Suppressing the Erasure Error of Fusion Operation in Photonic Quantum Computing
Xiangyu Ren 2 , Yuexun Huang 3 , Zhemin Zhang 4 , Yuchen Zhu 6 , Tsung-Yi Ho 4 , Antonio Barbalace 2 , Zhiding Liang 4 1 · 2026
Read Suppressing the Erasure Error of Fusion Operation in Photonic Quantum Computing by Xiangyu Ren 2 , Yuexun Huang 3 , Zhemin Zhang 4 , Yuchen Zhu 6 , Tsung-Yi Ho 4 , Antonio Barbalace 2 , Zhiding Liang 4 1 (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-21475.txt.
Introduction
Xiangyu Ren and six colleagues suppress fusion erasure errors in photonic measure-based work, where graph-state building hinges on chance fusion operations. Prior compiler OneAdapt handles fusion failures but ignores erasures from photon loss, which can damage far more. Their tree-encoded fusion scheme, built on spin-qubit quantum memory, suppresses erasures during graph-state building, inside a tool system with methods cutting run overhead. On six benchmarks across program scales it beats other fusion encodings in robustness and improves vastly over OneAdapt, with feasibility shown on real hardware.
Catalog record
| CREATOR | Xiangyu Ren 2 , Yuexun Huang 3 , Zhemin Zhang 4 , Yuchen Zhu 6 , Tsung-Yi Ho 4 , Antonio Barbalace 2 , Zhiding Liang 4 1 |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 62805 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.21475
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Suppressing the Erasure Error of Fusion Operation in Photonic Quantum Computing Xiangyu Ren 2 , Yuexun Huang 3 , Zhemin Zhang 4 , Yuchen Zhu 6 , Tsung-Yi Ho 4 , Antonio Barbalace 2 , Zhiding Liang 4 1 Abstract Photonic quantum computing (PQC) provides a promising route toward quantum computation by naturally supporting the measurement-based quantum computation (MBQC) model. In MBQC, programs are executed through measurements on a pre-generated graph state, whose construction largely depends on probabilistic fusion operations. However, fusion operations in PQC are vulnerable to two major error sources: fusion failure and fusion erasure . As a result, MBQC compilation must account for both error mechanisms to generate reliable and efficient photonic executions. Prior state-of-the-art MBQC compilation, represented by OneAdapt, is designed for all-photonic architectures and mainly focuses on handling fusion failures. Nevertheless, it does not explicitly model fusion erasures induced by photon loss, which can be substantially more damaging than fusion failures. To mitigate fusion erasure errors, we introduce a new MBQC compilation scheme built upon the spin qubit quantum memory architecture. We propose tree-encoded fusion, an encoding strategy that suppresses erasure errors during graph-state generation. We further incorporate this scheme into a compiler framework with algorithms that reduce the execution overhead of quantum programs. We evaluate the proposed framework using a realistic PQC simulator on six representative quantum algorithm benchmarks across multiple program scales. The results show that tree-encoded fusion achieves better robustness than alternative fusion-encoding strategies, and that our compiler provides exponential improvement over OneAdapt. In addition, we validate the feasibility of our approach through a proof-of-concept demonstration on real PQC hardware. I Introduction Photonic quantum computer (PQC) is one of the highly potential quantum computing systems that pave the way to quantum supremacy [ 48 ] . Operated at room temperature, the photonic qubits have long decoherence time and great scalability [ 55 ] ; also, their natural distributed characteristic makes PQC easy for quantum network integration [ 71 , 7 ] . Various companies have demonstrated their PQC hardware platforms: PsiQuantum introduces their manufacturable platform of PQC with 99.72% two qubit fusion fidelity [ 51 ] , and Quandela provides cloud platform access with a 24-photon modes PQC system [ 23 ] . Despite the rapid progress of photonic quantum hardware, compilation techniques of photonic quantum computer (PQC) is yet to be fully explored, while there are only a few novel PQC compilers [ 73 , 74 ] evaluated on a simplified quantum simulation model. MBQC and graph state. Different from other typical quantum computing hardware (e.g., superconducting or neutral-atoms), photonic quantum computing adapts a measurement-based computation (MBQC) model [ 8 ] rather than the usual gate-based model. In the paradigm of measurement-based quantum computation, a quantum program is represented as a highly entangled quantum state – the graph state, and sequential measurements are performed on qubits of the graph state to execute corresponding