Pith. sign in

REVIEW 4 major objections 4 minor 38 references

Constant Weighted Maximin Share Approximations for Chores

T0 review · 4 major / 4 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read The paper establishes the first constant-factor (12) approximation algorithm for weighted maximin share allocations of indivisible chores, and proves no algorithm can beat factor 2.

desk verdict First constant-factor WMMS for chores, plausible but not fully verified; the advertised two-agent characterization is missing from the body. read the letter →

arxiv 2510.06581 v2 pith:OWWNVGD2 submitted 2025-10-08 cs.GT

Bo Li , Fangxiao Wang , Shiji Xing This is my paper · ORCID
classification cs.GT MSC 91B32
keywords weightedmaximinsharechoresindivisibleitemsfairdivisionapproximationalgorithmcanonicalinstanceproxycostfunctionlowerbound
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 weighted maximin share (WMMS), the standard fairness notion for dividing indivisible chores among agents with different weights. Its central result is that for every number of agents and every weight distribution, an allocation exists in which no agent's cost exceeds 12 times her WMMS. This is the first constant-factor guarantee for WMMS on chores, replacing the previous logarithmic bound with a fixed number. The proof reduces arbitrary instances to highly structured canonical instances, then shows a greedy item-by-item rule with careful tie-breaking always completes. It also proves a lower bound: no algorithm can guarantee better than a factor of 2, leaving the optimal ratio between 2 and 12.

What carries the argument

The load-bearing object is the canonical instance: normalized weights and item values that are powers of 1/2 with WMMS equal to weight. The reduction theorem maps any instance to a canonical one at a factor-4 cost. Inside the canonical algorithm, the central mechanism is the proxy cost function v*_i for each delegating agent, a step function built only from group weights that is claimed to upper-bound the agent's true cost on every item tail. Lemma 1—the union of delegating agents' life windows is the whole item set—guarantees the greedy loop never runs out of active agents, and Claim 6 (proxy cost bounds true cost) is the inequality that makes Lemma 1 go through.

What would settle it

Construct a canonical instance and test Claim 6 directly: find a delegating agent a_i and an index t > d_i such that v_i([d_i+1, t]) exceeds v*_i([d_i+1, t]). Since the proof of Claim 6 is the deferred step in Lemma 1, any such pair would mean Algorithm 1 can run out of active agents, and the 3-WMMS (hence 12-WMMS) guarantee would fail. A simpler check is to run Algorithm 1 on random canonical instances and see whether any round has no active agent.

Watch

Extended reading notes

Core claim

The core discovery is that WMMS for chores admits a constant-factor approximation. Any instance can be transformed into a canonical instance—weights and item values are powers of 1/2, and every agent's WMMS equals her weight—at a factor-4 loss in the approximation ratio. On canonical instances, a deterministic greedy rule that hands each item to an active agent who values it least (and removes an agent once her cost reaches 3 times her weight) guarantees 3-WMMS. The proof's hinge is Lemma 1: at every step some active agent remains, which is shown by introducing delegating agents whose combined 'life windows' cover all items and whose proxy cost functions upper-bound true costs. The existenti

Load-bearing premise

The claim that every item eventually gets allocated depends on Claim 6: for every delegating agent, the proxy cost function built from group weights alone never underestimates the agent's true cost on an item tail; if that inequality fails for some canonical instance, the greedy algorithm could stall and the 12-factor guarantee collapses.

Editorial extensions

If this is right

  • Every weighted-chores instance, even with wildly asymmetric weights, has an allocation where no agent pays more than 12 times her WMMS.
  • A (24+epsilon)-WMMS allocation can be computed in polynomial time for any fixed epsilon.
  • No allocation can guarantee a factor better than 2 in the worst case, so the optimal approximation ratio is between 2 and 12.
  • Future improvements need only be demonstrated on canonical instances with power-of-two structure, since the reduction to that form loses only a fixed factor of 4.

