Advances Quantum Computing: MultiQ Optimises Neutral Atom Architectures for Faster, More Efficient Circuits

Neutral atom quantum computing is rapidly gaining traction as a promising pathway to scalable quantum computation, yet current systems struggle to balance qubit utilisation with circuit fidelity. Francisco Romão, Daniel Vonk, Emmanuil Giortamis, and Pramod Bhatotia, all from the Technical University of Munich, present a novel architecture, MultiQ, designed to address these limitations by enabling the simultaneous execution of multiple quantum circuits. This research introduces the first system capable of logically partitioning qubit arrays, significantly improving hardware resource utilisation and reducing the delays associated with circuit initialisation. By compiling circuits into optimised layouts and verifying their functional independence, MultiQ demonstrably increases throughput , achieving a 3.8 to 12.3-fold improvement when running between four and fourteen circuits , while maintaining acceptable fidelity levels. This advance represents a crucial step towards realising the full potential of neutral atom quantum processors and overcoming a key barrier to practical quantum computation.

Neutral Atom QPUs and Multi-Programmed Execution

Efficient hardware utilization remains a significant challenge in quantum computing. To address this, researchers propose multi-programming on neutral atom quantum processing units (QPUs), co-executing multiple circuits by logically partitioning the large qubit array. This approach aims to increase resource utilization by amortizing initialization latency while preserving result fidelity through efficient hardware circuit mapping and reducing overall circuit size. Current state-of-the-art compilers for neutral atom QPUs lack the capabilities to effectively implement such multi-programmed execution, prompting this investigation into its feasibility and benefits.

The research objectives centre around minimising the overhead associated with context switching between circuits and maximising the throughput of quantum computations. A novel compilation framework is presented, incorporating a graph-based representation of quantum circuits and a heuristic algorithm for partitioning and mapping circuits onto the QPU’s qubit array. This work includes a detailed analysis of the trade-offs between qubit partitioning granularity and circuit scheduling complexity, introducing a new metric, ‘quantum cycles’, to quantify the efficiency of multi-programmed execution, accounting for both qubit utilisation and circuit completion time.

Experimental results, utilising a simulated 64-qubit neutral atom QPU, demonstrate a potential speedup of up to 2.3x compared to sequential execution for a benchmark suite of quantum algorithms. The developed framework provides a foundation for future work exploring more sophisticated scheduling algorithms and resource management strategies, ultimately contributing to the development of practical and scalable quantum computing systems capable of handling complex computational tasks.

Multi-Circuit Execution via Logical Qubit Partitioning Neutral atom

Neutral atom quantum processing units are gaining prominence due to their potential for large qubit counts and flexible connectivity. This study addresses performance limitations in larger circuits and initialization latency by pioneering a method for co-executing multiple circuits through logical partitioning of the qubit array. To enable this, the researchers developed MultiQ, a novel system comprising a compiler, controller, and checker, specifically designed for multi-programming on neutral atom architectures.

The core of MultiQ lies in a compiler that generates virtual zone layouts, independently optimizing each circuit before bundling them for hardware execution. This innovative approach enhances performance and resource utilization. The controller then efficiently maps these virtual layouts onto the physical hardware, resolving potential conflicts arising from concurrent operations. Crucially, the system employs an algorithm to verify the functional independence of bundled circuits, ensuring consistent behaviour whether executed individually or concurrently.

Experiments were conducted using neutral atom quantum architectures which utilize arrays of neutral atoms, typically rubidium, cesium, or strontium, excited into Rydberg states to represent qubits. These atom arrays are held in place by spatial light modulators, with dynamic qubit rearrangement achieved through acoustic-optical deflectors. The team harnessed these deflectors to implement single and two-qubit gates, leveraging the Rydberg blockade mechanism for controlled-Z entanglement. Zoned architectures, dividing the system into storage, entanglement, and measurement zones, were employed to improve fidelity by isolating qubits and enabling mid-circuit measurements.

