Abstract: Distributed quantum computing (DQC) is a promising technique for scaling up quantum systems. While significant progress has been made in DQC for quantum circuit models, there exists much less research on DQC for measurement-based quantum computing (MBQC), which is a universal quantum computing model that is essentially different from the circuit model and particularly well-suited to photonic quantum platforms. In this paper, we propose DC-MBQC, the first distributed quantum compilation framework tailored for MBQC. We identify and address two key challenges in enabling DQC for MBQC. First, for task allocation among quantum processing units (QPUs), we develop an adaptive graph partitioning algorithm that preserves the structure of the graph state while balancing the workload across QPUs. Second, for inter-QPU communication, we introduce the layer scheduling problem and propose an algorithm to solve it. Regrading realistic hardware requirements, we optimize the execution time of running quantum programs and the corresponding required photon lifetime to avoid fatal failures caused by photon loss. Our experiments demonstrate a $7.46\times$ improvement on required photon lifetime and $6.82\times$ speedup with 8 fully-connected QPUs, which further confirm the advantage of distributed quantum computing in photonic systems. The source code is publicly available at https://github.com/qfcwj/DC-MBQC.
The Challenge of Scaling Photonic MBQC
While the quantum circuit model is the most common paradigm, Measurement-Based Quantum Computing (MBQC) offers a compelling alternative, particularly for photonic systems. In MBQC, computation proceeds by performing a sequence of measurements on a large, pre-entangled "graph state" of photons. Scaling this model to handle complex algorithms requires a Distributed Quantum Computing (DQC) architecture, where multiple small Quantum Processing Units (QPUs) are interconnected via optical fibers.
However, photonic systems face a critical physical bottleneck: photon loss. Photons must be stored in fiber delay lines while awaiting measurement results or synchronization signals. As shown in Figure 1, the probability of photon loss increases significantly with the number of clock cycles a photon must spend in storage. If a photon is lost, the entanglement structure of the graph state is broken, leading to fatal computational failure. Existing monolithic compilers do not account for these temporal constraints in a distributed environment, often resulting in schedules that exceed the physical capabilities of current hardware.
The DC-MBQC Compilation Framework
To address these challenges, we introduce DC-MBQC, the first compilation framework specifically designed for distributed MBQC. The framework unifies synchronization, measurement dependencies, and routing into a single formal metric: Required Photon Lifetime ($\tau_{photon}$). By minimizing this metric, DC-MBQC makes large-scale photonic computing feasible on realistic hardware.
As illustrated in Figure 2, the framework decomposes the compilation process into two primary stages: adaptive graph partitioning and layer scheduling.
1. Adaptive Graph Partitioning
Instead of a simple uniform split, DC-MBQC employs an adaptive partitioning algorithm. It uses a heuristic search to balance the computational workload (number of photons) across QPUs while simultaneously minimizing inter-QPU communication (cuts). A key innovation is the preservation of the graph state's local "modularity." By keeping tightly-knit sub-graphs together within a single QPU, the framework enables more compact local layouts, which further reduces the time photons must spend in storage.
2. Bottleneck-Driven Iterative Refinement (BDIR)
Scheduling tasks to minimize the maximum photon wait time is an NP-hard problem. DC-MBQC addresses this through the Bottleneck-Driven Iterative Refinement (BDIR) algorithm. Unlike general-purpose optimization techniques like simulated annealing, BDIR specifically identifies the "bottleneck" photon—the one with the longest required lifetime—and iteratively reschedules its associated tasks toward a "temporal equilibrium point." This targeted approach allows the compiler to find high-quality schedules for large-scale programs much faster than monolithic alternatives.
Performance Evaluation
We evaluated DC-MBQC using standard quantum benchmarks, including VQE, QAOA, QFT, and Ripple-Carry Adders, with programs ranging up to 196 qubits and over 68,000 fusion operations.
Our results demonstrate significant improvements over state-of-the-art monolithic compilers:
- Photon Lifetime Reduction: DC-MBQC achieved up to a 7.46× reduction in the required photon lifetime when using 8 QPUs. This reduction is transformative, potentially allowing the use of existing fiber technology for computations that previously required near-perfect storage.
- Execution Speedup: The framework provided a 6.82× speedup in total execution time with 8 QPUs.
- Scalability: The distributed nature of the framework allows it to scale more efficiently than monolithic approaches, handling larger quantum programs within reasonable compilation time budgets.
Conclusion and Future Directions
DC-MBQC demonstrates that distributed computing is not just a way to increase qubit counts, but a vital tool for relaxing the stringent hardware constraints of photonic systems. By spreading the computational depth across multiple QPUs, we can significantly lower the risk of photon loss.
Future work will focus on extending the framework to handle more restrictive network topologies (such as rings or grids) and integrating explicit support for Quantum Error Correction (QEC) codes. Additionally, incorporating "online" re-compilation strategies could help mitigate the probabilistic nature of photonic fusion operations, further enhancing the reliability of distributed MBQC systems.
Summary updated