pith. machine review for the scientific record. sign in

arxiv: 2605.12308 · v1 · submitted 2026-05-12 · 💻 cs.LG

Recognition: 2 theorem links

· Lean Theorem

In-context learning to predict critical transitions in dynamical systems

Yunus Sevinchan , Juan Nathaniel , Kai Ueltzh\"offer , Carla Roesch , Tobias Weber , Vaios Laschos , Hang Fan , Gregor Ramien
show 3 more authors
Johannes Haux Pierre Gentine Benjamin Herdeanu
Authors on Pith no claims yet

Pith reviewed 2026-05-13 07:03 UTC · model grok-4.3

classification 💻 cs.LG
keywords critical transitionsin-context learningbifurcationsearly warning systemsdynamical systemssynthetic datadeep learning
0
0 comments X

The pith

In-context learning on synthetic bifurcation data detects critical transitions in unseen dynamical systems.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Critical transitions are abrupt shifts in system dynamics that can have irreversible effects, yet real-world examples remain rare. Standard statistical indicators like rising variance often fail with short noisy records, and typical deep learning models cannot handle regimes outside their training data. This paper introduces TipPFN, a prior-data fitted network trained on a generator that produces many canonical bifurcation scenarios paired with varied randomized stochastic dynamics. The model reads contexts of different lengths and dimensions to estimate how close a system is to a tipping point. It achieves strong detection performance on new tipping types, simulated-to-real transfers, and actual observations, including in zero-shot mode.

Core claim

TipPFN, trained on synthetic data from canonical bifurcation scenarios coupled to diverse randomized stochastic dynamics, infers a system's proximity to a critical transition and delivers robust early detection in previously unseen tipping regimes, sim-to-real examples, and real-world observations under both in-context and zero-shot conditions.

What carries the argument

TipPFN, a prior-data fitted network that uses in-context learning to infer proximity to a critical transition from input contexts of varying sizes, complexity, and dimensionalities.

If this is right

  • Reliable early warning becomes possible for systems where real transition data are scarce.
  • Detection works for tipping regimes absent from the training distribution.
  • Performance holds under realistic conditions of limited samples and correlated noise.
  • Both in-context learning with examples and pure zero-shot inference are supported.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same synthetic-generator approach might be tested on other rare-event prediction tasks such as financial crashes or epidemic outbreaks.
  • Extending the context window or adding multi-scale dynamics to the generator could further improve detection lead time.
  • Hybrid use with traditional indicators might provide uncertainty estimates that pure deep-learning outputs lack.

Load-bearing premise

The novel synthetic data generator based on canonical bifurcation scenarios coupled to diverse randomized stochastic dynamics produces training distributions that allow the model to generalize to real-world critical transitions.

What would settle it

A concrete falsifier would be a documented real-world critical transition, such as a specific lake eutrophication or climate regime shift, where TipPFN fails to give an early warning signal even when the underlying bifurcation type matches those used in training.

Figures

Figures reproduced from arXiv: 2605.12308 by Benjamin Herdeanu, Carla Roesch, Gregor Ramien, Hang Fan, Johannes Haux, Juan Nathaniel, Kai Ueltzh\"offer, Pierre Gentine, Tobias Weber, Vaios Laschos, Yunus Sevinchan.

Figure 1
Figure 1. Figure 1: TipPFN detects critical transi￾tions earlier and more accurately than clas￾sical early warning signals (EWS), state-of￾the-art deep learning and ICL-based meth￾ods across 14 semi-real, sim-to-real, and real-world systems spanning climate, engi￾neering, and biology. Lines show across￾system mean and standard deviation. Tipping points occur when variation in forcing or sys￾tem parameters triggers an abrupt, … view at source ↗
Figure 2
Figure 2. Figure 2: Overview. TipPFN is trained on synthetic data, based on embedding canonical b-tipping systems into high-dimensional, randomized stochastic dynamics. Primary training target is the relative distance to criticality (RDTC), a measure of how close a system is to a critical transition. Although the synthetic dynamics were only driven by b-tipping systems, TipPFN successfully generalizes to other classes of tipp… view at source ↗
Figure 3
Figure 3. Figure 3: Tipping Prediction with TipPFN. (a) Example multi-variate time series from a critical episode and underlying RDTC Λ. Green bars mark candidate observation windows ending ∆ time steps before the critical time tcrit. (b) Example query episode at ∆ = 30 and context composed from a critical (red) and non-critical (blue) episode. TipPFN and TabPFN are conditioned on the shaded context and observed window, inclu… view at source ↗
Figure 4
Figure 4. Figure 4: Context matters. (a) Multi-parameter sweep over number of context episodes, observed feature channels, and window size W for TipPFN and TabPFN. Color denotes AUROC averaged over all ∆ > 0 and all datasets in [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Zero-shot TipPFN RDTC nowcasts on three uni-variate real-world time series, without per-system fine-tuning. For each system, the top row shows the observed empirical signal, the middle row shows an ensemble of TipPFN predictions for Λ ∗ (median and 10–90% predictive bands over 100 stochastic retained time points), and the bottom row shows the distribution of the first predicted crossings Λ ∗ thrs = tanh(5Λ… view at source ↗
read the original abstract

