Photonic Quantum Computing on Spin Memory Architecture with Tree-Encoded Fusion
Year: 2026
Type: conference-paper
Abstract: Photonic quantum computer (PQC) is a promising quantum computation platform, realizing the measurementbased quantum computation (MBQC) model. In MBQC, computation proceeds by preparing a graph state, and this preparation mainly relies on fusion operations. However, fusion operations on a PQC are prone to two types of errors: fusion failure and fusion erasure. Therefore, the MBQC compiler must be carefully designed to tolerate these errors. Previous state-of-the-art MBQC compiler - OneAdapt, is tailored to all-photonic architectures and primarily address fusion failures. However, it neglects fusion erasure errors caused by photon loss, which are more detrimental than fusion failures. To address the challenge of fusion erasure, we propose a novel MBQC scheme that is based on quantum spin memory architecture. We design a tree-encoded fusion scheme that effectively suppresses erasure errors. Then, we integrate this scheme into our compiler framework, with compilation algorithms that reduce execution overhead of quantum programs. We evaluate our framework on a realistic PQC simulator across six typical quantum algorithm benchmarks with various program sizes. Our results show that the proposed tree-encoding scheme outperforms other fusion encoding schemes, and that our compilation framework outperforms OneAdapt with exponential improvement. Moreover, we demonstrate a small-scale QAOA experiment on real PQC hardware, where it outperforms the latest superconducting hardware.
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7172443523