Quantum Workflows Now Bypass Lengthy Delays with New Scheduling System

Sinan Pehlivanoglu and colleagues at the Indiana University Bloomington in collaboration with University of Notre Dame, have created Qurator, a new task scheduler that optimises both queue times and circuit fidelity when using diverse quantum cloud providers. The architecture-agnostic system models complex hybrid workflows, accounting for uniquely quantum factors such as entanglement and synchronisation, which traditional scheduling methods cannot address. By unifying calibration data from providers including IBM, IonQ, IQM, Rigetti, AQT, and QuEra, Qurator estimates fidelity and demonstrates reductions in queue times of between 30 and 75 percent under heavy load, while maintaining performance comparable to the highest-fidelity baseline at lower loads.

Substantial Queue Time Reduction via Dynamic Quantum Workload Scheduling

Across workloads of 5,000 to 35,000 tasks, Qurator achieves a 30 to 75 per cent reduction in queue time, a strong improvement over existing methods which struggled to deliver such gains at scale. This performance leap crosses a key threshold for practical quantum computation, enabling workflows previously hampered by excessive waiting times; circuits that once faced queues lasting minutes or days can now begin executing much faster. Qurator’s success stems from its new approach to modelling hybrid quantum-classical workloads as dynamic Directed Acyclic Graphs, or DAGs, which represent task dependencies, allowing for efficient scheduling across diverse quantum cloud providers. Maintaining within one per cent of the highest-fidelity baseline when quantum workload is low, Qurator demonstrates its ability to preserve performance under ideal conditions. The system successfully estimates fidelity using a logarithmic success score, reconciling differing calibration data, including one and two-qubit gate errors, readout fidelity, and decoherence times, into standardised variables; for example, IBM provides per-qubit calibration data, while QuEra requires accounting for atom loss during pulse sequences.

Dynamic Dependency Graph Scheduling for Joint Optimisation of Queue Time and Fidelity

Qurator represents complex workflows as dynamic DAGs, a technique similar to a flowchart where tasks depend on each other, much like a recipe requiring ingredients to be mixed before baking. This approach jointly optimises both queue time and circuit fidelity when distributing tasks across different quantum computers, moving beyond treating them as separate problems. The system accounts for uniquely quantum phenomena such as entanglement dependencies, where two quantum bits are linked, instantly revealing information about each other, thus requiring careful coordination. Qurator was evaluated using circuits from the Munich Quantum Toolkit benchmark suite and a simulator mirroring eleven real quantum devices, including those with qubit counts ranging from 12, as found in AQT’s IBEX Q1, to 156, as in IBM’s Heron processor. Incorporating four months of real queue data, the simulator tested loads from 5 to 35,000 tasks, and used calibration data to estimate circuit fidelity and run times.

Joint optimisation of queue times and circuit reliability in remote quantum computing

Researchers at institutions including the University of Edinburgh and Oxford Ionics are striving to integrate quantum computers into existing high-performance computing infrastructure, but accessing these shared, remote devices often means enduring substantial delays. Qurator tackles this bottleneck by jointly optimising both queue times and the reliability of quantum circuits, a feat previously attempted as separate challenges. The evaluation relies on simulating real-world queue data, prompting consideration of how accurately a simulator can predict performance when faced with the unpredictable fluctuations of live quantum hardware and the inevitable variations in cloud provider systems.

Despite currently demonstrating its efficacy via simulation, Qurator’s potential remains significant for near-term quantum computing development. Real quantum hardware introduces unpredictable variations, with queue lengths and device stability fluctuating constantly, making precise prediction difficult. However, this establishes a key framework for optimising quantum workflows, addressing both speed and reliability simultaneously. By simultaneously optimising queue times and circuit fidelity, Qurator establishes a new approach to managing quantum computing workflows, a departure from previous methods that treated these as separate goals. Accounting for uniquely quantum phenomena, such as entanglement where linked quantum bits instantly share information, Qurator unlocks more efficient resource allocation.

Qurator, a new task scheduler, successfully optimised both queue times and circuit fidelity for quantum computing tasks. This matters because accessing remote quantum computers often involves significant waiting times, hindering efficient use of these resources. The system achieved performance within 1% of the highest-fidelity baseline under low load, while reducing queue times by 30-75% across simulations using up to 35,000 tasks and devices with qubit counts ranging from 12 to 156. Researchers evaluated Qurator using four months of real queue data and calibration data from six quantum providers.