Cross-LoRA: A Data-Free LoRA Transfer Framework across Heterogeneous LLMs

Defang Li; Deguo Xia; Feifan Xia; Jizhou Huang; Mingyang Liao; Weikang Li; Yang Li; Yantong Xie; Yuyang Fang

REVIEW 4 major objections 3 minor 1 cited by

Cross-LoRA: A Data-Free LoRA Transfer Framework across Heterogeneous LLMs

T0 review · 4 major / 3 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read Cross-LoRA claims to transfer LoRA adapters between different large language models without any task data, using only the geometry of the base models' weight matrices.

desk verdict Can't actually review this paper — the body we got is an unrelated math.AP article, so the only reviewable evidence is an abstract that describes a plausible but unverified LoRA transfer method. read the letter →

arxiv 2508.05232 v1 pith:GPV22SUQ submitted 2025-08-07 cs.LG

Feifan Xia , Mingyang Liao , Yuyang Fang , Defang Li , Yantong Xie , Weikang Li , Yang Li , Deguo Xia

show 1 more author

Jizhou Huang

This is my paper · ORCID

classification cs.LG

keywords LoRAtransferparameter-efficientfine-tuningsubspacealignmentsingularvaluedecompositiondata-freecross-modeladaptationcommonsensereasoning

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

Cross-LoRA aims to make LoRA adapters portable across different large language models—models that differ in size, architecture, or tokenizer—without using any task data. It does this by rewriting a source model's LoRA update in the target model's parameter space, after aligning the two base models' dominant singular subspaces. The paper reports that this transferred adapter improves over the target base model by up to 5.26% on ARCs, OBOA, and HellaSwag, and matches directly trained LoRA adapters on other commonsense reasoning benchmarks. The whole procedure is data-free and training-free, running in about 20 minutes on a commodity GPU.

What carries the argument

Two components form the method. LoRA-Align performs rank-truncated singular value decomposition on the source and target base-model weight matrices and finds the Frobenius-optimal linear map between their dominant singular subspaces, so the two models' update directions become comparable despite dimension mismatch. LoRA-Shift applies that aligned basis to the source LoRA weight updates, projecting each low-rank factor into the target model's parameter space. Together they convert a source adapter into a target adapter with no training and no data.

What would settle it

Take a source model with a task-trained LoRA, compute Cross-LoRA's alignment to a target model, and compare the transferred adapter against the same LoRA projected onto random subspaces of the target base model. If the gain over the random projection falls to zero on held-out examples, then the singular-subspace alignment carries no task signal.

Watch

Extended reading notes

Core claim

The central claim is that task behavior encoded in a LoRA adapter is not tied to the specific base model it was trained on. Cross-LoRA decomposes the source base model's weight matrix by rank-truncated SVD, computes a Frobenius-optimal linear transformation that maps the source's top singular subspaces onto the target's top singular subspaces, and then uses that map (LoRA-Align) to project the source LoRA update into the target parameter space (LoRA-Shift). The result is an adapter that can be applied directly to the target model. The paper argues this works without any target-task training data, and the experiments on commonsense reasoning benchmarks support the claim that the transferred a

Load-bearing premise

The method assumes that the dominant singular directions of the source and target base models encode the same learned features, so that a linear map computed from base-model geometry can faithfully transplant a task adapter; if the directions do not match, the transfer silently distorts the adapter.

Editorial extensions

If this is right

Adapters tuned on one model can be dropped onto a different model, so task knowledge survives model upgrades.
Organizations can transfer fine-tuned behavior without sharing or generating data, which matters for privacy and licensing.
The 20-minute, single-GPU budget makes the transfer usable as a routine operation.
If the alignment works across architectures, the same idea could apply to other low-rank adapters or to submodules beyond attention weights.

