Pith. sign in

REVIEW 2 major objections 4 minor 36 references

Reduced Network Cumulative Constraint Violation for Distributed Bandit Convex Optimization under Slater Condition

T0 review · 2 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Pith's one-line read A distributed bandit algorithm with an unclipped dual update achieves O(log T) cumulative constraint violation under Slater's condition.

desk verdict Solid rate results for distributed bandit constrained optimization under Slater; the main theorems hold, but the stated parameter range for the known-strong-convexity result needs a small fix. read the letter →

arxiv 2411.11574 v2 pith:VXQCPXIZ submitted 2024-11-18 eess.SY cs.SY

classification eess.SYcs.SY
keywords banditconvexoptimizationdistributedcumulativeconstraintviolationSlater'sconditiontime-varyingconstraintsprimal-dualalgorithmtwo-pointfeedbacknetworkregret
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 tackles distributed online decision-making in which networks of agents choose points, see only noisy function values (bandit feedback), and are penalized both for regret and for violating time-varying inequality constraints. Its central claim is that a primal–dual algorithm that updates dual variables by maximizing the regularized Lagrangian in closed form—rather than by plugging in the clipped constraint function—keeps Slater's condition effective. Under that condition, if local losses are strongly convex with known parameter, the algorithm attains O(log T) network regret and O(log T) network cumulative constraint violation; without Slater, the violation bound is only O($\sqrt$(T log T)). For merely convex losses, the violation bound improves from O($T^{{1-c/2}}$) to O($T^{{1-c}}$) when Slater holds. This matters because earlier distributed bandit algorithms that clipped constraints could not exploit Slater's condition at all.

What carries the argument

The load-bearing object is the regularized Lagrangian dual update q_{i,t+1} = γ_t[g_{i,t}(x_{i,t})]_+, combined with two-point stochastic subgradient estimators and consensus mixing over a time-varying directed graph. Its role is to let Slater's fixed point x_s enter the analysis through inequality (106), converting a uniform feasibility margin into a linear lower bound on the dual objective; this is the step that turns square-root violation rates into direct rates.

What would settle it

Run Algorithm 1 on a network with known strongly convex quadratic losses and time-varying linear constraints for which a fixed x_s with g_t(x_s) ≤ -ς_s 1_m exists, using the step sizes of Theorem 4, and measure average Net-CCV(T) over many random seeds. If it grows faster than C log T for T up to $10^{5}$, Theorem 4's bound (28) is contradicted; an algebraic check of inequality (106) along a simulated trajectory would locate whether the margin inequality or the subsequent bound fails.

Watch

Extended reading notes

Core claim

The paper's discovery is that the clipping operation [g_t(x)]_+, used by earlier algorithms to measure cumulative constraint violation, is what destroys Slater's condition: after clipping, a strictly feasible point x_s no longer gives a negative term in the dual update, and inequality (106) becomes nonnegative. Algorithm 1 instead keeps the original constraint function inside the regularized Lagrangian L_t(x,q) = (1/n)Σ f_{i,t}(x) + q^T g_t(x) - (1/(2γ_t))||q||^2, so the dual update has the closed form q_{i,t+1}=γ_t[g_{i,t}(x_{i,t})]_+. Slater's margin then yields inequality (106), Σ_{i,t} q_{i,t+1}^T g_{i,t}(x_s)/γ_t ≤ -ς_s Σ_{i,t} ||[g_{i,t}(x_{i,t})]_+||, which makes the cumulative violation linearly bounded by the regularized dual objective instead of by its square root. The theorems state the resulting rates: convex losses give Net-CCV O($T^{{1-c}}$) under Slater; strongly convex losses with known μ give Net-Reg O(log T) and Net-CCV O(log T) under Slater, versus O($\sqrt$(T log T)) without.

Load-bearing premise

The whole reduction depends on one fixed decision that strictly satisfies every constraint at every time step with a positive margin; if the feasible region drifts so no such point exists, the improved violation bounds no longer follow.

Editorial extensions

If this is right

  • When a fixed strictly feasible point exists, distributed bandit networks with strongly convex known losses can achieve logarithmic regret and logarithmic cumulative constraint violation simultaneously, matching full-information rates in the prior work this paper builds on.
  • For convex losses, Slater's condition improves the cumulative constraint violation exponent by c/2, from O(T^{1-c/2}) to O(T^{1-c}), without worsening the regret bound.
  • The unclipped dual update also removes the need for uniformly bounded constraint functions, an assumption the prior clipped algorithm required.
  • In the strongly convex case with unknown parameter, the Slater improvement still holds for violations, giving O(T^{1-c}) instead of O(T^{1-c/2}) while regret improves to O(T^{1-c}).