Critical transitions - abrupt, often irreversible changes in system dynamics - arise across human and natural systems, often with catastrophic consequences. Real-world observations of such shifts remain scarce, preventing the development of reliable early warning systems. Conventional statistical and spectral indicators, such as increasing variance, tend to fail under realistic conditions of limited data and correlated noise, whereas existing deep learning classifiers do not extrapolate beyond their training data distribution. In this work, we introduce TipPFN, an in-context learning (ICL) framework that uses a prior-data fitted network to infer a system's proximity to a critical transition. Trained on our novel synthetic data generator, which is based on canonical bifurcation scenarios coupled to diverse, randomized stochastic dynamics, TipPFN flexibly capitalizes on contexts of various sizes, complexity and dimensionalities. We demonstrate robust, state-of-the-art early detection of critical transitions in previously unseen tipping regimes, sim-to-real examples, and real-world observations in both ICL and zero-shot settings.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper introduces TipPFN, a prior-data fitted network (PFN) for in-context learning (ICL) that infers proximity to critical transitions from contexts of varying length and dimensionality. It is trained exclusively on a novel synthetic generator consisting of canonical bifurcation scenarios (e.g., fold, Hopf) coupled to randomized stochastic dynamics, and claims state-of-the-art early detection performance on previously unseen tipping regimes, sim-to-real transfers, and selected real-world time series, both in ICL and zero-shot regimes.

Significance. If the generalization claims hold under rigorous validation, the work would be significant: it offers a data-efficient route to early-warning systems for critical transitions where real labeled examples are scarce, and demonstrates that ICL on carefully constructed synthetic priors can outperform both classical indicators (variance, autocorrelation) and standard supervised deep-learning classifiers that fail to extrapolate outside their training distribution.

major comments (3)
  1. [Abstract and §4] Abstract and §4 (Experiments): the headline claim of 'state-of-the-art' performance on unseen regimes and real observations is load-bearing yet unsupported by any reported quantitative metrics, error bars, exact definitions of success (e.g., lead time, AUC, or false-positive rate), or explicit data-exclusion rules; without these, the generalization statement cannot be evaluated.
  2. [§3.2] §3.2 (Synthetic data generator): the central sim-to-real and zero-shot claims rest on the assumption that contexts drawn from the bifurcation-plus-randomized-stochastic generator lie sufficiently close to the tested real-world series; no distributional diagnostic (MMD, Wasserstein distance on autocorrelation spectra, variance scaling, or power-law exponents) is supplied to confirm that the reported real/sim-to-real successes are inside the training support rather than selected overlaps.
  3. [§4.3] §4.3 (Real-world results): the zero-shot and ICL results on real observations are presented without ablation on context length, noise structure, or non-stationary forcing; if these factors lie outside the generator's support, the reported robustness may not generalize, directly undermining the 'robust' claim.
minor comments (2)
  1. [§3.1] Notation for the PFN prior and the exact form of the in-context prompt (context length, embedding of time series) is introduced without a compact mathematical definition; a single equation or pseudocode block would improve clarity.
  2. [Figures 5-7] Figure captions for the real-world examples should explicitly state the source dataset, sampling rate, and any preprocessing steps applied before feeding the series to TipPFN.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback. We appreciate the emphasis on rigorous validation of our generalization claims. We address each major comment below and outline revisions to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Abstract and §4] Abstract and §4 (Experiments): the headline claim of 'state-of-the-art' performance on unseen regimes and real observations is load-bearing yet unsupported by any reported quantitative metrics, error bars, exact definitions of success (e.g., lead time, AUC, or false-positive rate), or explicit data-exclusion rules; without these, the generalization statement cannot be evaluated.

