Flow-critical graphs

Lovász et al. proved that every $6$-edge-connected graph has a nowhere-zero $3$-flow. In fact, they proved a more technical statement which says that there exists a nowhere zero $3$-flow that extends the flow prescribed on the incident edges of a single vertex $z$ with bounded degree. We extend this theorem of Lovász et al. to allow $z$ to have arbitrary degree, but with the additional assumption that there is another vertex $x$ with large degree and no small cut separating $x$ and $z$. Using this theorem, we prove two results regarding the generation of minimal graphs with the property that prescribing the edges incident to a vertex with specific flow does not extend to a nowhere-zero $3$-flow. We use this to further strengthen the theorem of Lovász et al., as well as make progress on a conjecture of Li et al.

💡 Research Summary

This paper investigates flow‑critical graphs and extends a key result of Lovász, Thomassen, Wu, and Zhang concerning nowhere‑zero 3‑flows. The classical 3‑flow conjecture (every 4‑edge‑connected graph admits a nowhere‑zero 3‑flow) remains open; the best known result is that every 6‑edge‑connected graph has such a flow, proved by Lovász et al. Their proof actually shows a stronger statement: given a vertex z of bounded degree and a prescribed flow on the edges incident to z, one can extend this “pre‑flow” to a full nowhere‑zero 3‑flow of the whole graph.

The authors generalize this theorem in two directions. First, they remove the bounded‑degree restriction on z, allowing z to have arbitrarily large degree. To compensate, they require the existence of another vertex x with large degree and the condition that there is no small cut (size ≤ 3) separating x from z. Under these hypotheses, every pre‑flow around z extends to a nowhere‑zero 3‑flow. The proof proceeds by a careful inductive argument on a specially defined structure called a “canvas” (a graph together with a distinguished tip z). They introduce the τ‑function, which encodes the parity of the degree of a vertex set together with the sum of prescribed boundary values, and show that the τ‑conditions guarantee the existence of an extension. The argument is organized into four steps: (1) establishing a minimal counterexample order, (2) proving the absence of small cuts around x, (3) showing that any minimal counterexample must be x‑homogeneous (all neighbours of x behave identically with respect to τ), and (4) deriving a contradiction by demonstrating that such a minimal counterexample cannot exist.

The second major contribution is a systematic study of flow‑critical graphs themselves. A graph G is called flow‑critical if it has no nowhere‑zero 3‑flow, but for every non‑trivial partition P of its vertex set (each part inducing a connected subgraph) the contracted graph G/P does admit a nowhere‑zero 3‑flow. This notion generalizes the previously studied connected‑flow‑critical graphs. The authors develop a suite of operations (vertex insertion, edge duplication, “splitting off” edges while preserving τ‑conditions) that generate all minimal flow‑critical canvases. They prove that these operations preserve flow‑criticality, allowing them to construct all minimal counterexamples and then eliminate them via the earlier extension theorem.

Using this machinery, the paper derives new density bounds for flow‑critical graphs. Theorem 1.6 states that if a flow‑critical graph has at most one vertex of degree at least 7, then its number of edges satisfies |E| ≤ 3|V| − 5. This improves upon the conjectured bound |E| ≤ 3|V| − 8 of Li, Luo, and Zhang for connected‑flow‑critical graphs, and it also strengthens a second conjecture involving the number n₃ of degree‑3 vertices. The authors’ method shows that the presence of a high‑degree vertex together with the absence of small cuts forces the graph to be “dense enough” to satisfy the stronger inequality.

Overall, the paper makes three significant advances: (i) it extends the Lovász‑et al. extension theorem to arbitrary‑degree tips under a mild global connectivity condition; (ii) it provides a constructive framework for generating and eliminating minimal flow‑critical canvases; and (iii) it establishes sharper edge‑density bounds for flow‑critical graphs with limited high‑degree vertices. These results bring the community closer to resolving the 3‑flow conjecture for 5‑edge‑connected graphs and open new avenues for applying τ‑based techniques to other modular flow problems and to graphs embedded on surfaces.