Weighted Besov spaces on Heisenberg groups and applications to the Parabolic Anderson model

Che-Hung Huang; Cheng Ouyang; Fabrice Baudoin; Jing Wang; Li Chen; Samy Tindel

REVIEW 1 major objections 5 minor 2 cited by

Weighted Besov spaces on Heisenberg groups and applications to the Parabolic Anderson model

T0 review · 1 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 Unique solution found for parabolic Anderson on Heisenberg groups

desk verdict Main theorem likely true, but Corollary 3.25 is a genuine missing proof: product localization is radial-only and the transfer to Besov blocks is not shown. read the letter →

arxiv 2501.04593 v1 pith:YFCCYB4S submitted 2025-01-08 math.PR math.FA

Fabrice Baudoin , Li Chen , Che-Hung Huang , Cheng Ouyang , Samy Tindel , Jing Wang This is my paper · ORCID

classification math.PRmath.FA MSC 60H1535R6043A8046E35

keywords parabolicAndersonmodelHeisenberggroupweightedBesovspacesprojectiveFouriertransformparaproductStratonovich/Youngintegrationsub-LaplacianGaussiannoise

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's aim is to give a rigorous pathwise (Stratonovich) meaning to the parabolic Anderson model on the Heisenberg group $\mathbf{H}_n$ and to determine exactly for which noises a unique solution exists. The noise is smoother than white noise in time, with spatial covariance given by a negative power of the sub-Laplacian, and the solution lives in a new family of weighted Besov spaces tailored to the group. These spaces, defined through a projective Fourier transform whose frequency variable is the pair $(m,\ell,\lambda)$ of Hermite-mode indices and a real weight, are developed in detail and are claimed to be new and of independent interest. The payoff is an explicit condition on the noise parameters under which existence and uniqueness hold.

What carries the argument

The load-bearing object is the weighted Besov scale $B^{\gamma,\nu}_{\alpha,\beta}(\mathbf{H}_n)$, defined by frequency blocks $\sigma_k f$ whose projective Fourier transform is supported on the diagonal $\{m=\ell\}$ with frequency $|\lambda|(2|m|+n)$ comparable to $2^k$; the exponential weight $e^{-\nu|q|_*^\eta}$ is what lets the method handle spatially unbounded noises. The projective Fourier transform itself — Fourier modes indexed by Hermite indices $(m,\ell)$ and $\lambda\in\mathbb{R}^*$, with multiplication turned into a discrete convolution — supplies the analogue of Euclidean frequency localization. The argument then runs on two rails: Bernstein-type estimates and heat-semigroup smoothing in these weighted spaces, and a paraproduct continuity theorem whose key input is a frequency-localization statement for products of two localized functions, imported from an existing paraproduct on $\mathbf{H}_n$ and transferred through radial test functions.

What would settle it

Take a non-radial Schwartz pair on $\mathbf{H}_n$, for instance $f(q)=x_1 e^{-|q|_h^2}$ and $g(q)=e^{-|q|_h^2}$, and check whether the projective Fourier coefficients of $h=S_{k-1}f\cdot \sigma_k g$ vanish outside $(2|m|+n)^{-1}2^k$ times the annulus $C_0'$ asserted in Corollary 3.25; a single pair with support leaking outside that annulus would invalidate Lemma 3.29 and the fixed-point contraction.

Watch

Extended reading notes

Core claim

The central discovery is that the parabolic Anderson model $\partial_t u_t = \tfrac12 \Delta u_t + u_t \dot W_{\zeta,\alpha}$ on $\mathbf{H}_n$ admits a unique mild solution, in the Young/Stratonovich sense, whenever the noise has temporal exponent $\zeta\in(0,1)$ and spatial regularity $\alpha$ satisfying $\frac{n+1}{2}-(1-\zeta)<\alpha<\frac{n+1}{2}$. The noise $\dot W_{\zeta,\alpha}$ is the centered Gaussian field whose time covariance is $|t-s|^{-\zeta}$ and whose space covariance is the resolvent kernel $G_{2\alpha}$ of $(-\Delta)^{-2\alpha}$; the paper proves that such a noise falls into the Hölder-in-time distribution class $\mathcal{C}^{\vartheta,-\gamma,\rho_b}_{\infty,\infty}$ required by its general fixed-point theorem. Along the way it constructs weighted Besov spaces $B^{\gamma,\nu}_{\alpha,\beta}$ on $\mathbf{H}_n$ via Littlewood–Paley blocks built from Gevrey partitions of unity in the projective Fourier picture, and proves the Bernstein lemma, heat-flow smoothing, and paraproduct bounds needed for the contraction argument.

