Distributing quantum workloads over many Quantum Processing Units (QPUs) is a crucial step in scaling up quantum computers toward practical quantum advantage due to the limitations in size of a single QPU. In the absence of high-fidelity quantum interconnects, circuit knitting could provide a path to computing certain properties of large quantum systems on many QPUs of limited size in a distributed fashion using only classical communication. Circuit knitting partitions large quantum circuits into manageable sub-circuits, however, reconstructing observables in a straightforward manner comes at an exponential cost in sampling and classical post-processing. To mitigate the overhead this technique incurs, we introduce an Adaptive Circuit Knitting (ACK) method that finds efficient partitions of quantum circuits by discovering regions of minimal entanglement between subsystems. We simulate 1D and 2D disordered mixed-field Ising models up to 60 qubits and show that the ACK approach can reduce circuit knitting sampling overheads by up to four orders of magnitude for observables of interest. We highlight our parallel GPU-accelerated implementation and discuss the need for efficient classical simulators to enable distributed quantum algorithm development. Our techniques could enable efficient distribution of quantum simulation for both near-term and fault-tolerant architectures.
Distributing quantum workloads over many Quantum Processing Units (QPUs) is a crucial step in scaling up quantum computers toward practical quantum advantage due to the limitations in size of a single QPU. In the absence of high-fidelity quantum interconnects, circuit knitting could provide a path to computing certain properties of large quantum systems on many QPUs of limited size in a distributed fashion using only classical communication. Circuit knitting partitions large quantum circuits into manageable sub-circuits, however, reconstructing observables in a straightforward manner comes at an exponential cost in sampling and classical post-processing. To mitigate the overhead this technique incurs, we introduce an Adaptive Circuit Knitting (ACK) method that finds efficient partitions of quantum circuits by discovering regions of minimal entanglement between subsystems. We simulate 1D and 2D disordered mixed-field Ising models up to 60 qubits and show that the ACK approach can reduce circuit knitting sampling overheads by up to four orders of magnitude for observables of interest. We highlight our parallel GPU-accelerated implementation and discuss the need for efficient classical simulators to enable distributed quantum algorithm development. Our techniques could enable efficient distribution of quantum simulation for both near-term and fault-tolerant architectures.
Circuit knitting, a method for connecting quantum circuits across multiple processors to simulate nonlocal quantum operations, is a promising approach for distributed quantum computing. While various techniques have been developed for circuit knitting, we uncover fundamental limitations to the scalability of this technology. We prove that the sampling overhead of circuit knitting is exponentially lower bounded by the exact entanglement cost of the target bipartite dynamic, even with the asymptotic amount of resources using the parallel cut strategy. Specifically, we prove that the regularized sampling overhead assisted with local operations and classical communication of any bipartite quantum channel is lower bounded by the exponential of its exact entanglement cost under separable preserving operations. Furthermore, we develop the faithful lower bounds for the regularized sampling overhead based on channels' $\ensuremath{\kappa}$ entanglement and max-Rains information, providing efficiently computable benchmarks. Our work reveals a profound connection between virtual quantum information processing via quasiprobability decomposition and quantum Shannon theory, highlighting the critical role of entanglement in distributed quantum computing.
Abstract As quantum computing rapidly evolves, scaling quantum algorithms to utilize multiple quantum processing units (QPUs) becomes crucial for overcoming the limitations of current noisy intermediate-scale quantum (NISQ) devices. This paper focuses on distributed quantum computing (DQC), specifically targeting the challenges associated with circuit cutting and circuit distribution in multi-QPU environments. By leveraging techniques such as hypergraph partitioning and threading models, this paper presents alternative strategies for dividing and managing both static and adaptive quantum circuits across multiple QPUs. A central research question is how the partitioning and distribution of quantum circuits can be optimized to minimize communication overhead and maximize computational performance in multi-QPU systems, for static and adaptive quantum circuits. To answer this question, a hypergraph partitioning method is proposed to effectively segment large static quantum circuits into manageable subcircuits, ensuring minimal inter-QPU gate operations and reduced time for the partitioning process. Additionally, a co-design threading model is presented, tailored for adaptive quantum circuits. This category of circuits can dynamically adjust their sequence of gates in runtime, based on intermediate measurements and classical control flows, creating unique challenges for circuit partition. Finally, to support the coordination between circuit partitions with static and adaptive circuits, we propose a quantum resource manager (QRM) architecture, bridging the gap between partitioning techniques and practical coordination for a scalable quantum computing system with multi-QPU architecture.
