Pith. sign in

REVIEW 4 major objections 6 minor 30 references

Properties of Quasi-synchronization Time of High-dimensional Hegselmann-Krause Dynamics

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read This paper claims that for high-dimensional noisy Hegselmann-Krause opinion dynamics, bounded state space guarantees quasi-synchronization in finite time with finite expected waiting time for every dimension, while in unbounded space the…

desk verdict Interesting and plausible high-dimensional HK dichotomy, but the main theorem's proof rests on a false cube-vs-ball projection equivalence and the d=1 proof misapplies Lemma 3.8. read the letter →

arxiv 2507.08900 v1 pith:B664IO4R submitted 2025-07-11 math.DS cs.MAnlin.AOphysics.soc-ph

Wei Su , Meiru Jiang , Yongguang Yu , Ge Chen This is my paper · ORCID
classification math.DScs.MAnlin.AOphysics.soc-ph MSC 34D0660G4060G5091D30
keywords opiniondynamicsHegselmann-Krausemodelquasi-synchronizationnoise-inducedsynchronizationstoppingtimerandomwalkrecurrencehigh-dimensionalconsensus
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

The paper asks when noisy multi-agent opinion dynamics reach quasi-synchronization, meaning all agents' opinions lie within the confidence radius, and how long the random waiting time is. It claims the answer is determined by two features of the state space: whether it is bounded and what dimension it has. In the bounded model with states confined to $[-1,1]^d$, it proves for all $d\ge 1$ and every initial state that the stopping time $T$ has finite expectation. In the unbounded model, it proves a sharp split: for $d=1,2$ quasi-synchronization occurs almost surely from every initial state but with infinite expected time for some initial states, whereas for $d\ge 3$ with symmetric noise there are initial states from which synchronization is not certain. The upshot, if correct, is that spatial boundedness is a genuine driver of noise-induced order and that the one-dimensional picture does not extend to higher-dimensional unbounded spaces.

What carries the argument

The load-bearing object is the stopping time $T$, the first time the diameter $d_V(t)$ falls to at most $\epsilon$; by Proposition 2.2, under $\delta \le \epsilon/2$ this coincides almost surely with the permanent quasi-synchronization time $T'$. For the bounded result, the key identity is the rewriting $x(t+1)=P_{B(\epsilon/2)}(F(x(t))+\xi(t+1))$, which turns coordinate-wise clipping into projection onto a ball, together with the contraction condition (3.3) with $\alpha=1$, meaning the update shortens distance to the target ball. Lemma 3.2 and Corollary 3.4 convert a positive probability of making progress into a geometric tail bound and hence a finite expectation. For the unbounded result, the key mechanism is a random-walk decomposition of the gap between two clusters: the between-cluster distance evolves as a centered random walk plus bounded increments, and recurrence in $d=1,2$ versus transience in $d\ge 3$ decides whether that gap must eventually hit the confidence threshold.

What would settle it

For the bounded model (2.2), a Monte Carlo estimate of the expected quasi-synchronization time in dimension $d=3$ from adversarial initial states should show a finite mean; if the empirical mean grows without bound as the number of samples increases, the theorem is contradicted. Separately, for the unbounded model in dimension 3 with symmetric bounded noise, simulate many runs starting from two separated clusters; the paper predicts a strictly positive fraction of non-synchronizing trajectories, so observing every run synchronize would contradict Theorem 3.5(b).

Watch

Extended reading notes

Core claim

The central discovery is a dimensional phase transition in the quasi-synchronization time $T=\inf\{t\ge 0: \max_{i,j}\|x_i(t)-x_j(t)\|≤ \epsilon\}$. Theorem 3.1 asserts that for the bounded noisy HK model (2.2), under zero-mean nondegenerate noise with $\delta \le \epsilon/2$ and $\epsilon \in (0,2\sqrt{d}]$, one has $E T<\infty$ for all $d\ge 1$ and all initial states. Theorem 3.5 asserts that for the unbounded model (2.3), for $d=1,2$, $P\{T<\infty\}=1$ for all initial states while $E T=\infty$ for some initial states; and, when the noise is symmetric, for $d\ge 3$ there are initial states with $P\{T<\infty\}<1$. The bounded-space proof rewrites the clipped average update as projection onto a Euclidean ball of radius $\epsilon/2$ and shows the distance to that ball contracts, while the unbounded proof reduces the two-cluster case to the hitting behavior of a centered random walk, which is recurrent in dimensions 1 and 2 and transient in dimension 3 and higher.