Load-bearing premise

The load-bearing step is the assertion, via Corollary 3.25, that the product of two frequency-localized functions—not merely two radial ones—remains supported in the expected projective frequency annulus or ball; the paper imports this from the radial case without giving a fully detailed proof for arbitrary Littlewood–Paley blocks.

Editorial extensions

If this is right

In the white-noise-in-time limit $\zeta\to 1$, the interval in condition (1.15) collapses; the model is solvable only if the spatial noise is a function rather than a distribution, matching known Stratonovich phenomena.
With $Q=2n+2$, condition (1.15) takes the shape $\alpha>Q/4-(1-\zeta)$, the analogue of the Euclidean Bessel-kernel condition $d/4-(1-\zeta)$.
The weighted Besov spaces and their paraproduct give a setting for other semilinear SPDEs on noncompact sub-Riemannian manifolds with spatially unbounded distributions.
Because the time integral is pathwise Young/Stratonovich, the solution map supports the polymer-type measures that motivate the paper.

Reading between the lines

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

If the block-localization step of Corollary 3.25 is genuinely valid for arbitrary blocks, the same paraproduct machinery should transfer to any homogeneous Lie group with a projective Fourier calculus, giving analogous exponent windows with the group's homogeneous dimension in place of $Q=2n+2$.
The paper stops at existence and uniqueness; a natural next test would be whether the boundary $\alpha=(n+1)/2-(1-\zeta)$ also governs moment growth and intermittency exponents.
Below the Young threshold one would expect regularity structures or paracontrolled calculus on $\mathbf{H}_n$ to extend the result; the weighted Besov scale constructed here would be the natural base scale.
The radial-to-arbitrary transfer in Corollary 3.25 could be checked on $n=1$ by explicit Hermite-coefficient computation; a counterexample would require a different localization argument for the paraproduct.

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

Colleague,

Here's my take on Baudoin-Chen-Huang-Ouyang-Tindel-Wang, arXiv:2501.04593. The paper builds a weighted Besov calculus on the Heisenberg group via the projective Fourier transform and uses it to solve a pathwise (Stratonovich-Young) parabolic Anderson model under the exponent condition α > (n+1)/2 − (1−ζ). The Besov machinery is a real contribution: new weighted spaces, heat-flow smoothing, Bernstein inequalities, and a noise embedding that uses no fitted parameters. The exponent threshold matching the flat-space Bessel-kernel result with Hausdorff dimension Q=2n+2 is a nice sanity check, and the comparison with the earlier Itô paper is honest.

I agree with the reader's conditional verdict, but I want to put the stress-test concern right at the center. Corollary 3.25 is not justified. Proposition 3.23 proves product localization only for radial functions, using the radial diagonalization of the projective Fourier transform. Corollary 3.25 asserts the same for arbitrary Besov blocks by "direct application." That is a non-sequitur: the blocks are not radial, their projective Fourier transforms are off-diagonal, and the pointwise product on H_n is not the matrix product in (2.17)-(2.18). The off-diagonal terms are exactly what [4] controls only for radial inputs. This step feeds Lemma 3.29, Proposition 3.11, and the fixed-point argument in Lemma 4.6, so it is load-bearing. A referee should demand a proof of Corollary 3.25 for non-radial blocks, or a precise restatement of the support conditions that avoids the radial assumption.

There is also a minor sign error in the proof of Theorem 4.16: the compatibility inequality is displayed with the wrong direction, although the final condition (1.15) is correct. That is easy to fix. Some estimates in the Young-integration section are left to the reader, but the structure is standard.

The paper is serious and not circular: no fitted parameters, no conclusion-as-input. Self-citation is contextual. If the paraproduct gap is repaired, the theorem is likely to stand.

