Homogenization of non-symmetric convolution type operators

Andrey Piatnitski; Elena Zhizhina; Tatiana Suslina; Vladimir Sloushch

REVIEW 1 major objections 5 minor 30 references

Homogenization of non-symmetric convolution type operators

T0 review · 1 major / 5 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 Non-symmetric convolution operators admit O(ε) resolvent homogenization.

desk verdict Solid non-self-adjoint extension of operator homogenization; verify the cited convolution-power lemma before accepting. read the letter →

arxiv 2506.07176 v1 pith:ADLU7F36 submitted 2025-06-08 math.FA math.AP

Andrey Piatnitski , Vladimir Sloushch , Tatiana Suslina , Elena Zhizhina This is my paper

classification math.FAmath.AP MSC 35B2745K0547G1047A55

keywords convolutiontypeoperatorsperiodichomogenizationoperatorestimatesofdiscrepancyeffectivenon-self-adjointdriftGelfandtransformresolventapproximation

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 a sharp operator-norm approximation for a family of non-symmetric convolution type operators with periodic coefficients on L2(R^d). For small ε, the resolvent (A_ε+I)^{-1} is approximated to order O(ε) by the resolvent of a constant-coefficient elliptic operator with an added drift of order $ε^{{-1}}$, multiplied by a rapidly oscillating periodic factor q0(x/ε). The effective diffusion matrix and drift vector are computed from cell problems, and q0 is the positive kernel of the adjoint periodic operator, normalized to mean one. This extends the operator-theoretic homogenization approach from self-adjoint to non-self-adjoint nonlocal operators, and shows that the associated Cauchy problem homogenizes only in a moving frame.

What carries the argument

The load-bearing mechanism is the threshold approximation near the spectral edge of the fibered operator family A(ξ) obtained by the Gelfand transform. At ξ=0, the operator has an isolated simple eigenvalue λ0=0, with kernel spanned by constants and adjoint kernel spanned by a strictly positive periodic function q0. For small ξ, the Riesz projection F(ξ) and the product A(ξ)F(ξ) are expanded to order O(|ξ|) and O(|ξ|^3) respectively; the linear coefficient yields the effective drift i⟨α,ξ⟩P and the quadratic coefficient yields the effective matrix g0. A quadratic-form lower bound for Re([q0]A(ξ)) then controls the resolvent (A(ξ)+$ε^{2}$ I)^{-1} with precision O($ε^{{-1}}$), which after unitary scaling becomes the O(ε) resolvent estimate for A_ε.

What would settle it

Take a concrete non-symmetric pair (a,µ), compute α and g0 from (2.40) and (2.47), and evaluate the L2→L2 norm of the difference (A_ε+I)^{-1} − ($A^{0}$+$ε^{{-1}}$⟨α,∇⟩+I)^{-1}[q_0^ε] numerically for ε = $2^{{-n}}$; if the norms do not stay below C ε for some fixed C, or if the bound fails when q0 is replaced by a sign-changing solution of the adjoint equation, the main theorem is false.

Watch

Extended reading notes

Core claim

Under conditions (1.1)–(1.3) and the finite third moment M3(a)<∞, the paper establishes ‖(A_ε+I)^{-1} − ($A^{0}$ + $ε^{{-1}}$⟨α,∇⟩+I)^{-1}[q_0^ε]‖_{L2→L2} ≤ C ε for all ε>0, where $A^{0}$ = −div g0∇ is the effective diffusion operator with positive definite constant matrix g0, α is a constant vector called the effective drift, and [q_0^ε] is multiplication by the ε-periodic positive function q0(x/ε). The estimate is sharp in order, and the constant depends only on the listed data of a and µ. In particular, the resolvent of A_ε does not converge in the usual sense to the resolvent of a fixed limit operator; the leading approximation requires both the large drift $ε^{{-1}}$⟨α,∇⟩ and the oscillating factor q0. The proof combines a Gelfand-transform direct integral with threshold approximations of the spectral projection near the isolated eigenvalue zero of the fibered operator A(ξ).

