上善若水

天道酬勤

LeetCode127—Word Ladder

LeetCode127—Word Ladder

原题

Given two words (beginWord and endWord), and a dictionary’s word list, find the length of shortest transformation sequence from beginWord to endWord, such that:

Only one letter can be changed at a time
Each intermediate word must exist in the word list
For example,

Given:
beginWord = “hit”
endWord = “cog”
wordList = [“hot”,”dot”,”dog”,”lot”,”log”]
As one shortest transformation is “hit” -> “hot” -> “dot” -> “dog” -> “cog”,
return its length 5.

Note:
Return 0 if there is no such transformation sequence.
All words have the same length.
All words contain only lowercase alphabetic characters.


分析1

要把这题转化成图论的问题,下述思路构建图:

1.beginWord、endWord、wordList里面的词作为图中的节点。
2.以确定这些节点是否可达:如果变换一个字符能够到另一个单词则说明可达
3.对图进行广度优先搜索,找出起点到终点的长度。

问题

这是我最开始的思路,并且实现了,但是遇到了一个问题,就是 当wordList中单词太多则图的规模就会变得很大,最终就会超时(Time Limit Exceeded),这里也贴出超时代码吧:

class Solution {
private:
    bool isConnect(const string &a, const string&b)
    {//判断两个单词是否可达关系
        int count = 0;
        for (int i = 0; i < a.size(); i++)
        {
            if (a[i] != b[i])
                count++;
            if (count > 1)
                return false; 
        }
        if (count == 1)//只有一个字符不同
            return true;
        else
            return false;
    }
    void buildMap(vector<vector<int>>&wordMap, map<string,int>wordTable)
    {
        int mapSize = wordTable.size();
        for (auto it1 = wordTable.begin(); it1 != wordTable.end(); it1++)
        {
            for (auto it2 = it1; it2 != wordTable.end(); it2++)
            {
                if (isConnect(it1->first, it2->first))
                {
                //初始化邻接矩阵(1-可达 0-不可达)
                    wordMap[it1->second][it2->second] = 1;
                    wordMap[it2->second][it1->second] = 1;
                }
            }
        }
    }
    int BFS(vector<vector<int>>&wordMap, string start, string end, map<string, int>wordTable)
    {
        vector<bool>visit(wordTable.size(), false);//访问表
        queue<int>q;
        queue<int>d;//记录距离distance的辅助队列
        q.push(wordTable[start]);
        d.push(1);
        visit[wordTable[start]] = true;
        while (!q.empty())
        {
            int t = d.front();
            int i = q.front();
            d.pop();
            q.pop();
            if (i == wordTable[end])//找到终点
            {
                return t;
            }
            for (int j = 0; j < wordMap.size(); j++)
            {
                if (!visit[j] && wordMap[i][j] == 1)
                {
                    visit[j] = true;
                    q.push(j);
                    d.push(t + 1);
                }
            }
        }//end while
        return 0;
    }
public:
    int ladderLength(string beginWord, string endWord, unordered_set<string>& wordList) {
        map<string, int>wordTable;//单词索引表
        wordTable[beginWord] = 0;
        wordTable[endWord] = 1;
        auto it = wordList.begin();
        int index = 2;
        for (; it != wordList.end(); it++)//
        {
            if (wordTable.find(*it)==wordTable.end())
                wordTable[*it] = index++;//建立一个单词-索引表方便建图
        }
        int size = wordTable.size();
        vector<vector<int>> wordMap(size,vector<int>(size,0));//初始化邻接矩阵
        buildMap(wordMap, wordTable);//建图
        return BFS(wordMap, beginWord, endWord, wordTable);
        //return 0;
    }
};

分析2

其实不用把整个图建出来,可以走一步看一步,也就是说:

当访问到节点A时,可以先求出所有跟节点A临接且未被访问的节点入队,然后按照BFS的思路做即可。

但是,修改后的代码依然是超时,后来找原因发现是我在“判断两个节点是否临接”这个步骤中使用的算法效率太低,对比网上的思路:

我的方法是:把当前节点的单词current和wordList里面的所有词比较,不相同的字符的个数为1的时候return true,否则return false。这样做的缺点是,不论如何都要遍历完整个wordList,查找时间复杂度是O(n)

网上一种思路是:替换当前节点的单词current每个字符(从a~z),然后看看替换过后的单词是否在单词表中,这里我不理解的是,由于我们是用unordered_set来保存单词表,虽然最坏情况下的时间复杂度会到O(n),但一般情况下是可以在常数时间O(1)下访问的,因此使用unordered_set::find的时间复杂度是要低于线性时间复杂度的,这样就提高了效率。


代码

class Solution {
private:
    void buildMap(string end,vector<string>&connect,unordered_set<string>&visit,string& current,const unordered_set<string>&wordList)
    {
        connect.clear();
        string cur = current;
        /*
        超时:时间复杂度O(n)
        for(auto i=wordList.begin();i!=wordList.end();i++)
        {
            if(visit.find(*i)!=visit.end())
                continue;
            if(isConnect(cur,*i))
                connect.push_back(*i);
        }
        */

        #if 1
        for (int i = 0; i < cur.size(); i++)
        {
            char t = cur[i];
            for (char c = 'a'; c < 'z'; c++)
            {
                if (c == t)
                {
                    continue;
                }
                cur[i] = c;
                if ((cur == end || wordList.find(cur) != wordList.end()) && (visit.find(cur) == visit.end()))
                {
                    connect.push_back(cur);
                }
            }
            cur[i]=t;
        }
        #endif
    }
    int BFS(string beginWord, string endWord, unordered_set<string>& wordList)
    {
        queue<string>q;
        queue<int>d;//路径distance的辅助队列
        unordered_set<string>visit;
        vector<string>connect;
        q.push(beginWord);
        d.push(1);
        while (!q.empty())
        {
            string current = q.front();
            int tmpDist = d.front();
            q.pop();
            d.pop();
            buildMap(endWord,connect, visit, current, wordList);//获取临接单词
            if (current == endWord)//找到终点
            {
                return tmpDist;
            }
            for (int i = 0; i < connect.size(); i++)
            {
                if (visit.find(connect[i]) == visit.end())//未被访问
                {
                    visit.insert(connect[i]);
                    q.push(connect[i]);
                    d.push(tmpDist + 1);
                }
            }
        }
        return 0;//没有找到路径
    }

public:
    int ladderLength(string beginWord, string endWord, unordered_set<string>& wordList) {
        int res= BFS(beginWord, endWord, wordList);
        return res;
    }
};

参考

unordered_set::find时间复杂度
http://www.cplusplus.com/reference/unordered_set/unordered_set/find/

Complexity
Average case: constant.
Worst case: linear in container size.

阅读更多
版权声明:本文为博主原创文章,未经博主允许不得转载。 https://blog.csdn.net/zhangxiao93/article/details/51354736
文章标签: leetcode bfs
个人分类: leetcode
想对作者说点什么? 我来说一句

没有更多推荐了,返回首页

加入CSDN,享受更精准的内容推荐,与500万程序员共同成长!
关闭
关闭