Pith. sign in

REVIEW 2 major objections 5 minor 81 references

The emergence of chaos in population game dynamics induced by comparisons

T0 review · 2 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read For every 2x2 anti-coordination game, there exist revision protocols and a time-step size such that the discrete-time dynamics is Li-Yorke chaotic, with periodic orbits of every period, even though the continuous-time dynamics converges…

desk verdict Theorems 2 and 3 are the real contribution, and they look right; but Theorem 3's key lemma is a proof sketch, so the main new claim needs referee verification. read the letter →

arxiv 2412.06037 v2 pith:KOH6C6CI submitted 2024-12-08 math.DS cs.GTecon.TH

Jakub Bielawski , {L}ukasz Cholewa , Fryderyk Falniowski This is my paper · ORCID
classification math.DScs.GTecon.TH MSC 37E0537D4591A22
keywords anti-coordinationgamesrevisionprotocolsdiscrete-timegamedynamicsLi-Yorkechaospairwiseproportionalimitationperiod-3orbitsintervalmapsNashequilibriuminstability
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 is trying to establish that discrete-time population game dynamics driven by revision protocols can be chaotic even in the simplest nontrivial case: a symmetric 2x2 anti-coordination game with payoff inequalities $a

What carries the argument

The carrying object is the one-dimensional interval map $F:[0,1]\to[0,1]$ obtained from the inflow-outflow equation $F(x)=x+\delta[(1-x)\rho_{BA}(x)-x\rho_{AB}(x)]$. The load-bearing dynamical criterion is Proposition 1: if two points $0\le z_l<z_r\le1$ satisfy $F^2(z_l)>z_r$ and one of two geometric crossing conditions ($F(z_r)<z_l<F(z_l)>z_r$, or $F(z_r)>z_l$ and $F(z_l)<z_l$), then some $x\in(z_l,z_r)$ has $F(x)<x<F^3(x)$; a form of the odd-chaos theorem then yields a period-3 orbit, and the Li-Yorke and Sharkovsky theorems yield Li-Yorke chaos with all periods. For imitative protocols the same map reads $F(x)=x(1+\delta(1-x)(r_{BA}(x)-r_{AB}(x)))$, with fixed points $0$, $1$ and $p$, so chaos is controlled by the sign of the difference of the two conditional imitation rates. The explicit constructions place the two critical points at $c_l=p/2$ and $c_r=(p+1)/2$ (or their mirrors under $x\mapsto1-x$) and verify Proposition 1 with formulas; the truncation in Section 6 replaces one rate by a constant past $\gamma$, enlarging the admissible $\eta$ and making $p$ repelling while preserving the self-map property $F([0,1])=[0,1]$.

What would settle it

For a specific test, take payoffs with $p=1/4$, $b-d=1$, and the maximum perturbations $\eta=4p/((1-p)^2(b-d))$, $\xi=4/(p(b-d))$ from Proposition 5 with $\delta=1$; the proof asserts that the map $F$ in (15) satisfies the crossing conditions of Proposition 1 at $c_l=p/2$ and $c_r=(p+1)/2$. One can plot the third iterate $F^3$ and check whether there is an $x\in(c_l,c_r)$ with $F(x)<x<F^3(x)$; if no such crossing exists, the period-3 route is absent. A second falsification target is the Euler-map assumption itself: simulate the asynchronous Poisson process with the same rates and check whether the long-run distribution matches the deterministic chaotic map.

Watch

Extended reading notes

Core claim

The paper states the result as three theorems. Theorem 1: for any $2\times2$ anti-coordination game with $a<c$ and $d<b$ there is a revision protocol—either innovative or imitative—and a choice of $\delta$ for which the map $F(x)=x+\delta[(1-x)\rho_{BA}(x)-x\rho_{AB}(x)]$ is Li-Yorke chaotic, has periodic orbits of any period, and has the unique Nash equilibrium $p$ repelling. Theorem 2 specializes to a perturbation of pairwise proportional imitation: with conditional imitation rates $r_{AB}(x)=\eta[(b-d)/p(x-p)]_+$ and $r_{BA}(x)=\xi[(b-d)/p(p-x)]_+$, explicit multipliers $\eta,\xi$ are given such that for every $\delta$ above a threshold $\delta_p$ the dynamics is Li-Yorke chaotic with all periods; for $p\in(0,1/2)$ the values are $\eta=4p/((1-p)^2(b-d))$ and $\xi=4/(p(b-d))$, and the case $p>1/2$ follows by the topological conjugacy $x\mapsto 1-x$. Theorem 3 modifies the rates by truncating one of them to a constant beyond a level $\gamma\in(0,1)\setminus\{p\}$; with $\gamma=p+p^2/2$ for $p<1/2$, the same chaotic conclusions hold and, for $\delta>\delta_p$, $p$ is repelling, so the chaotic behavior cannot be masked by almost-sure convergence to equilibrium.

Load-bearing premise

The load-bearing premise is that the discrete-time law of motion is exactly the first-order Euler map $F(x)=x+\delta[(1-x)\rho_{BA}(x)-x\rho_{AB}(x)]$, meaning each agent has at most one revision opportunity in a period and there is no within-period compounding; if agents can revise several times inside a period, the map and the chaos thresholds change. A second load-bearing point is that the proof of the repelling-fixed-point part of Theorem 3 uses Proposition 7, whose Appendix D proof explicitly skips calculations and gives only an outline.