Load-bearing premise

The argument depends on the periodic adjoint kernel q0 being strictly positive and bounded; if q0 could vanish on a set of positive measure, the quadratic-form lower bound and the O(ε) resolvent estimate would fail.

Editorial extensions

If this is right

The resolvent estimate (4.1) holds uniformly for all ε>0 with a constant depending only on d0, K, q−, q+, d, µ±, M1(a), M2(a), M3(a), M(a), Cπ(a), and Cr(a)(a).
If only M2(a)<∞ is assumed, the same approximating operator still gives convergence as ε→0, but without a rate; if M_k(a)<∞ for some 2<k<3, the rate becomes O(ε^{k−2}).
For the parabolic Cauchy problem ∂_t u = −A_ε u, the result implies homogenization in the moving frame (x,t) ↦ (x − α t/ε, t); in the original frame the semigroup does not converge in the usual strong topology.
The statement extends to arbitrary periodic lattices in R^d, with constants depending on the lattice parameters.

Reading between the lines

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

The same threshold-expansion strategy should transfer to non-self-adjoint periodic differential operators, where the effective drift would enter as a first-order symbol term and the adjoint ground state would play the role of q0.
Because the approximation keeps the oscillating factor q0(x/ε), the leading-order asymptotics carries a two-scale structure; a single autonomous effective operator cannot describe the resolvent on L2(R^d).
For non-symmetric Lévy-type operators with stable-like kernels, an analogous operator-norm estimate with a drift term should hold, with q0 corresponding to the invariant density of the periodic process; this is a testable extension of the present method.

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

This is the first operator-norm homogenization estimate for non-symmetric convolution-type operators, and the main theorem looks right. The approximation requires both an effective drift ε^{-1}⟨α,∇⟩ and multiplication by the periodic factor q0(x/ε) — that is genuinely new and not just a cosmetic addition. The effective coefficients come from cell problems, not from fitting, and the proof is careful and mostly self-contained.

The method is coherent: rescale, decompose into a direct integral, and use finite smoothness from the moment assumptions to do a threshold analysis near ξ=0. The quadratic-form identity in Lemma 1.6 is a neat trick, and the appendices on the auxiliary operator G are substantial and honest work.

The one point I would send a referee to check is the strict positivity of q0. That rests entirely on [27, Lemma 4.2], which asserts that some convolution power of the periodized L1 kernel is bounded below by a positive constant. The paper cites it without proof. If that lemma needs extra structure beyond (1.1)–(1.3) — say boundedness or a support with nonempty interior — the main estimate collapses. I did not find a counterexample, and the proof's use of the lower bound is not obviously excessive, but this is load-bearing and should be verified. The stress-test note flags exactly the right risk.

Also, the phrase 'sharp in order' is not backed by a matching lower bound. The O(ε) upper bound is proved, and the remark giving O(ε^{k-2}) for 2

If you are in quantitative homogenization, this is worth a serious referee. The architecture is sound, the novelty is real, and the cited lemma is likely checkable. I would send it to review with a request to confirm [27, Lemma 4.2] (either prove it or give a full reference) and to soften the 'sharp' claim. My own verdict is conditional, leaning accept after that verification.

Referee Report

1 major / 5 minor

Summary. This paper studies homogenization of non-symmetric convolution type operators of the form (0.1) in L2(Rd). Under assumptions (1.1)–(1.3) and the third moment condition M3(a)<∞, Theorem 4.1 asserts an O(ε) operator-norm approximation of the resolvent (Aε+I)^{-1} by the operator (A0+ε^{-1}⟨α,∇⟩+I)^{-1} multiplied on the right by q0(x/ε), where A0=-div g0∇ is the effective elliptic operator and α is the effective drift. The proof combines the Gelfand transform, a reduction to the fiber operators A(ξ), threshold expansions of the Riesz projector, accretivity estimates for [q0^{1/2}]A(ξ)[q0^{-1/2}], and cell problems for the effective coefficients. Sections 5 and 6 contain supporting material on stability of isolated eigenvalues and on the auxiliary operator G, including the proof that the adjoint kernel q0 is strictly positive and bounded.

