QaiJi IR: An Eight-Layer Intermediate Representation Family for Hybrid Quantum-Classical Compilation

Abstract:Hybrid quantum-classical compilers exchange programs among circuit, control-flow, pulse, device, and physical representations. Existing formats make different abstraction choices, so the properties that must survive a lowering step are often enforced by tool-specific code rather than stated in a common intermediate representation (IR). We present QaiJi IR, an eight-layer family organized by a \emph{five-axis semantic contract} for data, semantics, lowering, runtime, and verification. Its L4 SemanticIR records the operation class, declared equivalence level, optimization role, and classical-feedback model as attributes, and maps operation classes to ISA-neutral hardware actions. The prototype provides typed quantum and classical circuit nodes, OpenQASM~2 compatibility input, canonical OpenQASM~3 output, explicit rejection of unsupported constructs, and executable gate-convention checks. A representative measurement-conditioned program reaches a canonical QASM fixed point, acquires a measurement-to-condition edge, passes an exact aggregate consistency check, and produces a symbolic pulse template. For a one-bit active-reset case, device lowering materializes ALU/FPROC records and a conditional branch. Running the resulting virtual pulse processing unit (VPPU) program with externally supplied outcomes 0 and 1 respectively skips and executes the conditional drive. These tests validate the represented compiler path and binary branch behavior. They do not establish a physical feedback loop, because the simulator cannot yet apply mid-circuit measurement collapse and continue the same dynamical trajectory.

arXiv:2609.29305 [quant-ph]

(or arXiv:2609.29305v1 [quant-ph] for this version)

https://doi.org/10.48550/arXiv.2609.29305 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.