Editorial extensions

If this is right

  • For any $2\times2$ anti-coordination game, a sufficiently large revision step $\delta$ can destroy equilibrium prediction: the discrete-time dynamics can have periodic orbits of every period and an uncountable scrambled set of initial conditions.
  • The perturbed pairwise proportional imitation protocol of Theorem 2 is close to the microeconomic protocol that yields replicator dynamics in continuous time, so the instability is not confined to exotic or hard-to-interpret switch rates.
  • In the symmetric case $p=1/2$, Proposition 4 shows that any imitative protocol satisfying the symmetry condition $r_{AB}(x)+r_{AB}(1-x)=r_{BA}(x)+r_{BA}(1-x)$ is chaotic for large $\delta$; chaos is thus an inherent property of imitation, not a special construction.
  • The truncated rates of Theorem 3 make the Nash equilibrium repelling for sufficiently large $\delta$ for every equilibrium position $p$, so the chaotic set cannot be dismissed as a measure-zero phenomenon that almost all trajectories avoid.
  • All conclusions for equilibria above $1/2$ are obtained by topological conjugacy, so the result covers the whole range $p\in(0,1)$.

Reading between the lines

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

  • Beyond the paper, the same period-3 crossing criterion should be testable in larger anti-coordination games by projecting onto a one-dimensional factor of the state space; a numerical search for period-3 points would show whether the chaotic phenomenon extends beyond $2\times2$ games.
  • Beyond the paper, because the deterministic map is the exact Euler step, simulating the underlying asynchronous Poisson process would reveal whether allowing multiple revision opportunities within one period shifts or suppresses the chaos thresholds; this is a direct robustness check on the model.
  • Beyond the paper, the truncation parameter $\gamma$ has a behavioral interpretation as a satisficing threshold beyond which agents ignore further payoff differences, which suggests an experimental prediction: capping payoff information should make volatile, chaotic swings more likely rather than less.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 5 minor

Summary. The paper studies discrete-time revision-protocol dynamics for 2×2 anti-coordination games with payoff matrix (game) and a unique mixed Nash equilibrium p. The state map is the first-order Euler map F(x)=x+δ[(1−x)ρ_BA(x)−xρ_AB(x)]. The main results are: Theorem 1, which constructs innovative and imitative revision protocols for every such game so that, for large δ, the dynamics is Li-Yorke chaotic, has periodic orbits of all periods, and has repelling Nash equilibrium; Theorem 2, which shows that a two-parameter perturbation of pairwise proportional imitation (rates (14)) yields Li-Yorke chaos for every game; and Theorem 3, which uses truncated rates (18)-(19) to obtain Li-Yorke chaos together with a repelling Nash equilibrium. The proofs are constructive, based on the period-three criterion of Proposition 1 and explicit piecewise-linear bimodal maps. Most technical estimates are collected in the appendices; the exception is Proposition 7, whose proof is explicitly only an outline.

Significance. If the missing verification is supplied, this is a valuable contribution to the discrete-time theory of population games. The paper gives explicit, game-dependent constructions rather than abstract existence results, and it shows that a natural perturbation of pairwise proportional imitation—the standard microfoundation of the replicator dynamics—can produce Li-Yorke chaos with a repelling Nash equilibrium, in sharp contrast to the convergent continuous-time picture. The explicit computations in Propositions 2 and 5, Lemma 7, and Lemma 3 are detailed and check out, and the topological-conjugacy reductions are elegant. The main new claim concerning observable chaos, however, currently rests on an incomplete proof, so the result is not yet fully established as written.

major comments (2)
  1. [Appendix D, proof of Proposition 7] The proof of Proposition 7 states that it will 'skip all calculations and present only its outline' and then asserts the inequalities F*(γ)<cl, F*(cl)>γ, (F*)²(cl)>γ for δ>δ*_p, as well as the one-sided derivative bounds (F*)'_−(p)<−1 and (F*)'_+(p)<−1 for δ>max{δ*_4,δ*_5}. This proposition is the sole proof of Theorem 3 for p∈(0,1/2) and, through Proposition 8, for p>1/2. Since Theorem 3 is the paper's main added value beyond [23], these omitted calculations are load-bearing. Please supply the complete algebra for the thresholds δ*_1,...,δ*_5, the Lipschitz estimate with L=1+2δ/(2−2p−p²), and the derivative bounds, or provide a machine-checked verification.
  2. [Abstract and Section 7] The abstract and the concluding section state that unpredictability 'is encoded into any imitative revision protocol.' The supporting statement, Proposition 4, requires the special case d−b=a−c (i.e., p=1/2) and the extra symmetry condition (13). For general p and general imitative protocols the paper proves existence of some protocol with chaotic dynamics, not inevitability across all imitative protocols. Please qualify the abstract and conclusions accordingly.
minor comments (5)
  1. [Section 5.1, Step 3] The assertion that (η,ξ,δ)∈∆p is equivalent to (ξ,η,δ)∈∆ep is not true as stated; for example, with p=0.4 and ep=0.6, (η,ξ)=(5,5) satisfies the bounds of ∆ep but not those of ∆p. The subsequent conjugacy argument can be formulated directly without this equivalence and should be restated.
  2. [Section 6.1, Step 3] The analogous equivalence between ∆*_p and Γ*_ep also appears to be one-directional at best. Please state precisely which inclusion is needed for the proof of Proposition 8 and verify it for the specific maximal parameters used there.
  3. [Appendix C, proof of Proposition 4] The notation c^δ_l appears to be a typo for z^δ_l, and the displayed formula for δ* does not solve F(z)=1. From F(z)=z[1+δ(1−z)h(z)], the correct value is δ*=1/[z h(z)] (with z=z^δ_l), so the formula should be corrected.
  4. [Appendix A, proof of Proposition 2 and Proposition 9] There are typos in displayed formulas: in the expression for ρAB(x) on [2p,1], 'b(1−2b)' should be 'p(1−2p)', and in Proposition 9 the interval '[b + p²/2, 1]' should read '[p + p²/2, 1]'.
  5. [Appendix D, proof of Proposition 8] The sentence 'application of Lemma 4, Lemma 5 and Proposition 7 completes the proof' should explicitly verify that the truncated rates defined by (22) satisfy the conjugacy condition (29). The verification is short but is currently omitted.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular step: constructions are explicit and the only self-citation (the p=1/2 base case from [23]) is not load-bearing; the main caveat is an omitted verification in Proposition 7, which is a completeness risk rather than circularity.