Significance. If the main estimate is correct, this is a substantial extension of the Birman–Suslina operator-theoretic homogenization method to non-self-adjoint, nonlocal convolution-type operators. The explicit cell problems for α and g0, the moving-frame interpretation of the ε^{-1} drift term, the absence of fitted parameters, and the operator-norm nature of the error estimate are genuine strengths. The result is conditional on the positivity of q0, which in turn depends on a convolution-power lower bound quoted from a previous work; this point needs to be clarified before the significance can be fully certified.

major comments (1)

[§6.2, Lemma 6.2; Prop. 1.1(3)] The proof of Lemma 6.2 relies on the assertion, quoted from [27, Lemma 4.2], that there exist N and γ>0 such that the N-th convolution power F_N of the L1 periodization ea is bounded below by γ on Ω. This assertion is load-bearing: it is the only input that gives the strict positivity (1.10) of q0, which is then used in identity (1.15), Corollaries 1.7, 1.9 and 1.14, the quadratic-form lower bound (1.39), and the positive definiteness of g0 in Section 3.1. The manuscript neither states the hypotheses of [27, Lemma 4.2] nor proves the lower-bound statement under (1.1)–(1.3). For a general nonnegative L1 kernel with positive measure whose periodization has support with empty interior, convolution powers need not be uniformly positive on Ω. The authors should state the lemma in full and verify its assumptions under (1.1)–(1.3), or supply a self-contained proof; if an additional support or regularity condition is needed, the hypotheses of Theorem 4.1 must be modified accordingly.

minor comments (5)

[§0.2] The heading contains the typos 'Introdiction' and 'Secrions'; 'six Secrions' should read 'six Sections'.
[§2.2, proof of Prop. 2.4] There are two typos: 'conrour' should be 'contour', and in (2.39) the left-hand side is repeated as 'w_j(x) = w_j(x) ='.
[§6.2, Lemma 6.4] In the sentence 'for any n∈N' and in formula (6.7), the symbol should be N (the same letter as in G^N) rather than n.
[§6.6] The sentence 'From this inequality, taking into account (6.17)...' should refer to the almost-everywhere convergence established just before, so 'From this convergence' is more accurate.
[Abstract and §4.1] The phrase 'sharp in order' is stronger than what is proved: Theorem 4.1 establishes an O(ε) upper bound but no matching lower bound or example showing that the order cannot be improved. Please either add a lower-bound result or replace 'sharp in order' by 'with discrepancy of order O(ε)' throughout.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: effective coefficients are derived from cell problems, not fitted; the approximation does not encode its conclusion.

full rationale

The claimed O(eps) estimate in Theorem 4.1 follows from Theorem 3.4 by unitary scaling (Eqs. (4.2)-(4.3)), and Theorem 3.4 is obtained from the threshold approximations of A(xi) near the spectral edge. The effective drift alpha and the effective matrix g0 are defined by explicit cell-problem formulas (2.40) and (2.47), involving a, mu and the adjoint kernel q0, rather than by matching the resolvent that is to be approximated. The approximating operator (A0 + eps^{-1}<alpha, grad> + I)^{-1}[q0] is constructed from the spectral Taylor data of A(xi)F(xi) in Propositions 2.4-2.5, so it is not a restatement of the theorem. Positivity of q0 (Proposition 1.1(3)) relies on Theorem 6.1, whose Lemma 6.2 invokes [27, Lemma 4.2]; this is a genuine external published lemma and is not equivalent to the homogenization estimate, and the paper supplies the surrounding proof. No equation in the paper reduces to the target estimate by definition, and no fitted parameter is relabelled as a prediction. The reuse of the authors' own operator-theoretic technique [22,23,24] is methodological rather than circular.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No free parameters are fitted to data. The effective matrix g^0 and drift α are defined by cell problems (2.45) and (2.47), not tuned to match the approximation. The only auxiliary object is the function q0, which is an eigenfunction of the adjoint operator and is proven positive and bounded; it is not a new physical entity.