Bottom line: this deserves a serious referee, but with a specific mandate to verify Corollary 3.25. I would not cite the paraproduct estimates in their current form.

Referee Report

1 major / 5 minor

Summary. The paper develops weighted Besov spaces on Heisenberg groups H_n using the projective Fourier transform of Bahouri--Chemin--Danchin, proves Bernstein-type inequalities, heat-flow smoothing estimates, and a paraproduct calculus in these spaces, and then applies the framework to the pathwise (Stratonovich--Young) parabolic Anderson model driven by a Gaussian noise with time covariance |t-s|^{-\zeta} and spatial covariance given by the kernel G_{2\alpha}. The main existence-uniqueness result is Theorem 4.7 together with Theorem 4.16, which identifies the admissible range (n+1)/2-(1-\zeta)<\alpha<(n+1)/2 for the noise parameters. The authors emphasize that the weighted Besov space construction is new and of independent interest.

Significance. If the technical gaps are repaired, the paper makes a substantial contribution: it gives a genuinely new weighted Besov-space framework on Heisenberg groups that is compatible with heat-flow estimates, and it provides a pathwise well-posedness result for the parabolic Anderson model whose exponent condition matches the Hausdorff dimension Q=2n+2, in line with the authors' earlier It\^o-setting results. The paper contains many explicit and checkable estimates: the Gevrey-class construction of Littlewood--Paley blocks, the scaling arguments in Section 2.4, the heat-flow smoothing in Proposition 3.8, and the variance computation in Lemma 4.15 are all concrete and mostly coherent. The main unresolved point is a single but load-bearing localization step in the paraproduct proof, and the paper would be acceptable after that step is supplied.

major comments (1)

[§3.3.3, Corollary 3.25] Corollary 3.25 is asserted as a "direct application" of Proposition 3.23, but Proposition 3.23 is proved only for radial functions f,g; its proof uses the diagonalization (3.55) and imports [4, Proposition 4.2]. The blocks S_{k-1}f and \sigma_k g are not radial for general f, and the projective convolution identity (2.17)--(2.18) couples all indices j\in\mathbb{N}^n, so the claimed support localization of \widehat{h^S_k}(m,\ell,\lambda) and \widehat{h^{\sigma,\varepsilon}_k}(m,\ell,\lambda) does not follow from Proposition 3.23 without an additional argument. This step is load-bearing: Lemma 3.29 invokes Corollary 3.25 to apply Lemmas 3.27--3.28, Proposition 3.11 then uses Lemma 3.29, and Lemma 4.6 and Theorem 4.7 rely on Proposition 3.11. Please provide a complete proof of the non-radial frequency localization, or state and prove the needed non-radial version of [4, Proposition 4.2] in the projective Fourier setting.

minor comments (5)

[§4.3, proof of Theorem 4.16] The displayed compatibility inequality "2-\zeta<1+(n+1)/2-\alpha" has the wrong direction: the argument requires 1+(n+1)/2-\alpha<2-\zeta, which is equivalent to the stated condition \alpha>(n+1)/2-(1-\zeta). The final condition (1.15)/(4.66) is correct, but the displayed line should be corrected.
[§1.1 and §1.2, versus §4.3] The abstract and introduction say the spatial covariance is generated by negative powers (-\Delta)^{-\alpha}, while Definition 4.8 and equation (1.14) use G_{2\alpha}, i.e. (-\Delta)^{-2\alpha}. Please harmonize the wording so that the exponent convention is unambiguous.
[§1.1, (1.11), and §3.2, Proposition 3.6] The heat semigroup P_t is defined in (1.11) as the semigroup generated by \Delta, but Proposition 3.6 calls it e^{t\Delta/2}. Since the main SPDE (1.13) contains the factor 1/2, this normalization should be fixed consistently (for instance by defining p_t for the operator \Delta/2 or by writing the mild solution with P_{(t-s)/2}). This is a notational issue and does not change the estimates, because the constants can absorb a factor 2.
[§3.3, Definition 3.9, and §4.1, Definition 4.2] The weight \rho_b is defined inconsistently: Definition 3.9 sets \rho_b(q)=|q|_*^{-b} with b\in\mathbb{R}, while Definition 4.2 and Theorem 4.16 use \rho_b(q)=c(1+|q|_*^b) with b>0. The integrability condition in Lemma 4.15 depends on which convention is used. Please align the notation and state the exact class of polynomial weights used in Hypothesis 4.5.
[§4.2, Theorem 4.7] Theorem 4.7 states existence and uniqueness in D^{\theta,\kappa,\nu,b}_{\alpha,\beta} for arbitrary exponents satisfying \theta+\vartheta>1 and \gamma<\kappa<1, but Lemma 4.6 proves contraction only for the particular pair constructed with \theta=1-\vartheta+\varepsilon and \kappa=\gamma+2\varepsilon. Please either prove the more general statement or state Theorem 4.7 with the specific pair produced in Lemma 4.6, since the application in Theorem 4.16 only needs the existence of some such pair.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the main existence theorem is obtained from noise block-variance estimates plus a fixed-point argument, not from the target result.

