How to run a flow or matching query 01 Build the network. The Flow network S→T preset is a directed 6-node network; bipartite tests also accept the Bipartite K₃,₃ preset and any grid. 02 Choose Dinic max-flow or Edmonds–Karp , set Source S and Sink T, and read max flow, min-cut capacity, the cross-check row and the per-edge flow/capacity table. 03 For Bipartite check read the verdict, the left and right sets and — when the answer is No — the odd-cycle witness. 04 For Hungarian matching load a bipartite graph and read the matching size, the pairs, the minimum vertex cover and any unmatched vertices. 05 Step the run on the flow algorithms to watch augmenting paths raise the flow value.
Worked readouts
Setup
Readout
Flow preset, Dinic S → T
max flow 19 ; min-cut capacity 19 ; cross-check (Edmonds–Karp) 19 ; cut edges S→A, B→D ; source side S, B
Flow preset, edge flows
S→A 10/10 , S→B 9/10 , A→C 4/4 , A→D 6/8 , B→D 9/9 , C→T 9/10 , D→C 5/6 , D→T 10/10
Bipartite K₃,₃, Hungarian matching
maximum matching 3 ; pairs L1–R3, L2–R2, L3–R1 ; minimum vertex cover L1, L2, L3 ; unmatched —
Cycle C₅, Bipartite check
bipartite No ; odd-cycle witness C → B → A → E → D → C
Grid 4 × 5, Bipartite check
bipartite Yes ; left 10 nodes starting 1-1, 1-3, 1-5, 2-2; right 10 starting 1-2, 1-4, 2-1
The two flow algorithms agree at 19 , and the cut edges confirm the bound: S→A carries 10 and B→D carries 9 , summing to the max flow. The sink receives 9 through C→T and 10 through D→T, again 19 .
Why the cut proves the flow Max-flow min-cut says the best flow equals the cheapest bottleneck. Dinic stops when no augmenting path is left; the panel then shows the nodes still reachable from S — the source side S, B — and the edges leaving it, S→A and B→D , whose capacities 10 and 9 add up to the flow. Every unit must cross that cut, so no larger flow exists. On the matching side, Kőnig's theorem gives the same style of proof: a matching of size 3 and a vertex cover of size 3 pin each other down.
Limits
Matching needs a bipartite graph. Run Hungarian matching on the Petersen preset and the panel refuses with "Graph is not bipartite (odd cycle O3 → O2 → O1 → O5 → O4 → O3) — the Hungarian/Kuhn algorithm does not apply" instead of returning a partial answer. General-graph matching is not offered.
One source–sink pair at a time, no costs. There is no min-cost flow, no multi-commodity flow and no lower bounds on edges; run pairs in turn and compare the source sides. What it doesn't do. It does not enumerate cuts beyond the reported minimum, and it does not import network files. Build the network on the canvas, then take the adjacency matrix to math tools for further algebra or list reachability with components .
The direction trap Cut edges are printed as directed pairs. On the directed Flow preset they are S→A and B→D ; on the undirected default graph, Dinic A→F prints A→B, A→C with source side A . The source side is the stable reading across runs — everything still reachable from S when the algorithm stops — so compare that row, not the arrow direction, when you re-check a cut by hand.
Where it is useful Capacity planning The Flow preset shows how much of a 19 -unit demand can pass and where the bottleneck sits: S→B keeps 1 unit of headroom while A→D leaves 2 , but the cut S→A / B→D is saturated. Raise a cut-edge capacity and the flow value moves; raise an edge inside the network and it usually does not.
Assignment problems On Bipartite K₃,₃ the Hungarian panel pairs all three left nodes and prints a cover of the same size, so the assignment is provably maximal. Grids answer Yes to the bipartite check with an equal 10/10 split, while C₅ answers No and hands back the odd cycle — the certificate that no 2-colouring exists. The traversal panel can then show the colour groups node by node.
Privacy All flow, colouring and matching computation runs inside your browser; nothing is uploaded.
References
Wikipedia, Maximum flow problem , en.wikipedia.org (访问日期:2026-10-07)— augmenting paths and flow value.
Wikipedia, Max-flow min-cut theorem , en.wikipedia.org (访问日期:2026-10-07)— flow equals cut capacity.
Wikipedia, Bipartite graph , en.wikipedia.org (访问日期:2026-10-07)— 2-colouring and odd cycles.
Wikipedia, Maximum cardinality matching , en.wikipedia.org (访问日期:2026-10-07)— augmenting-path matching.
Wikipedia, Kőnig's theorem (graph theory) , en.wikipedia.org (访问日期:2026-10-07)— matching equals vertex cover in bipartite graphs.
Calculators in this hub
Hand-picked tools, one click away. The mini versions compute live and carry your values into the full calculator.
Sources & review
Reviewed by CalcX Editorial Team
Updated 2026-10-07