Stochastic scattering control of spider diffusion governed by an optimal diffraction probability measure selected from its own local-time

Isaac Ohavi

REVIEW 4 major objections 5 minor 47 references

Stochastic scattering control of spider diffusion governed by an optimal diffraction probability measure selected from its own local-time

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read The value function of a spider diffusion controlled at its junction is the unique viscosity solution of a Hamilton–Jacobi–Bellman system with a nonlinear local-time Kirchhoff boundary condition.

desk verdict New control problem, honest about its debts, but the vertex argument rests on an unproved local-time estimate from an unpublished preprint. read the letter →

arxiv 2501.18057 v1 pith:QEHJNZWR submitted 2025-01-29 math.AP math.OC

Isaac Ohavi This is my paper · ORCID

classification math.APmath.OC MSC 34B4535R0249L2049L2560G4678A4593E20

keywords stochasticscatteringcontrolWalsh'sspiderdiffusionnonlinearlocal-timeKirchhoffboundaryconditionHamilton-Jacobi-Bellmansystemviscositysolutionscomparisonprincipledynamicprogramminglocaltime

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 establishes a complete well-posedness theory for a new stochastic control problem: steering a Walsh spider diffusion by choosing, at each visit to the junction, the probability measure with which the process is scattered onto the branches. The value function of this scattering control problem is shown to be characterized as the unique continuous viscosity solution of a Hamilton–Jacobi–Bellman (HJB) system posed on a star-shaped network, with a new boundary condition at the junction called the nonlinear local-time Kirchhoff transmission. The boundary condition arises from the controller's ability to choose the diffraction measure as a function of the process's own local time at the vertex. A comparison theorem, a weak dynamic programming principle, and compactness of admissible rules are proved, yielding the unique characterization. If correct, the result turns an apparently singular control problem at a network junction into a tractable PDE problem.

What carries the argument

The machinery is the Walsh spider HJB system with the nonlinear local-time Kirchhoff transmission at the vertex: alongside the ray-wise HJB equations, the value u must satisfy ∂_l u(t,0,l) + sup_ϑ { Σ_i S_i(t,l,ϑ) ∂_x u_i(t,0,l) + h_0(t,l,ϑ) } = 0, where l is the accumulated local time and (S_i) is a controlled probability vector on the branches. The proof of well-posedness rests on three supports: the weak dynamic programming principle in the spirit of [32], which needs a compactness theorem for admissible rules at the junction; a comparison theorem adapted from [40] that handles the nonlinear boundary transmission; and the non-stickiness estimate (24) from [38]—expected local time increase ≈ h before the radial coordinate first exits [0,h]—which is what converts the dynamic programming principle into the boundary equation in the limit h → 0.

What would settle it

Simulate the spider with a deliberately sticky spinning measure at the vertex (or any measure violating the non-stickiness assumption) and measure, for small h, the ratio (1/h)E[l(τ_h)-l] where τ_h is the first time the radial part reaches h; if the limit is not 1, or if E[τ_h-t] does not decay like h², then the boundary condition in Theorem 2.7 is not the correct characterization for that process.

Watch

Extended reading notes

Core claim

The central discovery is that the value function defined through the martingale formulation of the controlled spider is the unique continuous viscosity solution of the Walsh spider backward HJB system. On each ray, the usual HJB equation holds; at the vertex, a new transmission condition appears: ∂_l u(t,0,l) + sup_ϑ { Σ_i S_i(t,l,ϑ) ∂_x u_i(t,0,l) + h_0(t,l,ϑ) } = 0, where l is the local time accumulated at the junction and S(t,l,ϑ) is the controlled diffraction probability. The proof passes to the limit at the vertex using the estimate that local time grows linearly before the radial process first exits a small interval, and it relies on a comparison theorem for this class of HJB systems. The paper also proves the weak dynamic programming principle under junction controls and the compactness of the admissible rules, so that the supremum defining the value function is attained.

Load-bearing premise

The whole vertex argument rests on the estimate (24) from [38]—that before the radial coordinate first exits [0,h] the expected local-time increase is approximately h and the expected time is at most a constant times h²—so if local time grows sublinearly or superlinearly in that window, the derived Kirchhoff boundary condition fails.

