Skip to main content

Priority_queue - Question 3

Leetcode 407. Trapping Rain Water II

Given an m x n integer matrix heightMap representing the height of each unit cell in a 2D elevation map, return the volume of water it can trap after raining.

Constraints:

m == heightMap.length

n == heightMap[i].length

1 <= m, n <= 200

0 <= heightMap[i][j] <= 2 * 10^4


Analysis:

If you have a bucket built with pieces of woods, then the amount of water is determined by the wood with the lowest height.

So we view the two D matrix as a square bucket. 

The first step is to put all the elements in the peripheral into a reversed prority_queue (pq), which serves as the "wall" of the bucket.

Then start with the smallest elements, we can do a bfs search;

1. all the smaller elements should have contributions, the magnitude is the height diff;

2.  all the larger or equal elements become the "new walls", so keep them in the pq.

In step 1, we can do a second bfs, to make sure all the connected lower elements are counted. The higher ones becomes the "new walls" of the bucket, which are saved in the pq.

To avoid redundant visit, we can use a visit matrix to label the visit information.


See the code below:


class Solution {
public:
    int trapRainWater(vector<vector<int>>& heightMap) {
        int res = 0, m = heightMap.size(), n = heightMap.front().size();
        priority_queue<vector<int>, vector<vector<int>>, greater<vector<int>>> pq;
        vector<vector<int>> vis(m, vector<int>(n, 0));
        for(int i=0; i<m; ++i) {
            pq.push({heightMap[i][0], i, 0});
            pq.push({heightMap[i][n-1], i, n-1});
            vis[i][0] = 1;
            vis[i][n-1] = 1;
        }
        for(int i=1; i<n-1; ++i) {
            pq.push({heightMap[0][i], 0, i});
            pq.push({heightMap[m-1][i], m-1, i});
            vis[0][i] = 1;
            vis[m-1][i] = 1;
        }
        vector<int> dirs = {-1, 0, 1, 0, -1};
        while(pq.size()) {
            auto t = pq.top();
            pq.pop();
            for(int i=0; i+1<dirs.size(); ++i) {
                int x = t[1] + dirs[i], y = t[2] + dirs[i+1];
                if(x<0 || x>=m || y<0 || y>=n || vis[x][y]) continue;
                int h = heightMap[x][y];
                if(h < t[0]) {
                    queue<vector<int>> pq2;
                    pq2.push({h, x, y});
                    res += t[0] - h;
                    vis[x][y] = 1;
                    while(pq2.size()) {
                        auto t2 = pq2.front();
                        pq2.pop();
                        for(int i2=0; i2+1<dirs.size(); ++i2) {
                            int x2 = t2[1] + dirs[i2], y2 = t2[2] + dirs[i2+1];
                            if(x2<0 || x2>=m || y2<0 || y2>=n || vis[x2][y2]) continue;
                            int h2 = heightMap[x2][y2];
                            if(h2 < t[0]) {
                                res += t[0] - h2;
                                pq2.push({h2, x2, y2});
                            }
                            else pq.push({h2, x2, y2});
                            vis[x2][y2] = 1;
                        }
                    }
                    continue;
                }
                pq.push({h, x, y});
                vis[x][y] = 1;
            }
        }
        return res;
    }
};

 

[code update, 2021-12-25]

The above code is not clean enough, see the updated one below:


class Solution {
public:
    int trapRainWater(vector<vector<int>>& heightMap) {
        int res = 0, m = heightMap.size(), n = heightMap.front().size();
        priority_queue<vector<int>, vector<vector<int>>, greater<vector<int>>> pq;
        vector<vector<int>> vis(m, vector<int>(n, 0));
        for(int i=0; i<m; ++i) {
            pq.push({heightMap[i][0], i, 0});
            pq.push({heightMap[i][n-1], i, n-1});
            vis[i][0] = 1;
            vis[i][n-1] = 1;
        }
        for(int i=1; i<n-1; ++i) {
            pq.push({heightMap[0][i], 0, i});
            pq.push({heightMap[m-1][i], m-1, i});
            vis[0][i] = 1;
            vis[m-1][i] = 1;
        }
        vector<int> dirs = {-1, 0, 1, 0, -1};
        while(pq.size()) {
            auto t = pq.top();
            pq.pop();
            queue<vector<int>> q;
            q.push({t[1], t[2]});
            while(q.size()) {
                auto t2 = q.front();
                q.pop();
                for(int i=0; i+1<dirs.size(); ++i) {
                    int x = t2[0] + dirs[i], y = t2[1] + dirs[i+1];
                    if(x<0 || x>=m || y<0 || y>=n || vis[x][y]) continue;
                    int h = heightMap[x][y];
                    if(h < t[0]) {
                        res += t[0] - h;
                        q.push({x, y});
                    }
                    else pq.push({h, x, y});
                    vis[x][y] = 1;
                }
            }
        }
        return res;
    }
};



Upper Layer

Comments

Popular posts from this blog

Sweep Line

Sweep (or scanning) line algorithm is very efficient for some specific questions involving discrete intervals. The intervals could be the lasting time of events, or the width of a building or an abstract square, etc. In the scanning line algorithm, we usually need to distinguish the start and the end of an interval. After the labeling of the starts and ends, we can sort them together based on the values of the starts and ends. Thus, if there are N intervals in total, we will have 2*N data points (since each interval will contribute 2). The sorting becomes the most time-consuming step, which is O(2N*log(2N) ~ O(N*logN). After the sorting, we usually can run a linear sweep for all the data points. If the data point is labeled as a starting point, it means a new interval is in the processing; when an ending time is reached, it means one of the interval has ended. In such direct way, we can easily figure out how many intervals are in the processes. Other related information can also be obt...

Bit Manipulation - Example

  Leetcode 136 Single Number Given a non-empty array of integers nums, every element appears twice except for one. Find that single one. You must implement a solution with a linear runtime complexity and use only constant extra space. Constraints: 1 <= nums.length <= 3 * 10^4 -3 * 10^4 <= nums[i] <= 3 * 10^4 Each element in the array appears twice except for one element which appears only once. Analysis: If there is no space limitation, this question can be solved by counting easily. But counting requires additional space. Here we can use xor (^) operation based on some interesting observations:  A^A = 0, here A is any number A^0 = A, here A is any number Since all the number appears twice except one, then all the number appear even numbers will be cancelled out, and only the number appears one time is left, which is what we want. See the code below: class Solution { public: int singleNumber(vector<int>& nums) { int res = 0; for(auto ...

Algorithm Advance Outline

Who wants to read these notes? 1. the one who wants to learn algorithm; 2. the one who has a planned interview in a short amount of time, such as one month later; 3. purely out of curiosity; 4. all the rest. The primary purpose for these posts is to help anyone who wants to learn algorithm in a short amount of time and be ready for coding interviews with tech companies, such as, Amazon, Facebook, Microsoft, etc. Before you start, you are expected to already know:  1. the fundamentals of at least one programming language; 2. the fundamentals of data structures. If you do not have these basics, it is better to "google and read" some intro docs, which should NOT take large amount of time. Of course, you can always learn whenever see a "unknown" data structure or line in the codes. Remember, "google and read" is always one of keys to learning. Common algorithms: 1. Recursion 2. Binary search 3. Dynamic programming 4. Breadth-first search 5. Depth-first search...