Hybrid Quantum-Classical Optimization Workflows for the Shipment Selection Problem
Miguel Angel Lopez-Ruiz, Daiwei Zhu, Jonas Hatzenbuhler, Shudian Zhao, Claudio Girotto, Willie Aboumrad, Jonas Alm, Julia Kompalla
Abstract
We present a quantum optimization framework for the Shipment Selection Problem (SSP) in electric freight logistics, developed jointly by IonQ and Einride. Idle gaps arising from stochastic shipment cancellations reduce fleet utilization and revenue; filling them optimally requires solving a combinatorial assignment problem with quadratic inter-gap dependencies. We formulate the SSP as a Mixed-Integer Quadratic Program, map it to an Ising cost Hamiltonian, and solve it using Iterative-QAOA, a non-variational warm-start extension of the Quantum Approximate Optimization Algorithm (QAOA) with a fixed linear-ramp parameter schedule. An end-to-end hybrid workflow integrates Einride's vehicle routing problem (VRP) solver with IonQ's quantum simulations, enabling evaluation on real, anonymized logistics data spanning up to 130 qubits. We assess solution quality through application-level performance metrics, including Shipments Delivered (SD), Schedule Compatibility Score (SCS), and Total Drive Distance (TDD). When the quantum assignment is passed to the classical solver as a warm start, the resulting hybrid workflow achieves improvements of up to 12\% in SD and a reduction of up to 6\% in total drive distance per shipment for specific instances, while total operational cost remains effectively unchanged. These results show that Iterative-QAOA can generate compatibility-aware assignments that become operationally valuable when embedded in a hybrid logistics optimization workflow.
AI Impact Assessments
(3 models)Scientific Impact Assessment
Core Contribution
This paper presents an end-to-end hybrid quantum-classical workflow for the Shipment Selection Problem (SSP) in electric freight logistics. The core novelty lies in formulating the gap-filling problem arising from stochastic shipment cancellations as a Mixed-Integer Quadratic Program (MIQP), mapping it to an Ising Hamiltonian, and solving it using Iterative-QAOA — a non-variational warm-start extension of QAOA with a fixed linear-ramp parameter schedule. The key architectural insight is that quantum optimization is not used as a standalone solver but rather as a structured initialization module within Einride's existing classical Vehicle Routing Problem (VRP) pipeline. The quantum assignment is passed to the classical EOS solver as a warm start, creating a three-tier evaluation: baseline (BL), direct quantum reconstruction (Q), and quantum warm-start (QWS).
The problem formulation itself is a meaningful contribution — the quadratic terms capture inter-gap dependencies (geographic and temporal interactions between assignments on different vehicles) that classical per-gap local optimization methods inherently miss. This structural observation motivates the quantum approach and distinguishes this work from simpler routing formulations.
Methodological Rigor
The methodology has several strengths but also notable limitations:
Strengths: The paper uses real, anonymized logistics data from Einride's operations across eight monthly schedules spanning over a year, with multiple cancellation scenarios per schedule (32 total). The three-way comparison (BL/Q/QWS) is well-designed and clearly separates the quantum assignment's direct contribution from the hybrid workflow's benefit. Application-level KPIs (SD, SCS, TDD) provide operationally meaningful evaluation beyond abstract optimization metrics.
Limitations: All quantum results are from noiseless simulations — statevector for ≤32 qubits and MPS (bond dimension χ=256) for larger instances up to 130 qubits. This is a significant caveat because MPS with fixed bond dimension introduces systematic approximation errors that may conflate algorithmic limitations with simulation artifacts, as the authors themselves acknowledge. The paper contains no actual quantum hardware execution, which substantially limits claims about near-term quantum utility.
The parameter choices involve considerable heuristic engineering: two fixed Δ values, pruning of ZZ couplings to ⌊30n/p⌋ per layer, an adaptive β_T schedule with tolerance thresholds, and a classical refinement heuristic applied post-quantum. The refinement heuristic (Algorithm 1) appears to do substantial work — Figure 10 shows it often closes the gap between raw QAOA solutions and SCIP optima. This raises the question of how much operational value stems from the quantum component versus the classical post-processing.