Editorial extensions

If this is right

The optimal diffraction probabilities at the junction are determined by the unique solution of a PDE system; solving the HJB system once gives the value function everywhere without simulating the controlled process.
The comparison theorem ensures that any approximation scheme whose limits are viscosity solutions of the HJB system converges to the true value function.
Because the value function is continuous and admissible rules are compact, optimal controls exist in the space of relaxed controls; the paper identifies that optimal feedback controls at the vertex can be built from the jets of u, a point deferred to future work.
For problems whose data do not depend on local time, the same arguments yield a simpler well-posedness result for a nonlinear Kirchhoff transmission without the local-time derivative, as discussed in Section 6.
The characterization supplies a candidate verification theorem: to certify a control is optimal it suffices to check that the candidate value function is a viscosity solution of the HJB system with the local-time Kirchhoff condition.

Reading between the lines

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

If the local-time growth estimate (24) can be verified for sticky or degenerate spinning measures, the same HJB characterization may extend to processes that spend positive time at the vertex; the paper explicitly leaves the degenerate case to future work.
The physics analogy drawn in the introduction suggests a concrete experimental test: a particle moving on a star-shaped photonic network whose ray affinities are updated by the local time at the junction should exhibit diffraction probabilities close to the ones maximizing the reward functional; such a prediction is not derived in the paper.
The comparison theorem opens a path to numerical solution of the problem via monotone schemes, though the paper does not report any computational experiment.
Because the HJB system is posed with l as an external variable, the same technique may apply to other reflected or boundary-controlled diffusions where the control acts only through the local time.

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

Short version: the problem is new and worth a referee's time, but the paper is not self-contained at its key vertex step.

The genuinely new thing here is the formulation: a stochastic control problem where the controller picks the diffraction probabilities at the spider's junction as a function of the current local time, with running and terminal rewards. That is not in the prior spider literature, which is about stopping or MFG with fixed spinning measures. The authors are also straight about the architecture: the weak DPP is from El Karoui–Nguyen–Jeanblanc, the comparison theorem is the author's own [40] adapted to the parabolic case, and the process estimates come from [37]. That is legitimate, but it means the reader is signing up for a chain of external results.

The central theorem, Thm 2.7, says the value function is the unique viscosity solution of the HJB system with the nonlinear local-time Kirchhoff transmission. The interior part is standard. The vertex part is where the real action is, and it depends on estimate (24): E[τ_h − t*] ≤ C h^2 and h^{-1}E[l(τ_h) − l*] → 1. That estimate is quoted from [38], which is marked in preparation. It is not proved here, and it is not a cosmetic gap: dividing by h and passing to the limit is exactly how the ∂_l u term appears in the Kirchhoff condition. If the local-time increment scaled differently, the boundary condition would come out with the wrong constant or not at all. So the characterization theorem hangs on an unpublished bootstrap.

Some smaller soft spots: Propositions 3.4–3.5 and the stability-by-conditioning/concatenation results are stated without proof. Those are less troubling because they are routine adaptations and the paper says so. The comparison proof in Section 4 is a sketch, but [40] is accepted and the adaptation is plausible.

Overall, this is a solid preprint with a real new formulation and an honest map of its dependencies. It is not yet a standalone proof of Theorem 2.7. The fix is also clear: either include a proof of (24) or wait for [38] to appear and be verified.

Who is this for? People working on stochastic control on networks and spider diffusions. It deserves a serious referee; a good referee could check whether (24) follows from the martingale problem in [37] or needs genuinely new work.

Recommendation: send it to peer review, with the expectation that the (24) gap gets closed or the paper comes with [38].

Referee Report

4 major / 5 minor

Summary. The paper formulates a finite-horizon stochastic optimal scattering control problem for a Walsh spider diffusion whose spinning measure at the junction depends on the process's own local time and on time. The value function is defined as a supremum over suitably defined weak admissible rules, and the main results are: a weak dynamic programming principle (Theorem 2.4), a comparison theorem for continuous viscosity solutions of the associated backward HJB system with nonlinear local-time Kirchhoff transmission (Theorem 2.5), and a unique viscosity characterization of the value function by that HJB system (Theorem 2.7). The proofs rely heavily on prior work by the same group: the spider martingale problem in [37], the companion local-time analysis in [38], the linear parabolic theory in [39], and the comparison principle in [40]. The central novelty is the Kirchhoff-type boundary condition at the vertex involving the local-time derivative and an optimized convex combination of spatial derivatives.