    Authors: We agree that the abstract and experimental section would benefit from more explicit quantitative support. The manuscript presents comparative results via figures in §4 showing superior performance over baselines on unseen regimes and real data, but we will revise to include dedicated tables with AUC, lead-time metrics, false-positive rates, error bars from repeated runs, precise success definitions, and data-exclusion criteria. This will make the state-of-the-art claims directly evaluable. revision: yes

  2. Referee: [§3.2] §3.2 (Synthetic data generator): the central sim-to-real and zero-shot claims rest on the assumption that contexts drawn from the bifurcation-plus-randomized-stochastic generator lie sufficiently close to the tested real-world series; no distributional diagnostic (MMD, Wasserstein distance on autocorrelation spectra, variance scaling, or power-law exponents) is supplied to confirm that the reported real/sim-to-real successes are inside the training support rather than selected overlaps.

    Authors: The generator was designed to span diverse bifurcation and stochastic regimes to approximate real-world variability. While qualitative matches are shown, we acknowledge the absence of formal distributional checks. In revision we will add MMD distances, autocorrelation spectrum comparisons, variance scaling, and power-law exponent analyses between synthetic training contexts and the real/sim-to-real test series to quantify overlap and support the generalization claims. revision: yes

  3. Referee: [§4.3] §4.3 (Real-world results): the zero-shot and ICL results on real observations are presented without ablation on context length, noise structure, or non-stationary forcing; if these factors lie outside the generator's support, the reported robustness may not generalize, directly undermining the 'robust' claim.

    Authors: Some context-length sensitivity was examined internally during development, but we concur that systematic ablations are required to substantiate robustness. The revised §4.3 will incorporate explicit ablations varying context length, noise correlation structures, and non-stationary forcing amplitudes, reporting performance changes to demonstrate where the method remains effective and where limits appear. revision: yes

Circularity Check

0 steps flagged

No circularity: training on independent synthetic generator and evaluation on external real-world data

full rationale

The paper trains TipPFN on a novel synthetic data generator (canonical bifurcations + randomized stochastic dynamics) and evaluates generalization to previously unseen tipping regimes, sim-to-real transfers, and real-world observations. No derivation, equation, or central claim reduces by construction to fitted parameters, self-citations, or ansatzes within the reported setup. The use of held-out real observations provides an independent external benchmark, so the headline claim of robust early detection does not collapse to a tautology or self-referential fit.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 1 invented entities

The central claim depends on the assumption that the synthetic generator captures enough of real tipping dynamics for generalization; the model itself introduces no new physical entities but relies on standard neural network training assumptions.

free parameters (1)
  • network architecture and training hyperparameters
    The prior-data fitted network requires choices of layers, context length, and optimization settings that are fitted during training on the synthetic corpus.
axioms (1)
  • domain assumption Canonical bifurcation models plus randomized stochastic dynamics sufficiently approximate real-world critical transition statistics
    Invoked to justify training on synthetic data for real-world generalization.
invented entities (1)
  • TipPFN no independent evidence
    purpose: In-context learning model for inferring proximity to critical transitions
    New model name and architecture introduced for this task.

pith-pipeline@v0.9.0 · 5500 in / 1318 out tokens · 69099 ms · 2026-05-13T07:03:52.125917+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

68 extracted references · 68 canonical work pages

  1. [1]

    Tipping points and cascading transitions: Methods, principles, and evidences.arXiv preprint arXiv:2511.01168, 2025

    Sheng Fang, Ziyan Wang, Jürgen Kurths, and Jingfang Fan. Tipping points and cascading transitions: Methods, principles, and evidences.arXiv preprint arXiv:2511.01168, 2025

    work page arXiv 2025

  2. [2]

    Peter Ashwin, Sebastian Wieczorek, Renato Vitolo, and Peter Cox. Tipping points in open systems: bifurcation, noise-induced and rate-dependent examples in the climate system.Philo- sophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 370(1962):1166–1184, 2012

    work page 1962

  3. [3]

    Stochastic bifurcation of the north atlantic circulation under a midrange future climate scenario with the nasa-giss modele

    Anastasia Romanou, David Rind, Jeff Jonas, Ron Miller, Maxwell Kelley, Gary Russell, Clara Orbe, Larissa Nazarenko, Rebecca Latto, and Gavin A Schmidt. Stochastic bifurcation of the north atlantic circulation under a midrange future climate scenario with the nasa-giss modele. Journal of Climate, 36(18):6141–6161, 2023

    work page 2023

  4. [4]

    Recovery rates reflect distance to a tipping point in a living system.Nature, 481(7381):357–359, 2012

    Annelies J Veraart, Elisabeth J Faassen, Vasilis Dakos, Egbert H Van Nes, Miquel Lürling, and Marten Scheffer. Recovery rates reflect distance to a tipping point in a living system.Nature, 481(7381):357–359, 2012

    work page 2012

  5. [5]