Reading between the lines

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

  • The 12-factor is the product of the factor-4 reduction and the factor-3 canonical algorithm; any tightening of the canonical guarantee automatically improves the overall ratio.
  • The lower-bound construction uses groups of exponentially growing size, suggesting the true worst-case ratio may lie close to 2 for instances with many distinct weight levels.
  • Because Claim 6's proof is deferred and is the load-bearing inequality, a numerical search over random canonical instances for violations of the proxy bound would be a direct and cheap stress test of the 12-approximation.
  • The abstract's two-agent characterization is not developed in the body of this version; readers seeking that curve should find it in the cited two-agent algorithm rather than in the proofs here.
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

4 major / 4 minor

Summary. The paper studies weighted maximin share (WMMS) for the allocation of indivisible chores. Its headline claim is the first constant-factor approximation: the authors introduce canonical instances (power-of-two weights and item values, with each agent's WMMS equal to her weight), prove that any instance can be reduced to a canonical instance at a factor-4 loss (Theorem 1), and give a simple greedy algorithm (Algorithm 1) that is 3-WMMS on canonical instances (Theorem 3). Combining these yields a 12-WMMS guarantee, with a polynomial-time (24+ε)-WMMS corollary via a rounding argument (Theorem 2). The paper also presents a lower bound showing that no algorithm can beat (2−ε)-WMMS, improving the previous 1.366 bound.

Significance. If the main claims were correct, this would resolve a well-known open problem in weighted fair division, improving the best previous upper bound from O(log n) to a constant. The canonical-instance framework, delegation of agents, and proxy cost functions are genuinely interesting technical tools that could be useful beyond this paper. The improved lower bound of 2−ε is also a worthwhile contribution, and the simple 3-WMMS algorithm for canonical instances is elegant. However, I find two load-bearing gaps in the proof of the central result—the I3→I4 reduction in Theorem 1 and the feasibility claim in Theorem 2—and an advertised two-agent characterization that does not appear in the body. The paper is not yet in a publishable form, though the approach may be repairable, in particular by accepting an arbitrarily small additive loss in the approximation ratio.

major comments (4)
  1. [Appendix A, I3→I4 (Claim 13)] The reduction from I3 to I4 rounds each item value down to the nearest power-of-(1/2) fraction of w1 and then claims that the residual v_i(B_i^j) − v'_i(B_i^j) can be written as a finite sum Σ_t w1/2^{q_t}. This is false for arbitrary additive valuations. For example, take w1 = 1/2 and an item with value 1/6. The round-down gives 1/8, leaving residual 1/24, which is not a finite dyadic rational multiple of w1. No finite set of powers of 1/2 sums to 1/24. Consequently, the auxiliary-item construction cannot simultaneously preserve the total value and the canonical power-of-two item-value property. Since this step is essential to Theorem 1, the factor-4 reduction—and hence the 12-approximation for arbitrary instances—is unsupported as written. The proof might be repaired by restricting to dyadic valuations, or by accepting an (1+ε) loss and proving a 12+ε guarantee, but the current text do
  2. [Theorem 2 / Appendix B] The linear program used in the proof restricts each item to agents with v_i(e_j) ≤ WMMS_i (or ≤ (1+ε)WMMS_i in the second version). The proof asserts that the existence of an integral α-approximate WMMS allocation implies feasibility of this LP. This implication is not valid for arbitrary α: an α-WMMS allocation may assign an item to an agent for whom that single item has value greater than WMMS_i, since α-WMMS only bounds total bundle cost, not per-item eligibility. Indeed, for chores, for every agent i and every item e, one has v_i(e) ≥ WMMS_i (put e in i's own bundle in the WMMS-defining partition), so the eligibility constraint can hold only at equality. The threshold in the LP must involve α (e.g., α(1+ε)WMMS_i) and the rounding factor must be re-derived accordingly. As written, the polynomial-time corollary (24+ε)-WMMS does not follow.
  3. [Abstract and Section 1] The abstract advertises a precise characterization of the optimal approximation ratio curve for every possible weight distribution in the two-agent case, and states that the results imply non-existence of WMMS for two agents with different weights. However, the body contains no such characterization. The two-agent case is discussed only by citing Wang et al. [WLL24] as already optimal (ratio 1.366). The example with weights 1−ε and ε does show non-existence of an exact WMMS allocation, but it is not a characterization of the full approximation-ratio curve. This advertised contribution must either be added or removed from the abstract and introduction.
  4. [Appendix C (proof of Claim 6)] Claim 6 is a key ingredient in Lemma 1, which guarantees that Algorithm 1 never stalls. Its proof in Appendix C is difficult to verify as written: the notation l_{i,j} is not defined in the main text or the appendix, the text switches between l_{i,j} and l_i without explanation, the inequality 'v_{i,j}(e_h) > l_i' appears to be a typo for '> w_i', and the phrase 'k > l_i + n_i' introduces an undefined k. Since the correctness of the entire canonical-instance algorithm depends on this claim, the proof needs to be rewritten with consistent notation and a complete, checkable argument.
minor comments (4)
  1. [Section 5.2, Claim 9] In the paragraph after the first estimate, 'i−β_{p−1} of the items in M^2_p' should be '1−β_{p−1}'.
  2. [Section 5.2, LP display] The displayed linear program has a likely index error: the constraint α_2 ≥ 2−4/Δ + β_2 − β_{p−1} should presumably be β_2 − β_1. The subsequent sentence 'the bottleneck of α_2 through α_{k−1} is the largest difference between two consecutive β' confirms this.
  3. [Section 5.1] The notation 'n_i = |N_i|' is inconsistent with the rest of the paper, where groups are denoted G_i; this should be n_i = |G_i|.
  4. [Claim 7] The proof of Claim 7 would benefit from stating explicitly that d_{r2} ≤ L_{r2} (from Claim 1) and from clarifying the relationship between the relabeling of items after d_{r2} and the set [1, L_{r2}].

