【Leetcode】1515. Best Position for a Service Centre

记录算法题解

于 2024-08-15 22:29:49 发布

阅读量224

点赞数 7

分类专栏： LC 二分、位运算与数学文章标签： leetcode 算法 c++

本文链接：https://blog.csdn.net/qq_46105170/article/details/141224692

版权

LC 二分、位运算与数学专栏收录该内容

398 篇文章 2 订阅

订阅专栏

题目地址：

https://leetcode.com/problems/best-position-for-a-service-centre/description/

给定 $n$ 个平面坐标系的坐标 $x_0,y_0),...,(x_{n-1},y_{n-1})$ ，求： $\min_{(x,y)} \sum_i\sqrt{(x_i-x)^2+(y_i-y)^2}$

令 $f(x,y)=\sum_i\sqrt{(x_i-x)^2+(y_i-y)^2}$ ，其海森矩阵为 $\begin{bmatrix} \frac{\partial^2 f}{\partial x^2} & \frac{\partial^2 f}{\partial x \partial y} \\ \frac{\partial^2 f}{\partial y \partial x} & \frac{\partial^2 f}{\partial y^2} \end{bmatrix}=\begin{bmatrix} \sum_{i} \frac{(y_i - y)^2}{\left[(x_i - x)^2 + (y_i - y)^2\right]^{3/2}} & \sum_{i} \frac{(x_i - x)(y_i - y)}{\left[(x_i - x)^2 + (y_i - y)^2\right]^{3/2}} \\ \sum_{i} \frac{(x_i - x)(y_i - y)}{\left[(x_i - x)^2 + (y_i - y)^2\right]^{3/2}} & \sum_{i} \frac{(x_i - x)^2}{\left[(x_i - x)^2 + (y_i - y)^2\right]^{3/2}} \end{bmatrix}$ 是半正定的，从而 $f$ 是凸函数，从而 $g(x)=\inf_y f(x,y)$ 对于 $x$ 来讲也是凸函数。单元凸函数求最小值可以用三分法，而二维凸函数求最小值可以用三分套三分的方法。代码如下：

class Solution {
 public:
  double getMinDistSum(vector<vector<int>>& ps) {
    constexpr double eps = 1e-8;
    double l = 0, r = 100.0;
    auto get_sum = [&](double x, double y) {
      double sum = 0.0;
      for (auto& p : ps) {
        double a = p[0], b = p[1];
        sum += sqrt((x - a) * (x - a) + (y - b) * (y - b));
      }
      return sum;
    };
    auto calc = [&](double x) -> double {
      double l = 0, r = 100;
      while (r - l > eps) {
        double y1 = l + (r - l) / 3, y2 = l + (r - l) / 3 * 2;
        if (get_sum(x, y1) >= get_sum(x, y2))
          l = y1;
        else
          r = y2;
      }
      return get_sum(x, l);
    };
    while (r - l > eps) {
      double x1 = l + (r - l) / 3, x2 = l + (r - l) / 3 * 2;
      if (calc(x1) > calc(x2))
        l = x1;
      else
        r = x2;
    }
    return calc(l);
  }
};