Reading between the lines

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

  • Because the proof's key inequality depends on one fixed x_s common to all t, the reduction should fail exactly when the strictly feasible region moves over time; testing a shifting-feasibility instance would likely show the O(log T) violation bound reverting to the unimproved rate.
  • The same unclipped regularized-Lagrangian trick may transfer to other bandit settings with cumulative penalties, such as distributed online learning with communication compression, where clipping currently blocks Slater-type arguments.
  • The two-point estimator's smoothing bias introduces O(δ_t) error, so a one-point feedback version would face larger gaps; an analogous reduction under Slater may require different estimators or slower rates.
  • A practical reading is that the gap between O(log T) and O(sqrt(T log T)) is the quantitative price of not knowing a strictly feasible point, suggesting system designers could enforce Slater's margin as a safety requirement.
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

2 major / 4 minor

Summary. This paper studies distributed bandit convex optimization with time-varying inequality constraints under two-point bandit feedback for both losses and constraints. The authors propose a distributed primal-dual algorithm in which dual variables are updated by maximizing the regularized Lagrangian rather than by substituting the clipped constraint function, while primal variables are updated by projected gradient steps using two-point stochastic subgradient estimators and consensus mixing. The main theoretical results are: for convex losses, network regret O(T^{max{c,1-c}}) and network cumulative constraint violation O(T^{1-c/2}), reduced to O(T^{1-c}) under Slater's condition (Theorems 1 and 2); for strongly convex losses with unknown parameter, regret O(T^{1-c}) and violation O(T^{1-c/2}), reduced to O(T^{1-c}) under Slater's condition (Theorem 3); and for known strong convexity parameter, regret O(log T) and violation O(sqrt(T log T)), reduced to O(log T) under Slater's condition (Theorem 4). Detailed proofs with explicit constants are given in appendices, and a numerical example compares the proposed algorithm with prior work.

Significance. This is a potentially significant contribution. The paper identifies a real limitation of prior distributed bandit algorithms that use clipped constraint functions in the dual update, namely that Slater's condition cannot improve cumulative constraint violation bounds under that design. The proposed dual update via the regularized Lagrangian is a clean fix, and the key inequality (106) genuinely converts the uniform Slater margin into a linear lower bound on the dual objective, enabling the improved violation rates. The proof structure follows and extends known techniques from [28] and [29], with detailed appendix proofs and explicit constants. If the parameter-range issue in Theorem 4 is corrected, the results would be the first reduced cumulative constraint violation bounds under Slater's condition for distributed bandit convex optimization, matching the full-information rates of [29].

major comments (2)
  1. [Section III-C, Theorem 4, Eq. (25) and Algorithm 1] Theorem 4 as stated does not cover all mu>0 allowed by Assumption 6. With alpha_t = 1/(mu t), for any mu<1 we have alpha_1 = 1/mu > 1, so xi_1 > 1, contradicting the Algorithm 1 input condition xi_t in (0,1). Consequently the set (1 - xi_1)X is not contained in X, and for a general convex X the observation points x_i,1 + delta_1 u_i,1 need not lie in X, so Assumptions 2 and 3 cannot be applied at t=1. The proof's telescoping identities (118) and the cancellation in (135) rely on the exact choice alpha_t = 1/(mu t), so this is not merely a cosmetic issue. The theorem statement and proof must be revised, for example by restricting to mu >= 1 or by shifting the stepsize to alpha_t = 1/(mu(t + t0)) and reworking the affected estimates.
  2. [Appendix A, Lemma 1] Lemma 1 as stated is false for arbitrary b in R^p. Taking K = [0,1], b = 10, and a = 0 gives x_c = 1 and ||x_c - b|| = 9 > 0 = ||a||, violating (30). The proof invokes the optimality condition with b, which is valid only when b belongs to K. In all applications in the paper, including (42) and (64), b = x_i,t does belong to K = (1 - xi_{t+1})X because z_j,t in (1 - xi_t)X and xi_t is non-increasing, so the applications are sound. Nevertheless, the lemma statement should be corrected to include the hypothesis b in K, and the proof should be adjusted accordingly.
minor comments (4)
  1. [Algorithm 1 input and Theorems 1, 3, 4] In Theorems 1 and 3, Eq. (16) gives xi_1 = alpha_1 = 1, and in Theorem 4, Eq. (25) gives xi_1 = 1 when mu = 1, both violating the stated condition xi_t in (0,1). Please either explicitly allow xi_1 = 1 or redefine the sequences for t >= 2.
  2. [Appendix A, Lemma 1 proof] The strong convexity inequality in the proof of Lemma 1 is missing the square on ||b - x_c||; the displayed argument should be corrected. A simpler proof of (30), under the added hypothesis b in K, is to specialize (29) to y = b.
  3. [Eqs. (78) and (118)] The notation 1/alpha_0 is used without defining alpha_0; please state the convention, presumably 1/alpha_0 = 0.
  4. [Lemma 6 and Appendix B] The constant epsilon_1 is used in (65) and (85) but only tilde{epsilon}_1 is defined in Lemma 5; the notation should be aligned to avoid ambiguity.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the Slater-based violation reduction follows from a genuine inequality (106) and the proofs in the appendices are self-contained.

