
testcase_20 WA -
testcase_21 WA -
testcase_22 TLE -
testcase_23 TLE -
testcase_24 TLE -
testcase_25 TLE -
testcase_26 TLE -
testcase_27 WA -
testcase_28 TLE -
testcase_29 WA -
testcase_30 TLE -
testcase_31 TLE -
testcase_32 TLE -
testcase_33 TLE -
testcase_34 TLE -
testcase_35 WA -
testcase_36 TLE -
testcase_37 WA -
testcase_38 TLE -
testcase_39 TLE -
testcase_40 TLE -
testcase_41 TLE -
権限があれば一括ダウンロードができます
コンパイルメッセージ
In file included from /home/linuxbrew/.linuxbrew/Cellar/gcc@12/12.3.0/include/c++/12/x86_64-pc-linux-gnu/bits/c++allocator.h:33,
                 from /home/linuxbrew/.linuxbrew/Cellar/gcc@12/12.3.0/include/c++/12/bits/allocator.h:46,
                 from /home/linuxbrew/.linuxbrew/Cellar/gcc@12/12.3.0/include/c++/12/string:41,
                 from /home/linuxbrew/.linuxbrew/Cellar/gcc@12/12.3.0/include/c++/12/bits/locale_classes.h:40,
                 from /home/linuxbrew/.linuxbrew/Cellar/gcc@12/12.3.0/include/c++/12/bits/ios_base.h:41,
                 from /home/linuxbrew/.linuxbrew/Cellar/gcc@12/12.3.0/include/c++/12/ios:42,
                 from /home/linuxbrew/.linuxbrew/Cellar/gcc@12/12.3.0/include/c++/12/istream:38,
                 from /home/linuxbrew/.linuxbrew/Cellar/gcc@12/12.3.0/include/c++/12/sstream:38,
                 from /home/linuxbrew/.linuxbrew/Cellar/gcc@12/12.3.0/include/c++/12/complex:45,
                 from /home/linuxbrew/.linuxbrew/Cellar/gcc@12/12.3.0/include/c++/12/ccomplex:39,
                 from /home/linuxbrew/.linuxbrew/Cellar/gcc@12/12.3.0/include/c++/12/x86_64-pc-linux-gnu/bits/stdc++.h:54,
                 from main.cpp:4:
In member function 'void std::__new_allocator<_Tp>::construct(_Up*, _Args&& ...) [with _Up = int; _Args = {const int&}; _Tp = std::_List_node<int>]',
    inlined from 'static void std::allocator_traits<std::allocator<_CharT> >::construct(allocator_type&, _Up*, _Args&& ...) [with _Up = int; _Args = {const int&}; _Tp = std::_List_node<int>]' at /home/linuxbrew/.linuxbrew/Cellar/gcc@12/12.3.0/include/c++/12/bits/alloc_traits.h:516:17,
    inlined from 'std::__cxx11::list<_Tp, _Alloc>::_Node* std::__cxx11::list<_Tp, _Alloc>::_M_create_node(_Args&& ...) [with _Args = {const int&}; _Tp = int; _Alloc = std::allocator<int>]' at /home/linuxbrew/.linuxbrew/Cellar/gcc@12/12.3.0/include/c++/12/bits/stl_list.h:713:33,
    inlined from 'void std::__cxx11::list<_Tp, _Alloc>::_M_insert(iterator, _Args&& ...) [with _Arg

ソースコード

diff #

// #define _GLIBCXX_DEBUG 
#pragma GCC optimize ( "O3" )
#pragma GCC target ( "avx" )
#include <bits/stdc++.h>
using namespace std;

using uint = unsigned int;
using ll = long long;

