QIRO: A Static Single Assignment-based Quantum Program Representation for Optimization

We propose an IR for quantum computing that directly exposes and classical data dependencies the purpose of optimization. The Quantum Intermediate Representation Optimization (QIRO) consists two dialects, one input dialect is specifically tailored to enable quantum-classical co-optimization. While first employs a perhaps more intuitive memory-semantics (quantum operations act as side-effects), latter uses value-semantics (operations consume produce states). Crucially, this encodes dataflow in IR, allowing host optimizations leverage analysis. discuss how map existing programming languages lower resulting optimization dialect. present prototype implementation based on MLIR includes several quantum-specific passes. Our benchmarks show significant improvements resource requirements are possible even through static In contrast circuit at run time, achieved while incurring only small constant overhead compilation making compelling approach program application scale.