Skip to main content

Depth-first-search - medium level - question 2

Depth-first-search - medium level - question 2


2115. Find All Possible Recipes from Given Supplies

You have information about n different recipes. You are given a string array recipes and a 2D string array ingredients. The ith recipe has the name recipes[i], and you can create it if you have all the needed ingredients from ingredients[i]. Ingredients to a recipe may need to be created from other recipes, i.e., ingredients[i] may contain a string that is in recipes.

You are also given a string array supplies containing all the ingredients that you initially have, and you have an infinite supply of all of them.

Return a list of all the recipes that you can create. You may return the answer in any order.

Note that two recipes may contain each other in their ingredients.


Constraints:


n == recipes.length == ingredients.length

1 <= n <= 100

1 <= ingredients[i].length, supplies.length <= 100

1 <= recipes[i].length, ingredients[i][j].length, supplies[k].length <= 10

recipes[i], ingredients[i][j], and supplies[k] consist only of lowercase English letters.

All the values of recipes and supplies combined are unique.

Each ingredients[i] does not contain any duplicate values.



Analysis:

The main concern to this problem is time complexity.

One logic to this question is that once recipe may becomes an available ingredient to other recipes, which may make the previously un-created recipes becomes creatable.

If we handle the above situation to scan all the ingredients again, the time complexity would be too large, since for each newly created recipe, we need to a scan of all the ingredients.

But one scan needs O(numb of ingredients * numb of each ingredients) which is ~ 10^4.

The overall time complexity is O(numb of recipes * num of scans) ~ 10^6, which seems to be Okay for the OJ of Leetcode. But somehow cannot pass it and observed the following output with the following code,

112 / 112 test cases passed, but took too long.

class Solution {
public:
    vector<string> findAllRecipes(vector<string>& recipes, vector<vector<string>>& ingredients, vector<string>& supplies) {
        vector<string> res;
        int n = recipes.size();
        unordered_set<string> st;
        for(auto &a : supplies) st.insert(a);
        queue<int> q;
        q.push(-1);
        while(q.size()) {
            auto t = q.front();
            q.pop();
            cout<<t<<endl;
            for(int i=0; i<n; ++i) {
                if(st.count(recipes[i]) > 0) continue;
                bool flag = true;
                for(auto &a : ingredients[i]) {
                    if(st.count(a) == 0) {
                        flag = false;
                        break;
                    }
                }
                if(flag) {
                    st.insert(recipes[i]);
                    q.push(i);
                }
            }
        }
        for(int i=0; i<n; ++i) {
            if(st.count(recipes[i]) > 0) res.push_back(recipes[i]);
        }
        return res;
    }
};


How can we make it faster?

We do not need to rescan every ingredient once creating a recipe. If we consider the relationship between recipes & ingredients, it looks like a "build graph": the recipe is the build target, and all the ingredients are the dependencies.

So we try to use a top-down dfs method: when all the dependencies can be built, then the initial (or root) built target can be built as well. So it is a dfs problem in nature.

One place is that: we could have cyclic dependencies, so need to take care of this case by tracking the path when do the dfs.


See the code below:


class Solution {
public:
    vector<string> findAllRecipes(vector<string>& recipes, vector<vector<string>>& ingredients, vector<string>& supplies) {
        vector<string> res;
        int n = recipes.size();
        unordered_map<string, int> mp;
        for(int i=0; i<n; ++i) mp[recipes[i]] = i;
        unordered_set<string> st;
        for(auto &a : supplies) st.insert(a);
        for(auto &s : recipes) {
            // cout<<s<<endl;
            vector<int> visit(n, 0);
            dfs(s, mp, ingredients, st, res, visit);
        }
        return res;
    }
private:
    bool dfs(string& s, unordered_map<string, int>& mp, vector<vector<string>>& is, unordered_set<string>& st, vector<string>& res, vector<int>& visit) {
        if(mp.count(s) == 0) return false;
        if(mp[s] == -1) return true;
        int id = mp[s];
        // cout<<id<<endl;
        if(visit[id]) return false;
        visit[id] = 1;
        for(auto &a : is[id]) {
            if(st.count(a) == 0) {
                if(mp.count(a) == 0) return false;
                if(!dfs(a, mp, is, st, res, visit)) return false;
            }
        }
        res.push_back(s);
        mp[s] = -1;
        st.insert(s);
        return true;
    }
};



Upper Layer


Comments

Popular posts from this blog

Trie

Trie is a data structure belonging to n-ary trees, which is designed for some specific questions, such as, prefix of strings. (If you have never known what binary tree is, maybe you can check binary tree first before read more about Trie). The fact that there are only 26 different letters greatly reduce the possible choices for each position. Or to make it clear, for each position, there are 26 different choices. In mathematics, it seems to be very large (height^26 for the full tree), but in reality, it is rare to be a full tree for the set of meaningful strings (needs at least height^26 strings to be a full tree!), since the height or length of English words are less than 10 in most cases. Therefore, it is popular in string matching or related operations. Please be noted that the trie structure is not only limited to strings, the same idea can be extended to other data types, such as integer (then it becomes a binary tree, since for bit, there are only two possible values). In the cod...

Dynamic Programming - Hard Level - Question 1

Dynamic Programming - Hard Level - Question 1  Leetcode 1937  Maximum Number of Points with Cost You are given an m x n integer matrix points (0-indexed). Starting with 0 points, you want to maximize the number of points you can get from the matrix. To gain points, you must pick one cell in each row. Picking the cell at coordinates (r, c) will add points[r][c] to your score. However, you will lose points if you pick a cell too far from the cell that you picked in the previous row. For every two adjacent rows r and r + 1 (where 0 <= r < m - 1), picking cells at coordinates (r, c1) and (r + 1, c2) will subtract abs(c1 - c2) from your score. Return the maximum number of points you can achieve. abs(x) is defined as: x for x >= 0. -x for x < 0. Constraints: m == points.length n == points[r].length 1 <= m, n <= 10^5 1 <= m * n <= 10^5 0 <= points[r][c] <= 10^5 Analysis: Based on the description, we can define dp[i][j] as the maximum number of points wit...