full rationale

No circular step identified. The theorems are existence constructions: the switch rates (10)-(11), (14), and the truncated rates (18)-(19) are defined explicitly from the game data, and the parameters η, ξ, γ are fixed constants (e.g., η=4p/((1−p)^2(b−d)), ξ=4/(p(b−d)), γ=p+p²/2), not fitted to the desired chaotic conclusion. Chaos is established by verifying the inequalities of Proposition 1 directly for the constructed maps (Propositions 5, 7 and Lemmas 7, 8); the map is not defined in terms of the target property. The only self-citation is the p=1/2 base case attributed to [23], whose authors include F. Falniowski (Section 4.1: 'existence of the revision protocol which introduces chaotic behavior for sufficiently large δ for p=1/2 was shown already in [23]'; Appendix B: 'Since the case p=1/2 was shown in [23]'). This is not load-bearing: Proposition 4 gives an independent proof for the imitative p=1/2 case, the innovative p=1/2 case is a boundary point of parameter space, and all p≠1/2 results are derived from explicit piecewise-linear maps and topological conjugacy. The genuine caveat is that the proof of Proposition 7 is only an outline and explicitly omits the threshold computations (Appendix D: 'we will skip all calculations and present only its outline'); Theorem 3's repelling-Nash conclusion depends on those unshown inequalities. This is a verification/completeness risk, not circularity, because the thresholds and inequalities are stated openly and can be checked independently rather than being assumed as inputs.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central constructions introduce several free parameters (β2, β3, η, ξ, γ) that are chosen by hand, not fitted to data; each one is a degree of freedom in the protocol or map being constructed. The main domain assumptions are the Euler-style discrete-time update and the anti-coordination payoff structure. Proposition 4's condition (13) is an extra ad hoc symmetry not implied by the definition of imitation.

free parameters (5)
  • β2 (innovative construction) = β2 ∈ (1/(1-2p), 2(1-p)/(1-2p))
    Chosen in Proposition 2 to make the piecewise linear map F satisfy Lemma 1 and have p as its unique interior fixed point.
  • β3 (innovative construction) = β3 ∈ [-p/((1-2p)(1-p)), 0]
    Chosen in Proposition 2 so the third lap of F keeps the map in [0,1] and satisfies the period-3 conditions.
  • η (perturbed PPI) = 4p/((1-p)^2(b-d)) for p∈(0,1/2); 4/(ep(b-d)) for ep∈(1/2,1)
    Multiplier on the A-to-B imitation rate; set to the maximal value allowed by the interval-map constraint so the inequalities F(cr)<cl and F^2(cl)>cr hold.
  • ξ (perturbed PPI) = 4/(p(b-d)) for p∈(0,1/2); 4ep/((1-ep)^2(b-d)) for ep∈(1/2,1)
    Multiplier on the B-to-A imitation rate; chosen symmetrically to η to make the map satisfy Proposition 1.
  • γ (truncation level) = p+p^2/2 for p∈(0,1/2); 1-γ for ep∈(1/2,1); γ=0 or 1 for p=1/2
    Truncation point for the imitation rates in Theorem 3; selected so the truncated map F* has a local minimum at γ and remains an interval map with enlarged η.
assumptions (4)
  • standard math Sharkovsky's theorem, Li-Yorke theorem, and the Li-Misiurewicz-Pianigiani-Yorke odd-chaos theorem for continuous interval maps.
    Used in Section 3 to convert the period-3 condition from Proposition 1 into Li-Yorke chaos and periodic orbits of every period.
  • domain assumption The inflow-outflow equation (3) exactly defines the discrete-time dynamics; each revising agent switches with probability δρ_ij(x) within one period.
    This is the discrete-time revision protocol model from [23]; the chaos theorems are statements about this map.
  • domain assumption The game is a 2x2 anti-coordination game with payoff matrix (game), a<c, d<b, so the payoffs uA and uB cross once at p given by (eq).
    All theorems are proved for this class; the paper does not address coordination games.
  • ad hoc to paper Symmetry condition (13): rAB(x)+rAB(1-x)=rBA(x)+rBA(1-x) for all x, assumed in Proposition 4.
    Required to extend the p=1/2 chaos result to 'any' imitative protocol; not a consequence of the imitative protocol definition (2), so the abstract's 'any imitative revision protocol' overstates the theorem.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The emergence of chaos in population game dynamics induced by comparisons." pith.science (2026). https://pith.science/paper/KOH6C6CI