Reading between the lines

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

The paper reports gains on multiple benchmarks, but a systematic study of failure cases—pairs of models whose singular subspaces are semantically far apart—would map the boundary of the assumption.
If the singular-subspace alignment genuinely captures shared feature structure, the same LoRA-Align map could be reused to transfer several adapters from the same source model, amortizing the one-time alignment cost.
A harder test would transfer to a model with a different tokenizer while controlling for architecture and size, isolating whether vocabulary mismatch or geometric mismatch is the bigger obstacle.

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.

Desk Editor's Note

read the letter

Quick take: we can't review this paper as submitted. The full text attached to arXiv:2508.05232 is actually arXiv:2508.05220v2, a math.AP paper by Romain Joly on parabolic evolution equations in uniformly local Sobolev spaces. It has nothing to do with LoRA, adapters, or LLMs. So the only in-scope evidence is the abstract.

What the abstract describes is coherent and worth taking seriously as an idea. Transferring a LoRA adapter across heterogeneous base models via rank-truncated SVD subspace alignment plus a Frobenius-optimal (Procrustes-style) linear map is a clean combination of standard linear algebra tools. It is genuinely data-free and training-free, which is a meaningful practical target for PEFT and model serving. The construction does not fit to task labels, so circularity is not a concern. The proposal is not a restatement of a named prior result, as far as that can be judged from the abstract.

The problems start with evidence. The headline number is "up to 5.26%" — that is a best case, not a spread. The claim that Cross-LoRA "maintains performance comparable to directly trained LoRA adapters" comes without per-benchmark numbers, baselines, standard deviations, or even the model pairs used. Only three benchmarks are named, all English multiple-choice commonsense tasks. No code or artifact is referenced. Most importantly, the load-bearing assumption is unexamined: that the top-r singular subspaces of different base models are semantically comparable, so a Frobenius-optimal map computed from base-model geometry can transfer task-relevant LoRA directions. That assumption can silently fail across different tokenizers, training objectives, and model scales, and because the method is data-free there is no signal to detect the failure.

So the honest verdict is: a plausible abstract, but nothing that can be checked. The missing body is not a minor referee concern; it is the entire paper. I would not send this artifact to peer review, because there is no manuscript to review. If the authors' actual paper exists, it may well deserve a look — the idea is worth a rigorous test. But from what we were given, I can't recommend engagement, citation, or reading-group time.

My recommendation: desk reject this artifact, and if the authors resubmit the correct full text, do a proper eval then.

Referee Report

4 major / 3 minor

Summary. The manuscript as submitted presents an abstract for 'Cross-LoRA', a data-free, training-free framework for transferring LoRA adapters between heterogeneous base LLMs. The proposed mechanism has two components: LoRA-Align (rank-truncated SVD of source and target base weights plus a Frobenius-optimal linear map) and LoRA-Shift (projecting source LoRA deltas into the target subspace). The abstract reports relative gains up to 5.26% over base models on ARCs, OBOA, and HellaSwag, and parity with directly trained LoRA adapters on other commonsense reasoning benchmarks. The full text supplied, however, is arXiv:2508.05220v2, a mathematics paper on parabolic abstract evolution equations in uniformly local Sobolev spaces; it contains no mention of LoRA, LLMs, subspace alignment, or any of the experiments. The only in-scope evidence for the claimed contribution is the abstract.

Significance. If the claimed transfer method works, the contribution could be practically useful: the data-free and training-free property is attractive, and the high-level construction (rank-truncated SVD plus optimal linear map) is concrete enough to be falsifiable. The paper should receive credit for proposing a clearly specified, parameterizable mechanism and for not relying on target-task labels. Nevertheless, as submitted, the manuscript cannot be technically evaluated. There are no derivations, algorithm definitions, benchmark tables, baselines, variance estimates, or artifact links. The body is an unrelated mathematics paper, so the central claim is unsupported by any in-scope evidence. This is a submission-integrity problem, not a normal scientific disagreement with the field's consensus.

major comments (4)

