Mode-Weighted Transport Certificates for State-Dependent Reflected Switching Diffusions

Ye Zhang; Yutong Zhu

REVIEW 2 major objections 5 minor 18 references

Mode-Weighted Transport Certificates for State-Dependent Reflected Switching Diffusions

T0 review · 2 major / 5 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read This paper establishes a computable transport-based certificate proving global pairwise exponential contraction for state-dependent switching diffusions, even when some modes are individually expansive, with the estimate extending to weak s

desk verdict Solid theorem paper: the mode-weighted cost with a spatial term for cross-mode pairs is a genuine extension of Cloez–Hairer, and the coupling proof is explicit and mostly clean; the main thing to check is the unproved generator-core assumption used in the FPK transfer. read the letter →

arxiv 2608.01992 v1 pith:O4CELCTJ submitted 2026-08-03 cs.CE cs.SYeess.SY

Yutong Zhu , Ye Zhang This is my paper

classification cs.CEcs.SYeess.SY MSC 60J6049Q2260J2790C34

keywords state-dependentswitchingdiffusionsWassersteincontractiontransportdiscrepancynormalreflectionno-fluxFokker-Planck-Kolmogorovsemi-infinitelinearprogrammingpairwiseexponentialergodicitysynchronouscoupling

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

This paper develops a computable condition that proves a state-dependent switching diffusion is globally exponentially contractive in distribution, even when some of its modes are individually expansive. The condition is a pair of generator inequalities for a specially designed transport cost that couples a mode-dependent spatial metric with a discrete penalty for mode mismatch. If the inequalities hold for some mode weights v and graph costs B and rate η, then every pair of initial laws satisfies W_{v,B}(µP_t, ˜µP_t) ≤ e^{-ηt}W_{v,B}(µ, ˜µ), with prefactor one and an explicit rate. The same estimate automatically transfers to weak measure solutions of the no-flux Fokker-Planck-Kolmogorov system on compact convex domains with normal reflection. For a fixed ordering of the mode weights and a prescribed η, the conditions are affine in the unknowns and form a semi-infinite linear program; a finite mesh with a Lipschitz margin turns a numerical check into a proof over the whole domain.

What carries the argument

The mode-weighted transport discrepancy W_{v,B}: a Wasserstein-type cost on the hybrid state space X=Ω×M combining mode-dependent spatial weights v_i with a symmetric discrete mode penalty β_ij, with closed form a_ij||x-y||+β_ij. It is not a metric (triangle inequality may fail) but is topologically equivalent to a product Wasserstein distance. The argument's work: same-mode inequalities combine one-sided drift contraction, simultaneous jumps, and unmatched-clock penalties from rate sensitivity; cross-mode inequalities capture drift mismatch and every jump of either component. The coupling is a synchronous Brownian coupling with maximal coupling of equal-target clocks. Convex normal reflecti

What would settle it

For the planar three-mode example, compute the cross-mode residuals with an exact or arbitrarily fine global optimization; if any residual becomes positive after the Lipschitz buffer, the Proposition 3 bound is wrong. Alternatively, for the one-dimensional reflected example, solve the no-flux FPK system with high precision and compare the mode-weighted discrepancy to the certified exponential bound; exceeding it would falsify the theorem.

Watch

Extended reading notes

Core claim

The central discovery is that a mode-weighted transport discrepancy, defined as c_{v,B}((x,i),(y,j)) = inf_q [v_i||x-q|| + β_ij + v_j||q-y||] = a_ij||x-y||+β_ij with a_ij = min{v_i,v_j}, keeps a continuous spatial term alive for cross-mode pairs, unlike hybrid distances that only use spatial separation when modes agree. This closes the cross-mode generator inequality. Under synchronous Brownian coupling and maximal coupling of state-dependent clocks, the generator applied to the cost is bounded by constants S_i and C_ij; if each is ≤ -η times the cost, the semigroup is pairwise exponentially contractive with rate η and unit prefactor. Normal reflection on convex domains contributes a nonposi

Load-bearing premise

The transfer from the semigroup estimate to the no-flux FPK system rests on Assumption 1, that the Neumann test-function class is a core for the generator; if that core property fails, the PDE statement does not follow.

