Pith. sign in

REVIEW 1 major objections 8 minor 44 references

Forgetting-Factor Regret for Online Zero-Sum Games

T0 review · 1 major / 8 minor · reviewed 2026-07-09 · glm-5.2

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

Pith's one-line read Exponential forgetting certifies real-time Nash tracking in zero-sum games

desk verdict Solid regret metric for time-varying zero-sum games; the main limitation (strong convexity + interior NE) is real but standard and acknowledged. read the letter →

arxiv 2607.07078 v1 pith:CVQH7IJI submitted 2026-07-08 math.OC

Yuhang Liu , Zi'ang Yan , Wenjun Mei , Wenxiao Zhao This is my paper · ORCID
classification math.OC
keywords regretonlinetrackingtime-varyingzero-sumalgorithmscharacterizecomputational
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 introduces a new performance metric for online two-player zero-sum games where the payoff function changes over time. Standard regret measures aggregate all past performance with equal weight, so an algorithm can look good on average while failing to track the current Nash equilibrium. The authors propose a forgetting-factor game regret that weights each past saddle gap by an exponentially decaying factor, so recent rounds dominate. When this weighted regret converges to zero, the terminal saddle gap must vanish, which means the players' strategies at the final time actually approach the equilibrium of the game being played at that moment. The paper proves this for three algorithms: projected gradient descent-ascent, a projection-free Frank-Wolfe method, and a zeroth-order finite-difference method that uses only function values. For each, the regret bound decomposes into explicit terms involving how fast the Nash equilibrium moves, how fast the payoff function itself changes, and the gradient-estimation error. If the equilibrium drift and (for some algorithms) the payoff drift vanish, the regret goes to zero and tracking is certified.

What carries the argument

The contraction factor beta(alpha) for projected gradient descent-ascent, derived from a refined gradient estimate (Lemma 2) that splits the gradient difference into a strong-convexity component and a smoothness residual, combined with a cross-gradient perturbation term bounded by Assumption 4. The same contraction structure propagates to the zeroth-order case with an additional finite-difference error term. For Frank-Wolfe, the key machinery is a duality-gap contraction (beta_FW) that uses the interiority of the Nash equilibrium to lower-bound the Frank-Wolfe gap and upper-bound the saddle gap, yielding a scalar recursion.

What would settle it

Construct a time-varying zero-sum game where the Nash equilibrium sits on the boundary of the feasible set and moves slowly, run gradient descent-ascent, and check whether the forgetting-factor regret still converges to zero. If the interior-NE condition is essential, the regret should fail to vanish even with arbitrarily slow equilibrium drift.

Watch

Extended reading notes

Core claim

The central object is the forgetting-factor game regret, defined as the sum over time of exponentially discounted two-sided saddle gaps against the current Nash equilibrium. The key mechanism is that because the forgetting factor lies strictly between zero and one, old errors are geometrically suppressed, so convergence of the weighted sum forces the most recent (undiscounted) saddle gap to zero. This converts an aggregate regret bound into a terminal tracking guarantee. The paper then shows that for gradient descent-ascent, Frank-Wolfe, and zeroth-order finite differences, the per-step tracking error contracts geometrically with a factor that depends on the strong convexity, smoothness, and

Load-bearing premise

Every payoff function must be strongly convex in one player's variable and strongly concave in the other's, and the Nash equilibrium must lie in the interior of the feasible set so that the gradient at the equilibrium is exactly zero. Without interiority, the stationarity conditions become inequalities and the contraction analysis does not go through.

Editorial extensions

If this is right

  • Any online algorithm whose tracking error contracts geometrically per step can be certified under the forgetting-factor regret framework, provided the equilibrium variation is summable or vanishing.
  • The decomposition of regret into NE variation, payoff variation, and gradient error terms suggests that adaptive step sizes or forgetting factors could be tuned to the observed rate of environmental change.
  • The zeroth-order result implies that tracking can be maintained with only function-value queries, which is relevant for adversarial or black-box settings where gradients are not available.
  • The interior-NE requirement for the Frank-Wolfe analysis suggests that constrained games with boundary equilibria would need a different gap-based contraction argument.