Significance. If the main theorem is established, this appears to be the first stochastic control characterization for spider diffusions in which the diffraction probability itself is controlled and depends on the local time. The paper also provides a compactness result for admissible rules and a weak DPP in the presence of vertex controls, which are nontrivial extensions of [32]. The comparison theorem would extend [40] to the parabolic backward setting. The work is transparent in its reliance on prior results and does not engage in data fitting or circularity by construction. The significance is, however, contingent on closing a central analytic gap: the local-time scaling estimate (24) is quoted from an unpublished source and is the key ingredient that recovers the nonlinear local-time Kirchhoff transmission in Theorem 2.7.

major comments (4)

[Section 5 (proof of Theorem 2.7), Eq. (24)] The vertex part of Theorem 2.7 is load-bearing on the estimate E^{Q}[(l(τ_h)-l*)/h] → 1 and E^{Q}[τ_h - t*] ≤ C h^2, quoted from [38, Proposition 8.1], which is listed as "in preparation" in the references. The present paper gives no proof of this estimate. The passage h → 0 in inequalities (25) and (28) divides by h and uses both the time bound and the local-time asymptotic; if the true asymptotic were C h with C ≠ 1, the recovered vertex condition would have a different coefficient, and if the growth were not linear, no Kirchhoff-type transmission would be obtained. Because Theorem 2.7 claims exactly the unique characterization by system (6), this gap is central and must be resolved either by proving the estimate in this paper or by supplying a fully citable proof of the quoted result.
[Section 4 (proof of Theorem 2.5)] The comparison theorem is only sketched. After reducing to the exponentially weighted system (8), the interior case is dismissed with a reference to standard viscosity arguments, and the vertex case is said to follow the proof of Theorem 2.2 of [40] with the statement that the time variable "will have no impact." The construction of the ODE test functions φ and φ in (16)-(17), the uniform bounds on their local-time derivatives, and the limiting passages (20)-(21) are not proved here, despite being essential to the contradiction argument. Since Theorem 2.7 uses Theorem 2.5 for uniqueness, the comparison theorem should either be stated as a black-box theorem imported verbatim from a published or accepted source with all hypotheses verified, or its proof should be completed in this paper.
[Section 3.2 and 3.3 (Propositions 3.4, 3.5, 3.7, 3.8, 3.9)] Several results that underlie the compactness of admissible rules and the dynamic programming principle are stated without proof. Proposition 3.4 (path estimates) and Proposition 3.5 (non-stickiness) are quoted as adaptations from [37], while Propositions 3.7-3.9 (stability by conditioning and concatenation) are stated as direct consequences of [44]. The compactness theorem 3.6 and the proof of Theorem 2.4 depend on these properties. If these statements do not hold verbatim in the controlled setting with vertex-dependent spinning measures and local-time-dependent coefficients, the DPP is not established. Please either prove them or give precise references with theorem numbers and verify that the assumptions of the current paper match those references.
[Section 5, Eqs. (25) and (28)] In the displayed inequalities used to pass to the limit at the vertex, the terms involving the spinning measure are written as S_i(u, l(u), ϑ) without the factor ∂_x φ_i, e.g. "Σ_i S_i(u, l(u), ϑ) + h_0" in (25) and a similar expression in (28). The correct vertex Hamiltonian in (6) contains Σ_i S_i(t,l,ϑ) ∂_x φ_i(t,0,l). This appears to be a typographical omission, but it occurs exactly in the limiting argument that proves the Kirchhoff transmission, so the displayed computation should be corrected and the limit rechecked with the factor present.

minor comments (5)

[Section 2.1, Assumption (H), (R-iii)] In the last Lipschitz modulus of (R-iii), the expression uses |b_i(t,x,l,β_i) - b_i(t,x,l',β_i)|/|l-l'|, but the displayed bound is attributed to |h|; it should presumably involve h_i. Please correct the typo.
[Section 3.2, Proposition 3.4] The notation ℓ^{ν_0(·)}(·)^2 is confusing; it should be made clear that the square applies to the local-time increment process over the interval [t,·].
[Section 5, proof of Theorem 2.7] The phrase "we will obtain as soon as h ց 0" in the subsolution case is grammatically incomplete; it should read "as h → 0, we obtain."
[References] Reference [38] is marked "In preparation" and reference [40] is marked accepted but without a volume/year; since crucial results are imported from these works, the author should update the reference list to published or fully accessible versions, or state which parts of [38] are available.
[Section 6] The discussion of the local-time-independent case is informal and does not state a theorem; if this section is intended as a mathematical contribution, the claims in items (a)-(c) should be formulated as propositions with hypotheses and proofs or precise references.