assumptions (4)

standard math The Gelfand transform maps periodic operators to direct integrals of fiber operators.
Used in Section 1.2, formula (1.5), to diagonalize the operator and reduce the problem to a family A(ξ) on the cell.
standard math The analytic Fredholm theorem guarantees the discrete nature of the spectrum of A(ξ) outside the essential spectrum.
Invoked in Section 1.2 to analyze the spectrum and Riesz projectors.
domain assumption The coefficient a is nonnegative, integrable, with finite third moment M3(a).
Condition (1.1) and the main theorem's hypothesis; provides C^3 regularity of the fiber family and the threshold expansions in Section 2.
domain assumption The coefficient μ is bounded and uniformly positive.
Condition (1.2) yields the accretivity estimates (1.39) and the spectral gap near zero.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Homogenization of non-symmetric convolution type operators." pith.science (2026). https://pith.science/paper/ADLU7F36

@misc{pith2026250607176,
  author       = {Pith},
  title        = {Pith review of: Homogenization of non-symmetric convolution type operators},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ADLU7F36}},
  note         = {Machine review of arXiv:2506.07176}
}

abstract

The paper studies homogenization problem for a bounded in $L_2(\mathbb R^d)$ convolution type operator ${\mathbb A}_\eps$, $\eps >0$, of the form $$ ({\mathbb A}_\eps u) (\x) = \eps^{-d-2} \int_{\R^d} a((\x-\y)/\eps) \mu(\x/\eps, \y/\eps) \left( u(\x) - u(\y) \right)\,d\y. $$ It is assumed that $a(\x)$ is a non-negative function from $L_1(\R^d)$, and $\mu(\x,\y)$ is a periodic in $\x$ and $\y$ function such that $0< \mu_- \leqslant \mu(\x,\y) \leqslant \mu_+< \infty$. No symmetry assumption on $a(\cdot)$ and $\mu(\cdot)$ is imposed, so the operator ${\mathbb A}_\eps$ need not be self-adjoint. Under the assumption that the moments $M_k = \int_{\R^d} |\x|^k a(\x)\,d\x$, $k=1,2,3$, are finite we obtain, for small $\eps>0$, sharp in order approximation of the resolvent $({\mathbb A}_\eps + I)^{-1}$ in the operator norm in $L_2(\mathbb R^d)$, the discrepancy being of order $O(\eps)$. The approximation is given by an operator of the form $({\mathbb A}^0 + \eps^{-1} \langle \boldsymbol{\alpha},\nabla \rangle + I)^{-1}$ multiplied on the right by a periodic function $q_0(\x/\eps)$; here ${\mathbb A}^0 = - \operatorname{div}g^0 \nabla$ is the effective operator, and $\boldsymbol{\alpha}$ is a constant vector.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

30 extracted references · 27 canonical work pages

[1]

ESAIM: Control, Optimisation and Calculus of Variations 13(4), 735–749 (2007)

Allaire, G., Orive, R.: Homogenization of periodic non self-adjoint problems with large drift and potential. ESAIM: Control, Optimisation and Calculus of Variations 13(4), 735–749 (2007)

work page 2007
[2]

Kluwer Academic Pub- lishers, Dordrecht, Boston, London (1989)

Bakhvalov, N., Panasenko, G.: Homogenization: averaging processes in periodic media. Kluwer Academic Pub- lishers, Dordrecht, Boston, London (1989)

work page 1989
[3]

Bensoussan, A., Lions, J.-L., Papanicolaou, G.: Asymptotic analysis for periodic structures. Stud. Math. Appl., vol. 5, North-Holland Publishing Co., Amsterdam, New York (1978)

work page 1978
[4]

Sh., Suslina, T

Birman, M. Sh., Suslina, T. A.: Second order periodic differential operators. Threshold properties and homoge- nization. St. Petersburg Math. J. 15(5), 639–714 (2004)

work page 2004
[5]

Sh., Suslina T

Birman, M. Sh., Suslina T. A.: Homogenization with corrector term for periodic elliptic differential operators. St. Petersburg Math. J. 17(6), 897–973 (2006)

work page 2006
[6]