@misc{pith2026241206037,
  author       = {Pith},
  title        = {Pith review of: The emergence of chaos in population game dynamics induced by comparisons},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KOH6C6CI}},
  note         = {Machine review of arXiv:2412.06037}
}
abstract

Precise description of population game dynamics introduced by revision protocols - an economic model describing the agent's propensity to switch to a better-performing strategy - is of importance in economics and social sciences in general. In this setting innovation or imitation of others is the force which drives the evolution of the economic system. As the continuous-time game dynamics is relatively well understood, the same cannot be said about revision driven dynamics in the discrete time. We investigate the behavior of agents in a $2\times 2$ anti-coordination game with symmetric random matching and a unique mixed Nash equilibrium. In continuous time the Nash equilibrium is attracting and induces a global evolutionary stable state. We show that in the discrete time one can construct (either innovative or imitative) revision protocol and choose a level of the time step, under which the game dynamics is Li-Yorke chaotic, inducing complex and unpredictable behavior of the system, precluding stable predictions of equilibrium. Moreover, we reveal that this unpredictability is encoded into any imitative revision protocol. Furthermore, we show that for any such game there exists a perturbed pairwise proportional imitation protocol introducing chaotic behavior of the agents for sufficiently large time step.

Figures

Figures reproduced from arXiv: 2412.06037 by the authors.

Figure 1
Figure 1. Diagrams illustrating conditions (1) and (2) of Proposition 1 are in the first row. Diagrams illustrating conditions (1′ ) and (2′ ) are in the second row. Cobweb diagrams (columns 1. and 3.) and the first 50 iterations of a starting point (columns 2. and 4.). Here we test conditions of Proposition 1 for zl = 0.2 and zr = 0.6 when they are critical points of piecewise linear maps: (a) None of the conditions is satis… view at source ↗
Figure 2
Figure 2. Switch rates ρAB (on the left) and ρBA (on the right) from Proposition 2 for the game (game) with Nash equilibrium p = 0.2, and parameters β2 = 2, β3 = − 1 3 . Step 3. Proof of Theorem 1 for pe ∈ ( 1 2 , 1). Now, consider the game G ≡e Ge(A, ve), given be (game), where ve(x) = (ue1(x), ue2(x)) is its payoff vector, with the unique (symmetric) Nash equilibrium pe ∈ ( 1 2 , 1). Then there exists a population game G ≡ … view at source ↗
Figure 3
Figure 3. The graphs of maps F induced by the switch rates ρAB and ρBA from Proposition 2 for p = 0.2 (on the left) and Fe from Proposition 3 for pe = 0.8 (on the right) and for parameters β2 = 2, β3 = − 1 3 . Remark 4. One can also construct an appropriate imitative revision protocol introducing chaotic game dynamics. This construction follows similar steps as presented above, that is, it is based on analogs of Lemma 1 and P… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Graphs of the elements of the dynamical system induced by an imitative revision protocol. [(a) and (b)] Graphs of the switch rates for Nash equilibrium p = 0.4. (c) Graph of the update map for Nash equilibrium p = 0.4. (d) Graph of the update map for Nash equilibrium p…
Figure 5
Figure 5. Figure 5: Bifurcation diagrams for the dynamical system of the map F in (15) for the maximal values of η and ξ and for p = 0.4. The horizontal axis is the parameter δ ∈ (0, 1]. For each δ ∈ (0, 1], 20000 iterations of the starting points were made and then next 100 iterations of…
Figure 6
Figure 6. Figure 6: Bifurcation diagram for the dynamical system of the map F in (15) for the maximal values of η and ξ and for p = 0.25. The horizontal axis is the parameter δ ∈ [0, 1]. For each δ ∈ [0, 1], 20000 iterations of the starting points were made and then next 100 iterations of…
Figure 7
Figure 7. Figure 7: Truncation of imitation revision rates of a game with Nash equilibrium p = 1 4 . [(a) and (b)] The perturbed imitation rates and the update map with ξ = 6, δ = 1 and maximal possible value of η for (15) (η = 16 9 ). [(c) and (d)] Truncation of (a) at level γ = 3 4 and …
Figure 8
Figure 8. Figure 8: Piecewise linear bimodal maps g of the form (31) (on the left) and f of the form (30) (on the right) for the parameters cl = 0.2, cr = 0.6 and β2 = 2.3. Step 2. Proof of Theorem 1 for p ∈ (0, 1 2 ). We will again consider a 2 × 2 anti-coordination game G ≡ G(A, v) defi…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

