Pith. sign in

REVIEW 3 minor 50 references

Same physical state, different collective dynamics: state encodings select synchronization outcomes in language-model agents

T0 review · 0 major / 3 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read How a state is described to a language-model agent is part of its effective interaction law, not a neutral interface: moment encodings synchronized GPT populations 6/6 seeds and histogram encodings 0/6, with the effect reversed in Claude.

desk verdict Careful, well-controlled demonstration that observation encodings change collective LLM-agent dynamics; the API-stationarity worry is real but the controls contain it. read the letter →

arxiv 2608.06968 v1 pith:JGPXNVXL submitted 2026-08-07 physics.soc-ph cs.AIcs.CY

Takahiro Ezaki , Naoto Imura , Katsuhiro Nishinari This is my paper · ORCID
classification physics.soc-phcs.AIcs.CY PACS 05.45.Xt89.65.-f
keywords language-modelagentsstateencodingobservationmapsynchronizationcollectivedynamicsphaseoscillatorsperformativefeedbackmulti-agentsystems
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

Coupled language-model agents never act on an environment directly; they act on a text description of it, and this paper argues that the description — the observation map — is part of the agents' effective interaction law rather than a neutral interface. To test this it built a minimal circular-synchronization system in which each agent sees only an encoding of its neighbors' relative phases and chooses to advance, stay, or retard, while a deterministic engine applies the coupling. In GPT, encoding the same physical field as low-order circular moments locked the population in 6 of 6 seeds at positive coupling, while two histogram encodings locked in 0 of 6; in Claude the encoding effect replicated but reversed direction. On 48 frozen fields replayed from real runs, re-encoding alone moved each agent's action probabilities by a mean total-variation distance of 0.344, against a 0.092 test-retest floor, an effect seen in all three model families tested. A sympathetic reader takes the conclusion to be that the observation serializer is a model-dependent component of collective dynamics, so evaluations of LLM agent populations must report and re-validate it together with the model.

What carries the argument

The load-bearing object is the circular-synchronization assay, a population of $N = 17$ phase oscillators whose only coupling is the sampled action of a pretrained language model: the engine applies $x_i(t+1) = x_i(t) + \omega_i + K f_i(t)$ with $f_i(t) \in \{-1, 0, +1\}$, and the model sees only a text serialization of the relative-phase field. The companion object is the response operator $p_R(\rho)$, the advance/stay/retard probability distribution measured from repeated calls at a fixed field $\rho$ under observation map $R$; the three maps are the first three circular moments, a 24-bin histogram labelled by bin centers, and the same 24 masses labelled by bin intervals. The decisive mechanism is identical-field replay, which takes 48 frozen fields generated by the closed loop and re-encodes each under all three maps, so any response difference can only be attributed to the encoding. Supporting machinery includes the order parameters $r_m(t)$, the Fourier decomposition of the signed response curve as a data-driven phase interaction function, and the $K = 0$ control arm in which the coupling term vanishes, making trajectories coincide exactly while actions still differ.

What would settle it

A prospective replication that pins provider model versions and records all sampling parameters, reruns the identical-field replay on fresh fields, and finds the between-encoding total-variation distance fallen to the within-encoding test-retest floor would falsify the claim that the encoding itself moves the operator. A second decisive test is the operator-swap the paper leaves open: if the closed-loop trajectories were unchanged when the action distribution of one encoding is substituted for another on a fixed state sequence, the measured microscopic differences could not be what selects the collective outcome.

Watch

Extended reading notes

Core claim

The paper's central claim is that the observation map, the transformation from physical state to model input, is a constituent of the effective policy of a language-model agent, so that different encodings of the same state can drive a population to qualitatively different collective outcomes. The demonstration proceeds in three layers. On identical replayed fields, the three encodings — moments, bin centers, bin intervals — produced action distributions separated by a mean pairwise total-variation distance of 0.344, against a between-block test-retest floor of 0.092, with no separation this large in 5,000 field-blocked relabellings ($p = 0.0002$), and the two information-matched histogram serializations alone separated by 3.2 times the floor. In the closed loop, this microscopic difference redirected collective outcomes: in GPT the moments encoding locked in 6/6 seeds at both positive couplings while the histogram encodings locked in 0/6 (exact paired sign test $p = 0.03125$); in Claude the same design produced the reverse ordering, with moments locking 0/6 and intervals 6/6 (exact within-seed enumeration $p = 0.00103$ for terminal lock). The exact $K = 0$ control, in which the coupling term vanishes, gave identical trajectories across encodings while actions still differed, excluding matched-engine and initial-condition confounds. The stated conclusion is that state encodings form part of a model-dependent effective interaction law, not a neutral interface.

