PrevNext
Not Frequent
 0/17

Shortest Paths with Unweighted Edges

Authors: Benjamin Qi, Andi Qu, Neo Wang

Contributor: Qi Wang

Introduces how BFS can be used to find shortest paths in unweighted graphs.

Unweighted Shortest Path

Focus Problem – try your best to solve this problem before continuing!

Solution - Message Route

Time Complexity: O(V+E)\mathcal{O}(V+E)

We can observe is that there are many possible shortest paths to output. Fortunately, the problem states that we can print any valid solution. Notice that like every other BFS problem, the distance of each node increases by 11 when we travel to the next level of unvisited nodes. However, the problem requires that we add additional information - in this case, the path. When we traverse from node aa to bb, we can set the parent of bb to aa. After the BFS is complete, this allows us to backtrack through the parents which ultimately leads us to our starting node. We know to terminate at node 11 because it's the starting node. If there is no path to our end node, then its distance will remain at INT_MAX.

For the test input, we start with the following parent array.

Node 1 2 3 4 5
Parent 0 0 0 0 0
Distance 0 INT_MAX INT_MAX INT_MAX INT_MAX

After visiting children of node 11:

Node 1 2 3 4 5
Parent 0 1 1 1 0
Distance 0 1 1 1 INT_MAX

After visiting node 55 from node 44:

Node 1 2 3 4 5
Parent 0 1 1 1 4
Distance 0 1 1 1 2

To determine the path, we can backtrack from node n→1n \rightarrow 1, in this case 5→15 \rightarrow 1, pushing each value that we backtrack into a vector. The path we take is 5→parent[5]=4→parent[4]=15 \rightarrow \texttt{parent}[5]=4 \rightarrow \texttt{parent}[4] =1 which corresponds to the vector [5,4,1][5, 4, 1]. We break at node 11 because it was the initial starting node. Finally, we reverse the vector and print out its length (in this case, 33).

C++

#include <bits/stdc++.h>
using namespace std;
using vi = vector<int>;
#define pb push_back
int main() {
int N, M;
cin >> N >> M;
vi dist(N + 1, INT_MAX), parent(N + 1);

Java

import java.io.*;
import java.util.*;
public class Solution {
Code Snippet: Kattio (Click to expand)
private static Map<Integer, LinkedList<Integer>> adj = new HashMap<>();
public static void main(String[] args) {
Kattio io = new Kattio();

Pro Tip

In the gold division, the problem statement will almost never directly be, "Given an unweighted graph, find the shortest path between node uu and vv." Instead, the difficulty in many BFS problems are converting the problem into a graph on which we can run BFS and get the answer.

Extension - 0/1 BFS

Focus Problem – try your best to solve this problem before continuing!

A 0/1 BFS finds the shortest path in a graph where the weights on the edges can only be 0 or 1, and runs in O(V+E)\mathcal{O}(V + E) using a deque. Read the resource below for an explanation of how the algorithm works.

Resources
cp-algo

common applications

Solution - Tracks in the Snow

Complexity: O(NM)\mathcal O(NM)

We can use the following greedy strategy to find our answer:

  • Run flood fill to find each connected component with the same tracks.
  • Construct a graph where the nodes are the connected components and there are edges between adjacent connected components.
  • The answer is the maximum distance from the node containing (1,1)(1, 1) to another node. We can use BFS to find this distance.

For a detailed proof of why this works, see the official editorial.

Although this gives us an O(NM)\mathcal O(NM) solution, there is a simpler solution using 0/1 BFS!

Consider the graph with an edge between each pair of adjacent cells with tracks, where the weight is 0 if the tracks are the same and 1 otherwise. The answer is simply the longest shortest-path from the top left cell. This is because going from one track to another same one is like not leaving a node (hence the cost is 00), while going from one track to a different one is like traversing the edge between two nodes (hence the cost is 11).

Since the weight of each edge is either 0 or 1 and we want the shortest paths from the top left cell to each other cell, we can apply 0/1 BFS. The time complexity of this solution is O(NM)\mathcal O(NM).

C++

#include <bits/stdc++.h>
using namespace std;
int dx[4]{1, -1, 0, 0}, dy[4]{0, 0, 1, -1};
int n, m, depth[4000][4000], ans = 1;
string snow[4000];
bool inside(int x, int y) {
return (x > -1 && x < n && y > -1 && y < m && snow[x][y] != '.');

Java

import java.io.*;
import java.util.*;
public class tracks {
static final int[] dx = {0, 0, -1, 1};
static final int[] dy = {-1, 1, 0, 0};
static int N = 1, H, W;
static int[][] grid, count;
public static void main(String[] args) {
FastIO io = new FastIO();

Warning!

Due to oj.uz's grading constraints for Java, this solution will TLE on the judge.

Problems

StatusSourceProblem NameDifficultyTags
CSESEasy
Show TagsBFS
Old SilverEasy
Show TagsBFS
CSESEasy
Show TagsBFS
SilverEasy
Show TagsBFS
CSESNormal
Show TagsCycle
CSANormal
Show TagsBFS, DFS
GoldNormal
Show TagsBFS
ACNormal
Show TagsBFS
GoldNormal
Show TagsBFS
ACNormal
Show TagsBFS
IOINormal
Show TagsBFS, Binary Search
CSESNormal
Show TagsBFS
IOINormal
Show TagsBFS
GoldHard
Show TagsBFS
GoldHard
Show TagsBFS

Module Progress:

Join the USACO Forum!

Stuck on a problem, or don't understand a module? Join the USACO Forum and get help from other competitive programmers!

PrevNext