A Pathwise Approach to the Strong Feller Property and Irreducibility of Nonlinear Branching Processes

Pei-Sen Li; Xiangqi Zheng; Xiaowen Zhou

A Pathwise Approach to the Strong Feller Property and Irreducibility of Nonlinear Branching Processes

T0 review · 0 major / 4 minor · reviewed 2026-07-13 · grok-4.5

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A pathwise method proves strong Feller and irreducibility for nonlinear branching processes despite boundary degeneracy and discontinuous jumps, yielding exponential quasi-stationarity or uniform ergodicity.

desk verdict Solid pathwise fix for strong Feller + irreducibility of nonlinear CSNBPs under boundary degeneracy; the applications then fall out cleanly. read the letter →

arxiv 2606.24821 v2 pith:CJNP3LB3 submitted 2026-06-23 math.PR

Pei-Sen Li , Xiangqi Zheng , Xiaowen Zhou This is my paper

classification math.PR MSC 60J8060H1060J76

keywords continuous-statebranchingprocessesdegeneratejump-diffusionsstrongFellerpropertyirreducibilitystate-dependenttime-changequasi-stationarydistributionsuniformexponentialergodicity

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Continuous-state nonlinear branching processes model density-dependent population dynamics via jump-diffusions that can degenerate at zero and have discontinuous jump rates. Classical analytic or coupling tools fail under those features, and the usual branching structure is absent. This paper builds a pathwise framework—state-dependent time change, truncated auxiliary processes, and localized refined coupling—to establish the strong Feller property and irreducibility on the positive reals under mild noise conditions. Those topological properties, plus Lyapunov drift, give a unique quasi-stationary distribution with exponential total-variation convergence when zero is absorbing, and a unique invariant measure with uniform exponential ergodicity when zero is reflecting. The same pathwise ideas apply to a wider class of degenerate jump-diffusions without special coefficient structure.

What carries the argument

The pathwise package of state-dependent time change (normalizing jump intensity), truncated auxiliary processes (retaining compensators while removing large jumps), and localized refined basic coupling (comparing approximate coupling times with local exit times via a local Lyapunov function).

What would settle it

Exhibit continuous coefficients satisfying the local Lipschitz and boundary conditions of the paper for which the SDE either fails to have pathwise uniqueness or produces a process that is not strong Feller (or not irreducible) under condition (C1) or (C2)/(C3).

Watch

Extended reading notes

Core claim

Under either strictly positive interior diffusion or stable-like small-jump activity the transition semigroup of the continuous-state nonlinear branching process is strong Feller on (0,∞); under interior diffusion or infinite small-jump variation it is irreducible on (0,∞). Combined with a super-linear negative drift and local control near zero, these properties produce exponential convergence to a unique quasi-stationary distribution when the origin is absorbing, and uniform exponential ergodicity when the origin is non-absorbing.

Load-bearing premise

The paper assumes that the defining stochastic differential equation always admits a pathwise-unique non-explosive strong solution; every comparison and time-change argument rests on that uniqueness.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

This paper does the technical job that was missing for continuous-state nonlinear branching processes: it proves strong Feller on (0,∞) under either interior diffusion (C1) or stable-like small jumps (C2), and irreducibility under (C1) or infinite small-jump variation (C3). Once those two properties are in hand, the QSD exponential convergence (absorbing case) and uniform exponential ergodicity (non-absorbing case) follow from standard Lyapunov + Meyn–Tweedie / Guillin–Nectoux–Wu arguments under the usual drift and local-boundary conditions.

What is actually new is the pathwise toolkit. Classical Bismut/Krylov/global-coupling methods need non-degeneracy and continuous jump maps; affine/Lamperti arguments need the branching structure. Here the coefficients vanish at 0 and the jump intensity carries a discontinuous indicator. The authors localize the refined basic coupling of Luo–Wang, control exit times with a simple martingale estimate, and for downward reachability they time-change by 1/R2, truncate large jumps while keeping the compensator, and exploit pathwise uniqueness to get coincidence with positive probability. The proofs in Sections 3–4 are self-contained and careful; the local Lyapunov condition (C*) is verified cleanly under both noise regimes.

