Pith. sign in

REVIEW 5 minor 21 references

Characterizations of the Hardy space $\mathcal{H}_{FIO}^{1}(\mathbb{R}^{n})$ for Fourier integral operators

T0 review · 0 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read This paper proves that the Hardy space for Fourier integral operators on R^n is characterized by Littlewood–Paley g functions, maximal functions, and Lusin-type square functions, with equivalent norms.

desk verdict Solid endpoint result answering the p=1 Hardy-space question for FIOs, with a clean direct proof and only one load-bearing imported bound to watch. read the letter →

arxiv 1908.01448 v5 pith:6PT3QVA4 submitted 2019-08-05 math.AP

Zhijie Fan , Naijia Liu , Jan Rozendaal , Liang Song This is my paper · ORCID
classification math.AP MSC 42B3535S3042B30
keywords FourierintegraloperatorsHardyspacesLittlewood-PaleygfunctionmaximalcharacterizationwavepacketsconicalsquareparabolicfrequencylocalizationLusinarea
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 proves that $H^1_{FIO}(\mathbb{R}^n)$, the Hardy space adapted to Fourier integral operators, can be described by tools from classical harmonic analysis. Specifically, an $f$ lies in $H^1_{FIO}(\mathbb{R}^n)$ exactly when its low-frequency part $q(D)f$ is in $L^1$ and, for almost every direction $\omega$, the parabolically localized piece $\phi_\omega(D)f$ lies in the classical Hardy space $H^1(\mathbb{R}^n)$, with finite integral of those $H^1$ norms over the sphere. The same conclusion holds for the Littlewood–Paley $g$-function description, for a maximal-function description, and for the Lusin-type square function $G^*_\alpha$ when $\alpha>2$, all with equivalent norms. This answers an open question from earlier work on these spaces, and it matters because the new descriptions make $L^1$-based harmonic analysis tools available for studying Fourier integral operators and wave equations with rough coefficients.

What carries the argument

The machinery consists of wave packet projections $\theta_{\omega,\sigma}(D)$, $\chi_{\omega,\sigma}(D)$, and $\phi_\omega(D)$, where $\theta_{\omega,\sigma}$ localizes to frequencies of size $\sigma^{-1}$ inside a parabolic cone around the direction $\omega$. The hard inclusion is carried by a Peetre-type maximal function $M^*_\alpha$ and a lacunary decomposition identity. The load-bearing estimate is Corollary 2.3: the kernel of $\theta_{\omega,\sigma}(D)\chi_{\nu,\tau}(D)$ decays like $\min(\sigma/\tau,\tau/\sigma)^N$ for every $N$, which makes dyadic-lacunary sums converge. A geometric-convolution lemma (Lemma 3.4) converts power-type inequalities into $L^r$ estimates, Lemma 3.7 controls the resulting weighted sums, and the vector-valued Hardy–Littlewood maximal operator completes the argument.

What would settle it

A concrete way to test the central claim is to exhibit $f\in \mathcal{S}'(\mathbb{R}^n)$ with $q(D)f\in L^1$ and $G(f)\in L^1(S^*(\mathbb{R}^n))$ but with the conical square function $S(f)$ infinite; Theorem 3.8 says no such $f$ exists, so one would refute the main theorem. Alternatively, compute the kernel bound (2.12) explicitly for $n=2$ with $\sigma=2^{-k}$ and $\tau=2^{-l}$, and look for a failure of the $\min(\sigma/\tau,\tau/\sigma)^N$ decay, which would break the engine of the proof.