#define TYPE_OF( VAR ) remove_const<remove_reference<decltype( VAR )>::type >::type
#define UNTIE ios_base::sync_with_stdio( false ); cin.tie( nullptr ) 
#define CEXPR( LL , BOUND , VALUE ) constexpr const LL BOUND = VALUE 
#define CIN( LL , A ) LL A; cin >> A 
#define ASSERT( A , MIN , MAX ) assert( MIN <= A && A <= MAX ) 
#define CIN_ASSERT( A , MIN , MAX ) CIN( TYPE_OF( MAX ) , A ); ASSERT( A , MIN , MAX ) 
#define GETLINE( A ) string A; getline( cin , A ) 
#define GETLINE_SEPARATE( A , SEPARATOR ) string A; getline( cin , A , SEPARATOR ) 
#define FOR( VAR , INITIAL , FINAL_PLUS_ONE ) for( TYPE_OF( FINAL_PLUS_ONE ) VAR = INITIAL ; VAR < FINAL_PLUS_ONE ; VAR ++ ) 
#define FOREQ( VAR , INITIAL , FINAL ) for( TYPE_OF( FINAL ) VAR = INITIAL ; VAR <= FINAL ; VAR ++ ) 
#define FOREQINV( VAR , INITIAL , FINAL ) for( TYPE_OF( INITIAL ) VAR = INITIAL ; VAR >= FINAL ; VAR -- ) 
#define FOR_ITR( ARRAY , ITR , END ) for( auto ITR = ARRAY .begin() , END = ARRAY .end() ; ITR != END ; ITR ++ ) 
#define REPEAT( HOW_MANY_TIMES ) FOR( VARIABLE_FOR_REPEAT , 0 , HOW_MANY_TIMES ) 
#define QUIT return 0 
// #define COUT( ANSWER ) cout << ( ANSWER ) << "\n"; 
#define COUT( ANSWER ) cout << ( ANSWER ) << endl; 
#define RETURN( ANSWER ) COUT( ANSWER ); QUIT 
#define DOUBLE( PRECISION , ANSWER ) cout << fixed << setprecision( PRECISION ) << ( ANSWER ) << "\n"; QUIT 

#define POWER( ANSWER , ARGUMENT , EXPONENT )				\
  TYPE_OF( ARGUMENT ) ANSWER{ 1 };					\
  {									\
    TYPE_OF( ARGUMENT ) ARGUMENT_FOR_SQUARE_FOR_POWER = ( ARGUMENT );	\
    TYPE_OF( EXPONENT ) EXPONENT_FOR_SQUARE_FOR_POWER = ( EXPONENT );	\
    while( EXPONENT_FOR_SQUARE_FOR_POWER != 0 ){			\
      if( EXPONENT_FOR_SQUARE_FOR_POWER % 2 == 1 ){			\
	ANSWER *= ARGUMENT_FOR_SQUARE_FOR_POWER;			\
      }									\
      ARGUMENT_FOR_SQUARE_FOR_POWER *= ARGUMENT_FOR_SQUARE_FOR_POWER;	\
      EXPONENT_FOR_SQUARE_FOR_POWER /= 2;				\
    }									\
  }									\


#define POWER_MOD( ANSWER , ARGUMENT , EXPONENT , MODULO )		\
  TYPE_OF( ARGUMENT ) ANSWER{ 1 };					\
  {									\
    TYPE_OF( ARGUMENT ) ARGUMENT_FOR_SQUARE_FOR_POWER = ( MODULO + ( ARGUMENT ) % MODULO ) % MODULO; \
    TYPE_OF( EXPONENT ) EXPONENT_FOR_SQUARE_FOR_POWER = ( EXPONENT );	\
    while( EXPONENT_FOR_SQUARE_FOR_POWER != 0 ){			\
      if( EXPONENT_FOR_SQUARE_FOR_POWER % 2 == 1 ){			\
	ANSWER = ( ANSWER * ARGUMENT_FOR_SQUARE_FOR_POWER ) % MODULO;	\
      }									\
      ARGUMENT_FOR_SQUARE_FOR_POWER = ( ARGUMENT_FOR_SQUARE_FOR_POWER * ARGUMENT_FOR_SQUARE_FOR_POWER ) % MODULO; \
      EXPONENT_FOR_SQUARE_FOR_POWER /= 2;				\
    }									\
  }									\


#define FACTORIAL_MOD( ANSWER , ANSWER_INV , MAX_I , LENGTH , MODULO )	\
  ll ANSWER[LENGTH];							\
  ll ANSWER_INV[LENGTH];						\
  {									\
    ll VARIABLE_FOR_PRODUCT_FOR_FACTORIAL = 1;				\
    ANSWER[0] = VARIABLE_FOR_PRODUCT_FOR_FACTORIAL;			\
    FOREQ( i , 1 , MAX_I ){						\
      ANSWER[i] = ( VARIABLE_FOR_PRODUCT_FOR_FACTORIAL *= i ) %= MODULO; \
    }									\
    POWER_MOD( FACTORIAL_MAX_INV , ANSWER[MAX_I] , MODULO - 2 , MODULO ); \
    ANSWER_INV[MAX_I] = FACTORIAL_MAX_INV;				\
    FOREQINV( i , MAX_I - 1 , 0 ){					\
      ANSWER_INV[i] = ( FACTORIAL_MAX_INV *= i + 1 ) %= MODULO;		\
    }									\
  }									\
									\

