Legalarxiv.

research collection · exact source version

Accelerating Scientific Research with Gemini: Case Studies and Common Techniques

David P. Woodruff, Vincent Cohen-Addad, Lalit Jain, and 33 others

All 36 authors and source roles
  1. David P. Woodruffauthor · individual_section_contributor · advanced_model_contributor
  2. Vincent Cohen-Addadauthor · individual_section_contributor · advanced_model_contributor
  3. Lalit Jainauthor · advanced_model_contributor
  4. Jieming Maoauthor · advanced_model_contributor
  5. Song Zuoauthor · individual_section_contributor · advanced_model_contributor
  6. MohammadHossein Bateniauthor · individual_section_contributor
  7. Simina Brânzeiauthor · individual_section_contributor
  8. Michael P. Brennerauthor · individual_section_contributor
  9. Lin Chenauthor · individual_section_contributor
  10. Ying Fengauthor · individual_section_contributor
  11. Lance Fortnowauthor · individual_section_contributor
  12. Gang Fuauthor · individual_section_contributor
  13. Ziyi Guanauthor · individual_section_contributor
  14. Zahra Hadizadehauthor · individual_section_contributor
  15. Mohammad T. Hajiaghayiauthor · individual_section_contributor
  16. Mahdi JafariRavizauthor · individual_section_contributor
  17. Adel Javanmardauthor · individual_section_contributor
  18. Karthik C. S.author · individual_section_contributor
  19. Ken-ichi Kawarabayashiauthor · individual_section_contributor
  20. Ravi Kumarauthor · individual_section_contributor
  21. Silvio Lattanziauthor · individual_section_contributor
  22. Euiwoong Leeauthor · individual_section_contributor
  23. Yi Liauthor · individual_section_contributor
  24. Ioannis Panageasauthor · individual_section_contributor
  25. Dimitris Paparasauthor · individual_section_contributor
  26. Benjamin Przybockiauthor · individual_section_contributor
  27. Bernardo Subercaseauxauthor · individual_section_contributor
  28. Ola Svenssonauthor · individual_section_contributor
  29. Shayan Taherijamauthor · individual_section_contributor
  30. Xuan Wuauthor · individual_section_contributor
  31. Eylon Yogevauthor · individual_section_contributor
  32. Morteza Zadimoghaddamauthor · individual_section_contributor
  33. Samson Zhouauthor · individual_section_contributor
  34. Yossi Matiasauthor
  35. James Manyikaauthor
  36. Vahab Mirrokniauthor · individual_section_contributor · advanced_model_contributor

Authors marked as individual-section contributors are responsible only for the correctness of their attributed section; model-variant contribution is a separate source role.

v3 · Mar 06, 2026 · Legacy record · quality not assessed

Contribution
Selected source-authored cases report material model contributions within expert-directed theoretical research.
Limitation
The collection does not estimate reliability, frequency, causal acceleration, comparative quality, or generalization.

Abstract

Source-authored

Recent advances in large language models (LLMs) have opened new avenues for accelerating scientific research. While models are increasingly capable of assisting with routine tasks, their ability to contribute to novel, expert-level mathematical discovery is less understood.

We present a collection of case studies demonstrating how researchers have successfully collaborated with advanced AI models, specifically Google’s Gemini-based models (in particular Gemini Deep Think and its advanced variants), to solve open problems, refute conjectures, and generate new proofs across diverse areas in theoretical computer science, as well as other areas such as economics, optimization, and physics.

Based on these experiences, we extract common techniques for effective human-AI collaboration in theoretical research, such as iterative refinement, problem decomposition, and cross-disciplinary knowledge transfer. While the majority of our results stem from this interactive, conversational methodology, we also highlight specific instances that push beyond standard chat interfaces.

These include deploying the model as a rigorous adversarial reviewer to detect subtle flaws in existing proofs, and embedding it within a “neuro-symbolic” loop that autonomously writes and executes code to verify complex derivations. Together, these examples highlight the potential of AI not just as a tool for automation, but as a versatile, genuine partner in the creative process of scientific discovery.

Inspect the identified source ↗

Evidence map

1 mapped claim · v3

Follow a claim to its source passages, reported evidence and recorded checks.

Claim 1The source presents a three-item, two-agent submodular-welfare instance whose calculated residual gain violates the cited Move-vs-Copy conjecture.

platform interpreted · mathematical counterexample

  1. Source passages

    The places this claim is traced to.