Sh., Suslina, T

Birman, M. Sh., Suslina, T. A.: Homogenization with corrector for periodic differential operators. Approximation of solutions in the Sobolev classH 1(Rd). St. Petersburg Math. J. 18(6), 857–955 (2007)

work page 2007
[7]

Braides, A., Piatnitski, A.:Homogenization of quadratic convolution energies in periodically perforated domains. Adv. Calc. Var. 15(3), 351–368 (2022). https://doi.org/10.1515/acv-2019-0083

doi:10.1515/acv-2019-0083 2022
[8]

Chen, X.; Chen, Z.-Q.; Kumagai, T.; Wang, J.,Quantitative periodic homogenization for symmetric non-local stable-like operators, arXiv: 2409.08120v1 (12 Sep 2024)

arXiv 2024

Show all 30 references

[9]

Chen, X., Chen, Z.-Q., Kumagai, T., Wang, J.: Periodic homogenization of nonsymmetric L´ evy-type processes. Ann. Probab. 49(6), 2874–2921 (2021)

2021
[10]

Multi scale problems and asymptotic analysis

Donato, P., Piatnitski, A.: Averaging of nonstationary parabolic operators with large lower order terms. Multi scale problems and asymptotic analysis. 153–165, GAKUTO Internat. Ser., Math. Sci. Appl.,24, Gakk¯ otosho, Tokyo (2005)

2005
[11]

Garnier, J.: Homogenization in a periodic and time-dependent potential. SIAM J. Appl. Math. 57(1), 95–111 (1997)

1997
[12]

A.: Classes of Linear Operators

Gohberg, I., Goldberg, S., Kaashoek, M. A.: Classes of Linear Operators. Vol. I, Operator Theory: Advances and Applications. (OT, Vol. 49), Springer, Basel (1990)

1990
[13]

Kassmann, M., Piatnitski, A., Zhizhina, E.: Homogenization of L´ evy type operators with oscillating coefficients. SIAM J. Math. Anal. 51, 3641–3665 (2019)

2019
[14]

Kesavan S.: Homogenization of elliptic eigenvalue problems: part I. Appl. Math. Optim., 5, 153–167 (1979)

1979
[15]

Kesavan S.: Homogenization of elliptic eigenvalue problems: part II. Appl. Math. Optim. 5 197–216 (1979)

1979
[16]

Springer-Verlag, Berlin (1980)

Kato T.:Perturbation theory for linear operators. Springer-Verlag, Berlin (1980)

1980
[17]

Studies in mathematics and its applications

Oleinik, O.A., Shamaev, A.S., Yosifian, G.A.: Mathematical Problems in Elasticity and Homogenization. Studies in mathematics and its applications. V.26, North-Holland (1992)

1992
[18]

V.: On operator estimates in homogenization theory

Zhikov,V. V.: On operator estimates in homogenization theory. Dokl. Math. 72(1), 534–538 (2005)

2005
[19]

V., Kozlov, S

Zhikov, V. V., Kozlov, S. M., Oleinik, O. A.: Homogenization of differential operators and integral functionals. Springer-Verlag, Berlin (1994)

1994
[20]

V., Pastukhova, S

Zhikov, V. V., Pastukhova, S. E.: On operator estimates for some problems in homogenization theory. Russ. J. Math. Phys. 12(4), 515–524 (2005)

2005
[21]

V., Pastukhova, S

Zhikov, V. V., Pastukhova, S. E.: Operator estimates in homogenization theory. Russian Math. Surveys 71(3), 417–511 (2016)

2016
[22]

Piatnitski, A., Sloushch, V., Suslina, T., Zhizhina, E.: On operator estimates in homogenization of nonlocal operators of convolution type. J. Diff. Equ. 352, 153–188 (2023)

2023
[23]

Piatnitski, A., Sloushch, V., Suslina, T., Zhizhina, E.: On the homogenization of nonlocal convolution type operators. Russ. J. Math. Phys. 31, 137–145 (2024). https://doi.org/10.1134/S106192084010114