// 通常の二分探索
#define BS( ANSWER , MINIMUM , MAXIMUM , EXPRESSION , TARGET )		\
  ll ANSWER = MAXIMUM;							\
  {									\
    ll VARIABLE_FOR_BINARY_SEARCH_L = MINIMUM;				\
    ll VARIABLE_FOR_BINARY_SEARCH_U = ANSWER;				\
    ll VARIABLE_FOR_DIFFERENCE_FOR_BINARY_SEARCH = ( TARGET ) - ( EXPRESSION ); \
    if( VARIABLE_FOR_DIFFERENCE_FOR_BINARY_SEARCH == 0 ){		\
      VARIABLE_FOR_BINARY_SEARCH_L = ANSWER;				\
    } else {								\
      ANSWER = ( VARIABLE_FOR_BINARY_SEARCH_L + VARIABLE_FOR_BINARY_SEARCH_U ) / 2; \
    }									\
    while( VARIABLE_FOR_BINARY_SEARCH_L != ANSWER ){			\
      VARIABLE_FOR_DIFFERENCE_FOR_BINARY_SEARCH = ( TARGET ) - ( EXPRESSION ); \
      if( VARIABLE_FOR_DIFFERENCE_FOR_BINARY_SEARCH == 0 ){		\
	VARIABLE_FOR_BINARY_SEARCH_L = ANSWER;				\
	break;								\
      } else {								\
	if( VARIABLE_FOR_DIFFERENCE_FOR_BINARY_SEARCH > 0 ){		\
	  VARIABLE_FOR_BINARY_SEARCH_L = ANSWER;			\
	} else {							\
	  VARIABLE_FOR_BINARY_SEARCH_U = ANSWER;			\
	}								\
	ANSWER = ( VARIABLE_FOR_BINARY_SEARCH_L + VARIABLE_FOR_BINARY_SEARCH_U ) / 2; \
      }									\
    }									\
  }									\
									\


// 二進法の二分探索
#define BS2( ANSWER , MINIMUM , MAXIMUM , EXPRESSION , TARGET )		\
  ll ANSWER = MINIMUM;							\
  {									\
    ll VARIABLE_FOR_POWER_FOR_BINARY_SEARCH_2 = 1;			\
    ll VARIABLE_FOR_DIFFERENCE_FOR_BINARY_SEARCH = ( MAXIMUM ) - ANSWER; \
    while( VARIABLE_FOR_POWER_FOR_BINARY_SEARCH_2 <= VARIABLE_FOR_DIFFERENCE_FOR_BINARY_SEARCH ){ \
      VARIABLE_FOR_POWER_FOR_BINARY_SEARCH_2 *= 2;			\
    }									\
    VARIABLE_FOR_POWER_FOR_BINARY_SEARCH_2 /= 2;			\
    ll VARIABLE_FOR_ANSWER_FOR_BINARY_SEARCH_2 = ANSWER;		\
    while( VARIABLE_FOR_POWER_FOR_BINARY_SEARCH_2 != 0 ){		\
      ANSWER = VARIABLE_FOR_ANSWER_FOR_BINARY_SEARCH_2 + VARIABLE_FOR_POWER_FOR_BINARY_SEARCH_2; \
      VARIABLE_FOR_DIFFERENCE_FOR_BINARY_SEARCH = ( TARGET ) - ( EXPRESSION ); \
      if( VARIABLE_FOR_DIFFERENCE_FOR_BINARY_SEARCH == 0 ){		\
	VARIABLE_FOR_ANSWER_FOR_BINARY_SEARCH_2 = ANSWER;		\
	break;								\
      } else if( VARIABLE_FOR_DIFFERENCE_FOR_BINARY_SEARCH > 0 ){	\
	VARIABLE_FOR_ANSWER_FOR_BINARY_SEARCH_2 = ANSWER;		\
      }									\
      VARIABLE_FOR_POWER_FOR_BINARY_SEARCH_2 /= 2;			\
    }									\
    ANSWER = VARIABLE_FOR_ANSWER_FOR_BINARY_SEARCH_2;			\
  }									\
									\


template <typename T> inline T Absolute( const T& a ){ return a > 0 ? a : - a; }
template <typename T> inline T Residue( const T& a , const T& p ){ return a >= 0 ? a % p : p - ( - a - 1 ) % p - 1; }


