In a paper published in the journal Nature, researchers demonstrated error-mitigated dynamic circuits using 142 qubits across two quantum processing units linked by a real-time classical connection. They showed that quantum gates could be controlled based on mid-circuit measurements, enhancing qubit connectivity. This approach allowed multiple processors to function as unified, improving hardware versatility. The work marked a significant step in advancing quantum computing capabilities.
Study: Combining quantum processors with real-time classical communication. Image Credit: UladzimirZuyeu/Shutterstock.com
Related Work
Past work addressed scalability and connectivity challenges in noisy quantum processors by exploring modular architectures and virtual gates for long-range interactions. Researchers compared local operations (LO) and LO with classical communication (LOCC), finding LOCC more efficient due to its use of teleportation circuits with virtual Bell pairs. While LO was simpler, LOCC reduced compilation costs and improved performance. These advancements aimed to enhance quantum processor utility.
Efficient Modular Quantum Computing
Circuit cutting enables the execution of large quantum circuits beyond hardware limitations by decomposing them into subcircuits, which are individually run and classically recombined. This approach increases the sampling overhead but allows for flexibility in creating virtual gates.
In this work, cut bell pairs are engineered through parameterized quantum circuits, forming a "cut bell pair factory." These pairs enable teleportation circuits, facilitating operations like controlled-NOT (CNOT) gates on long-range qubits while minimizing overhead and improving scalability.
Researchers implemented graph states with periodic boundary conditions on an International Business Machines Corporation (IBM) eagle processor to overcome physical connectivity constraints using SWAP gates, LOCC, and LO. The SWAP-based approach added substantial error due to increased CNOT gates, while LOCC and LO showed significantly lower errors, demonstrating their efficiency in long-range operations. The results also highlighted that LOCC achieved a 99% confidence in detecting bipartite entanglement across all edges, surpassing SWAP in error metrics and stabilizer quality.
Additional experiments integrated two 127-qubit Eagle processors into a single 254-qubit quantum processor using a real-time classical link, facilitating the execution of dynamic quantum circuits. A 134-qubit graph state with periodic boundaries was created, implementing long-range gates with LO and LOCC. Both approaches demonstrated high stabilizer quality and reduced error rates compared to benchmarks without long-range gates. While LOCC required additional resources like cut bell pairs, it maintained entanglement across all edges, showcasing the power of dynamic circuits in modular architectures.
Overall, LO and LOCC approaches allowed efficient quantum circuit execution across modular processors. They mitigated connectivity and resource constraints, outperforming traditional methods like SWAP gates. By leveraging virtual gates and real-time classical links, this work opens pathways for scalable and error-mitigated quantum computing with modular quantum processors.
Quantum Circuit Cutting Methodology
This section provides a detailed methodology for quantum circuit cutting, focusing on techniques like quantum process decomposition (QPD) and its implementation using LO and LOCC. For experimental feasibility, circuit cutting is introduced to partition quantum channels into simpler components. By expressing a quantum channel as a sum of multiple channels with specific coefficients, the QPD framework allows unbiased estimation of observables but incurs a sampling overhead, necessitating careful optimization.
Implementing virtual gates using LO involves decomposing controlled-Z (CZ) gates into six weighted circuits built from LO, enabling probabilistic estimation. In practice, QPDs are often fully enumerated, minimizing sampling costs. The optimal decomposition of a single gate has an overhead of 9. LOCC extends this approach, utilizing dynamic circuits with error mitigation techniques such as dynamical decoupling and zero-noise extrapolation to address latency-induced errors during feed-forward operations.
To create entangled Bell pairs essential for LOCC, QPDs are employed with a hierarchy of probabilistic mixtures, incorporating variational quantum circuits for efficiency. This method ensures separability within qubit partitions while achieving target entanglement. Hardware-dependent factors, such as qubit relaxation and measurement fidelity, influence the success of these operations, emphasizing the need for pre-experimental benchmarking to select suitable qubits.
Benchmarking strategies, such as evaluating stabilizer measurements on smaller qubit chains, offer valuable insights into device-specific performance metrics. These methodologies collectively enable scalable quantum experiments, including large-scale graph state implementations, leveraging advanced circuit design and error mitigation techniques.
Conclusion
To sum up, the implementation of long-range gates with LO and LOCC enabled the creation of a graph state on 134 qubits, surpassing the capacity of a single chip. Circuit cutting, while increasing variance, was controlled through statistical tests. The research focused on reducing sampling overhead and proposed alternative methods like entanglement distribution across chips.
This approach could benefit applications like Hamiltonian simulation and measurement-based quantum computing. Error-mitigated dynamic circuits addressed noise sources, enabling scalable quantum processors. Overall, the work demonstrated the feasibility of using multiple quantum processors.
Journal Reference
Carrera Vazquez, A., et al. (2024). Combining quantum processors with real-time classical communication. Nature, 1-5. DOI: 10.1038/s41586-024-08178-2, https://www.nature.com/articles/s41586-024-08178-2
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.