Circularity Check

1 steps flagged · score 4.0 of 10

The vertex half of the unique characterization rests on the unproved local-time scaling estimate (24) imported from the in-preparation companion paper [38]; the comparison/uniqueness half is inherited from the author's accepted [40]. This is a load-bearing self-citation gap rather than an algebraic circle, giving partial circularity.

self citation load bearing [Section 5, Proof of Theorem 2.7, Eq. (24)-(28)]
"This important fact has been proven in [38] Proposition 8.1, and will be used in the sequel: Ch 2 ≥ EQt,x,i,l β,O [τh − t⋆] ≥ 0, lim hց0 1 h EQt,x,i,l β,O [l(τh) − l⋆] = 1 . (24) where C > 0 is a uniform constant (depending on the data of the system) in dependent of h."

Inequality (25) divides the vertex integral by h and then uses (24) to replace h^{-1}E^{Q}[l(τ_h)-l_*] by 1, which is exactly what converts the bracket into ∂_l φ(t_*,0,l_*) + sup_ϑ{Σ_i S_i ∂_x φ_i + h_0} ≥ 0. The limiting coefficient in the Kirchhoff transmission is therefore imported, without proof, from [38, Prop. 8.1], an in-preparation companion paper by the same authors. If the true limit were C ≠ 1, the recovered condition would be ∂_l φ + C sup{...} = 0; if the local time did not grow linearly in h, no Kirchhoff-type transmission would be obtained at all. The central claim that the value function solves (6) at the vertex thus reduces to an unverified self-citation rather than to a proof contained in the present paper.

full rationale

The paper is not circular in the strict sense of defining X via Y or fitting a parameter and renaming it a prediction. The value function (Definition 2.2) is a genuine sup over controlled martingale problems, and the HJB system (6) is derived from the controlled generator, not assumed equal to the value function. The DPP is adapted from El Karoui-Nguyen-Jeanblanc-Picqué [32]; the comparison theorem is inherited from the author's accepted [40], which is an external published proof and therefore real evidence under the self-citation rule. The one genuinely load-bearing gap is Eq. (24): the h^{-1}E[l(τ_h)-l_*]→1 and E[τ_h-t_*]≤Ch^2 estimates are quoted from [38, Prop. 8.1], an in-preparation companion paper with overlapping authorship, and the vertex viscosity proof divides by h and uses the unit constant to produce the nonlinear local-time Kirchhoff transmission. This is a self-citation load-bearing step with no proof in the present text; if the constant were not 1, or the growth not linear, the recovered boundary condition would change or disappear. This raises the circularity score to 4 (some self-citation; central claim has independent content), but it is not an Eq.-equals-Eq. construction and not a fitted prediction.

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

No free parameters are fitted; all coefficients are inputs. The central claim rests on five external or unproved inputs: the local-time spider martingale problem [37], linear PDE regularity [39], the comparison theorem [40], the exit-time asymptotics [38] (in preparation), and the conditioning/concatenation stability [44]. No new physical entities are postulated; the nonlinear local-time Kirchhoff boundary condition is a mathematical construction derived from the control problem, not an independent entity.

assumptions (5)

domain assumption Well-posedness of the spider martingale problem with local-time-dependent coefficients and spinning measure (Thm 3.1 and Remark 2.1 in [37])
Used to ensure the set of admissible rules A(t,x,i,l) is nonempty (Remark 2.1) and to derive path estimates in Proposition 3.4 and 3.5.
domain assumption Regularity and comparison for linear parabolic systems with local-time Kirchhoff boundary condition ([39])
This PDE well-posedness is the basis for uniqueness in law of the martingale problem and for the linear-system comparison used in [37].
domain assumption Comparison theorem for spider HJB equations with nonlinear local-time Kirchhoff transmission ([40]; restated here as Theorem 2.5)
The paper's uniqueness step in Theorem 2.7 rests on this comparison result. Section 4 only sketches the proof and explicitly refers to [40] for the main construction.
domain assumption Local-time exit estimate: E[tau_h - t] <= C h^2 and lim_{h->0} (1/h)E[l(tau_h)-l] = 1 ([38] Prop 8.1)
Used in Section 5 to pass to the limit h -> 0 in the viscosity sub/supersolution proof at the vertex. This is from an 'in preparation' reference and is not proved here.
standard math Stability by conditioning and concatenation of admissible rules (Propositions 3.7-3.9, following [44])
Foundation of the weak dynamic programming principle; the propositions are stated without proof, justified by a reference to Stroock-Varadhan.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Stochastic scattering control of spider diffusion governed by an optimal diffraction probability measure selected from its own local-time." pith.science (2026). https://pith.science/paper/QEHJNZWR

