Pith. sign in

REVIEW 3 major objections 2 minor 36 references

Existence of the longest arcs for left-invariant three-dimensional contact sub-Lorentzian structures

T0 review · 3 major / 2 minor · reviewed 2026-07-15 · grok-4.5

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

Pith's one-line read Longest arcs exist for some left-invariant three-dimensional contact sub-Lorentzian structures, under sufficient conditions on solvable Lie groups and the universal cover of SL(2, R).

desk verdict We only have the abstract for the sub-Lorentzian existence paper; the supplied body is a different AVSR manuscript, so the claims cannot be audited. read the letter →

arxiv 2603.07262 v2 pith:UCZFGWTV submitted 2026-03-07 math.DG math.MGmath.OC

A.V. Podobryaev This is my paper · ORCID
classification math.DGmath.MGmath.OC MSC 53C1749J1522E2553C50
keywords sub-Lorentziangeometrylongestarcsoptimalcontrolleft-invariantstructurescontactLiegroupsexistenceofmaximizers
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

Finding longest arcs for sub-Lorentzian structures is an optimal-control problem with an unbounded control set and a concave cost, so existence of a maximizer is not automatic. The paper settles existence for a class of left-invariant three-dimensional contact sub-Lorentzian structures whose classification is already known. It also proposes sufficient conditions that guarantee longest arcs for left-invariant (sub-)Lorentzian structures on solvable Lie groups and on the universal cover of SL(2, R). A reader who works with geometric control or Lorentzian geometry on Lie groups gets concrete existence results rather than only necessary conditions from first-order optimality.

What carries the argument

Reduction, via the known classification of left-invariant three-dimensional contact sub-Lorentzian structures, of the existence question to proposed sufficient conditions on solvable Lie groups and on the universal cover of SL(2, R); those conditions make the unbounded-control concave problem admit maximizers.

What would settle it

Produce a left-invariant three-dimensional contact sub-Lorentzian structure covered by the classification for which no longest arc exists between some pair of points, or exhibit a solvable Lie group (or the cover of SL(2, R)) where the stated sufficient conditions hold yet a maximizer fails, or where a maximizer fails while the conditions are the only obstruction claimed.

Watch

Extended reading notes

Core claim

For some left-invariant three-dimensional contact sub-Lorentzian structures (with a known classification), longest arcs exist as solutions of the associated unbounded-control concave optimal-control problem; more generally, the authors give sufficient conditions ensuring existence of longest arcs for left-invariant (sub-)Lorentzian structures on solvable Lie groups and on the universal cover of SL(2, R).

Load-bearing premise

The known classification of left-invariant three-dimensional contact sub-Lorentzian structures, together with left-invariance and the contact assumption, is enough to reduce existence to the proposed conditions on solvable groups and the universal cover of SL(2, R).

Editorial extensions

If this is right

  • On the listed groups and structures the longest-arc problem is well-posed: maximizers exist, not merely formal critical points.
  • Classification-based case analysis can settle existence without first solving every boundary-value problem.
  • Existence statements extend from ordinary Lorentzian structures to contact sub-Lorentzian left-invariant structures in dimension three.
  • Under the proposed conditions one may search for maximizers on solvable groups and on the cover of SL(2, R) knowing they exist.

Reading between the lines

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

  • The same sufficient conditions may carry over to higher-dimensional left-invariant contact sub-Lorentzian structures once usable normal forms exist.
  • A solvable group that violates the conditions would give a concrete class of counter-examples where longest arcs need not exist despite left-invariance.
  • The role of the contact assumption in securing existence may suggest parallel results for related sub-Riemannian energy or length problems with unbounded controls.
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

3 major / 2 minor

Summary. The abstract claims that existence of longest arcs (optimal curves) is established for some left-invariant three-dimensional contact sub-Lorentzian structures, using a known classification, and that sufficient conditions are given for left-invariant (sub-)Lorentzian structures on solvable Lie groups and on the universal cover of SL(2,R). The problem is framed as an optimal-control problem with unbounded controls and a concave cost, for which existence is nontrivial. The supplied full manuscript body, however, is an unrelated paper on context-aware audio-visual speech recognition (VASR / AV-CoT, multimodal LLMs, CER benchmarks). No statements of the sufficient conditions, no reduction from the classification, no control-theoretic arguments, and no proofs of existence appear in the provided text.