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the canonical-instance reduction, Algorithm 1, and the lower bound are self-contained; self-citations are background.

full rationale

The paper's central derivation is not circular. The canonical-instance property WMMS_i = w_i is not assumed as the main result; it is manufactured in the reduction I1->I2->I3 of Appendix A by adding zero/auxiliary items and scaling, so WMMS values in the canonical instance are known by construction from the original WMMS partitions. Theorem 3's 3-WMMS guarantee for Algorithm 1 is then proved from canonical properties (power-of-two item values, IDO, WMMS_i=w_i) via Claims 1-8; Claim 6's deferred proof uses only the definition of a WMMS-defining partition to bound the capacity of early bundles, not the desired 12-approximation. The constants 3, 4, and 12 are analytic bounds, not fitted parameters later renamed as predictions. Self-citations to [WLL24] and [LWZ23] are used for background (previous O(log n) bound), for motivation, and as one of several citations for the standard identical-ordering reduction ([BL16, BK20, LWZ23]); they do not carry the load-bearing uniqueness or ansatz. The dyadic-rounding construction in Appendix A is a potential correctness detail, but it does not make the theorem's output equal to its input by construction.

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

The paper introduces mathematical devices (canonical instances, delegating agents, proxy cost functions) but these are proof constructs, not new physical or empirical entities requiring independent evidence. There are no data-fitted parameters; lower-bound constants k and Δ are adversarial-instance parameters, not fitted values.

assumptions (3)
  • domain assumption An arbitrary weighted-chores instance can be transformed into an identical-ordering instance without changing the best WMMS approximation ratio.
    Invoked in Section 3 (step I4→I5) and cited to [BL16, BK20, LWZ23]; load-bearing for Theorem 1.
  • standard math WMMS_i equals w_i times the minimum makespan of a related-machine scheduling instance, and this makespan has a PTAS (Hochbaum–Shmoys).
    Used in Theorem 2 / Appendix B to compute (1+ε)-approximate WMMS values in polynomial time.
  • domain assumption Valuations are additive, weights can be normalized to sum to 1, and WMMS is defined via the minimum maximum cost ratio.
    Section 2 definitions; these are standard model assumptions for the paper's setting.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Constant Weighted Maximin Share Approximations for Chores." pith.science (2026). https://pith.science/paper/OWWNVGD2

@misc{pith2026251006581,
  author       = {Pith},
  title        = {Pith review of: Constant Weighted Maximin Share Approximations for Chores},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OWWNVGD2}},
  note         = {Machine review of arXiv:2510.06581}
}
abstract