Load-bearing premise

The load-bearing premise is that each proprietary model's response behavior is stationary over the acquisition windows, so that differences between encodings reflect the encoding rather than drift in the model or its provider backend; the paper locks model identities and most generation parameters, but temperature is not echoed for the macroscopic collective runners and provider-side updates during acquisition cannot be fully excluded.

Editorial extensions

If this is right

  • The observation serializer becomes a versioned component of any LLM agent evaluation: a result obtained with one encoding is a property of the encoding–model pair, not of the task.
  • Cross-model comparisons of agent populations cannot be transferred by analogy, because the encoding-to-outcome map reversed between GPT and Claude under an identical design.
  • Single-turn prompt-sensitivity measurements understate the effect of formatting on collective dynamics, because feedback can amplify, suppress, or invert a small operator difference.
  • Surrogate models and simulators that stand in for LLM agents must be validated on the field distribution the interacting population actually generates, not only on the acquisition distribution.
  • Because centers and intervals carried identical bin masses and still separated the operator, information-equivalent serializations are not behaviorally equivalent, so a faithful encoding cannot be defined by information content alone.

Reading between the lines

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

  • An untested corollary of the model-dependence result is that the same encoding-to-outcome map should be mutable by training: fine-tuning or instruction-tuning a single model family would be predicted to shift, and possibly reverse, which encoding synchronizes, offering a cheaper test of the mechanism than adding new model families.
  • A quantitative gap the authors flag is whether the measured operator differences are sufficient to reproduce the encoding-specific trajectories; an operator-swap experiment at a fixed state sequence would close it and would also separate the contribution of the operator from that of trajectory feedback.
  • If the phenomenon generalizes beyond this minimal assay, then design choices such as memory, message-passing interfaces, and structured observation schemas in deployed multi-agent systems acquire the same status as the reward or instruction contract, and neutral-interface assumptions in agent scaffolding should be abandoned.
  • The sharp, grid-limited activation of the moments operator near zero directional imbalance hints at a possible threshold-like sensitivity to peer asymmetry; testing finer imbalance grids would reveal whether the response is continuous or genuinely discontinuous.
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 / 3 minor

Summary. The paper reports a circular-synchronization experiment in which pretrained language-model agents observe only an encoded summary of their neighbours' relative phases and choose to advance, stay, or retard. The authors manipulate the state encoding (first three circular moments vs 24-bin histograms labelled by bin centers or bin intervals) while holding the physical field fixed. In GPT, the moments encoding synchronizes the population in 6/6 seeds at two positive couplings, whereas both histogram encodings lock in 0/6; in Claude, the encoding effect replicates but reverses, with the histogram encodings locking. Identical-field replay on 48 frozen fields shows that the presented encoding alone changes the advance/stay/retard action distribution in GPT, Claude, and Gemini. Additional controls show that presentation changes with moment values fixed shift the GPT operator. The authors conclude that state encodings form part of a model-dependent effective interaction law, not a neutral interface.

Significance. If the result holds, it has direct implications for the design and evaluation of multi-agent LLM systems, for performative prediction, and for the interpretation of state-representation choices in agent scaffolds. The study is methodologically careful: matched physical seeds across encodings, an exact K=0 negative control, identical-field replay with field-blocked label permutations, prespecified thresholds and sensitivity sweeps, and full release of raw traces and code. The cross-family reversal is a strong falsification of any claim that one encoding is intrinsically superior. The main limitation is reliance on proprietary API behaviour, which the internal controls mitigate but cannot fully eliminate.