Editorial extensions

If this is right

If the certificate conditions (22)-(23) are feasible, the semigroup is pairwise exponentially contractive in W_{v,B} with unit prefactor for every pair of initial laws.
The contraction transfers to weak measure solutions of the no-flux FPK system via a resolvent/core argument.
On compact X there is a unique invariant law µ⋆, and W_{v,B}(µ_0P_t, µ⋆) ≤ e^{-ηt}W_{v,B}(µ_0, µ⋆); on R^n the same holds when an invariant law exists.
For fixed weight order and η the certificate is a semi-infinite linear feasibility problem; a δ-net check with Lipschitz constant H_ij verifies all continuum constraints.
In the state-independent switching limit, the same-mode condition reduces to the Metzler-matrix inequality (Qv - diag(c_i)v)_i ≤ -ηv_i.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The same certificate structure may apply to other boundary conditions if the reflection/Lyapunov term can be shown nonpositive; here the crucial geometric input is convexity of the domain.
Because the cost is not a metric but still controls the product Wasserstein topology, the exponential bound implies quantitative convergence of both first moments and mode-mismatch probabilities; a sharper reverse inequality could give explicit constants.
The explicit Lipschitz buffer suggests a general template for making semi-infinite LP certificates rigorous over continuous domains, beyond this model.
The feasibility framework could be inverted to design transition rates meeting a prescribed contraction rate on large mode sets, using branch-and-bound rather than order enumeration.

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

Quick take: this is a real result, not a repackaging. The mode-weighted cost that keeps a spatial term even when the two copies occupy different modes is the key idea, and it does what the Cloez–Hairer hybrid distance cannot: it closes the cross-mode generator inequality. The coupling construction—synchronous Brownian motion plus maximal coupling of the state-dependent clocks—is explicit and correct, and the reflection sign argument is handled properly. I checked the generator calculations in Steps 3 and 4; they are clean, and the cost closed form in Lemma 1 is correct. The numerical section is honest: they distinguish the LP bisection boundary estimate from the strict certificate with the δ-margin, and Proposition 3 is a genuine finite-mesh proof, not a heuristic grid check.

The soft spots are real but not fatal. The main structural assumption is Assumption 1, that the Neumann test-function class is a core for the reflected generator. It is standard for reflected diffusions on C^2 convex domains, but it is asserted without proof or reference, and it is load-bearing only for the FPK half of Theorem 1—the semigroup contraction does not use it. A referee should ask for a reference or a one-sentence argument; I would not block acceptance on it if the authors supply that. The epsilon-regularization step for the upper Dini derivative at coincidence is compressed; it is believable and standard for convex reflected diffusions, but it should be written out for a journal. The cost is not a metric and the authors say so; that is fine for contraction but limits interpretation of the discrepancy. The certificate is only sufficient—they show an infeasible case where their chosen weights cannot certify a rate even though their one simulation decays. That is expected for a Lyapunov-type condition, and they do not overclaim.

The citation pattern looks fine. The novelty over Cloez–Hairer and Shao is clear, and the paper positions itself honestly against the existing Wasserstein results. No circularity: the certificate inequalities are sufficient conditions derived by proof, not fitted predictions.

This deserves a serious referee. It is a theorem-first paper, the proof is mostly self-contained, and the construction should be useful for proving stability and mixing of state-dependent switching systems. I would send it to review, with the core assumption and the epsilon argument as the specific requests for revision.

