Question

I need to generate random single-source/single-sink flow networks of different dimensions so that I can measure the performance of some algorithms such as the Ford-Fulkerson and Dinic.

Is the Kruskal algorithm a way to generate such graphs?

Answer 1

To create a generic flow network you just need to create an adjancency matrix.

adj[u][v] = capacity from node u to node v

So, you just have to randomly create this matrix.

For example, if n is the number of vertices that you want ( you could make that random too ):

for u in 0..n-1:
    for v in 0..u-1:
      if (rand() % 2 and u != sink and v != source or u == source):
         adj[u][v] = rand()
         adj[v][u] = 0
      else:
         adj[u][v] = 0
         adj[v][u] = rand()