Significance. If the mathematical claims in the abstract hold, the work would address a genuine gap: existence of maximizers for sub-Lorentzian length with unbounded controls is not automatic, and a clean treatment for the classified left-invariant 3D contact family plus sufficient conditions on solvable groups and the universal cover of SL(2,R) would be a useful contribution to geometric control and sub-Lorentzian geometry. That significance cannot be assessed from the supplied body, which contains none of the claimed results.

major comments (3)
  1. The full manuscript text does not match the title, abstract, or arXiv identifier of the paper under review. The body is a complete, self-contained paper on CA VSR / VASR (Audio-Visual Chain-of-Thought, data pipeline, Chinese-LiPS and VASR test sets, Tables 2–3, CER results). No sub-Lorentzian structures, Lie groups, longest arcs, or optimal-control existence arguments appear. The central existence claim is therefore uncheckable from the submission as provided.
  2. Because the body is the wrong paper, the abstract’s load-bearing reduction—that left-invariance, contact, and the known classification of 3D contact sub-Lorentzian structures reduce existence to the proposed sufficient conditions on solvable groups and the universal cover of SL(2,R)—cannot be verified. Degenerate, non-contact, or non-solvable cases may or may not be covered; the manuscript supplies no lemmas or statements with which to decide.
  3. No equations, theorems, or proofs related to the claimed existence result are present. A referee report on soundness of the longest-arc existence argument is impossible until the correct mathematical manuscript is supplied.
minor comments (2)
  1. The abstract alone is internally coherent as a problem statement, but it is not a substitute for the missing theorems and proofs.
  2. The provided body (VASR) has its own presentation issues (e.g., OCR/garbled math tokens in displayed equations, figure placeholders), but those are irrelevant to the paper that was supposed to be reviewed.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity can be exhibited: the supplied full text is an unrelated AVSR paper, and the target math abstract shows no self-definitional or fitted-as-prediction reduction.

full rationale

The load-bearing claim of arXiv:2603.07262 is existence of longest arcs for some left-invariant 3D contact sub-Lorentzian structures (via a known classification) plus sufficient conditions on solvable Lie groups and the universal cover of SL(2,R). The CACHEABLE full manuscript is not that paper; it is an unrelated multimodal speech-recognition work (VASR / Context AVSR). Consequently no equations, reductions, self-citations, uniqueness theorems, or control-theoretic arguments from 2603.07262 are available to quote. From the abstract alone there is no self-definitional loop (existence is not defined as the conditions), no fitted parameter renamed as a prediction, and no uniqueness imported from the authors. Per the hard rules, circularity may be claimed only when a specific reduction can be quoted; absence of the correct body is not circularity. Score 0 with empty steps is therefore the only honest outcome.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

Abstract-only review of a pure existence paper in geometric control. No free parameters or fitted constants appear. Background axioms are standard differential geometry / Lie theory / optimal control; the paper invents no new physical entities. The main unproved (from our side) inputs are the known classification of 3D contact left-invariant sub-Lorentzian structures and the standard setup of left-invariant sub-Lorentzian optimal control.

assumptions (3)
  • domain assumption Classification of left-invariant three-dimensional contact sub-Lorentzian structures is known and can be used as the starting point for existence analysis.
    Invoked in the abstract as the setting in which the existence question is solved; the classification itself is prior literature, not proved here.
  • domain assumption Finding longest arcs is an optimal control problem with unbounded control set and concave cost functional, for which existence is nontrivial.
    Stated as the reason the problem is open; standard in geometric control of sub-Lorentzian structures.
  • standard math Left-invariant (sub-)Lorentzian structures on solvable Lie groups and on the universal cover of SL(2,R) admit analysis via Lie-algebraic data.
    Standard reduction for left-invariant geometric structures on Lie groups.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Existence of the longest arcs for left-invariant three-dimensional contact sub-Lorentzian structures." pith.science (2026). https://pith.science/paper/UCZFGWTV

@misc{pith2026260307262,
  author       = {Pith},
  title        = {Pith review of: Existence of the longest arcs for left-invariant three-dimensional contact sub-Lorentzian structures},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UCZFGWTV}},
  note         = {Machine review of arXiv:2603.07262}
}
read the original abstract

The problem of finding optimal curves (the longest arcs) for sub-Lorentzian structures is an optimal control problem with an unbounded control set and a concave cost functional. The question of existence of an optimal solution is nontrivial for such problems. We solve here this question for some left-invariant three-dimensional contact sub-Lorentzian structures, whose classification is known. We propose sufficient conditions for the existence of the longest arcs for left-invariant (sub-)Lorentzian structures on solvable Lie groups and on the universal cover of the Lie group SL(2, R).

