Strong Coupling

Strong coupling means that the interaction between a retained system and its environment cannot be treated as a small perturbation around an almost unchanged bath. In that regime, weak-coupling master equations can fail qualitatively: the relevant states are dressed, the bath is displaced or depleted, interaction energy matters, and the reduced dynamics may have memory or preparation dependence.

The most important warning is:

large decay rate is not the definition of strong coupling

A decay rate can be large because a Markovian bath has a large density of states. Conversely, a system can be strongly coupled to a slow or structured environmental coordinate even when a particular observed decay is not fast. Strong coupling is about the failure of the perturbative separation between system, bath, and interaction.

Weak-Coupling Reference Point

The usual weak-coupling setup writes

$$H = H_S+H_B+\lambda H_I, \qquad H_I = \sum_\alpha A_\alpha\otimes B_\alpha.$$

The Born Approximation treats the bath as staying close to a reference state and replaces, inside the retained order,

$$\rho_{SB}(t) \approx \rho_S(t)\otimes\rho_B.$$

The Markov Approximation then assumes bath correlations decay fast compared with the reduced system evolution. After secularization, one often obtains a Thermal Master Equation or another Lindblad–GKSL generator.

Strong coupling is the regime where one or more of these moves becomes unreliable.

What Can Go Wrong

Weak-coupling derivations can fail in several distinct ways.

Failure mode	What breaks
Bath backaction	The bath state changes enough that fixed bath correlations are not adequate.
System–bath correlations	Correlations become independent dynamical variables, not small corrections.
Dressed eigenstates	The eigenstates of $H_S$ are not the right transition basis.
Interaction energy	The term $H_I$ contributes to equilibrium, heat, work, and spectra.
Slow or structured modes	The environment returns information or excitation to the system.
Initial slip	A factorized initial state rapidly builds dressing correlations before slower dynamics begins.
Perturbative positivity failure	Weak-coupling equations are pushed outside their validity domain.

These problems often occur together, but they are not identical. A structured but weakly coupled bath is a memory problem. A broadband but very strongly coupled bath can be a dressing and thermodynamic-consistency problem. A finite environment can be a recurrence problem.

Dressed States

At strong coupling, the physically relevant eigenstates are eigenstates of the total Hamiltonian or of an enlarged system Hamiltonian, not bare eigenstates of $H_S$ alone.

If a prominent environmental mode is important, one may split

$$H = H_{S'} + H_R + H_{S'R}, \qquad S'=S+\text{mode}.$$

The enlarged Hamiltonian $H_{S'}$ contains the original system and the strongly coupled mode. Dissipation is then derived for $S'$ weakly coupled to the residual bath $R$.

This is the logic behind Reaction-Coordinate Mapping and Pseudomode Methods. The memory has not vanished. It has been moved from an implicit bath into explicit degrees of freedom.

Renormalization and Polaron Ideas

Strong coupling can renormalize the system Hamiltonian. In the spin–boson model, environmental displacement can suppress tunneling, shift transition frequencies, and change the preferred basis. A common preview of the idea is a polaron-type unitary transformation.

For a longitudinal spin–boson coupling, a schematic transformation is

$$U_P = \exp \left[ - \sigma_z \sum_k \frac{g_k}{2\hbar\omega_k} \left( b_k^\dagger-b_k \right) \right].$$

This transformation displaces the bath conditioned on the system state. In the transformed frame, the “small” perturbation may be a residual tunneling or residual bath fluctuation rather than the original coupling.

The point is not that the polaron method always solves the problem. It changes the perturbative split. A weak-coupling equation in the wrong frame can fail, while an equation in a dressed or displaced frame may capture the dominant physics.

For the canonical model where these issues appear, see Spin–Boson Model.

Reorganization Energy

For oscillator baths, a useful scale is the reorganization energy: the energy associated with shifting environmental coordinates between system configurations. With one common spectral-density convention,

$$\lambda_{\mathrm{reorg}} \sim \int_0^\infty d\omega\, \frac{J(\omega)}{\omega}.$$

The prefactor depends on the definition of $J(\omega)$. The comparison is the important part:

$$\lambda_{\mathrm{reorg}} \sim \text{system gaps, thermal energy, or driving scales}$$

signals that treating the interaction as a tiny correction may be unsafe.

In molecular and condensed-matter settings, this scale can compete with tunneling amplitudes, exciton splittings, or vibrational frequencies. Then the bath is not just a source of rates; it reshapes the effective system.

