An Efficient Compilation Architecture for Quantum Experimentation, Calibration and Control Tasks
Lukas Scheller, Cedric Becker, Marvin Fuchs +4 more
Quantum compiler research has largely focused on high-level circuit and gate compilation, while the translation of quantum experiments into low-level control code remains less studied in academia and often confined to vendor-specific toolchains. This “last-mile” compilation stage is critical for calibration, hardware bring-up, and algorithm execution, but may become a major performance and maintainability bottleneck as systems scale. We present a compiler architecture for quantum experiments that decouples frontend experiment languages from backend control hardware through two design choices: (i) a schema-defined abstract syntax tree (AST) represented using Protocol Buffers as a stable, language-agnostic interchange format, and (ii) a compilation pipeline built on the Multi-Level Intermediate Representation (MLIR) framework. Our approach introduces multiple intermediate representations, enabling domain-specific optimizations and systematic lowering from high-level experiment descriptions to binary code, targeting a RISC-V-inspired hardware sequencer. We demonstrate the approach using the QiCode Python-based experiment language and the QiController control platform. Benchmarks on representative calibration workloads show comparable compilation latency and scalability relative to a vendor-API baseline, while synthetic scaling experiments demonstrate sub-second compilation latency for intermediate representation sizes exceeding 1000 qubit control cells, with minimal serialization overhead. These results suggest that combining schema-based AST definitions with MLIR-based multi-stage lowering provides a scalable foundation for portable quantum control compilation pipelines.
IEEE International Conference on Quantum Software (QSW)•Jul 1, 2026•0 citations