Branched harmonic majorants: representations for multidimensional optimal stopping
Pith reviewed 2026-05-22 14:54 UTC · model grok-4.3
The pith
The optimal stopping region is the contact set between a gain function and the infimum of branched harmonic majorants on the unit ball.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
The least superharmonic majorant of g is the pointwise infimum of a family of branched harmonic majorants obtained by arranging classical harmonic functions on smoothly bounded domains in a finite, depth-indexed branching structure. The optimal stopping region is the contact set between g and this infimum, while the value function equals the expected gain of g at the first exit time from the non-contact set. This yields a multidimensional generalisation of the Dynkin-Yushkevich theorem in which the branching structure overcomes the localisation obstruction for Brownian paths in dimensions d greater than or equal to 2.
What carries the argument
branched harmonic majorants: classical harmonic functions on smoothly bounded subdomains arranged in a finite, depth-indexed branching structure that relaxes global majorisation to local constraints on decreasing non-contact sets
If this is right
- The value function of the optimal stopping problem equals the expected gain at the first exit time from the non-contact set.
- Truncation of the branching depth produces a decreasing sequence of envelopes that converges pointwise to the infimum.
- Local majorisation constraints on a decreasing sequence of non-contact sets suffice to recover the global Perron envelope.
- The sequential composition of stopping times overcomes the localisation obstruction arising from thin Brownian paths in dimensions d at least 2.
Where Pith is reading between the lines
- Numerical truncation of the branching depth could yield practical approximation algorithms for computing value functions and free boundaries in multidimensional optimal stopping.
- The construction may extend to other Markov processes or irregular domains where classical harmonic majorants are unavailable.
- It suggests a route to numerical solution of free-boundary problems in higher-dimensional stochastic control by composing local harmonic solutions.
Load-bearing premise
A branching structure can be chosen so that local majorisation constraints on the decreasing sequence of non-contact sets recover the global Perron envelope and sequential stopping times overcome the thinness of Brownian paths in dimensions two and higher.
What would settle it
A concrete continuous gain function g on the two-dimensional unit disk for which the pointwise infimum of all possible branched harmonic majorants differs from the true least superharmonic majorant computed by another method such as direct solution of the obstacle problem.
Figures
read the original abstract
We construct the least superharmonic majorant of a continuous function $g$ on the $d$-dimensional unit ball ($d \geq 2$) via a canonical sequential scheme. While classical theory identifies this majorant with the value function of the optimal stopping problem for Brownian motion absorbed at the domain boundary, no comparable constructive approximation scheme has been available. We introduce branched harmonic majorants, obtained by arranging classical harmonic functions on smoothly bounded domains in a finite, depth-indexed branching structure, and prove two main results. First, the optimal stopping region is identified as the contact set between the gain function $g$ and the pointwise infimum of this family; the value function is recovered as the expected gain at the first exit time from the non-contact set. This yields a multidimensional generalisation of the Dynkin--Yushkevich concave-envelope theorem in which affine functions are replaced by branched harmonic majorants. Second, truncation in the branching depth produces a decreasing sequence of envelopes that converges pointwise to this infimum, yielding an explicit approximation scheme not present in classical formulations. Analytically, the branching structure relaxes the global majorisation constraint to a local constraint imposed on a decreasing sequence of non-contact sets, yielding a representation of the Perron envelope in terms of harmonic functions on smoothly bounded domains. Probabilistically, the construction corresponds to the sequential composition of stopping times and overcomes the localisation obstruction arising from the thinness of Brownian paths in dimensions $d \geq 2$.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The manuscript constructs the least superharmonic majorant of a continuous function g on the d-dimensional unit ball (d ≥ 2) via a canonical sequential scheme using branched harmonic majorants arranged in a finite, depth-indexed branching structure on smoothly bounded subdomains. It proves two main results: the optimal stopping region is the contact set between g and the pointwise infimum of this family, with the value function recovered as the expected gain at the first exit time from the non-contact set (a multidimensional generalization of the Dynkin–Yushkevich theorem); and truncation in branching depth produces a decreasing sequence of envelopes converging pointwise to the infimum, yielding an explicit approximation scheme. Analytically, the branching relaxes the global majorisation constraint to local constraints on a decreasing sequence of non-contact sets; probabilistically, it corresponds to sequential composition of stopping times.
