首页 > 代码库 > UVa 1451 Average (斜率优化)

UVa 1451 Average (斜率优化)

题意:给定一个01序列,让你找出一个长度大于等于F的连续子序列使得平均值最大。

析:直接枚举肯定是O(n^3),超时,然后用前缀和来优化,O(n^2),还是太大,这个要求的平均值是 i > j (sum[i] - sum[j-1]) / (i-(j-1)),这正好就是一个斜率的表示形式,可以考虑用优化,每次在加入一个新点的时候把不成立的全部删除,然后再加入,维护一个下凸线。

#pragma comment(linker, "/STACK:1024000000,1024000000")
#include <cstdio>
#include <string>
#include <cstdlib>
#include <cmath>
#include <iostream>
#include <cstring>
#include <set>
#include <queue>
#include <algorithm>
#include <vector>
#include <map>
#include <cctype>
#include <cmath>
#include <stack>
#include <sstream>
#define debug() puts("++++");
#define gcd(a, b) __gcd(a, b)
#define lson l,m,rt<<1
#define rson m+1,r,rt<<1|1
#define freopenr freopen("in.txt", "r", stdin)
#define freopenw freopen("out.txt", "w", stdout)
using namespace std;

typedef long long LL;
typedef unsigned long long ULL;
typedef pair<int, int> P;
const int INF = 0x3f3f3f3f;
const LL LNF = 1e16;
const double inf = 0x3f3f3f3f3f3f;
const double PI = acos(-1.0);
const double eps = 1e-8;
const int maxn = 1e5 + 10;
const int mod = 1e9 + 7;
const int dr[] = {-1, 0, 1, 0};
const int dc[] = {0, 1, 0, -1};
const char *de[] = {"0000", "0001", "0010", "0011", "0100", "0101", "0110", "0111", "1000", "1001", "1010", "1011", "1100", "1101", "1110", "1111"};
int n, m;
const int mon[] = {0, 31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31};
const int monn[] = {0, 31, 29, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31};
inline bool is_in(int r, int c){
  return r >= 0 && r < n && c >= 0 && c < m;
}

int q[maxn];
char s[maxn];
int sum[maxn];

bool judge(int i, int j, int k){
  return (LL)(sum[i]-sum[j])*(i-k) >= (LL)(sum[i]-sum[k])*(i-j);
}

int main(){
  int T;  cin >> T;
  while(T--){
    scanf("%d %d", &n, &m);
    scanf("%s", s+1);
    for(int i = 1; i <= n; ++i)  sum[i] = sum[i-1] + s[i] - ‘0‘;
    int fro = 0, rear = 0;
    int ansL = 1, ansR = m;
    double ans = 0.0;
    for(int i = m; i <= n; ++i){
      while(fro + 1 < rear && judge(i-m, q[rear-1], q[rear]))  --rear;
      q[++rear] = i - m;
      while(fro + 1 < rear && judge(i, q[fro+2], q[fro+1]))  ++fro;
      double c = (double)(sum[i]-sum[q[fro+1]])/(i-q[fro+1]);
      if(ans < c || ans == c && ansR - ansL + 1> i - q[fro+1]){
        ans = c;
        ansL = q[fro+1] + 1;  ansR = i;
      }
    }
    printf("%d %d\n", ansL, ansR);
  }
  return 0;
}

  

 

简单的枚举算法可以这样描述:每次枚举一对满足条件的(a, b),即 a≤b-F+1,检查 ave(a, b),并更新当前最大值。然而这题中 N 很大,N 2 的枚举算法显然不能使用,但是能不能优化一下这个效率不高的算法呢?答案是肯定的。目标图形化首先一定会设序列 a i 的部分和:S i =a 1 +a 2 +…+a i , ,特别的定义 S 0 =0。这样可以很简洁的表示出目标函数) 1 () , (1? ??=?i jS Sj i avei j!如果将 S 函数绘在平面直角坐标系内,这就是过点 S j 和点 S i-1 直线的斜率!于是问题转化为:平面上已知 N+1 个点,P i (i, S i ),0≤i≤N,求横向距离大于等于 F 的任意两点连线的最大

UVa 1451 Average (斜率优化)