TheoremComplete

Max-Flow Min-Cut Theorem

Theorem7.1Max-Flow Min-Cut Theorem (Ford-Fulkerson)

In a flow network G=(V,E)G = (V, E) with source ss, sink tt, and capacity function cc, the maximum value of an ss-tt flow equals the minimum capacity of an ss-tt cut: max⁑f∣f∣=min⁑(S,T)c(S,T)\max_f |f| = \min_{(S,T)} c(S, T).


Proof

Proof

Weak duality (≀\leq). For any flow ff and cut (S,T)(S,T): ∣f∣=βˆ‘v∈S(netΒ flowΒ outΒ ofΒ v)=βˆ‘u∈S,v∈Tf(u,v)βˆ’βˆ‘u∈T,v∈Sf(u,v)β‰€βˆ‘u∈S,v∈Tc(u,v)=c(S,T)|f| = \sum_{v\in S} (\text{net flow out of } v) = \sum_{\substack{u\in S, v\in T}} f(u,v) - \sum_{\substack{u\in T, v\in S}} f(u,v) \leq \sum_{\substack{u\in S, v\in T}} c(u,v) = c(S,T).

Strong duality (β‰₯\geq). We show that when no augmenting path exists in GfG_f, the current flow achieves the bound.

Let fβˆ—f^* be a flow with no augmenting ss-tt path in the residual graph Gfβˆ—G_{f^*}. Define S={v∈V:vΒ isΒ reachableΒ fromΒ sΒ inΒ Gfβˆ—}S = \{v \in V : v \text{ is reachable from } s \text{ in } G_{f^*}\} and T=Vβˆ–ST = V \setminus S.

Since no augmenting path exists, tβˆ‰St \notin S, so (S,T)(S, T) is a valid ss-tt cut.

Claim: ∣fβˆ—βˆ£=c(S,T)|f^*| = c(S, T).

For any edge (u,v)∈E(u, v) \in E with u∈S,v∈Tu \in S, v \in T: if fβˆ—(u,v)<c(u,v)f^*(u,v) < c(u,v), then cfβˆ—(u,v)>0c_{f^*}(u,v) > 0, so vv would be reachable from ss in Gfβˆ—G_{f^*}, contradicting v∈Tv \in T. Thus fβˆ—(u,v)=c(u,v)f^*(u,v) = c(u,v).

For any edge (v,u)∈E(v, u) \in E with u∈S,v∈Tu \in S, v \in T: if fβˆ—(v,u)>0f^*(v,u) > 0, then cfβˆ—(u,v)=fβˆ—(v,u)>0c_{f^*}(u,v) = f^*(v,u) > 0, so vv would be reachable, contradicting v∈Tv \in T. Thus fβˆ—(v,u)=0f^*(v,u) = 0.

Therefore: ∣fβˆ—βˆ£=βˆ‘u∈S,v∈Tfβˆ—(u,v)βˆ’βˆ‘v∈T,u∈Sfβˆ—(v,u)=βˆ‘u∈S,v∈Tc(u,v)βˆ’0=c(S,T)|f^*| = \sum_{\substack{u\in S,v\in T}} f^*(u,v) - \sum_{\substack{v\in T,u\in S}} f^*(v,u) = \sum_{\substack{u\in S,v\in T}} c(u,v) - 0 = c(S,T).

Combined with weak duality: max⁑∣f∣=min⁑c(S,T)\max|f| = \min c(S,T). β–‘\square

β– 

ExampleIntegrality Theorem

If all capacities are integers, the Ford-Fulkerson algorithm (with any augmenting path strategy) terminates with an integer-valued maximum flow. Proof: each augmentation increases ∣f∣|f| by a positive integer (the minimum residual capacity on an augmenting path). This integer flow integrality is crucial for applications to matching, connectivity, and covering.

RemarkGeneralizations

The max-flow min-cut theorem generalizes to: multi-commodity flows (but integral multi-commodity flow is NP-hard), flows with lower bounds, flows in undirected networks (replace each edge by two directed edges), and submodular flow (Edmonds-Giles theorem).