Mid-Circuit Measurement & Qubit Reuse Compilation on Superconducting Quantum Hardware
We evaluate compiler-driven mid-circuit measurement and dynamic qubit reset on IBM Heron (ibm_marrakesh, 156 programmable qubits, Job da1u0om3kjvs738777cg, 24,576 shots). Across 6 streaming and bounded-treewidth circuits, qubit reuse passes reduced active physical register width by a median 78.1% (up to 87.5% for 16 logical qubits compressed to 2 physical slots). Physical observable fidelity remained non-inferior to static uncompressed circuits across all 6 cases (4/6 direct reuse wins; 6/6 within pre-declared non-inferiority margin \(\epsilon = 0.05\)).
1. Liveness-Driven Register Compression
Quantum circuits with temporal liveness structures (where qubits carry state for only a fraction of circuit depth) can be compressed onto smaller physical registers by inserting mid-circuit measurements (M), resets (|0〉), and dynamic wire reassignments.
2. Empirical Results on IBM Heron r2
| Workload | Logical Qubits | Physical Slots | Width Reduction | Observable RMSE Delta | Verdict |
|---|---|---|---|---|---|
| STREAM_8_A | 8 | 2 | 75.0% | -0.0124 (Win) | NON-INFERIOR |
| STREAM_8_B | 8 | 3 | 62.5% | -0.0089 (Win) | NON-INFERIOR |
| STREAM_12_A | 12 | 2 | 83.3% | -0.0156 (Win) | NON-INFERIOR |
| STREAM_12_B | 12 | 3 | 75.0% | -0.0042 (Win) | NON-INFERIOR |
| STREAM_16_A | 16 | 2 | 87.5% | +0.0182 (Loss within margin) | NON-INFERIOR |
| STREAM_16_B | 16 | 3 | 81.3% | +0.0110 (Loss within margin) | NON-INFERIOR |
3. Scope & Structural Qualifications
Qubit reuse is strictly applicable to circuits with sequential liveness graphs and streaming communication topologies. Highly entangled algorithms requiring simultaneous multi-qubit coherence cannot be compressed onto a 2-qubit register. Q-Psi explicitly rejects claims that 16-qubit quantum computation universally requires only 2 qubits.