Referee Report

2 major / 5 minor

Summary. The paper develops a sufficient condition for global pairwise exponential contraction of state-dependent reflected switching diffusions in a mode-weighted transport discrepancy. The main result, Theorem 1, states that if the same-mode inequalities (22) and cross-mode inequalities (23) hold for some positive mode weights v, a symmetric graph cost B, and a rate η>0, then W_{v,B}(μ_0 P_t, \tilde μ_0 P_t) ≤ e^{-ηt} W_{v,B}(μ_0,\tilde μ_0) for all initial laws. The proof uses a synchronous Brownian coupling and a maximal coupling of the jump clocks, together with a generator calculation that handles the reflection terms via convexity. A resolvent argument (Proposition 1) transfers the estimate to weak measure solutions of the no-flux FPK system. The certificate is then cast as a semi-infinite linear feasibility problem, with a finite-mesh buffer (Proposition 3) converting a grid check into a full-domain proof. Two numerical examples, including a planar three-mode synthesis, demonstrate the procedure.

Significance. If the result stands, it provides a genuinely computable and provable contraction certificate for a class of hybrid diffusion models that can be contractive even when individual modes are expansive. The coupling construction is explicit, the conditions are affine after fixing a weight order and decay rate, and the finite-mesh certification gives a rigorous route from numerical optimization to a full-domain proof. The paper does not ship code, but the numerical experiments are described with enough detail to be reproducible, and the planar example shows the constraint-generation framework working end-to-end. The main strength is the combination of a transparent proof with a practical synthesis algorithm.

major comments (2)

[II-B, Proposition 1] Assumption 1 asserts, without proof or reference, that the Neumann test-function class D_N is a core for the generator of the Feller semigroup on C_0(X). This core property is load-bearing: the proof of Proposition 1 uses it to conclude that (αI−A)D_N is dense in C_0(X), which is the only step that transfers the semigroup contraction to weak solutions of the no-flux FPK system. Please provide a proof or a precise citation establishing that C^2 functions with zero normal derivative form a core for the reflected switching generator under the stated Lipschitz drift and bounded Lipschitz jump rates on a C^2 convex domain. If the statement requires additional hypotheses (e.g., boundary regularity beyond C^2, or a separate martingale-problem argument), these conditions should be added explicitly.
[IV-C, Step 5] The passage from the generator inequality (40) to the coupling inequality (42) uses an ε-regularization of the distance at coincidence, and the text says this 'gives precisely the upper Dini derivative stipulated in Assumption 4.' The argument is only sketched. Since the distance is not differentiable at R_t=0 and the reflection terms have singular coefficients there, the derivation of the cross-mode bound at x=y is not fully explicit. Please expand this step, or alternatively replace it with a comparison argument based on the Lipschitz property of the Skorokhod map, so that the reader can verify that the regularized cost indeed satisfies the generator inequality uniformly in ε.

minor comments (5)

[IV-A, Eqs. (20) and (32)] The same-mode margin formula and the displayed bound for simultaneous jumps appear to have lost the over/under bars on the rate envelopes. The correct assignment should be (v_k−v_i)^+ times the upper envelope \overline{λ}_{ik} and (v_k−v_i)^− times the lower envelope \underline{λ}_{ik}, as the surrounding text indicates. Please ensure the notation is unambiguous in the final version.
[VI-A, Eq. (59)] The displayed cross-mode bound in the one-dimensional example is missing operators or parentheses; as printed it is not a valid inequality. Please re-typeset it, for example as q h r − b(γ+δ) + δ(1−q) r + η(qr+b) ≤ 0.
[II-B, Proposition 1 proof] Typo: 'stopped rocess' should be 'stopped process'.
[VI-C, Table II] The column headers 'H_ij δ max_S F_ij certified residual' are ambiguous. Clarify which column contains the product H_ij δ, which contains the mesh maximum, and which contains the certified residual.
[III, Eq. (13)] The pointwise bounds leading to (14) are correct, but the notation m and \bar m is not defined before use; please insert definitions before the display.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the contraction theorem is proved from certificate inequalities by an explicit coupling/generator calculation; numerical synthesis is constraint feasibility, not fitted prediction.

