• [LeetCode 973.] K Closest Points to Origin


    LeetCode 973. K Closest Points to Origin

    题目描述

    Given an array of points where points[i] = [xi, yi] represents a point on the X-Y plane and an integer k, return the k closest points to the origin (0, 0).

    The distance between two points on the X-Y plane is the Euclidean distance (i.e., √((x1 - x2)^2 + (y1 - y2)^2)).

    You may return the answer in any order. The answer is guaranteed to be unique (except for the order that it is in).

    Example 1:

    Input: points = [[1,3],[-2,2]], k = 1
    Output: [[-2,2]]
    Explanation:
    The distance between (1, 3) and the origin is sqrt(10).
    The distance between (-2, 2) and the origin is sqrt(8).
    Since sqrt(8) < sqrt(10), (-2, 2) is closer to the origin.
    We only want the closest k = 1 points from the origin, so the answer is just [[-2,2]].

    Example 2:

    Input: points = [[3,3],[5,-1],[-2,4]], k = 2
    Output: [[3,3],[-2,4]]
    Explanation: The answer [[-2,4],[3,3]] would also be accepted.

    Constraints:

    • 1 <= k <= points.length <= 104
    • -104 < xi, yi < 104

    解题思路

    前k大元素和第k大元素可以视为同一类题,都是 topK 问题,经典解法有堆和快速选择算法两种。

    思路一:堆
    C++ 中有现成的 priority_queue 可用,注意默认是大根堆,并且自定义 comparator 注意写法。
    时间复杂度 O(K+NlogK),空间复杂度 O(K)。

    思路二:快速选择算法
    C++中有 nth_element 可用,同样要注意函数参数、以及 comparator 的写法。
    我们也可以选择手写快速选择算法。
    时间复杂度 O(N),空间复杂度 O(1)。

    参考代码

    这里比较的是到原点的距离,我们可以直接比较 x^2 + y^2 而不必开平方,比较结果是一样的。

    /*
     * @lc app=leetcode id=973 lang=cpp
     *
     * [973] K Closest Points to Origin
     */
    
    // @lc code=start
    class Solution {
    public:
        // 堆
        vector<vector<int>> kClosest(vector<vector<int>>& points, int k) {
            assert(1 <= k && k <= points.size());
    
            using P = vector<int>;
            auto cmp = [&](const P& a, const P& b) {
                return a[0]*a[0] + a[1]*a[1] < b[0]*b[0] + b[1]*b[1];
            };
            priority_queue<P, deque<P>, decltype(cmp)> q(cmp);
            for (auto&& p : points) {
                q.push(p);
                if (q.size() > k) {
                    q.pop();
                }
            }
            vector<vector<int>> res;
            while (!q.empty()) {
                res.push_back(q.top());
                q.pop();
            }
            return res;
        } // AC
    };
    // @lc code=end
    

    快速选择 nth_element

        // nth_element(begin, nth, end)
        vector<vector<int>> kClosest(vector<vector<int>>& points, int k) {
            assert(1 <= k && k <= points.size());
    
            using P = vector<int>;
            auto cmp = [&](const P& a, const P& b) {
                return a[0]*a[0] + a[1]*a[1] < b[0]*b[0] + b[1]*b[1];
            }; // OK
            nth_element(points.begin(), points.begin()+k, points.end(), cmp);
            // nth_element(points.begin(), points.begin()+k, points.end(), 
            //     [](auto&& a, auto&& b){
            //         return a[0]*a[0] + a[1]*a[1] < b[0]*b[0] + b[1]*b[1];
            //     }); // OK
            points.resize(k);
            return points;
        } // AC
    

    手写快速选择算法

        static inline int dist(vector<int>& p) {
            return p[0]*p[0] + p[1]*p[1];
        }
        int partation(vector<vector<int>>& points, int l, int r) {
            auto pivot = points[l];
            while (l < r) {
                while (l < r && dist(points[r]) >= dist(pivot)) r--;
                if (l < r) points[l++] = points[r];
                while (l < r && dist(points[l]) < dist(pivot)) l++;
                if (l < r) points[r--] = points[l];
            }
            points[l] = pivot;
            return l;
        }
        void find_kth(vector<vector<int>>& points, int k) {
            int l = 0, r = points.size() - 1;
            while (l < r) {
                int t = partation(points, l, r);
                if (t == k) return;
                else if (t < k) l = t + 1;
                else r = t - 1;
            }
        }
        vector<vector<int>> kClosest(vector<vector<int>>& points, int k) {
            assert(1 <= k && k <= points.size());
    
            find_kth(points, k-1);
            points.resize(k);
            return points;
        } // AC
    
  • 相关阅读:
    Win8常用快捷键
    清除远程桌面连接记录
    通过注册表改变“我的文档”等的默认位置,防止系统重装造成数据丢失
    HTML 转义字符对照表
    r语言 函数
    sparkr——报错
    R语言--saprkR基本使用
    r绘图基本
    R绘制中国地图,并展示流行病学数据
    r画饼图
  • 原文地址:https://www.cnblogs.com/zhcpku/p/14636739.html
Copyright © 2020-2023  润新知