Watch

Extended reading notes

Core claim

The central discovery is that, for $p=1$, the Hardy space for Fourier integral operators $H^1_{FIO}(\mathbb{R}^n)$ coincides with $H^1_{FIO,G}(\mathbb{R}^n)$, $H^1_{FIO,\max}(\mathbb{R}^n)$, and $H^1_{FIO,G^*_\alpha}(\mathbb{R}^n)$ for $\alpha>2$, with equivalence of norms. Equivalently, an $f$ belongs to $H^1_{FIO}(\mathbb{R}^n)$ if and only if $q(D)f$ is in $L^1(\mathbb{R}^n)$, $\phi_\omega(D)f$ is in the classical Hardy space $H^1(\mathbb{R}^n)$ for almost every $\omega\in S^{n-1}$, and the integral over $S^{n-1}$ of $\|\phi_\omega(D)f\|_{H^1}\,d\omega$ is finite. The difficult inclusion $H^1_{FIO,G}\subseteq H^1_{FIO}$ is proved directly rather than by duality, using parabolic frequency localizations and a maximal function of Peetre type.

Load-bearing premise

The argument depends on an imported, unproved kernel estimate: when two wave-packet projections at frequency scales $\sigma$ and $\tau$ are composed, their kernel decays by a fixed power $\min(\sigma/\tau,\tau/\sigma)^N$ for every $N$. If that decay were weaker, the lacunary summation driving the main $L^1$ estimate would not converge, and the inclusions would not follow by this route.

Editorial extensions

If this is right

  • The conical square function in the definition of $H^1_{FIO}$ can be replaced by the simpler vertical Littlewood–Paley $g$ function, giving $H^1_{FIO}(\mathbb{R}^n)=H^1_{FIO,G}(\mathbb{R}^n)$ with comparable norms.
  • An $f$ belongs to $H^1_{FIO}(\mathbb{R}^n)$ precisely when its parabolically localized pieces $\phi_\omega(D)f$ are in $H^1(\mathbb{R}^n)$ for almost every direction and the average of their $H^1$ norms over the sphere is finite.
  • For $\alpha>2$, the Lusin-type square function $G^*_\alpha$ gives a norm-equivalent description, so changes of aperture are harmless in $H^1_{FIO}$.
  • Any bounded Fourier multiplier that maps the local Hardy space $H^1(\mathbb{R}^n)$ to $L^1(\mathbb{R}^n)$ is automatically bounded on $H^p_{FIO}(\mathbb{R}^n)$ for every $1\le p\le \infty$, extending earlier smooth-symbol results.
  • For functions with frequency support in a dyadic-parabolic region, the $H^p_{FIO}$ norm is equivalent to a shifted Sobolev norm, and the shift $s_p=\frac{n-1}{2}|\tfrac12-\tfrac1p|$ is optimal.

Reading between the lines

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

  • Because the characterization reduces $H^1_{FIO}$ to an average of classical $H^1$ norms of directionally localized pieces, many $L^1$-based harmonic analysis tools—maximal inequalities, Riesz transforms, interpolation—can be transplanted to Fourier integral operator problems without re-proving them for the conical square function.
  • A natural test is whether the endpoint $p=\infty$ admits a BMO description of the same shape; the paper proves only the easy inclusion there, so the converse is a plausible but unproved extension.
  • The same off-singularity decay that powers the proof suggests the characterizations should survive for more general phases or rough-coefficient wave equations, provided the kernel bound remains valid.
  • The explicit norm computation for wave packets makes these spaces amenable to numerical checks: one can approximate $\phi_\omega(D)f$ and compare the directional $H^1$ average against the conical square function on test functions.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

0 major / 5 minor

Summary. The paper studies the Hardy space H^1_FIO(R^n) introduced by Smith and by Hassell, Portal, and Rozendaal for Fourier integral operators. Its main result (Theorem 1.4) asserts that, for alpha > 2, H^1_FIO(R^n) coincides, with equivalent norms, with the Littlewood--Paley g-function space H^1_FIO,G(R^n), the maximal-function space H^1_FIO,max(R^n), and the Lusin-type square-function space H^1_FIO,G*_alpha(R^n). The proof of the hardest inclusion H^1_FIO,G subset H^1_FIO is carried out in Section 3 via a Peetre-type maximal function, a technical lemma of Rychkov type, off-singularity kernel bounds, and vector-valued Hardy--Littlewood maximal estimates. Section 4 gives the maximal-function characterization, Section 5 gives the G*_alpha characterization, and Section 6 applies these results to Fourier multiplier boundedness on H^p_FIO and to sharp norm comparisons for frequency-localized functions. The paper answers an open question posed in [15, Remark 4.3].

