A QUBO Formulation for Nowhere-Zero $k$-Flows

Abstract: We consider the encoding of graph problems as Quadratic Unconstrained Binary Optimization (QUBO) problems, which are solvable by either quantum or classical annealers. Yet, the class of problems encoded as QUBO problems has not previously included nowhere-zero flows. Nowhere-zero flows are related to Tutte's $5$-flow conjecture and appear in many contexts in graph theory. We provide an encoding of nowhere-zero flows as a QUBO Hamiltonian and prove the correctness of the construction. Our construction yields a Hamiltonian $H_{\mathrm{mod},k}$ whose ground state has zero energy if and only if the graph $G$ has a nowhere-zero $\mathbb Z_k$-flow. By Tutte's equivalence theorem, zero ground energy is equivalent to $\varphi(G)\le k$, and the zero-energy degeneracy is given by the flow polynomial $F(G;k)$. In particular, when the ground-state energy is zero, this is also the ground-state degeneracy. The construction uses one-hot variables to represent the edge flow residues modulo $k$ and auxiliary variables to represent the per-vertex modular quotient. We prove that the correctness of the construction is independent of the choice of orientation, root vertex, and positive penalty weights. We verify the construction on $59$ examples of graphs and values of $k$ that include both yes-instances and no-instances. We exhaustively sweep orientations and root choices on selected robustness instances and test a finite suite of positive penalty weights. The resulting Hamiltonian is implemented using the dimod.BinaryQuadraticModel class, which is compatible with the D-Wave Ocean SDK. Quantum-hardware runs and claims about potential speedup using these devices are left to follow-up work.

• The paper introduces a novel QUBO reduction for nowhere-zero k-flows, establishing a bijection between zero-energy states and the flow polynomial of a graph.
• It details a Hamiltonian construction using one-hot edge encoding and auxiliary variables for vertex conservation, with rigorous proofs of parameter robustness.
• The method is empirically validated on challenging graphs, demonstrating its practical applicability for both quantum annealing and classical optimization.

QUBO Formulation for Nowhere-Zero $k$-Flows: An Analytical Essay

Introduction

This paper provides a formal reduction of the nowhere-zero $k$-flow feasibility problem on graphs to Quadratic Unconstrained Binary Optimization (QUBO), establishing a rigorous and robust QUBO Hamiltonian encoding for detecting nowhere-zero $Z$-flows in (potentially multiedged and disconnected) nonempty loopless multigraphs. The reduction generalizes and completes the set of classical combinatorial graph problems (such as coloring, Hamiltonicity, and matching) with explicit QUBO mappings, using a construction fundamentally distinct from those for colorings by the modular conservation and auxiliary variable architecture. The paper delivers theoretical completeness proofs, implementation-relevant details regarding variable allocations, and comprehensive empirical validation on a benchmark suite including challenging graph families for the $k$-flow problem.

The Nowhere-Zero $k$-Flow Problem

A $k$-flow assigns to each edge of an oriented multigraph a nonzero integer in $\{\pm1, \ldots, \pm(k-1)\}$ such that the signed sum of flow values at every vertex satisfies Kirchhoff’s law (i.e., it is zero in the integers). The corresponding modular formulation (over $\mathbb{Z}_k$), due to Tutte, is to assign a residue $a_e \in \{1, \ldots, k-1\}$ to each directed edge such that for all vertices $v$, the total sum of $k$0 over all incident edges $k$1, with $k$2 indicating the orientation, vanishes modulo $k$3. The existence of such a $k$4-flow is a central question in combinatorial graph theory, tightly related to several open conjectures (Tutte's 5-flow conjecture, the 4-flow conjecture, and reconfiguration questions).

QUBO Encoding: Hamiltonian Construction

The authors construct a QUBO Hamiltonian $k$5 incorporating two key constraints:

1. Edge Labeling (One-Hot Encoding): For each edge, exactly one binary variable (out of $k$6 choices) is set, representing the assignment of a residue value.
2. Vertex Conservation: For each non-root vertex, the modular conservation constraint is encoded using an auxiliary block of binary variables formulating the per-vertex modular quotient. The outcome is a sum-of-squares Hamiltonian, where vanishing energy implies exactly one selected residue for each edge and satisfaction of modular conservation at each relevant vertex.

