Remainders of generalised Taylor expansions and a priori bounds for rough differential equations

Alberto Bonicelli; Lorenzo Agabiti; Lorenzo Zambotti

REVIEW 1 major objections 3 minor 36 references

Remainders of generalised Taylor expansions and a priori bounds for rough differential equations

T0 review · 1 major / 3 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read This paper proves global a priori bounds for rough differential equations assuming only Lipschitz continuity of the coefficient combinations called elementary differentials, for any Hölder exponent in (0,1].

desk verdict Real advance in rough path a priori bounds, but the central combinatorial identity (Cor 4.31) has a wrong leaf label as written; likely repairable, but needs correction before the proof chain can be trusted. read the letter →

arxiv 2607.18635 v2 pith:ME3X5OAF submitted 2026-07-21 math.PR math.COmath.RA

Lorenzo Agabiti , Alberto Bonicelli , Lorenzo Zambotti This is my paper · ORCID

classification math.PRmath.COmath.RA MSC 60L2016T0505C05

keywords roughdifferentialequationsbranchedpathsaprioriestimateselementarydifferentialsgeneralisedTaylorexpansionspre-Liealgebrasplanarbinarytreesgraftingoperation

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

The paper establishes that every solution remainder in the paper's local truncated expansion of a rough differential equation is controlled by the driver, the initial condition, and the elementary differentials — no boundedness or coercivity of the coefficients is needed. The engine is a closed formula for the remainder of a generalised Taylor expansion, written purely in terms of gradients of those elementary differentials, which exploits cancellations optimally. If the proof is right, it provides the quantitative control needed for a global solution theory and for continuity of the solution map in the full Hölder range.

What carries the argument

The central object is the explicit expansion of the forest-grafting operation — the canonical pre-Lie grafting extended to forests — as a signed sum over planar binary trees, with combinatorial coefficients C(ϱ) that count reduction chains and multi-indices J_{ϱ,ℓ} that choose evaluation points. This expansion rewrites the remainder B^{m+1}(τ) so that only first derivatives ∇Υτ appear, making global Lipschitz assumptions on the elementary differentials sufficient for the bounds.

What would settle it

Expand (τ1τ2τ3)↷τ0 directly from the recursive definition of the grafting extension and compare it, term by term, with the formula's signed sum over planar binary trees for k=3; any mismatch in a coefficient, sign, or evaluation point disproves the central identity and therefore the remainder representation.

Watch

Extended reading notes

Core claim

The main result is a quantitative a priori bound: under global Lipschitz continuity of the elementary differentials of order at most N-1 and Hölder continuity of those of order N, every weighted Hölder norm of the solution remainders is bounded by a constant times sums of |τ| |Υτ(Z0)| ‖X(τ)‖_{|τ|α} over all decorated trees up to order N. The bound holds for sufficiently small interval length T or weight parameter μ. The proof rests on an explicit algebraic formula, Theorem 4.36, expressing the generalised Taylor remainder as a sum over planar binary trees of products of first derivatives of elementary differentials evaluated at carefully chosen interpolation points.

Load-bearing premise

The argument collapses if the explicit combinatorial identity for the grafted forest expansion — the signed sum over planar binary trees with coefficients C(ϱ) and evaluation points J_{ϱ,ℓ} — contains any error in signs, coefficients, or evaluation points.

Editorial extensions

If this is right

A full well-posedness theory for local (Davie-type) solutions of rough differential equations becomes possible without bounded or coercive coefficients.
The solution map (initial condition, driving path) to the solution is continuous in the natural Hölder topologies.
The a priori control of the highest-order remainder reduces to control of lower-order remainders, yielding a recursive estimate scheme that is uniform in the initial condition.
The result covers the entire range α∈(0,1], including α=1, and requires no separate treatment of linear-growth or unbounded coefficients.
The same remainder representation could serve as a quantitative tool for numerical error analysis of rough-path integrators.

Reading between the lines

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

Editorial extension: the cancellation mechanism encoded by the planar-binary-tree expansion may transfer to regularity structures, where the analogous post-Lie grafting could yield a priori bounds for singular SPDEs with unbounded coefficients.
Editorial extension: the combinatorial coefficients C(ϱ) are defined by a natural reduction-chain recurrence; a concrete test is to compare them against direct expansions for forests of size 2 and 3 — a symbolic computation could settle the identity independently.
Editorial extension: the smallness condition on T or μ, needed to absorb error terms, might be removable in many concrete equations by a continuation/bootstrap argument, giving genuinely global-in-time bounds.
Editorial extension: the explicit formula may allow computing the proportionality constants, enabling quantitative rather than merely qualitative a priori estimates.