[Full Text (entire document)] The body of the submission is arXiv:2508.05220v2, 'Parabolic abstract evolution equations in cylindrical domains and uniformly local Sobolev spaces' by Romain Joly. This text has no overlap with the abstract's topic: there is no LoRA, no SVD-based subspace alignment, no LoRA-Shift projection, and no LLM experiments. The manuscript therefore contains no definitions, equations, pseudo-code, or experimental protocol for the claimed Cross-LoRA framework. The central claim cannot be checked from the submitted material.
[Abstract] The headline result, 'relative gains of up to 5.26%', is reported as a selected maximum rather than as a distribution over benchmarks and model pairs. No per-benchmark results, model architectures, LoRA ranks, base-model pairs, or baselines are provided. A single 'up to' figure is not an evaluable empirical claim, especially when the sentence covers three different benchmarks (ARCs, OBOA, HellaSwag) with presumably different gains.
[Abstract (parity claim)] The statement that on 'other commonsense reasoning benchmarks' Cross-LoRA 'maintains performance comparable to that of directly trained LoRA adapters' is made without naming those benchmarks, giving their metrics, or defining 'comparable.' No confidence intervals, standard deviations, or significance tests are reported. Because the method is data-free, there is also no discussion of how a silent transfer failure would be detected or bounded; this is especially important given the unsupervised nature of the projection.
[Abstract (LoRA-Align)] The method rests on the assumption that the top-r singular subspaces of source and target base-model weight matrices are semantically comparable. The abstract provides no evidence for this assumption across heterogeneous architectures, tokenizers, or pretraining runs. The Frobenius-optimal linear map is optimal in a matrix-norm sense, but optimality in that sense does not imply preservation of task-relevant directions. With no ablations over the rank-truncation level r and no layer-wise analysis, this load-bearing premise remains unsupported.

minor comments (3)

[Title/identifier] The first page of the supplied full text displays arXiv:2508.05220v2 with the title of the mathematics paper, while the submission is identified as arXiv:2508.05232 and titled 'Cross-LoRA'. This identifier/title mismatch must be resolved before the manuscript can be handled as the intended submission.
[Abstract] The benchmark names ARCs and OBOA are not expanded or cited, and the 'other commonsense reasoning benchmarks' are not enumerated. Full benchmark names, licenses, and evaluation settings would be needed even in a short abstract.
[Reproducibility] The abstract mentions 'lightweight adaptation on a commodity GPU in 20 minutes' but gives no hardware, framework, code, or seed information. A reproducibility statement and an artifact link are needed for the claimed experiments.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found; the claimed transfer construction is data-free and no fitted-input loop is visible, though the supplied full text is an unrelated paper and the central empirical claim is unverifiable rather than circular.

full rationale

The Cross-LoRA abstract describes a deterministic, data-free construction: LoRA-Align computes a rank-truncated SVD of the source and target base-model weight matrices and a Frobenius-optimal linear transformation between subspaces, and LoRA-Shift projects the source LoRA update into the target parameter space. There is no target-task label fitting, no benchmark outcome used to set parameters, and no equation in the supplied material that makes the reported performance equal to an input by construction. The claimed relative gains and 'comparable to directly trained LoRA' statements are empirical assertions, not derived identities. Under the hard rules, circularity requires quoting the paper and exhibiting a specific reduction; none can be exhibited here. The supplied full text is arXiv:2508.05220v2, a mathematics paper on parabolic abstract evolution equations by Romain Joly, which contains no definitions, equations, experiments, or code for Cross-LoRA. This is a serious evidentiary gap: the central claim cannot be checked from the in-scope text. However, unverifiability and missing support are not circularity. No self-citation chain, uniqueness import, ansatz-smuggling, or renaming of a known result appears in the supplied material. Therefore the appropriate circularity score is 0, with the caveat that the manuscript-as-supplied does not contain the actual method description needed to fully audit it.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The framework's contribution rests on two domain assumptions that the abstract cannot establish: that base-model singular subspaces are semantically comparable across heterogeneous architectures, and that LoRA updates live within the truncated principal subspace of the base weights. The only clearly identifiable free parameter is the SVD rank-truncation level; other design choices (layer scope, scaling) are unspecified in the abstract. No invented entities are introduced; LoRA-Align and LoRA-Shift are algorithmic stages, not ontological postulates.

