Measurement-based Dynamical Decoupling Achieves Improved Fidelity for 9-Qubit Fourier Transforms on Quantum Processors

Quantum computers promise revolutionary computational power, but maintaining the delicate quantum states necessary for calculations remains a significant challenge, as environmental noise causes rapid decoherence. Jeongwoo Jae, Changwon Lee, Juzar Thingna, and colleagues at Samsung SDS, Yonsei University, and the Institute for Basic Science now demonstrate a new technique called measurement-based dynamical decoupling that actively combats this decoherence. Their method intelligently determines the optimal control sequences from partial measurements of the quantum system, scaling efficiently with increasing complexity, and crucially, achieves the highest possible fidelity for preserving quantum entanglement. Experiments on a state-of-the-art quantum processor reveal that this approach dramatically improves the performance of complex algorithms, boosting the success rate of a multi-qubit Fourier transform and enhancing the accuracy of ground-state energy estimations, establishing a pathway towards robust and scalable quantum computation.

Measurement-Based Decoupling Preserves Qubit Fidelity

Researchers are tackling the challenge of maintaining qubit stability in increasingly large quantum processors, a problem caused by environmental noise and imperfections in control systems. This work investigates measurement-based dynamical decoupling (MBDD), a method that actively monitors and corrects qubit states through repeated measurements and feedback, effectively reducing the accumulation of errors over time. The approach works by interleaving quantum gates with projective measurements, creating a closed-loop control system that suppresses decoherence and significantly improves the performance of quantum algorithms on large-scale processors. By carefully designing the measurement process and feedback strategy, the team achieved substantial improvements in qubit coherence and gate fidelity. The method leverages principles of quantum control to optimise measurements, minimising disturbance to the quantum state while maximising information gained about the error environment, allowing for precise correction of qubit errors and extending the duration of coherent quantum computation. This practical MBDD scheme is tailored for large-scale quantum processors, effectively suppressing both low-frequency and high-frequency noise common in real-world quantum devices.

Eagle and Heron Qubit Performance Demonstrated

This research details the experimental setup and device characteristics used in quantum computations, specifically focusing on Single-Shot Quantum Error Detection (SQD) experiments. The study examined devices based on two architectures: ibm_yonsei (Eagle) and ibm_fez (Heron r2), both featuring fixed-frequency transmon qubits and utilising the controlled-Z (CZ) gate. The Eagle architecture features 35 and 56 qubits, while the Heron r2 architecture features 35 and 55 qubits with tunable couplers, which mitigate parasitic interactions and always-on ZZ crosstalk errors. Device characterisation revealed energy relaxation times (T1) ranging from approximately 90µs (Eagle, 35 qubits) to 53µs (Heron, 55 qubits), and dephasing times (T2) ranging from approximately 170µs (Eagle, 35 qubits) to 100µs (Heron, 55 qubits).

Single-qubit gate errors were generally low, ranging from 0. 015% to 0. 032%, while two-qubit gate errors were higher, ranging from 0. 42% to 2. 29%, and readout errors were significant, ranging from 1.12% to 12. 10%. The Heron architecture’s tunable couplers and faster gate times represent key improvements, with the data highlighting that readout errors are the most significant source of error in these experiments.

Scalable Dynamical Decoupling Boosts Quantum Algorithms

This research introduces a measurement-based dynamical decoupling (MDD) protocol, a technique designed to preserve the performance of quantum algorithms by actively suppressing decoherence. The team demonstrated that MDD optimally protects qubits from local decoherence, achieving the maximum entanglement fidelity possible with current control operations. Experiments conducted on the 127-qubit IBM Eagle processor yielded significant improvements, with up to a 450-fold enhancement in the success probability of a 14-qubit Fourier transform and improved accuracy in ground-state energy estimation for molecular simulations. These results establish MDD as a scalable and efficient method for mitigating noise in large-scale quantum algorithms, even on hardware with limited coherence times. Notably, the team showed that MDD enabled the Eagle processor to achieve accuracy comparable to the more advanced Heron device, highlighting its ability to extract near state-of-the-art performance from existing hardware. Future work will focus on adapting MDD to address a wider range of noise types and further optimise its performance in complex quantum computations, potentially combining it with other error mitigation techniques to achieve even greater levels of accuracy and reliability.