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 13 linked inside Pith

  1. [1]

    single-modality dominance

    Introduction Automatic Speech Recognition (ASR) [1, 2, 3] has achieved re- markable progress in recent years, largely driven by scaling up training data and model capacity. However, such audio-only systems still struggle in scenarios requiring context-aware dis- ambiguation, such as recognizing homophones, named entities, and domain-specific terms. This l...

    arXiv 2026

  2. [2]

    Overall Architecture V ASR aims to achieve CA VSR in diverse real-world scenarios through structured multimodal reasoning

    Framework 2.1. Overall Architecture V ASR aims to achieve CA VSR in diverse real-world scenarios through structured multimodal reasoning. Equipped with A V- CoT, V ASR reformulates CA VSR task as a structured Percep- tion–Reasoning–Transcription pipeline. As illustrated in Fig- ure 2, given multimodal inputs consisting of video�and au- dio signals�, the f...

  3. [3]

    Data Pipeline To overcome the scarcity of open-source datasets for CA VSR, we design an automated, scalable data curation pipeline

    Dataset 3.1. Data Pipeline To overcome the scarcity of open-source datasets for CA VSR, we design an automated, scalable data curation pipeline. The objective is to identify challenging, visually rich data in lin- guistic ambiguity and to generate high-quality multimodal an- notations. As shown on the right of Figure 2, we split the data process into two ...

  4. [4]

    w/o A V-CoT

    Experiments 4.1. Datasets We utilize open-source chinese datasets, including MEIJU [28] and MER25 [29], ch-sims-v2 [30], and the Chinese-LiPS [16] training set. All open-source data are uniformly processed using our proposed data processing pipeline, totaling approximately 230 hours. We use Chinese-LiPS test set as the open-source test set. For general-sc...

  5. [5]

    We propose V ASR, which leverages A V- CoT to explicitly model the transcription process to address the single-modality dominance

    Conclusion and Limitation In this work, we introduce the CA VSR task, extending A VSR beyond lip-reading to leverage rich visual context for resolving linguistic ambiguities. We propose V ASR, which leverages A V- CoT to explicitly model the transcription process to address the single-modality dominance. In addition, we develop and release a scalable data...

  6. [6]

    They play no role in methodology, experimenta- tion, interpretation, or the production of scientific results

    Generative AI Use Disclosure Generative AI tools are used for writing refinement and data generation. They play no role in methodology, experimenta- tion, interpretation, or the production of scientific results. The authors bear full intellectual responsibility for all content in this manuscript. All authors agree with the submission of this paper

  7. [7]

    Qwen3-asr technical report,

    X. Shi, X. Wang, Z. Guo, Y . Wang, P. Zhang, X. Zhang, Z. Guo, H. Hao, Y . Xi, B. Yang, J. Xu, J. Zhou, and J. Lin, “Qwen3-asr technical report,”CoRR, vol. abs/2601.21337, 2026

    arXiv 2026

  8. [8]

    Fireredasr: Open- source industrial-grade mandarin speech recognition mod- els from encoder-decoder to LLM integration,

    K. Xu, F. Xie, X. Tang, and Y . Hu, “Fireredasr: Open- source industrial-grade mandarin speech recognition mod- els from encoder-decoder to LLM integration,”CoRR, vol. abs/2501.14350, 2025

    arXiv 2025

Show all 36 references
  1. [9]

    Paraformer: Fast and accurate parallel transformer for non-autoregressive end-to- end speech recognition,

    Z. Gao, S. Zhang, I. McLoughlin, and Z. Yan, “Paraformer: Fast and accurate parallel transformer for non-autoregressive end-to- end speech recognition,” inProc.Interspeech, 2022, pp. 2063– 2067

    2022

  2. [10]

    Omni- avsr: Towards unified multimodal speech recognition with large language models,

    U. Cappellazzo, X. Liu, P. Ma, S. Petridis, and M. Pantic, “Omni- avsr: Towards unified multimodal speech recognition with large language models,”CoRR, vol. abs/2511.07253, 2025

    2025

  3. [11]

    AD- A VSR: asymmetric dual-stream enhancement for robust audio- visual speech recognition,

    J. Xue, X. Liu, X. Wu, X. Yin, D. Huang, and F. Yu, “AD- A VSR: asymmetric dual-stream enhancement for robust audio- visual speech recognition,”CoRR, vol. abs/2508.07608, 2025

    2025 arXiv

  4. [12]

    MLCA-A VSR: multi-layer cross attention fusion based audio-visual speech recognition,

    H. Wang, P. Guo, P. Zhou, and L. Xie, “MLCA-A VSR: multi-layer cross attention fusion based audio-visual speech recognition,” in Proc.ICASSP, 2024, pp. 8150–8154