Circuit knitting, a method for connecting quantum circuits across multiple processors to simulate nonlocal quantum operations, is a promising approach for distributed quantum computing. While various techniques have been developed for circuit knitting, we uncover fundamental limitations to the scalability of this technology. We prove that the sampling overhead of circuit knitting is exponentially lower bounded by the exact entanglement cost of the target bipartite dynamic, even for asymptotic overhead in the parallel cut regime. Specifically, we prove that the regularized sampling overhead assisted with local operations and classical communication (LOCC), of any bipartite quantum channel is lower bounded by the exponential of its exact entanglement cost under separable preserving operations. Furthermore, we show that the regularized sampling overhead for simulating a general bipartite channel via LOCC is lower bounded by κ-entanglement and max-Rains information, providing efficiently computable benchmarks. Our work reveals a profound connection between virtual quantum information processing via quasi-probability decomposition and quantum Shannon theory, highlighting the critical role of entanglement in distributed quantum computing.
A quantum processing unit (QPU) must contain a large number of high quality qubits to produce accurate results for problems at useful scales. In contrast, most scientific and industry classical computation workloads happen in parallel on distributed systems, which rely on copying data across multiple cores. Unfortunately, copying quantum data is theoretically prohibited due to the quantum non-cloning theory. Instead, quantum circuit cutting techniques cut a large quantum circuit into multiple smaller subcircuits, distribute the subcircuits on parallel QPUs and reconstruct the results with classical computing. Such techniques make distributed hybrid quantum computing (DHQC) a possibility but also introduce an exponential classical co-processing cost in the number of cuts and easily become intractable. This paper presents TensorQC, which leverages classical tensor networks to bring an exponential runtime advantage over state-of-the-art parallelization post-processing techniques. As a result, this paper demonstrates running benchmarks that are otherwise intractable for a standalone QPU and prior circuit cutting techniques. Specifically, this paper runs six realistic benchmarks using QPUs available nowadays and a single GPU, and reduces the QPU size and quality requirements by more than 10× over purely quantum platforms.
Today's quantum processors have tens to hundreds of physical qubits, but reliable execution of arbitrary circuits remains limited to fewer than 30 entangled qubits across hardware modalities. Building upon prior work, we introduce an error- and topology-aware method for mapping multiple independent circuits onto disjoint regions of a single large-scale QPU for parallel circuit execution. For applications with many similarly sized circuits, such as observable estimation for Hamiltonian simulation, this approach can reduce billed QPU execution time, with ideal speedup proportional to the number of usable partitions. We demonstrate the approach on IBM's 156-qubit ibm_boston processor using standard QED-C benchmark and Hamiltonian-based observable-estimation workloads. Compared with standard sequential execution, parallel execution reduces billed execution time by 3.5-5.5x while retaining 83-92% of the sequential fidelity. We further evaluate the scaling of parallel circuit execution using GPU-accelerated classical simulation, distributing measurement circuits across GPUs via MPI. Using CUDA-Q on the NERSC Perlmutter system, we achieve up to 13.8x speedup on 16 GPUs (86% parallel efficiency) for an H2 electronic-structure simulation, with scaling evaluated across multiple Hamiltonians and circuit counts. These results provide an indication of the performance ceiling that parallel execution on future quantum hardware may eventually approach. Both execution modes are implemented as a runtime option within the QED-C Application-Oriented Benchmark suite. Together, the results show that circuit-level parallelism can reduce execution cost on current quantum hardware and simulation time on GPU clusters, with the potential for greater benefits as device quality and qubit counts increase.
Circuit knitting is the process of partitioning large quantum circuits into smaller subcircuits such that the result of the original circuits can be deduced by only running the subcircuits. Such techniques will be crucial for near-term and early fault-tolerant quantum computers, as the limited number of qubits is likely to be a major bottleneck for demonstrating quantum advantage. One typically distinguishes between gate cuts and wire cuts when partitioning a circuit. The cost for any circuit knitting approach scales exponentially in the number of cuts. One possibility to realize a cut is via the quasiprobability simulation technique. In fact, we argue that all existing rigorous circuit knitting techniques can be understood in this framework. Furthermore, we characterize the optimal overhead for wire cuts where the subcircuits can exchange classical information or not. We show that the optimal cost for cuttingnwires without and with classical communication between the subcircuits scales asO(16n) andO(4n), respectively.