81 extracted references · 63 canonical work pages

  1. [23]

    Falniowski and P

    F. Falniowski and P. Mertikopoulos. On the discrete-time origins of replicator dynamics: from con- vergence to instability and chaos. Int J Game Theory, 54:7, 2025. doi: https://doi.org/10.1007/ s00182-025-00929-3

    work page 2025

  2. [1]

    E. Akin. Domination or equilibrium.Mathematical Biosciences, 50(3-4):239–250, 1980. doi: 10.1016/ 0025-5564(80)90039-5

    1980

  3. [2]

    Argasinski and M

    K. Argasinski and M. Broom. Evolutionary stability under limited population growth: Eco-evolutionary feedbacks and replicator dynamics.Ecological Complexity, 34:198–212, 2018. THE EMERGENCE OF CHAOS IN POPULATION GAME DYNAMICS 23

    work page 2018

  4. [3]

    Argasinski and M

    K. Argasinski and M. Broom. Interaction rates, vital rates, background fitness and replicator dynamics: how to embed evolutionary game structure into realistic population dynamics.Theory in Biosciences, 137:33–50, 2018

    work page 2018

  5. [4]

    P. Auer, N. Cesa-Bianchi, Y. Freund, and R. E. Schapire. Gambling in a rigged casino: The adversarial multi-armed bandit problem. InProceedings of the 36th Annual Symposium on Foundations of Computer Science, 1995. doi: 10.1109/SFCS.1995.492488

    arXiv 1995

  6. [5]

    R. A. Becker, S. K. Chakrabarti, W. Geller, B. Kitchens, and M. Misiurewicz. Dynamics of the Nash map in the game of Matching Pennies.Journal of Difference Equations and Applications, 13(2-3): 223–235, 2007. doi: 10.1080/10236190601069275

    work page doi:10.1080/10236190601069275 2007

  7. [6]

    Bielawski, T

    J. Bielawski, T. Chotibut, F. Falniowski, G. Kosiorowski, M. Misiurewicz, and G. Piliouras. Follow-the- regularized-leader routes to chaos in routing games. InProceedings of the 38th International Conference on Machine Learning, volume 139 ofPMLR, pages 925–935, 2021

    work page 2021

  8. [7]

    Bielawski, T

    J. Bielawski, T. Chotibut, F. Falniowski, M. Misiurewicz, and G. Piliouras. Memory loss can prevent chaos in games dynamics.Chaos: An Interdisciplinary Journal of Nonlinear Science, 34(1):013146,

    work page

Show all 81 references
  1. [8]

    Binmore and L

    K. Binmore and L. Samuelson. Muddling through: Noisy equilibrium selection.Journal of Economic Theory, 74(2):235–265, June 1997. doi: 10.1006/jeth.1996.2255

    1997

  2. [9]

    Björnerstedt and J

    J. Björnerstedt and J. Weibull. Nash equilibrium and evolution by imitation. The rational foundations of economic behavior, KJ Arrow, E. Colombatto, M. Perlman, and C. Schmidt, eds, 1996

    1996

  3. [10]

    Boone and P

    V. Boone and P. Mertikopoulos. The equivalence of dynamic and strategic stability under regularized learning in games. In NeurIPS ’23: Proceedings of the 37th International Conference on Neural Information Processing Systems, 2023

    2023

  4. [11]

    G. W. Brown and J. Von Neumann.Solutions of games by differential equations. Rand Corporation, 1950

    1950

  5. [12]

    Bruin and V

    H. Bruin and V. Jiménez López. On the lebesgue measure of li-yorke pairs for interval maps.Commu- nications in Mathematical Physics, 299(2):523–560, 2010. doi: 10.1007/s00220-010-1085-9

    2010 doi

  6. [13]

    Burke and A

    A. Burke and A. van Stel. Entry and exit in disequilibrium.Journal of Business Venturing, 29(1): 174–192, 2014

    2014

  7. [14]

    Cesa-Bianchi and G

    N. Cesa-Bianchi and G. Lugosi.Prediction, Learning, and Games. Cambridge University Press, 2006

    2006

  8. [15]

    Algorithms, games, andevolution.Proceedings of the National Academy of Sciences, 111(29):10620–10623, 2014

    E.Chastain, A.Livnat, C.Papadimitriou, andU.Vazirani. Algorithms, games, andevolution.Proceedings of the National Academy of Sciences, 111(29):10620–10623, 2014. doi: 10.1073/pnas.1406556111

    2014 doi

  9. [16]

    Chotibut, F

    T. Chotibut, F. Falniowski, M. Misiurewicz, and G. Piliouras. The route to chaos in routing games: When is price of anarchy too optimistic?Advances in Neural Information Processing Systems, 33: 766–777, 2020

    2020

  10. [17]

    Chotibut, F

    T. Chotibut, F. Falniowski, M. Misiurewicz, and G. Piliouras. Family of chaotic maps from game theory. Dynamical Systems, 36(1):48–63, 2021. https://doi.org/10.1080/14689367.2020.1795624

    2021

  11. [18]

    Collet and J.-P

    P. Collet and J.-P. Eckmann. On the abundance of aperiodic behaviour for maps on the interval. 1980

    1980

  12. [19]