minor comments (3)
  1. [Introduction] In the first paragraph, the string 'reserveserializationfor' should read 'reserve serialization for'; the missing spaces make the sentence difficult to parse on first reading.
  2. [Supplementary Table S2] The temperature is not echoed in the resolved configuration for the macroscopic collective runners; because the stationarity of the proprietary API is the main external assumption, consider reporting the effective temperature used or adding a time-indexed stability analysis of action distributions over each acquisition window (e.g., first versus second half of calls) to further rule out drift as a confound.
  3. [Fig. 5 caption] The reformatting contrast (original–reformatted mean 0.145, 95% CI 0.079–0.219) overlaps the test–retest floor (0.049–0.090), and the text correctly notes this; consider adding a similar explicit qualifier in the abstract so that 'presentation alone shifted the operator' is not misread as being driven by the weak reformatting effect, when the padding condition carries the effect.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the encoding intervention is a controlled measurement, and the central synchronization claim is not constructed from fitted parameters or self-citations.

full rationale

The paper's central claim is an experimental finding: holding the physical field fixed and changing only the serialized observation changes action distributions (identical-field replay, Fig. 3) and, in closed loop, selects different synchronization outcomes (GPT and Claude collective runs, Figs. 1 and 4). No fitted parameter is renamed as a prediction: the Fourier coefficients in Eq. (6) describe the measured response curves and are not used to construct the collective outcome, and the K=0 control shows identical trajectories under differing action distributions, so the comparison does not reduce to the input encoding. The identical-field replay permutes presented-encoding labels within fields, making the statistical test conditional on the physical state rather than on the encoding. Ref. 12 is a self-citation but only contextualizes market-concentration applications; it is not load-bearing for the observation-map claim. The surrogate analysis is explicitly secondary and is not used to support the main claim. The only non-circular but environment-dependent assumption is stationarity of the proprietary APIs, which the block-noise floors and the cross-family reversal address empirically rather than by assumption.

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

The central claim requires no free parameters fitted to the synchronization outcome; the qualitative separation appears at prespecified thresholds and is robust to threshold sweeps. The main unverifiable premises concern proprietary model stability and the prompt contract fully controlling behavior.

free parameters (3)
  • Polar-lock threshold r1 >= 0.9 = 0.9
    Prespecified endpoint threshold for labeling a run locked; robustness swept in Supplementary Fig. S4, so it is not load-bearing for the qualitative separation.
  • Coupling constants K = -0.15, 0, 0.08, 0.15
    Hand-chosen experimental couplings; not fitted to outcomes. The qualitative effect is shown at two positive couplings and the K=0 control excludes an engine artifact.
  • Fourier coefficients (a_m, b_m) of g_R = Encoding- and field-specific estimates
    Fitted to measured mean signed actions in Eq. (6); used to characterize the response operator, not to construct the central claim.
assumptions (3)
  • domain assumption LLM responses to one fixed prompt within an acquisition block are samples from a stable distribution.
    Repeated calls estimate p_R(rho); if the model drifted mid-acquisition, block noise would be underestimated. The paper records acquisition windows and model IDs, but this assumption is not externally testable.
  • domain assumption The deterministic engine in Eq. (1) fully mediates coupling and the model never receives K, identity, time, or history.
    Central to causal isolation; the engine audit in Supplementary Fig. S1 verifies the implementation but cannot verify the model's ignorance beyond the prompt text.
  • standard math The three observation maps are deterministic functions of the same 24-bin field.
    Encoding invariance and peer-order invariance are verified numerically; the centers-versus-intervals comparison isolates serialization with identical retained information.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Same physical state, different collective dynamics: state encodings select synchronization outcomes in language-model agents." pith.science (2026). https://pith.science/paper/JGPXNVXL

@misc{pith2026260806968,
  author       = {Pith},
  title        = {Pith review of: Same physical state, different collective dynamics: state encodings select synchronization outcomes in language-model agents},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JGPXNVXL}},
  note         = {Machine review of arXiv:2608.06968}
}
read the original abstract

Language-model agents act on state encodings of their environment, yet these are treated as interchangeable interfaces. Using pretrained language models, we designed a circular-synchronization experiment applying a state-encoding intervention while holding the physical system fixed: each agent sees only a summary of its neighbours' relative phases and chooses to advance, stay or retard. Encoding that state as low-order circular moments rather than as a histogram selected different collective outcomes. In GPT the moment encoding synchronized the population in 6/6 seeds and the histogram encodings in 0/6; the effect replicated in Claude but reversed direction. Replaying identical fields shifted each agent's advance/stay/retard probabilities far beyond within-encoding repeat variation, in GPT, Claude and Gemini; in GPT, presentation alone shifted the operator with the moment values fixed. State encodings therefore form part of a model-dependent effective interaction law, not a neutral interface.