    2024

  5. [13]

    Adaptive audio-visual speech recognition via matryoshka-based multimodal llms,

    U. Cappellazzo, M. Kim, and S. Petridis, “Adaptive audio-visual speech recognition via matryoshka-based multimodal llms,” CoRR, vol. abs/2503.06362, 2025

    2025 arXiv

  6. [14]

    Mms-llama: Ef- ficient llm-based audio-visual speech recognition with minimal multimodal speech tokens,

    J. H. Yeo, H. Rha, S. J. Park, and Y . M. Ro, “Mms-llama: Ef- ficient llm-based audio-visual speech recognition with minimal multimodal speech tokens,” inFindings of the Association for Computational Linguistics, ACL 2025, Vienna, Austria, July 27 - August 1, 2025, ser. Findin...

    2025

  7. [15]

    Zero- avsr: Zero-shot audio-visual speech recognition with llms by learning language-agnostic speech representations,

    J. H. Yeo, M. Kim, C. W. Kim, S. Petridis, and Y . M. Ro, “Zero- avsr: Zero-shot audio-visual speech recognition with llms by learning language-agnostic speech representations,”CoRR, vol. abs/2503.06273, 2025

    2025 arXiv

  8. [16]

    Qwen2.5-omni technical report,

    J. Xu, Z. Guo, J. He, H. Hu, T. He, S. Bai, K. Chen, J. Wang, Y . Fan, K. Dang, B. Zhang, X. Wang, Y . Chu, and J. Lin, “Qwen2.5-omni technical report,”CoRR, vol. abs/2503.20215, 2025

    2025 arXiv

  9. [17]

    Qwen3-omni technical report,

    J. Xu, Z. Guo, H. Hu, Y . Chu, X. Wang, J. Heet al., “Qwen3-omni technical report,”CoRR, vol. abs/2509.17765, 2025

    2025 arXiv

  10. [18]

    Intern-s1: A scientific multimodal foundation model,

    L. Bai, Z. Cai, Y . Cao, M. Cao, W. Cao, C. Chenet al., “Intern-s1: A scientific multimodal foundation model,”CoRR, vol. abs/2508.15763, 2025

    2025

  11. [19]

    Minicpm-v: A GPT-4V level MLLM on your phone,

    Y . Yao, T. Yu, A. Zhang, C. Wang, J. Cui, H. Zhu, T. Cai, H. Li, W. Zhao, Z. He, Q. Chen, H. Zhou, Z. Zou, H. Zhang, S. Hu, Z. Zheng, J. Zhou, J. Cai, X. Han, G. Zeng, D. Li, Z. Liu, and M. Sun, “Minicpm-v: A GPT-4V level MLLM on your phone,” CoRR, vol. abs/2408.01800, 2024

    2024 arXiv

  12. [20]

    Gemini 2.5: Pushing the frontier with ad- vanced reasoning, multimodality, long context, and next genera- tion agentic capabilities,

    G. Comanici, E. Bieber, M. Schaekermann, I. Pasupat, N. Sachdevaet al., “Gemini 2.5: Pushing the frontier with ad- vanced reasoning, multimodality, long context, and next genera- tion agentic capabilities,”CoRR, vol. abs/2507.06261, 2025

    2025 arXiv

  13. [21]

    Slidespeech: A large scale slide-enriched audio-visual corpus,

    H. Wang, F. Yu, X. Shi, Y . Wang, S. Zhang, and M. Li, “Slidespeech: A large scale slide-enriched audio-visual corpus,” inProc.ICASSP, 2024, pp. 11 076–11 080

    2024

  14. [22]

    Chinese- lips: A chinese audio-visual speech recognition dataset with lip- reading and presentation slides,

    J. Zhao, Y . Jia, S. Wang, J. Zhou, H. Wang, and Y . Qin, “Chinese- lips: A chinese audio-visual speech recognition dataset with lip- reading and presentation slides,” inProc.ICME, 2025, pp. 1–6

    2025

  15. [23]

    How2: A large-scale dataset for mul- timodal language understanding,

    R. Sanabria, O. Caglayan, S. Palaskar, D. Elliott, L. Barrault, L. Specia, and F. Metze, “How2: A large-scale dataset for mul- timodal language understanding,”CoRR, vol. abs/1811.00347, 2018

    2018 arXiv

  16. [24]

