The Daily Programmer: "Bellman's Algorithm"
An alternative approach to path finding is Bellman's algorithm. This page explains that approach and its implementation.
In this post, we will study an algorithm for single source shortest path on a graph with negative weights but no negative cycles.
Here are some points to keep in mind regarding the different algorithms studied so far:
- Breadth First Search: For graphs where the edge-weights are either zero or all same.
- Dijkstra's Algorithm: For graphs where the edge-weights are non-negative. Uses greedy strategy.
- Bellman-Ford Algorithm: For graphs where the edge-weights may be negative, but no negative weight cycle exists. Uses dynamic programming.
This post contains array - based implementation for simplicity. Another way is to use linked lists using dynamic allocation.
An example graph taken from Introduction to Algorithms:
The code in C is as follows. Its time complexity is O (VE).
#include <stdio.h> #include <stdlib.h> int Bellman_Ford(int G[20][20] , int V, int E, int edge[20][2]) { int i,u,v,k,distance[20],parent[20],S,flag=1; for(i=0;i<V;i++) distance[i] = 1000 , parent[i] = -1 ; printf("Enter source: "); scanf("%d",&S); distance[S-1]=0 ; for(i=0;i<V-1;i++) { for(k=0;k<E;k++) { u = edge[k][0] , v = edge[k][1] ; if(distance[u]+G[u][v] < distance[v]) distance[v] = distance[u] + G[u][v] , parent[v]=u ; } } for(k=0;k<E;k++) { u = edge[k][0] , v = edge[k][1] ; if(distance[u]+G[u][v] < distance[v]) flag = 0 ; } if(flag) for(i=0;i<V;i++) printf("Vertex %d -> cost = %d parent = %d\n",i+1,distance[i],parent[i]+1); return flag; } int main() { int V,edge[20][2],G[20][20],i,j,k=0; printf("BELLMAN FORD\n"); printf("Enter no. of vertices: "); scanf("%d",&V); printf("Enter graph in matrix form:\n"); for(i=0;i<V;i++) for(j=0;j<V;j++) { scanf("%d",&G[i][j]); if(G[i][j]!=0) edge[k][0]=i,edge[k++][1]=j; } if(Bellman_Ford(G,V,k,edge)) printf("\nNo negative weight cycle\n"); else printf("\nNegative weight cycle exists\n"); return 0; }
It's output on the above graph is:
Source: http://www.thedailyprogrammer.com/2015/07/bellman-ford-algorithm.html
This work is licensed under a Creative Commons Attribution 4.0 License.