The reporting of results as "average (best)" obscures a concerning pattern: averages for direct quantum solutions (Q) are consistently negative for SD (mean -5.1%), meaning the quantum assignment alone typically performs worse than the classical baseline. The positive results emerge primarily through the warm-start mechanism, and the headline improvements (+12.1% SD, -6% TDD/SD) are cherry-picked best-case scenarios from individual instances.
Potential Impact
The paper addresses a genuine industrial need — dynamic schedule repair in electric freight logistics with inter-gap dependencies. The formulation as a QAP-like problem with quadratic interactions is natural and could apply to other fleet management contexts. However, the practical impact is currently limited by:
1. No hardware demonstration: Without quantum hardware results, the pathway to operational deployment remains speculative.
2. Modest average improvements: The mean QWS improvement of +1.7% in SD and -0.2% in TDD/SD, while positive, is marginal and may not justify the complexity overhead of a quantum workflow.
3. Classical competitiveness: The SCIP solver finds optimal or near-optimal solutions for most instances within reasonable time, and the classical refinement heuristic alone closes most of the quality gap.
The classical scaling analysis (Figs. 11-12) attempts to motivate quantum approaches by showing SCIP's runtime sensitivity to quadratic weight and instance size, but this argument would be stronger with larger instances where classical methods demonstrably fail.
Timeliness & Relevance
The paper is timely in the sense that hybrid quantum-classical workflows represent the most pragmatic near-term approach to quantum optimization. The focus on embedding quantum solvers within existing industrial pipelines rather than standalone replacement is a mature architectural perspective. The use of Iterative-QAOA with fixed schedules (avoiding variational optimization) is relevant to addressing known challenges with barren plateaus and parameter optimization overhead in QAOA.
However, the absence of hardware execution makes this more of a workflow design paper than a quantum computing demonstration. The field is increasingly demanding hardware results for credibility.
Strengths & Limitations
Key Strengths:
Notable Weaknesses:
Overall Assessment
This paper presents a thoughtfully designed hybrid workflow for a real logistics problem, with careful attention to the integration between quantum and classical components. The formulation is clean, and the evaluation methodology is sound. However, the scientific impact is limited by the simulation-only nature of the study, the marginal average improvements, and the heavy lifting done by classical post-processing. The paper represents competent applied quantum computing research but falls short of demonstrating clear quantum advantage or even quantum utility for this application class.
Generated Apr 14, 2026
Comparison History (54)
Paper 1 addresses a fundamental open problem in physics—the intersection of general relativity and quantum mechanics. By extending the Diósi-Penrose collapse model to include rotational degrees of freedom via gravitoelectromagnetism, it offers profound theoretical novelty that could influence foundational physics. While Paper 2 presents a valuable practical application of quantum computing to logistics, its scientific scope is narrower and more applied compared to the paradigm-shifting potential of a new quantum-gravity collapse mechanism.
Paper 2 addresses a fundamental and critical bottleneck in quantum computing: error accumulation. By demonstrating a 54% reduction in logical error rates using mid-circuit measurements on actual hardware, it provides a broadly applicable foundation for advancing toward fault-tolerant quantum computing. While Paper 1 presents an excellent real-world application, Paper 2's methodological advancements in error mitigation have a wider potential impact across all quantum simulation and computation fields.
Paper 2 offers a broadly enabling, mathematically constructive design framework for RF trap networks, expanding the achievable topology of field-free guides (including non-smooth features) with explicit parametrizations and periodic/Fourier formulations. This is novel and methodologically rigorous (analytic continuation/Laplace-based construction) and directly targets scalable QCCD architectures, a timely area in quantum computing hardware with cross-impact in ion trapping, microfabrication, and electromagnetics. Paper 1 is application-driven and valuable as a hybrid workflow demonstration, but its scientific novelty is more incremental within QAOA/warm-start heuristics and impact is narrower and near-term.
Paper 1 offers a broadly reusable, mathematically grounded design framework (harmonic continuation from Cauchy data) that directly expands the attainable geometry of RF field-free guide networks (cusps, cotangencies, lattices) with explicit constructions and Fourier formulas. This is novel and likely impactful for ion-trap/QCCD architecture design, a core bottleneck in scalable quantum computing, with applicability across trap engineering and potential theory. Paper 2 is timely and application-driven, but its scientific novelty is more incremental (QAOA workflow/warm-starting) and its claims depend on simulations and instance-specific gains without clear generalizable algorithmic guarantees.