full rationale

The central claims, Theorems 1–4, are derived in the appendices from the algorithm and stated assumptions rather than assumed. The key Slater improvement is inequality (106), obtained by combining Assumption 5, g_t(x_s) ≤ -ς_s 1_m, with the explicit dual update q_{i,t+1}=γ_t[g_{i,t}(x_{i,t})]_+, yielding h_T(x_s) ≤ -ς_s ∑_i∑_t ‖[g_{i,t}(x_{i,t})]_+‖. This is a direct calculation, not a restatement of the desired CCV bound. The O(log T) regret and CCV results in Theorem 4 follow from the stated stepsize choice α_t=1/(μt) through inequalities (115), (116), (121), (128)–(139), with explicit constant expressions; no quantity is fitted to data and no conclusion is used as an input. The paper does import Lemmas 2 and 3 from the authors' earlier work [28], and the algorithmic idea of keeping Slater's condition effective is credited to [29], but these are technical estimator/consensus bounds and a design template, not the target logarithmic or reduced-violation theorems; the cited lemmas have stated assumptions that do not include the target result, and the new proof sections carry the load. The numerical example is illustrative and does not claim to predict fitted data. A separate concern that Theorem 4's initialization may be ill-posed when μ<1 because ξ_1=α_1=1/μ can exceed 1, so (1−ξ_1)X need not lie inside X, is a correctness/parameter-range matter rather than circularity, since the theorem's conclusions are not assumed in the stepsize specification. Overall, the derivation chain is not circular.

Assumptions & free parameters 4 free parameters · 7 assumptions · 0 invented entities

The central rates depend on standard explicit assumptions: bounded losses, Lipschitz subgradients, graph connectivity, Slater's condition, and strong convexity. Two technical lemmas are imported from the authors' earlier paper [28], which is a self-citation but not circular because those lemmas do not contain the target result. The user parameter c and constant gamma0 are design choices, not fitted values. No new entities are introduced.

free parameters (4)
  • c = user-defined in (0,1)
    Trade-off parameter in stepsize alpha_t = t^{-c}; all bounds depend on c; chosen by the user to balance regret and violation, not fitted to data.
  • gamma0 = in (0, 1/(4(p^2+1)G2^2)]
    Constant in dual regularization and stepsize gamma_t = gamma0/alpha_t; chosen from known G2 and ensures inequality (52); affects constants but not rates.
  • r(X), R(X)
    Geometry constants from Assumption 1; used in exploration step and in bounds; assumed known inputs, not fitted.
  • mu = unknown in Theorem 3, known in Theorem 4
    Strong convexity parameter; used to set alpha_t = 1/(mu t) in Theorem 4 and appears in constants in Theorem 3; a problem parameter, not fitted.
assumptions (7)
  • domain assumption Assumption 1: X contains a ball of radius r(X) and is contained in a ball of radius R(X).
    Ensures exploration points x+delta u remain in X and bounds are finite; standard for bandit convex optimization.
  • domain assumption Assumption 2: |f_i,t(x)| <= F for all i,t,x.
    Uniform boundedness of losses is used in the dual bound (53).
  • domain assumption Assumption 3: f_i,t and g_i,t are convex and have subgradients bounded by G1 and G2.
    Lipschitz-type condition on functions; used throughout the regret and violation analysis.
  • domain assumption Assumption 4: doubly stochastic mixing matrices with B-connected union graph.
    Controls consensus error and is standard in distributed optimization; used in Lemma 3.
  • domain assumption Assumption 5: Slater's condition with a fixed x_s and uniform margin ς_s.
    Critical for reduced violation; provides the negative margin in (106); without it only the non-Slater bounds hold.
  • domain assumption Assumption 6: strong convexity of local losses with parameter mu.
    Used in Theorems 3 and 4 to improve regret via telescoping cancellations.
  • domain assumption Lemma 2 and Lemma 3 from [28] (two-point estimator properties and consensus error bound).
    The paper quotes these lemmas from its own earlier work without reproducing their proofs; they are published results but are self-cited and not contained in this paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Reduced Network Cumulative Constraint Violation for Distributed Bandit Convex Optimization under Slater Condition." pith.science (2026). https://pith.science/paper/VXQCPXIZ