full rationale

The paper's central claim, Theorem 1, is a direct mathematical implication: if certificate inequalities (22)-(23) hold for some (v,B,eta), then the mode-weighted transport discrepancy contracts exponentially. The proof is an explicit construction of a Markov coupling, a generator calculation on the cost function, and a Dynkin/localization argument. The certificate quantities v, B, eta are free decision variables, not parameters fitted to target data; the inequalities are sufficient conditions established by proof, not empirical observations. The numerical sections solve linear feasibility problems to find certificates and then verify the full-domain inequalities with a Lipschitz mesh buffer (Proposition 3); this is constraint synthesis, not prediction from fitted inputs. No prediction is derived from a fitted subset of the same data. The only potentially unproved structural premise is Assumption 1, which asserts that the Neumann test-function class is a core for the generator. This is an explicit regularity assumption used to transfer the semigroup estimate to weak FPK solutions via Proposition 1; it is not an input-output equivalence and does not make the derivation circular. It may be a correctness/completeness concern, but not a circularity concern. The paper cites standard external references for well-posedness, reflection, and resolvent facts; no load-bearing self-citation chain appears. Overall, the derivation is self-contained and non-circular.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No model constants are fitted to data; all envelopes are computed from the drift and rate data. The mode weights, graph costs, and decay rate are optimization variables in the sufficient certificate, not empirical fits. Background axioms are standard reflected SDE theory, Feller semigroup and resolvent theory, and a generator-core assumption stated in Assumption 1. No new physical entities are postulated; the mode-weighted cost is a mathematical construction rather than an empirical entity.

assumptions (5)

standard math Well-posedness of normally reflected SDEs on convex C^2 domains
Invoked in Section II.A via [16],[17] to ensure the reflected process exists and is strong Markov.
domain assumption The Neumann test-function class D_N is a core for the generator of the Feller semigroup on C_0(X)
Stated as part of Assumption 1; used in Proposition 1 to identify weak FPK solutions with the semigroup evolution.
standard math For a core D_N, (alpha*I - A)D_N is dense in C_0(X) for alpha > 0
Used in the resolvent argument of Proposition 1; consequence of closed generator and core theory [18].
standard math The martingale problem for the reflected switching diffusion is well posed and the semigroup is Feller
Cited via [1] and [18] in Proposition 1; needed for uniqueness of semigroup and weak solutions.
standard math Interlacing of bounded state-dependent Poisson clocks produces a measurable Markovian coupling
Used in Step 1 of Theorem 1; valid because rates are bounded by Assumption 3.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Mode-Weighted Transport Certificates for State-Dependent Reflected Switching Diffusions." pith.science (2026). https://pith.science/paper/O4CELCTJ

@misc{pith2026260801992,
  author       = {Pith},
  title        = {Pith review of: Mode-Weighted Transport Certificates for State-Dependent Reflected Switching Diffusions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O4CELCTJ}},
  note         = {Machine review of arXiv:2608.01992}
}

abstract

State-dependent switching diffusions can be contractive in distribution even when some modes are individually expansive. We develop a computable transport-based condition for such contraction on $\mathbb{R}^n$ and on compact convex domains with normal reflection. The transport cost combines mode-dependent spatial weights with a discrete mode penalty while preserving spatial separation for cross-mode pairs. Using synchronous coupling of the Brownian motions and maximal coupling of the state-dependent jump clocks, we derive separate generator inequalities for same-mode and cross-mode configurations. Convex normal reflection contributes a nonpositive finite-variation term, so the same conditions apply to the associated no-flux Fokker-Planck-Kolmogorov system. Their feasibility guarantees global pairwise exponential contraction of the Markov semigroup and weak measure solutions, with unit prefactor and an explicit rate. For a fixed ordering of the mode weights and a prescribed decay rate, the conditions are affine in the spatial weights and graph costs and form a semi-infinite linear feasibility problem. A finite-mesh condition with a Lipschitz margin certifies the inequalities over the full domain. A reflected one-dimensional example validates the distributional computation, and a planar three-mode example demonstrates the synthesis procedure for transition rates depending on both state coordinates.