Reading between the lines

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

  • The forgetting-factor approach could extend to general-sum games or potential games, where the relevant benchmark would be a time-varying coarse correlated equilibrium rather than a Nash equilibrium.
  • If the forgetting factor is adapted online based on observed payoff variation, the metric could interpolate between average regret (slow environments) and tracking regret (fast environments) without prior knowledge of the drift rate.
  • The cross-gradient coupling condition (Assumption 4) likely has an information-theoretic interpretation: it bounds how much information about one player's strategy leaks through the other player's gradient, and its absence would make decentralized tracking fundamentally harder.
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

1 major / 8 minor

Summary. This paper introduces a forgetting-factor regret for online two-player zero-sum games with time-varying convex-concave payoff functions. The metric G-Regret_F_T = Σ ρ^{T-t} (f_t(x_t, y*_t) - f_t(x*_t, y_t)) assigns exponentially decaying weights to past saddle gaps, emphasizing recent performance and directly linking regret minimization to terminal Nash equilibrium (NE) tracking. Three algorithms are analyzed: projected online gradient descent-ascent (OGDA), a projection-free online Frank-Wolfe method, and a deterministic finite-difference zeroth-order method. For each, the authors establish regret bounds that explicitly decompose the effects of NE variation, payoff variation, and gradient-estimation error, and provide sufficient conditions under which G-Regret_F_T → 0, certifying asymptotic tracking of time-varying NEs.

Significance. The paper addresses a genuine gap in the online saddle-point literature: existing regret notions (individual regret, saddle-point regret, NE-regret, duality gap) aggregate historical performance with uniform weights and do not guarantee terminal tracking of a moving NE. The forgetting-factor regret metric is well-motivated and the connection to terminal tracking (Eqs. 12–13, 16) is clean and correct. The technical contributions are substantial: Lemma 2 provides a refined gradient estimate for smooth strongly convex functions (proved from first principles in Appendix B) that enables a sharp contraction factor β(α) for OGDA (Eq. 28), explicitly capturing the interplay between strong convexity, smoothness, and cross-gradient coupling. The Frank-Wolfe analysis (Theorem 2) is notable for deriving contraction on the saddle gap rather than on distance to NE, with an adaptive step size η_t = νg_t/C. The zeroth-order extension (Theorem 3) cleanly separates equilibrium variation from gradient-estimation error. The proofs are detailed and technically careful; the key identity (81) and the contraction derivation (79–95) were verified and are correct.