We study the fair allocation of indivisible chores among agents with asymmetric weights. Among the various fairness notions, weighted maximin share (WMMS) stands out as particularly compelling. Despite its appeal, the existence of a constant-factor approximation for WMMS has remained an important open problem in weighted fair division [Aziz et al., 2022, Suksompong, 2025]. Prior to our work, the best known approximation ratio was $O(\log n)$, where $n$ is the number of agents. In this paper, we make significant progress by presenting the first constant-factor approximation algorithm for WMMS. Our main contributions are as follows: [1] We design the first algorithm that guarantees a 12-approximate WMMS allocation, substantially improving upon the previous $O(\log n)$ upper bound. Our approach introduces a novel analytical framework based on canonical instance reductions, agent delegation, and proxy cost functions to effectively bound agents' costs. Additionally, we provide a polynomial-time implementation for any approximate WMMS algorithm, incurring a factor of 2 loss in the approximation ratio. [2] We present an improved worst-case lower bound, showing that no algorithm can achieve better than 2-approximate WMMS, thereby strengthening the previous best lower bound of 1.366. We further construct a general hard instance, which provides lower bounds for an arbitrary number of agents. [3] Beyond worst-case bounds, we precisely characterize the optimal approximation ratio curve for every possible weight distribution in the two-agent case. Notably, our results imply that a WMMS allocation may not exist for any two agents with different weights, in sharp contrast to the symmetric case where an MMS allocation always exists.

Figures

Figures reproduced from arXiv: 2510.06581 by the authors.

Figure 1
Figure 1. An illustration of delegating agents N ∗ . For each interval ari , the left dashed boundary means that ai becomes active at item dri+1, and the right double lines mean ai becomes inactive after receiving this item. In the following, we prove that the collective life window of delegating agents can cover all items, which accordingly proves Theorem 3. Lemma 1. S ai∈N ∗ Φi = [1, m]. 9 [PITH_FULL_IMAGE:figures/full_fig… view at source ↗
Figure 2
Figure 2. An Illustration of vi(·). Next, we prove that each agent’s weighted maximin share is exactly her weight. Lemma 2. For any group of agents Gi , WMMSi = wi . Proof. For the sole agent in G1, one WMMS-defining partition is as follows: • The agent in G1 gets the sole item in M1; • Each agent ai,j ∈ Gi for i ≥ 2 gets either exactly one item in M1 i or exactly two items in M2 i . This way, all the items are allocated. Bes… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 1 linked inside Pith

  1. [1]

    Voudouris

    Georgios Amanatidis, Georgios Birmpas, Aris Filos - Ratsikas, and Alexandros A. Voudouris. Fair division of indivisible goods: A survey. In IJCAI , pages 5385--5393. ijcai.org, 2022

    2022

  2. [2]

    On truthful mechanisms for maximin share allocations

    Georgios Amanatidis, Georgios Birmpas, and Evangelos Markakis. On truthful mechanisms for maximin share allocations. In IJCAI , pages 31--37. IJCAI/AAAI Press, 2016

    2016

  3. [3]

    Weighted maxmin fair share allocation of indivisible chores

    Haris Aziz, Hau Chan, and Bo Li. Weighted maxmin fair share allocation of indivisible chores. In IJCAI , pages 46--52. ijcai.org, 2019

    2019

  4. [4]

    Breaking the 3/4 barrier for approximate maximin share

    Hannaneh Akrami and Jugal Garg. Breaking the 3/4 barrier for approximate maximin share. In SODA , pages 74--91. SIAM , 2024

    2024

  5. [5]

    Algorithmic fair allocation of indivisible items: a survey and new questions

    Haris Aziz, Bo Li, Herv \' e Moulin, and Xiaowei Wu. Algorithmic fair allocation of indivisible items: a survey and new questions. SIGecom Exch. , 20(1):24--40, 2022

    2022

  6. [6]

    A polynomial-time algorithm for computing a pareto optimal and almost proportional allocation

    Haris Aziz, Herv \' e Moulin, and Fedor Sandomirskiy. A polynomial-time algorithm for computing a pareto optimal and almost proportional allocation. Oper. Res. Lett. , 48(5):573--578, 2020

    2020

  7. [7]

