Three ways to share a QPU: Scheduling strategies for hybrid Quantum-HPC applications
Marco Cipollini, Simone Rizzo, Sergio Iserte, Paolo Viviani, Giacomo Vitali, Matteo Barbieri, Gabriella Bettonte, Elisabetta Boella
Abstract
As quantum computing (QC) technologies mature, their integration into established high-performance computing (HPC) infrastructures is becoming a central objective for next-generation computing systems. However, unlocking the potential of hybrid platforms for computationally demanding workloads remains challenging. The mismatch between quantum and classical programming models, the limited maturity of quantum software stacks, and the scarcity of quantum processing units (QPUs) above all, necessitate scheduling strategies that go beyond standard HPC mechanisms to manage such heterogeneous and constrained resources. To address this issue, we investigate three distinct methodologies for HPC-QC resource scheduling: time-based multiplexing, dynamic resource management, and workflow decomposition. Experimental validation on production HPC clusters and real quantum hardware demonstrates the effectiveness of these approaches under different workload scenarios. Malleability and workflow strategies significantly optimize classical resource utilization, reducing consumption by up to 45.7% and 64% respectively, proving to be best fitted for hybrid jobs where quantum and classical workloads are evenly balanced. Conversely, time-multiplexing enhances QPU utilization and reduces execution time at the cluster level, making it the optimal strategy for the opposite context, which is characterized by high classical-quantum workload imbalances. These findings underscore the practical viability of tailored scheduling strategies for hybrid HPC-QC environments and highlight their complementarity in building efficient, scalable software stacks for next-generation quantum-accelerated facilities.
AI Impact Assessments
(3 models)Scientific Impact Assessment
1. Core Contribution
This paper consolidates and experimentally validates three complementary scheduling strategies for hybrid HPC-quantum computing (HPC-QC) environments: (1) virtual QPUs (vQPUs) enabling time-based multiplexing of a physical QPU among concurrent jobs, (2) dynamic resource management via MPI malleability (using the DMR framework) to resize classical allocations at runtime, and (3) workflow decomposition (using StreamFlow WMS) to provision resources only when tasks are active.
The key insight is that these strategies are not competing but complementary, occupying distinct regions along two orthogonal axes: time-scale granularity and programming-model transparency. The paper maps each strategy to its ideal workload profile—vQPUs for high classical-to-quantum imbalance (typical of superconducting QPUs), and malleability/workflow approaches for more balanced workloads or longer quantum phases (e.g., neutral-atom QPUs).
2. Methodological Rigor
The experimental design is reasonable but has notable limitations:
Strengths: The vQPU experiments use a real 5-qubit IQM quantum computer (Lagrange) and production HPC infrastructure. The authors run 10 independent samples per configuration and report standard deviations. The clustering-aggregation experiments are validated on both a controlled testbed (Qluster) and a production EuroHPC Tier-0 system (Leonardo), with 32 runs on Leonardo to account for production variability.
Weaknesses: The vQPU experiments use artificially injected sleep times to vary classical-to-quantum ratios rather than naturally varying workloads. The 5-qubit QPU is extremely small, and circuit cutting is used to handle 9-qubit problems—this raises questions about scalability to realistic quantum workloads. The malleability and workflow experiments use simulated annealing as a quantum proxy rather than actual quantum hardware, which, while the authors justify this, limits the completeness of the validation. The workload scenarios are limited to a single application type per strategy (graph coloring for vQPUs, clustering aggregation for the others), making generalizability unclear.
The comparison across strategies is conceptual rather than empirical—the three strategies are never directly compared on the same workload, which weakens the claimed complementarity analysis. The reported resource savings (up to 45.7% and 64%) are specific to particular configurations and may not generalize broadly.
3. Potential Impact
Practical relevance: The paper addresses a genuinely pressing problem. As quantum computers are installed alongside HPC systems (e.g., in EuroHPC facilities), efficient scheduling of scarce QPU resources becomes critical. The pragmatic design philosophy—augmenting existing HPC infrastructure rather than replacing it—is well-aligned with real operational constraints.