    Western systems coordinating council disturbance report, 1996

    Western Electricity Coordinating Council. Western systems coordinating council disturbance report, 1996

    work page 1996

  6. [6]

    Lange, Pascal P

    Luis Gómez-Nava, Robert T. Lange, Pascal P. Klamser, Juliane Lukas, Lenin Arias-Rodriguez, David Bierbach, Jens Krause, Henning Sprekeler, and Pawel Romanczuk. Fish shoals re- semble a stochastic excitable system driven by environmental perturbations.Nature Physics, 19(5):663–669, February 2023

    work page 2023

  7. [7]

    Rate-induced tipping in natural and human systems.Earth System Dynamics, 14(3):669–683, 2023

    Paul DL Ritchie, Hassan Alkhayuon, Peter M Cox, and Sebastian Wieczorek. Rate-induced tipping in natural and human systems.Earth System Dynamics, 14(3):669–683, 2023

    work page 2023

  8. [8]

    Collective decision-making with heterogeneous biases: Role of network topology and suscepti- bility.Physical Review Research, 7(1), March 2025

    Yunus Sevinchan, Petro Sarkanych, Abi Tenenbaum, Yurij Holovatch, and Pawel Romanczuk. Collective decision-making with heterogeneous biases: Role of network topology and suscepti- bility.Physical Review Research, 7(1), March 2025

    work page 2025

  9. [9]

    Robustness of variance and autocorrelation as indicators of critical slowing down.Ecology, 93(2):264–271, 2012

    Vasilis Dakos, Egbert H Van Nes, Paolo d’Odorico, and Marten Scheffer. Robustness of variance and autocorrelation as indicators of critical slowing down.Ecology, 93(2):264–271, 2012

    work page 2012

  10. [10]

    A universal law of the characteristic return time near thresholds.Oecologia, 65(1):101– 107, 1984

    C Wissel. A universal law of the characteristic return time near thresholds.Oecologia, 65(1):101– 107, 1984

    work page 1984

  11. [11]

    Slow recovery from perturbations as a generic indicator of a nearby catastrophic shift.The American Naturalist, 169(6):738–747, 2007

    Egbert H Van Nes and Marten Scheffer. Slow recovery from perturbations as a generic indicator of a nearby catastrophic shift.The American Naturalist, 169(6):738–747, 2007

    work page 2007

  12. [12]

    Slowing down as an early warning signal for abrupt climate change.Proceedings of the National Academy of Sciences, 105(38):14308–14312, 2008

    Vasilis Dakos, Marten Scheffer, Egbert H Van Nes, Victor Brovkin, Vladimir Petoukhov, and Hermann Held. Slowing down as an early warning signal for abrupt climate change.Proceedings of the National Academy of Sciences, 105(38):14308–14312, 2008

    work page 2008

  13. [13]

    Rising variance: a leading indicator of ecological transition.Ecology letters, 9(3):311–318, 2006

    Stephen R Carpenter and William A Brock. Rising variance: a leading indicator of ecological transition.Ecology letters, 9(3):311–318, 2006

    work page 2006

  14. [14]

    Changing skewness: an early warning signal of regime shifts in ecosystems.Ecology letters, 11(5):450–460, 2008

    Vishwesha Guttal and Ciriyam Jayaprakash. Changing skewness: an early warning signal of regime shifts in ecosystems.Ecology letters, 11(5):450–460, 2008

    work page 2008

  15. [15]

    Resilience indica- tors: prospects and limitations for early warnings of regime shifts.Philosophical Transactions of the Royal Society B: Biological Sciences, 370(1659), 2015

    Vasilis Dakos, Stephen R Carpenter, Egbert H van Nes, and Marten Scheffer. Resilience indica- tors: prospects and limitations for early warnings of regime shifts.Philosophical Transactions of the Royal Society B: Biological Sciences, 370(1659), 2015

    work page 2015

  16. [16]

    Early warning signals also precede non-catastrophic transitions.Oikos, 122(5):641–648, 2013

    Sonia Kéfi, Vasilis Dakos, Marten Scheffer, Egbert H Van Nes, and Max Rietkerk. Early warning signals also precede non-catastrophic transitions.Oikos, 122(5):641–648, 2013

    work page 2013

  17. [17]

    Effect of rate of change of parameter on early warning signals for critical transitions.Chaos: An Interdisciplinary Journal of Nonlinear Science, 31(1), 2021

    Induja Pavithran and RI Sujith. Effect of rate of change of parameter on early warning signals for critical transitions.Chaos: An Interdisciplinary Journal of Nonlinear Science, 31(1), 2021

    work page 2021

  18. [18]

