Environment engineering and dissipative steady states

Choose an orthonormal target–error basis for a qubit. A jump proportional to the target ket times the error bra transfers population from error to target. It annihilates the target, making it dark. In that basis this is ordinary amplitude damping, even if the target is a superposition of the energy eigenstates. The required coupling is therefore physically different from merely cooling the original Hamiltonian: it must select the desired superposition and its phase.

For a pure target projector to remain stationary, every dissipative contribution and the Hamiltonian action must be compatible with it. Annihilation by all jumps is a simple sufficient condition when the Hamiltonian also preserves the target ray. More generally a pure stationary state can be a common eigenvector with nonzero jump eigenvalues, with appropriate Hamiltonian compatibility. Confusing the sufficient dark condition with a necessary universal criterion would exclude valid driven or displaced steady-state constructions.

In this two-dimensional sink model with no additional Hamiltonian mixing, the error population decays at the engineering rate and target–error coherence at half that rate. The target is the unique stationary density operator. The slowest decaying mode sets the asymptotic preparation timescale, so checking only population convergence can overstate how quickly an arbitrary coherent input approaches the target. The relevant generator eigenvalues are complex in more general driven models, where damped oscillations can accompany preparation.

With a larger Hilbert space, several dark states may survive, or an invariant subspace may never reach the target. Common dark-state equations then identify possible stationary states rather than their basin of attraction. A finite-dimensional time-independent GKSL generator always has stationary states, but uniqueness and a nonzero relaxation gap require separate analysis. Even a small gap can make preparation impractically slow; increasing one local jump rate need not improve the slowest collective mode in a many-body system.

Let the target be the plus state and the error state the minus state. Ordinary Z phase flips exchange those two states. Adding that noise creates a reverse population flow that competes with the engineered sink. Positive noise rates still give a legitimate generator, but the stationary state is no longer perfectly dark or pure. Solving the two population balances quantifies the achievable fidelity and makes the required rate ratio explicit instead of claiming immunity to arbitrary decoherence.

Preparing a coherent target relative to the energy basis can require external driving or controlled work resources; it should not be described as free equilibration with a single passive thermal bath. Thermal detailed balance and an engineered nonthermal reservoir answer different physical design questions. Laboratory implementation also requires verifying bandwidth, elimination approximations, and unwanted channels. Theoretical constructions of many-body dissipative preparation establish possibilities under stated assumptions, while robustness and scalable preparation times depend on the particular model and experiment.