Load-bearing premise

The load-bearing premise of the bounded-space proof is that the model's coordinate-wise clipping to the cube is equivalent to projecting onto the Euclidean ball $B(\epsilon/2)$ as written in equation (3.15); if these two projections differ, the contraction argument behind Theorem 3.1 does not apply to the stated model.

Editorial extensions

If this is right

  • The bounded HK model reaches quasi-synchronization almost surely in finite time in every dimension, and the expected waiting time is finite, so bounded opinion ranges guarantee noise-induced order.
  • In unbounded spaces of dimension 1 and 2, quasi-synchronization still occurs almost surely from any initial state, but the waiting time has infinite expectation for some initial configurations, so almost-sure eventual synchronization does not mean typically fast synchronization.
  • In unbounded dimension 3 and higher with symmetric noise, the system can fail to synchronize with positive probability from separated initial clusters, so high-dimensional unbounded opinion spaces can prevent consensus.
  • The unbounded behavior is governed by recurrence versus transience of a random walk, connecting the synchronization question to classical random-walk return probabilities.
  • In the bounded case the dimension plays no role in the qualitative conclusion, since all dimensions behave the same way.

Reading between the lines

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

  • A direct check of equation (3.15) is warranted: the model's coordinate-wise cube clipping is not obviously the same as projection onto the Euclidean ball of radius $\epsilon/2$, and if the two projections differ, the contraction argument does not automatically cover the stated model even though the theorem might still be true.
  • The proof establishes finiteness of $E T$ in the bounded case but leaves the rate implicit; a natural next step is to derive explicit bounds on $E T$ in terms of the number of agents $n$, dimension $d$, confidence radius $\epsilon$, and noise bound $\delta$.
  • The recurrence-versus-transience mechanism suggests that other bounded-confidence opinion models, such as asynchronous or inertial variants, may exhibit the same dimensional split in unbounded space.
  • The paper assumes $\delta \le \epsilon/2$ throughout, which is what makes the absorption property in Proposition 2.2 work; an extension could explore what happens with stronger noise, where leaving permanent quasi-synchronization becomes possible.
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

4 major / 6 minor

Summary. This paper studies the noisy Hegselmann-Krause model in R^d, in both bounded and unbounded state spaces. It defines a stopping time T at which the configuration diameter first falls below the confidence threshold ε and claims: (i) in bounded space, E T < ∞ for every dimension d; (ii) in unbounded space, P{T < ∞} = 1 for d = 1, 2 with E T = ∞ for some initial states; and (iii) for d ≥ 3 with symmetric noise, P{T < ∞} < 1 for some initial states. The proofs combine a general contraction lemma for projected stochastic systems with random-walk recurrence and transience arguments.

Significance. If the results were correct, the bounded/unbounded and low/high-dimensional dichotomy would be a valuable contribution to the theory of noisy opinion dynamics, and the distinction between finite almost-sure hitting and integrability is natural. The paper correctly builds on the authors' earlier one-dimensional results for the almost-sure claims, and the d ≥ 3 transience argument contains a substantive idea. However, the current manuscript does not establish the main claims: the bounded-space proof rests on a false identification of the update rule with a ball projection, the one-dimensional non-integrability proof misapplies a key lemma, and the two-dimensional almost-sure claim is deferred to a reference to a similar argument. The central results are therefore unproven in this version.

