Exact Signature Tail Asymptotics for Pure Rough Paths

Nannan Li; Xing Gao

arxiv: 2606.30055 · v1 · pith:I4XNQTIY · submitted 2026-06-29 · math.PR

Exact Signature Tail Asymptotics for Pure Rough Paths

Nannan Li , Xing Gao This is my paper

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel 2026-06-30 05:24 UTCgrok-4.3pith:I4XNQTIY record.json open to challenge →

classification math.PR

keywords rough pathssignature asymptoticspure rough pathstensor algebra normsLie seriestail asymptoticsfinite-dimensional developments

0 comments

The pith

For a pure m-rough path exp(t l), the limsup of scaled signature tail norms equals the m-norm of the top Lie component under any reasonable tensor norms.

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

The paper proves that for the pure m-rough path given by X_t = exp(t l) where l sums Lie elements up to degree m, the quantity limsup as n to infinity of [(n/m)! times the n-norm of the degree-n projection of exp(l)] raised to m/n equals the m-norm of l_m. This holds when the tensor powers carry arbitrary reasonable tensor algebra norms. The identification shows that the signature tail is governed exactly by the local m-variation of the path. The upper bound was previously known; the new work supplies the matching lower bound by reducing to finite-dimensional developments and constructing, for each top tensor, a contractive development in which that norm appears as an eigenvalue of degree m.

Core claim

Let X_t = exp(t l) with l = l_1 + ... + l_m and l_r in the Lie algebra of degree r. Then limsup_{n→∞} [(n/m)! ||π_n(exp l)||_n ]^{m/n} = ||l_m||_m when the tensor algebra is equipped with any reasonable norms. The result confirms the conjecture for these pure paths and extends the statement beyond previously treated norms.

What carries the argument

The norming cyclic construction that produces, for any given top-level tensor l_m, a contractive finite-dimensional development in which ||l_m||_m appears as an eigenvalue at degree m.

If this is right

The signature tail of a pure m-rough path is determined solely by the norm of its highest-degree Lie component.
The same asymptotic holds for every choice of reasonable tensor algebra norms.
Finite-dimensional developments suffice to obtain the sharp lower bound on the tail growth.
Every top-level tensor admits a contractive development realizing its norm as a degree-m eigenvalue.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same reduction technique could be tested on rough paths whose leading term is not strictly the highest level.
Numerical checks for low m and low-dimensional spaces would directly verify the eigenvalue construction.
The cyclic construction may supply explicit truncations useful for simulating the tail behavior of rough differential equations.

Load-bearing premise

For every top-level tensor l_m there exists a norming cyclic construction that yields a contractive development in which the norm of l_m appears as an eigenvalue at degree m.

What would settle it

An explicit pure m-rough path, choice of reasonable tensor norms, and computed sequence of signature norms whose limsup after scaling is strictly smaller than the m-norm of l_m.

read the original abstract

We prove~\cite[Conjecture 2.12]{BGS20} on the signature tail asymptotics of pure rough paths and extend it to arbitrary reasonable tensor norms. In more details, let \[ \mathbf X_t=\exp(tl) \,\text{ with }\, l=l_1+\cdots+l_m\,\text{ and }\, l_r\in\mathcal L_r(V), \] be a pure $m$-rough path over a finite dimensional real or complex Banach space, and equip the tensor powers of $V$ with arbitrary reasonable tensor algebra norms. We prove that \[ \limsup_{n\to\infty}\left(\left(\frac{n}{m}\right)!\left\|\pi_n(\exp l)\right\|_n\right)^{m/n}=\|l_m\|_m . \] In particular, this identifies the signature tail with the local $m$-variation of the pure rough path. The upper bound was obtained in~\cite{BGS20}; the main contribution of the paper is the matching lower bound. Its proof is based on finite dimensional developments and a norming cyclic construction. For every top-level tensor $l_m$, we also build a contractive development in which $\|l_m\|_m$ appears as an eigenvalue at degree $m$.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This paper completes the BGS20 conjecture by proving the matching lower bound on signature tail asymptotics for pure m-rough paths via a norming cyclic construction, and extends the result to arbitrary reasonable tensor norms.

read the letter

The core advance is the lower bound: for X_t = exp(t l) with top term l_m, they show limsup [(n/m)! ||π_n(exp l)||_n]^{m/n} equals ||l_m||_m under any reasonable tensor algebra norm. The upper bound and the conjecture itself came from BGS20; this work supplies the missing piece with an independent construction rather than algebraic rearrangement.

