# Ergodicity and convergence to equilibrium of continuous-time Markov processes

A continuous-time Markov process is ergodic when its distribution converges, as time grows, to a stationary distribution that the process then keeps forever. This article covers how recurrence and transience are defined for continuous-time Markov chains (CTMCs), when a stationary distribution exists, the ergodic theorem in its transition-probability and time-average forms, and the rates at which convergence happens, from exponential decay governed by a spectral gap (gap in operator eigenvalues controlling exponential convergence speed) to the subgeometric rates forced by rare excursions on unbounded state spaces.

| Key fact | Statement |
|---|---|
| Recurrence | A CTMC state is recurrent if it is recurrent for the embedded jump chain, i.e. \(P_i(\tau_i < \infty) = 1\); otherwise transient<sup>[1](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)</sup><sup> • </sup><sup>[2](https://hairer.org/notes/Convergence.pdf)</sup> |
| Positive recurrence | State i is positive recurrent if \(E_i[R_i] < \infty\) for the embedded chain's return time; otherwise null recurrent<sup>[1](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)</sup> |
| Stationary distribution | An irreducible CTMC with generator Q is positive recurrent if and only if a distribution π with π⊤Q = 0 exists<sup>[1](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)</sup> |
| Ergodic theorem | For irreducible positive recurrent CTMCs, \(p_{ij}(t) \to \pi(j)\) for all states i, j<sup>[1](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)</sup> |
| Exponential convergence | A Poincaré inequality with constant C_P gives \(\|P_t f − \int f \, d\mu\|^2 \le e^{−t/C_P}\|f − \int f \, d\mu\|^2\)<sup>[3](https://ar5iv.labs.arxiv.org/html/math/0703355)</sup> |
| Drift recipe | LW ≤ α − βW (with small-set/irreducibility companions) is the standard Foster–Lyapunov condition for exponential ergodicity<sup>[4](https://luc-umass.github.io/pdf/EMP.pdf)</sup> |
| Recent result | A 2025 path method proves positive spectral gaps for CTMCs on \(\mathbb{Z}^d\) without reversibility, and shows every open complex-balanced reaction network is exponentially ergodic<sup>[5](https://people.math.wisc.edu/~dfanderson/papers/2025/ACFK_Mixing.pdf)</sup> |

## Recurrence and transience in continuous time

Recurrence for a continuous-time chain is most cleanly read off the embedded jump chain, the discrete-time chain of successive visited states. A state is recurrent for the [CTMC](https://www.edgepedia.org/continuous-time-markov-chain) if it is recurrent for this embedded process; if not, it is transient<sup>[1](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)</sup>. Equivalently, with \(\tau_i\) the return time to \(i\), state \(i\) is transient if \(P_i(\tau_i = \infty) > 0\), recurrent if \(P_i(\tau_i < \infty) = 1\), and positive recurrent if it is recurrent and \(E_i \, \tau_i < \infty\)<sup>[2](https://hairer.org/notes/Convergence.pdf)</sup>. Recurrence and transience are class properties: within a communicating class, all states share the same status<sup>[1](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)</sup>.

Continuous time introduces a subtlety absent in discrete time. There are two natural return-time quantities, the return time \(T_i\) measured by the CTMC itself and the return time \(R_i\) of the embedded chain, and it is quite possible to have \(E_i[T_i] < \infty\) while \(E_i[R_i] = \infty\), or the reverse<sup>[1](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)</sup>. Which mean is finite therefore matters for the definitions. A further quirk under this embedded-chain convention: an absorbing state of a CTMC counts as transient, since the jump chain never makes a return jump to it, although the CTMC itself stays at the absorbing state forever, the standard continuous-time definition of recurrence<sup>[1](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)</sup>.<sup>[13](https://www.sjsu.edu/faculty/guangliang.chen/Math263/lec3classification.pdf)</sup>

Generator-based criteria exist for infinite state spaces: two sets of conditions on the Q-matrix of an irreducible countably infinite Markov process ensure that Q is regular (no explosion), one set additionally implying ergodicity and the other recurrence; these parallel discrete-time chain conditions<sup>[6](https://doi.org/10.1017/s0305004100051562)</sup>.

## Positive and null recurrence, and existence of a stationary distribution

The bridge between recurrence and equilibrium is the mean return time. A recurrent state is positive recurrent if the embedded process's mean return time \(E_i[R_i]\) is finite, and null recurrent otherwise<sup>[1](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)</sup>.