#define first1 first.first
#define second1 first.second
#define third1 second
#define first2 first1
#define second2 second1
#define third2 second.first
#define fourth2 second.second

inline CEXPR( int , bound_N , 50000 );
// lazy_segtree用のデータ1
using S1_12 = pair<ll,ll>;
using S1 = pair<S1_12,ll>;
inline S1 op1( S1 a , S1 b ) { return S1( S1_12( a.first1 + b.first1 , a.second1 + b.second1 ) , a.third1 + b.third1 ); }
inline S1 unit1() { return S1(); }
using F1 = int;
// vが(A,Ax,Ax^2)の形のベクトルの時にfvが(A,A(x+f),A(x+f)^2)=(A,Ax+fA,Ax^2+2fAx+f^2A)となるようなモノイド作用
inline S1 act1( F1 f , S1 v ) { return S1( S1_12( v.first1 , v.second1 + f * v.first1 ) , v.third1 + ( f * 2 ) * v.second1 + ( f * f ) * v.first1 ); }
inline F1 comp1( F1 f1 , F1 f2 ) { return f1 + f2; }
inline F1 id1() { return 0; }

// lazy_segtree用のデータ2
using S2_12 = pair<ll,ll>;
using S2_34 = pair<ll,ll>;
using S2 = pair<S2_12,S2_34>;
inline S2 op2( S2 a , S2 b ) { return S2( S2_12( a.first2 + b.first2 , a.second2 + b.second2 ) , S2_34( a.third2 + b.third2 , a.fourth2 + b.fourth2 ) ); }
inline S2 unit2() { return S2(); }
using F2 = pair<int,int>;
// vが(A,Ax,Ay,Axy)の形のベクトルの時に(f_1,f_2)vが(A,A(x+f_1),A(y+f_2),A(x+f_1)(y+f_2))=(A,Ax+f_1A,Ay+f_2A,Axy+f_1Ay+f_2Ax+f_1f_2A)となるようなモノイド作用
inline S2 act2( F2 f , S2 v ) { return S2( S2_12( v.first2 , v.second2 + f.first * v.first2 ) , S2_34( v.third2 + f.second * v.first2 , v.fourth2 + f.first * v.third2 + f.second * v.second2 + ( f.first * f.second ) * v.first2 ) ); }
inline F2 comp2( F2 f1 , F2 f2 ) { return F2( f1.first + f2.first , f1.second + f2.second ); }
inline F2 id2() { return F2(); }

// lazy_segtree用のデータ3
using S3 = ll;
inline S3 op3( S3 a , S3 b ) { return a + b; }
inline S3 unit3() { return 0; }
using F3 = int;
inline S3 act3( F3 f , S3 v ) { return f + v; }
inline F3 comp3( F3 f1 , F3 f2 ) { return f1 + f2; }
inline F3 id3() { return 0; }

using Interval = pair<int,int>;

#include <atcoder/lazysegtree>
using namespace atcoder;


// Xの総和を計算するための、部分木におけるS内の総和の計算
ll sum_v3( int v , list<int> ( &children )[bound_N+1] , int ( &HL_inv )[bound_N+1] , int ( &leftmost_descendant )[bound_N] , lazy_segtree<S3,op3,unit3,F3,act3,comp3,id3>& v3 )
{
  int& HL_inv_v = HL_inv[v];
  ll answer = v3.prod( leftmost_descendant[HL_inv_v] , HL_inv_v + 1 );
  list<int>* p_children_v = &( children[v] );
  while( ! p_children_v->empty() ){
    auto itr_v = p_children_v->begin() , end_v = p_children_v->end();
    v = *itr_v;
    p_children_v = &( children[v] );
    while( ++itr_v != end_v ){
      answer += sum_v3( *itr_v , children , HL_inv , leftmost_descendant , v3 );
    }
  }
  return answer;
}