free parameters (2)

SVD rank-truncation level r = not reported in abstract
LoRA-Align performs "rank-truncated SVD" of source and target base models; the truncation rank determines how much of each base model's subspace is retained for the Frobenius-optimal alignment. The abstract gives no rule for choosing it, so it is a hand-chosen hyperparameter of the method.
Layer-wise scope of alignment = not reported in abstract
The abstract does not say whether every LoRA/low-rank weight matrix of every layer is aligned identically or whether specific modules are selected; this choice changes the effective transfer and is not determined by the data-free framework.

assumptions (4)

domain assumption The top-r singular subspaces of source and target base-model weight matrices are semantically comparable, so a Frobenius-optimal linear map between them transfers task-relevant directions.
This is the load-bearing premise of LoRA-Align. If two base models' dominant singular directions do not correspond to the same learned features, projecting the LoRA update along them will corrupt the adapter, and the data-free framework provides no correction signal. Stated implicitly in the abstract's "subspace alignment... ensuring compatibility under dimension mismatch."
domain assumption LoRA fine-tuning updates are well represented in the rank-truncated principal subspace of the base weights.
The transfer re-expresses source LoRA updates through source base-model subspaces; if the update's relevant directions lie outside the truncated subspace, information is lost before the alignment. Plausible for low-rank adaptation but neither proven nor tested in the abstract.
standard math The Frobenius-optimal linear map (orthogonal Procrustes style) preserves the semantics of the LoRA update under rotation and reflection.
The optimality of the mapping is a standard linear algebra result; the nontrivial semantic claim is captured by the previous axiom. Cited in the abstract as "Frobenius-optimal linear transformation."
domain assumption Performance parity on ARC, OpenBookQA, and HellaSwag transfers to the other capabilities claimed by the paper ("other commonsense reasoning benchmarks").
The evaluation is multiple-choice commonsense QA; the abstract generalizes to "other commonsense reasoning benchmarks" without reporting them. The representativeness of the three named tasks for the general claim is assumed.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Cross-LoRA: A Data-Free LoRA Transfer Framework across Heterogeneous LLMs." pith.science (2026). https://pith.science/paper/GPV22SUQ

@misc{pith2026250805232,
  author       = {Pith},
  title        = {Pith review of: Cross-LoRA: A Data-Free LoRA Transfer Framework across Heterogeneous LLMs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GPV22SUQ}},
  note         = {Machine review of arXiv:2508.05232}
}

read the original abstract

Traditional parameter-efficient fine-tuning (PEFT) methods such as LoRA are tightly coupled with the base model architecture, which constrains their applicability across heterogeneous pretrained large language models (LLMs). To address this limitation, we introduce Cross-LoRA, a data-free framework for transferring LoRA modules between diverse base models without requiring additional training data. Cross-LoRA consists of two key components: (a) LoRA-Align, which performs subspace alignment between source and target base models through rank-truncated singular value decomposition (SVD) and Frobenius-optimal linear transformation, ensuring compatibility under dimension mismatch; and (b) LoRA-Shift, which applies the aligned subspaces to project source LoRA weight updates into the target model parameter space. Both components are data-free, training-free, and enable lightweight adaptation on a commodity GPU in 20 minutes. Experiments on ARCs, OBOA and HellaSwag show that Cross-LoRA achieves relative gains of up to 5.26% over base models. Across other commonsense reasoning benchmarks, Cross-LoRA maintains performance comparable to that of directly trained LoRA adapters.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

Parametric Memory Decoding for Zero-Shot Routing in LoRA-Based External Parametric Memory
cs.LG 2026-07 conditional novelty 6.0 of 10

PMDRouter selects LoRAs zero-shot by decoding scale-normalized linear response energy from one adapter-free backbone prefill, and leads most internal-signal baselines on a new multi-granularity EPM bench.

