---
title: Submarkovian Semigroups
url: https://www.emergentmind.com/topics/submarkovian-semigroups
type: topic
---

# Submarkovian Semigroups

A submarkovian semigroup is a one-parameter family of linear or nonlinear operators on a function space that encodes the time-evolution of processes preserving positivity, non-expansiveness in norm, and contractivity on $L^\infty$. These semigroups generalize Markovian dynamics by allowing for a form of non-conservativity (i.e., $T_t1\leq1$ rather than $T_t1=1$), and arise as fundamental objects in the study of partial differential equations, Dirichlet forms, stochastic processes (including both diffusions and jump processes), and nonlocal and nonlinear phenomena. The concept bridges analysis, probability, and geometry, appearing in both commutative and noncommutative setups, linear and nonlinear operator theory, and on spaces as diverse as Euclidean domains, manifolds, Lie groups, graphs, and metric measure spaces.

## 1. Formal Definitions and Core Properties

A family $\{T_t\}_{t\ge0}$ of linear operators on $L^2(\Omega, \mu)$, for a $\sigma$-finite measure space $(\Omega,\mu)$, is a submarkovian semigroup if the following conditions hold:
- $T_t$ is positivity-preserving: $f\geq0$ implies $T_t f\geq0$.
- $T_t$ is $L^\infty$-contractive on $L^2$, i.e., $\|T_t f\|_\infty\leq\|f\|_\infty$ for $f\in L^2\cap L^\infty$.
- $T_t$ is a contraction: $\|T_t f\|_2\leq\|f\|_2$ for all $f\in L^2$.
- $T_t1\leq1$ almost everywhere.

On $L^p(\Omega, \mu)$, $1<p<\infty$, such semigroups interpolate to strongly continuous contraction semigroups $T_p(t)$ with a generator $-A_p$. For a symmetric submarkovian semigroup, each $T_t$ is self-adjoint on $L^2$. For the nonlinear setting on a real Hilbert space $H$, the semigroup $S(t)$ is submarkovian if it is order-preserving and $L^\infty$-contractive:
\[
u \leq v \implies S(t)u \leq S(t)v, \quad \|S(t)u - S(t)v\|_{L^\infty} \leq \|u-v\|_{L^\infty}.
\]
If further $T_t1=1$, the semigroup is called Markovian [1603.08862, 1910.03979, 2601.18028].

## 2. Generators, Functional Calculus, and Numerical Range

Given a submarkovian semigroup $T_p(t)$, its generator $-A_p$ is sectorial, and its numerical range $W(A_p)$ can be precisely localized. For $1<p<\infty$, define the mapping 
\[
F_p(z) = 
\begin{cases}
z|z|^{p-2}, & z\neq0,\\
0, & z=0,
\end{cases}
\]
and the sector $\Sigma(\varphi_0)$ with
\[
\varphi_0 = \arcsin|1-2/p|, \qquad \Sigma_p:=\Sigma(\varphi_0).
\]
Key result: for symmetric $L_\infty$-contractive semigroups, $W(A_p)\subseteq\Sigma_p$, so $T_p(t)$ extends to a bounded analytic semigroup on $\Sigma(\pi/2 - \varphi_0)$. The angle $\varphi_0$ is optimal and critical for maximal $L^p$-regularity and for the boundedness of the $H^\infty$ functional calculus on sectors [1603.08862].

In weighted $L^2(\Omega, wd\mu)$ spaces, the $Q^A_2(w)$ characteristic controls the $H^\infty$-calculus for the generator. For $w:\Omega\to(0,\infty)$,
\[
Q^A_2(w) = \sup_{t>0} \|T_t(w)T_t(w^{-1})\|_{L^\infty(\Omega)}.
\]
For $\theta>\pi/2$, the generator $A$ admits a bounded $H^\infty(\Sigma_\theta)$-calculus with norm linearly depending on $Q_2^A(w)$ [1910.03979].

## 3. Submarkovian Semigroups on Groups, Manifolds, and Graphs

On compact Lie groups, submarkovian semigroups naturally arise as the $L^2$-extensions of Feller semigroups on $C(G)$. The Ruzhansky–Turunen theory identifies the generators as pseudo-differential operators with matrix-valued symbols, explicitly characterized for Lévy-type operators:
\[
\sigma(x,\pi) = i b^i(x)d\pi(X_i) - \frac{1}{2}a^{ij}(x)d\pi(X_i)d\pi(X_j) + \int_{G\setminus\{e\}} (\pi(x^{-1}y)-I-x^i(x^{-1}y)d\pi(X_i))\nu(x,dy).
\]
Symmetry and the structure of the Dirichlet form follow from specific conditions on the drift $b^i(x)$ and symmetry of the Lévy kernel $\nu(x, dy)$. The resulting Dirichlet form exhibits both diffusion and jump components:
\[
\mathcal{E}(f,g) = \int_G a^{ij}(x)X_i f(x) X_j g(x)\, dx + \iint_{G\times G \setminus \{e\}} [f(y)-f(x)][g(y)-g(x)]\nu(x,dy)dx.
\]
This construction encompasses isotropic stable processes, combined diffusion-jump processes, and pseudo-Poisson processes [1101.5030].