They achieve the lower bound by building, for each l_m, a finite-dimensional contractive development in which ||l_m||_m appears as an eigenvalue at degree m, then reading off the growth rate. This also identifies the tail with local m-variation. The extension to arbitrary reasonable norms is explicit and stated cleanly.

The construction is the load-bearing step. It must produce a contractive development that works uniformly across all reasonable norms and all top-level tensors; if the eigenvalue fails to control the limsup under the original norm for some choice, the equality does not follow. The abstract outlines the strategy but leaves the verification details to the body. No free parameters or invented entities appear in the statement.

This is narrow but cleanly executed work inside rough path theory. Readers already following signature asymptotics or BGS20 will find the completion useful; outsiders will not. The argument is self-contained once the cyclic construction is accepted, and the citation pattern is appropriate.

I would bring it to a reading group only if the group already works in this area. I would not cite it in my own papers. It deserves peer review because it resolves a stated conjecture with a concrete new device.

Referee Report

1 major / 2 minor

Summary. The manuscript proves Conjecture 2.12 of BGS20 on signature tail asymptotics for pure m-rough paths X_t = exp(t l) with l = l_1 + ⋯ + l_m. It extends the result to arbitrary reasonable tensor algebra norms on the tensor powers of V and establishes the matching lower bound, yielding limsup_{n→∞} [(n/m)! ‖π_n(exp l)‖_n]^{m/n} = ‖l_m‖_m. The upper bound is taken from BGS20; the new contribution is the lower bound obtained via finite-dimensional developments together with a norming cyclic construction that produces, for every top-level tensor l_m, a contractive development in which ‖l_m‖_m appears as an eigenvalue at degree m.

Significance. If the lower-bound construction is valid for every reasonable norm, the result supplies an exact identification between the signature tail and the local m-variation of a pure rough path. This is a precise asymptotic statement in rough-path theory. The paper receives credit for supplying an independent construction that avoids circularity with the upper bound and for extending the statement beyond the norms treated in BGS20.

major comments (1)

[Proof of the lower bound (norming cyclic construction)] The norming cyclic construction (proof of the lower bound): the manuscript must verify that, for an arbitrary reasonable tensor algebra norm and an arbitrary top-level tensor l_m, the constructed finite-dimensional development remains contractive and that the eigenvalue ‖l_m‖_m at degree m controls the limsup of the signature growth under the original norm. Without an explicit check that contractivity and eigenvalue control hold uniformly, the equality does not follow for the full class of norms claimed.

minor comments (2)

Clarify the precise definition of “reasonable tensor algebra norm” at the first appearance and confirm that every norm appearing in the construction satisfies the definition.
Add a short remark comparing the growth rate obtained from the cyclic construction with the growth rate of the original signature under the given norm.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the careful reading of the manuscript and for identifying the need for greater explicitness in the verification of the lower-bound construction. We address the major comment below and will revise the manuscript accordingly.

read point-by-point responses

Referee: [Proof of the lower bound (norming cyclic construction)] The norming cyclic construction (proof of the lower bound): the manuscript must verify that, for an arbitrary reasonable tensor algebra norm and an arbitrary top-level tensor l_m, the constructed finite-dimensional development remains contractive and that the eigenvalue ‖l_m‖_m at degree m controls the limsup of the signature growth under the original norm. Without an explicit check that contractivity and eigenvalue control hold uniformly, the equality does not follow for the full class of norms claimed.