Near-term applicability: The vQPU approach using Slurm GRES or license mechanisms is immediately deployable. The DMR-based malleability requires minimal code changes. The workflow approach requires restructuring applications but offers the greatest resource savings.
Broader influence: The paper provides a useful framework for HPC center operators deciding how to integrate quantum resources. The taxonomy and positioning along the two-axis framework (Figure 1) could become a reference for the community.
However, the impact is constrained by the current state of quantum computing—the problems being solved are toy-scale, and the scheduling gains are largely about managing idle time during quantum offloading, which may become less relevant as quantum hardware matures and tighter integration modes become feasible.
4. Timeliness & Relevance
The paper is highly timely. Multiple EuroHPC centers are actively integrating quantum hardware, and the scheduling challenge is recognized as a critical bottleneck. The paper explicitly cites the EuroHPC Joint Undertaking and positions itself within this institutional ecosystem. The claim of being "the first practical demonstrations of HPC-Quantum hybrid jobs fully executed on hardware deployed in Italy" adds contextual significance.
The focus on near-term, NISQ-era challenges is appropriate—error-corrected quantum computing is explicitly excluded from scope. The three strategies address real gaps in current HPC-QC software stacks.
5. Strengths & Limitations
Key Strengths:
Notable Weaknesses:
Additional Observations
The paper involves a large multi-institutional team (10+ affiliations), which demonstrates collaborative breadth but also suggests the work aggregates multiple ongoing efforts rather than presenting a cohesive, deeply investigated single contribution. The paper would benefit from a formal cost model that could predict optimal strategy selection based on workload characteristics, rather than relying solely on empirical observations from specific test cases.
Generated Apr 17, 2026
Comparison History (40)
Paper 2 addresses a critical bottleneck in quantum computing: the integration of QPUs with classical HPC systems. Its proposed scheduling strategies offer broad, immediate real-world applicability for hybrid workloads and significantly improve resource utilization. While Paper 1 is innovative in quantum optics, it solves a more specialized hardware challenge, giving it a narrower potential impact compared to the broad system-level advances presented in Paper 2.
Paper 1 presents a novel theoretical result proving that quantum walk characteristic polynomials can distinguish all strongly regular graphs of prime order, yielding a polynomial-time graph isomorphism algorithm for this class—a significant advance in algebraic graph theory and computational complexity. This connects quantum walks to a longstanding problem (graph isomorphism) with deep mathematical implications. Paper 2 addresses practical scheduling for hybrid HPC-quantum systems, which is useful engineering work but incremental in nature, with findings that are more operational than foundational. Paper 1's novelty, mathematical depth, and cross-field impact (graph theory, quantum computing, complexity theory) give it higher potential scientific impact.
Paper 1 has higher likely scientific impact due to clear theoretical novelty (IQP solving 2-Forrelation with minimal queries), resolution of a recent open question, and strengthened oracle separations (implicating core complexity-theoretic boundaries like BQP vs PH). Its results can influence multiple subareas (restricted quantum models, quantum advantage, Fourier analysis/Boolean functions). Paper 2 is timely and practically relevant for hybrid HPC–QPU operations, but its contributions are more engineering-incremental and likely narrower in long-term foundational impact.
Paper 2 presents fundamental theoretical advances in quantum circuit complexity with tight characterizations and novel constant-depth constructions for Dicke states and arbitrary symmetric states. It resolves open questions about QAC^0 capabilities, provides results directly applicable to NISQ-era hardware (e.g., trapped ions), and connects to practical applications like Decoded Quantum Interferometry. Paper 1 addresses important but more incremental engineering challenges in HPC-QC scheduling. Paper 2's theoretical contributions are more broadly impactful, establishing foundational results that will influence quantum computing theory and practice for years.