Hamiltonian of Mean Force

At weak coupling to a single thermal bath, one expects the system to relax toward the bare Gibbs state

$$\rho_\beta^{(0)} = \frac{e^{-\beta H_S}} {\operatorname{Tr}_S(e^{-\beta H_S})}.$$

At strong coupling, the equilibrium reduced state of the system is instead obtained from the total Gibbs state:

$$\rho_S^\star = \operatorname{Tr}_B \left[ \frac{e^{-\beta H}} {\operatorname{Tr}_{SB}(e^{-\beta H})} \right].$$

This can be written using a Hamiltonian of mean force:

$$H_{\mathrm{MF}} = - \beta^{-1} \log \left( \frac{ \operatorname{Tr}_B(e^{-\beta H}) }{Z_B} \right),$$

so that

$$\rho_S^\star = \frac{e^{-\beta H_{\mathrm{MF}}}} {\operatorname{Tr}_S(e^{-\beta H_{\mathrm{MF}}})}.$$

The difference between $H_{\mathrm{MF}}$ and $H_S$ is not bookkeeping. It affects steady states, thermodynamic potentials, and the meaning of heat and work. A local dissipator that forces the bare Gibbs state can be thermodynamically inconsistent in a strong-coupling regime.

Initial Slips

A factorized initial state

$$\rho_{SB}(0) = \rho_S(0)\otimes\rho_B$$

may be physically convenient but dynamically unnatural at strong coupling. The exact evolution quickly builds dressing correlations and bath displacement. The reduced state can show a rapid transient, often called an initial slip, before slower relaxation begins.

Ignoring the slip can lead to misleading fitted rates or apparent violations of positivity. In some calculations, the weak-coupling equation is intended only after a short correlation-building layer has passed. In other calculations, one should choose an initially correlated state compatible with the strong-coupling equilibrium or preparation procedure.

See Initial Correlations for the map-level consequences of correlated preparations.

Thermodynamic Consistency

Strong coupling changes thermodynamic bookkeeping. The interaction energy

$$\langle H_I\rangle$$

may be comparable to changes in $\langle H_S\rangle$. Then assigning all energy change of $S$ to heat or work using only $H_S$ can be ambiguous.

Consistent treatments must specify:

• the system boundary;
• the internal energy convention;
• whether interaction energy is included;
• the equilibrium state being used;
• whether the master equation satisfies the correct detailed-balance relation in the dressed basis;
• how external driving changes the interaction term.

The page Energy, Heat, and Work owns the thermodynamic definitions. The strong-coupling lesson here is simpler: bare weak-coupling thermal intuition should not be imported without checking the boundary and equilibrium state.

Modeling Strategies

There is no universal strong-coupling master equation. Common strategies include:

Strategy	Use when
Enlarged system	A small number of modes or coordinates carry most of the memory.
Reaction-coordinate mapping	A collective bath coordinate is strongly coupled or slow.
Pseudomodes	A structured spectrum has a few dominant poles or resonances.
Polaron transformation	Bath displacement or tunneling renormalization is central.
HEOM	Gaussian bath correlations can be expanded into controlled exponentials.
Chain mapping or tensor networks	Many modes with structured correlations matter.
Exact diagonalization	A small finite environment can be retained explicitly.
Collision models	Memory carriers can be represented as repeated ancillary interactions.

The best strategy depends on the structure of the environment, not only on the size of a coupling constant.

Positivity and Complete Positivity

When a weak-coupling equation is used outside its regime, it may produce negative populations or a nonpositive density matrix. That is a red flag, but the interpretation requires care.

• A Redfield Equation is not generally in Lindblad form and can fail positivity when pushed too far.
• A time-local equation with temporarily negative canonical rates may still give completely positive finite-time maps over some intervals.
• A strong-coupling exact reduced map can be completely positive on its preparation domain while not admitting a weak-coupling semigroup generator.

The cure is not to force every model into a Lindblad equation. The cure is to choose a representation whose assumptions match the physics, then check positivity, steady states, and convergence.

Practical Warning Signs

Be suspicious of a weak-coupling master equation when:

• the Lamb shift or reorganization energy is comparable to system gaps;
• relaxation rates are comparable to Bohr frequencies or drive frequencies;
• the bath has a sharp mode, band edge, or long correlation time;
• steady states disagree with expected dressed or strong-coupling equilibrium states;
• a local master equation predicts heat flow at equilibrium;
• exact small-environment simulations show revivals or large dressing transients;
• fitted rates depend strongly on the initial preparation;
• positivity failures appear under modest parameter changes.

These signs do not automatically prove a model is wrong. They tell you where to test the approximation.