Significance. If the results hold, the work supplies a constructive approximation scheme for the value function of multidimensional optimal stopping problems for Brownian motion, where classical theory guarantees existence of the least superharmonic majorant but provides no comparable explicit scheme. The representation of the Perron envelope via harmonic functions on smoothly bounded domains, together with the sequential stopping-time construction, addresses the localisation obstruction arising from the thinness of paths in d ≥ 2. Credit is due for the parameter-free character of the construction and for the explicit convergence statement under depth truncation.
major comments (1)
- [Proof of the identification of the optimal stopping region (following the definition of the branching structure)] The central claim that the pointwise infimum of the branched harmonic majorants coincides with the least superharmonic majorant of g (and hence with the value function) rests on the assertion that local majorisation constraints on the decreasing sequence of non-contact sets suffice to recover the global Perron envelope. The precise mechanism ensuring that no superharmonic function dominating g can lie strictly between this infimum and the true envelope—particularly when the branching tree is chosen independently of g and when finite depth must control path thinness uniformly in d ≥ 2—requires a self-contained argument in the proof of the first main theorem.
minor comments (2)
- [Construction of branched harmonic majorants] The recursive definition of the branching structure and the precise rule for composing the exit times across branches should be stated formally (e.g., as an inductive definition or displayed equation) rather than described only in prose.
- [Introduction] A short remark on the dependence (or independence) of the branching tree on the gain function g would clarify the scope of the approximation scheme.
Simulated Author's Rebuttal
We thank the referee for the positive assessment of the significance of our results and for the detailed reading of the manuscript. We address the single major comment below and will incorporate clarifications into a revised version.
read point-by-point responses
-
Referee: [Proof of the identification of the optimal stopping region (following the definition of the branching structure)] The central claim that the pointwise infimum of the branched harmonic majorants coincides with the least superharmonic majorant of g (and hence with the value function) rests on the assertion that local majorisation constraints on the decreasing sequence of non-contact sets suffice to recover the global Perron envelope. The precise mechanism ensuring that no superharmonic function dominating g can lie strictly between this infimum and the true envelope—particularly when the branching tree is chosen independently of g and when finite depth must control path thinness uniformly in d ≥ 2—requires a self-contained argument in the proof of the first main theorem.
Authors: We agree that the identification of the pointwise infimum with the least superharmonic majorant is the core of Theorem 1 and that the transition from local majorisation on the nested non-contact sets to the global Perron envelope merits a more isolated and self-contained exposition. In the current proof we establish superharmonicity of the infimum by verifying the mean-value inequality on each harmonic branch and using the monotonicity of the non-contact sets to pass to the limit; we then show minimality by applying optional sampling to the sequence of exit times generated by the branching structure, which yields that any superharmonic majorant dominates every finite-depth approximant and hence their infimum. The independence of the branching tree from g is achieved by fixing a countable dense collection of smoothly bounded subdomains that exhaust the ball, while uniform control on path thinness for finite depth follows from the boundary regularity and the fact that Brownian motion exits each subdomain in finite time almost surely. Nevertheless, we acknowledge that these steps are currently interwoven with the construction and that an explicit lemma separating the localisation argument (including its dimension-independent uniformity) would improve readability. We will therefore add such a lemma immediately after the definition of the branching structure and before the statement of Theorem 1, together with a short comparison to the classical Perron process. revision: yes
Circularity Check
No circularity: construction from classical harmonic functions and exit times is self-contained
full rationale
The paper defines branched harmonic majorants explicitly via classical harmonic functions arranged on smoothly bounded subdomains in a finite-depth branching structure, then identifies the value function as the pointwise infimum and the optimal stopping region as the contact set with g. This identification rests on standard properties of superharmonic functions, the Perron envelope, and Brownian motion exit times rather than any fitted parameter, self-referential definition, or load-bearing self-citation. The claimed generalization of the Dynkin-Yushkevich theorem is presented as an extension using these classical objects, with no reduction of the central claim to its own inputs by construction. The sequential composition of stopping times is offered as a probabilistic representation that addresses localisation, but the argument does not collapse into a renaming or ansatz smuggled via prior work by the same authors.