Paper 1 is likely to have higher impact due to strong timeliness and direct applicability: hybrid QC–HPC integration is an immediate bottleneck for real deployments, and the paper provides experimentally validated scheduling strategies on production clusters and real QPUs with sizable resource savings. Its results can influence system software, resource managers, and operational policies across many quantum-accelerated facilities, giving broad cross-field impact (HPC, quantum computing, scheduling). Paper 2 is novel for metrology-oriented variational state prep under noise, but its impact is narrower and appears more limited in demonstrated scale.
Paper 2 introduces a novel theoretical framework connecting quantum Fourier transforms with equivariant quantum neural networks, providing constructive characterizations and trainability guarantees (ruling out barren plateaus). This addresses fundamental challenges in quantum machine learning—a rapidly growing field. The mathematical rigor (proving equivariance characterizations and gradient bounds) and the bridging of symmetry principles from classical CNNs to quantum architectures give it broader theoretical impact. Paper 1, while practically useful for hybrid HPC-QC scheduling, is more incremental and engineering-focused with narrower applicability to current quantum computing infrastructure.
Paper 1 offers a more novel, technically deep contribution: an exact, symbolic, memory-efficient universal simulator tailored to QEC, enabling maximum-likelihood decoding, exact logical error-rate expressions, and fault-distance verification—capabilities likely to influence both QEC research and quantum software tooling broadly. Its methodological innovation (auxiliary-qubit representation, symbolic probabilities) and open-source release increase uptake potential. Paper 2 is timely and practically relevant for hybrid QC-HPC operations, but its scheduling strategies are more incremental/engineering-oriented with narrower cross-field scientific novelty and less fundamental methodological advancement.
Paper 2 addresses a critical, timely bottleneck in quantum computing: integrating QPUs with existing HPC infrastructure. Its practical, experimentally validated scheduling strategies offer broad, immediate real-world applications across the rapidly growing quantum-HPC field. In contrast, Paper 1 presents highly theoretical work focused on a narrower subfield of optomechanics, which, while rigorous, has a more limited and longer-term scope for widespread scientific impact.
Paper 2 has higher likely scientific impact due to strong real-world applicability and timeliness: efficient QPU sharing and hybrid QC-HPC scheduling is an immediate bottleneck for deployed quantum hardware. It reports experimental validation on production HPC clusters and real QPUs, suggesting methodological maturity and near-term adoption potential across computing centers, systems software, and quantum workflows. Paper 1 is novel and theoretically interesting for quantum sensing, but its impact is narrower (open quantum systems/thermometry) and may depend on specific solid-state implementations and experimental feasibility.
Paper 2 has higher likely scientific impact due to actionable, experimentally validated scheduling strategies on production HPC clusters and real quantum hardware, directly addressing an immediate bottleneck (scarce QPUs and hybrid orchestration). It offers clear performance/utilization gains and guidance for deploying quantum-accelerated facilities, with relevance to HPC operations, systems software, and quantum workflows. Paper 1 is a rigorous, potentially influential review bridging UQ and QC, but as a survey it is less immediately transformative and provides fewer concrete, validated methods or deployable outcomes.
Paper 2 addresses fault-tolerant quantum computation and error correction, which are the fundamental bottlenecks for scalable quantum computing. By significantly reducing space/time overheads and logical error rates in lattice surgery, it tackles a core challenge for long-term quantum viability. Paper 1 offers valuable systems engineering solutions for integrating QPUs into HPCs, but Paper 2's focus on error correction optimization has a deeper, more transformative impact on the realization of practical, large-scale quantum computers.
Paper 2 likely has higher scientific impact: it provides a large-scale, generalizable empirical dataset (394 confirmed bugs across 12 simulators) and actionable taxonomy/results that can influence testing, tool design, and reliability practices across much of the quantum software stack. Its findings are broadly relevant to researchers and practitioners relying on simulators as “ground truth,” making it timely and cross-cutting. Paper 1 is practically valuable for hybrid HPC-QPU operations, but its impact is more domain-specific and dependent on near-term availability/deployment of scarce QPU resources and particular scheduling environments.