The soft spot is real but ordinary for the area: pathwise uniqueness and non-explosion of the SDE are imported wholesale from their earlier paper [21]. Every comparison and time-change identity rests on that black box. If uniqueness fails for some of the nonlinear coefficients one cares about, the coincidence arguments collapse. That is not a hidden flaw in the present work; it is simply the standing assumption. Self-citation to [19,20,21] is heavy but functional—they reuse the coupling generator and well-posedness criteria as tools, not as circular evidence for the new theorems.

This is for people who work on continuous-state branching, quasi-stationary distributions, or degenerate jump-diffusions. The method is written so it can be ported. I would send it to peer review without hesitation; the central claims are supported by the written proofs and the contribution is genuine.

Referee Report

0 major / 4 minor

Summary. The paper studies continuous-state nonlinear branching processes given by the jump-diffusion SDE (1.1) with state-dependent rates R0, R1, R2. Classical strong-Feller and irreducibility methods fail because of boundary degeneracy of R1 and R2, the discontinuous indicator in the jump intensity, and the absence of affine/branching structure. The authors develop a pathwise framework: a localized refined basic coupling with a local Lyapunov condition (C*) that yields the strong Feller property on (0,∞) under either interior diffusion (C1) or stable-like small jumps (C2) (Theorem 2.1, via Theorem 3.2 and Proposition 3.4); and a state-dependent time change plus truncated auxiliary processes that establish upward and downward reachability, hence irreducibility under (C1) or infinite small-jump variation (C3) (Theorem 2.2, via Propositions 4.4 and 4.8). Trajectory Feller properties (Theorems 3.6–3.8) are obtained as by-products. Combined with Lyapunov drift (C4) and local boundary control (C5), these yield unique QSD with exponential TV convergence when R0(0)=0 (Theorem 2.4) and unique invariant measure with uniform exponential ergodicity when R0(0)>0 (Theorem 2.5).

Significance. The contribution is a genuine technical advance for a class of degenerate jump-diffusions that arise naturally in population models with density dependence. The pathwise combination of localized refined coupling, state-dependent time change, and truncated auxiliaries bypasses both uniform ellipticity and affine structure, and the abstract correctly notes that the method is not tied to the branching form of the coefficients. The applications recover and extend known QSD and ergodicity results for competition models (Example 2.1) under weaker noise assumptions. The proofs in Sections 3–5 are self-contained once pathwise uniqueness is granted, and the conditions (C1)–(C5) are stated cleanly. This is solid, publishable work in the theory of continuous-state branching processes and degenerate jump SDEs.

minor comments (4)

Section 2.1: the standing pathwise-uniqueness/non-explosion assumption is imported wholesale from [21]. A one-sentence pointer to the precise local-Lipschitz-plus-growth hypotheses of [21] that cover the coefficients used in Example 2.1 would make the paper more self-contained for readers who do not have that reference open.
Lemma 4.5 (time-change identity): the construction of the Poisson measure N via (4.33) and the appeal to Ikeda–Watanabe characterizations are correct but dense. A short remark that the identity holds pathwise up to every τ_ε^-(X) would help readers who only need the law-equivalence for the subsequent reachability arguments.
Proposition 3.4: the choice ρ < (α ∧ 1)/2 for the test function φ under (C2) is stated after the estimate for I4; moving the admissible range of ρ to the beginning of the (C2) case would improve readability.
Throughout: a few typographical inconsistencies appear (e.g., “Lévy” vs. “Lévy”, occasional missing spaces around “(C1)”). A light copy-edit pass would clean these.

Circularity Check

1 steps flagged · score 2.0 of 10

No significant circularity in the strong-Feller/irreducibility derivations; only a standard standing well-posedness assumption imported from overlapping-author prior work [21].

self citation load bearing [Section 2.1 (Standing Assumptions) and Lemma 4.5]
"we shall not discuss the criteria for the existence of a strong solution. Instead, we operate under the standing assumption that for any initial value X0 ≥ 0, the SDE (1.1) admits a pathwise unique and non-explosive strong solution taking values in [0,∞) (see [21] for such well-posedness criteria). ... By the standing pathwise uniqueness assumption for (1.1), this time-changed solution has the same law as the original solution started from x."