major comments (1)
  1. The interior NE condition (used at Eq. 82 to set ∇f_t(x*_t, y*_t) = 0) is required in all three theorems but is stated only in the theorem hypotheses, not as a standalone assumption. This is a load-bearing condition: without it, the gradient decomposition (84) fails and the contraction analysis breaks down. The authors acknowledge in Section VI that weakening strong convexity-strong concavity is future work, but the interiority limitation itself is not discussed. A brief remark (e.g., after Assumption 4 or in Section II) explaining that interiority is needed for the first-order stationarity conditions, and noting that it excludes games with active constraints at the NE, would improve transparency and help readers assess applicability.
minor comments (8)
  1. The role of the forgetting factor ρ deserves more discussion. The regret bounds (Eqs. 34, 40, 46) hold for any ρ ∈ (0,1), but the choice of ρ affects the tightness of the bound and the tracking timescale. A remark on how ρ trades off historical smoothing against tracking responsiveness would help practitioners. Additionally, the simulation uses ρ = 0.5 without justification.
  2. Section V (Simulations): The numerical example uses a payoff function (Eq. 47) with L_× = 0.1, which is a weakly coupled game. It would strengthen the paper to include a scenario with larger L_× to demonstrate that the algorithms remain effective when cross-gradient coupling is significant, since this is a central feature of the analysis. Also, the Frank-Wolfe algorithm (Algorithm 2) is included in the simulation but the payoff variation F_sup_t does not appear in the example since the payoff is piecewise constant; a case with continuous payoff variation would better illustrate the regime where F_sup_t enters the bound.
  3. Theorem 2 requires the condition ν > 0 (Eq. 38), which involves the interior distances δ_X, δ_Y and the cross-gradient constants. This condition is somewhat restrictive and its verifiability is not discussed. A remark on when ν > 0 is satisfied in practice, or at least noting that it requires the NE to be well-interior and cross-gradient coupling to be moderate, would help readers assess applicability.
  4. In Definition 1 (Eq. 15), the regret is defined as a sum from t=1 to T, but the algorithms start from t=0. The indexing convention should be clarified (e.g., whether the t=0 gap is excluded by design or folded into initialization).
  5. The convergence argument 'a vanishing sequence convolved with a stable geometric kernel remains vanishing' is invoked at the end of each proof (Appendices C–E) without explicit justification. While standard, a one-line lemma or reference would improve rigor and readability.
  6. Assumption 5 (zeroth-order) requires query access on an expanded domain (X + δ̄_x B) × (Y + δ̄_y B). For sets like probability simplices or structured polytopes, this expansion may not be natural. A brief remark on what sets satisfy this, or whether projection onto the original set suffices, would clarify the assumption's scope.
  7. The paper does not provide a formal comparison table summarizing the three algorithms' regret bounds, required assumptions, and convergence conditions side by side. Such a table would help readers quickly compare the results.
  8. Reference [42] is cited as motivation but the key differences between the OCO setting (single-agent, no cross-gradient coupling) and the game setting (two-agent, cross-gradient terms L_xy, L_yx) are not explicitly enumerated. A sentence or two highlighting what is genuinely new beyond porting the forgetting-factor idea to games would strengthen the novelty claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity found; the derivation chain is self-contained.

full rationale

The paper introduces a new regret metric (Definition 1, Eq. 15) as a definition, not a fitted quantity, and derives bounds for three algorithms from stated assumptions (Assumptions 1-5) without circular reasoning. The key identity (Eq. 81) that cross-gradient perturbations satisfy ⟨d_x^t, a_t⟩ + ⟨d_y^t, b_t⟩ = 0 is proven from the fundamental theorem of calculus, not assumed. The contraction factor β(α) in Eq. (28) is derived from Lemma 2 (a Baillon-Haddad-type gradient bound) and the update rules, not defined to guarantee the result. The sufficient conditions for G-Regret_F_T → 0 (V_t → 0, F_sup_t → 0, c_t → 0) are stated in terms of problem quantities (NE variation, payoff variation, gradient-estimation error) that are independent of the regret metric itself. The paper builds on the authors' prior work [42] (forgetting-factor regret for OCO), but this citation provides the conceptual motivation for extending the forgetting-factor idea to the game setting; the game-specific analysis (cross-gradient coupling, saddle-gap contraction, Frank-Wolfe step-size design) is developed independently within this paper. The interior-NE condition used at Eq. (82) to set ∇f_t(x*_t, y*_t) = 0 is a stated hypothesis, not a conclusion smuggled into the premises. No step in the derivation chain reduces to its inputs by construction, and no 'prediction' is a renamed fit. The derivation is self-contained against external benchmarks, so the circularity score is 0.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

No new entities, particles, forces, or dimensions are introduced. The forgetting-factor regret is a new metric definition, not a new physical or mathematical object.