Paper 1 addresses a broadly relevant and timely problem—scheduling strategies for hybrid quantum-HPC systems—with practical validation on production hardware and clear, quantifiable improvements (up to 64% resource reduction). It has wide applicability as quantum-HPC integration is a major infrastructure challenge affecting many fields. Paper 2 presents a narrow hardware validation of a specific framework (DAGI) with a small-scale experiment (4-qubit registers), limited practical applicability, and incremental contributions to information-theoretic analysis. Paper 1's systemic impact on quantum computing infrastructure gives it significantly broader and more lasting influence.
Paper 2 addresses a foundational challenge in quantum computing—error mitigation—by establishing fundamental physical limits and optimal protocols under energy constraints. While Paper 1 offers highly practical engineering solutions for hybrid HPC-QC systems, Paper 2's theoretical breakthroughs provide fundamental insights into quantum state distillation that will likely have a deeper, longer-lasting scientific impact across quantum information theory and physics.
While Paper 1 offers novel theoretical insights into quantum thermodynamics and quantum batteries, Paper 2 addresses a critical, immediate bottleneck in the field: integrating QPUs into existing HPC infrastructures. By validating practical scheduling strategies for hybrid Quantum-HPC applications on real hardware, Paper 2 provides highly applicable, near-term solutions. This systems-level approach will broadly impact both computer science and quantum research by enabling efficient, scalable use of scarce quantum resources in next-generation computing facilities.
Paper 2 likely has higher impact due to greater novelty and broader cross-field relevance: it introduces a queueing-theoretic, congestion-aware adaptive control framework that tightly couples quantum-memory physics with network stability/delay/fidelity trade-offs, extending to multi-user shared resources. This targets a core bottleneck for scalable quantum networks (repeaters, entanglement distribution) with clear real-world applicability to quantum internet design and control. Paper 1 is practical and timely for near-term hybrid HPC-QPU operations, but its contributions are more incremental within scheduling systems and narrower in breadth compared to foundational network-control results.
Paper 1 presents a novel theoretical framework (loop-shaping for coherent feedback control) that addresses a fundamental challenge in quantum physics—ground-state cooling in the unresolved-sideband regime. It introduces a generalizable methodology applicable to a wide class of quantum systems, combining conceptual novelty with practical significance. Paper 2 addresses an important engineering problem (hybrid HPC-QC scheduling) but is more incremental, adapting existing HPC scheduling concepts to quantum settings. Paper 1's deeper theoretical contribution and broader applicability to quantum control give it higher long-term scientific impact.
Paper 1 addresses a critical and highly practical bottleneck in modern computing: integrating quantum processors into high-performance computing infrastructures. Its empirical validation of scheduling strategies demonstrates significant resource savings, offering immediate real-world utility for scalable hybrid systems. In contrast, Paper 2 provides a highly specialized mathematical improvement to a trace inequality. While valuable for theoretical quantum information theory, Paper 1 has much broader potential impact across experimental physics, computer science, and industry applications.
Paper 2 presents a novel fundamental physics result—unconventional photon blockade in a simple symmetric configuration—with clear advantages over prior schemes (no post-fabrication trimming, standard detector compatibility, disorder tolerance). This has broad implications for quantum photonics, single-photon sources, and quantum information processing. Paper 1 addresses practical scheduling for hybrid HPC-QC systems, which is timely but more incremental and narrower in scope, essentially comparing engineering strategies rather than introducing fundamentally new concepts. Paper 2's potential to enable new quantum photonic devices gives it broader and deeper scientific impact.
Paper 1 addresses a fundamental theoretical problem—deriving consistent classical-quantum dynamics from first principles using open quantum systems theory. This has broad implications across quantum foundations, quantum gravity (semiclassical gravity), quantum chemistry, and quantum information. It offers conceptual clarity on a long-standing issue (the classical-quantum divide) with potential to reshape semiclassical methods across multiple fields. Paper 2, while practically useful for hybrid HPC-QC scheduling, addresses a more niche engineering problem whose relevance is tied to current-generation quantum hardware constraints and may become obsolete as the technology evolves.