For an irreducible CTMC with generator Q, the classification theorem is exact: the chain is positive recurrent **if and only if** there exists a probability distribution π satisfying π⊤Q = 0<sup>[1](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)</sup>. This answers why a chain can be recurrent yet have no equilibrium: null recurrence means returns happen with probability one but their mean is infinite, and no stationary probability distribution exists. For comparison, in the discrete-time setting, an irreducible discrete-time chain that is recurrent has a unique invariant probability measure, strictly positive at every state<sup>[7](https://perso.math.univ-toulouse.fr/cattiaux/files/2013/11/Cattiaux_tempslong_revise.pdf)</sup>.

## The ergodic theorem

For an irreducible positive recurrent CTMC, the transition probabilities converge: \(p_{ij}(t) \rightarrow \pi(j)\) as \(t \rightarrow \infty\) for any states \(i\), \(j\), where \(\pi\) is the invariant distribution<sup>[1](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)</sup>. The starting state is forgotten completely.

There is a pathwise version. For a stationary version of the process with law \(P_{\mu}\), ergodicity is equivalent to the time-average convergence \(\int_0^t f(X_s)\,ds \to \int f\,d\mu\) almost surely as \(t \to \infty\)<sup>[8](https://wt.iam.uni-bonn.de/fileadmin/WT/Inhalt/people/Andreas_Eberle/MarkovProcesses1920/MarkovProcesses1920.pdf)</sup>. The limit law is not arbitrary: any distribution approached in the long-time limit must be stationary, satisfying \(\mu(B) = \int \mu(dx)\, p_t(x,B)\) for all \(t\)<sup>[8](https://wt.iam.uni-bonn.de/fileadmin/WT/Inhalt/people/Andreas_Eberle/MarkovProcesses1920/MarkovProcesses1920.pdf)</sup>.

For general state spaces, a Markov process is called ergodic if an invariant probability π exists and \(\|P_t(x,\cdot) - \pi\| \to 0\) as \(t \to \infty\) for all x<sup>[9](https://faculty.eng.ufl.edu/meyn/wp-content/uploads/sites/671/archive/spm_files/Papers_pdf/meytweIII.pdf)</sup>, where the norm denotes a distributional distance such as total variation.

## Lyapunov drift versus Poincaré and spectral methods

Two broad machineries dominate the analysis of convergence rates, and they answer slightly different questions.

**The Foster–Lyapunov route** works with a test (Lyapunov) function W on the state space and conditions on the generator's action. In a standard ladder<sup>[4](https://luc-umass.github.io/pdf/EMP.pdf)</sup>:

- LW ≤ α + βW guarantees non-explosive solutions;
- LW ≤ −α implies existence of an invariant measure;
- LW ≤ α − βW is the drift condition for exponential ergodicity.

Meyn and Tweedie's continuous-time theory develops such test-function criteria for non-explosivity, non-evanescence, Harris recurrence, and positive Harris recurrence, proved systematically via Dynkin's formula<sup>[9](https://faculty.eng.ufl.edu/meyn/wp-content/uploads/sites/671/archive/spm_files/Papers_pdf/meytweIII.pdf)</sup>. Combined with small sets, a geometric drift condition yields convergence to a unique invariant measure at an exponential rate; this theorem dates to Harris (1956) and extends Doeblin's ideas to unbounded state spaces<sup>[2](https://hairer.org/notes/Convergence.pdf)</sup>. More generally, if a \(\varphi\)-[Lyapunov function](https://www.edgepedia.org/lyapunov-function) bounded on a petite set is combined with irreducibility, one obtains \(\|P_t(x,\cdot) - \mu\|_{\mathrm{TV}} \le c \cdot V(x) \cdot \psi(t)\) for all \(x\); when \(\varphi\) is linear, \(\psi(t) = e^{-\rho t}\) for some positive explicit \(\rho\), and sub-geometric or polynomial rates can also be handled<sup>[3](https://ar5iv.labs.arxiv.org/html/math/0703355)</sup>. Lyapunov techniques essentially reduce a complicated Markov process to a one-dimensional one, with all information encoded in \(V(x_t)\), and subgeometric convergence is dominated by relatively rare excursions<sup>[2](https://hairer.org/notes/Convergence.pdf)</sup>. Under the T-condition, the Foster–Lyapunov criterion is not merely sufficient but necessary and sufficient for existence of an invariant probability measure<sup>[10](https://doi.org/10.1017/s0021900200002096)</sup>.