free parameters (3)
  • ρ (forgetting factor) = 0.5 (in simulation)
    Algorithm parameter controlling the decay rate of historical weights. Not fitted to data; chosen by the user. Appears in Definition 1.
  • α (step size for OGDA and zeroth-order) = 1 (in simulation)
    Step size satisfying condition (29). Not fitted; constrained by problem constants μ, L, L_×.
  • c_t (perturbation radius sequence for zeroth-order) = 1/(t+1) (in simulation)
    User-chosen sequence controlling gradient estimation accuracy. Must vanish for convergence.
assumptions (6)
  • standard math Assumption 1: X, Y nonempty, compact, convex; f_t continuous and convex-concave.
    Standard assumption for zero-sum games; guarantees NE existence via Kakutani's theorem (Lemma A2).
  • domain assumption Assumption 2: f_t is μ_x-strongly convex in x and μ_y-strongly concave in y.
    Ensures unique NE (Lemma 1) and enables contraction analysis. Restrictive; acknowledged as limitation in Section VI.
  • standard math Assumption 3: f_t differentiable with L_x, L_y-smooth gradients.
    Standard smoothness assumption for gradient-based methods.
  • domain assumption Assumption 4: Cross-gradient Lipschitz conditions with constants L_xy, L_yx.
    Peculiar to game setting; controls cross-player coupling. Essential for contraction factor β(α) < 1. Discussed in Remark 1.
  • ad hoc to paper Interior NE condition: NE trajectory lies in interior of X×Y (Theorems 1-3).
    Used at equation (82) to set ∇f_t(x*_t, y*_t) = 0. Load-bearing for the contraction analysis. Not assumed in all prior work on saddle-point problems.
  • domain assumption Assumption 5: f_t queryable on expanded domain (X+δ̄_x B)×(Y+δ̄_y B).
    Needed only for zeroth-order algorithm; standard for finite-difference gradient estimation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Forgetting-Factor Regret for Online Zero-Sum Games." pith.science (2026). https://pith.science/paper/CVQH7IJI

@misc{pith2026260707078,
  author       = {Pith},
  title        = {Pith review of: Forgetting-Factor Regret for Online Zero-Sum Games},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CVQH7IJI}},
  note         = {Machine review of arXiv:2607.07078}
}
read the original abstract

This paper studies dynamic equilibrium tracking in online two-player zero-sum games with time-varying convex-concave payoff functions. Existing regret metrics for online saddle-point problems usually aggregate historical payoffs with uniform weights, and hence may fail to characterize the real-time tracking performance with respect to the current Nash equilibrium (NE). To address this issue, we introduce a zero-sum game regret function with a forgetting factor, which assigns exponentially decaying weights to past saddle gaps and emphasizes recent performance. This metric directly links regret minimization to the tracking of time-varying NEs. Within this framework, we investigate three online algorithms under different computational and information settings. For first-order feedback, we analyze projected gradient descent-ascent and design a projection-free online Frank-Wolfe method to reduce the computational cost of projections. For zeroth-order feedback, we develop a deterministic finite-difference method that only uses function-value queries. For all three algorithms, we establish forgetting-factor regret bounds that explicitly characterize the effects of NE variation, payoff variation, and gradient-estimation error. We further provide sufficient conditions under which the proposed regret converges to zero, thereby certifying asymptotic tracking of time-varying NEs. The numerical example validates the theoretical results and illustrates the tracking advantage of the proposed regret metric.

Figures

Figures reproduced from arXiv: 2607.07078 by the authors.

Figure 1
Figure 1. Estimates of saddle-point Follow-The-Leader and Algorithms 1–3 with respect to [PITH_FULL_IMAGE:figures/full_fig_p016_1.png] view at source ↗
Figure 2
Figure 2. Estimates of saddle-point Follow-The-Leader and Algorithms 1–3 with respect to [PITH_FULL_IMAGE:figures/full_fig_p017_2.png] view at source ↗
Figure 3
Figure 3. Forgetting-factor regrets of saddle-point Follow-The-Leader and Algorithms 1–3. [PITH_FULL_IMAGE:figures/full_fig_p018_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 44 canonical work pages

  1. [1]

    Equilibrium points in n-person games,

    J. F. Nash Jr, “Equilibrium points in n-person games,”Proceedings of the national academy of sciences, vol. 36, no. 1, pp. 48–49, 1950

    work page 1950

  2. [2]

    Theory of games and economic behavior: 60th anniversary commemorative edition,

    J. V on Neumann and O. Morgenstern, “Theory of games and economic behavior: 60th anniversary commemorative edition,” inTheory of games and economic behavior, Princeton university press, 2007

    work page 2007

  3. [3]

    Online convex programming and generalized infinitesimal gradient ascent,

    M. Zinkevich, “Online convex programming and generalized infinitesimal gradient ascent,” inProceedings of the 20th International Conference on Machine Learning, pp. 928–936, 2003. July 9, 2026 DRAFT 36

    work page 2003

  4. [4]