Reference graph

Works this paper leans on

61 extracted references · 61 canonical work pages · cited by 1 Pith paper

[1]

Indeed, an initial data as �0(�) = sin( �2) exhibits very fast oscillation at infinity

�� 2 ul� w(� �� ). Indeed, an initial data as �0(�) = sin( �2) exhibits very fast oscillation at infinity. These oscillations are homogenized by the heat semigroup (defined via the convolution by its kernel) and, as soon as � is positive, �(�) = �∆t�0 must converge to 0 as � goes to �� , see Figure 2. Thus ��(�)� �0�L2 ul d...

work page
[2]

This excludes the problematic � �� sin(�2) and can be seen as replacing � 0(� �� ) by the space of uniformly continuous functions

�� 2 ul� s(� �� )� The space � 2 ul� s(� �� ) is a stronger version of the uniformly local space where the translation � �� (� � �) is continuous for all functions �, see Section 3. This excludes the problematic � �� sin(�2) and can be seen as replacing � 0(� �� ) by the space of uniformly continuous functions. It is ...

work page
[3]

Indeed, we can use the oscillations in the � direction to construct an initial data for which the heat semigroup is not continuous at � = 0 +, see Figure

�� 2 ul� s(� � �2((� 1�1)))� In our cylindrical set- ting, choosing the strong version of the uniformly local spaces is not sufficient to define a proper Cauchy problem, even for the heat equation in � � (� 1�1). Indeed, we can use the oscillations in the � direction to construct an initial data for which the heat semigroup...

work page
[4]

abstract parabolic operator

This shows that the strong version of the uniformly local spaces is not so helpful in a general context of vector valued functions. To our point of view, one of the main interests of the present paper will be to clarify the issues raised in the above remarks. Figure 2: �� sin(�2) �� ...

work page
[6]

Arrieta, J.W

J.M. Arrieta, J.W. Cholewa, T. Dlotko and A. Rodr ´ ıguez-Bernal, �� , Mathematische Nachrichten no280 (2007), pp. 1643–1663

work page 2007
[7]

Baillon and Ph

J.-B. Baillon and Ph. Cl´ ement, �� , J. Funct. Anal. n o100 (1991), pp. 419–434

work page 1991
[8]

Bergh and J

J. Bergh and J. L¨ ofstr¨ om,�� . Grundlehren der mathematischen Wissenschaften n o223. Berlin-Heidelberg-New York: Springer-Verlag, 1976

work page 1976
[9]

Berestycki and L

H. Berestycki and L. Nirenberg, �� , Annales de l’Institut Henri Poincar´ e C, Analyse non lin´ eaire no9 (1992), pp. 497–572

work page 1992

Show all 61 references

[10]

Bertoin, ��

J. Bertoin, �� . Cambridge Tracts in Mathematics no121. Cambridge University Press, 1996

1996
[11]

Bochner, �� , Fundamenta Mathematicae no20 (1933), pp

S. Bochner, �� , Fundamenta Mathematicae no20 (1933), pp. 262–276

1933
[12]

Bochner and R.E

S. Bochner and R.E. Taylor, �� , Annals of Mathematics, second series, n o39 (1938), pp. 913– 944

1938
[13]

Br´ ezis, �� Collection Math´ ematiques Appliqu´ ees pour la Ma ˆ ıtrise

H. Br´ ezis, �� Collection Math´ ematiques Appliqu´ ees pour la Ma ˆ ıtrise. Paris: Masson, 1994. Translation: �� . Universi- text. New York, NY: Springer, 2011

1994
[14]

Calder´ on,�� , Studia Mathematica, Se- ria Specjalna n o1 (1963), pp

A.P. Calder´ on,�� , Studia Mathematica, Se- ria Specjalna n o1 (1963), pp. 31–34

1963
[15]