major comments (4)
  1. [3.1, Eq. (3.15), proof of Theorem 3.1] The equality x(t+1) = P_{B(ε/2)}(F(x(t)) + ξ(t+1)) is false. The bounded update (2.2) is the coordinate-wise projection onto the cube C = [-1,1]^{n×d}, not the Euclidean projection onto the ball B(ε/2) in R^{n×d}. For example, with n = 2, d = 1, ε = 0.5, x(0) = (1,1) and ξ(1) = 0, the left side is (1,1), while P_{B(0.25)}(1,1) = (0.25/√2, 0.25/√2). The two projections also differ for ε > 2, where B(ε/2) is not even contained in C. Consequently, Corollary 3.4, which applies to the ball-projection system (3.2), does not apply to the actual bounded HK model, and Theorem 3.1 is unproven.
  2. [3.1, application of Corollary 3.4] Even if (3.15) were replaced by the correct cube projection, the hypotheses of Corollary 3.4 are not met by the bounded HK model. The corollary assumes ∥ξ_i(t)∥ ≤ r0/2 a.s.; with D = B(ε/2) this means r0 = ε/2, hence δ ≤ ε/4, whereas Theorem 3.1 only assumes δ ≤ ε/2. In addition, Lemma 3.2's condition (3.4) requires the noise to charge every measurable subset of B(r), which cannot hold when the noise is a.s. bounded by δ < r. Corollary 3.4 asserts without proof that bounded nondegenerate noise suffices for (3.11)-(3.12), but the sets appearing there can lie outside the noise support for states far from D. The proof of Theorem 3.1 therefore needs a different argument even beyond the correction of (3.15).
  3. [3.2, Lemma 3.8 and Eqs. (3.27)-(3.31)] Lemma 3.8 is not applicable to the process Q_ij. The lemma requires S(t+1) = g(U(t)) + h(ξ(t+1)), where U(t) = ∑_{k=1}^t ξ(k) and h is a function of the same increment ξ(t+1) that drives U. In the application, U(t) = Z(t) = ∑_{k=1}^t y(k), while the additive term is ξ_i(t+1) - ξ_j(t+1). This term is not a function of y(t+1), nor is it the increment of Z. In particular, the proof of Lemma 3.8 uses the event {ξ(1)>a, ..., ξ(L)>a} to conclude h(ξ(j)) > 0 for j ≤ L; no such conclusion is valid when the conditioning is on y(j) > a and the noise entering the additive term is ξ_i(j) - ξ_j(j). Hence E T_Q = ∞ is not established, and the non-integrability statements for d = 1, as well as the d = 2 reduction, are unsupported.
  4. [3.2, proof of Theorem 3.5 for d = 2] The proof of P{T < ∞} = 1 for d = 2 consists of the sentence 'using the recurrence of random walk in R^2 (Lemma 3.9) and the homogenous ε of HK model (2.3), then following a similar argument of Proposition 3.1 of [11], we can obtain the conclusion.' This is not a proof. The reduction to a two-dimensional random walk, the role of the confidence threshold ε, and the interaction structure between the two groups are not written down. Since the two-dimensional almost-sure claim is a central part of Theorem 3.5(a), it requires a complete argument.
minor comments (6)
  1. [Throughout] The phrase 'Morkov property' should be 'Markov property'; it appears at least twice in Section 3.2.
  2. [3.2, Eqs. (3.24) and (3.27)] The time indexing is inconsistent: d_ij(t) in (3.24) contains ξ_i(t) - ξ_j(t), while (3.27) defines Q_ij(t+1) = Z(t) + (ξ_i(t+1) - ξ_j(t+1)). Please align the indices so that (3.28) holds as written.
  3. [3.2, Lemma 3.8 proof] The notation U_k^L and U^{j-L}_L is used without a clear definition of the shifted sums; please define these partial sums explicitly before (3.16).
  4. [References] Reference [30] contains a typo: 'mordern probability' should be 'modern probability'.
  5. [3.2, d ≥ 3 paragraph] The sentence 'Give any O1 ∈ R^{n×d}' should read 'Given any ...'.
  6. [2, Eq. (2.2)] The vector clipping notation (y_1,...,y_d)^T_{[-1,1]} is introduced with a missing bracket in the displayed definition; please make clear that each coordinate is clipped separately.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation chain; self-citations are to prior published results, and the apparent flaw in Eq. (3.15) is a correctness gap rather than a circular reduction.