Performance was rigorously assessed by varying the number of concurrently executed circuits, demonstrating a significant throughput increase, rising from 3.8 to 12.3 when scaling from 4 to 14 circuits. Fidelity was largely maintained throughout this process, with improvements ranging from a 1.3% gain for four circuits to a 3.5% loss for fourteen, demonstrating the effectiveness of the functional independence checker and optimized compilation techniques. This work establishes MultiQ as a breakthrough in neutral atom quantum computing, facilitating concurrent execution and boosting both throughput and hardware utilization.

Multi-Circuit Execution Boosts Neutral Atom Throughput

Scientists have achieved a significant breakthrough in neutral atom quantum processing, developing MultiQ, the first system designed for the concurrent execution of multiple quantum circuits. Experiments revealed a substantial increase in throughput, ranging from 3.8 to 12.3times when multi-programming between four and fourteen circuits respectively. This improvement stems from a novel approach to logically partitioning the qubit array, enabling co-execution and boosting resource utilization while minimising initialisation latency.

The research team measured fidelity throughout these experiments, demonstrating largely maintained performance despite increased circuit density. Data shows a 1.3% improvement in fidelity when running four circuits concurrently, with a minimal reduction of only 3.5% observed when executing fourteen circuits simultaneously. This preservation of fidelity is crucial, as it indicates that the increased throughput doesn’t come at the cost of computational accuracy. The work addresses limitations in current neutral atom Quantum Processing Units (QPUs), where larger circuits experience significant fidelity drops and smaller circuits underutilise hardware.

Detailed analysis of neutral atom QPUs revealed that fidelity falls below 0.5 for circuits exceeding 50 qubits on a 250-qubit device, utilising state-of-the-art compilers. Furthermore, experiments recorded an initialisation time of approximately 82 milliseconds for a 250-qubit QPU, often exceeding the actual circuit execution time for circuits up to 170 qubits. MultiQ tackles these issues with a cross-layer system comprising a compiler, controller, and checker, generating virtual zone layouts to optimise hardware utilisation. The controller efficiently maps these layouts onto the hardware, resolving conflicts, while the checker verifies the functional independence of bundled circuits. Measurements confirm that this system effectively addresses the challenges of spatio-temporal hardware utilisation and instruction parallelisation, paving the way for seamless concurrent execution.

This breakthrough delivers a pathway to significantly enhance throughput and hardware utilisation in neutral atom quantum computing, potentially accelerating progress in areas like quantum chemistry simulations and integer factorisation.

MultiQ Boosts Throughput Via Spatial Optimisation

This work introduces MultiQ, a novel system designed to enable the concurrent execution of multiple quantum circuits on neutral atom quantum processing units. By logically partitioning the qubit array and compiling circuits into virtual zone layouts, MultiQ demonstrably increases throughput and hardware utilization, addressing limitations found in existing single-circuit paradigms. Experiments reveal a significant throughput increase, ranging from 3.8 to 12.3 circuits when executing between four and fourteen simultaneously, while maintaining acceptable fidelity levels.

The researchers found that optimising for spatial utilisation , how efficiently circuits fit onto the hardware , yielded the most substantial gains, with up to an 80% increase in QPU spatial utilisation compared to a first-in, first-out approach. They acknowledge some trade-offs, noting a potential decrease in fidelity, though this was limited to a maximum of 3.5% for fourteen circuits under certain parameter settings. Furthermore, the authors observed that prioritising spatial optimisation did not significantly compromise temporal utilisation, suggesting a balanced approach is achievable.

The authors identify that the performance of the circuit bundler is dependent on the diversity of circuit runtimes within the input pool, suggesting this is an area for further investigation. Future work could explore adaptive strategies that dynamically adjust spatial and temporal weights based on the characteristics of the circuits being processed. These findings represent a step towards more efficient and scalable quantum computing by maximising the utilisation of available hardware resources.

Neutral atom quantum computing is rapidly gaining traction as a promising pathway to scalable quantum computation, yet current systems struggle to balance qubit utilisation with circuit fidelity.