// 解説におけるf(1,v)の表示の第1項の計算
ll sum_v1( int v , list<int> ( &children )[bound_N+1], int ( &HL_inv )[bound_N+1] , int ( &leftmost_descendant )[bound_N] , lazy_segtree<S1,op1,unit1,F1,act1,comp1,id1>& v1 )
{
  int& HL_inv_v = HL_inv[v];
  // 解説ではvの親に言及しているため暗にv != 1を課しているが、ここではrootの親を便宜上0とすることで場合分けを回避
  ll answer = v1.prod( leftmost_descendant[HL_inv_v] , HL_inv_v + 1 ).third1;
  list<int>* p_children_v = &( children[v] );
  while( ! p_children_v->empty() ){
    auto itr_v = p_children_v->begin() , end_v = p_children_v->end();
    v = *itr_v;
    p_children_v = &( children[v] );
    while( ++itr_v != end_v ){
      answer += sum_v1( *itr_v , children , HL_inv , leftmost_descendant , v1 );
    }
  }
  return answer;
}

// 解説におけるf(1,v)の計算
inline ll f1v( int v , list<int> ( &children )[bound_N+1] , int ( &parent )[bound_N+1] , int ( &HL_inv )[bound_N+1] , int ( &leftmost_descendant )[bound_N] , lazy_segtree<S1,op1,unit1,F1,act1,comp1,id1>& v1 , lazy_segtree<S3,op3,unit3,F3,act3,comp3,id3>& v3 )
{
  list<int>& children_v = children[v];
  int& HL_inv_v = HL_inv[v];
  ll answer = sum_v1( v , children , HL_inv , leftmost_descendant , v1 );
  // 解説におけるXの総和は(Ri,Vi)を(1,v)に置き換えてXを定義し直した時の値であることに注意。
  // 特に(Ri,Vi)=(v,v)!=(1,1)の時のf(v,v)(v!=1)の計算で呼び出されるf(1,1)の計算で参照されるXは問題文中で定義された集合とは別物であり(Ri,Vi)=(1,1)に置き換えて定義し直したものとなる。
  answer -= sum_v3( v , children , HL_inv , leftmost_descendant , v3 );
  int& parent_v = parent[v];
  answer += ( v1.get( HL_inv_v ).third1 / ( v - parent_v ) ) * parent_v;
  answer /= 2;
  return answer;
}

// 解説におけるf(v,v)の計算
inline ll fvv( int v , list<int> ( &children )[bound_N+1] , int ( &parent )[bound_N+1] , int ( &HL_inv )[bound_N+1] , int ( &leftmost_descendant )[bound_N] , list<Interval> ( &path )[bound_N] , lazy_segtree<S1,op1,unit1,F1,act1,comp1,id1>& v1 , lazy_segtree<S2,op2,unit2,F2,act2,comp2,id2>& v2 , lazy_segtree<S3,op3,unit3,F3,act3,comp3,id3>& v3 )
{
  ll answer = f1v( 1 , children , parent , HL_inv , leftmost_descendant , v1 , v3 );
  list<Interval>& path_v = path[HL_inv[v]];
  FOR_ITR( path_v , itr_v , end_v ){
    // |S|-c(root,root) = 0であるので本来不要なrootの寄与は自動的に消える
    answer += v2.prod( itr_v->first , itr_v->second + 1 ).fourth2;
  }
  return answer;
}


