Quantum Circuit Pruning Achieves 47.7% Fidelity Gain Via Smart Approximation

Scientists are tackling the critical challenge of improving quantum computation fidelity on today’s noisy quantum devices. Pau Escofet, Santiago Rodrigo, and Rohit Sarma Sarkar, from Universitat Politècnica de Catalunya, alongside Carmen G. Almudéver (Universitat Politècnica de València) and Eduard Alarcón and Sergi Abadal, demonstrate a novel circuit pruning strategy that intelligently reduces errors by approximating quantum circuits with routing costs in mind. Their research introduces a method for removing unnecessary quantum gates , specifically, small-angle controlled rotations , thereby minimising the fidelity loss caused by the SWAP operations needed during circuit compilation. Simulations reveal significant reductions in two-qubit gate counts (up to 48.6%) and, crucially, improvements in final state fidelity (up to 47.7%), suggesting this approach could be vital for scaling up near-term quantum algorithms.

Routing-aware pruning boosts NISQ circuit fidelity

Scientists have unveiled a novel routing-aware pruning strategy designed to enhance the fidelity of quantum circuits executed on Noisy Intermediate-Scale Quantum (NISQ) devices. The research team at Universitat Politècnica de Catalunya and Universitat Politècnica de València developed a method to selectively remove parametric controlled rotations, quantum gates, when their small rotation angles do not justify the routing overhead required for their implementation. This innovative approach directly addresses fidelity loss stemming from the additional SWAP operations inevitably introduced during the compilation process, a critical limitation of current quantum hardware. By carefully evaluating whether executing a gate leads to a greater fidelity loss than simply omitting it, the scientists have demonstrated a significant improvement in quantum computation reliability.
The core of this breakthrough lies in a meticulous assessment of the trade-off between gate execution and routing costs. Researchers established a methodology for pruning gates based on the fidelity impact of routing quantum states, comparing it to the potential loss incurred by omitting small-angle parametric controlled rotations. They defined the fidelity of a θ-rotation as the minimum change over all quantum states, utilising a quantum fidelity calculation to quantify the impact of rotations. Simultaneously, the team modelled the fidelity loss associated with SWAP gates required for qubit routing, accounting for error rates and the physical distance between qubits, approximating the number of SWAP gates needed for interaction as 1.25times the distance between qubits.

This detailed analysis allows for a precise determination of whether a gate’s contribution to the overall circuit fidelity outweighs the cost of implementing it. Experiments conducted on benchmark circuits, compiled for grid-based architectures and incorporating a realistic noise model, reveal substantial performance gains. The study demonstrates a reduction in two-qubit gate counts of up to 48.6%, coupled with an impressive improvement in final state fidelity reaching up to 47.7%. Notably, these enhancements are particularly pronounced in larger circuits where routing costs become increasingly dominant, highlighting the scalability of the proposed method.

The team’s simulations confirm that integrating pruning decisions directly into the compilation flow, specifically targeting small-angle rotations with high routing overhead, yields superior results compared to traditional pre-routing pruning strategies. This work establishes a practical and scalable approach to boosting computation reliability on NISQ hardware. The research team’s methodology considers the fidelity of a θ-rotation, defined by the minimum change over all quantum states, and compares it to the fidelity loss from imperfect routing operations, calculated using a model incorporating error rates and qubit distances. By focusing on parametric controlled rotations, fundamental building blocks for Variational Quantum Algorithms and Quantum Machine Learning, the scientists have identified a key area for optimisation. The findings suggest that this routing-aware pruning strategy could significantly accelerate progress towards practical quantum computation, paving the way for more complex and reliable quantum algorithms .

Pruning Small-Angle Gates to Reduce SWAP Errors

Scientists developed a routing-aware pruning strategy to enhance fidelity in quantum circuits executed on Noisy Intermediate-Scale Quantum (NISQ) devices. The research team engineered a method to selectively remove parametric controlled rotations, focusing on those with small rotation angles that introduce significant routing overhead during implementation. This innovative approach directly addresses fidelity loss stemming from the additional SWAP operations required during circuit compilation, a major limitation of current quantum hardware. Experiments employed a detailed evaluation process, assessing whether executing a specific gate would lead to a greater reduction in fidelity than simply omitting it.

The study pioneered a technique that integrates pruning decisions directly into the compilation flow, unlike existing methods which typically operate before routing. Researchers meticulously compiled benchmark circuits to grid-based architectures, simulating realistic noise models to accurately reflect the behaviour of NISQ devices. This setup enabled precise measurement of the impact of pruning on both gate count and final state fidelity. The team harnessed a suite of benchmark circuits, subjecting them to the pruning strategy and then analysing the resulting performance metrics. Specifically, the method quantifies the trade-off between gate removal and fidelity preservation, identifying gates where the routing cost outweighs their contribution to the quantum state.