    Deep learning for predicting rate-induced tipping.Nature Machine Intelligence, 6(12):1556–1565, 2024

    Yu Huang, Sebastian Bathiany, Peter Ashwin, and Niklas Boers. Deep learning for predicting rate-induced tipping.Nature Machine Intelligence, 6(12):1556–1565, 2024

    work page 2024

  19. [19]

    Towards out-of-distribution generalization: A survey, 2023

    Jiashuo Liu, Zheyan Shen, Yue He, Xingxuan Zhang, Renzhe Xu, Han Yu, and Peng Cui. Towards out-of-distribution generalization: A survey, 2023

    work page 2023

  20. [20]

    Domain generalization: A survey.IEEE transactions on pattern analysis and machine intelligence, 45(4):4396–4415, 2022

    Kaiyang Zhou, Ziwei Liu, Yu Qiao, Tao Xiang, and Chen Change Loy. Domain generalization: A survey.IEEE transactions on pattern analysis and machine intelligence, 45(4):4396–4415, 2022. 11

    work page 2022

  21. [21]

    Out-of-distribution generalization in time series: A survey, 2025

    Xin Wu, Fei Teng, Xingwang Li, Ji Zhang, Tianrui Li, and Qiang Duan. Out-of-distribution generalization in time series: A survey, 2025

    work page 2025

  22. [22]

    Deep learning for early warning signals of tipping points

    Thomas M Bury, RI Sujith, Induja Pavithran, Marten Scheffer, Timothy M Lenton, Mad- hur Anand, and Chris T Bauch. Deep learning for early warning signals of tipping points. Proceedings of the National Academy of Sciences, 118(39):e2106140118, 2021

    work page 2021

  23. [23]

    Ultra- early prediction of tipping points: Integrating dynamical measures with reservoir computing

    Xin Li, Qunxi Zhu, Chengli Zhao, Bolin Zhao, Xue Zhang, Xiaojun Duan, and Wei Lin. Ultra- early prediction of tipping points: Integrating dynamical measures with reservoir computing. arXiv preprint arXiv:2603.14944, 2026

    work page arXiv 2026

  24. [24]

    Deep learning for predicting the occurrence of tipping points.Royal Society Open Science, 12(7), 2025

    Chengzuo Zhuge, Jiawei Li, and Wei Chen. Deep learning for predicting the occurrence of tipping points.Royal Society Open Science, 12(7), 2025

    work page 2025

  25. [25]

    Machine learning prediction of tipping in complex dynamical systems.Physical Review Research, 6(4):043194, 2024

    Shirin Panahi, Ling-Wei Kong, Mohammadamin Moradi, Zheng-Meng Zhai, Bryan Glaz, Mulugeta Haile, and Ying-Cheng Lai. Machine learning prediction of tipping in complex dynamical systems.Physical Review Research, 6(4):043194, 2024

    work page 2024

  26. [26]

    Transformers can do bayesian inference

    Samuel Müller, Noah Hollmann, Sebastian Pineda Arango, Josif Grabocka, and Frank Hutter. Transformers can do bayesian inference. InInternational Conference on Learning Representa- tions, 2022

    work page 2022

  27. [27]

    Tabpfn: A transformer that solves small tabular classification problems in a second, 2023

    Noah Hollmann, Samuel Müller, Katharina Eggensperger, and Frank Hutter. Tabpfn: A transformer that solves small tabular classification problems in a second, 2023

    work page 2023

  28. [28]

    Accurate predictions on small data with a tabular foundation model.Nature, 637(8045):319–326, 2025

    Noah Hollmann, Samuel Müller, Lennart Purucker, Arjun Krishnakumar, Max Körfer, Shi Bin Hoo, Robin Tibor Schirrmeister, and Frank Hutter. Accurate predictions on small data with a tabular foundation model.Nature, 637(8045):319–326, 2025

    work page 2025

  29. [29]

    Tabiclv2: A better, faster, scalable, and open tabular foundation model, 2026

    Jingang Qu, David Holzmüller, Gaël Varoquaux, and Marine Le Morvan. Tabiclv2: A better, faster, scalable, and open tabular foundation model, 2026

    work page 2026

  30. [30]

    From tables to time: Extending tabpfn-v2 to time series forecasting, 2026

    Shi Bin Hoo, Samuel Müller, David Salinas, and Frank Hutter. From tables to time: Extending tabpfn-v2 to time series forecasting, 2026

    work page 2026

  31. [31]