Chapuisat, �� , ESAIM: Proceedings n o18 (2007), pp

G. Chapuisat, �� , ESAIM: Proceedings n o18 (2007), pp. 87–98

2007
[16]

Chapuisat and E

G. Chapuisat and E. Grenier, �� , Communications in Partial Differential Equa- tions no30 (2005), pp. 1805–1816

2005
[17]

Chapuisat and R

G. Chapuisat and R. Joly, �� , Mathematical Models & Methods in Applied Sciences n o21 (2011), pp. 2155–2177

2011
[18]

Cwikel, �� (� p0(� 0)� �p1(� 1))θ,q, Proceedings of the American Mathemati- cal Society n o44 (1974), pp

M. Cwikel, �� (� p0(� 0)� �p1(� 1))θ,q, Proceedings of the American Mathemati- cal Society n o44 (1974), pp. 286–292

1974
[19]

da Prato and P

G. da Prato and P. Grisvard, �� , Journal de Math´ ematiques Pures et Appliqu´ ees IXe s´ erie, no54 (1975), pp. 305–387. 53

1975
[20]

da Prato and E

G. da Prato and E. Sinestrari, �� , Annali della Scuola Normale Superiore di Pisa. Classe di Scienze. Serie IV, no14 (1987), pp. 285–344

1987
[21]

Diestel and J.J

J. Diestel and J.J. Uhl, �� . Mathematical Surveys n o15. Provi- dence, R.I.: American Mathematical Society, 1977

1977
[22]

Di Nezza, G

E. Di Nezza, G. Palatucci and E. Valdinoci, �� , Bulletin des Sciences Math´ ematiques no136 (2012), pp. 521–573

2012
[23]

Dore and A

G. Dore and A. Venni, �� , Mathematische Zeitschrift no196 (1987), pp. 189-201

1987
[24]

Ducrot, P

A. Ducrot, P. Magal and K. Prevost, �� , Journal of Evolution Equations no10 (2010), pp. 263–291

2010
[25]

Ducrot, P

A. Ducrot, P. Magal and A. Thorel, �� , Nonlinear Differential Equations and Applications n o28 (2021), 49

2021
[26]

Dunford and J.T

N. Dunford and J.T. Schwartz, �� . Pure and Applied Mathematics Vol. 7. New York and London: Interscience Publishers, 1958

1958
[27]

Fackler, ��

S. Fackler, �� . PhD thesis, Fakult¨ at f¨ ur Mathematik und Wirtschaftswissenschaften der Universit¨ at Ulm, 2014

2014
[28]

Fackler, ��

S. Fackler, �� . In Operator semigroups meet complex analysis, harmonic anal- ysis and mathematical physics. Proceedings of the conference, Herrnhut, Ger- many, June 3–7, 2013. Operator Theory: Advance...

2013
[29]

Gallay and E

Th. Gallay and E. Risler, � �� , Differential and Integral Equations n o20 (2007), pp. 901–926

2007
[30]

T. Giletti, �� , Communications in Mathematical Sciences n o9 (2011), pp. 1177–1201

2011
[31]

Haque, N

Md. Haque, N. Ioku, T. Ogawa and R. Sato, �� , Differential and Integral Equations n o34 (2021), pp. 223–244

2021
[32]

Henry, �� , Lecture Notes In Mathematics no840

D. Henry, �� , Lecture Notes In Mathematics no840. Springer-Verlag, New York, 1981. 54

1981
[33]

Hyt¨ onen, J

T. Hyt¨ onen, J. van Neerven, M. Veraar and L. Weis,�� . Volume I. Martingales and Littlewood-Paley theory. Ergebnisse der Mathematik und ihrer Grenzgebiete n o3, Folge 63. Springer Verlag, 2016

2016
[34]

Hyt¨ onen, J

T. Hyt¨ onen, J. van Neerven, M. Veraar and L. Weis,�� . Volume III. Harmonic analysis and spectral theory. Ergebnisse der Mathematik und ihrer Grenzgebiete n o3, Folge 76. Springer Verlag, 2023

