Pith. sign in

REVIEW 2 major objections 4 minor 15 references

The Fourier transform in variable exponent Lebesgue spaces

T0 review · 2 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A Fourier transform is defined for every function in a variable-exponent Lebesgue space, and the space of transforms is shown to be a Banach space isometrically isomorphic to the original space, with norm inversion and an exchange theorem.

desk verdict Solid isometry construction for variable-exponent Fourier transforms, but the headline inversion theorem is false as stated for exponents that take the value infinity on a positive-measure set. read the letter →

arxiv 2506.08891 v1 pith:KVAGQV3E submitted 2025-06-10 math.CA

André Pedroso Kowacs , Wagner Augusto Almeida de Moraes This is my paper
classification math.CA MSC 42A3846E3026A4246B04
keywords variableexponentLebesguespaceFouriertransformtempereddistributioncontinuousprimitiveintegralBanachlog-Höldercontinuitydistributionalderivativeisometricisomorphism
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 extends the Fourier transform to variable-exponent Lebesgue spaces $L^{p(\cdot)}(\mathbb{R})$, where the exponent $p(\cdot)$ can vary over $[1,\infty]$ rather than being a fixed number. For each $f \in L^{p(\cdot)}(\mathbb{R})$ it constructs the auxiliary function $\Psi_f(s)=\int_{\mathbb{R}} \frac{1-e^{-ist}}{it} f(t)\,dt$ and defines the Fourier transform as the distributional derivative $\hat{f}=\Psi'_f$. The main theorem shows that, whenever $p_+<\infty$ or $1/p(\cdot)$ is log-Hölder continuous at infinity with $p_\infty=1$, the spaces of these transforms are Banach spaces isometrically isomorphic to $L^{p(\cdot)}(\mathbb{R})$ with the inherited norm. The paper goes on to prove that the transform agrees with the classical Fourier transform on tempered distributions, that inversion holds in norm, and that an exchange formula extends Parseval-type identities. If the results are correct, every function in a variable-exponent Lebesgue space of this type has a genuine Fourier calculus, recovering the constant-exponent theory as the special case where $p(\cdot)$ is constant.

What carries the argument

The load-bearing object is the auxiliary function $\Psi_f(s)=\int_{\mathbb{R}} \frac{1-e^{-ist}}{it} f(t)\,dt$, shown to be Hölder continuous with exponent $1/p_+$ when $p_+<\infty$ and Lipschitz continuous when $1/p(\cdot)$ is log-Hölder at infinity with limit $1$. The proof of these estimates uses the generalized Hölder inequality for variable-exponent spaces and bounds the $L^{q(\cdot)}$ norm of the kernel $u_s(t)=(1-e^{-ist})/it$, producing constants that involve the integral $C_q$ of $|\sin y / y|^q$ and the Lambert $W$ function in the unbounded-exponent case. With $\Psi_f$ continuous, the Fourier transform is defined as the distributional derivative $\Psi'_f$, which allows the continuous primitive integral and Stieltjes integration to be used for exchange and inversion results. The isometry from $L^{p(\cdot)}$ to the transform space is what turns these estimates into a Banach space isomorphism.

What would settle it

Construct a specific $p(\cdot)$ that equals $\infty$ on a set of positive measure and satisfies the log-Hölder decay condition at infinity with $p_\infty=1$. For a function $g$ of bounded variation with $g(\pm\infty)=0$ and $\int |s|^{1/p_-}\,d|g|<\infty$, evaluate the functional $f \mapsto \int \hat{f}\,g$; if this functional is not given by integration against any $h \in L^{q(\cdot)}(\mathbb{R})$, the exchange theorem's proof step fails and the theorem would need repair for that exponent.

Watch

Extended reading notes

Core claim

The central discovery is that a single family of Hölder continuous primitives $\Psi_f$ carries the entire Fourier analysis of variable-exponent Lebesgue spaces. For exponents with finite essential supremum, or with $1/p(\cdot)$ log-Hölder continuous at infinity and $p_\infty=1$, the space $\{\Psi_f : f \in L^{p(\cdot)}(\mathbb{R})\}$ and its distributional derivative space $\{\Psi'_f : f \in L^{p(\cdot)}(\mathbb{R})\}$ are Banach spaces isometrically isomorphic to $L^{p(\cdot)}(\mathbb{R})$ under the norm $\|\Psi_f\| = \|f\|_{p(\cdot)}$. The transform defined this way coincides with the tempered-distribution Fourier transform, and the paper establishes an exchange formula $\int_{\mathbb{R}} \hat{f}\, g = \int_{\mathbb{R}} f\, \hat{g}$ for suitable $g$ and norm-convergent inversion with standard approximate identities. This gives a Fourier theory that works even when the exponent is unbounded, a regime where many tools of the constant-exponent theory fail.