    Cressman

    R. Cressman. Evolutionary dynamics and extensive form games, volume 5. MIT Press, 2003

    2003

  13. [20]

    Cressman

    R. Cressman. The stability concept of evolutionary game theory: a dynamic approach, volume 94. Springer Science & Business Media, 2013

    2013

  14. [21]

    Cressman and Y

    R. Cressman and Y. Tao. The replicator equation and other game dynamics.Proceedings of the National Academy of Sciences, 111(supplement_3):10810–10817, 2014. doi: 10.1073/pnas.1400823111

    2014 doi

  15. [22]

    Daskalakis and I

    C. Daskalakis and I. Panageas. Last-iterate convergence: Zero-sum games and constrained min-max optimization. arXiv preprint arXiv:1807.04252, 2018

    2018

  16. [24]

    Fudenberg and L

    D. Fudenberg and L. A. Imhof. Imitation processes with small mutations.Journal of Economic Theory, 131(1):251–262, 2006. doi: 10.1016/j.jet.2005.04.006

    2006 doi

  17. [25]

    Fudenberg and L

    D. Fudenberg and L. A. Imhof. Monotone imitation dynamics in large populations.Journal of Economic Theory, 140(1):229–245, 2008. doi: 10.1016/j.jet.2007.08.002

    2008 doi

  18. [26]

    Fudenberg and D

    D. Fudenberg and D. K. Levine.The Theory of Learning in Games, volume 2 ofEconomic learning and social evolution. MIT Press, Cambridge, MA, 1998. 24 J. BIELA WSKI, Ł. CHOLEW A, AND F. F ALNIOWSKI

    1998

  19. [27]

    Galla and J

    T. Galla and J. D. Farmer. Complex dynamics in learning complicated games.Proceedings of the National Academy of Sciences, 110(4):1232–1236, 2013. ISSN 0027-8424. doi: 10.1073/pnas.1109672110

    2013 doi

  20. [28]

    Geller, B

    W. Geller, B. Kitchens, and M. Misiurewicz. Microdynamics for Nash maps.Discrete & Continuous Dynamical Systems, 27(3):1007–1024, 2010. doi: 10.3934/dcds.2010.27.1007

    2010 doi

  21. [29]

    Gilboa and A

    I. Gilboa and A. Matsui. Social stability and equilibrium.Econometrica: Journal of the Econometric Society, pages 859–867, 1991

    1991

  22. [30]

    Hadikhanloo, R

    S. Hadikhanloo, R. Laraki, P. Mertikopoulos, and S. Sorin. Learning in nonatomic games, Part I: Finite action spaces and population games.Journal of Dynamics and Games, 9(4, William H. Sandholm memorial issue):433–460, October 2022

    2022

  23. [31]

    D. Helbing. A mathematical model for behavioral changes by pair interactions. In G. Haag, U. Mueller, and K. G. Troitzsch, editors,Economic Evolution and Demographic Change: Formal Models in Social Sciences, pages 330–348. Springer, Berlin, 1992

    1992

  24. [32]

    Hofbauer and W

    J. Hofbauer and W. H. Sandholm. Survival of dominated strategies under evolutionary dynamics. Theoretical Economics, 6(3):341–377, 2011. doi: 10.3982/TE771

    2011 doi

  25. [33]

    Hofbauer and K

    J. Hofbauer and K. Sigmund.The Theory of Evolution and Dynamical Systems. Cambridge University Press, 1988

    1988

  26. [34]

    Hofbauer and K

    J. Hofbauer and K. Sigmund.Evolutionary Games and Population Dynamics. Cambridge University Press, Cambridge, UK, 1998

    1998

  27. [35]

    Hsieh, K

    Y.-G. Hsieh, K. Antonakopoulos, V. Cevher, and P. Mertikopoulos. No-regret learning in games with noisy feedback: Faster rates and adaptivity via learning rate separation. InNeurIPS ’22: Proceedings of the 36th International Conference on Neural Information Processing Systems, 2022

    2022

  28. [36]

    Hsieh, P

    Y.-P. Hsieh, P. Mertikopoulos, and V. Cevher. The limits of min-max optimization algorithms: Convergence to spurious non-critical sets. InICML ’21: Proceedings of the 38th International Conference on Machine Learning, 2021

    2021

  29. [37]

    Li and J

    T.-Y. Li and J. A. Yorke. Period three implies chaos.The American Mathematical Monthly, 82(10): 985–992, 1975

    1975

  30. [38]

    T.-Y. Li, M. Misiurewicz, G. Pianigiani, and J. A. Yorke. Odd chaos.Physics Letters A, 87(6):271–273,

  31. [39]

    Littlestone and M

    N. Littlestone and M. K. Warmuth. The weighted majority algorithm.Information and Computation, 108(2):212–261, 1994. doi: 10.1006/inco.1994.1009

    1994

  32. [40]

    Lotidis, P

    K. Lotidis, P. Mertikopoulos, and N. Bambos. The stability of matrix multiplicative weights dynamics in quantum games. InCDC ’23: Proceedings of the 62nd IEEE Annual Conference on Decision and Control, 2023. doi: 10.1109/CDC49753.2023.10384169

    2023

  33. [41]

