LeetCode 399. Evaluate Division

Problem Statement

(Source) Equations are given in the format A / B = k, where A and B are variables represented as strings, and k is a real number (floating point number). Given some queries, return the answers. If the answer does not exist, return -1.0.

Example:
Given a / b = 2.0, b / c = 3.0.
queries are: a / c = ?, b / a = ?, a / e = ?, a / a = ?, x / x = ? .
return [6.0, 0.5, -1.0, 1.0, -1.0 ].

The input is: vector<pair<string, string>> equations, vector<double>& values, vector<pair<string, string>> queries, where equations.size() == values.size(), and the values are positive. This represents the equations. Return vector<double>.

According to the example above:

equations = [ ["a", "b"], ["b", "c"] ],
values = [2.0, 3.0],
queries = [ ["a", "c"], ["b", "a"], ["a", "e"], ["a", "a"], ["x", "x"] ]. 

The input is always valid. You may assume that evaluating the queries will result in no division by zero and there is no contradiction.

Solution 1 - Depth-first Search

Tags: Graph.

class Solution(object):
    def calcEquation(self, equations, values, queries):
        """
        :type equations: List[List[str]]
        :type values: List[float]
        :type queries: List[List[str]]
        :rtype: List[float]
        """
        # Graph.
        g = {}
        n = len(values)
        for i in xrange(n):
            A, B = equations[i]
            k = values[i]
            g.setdefault(A, {})[B] = k
            g.setdefault(B, {})[A] = 1 / k

        # Depth-first Search.
        from collections import deque


        m = len(queries)
        res = [-1.0] * m
        for i in xrange(m):
            A, B = queries[i]
            if (A not in g) or (B not in g):
                continue
            elif A == B:
                res[i] = 1
                continue
            sta = deque([A])
            explored = set([A])
            ans = 1.0
            while sta:
                top = sta[-1]
                for C in g[top]:
                    y = g[top][C]
                    if C not in explored:
                        explored.add(C)
                        sta.append(C)
                        ans *= y
                        break
                if sta[-1] == top:
                    y = sta.pop()
                    if sta:
                        x = sta[-1]
                        ans /= g[x][y]
                else:
                    if sta[-1] == B:
                        res[i] = ans
                        break
        return res

Complexity Analysis:

• Time Complexity: $O(mn)$, where $m$ is the number of variables (or equations), and $n$ is the number of queries.
• Space Complexity: $O(mn)$, where $m$ is the number of variables (or equations), and $n$ is the number of queries.

Solution 2 - Amazing

A variation of Floyd–Warshall, computing quotients instead of shortest paths. An equation A/B=k is like a graph edge A->B, and (A/B)(B/C)(C/D) is like the path A->B->C->D. Submitted once, accepted in 35 ms (Credits @StephanPochmann).

def calcEquation(self, equations, values, queries):
    quot = collections.defaultdict(dict)
    for (num, den), val in zip(equations, values):
        quot[num][num] = quot[den][den] = 1.0
        quot[num][den] = val
        quot[den][num] = 1 / val
    for k, i, j in itertools.permutations(quot, 3):
        if k in quot[i] and j in quot[k]:
            quot[i][j] = quot[i][k] * quot[k][j]
    return [quot[num].get(den, -1.0) for num, den in queries]

Variation without the if (submitted twice, accepted in 68 and 39 ms):

def calcEquation(self, equations, values, queries):
    quot = collections.defaultdict(dict)
    for (num, den), val in zip(equations, values):
        quot[num][num] = quot[den][den] = 1.0
        quot[num][den] = val
        quot[den][num] = 1 / val
    for k in quot:
        for i in quot[k]:
            for j in quot[k]:
                quot[i][j] = quot[i][k] * quot[k][j]
    return [quot[num].get(den, -1.0) for num, den in queries]

Could save a line with for i, j in itertools.permutations(quot[k], 2) but it’s longer and I don’t like it as much here.

Explanation (Credits @ww123):

Nice solution. This is a floyd-warshall-y solution because of the triple nested loops. However it is difficult for me to understand it following the floyd-warshall reasoning. I will share my $.02 here. Correct me if I’m wrong.

Concretely the triple loops create a complete graph. For the first node, connect every pair of its neighbors, a complete sub-graph containing the first node. then the second, then the third, …. so after the loops, all related variables will be in one big complete graph.