Authors: The norming cyclic construction is carried out with respect to an arbitrary reasonable tensor algebra norm by selecting a finite-dimensional subspace of the tensor algebra and defining a cyclic development operator whose scaling is chosen so that the resulting map is contractive under that norm. The top-level component l_m is embedded so that the induced operator at degree m has eigenvalue exactly ||l_m||_m with respect to the given norm. Contractivity of the development then ensures that the growth rate of the signature coefficients in the developed path is at least the eigenvalue, and the lower bound transfers to the original norm because the development map is contractive (hence norm-nonincreasing on the relevant components). We agree, however, that an explicit auxiliary statement isolating this uniformity argument would remove any ambiguity. We will therefore add a short lemma in the revised manuscript that records the contractivity and eigenvalue control for general reasonable norms. revision: yes

Circularity Check

0 steps flagged

No circularity; lower bound via independent construction

full rationale

The paper cites BGS20 only for the upper bound and the original conjecture. The new contribution is the matching lower bound, obtained by an explicit finite-dimensional contractive development and norming cyclic construction that places ||l_m||_m as an eigenvalue at degree m. This construction is not obtained by rearranging the upper bound, by fitting parameters to the target quantity, or by any self-citation chain; it is an independent existence argument that works for arbitrary reasonable tensor algebra norms. No self-definitional, fitted-input, or ansatz-smuggling steps appear in the derivation.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The paper relies on standard background from rough path theory and the definition of reasonable tensor norms; no free parameters are fitted and no new entities are postulated.

axioms (2)

domain assumption Tensor powers of V admit reasonable tensor algebra norms under which the stated developments and norm equivalences hold.
Invoked to extend the result beyond the norms used in BGS20.
domain assumption Finite-dimensional developments of the pure rough path exist and can be equipped with the cyclic construction.
Central device for proving the lower bound.

pith-pipeline@v0.9.1-grok · 5751 in / 1307 out tokens · 55378 ms · 2026-06-30T05:24:53.042671+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

17 extracted references

[1]

Boedihardjo and X

H. Boedihardjo and X. Geng, Tail asymptotics of the Brownian signature, Trans. Am. Math. Soc.,372(2019), 585–614. 2

2019
[2]

Boedihardjo and X

H. Boedihardjo and X. Geng, SL 2(R)-developments and signature asymptotics for planar paths with bounded variation, Rev. Mat. Iberoam.,39(5)(2023), 1973–2006. 2

2023
[3]

Boedihardjo, X

H. Boedihardjo, X. Geng, T. Lyons and D. Yang, The signature of a rough path: uniqueness, Adv. Math.,293 (2016), 720–737. 2

2016
[4]

Boedihardjo, X

H. Boedihardjo, X. Geng and N. P. Souris, Path developments and tail asymptotics of signature for pure rough paths, Adv. Math.,364(2020), 107043. 1, 2, 3, 5, 6, 9

2020
[5]

Chen, Iterated integrals of differential forms and loop space homology, Ann

K.-T. Chen, Iterated integrals of differential forms and loop space homology, Ann. of Math.,97(1973), 217–

1973
[6]

Chang, Effective algorithms for inverting the signature of a path, Ph.D

J. Chang, Effective algorithms for inverting the signature of a path, Ph.D. Thesis, Oxford, 2018. 2

2018
[7]

Chang, T

J. Chang, T. Lyons and H. Ni, Super-multiplicativity and a lower bound for the decay of the signature of a path of finite length, C. R. Acad. Sci. Paris,356 (1)(2018), 720–724. 2

2018
[8]

Friz and M

P. Friz and M. Hairer, A course on rough paths, Springer, 2020. 1

2020
[9]

Friz and N

P. Friz and N. Victoir, Multidimensional Stochastic Processes as Rough Paths, Cambridge University Press,
[10]

Geng, Reconstruction for the signature of a rough path, Proc

X. Geng, Reconstruction for the signature of a rough path, Proc. Lond. Math. Soc.,114 (3)(2017), 495–526. 2

2017
[11]

Gubinelli, Controlling rough paths, J

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

2004
[12]

Gubinelli, Ramifications of rough paths, J

M. Gubinelli, Ramifications of rough paths, J. Differential Equations,248 (4)(2010), 693–721. 1

2010
[13]

Hambly and T

B. Hambly and T. Lyons, Uniqueness for the signature of a path of bounded variation and the reduced path group, Ann. Math.,171(1)(2010), 109–167. 2

2010
[14]

Lyons, Differential equations driven by rough signals, Rev

T. Lyons, Differential equations driven by rough signals, Rev. Mat. Iberoam.,14 (2)(1998), 215–310. 1, 2, 4

1998
[15]

Lyons and Z

T. Lyons and Z. Qian, System Control and Rough Paths, Oxford University Press, 2002. 1

2002
[16]

Lyons and W

T. Lyons and W. Xu, Hyperbolic development and the inversion of signature, J. Funct. Anal.,272 (7)(2017), 2933–2955. 2

2017
[17]

Lyons and W

T. Lyons and W. Xu, Inverting the signature of a path, J. Eur. Math. Soc.,20 (7)(2018) 1655–1687. 2 School ofMathematics andStatistics, LanzhouUniversityLanzhou, 730000, China Email address:linn2024@lzu.edu.cn School ofMathematics andStatistics, LanzhouUniversityLanzhou, 730000, China; GansuProvincialRe- searchCenter forBasicDisciplines ofMathematics andS...