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: this is a substantial paper and worth one full referee cycle. The new a priori estimates for Davie solutions under Lipschitz elementary differentials — no boundedness or coercivity, all α∈(0,1] — look right, and the analytic skeleton (Sewing bound plus a bootstrap inequality absorbed for small ε) is sound. The genuinely new ingredient is the closed-form remainder formula that only involves gradients, and the planar-binary-tree expansion of Guin–Oudom grafting. The authors are honest about what they do and do not prove, and the comparison with [BCMW22, CG25, Lej12] is accurate.

The soft spot is exactly where the reader's report suspected: the 30-page combinatorial section 4 is not machine-checked and is extremely intricate. I found a concrete error in Corollary 4.31 as displayed. The sum is over ϱ∈T^{p(I_{k−1})}_b but the insertion is written as •_k→_ℓϱ. To build a tree in T^{p(I_k)}_b from one in the (k−1)-tree set, the new leaf must carry decoration p(k), not k. The displayed formula is false as written: for k=2, p=(2,1), it produces spurious τ2↷(τ2↷τ0) terms. The proof of Proposition 4.38 explicitly uses all possible insertions of •_{p(k)}, so the intended statement is clear and the mistake is a typo with consequences. But a false statement in the middle of the main theorem's proof chain is not cosmetic; the manuscript must be corrected before the chain can be trusted.

Also minor: the metadata abstract says α∈(0,1), the full text says (0,1]; normalize it. The paper defers global well-posedness to a companion paper — that's a limitation, but clearly stated and not an overclaim.

If the Corollary is fixed (or a referee confirms the intended statement is the one used in Prop 4.38), the main theorem stands and the contribution is significant: it removes a standard boundedness/coercivity crutch and gives quantitative control of all solution remainders in terms of the driver and initial data.

I would send this to peer review. The referee should verify §4.1–4.4, especially the signs in Lemma 4.11 / Corollary 4.31 and the evaluation multi-indices in Lemma 4.25. The analytic part of §5 is in good shape.

Referee Report

1 major / 3 minor

Summary. The paper develops an explicit algebraic representation for the remainder of a generalised Taylor expansion of elementary differentials, expressed through planar binary trees, reduction chains, and a family of multi-indices. This representation (Theorem 4.36) is then used to prove global a priori bounds for Davie solutions of branched rough differential equations under only Lipschitz/Hölder regularity of the elementary differentials, with no boundedness or coercivity assumption. The central result is Theorem 5.4 (Theorem 1.1), which controls all solution remainders by the driver, the initial condition, and the initial values of the elementary differentials. The paper is largely combinatorial in character and the analytic part is comparatively short once the algebraic formula is accepted.

Significance. If the main result is correct, it is a substantial contribution: it removes boundedness/coercivity assumptions that appear in earlier treatments and provides quantitative global control of Davie solution remainders. The combinatorial machinery — especially Lemma 4.11 and Theorem 4.36 — is novel and likely to be of independent interest. The proof is detailed and self-contained, with definitions and lemmas checked locally; the paper also contains useful appendices on grafting and weighted norms. The principal risk is the correctness of the extremely intricate combinatorial identity chain; the argument is not machine-checked and independent verification is needed.

major comments (1)

[Corollary 4.31] The displayed formula has a wrong leaf decoration: the insertion should be •_{p(k)}, not •_k. For a fixed p∈Σ_k, the tree ϱ is in T^{p(I_{k−1})}_b, so its leaves carry decorations p(1),...,p(k−1); the missing label needed to obtain a tree in T^{p(I_k)}_b is p(k). As stated, for k=2, τ1≠τ2 and p=(2,1), the formula produces terms G(•_2→ℓ ϱ), i.e. insertions of τ2 into a tree already labelled by τ1, rather than the correct G(•_1→ℓ ϱ). This is not cosmetic: Proposition 4.38 explicitly invokes this corollary, and its proof at equations (4.34)–(4.36) uses •_{p(k)} with the explicit statement that insertions of •_{p(k)} span T^{p(I_k)}_b. Thus the manuscript as written contains a false statement in the load-bearing combinatorial chain Lemma 4.11 → Corollary 4.31 → Proposition 4.38 → Theorem 4.36 → Lemma 5.10 → Theorem 5.4. The intended statement is clearly repairable by replacing •_k with •_{p(

minor comments (3)