2023
[35]

Kalton and G

N.J. Kalton and G. Lancien, � �� p�� , Mathematische Zeitschrift no235 (2000), pp. 559–568

2000
[36]

Kalton and L

N.J. Kalton and L. Weis, �� , Math- ematische Annalen n o321 (2001), pp. 319–345

2001
[37]

Kato, �� , Archive for Rational Mechanics and Analysis n o58 (1975), pp

T. Kato, �� , Archive for Rational Mechanics and Analysis n o58 (1975), pp. 181–205

1975
[38]

Kre ˘ ın,�� , Doklady Akademii Nauk SSSR n o132 (1960), pp

S.G. Kre ˘ ın,�� , Doklady Akademii Nauk SSSR n o132 (1960), pp. 510–513. Translated in Soviet Mathematics, Doklady no1 (1960), pp. 586–589

1960
[39]

Kreuter, ��

M. Kreuter, �� . Master Thesis in Math- ematics. Universit¨ at Ulm, 2015

2015
[40]

Lions, �� , Annali della Scuola Normale Superiore di Pisa

J.L. Lions, �� , Annali della Scuola Normale Superiore di Pisa. Scienze Fisiche e Matematiche. III. Ser. n o13 (1960), pp. 389–403

1960
[41]

Lions, �� , Comptes Rendus de l’Acad´ emie des Sciences de Paris no251 (1960), pp

J.L. Lions, �� , Comptes Rendus de l’Acad´ emie des Sciences de Paris no251 (1960), pp. 1853–1855

1960
[42]

Lions and J

J.L. Lions and J. Peetre, �� , Publications Math´ ematiques de l’IHES no19 (1964), pp. 5–68

1964
[43]

Lunardi, �� , Mathematische Annalen no267 (1984), pp

A. Lunardi, �� , Mathematische Annalen no267 (1984), pp. 395–415

1984
[44]

Magal and S

P. Magal and S. Ruan, �� . Applied Mathematical Sciences n o201. Springer-Verlag, 2018

2018
[45]

Magal, O

P. Magal, O. Seydi and F.-B. Wang, �� , Journal of Mathematical Analysis and Applications n o479 (2019), pp. 450–481

2019
[46]

Mart ´ ınez and M

C. Mart ´ ınez and M. Celso,�� , Annali della Scuola Normale Superiore di Pisa. Classe di Scienze. Serie IV, n o3 (1991), pp. 443-454

1991
[47]

Matos and Ph

J. Matos and Ph. Souplet, �� , Houston Journal of Mathematics n o30 (2004), pp. 879–890. 55

2004
[48]

Meyers and J

N.G. Meyers and J. Serrin, � � � , Proceedings of the National Academy of Science no51 (1964), pp. 1055–1056

1964
[49]

Mich´ alek, D

M. Mich´ alek, D. Praˇ z´ ak and J. Slav ´ ık,�� , Communications in Pure and Applied Analysis n o16 (2017), pp. 1673–1695

2017
[50]

Muratov and M

C.B. Muratov and M. Novaga, �� , Communications in Mathematical Sciences n o6 (2008), pp. 799–826

2008
[51]

Muratov and M

C.B. Muratov and M. Novaga, �� , Calculus of Variations and Partial Differential Equa- tions no31 (2008), pp. 521–547

2008
[52]

Panfilov, H

A.V. Panfilov, H. Dierckx and V. Volpert, �� , Physica D n o399 (2019), pp. 1–34

2019
[53]

Pang and S.-L

L. Pang and S.-L. Wu, �� , Applied Mathematics Letters n o129 (2022), Article ID 107963, 6 p

2022
[54]

Pazy, ��

A. Pazy, �� . Applied Mathematical Sciences n o44. Springer-Verlag, 1983

1983
[55]

Reed and B

M. Reed and B. Simon, �� . New York-London: Academic Press, 1972

1972
[56]