Common Mistakes

• Defining strong coupling only by saying that a rate is large.
• Using bare-system transition frequencies when the physical eigenstates are dressed.
• Treating the bath as stationary when the coupling visibly displaces or depletes it.
• Assuming that a Lindblad-form local dissipator is thermodynamically consistent.
• Ignoring the interaction energy in heat and work accounting.
• Calling every negative Redfield population a fundamental problem rather than a regime warning.
• Starting from a factorized state when the preparation should already be dressed or correlated.
• Promoting a reaction coordinate but then using an inconsistent residual-bath temperature or coupling.

Exercises

Bare Gibbs versus mean force

Why need the strong-coupling equilibrium state of $S$ not equal $e^{-\beta H_S}/Z_S$?

Solution

At strong coupling, the equilibrium state of the total system is

$$\rho_{SB} = \frac{e^{-\beta(H_S+H_B+H_I)}} {\operatorname{Tr}_{SB}e^{-\beta(H_S+H_B+H_I)}}.$$

Tracing out the bath does not generally remove the effect of $H_I$. The reduced state is

$$\rho_S^\star = \operatorname{Tr}_B\rho_{SB},$$

which can be written using the Hamiltonian of mean force. Only when the coupling is negligible, or when special commutation and cancellation conditions hold, does this reduce to the bare Gibbs state of $H_S$.

Initial slip

Explain why a factorized initial state can produce a rapid transient at strong coupling.

Solution

At strong coupling, physical eigenstates and equilibrium states generally contain system–bath correlations or bath displacement. A product state $\rho_S\otimes\rho_B$ lacks that dressing. The exact dynamics rapidly builds the missing correlations on a bath or dressing time scale. The reduced system can therefore move quickly before the slower relaxation described by coarse-grained rates begins.

Enlarging the system

A qubit strongly exchanges excitation with one cavity mode, and the cavity leaks weakly into a broadband background. Which system boundary is more natural for a Markovian master equation?

Solution

The natural retained system is qubit plus cavity mode. The strong coherent exchange is then part of the system Hamiltonian, while the residual broadband leakage can often be treated by a Markovian dissipator. If the cavity is traced out, the qubit-only dynamics can show memory and excitation revivals.

Reorganization scale

Suppose a molecular two-level system has tunneling scale $\Delta$ and bath reorganization energy $\lambda_{\mathrm{reorg}}$ with $\lambda_{\mathrm{reorg}}\gg\Delta$. Why is a bare weak-coupling relaxation equation suspicious?

Solution

If the reorganization energy is much larger than the tunneling scale, the bath displacement associated with the system state is not a small correction. The bath reshapes the effective tunneling and preferred basis. A weak-coupling equation that treats the bath only as a perturbative source of transition rates around the bare $H_S$ can miss the dominant dressed physics.

Cross-Links

• Non-Markovian Dynamics
• Pitfalls in Non-Markovian Modeling
• Born Approximation
• Markov Approximation
• Redfield Equation
• Initial Correlations
• Thermal Master Equations
• Reaction-Coordinate Mapping
• Pseudomode Methods
• Hierarchical Equations of Motion
• Collision Models
• Spin–Boson Model
• Caldeira–Leggett Model
• Energy, Heat, and Work
• Approximation Checklist

References

• H.-P. Breuer and F. Petruccione, The Theory of Open Quantum Systems, Oxford University Press (2002).
• U. Weiss, Quantum Dissipative Systems, 4th ed., World Scientific (2012).
• A. J. Leggett, S. Chakravarty, A. T. Dorsey, M. P. A. Fisher, A. Garg, and W. Zwerger, “Dynamics of the dissipative two-state system,” Reviews of Modern Physics 59, 1–85 (1987).
• I. de Vega and D. Alonso, “Dynamics of non-Markovian open quantum systems,” Reviews of Modern Physics 89, 015001 (2017).
• J. Iles-Smith, N. Lambert, and A. Nazir, “Environmental dynamics, correlations and the emergence of noncanonical equilibrium states in open quantum systems,” Physical Review A 90, 032114 (2014).
• P. Strasberg, G. Schaller, N. Lambert, and T. Brandes, “Nonequilibrium thermodynamics in the strong coupling and non-Markovian regime based on a reaction coordinate mapping,” New Journal of Physics 18, 073007 (2016).
• A. Nazir and G. Schaller, “The reaction coordinate mapping in quantum thermodynamics,” in Thermodynamics in the Quantum Regime, Springer (2018).
• P. Talkner and P. Hänggi, “Colloquium: Statistical mechanics and thermodynamics at strong coupling: Quantum and classical,” Reviews of Modern Physics 92, 041002 (2020).