Quantum Feedback and Trajectories

Quantum feedback uses a measurement record to estimate a system’s conditional state and change its subsequent dynamics. The research challenge is to make the estimator, controller, detector, and physical system work together when the model is imperfect and the feedback is delayed. A controller that succeeds with access to the exact simulated state may fail when it receives only the actual record.

Required background. Stochastic master equations describes conditional evolution; measurement-based feedback owns the stable control framework.

The measurement record is the available information

For a common diffusive measurement convention, the record has the form

$$dY_t=\sqrt{\eta}\,\operatorname{Tr}[(c+c^\dagger)\rho_t]dt+dW_t,$$

where $c$ includes the measurement strength, $0\le\eta\le1$ is detection efficiency, and $dW_t$ is an innovation increment with variance $dt$. The conditional state update must use the same convention and record. Substituting independent noise into the estimator changes the inference problem.

Established: filtering and feedback can stabilize selected states and dynamics under explicit models. Active: robust estimation, sampled and delayed control, restricted observers, and scalable feedback. Liang et al.’s September 2026 preprint reviews these issues for principally finite-dimensional diffusive systems; it is a tutorial preprint, not evidence that every proposed implementation is experimentally achieved. Manzano and Zambrini connect monitoring to thermodynamic trajectories.

Sources were checked on 2026-10-11. The evidence horizon includes the 5 September 2026 review preprint, with its stated scope.

Compare controllers with the same information

An open-loop pulse, a state-feedback law, and a learned controller receive different information unless the comparison is designed carefully. Count measurement strength, efficiency, record bandwidth, estimation cost, actuator limits, and latency. Feedback may improve stabilization while increasing control energy or measurement-induced disturbance.

A useful question is whether a low-order observer retains enough information for a specified task under model mismatch. Another is whether a stability guarantee survives realistic sampling and delay. Do not identify average fidelity with a guarantee of rare-event reliability; long excursions can matter even when the mean looks favorable.

Starter project: measure the price of an imperfect observer

Simulate a continuously monitored qubit with a bounded feedback drive. Maintain a physical state and an estimated state driven by the same measurement record. Begin with matched parameters and unit efficiency, then introduce one mismatch: detector efficiency, Hamiltonian drift, or a fixed delay.

Compare no feedback, feedback using the true state as an explicitly unattainable reference, and feedback using the estimator. Refine the stochastic time step and increase the number of trajectories. Verify that ensemble-averaged no-feedback trajectories agree with the corresponding master equation.

The laptop-scale deliverable is a map of stabilization error, excursion probability, and control effort versus mismatch or delay. Hold out parameter values from any controller tuning. Stop interpreting a gain if it relies on future measurement increments, the true hidden state, or inconsistent record normalization. The proposed study is a baseline; a research extension needs a specific robustness or observer-design gap confirmed in current literature.

Open directions and useful negative results

Scalable estimation. Can a reduced observer predict the variables required for control without reconstructing the full density operator? State the approximation and quantify where it fails.

Delay and hybrid dynamics. Finite processing time, jumps, resets, and saturation can change stability. Derive a bounded stability region or identify a failure mechanism rather than present only a successful trajectory.

Thermodynamic accounting. Feedback changes the information available to the controller. Compare work or entropy statements only after specifying the measurement and information ledger; see fluctuation relations.

A decisive advance would survive independent record-based validation under the stated hardware constraints. A failure boundary is useful if it tells an experiment which efficiency, bandwidth, or latency must improve.

Entry points and references

Read the stable measurement and filtering pages, then Liang et al. for the control-theoretic questions. Reassess when an observer or delay theorem covers your model, or when measured detector behavior invalidates the assumed record statistics.

References

• W. Liang, J. Zhang, G. Guo, and D. Dong, “Measurement-Based Feedback of Open Quantum Systems: A Control-Theoretic Review and Tutorial,” arXiv:2609.06111v1 (2026). Preprint.
• G. Manzano and R. Zambrini, “Quantum thermodynamics under continuous monitoring: a general framework,” AVS Quantum Science 4, 025302 (2022). Preprint.