Significance. If correct, the result completes the p = 1 case of the characterization program started in [15] for 1 < p < infinity and provides practically useful descriptions of H^1_FIO in terms of more classical harmonic-analysis objects. The proof is nontrivial, and the main chain of estimates is internally consistent: I checked the support decomposition in Proposition 3.6, the application of Corollary 2.3, and the summation and maximal-function arguments in Theorem 3.8. The applications in Section 6 are reasonable first demonstrations of the utility of the new characterizations. The paper is also transparent about its dependence on the authors' earlier work for the off-singularity kernel bound; the reference is precise and the proof strategy is indicated, so I do not regard this as an internal gap.

minor comments (5)
  1. [Section 2.3, Corollary 2.3] The off-singularity bound (2.12) is the single most load-bearing estimate in the paper, since it drives the lacunary summation in Proposition 3.6, but its proof here is only a reference to [15, Proposition 3.6] and [12, Theorem 5.1]. I suggest including the full proof or at least a precise statement of the cited result, so that the present paper is self-contained at this key point; the current sentence "repeat the arguments" is acceptable as a citation but would benefit from a short explanation of the mechanism.
  2. [Section 3.2, Theorem 3.8] In the display following the choice of r and N, the exponent on 2^{-|j-l|} is written as N, but Proposition 3.6's proof produces the exponent (N-alpha)r before the final relabeling of N. This is not an error, but the notation should be clarified by using distinct constants (for example N_0 and N) to avoid confusing the reader.
  3. [Section 6.1, Theorem 6.1] The adjoint computation for m(D) is difficult to read in the current typesetting (the display appears as "m(D)^*g(x) = m(D)g(x) = m(D)\tilde g(-x)"). Please state explicitly the convention for the distributional duality pairing used in (1.2) and rewrite the identity in a way that is unambiguous about complex conjugation and the action on g.
  4. [Section 5, Theorem 5.2] The use of V(B_\sqrt{\sigma}(x,omega)) \simeq \sigma^n may surprise readers, since Lemma 2.1 states that balls of radius tau < 1 have volume of order tau^{2n}. A brief parenthetical noting that the radius here is \sqrt{\sigma} and hence the volume is (\sqrt{\sigma})^{2n} = \sigma^n would remove any ambiguity.
  5. [Throughout] The preprint typesetting contains several typographical artifacts, including "m aximal" in the abstract, "\greaterorsimilar" and "\nelement" in place of symbols, and assorted spacing issues. Please ensure the final published version is typeset cleanly.

Circularity Check

0 steps flagged · score 1.0 of 10

There is no significant circularity: Theorem 1.4 is proved by direct estimates, and the prior-work citations supply independent technical lemmas rather than the conclusion.

full rationale

The central equivalence is not produced by fitting or definitional identity. Theorem 1.4 is assembled from three independent inclusions: H1_FIO(R^n) ⊆ H1_FIO,G(R^n) is quoted from [4] (external vertical-versus-conical square function estimates); the hard inclusion H1_FIO,G(R^n) ⊆ H1_FIO(R^n) is proved directly in Theorem 3.8 via Proposition 3.6, which uses pointwise Peetre-type maximal inequalities, Rychkov's lemma from [16], the Hardy-Littlewood maximal operator on the doubling cosphere bundle, and the off-singularity kernel bound of Corollary 2.3. Corollary 2.3 is the only load-bearing input imported from the authors' own prior work ([15, Proposition 3.6] and [12, Theorem 5.1]); it is stated without proof here, but it is an independent kernel estimate derived by integration by parts from the wave-packet bounds in Lemma 2.2, and it does not assume the equality of spaces under investigation. The maximal-function and G*_alpha characterizations are then derived from Proposition 3.1 and Theorem 3.8 together with classical H1(R^n) characterizations (Stein, Grafakos, Uchiyama) and a change-of-aperture lemma from [15] and [3]. No parameter is fitted to a subset of data and then called a prediction, and no uniqueness theorem is imported to force the choice of the spaces. The paper is self-contained in its argument structure; the self-citations are technical inputs with independent content, so the circularity score is low.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