@misc{pith2026241111574,
  author       = {Pith},
  title        = {Pith review of: Reduced Network Cumulative Constraint Violation for Distributed Bandit Convex Optimization under Slater Condition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VXQCPXIZ}},
  note         = {Machine review of arXiv:2411.11574}
}
read the original abstract

This paper studies the distributed bandit convex optimization problem with time-varying inequality constraints, where the goal is to minimize network regret and cumulative constraint violation. To calculate network cumulative constraint violation, existing distributed bandit online algorithms solving this problem directly use the clipped constraint function to replace its original constraint function. However, the use of the clipping operation renders Slater condition (i.e, there exists a point that strictly satisfies the inequality constraints at all iterations) ineffective to achieve reduced network cumulative constraint violation. To tackle this challenge, we propose a new distributed bandit online primal-dual algorithm. If local loss functions are convex, we show that the proposed algorithm establishes sublinear network regret and cumulative constraint violation bounds. When Slater condition holds, the network cumulative constraint violation bound is reduced. In addition, if local loss functions are strongly convex, for the case where strongly convex parameters are unknown, the network regret bound is reduced. For the case where strongly convex parameters are known, the network regret and cumulative constraint violation bounds are further reduced. To the best of our knowledge, this paper is among the first to establish reduced (network) cumulative constraint violation bounds for (distributed) bandit convex optimization with time-varying constraints under Slater condition. Finally, a numerical example is provided to verify the theoretical results.

Figures

Figures reproduced from arXiv: 2411.11574 by the authors.

Figure 1
Figure 1. Evolutions of 1 n Pn i=1 PT t=1 ft(xi,t). 0 200 400 600 800 1000 101 102 103 104 Algorithm 1 in this paper Algorithm 2 in [28] Algorithm 1 in [29] [PITH_FULL_IMAGE:figures/full_fig_p019_1.png] view at source ↗
Figure 2
Figure 2. Evolutions of 1 n Pn i=1 PT t=1 k[gt(xi,t)]+ k. V. CONCLUSIONS This paper studied the distributed bandit convex optimization problem with time-varying inequality constraints. We proposed a new distributed bandit online primal–dual algorithm, and established network regret and cumulative constraint violation bounds for convex and strongly [PITH_FULL_IMAGE:figures/full_fig_p019_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 33 canonical work pages

  1. [29]

    Distributed online convex optimization with adversarial constraints: Reduced cumulative constra int violation bounds under slater’s condition,

    X. Yi, X. Li, T. Yang, L. Xie, Y . Hong, T. Chai, and K. H. Joh ansson, “Distributed online convex optimization with adversarial constraints: Reduced cumulative constra int violation bounds under slater’s condition,” arXiv preprint arXiv:2306.00149, 2023

  2. [28]

    Regret and cumulative constraint violation analysis for distributed online constrained convex optimization,

    X. Yi, X. Li, T. Yang, L. Xie, T. Chai, and K. H. Johansson, “Regret and cumulative constraint violation analysis for distributed online constrained convex optimization,” IEEE Transactions on Automatic Control , vol. 68, no. 5, pp. 2875– 2890, 2023

  3. [1]

    Introduction to online convex optimization,

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

  4. [2]

    A survey on distributed online op timization and online games,

    X. Li, L. Xie, and N. Li, “A survey on distributed online op timization and online games,” Annual Reviews in Control , vol. 56, p. 100904, 2023

  5. [3]

    Online energ y generation scheduling for microgrids with intermittent energy sources and co-generation,