    Algorithms for max-min share fair allocation of indivisible chores

    Haris Aziz, Gerhard Rauchecker, Guido Schryen, and Toby Walsh. Algorithms for max-min share fair allocation of indivisible chores. In AAAI , pages 335--341. AAAI Press, 2017

    2017

  8. [8]

    Fair-share allocations for agents with arbitrary entitlements

    Moshe Babaioff, Tomer Ezra, and Uriel Feige. Fair-share allocations for agents with arbitrary entitlements. Math. Oper. Res. , 49(4):2180--2211, 2024

    2024

Show all 38 references
  1. [9]

    Approximation algorithms for maximin fair division

    Siddharth Barman and Sanath Kumar Krishnamurthy. Approximation algorithms for maximin fair division. ACM Trans. Economics and Comput. , 8(1):5:1--5:28, 2020

    2020

  2. [10]

    Characterizing conflicts in fair division of indivisible goods using a scale of criteria

    Sylvain Bouveret and Michel Lema \^ tre. Characterizing conflicts in fair division of indivisible goods using a scale of criteria. Auton. Agents Multi Agent Syst. , 30(2):259--290, 2016

    2016

  3. [11]

    The combinatorial assignment problem: approximate competitive equilibrium from equal incomes

    Eric Budish. The combinatorial assignment problem: approximate competitive equilibrium from equal incomes. In BQGT , page 74:1. ACM , 2010

    2010

  4. [12]

    Fair public decision making

    Vincent Conitzer, Rupert Freeman, and Nisarg Shah. Fair public decision making. In EC , pages 629--646. ACM , 2017

    2017

  5. [13]

    Weighted envy-freeness in indivisible item allocation

    Mithun Chakraborty, Ayumi Igarashi, Warut Suksompong, and Yair Zick. Weighted envy-freeness in indivisible item allocation. ACM Trans. Economics and Comput. , 9(3):18:1--18:39, 2021

    2021

  6. [14]

    Procaccia, Nisarg Shah, and Junxing Wang

    Ioannis Caragiannis, David Kurokawa, Herv \' e Moulin, Ariel D. Procaccia, Nisarg Shah, and Junxing Wang. The unreasonable fairness of maximum nash welfare. ACM Trans. Economics and Comput. , 7(3):12:1--12:32, 2019

    2019

  7. [15]

    Pennock, Masoud Seddighin, Saeed Seddighin, and Hadi Yami

    Alireza Farhadi, Mohammad Ghodsi, Mohammad Taghi Hajiaghayi, S \' e bastien Lahaie, David M. Pennock, Masoud Seddighin, Saeed Seddighin, and Hadi Yami. Fair allocation of indivisible goods to asymmetric agents. J. Artif. Intell. Res. , 64:1--20, 2019

    2019

  8. [16]

    On picking sequences for chores

    Uriel Feige and Xin Huang. On picking sequences for chores. In EC , pages 626--655. ACM , 2023

    2023

  9. [17]

    Resource allocation and the public sector

    Duncan Karl Foley. Resource allocation and the public sector. Yale Economics Essays , 7(1):45--98, 1967

    1967

  10. [18]

    Fair allocation of indivisible goods: Improvements and generalizations

    Mohammad Ghodsi, Mohammad Taghi Hajiaghayi, Masoud Seddighin, Saeed Seddighin, and Hadi Yami. Fair allocation of indivisible goods: Improvements and generalizations. In EC , pages 539--556. ACM , 2018

    2018

  11. [19]

    A survey on fair allocation of chores

    Hao Guo, Weidong Li, and Bin Deng. A survey on fair allocation of chores. Mathematics , 11(16):3616, 2023

    2023

  12. [20]

    Near fairness in matroids

    Laurent Gourv \` e s, J \' e r \^ o me Monnot, and Lydia Tlilane. Near fairness in matroids. In ECAI , volume 263 of Frontiers in Artificial Intelligence and Applications , pages 393--398. IOS Press, 2014

    2014

  13. [21]

    An algorithmic framework for approximating maximin share allocation of chores

    Xin Huang and Pinyan Lu. An algorithmic framework for approximating maximin share allocation of chores. In EC , pages 630--631. ACM , 2021