Simulations revealed that the approach reduces two-qubit gate counts by up to 48.6%, demonstrating a substantial reduction in circuit complexity. Crucially, this pruning also improved final state fidelity by up to 47.7%, particularly in larger circuits where routing costs are most dominant. This work’s methodology achieves a significant improvement by considering the actual connectivity constraints and routing overhead, a factor often ignored in previous pruning strategies. The system delivers a practical and scalable solution for enhancing computation reliability on NISQ hardware, offering a promising pathway towards more robust quantum computations. The technique reveals that integrating pruning into the routing process, targeting small-angle rotations with high routing costs, can demonstrably improve circuit fidelity compared to pre-routing approaches.

Pruning reduces gate counts and boosts fidelity, ultimately

Scientists achieved a significant reduction in two-qubit gate counts, up to 48.6%, through a novel routing-aware pruning strategy for quantum circuits. This work introduces a method to selectively remove parametric controlled rotations with small rotation angles, mitigating fidelity loss caused by the SWAP operations required during compilation on Noisy Intermediate-Scale Quantum (NISQ) devices. The team meticulously evaluated whether executing a gate would lead to a greater fidelity loss than omitting it, demonstrating a practical approach to enhancing computation reliability. Results demonstrate that this pruning technique is particularly effective for larger circuits where routing costs become dominant, paving the way for more efficient quantum computations.

Experiments revealed that the method improves final state fidelity by up to 47.7%, a substantial gain achieved by carefully balancing gate removal with the potential for increased routing overhead. Researchers defined the fidelity of a θ-rotation as the minimum change over all quantum states, establishing a quantifiable metric for assessing the impact of gate omission. The fidelity loss associated with routing was calculated considering the error rate of each quantum gate and the distance between physical qubits, using the equation Fswap := (1 −p2) 3· dqi,qj 2 + 1 −(1 −p2) 3· dqi,qj 2 4 2, where dqi,qj represents the distance between qubits. This detailed analysis allowed for a precise determination of when pruning would actually improve overall circuit performance.

Measurements confirm that the approach effectively targets small-angle parametric controlled rotations, recognizing that their implementation can incur significant routing effort. The team quantified the fidelity of these rotations using the equation F(|ψ⟩, Rn(θ)|ψ⟩) ≥cos2 θ 2, providing a lower bound on the fidelity loss associated with a given rotation angle θ. By integrating pruning decisions directly into the compilation flow, the research circumvents the limitations of pre-routing strategies that often ignore connectivity constraints. The study approximated the number of SWAP gates required for qubit interaction as 1.25·dqi,qj, accounting for the non-optimal nature of current routing algorithms.

The breakthrough delivers a scalable solution for enhancing quantum computation on NISQ hardware, addressing the critical challenge of balancing gate complexity with routing overhead. Tests prove that the methodology maximizes overall circuit fidelity by comparing the expected fidelity loss from imperfect routing operations to the loss incurred by omitting parametric controlled rotations. This work focuses on parametric controlled rotation gates, fundamental building blocks for Variational Quantum Algorithms (VQAs) and Quantum Machine Learning (QML), where the rotation angle θ dictates the magnitude of the transformation. The team’s findings suggest that a π/6 rotation between qubits is more likely to be pruned when the distance between them is substantial, as illustrated in Figure 0.1.

Pruning reduces gates and boosts fidelity, ultimately improving

Scientists have developed a routing-aware pruning strategy designed to optimise quantum circuits for execution on Noisy Intermediate-Scale Quantum (NISQ) devices. This method selectively removes two-qubit parametric gates when their contribution to the overall computation is less than the fidelity loss caused by the routing overhead required to implement them. By carefully pruning these gates, researchers aim to reduce the impact of SWAP operations, necessary for mapping logical qubits to physical qubits, which introduce errors into the computation. The proposed approach demonstrated reductions in two-qubit gate counts of up to 48.6% and improvements in final state fidelity of up to 47.7% in simulations using benchmark circuits and realistic noise models.

These gains were particularly pronounced in larger circuits where the costs associated with routing become more significant. Notably, the technique adapts to both circuit structure and hardware connectivity without requiring pre-tuning of approximation parameters, making it a versatile tool for quantum compilation. The authors acknowledge that the current work focuses on parametric gates and doesn’t address pruning of other gate types. Future research will focus on extending the method and evaluating its performance on actual quantum processors, potentially paving the way for more reliable quantum computations. This routing-aware pruning offers a promising strategy for improving the performance of quantum algorithms on NISQ hardware, mitigating the effects of noise and resource limitations inherent in current quantum technology.