#####   ######  #####    ###    #   #  ###  #   #  ######
##  ##  ##      ##  ##  ## ##   #   #   #   #   #  ##
#####   ####    #####   #   #   #   #   #   #   #  ####
##  #   ##      ##      ## ##    # #    #    # #   ##
##   #  ######  ##       ###      #    ###    #    ######

$ curl repovive.com/roadmaps/graph-theory/strongly-connected-components/kosaraju-implementation

░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░███████████████████████████████████████████████████████████████████████████████████████████████

#####   ######  #####    ###    #   #  ###  #   #  ######
##  ##  ##      ##  ##  ## ##   #   #   #   #   #  ##
#####   ####    #####   #   #   #   #   #   #   #  ####
##  #   ##      ##      ## ##    # #    #    # #   ##
##   #  ######  ##       ###      #    ###    #    ######

$ curl repovive.com/roadmaps/graph-theory/strongly-connected-components/kosaraju-implementation

Repovive

Kosaraju - Implementation - Strongly Connected Components | Graph Theory | Repovive

Repovive.

Graph Theory37 sections · 1633 units81h total

Open in Course

Kosaraju - Implementation

(C++ code)

Here is Kosaraju:

function dfs1(v, adj, vis, st):
    vis[v] = true
    for u in adj[v]:
        if not vis[u]:
            dfs1(u, adj, vis, st)
    st.push(v)

function dfs2(v, radj, vis, comp):
    vis[v] = true
    scc[v] = comp
    for u in radj[v]:
        if not vis[u]:
            dfs2(u, radj, vis, comp)

Build transpose graph, run dfs1 on all nodes, then run dfs2 in stack order.

This runs in $O(V + E)$ time and uses $O(V)$ space.