    Logarithmic regret algorithms for online convex optimization,

    E. Hazan, A. Agarwal, and S. Kale, “Logarithmic regret algorithms for online convex optimization,”Machine Learning, vol. 69, no. 2-3, pp. 169–192, 2007

    work page 2007

  5. [5]

    Zur theorie der gesellschaftsspiele,

    J. Neumann, “Zur theorie der gesellschaftsspiele,”Mathematische Annalen, vol. 100, pp. 295–320, 1928

    work page 1928

  6. [6]

    V on Neumann and O

    J. V on Neumann and O. Morgenstern,Theory of Games and Economic Behavior. Princeton: Princeton University Press, 2004

    work page 2004

  7. [7]

    Ben-Tal, L

    A. Ben-Tal, L. El Ghaoui, and A. Nemirovski,Robust optimization. Princeton University Press, 2009

    work page 2009

  8. [8]

    Game theory, online prediction and boosting,

    Y . Freund and R. E. Schapire, “Game theory, online prediction and boosting,” inProceedings of the 9th Annual Conference on Computational Learning Theory, p. 325–332, 1996

    work page 1996

Show all 44 references
  1. [9]

    Generative adversarial nets,

    I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y . Bengio, “Generative adversarial nets,” inProceedings of the 27th Advances in Neural Information Processing Systems, p. 2672–2680, 2014

    2014

  2. [10]

    Heads-up limit hold’em poker is solved,

    M. Bowling, N. Burch, M. Johanson, and O. Tammelin, “Heads-up limit hold’em poker is solved,”Science, vol. 347, no. 6218, p. 145–149, 2015

    2015

  3. [11]

    A generalization of brouwer’s fixed point theorem,

    S. Kakutani, “A generalization of brouwer’s fixed point theorem,”Duke Mathematical Journal, vol. 8, no. 3, pp. 457–459, 1941

    1941

  4. [12]

    Subgradient methods for saddle-point problems,

    A. Nedi ´c and A. Ozdaglar, “Subgradient methods for saddle-point problems,”Journal of optimization theory and applications, vol. 142, no. 1, pp. 205–208, 2009

    2009

  5. [13]

    The limit points of (optimistic) gradient descent in min-max optimization,

    C. Daskalakis and I. Panageas, “The limit points of (optimistic) gradient descent in min-max optimization,”Advances in neural information processing systems, vol. 31, 2018

    2018

  6. [14]

    Competitive gradient descent,

    F. Sch ¨afer and A. Anandkumar, “Competitive gradient descent,”Advances in Neural Information Processing Systems, vol. 32, 2019

    2019

  7. [15]

    Local saddle point optimization: A curvature exploitation approach,

    L. Adolphs, H. Daneshmand, A. Lucchi, and T. Hofmann, “Local saddle point optimization: A curvature exploitation approach,” inThe 22nd International Conference on Artificial Intelligence and Statistics, pp. 486–495, 2019

    2019

  8. [16]

    A primal-dual algorithm with line search for general convex-concave saddle point problems,

    E. Y . Hamedani and N. S. Aybat, “A primal-dual algorithm with line search for general convex-concave saddle point problems,”SIAM Journal on Optimization, vol. 31, no. 2, pp. 1299–1329, 2021

    2021

  9. [17]

    On linear convergence of iterative methods for the variational inequality problem,

    P. Tseng, “On linear convergence of iterative methods for the variational inequality problem,”Journal of Computational and Applied Mathematics, vol. 60, no. 1-2, p. 237–252, 1995

    1995

  10. [18]

    Interaction matters: a note on non-asymptotic local convergence of generative adversarial networks,

    T. Liang and J. Stokes, “Interaction matters: a note on non-asymptotic local convergence of generative adversarial networks,” inProceedings of the 22nd International Conference on Artificial Intelligence and Statistics, p. 907–915, 2019

    2019

  11. [19]

    A unified analysis of extra-gradient and optimistic gradient methods for saddle point problems: proximal point approach,

