Minimum cost path from source node to destination node via an intermediate node

• Difficulty Level : Medium
• Last Updated : 13 Jan, 2022

Given an undirected weighted graph. The task is to find the minimum cost of the path from source node to the destination node via an intermediate node.
Note: If an edge is traveled twice, only once weight is calculated as cost. Examples:

Input: source = 0, destination = 2, intermediate = 3;
Output:
The minimum cost path 0->1->3->1->2
The edge (1-3) occurs twice in the path, but its weight is
Input: source = 0, destination = 2, intermediate = 1;
Output:
The minimum cost path is 0->1>2

Approach: Let suppose take a path P1 from Source to intermediate, and a path P2 from intermediate to destination. There can be some common edges among these 2 paths. Hence, the optimal path will always have the following form: for any node U, the walk consists of edges on the shortest path from Source to U, from intermediate to U, and from destination to U. Hence, if dist(a, b) is the cost of shortest path between node a and b, the required minimum cost path will be min{ dist(Source, U) + dist(intermediate, U) + dist(destination, U) } for all U. The Minimum distance of all nodes from Source, intermediate, and destination can be found by doing Dijkstra’s Shortest Path algorithm from these 3 nodes.
Below is the implementation of the above approach.

CPP

 // CPP program to find minimum distance between// source and destination node and visiting// of intermediate node is compulsory#include using namespace std;#define MAXN 100005 // to store mapped values of graphvector > v[MAXN]; // to store distance of// all nodes from the source nodeint dist[MAXN]; // Dijkstra's algorithm to find// shortest path from source to nodevoid dijkstra(int source, int n){    // set all the vertices    // distances as infinity    for (int i = 0; i < n; i++)        dist[i] = INT_MAX;     // set all vertex as unvisited    bool vis[n];    memset(vis, false, sizeof vis);     // make distance from source    // vertex to source vertex is zero    dist = 0;     // // multiset do the job    // as a min-priority queue    multiset > s;     // insert the source node with distance = 0    s.insert({ 0, source });     while (!s.empty()) {        pair p = *s.begin();        // pop the vertex with the minimum distance        s.erase(s.begin());         int x = p.second;        int wei = p.first;         // check if the popped vertex        // is visited before        if (vis[x])            continue;         vis[x] = true;         for (int i = 0; i < v[x].size(); i++) {            int e = v[x][i].first;            int w = v[x][i].second;             // check if the next vertex            // distance could be minimized            if (dist[x] + w < dist[e]) {                 dist[e] = dist[x] + w;                 // insert the next vertex                // with the updated distance                s.insert({ dist[e], e });            }        }    }} // function to add edges in graphvoid add_edge(int s, int t, int weight){    v[s].push_back({ t, weight });    v[t].push_back({ s, weight });} // function to find the minimum shortest pathint solve(int source, int destination,               int intermediate, int n){    int ans = INT_MAX;     dijkstra(source, n);     // store distance from source to    // all other vertices    int dsource[n];    for (int i = 0; i < n; i++)        dsource[i] = dist[i];     dijkstra(destination, n);    // store distance from destination    // to all other vertices    int ddestination[n];    for (int i = 0; i < n; i++)        ddestination[i] = dist[i];     dijkstra(intermediate, n);    // store distance from intermediate    // to all other vertices    int dintermediate[n];    for (int i = 0; i < n; i++)        dintermediate[i] = dist[i];     // find required answer    for (int i = 0; i < n; i++)        ans = min(ans, dsource[i] + ddestination[i]                                  + dintermediate[i]);     return ans;} // Driver codeint main(){     int n = 4;    int source = 0, destination = 2, intermediate = 3;     // add edges in graph    add_edge(0, 1, 1);    add_edge(1, 2, 2);    add_edge(1, 3, 3);     // function call for minimum shortest path    cout << solve(source, destination, intermediate, n);     return 0;}
Output:
6

Time complexity: O((N + M) * logN), where N is number of nodes, M is number of edges.
Auxiliary Space: O(N+M)

My Personal Notes arrow_drop_up