Figures

Figures reproduced from arXiv: 2608.06968 by the authors.

Figure 0
Figure 0. [PITH_FULL_IMAGE:figures/full_fig_p017_0.png] view at source ↗
Figure 1
Figure 1. Observation maps select distinct collective outcomes in matched GPT agents. a [PITH_FULL_IMAGE:figures/full_fig_p017_1.png] view at source ↗
Figure 2
Figure 2. Controlled fields elicit encoding-dependent microscopic response operators. [PITH_FULL_IMAGE:figures/full_fig_p018_2.png] view at source ↗
Figures from the paper (3 more)
Figure 3
Figure 3. Figure 3: Identical-field replay isolates a presented-encoding effect. a [PITH_FULL_IMAGE:figures/full_fig_p019_3.png]
Figure 4
Figure 4. Figure 4: Observation maps select model-family-specific collective outcomes. a [PITH_FULL_IMAGE:figures/full_fig_p020_4.png]
Figure 5
Figure 5. Figure 5: Presentation changes alter the GPT response operator when task-relevant numerical [PITH_FULL_IMAGE:figures/full_fig_p021_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 12 canonical work pages

  1. [1]

    & Suhr, A

    Sclar, M., Choi, Y., Tsvetkov, Y. & Suhr, A. Quantifying language models’ sensitivity to spurious features in prompt design, or: how I learned to start worrying about prompt formatting. InProc. International Conference on Learning Representations (ICLR)(2024). Preprint athttps://arxiv.org/abs/2310.11324

    arXiv 2024

  2. [2]

    & Stenetorp, P

    Lu, Y., Bartolo, M., Moore, A., Riedel, S. & Stenetorp, P. Fantastically ordered prompts and where to find them: overcoming few-shot prompt order sensitivity. InProc. 60th Annual Meeting of the Association for Computational Linguistics (ACL)8086–8098 (2022). https://doi.org/10.18653/v1/2022.acl-long.556

    doi:10.18653/v1/2022.acl-long.556 2022

  3. [3]

    Liu, N. F. et al. Lost in the middle: how language models use long contexts.Trans. Assoc. Comput. Linguist. 12, 157–173 (2024).https://doi.org/10.1162/tacl_a_00638

    doi:10.1162/tacl_a_00638 2024

  4. [4]

    & Huang, M

    Zheng, C., Zhou, H., Meng, F., Zhou, J. & Huang, M. Large language models are not robust multiple choice selectors. InProc. International Conference on Learning Representations (ICLR)(2024). Preprint at https://arxiv.org/abs/2309.03882

    arXiv 2024

  5. [5]

    Park, J. S. et al. Generative agents: interactive simulacra of human behavior. InProc. 36th Annual ACM Symposium on User Interface Software and Technology (UIST ’23)(ACM, 2023).https://doi.org/10. 1145/3586183.3606763

    arXiv 2023

  6. [6]

    Li, H. et al. Theory of mind for multi-agent collaboration via large language models. InProc. 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP)180–192 (2023).https://doi.org/10. 18653/v1/2023.emnlp-main.13

    work page 2023

  7. [7]

    Du, Y., Li, S., Torralba, A., Tenenbaum, J. B. & Mordatch, I. Improving factuality and reasoning in language models through multiagent debate. InProc. 41st International Conference on Machine Learning, PMLR 235, 11733–11763 (2024). Preprint athttps://arxiv.org/abs/2305.14325

    arXiv 2024

  8. [8]

    Chen, W. et al. AgentVerse: facilitating multi-agent collaboration and exploring emergent behaviors. In Proc. International Conference on Learning Representations (ICLR)(2024). Preprint athttps://arxiv. org/abs/2308.10848

    arXiv 2024

Show all 50 references
  1. [9]

