Equilibria Under the Probabilistic Serial Rule

Authors: Haris Aziz, Serge Gaspers, Simon Mackenzie, Nicholas Mattei, Nina Narodytska, Toby Walsh

IJCAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We conducted a series of experiments to understand the number and quality of equilibria that are possible under the PS rule.
Researcher Affiliation Academia Haris Aziz and Serge Gaspers and Simon Mackenzie and Nicholas Mattei NICTA and UNSW, Sydney, Australia {haris.aziz, serge.gaspers, simon.mackenzie, nicholas.mattei}@nicta.com.au Nina Narodytska Carnegie Mellon University, Pittsburgh, USA ninan@cs.cmu.edu Toby Walsh NICTA and UNSW, Sydney, Australia toby.walsh@nicta.com.au
Pseudocode Yes Algorithm 1: Threat profile DL-Nash equilibrium for 2 agents (which also is an EU-Nash equilibrium) which provides the same allocation as the truthful profile.
Open Source Code No The paper does not provide a direct link to the source code for its described methodology, nor does it explicitly state that its code is open-source or publicly available.
Open Datasets Yes For generating all model data we used the PREFLIB Tool Suite [Mattei and Walsh, 2013]. We also used real world data from PREFLIB [Mattei and Walsh, 2013]: AGH Course Selection (ED-00009).
Dataset Splits No The paper does not specify training, validation, or test dataset splits or mention cross-validation. It discusses generating and using '1000 samples for each combination'.
Hardware Specification Yes Each individual sample with 4 agents and 4 houses took about 15 minutes to complete using one core on an Intel Xeon E5405 CPU running at 2.0 GHz with 4 GB of RAM running Debian 6.0 (build 2.6.32-5-amd64 Squeeze10).
Software Dependencies No The paper mentions 'Debian 6.0 (build 2.6.32-5-amd64 Squeeze10)' for the operating system and 'PREFLIB Tool Suite' for data generation, but it does not specify version numbers for other key software components, libraries, or solvers directly used in their simulations.
Experiment Setup Yes We generated 1000 samples for each combination of preference model, number of agents, and number of items... In our experiments we set the probability that the second order is equivalent to the first to 0.5.