@misc{pith2026250118057,
  author       = {Pith},
  title        = {Pith review of: Stochastic scattering control of spider diffusion governed by an optimal diffraction probability measure selected from its own local-time},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QEHJNZWR}},
  note         = {Machine review of arXiv:2501.18057}
}

read the original abstract

The purpose of this article is to study a new problem of stochastic control, related to Walsh's spider diffusion, named: stochastic optimal scattering control. The optimal scattering control of the spider diffusion at the junction point is governed by an appropriate and highly non-trivial condition of the Kirchhoff Law type, involving an optimal diffraction probability measure selected from the own local time of the spider process at the vertex. In this work, we prove first the weak dynamic programming principle in the spirit of [32], adapted to the new class of spider diffusion introduced recently in [37]-[38]. Thereafter, we show that the value function of the problem is characterized uniquely in terms of a Hamilton Jacobi Bellman (HJB) system posed on a star-shaped network, having a new boundary condition at the vertex called : non linear local-time Kirchhoff's transmission. The key main point is to use the recent comparison theorem obtained in [40], that has significantly unlocked the study of this type of problem. We conclude by discussing the formulation of stochastic scattering control problems, where there is no dependency w.r.t. the local-time variable, for which their well-posedness appear as a simpler consequence of the results of this work and the advances contained in [40].

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

47 extracted references · 46 canonical work pages

[40]

Comparison principle for Walsh's spider HJB equations with non linear local time Kirchhoff's boundary transmission

I.Ohavi. Comparison principle for Walsh’s spider HJB e quations with non linear local time Kirchhoﬀ’s boundary transmission. arXiv:2312.01362. Acc epted, to appear in Journal of Mathematical Analysis and Applications , (2025)

work page Pith review arXiv 2025
[37]

Martingale problem for a Walsh ’s spider diﬀusion with spinning measure selected from its own local time

M.Martinez and I.Ohavi. Martingale problem for a Walsh ’s spider diﬀusion with spinning measure selected from its own local time. Accepted, to appea r in Electronic Journal of Prob- ability, 2025

work page 2025
[38]

On spider diﬀusion having a spi nning measure selected from its own local-time

M.Martinez and I.Ohavi. On spider diﬀusion having a spi nning measure selected from its own local-time. In preparation

work page
[39]

and I.Ohavi

Martinez M. and I.Ohavi. Well posedness of linear Parab olic partial diﬀerential equation posed on a star-shaped network with local-time Kirchhoﬀ’s b oundary condition at the vertex. Journal of Mathematical Analysis and Applications , 537, (2024), pp. 128–294

work page 2024
[32]

Compac tiﬁcation methods in the control of degenerate diﬀusions: Existence of an optimal control

N.El Karoui, D.Nguyen and M.Jeanblanc-Picqué. Compac tiﬁcation methods in the control of degenerate diﬀusions: Existence of an optimal control. Journal Stochastics: An International Journal of Probability and Stochastic Processes , Volume 20, Issue 3, p. 169-219, (1987)

work page 1987
[44]

Multidimensional Diﬀus ion Processes, (2004)

D.W.Stroock, S.R.S Varadhan. Multidimensional Diﬀus ion Processes, (2004)

work page 2004
[1]

Serve the shortest queue and Walsh Brown ian motion, The Annals of Applied Probability, 29, 1, (2019) pp

Atar R., Cohen A. Serve the shortest queue and Walsh Brown ian motion, The Annals of Applied Probability, 29, 1, (2019) pp. 613–651

work page 2019
[2]