Pure mathematics: there are no data, no fitted constants, and no empirical inputs. The only tunable constants are theorem hypotheses, not fitted values: alpha > 2 in Theorem 1.4 (needed for the geometric series in Theorem 5.2), alpha > n in the proof of Theorem 3.8 (so that r in (n/alpha, 1) exists), and auxiliary exponents N and r chosen inside the proofs. All load-bearing inputs are listed as axioms; they are either standard results or theorems imported from [12, 15, 4, 9, 16]. No new entities are postulated; the wave packets theta, phi, chi, and eta are technical localization tools inherited from [12, 15].

assumptions (7)
  • domain assumption The cosphere bundle (S*(R^n), d, dxdomega) with the sub-Riemannian contact metric is a doubling metric measure space, with V(B_tau(x,omega)) comparable to tau^{2n} for tau < 1 and comparable to tau^n for tau >= 1.
    Imported from [12, Lemma 2.3] and stated as Lemma 2.1. Used in the annulus estimates of (3.21), the volume normalizations V(B_{sqrt(sigma)}) comparable to sigma^n in Theorem 5.2, and the applicability of Lemma 3.7. Different volume growth would break the polynomial decay in the maximal-function argument for (3.21).
  • domain assumption Wave packet estimates (2.7)-(2.9): derivative bounds for theta_{omega,sigma} and chi_{omega,sigma}, the kernel decay (1 + sigma^{-1}|x|^2 + sigma^{-2}<omega,x>^2)^{-N}, and the L^infty bounds for (integral phi_nu dnu)^{-1}.
    For theta these were proved in [15, Lemma 3.2]; for the new packets chi and eta they are proved in Lemma 2.2. The kernel decay drives Lemma 3.5 (growth condition) and enters Corollary 2.3. These bounds encode the parabolic geometry of the wave packets.
  • domain assumption Off-singularity kernel bound, Corollary 2.3: the integral kernel of theta_{omega,sigma}(D) chi_{nu,tau}(D) decays like min(sigma/tau, tau/sigma)^N rho^{-n}(1 + rho^{-1} d^2)^{-N} for each N, with rho = min(sigma,tau).
    Stated without proof in this paper, deferred to [15, Proposition 3.6] and [12, Theorem 5.1]. This is the engine of Proposition 3.6: the fast dyadic decay makes the lacunary sum converge and yields the pointwise inequality behind (3.21). Identified as the weakest assumption of the paper.
  • domain assumption Definition 1.1 of H^p_{FIO} via the conical square function (1.1) is equivalent to the original definition in [12].
    Cited from [15, Corollary 3.8]. This anchors all characterizations to the established H^p_{FIO} spaces; Theorem 1.4 characterizes this space, not a new variant.
  • standard math The vector-valued Hardy-Littlewood maximal operator is bounded on L^p(ell^q) for p, q > 1, and the Rychkov lacunary convolution lemma (Lemma 3.7) holds on doubling metric measure spaces.
    Cited from [9, Section 6.6] and [16]. Used in (3.21) with p = 1/r, q = 2/r for r in (0,1); the maximal inequality is applied componentwise to the sequence {theta_{.,2^{-j}sigma}(D)h}.
  • standard math Classical H^1(R^n) theory: Littlewood-Paley g-function, maximal function, and Riesz transform characterizations; H^infty_{FIO} = (H^1_{FIO})*; density of S(R^n) in H^1_{FIO}.
    Used in Propositions 3.1, Theorem 4.1, and Theorem 6.1. Standard references [20, 21, 11, 12, Proposition 6.6 and 6.8]. The paper's p = 1 characterization reduces H^1_{FIO} membership to these classical facts.
  • domain assumption Change of aperture formula, Lemma 5.1: replacing the cone B_{sqrt(sigma)} by B_{lambda sqrt(sigma)} costs a factor lambda^n in the L^1 square-function norm.
    From [15, Lemma 2.2] (after [3]). Needed in Theorem 5.2 to handle the aperture widening in the G*_alpha comparison. The cost factor is what forces the condition alpha > 2.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Characterizations of the Hardy space $\mathcal{H}_{FIO}^{1}(\mathbb{R}^{n})$ for Fourier integral operators." pith.science (2026). https://pith.science/paper/6PT3QVA4

