Researchers build intent-level quantum programming with live debugging

Quantum programs are hard to validate, but a new paper presents a quantum domain-specific language built around three key mechanisms: intent-execution separation, first-class IR introspection, and assertion-guided modality inference. Algorithmic constructs like preparation, superposition, entanglement, and measurement become inspectable objects that are compiled into backend-specific circuits only after pre-execution intermediate representation validation. Developers can inspect circuit width, wire mapping, operation order, and measurement intent before running. Assertions automatically select sampling, statevector, or dual execution without manual intervention, and structured logs support reproducible regression testing. Fault-injection experiments on Bell and 3-qubit GHZ circuits achieved a true positive rate of 1.0 across all evaluated shot budgets, with false positives only at 128 shots. Differential testing showed numerical agreement across PennyLane and Qiskit compilation targets after endianness canonicalization, and IR generation costs were sub-millisecond.