[Theorem 5.4] In the first display, the summand 'Σ_{m+1}^{ℓ=2} ∥Z[ℓ]∥_{ℓα,µ}' appears to contain a typo: the upper limit should presumably be N+1 (or the sum should be over ℓ=2,...,N+1), since m is not defined in the statement.
[Abstract and Introduction] The abstract says α∈(0,1] while the Introduction says α∈(0,1); the main theorem is stated for α∈(0,1] in Assumption 5.1, so the abstract should be aligned.
[Proof of Theorem 4.36] In the base case m=0, the text writes 'R_t^s ∇G(•0)(u1)·z[1] dt1' with an undefined t,s; this is a notational slip from the rough-path specialization and should be written as an integral over u1∈[0,1].

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the a priori bound follows from a self-contained algebraic identity and a directly proved Sewing bound.

full rationale

The central claim (Theorem 5.4) is derived from the generalized Taylor remainder identity (Theorem 4.36 / equation (4.26)), which is proved by induction from the algebraic definitions of Guin–Oudom grafting, the multi-indices J, and the evaluation map e. The proof does not assume the bound it is later used to prove. Lemma 5.10 and Theorem 5.4 then perform a standard small-epsilon absorption argument: with finiteness of the remainders supplied by Proposition 5.9, the estimate takes the form A <= data + c epsilon^{beta alpha} A, which is closed by choosing epsilon small. No parameter is fitted to the target quantities, and no definition identifies the conclusion with an assumption. The Sewing bound is attributed to [CGZ25], which shares an author with the present paper, but the paper also proves the bound directly in Lemma 5.6 and Theorem 5.7, so the citation is not load-bearing. Other self-citations, such as [BHZ19] and [JDJZ25], appear only in contextual remarks about SPDE extensions and play no role in the proof of Theorem 5.4. Any possible index error in Corollary 4.31 would be a correctness issue, not an instance of circular reasoning.

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

The paper introduces no empirical postulates. The planar binary trees, reduction chains, multi-indices J_{ϱ,ℓ}, and the map A are internal proof devices with no falsifiable handle outside the mathematics. The only free-standing input is the regularity assumption on the elementary differentials, which is the theorem's hypothesis rather than a fitted parameter.

assumptions (5)

domain assumption The driving path X admits a branched α-rough path lift X satisfying Chen's relation (2.10) with the paper's nonstandard product ⋆ (Definition 2.5).
The entire estimate is conditional on the existence of such a rough path lift; the paper cites standard existence theory but does not prove it here. Lemma A.4 verifies the Chen relation for smooth lifts.
domain assumption The Davie solution condition (2.14): Z satisfies δZ_st = Σ_{τ∈T≤N} Υτ(Z_s) X_st(τ) + o(t−s).
Used to apply the Sewing bound, which requires the top-order remainder to be o(t−s). This is the definition of solution à la Davie, not an extra hypothesis.
domain assumption Assumption 5.1: Υτ globally Lipschitz for τ∈T≤N−1 and Υτ∈C^β for τ∈T^=N, with β∈(1/α−N,1].
The theorem hypothesis. The a priori bound is proved under these regularity conditions; they replace the stronger boundedness/coercivity assumptions of earlier works.
standard math Guin–Oudom extension of grafting (2.6) and the morphism property Υ_{(τ1⋯τk)↷τ} = ∇^k Υτ · [Υτ1⋯Υτk] (Lemma 2.8).
Taken from [OG08] and proved in Appendix A.1. The morphism property is essential to rewrite grafting of forests in terms of gradients of Υ.
standard math Sewing bound (Theorem 1.2) and the weighted-norm bounds (B.3)–(B.7).
Classical analytic tools used to convert three-point increments ∥δR∥_η into two-point bounds ∥R∥_η, and to control sup norms in the weighted spaces.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Remainders of generalised Taylor expansions and a priori bounds for rough differential equations." pith.science (2026). https://pith.science/paper/ME3X5OAF