On metric random walk spaces, graphs, or spaces associated with pseudodifferential operators of nonlocal or fractional type, submarkovian semigroups are produced by self-adjoint, order-preserving nonlinear (or linear) generators under convexity and monotonicity criteria on the associated energy functionals [2601.18028].

## 4. Nonlinear, Nonlocal, and Fractional Submarkovian Semigroups

Submarkovianity extends to nonlinear (monotone) semigroups associated with subdifferentials of convex lower semicontinuous functionals in Hilbert spaces. If $\varphi:H\to(-\infty, +\infty]$ is proper, convex, and lower semicontinuous, then the negative subgradient $-\partial\varphi$ is $m$-accretive and generates a nonlinear contraction semigroup via the evolution equation
\[
u'(t) + \partial\varphi(u(t)) \ni 0.
\]
When $\varphi$ also satisfies the Dirichlet form-like inequality with respect to normal contractions,
\[
\varphi(u-p(u-v)) + \varphi(v+p(u-v)) \leq \varphi(u) + \varphi(v),
\]
the resulting nonlinear semigroup is submarkovian and order-preserving. This framework includes fractional Laplacians, graph $p$-Laplacians, metric random walks, Lévy processes, and integral operators associated with nonlocal energies [2601.18028].

Ultracontractivity, domination, and regularization estimates are obtainable under additional functional inequalities (logarithmic-Sobolev or Nash inequalities), providing strong analytic regularization, $L^p\to L^q$ bounds, and Hölder regularization for the semigroup.

## 5. Analyticity, Optimal Angles, and Functional Calculus Limits

The angle of analyticity for symmetric $L_\infty$-contractive semigroups on $L_p$ is given by $\varphi_0=\arcsin|1-2/p|$ and is provably sharp, with the extremal example furnished by the symmetric two-point semigroup. In weighted $L^2$ settings, the optimal angle for bounded $H^\infty$-calculus is generally no smaller than $\pi/2$, with angle reduction possible only under further "power-bump" conditions on the weight. There are explicit counterexamples showing the impossibility of further reduction of the calculus angle for generic semigroup-weight pairs; namely, there exist semigroups and weights for which no Hörmander functional calculus of any order below $\theta=\pi/2$ holds [1603.08862, 1910.03979].

Such sectoriality and analyticity results are central to $H^\infty$ spectral multiplier theorems, maximal $L^p$-regularity for parabolic evolution equations, and operator-theoretic approaches to PDEs with rough coefficients.

## 6. Illustrative Examples and Applications

Prominent families of submarkovian semigroups include:
- The heat semigroup (and subordinated variants) on Euclidean domains, manifolds, or Lie groups.
- Pure jump processes governed by stable Lévy flights or pseudo-differential generators.
- Sub-Laplacian semigroups on doubling metric spaces with Gaussian kernel bounds.
- Schrödinger and Ornstein–Uhlenbeck semigroups in analysis and probability.
- Nonlocal, nonlinear semigroups on graphs, metric random walk spaces, or Orlicz spaces.

Applications span maximal regularity theory for parabolic PDEs, quantitative spectral multiplier estimates, probabilistic potential theory, and regularity for nonlocal and nonlinear diffusion equations arising in models of fractional diffusion, anomalous transport, and random environments [1101.5030, 1910.03979, 2601.18028].

## 7. Open Problems and Extensions

Active research continues in the following directions:
- Characterization of submarkovian semigroups for general non-symmetric, nonlocal Dirichlet forms, and for operators with non-trivial boundary/exterior conditions (intrinsic capacity).
- Analysis of semigroup regularization, Hölder and Harnack inequalities on rough and fractal domains.
- Extension of the Beurling–Deny and domination principles to broader classes of nonlinear and nonlocal evolution equations.
- Numerical analysis and stability of discretizations for nonlinear and nonlocal submarkovian semigroups, especially on combinatorial structures such as graphs.

The structure, analytic theory, and applications of submarkovian semigroups continue to provide a vital analytical toolkit for a wide range of linear and nonlinear, local and nonlocal, deterministic and stochastic evolution phenomena [1603.08862, 1101.5030, 1910.03979, 2601.18028].

Source: https://www.emergentmind.com/topics/submarkovian-semigroups