2018

[1] [1]

Boedihardjo and X

H. Boedihardjo and X. Geng, Tail asymptotics of the Brownian signature, Trans. Am. Math. Soc.,372(2019), 585–614. 2

2019

[2] [2]

Boedihardjo and X

H. Boedihardjo and X. Geng, SL 2(R)-developments and signature asymptotics for planar paths with bounded variation, Rev. Mat. Iberoam.,39(5)(2023), 1973–2006. 2

2023

[3] [3]

Boedihardjo, X

H. Boedihardjo, X. Geng, T. Lyons and D. Yang, The signature of a rough path: uniqueness, Adv. Math.,293 (2016), 720–737. 2

2016

[4] [4]

Boedihardjo, X

H. Boedihardjo, X. Geng and N. P. Souris, Path developments and tail asymptotics of signature for pure rough paths, Adv. Math.,364(2020), 107043. 1, 2, 3, 5, 6, 9

2020

[5] [5]

Chen, Iterated integrals of differential forms and loop space homology, Ann

K.-T. Chen, Iterated integrals of differential forms and loop space homology, Ann. of Math.,97(1973), 217–

1973

[6] [6]

Chang, Effective algorithms for inverting the signature of a path, Ph.D

J. Chang, Effective algorithms for inverting the signature of a path, Ph.D. Thesis, Oxford, 2018. 2

2018

[7] [7]

Chang, T

J. Chang, T. Lyons and H. Ni, Super-multiplicativity and a lower bound for the decay of the signature of a path of finite length, C. R. Acad. Sci. Paris,356 (1)(2018), 720–724. 2

2018

[8] [8]

Friz and M

P. Friz and M. Hairer, A course on rough paths, Springer, 2020. 1

2020

[9] [9]

Friz and N

P. Friz and N. Victoir, Multidimensional Stochastic Processes as Rough Paths, Cambridge University Press,

[10] [10]

Geng, Reconstruction for the signature of a rough path, Proc

X. Geng, Reconstruction for the signature of a rough path, Proc. Lond. Math. Soc.,114 (3)(2017), 495–526. 2

2017

[11] [11]

Gubinelli, Controlling rough paths, J

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

2004

[12] [12]

Gubinelli, Ramifications of rough paths, J

M. Gubinelli, Ramifications of rough paths, J. Differential Equations,248 (4)(2010), 693–721. 1

2010

[13] [13]

Hambly and T

B. Hambly and T. Lyons, Uniqueness for the signature of a path of bounded variation and the reduced path group, Ann. Math.,171(1)(2010), 109–167. 2

2010

[14] [14]

Lyons, Differential equations driven by rough signals, Rev

T. Lyons, Differential equations driven by rough signals, Rev. Mat. Iberoam.,14 (2)(1998), 215–310. 1, 2, 4

1998

[15] [15]

Lyons and Z

T. Lyons and Z. Qian, System Control and Rough Paths, Oxford University Press, 2002. 1

2002

[16] [16]

Lyons and W

T. Lyons and W. Xu, Hyperbolic development and the inversion of signature, J. Funct. Anal.,272 (7)(2017), 2933–2955. 2

2017

[17] [17]

Lyons and W

T. Lyons and W. Xu, Inverting the signature of a path, J. Eur. Math. Soc.,20 (7)(2018) 1655–1687. 2 School ofMathematics andStatistics, LanzhouUniversityLanzhou, 730000, China Email address:linn2024@lzu.edu.cn School ofMathematics andStatistics, LanzhouUniversityLanzhou, 730000, China; GansuProvincialRe- searchCenter forBasicDisciplines ofMathematics andS...

2018