Algorithms for Max-Min Share Fair Allocation of Indivisible Chores

Authors: Haris Aziz, Gerhard Rauchecker, Guido Schryen, Toby Walsh

AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we report the results of our computational study where we tested the performance of our SCHED heuristic (Algorithm 3). We show that SCHED returns near-optimal solutions and greatly outperforms the simpler greedy round-robin protocol (Algorithm 1).
Researcher Affiliation Academia Haris Aziz Data61 and UNSW Sydney, NSW 2052, Australia haris.aziz@data61.csiro.au; Gerhard Rauchecker University of Regensburg 93053 Regensburg, Germany gerhard.rauchecker@ur.de; Guido Schryen University of Regensburg 93053 Regensburg, Germany guido.schryen@ur.de; Toby Walsh UNSW, Data61 and TU Berlin Sydney, NSW 2052, Australia toby.walsh@data61.csiro.au
Pseudocode Yes Algorithm 1 Greedy round-robin protocol; Algorithm 2 PTAS for optimal Mm S; Algorithm 3 SCHED heuristic for optimal Mm S
Open Source Code No The paper does not provide any statement or link indicating that the source code for the described methodology is publicly available.
Open Datasets No Utilities were drawn from a uniform distribution ui(j) U(0, 100) (and negated for obtaining chores instances) as it is common in the literature for fair division of goods (Amanatidis et al. 2015; Bouveret and Lemaˆıtre 2016; Kurokawa, Procaccia, and Wang 2016). This describes data generation, not the use of a publicly accessible dataset with concrete access info.
Dataset Splits No The paper does not explicitly provide information about training, validation, or test dataset splits. It describes generating instances for computational experiments but not how they are partitioned for model evaluation.
Hardware Specification Yes We coded all algorithms in C++ on a Linux Cent OS 7 based 12-core processor with a clock speed of 3.07 GHz and 24 Gi B memory.
Software Dependencies Yes Integer linear programs were solved via the Gurobi 6 C++ API.
Experiment Setup Yes The time limit in step 4 of SCHED was set to Tmax = 300s.