    L. Lu, J. Tu, C.-K. Chau, M. Chen, and X. Lin, “Online energ y generation scheduling for microgrids with intermittent energy sources and co-generation,” ACM SIGMETRICS Performance Evaluation Review , vol. 41, no. 1, pp. 53–66, 2013

  6. [4]

    P eak-aware online economic dispatching for microgrids,

    Y . Zhang, M. H. Hajiesmaili, S. Cai, M. Chen, and Q. Zhu, “P eak-aware online economic dispatching for microgrids,” IEEE Transactions on Smart Grid , vol. 9, no. 1, pp. 323–335, 2018

  7. [5]

    Dynami c right-sizing for power-proportional data centers,

    M. Lin, A. Wierman, L. L. Andrew, and E. Thereska, “Dynami c right-sizing for power-proportional data centers,” IEEE/ACM Transactions on Networking , vol. 21, no. 5, pp. 1378–1391, 2012

  8. [6]

    Data ce nter demand response: Avoiding the coincident peak via workload shifting and local generation,

    Z. Liu, A. Wierman, Y . Chen, B. Razon, and N. Chen, “Data ce nter demand response: Avoiding the coincident peak via workload shifting and local generation,” in the ACM SIGMETRICS/International Conference on Measureme nt and Modeling of Computer Systems , 2013, pp. 341–342

Show all 36 references
  1. [7]

    Pricing data cente r demand response,

    Z. Liu, I. Liu, S. Low, and A. Wierman, “Pricing data cente r demand response,” ACM SIGMETRICS Performance Evaluation Review, vol. 42, no. 1, pp. 111–123, 2014

  2. [8]

    Online conv ex optimization in the bandit setting: Gradient descent without a gradient,

    A. D. Flaxman, A. T. Kalai, and H. B. McMahan, “Online conv ex optimization in the bandit setting: Gradient descent without a gradient,” in the Sixteenth Annual ACM-SIAM Symposium on Discrete Algori thms, 2005, pp. 385–394

  3. [9]

    Improved regret guarantees for on line smooth convex optimization with bandit feedback,

    A. Saha and A. Tewari, “Improved regret guarantees for on line smooth convex optimization with bandit feedback,” in International Conference on Artificial Intelligence and St atistics, 2011, pp. 636–642

  4. [10]

    The price of bandit i nformation for online optimization,

    V . Dani, S. M. Kakade, and T. Hayes, “The price of bandit i nformation for online optimization,” in Advances in Neural Information Processing Systems , 2008, pp. 345–352

  5. [11]

    Logarithmic regret a lgorithms for online convex optimization,

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

  6. [12]

    Optimal algorithms f or online convex optimization with multi-point bandit feed back,

    A. Agarwal, O. Dekel, and L. Xiao, “Optimal algorithms f or online convex optimization with multi-point bandit feed back,” in Conference on Learning Theory , 2010, pp. 28–40

  7. [13]

    Online convex programming and generali zed infinitesimal gradient ascent,

    M. Zinkevich, “Online convex programming and generali zed infinitesimal gradient ascent,” in International Conference on Machine Learning , 2003, pp. 928–936. 40

  8. [14]

    Trading regret for effic iency: Online convex optimization with long term constrain ts,

    M. Mahdavi, R. Jin, and T. Yang, “Trading regret for effic iency: Online convex optimization with long term constrain ts,” Journal of Machine Learning Research , vol. 13, no. 1, pp. 2503–2528, 2012

  9. [15]

    Online convex optimization with ti me-varying constraints and bandit feedback,

    X. Cao and K. R. Liu, “Online convex optimization with ti me-varying constraints and bandit feedback,” IEEE Transactions on Automatic Control , vol. 64, no. 7, pp. 2665–2680, 2019

  10. [16]

    Distributed optimization for con trol,

    A. Nedi´ c and J. Liu, “Distributed optimization for con trol,” Annual Review of Control, Robotics, and Autonomous Systems , vol. 1, pp. 77–103, 2018

  11. [17]

    A survey of distributed optimization,

    T. Yang, X. Yi, J. Wu, Y . Y uan, D. Wu, Z. Meng, Y . Hong, H. Wa ng, Z. Lin, and K. H. Johansson, “A survey of distributed optimization,” Annual Reviews in Control , vol. 47, pp. 278–305, 2019

  12. [18]