Achdou, M.-K

Y. Achdou, M.-K. Dao, O. Ley, and N. Tchou. A Class of Inﬁni te Horizon Mean Field Games on Networks. Networks and Heterogeneous Media , 14.3 (2019), pp. 537–566

work page 2019

Show all 47 references

[3]

Achdou, M.-K

Y. Achdou, M.-K. Dao, O. Ley, and N. Tchou. Finite horizon mean ﬁeld games on networks. Calc. Var. Partial Diﬀerential Equations 59.5 (2020), Paper No. 157, 34

2020
[4]

Barles, A

G. Barles, A. Briani and E. Chasseigne. A Bellman approac h for two-domains optimal control problems in RN . ESAIM Control Optim. Calc. Var. 19, no. 3, (2013)

2013
[5]

Barles, A

G. Barles, A. Briani and E. Chasseigne. A Bellman approac h for regional optimal control problems in RN . SIAM J. Control Optim. 52 , no. 3, (2014)

2014
[6]

Barles and E

G. Barles and E. Chasseigne. (Almost) everything you alw ays wanted to know about deter- ministic control problems in stratiﬁed domains. Netw. Heterog. Media 10 , no. 4, (2015)

2015
[7]

On Modern Approaches of Hami lton-Jacobi Equations and Con- trol Problems with Discontinuities

G.Barles and E.Chasseigne. On Modern Approaches of Hami lton-Jacobi Equations and Con- trol Problems with Discontinuities. A Guide to Theory, Appl ications, and Some Open Prob- lems. Progress in non linear diﬀerential equations and their Appl ications, (104), 2024

2024
[8]

Barles, O

G. Barles, O. Ley, and E. Topp. Non local Hamilton-Jacobi equations on a network with Kirchhoﬀ type conditions. arXiv:2411.13126, (2024)

2024 arXiv
[9]

Bressan and Y

A. Bressan and Y. Hong. Optimal control problems on strat iﬁed domains. Netw. Heterog. Media., 2, (2007)

2007
[10]

Cardaliaguet, N

P. Cardaliaguet, N. Forcadel, T. Girard and R.Monneau. Conservation laws and Hamilton- Jacobi equations on a junction: The convex case. Discrete and Continuous Dynamical Sys- tems, Vol. 44, No. 12, pp. 3920-3961, (2024)

2024
[11]

Cardaliaguet and P

P. Cardaliaguet and P. E. Souganidis. An optimal contro l problem of traﬃc ﬂow on a junction. ESAIM Control Optim. Calc. Var , 30 (2024) 88. STOCHASTIC OPTIMAL SCATTERING CONTROL 51

2024
[12]

Scattering Theory

Y.B.Band and Y.A vishai. Scattering Theory. Quantum Mechanics with Applications to Nan- otechnology and Information Science (2013), Pages 605-747

2013
[13]

Emery F.B

M.T.Barlow, M. Emery F.B. Knight, S.Song and M.Yor. Aut our d’un théorème de Tsirelson sur des ﬁltrations browniennes et non browniennes. Séminaire de Probabilités , XXXII, vol 1686 of Lecture Notes in Math., pp.264-305, Springer, (1998 )

1998
[14]

On Walsh’s Brownian mot ions

M.Barlow, J.Pitman, and M.Yor. On Walsh’s Brownian mot ions. Séminaire de Probabilités . Volume 23 (1989), pp. 275-293

1989
[15]

Embedding of Walsh Brownian m otion

E.Bayraktar and X.Zhang. Embedding of Walsh Brownian m otion. Stochastic Processes and their Applications ; (2021), 134, pp. 1-28

2021
[16]

Walsh diﬀusions as time changed multi-parameter pro- cess, preprint (2023), Arxiv.2204.07101

E.Bayraktar, J.Zhang, and X.Zhang. Walsh diﬀusions as time changed multi-parameter pro- cess, preprint (2023), Arxiv.2204.07101

2023 arXiv
[17]

Crandall, H

M.G. Crandall, H. Ishii and P.L. Lions. User’s guide to v iscosity solutions of second order Partial Diﬀerential Equations. Bull. Amer. Soc. , 27, (1992)

1992
[18]

Stationary Mean Field Games on n etworks with sticky transition conditions

J.Berry and F.Camilli. Stationary Mean Field Games on n etworks with sticky transition conditions. arXiv:2406.19739, (2024)

2024 arXiv
[19]

