---
title: Quantum Markov Semigroups
url: https://www.emergentmind.com/topics/quantum-markov-semigroups
type: topic
---

# Quantum Markov Semigroups

A quantum Markov semigroup (QMS) is a one-parameter family of completely positive, identity-preserving maps acting on a von Neumann (or $C^*$-) algebra, generalizing classical Markov semigroups to the noncommutative (quantum) setting. QMSs model the reduced evolution of open quantum systems, including irreversible decoherence, relaxation, and thermalization. Their mathematical structure and physical interpretation are fundamentally linked to quantum stochastic calculus, operator algebras, noncommutative probability, and quantum statistical mechanics.

## 1. Formal Definition and Generator Structure

Let $\mathcal{M}$ be a (not necessarily finite-dimensional) von Neumann algebra with predual $\mathcal{M}_*$. A quantum Markov semigroup (QMS) is a family $(T_t)_{t \ge 0}$ of normal, completely positive, unital maps $T_t:\mathcal{M}\to\mathcal{M}$ satisfying
- $T_0 = \mathrm{id}$
- $T_{s+t} = T_s \circ T_t$ for all $s, t \geq 0$
- $t \mapsto T_t(A)$ is $\sigma$-weakly continuous for each $A \in \mathcal{M}$
- $T_t(1) = 1$ for all $t \geq 0$.

The generator $\mathcal{L}$ (when densely defined) is given by the strong limit
\[
\mathcal{L}(A) = \lim_{t \downarrow 0} \frac{T_t(A) - A}{t}
\]
and, on a suitable dense $*$-subalgebra, it admits the Gorini–Kossakowski–Sudarshan–Lindblad (GKSL) form:
\[
\mathcal{L}(A) = i[H, A] + \sum_j \left(V_j^* A V_j - \tfrac12 \{V_j^* V_j, A\}\right)
\]
where $H = H^*$ is the "Hamiltonian" part and $\{V_j\}$ are bounded or (in infinite dimension) possibly unbounded "noise" operators [2508.06396, 1406.3417].

In finite dimensions ($M_n(\mathbb{C})$), every uniformly continuous QMS generator is of this form, and the associated predual semigroup $T_t^*$ acts on states (density matrices), evolving them according to $\mathcal{L}^*$:
\[
\mathcal{L}^*(\rho) = -i[H, \rho] + \sum_j \left(V_j \rho V_j^* - \tfrac12\{V_j^* V_j, \rho\}\right)
\]
[2508.06396, 2201.05094, 1405.6374].

## 2. Irreducibility, Ergodicity, and Positive Recurrent Subspaces

Irreducibility of a QMS is equivalent (under suitable conditions) to the absence of nontrivial subharmonic projections, or, physically, the inability to confine dynamics to a proper, nontrivial invariant subspace. In finite dimensions, irreducible QMSs are primitive, admitting a unique faithful invariant state and a strict spectral gap [1405.6374, 2509.20133]. For infinite-dimensional systems, the relevant structure is given by the positive recurrent subspace $\mathcal{R}_+$: the supremum of supports of invariant normal states. The Frigerio–Verri–Carbone–Girotti ergodic theorem characterizes ergodicity in terms of the absorption and invariance properties of $\mathcal{R}_+$, granting an explicit description of convergence to equilibrium or reduction to minimal globally asymptotically stable subspaces [2509.20133].

The spectrum of the predual generator $\mathcal{L}^*$ determines relaxation rates. Under irreducibility/primitivity, $0$ is a simple eigenvalue (stationary state), and all other eigenvalues have negative real part, with a spectral gap $\lambda_1>0$ controlling exponential convergence in trace distance and various divergences [2206.11923].

## 3. Stationary, Quasi-Stationary, and Quasi-Stationary States (QSS)

- A **stationary state** $\rho_0$ satisfies $T_t^*(\rho_0) = \rho_0$ for all $t$, i.e., it is a fixed point: $\mathcal{L}^*(\rho_0) = 0$.
- A **quasi-stationary state** (QSS) $\rho$ with decay rate $\lambda>0$ is a positive normal state satisfying
  \[
  T_t^*(\rho) = e^{-\lambda t} \rho + o(e^{-\lambda t}), \quad \text{as } t \to \infty
  \]
  with $||T_t^*(\rho) - e^{-\lambda t} \rho||_1 = o(e^{-\lambda t})$. Typically, $\operatorname{supp}\rho$ is disjoint from any stationary state's support [2508.06396].

QSS characterize the leading corrections to equilibrium; they correspond to eigencomponents of $\mathcal{L}^*$ with strictly negative real eigenvalue, controlling subexponential relaxation and decay of "meta-stable" structures.

## 4. Spectral Characterization and Asymptotic Behavior

Let $\mathcal{L}^*$ be the generator acting on normal states. For primitive QMS:
- The spectrum of $\mathcal{L}^*$ is contained in $\{z: \operatorname{Re} z \leq 0\}$.
- $0$ is a simple eigenvalue with positive eigenvector $\rho_{\mathrm{stat}}$ (the unique stationary state).
- There is $\lambda_1>0$ such that
  \[
  \sigma(\mathcal{L}^*) \cap \{\operatorname{Re} z \geq -\lambda_1\} = \{0, -\lambda_1\}
  \]
- The corresponding eigenprojector $P_{-\lambda_1}$ targets the unique (up to scalar) quasi-stationary state.