    A V ATAR: unconstrained audiovisual speech recog- nition,

    V . Gabeur, P. H. Seo, A. Nagrani, C. Sun, K. Alahari, and C. Schmid, “A V ATAR: unconstrained audiovisual speech recog- nition,” inProc.Interspeech, 2022, pp. 2818–2822

    2022

  17. [25]

    Slideavsr: A dataset of paper explanation videos for audio-visual speech recog- nition,

    H. Wang, S. Kurita, S. Shimizu, and D. Kawahara, “Slideavsr: A dataset of paper explanation videos for audio-visual speech recog- nition,”CoRR, vol. abs/2401.09759, 2024

    2024 arXiv

  18. [26]

    Multi-modality speech recognition driven by background visual scenes,

    C. Luo, Y . Liu, W. Sun, and Z. Sun, “Multi-modality speech recognition driven by background visual scenes,” inProc.ICASSP, 2024, pp. 10 926–10 930

    2024

  19. [27]

    Avformer: Injecting vision into frozen speech models for zero-shot av-asr,

    P. H. Seo, A. Nagrani, and C. Schmid, “Avformer: Injecting vision into frozen speech models for zero-shot av-asr,” inProc.CVPR, 2023, pp. 22 922–22 931

    2023

  20. [28]

    Lip reading sentences in the wild,

    J. S. Chung, A. W. Senior, O. Vinyals, and A. Zisserman, “Lip reading sentences in the wild,” inProc.CVPR, 2017, pp. 3444– 3453

    2017

  21. [29]

    A cascade sequence-to-sequence model for chinese mandarin lip reading,

    Y . Zhao, R. Xu, and M. Song, “A cascade sequence-to-sequence model for chinese mandarin lip reading,” inProc.MMAsia, 2019, pp. 32:1–32:6

    2019

  22. [30]

    CN-CVS: A mandarin audio- visual dataset for large vocabulary continuous visual to speech synthesis,

    C. Chen, D. Wang, and T. F. Zheng, “CN-CVS: A mandarin audio- visual dataset for large vocabulary continuous visual to speech synthesis,” inProc.ICASSP, 2023, pp. 1–5

    2023

  23. [31]

    V oxceleb2: Deep speaker recognition,

    J. S. Chung, A. Nagrani, and A. Zisserman, “V oxceleb2: Deep speaker recognition,” inProc.Interspeech, 2018, pp. 1086–1090

    2018

  24. [32]

    Robust speech recognition via large-scale weak su- pervision,

    A. Radford, J. W. Kim, T. Xu, G. Brockman, C. McLeavey, and I. Sutskever, “Robust speech recognition via large-scale weak su- pervision,” inProc.ICML, ser. Proceedings of Machine Learning Research, vol. 202, 2023, pp. 28 492–28 518

    2023

  25. [33]

    Qwen2.5-vl technical report,

    S. Bai, K. Chen, X. Liu, J. Wang, W. Ge, S. Song, K. Dang, P. Wang, S. Wang, J. Tang, H. Zhong, Y . Zhu, M. Yang, Z. Li, J. Wan, P. Wang, W. Ding, Z. Fu, Y . Xu, J. Ye, X. Zhang, T. Xie, Z. Cheng, H. Zhang, Z. Yang, H. Xu, and J. Lin, “Qwen2.5-vl technical report,”CoRR, vol. a...

    2025 arXiv

  26. [34]

    MEIJU - the 1st multimodal emotion and intent joint understand- ing challenge,

    R. Liu, X. Xing, Z. Lian, H. Li, B. W. Schuller, and H. Zuo, “MEIJU - the 1st multimodal emotion and intent joint understand- ing challenge,” inProc.ICASSP, 2025, pp. 1–2

    2025

  27. [35]

    MER 2025: When affective computing meets large language models,

    Z. Lian, R. Liu, K. Xu, B. Liu, X. Liu, Y . Zhang, X. Liu, Y . Li, Z. Cheng, H. Zuo, Z. Ma, X. Peng, X. Chen, Y . Li, E. Cam- bria, G. Zhao, B. W. Schuller, and J. Tao, “MER 2025: When affective computing meets large language models,”CoRR, vol. abs/2504.19423, 2025

    2025 arXiv

  28. [36]

    Make acous- tic and visual cues matter: CH-SIMS v2.0 dataset and av-mixup consistent module,

    Y . Liu, Z. Yuan, H. Mao, Z. Liang, W. Yanget al., “Make acous- tic and visual cues matter: CH-SIMS v2.0 dataset and av-mixup consistent module,” inProc.ICMI, 2022, pp. 247–258

    2022

Pith tools

Reviewed July 15, 2026 · model on record in the stance chip above.