full rationale

The central derivation chain is self-contained with respect to its conclusion. Theorem 4.7 follows from the contraction property of the fixed-point map in Lemma 4.6, whose estimates use only heat-flow smoothing (Proposition 3.8), product estimates (Proposition 3.11 and Lemma 3.29), and the noise regularity encoded in Hypothesis 4.5. The verification that the Gaussian family V_{zeta,alpha} satisfies this hypothesis is performed in Lemma 4.15 by directly computing block variances from the covariance structure (4.37) and the projective Fourier diagonalization (2.13)-(2.16); no fitted parameter is introduced and the target regularity is not assumed. Citations to [3], [4] and the authors' [6] import Fourier/paraproduct machinery and facts about the noise in the Ito setting; these are external inputs, not conclusions equivalent to Theorem 4.16. The only flagged weakness is Corollary 3.25: Proposition 3.23 is proved for radial functions, and the extension to arbitrary Besov blocks is asserted by 'direct application', which is a proof gap rather than circularity. There is also a '<' versus '>' typo in the displayed compatibility line of the proof of Theorem 4.16, although the final condition (4.66) is correct. Neither issue makes the derivation reduce to its own inputs.

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

The paper introduces no fitted parameters and no new physical entities. Its functional-analytic objects are definitions, and its main proof depends on substantial borrowed machinery from [3], [4], and [6]. The ledger records those external building blocks as domain assumptions.

assumptions (5)

domain assumption The projective Fourier transform from [3] is a bi-continuous isomorphism between S(H_n) and S(tilde H_n), with inversion (2.11), Plancherel (2.12), and the duality identities M_2 f = -hatDelta hat f, M_0 f = hatD_lambda hat f.
Invoked in Definition 2.6, Proposition 2.8 and Lemma 2.9, all borrowed from [3] rather than proved in this paper. The entire Besov construction depends on these facts.
domain assumption Bahouri-Gallagher Proposition 4.2 supplies the frequency localization of products of functions with Fourier support in suitable sets in the tilde-F Fourier representation.
Imported in Proposition 3.23 and Remark 3.24; the paper does not reproduce those computations. Corollary 3.25 and Lemma 3.29 rest on this localization.
domain assumption A centered Gaussian family W_{zeta,alpha} with covariance (4.37)-(4.38) exists, with spatial component identified as the Sobolev space W^{-alpha,2}.
Definition 4.8 refers to the authors' previous paper [6, Section 3] and to [13] for the time structure. The variance estimates in Lemma 4.15 start from this noise model.
standard math Gaveau's explicit heat kernel formula (1.9) and the diagonalization of the sub-Laplacian under the projective Fourier transform hold.
Used in Sections 1.1 and 2 to obtain (2.13)-(2.16) and heat-kernel estimates; these are classical results in sub-Riemannian analysis.
standard math Standard Young inequality on locally compact groups and existence of compactly supported Gevrey cutoff functions equal to 1 near the origin.
Used in Proposition 2.15 and Lemma 2.20; the paper cites [2, Lemma 1.4] and [14].

