A physical decision boundary for dynamic versus unitary quantum circuits
LapYuen LEE
Abstract
Classical control flow via mid-circuit measurement is mathematically permitted in quantum computation and is the standard execution model for current hardware. We show that it is not, however,the physically optimal implementation strategy for deep quantum circuits. The cost of each mid-circuitmeasurement decomposes into a physics-bounded component (projection destroys coherence) and anengineering-bounded component (feedback latency introduces decoherence); the former is irreducible,while the latter tracks hardware progress but retains a physical floor set by dispersive readout. Gatefidelity, by contrast, improves with materials engineering and carries no measurement penalty. We formalize this asymmetry as a strategy bifurcation criterion: for any hardware generation, there exists acircuit depth N∗ above which pure unitary execution achieves strictly higher fidelity than dynamic execution. We measure N∗ ≈ 4 on IBM’s 156-qubit Heron processor and verify the predicted monotonicdivergence to N = 31, where the dynamic approach yields zero counts while the pure unitary approachretains 0.9% fidelity. The per-step fidelity ratio is ∼1.20, independent of the controlled operation’s depth.Projecting across hardware generations, the absolute executable depth advantage remains ∼30×, but N∗trends toward zero: by 2036, pure unitary execution is the optimal strategy at all depths. We provide acompilation framework that translates all four algorithmic control structures (for, if, switch, while) intopure unitary circuits, validated on four quantum platforms spanning three hardware modalities. The result is a hardware-aware control strategy selection: for algorithms with conditional depth exceeding N∗,compilers should default to static unitary expansion rather than dynamic measurement-feedback.