    Do-pfn: In-context learning for causal effect estimation, 2025

    Jake Robertson, Arik Reuter, Siyuan Guo, Noah Hollmann, Frank Hutter, and Bernhard Schölkopf. Do-pfn: In-context learning for causal effect estimation, 2025

    work page 2025

  32. [32]

    Cresswell, and Rahul G

    Vahid Balazadeh, Hamidreza Kamkari, Valentin Thomas, Benson Li, Junwei Ma, Jesse C. Cresswell, and Rahul G. Krishnan. Causalpfn: Amortized causal effect estimation via in-context learning, 2025

    work page 2025

  33. [33]

    Gradient free deep reinforcement learning with tabpfn, 2025

    David Schiff, Ofir Lindenbaum, and Yonathan Efroni. Gradient free deep reinforcement learning with tabpfn, 2025

    work page 2025

  34. [34]

    Generalized stochastic resilience for early warning signals based on koopman operator.Nonlinear Dynamics, 114(4):246, 2026

    Yuta Miyauchi, Masahiro Ikeda, and Yoshinobu Kawahara. Generalized stochastic resilience for early warning signals based on koopman operator.Nonlinear Dynamics, 114(4):246, 2026

    work page 2026

  35. [35]

    Anticipating the occurrence and type of critical transitions.Science Advances, 9(1):eabq4558, 2023

    Florian Grziwotz, Chun-Wei Chang, Vasilis Dakos, Egbert H van Nes, Markus Schwarzländer, Oliver Kamps, Martin Heßler, Isao T Tokuda, Arndt Telschow, and Chih-hao Hsieh. Anticipating the occurrence and type of critical transitions.Science Advances, 9(1):eabq4558, 2023

    work page 2023

  36. [36]

    A scale-free graph model based on bipartite graphs.Discrete Applied Mathematics, 157(10):2267–2284, 2009

    Étienne Birmelé. A scale-free graph model based on bipartite graphs.Discrete Applied Mathematics, 157(10):2267–2284, 2009

    work page 2009

  37. [37]

    The meaning and use of the area under a receiver operating characteristic (roc) curve.Radiology, 143(1):29–36, 1982

    James A Hanley and Barbara J McNeil. The meaning and use of the area under a receiver operating characteristic (roc) curve.Radiology, 143(1):29–36, 1982

    work page 1982

  38. [38]

    An introduction to roc analysis.Pattern recognition letters, 27(8):861–874, 2006

    Tom Fawcett. An introduction to roc analysis.Pattern recognition letters, 27(8):861–874, 2006

    work page 2006

  39. [39]

    The theory of signal detectability.Transactions of the IRE professional group on information theory, 4(4):171–212, 1954

    WWTG Peterson, T Birdsall, and We Fox. The theory of signal detectability.Transactions of the IRE professional group on information theory, 4(4):171–212, 1954

    work page 1954

  40. [40]

    The use of the area under the roc curve in the evaluation of machine learning algorithms.Pattern recognition, 30(7):1145–1159, 1997

    Andrew P Bradley. The use of the area under the roc curve in the evaluation of machine learning algorithms.Pattern recognition, 30(7):1145–1159, 1997

    work page 1997

  41. [41]

    A new measure of rank correlation.Biometrika, 30(1-2):81–93, 1938

    Maurice G Kendall. A new measure of rank correlation.Biometrika, 30(1-2):81–93, 1938

    work page 1938

  42. [42]

    W. E. Ricker. Stock and recruitment.Journal of the Fisheries Research Board of Canada, 11(5):559–623, May 1954

    work page 1954

  43. [43]

    Early warning signals of extinction in deteriorating environments.Nature, 467(7314):456–459, 2010

    John M Drake and Blaine D Griffen. Early warning signals of extinction in deteriorating environments.Nature, 467(7314):456–459, 2010. 12

    work page 2010

  44. [44]

    Experiments and modelling of rate-dependent transition delay in a stochastic subcritical bifurcation.Royal Society open science, 5(3), 2018

    Giacomo Bonciolini, Dominik Ebi, Edouard Boujo, and Nicolas Noiray. Experiments and modelling of rate-dependent transition delay in a stochastic subcritical bifurcation.Royal Society open science, 5(3), 2018

    work page 2018

  45. [45]

    A pause in the weakening of the atlantic meridional overturning circulation since the early 2010s.Nature Communications, 15(1):10642, 2024

    Sang-Ki Lee, Dongmin Kim, Fabian A Gomez, Hosmay Lopez, Denis L V olkov, Shenfu Dong, Rick Lumpkin, and Stephen Yeager. A pause in the weakening of the atlantic meridional overturning circulation since the early 2010s.Nature Communications, 15(1):10642, 2024

    work page 2024

  46. [46]