Stochastic targets with mixed diﬀusion pr ocesses

B.Bouchard. Stochastic targets with mixed diﬀusion pr ocesses. Stochastic Processes and their Applications, 101:273–302, (2002)

2002
[20]

A study of Browni an motion using light scattering

N.A Clark, J.H Lunacek and G.B Benedek. A study of Browni an motion using light scattering. American Journal of Physics , Vol. 38, number 5, (1970)

1970
[21]

De Meyer

B. De Meyer. Une simpliﬁcation de l’argument de Tsirels on sur le caractère non-brownien des processus de Walsh, Séminaire de Probabilités, XXXIII , Lecture Notes in Math., 1709 217–220, Springer, Berlin, (1999)

1999
[22]

Diﬀusion processes on graphs : stochastic diﬀerential equations, large deviation principle

M.Freidlin and S-J.Sheu. Diﬀusion processes on graphs : stochastic diﬀerential equations, large deviation principle. Probability Theory and Related Fields , 116(2), p. 181-220, (2000)

2000
[23]

Diﬀusion processes on gr aphs and the averaging principle, Ann

M.Freidlin and A.D.Wentzell. Diﬀusion processes on gr aphs and the averaging principle, Ann. Probab.. The Annals of Probability , 21, (1993)

1993
[24]

Wentzell

M.Freidlin and A.D. Wentzell. Diﬀusion processes on an open book and the averaging princi- ple. Stochastic Process. Appl., 113, (1), (2004), p. 101-126

2004
[25]

Stochastic integral equations for Walsh semi- martingales, Annales de l’Institut Henri Poincaré Probabilités et Statist iques, 54, (2018)

T.Ichiba, I.Karatzas, V.Prokaj and M.Yan. Stochastic integral equations for Walsh semi- martingales, Annales de l’Institut Henri Poincaré Probabilités et Statist iques, 54, (2018)

2018
[26]

Stationary distributions a nd convergence for Walsh diﬀusions, Oﬃcial Journal of the Bernoulli Society for Mathematical Stat istics and Probability, 25 (2019)

T.Ichiba and A.Sarantsev. Stationary distributions a nd convergence for Walsh diﬀusions, Oﬃcial Journal of the Bernoulli Society for Mathematical Stat istics and Probability, 25 (2019)

2019
[27]

Imbert and R

C. Imbert and R. Monneau. Flux-limited solutions for qu asi-convex Hamilton Jacobi equations on networks preprint. Annales scientiﬁques de l’école normale supérieure , 50, fascicule 2, (2017). 52 ISAAC OHA VI

2017
[28]

Imbert and V

C. Imbert and V. D. Nguyen. Generalized junction condit ions for degenerate parabolic equa- tions. Calculus of Variations and PDE’s , 56, (2017)

2017
[29]

Sur un type de convergence intermé diaire entre la convergence en loi et la convergence en probabilité

J.Jacod and J.Memin. Sur un type de convergence intermé diaire entre la convergence en loi et la convergence en probabilité. Strasbourg seminary, (tome 15), p. 529-546, (1981)

1981
[30]

Invariant measures and disintegration s with applications to Palm and related kernels

O.Kallenberg. Invariant measures and disintegration s with applications to Palm and related kernels. Probability Theory and Related Fields . Volume 139, Issue 1–2, p. 285-310, (2007)

2007
[31]

Semimartingales on rays, Walsh diﬀusions, and related problems of control and stopping

I.Karatzas, and M.Yan. Semimartingales on rays, Walsh diﬀusions, and related problems of control and stopping. Stochastic Processes and their Applications, Volume 129, Issue 6, 1921-1963, June 2019

1921
[33]

On the constructions of the skew Brownian moti on

A.Lejay. On the constructions of the skew Brownian moti on. Probab. Surveys 3(none): 413- 466, (2006)

2006
[34]

Viscosity solutions for j unctions: well posedness and stability

P.L.Lions and P.Souganidis. Viscosity solutions for j unctions: well posedness and stability. Rend. Lincei Mat. Appl . 27, (2016)

2016
[35]

Well-posedness for multi -dimensional junction problems with Kirchhoﬀ-type conditions

P.L.Lions and P.Souganidis. Well-posedness for multi -dimensional junction problems with Kirchhoﬀ-type conditions. Rend. Lincei Mat. Appl . 28, 2017

2017
[36]