Pathwise uniqueness of the nonlinear SDE is imported wholesale from the authors' prior paper [21] and is used as a black-box premise for every subsequent pathwise comparison, time-change identity, and coupling construction. While standard and not a tautology of the strong-Feller/irreducibility claims themselves, the entire derivation chain is conditional on this self-cited uniqueness; if it fails, the coincidence arguments collapse. No independent verification is supplied here.

full rationale

The paper's core claims (Theorems 2.1–2.2 on strong Feller and irreducibility, then QSD/ergodicity applications) are derived pathwise from the explicitly listed analytic conditions (C1)–(C5) plus local Lipschitz/boundary assumptions on the coefficients. The localized refined coupling (generator (3.4)–(3.6), Lyapunov (C*), Theorem 3.2 + Prop. 3.4), state-dependent time change (Lemma 4.5), truncated auxiliaries, and one-sided reachability (Props. 4.4, 4.8) close under those conditions without redefining the target properties in terms of themselves or fitted parameters. The sole self-citation that is load-bearing for the setup is the standing pathwise uniqueness/non-explosion of (1.1), deferred to [21] (same lead author). This is ordinary for the subfield and does not force the new topological conclusions by construction; once uniqueness is granted, the coincidence and coupling arguments are independent. External criteria ([10], [5]) are used for the applications. No self-definitional loops, fitted-as-prediction, ansatz smuggling, or renaming of known results appear.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

The paper is a pure existence/uniqueness/ergodicity theorem for a class of SDEs. It introduces no free numerical parameters and no new physical entities. Load-bearing content consists of standard stochastic-calculus background, domain assumptions on the coefficients (local Lipschitz, boundary degeneracy, monotonicity of R2, Lévy measure integrability), the standing pathwise well-posedness hypothesis imported from prior work, and the noise/Lyapunov conditions (C1)–(C5) that delimit the theorems. The methodological objects (localized refined coupling, truncated auxiliaries, state-dependent time change) are constructions, not postulated entities.

assumptions (7)

domain assumption For every X0≥0 the SDE (1.1) admits a pathwise-unique non-explosive strong solution in [0,∞) (standing assumption, §2.1, deferred to [21]).
All pathwise coincidence and time-change identities rest on uniqueness; the paper does not re-prove it.
domain assumption Coefficients R0,R1,R2 are continuous, locally Lipschitz on compact subintervals of (0,∞), R1(0)=R2(0)=0, R0(0)≥0, and R2 is non-decreasing and strictly positive on (0,∞) (§2.1).
Used throughout generator calculations, exit estimates, and jump-intensity comparisons.
standard math Lévy measure satisfies ∫(z∧z²)π(dz)<∞; Poisson random measure and Brownian motion are independent (§1).
Standard setup for jump-diffusions; needed for Itô formula and compensators.
domain assumption Noise alternatives (C1) interior diffusion positivity, (C2) stable-like small-jump lower bound, or (C3) infinite small-jump variation (§2.2).
These replace global nondegeneracy; each main theorem is conditioned on one of them.
domain assumption Lyapunov drift (C4) superlinear negative growth of R0 and local boundary control (C5) near 0 (§2.3).
Required only for the QSD and uniform ergodicity applications, not for strong Feller/irreducibility.
standard math Refined basic coupling generator of Luo–Wang [23] and the two-dimensional coupling process constructed in [19, Prop. 2.2] (§3.1).
Imported coupling machinery; the paper’s novelty is the localization argument around it.
standard math QSD criterion of Guillin–Nectoux–Wu [10, Thm 2.2] and continuous-time Meyn–Tweedie drift criterion of Down–Meyn–Tweedie [5] (§5).
Black-box ergodicity theorems applied after topological inputs are verified.

how reviews work

0 comments

Cite this review

Pith. "Pith review of A Pathwise Approach to the Strong Feller Property and Irreducibility of Nonlinear Branching Processes." pith.science (2026). https://pith.science/paper/CJNP3LB3

@misc{pith2026260624821,
  author       = {Pith},
  title        = {Pith review of: A Pathwise Approach to the Strong Feller Property and Irreducibility of Nonlinear Branching Processes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CJNP3LB3}},
  note         = {Machine review of arXiv:2606.24821}
}

read the original abstract

We study the strong Feller property and irreducibility for continuous-state nonlinear branching processes defined as solutions to stochastic differential equations with jumps. Due to boundary degeneracy and discontinuous jump coefficients, classical methods do not apply. We develop a pathwise approach combining state-dependent time change, truncated auxiliary processes, and localized coupling to establish these two properties. As applications, we obtain exponential convergence to a unique quasi-stationary distribution in the absorbing case, and uniform exponential ergodicity in the non-absorbing case. This pathwise approach is flexible and can be adapted to a broader class of jump-diffusions without relying on specific coefficient structures.