@misc{pith2026260718635,
  author       = {Pith},
  title        = {Pith review of: Remainders of generalised Taylor expansions and a priori bounds for rough differential equations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ME3X5OAF}},
  note         = {Machine review of arXiv:2607.18635}
}

abstract

In this article we establish global a priori estimates on the solution of a generic rough differential equation driven by an $\alpha$-H\"older path covering the full range of regularity $\alpha\in(0,1)$, under the hypothesis of Lipschitz continuity (but neither boundedness nor coercivity) of specific combinations of the coefficients and their derivatives, called elementary differentials. Our main tool is a closed formula for the remainder of a generalised Taylor expansion in terms of products of gradients of the elementary differentials, which allows to take advantage of cancellations in an optimal way.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 1 linked inside Pith

[1]

Bailleul

I. Bailleul. Flows driven by rough paths. Revista matem \'a tica iberoamericana , 31(3):901--934, 2015

2015
[2]

Bailleul and R

I. Bailleul and R. Catellier. Non-explosion criteria for rough differential equations driven by unbounded vector fields. In Annales de la Facult \'e des sciences de Toulouse: Math \'e matiques , volume 29, pages 721--759, 2020

2020
[3]

Bonnefoi, A

T. Bonnefoi, A. Chandra, A. Moinat, and H. Weber. A priori bounds for rough differential equations with a non-linear damping term. Journal of Differential Equations , 318:58--93, 2022

2022
[4]

Bruned, M

Y. Bruned, M. Hairer, and L. Zambotti. Algebraic renormalisation of regularity structures. Invent. Math. , 215:1039--1156, 2019

2019
[5]

Bruned and F

Y. Bruned and F. Katsetsiadis. Post- L ie algebras in regularity structures. Forum Math. Sigma , 11:Paper No. e98, 20, 2023

2023
[6]

Bruned and D

Y. Bruned and D. Manchon. Algebraic deformation for ( S ) PDE s. J. Math. Soc. Japan , 75(2):485--526, 2023

2023
[7]

A. Cayley. On the theory of the analytical forms called trees. Philos. Mag. (5) , 13(85):172--176, 1857
[8]

Chevyrev and M

I. Chevyrev and M. Gubinelli. Large field problem in coercive singular PDE s. arXiv preprint arXiv:2510.20716 , 2025

arXiv 2025

Show all 36 references

[9]

Caravenna, M

F. Caravenna, M. Gubinelli, and L. Zambotti. Lectures on rough paths. Lecture notes, 2025. https://www.lpsm.paris/_media/users/zambotti/bookcgz.pdf

2025
[10]

Chapoton

F. Chapoton. Operads and algebraic combinatorics of trees. S \'e minaire Lotharingien de Combinatoire , 58:B58c--27, 2007

2007
[11]

A. Davie. Differential equations driven by rough paths: an approach via discrete approximation. Applied Mathematics Research eXpress , 2008

2008
[12]

Munthe-Kaas

Kurusch Ebrahimi-Fard, Alexander Lundervold, and Hans Z. Munthe-Kaas. On the L ie enveloping algebra of a post- L ie algebra. J. Lie Theory , 25(4):1139--1165, 2015

2015
[13]

Feyel and A

D. Feyel and A. de La Pradelle. Curvilinear integrals along enriched paths. Electron. J. Probab. , 11:no. 34, 860--892, 2006

2006
[14]

P. K. Friz and M. Hairer. A course on rough paths . Springer, 2020

2020
[15]

Friz and N

P. Friz and N. Victoir. Euler estimates for rough differential equations. Journal of Differential Equations , 244(2):388--412, 2008

2008
[16]

Gubinelli

M. Gubinelli. Controlling rough paths. J. Funct. Anal. , 216(1):86--140, 2004

2004
[17]

Gubinelli

M. Gubinelli. Ramification of rough paths. J. Differ. Equ. , 248(4):693--721, 2010

2010
[18]

M. Hairer. A theory of regularity structures. Invent. Math. , 198(2):269--504, 2014

2014
[19]

Hairer and D

M. Hairer and D. Kelly. Geometric versus non-geometric rough paths. In Annales de l'IHP Probabilit \'e s et statistiques , volume 51, pages 207--251, 2015

2015
[20]

M. Hoffman. Combinatorics of rooted trees and H opf algebras. Transactions of the American Mathematical Society , 355(9):3795--3811, 2003

2003
[21]

Jean-David Jacques and L

J.-D. Jean-David Jacques and L. Zambotti. Post-lie algebras of derivations and regularity structures. J. Comb. Algebra , online first, 2025