Paper 2 likely has higher impact: it introduces a novel generative/posterior-decoding framework for QEC using discrete diffusion models, addressing a core bottleneck for scalable quantum computing. It demonstrates methodological rigor with benchmarks against strong baselines (MWPM, tensor networks), uses experimental Google processor data, and shows gains across code distances and circuit depths. The approach is broadly relevant to quantum computing, machine learning, and probabilistic inference, and its posterior/confidence outputs enable new operational uses (post-selection, diagnostics). Paper 1 is application-specific and tied to near-term quantum optimization with more limited, instance-dependent benefits.
Paper 1 addresses a foundational bottleneck in quantum computing (quantum error correction) by introducing a highly novel generative diffusion model. Demonstrating performance improvements on real experimental hardware over state-of-the-art decoders (MWPM and tensor networks) represents a significant methodological breakthrough. In contrast, Paper 2 applies existing quantum optimization techniques to a specific logistics problem, relying on simulation rather than real quantum hardware. The breadth of impact for Paper 1 is vastly greater, as improving QEC is essential for realizing fault-tolerant quantum computing across all applications.
Paper 1 presents a highly novel hybrid quantum-classical workflow applied to a real-world logistics problem, demonstrating tangible performance improvements. Its focus on practical quantum advantage and scalability to 130 qubits offers broad industrial and scientific impact. In contrast, Paper 2 is primarily conceptual and pedagogical, revisiting a well-known experiment without introducing groundbreaking new methodologies or practical applications.
Paper 1 is more likely to have higher scientific impact: it derives fundamental, platform-independent bounds from canonical commutation relations and stability, offering broadly applicable limits for dissipative squeezing and entanglement criteria across quantum optics, electromechanics, and quantum information. The results are novel, theoretically rigorous, and timely for ongoing reservoir-engineering experiments, with clear guidance for achievable performance. Paper 2 is application-driven and valuable for hybrid workflows, but its impact is narrower and more contingent on near-term quantum advantage, relying on heuristics and instance-specific gains largely demonstrated via simulations rather than new fundamental methodology.
Paper 1 provides rigorous mathematical proofs about convergence and absence of spectral pollution for HEOM truncations, which are fundamental tools in open quantum systems theory. These results have broad, lasting impact across quantum chemistry, condensed matter physics, and quantum information. Paper 2 presents an applied industry case study using QAOA for logistics optimization, but its impact is narrower: the quantum advantage shown is marginal (up to 12% improvement on specific instances), the problem scale is modest, and the results are tied to a specific commercial application rather than advancing fundamental methodology.
Paper 1 offers a fundamental methodological advancement by eliminating slack variables in QUBO formulations for quantum optimization. This addresses a major scaling bottleneck in quantum computing (qubit requirements for constraints). Its generalized approach impacts a broad range of constrained combinatorial optimization problems across operations research. While Paper 2 provides an excellent real-world industry application, its scientific contribution is narrower and highly domain-specific. Paper 1's foundational algorithmic innovation and broader cross-domain applicability give it a higher potential for widespread scientific impact.
Paper 2 demonstrates higher potential scientific impact due to its clear real-world application in logistics optimization, its novel hybrid quantum-classical framework tested on real anonymized data at meaningful scale (130 qubits), and its industry collaboration (IonQ and Einride). It addresses the timely intersection of quantum computing and practical combinatorial optimization, with measurable improvements (12% in shipments delivered, 6% distance reduction). Paper 1, while technically rigorous in quantum information theory, addresses a more specialized mathematical question about coherence measures with narrower appeal and less immediate practical applicability.
Paper 1 offers higher potential scientific impact due to stronger novelty and broader cross-field relevance: it advances quantum algorithms for nonlinear dynamics on higher-order (simplicial) networks, moving beyond structural tasks to dynamical diagnostics with claimed polynomial/super-polynomial advantages under explicit assumptions. This could influence network science, dynamical systems, and quantum algorithms. Paper 2 is timely and application-driven with real-world logistics data, but it is closer to an engineering demonstration of a QAOA-based hybrid workflow (and appears simulation-based), with more limited methodological novelty and narrower breadth of impact.