    A. Mokhtari, A. Ozdaglar, and S. Pattathil, “A unified analysis of extra-gradient and optimistic gradient methods for saddle point problems: proximal point approach,” inProceedings of the 23rd International Conference on Artificial Intelligence and Statistics, p. 1497–1507, 2020

    2020

  12. [20]

    Convergence rate of o(1/k) for optimistic gradient and extragradient methods in smooth convex-concave saddle point problems,

    A. Mokhtari, A. E. Ozdaglar, and S. Pattathil, “Convergence rate of o(1/k) for optimistic gradient and extragradient methods in smooth convex-concave saddle point problems,”SIAM Journal on Optimization, vol. 30, no. 4, pp. 3230–3251, 2020

    2020

  13. [21]

    Accelerated algorithms for smooth convex-concave minimax problems with o (1/kˆ 2) rate on squared gradient norm,

    T. Yoon and E. K. Ryu, “Accelerated algorithms for smooth convex-concave minimax problems with o (1/kˆ 2) rate on squared gradient norm,” inInternational Conference on Machine Learning, pp. 12098–12109, 2021

    2021

  14. [22]

    Linear last-iterate convergence in constrained saddle-point optimization,

    C.-Y . Wei, C.-W. Lee, M. Zhang, and H. Luo, “Linear last-iterate convergence in constrained saddle-point optimization,” inProceedings of the International Conference on Learning Representations, 2021

    2021

  15. [23]

    Min-max optimization without gradients: Convergence and applications to black-box evasion and poisoning attacks,

    S. Liu, S. Lu, X. Chen, Y . Feng, K. Xu, A. Al-Dujaili, M. Hong, and U.-M. O’Reilly, “Min-max optimization without gradients: Convergence and applications to black-box evasion and poisoning attacks,” inInternational conference on machine learning, pp. 6282–6293, 2020

    2020

  16. [24]

    Zeroth-order algorithms for nonconvex-strongly-concave minimax problems with improved complexities,

    Z. Wang, K. Balasubramanian, S. Ma, and M. Razaviyayn, “Zeroth-order algorithms for nonconvex-strongly-concave minimax problems with improved complexities,”Journal of Global Optimization, vol. 87, no. 2, pp. 709–740, 2023. July 9, 2026 DRAFT 37

    2023

  17. [25]

    Frank-wolfe algorithms for saddle point problems,

    G. Gidel, T. Jebara, and S. Lacoste-Julien, “Frank-wolfe algorithms for saddle point problems,” inArtificial Intelligence and Statistics, pp. 362–371, 2017

    2017

  18. [26]

    On frank-wolfe and equilibrium computation,

    J. D. Abernethy and J.-K. Wang, “On frank-wolfe and equilibrium computation,”Advances in Neural Information Processing Systems, vol. 30, 2017

    2017

  19. [27]

    Faster rates for convex-concave games,

    J. Abernethy, K. A. Lai, K. Y . Levy, and J.-K. Wang, “Faster rates for convex-concave games,” inConference On Learning Theory, pp. 1595–1625, PMLR, 2018

    2018

  20. [28]

    Efficient regret minimization in non-convex games,

    E. Hazan, K. Singh, and C. Zhang, “Efficient regret minimization in non-convex games,” inInternational Conference on Machine Learning, pp. 1433–1441, 2017

    2017

  21. [29]

    On gradient descent ascent for nonconvex-concave minimax problems,

    T. Lin, C. Jin, and M. Jordan, “On gradient descent ascent for nonconvex-concave minimax problems,” inInternational Conference on Machine Learning, pp. 6083–6093, PMLR, 2020

    2020

  22. [30]

    Online learning and online convex optimization,

    S. Shalev-Shwartz, “Online learning and online convex optimization,”Foundations and Trends in Machine Learning, vol. 4, no. 2, pp. 107–194, 2012

    2012

  23. [31]

    Online convex optimization in the bandit setting: gradient descent without a gradient,

