IBM Patent Application Targets More Efficient Sharing of Quantum Computers
AI-summarised brief · reviewed before publication
IBM has filed a patent for a quantum‑job scheduler that groups compatible circuits from multiple users to run concurrently on a single processor while reusing stored noise models for error mitigation. The software evaluates qubit availability, connection topology, execution time and user preferences, then matches jobs whose combined operations align with existing noise characterizations, reducing the need for fresh calibration measurements. The system can also prioritize jobs that can exploit still‑valid noise data, updating models as hardware noise drifts. The filing illustrates a scenario on a 127‑qubit chip where two programs using 52 and 30 qubits, respectively, are run together with four idle qubits separating them to limit crosstalk. IBM proposes billing rebates for users who permit sharing, though no performance metrics are disclosed.
💡 Why It Matters
- · By turning idle qubits into productive capacity, the approach could dramatically increase the usable throughput of costly quantum hardware, accelerating research cycles and lowering access barriers.