Load-bearing premise

The proof of the exchange formula assumes that every bounded linear functional on $L^{p(\cdot)}(\mathbb{R})$ is integration against a function of the conjugate variable-exponent space $L^{q(\cdot)}(\mathbb{R})$, a duality result whose standard form requires $p_+<\infty$; the paper applies it to exponents that may take the value $\infty$ on a set of positive measure, where the dual space is larger.

Editorial extensions

If this is right

  • Every $f \in L^{p(\cdot)}(\mathbb{R})$ with $p_+<\infty$ or with $1/p(\cdot)$ log-Hölder continuous at infinity and $p_\infty=1$ has a generalized Fourier transform $\hat{f}=\Psi'_f$, and the map $f \mapsto \hat{f}$ is an isometric isomorphism onto a Banach space.
  • The generalized transform agrees with the classical Fourier transform on tempered distributions, so it inherits standard identities for translations, modulations, and differentiation.
  • For the Cesàro–Fejér, Abel–Poisson, and Gauss–Weierstrass kernels, the approximation $I_a[f]$ converges to $f$ in $L^{p(\cdot)}$ norm as $a \to 0^+$.
  • The exchange formula $\int \hat{f}\, g = \int f\, \hat{g}$ holds under mild assumptions on $g$, yielding a variable-exponent analog of Parseval's identity.
  • When $p(\cdot)$ is constant, all results reduce to the known Fourier transform theory for $L^p$ spaces.

Reading between the lines

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

  • The isometric isomorphism suggests the transform could be iterated, but the image space $A^{p(\cdot)}$ is not shown to be closed under the transform, so a true Fourier inversion on the image is an open question rather than a corollary.
  • If the duality step in the exchange theorem fails for exponents with $p=\infty$ on positive measure, the inversion theorem may still be salvageable by adding a separate restriction on the set where the exponent is infinite.
  • The explicit constants involving $C_q$ and the Lambert $W$ function could be optimized to give sharp operator norms for the maps $f \mapsto \Psi_f$ in simple cases such as $p(x)=1+\kappa \ln(e+|x|)$.
  • The pointwise-kernel construction suggests a route to variable-exponent Fourier analysis on $\mathbb{R}^n$ by taking products of the one-dimensional kernel, though the distributional derivative step would need a higher-dimensional Hölder condition.
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.

Referee Report

2 major / 4 minor

Summary. The manuscript defines, for f in L^{p(·)}(R) with p+<∞ or 1/p(·) log-Hölder continuous at infinity with limit 1 (the LH1∞ class), an auxiliary function Ψ_f(s)=∫(1−e^{-ist})/(it) f(t)dt, proves Hölder or Lipschitz estimates for Ψ_f, and then defines the Fourier transform of f as the distributional derivative Ψ'_f. Theorem 4.1 equips the spaces B^{p(·)}(R) and A^{p(·)}(R) of such transforms with the norm ||Ψ_f||=||f||_{p(·)} and proves that they are isometrically isomorphic to L^{p(·)}(R). Section 5 shows consistency with the tempered-distribution Fourier transform and derives standard operational rules. Section 6 introduces continuous-primitive integrals of the transform and states an exchange theorem and a norm-inversion theorem. The main difficulty is in Section 6: the exchange and inversion results are claimed for the full LH1∞ class, but their proofs require p+<∞ at a load-bearing point, and the inversion theorem is false as stated for exponents that take the value ∞ on a set of positive measure.

Significance. The construction in Theorem 4.1 is a genuine and clean extension of Talvila's Fourier transform to variable exponent Lebesgue spaces, with an explicit norm and a careful injectivity argument via the heat kernel. The estimates in Section 3 are explicit and the reduction of the isometry to injectivity is well motivated. If the scope were restricted to exponents with p+<∞, the exchange and inversion results would be plausible and would give a useful new framework. As written, however, the paper advertises the LH1∞ class in the abstract and in Theorems 6.2 and 6.4, and in that class the inversion claim is false and the exchange proof has a duality gap. The central isometry theorem is not the problem; the Section 6 claims need correction before the paper can be accepted.