    Metaflux: Meta-learning global carbon fluxes from sparse spatiotemporal observations.Scientific Data, 10(1):440, 2023

    Juan Nathaniel, Jiangong Liu, and Pierre Gentine. Metaflux: Meta-learning global carbon fluxes from sparse spatiotemporal observations.Scientific Data, 10(1):440, 2023

    work page 2023

  47. [47]

    Spatiotemporal upscaling of sparse air-sea pco2 data via physics-informed transfer learning.Scientific data, 11(1):1098, 2024

    Siyeon Kim, Juan Nathaniel, Zhewen Hou, Tian Zheng, and Pierre Gentine. Spatiotemporal upscaling of sparse air-sea pco2 data via physics-informed transfer learning.Scientific data, 11(1):1098, 2024

    work page 2024

  48. [48]

    Lora: Low-rank adaptation of large language models.Iclr, 1(2):3, 2022

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Liang Wang, Weizhu Chen, et al. Lora: Low-rank adaptation of large language models.Iclr, 1(2):3, 2022

    work page 2022

  49. [49]

    Test-time training with self-supervision for generalization under distribution shifts

    Yu Sun, Xiaolong Wang, Zhuang Liu, John Miller, Alexei Efros, and Moritz Hardt. Test-time training with self-supervision for generalization under distribution shifts. InInternational conference on machine learning, pages 9229–9248. PMLR, 2020

    work page 2020

  50. [50]

    End-to-end test-time training for long context.arXiv preprint arXiv:2512.23675, 2025

    Arnuv Tandon, Karan Dalal, Xinhao Li, Daniel Koceja, Marcel Rød, Sam Buchanan, Xiaolong Wang, Jure Leskovec, Sanmi Koyejo, Tatsunori Hashimoto, et al. End-to-end test-time training for long context.arXiv preprint arXiv:2512.23675, 2025

    work page arXiv 2025

  51. [51]

    Chaosbench: A multi-channel, physics-based benchmark for subseasonal-to- seasonal climate prediction.Advances in Neural Information Processing Systems, 37:43715– 43729, 2024

    Juan Nathaniel, Yongquan Qu, Tung Nguyen, Sungduk Yu, Julius Busecke, Aditya Grover, and Pierre Gentine. Chaosbench: A multi-channel, physics-based benchmark for subseasonal-to- seasonal climate prediction.Advances in Neural Information Processing Systems, 37:43715– 43729, 2024

    work page 2024

  52. [52]

    Auto-07p: Continuation and bifurcation software for ordinary differential equations

    Eusebius J Doedel, Alan R Champneys, Fabio Dercole, Thomas F Fairgrieve, Yuri A Kuznetsov, B Oldeman, RC Paffenroth, B Sandstede, XJ Wang, and CH Zhang. Auto-07p: Continuation and bifurcation software for ordinary differential equations. 2007

    work page 2007

  53. [53]

    CausalDynamics: A large-scale benchmark for structural discovery of dynamical causal models

    Benjamin Herdeanu, Juan Nathaniel, Carla Roesch, Jatan Buch, Gregor Ramien, Johannes Haux, and Pierre Gentine. CausalDynamics: A large-scale benchmark for structural discovery of dynamical causal models. InAdvances in Neural Information Processing Systems 38 (NeurIPS 2025), Track on Datasets and Benchmarks, 2025. arXiv preprint arxiv:2505.16620

    work page arXiv 2025

  54. [54]

    Deep koopman operators for causal discovery.Communications Physics, 8(1):513, 2025

    Juan Nathaniel, Carla Roesch, Jatan Buch, Derek DeSantis, Adam Rupe, Kara D Lamb, and Pierre Gentine. Deep koopman operators for causal discovery.Communications Physics, 8(1):513, 2025

    work page 2025

  55. [55]

    Thresholds and breakpoints in ecosystems with a multiplicity of stable states

    Robert M May. Thresholds and breakpoints in ecosystems with a multiplicity of stable states. Nature, 269(5628):471–477, 1977

    work page 1977

  56. [56]

    Hassan Alkhayuon, Peter Ashwin, Laura C Jackson, Courtney Quinn, and Richard A Wood. Basin bifurcations, oscillatory instability and rate-induced thresholds for atlantic meridional overturning circulation in a global oceanic box model.Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 475(2225), 2019

    work page 2019

  57. [57]

    Soil carbon and climate change: from the jenkinson effect to the compost-bomb instability.European journal of soil science, 62(1):5–12, 2011