int main()
{
  UNTIE;
  CIN_ASSERT( N , 1 , bound_N );
  CEXPR( int , bound_Q , 100000 );
  CIN_ASSERT( Q , 1 , bound_Q );
  // 一旦子と親を全て同じ配列に格納
  list<int> children[bound_N+1] = {};
  FOR( i , 1 , N ){
    CIN_ASSERT( Ai , 1 , N );
    CIN_ASSERT( Bi , 1 , N );
    assert( Ai != Bi );
    children[Ai].push_back( Bi );
    children[Bi].push_back( Ai );
  }
  // 子と親の決定
  int parent[bound_N+1];
  parent[1] = 0;
  list<int> layer[bound_N] = {};
  int children_query[bound_N];
  children_query[0] = 1;
  int start_children_query = 0;
  int size_children_query = 1;
  int size_children_query_copy = size_children_query;
  int depth = 0;
  while( start_children_query < size_children_query ){
    list<int>& layer_curr = layer[depth];
    FOR( i , start_children_query , size_children_query ){
      int& children_query_i = children_query[i];
      layer_curr.push_back( children_query_i );
      list<int>& children_i = children[children_query_i];
      FOR_ITR( children_i , itr_i , end_i ){
	int& j = *itr_i;
	parent[j] = children_query_i;
	list<int>& children_ij = children[j];
	FOR_ITR( children_ij , itr_ij , end_ij ){
	  if( children_query_i == *itr_ij ){
	    children_ij.erase( itr_ij );
	    break;
	  }
	}
	children_query[size_children_query_copy++] = j;
      }
    }
    depth++;
    start_children_query = size_children_query;
    size_children_query = size_children_query_copy;
  }
  // HL分解のための重さ計算
  int weight[bound_N + 1];
  FOREQINV( d , depth - 1 , 0 ){
    list<int>& layer_curr = layer[d];
    FOR_ITR( layer_curr , itr_d , end_d ){
      int& i = *itr_d;
      list<int>& children_curr = children[i];
      int& weight_i = weight[i];
      weight_i = 1;
      FOR_ITR( children_curr , itr_i , end_i ){
	weight_i += weight[*itr_i];
      }
    }
  }
  // HL分解のためのheight計算
  int children_H_i;
  int weight_max;
  FOREQ( i , 1 , N ){
    list<int>& children_i = children[i];
    weight_max = -1;
    FOR_ITR( children_i , itr , end ){
      int& j_curr = *itr;
      int& weight_j_curr = weight[j_curr];
      if( weight_max < weight_j_curr ){
	weight_max = weight_j_curr;
	children_H_i = j_curr;
      }
    }
    FOR_ITR( children_i , itr , end ){
      if( children_H_i == *itr ){
	children_i.erase( itr );
	children_i.push_front( children_H_i );
	break;
      }
    }
  }
  // HL分解(添え字は0始まり、葉に小さい番号づけ)
  int HL[bound_N];
  int HL_inv[bound_N+1];
  list<int> vertex_stack{};
  list<int> root{ { 1 } };
  list<list<int>::iterator> children_itr_stack{};
  children_itr_stack.push_back( root.begin() );
  list<list<int>::iterator> children_end_stack{};
  children_end_stack.push_back( root.end() );
  int length = 0;
  while( ! children_itr_stack.empty() ){
    list<int>::iterator& itr_curr = children_itr_stack.back();
    list<int>::iterator& end_curr = children_end_stack.back();
    if( itr_curr == end_curr ){
      children_itr_stack.pop_back();
      children_end_stack.pop_back();
    } else {
      int& i_curr = *itr_curr;
      list<int>& children_curr = children[i_curr];
      auto itr_new = children_curr.begin() , end_new = children_curr.end();
      if( itr_new == end_new ){
	HL_inv[i_curr] = length;
	HL[length++] = i_curr;
	FOR_ITR( vertex_stack , itr_v , end_v ){
	  int& i_prev = *itr_v;
	  HL_inv[i_prev] = length;
	  HL[length++] = i_prev;
	}
	vertex_stack.clear();
      } else {
	vertex_stack.push_front( i_curr );
	children_itr_stack.push_back( itr_new );
	children_end_stack.push_back( end_new );