2025
[22]

On rough differential equations

Antoine Lejay. On rough differential equations. Electron. J. Probab. , 14:no. 12, 341--364, 2009

2009
[23]

A. Lejay. Global solutions to rough differential equations with unbounded vector fields. In S \'e minaire de probabilit \'e s XLIV , pages 215--246. Springer, 2012

2012
[24]

Linares, F

P. Linares, F. Otto, and M. Tempelmayr. The structure group for quasi-linear equations via universal enveloping algebras. CAMS , 3(01):1--64, 2023

2023
[25]

Li and K

X.M. Li and K. Ying. Strong completeness of SDE s and non-explosion for RDE s with coefficients having unbounded derivatives. arXiv preprint arXiv:2502.08799 , 2025

2025
[26]

T. J. Lyons. Differential equations driven by rough signals. Rev. Ma. Iberoam. , 14(2):215--310, 1998

1998
[27]

Rough paths and H opf algebras

Dominique Manchon. Rough paths and H opf algebras. J. Comput. Dyn. , 12(1):160--177, 2025

2025
[28]

Manchon and A

D. Manchon and A. Sa \" di. Lois pr \'e - L ie en interaction. Communications in Algebra , 39(10):3662--3680, 2011

2011
[29]

J. C. Mourrat and H. Weber. The dynamic ^4_3 model comes down from infinity. Commun. Math. Phys. , 356(3):673–753, October 2017

2017
[30]

J. C. Mourrat and H. Weber. Global well-posedness of the dynamic ^ 4 model in the plane . Ann. Probab. , 45(4):2398 -- 2476, 2017

2017
[31]

Oudom and D

J.M. Oudom and D. Guin. On the L ie enveloping algebra of a pre- L ie algebra. Journal of K-theory , 2(1):147--167, 2008

2008
[32]

F. Otto, J. Sauer, S. A Smith, and H. Weber. A priori bounds for quasi-linear SPDE s in the full subcritical regime. J. Eur. Math. Soc. , 27(1):71--118, 2024

2024
[33]

Otto and H

F. Otto and H. Weber. Quasi-linear SPDE s in divergence form. Stoch. Partial Differ. Equ. Anal. Comput. , 7(1):64--85, 2019

2019
[34]

Otto and H

F. Otto and H. Weber. Quasilinear SPDE s via rough paths. Arch. Ration. Mech. Anal. , 232(2):873--950, 2019

2019
[35]

R. P. Stanley. Enumerative combinatorics. V ol. 2 . 1999

1999
[36]

K. Ying. Non-explosion principles for branched rough differential equations with unbounded coefficients. arXiv preprint arXiv:2606.09904 , 2026

2026 arXiv

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.

[1] [1]

Bailleul

I. Bailleul. Flows driven by rough paths. Revista matem \'a tica iberoamericana , 31(3):901--934, 2015

2015

[2] [2]

Bailleul and R

I. Bailleul and R. Catellier. Non-explosion criteria for rough differential equations driven by unbounded vector fields. In Annales de la Facult \'e des sciences de Toulouse: Math \'e matiques , volume 29, pages 721--759, 2020

2020

[3] [3]

Bonnefoi, A

T. Bonnefoi, A. Chandra, A. Moinat, and H. Weber. A priori bounds for rough differential equations with a non-linear damping term. Journal of Differential Equations , 318:58--93, 2022

2022

[4] [4]

Bruned, M

Y. Bruned, M. Hairer, and L. Zambotti. Algebraic renormalisation of regularity structures. Invent. Math. , 215:1039--1156, 2019

2019

[5] [5]

Bruned and F

Y. Bruned and F. Katsetsiadis. Post- L ie algebras in regularity structures. Forum Math. Sigma , 11:Paper No. e98, 20, 2023

2023

[6] [6]

Bruned and D

Y. Bruned and D. Manchon. Algebraic deformation for ( S ) PDE s. J. Math. Soc. Japan , 75(2):485--526, 2023

2023

[7] [7]

A. Cayley. On the theory of the analytical forms called trees. Philos. Mag. (5) , 13(85):172--176, 1857

[8] [8]

Chevyrev and M

I. Chevyrev and M. Gubinelli. Large field problem in coercive singular PDE s. arXiv preprint arXiv:2510.20716 , 2025

arXiv 2025