how reviews work

0 comments

Cite this review

Pith. "Pith review of Weighted Besov spaces on Heisenberg groups and applications to the Parabolic Anderson model." pith.science (2026). https://pith.science/paper/YFCCYB4S

@misc{pith2026250104593,
  author       = {Pith},
  title        = {Pith review of: Weighted Besov spaces on Heisenberg groups and applications to the Parabolic Anderson model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YFCCYB4S}},
  note         = {Machine review of arXiv:2501.04593}
}

abstract

This article aims at a proper definition and resolution of the parabolic Anderson model on Heisenberg groups $\mathbf{H}_{n}$. This stochastic PDE is understood in a pathwise (Stratonovich) sense. We consider a noise which is smoother than white noise in time, with a spatial covariance function generated by negative powers $(-\Delta)^{-\alpha}$ of the sub-Laplacian on $\mathbf{H}_{n}$. We give optimal conditions on the covariance function so that the stochastic PDE is solvable. A large portion of the article is dedicated to a detailed definition of weighted Besov spaces on $\mathbf{H}_{n}$. This definition, related paraproducts and heat flow smoothing properties, forms a necessary step in the resolution of our main equation. It also appears to be new and of independent interest. It relies on a recent approach, called projective, to Fourier transforms on $\mathbf{H}_{n}$.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Weighted Besov Spaces on Homogeneous Lie Groups and Applications to Parabolic Anderson Models
math.AP 2026-08 reject novelty 7.0 of 10

A unified weighted Besov theory on homogeneous Lie groups with a multiscale characterization is developed and applied to prove well-posedness of parabolic Anderson models in the Young and first singular regimes.
Renormalised Models for Variable Coefficient Singular SPDEs
math.AP 2025-07 conditional novelty 7.0 of 10

The paper proves convergence of renormalised models in regularity structures for variable coefficient singular SPDEs across full subcritical regimes, with renormalisation functions depending only on a finite jet of th...

Reference graph

Works this paper leans on

17 extracted references · 17 canonical work pages · cited by 2 Pith papers

[4]

Bahouri, and I

H. Bahouri, and I. Gallagher. Paraproduit sur le groupe d e Heisenberg et applications. Revista Matem- atica Iberoamericana 17.1 (2001): 69-105

work page 2001
[1]

Bahouri, Hajer, D

H. Bahouri, Hajer, D. Barilari, and I. Gallagher. Strich artz estimates and Fourier restriction theorems on the Heisenberg group. Journal of Fourier Analysis and App lications 27.2 (2021): 1-41

work page 2021
[2]

Bahouri, J.-Y

H. Bahouri, J.-Y. Chemin, and R. Danchin, Fourier analysis and nonlinear partial diﬀerential equation s, Grundlehren der Mathematischen Wissenschaften 343, Springer (2011)

work page 2011
[3]

Bahouri, J.-Y

H. Bahouri, J.-Y. Chemin, and R. Danchin, Tempered distr ibutions and Fourier transform on the Heisenberg group, Ann. Henri Lebesgue 1 (2018) 1–46

work page 2018
[5]

Baudoin, C

F. Baudoin, C. Ouyang, L. Chen, C-H. Huang, S. Tindel, and J. Wang. Parabolic Anderson model in bounded domains of recurrent metric measure spaces. Arxiv preprint, 2024

work page 2024
[6]

Baudoin, C

F. Baudoin, C. Ouyang, S. Tindel, and J. Wang. Parabolic A nderson model on Heisenberg groups: The Itô setting. Journal of Functional Analysis 285.1 (2023): 1 09920

work page 2023
[7]

X. Chen, A. Deya, J. Song, and S. Tindel. Hyperbolic Ander son Model 2: Strichartz Estimates and Stratonovich Setting. International Mathematics Researc h Notices, Vol. 2023, Iss. 21 (2023): 18575- 18628

work page 2023
[8]

Friz, and N

P. Friz, and N. Victoir. Multidimensional stochastic pr ocesses as rough paths: theory and applications. Vol. 120. Cambridge University Press, 2010

work page 2010

Show all 17 references

[9]

B. Gaveau. Principe de moindre action, propagation de la chaleur et estimées sous-elliptiques sur certains groupes nilpotents. Acta Math. 139 (1977) 95-153