full rationale

I walked the derivation chain and found no step in which a prediction is equivalent to its inputs by construction. Theorem 3.1 is attempted through Lemma 3.2 and Corollary 3.4, which are new contraction-and-noise lemmas for general projected systems; no parameter is fitted and no quantity is defined in terms of the hitting time being proved. Theorem 3.5 uses prior results [9,11] for the one-dimensional unbounded case and Proposition 2.2, but these are published, externally checkable results, so citing them is legitimate support rather than a self-citation loop. The d=2 and d>=3 arguments are attempted through recurrence and transience of random walks, and they are not restatements of the conclusions. I did find a serious non-circular correctness concern: Eq. (3.15) rewrites the coordinate-wise clipping in (2.2) as projection onto B(epsilon/2), and the inequality with D=B(epsilon/2) is asserted 'by equation (2.2)' without proof; these projections are not equivalent, so Theorem 3.1 may be unproven. This is a false reduction, not a circular one: if the rewrite were true, the proof would still be a genuine derivation, and its falsity means the argument is invalid rather than self-referential. Accordingly the circularity score is 0.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The paper introduces no fitted parameters or new entities. Its proofs rely on standard random-walk facts plus several unproven or questionable modeling equivalences (ball vs cube projection) and unsupported uniform-probability assertions.

assumptions (5)
  • standard math Random walk in R^d is recurrent for d=1,2 and transient for d≥3 (Lemma 3.9, cited from [30]).
    Used in the proof of Theorem 3.5 to conclude a.s. recurrence in low dimensions and transience in high dimensions.
  • standard math Hitting times of one-dimensional zero-mean nondegenerate random walks have infinite expectation (Lemma 3.7, cited from [29]).
    Used to establish non-integrability of the quasi-synchronization time.
  • ad hoc to paper The bounded HK system (2.2) can be represented as projection onto the ball B(ε/2) (equation (3.15)).
    This equivalence is asserted without proof and is questionable; the cube clipping in (2.2) differs from ball projection. It is load-bearing for Theorem 3.1.
  • ad hoc to paper Bounded nondegenerate noise provides a uniform positive probability p for the contraction events in (3.11)-(3.12) for all states in B(r) (Corollary 3.4).
    The uniform lower bound is asserted as 'easy to show' but is not established; with bounded noise the required noise ball may have zero probability for states far from the target.
  • ad hoc to paper Lemma 3.8's hypotheses, including g(x)x>0 and h(x)x>0, hold for the process Q_ij(t) constructed in (3.27).
    The process uses h(ξ)=ξ_i−ξ_j, which is not a sign-preserving scalar function of the driving noise, so the lemma does not obviously apply.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Properties of Quasi-synchronization Time of High-dimensional Hegselmann-Krause Dynamics." pith.science (2026). https://pith.science/paper/B664IO4R

@misc{pith2026250708900,
  author       = {Pith},
  title        = {Pith review of: Properties of Quasi-synchronization Time of High-dimensional Hegselmann-Krause Dynamics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/B664IO4R}},
  note         = {Machine review of arXiv:2507.08900}
}
read the original abstract

The behavior of one-dimensional Hegselmann-Krause (HK) dynamics driven by noise has been extensively studied. Previous research has indicated that within no matter the bounded or the unbounded space of one dimension, the HK dynamics attain quasi-synchronization (synchronization in noisy case) in finite time. However, it remains unclear whether this phenomenon holds in high-dimensional space. This paper investigates the random time for quasi-synchronization of multi-dimensional HK model and reveals that the boundedness and dimensions of the space determine different outcomes. To be specific, if the space is bounded, quasi-synchronization can be attained almost surely for all dimensions within a finite time, whereas in unbounded space, quasi-synchronization can only be achieved in low-dimensional cases (one and two). Furthermore, different integrability of the random time of various cases is proved.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 30 canonical work pages

  1. [11]

    Noise-induced synchronization of Hegselmann- Krause dynamics in full space

    W. Su, J. Guo, X. Chen and G. Chen, “Noise-induced synchronization of Hegselmann- Krause dynamics in full space”, IEEE Trans. Auto. Control , vol. 64, no. 9, pp. 3804-3808, 2019

    work page 2019

  2. [1]

    Opinion dynamics and bounded confidence models, anal- ysis, and simulation

    R. Hegselmann and U. Krause, “Opinion dynamics and bounded confidence models, anal- ysis, and simulation”, J. Artif. Socie. Social Simul. , vol. 5, pp. 1-33, 2002

    work page 2002

  3. [2]