Review history (2 revisions) →

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

33 extracted references · 1 linked inside Pith

[21]

P.-S. Li, X. Yang, and X. Zhou. A general continuous-state nonlinear branching process.Ann. Appl. Probab., 29:2523–2555, 2019. 39

2019
[1]

Berestycki, M

J. Berestycki, M. C. Fittipaldi, and J. Fontbona. Ray–Knight representation of flows of branching processes with competition by pruning of L´ evy trees.Probab. Theory Related Fields, 172:725–788, 2018

2018
[2]

Chen and Z

S. Chen and Z. Li. Strong Feller and ergodic properties of the (1+1)-affine process. J. Appl. Probab., 60:812–834, 2023

2023
[3]

Da Prato, K

G. Da Prato, K. D. Elworthy, and J. Zabczyk. Strong Feller property for stochastic semilinear equations.Stochastic Anal. Appl., 13(1):35–45, 1995

1995
[4]

D. A. Dawson and Z. Li. Stochastic equations, flows and measure-valued processes. Ann. Probab., 40(2):813–857, 2012

2012
[5]

D. Down, S. P. Meyn, and R. L. Tweedie. Exponential and uniform ergodicity of Markov processes.Ann. Probab., 23(4):1671–1691, 1995

1995
[6]

W. Feller. Diffusion processes in genetics. InProceedings of the Second Berkeley Sym- posium on Mathematical Statistics and Probability, 1950, pages 227–246. University of California Press, Berkeley, 1951. 38

1950
[7]

Friesen, P

M. Friesen, P. Jin, J. Kremer, and B. R¨ udiger. Exponential ergodicity for stochastic equations of nonnegative processes with jumps.ALEA Lat. Am. J. Probab. Math. Stat., 20:593–627, 2023

2023

Show all 33 references

[8]

Friesen, P

M. Friesen, P. Jin, and B. R¨ udiger. Stochastic equation and exponential ergodicity in Wasserstein distances for affine processes.Ann. Appl. Probab., 30(5):2165–2195, 2020

2020
[9]

Fu and Z

Z. Fu and Z. Li. Stochastic equations of non-negative processes with jumps.Stochastic Process. Appl., 120:306–330, 2010

2010
[10]

Guillin, B

A. Guillin, B. Nectoux, and L. Wu. Quasi-stationary distribution for strongly Feller Markov processes by Lyapunov functions and applications to hypoelliptic Hamilto- nian systems.J. Eur. Math. Soc., 26(8):3047–3090, 2024

2024
[11]

Hausenblas, P

E. Hausenblas, P. Razafimandimby, and P. Fernando. Irreducibility and exponen- tial mixing of some stochastic hydrodynamical systems driven by pure jump noise. Commun. Math. Phys., 348(2):535–565, 2016

2016
[12]

Ikeda and S

N. Ikeda and S. Watanabe.Stochastic Differential Equations and Diffusion Processes. North-Holland/Kodansha, 1989

1989
[13]

Karatzas and S

I. Karatzas and S. E. Shreve.Brownian Motion and Stochastic Calculus. Springer, 1991

1991
[14]

Kunwai and C

K. Kunwai and C. Zhu. On Feller and strong Feller properties and irreducibility of regime-switching jump diffusion processes with countable regimes.Nonlinear Anal. Hybrid Syst., 38:100946, 2020

2020
[15]

Kwon and C

Y. Kwon and C. I. Park. Strong Feller property and irreducibility of diffusions with jumps.Stochastics, 67:147–157, 1999

1999
[16]

A. Lambert. The branching process with logistic growth.Ann. Appl. Probab., 15:1506–1535, 2005

2005
[17]

Lamperti

J. Lamperti. The limit of a sequence of branching processes.Z. Wahrscheinlichkeit- stheorie Verw. Gebiete, 7:271–288, 1967

1967
[18]

P.-S. Li. A continuous-state polynomial branching process.Stochastic Process. Appl., 129:2941–2967, 2019

2019
[19]

Li and J

P.-S. Li and J. Wang. Exponential ergodicity for general continuous-state nonlinear branching processes.Electron. J. Probab., 25:Article 125, 1–25, 2020

2020
[20]

P.-S. Li, J. Wang, and X. Zhou. Quasi-stationary distribution for continuous-state branching processes with competition.Stochastic Process. Appl., 177:104457, 2024

2024
[22]