    M. Lyubich. Almost every real quadratic map is either regular or stochastic.Annals of Mathematics, pages 1–78, 2002

    2002

  34. [42]

    Mertikopoulos and W

    P. Mertikopoulos and W. H. Sandholm. Riemannian game dynamics.Journal of Economic Theory, 177: 315–364, 2018

    2018

  35. [43]

    Mertikopoulos and W

    P. Mertikopoulos and W. H. Sandholm. Nested replicator dynamics, nested logit choice, and similarity- based learning. Journal of Economic Theory, 220:105881, 2024. doi: 10.1016/j.jet.2024.105881

    2024

  36. [44]

    Mertikopoulos and Y

    P. Mertikopoulos and Y. Viossat. Survival of dominated strategies under imitation dynamics.Journal of Dynamics and Games, 9(4, William H. Sandholm memorial issue):499–528, October 2022. doi: 10.3934/jdg.2022021

    2022 doi

  37. [45]

    Mertikopoulos, C

    P. Mertikopoulos, C. Papadimitriou, and G. Piliouras. Cycles in adversarial regularized learning. In Proceedings of the Twenty-Ninth Annual ACM-SIAM Symposium on Discrete Algorithms, pages 2703–2717. SIAM, 2018. doi: 10.1137/1.9781611975031.172

    2018 doi

  38. [46]

    Mertikopoulos, Y.-P

    P. Mertikopoulos, Y.-P. Hsieh, and V. Cevher. A unified stochastic approximation framework for learning in games. Mathematical Programming, 203:559–609, January 2024. doi: 10.1007/s10107-023-02001-y

    2024 doi

  39. [47]

    Milnor and C

    J. Milnor and C. Tresser. On entropy and monotonicity for real cubic maps: (with an appendix by adrien douady and pierrette sentenac).Communications in Mathematical Physics, 209:123–178, 2000. doi: 10.1007/s002200050018

    2000 doi

  40. [48]

    Misiurewicz

    M. Misiurewicz. Ergodic natural measures. InAlgebraic and topological dynamics, pages 1–6. AMS, 2005

    2005

  41. [49]

    Monderer and L

    D. Monderer and L. S. Shapley. Potential games.Games and Economic Behavior, 14(1):124–143, 1996. doi: 10.1006/game.1996.0044. THE EMERGENCE OF CHAOS IN POPULATION GAME DYNAMICS 25

    1996

  42. [50]

    Mukhopadhyay and S

    A. Mukhopadhyay and S. Chakraborty. Periodic orbit can be evolutionarily stable: Case study of discrete replicator dynamics. Journal of Theoretical Biology, 497:110288, 2020. doi: 10.1016/j.jtbi.2020.110288

    2020

  43. [51]

    Mukhopadhyay and S

    A. Mukhopadhyay and S. Chakraborty. Replicator equations induced by microscopic processes in nonoverlapping population playing bimatrix games.Chaos: An Interdisciplinary Journal of Nonlinear Science, 31(2):023123, 2021. doi: 10.1063/5.0032311

    2021 doi

  44. [52]

    evolutionary

    J. H. Nachbar. “evolutionary” selection dynamics in games: Convergence and limit properties.Interna- tional journal of game theory, 19:59–89, 1990. doi: 10.1007/BF01753708

    1990 doi

  45. [53]

    Obloj and T

    T. Obloj and T. Zenger. Organization design, proximity, and productivity responses to upward social comparison. Organization Science, 28(1):1–18, 2017

    2017

  46. [54]

    Palaiopanos, I

    G. Palaiopanos, I. Panageas, and G. Piliouras. Multiplicative weights update with constant step-size in congestion games: Convergence, limit cycles and chaos. InAdvances in Neural Information Processing Systems, pages 5872–5882, 2017

    2017

  47. [55]

    Pandit, A

    V. Pandit, A. Mukhopadhyay, and S. Chakraborty. Weight of fitness deviation governs strict physical chaos in replicator dynamics.Chaos: An Interdisciplinary Journal of Nonlinear Science, 28(3):033104,

  48. [56]

    Pangallo, J

    M. Pangallo, J. B. Sanders, T. Galla, and J. D. Farmer. Towards a taxonomy of learning dynamics in 2× 2 games. Games and Economic Behavior, 132:1–21, 2022. doi: 10.1016/j.geb.2021.11.015

    2022 doi

  49. [57]

    Piliouras and J

    G. Piliouras and J. S. Shamma. Optimization despite chaos: Convex relaxations to complex limit sets via poincaré recurrence. InProceedings of the twenty-fifth annual ACM-SIAM symposium on Discrete algorithms, pages 861–873. SIAM, 2014. doi: 10.1137/1.9781611973402.64

    2014 doi

  50. [58]

    relative concerns for consumption at the top

    C. Quintana-Domeque and J. Wohlfart. “relative concerns for consumption at the top”: An intertemporal analysis for the uk.Journal of Economic Behavior & Organization, 129:172–194, 2016

    2016

  51. [59]

    S. Ruette. Chaos on the interval, volume 67 ofUniversity Lecture Series. American Mathematical Society, 2017

    2017

  52. [60]

    Rustichini

    A. Rustichini. Optimal properties of stimulus-response learning models.Games and Economic Behavior, 29(1-2):244–273, 1999. doi: 10.1006/game.1999.0712