major comments (2)
  1. [Theorem 6.4] Theorem 6.4 is false for the LH1∞ class as stated. Let p(x)=∞ on [-1,1] and p(x)=1+1/log(e+|x|) for |x|>1; this exponent satisfies 1/p(·) log-Hölder continuity at infinity with p∞=1, so it belongs to the LH1∞ class. Take f=χ_{[0,1]} and let ψ be a standard Gaussian. Then ψ satisfies all the kernel hypotheses of the theorem: ψ∈L^1(R), ∫ψ=1, ψ^ is absolutely continuous, and ∫|s||ψ^(s)|ds and ∫|s||ψ^'(s)|ds are finite. The conclusion would require ||f∗ψ_a−f||_{p(·)}→0 as a→0+. But for x∈(0,a), f∗ψ_a(x)=Φ(x/a), the standard normal CDF, while f(x)=1, so ||f∗ψ_a−f||_{p(·)} ≥ esssup_{[-1,1]}|f∗ψ_a−f| = sup_{u∈(0,1)}(1−Φ(u)) = 1/2 for every a>0. Hence the claimed norm convergence fails. The proof's appeal to [3, Theorem 5.4] is an approximate-identity/norm-density result that is not valid when p+<∞ is not assumed, and the LH1∞ class explicitly allows p=∞ on a set of positive measure. The theorem, and the corresponding claim in the abstract, must be restricted, for example to p+<∞ or to exponents with |Ω∞|=0.
  2. [Theorem 6.2, final step] The proof of Theorem 6.2 concludes that bg∈L^{q(·)} by invoking [3, Proposition 2.79] after showing that f↦∫\hat f g is a bounded linear functional on L^{p(·)}. That duality identification requires p+<∞. In the LH1∞ case with p=∞ on a set of positive measure, the dual of L^{p(·)} is strictly larger than L^{q(·)}, so the boundedness of the functional does not imply bg∈L^{q(·)}. The exchange theorem is therefore unproved for the full stated class. This is load-bearing because Theorem 6.4 relies on the exchange formula to identify I_a[f] with f∗ψ_a; the LH1∞ version of the inversion theorem cannot inherit a valid proof from Theorem 6.2.
minor comments (4)
  1. [Definition 6.1(2)] The notation ∫_∞^∞ \hat f g is nonstandard and should be replaced by an explicitly defined improper integral over R, for example ∫_{−∞}^{∞} with the continuous-primitive convention explained in the preamble.
  2. [Theorem 6.2 proof] In equation (20) the expressions 'dg_s(s)' and 'dg_i(s)' appear to be typographical errors; the intended notation is dg_1(s) and dg_2(s).
  3. [Introduction] The claim that when restricted to constant exponents the results 'coincide precisely' with [12] should be qualified, since [12] treats 1≤p<∞ whereas the present paper also allows p=∞; the exact sense in which the constant-exponent case matches should be stated.
  4. [Example 5.5] In the first part of Example 5.5, the notation such as ∥\hat f∥_{p±}^{∞} is ambiguous; writing the relevant sup norms explicitly would improve readability.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the transform-space norm is defined as the source norm, and the substantive claims (injectivity, Banach-space structure, exchange/inversion) rest on external results, not on the paper's own conclusions.

full rationale

The paper defines the norm on B^{p(·)}(R) and A^{p(·)}(R) as ||Ψ_f|| = ||f||_{p(·)} (Section 4, Theorem 4.1). This makes the isometry a matter of definition, but the paper states this openly in the abstract ('A norm is defined ... so that it is isometrically isomorphic'), so it is not a hidden circular reduction. The nontrivial content of Theorem 4.1 is that the norm is well-defined, i.e. that Ψ_f = 0 implies f = 0, which is proved independently via Fubini and the heat kernel using the external pointwise convergence result [3, Theorem 5.8]. No parameters are fitted, and no prediction is inferred from a fitted quantity. The paper contains no self-citations: references [3], [12], [11], and others are external standard sources. The dependence on [3] for duality (Proposition 2.79) and approximate identities (Theorem 5.4) may be problematic for the LH1∞ class when p takes the value ∞ on a set of positive measure, since those external results typically require p+<∞; however, that is a correctness/applicability objection, not a circularity. The exchange and inversion theorems do not reduce to their conclusions by construction. Therefore no significant circularity is present.

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

No free parameters are fitted. The construction introduces the spaces B^{p(·)} and A^{p(·)} and the auxiliary function Psi_f, but these are mathematical objects, not unexplained entities. The main external input is the variable exponent Lebesgue space theory from [3] and the continuous primitive integral from [11,12]; the paper's contribution is the estimates and the isometric realization.

assumptions (5)
  • domain assumption Background theory of variable Lebesgue spaces from [3], including the modular definition, generalized Hölder inequality, and duality Proposition 2.79.
    Used throughout; the duality statement is load-bearing in Theorem 6.2 and is not valid for all exponents covered by the paper.
  • domain assumption Either p+<∞ or 1/p(·) is log-Hölder continuous at infinity with p∞=1 (LH1∞).
    These are the two hypotheses under which the growth estimates for Psi_f are proved.
  • standard math The continuous primitive integral theory of Talvila [11,12], including integration by parts for BV functions.
    Used to define integrals of the distribution-valued Fourier transform against BV functions.
  • standard math Classical heat kernel and approximate identity convergence for locally integrable functions.
    Used in Theorem 4.1 to conclude f=0 from vanishing convolutions.
  • standard math Fubini's theorem can be applied to the double integrals defining the distributional pairing of the Fourier transform.
    Justified in the paper by the estimates in equation (16).

