「分块系列」数列分块入门3 解题报告

数列分块入门3

题意概括

区间加法，区间求前驱。

写在前面

这题的方法与分块2方法极其类似，建议自行解决。

正题

和上一题类似，但是二分不是用来计数的，而是用来求小于c的最大值的。然后对于不完整快，将小于c的值求最大值，再与所有块中二分结果求最大值即可。(其他思路上一篇题解已经讲了，这里不再复述，代码注释也懒得打了，因为比较简单，很容易理解)

代码

#include<cstdio>
#include<vector>
#include<algorithm>
#include<cmath>
using namespace std;
#define MAXN 100005int n, m, a[MAXN], p[MAXN], b[1005], mm;
vector<int> v[1005];
int opt, l, r, c;int EF( vector<int> vec, int x ){int l, r, mid, ans(-1);l = 0; r = vec.size() - 1;while( l <= r ){mid = ( l + r ) >> 1;if ( vec[mid] < x ){ans = mid;l = mid + 1;}else r = mid - 1;}return ans;
}int query( int l, int r, int c ){int ans(-1);if ( p[l] == p[r] ){for ( int i = l; i <= r; ++i )if ( a[i] + b[p[l]] < c ) ans = max( ans, a[i] + b[p[l]] );return ans;}for ( int i = l; p[i] == p[l]; ++i )if ( a[i] + b[p[i]] < c ) ans = max( ans, a[i] + b[p[l]] );for ( int i = r; p[i] == p[r]; --i )if ( a[i] + b[p[i]] < c ) ans = max( ans, a[i] + b[p[r]] );for ( int i = p[l] + 1; i <= p[r] - 1; ++i ){int t(EF( v[i], c - b[i] ));if ( t >= 0 ) ans = max( ans, v[i][t] + b[i] );}return ans;
}void re( int x ){v[x].clear();int be(( x - 1 ) * m + 1);for ( int i = be; p[i] == p[be]; i++ ) v[x].push_back( a[i] );sort( v[x].begin(), v[x].end() );
}void Add( int l, int r, int c ){if ( p[l] == p[r] ){for ( int i = l; i <= r; ++i ) a[i] += c;re( p[l] ); return; }for ( int i = l; p[i] == p[l]; ++i ) a[i] += c;re(p[l]);for ( int i = r; p[i] == p[r]; --i ) a[i] += c;re(p[r]);for ( int i = p[l] + 1; i < p[r]; ++i ) b[i] += c;
}int main(){scanf( "%d", &n ); m = (int)sqrt(n);for ( int i = 1; i <= n; ++i ) p[i] = ( i - 1 ) / m + 1, mm = p[i];for ( int i = 1; i <= n; ++i ) scanf( "%d", &a[i] );for ( int i = 1; i <= n; ++i ) v[p[i]].push_back(a[i]);for ( int i = 1; i <= mm; ++i ) sort( v[i].begin(), v[i].end() );for ( int i = 1; i <= n; ++i ){scanf( "%d%d%d%d", &opt, &l, &r, &c );if ( opt ) printf( "%d\n", query( l, r, c ) );else Add( l, r, c );}return 0;
}