@misc{pith2026190801448,
  author       = {Pith},
  title        = {Pith review of: Characterizations of the Hardy space $\mathcalH_FIO^1(\mathbbR^n)$ for Fourier integral operators},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6PT3QVA4}},
  note         = {Machine review of arXiv:1908.01448}
}
abstract

The Hardy spaces for Fourier integral operators $\mathcal{H}_{FIO}^{p}(\mathbb{R}^{n})$, for $1\leq p\leq \infty$, were introduced by Smith in [Smith,1998] and Hassell et al. in [Hassell-Portal-Rozendaal,2020]. In this article, we give several equivalent characterizations of $\mathcal{H}_{FIO}^{1}(\mathbb{R}^{n})$, for example in terms of Littlewood--Paley $g$ functions and maximal functions. This answers a question from [Rozendaal,2021]. We also give several applications of the characterizations.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references · 21 canonical work pages

  1. [12]

    Off-singularity bounds and Hardy spaces for Fourier integral operators

    A. Hassell, P . Portal, and J. Rozendaal. O ff-singularity bounds and Hardy spaces for Fourier integral o perators. To appear in Trans. Amer. Math. Soc. Preprint available at https://arxiv.org/abs/1811.11376, 2018. 1, 2, 3, 6, 8, 9, 10, 14, 20, 21, 22, 23

    work page Pith review arXiv 2018

  2. [15]

    Characterizations of Hardy spaces for Fourier integral operators

    J. Rozendaal. Characterizations of Hardy spaces for Fo urier integral operators. To appear in Rev. Mat. Iberoam. Preprint available at https://arxiv.org/abs/1907.02680, 2019. 1, 2, 3, 4, 5, 6, 7, 9, 17, 21

    work page Pith review arXiv 1907

  3. [1]

    A. Amenta. Tent spaces over metric measure spaces under d oubling and related assumptions. In Operator theory in harmonic and non-commutative analysis , volume 240 of Oper . Theory Adv. Appl. , pages 1–29. Birkh¨ auser/Springer, Cham, 2014. 2

    work page 2014

  4. [2]

    P . Auscher. On necessary and su fficient conditions for Lp-estimates of Riesz transforms associated to elliptic operators on Rn and related estimates. Mem. Amer . Math. Soc., 186(871):xviii+75, 2007. 1

    work page 2007

  5. [3]

    P . Auscher. Change of angle in tent spaces. C. R. Math. Acad. Sci. Paris , 349(5-6):297–301, 2011. 17

    work page 2011

  6. [4]

    Auscher, S

    P . Auscher, S. Hofmann, and J.-M. Martell. V ertical vers us conical square functions. Trans. Amer . Math. Soc., 364(10):5469–5489, 2012. 3, 9, 14

    work page 2012

  7. [5]

