解码器与经典实时系统decoder
Windowed and streaming decoder
把持续产生的时空 syndrome 划分为带重叠缓冲区的有限窗口,只提交高置信区域的修正,并用并行或分布式执行维持实时吞吐。
- 成熟度
- research
- 重要度
- 核心主题
- 证据
- 3 篇代表来源
01
范围与辨析
必须记录 window、commit 与 buffer 大小、边界协调、inner decoder、吞吐和尾延迟;离线分块但不提交在线结果的流程不算 streaming。
02
核心机制与研究判断
核心机制
- overlapping spacetime windows
- commit and buffer regions
- bounded backlog
- parallel or distributed execution
适用与评测约束
- Noise Level
- must be declared
- Execution Mode
- batch or online must be declared
- Hardware Sensitivity
- latency and parallelism must be measured
- Input Representation
- syndrome or detector model specific
- Benchmark Comparability
- requires code noise syndrome hardware tuple
03
全栈位置与直接关系
decoder/real-time/windowed← 依赖窗口化实时输入长时间计算不能等待全部 syndrome 才解码,需要滑动窗口、边界传递与流式处理。← 依赖实时 Decoder 实现验证最坏情况延迟、带宽、量化精度和 backlog,而不只看 Python 平均运行时间。← 兼容Matching 系将 detection events 配对或在相关图上求近似最优纠错链。← 兼容Union-Find / clustering通过缺陷聚类和增长实现低复杂度拓扑码解码。← 兼容BP + localized statistics decoding以 belief propagation 输出的可靠度引导局部错误簇生长,并在各簇内并行执行 on-the-fly elimination,从而避免 BP+OSD 的全局矩阵求逆。
04
代表证据
定义2023
Parallel window decoding enables scalable fault tolerant quantum computationLuka Skoric, Dan E. Browne, Kenton M. Barnes, Neil I. Gillespie and Earl T. CampbellBenchmark2023
Scalable Surface-Code Decoders with Parallelization in TimeXinyu Tan, Fang Zhang, Rui Chao, Yaoyun Shi and Jianxin Chen前沿代表2026
Snowflake: A Distributed Streaming DecoderTim Chan05