Manhattan and Chebyshev flows
Lukáš Gáborik, Sascha Kurz, Giuseppe Mazzuoccolo, Jozef Rajník, Florian Rieg
TL;DR
This work extends multidimensional nowhere-zero flows to non-Euclidean geometries by introducing Manhattan and Chebyshev norm flows in $d$ dimensions, defining $\Phi_d^1(G)$ and $\Phi_d^\infty(G)$ as their flow numbers. The authors prove rationality of these invariants for all bridgeless graphs and show the two-dimensional case satisfies $\Phi_2^1(G)=\Phi_2^\infty(G)$, linking them to cycle covers and $t$-flow-pairs. They derive general upper bounds via dimension and construct Hadamard-matrix-based flows to achieve tight bounds in certain families, while also establishing a sharp 2D bound for cubic graphs and a lower bound for non-$3$-edge-colorable graphs. The introduction of $t$-flow-pairs connects to Seymour's $6$-flow theorem and Tutte's $5$-flow conjecture, with computational evidence supporting the existence of half-flow-pairs in many snarks and several conjectures that aim to strengthen classical flow results. Altogether, the paper provides a rational, structurally rich framework that links norms, cycle covers, and longstanding flow conjectures, and it proposes concrete avenues for progress toward Tutte-type bounds via non-Euclidean flows.
Abstract
We investigate multidimensional nowhere-zero flows of bridgeless graphs. By extending the established use of the Euclidean norm, this paper considers the Manhattan and Chebyshev norms, leading to the definition of the flow numbers $Φ_d^1(G)$ and $Φ_d^\infty(G)$, respectively. These flow numbers are always rational and in two dimensions, they distinguish between cubic graphs that are 3-edge-colourable and those that are not. We also prove that, for any bridgeless graph $G$, the two values $Φ^1_2(G)$ and $Φ^\infty_2(G)$ are the same. We give new upper and lower bounds and structural results, and we find connections with cycle covers. Finally, we introduce the idea of $t$-flow-pairs, which comes from a method used in Seymour's proof of the 6-flow theorem, and we propose new conjectures that could be stronger than Tutte's famous 5-flow conjecture.