    CM Luke and PM Cox. Soil carbon and climate change: from the jenkinson effect to the compost-bomb instability.European journal of soil science, 62(1):5–12, 2011

    work page 2011

  58. [58]

    Excitability in ramped systems: the compost-bomb instability.Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 467(2129):1243–1269, 2011

    Sebastian Wieczorek, Peter Ashwin, Catherine M Luke, and Peter M Cox. Excitability in ramped systems: the compost-bomb instability.Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 467(2129):1243–1269, 2011

    work page 2011

  59. [59]

    Laelaps: An energy-efficient seizure detection algorithm from long-term human iEEG recordings without false alarms

    Alessio Burrello, Lukas Cavigelli, Kaspar Schindler, Luca Benini, and Abbas Rahimi. Laelaps: An energy-efficient seizure detection algorithm from long-term human iEEG recordings without false alarms. In2019 Design, Automation & Test in Europe Conference & Exhibition (DATE), pages 752–757, 2019

    work page 2019

  60. [60]

    Jirsa, William C

    Viktor K. Jirsa, William C. Stacey, Pascale P. Quilichini, Anton I. Ivanov, and Christophe Bernard. On the nature of seizure dynamics.Brain, 137(8):2210–2230, 2014

    work page 2014

  61. [61]

    Phonation onset: vocal fold modeling and high-speed glottography.The Journal of the Acoustical Society of America, 104(1):464–470, 1998

    Patrick Mergell, Hanspeter Herzel, Thomas Wittenberg, Monika Tigges, and Ulrich Eysh- oldt. Phonation onset: vocal fold modeling and high-speed glottography.The Journal of the Acoustical Society of America, 104(1):464–470, 1998. 13

    work page 1998

  62. [62]

    Vibratory responses of synthetic, self-oscillating vocal fold models.The Journal of the Acoustical Society of America, 132(5):3428–3438, 2012

    Preston R Murray and Scott L Thomson. Vibratory responses of synthetic, self-oscillating vocal fold models.The Journal of the Acoustical Society of America, 132(5):3428–3438, 2012

    work page 2012

  63. [63]

    Ryo Shimamura and Isao T Tokuda. Effect of level difference between left and right vocal folds on phonation: Physical experiment and theoretical study.Journal of the Acoustical Society of America, 140(4_Supplement):3393–3394, 2016

    work page 2016

  64. [64]

    Multiparametric real-time sensing of cytosolic physiology links hypoxia responses to mitochondrial electron transport.New Phytologist, 224(4):1668–1684, 2019

    Stephan Wagner, Janina Steinbeck, Philippe Fuchs, Sophie Lichtenauer, Marlene Elsässer, Jos HM Schippers, Thomas Nietzel, Cristina Ruberti, Olivier Van Aken, Andreas J Meyer, et al. Multiparametric real-time sensing of cytosolic physiology links hypoxia responses to mitochondrial electron transport.New Phytologist, 224(4):1668–1684, 2019

    work page 2019

  65. [65]

    Ben I Moat, David Smeed, Darren Rayner, William E Johns, Ryan H Smith, Denis L V olkov, Shane Elipot, Tillys Petit, Jules B Kajtar, Molly O Baringer, et al. Atlantic meridional overturn- ing circulation observed by the rapid-mocha-wbts (rapid-meridional overturning circulation and heatflux array-western boundary time series) array at 26n from 2004 to 2023...

    work page 2004

  66. [66]

    Graphical representation and stability conditions of predator-prey interactions.The American Naturalist, 97(895):209–223, 1963

    Michael L Rosenzweig and Robert H MacArthur. Graphical representation and stability conditions of predator-prey interactions.The American Naturalist, 97(895):209–223, 1963

    work page 1963

  67. [67]

    The influence of social norms on the dynamics of vaccinating behaviour for paediatric infectious diseases.Proceedings of the Royal Society B: Biological Sciences, 281(1780), 2014

    Tamer Oraby, Vivek Thampi, and Chris T Bauch. The influence of social norms on the dynamics of vaccinating behaviour for paediatric infectious diseases.Proceedings of the Royal Society B: Biological Sciences, 281(1780), 2014

    work page 2014

  68. [68]

    sim-to- real

    A Demetri Pananos, Thomas M Bury, Clara Wang, Justin Schonfeld, Sharada P Mohanty, Brendan Nyhan, Marcel Salathé, and Chris T Bauch. Critical dynamics in population vaccinating behavior.Proceedings of the National Academy of Sciences, 114(52):13762–13767, 2017. 14 Appendix contents A Additional background 15 B Prior-Data Fitted Networks 17 B.1 Driver vari...

    work page 2017