E. Risler, �� , Annales de l’Institut Henri Poincar´ e, Analyse Non Lin´ eaire no25 (2008), pp. 381–424

2008
[57]

Roidos, �� , Journal of Functional Analysis no266 (2014), pp

N. Roidos, �� , Journal of Functional Analysis no266 (2014), pp. 6938–6953

2014
[58]

Sinestrari, �� , Journal of Mathematical Analysis and Applications no107 (1985), pp

E. Sinestrari, �� , Journal of Mathematical Analysis and Applications no107 (1985), pp. 16–66

1985
[59]

Suguro, �� , Journal of Evolution Equations no21 (2021), pp

T. Suguro, �� , Journal of Evolution Equations no21 (2021), pp. 4599–4618

2021
[60]

Triebel, ��

H. Triebel, �� . North- Holland Mathematical Library. Vol. 18. Amsterdam - New York - Oxford: North-Holland Publishing Company, 1978

1978
[61]

Venni, � ��

A. Venni, � �� . In Functional analysis and related topics, Kyoto, 1991. Lecture Notes in Math- ematics no1540, pp. 381–387. Springer, Berlin, 1993

1991
[62]

Xin, �� , SIAM Review n o42 (2000), pp.161–230

J. Xin, �� , SIAM Review n o42 (2000), pp.161–230. 56

2000

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [1]

Indeed, an initial data as �0(�) = sin( �2) exhibits very fast oscillation at infinity

�� 2 ul� w(� �� ). Indeed, an initial data as �0(�) = sin( �2) exhibits very fast oscillation at infinity. These oscillations are homogenized by the heat semigroup (defined via the convolution by its kernel) and, as soon as � is positive, �(�) = �∆t�0 must converge to 0 as � goes to �� , see Figure 2. Thus ��(�)� �0�L2 ul d...

work page

[2] [2]

This excludes the problematic � �� sin(�2) and can be seen as replacing � 0(� �� ) by the space of uniformly continuous functions

�� 2 ul� s(� �� )� The space � 2 ul� s(� �� ) is a stronger version of the uniformly local space where the translation � �� (� � �) is continuous for all functions �, see Section 3. This excludes the problematic � �� sin(�2) and can be seen as replacing � 0(� �� ) by the space of uniformly continuous functions. It is ...

work page

[3] [3]

Indeed, we can use the oscillations in the � direction to construct an initial data for which the heat semigroup is not continuous at � = 0 +, see Figure

�� 2 ul� s(� � �2((� 1�1)))� In our cylindrical set- ting, choosing the strong version of the uniformly local spaces is not sufficient to define a proper Cauchy problem, even for the heat equation in � � (� 1�1). Indeed, we can use the oscillations in the � direction to construct an initial data for which the heat semigroup...

work page

[4] [4]

abstract parabolic operator

This shows that the strong version of the uniformly local spaces is not so helpful in a general context of vector valued functions. To our point of view, one of the main interests of the present paper will be to clarify the issues raised in the above remarks. Figure 2: �� sin(�2) �� ...

work page

[5] [6]

Arrieta, J.W

J.M. Arrieta, J.W. Cholewa, T. Dlotko and A. Rodr ´ ıguez-Bernal, �� , Mathematische Nachrichten no280 (2007), pp. 1643–1663

work page 2007

[6] [7]

Baillon and Ph

J.-B. Baillon and Ph. Cl´ ement, �� , J. Funct. Anal. n o100 (1991), pp. 419–434

work page 1991

[7] [8]

Bergh and J

J. Bergh and J. L¨ ofstr¨ om,�� . Grundlehren der mathematischen Wissenschaften n o223. Berlin-Heidelberg-New York: Springer-Verlag, 1976

work page 1976

[8] [9]

Berestycki and L

H. Berestycki and L. Nirenberg, �� , Annales de l’Institut Henri Poincar´ e C, Analyse non lin´ eaire no9 (1992), pp. 497–572

work page 1992