What changes when the spin system remembers its environment?
The \(T_1/T_2\) picture is useful, but it hides the microscopic origin of relaxation. Open-system theory asks how a selected spin subsystem evolves when it is coupled to degrees of freedom that we do not explicitly keep.
Distinguish phenomenological Lindblad dynamics from microscopic weak-coupling relaxation theory.
State the physical content of the Born, Markov and secular approximations.
Explain what a memory kernel changes and when a time-nonlocal description becomes relevant.
System + environment
Start from a larger closed problem
Open-system language separates what you keep from what you average over
What it is: The degrees of freedom whose quantum state you want to predict explicitly—for example the electron and nuclear spins of a radical pair.
What it changes: Its Hamiltonian defines the coherent part of the dynamics.
What you observe: System observables such as spin populations, coherence, magnetization or reaction yield.
What it is: All other degrees of freedom that interact with the system but are not propagated explicitly, such as molecular vibrations, solvent motion or protein fluctuations.
What it changes: It can exchange energy with the spin system and randomize phases, generating relaxation and memory effects.
What you observe: Finite \(T_1\), \(T_2\), line broadening, stochastic shifts and non-exponential decay.
Conceptually, divide the full Hamiltonian into system, bath and coupling terms:
The reduced density operator is obtained by tracing over the bath,
The hard part is that eliminating the bath generally leaves both dissipation and memory in the remaining equation of motion.
Markovian dynamics
Lindblad form gives a controlled time-local generator
A Lindblad operator names a relaxation channel, while its rate says how strongly it acts
What it is: An operator specifying which state change or dephasing process the environment induces.
What it changes: It determines the structure of population transfer or coherence loss while preserving a valid density matrix in the GKSL form.
What you observe: Specific decay pathways, steady states and characteristic relaxation modes.
What it is: The timescale assigned to that environmental channel.
What it changes: It controls how rapidly the corresponding dissipative process competes with coherent Hamiltonian motion.
What you observe: Exponential or multi-exponential decay constants and linewidth contributions.
A widely used Markovian master equation has the Gorini–Kossakowski–Sudarshan–Lindblad form
The dissipator is constructed so that the dynamics remains trace preserving and completely positive. The operators \(L_k\) encode specific channels such as relaxation or dephasing.
Bloch–Redfield–Wangsness
Connect fluctuating interactions to relaxation rates
The common approximations are physical timescale statements
What it is: The system–bath interaction is weak enough that the bath is only weakly perturbed by the system and correlations can be treated perturbatively.
What it changes: It allows relaxation rates to be expressed to low order in the fluctuating interaction.
What you observe: A regime where relaxation is slow compared with the microscopic bath dynamics.
What it is: The bath loses memory much faster than the system state changes.
What it changes: The future depends effectively on the current reduced state rather than its detailed history.
What you observe: Approximately exponential relaxation with no pronounced memory-induced revival.
What it is: Rapidly oscillating couplings between well-separated transition frequencies are neglected.
What it changes: It decouples many density-matrix components and often yields a simpler, more stable relaxation generator.
What you observe: Failure can appear near degeneracies where coherences and populations remain dynamically coupled.
BRW theory starts from a weak system–bath interaction and expresses relaxation through correlation functions or spectral densities of the fluctuating Hamiltonian. In schematic form,
The standard derivation uses weak coupling and a Born–Markov approximation. A secular approximation is often added, but it is a separate approximation and should not be silently assumed when near-degenerate levels make non-secular terms important.
Memory kernels
Nakajima–Zwanzig keeps the past explicitly
Non-Markovianity means the environment can feed information back on the relevant timescale
What it is: A function that weights how strongly earlier reduced states influence the present derivative.
What it changes: It makes the dynamics time-nonlocal and can produce non-exponential decay, oscillations or partial revivals.
What you observe: History-dependent relaxation and deviations from simple single-rate kinetics.
What it is: Correlations already present between system and environment at the chosen initial time.
What it changes: They can contribute an inhomogeneous term and invalidate the assumption of a factorized initial state.
What you observe: Early-time transients that cannot be reproduced by a memoryless model started from the same reduced state.
Projection-operator methods can produce a time-nonlocal equation of the schematic form
The memory kernel \(\mathcal K\) says that the derivative at the current time can depend on the state at earlier times. The inhomogeneous term \(I(t)\) contains effects of initial system–bath correlations in the general formulation.
Interactive
What does finite memory do to a simple decay law?
Markovian versus finite-memory decay
This scalar model is not a complete quantum master equation. It is a deliberately simple way to see how a finite memory time changes an otherwise exponential decay.
The memory is finite but still short enough that the response remains overdamped.
For the exponential kernel, the scalar equation is equivalent to \(\tau_m\ddot x+\dot x+\gamma x=0\) with \(x(0)=1\) and \(\dot x(0)=0\). Negative values in the underdamped regime are why \(x\) should be read as an amplitude-like toy variable, not automatically as a population.
Stochastic unravelings
One density-matrix equation can correspond to many trajectory pictures
Some Markovian master equations can be represented by an ensemble of stochastic pure-state trajectories. Quantum-jump and diffusive stochastic Schrödinger equations are examples. Averaging the trajectories recovers the density operator,
This can be computationally attractive because each trajectory contains \(D\) amplitudes instead of \(D^2\) density-matrix elements. The tradeoff is stochastic sampling error.
Choosing a method
Match the approximation to the timescales
Good when the relevant decay channels and rates are known and a Markovian description is adequate.
Useful for weak fluctuating interactions with sufficiently short bath memory and known spectral densities.
Useful when memory is central and a time-nonlocal description is needed.
Useful when molecular trajectories directly provide the fluctuating interactions and you want to propagate that time dependence.
Selected reading
Examples from my work
Modeling spin relaxation in complex radical systems using MolSpin
Open-system spin dynamics and relaxation in complex radical systems.
J. Comput. Chem. (2023) →Spin Dynamics of Radical Pairs Using the Stochastic Schrödinger Equation in MolSpin
Stochastic state-vector propagation as an efficient open-system route.
J. Chem. Theory Comput. (2024) →Weak Radiofrequency Field Effects on Biological Systems Mediated through the Radical Pair Mechanism
A broader view of coherence, relaxation and RF perturbations in radical-pair systems.
Chemical Reviews (2025) →Key external literature
Where to read next
These are deliberately selected from outside my own work: foundational papers or reviews that are especially useful for this topic.
On Quantum Theory of Transport Phenomena: Steady Diffusion
S. Nakajima · Progress of Theoretical Physics (1958). One of the foundational projection-operator formulations behind non-Markovian reduced dynamics.
Open DOI →Ensemble Method in the Theory of Irreversibility
R. Zwanzig · The Journal of Chemical Physics (1960). The complementary projection-operator formulation leading to generalized kinetic equations with memory.
Open DOI →On the Generators of Quantum Dynamical Semigroups
G. Lindblad · Communications in Mathematical Physics (1976). The canonical characterization of completely positive Markovian quantum generators.
Open DOI →