TuniQ: Autotuning Compilation Passes for Quantum Workloads at Scale for Effectiveness and Efficiency

TL;DR

TuniQ is a reinforcement learning system that optimizes quantum circuit compilation by dynamically selecting passes, improving fidelity and efficiency across diverse workloads and hardware.

Contribution

It introduces novel RL-based techniques for adaptive pass selection in quantum compilation, outperforming fixed-sequence compilers like Qiskit.

Findings

TuniQ improves quantum circuit fidelity over Qiskit.

It reduces compilation time significantly.

It generalizes across different quantum hardware without retraining.

Abstract

Quantum processors are being integrated into HPC ecosystems as co-processors, where compilation of quantum circuits into hardware-executable form determines both output fidelity and runtime. Current compilers use a fixed pass sequence and ignore the fact that optimal pass selection varies with circuit, hardware, and noise conditions. We present TuniQ, a reinforcement learning-based system that selects compilation passes at each pipeline stage, adapting to circuit, backend, and current noise profile. TuniQ introduces several novel design components like a dual-encoder for stage-aware representation, shaped rewards for cross-stage credit assignment, and dynamic action masking for valid compilation. Evaluated across diverse quantum workloads on multiple IBM Quantum Cloud processors, TuniQ improves fidelity and reduces compilation time over the state-of-the-art IBM Qiskit transpiler, generalizes across backends without retraining, and scales strongly to utility-scale circuits with growing advantage.