2024 doi
[24]

arXiv:2311.16574

Piatnitski, A., Sloushch, V., Suslina, T., Zhizhina, E.: Homogenization of nonlocal convolution type operators: Approximation for the resolvent with corrector. arXiv:2311.16574

arXiv
[25]

Piatnitski, A., Sloushch, V., Suslina, T., Zhizhina, E.: Operator estimates in homogenization of L´ evy-type operators with periodic coefficients, arXiv:2412.20408

arXiv
[26]

Piatnitski, A., Zhizhina, E.: Periodic homogenization of nonlocal operators with a convolution-type kernel. SIAM J. Math. Anal. 49(1), 64–81 (2017) 34

2017
[27]

Asymptotic Anal

Piatnitski, A., Zhizhina, E.: Homogenization of biased convolution type operators. Asymptotic Anal. 115(3-4), 241–262 (2019)

2019
[28]

Markov Processes and Related Fields 29, 173–188 (2023)

Piatnitski, A., Zhizhina, E.: Homogenization of non-autonomous operators of convolution type in peri- odic media. Markov Processes and Related Fields 29, 173–188 (2023). https://doi.org/10.61102/1024-2953- mprf.2023.29.2.001

2023 doi
[29]

M.: Geometric and arithmetic methods in the spectral theory of multidimensional periodic oper- ators

Skriganov, M. M.: Geometric and arithmetic methods in the spectral theory of multidimensional periodic oper- ators. Proceedings of the Steklov Institute of Mathematics 171, AMS, Providence (1987)

1987
[30]

A.: Operator-theoretic approach to the homogenization of Schr¨ odinger-type equations with periodic coefficients

Suslina T. A.: Operator-theoretic approach to the homogenization of Schr¨ odinger-type equations with periodic coefficients. Russian Math. Surveys 78(6), 1023–1154 (2023)

2023

Pith tools

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

[1] [1]

ESAIM: Control, Optimisation and Calculus of Variations 13(4), 735–749 (2007)

Allaire, G., Orive, R.: Homogenization of periodic non self-adjoint problems with large drift and potential. ESAIM: Control, Optimisation and Calculus of Variations 13(4), 735–749 (2007)

work page 2007

[2] [2]

Kluwer Academic Pub- lishers, Dordrecht, Boston, London (1989)

Bakhvalov, N., Panasenko, G.: Homogenization: averaging processes in periodic media. Kluwer Academic Pub- lishers, Dordrecht, Boston, London (1989)

work page 1989

[3] [3]

Bensoussan, A., Lions, J.-L., Papanicolaou, G.: Asymptotic analysis for periodic structures. Stud. Math. Appl., vol. 5, North-Holland Publishing Co., Amsterdam, New York (1978)

work page 1978

[4] [4]

Sh., Suslina, T

Birman, M. Sh., Suslina, T. A.: Second order periodic differential operators. Threshold properties and homoge- nization. St. Petersburg Math. J. 15(5), 639–714 (2004)

work page 2004

[5] [5]

Sh., Suslina T

Birman, M. Sh., Suslina T. A.: Homogenization with corrector term for periodic elliptic differential operators. St. Petersburg Math. J. 17(6), 897–973 (2006)

work page 2006

[6] [6]

Sh., Suslina, T

Birman, M. Sh., Suslina, T. A.: Homogenization with corrector for periodic differential operators. Approximation of solutions in the Sobolev classH 1(Rd). St. Petersburg Math. J. 18(6), 857–955 (2007)

work page 2007

[7] [7]

Braides, A., Piatnitski, A.:Homogenization of quadratic convolution energies in periodically perforated domains. Adv. Calc. Var. 15(3), 351–368 (2022). https://doi.org/10.1515/acv-2019-0083

doi:10.1515/acv-2019-0083 2022

[8] [8]

Chen, X.; Chen, Z.-Q.; Kumagai, T.; Wang, J.,Quantitative periodic homogenization for symmetric non-local stable-like operators, arXiv: 2409.08120v1 (12 Sep 2024)

arXiv 2024