how reviews work

0 comments
Cite this review

Pith. "Pith review of The Fourier transform in variable exponent Lebesgue spaces." pith.science (2026). https://pith.science/paper/KVAGQV3E

@misc{pith2026250608891,
  author       = {Pith},
  title        = {Pith review of: The Fourier transform in variable exponent Lebesgue spaces},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KVAGQV3E}},
  note         = {Machine review of arXiv:2506.08891}
}
abstract

In this work we define a Fourier transform for each $f\in L^{p(\cdot)}(\mathbb{R})$, for a large class of exponent functions $p(\cdot)$, as the distributional derivative of a H\"older continuous function. A norm is defined in the space of such Fourier transforms so that it is isometrically isomorphic to $L^{p(\cdot)}(\mathbb{R})$. We also prove several properties of this Fourier transform, such as inversion in norm and an exchange theorem.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

15 extracted references · 15 canonical work pages

  1. [1]

    and Vergara-Hermosilla, G

    Chamorro, D. and Vergara-Hermosilla, G. Lebesgue spaces with variable exponent: some applications to the Navier–Stokes equations,Positivity28(2024), no. 24

    work page 2024

  2. [2]

    Cruz-Uribe, D. V. and Convertito, G.The Stieltjes Integral, 1st ed., Chapman and Hall/CRC, 2023

    work page 2023

  3. [3]

    Cruz-Uribe, D. V. and Fiorenza, A.Variable Lebesgue Spaces: Foundations and Harmonic Analysis, Birkh¨ auser/Springer, Heidelberg, 2013

    work page 2013

  4. [4]

    Cruz-Uribe, D., Hern´ andez, E., and Martell, J. M. Greedy bases in variable Lebesgue spaces,Monatsh. Math.179(3) (2016), 355–378

    work page 2016

  5. [5]

    Maximal function on generalized Lebesgue spacesL p(·),Math

    Diening, L. Maximal function on generalized Lebesgue spacesL p(·),Math. Inequal. Appl.7(2004), no. 2, 245–253

    work page 2004

  6. [6]

    Riesz potential and Sobolev embeddings of generalized Lebesgue and Sobolev spacesL p(·) and W k,p(·),Math

    Diening, L. Riesz potential and Sobolev embeddings of generalized Lebesgue and Sobolev spacesL p(·) and W k,p(·),Math. Nachr.263(2004), no. 1, 31–43

    work page 2004

  7. [7]

    and Schwarzacher, S

    Diening, L. and Schwarzacher, S. On the Key Estimate for Variable Exponent Spaces,Azerbaijan J. Math. 3(2013), no. 2, 62-69

    work page 2013

  8. [8]

    Local-to-global results in variable exponent spaces,Math

    H¨ ast¨ o, P. Local-to-global results in variable exponent spaces,Math. Res. Lett.16(2009), no. 2, 263–278

    work page 2009

Show all 15 references
  1. [9]

    and R´ akosn ´ ık, J

    Kov´ aˇ cik, O. and R´ akosn ´ ık, J. On spacesLp(x) andW k,p(x),Czech. Math. J.41(116) (1991), 592–618

    1991

  2. [10]

    ¨Uber eine gewisse Klasse von R¨ aumen vom Typus B,Bull

    Orlicz, W. ¨Uber eine gewisse Klasse von R¨ aumen vom Typus B,Bull. Int. Acad. Pol. Sci. Ser. A8(1932), 207–220

    1932

  3. [11]

    The distributional Denjoy integral,Real Anal

    Talvila, E. The distributional Denjoy integral,Real Anal. Exchange33(2007), no. 1, 51–84

    2007

  4. [12]

    The Fourier Transform in Lebesgue spaces,Czech

    Talvila, E. The Fourier Transform in Lebesgue spaces,Czech. Math. J.75(2025), 179–191

    2025

  5. [13]

    Samko, S. G. Convolution and potential type operators inL p(x)(Rn),Integral Transforms and Special Func- tions7(3–4) (1998), 261–284

    1998

  6. [14]

    On variable Lebesgue spaces and generalized nonlinear heat equations,Preprint, arXiv:2404.09588, 2024

    Vergara-Hermosilla, G. On variable Lebesgue spaces and generalized nonlinear heat equations,Preprint, arXiv:2404.09588, 2024

    2024 arXiv

  7. [15]

    and Yuan, W

    Yang, S., Yang, D. and Yuan, W. New characterizations of Musielak-Orlicz-Sobolev spaces via sharp ball averaging functions,Front. Math. China14(2019), 177–201. Universidade Federal do Paran´a, Departamento de Matem´atica, C.P.19096, CEP 81531-990, Cu- ritiba, Brazil Email addr...

    2019

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.