Ahuja-Orlin
General Information
Algorithmic problem: Max-Flow Problems
Type of algorithm: loop
Abstract View
Invariant: After [math]\displaystyle{ i \ge 0 }[/math] iterations:
- The flow [math]\displaystyle{ f }[/math] is a fleasible flow.
- If all upper bounds are integral, [math]\displaystyle{ f }[/math] is integral as well.
- There is a valid distance labeling [math]\displaystyle{ d }[/math] with respect to [math]\displaystyle{ f }[/math].
- Each node [math]\displaystyle{ v\in V\setminus\{t\} }[/math] has a current arc, which is either void or one of the outgoing arcs of [math]\displaystyle{ v }[/math].
- There is a current flow-augmenting path with respect to [math]\displaystyle{ f }[/math]. This path starts with [math]\displaystyle{ s }[/math] and ends at an arbitrary node of [math]\displaystyle{ G }[/math]. Each arc on this path is the current arc of its tail node.
Variant: If no arc is appended to tha path in the current iteration, the distance label of some node (the endnode of the path , in fact) is increased.
Break condition: [math]\displaystyle{ d(s)\geq n }[/math].
Induction basis
Induction step
Complexity
Remark
This algorithm may be seen as a "lazy" variant on Edmonds-Karp. In fact, the most expensive step there is the computation of a flow-augmenting [math]\displaystyle{ (s,t) }[/math]-path. This task amounts to computing the true distance from every node to [math]\displaystyle{ t }[/math]. A valid distance labeling may be seen as "lazily evaluated" true distances.