    1999

  53. [61]

    Rustichini

    A. Rustichini. Minimizing regret: The general case.Games and Economic Behavior, 29(1-2):224–243, October 1999. doi: 10.1006/game.1998.0690

    1999

  54. [62]

    J. B. T. Sanders, J. D. Farmer, and T. Galla. The prevalence of chaotic dynamics in games with many players. Scientific Reports, 8(1):1–13, 2018. doi: 10.1038/s41598-018-22013-5

    2018 doi

  55. [63]

    W. H. Sandholm.Population Games and Evolutionary Dynamics. MIT Press, Cambridge, MA, 2010

    2010

  56. [65]

    W. H. Sandholm. Pairwise comparison dynamics and evolutionary foundations for Nash equilibrium. Games, 1(1):3–17, 2010. doi: 10.3390/g1010003

    2010 doi

  57. [66]

    Sato and J

    Y. Sato and J. P. Crutchfield. Coupled replicator equations for the dynamics of learning in multiagent systems. Physical Review E, 67(1):015206, 2003. doi: 10.1103/PhysRevE.67.015206

    2003 doi

  58. [67]

    Y. Sato, E. Akiyama, and J. D. Farmer. Chaos in learning a simple two-person game.Proceedings of the National Academy of Sciences, 99(7):4748–4751, 2002. ISSN 0027-8424. doi: 10.1073/pnas.032086299

    2002 doi

  59. [68]

    K. H. Schlag. Why imitate, and if so, how?: A boundedly rational approach to multi-armed bandits. Journal of Economic Theory, 78(1):130–156, 1998. doi: 10.1006/jeth.1997.2347

    1998

  60. [69]

    A. N. Sharkovsky. Coexistence of the cycles of a continuous mapping of the line into itself.Ukrain. Math. Zh., 16:61–71, 1964. doi: 10.1142/S0218127495000934

    1964 doi

  61. [70]

    M. J. Smith. The stability of a dynamic model of traffic assignment - an application of a method of Lyapunov. Transportation Science, 18:245–252, 1984. doi: 10.1287/trsc.18.3.245

    1984 doi

  62. [71]

    Sparrow, S

    C. Sparrow, S. van Strien, and C. Harris. Fictitious play in 3x3 games: The transition between periodic and chaotic behaviour.Games and Economic Behavior, 63(1):259 – 291, 2008. ISSN 0899-8256. doi: 10.1016/j.geb.2007.08.005

    2008 doi

  63. [72]

    Stark, J

    O. Stark, J. Bielawski, and F. Falniowski. A class of proximity-sensitive measures of relative deprivation. Economics Letters, 160:105–110, 2017. doi: 10.1016/j.econlet.2017.08.002

    2017 doi

  64. [73]

    P. D. Taylor and L. B. Jonker. Evolutionary stable strategies and game dynamics.Mathematical Biosciences, 40(1-2):145–156, 1978. doi: 10.1016/0025-5564(78)90077-9

    1978 doi

  65. [74]

    van Strien and C

    S. van Strien and C. Sparrow. Fictitious play in 3x3 games: Chaos and dithering behaviour.Games and Economic Behavior, 73(1):262 – 286, 2011. ISSN 0899-8256. doi: 10.1016/j.geb.2010.12.004. 26 J. BIELA WSKI, Ł. CHOLEW A, AND F. F ALNIOWSKI

    2011 doi

  66. [75]

    Vilone, A

    D. Vilone, A. Robledo, and A. Sánchez. Chaos and unpredictability in evolutionary dynamics in discrete time. Physical Review Letters, 107(3):038101, 2011. doi: 10.1103/PhysRevLett.107.038101

    2011 doi

  67. [76]

    Y. Viossat. Evolutionary dynamics and dominated strategies.Economic Theory Bulletin, 3(1):91–113, April 2015. doi: 10.1007/s40505-014-0062-4

    2015 doi

  68. [77]

    V. G. Vovk. Aggregating strategies. InCOLT ’90: Proceedings of the 3rd Workshop on Computational Learning Theory, pages 371–383, 1990. doi: 10.1016/b978-1-55860-146-8.50032-1

    1990 doi

  69. [78]

    J. W. Weibull.Evolutionary Game Theory. MIT Press, Cambridge, MA, 1995. Appendix Appendix A. Proof of Theorem 1 Proof of Lemma 1.First, observe that if a continuous bimodal mapf : [0, 1] → [0, 1] with critical points cl < cr, satisfies conditions: (1′′) f (cr) > cl, (2′′) f (c...

    1995

  70. [82]

    Our next goal is to find a maph(x) for which the mapF can be presented in the form(6) where δ >0

    Therefore the fixed pointp is repelling. Our next goal is to find a maph(x) for which the mapF can be presented in the form(6) where δ >0. Let us putδ := 1 and observe that for every pointx ∈ (0, 1) simple calculations give h(x) = F (x) x(1 − x) − 1 (1 − x) = F (x) − x x(1 − x...

  71. [1982]

    doi: 10.1016/0375-9601(82)90692-2

    doi

  72. [2018]

    doi: 10.1063/1.5011955

    doi

  73. [2024]

    doi: 10.1063/5.0184318

    doi

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.