    Opinion dynamics and learning in social networks

    D. Acemoglu and A. Ozdaglar, “Opinion dynamics and learning in social networks”, Dy- namic Games and Applications , vol. 1, pp.3-49, 2011

    work page 2011

  4. [3]

    Game-theoretic analysis of the Hegselmann-Krause model for opinion dynamics in finite dimensions

    S. Etesami and T. Basar, “Game-theoretic analysis of the Hegselmann-Krause model for opinion dynamics in finite dimensions”, IEEE Trans. Auto. Contr. , vol. 60, no. 7, pp. 1886-1897, 2015. 14

    work page 2015

  5. [4]

    Population games methods for distributed control of microgrids

    E. Mojica-Nava, C. Barreto and N. Quijano, “Population games methods for distributed control of microgrids”, IEEE Trans. Smart Grid , vol. 6, no. 6, pp. 2586-2595, 2015

    work page 2015

  6. [5]

    A coordinated optimal operation of a grid- connected wind-solar microgrid incorporating hybrid energy storage management systems

    M. Abdelghany, A. Al-Durra and F. Gao, “A coordinated optimal operation of a grid- connected wind-solar microgrid incorporating hybrid energy storage management systems”, IEEE Trans. Sustainable Energy , vol. 15, no. 1, pp. 39-51, 2024

    work page 2024

  7. [6]

    Individualization as driving force of clustering phe- nomena in humans

    M. M¨ as, A. Flache and D. Helbing, “Individualization as driving force of clustering phe- nomena in humans”, PLoS Comput. Bio. , vol. 6, no. 10, e1000959, 2010

    work page 2010

  8. [7]

    The role of noise and initial conditions in the asymptotic solution of a bounded confidence, continuous-opinion model

    A. Carro, R. Toral and M. San Miguel, “The role of noise and initial conditions in the asymptotic solution of a bounded confidence, continuous-opinion model”, J. Statis. Phys. , vol. 151, no. 1-2, pp.131-149, 2013

    work page 2013

Show all 30 references
  1. [8]

    The noisy Hegselmann-Krause model for opinion dynamics

    M. Pineda, R. Toral and E. Hern´ andez-Garc ´ ıa, “The noisy Hegselmann-Krause model for opinion dynamics”, Eur. Phys. J. B , vol.86, no.12, pp.1-10, 2013

    2013

  2. [9]

    Noise leads to quasi-consensus of Hegselmann-Krause opinion dynamics

    W. Su, G. Chen and Y. Hong, “Noise leads to quasi-consensus of Hegselmann-Krause opinion dynamics”, Automatica, vol. 85, pp. 448-454, 2017

    2017

  3. [10]

    Noisy Hegselmann-Krause systems: Phase transi- tion and the 2R-conjecture

    C. Wang, Q. Li, W. E and B. Chazelle, “Noisy Hegselmann-Krause systems: Phase transi- tion and the 2R-conjecture”, J. Statis. Phys. , vol. 166, pp. 1209-1225, 2017

    2017

  4. [12]

    Mixing beliefs among interacting agents

    G. Deffuant, D. Neau, F. Amblard and G. Weisbuch, “Mixing beliefs among interacting agents”, Adv. Compl. Syst. , vol.3, pp. 87-98, 2000

    2000

  5. [13]

    Quasi-synchronization of bounded confidence opinion dynamics with a stochastic asynchronous rule

    W. Su, X. Wang, G. Chen and K. Shen, “Quasi-synchronization of bounded confidence opinion dynamics with a stochastic asynchronous rule”, Sci. China-Infor. Sci. , vol. 65, 119203, 2022

    2022

  6. [14]

    Pikovsky, M

    A. Pikovsky, M. Rosenblum and J. Kurths, Synchronization: a universal concept in non- linear sciences, Cambridge University Press, 2001

    2001

  7. [15]

    Balanov, N

    A. Balanov, N. Janson, D. Postnov and O. Sosnovtseva, Synchronization: From simple to complex, Springer, 2009

    2009

  8. [16]

    Bellman, Adaptive control processes: a guided tour , Princeton University Press, 1961

    R. Bellman, Adaptive control processes: a guided tour , Princeton University Press, 1961

    1961

  9. [17]