Axiom & Free-Parameter Ledger
axioms (2)
- standard math Harmonic functions satisfy the mean-value property on smoothly bounded domains
- domain assumption Brownian motion is absorbed at the boundary of the unit ball
invented entities (1)
-
branched harmonic majorants
no independent evidence
Reference graph
Works this paper leans on
-
[1]
E. B. Dynkin and A. A. Yushkevich.Markov processes: Theorems and problems. Nauka Press, Moscow, 1967
work page 1967
-
[2]
S. Dayanik and I. Karatzas. On the optimal stopping problem for one- dimensional diffusions.Stochastic Process. Appl., 107(2):173–212, 2003
work page 2003
-
[3]
Tomasz Kosmala and John Moriarty. Optimal stopping with nonlinear expectation: geometric and algorithmic solutions.The Annals of Applied Probability, 35(5):3310–3333, 2025
work page 2025
-
[4]
A. Bensoussan and J.-L. Lions.Applications of Variational Inequalities in Stochastic Control. North-Holland, 1982
work page 1982
-
[5]
Kristin Reikvam. Viscosity solutions of optimal stopping prob- lems.Stochastics and Stochastic Reports, 62(3-4):285–301, 1998. doi: 10.1080/17442509808834137. URLhttps://doi.org/10.1080/ 17442509808834137
-
[6]
Goran Peskir. Continuity of the optimal stopping boundary for two- dimensional diffusions.The Annals of Applied Probability, 29(1):505– 530, 2019
work page 2019
-
[7]
G. Peskir and A. N. Shiryaev.Optimal Stopping and Free-Boundary Problems. Lectures in Mathematics. ETH Z¨ urich. Birkh¨ auser Basel, Basel, 2006. ISBN 978-3-7643-2419-3. doi: 10.1007/978-3-7643-7390-0
-
[8]
S¨ oren Christensen, Fabi´ an Crocce, Ernesto Mordecki, and Paavo Salmi- nen. On optimal stopping of multidimensional diffusions.Stochastic Pro- cesses and their Applications, 129(7):2561–2581, 2019. ISSN 0304-4149. doi: https://doi.org/10.1016/j.spa.2018.07.014. URLhttps://www. sciencedirect.com/science/article/pii/S0304414918303612
-
[9]
Sigurd Assing, Saul Jacka, and Adriana Ocejo. Monotonicity of the value function for a two-dimensional optimal stopping problem.The Annals of Applied Probability, 24(4):1554 – 1584, 2014. doi: 10.1214/13-AAP956. URLhttps://doi.org/10.1214/13-AAP956
-
[10]
Deep optimal stopping.Journal of Machine Learning Research, 20(74):1–25, 2019
Sebastian Becker, Patrick Cheridito, and Arnulf Jentzen. Deep optimal stopping.Journal of Machine Learning Research, 20(74):1–25, 2019
work page 2019
-
[11]
Deep learning for the multiple op- timal stopping problem.arXiv preprint arXiv:2512.22961, 2025
Mathieu Lauri` ere and Mehdi Talbi. Deep learning for the multiple op- timal stopping problem.arXiv preprint arXiv:2512.22961, 2025. 27
-
[12]
I. Karatzas and S. E. Shreve.Brownian motion and stochastic calculus, volume 113 ofGraduate Texts in Mathematics. Springer-Verlag, New York, second edition, 1991
work page 1991
-
[13]
L.C. Evans and R.F. Gariepy.Measure Theory and Fine Properties of Functions, Revised Edition. Textbooks in Mathematics. CRC Press,
-
[14]
ISBN 9781482242393. 28
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.