    F., Aiello, L

    Ashery, A. F., Aiello, L. M. & Baronchelli, A. Emergent social conventions and collective bias in LLM populations.Sci. Adv.11, eadu9368 (2025).https://doi.org/10.1126/sciadv.adu9368

    2025 doi

  2. [10]

    & De Domenico, M

    Zomer, N. & De Domenico, M. Unraveling the emergence of collective behavior in networks of cognitive agents.npj Artif. Intell.2, 36 (2026).https://doi.org/10.1038/s44387-026-00091-5

    2026 doi

  3. [11]

    & Leibo, J

    Willis, R., Zhao, J., Du, Y. & Leibo, J. Z. Evaluating collective behaviour of hundreds of LLM agents. Preprint athttps://arxiv.org/abs/2602.16662(2026)

    2026 arXiv

  4. [12]

    & Nishinari, K

    Ezaki, T., Imura, N. & Nishinari, K. When shippers become algorithms: candidate exposure, information design, and the concentration of LLM-mediated freight markets. Preprint athttps://arxiv.org/abs/2607. 19967(2026)

    2026

  5. [13]

    La Malfa, E. et al. Large language models miss the multi-agent mark. InAdvances in Neural Information Processing Systems38, Position Paper Track (2025). Preprint athttps://arxiv.org/abs/2505.21298

    2025

  6. [14]

    C., Zrnic, T., Mendler-Dünner, C

    Perdomo, J. C., Zrnic, T., Mendler-Dünner, C. & Hardt, M. Performative prediction. InProc. 37th International Conference on Machine Learning, PMLR119, 7599–7609 (2020). Preprint athttps://arxiv. org/abs/2002.06673

    2020 arXiv

  7. [15]

    Ross, S., Gordon, G. J. & Bagnell, J. A. A reduction of imitation learning and structured prediction to no-regret online learning. InProc. 14th International Conference on Artificial Intelligence and Statistics (AISTATS)627–635 (2011). Preprint athttps://arxiv.org/abs/1011.0686

    2011 arXiv

  8. [16]

    & Levine, S

    Zhang, A., McAllister, R., Calandra, R., Gal, Y. & Levine, S. Learning invariant representations for reinforcement learning without reconstruction. InProc. International Conference on Learning Representations (ICLR)(2021). Preprint athttps://arxiv.org/abs/2006.10742. 22

    2021 arXiv

  9. [17]

    P., Littman, M

    Kaelbling, L. P., Littman, M. L. & Cassandra, A. R. Planning and acting in partially observable stochastic domains.Artif. Intell.101, 99–134 (1998).https://doi.org/10.1016/S0004-3702(98)00023-X

    1998 doi

  10. [18]

    Self-entrainment of a population of coupled non-linear oscillators

    Kuramoto, Y. Self-entrainment of a population of coupled non-linear oscillators. InInternational Symposium on Mathematical Problems in Theoretical Physics, Lecture Notes in Physics vol. 39, 420–422 (Springer, 1975). https://doi.org/10.1007/BFb0013365

    1975 doi

  11. [19]

    Winfree, A. T. Biological rhythms and the behavior of populations of coupled oscillators.J. Theor. Biol.16, 15–42 (1967).https://doi.org/10.1016/0022-5193(67)90051-3

    1967 doi

  12. [20]

    A., Bonilla, L

    Acebrón, J. A., Bonilla, L. L., Pérez Vicente, C. J., Ritort, F. & Spigler, R. The Kuramoto model: a simple paradigm for synchronization phenomena.Rev. Mod. Phys.77, 137–185 (2005).https://doi.org/ 10.1103/RevModPhys.77.137

    2005 doi

  13. [21]

    Z., Zhai, Y

    Kiss, I. Z., Zhai, Y. & Hudson, J. L. Emerging coherence in a population of chemical oscillators.Science 296, 1676–1678 (2002).https://doi.org/10.1126/science.1070757

    2002 doi

  14. [22]