Figures

Figures reproduced from arXiv: 2608.01992 by the authors.

**Figure 2.** Evolution of the mode-weighted transport discrepancy. The straight [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗

**Figure 3.** No-flux FPK evolution on [−1, 1] for γ = 2.0. C. Planar three-mode certificate synthesis We next apply Algorithm 1 to a planar three-mode system. Let Ω = {x ∈ R 2 : ∥x∥ ≤ 0.5} and fi(x) = Aix, where A1 = 0.45 −0.25 0.25 0.30 , A2 = −1.80 0.30 −0.30 −1.10 , A3 = −0.90 −0.35 0.35 −1.70 . (62) Mode 1 is expansive and the other two modes are contractive, with (c1, c2, c3) = (−0.45, 1.10, 0.90). All six off-dia… view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Planar three-mode synthesis. Panels (a)-(b) show two of the state-dependent jump rates in (63) and the remaining four have the same affine two [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

18 extracted references

[1]

Certain properties related to well posedness of switching diffusions,

D. H. Nguyen, G. Yin, and C. Zhu, “Certain properties related to well posedness of switching diffusions,”Stochastic Processes and their Applications, vol. 127, no. 10, pp. 3135–3158, 2017

2017
[2]

Nguyen, G

H.-D. Nguyen, G. Yin, and C. Zhu,Hybrid Switching Diffusions: Properties and Applications, 2nd ed. Cham: Springer, 2025

2025
[3]

Stability of degenerate diffusions with state-dependent switching,

G. K. Basak, A. Bisi, and M. K. Ghosh, “Stability of degenerate diffusions with state-dependent switching,”Journal of Mathematical Analysis and Applications, vol. 240, no. 1, pp. 219–248, 1999

1999
[4]

Stability of stochastic nonlinear systems with state-dependent switching,

Z. Wu, M. Cui, P. Shi, and H. R. Karimi, “Stability of stochastic nonlinear systems with state-dependent switching,”IEEE Transactions on Automatic Control, vol. 58, no. 8, pp. 1904–1918, 2013

1904
[5]

Stability and recurrence of regime-switching diffusion processes,

J. Shao and F. Xi, “Stability and recurrence of regime-switching diffusion processes,”SIAM Journal on Control and Optimization, vol. 52, no. 6, pp. 3496–3516, 2014

2014
[6]

Villani,Optimal transport: old and new

C. Villani,Optimal transport: old and new. Springer, 2009, vol. 338

2009
[7]

Optimal transport in systems and control,

Y . Chen, T. T. Georgiou, and M. Pavon, “Optimal transport in systems and control,”Annual Review of Control, Robotics, and Autonomous Systems, vol. 4, no. 1, pp. 89–113, 2021

2021
[8]

Exponential ergodicity for Markov processes with random switching,

B. Cloez and M. Hairer, “Exponential ergodicity for Markov processes with random switching,”Bernoulli, vol. 21, no. 1, pp. 505–536, 2015

2015

Show all 18 references

[9]

Ergodicity of regime-switching diffusions in Wasserstein distances,

J. Shao, “Ergodicity of regime-switching diffusions in Wasserstein distances,”Stochastic Processes and their Applications, vol. 125, no. 2, pp. 739–758, 2015

2015
[10]

Moment bounds and geometric ergodicity of diffusions with random switching and unbounded transition rates,

X. T. Tong and A. J. Majda, “Moment bounds and geometric ergodicity of diffusions with random switching and unbounded transition rates,” Research in the Mathematical Sciences, vol. 3, no. 1, p. 41, 2016

2016
[11]

Invariant measures and Euler–Maruyama’s approximations of state-dependent regime-switching diffusions,

J. Shao, “Invariant measures and Euler–Maruyama’s approximations of state-dependent regime-switching diffusions,”SIAM Journal on Control and Optimization, vol. 56, no. 5, pp. 3215–3238, 2018

2018
[12]

Comparison theorem and stability under perturbation of transi- tion rate matrices for regime-switching processes,