    On the surprising behavior of distance metrics in high dimensional space

    C. Aggarwal, A. Hinneburg and D. Keim, “On the surprising behavior of distance metrics in high dimensional space”, Inter. conf. on Database Theory , pp. 420-434, 2001

    2001

  10. [18]

    Continuous opinion dynamics under bounded confidence: A survey

    J. Lorenz, “Continuous opinion dynamics under bounded confidence: A survey”, Inter. J. Mod. Phys. C , vol. 18, no. 12, pp. 1819-1838, 2007

    2007

  11. [19]

    Fostering consensus in multidimensional continuous opinion dynamics under bounded confidence

    J. Lorenz, “Fostering consensus in multidimensional continuous opinion dynamics under bounded confidence”, In Managing Complexity: Insights, Concepts, Applications , (pp. 321- 334). Berlin/Heidelberg: Springer, 2008

    2008

  12. [20]

    Multi-dimensional Hegselmann-Krause dynamics

    A. Nedi´ c and B. Touri, “Multi-dimensional Hegselmann-Krause dynamics”,2012 IEEE 51st Conf. Deci. Contr. (CDC) , Maui, HI, USA, pp. 68-73, 2012 15

    2012

  13. [21]

    Novel multidimensional models of opinion dynamics in social networks

    S. Parsegov, A. Proskurnikov, R. Tempo and N. Friedkin, “Novel multidimensional models of opinion dynamics in social networks”, IEEE Trans. Auto. Contr. , vol. 62, no. 5, pp. 2270-2285, 2017

    2017

  14. [22]

    Inertial Hegselmann-Krause systems

    B. Chazelle and C. Wang, “Inertial Hegselmann-Krause systems”, IEEE Trans. Auto. Contr., vol. 62, no. 8, pp. 3905-3913, 2017

    2017

  15. [23]

    Multi-dimensional extensions of the Hegselmann-Krause model

    G. De Pasquale and M. Valcher, “Multi-dimensional extensions of the Hegselmann-Krause model”, 2022 IEEE 61st Conf. Deci. Contr. (CDC) , Cancun, Mexico, pp. 3525-3530, 2022

    2022

  16. [24]

    Bounded confidence opinion dynamics: A survey

    C. Bernardo, C. Altafini, A. Proskurnikov and F. Vasca, “Bounded confidence opinion dynamics: A survey”, Automatica, vol. 159, 111302, 2024

    2024

  17. [25]

    Probability laws in multi-dimensional noisy HK-models

    A. Bolychev and A. Manita, “Probability laws in multi-dimensional noisy HK-models”, J. Phys.: Conf. Ser. , vol. 1740, 012039, 2021

    2021

  18. [26]

    On the origins of cells: a hypothesis for the evolutionary tran- sitions from abiotic geochemistry to chemoautotrophic prokaryotes, and from prokaryotes to nucleated cells

    W. Martin and M. Russell, “On the origins of cells: a hypothesis for the evolutionary tran- sitions from abiotic geochemistry to chemoautotrophic prokaryotes, and from prokaryotes to nucleated cells”, Philo. Trans. Roy. Soci. B: Bio. Scie. , vol. 358, no. 1429, pp. 59-85, 2003

    2003

  19. [27]

    Experimental models of primitive cellular com- partments: encapsulation, growth, and division

    M. Hanczyc, S. Fujikawa and J. Szostak, “Experimental models of primitive cellular com- partments: encapsulation, growth, and division”, Science, vol. 302, no. 5645, pp. 618-622, 2003

    2003

  20. [28]

    Emergence of collective oscillations in massive human crowds

    F. Gu, B. Guiselin, N. Bain, I. Zuriguel and D. Bartolo, “Emergence of collective oscillations in massive human crowds”, Nature, vol. 638, pp. 112-119, 2025

    2025

  21. [29]

    Chow and H

    Y. Chow and H. Teicher, Probability theory: Independence, interchangeability, martingales, Springer Science & Business Media, 1997

    1997

  22. [30]

    Kallenberg, Foundations of mordern probability, Springer, 2002

    O. Kallenberg, Foundations of mordern probability, Springer, 2002. 16

    2002

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.