    & Bullo, F

    Dörfler, F., Chertkov, M. & Bullo, F. Synchronization in complex oscillator networks and smart grids.Proc. Natl Acad. Sci. USA110, 2005–2010 (2013).https://doi.org/10.1073/pnas.1212134110

    2013 doi

  15. [23]

    Yamaguchi, S. et al. Synchronization of cellular clocks in the suprachiasmatic nucleus.Science302, 1408–1412 (2003).https://doi.org/10.1126/science.1089287

    2003 doi

  16. [24]

    & Meunier, C

    Hansel, D., Mato, G. & Meunier, C. Clustering and slow switching in globally coupled phase oscillators. Phys. Rev. E48, 3470–3477 (1993).https://doi.org/10.1103/PhysRevE.48.3470

    1993 doi

  17. [25]

    Variety and generality of clustering in globally coupled oscillators.Physica D63, 424–436 (1993)

    Okuda, K. Variety and generality of clustering in globally coupled oscillators.Physica D63, 424–436 (1993). https://doi.org/10.1016/0167-2789(93)90121-G

    1993 doi

  18. [26]

    Order function and macroscopic mutual entrainment in uniformly coupled limit-cycle oscillators

    Daido, H. Order function and macroscopic mutual entrainment in uniformly coupled limit-cycle oscillators. Prog. Theor. Phys.88, 1213–1218 (1992).https://doi.org/10.1143/ptp/88.6.1213

    1992 doi

  19. [27]

    Z., Zhai, Y

    Kiss, I. Z., Zhai, Y. & Hudson, J. L. Predicting mutual entrainment of oscillators with experiment-based phase models.Phys. Rev. Lett.94, 248301 (2005).https://doi.org/10.1103/PhysRevLett.94.248301

    2005 doi

  20. [28]

    Xie, C. et al. Can large language model agents simulate human trust behavior? InAdvances in Neural Information Processing Systems37 (2024). Preprint athttps://arxiv.org/abs/2402.04559

    2024 arXiv

  21. [29]

    Cemri, M. et al. Why do multi-agent LLM systems fail? InAdvances in Neural Information Processing Systems38, Datasets and Benchmarks Track (2025). Preprint athttps://arxiv.org/abs/2503.13657

    2025 arXiv

  22. [30]

    & Shutova, E

    Leidinger, A., van Rooij, R. & Shutova, E. The language of prompting: what linguistic properties make a prompt successful? InFindings of the Association for Computational Linguistics: EMNLP 20239210–9232 (2023).https://doi.org/10.18653/v1/2023.findings-emnlp.618

    2023 doi

  23. [31]

    Min, S. et al. Rethinking the role of demonstrations: what makes in-context learning work? InProc. 2022 Conference on Empirical Methods in Natural Language Processing (EMNLP)11048–11064 (2022). https://doi.org/10.18653/v1/2022.emnlp-main.759

    2022 doi

  24. [32]

    & Greig, M

    Givan, R., Dean, T. & Greig, M. Equivalence notions and model minimization in Markov decision processes. Artif. Intell.147, 163–223 (2003).https://doi.org/10.1016/S0004-3702(02)00376-4

    2003 doi

  25. [33]

    & Filliat, D

    Lesort, T., Díaz-Rodríguez, N., Goudou, J.-F. & Filliat, D. State representation learning for control: an overview.Neural Netw.108, 379–392 (2018).https://doi.org/10.1016/j.neunet.2018.07.006

    2018 doi

  26. [34]

    Fish, S., Gonczarowski, Y. A. & Shorrer, R. I. Algorithmic collusion by large language models. Preprint at https://arxiv.org/abs/2404.00806(2024)

    2024

  27. [35]

    & Pastorello, S

    Calvano, E., Calzolari, G., Denicolò, V. & Pastorello, S. Artificial intelligence, algorithmic pricing, and collusion.Am. Econ. Rev.110, 3267–3297 (2020).https://doi.org/10.1257/aer.20190623. 23

    2020 doi

  28. [36]

    & Liao, Q

    Li, N., Gao, C., Li, M., Li, Y. & Liao, Q. EconAgent: large language model-empowered agents for simulating macroeconomic activities. InProc. 62nd Annual Meeting of the Association for Computational Linguistics (ACL)15523–15536 (2024).https://doi.org/10.18653/v1/2024.acl-long.829