      }
      itr_curr++;
    }
  }
  // 1からのパス計算(ここが各ノードでlogオーダーになることがHL分解特有の事情)
  int& HL_inv_root = HL_inv[1];
  list<Interval> path[bound_N];
  path[HL_inv_root] = list<Interval>( { Interval( HL_inv_root , HL_inv_root ) } );
  FOR( d , 1 , depth ){
    list<int>& layer_curr = layer[d];
    FOR_ITR( layer_curr , itr_d , end_d ){
      int& i = *itr_d;
      int& parent_i = parent[i];
      int& HL_inv_parent_i = HL_inv[parent_i];
      int& HL_inv_i = HL_inv[i];
      list<Interval>& path_parent_i = path[HL_inv_parent_i];
      list<Interval>& path_i = path[HL_inv_i];
      path_i = path_parent_i;
      if( children[parent_i].front() == i ){
	path_i.back().first = HL_inv_i;
      } else {
	path_i.push_back( Interval( HL_inv_i , HL_inv_i ) );
      }
    }
  }
  // ノードの子孫のなすHL上の区間の左端の決定
  int leftmost_descendant[bound_N];
  FOREQINV( d , depth - 1 , 0 ){
    list<int>& layer_curr = layer[d];
    FOR_ITR( layer_curr , itr_d , end_d ){
      int& i = *itr_d;
      list<int>& children_curr = children[i];
      int& HL_inv_i = HL_inv[i];
      leftmost_descendant[HL_inv_i] = children_curr.empty() ? HL_inv_i : leftmost_descendant[HL_inv[children_curr.front()]];
    }
  }
  // 解説における(x-(xの親),(x-(xの親))c(1,x),(x-(xの親))c(1,x)^2)の管理
  lazy_segtree<S1,op1,unit1,F1,act1,comp1,id1> v1{ N };
  FOR( i , 0 , N ){
    int& HL_i = HL[i];
    v1.set( i , S1( S1_12( HL_i - parent[HL_i] , 0 ) , 0 ) );
  }
  // 解説における(x-(xの親),(x-(xの親))c(1,x),(x-(xの親))(|S|-c(1,x)),(x-(xの親))c(1,x)(|S|-c(1,x)))の管理
  lazy_segtree<S2,op2,unit2,F2,act2,comp2,id2> v2{ N };
  FOR( i , 0 , N ){
    int& HL_i = HL[i];
    v2.set( i , S2( S2_12( HL_i - parent[HL_i] , 0 ) , S2_34( 0 , 0 ) ) );
  }
  // Xの総和を計算するための(x in S ? x : 0)の管理
  lazy_segtree<S3,op3,unit3,F3,act3,comp3,id3> v3{ N };
  set<int> S{};
  int sign , x , HL_inv_x;
  bool belong;
  // クエリ処理
  REPEAT( Q ){
    CIN_ASSERT( Ui , 1 , N );
    CIN_ASSERT( Ri , 1 , N );
    CIN_ASSERT( Vi , 1 , N );
    if( S.count( Ui ) == 0 ){
      sign = 1;
      S.insert( Ui );
    } else {
      sign = -1;
      S.erase( Ui );
    }
    int& HL_inv_Ui = HL_inv[Ui];
    v3.apply( HL_inv_Ui , HL_inv_Ui + 1, sign * Ui );
    v2.apply( 0 , N , F2( 0 , sign ) );
    list<Interval>& path_Ui = path[HL_inv[Ui]];
    FOR_ITR( path_Ui , itr_i , end_i ){
      Interval& path_Uij = *itr_i;
      v1.apply( path_Uij.first , path_Uij.second + 1 , sign );
      v2.apply( path_Uij.first , path_Uij.second + 1 , F2( sign , -sign ) );
    }
    int& HL_inv_Vi = HL_inv[Vi];
    if( Ri == Vi ){
      COUT( fvv( Vi , children , parent , HL_inv , leftmost_descendant , path , v1 , v2 , v3 ) );
    } else {
      int& HL_inv_Ri = HL_inv[Ri];
      belong = false;
      list<Interval>& path_Ri = path[HL_inv_Ri];
      FOR_ITR( path_Ri , itr_Ri , end_Ri ){
	Interval& path_Rij = *itr_Ri;
	if( path_Rij.first <= HL_inv_Vi && HL_inv_Vi <= path_Rij.second ){
	  belong = true;
	  if( path_Rij.first == HL_inv_Vi ){
	    itr_Ri++;
	    HL_inv_x = itr_Ri->second;
	  } else {
	    HL_inv_x = HL_inv_Vi - 1;
	  }
	  x = HL[HL_inv_x];
	  COUT( fvv( Vi , children , parent , HL_inv , leftmost_descendant , path , v1 , v2 , v3 ) - f1v( x , children , parent , HL_inv , leftmost_descendant , v1 , v3 ) - ( v2.prod( HL_inv_x , HL_inv_x + 1 ).fourth2 / ( x - Vi ) ) * Vi );
	  break;
	}
	if( ! belong ){
	  COUT( f1v( Vi , children , parent , HL_inv , leftmost_descendant , v1 , v3 ));
	}
      }
    }
  }
  QUIT;
}

0