1977
[10]

Y. Hu, J. Huang, D. Nualart, and S. Tindel, Stochastic he at equations with general multiplicative Gaussian noises: Hölder continuity and intermittency. Ele ctron. J. Probab. 20 (2015), no. 55, 50 pp

2015
[11]

Mayorcas and H

A. Mayorcas and H. Singh, Singular SPDEs on Homogeneous Lie Groups. Arxiv preprint, 2023

2023
[12]

Mourrat and H

J.-C. Mourrat and H. Weber, Global well-posedness of th e dynamic Φ 4 model in the plane, Ann. Probab. 45 (2017), no. 4, 2398-2476. 54 F. BAUDOIN, L. CHEN, C-H. HUANG, C. OUYANG, S. TINDEL, AND J . W ANG

2017
[13]

Pipiras and M

V. Pipiras and M. Taqqu, Integration questions related to fractional Brownian motion. Probab. Theory Related Fields 118 (2000), no. 2, 251-291

2000
[14]

Rodino, Linear partial diﬀerential operators in Gevrey spaces , World Scientiﬁc Publishing Co

L. Rodino, Linear partial diﬀerential operators in Gevrey spaces , World Scientiﬁc Publishing Co. (1993)

1993
[15]

V. S. Rychkov, Littlewood–Paley theory and function sp aces with Ap loc weights, Math. Nachr. 224, no. 1 (2001)

2001
[16]

Strasburger, and D

A. Strasburger, and D. Dziewa-Dawidczyk. Hermite poly nomials and representations of the unitary group. Journal of Physics: Conference Series. Vol. 597. No. 1. IOP Publishing, 2015

2015
[17]

Tessitore, and J

G. Tessitore, and J. Zabczyk. Wong-Zakai approximatio ns of stochastic evolution equations. Journal of Evolution Equations 6.4 (2006): 621-655. F abrice Baudoin: Department of Mathematics, Aarhus Universi ty, Denmark. Email address : fbaudoin@math.au.dk Li Chen: Department of ...

2006

Pith tools

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

[1] [4]

Bahouri, and I

H. Bahouri, and I. Gallagher. Paraproduit sur le groupe d e Heisenberg et applications. Revista Matem- atica Iberoamericana 17.1 (2001): 69-105

work page 2001

[2] [1]

Bahouri, Hajer, D

H. Bahouri, Hajer, D. Barilari, and I. Gallagher. Strich artz estimates and Fourier restriction theorems on the Heisenberg group. Journal of Fourier Analysis and App lications 27.2 (2021): 1-41

work page 2021

[3] [2]

Bahouri, J.-Y

H. Bahouri, J.-Y. Chemin, and R. Danchin, Fourier analysis and nonlinear partial diﬀerential equation s, Grundlehren der Mathematischen Wissenschaften 343, Springer (2011)

work page 2011

[4] [3]

Bahouri, J.-Y

H. Bahouri, J.-Y. Chemin, and R. Danchin, Tempered distr ibutions and Fourier transform on the Heisenberg group, Ann. Henri Lebesgue 1 (2018) 1–46

work page 2018

[5] [5]

Baudoin, C

F. Baudoin, C. Ouyang, L. Chen, C-H. Huang, S. Tindel, and J. Wang. Parabolic Anderson model in bounded domains of recurrent metric measure spaces. Arxiv preprint, 2024

work page 2024

[6] [6]

Baudoin, C

F. Baudoin, C. Ouyang, S. Tindel, and J. Wang. Parabolic A nderson model on Heisenberg groups: The Itô setting. Journal of Functional Analysis 285.1 (2023): 1 09920

work page 2023

[7] [7]

X. Chen, A. Deya, J. Song, and S. Tindel. Hyperbolic Ander son Model 2: Strichartz Estimates and Stratonovich Setting. International Mathematics Researc h Notices, Vol. 2023, Iss. 21 (2023): 18575- 18628

work page 2023

[8] [8]

Friz, and N

P. Friz, and N. Victoir. Multidimensional stochastic pr ocesses as rough paths: theory and applications. Vol. 120. Cambridge University Press, 2010

work page 2010