Quantum-Classical Hybrid Workflows in L0

Where a quantum processor would plug into the Kronos stack: strictly offline in L0, orchestrated by classical code, always with a classical baseline.

Quantum lives in L0, never in control

No quantum computation runs anywhere near the microsecond control plane. In the eight-layer Kronos architecture, quantum processors are, at most, an accelerator called from the offline L0 foundation layer, the same place classical supercomputing, surrogate training, and large sweeps run. Everything time-critical on the breeder and burner is deterministic classical hardware.

# L0 hybrid job: quantum used as a subroutine, classical drives + verifies
def materials_energy(active_space):
    classical = classical_solver(active_space)      # DMRG / coupled cluster
    if quantum_available and within_resource_budget(active_space):
        q = quantum_estimator(active_space)         # VQE or (future) QPE
        assert agrees(q, classical, tol='chemical') # gate on agreement
        return reconcile(q, classical)
    return classical                                # default is classical

The orchestration pattern

• A classical optimizer or driver owns the loop; the quantum device evaluates a cost or estimates a scalar.
• Every quantum result is checked against a classical baseline computed in parallel.
• Jobs are scheduled like any other L0 batch workload, with reproducibility and provenance tracked.

Two integration shapes

Variational methods (VQE, QAOA) are tight loops: many short quantum evaluations per classical optimizer step, latency-sensitive to the quantum-classical round trip. Estimation methods (QPE, QAE) are long single circuits: one deep coherent run returning a high-precision number, requiring fault tolerance.

variational:  classical_opt <--(scalar)--  QPU   (many fast round trips)
estimation :  classical_prep --> deep QPU circuit --> classical_readout
# both are OFFLINE; neither is on any real-time actuation path

This isolation is a safety property, not a limitation. A quantum result can be wrong, noisy, or unavailable without any consequence for machine operation, because operations never depend on it. That is precisely what lets us experiment with immature quantum methods responsibly. The governing stance is the honest NISQ assessment.