The spectral expansion:
\[
T_t^* = |\rho_{\mathrm{stat}}\rangle\langle \mathbf{1}| + e^{-\lambda_1 t} P_{-\lambda_1} + O(e^{-\lambda_2 t}), \quad \lambda_2 > \lambda_1
\]
shows that the QSS dominates long-time (substationary) decay [2508.06396, 2206.11923].

## 5. Detailed Balance, Symmetry, and Gradient Flow Structure

Detailed balance (in its several quantum variants) is a symmetry property relating the QMS and the modular structure associated to a reference state $\sigma$:
- **KMS symmetry (detailed balance):** $T_t$ is self-adjoint on the GNS Hilbert space $L^2(M, \sigma)$, or, for the generator, $\sigma^{1/2} \mathcal{L}(A) \sigma^{-1/2} = \mathcal{L}^*(\sigma^{1/2} A \sigma^{-1/2})$.
- **Bimodule (KMS or GNS) symmetry:** Inclusions $N \subset M$ and associated bimodule QMS allow a rich structure, with directional matrices and "hidden densities" associated to multi-dimensional symmetry backgrounds [2511.04881, 2504.09576].

For ergodic, detailed-balance QMSs, the generator $\mathcal{L}^*$ is the (modular) gradient flow of the quantum relative entropy $D(\cdot \| \sigma)$ with respect to a noncommutative 2-Wasserstein (or more general) Riemannian metric [1609.01254, 2201.05094, 2511.04881]. This structure underpins exponential decay of entropy and a suite of quantum functional and transport inequalities (e.g., modified logarithmic Sobolev, Talagrand, Poincaré).

## 6. Functional Inequalities, Convergence Rates, and Curvature-Dimension

The spectral data and symmetry properties enable the derivation of sharp functional inequalities:
- **Modified Logarithmic Sobolev Inequality (MLSI):** For $C>0$, $D(\rho(t) \| \sigma) \leq e^{-Ct} D(\rho(0)\|\sigma)$ is equivalent to the gradient-flow convexity of the entropy [2505.07549].
- **Transport Inequalities:** Quantum analogs of Talagrand and Ricci lower bounds provide geometric control on the quantum state space [2007.13506, 2105.08303].
- **Poincaré Inequalities:** Spectral gap estimates control the exponential convergence (mixing times) in trace norm and quantum $\chi^2$ divergence [2206.11923].

In suitable settings, the QMS satisfies noncommutative curvature-dimension ($CD(K,N)$) conditions, generalizing Bakry–Émery theory [2105.08303].

## 7. Examples and Advanced Structures

- **Quantum Walks:** The QMS corresponding to continuous-time quantum walks is purely unitary, with GKSL generator $\mathcal{L}(A) = i[H, A]$ and no dissipative part, so it does not exhibit relaxation; only the center persists under long-time averages [1305.1749].
- **Covariant/Relativistic QMS:** Poincaré-covariant QMS constructed via imprimitivity systems yield ergodic QMSs on unitary representations of relativistic systems, particularly with unique stationary states due to transitivity [2102.09984].
- **QMS on Compact Quantum Groups:** Translation-invariant QMS correspond one-to-one to Lévy processes, with symmetry and potential theory characterized via generating functionals invariant under antipode/unitary antipode. This yields a full classification of Dirichlet forms, derivations, and spectral triples for compact and discrete (quantum) groups [1210.6768].

---

**Summary Table: Classical/Quantum QMS Features**

| Feature                      | Classical Markov             | Quantum Markov (QMS)                      |
|------------------------------|------------------------------|-------------------------------------------|
| State space                  | Probability measures         | Density operators (trace class)           |
| Generator                    | $L(f) = Af$ (infinitesimal) | $\mathcal{L}(A) = i[H,A] + \text{diss.}$  |
| Stationary state             | $L^* \mu_0 = 0$              | $\mathcal{L}^*(\rho_0) = 0$               |
| Quasi-stationary state       | $L^* \mu = -\lambda \mu$     | $\mathcal{L}^*(\rho) = -\lambda \rho$     |
| Detailed balance             | $L$ reversible in $L^2(\mu)$ | GNS or KMS symmetry w.r.t. $\sigma$       |
| Gradient flow/entropy        | $H(\nu|\mu)$, Wasserstein-2  | $D(\rho\|\sigma)$, noncomm. metric        |
| Functional inequalities      | Log-Sobolev, Poincaré, etc.  | Quantum MLSI, Talagrand, Poincaré         |
| Ergodicity/sp. gap           | Unique $\mu_0$, gap $\lambda$| Unique $\rho_0$, gap, QSS, etc.           |

---

## References
- "Quasi-stationary normal states for quantum Markov semigroups" [2508.06396]
- "Thermodynamic formalism for continuous-time quantum Markov semigroups: the detailed balance condition, entropy, pressure and equilibrium quantum processes" [2201.05094]
- "On the relationship between a quantum Markov semigroup and its representation via linear stochastic Schr\"odinger equations" [1405.6374]
- "Generators of Quantum Markov Semigroups" [1406.3417]
- "Trace distance ergodicity for quantum Markov semigroups" [2206.11923]
- "Ergodic Properties of Quantum Markov Semigroups" [2509.20133]
- "Gradient flow and entropy inequalities for quantum Markov semigroups with detailed balance" [1609.01254]
- "Bimodule KMS Symmetric Quantum Markov Semigroups and Gradient Flows" [2511.04881]
- "Symmetries of Lévy processes on compact quantum groups, their Markov semigroups and potential theory" [1210.6768]
- Additional sources as cited throughout.

Source: https://www.emergentmind.com/topics/quantum-markov-semigroups