**The Poincaré/spectral route** works with functional inequalities. A [Poincaré inequality](https://www.edgepedia.org/poincare-inequality) with constant C_P implies exponential L² decay,

\(\|P_t f - \int f \, d\mu\|^2 \leq e^{-t/C_P} \|f - \int f \, d\mu\|^2\),

hence exponential convergence in total variation for initial laws with L² density with respect to μ<sup>[3](https://ar5iv.labs.arxiv.org/html/math/0703355)</sup>.

**Where they differ in sharpness.** An important drawback of the Meyn–Tweedie approach is that, in general, there is no explicit control of the constant c in the bound c V(x) ψ(t); Lyapunov–Poincaré inequalities were introduced precisely to give explicit constants starting from the same drift condition<sup>[3](https://ar5iv.labs.arxiv.org/html/math/0703355)</sup>. These new inequalities link the Meyn–Tweedie Lyapunov approach and Poincaré-type functional inequalities<sup>[3](https://ar5iv.labs.arxiv.org/html/math/0703355)</sup>.

## Exponential convergence, spectral gaps and mixing

[Total variation distance](https://www.edgepedia.org/total-variation-distance) is the standard gauge of convergence: \(\|P_t(x,\cdot) - \pi\|_{\mathrm{TV}}\), and the mixing time at level ε for initial state x is \(\tau_\varepsilon^x = \inf\{t \geq 0 : \|P_t(x,\cdot) - \pi\|_{\mathrm{TV}} \leq \varepsilon\}\)<sup>[5](https://people.math.wisc.edu/~dfanderson/papers/2025/ACFK_Mixing.pdf)</sup>.

For countable-state CTMCs, two standard routes to exponential ergodicity are Foster–Lyapunov functions and spectral-gap analysis<sup>[5](https://people.math.wisc.edu/~dfanderson/papers/2025/ACFK_Mixing.pdf)</sup>. When a spectral gap \(C > 0\) is established, one obtains bounds of the form \(\|P_t(x,\cdot) - \pi\|_{\mathrm{TV}} \leq \sqrt{2/\pi(x)} \, e^{-2Ct}\), giving mixing times of order \(|\ln \pi(x)|\)<sup>[5](https://people.math.wisc.edu/~dfanderson/papers/2025/ACFK_Mixing.pdf)</sup>.

A third, older route is minorization. If some time-T map S_T of a stochastic semigroup satisfies the Doeblin condition, the semigroup has a unique equilibrium μ* and converges to it<sup>[11](https://ar5iv.labs.arxiv.org/html/2110.09650)</sup>; Harris/Meyn–Tweedie theorems give both existence of a unique invariant measure and the speed of convergence, exponential or subgeometric, with constructive subgeometric estimates<sup>[11](https://ar5iv.labs.arxiv.org/html/2110.09650)</sup>. Strong stationary times offer a complementary practical tool: for continuous-time chains, they provide bounds on separation distance and so aid in the analysis of mixing rates, extending the Aldous–Diaconis (1987) discrete-time connection<sup>[12](https://www.cambridge.org/core/journals/probability-in-the-engineering-and-informational-sciences/article/abs/time-to-stationarity-for-a-continuoustime-markov-chain/BDCCC1463C6223EDA28BB166D4538862)</sup>.

## General state spaces and diffusions

On unbounded or continuous state spaces, recurrence must be redefined, and the Harris framework replaces state-wise return criteria. The convergence theorem quoted above, exponential convergence to a unique invariant measure under a geometric drift condition plus sufficiently large small sets, is exactly the Harris (1956) extension of Doeblin<sup>[2](https://hairer.org/notes/Convergence.pdf)</sup>, and stochastic-semigroup versions of these Harris-type results give both the unique equilibrium and the convergence speed, with new constructive estimates in the subgeometric case<sup>[11](https://ar5iv.labs.arxiv.org/html/2110.09650)</sup>.

For diffusions specifically, a substantial toolkit exists for determining at which speed, if at all, the law approaches stationarity, with particular interest in subexponential rates; the tools include Lyapunov techniques and, for degenerate diffusions, [Malliavin calculus](https://www.edgepedia.org/malliavin-calculus) and Hörmander sums-of-squares conditions<sup>[2](https://hairer.org/notes/Convergence.pdf)</sup>. Subgeometric rates in these settings are dominated by relatively rare excursions far from the center of the state space<sup>[2](https://hairer.org/notes/Convergence.pdf)</sup>.