    Distributed mirr or descent for online composite optimization,

    D. Y uan, Y . Hong, D. W. C. Ho, and S. Xu, “Distributed mirr or descent for online composite optimization,” IEEE Transactions on Automatic Control , vol. 66, no. 2, pp. 714–729, 2021

  13. [19]

    Push-sum distributed online optimization with bandit feedback,

    C. Wang, S. Xu, D. Y uan, B. Zhang, and Z. Zhang, “Push-sum distributed online optimization with bandit feedback,” IEEE Transactions on Cybernetics , vol. 52, no. 4, pp. 2263–2273, 2020

  14. [20]

    Decentralized online convex opti mization with event-triggered communications,

    X. Cao and T. Bas ¸ar, “Decentralized online convex opti mization with event-triggered communications,” IEEE Transactions on Signal Processing , vol. 69, pp. 284–299, 2021

  15. [21]

    Distributed online bandit learning in dynamic environments over unbalanced digraphs,

    J. Li, C. Li, W. Y u, X. Zhu, and X. Y u, “Distributed online bandit learning in dynamic environments over unbalanced digraphs,” IEEE Transactions on Network Science and Engineering , vol. 8, no. 4, pp. 3034–3047, 2021

  16. [22]

    Dist ributed online convex optimization with compressed communication,

    Z. Tu, X. Wang, Y . Hong, L. Wang, D. Y uan, and G. Shi, “Dist ributed online convex optimization with compressed communication,” in Advances in Neural Information Processing Systems , 2022, pp. 34 492–34 504

  17. [23]

    Distribute d online and bandit convex optimization,

    K. K. Patel, A. Saha, L. Wang, and N. Srebro, “Distribute d online and bandit convex optimization,” in OPT 2022: Optimization for Machine Learning , 2022

  18. [24]

    Event -triggered distributed online convex optimization with de layed bandit feedback,

    M. Xiong, B. Zhang, D. Y uan, Y . Zhang, and J. Chen, “Event -triggered distributed online convex optimization with de layed bandit feedback,” Applied Mathematics and Computation , vol. 445, p. 127865, 2023

  19. [25]

    Distributed online l inear regressions,

    D. Y uan, A. Proutiere, and G. Shi, “Distributed online l inear regressions,” IEEE Transactions on Information Theory , vol. 67, no. 1, pp. 616–639, 2021

  20. [26]

    Online convex optimization f or cumulative constraints,

    J. Y uan and A. Lamperski, “Online convex optimization f or cumulative constraints,” in Advances in Neural Information Processing Systems, 2018, pp. 6140–6149

  21. [27]

    Distributed online o ptimization with long-term constraints,

    D. Y uan, A. Proutiere, and G. Shi, “Distributed online o ptimization with long-term constraints,” IEEE Transactions on Automatic Control, vol. 67, no. 3, pp. 1089–1104, 2022

  22. [30]

    Boyd and L

    S. Boyd and L. V andenberghe, Convex Optimization. Cambridge University Press, 2004

  23. [31]

    Online convex optimization with stochastic constraints,

    H. Y u, M. Neely, and X. Wei, “Online convex optimization with stochastic constraints,” Advances in Neural Information Processing Systems, vol. 30, 2017

  24. [32]

    Online convex optimization with t ime-varying constraints,

    M. J. Neely and H. Y u, “Online convex optimization with t ime-varying constraints,” arXiv preprint arXiv:1702.04783 , 2017. 41

  25. [33]

    Online learning and online convex optimization,

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

  26. [34]

    Distributed on line convex optimization with time-varying coupled inequa lity constraints,

    X. Yi, X. Li, L. Xie, and K. H. Johansson, “Distributed on line convex optimization with time-varying coupled inequa lity constraints,” IEEE Transactions on Signal Processing , vol. 68, pp. 731–746, 2020

  27. [35]

    Distributed bandit online convex optimization with time- varying coupled inequality constraints,

    X. Yi, X. Li, T. Yang, L. Xie, T. Chai, and K. H. Johansson, “Distributed bandit online convex optimization with time- varying coupled inequality constraints,” IEEE Transactions on Automatic Control , vol. 66, no. 10, pp. 4620–4635, 2021

  28. [36]

    An optimal algorithm for bandit and zero-or der convex optimization with two-point feedback,

    O. Shamir, “An optimal algorithm for bandit and zero-or der convex optimization with two-point feedback,” Journal of Machine Learning Research , vol. 18, no. 1, pp. 1703–1713, 2017

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.