Li and C

Z. Li and C. Ma. Asymptotic properties of estimators in a stable Cox–Ingersoll–Ross model.Stochastic Process. Appl., 125(8):3196–3233, 2015

2015
[23]

Luo and J

D. Luo and J. Wang. Refined basic couplings and Wasserstein-type distances for SDEs with L´ evy noises.Stochastic Process. Appl., 129:3129–3173, 2019

2019
[24]

S. P. Meyn and R. L. Tweedie.Markov Chains and Stochastic Stability. Springer- Verlag, London, 1993

1993
[25]

Pardoux and A

E. Pardoux and A. Wakolbinger. Trees under attack: a Ray–Knight representation of Feller’s branching diffusion with logistic growth.Probab. Theory Related Fields, 155:583–619, 2013

2013
[26]

Pardoux and A

E. Pardoux and A. Wakolbinger. A path-valued Markov process indexed by the ancestral mass.ALEA Lat. Am. J. Probab. Math. Stat., 12:193–212, 2015

2015
[27]

Peszat and J

S. Peszat and J. Zabczyk. Strong Feller property and irreducibility for diffusions on Hilbert spaces.Ann. Probab., 23(1):157–172, 1995

1995
[28]

Priola, A

E. Priola, A. Shirikyan, L. Xu, and J. Zabczyk. Exponential ergodicity and regularity for equations with L´ evy noise.Stochastic Process. Appl., 122(1):106–133, 2012

2012
[29]

H. J. Qiao. Exponential ergodicity for SDEs with jumps and non-Lipschitz coeffi- cients.J. Theoret. Probab., 27:137–152, 2014

2014
[30]

Wang and C

F.-Y. Wang and C. Yuan. Harnack inequalities for functional SDEs with multiplica- tive noise and applications.Stochastic Process. Appl., 121(11):2692–2710, 2011

2011
[31]

J. Wang, H. Yang, J. Zhai, and T. Zhang. Irreducibility of SPDEs driven by pure jump noise, 2022. arXiv preprint arXiv:2207.11488

2022 arXiv
[32]

Xi and C

F. Xi and C. Zhu. Jump type stochastic differential equations with non-Lipschitz coefficients: non-confluence, Feller and strong Feller properties, and exponential er- godicity.J. Differential Equations, 266(8):4668–4711, 2019

2019
[33]

Xie and X

L. Xie and X. Zhang. Ergodicity of stochastic differential equations with jumps and singular coefficients.Ann. Inst. Henri Poincar´ e Probab. Stat., 56(1):175–229, 2020. 40

2020

Pith tools

Reviewed July 13, 2026 · model on record in the stance chip above.

[1] [21]

P.-S. Li, X. Yang, and X. Zhou. A general continuous-state nonlinear branching process.Ann. Appl. Probab., 29:2523–2555, 2019. 39

2019

[2] [1]

Berestycki, M

J. Berestycki, M. C. Fittipaldi, and J. Fontbona. Ray–Knight representation of flows of branching processes with competition by pruning of L´ evy trees.Probab. Theory Related Fields, 172:725–788, 2018

2018

[3] [2]

Chen and Z

S. Chen and Z. Li. Strong Feller and ergodic properties of the (1+1)-affine process. J. Appl. Probab., 60:812–834, 2023

2023

[4] [3]

Da Prato, K

G. Da Prato, K. D. Elworthy, and J. Zabczyk. Strong Feller property for stochastic semilinear equations.Stochastic Anal. Appl., 13(1):35–45, 1995

1995

[5] [4]

D. A. Dawson and Z. Li. Stochastic equations, flows and measure-valued processes. Ann. Probab., 40(2):813–857, 2012

2012

[6] [5]

D. Down, S. P. Meyn, and R. L. Tweedie. Exponential and uniform ergodicity of Markov processes.Ann. Probab., 23(4):1671–1691, 1995

1995

[7] [6]

W. Feller. Diffusion processes in genetics. InProceedings of the Second Berkeley Sym- posium on Mathematical Statistics and Probability, 1950, pages 227–246. University of California Press, Berkeley, 1951. 38

1950

[8] [7]

Friesen, P

M. Friesen, P. Jin, J. Kremer, and B. R¨ udiger. Exponential ergodicity for stochastic equations of nonnegative processes with jumps.ALEA Lat. Am. J. Probab. Math. Stat., 20:593–627, 2023

2023