  2. Reported evidence

    A valuation table and six-permutation calculation report LHS = 122.6/6 and RHS = 121.8/6.

    Access: public source html

  3. Recorded check

    The public source supplies the construction and reports independent arithmetic checking by the human researchers.

    Platform recheck: not performed

Limits of this claim

  • The platform has not independently re-run or formally verified the calculation.
  • The example proves existence and refutes the stated inequality; it is not evidence about model success frequency or comparative research acceleration.
  • The full prompt, sampling settings, rejected attempts, and attempt denominator are unavailable.
  • The result does not resolve the tight approximation bound discussed by the source.
Contributors & provenance

Gemini-based model, as reported by the section authors

Selected small dimensions, proposed the valuations and epsilon, and generated the verification calculations without further interaction, according to the source.

Responsibility: none

Morteza Zadimoghaddam and Vahab Mirrokni

Defined the task and search parameters, independently checked the arithmetic, adopted the result, and accepted section-scoped responsibility.

Responsibility: section correctness as declared by source

Evaluation status
source supported not independently reverified
Recorded by
agent:legalarxiv-pilot
As of
2026-08-28T00:00:00+02:00
Claim identifier
claim:arxiv-2602-03837-v3:move-vs-copy-counterexample
Exact version
version:arxiv-2602-03837-v3

Inspect the complete machine-readable record

Quality screening

Legacy record · quality not assessed

This record predates the evidence-based admission gate. Its presence in the archive does not establish that it passed the current quality policy.

Screening scope and limitations

  • This record predates the quality gate and has no assessment under the current admission policy.

Inspect the current screening receipt and assessment references ↗

Detailed record status

Status remains facet-specific. Deposit, reconstruction, evaluation, peer review, currentness, and completeness are not compressed into one score.

deposit
source preprint v3The exact source manuscript is arXiv version 3, submitted 6 March 2026.
archive record
pilot reconstructionThe platform reconstructed this record from public materials; authors have not attested it here.
validation
heterogeneous claim specificValidation is recorded separately for each episode; no aggregate correctness score is provided.
peer review
not assertedThis pilot makes no work-wide claim of independent peer review.
currentness
checked through 2026 08 28 for pilot sourcesPilot source and version relationships were reviewed through the as-of date; this is not continuous monitoring.
Source sections · 17
  1. 3.1

    Online Algorithms: Submodular Welfare

    Morteza Zadimoghaddam, Vahab Mirrokni

    Mapped note
  2. 3.2

    Cryptography: AI-Assisted Bug Detection in SNARGs

    Ziyi Guan, David P. Woodruff, Eylon Yogev

    Mapped note
  3. 4.1Mapped note
  4. 4.2Mapped note
  5. 4.3Mapped note
  6. 5.1Mapped note
  7. 6.1

    Physics: Cosmic String Spectra

    Michael P. Brenner, Vincent Cohen-Addad, David P. Woodruff

    Mapped note
  8. 7.1

    Graph Theory: Biclique Partitions

    Benjamin Przybocki, Bernardo Subercaseaux

    Source index
  9. 7.2

    Query Complexity: Local Search on General Graphs

    Simina Brânzei, Ioannis Panageas, Dimitris Paparas

    Source index
  10. 7.3

    Robust Coresets

    Yi Li, David P. Woodruff, Xuan Wu

    Source index
  11. 7.4Source index
  12. 7.5Source index
  13. 8.1Mapped note
  14. 8.2

    NP-hardness: Ratio Difference Maximization (RDM)

    Ravi Kumar and Silvio Lattanzi, on behalf of coauthors

    Source index
  15. 8.3Source index
  16. 8.4Source index
  17. 8.5

    Networked Information Aggregation for Binary Classification

    MohammadHossein Bateni, Zahra Hadizadeh, MohammadTaghi Hajiaghayi, Mahdi JafariRaviz, and Shayan Taherijam

    Source index
Source edition, rights & technical exports
As of
2026-08-28T00:00:00+02:00
Schema
legalarxiv.record.v3
Packet digest
sha256:91cd792ecb69363e30309e1950dde4ccd49da2f672b477477d6b04581f578e0c

The source license and applicable law control; this field is descriptive.

Research updates

A monthly digest of selected legal AI papers, datasets and benchmarks, plus published research checks.