    A. D. Flaxman, A. T. Kalai, and H. B. Mcmahan, “Online convex optimization in the bandit setting: gradient descent without a gradient,” inProceedings of the 16th Annual ACM-SIAM Symposium on Discrete Algorithms (SODA), pp. 385–394, 2005

    2005

  24. [33]

    Introduction to online convex optimization,

    E. Hazan, “Introduction to online convex optimization,”Foundations and Trends in Optimization, vol. 2, no. 3-4, pp. 157– 325, 2016

    2016

  25. [34]

    Dynamic regret of strongly adaptive methods,

    L. Zhang, T. Yang, and Z. Zhou, “Dynamic regret of strongly adaptive methods,” inProceedings of the 35th International Conference on Machine Learning, pp. 5882–5891, 2018

    2018

  26. [35]

    An online optimization approach for multi-agent tracking of dynamic parameters in the presence of adversarial noise,

    S. Shahrampour and A. Jadbabaie, “An online optimization approach for multi-agent tracking of dynamic parameters in the presence of adversarial noise,” in2017 American Control Conference, pp. 3306–3311, 2017

    2017

  27. [36]

    The online saddle point problem and online convex optimization with knapsacks,

    A. R. Cardoso, H. Wang, and H. Xu, “The online saddle point problem and online convex optimization with knapsacks,” Mathematics of Operations Research, vol. 50, no. 1, pp. 1–39, 2025

    2025

  28. [37]

    An online saddle point optimization algorithm with regularization,

    Y . Xu, Y . Jiang, X. Xie, and D. Li, “An online saddle point optimization algorithm with regularization,” inIOP Conference Series: Materials Science and Engineering, vol. 569, p. 052035, IOP Publishing, 2019

    2019

  29. [38]

    Competing against equilibria in zero-sum games with evolving payoffs,

    A. R. Cardoso, J. Abernethy, H. Wang, and H. Xu, “Competing against equilibria in zero-sum games with evolving payoffs,” arXiv preprint arXiv:1907.07723, 2019

    1907 arXiv

  30. [39]

    Online and bandit algorithms for nonstationary stochastic saddle-point optimization,

    A. Roy, Y . Chen, K. Balasubramanian, and P. Mohapatra, “Online and bandit algorithms for nonstationary stochastic saddle-point optimization,”arXiv preprint arXiv:1912.01698, 2019

    1912 arXiv

  31. [40]

    No-regret learning in time-varying zero-sum games,

    M. Zhang, P. Zhao, H. Luo, and Z.-H. Zhou, “No-regret learning in time-varying zero-sum games,” inProceedings of the International Conference on Machine Learning, pp. 26772–26808, 2022

    2022

  32. [41]

    Online saddle point problem and online convex-concave optimization,

    Q.-x. Meng and J.-w. Liu, “Online saddle point problem and online convex-concave optimization,”arXiv preprint arXiv:2312.06957, 2023

    2023 arXiv

  33. [42]

    Forgetting-factor regrets for online convex optimization,

    Y . Liu, W. Zhao, and G. Yin, “Forgetting-factor regrets for online convex optimization,”IEEE Transactions on Automatic Control, vol. 69, no. 8, pp. 5034–5048, 2024

    2024

  34. [43]

    Projection-free online learning,

    E. Hazan and S. Kale, “Projection-free online learning,” inProceedings of the 29th International Conference on Machine Learning, pp. 1843–1850, 2012

    2012

  35. [44]

    Distributed convergence to Nash equilibria in two-network zero-sum games,

    B. Gharesifard and J. Cort ´es, “Distributed convergence to Nash equilibria in two-network zero-sum games,”Automatica, vol. 49, no. 6, p. 1683–1692, 2013

    2013

  36. [45]

    Distributed stochastic subgradient projection algorithms for convex optimization,

    S. S. Ram, A. Nedi ´c, and V . Veeravalli, “Distributed stochastic subgradient projection algorithms for convex optimization,” Journal of Optimization Theory&Applications, vol. 147, no. 3, pp. 516–545, 2010. July 9, 2026 DRAFT

    2010

Pith tools

Reviewed July 9, 2026 · model on record in the stance chip above.