    2024 doi

  29. [37]

    & Schmidhuber, J

    Ha, D. & Schmidhuber, J. Recurrent world models facilitate policy evolution. InAdvances in Neural Information Processing Systems31 (2018). Preprint athttps://arxiv.org/abs/1803.10122

    2018 arXiv

  30. [38]

    & Norouzi, M

    Hafner, D., Lillicrap, T., Ba, J. & Norouzi, M. Dream to control: learning behaviors by latent imagination. InProc. International Conference on Learning Representations (ICLR)(2020). Preprint athttps://arxiv. org/abs/1912.01603

    2020 arXiv

  31. [39]

    & Levine, S

    Janner, M., Fu, J., Zhang, M. & Levine, S. When to trust your model: model-based policy optimization. In Advances in Neural Information Processing Systems32 (2019). Preprint athttps://arxiv.org/abs/1906. 08253

    2019

  32. [40]

    & Mendler-Dünner, C

    Hardt, M. & Mendler-Dünner, C. Performative prediction: past and future.Statist. Sci.40, 417–436 (2025). https://doi.org/10.1214/25-STS986

    2025 doi

  33. [41]

    & Ratliff, L

    Narang, A., Faulkner, E., Drusvyatskiy, D., Fazel, M. & Ratliff, L. J. Multiplayer performative prediction: learning in decision-dependent games.J. Mach. Learn. Res.24(202), 1–56 (2023). Preprint at https: //arxiv.org/abs/2201.03398

    2023 arXiv

  34. [42]

    & Kochenderfer, M

    Kelly, M., Sidrane, C., Driggs-Campbell, K. & Kochenderfer, M. J. HG-DAgger: interactive imitation learning with human experts. InProc. IEEE International Conference on Robotics and Automation (ICRA) 8077–8083 (2019).https://doi.org/10.1109/ICRA.2019.8793698

    2019

  35. [43]

    Koh, P. W. et al. WILDS: a benchmark of in-the-wild distribution shifts. InProc. 38th International Conference on Machine Learning, PMLR139, 5637–5664 (2021). Preprint athttps://arxiv.org/abs/ 2012.07421

    2021 arXiv

  36. [44]

    Strogatz, S. H. From Kuramoto to Crawford: exploring the onset of synchronization in populations of coupled oscillators.Physica D143, 1–20 (2000).https://doi.org/10.1016/S0167-2789(00)00094-4

    2000 doi

  37. [45]

    & Zou, J

    Chen, L., Zaharia, M. & Zou, J. How is ChatGPT’s behavior changing over time?Harvard Data Sci. Rev. 6(2) (2024).https://doi.org/10.1162/99608f92.5317da47

    2024 doi

  38. [46]

    Wu, Q. et al. AutoGen: enabling next-gen LLM applications via multi-agent conversations. InProc. Conference on Language Modeling (COLM)(2024). Preprint athttps://arxiv.org/abs/2308.08155

    2024 arXiv

  39. [47]

    block noise

    Wang, J., Wang, J., Athiwaratkun, B., Zhang, C. & Zou, J. Mixture-of-agents enhances large language model capabilities. InProc. International Conference on Learning Representations (ICLR)(2025). Preprint athttps://arxiv.org/abs/2406.04692. Author contributions T.E.: conceptual...

    2025 arXiv

  40. [48]

    It is a reproducibility check: the same field appears in training and in test

    Cross-block stability.Does the response law measured on a field in one acquisition block reproduce on a second, independent block of thesamefield? This is the acquisition_block_holdout scheme, and it is the criterion by which the production model class was selected. It is a re...

  41. [49]

    In-domain field generalization, referred to ascompressibility. Can the surrogate predict controlled fields that were excluded from training altogether? This is what the four field-preserving schemes measure (leave-one-profile-out, offset-group, sparse-realization and source-fa...

  42. [50]

    grouped out-of-fold, by field

    Closed-loop support.Do the fields an interacting population generates for itself fall inside the region the training set covers, and do their near neighbours in that set agree with one another about what the model does? 4.Prospective closed-loop accuracy, referred to astranspo...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.