The variable set includes $k$7 binary variables per edge (for one-hot residue choice) and per-vertex quotient blocks for all non-root vertices. The construction is robust to orientation and root choices owing to a formal symmetry argument; the parameterization with arbitrary positive weights for penalty terms does not affect ground-state structure.

Theoretical Results

Rigorous correctness proofs are given:

• Bijection of Zero-Energy States and Nowhere-Zero Flows: There exists a bijection between the nowhere-zero $k$8-flows of the input oriented multigraph and the zero-energy states of $k$9; the count of zero-energy states equals the value of the flow polynomial $Z$0, a fundamental Tutte invariant.
• Independence from Orientations and Roots: The emptiness and cardinality of the set of zero-energy states is independent of orientation, root selection, and penalty scaling, confirming parameter-robustness at the theorem-level.
• Sharp Energy Gap: For all assignments with nonzero energy, the energy is bounded below by the minimum penalty parameter, ensuring strong penalization outside the feasible set, which is essential for any annealing-based solver.
• Scope: The Hamiltonian is well-defined for all nonempty loopless multigraphs, including disconnected or bridged graphs; on such no-instances, the ground-state energy is strictly positive.

Experimental Validation and Benchmarks

The reduction and the encoding have been validated exhaustively over an extensive test suite of 59 $Z$1 pairs, covering both simple and multigraph cases, including acute worst-case graphs such as the Petersen graph and snark families. For yes-instances, all enumerated flows map bijectively to zero-energy states, while for no-instances all one-hot assignments yield strictly positive energies. Extensive parameter robustness sweeps (orientation, root, penalty weights) found no deviations from theorem-level predictions.

Empirical metrics concerning the growth of variable count and coupler density with respect to $Z$2 and graph structure are provided. The QUBO models are efficiently implementable, and the dominant scaling terms are explicitly characterized. Random-graph families and snark benchmarks confirm the scalability and generality of the construction.

Notably, no quantum hardware optimization or empirical quantum annealing performance results are reported; the paper focuses exclusively on the reduction and mathematical validation.

Practical and Theoretical Implications

The QUBO encoding explicitly integrates an important open graph-theoretical challenge into the standard toolbox for quantum/classical optimization by annealing, making nowhere-zero $Z$3-flow feasibility accessible to generic hardware and software workflows for QUBO instances. The precise characterization of the ground-state manifold in terms of flow polynomials makes this a powerful diagnostic for future studies exploiting the structure or degeneracies of the flow space.

On the practical side, the encoding enables the application of quantum annealers or classical simulated annealers to instances of the nowhere-zero $Z$4-flow problem, and allows for benchmarking heuristic and exact solver behavior against a mathematically certified zero-energy ground state structure. However, the authors explicitly abstain from any claims about quantum speedup, solution success rates, or empirical comparative performance; such investigations are deferred to future work.

On the theoretical side, the formal methodology solidifies the role of QUBO reductions for modular algebraic flow constraints, potentially guiding similar reductions for other algebraic or topological invariants in combinatorics and condensed matter. The auxiliary-variable strategy for encoding modular constraints in binary optimization is likely relevant for broader classes of problems requiring modular conservation.

Future Directions

Future research should address:

• The empirical performance of standard quantum annealers and classical heuristics on hard instances encoded via $Z$5, quantifying solution rates and scaling with problem parameters.
• Investigation of alternative encodings (e.g., signed-integer Hamiltonians or special cases for small $Z$6) to balance hardware-specific requirements (such as lower dynamic range or coupler layout) with ground-state degeneracy.
• Utilization of the encoding for sampling flows to study the geometry and connectivity of the flow-reconfiguration graph $Z$7, relevant for recent combinatorial reconfiguration research.
• Theoretical analysis of Hamiltonian gap and degeneracy structure for quantum optimization landscapes, potentially connecting with open conjectures in graph theory.

Conclusion

This work completes the reduction of the nowhere-zero $Z$8-flow feasibility problem to binary QUBO and provides a robust, parameter-insensitive Hamiltonian whose ground-state structure precisely reflects the existence and cardinality of modular flows. The contributions strongly facilitate the integration of flow-theoretic questions into quantum and classical binary optimization platforms and set a standard for encoding rigor and verifiability in the study of QUBO-representable combinatorial structures.