## How this compares with the stationarity and hitting-time siblings

This article's subject is convergence to equilibrium and its speed. The sibling leaf on stationarity, reversibility, and detailed balance treats the complementary characterization problem: finding and structurally characterizing π (for instance via detailed balance), rather than proving that P_t(x,·) approaches it. The distinction is visible in the theorems themselves: π⊤Q = 0 characterizes π for a positive recurrent chain<sup>[1](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)</sup>, while p_ij(t) → π(j)<sup>[1](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)</sup>, the definition of ergodicity via ||P_t(x,·) − π|| → 0<sup>[9](https://faculty.eng.ufl.edu/meyn/wp-content/uploads/sites/671/archive/spm_files/Papers_pdf/meytweIII.pdf)</sup>, and the spectral-gap bounds<sup>[3](https://ar5iv.labs.arxiv.org/html/math/0703355)</sup> all concern the approach to π. Boundary behavior, meaning hitting times, absorption, and extinction, belongs to the hitting-time and absorption sibling; the transience of absorbing CTMC states<sup>[1](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)</sup> marks precisely where that sibling's questions begin and this one's end.

## References

1. [Continuous-time homogeneous Markov chains, lecture notes, University of Copenhagen](https://web.math.ku.dk/~susanne/kursusstokproc/ContinuousTime.pdf)
2. [Convergence of Markov processes, Martin Hairer lecture notes](https://hairer.org/notes/Convergence.pdf)
3. [Rate of convergence for ergodic continuous Markov processes: Lyapunov versus Poincaré](https://ar5iv.labs.arxiv.org/html/math/0703355)
4. [Ergodic Properties of Markov Processes, lecture notes following Meyn–Tweedie](https://luc-umass.github.io/pdf/EMP.pdf)
5. [A new path method for exponential ergodicity of Markov processes on Z^d, with applications to stochastic reaction networks, Anderson et al., 2025](https://people.math.wisc.edu/~dfanderson/papers/2025/ACFK_Mixing.pdf)
6. [Sufficient conditions for regularity, recurrence and ergodicity of Markov processes, Math. Proc. Camb. Phil. Soc.](https://doi.org/10.1017/s0305004100051562)
7. [Long time behavior of Markov processes, P. Cattiaux, review](https://perso.math.univ-toulouse.fr/cattiaux/files/2013/11/Cattiaux_tempslong_revise.pdf)
8. [Markov Processes, Andreas Eberle lecture notes, Bonn](https://wt.iam.uni-bonn.de/fileadmin/WT/Inhalt/people/Andreas_Eberle/MarkovProcesses1920/MarkovProcesses1920.pdf)
9. [Stability of Markovian Processes III: Foster–Lyapunov Criteria for Continuous-Time Processes, Meyn & Tweedie](https://faculty.eng.ufl.edu/meyn/wp-content/uploads/sites/671/archive/spm_files/Papers_pdf/meytweIII.pdf)
10. [Ergodic properties and ergodic decompositions of continuous-time Markov processes, Journal of Applied Probability](https://doi.org/10.1017/s0021900200002096)
11. [Harris-type results on geometric and subgeometric convergence to equilibrium for stochastic semigroups](https://ar5iv.labs.arxiv.org/html/2110.09650)
12. [Time to Stationarity for a Continuous-Time Markov Chain, Probability in the Engineering and Informational Sciences](https://www.cambridge.org/core/journals/probability-in-the-engineering-and-informational-sciences/article/abs/time-to-stationarity-for-a-continuoustime-markov-chain/BDCCC1463C6223EDA28BB166D4538862)
13. [Lec3classification (sjsu.edu)](https://www.sjsu.edu/faculty/guangliang.chen/Math263/lec3classification.pdf)

---
*Topic: Encyclopedia › Physical world and mathematics › Mathematics and statistics › Statistics and probability › Stochastic processes › Markov chains and processes › Continuous-time Markov processes › Ergodicity, recurrence, and convergence to equilibrium*

*Initially written Sep 17, 2026 · Reviewed: Sep 30, 2026 · Edited: Sep 30, 2026; Oct 11, 2026; Oct 15, 2026 · Last review: Sep 30, 2026*

*Copyright 2026 EdgeChat AI, a subsidiary of Biostate AI.*

License: Edgepedia Community License 1.0, https://www.edgepedia.org/edgepedia/license