Paper 1 provides fundamental theoretical breakthroughs in quantum circuit complexity and state preparation, offering the first constant-depth construction for super-constant weight Dicke states without fanout. Its rigorous theoretical contributions broadly impact quantum algorithm design and complexity theory. Paper 2, while demonstrating a valuable real-world logistics application, relies on existing heuristic frameworks for a highly specific optimization problem, making its broader scientific impact likely lower than the foundational advances presented in Paper 1.
Paper 1 establishes a fundamental result connecting quantum cloning complexity to learning complexity for stabilizer states, proving tight Θ(n) bounds using novel representation-theoretic techniques. It bridges quantum foundations, learning theory, and cryptography with broad theoretical implications. Paper 2 presents an applied quantum-classical hybrid for logistics optimization with modest empirical improvements on a specific industrial problem, but lacks methodological novelty beyond combining existing techniques (QAOA variants with classical solvers) and demonstrates limited quantum advantage. Paper 1's foundational contributions will likely influence multiple research directions more broadly and durably.
Paper 1 likely has higher scientific impact due to its novelty in integrating a concrete hybrid quantum-classical workflow (Iterative-QAOA warm-start) with an industrial logistics pipeline and demonstrating measurable improvements on real-world data at sizable problem scales (up to ~130 qubits). Its applications to freight operations are immediate and broadly relevant to optimization, operations research, and near-term quantum computing. Paper 2 is methodologically careful and valuable within quantum coding theory, but its impact is narrower and primarily incremental (tightening upper bounds for a specific code family) with less immediate cross-domain applicability.
Paper 2 offers a broadly applicable, technically novel advance in fault-tolerant quantum computing: improved logical rotation-gate implementation with better error scaling and substantial spacetime savings, plus an architecture/compilation framework with bounds and resource estimates for impactful simulations. This targets a central bottleneck (non-Clifford/rotation costs) and could influence multiple domains (quantum algorithms, architecture, error correction, simulation). Paper 1 demonstrates a valuable hybrid workflow on real logistics data, but its impact is narrower, depends on near-term quantum advantage not yet established, and improvements are instance-specific with limited methodological generality.
Paper 1 addresses a fundamental theoretical gap in quantum machine learning by deriving the first PAC-Bayesian generalization bounds for quantum models. Its foundational nature provides a critical tool for model design and theory, offering broader and more profound scientific impact across the QML community compared to Paper 2, which focuses on applying existing quantum optimization algorithms to a specific industry use case.
Paper 2 has higher potential impact due to stronger real-world applicability and timeliness: it targets an operational logistics problem with demonstrated end-to-end hybrid integration and measurable improvements on real (anonymized) data at problem sizes up to 130 qubits. Its workflow bridges quantum algorithms, classical optimization, and industry deployment, enabling broader cross-field influence (quantum computing, operations research, transportation). Paper 1 is conceptually valuable for foundational quantum information and experimental rigor, but its immediate applications and audience breadth are narrower, and impact is more incremental within steering-inequality robustness studies.
Paper 2 demonstrates higher potential scientific impact due to its novel application of quantum computing to a real-world logistics problem with concrete, measurable improvements. It presents a hybrid quantum-classical framework tested on real anonymized data at scale (up to 130 qubits), bridging the gap between theoretical quantum algorithms and practical industry use. The collaboration between IonQ and Einride adds credibility and applicability. Paper 1, while methodologically sound, is an incremental extension of prior work on open quantum system dynamics, covering well-studied phenomena (discord vs. entanglement resilience, sudden death) in a relatively standard framework.
Paper 1 offers greater scientific impact through fundamental contributions to quantum battery network theory, establishing novel scaling laws for energy transport across different topologies and providing generalizable design principles. Its findings on reciprocal vs. nonreciprocal couplings, odd-even transport effects, and the roles of thermal/squeezed reservoirs advance foundational understanding applicable across quantum thermodynamics and quantum information. Paper 2, while practically interesting, presents an incremental application of existing QAOA methods to a specific logistics problem with modest improvements (6-12%), limited generalizability, and results that are simulation-based rather than demonstrating quantum advantage.