    2021

  14. [22]

    A polynomial approximation scheme for scheduling on uniform processors: Using the dual approximation approach

    Dorit S Hochbaum and David B Shmoys. A polynomial approximation scheme for scheduling on uniform processors: Using the dual approximation approach. SIAM journal on computing , 17(3):539--551, 1988

    1988

  15. [23]

    Fair and efficient resource allocation with partial information

    Daniel Halpern and Nisarg Shah. Fair and efficient resource allocation with partial information. In IJCAI , pages 224--230. ijcai.org, 2021

    2021

  16. [24]

    A reduction from chores allocation to job scheduling

    Xin Huang and Erel Segal - Halevi. A reduction from chores allocation to job scheduling. In EC , page 908. ACM , 2023

    2023

  17. [25]

    Procaccia, and Junxing Wang

    David Kurokawa, Ariel D. Procaccia, and Junxing Wang. Fair enough: Guaranteeing approximate maximin shares. J. ACM , 65(2):8:1--8:27, 2018

    2018

  18. [26]

    Almost (weighted) proportional allocations for indivisible chores

    Bo Li, Yingkai Li, and Xiaowei Wu. Almost (weighted) proportional allocations for indivisible chores. In WWW , pages 122--131. ACM , 2022

    2022

  19. [27]

    Lipton, Evangelos Markakis, Elchanan Mossel, and Amin Saberi

    Richard J. Lipton, Evangelos Markakis, Elchanan Mossel, and Amin Saberi. On approximately fair allocations of indivisible goods. In EC , pages 125--131. ACM , 2004

    2004

  20. [28]

    Fair allocation of indivisible chores: Beyond additive costs

    Bo Li, Fangxiao Wang, and Yu Zhou. Fair allocation of indivisible chores: Beyond additive costs. In NeurIPS , 2023

    2023

  21. [29]

    Almost envy-free allocations of indivisible goods or chores with entitlements

    Max Springer, MohammadTaghi Hajiaghayi, and Hadi Yami. Almost envy-free allocations of indivisible goods or chores with entitlements. In AAAI , pages 9901--9908. AAAI Press, 2024

    2024

  22. [30]

    Lower bound for online MMS assignment of indivisible chores

    Masoud Seddighin and Saeed Seddighin. Lower bound for online MMS assignment of indivisible chores. CoRR , abs/2507.12984, 2025

    2025 arXiv

  23. [31]

    Shmoys and \' E va Tardos

    David B. Shmoys and \' E va Tardos. An approximation algorithm for the generalized assignment problem. Math. Program. , 62:461--474, 1993

    1993

  24. [32]

    The problem of fair division

    Hugo Steinhaus. The problem of fair division. Econometrica , 16(1):101--104, 1948

    1948

  25. [33]

    Online MMS allocation for chores

    Jiaxin Song, Biaoshuai Tao, Wenqian Wang, and Yuhao Zhang. Online MMS allocation for chores. CoRR , abs/2507.14039, 2025

    2025

  26. [34]

    Weighted fair division of indivisible items: A review

    Warut Suksompong. Weighted fair division of indivisible items: A review. Inf. Process. Lett. , 187:106519, 2025

    2025

  27. [35]

    Equity, envy, and efficiency

    Hal R Varian. Equity, envy, and efficiency. 1973

    1973

  28. [36]

    Improved approximation of weighted MMS fairness for indivisible chores

    Fangxiao Wang, Bo Li, and Pinyan Lu. Improved approximation of weighted MMS fairness for indivisible chores. In IJCAI , pages 3014--3022. ijcai.org, 2024

    2024

  29. [37]

    Weighted EF1 allocations for indivisible chores

    Xiaowei Wu, Cong Zhang, and Shengwei Zhou. Weighted EF1 allocations for indivisible chores. In EC , page 1155. ACM , 2023

    2023

  30. [38]

    Multi-agent online scheduling: MMS allocations for indivisible items

    Shengwei Zhou, Rufan Bai, and Xiaowei Wu. Multi-agent online scheduling: MMS allocations for indivisible items. In ICML , volume 202 of Proceedings of Machine Learning Research , pages 42506--42516. PMLR , 2023

    2023

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.