computations. Hence, the graph state plays a key role in MBQC, and the main challenge of MBQC compilation on photonic systems is to generate the target graph state robustly and efficiently. Fusion operations for graph states are imperfect. In a photonic system, photonic qubits are emitted from the photon sources, and they are entangled through fusion operations to form a graph state. Fusion operations are prone to two types of errors: fusion failure and fusion erasure. Fusion failures arise from the natural characteristic of photonic qubits. When the fusion of two qubits fails, they are automatically measured out and have no impact on the rest of their entangled qubits. In contrast, fusion erasure is caused by photon loss: one of the fusion qubits cannot be detected by the system, thus the outcome of fusion operation remains unknown. Prior Works. A recent state-of-the-art MBQC compiler – OneAdapt [ 74 ] purpose an efficient framework to execute quantum programs on PQC, and it tackles the fusion failure by utilizing a normalization method. However, it neglects the occurrence of fusion erasure, which is unable to deal with only using normalization. Also, normalization method distillates a valid 2D graph state layer from much larger number of photonic qubits, which fails to efficiently utilize photon resources. Meanwhile, Li et al. [ 37 ] propose the RLGS compiler on emitter-based photonic architecture. It benefits from the deterministic nature of emitter-based photon source, and utilize the interactions between emitters to generate entanglements for graph state. Despite this, emitter-based solution has still not been fully demonstrated in experiment, due to the challenge of coupling between emitters efficiently [ 25 ] . As a result, the evaluations in [ 37 ] are based on assumptions from theoretical analysis [ 56 , 18 ] , so the proposed solution still has a gap from near-term photonic quantum computing systems. Quantum spin memory architecture. To fill this gap, we introduce our PQC framework to handle the fusion errors in the near-term PQC architecture with a realistic hardware model. The PQC architecture that we adapt [ 25 ] utilizes silicon-based quantum spin memory as photon source, which could generate a specific structure of resource graph state, namely caterpillar state. The caterpillar state has been demonstrated experimentally in a small scale [ 28 ] . Next, these caterpillar states are concatenated together by fusion operations, forming a large graph state corresponding to the target quantum program. Generally, the main problem we are addressing in this work, is finding an efficient solution that generates the target graph state from caterpillar state, while maintaining robustness against fusion failure and erasure. Firstly, we propose a novel tree-encoded scheme to boost the success rate of fusion operation , while effectively suppressing fusion failure and fusion erasure. We leverage the flexibility of caterpillar state, and are inspired by the quantum error correction (QEC) tree code [ 68 , 67 , 5 ] to design our unique fusion scheme. Specifically, we eliminate the photonic qubit that undergoes erasure (photon loss) utilizing the indirect Z-measurement . Meanwhile, our dedicated tree structure design provides multiple trials for the fusion operation, offering resistance to fusion failures. Compared to previous boosted-fusion schemes, e.g. repetition encoding [ 25 ] and repeat-until-success [ 19 ] approach, our tree-encoded scheme achieves significant error-tolerance against high erasure error rates (Fig. 4 (f)). Furthermore, the tree-encoding structure can be easily generated from the caterpillar state, proving its fitness for quantum spin memory architecture. Detailed theoretical analysis and simulation are given in Sec. IV . Secondly, we propose MemTree – a compilation framework for scalable and resource-efficient execution of quantum programs on quantum spin memory architecture. We implement a divide-and-conquer algorithm to separate the target graph state into multiple caterpillar states while reducing the total number of fusions to minimize execution overhead. Specifically, we design a hierarchical minimal-cut algorithm for dividing the target graph state. In the algorithm, we leverage mix-integer-programming (MIP) to obtain the solution, with dedicated constraints to guarantee the divided subgraphs conform to the structure of caterpillar states. Details of compiler design are given in Sec. V . Thirdly, we build a realistic error-aware simulator to evaluate our framework and compare it with SOTA [ 74 ] . Our simulator considers the impact of fusion failure and fusion erasure based on the hardware noise