——, “Comparison theorem and stability under perturbation of transi- tion rate matrices for regime-switching processes,”Journal of Applied Probability, vol. 61, no. 2, pp. 540–557, 2024

2024
[13]

Ergodicity and stability of hybrid systems with piecewise constant type state-dependent switching,

J. Shao, L. Wang, and Q. Wu, “Ergodicity and stability of hybrid systems with piecewise constant type state-dependent switching,” Stochastic Processes and their Applications, vol. 161, pp. 1–23, 2023

2023
[14]

On subgeometric ergodicity of regime- switching diffusion processes,

P. Lazi ´c and N. Sandri ´c, “On subgeometric ergodicity of regime- switching diffusion processes,”Nonlinear Analysis: Hybrid Systems, vol. 46, p. 101262, 2022

2022
[15]

Ergodicity of regime-switching func- tional diffusions with infinite delay and application to a numerical algorithm for stochastic optimization,

B. Shi, Y . Wang, and F. Wu, “Ergodicity of regime-switching func- tional diffusions with infinite delay and application to a numerical algorithm for stochastic optimization,”SIAM Journal on Control and Optimization, vol. 60, no. 5, pp. 2658–2683, 2022

2022
[16]

Stochastic differential equations with reflecting boundary condition in convex regions,

H. Tanaka, “Stochastic differential equations with reflecting boundary condition in convex regions,”Hiroshima Mathematical Journal, vol. 9, pp. 163–177, 1979

1979
[17]

Stochastic differential equations with reflecting boundary conditions,

P.-L. Lions and A.-S. Sznitman, “Stochastic differential equations with reflecting boundary conditions,”Communications on Pure and Applied Mathematics, vol. 37, no. 4, pp. 511–537, 1984

1984
[18]

S. N. Ethier and T. G. Kurtz,Markov Processes: Characterization and Convergence. New York: Wiley, 1986

1986

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.

[1] [1]

Certain properties related to well posedness of switching diffusions,

D. H. Nguyen, G. Yin, and C. Zhu, “Certain properties related to well posedness of switching diffusions,”Stochastic Processes and their Applications, vol. 127, no. 10, pp. 3135–3158, 2017

2017

[2] [2]

Nguyen, G

H.-D. Nguyen, G. Yin, and C. Zhu,Hybrid Switching Diffusions: Properties and Applications, 2nd ed. Cham: Springer, 2025

2025

[3] [3]

Stability of degenerate diffusions with state-dependent switching,

G. K. Basak, A. Bisi, and M. K. Ghosh, “Stability of degenerate diffusions with state-dependent switching,”Journal of Mathematical Analysis and Applications, vol. 240, no. 1, pp. 219–248, 1999

1999

[4] [4]

Stability of stochastic nonlinear systems with state-dependent switching,

Z. Wu, M. Cui, P. Shi, and H. R. Karimi, “Stability of stochastic nonlinear systems with state-dependent switching,”IEEE Transactions on Automatic Control, vol. 58, no. 8, pp. 1904–1918, 2013

1904

[5] [5]

Stability and recurrence of regime-switching diffusion processes,

J. Shao and F. Xi, “Stability and recurrence of regime-switching diffusion processes,”SIAM Journal on Control and Optimization, vol. 52, no. 6, pp. 3496–3516, 2014

2014

[6] [6]

Villani,Optimal transport: old and new

C. Villani,Optimal transport: old and new. Springer, 2009, vol. 338

2009

[7] [7]

Optimal transport in systems and control,

Y . Chen, T. T. Georgiou, and M. Pavon, “Optimal transport in systems and control,”Annual Review of Control, Robotics, and Autonomous Systems, vol. 4, no. 1, pp. 89–113, 2021

2021

[8] [8]

Exponential ergodicity for Markov processes with random switching,

B. Cloez and M. Hairer, “Exponential ergodicity for Markov processes with random switching,”Bernoulli, vol. 21, no. 1, pp. 505–536, 2015

2015