Eﬀective transmission co nditions for second-order elliptic equa- tions on networks in the limit of thin domains

P.L.Lions and P.Souganidis. Eﬀective transmission co nditions for second-order elliptic equa- tions on networks in the limit of thin domains. Compte rendus Mathematiques . Vol. 358, Issue 7, (2020)

2020
[41]

Quasi linear parabolic pde posed on a network w ith nonlinear Neumann boundary condition at vertices

I.Ohavi. Quasi linear parabolic pde posed on a network w ith nonlinear Neumann boundary condition at vertices. Journal of Mathematical Analysis and Applications . Volume 500, Issue 1, 2021

2021
[42]

Some Notes on Standard Borel and Related Spa ces

C.Preston. Some Notes on Standard Borel and Related Spa ces. arXiv:0809.3066, 2008. STOCHASTIC OPTIMAL SCATTERING CONTROL 53

2008 arXiv
[43]

Construction of right processes from ex cursions, Probability Theory and Related Fields, 73 (1986)

Salisbury T.S. Construction of right processes from ex cursions, Probability Theory and Related Fields, 73 (1986)

1986
[45]

Triple points: From non-Brownian ﬁltra tion to harmonic measures

B.Tsirel’son. Triple points: From non-Brownian ﬁltra tion to harmonic measures. Geometric and Functional Analysis , 7, 1096-1142, (1997)

1997
[46]

Classical solvability of linear paraboli c equations on networks

J.Von Below. Classical solvability of linear paraboli c equations on networks. J. Diﬀerential Equations, 72(2), 316-337, (1988)

1988
[47]

J.B. Walsh. A diﬀusion with a discontinuous local time. Temps locaux, Astérisques, pp. 37–45. Société Mathématique de France, (1978). Email address : isaac.ohavi@mail.huji.ac.il & isaac.ohavi@gmail.com

1978

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [40]

Comparison principle for Walsh's spider HJB equations with non linear local time Kirchhoff's boundary transmission

I.Ohavi. Comparison principle for Walsh’s spider HJB e quations with non linear local time Kirchhoﬀ’s boundary transmission. arXiv:2312.01362. Acc epted, to appear in Journal of Mathematical Analysis and Applications , (2025)

work page Pith review arXiv 2025

[2] [37]

Martingale problem for a Walsh ’s spider diﬀusion with spinning measure selected from its own local time

M.Martinez and I.Ohavi. Martingale problem for a Walsh ’s spider diﬀusion with spinning measure selected from its own local time. Accepted, to appea r in Electronic Journal of Prob- ability, 2025

work page 2025

[3] [38]

On spider diﬀusion having a spi nning measure selected from its own local-time

M.Martinez and I.Ohavi. On spider diﬀusion having a spi nning measure selected from its own local-time. In preparation

work page

[4] [39]

and I.Ohavi

Martinez M. and I.Ohavi. Well posedness of linear Parab olic partial diﬀerential equation posed on a star-shaped network with local-time Kirchhoﬀ’s b oundary condition at the vertex. Journal of Mathematical Analysis and Applications , 537, (2024), pp. 128–294

work page 2024

[5] [32]

Compac tiﬁcation methods in the control of degenerate diﬀusions: Existence of an optimal control

N.El Karoui, D.Nguyen and M.Jeanblanc-Picqué. Compac tiﬁcation methods in the control of degenerate diﬀusions: Existence of an optimal control. Journal Stochastics: An International Journal of Probability and Stochastic Processes , Volume 20, Issue 3, p. 169-219, (1987)

work page 1987

[6] [44]

Multidimensional Diﬀus ion Processes, (2004)

D.W.Stroock, S.R.S Varadhan. Multidimensional Diﬀus ion Processes, (2004)

work page 2004

[7] [1]

Serve the shortest queue and Walsh Brown ian motion, The Annals of Applied Probability, 29, 1, (2019) pp

Atar R., Cohen A. Serve the shortest queue and Walsh Brown ian motion, The Annals of Applied Probability, 29, 1, (2019) pp. 613–651

work page 2019

[8] [2]

Achdou, M.-K

Y. Achdou, M.-K. Dao, O. Ley, and N. Tchou. A Class of Inﬁni te Horizon Mean Field Games on Networks. Networks and Heterogeneous Media , 14.3 (2019), pp. 537–566

work page 2019