    H.-Q. Bui, M. Paluszy´ nski, and M. Taibleson. Character ization of the Besov-Lipschitz and Triebel-Lizorkin spaces. The case q < 1. In Proceedings of the conference dedicated to Professor Migue l de Guzm´ an (El Es- corial, 1996), volume 3, pages 837–846, 1997. 5

    work page 1996

  8. [6]

    H.-Q. Bui, M. Paluszy´ nski, and M. H. Taibleson. A maxima l function characterization of weighted Besov- Lipschitz and Triebel-Lizorkin spaces. Studia Math., 119(3):219–246, 1996. 5

    work page 1996

Show all 21 references
  1. [7]

    Coifman, Y

    R. Coifman, Y . Meyer, and E. M. Stein. Some new function sp aces and their applications to harmonic analysis. J. Funct. Anal., 62(2):304–335, 1985. 2

    1985

  2. [8]

    Duistermaat

    J. Duistermaat. F ourier integral operators. Modern Birkh¨ auser Classics. Birkh¨ auser/Springer, New Y ork, 2011. Reprint of the 1996 edition [MR1362544], based on the origin al lecture notes published in 1973 [MR0451313]. 1

    2011

  3. [9]

    Genebashvili, A

    I. Genebashvili, A. Gogatishvili, V . Kokilashvili, and M. Krbec. W eight theory for integral transforms on spaces of homogeneous type, volume 92 of Pitman Monographs and Surveys in Pure and Applied Mathemati cs. Longman, Harlow, 1998. 16

    1998

  4. [10]

    Goldberg

    D. Goldberg. A local version of real Hardy spaces. Duke Math. J., 46(1):27–42, 1979. 10

    1979

  5. [11]

    Grafakos

    L. Grafakos. Modern Fourier analysis, volume 250 of Graduate T exts in Mathematics. Springer, New Y ork, third edition, 2014. 1, 17 24 ZHIJIE FAN, NAIJIA LIU, JAN ROZENDAAL AND LIANG SONG

    2014

  6. [13]

    H¨ ormander

    L. H¨ ormander. Fourier integral operators. I. Acta Math., 127(1-2):79–183, 1971. 1

    1971

  7. [14]

    G. Hu. Littlewood-Paley characterization of weighted Hardy spaces associated with operators. J. Aust. Math. Soc., 103(2):250–267, 2017. 5

    2017

  8. [16]

    V . S. Rychkov. On a theorem of Bui, Paluszy´ nski, and Tai bleson. Proc. Steklov. Inst. Math. , 227(4):280–292,

  9. [17]

    Seeger, C

    A. Seeger, C. D. Sogge, and E. M. Stein. Regularity prope rties of Fourier integral operators. Ann. of Math. (2) , 134(2):231–251, 1991. 1

    1991

  10. [18]

    H. Smith. A Hardy space for Fourier integral operators. J. Geom. Anal., 8(4):629–653, 1998. 1, 2

    1998

  11. [19]

    C. Sogge. F ourier integrals in classical analysis, volume 210 of Cambridge Tracts in Mathematics . Cambridge University Press, Cambridge, second edition, 2017. 1

    2017

  12. [20]

    E. M. Stein. Harmonic analysis: real-variable methods, orthogonality , and oscillatory integrals , volume 43 of Princeton Mathematical Series . Princeton University Press, Princeton, NJ, 1993. With the assistance of Timothy S. Murphy, Monographs in Harmonic Analysis, III. 1, 9, 19, 20

    1993

  13. [21]

    Uchiyama

    A. Uchiyama. Characterization of H p(Rn) in terms of generalized Littlewood-Paley g-functions. Studia Math. , 81(2):135–158, 1985. 9 Zhijie Fan, Department of Mathematics, Sun Yat-sen University, Guangzhou, 510275, P .R. China E-mail address: fanzhj3@mail2.sysu.edu.cn Naijia L...

    1985

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.