All revision notes topics

Augmenting flows and refinementsAQA A-Level Further Maths: Revision notes

Section 1

Augmenting a flow

Given a feasible flow, you improve it by finding a flow-augmenting route from SS to TT. Along each arc used in its own direction (a forward arc) the spare capacity is capacity−flow\text{capacity}-\text{flow}. An arc may also be used backwards: if arc ABAB carries flow ff, then travelling from BB to AA is allowed and can cancel up to ff units of it. The amount by which you can increase the flow is the smallest spare capacity (or backward flow) along the route, the bottleneck. After augmenting, update every arc: add the amount on forward arcs, subtract it on backward arcs. Example: arc SASA with capacity 6 and flow 4 has forward spare capacity 2.

Key termsflow-augmenting routespare capacitybackward arc
Common mistake

Increasing the flow by the capacity of an arc rather than by the spare capacity left after the current flow.

Section 2

The augmenting algorithm and proving maximality

Start with any feasible flow (for example zero). Repeatedly find a flow-augmenting route and increase the flow by its bottleneck. Stop when no augmenting route exists. At that point, the nodes reachable from SS by forward arcs with spare capacity and backward arcs with positive flow form the set XX; every arc leaving XX is saturated and every arc entering XX carries no flow, so the flow equals the capacity of the cut X∣YX|Y. By the max-flow min-cut theorem the flow is maximal. Example: arcs SASA 6, SBSB 3, ABAB 5, ATAT 3, BTBT 4 with flow 4 on S→A→B→TS\to A\to B\to T. Route S→A→TS\to A\to T adds 2 (total 6). Route S→B→A→TS\to B\to A\to T, using ABAB backwards, adds 1 (total 7). The cut {S,A,B}∣{T}\{S,A,B\}|\{T\} has capacity 3+4=73+4=7, so 7 is the maximum.

Key termssaturatedbottleneck
Exam tip

Quote the cut when you stop: list its arcs, add the capacities and compare with your flow value.

Section 3

Upper and lower capacities

Some arcs have a lower capacity as well as an upper capacity, written [l,u][l,u]: the flow must satisfy l≤f≤ul\leq f\leq u. A flow is feasible only if every arc meets its limits and conservation holds at every intermediate node. Find a feasible flow first (it may need adjusting by trial), then augment it. On an arc with flow ff, the forward spare capacity is u−fu-f and the amount that can be cancelled by a backward move is f−lf-l. For a cut X∣YX|Y, the capacity is ∑u\sum u for arcs from XX to YY minus ∑l\sum l for arcs from YY to XX. Lower capacities also set a minimum flow: for example, if AT≥4AT\geq4 and BT≥3BT\geq3 into the sink, every feasible flow has value at least 7.

Key termslower capacityupper capacityfeasible flow
Common mistake

Reducing flow on an arc below its lower capacity when using it backwards. The backward amount is f−lf-l, not ff.

Section 4

Nodes of restricted capacity

Sometimes a node, not an arc, limits the flow, for example a depot that can handle at most 6 units per hour. Replace the node BB by two nodes B1B_1 and B2B_2 joined by an arc B1→B2B_1\to B_2 whose capacity is the restriction. Every arc that entered BB now enters B1B_1, and every arc that left BB now leaves B2B_2. The new arc is then treated like any other, and may appear in a minimum cut. Example: with SASA 8, SBSB 7, ABAB 4, ACAC 5, BCBC 6, BTBT 5, CTCT 9, the unrestricted maximum flow is 14. Limiting BB to 6 gives a minimum cut {AC,B1B2}\{AC,B_1B_2\} of capacity 5+6=115+6=11.

Key termsrestricted nodenode splitting
Common mistake

Putting the restriction on one of the original arcs rather than on a new arc between the two halves of the node.

Section 5

Refining and evaluating a model

Once the maximum flow is known you can evaluate changes. The minimum cuts show where the bottleneck is. Raising an arc's capacity helps only if that arc is in every minimum cut, and then the flow rises by the increase or until another cut becomes the minimum, whichever is smaller. Example: raising a capacity by 2 may raise the maximum flow by only 1 if another cut has capacity one more than the current minimum. Always check the next smallest cut before you give the new maximum, and state conclusions in context with units.

Exam tip

Compare options by their effect on the maximum flow, and name the cut that limits each one.

That's the notes covered.

Carry on to the next subtopic.

Exam questions on Augmenting flows and refinements

  1. A network has source SS, sink TT and nodes AA and BB. The directed arcs and their capacities are SASA 6, SBSB 3, ABAB 5, ATAT 3 and BTBT 4. An initial flow of 4 units is sent along the route S→A→B→TS\to A\to B\to T.
    Show how the flow of 6 can be increased further, and prove that the resulting flow is a maximum.2 marks
  2. Water flows from a source SS to a sink TT through junctions AA and BB. Each pipe has a lower and an upper limit on its flow, in litres per second, written [lower,upper][\text{lower},\text{upper}]: SASA [3,8][3,8], SBSB [2,5][2,5], ABAB [1,4][1,4], ATAT [4,7][4,7] and BTBT [3,6][3,6]. Water flows in the direction of each arc.
    Show that the value of every feasible flow is at least 7 litres per second.2 marks
  3. A sorting centre network has source SS, sink TT and nodes AA, BB and CC. The directed arcs and their capacities (parcels per hour, in thousands) are SASA 8, SBSB 7, ABAB 4, ACAC 5, BCBC 6, BTBT 5 and CTCT 9. Node BB can handle at most 6 thousand parcels per hour in total.
    Explain how the network can be modified to include the restriction on node BB.3 marks
See the full worksheet

Written by the Exaim team, led by Shaun Daswani (Head of Upper Secondary, Improve ME Institute; MSc Financial Mathematics, Imperial College London; BSc, UCL) and Jason Daswani (operational lead, Improve ME Institute; LSE).