Expected Variational Inequalities

Authors: Brian Hu Zhang, Ioannis Anagnostides, Emanuel Tewolde, Ratip Emin Berker, Gabriele Farina, Vincent Conitzer, Tuomas Sandholm

ICML 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental First, there exist games for which a CE need not be a solution to the ΦLIN-EVI. In this sense, ΦLIN-EVIs yield a computationally tractable superset of Nash equilibria that is tighter than CEs. Second, computation suggests that the set of solutions of the ΦLIN-EVI for the game need not be a polyhedron, unlike the set of CEs. We provide a graphical depiction of this phenomenon in Figure 1. The figure depicts the set of ΦLIN-EVI solutions to a simple Bach or Stravinsky game... The set of marginals of ΦLIN-EVI solutions appears to have a curved boundary corresponding, we believe, to the hyperbola 10x2 - 25xy + 10y2 - 6x + 11y = 0.
Researcher Affiliation Collaboration 1Carnegie Mellon University 2Foundations of Cooperative AI Lab (FOCAL) 3Massachusetts Institute of Technology 4University of Oxford 5Strategy Robot, Inc. 6Strategic Machine, Inc. 7Optimized Markets, Inc.
Pseudocode Yes Algorithm 1 Ellipsoid against hope (EAH) under SEPor GER oracle (Daskalakis et al., 2025) input Parameters Ry, ry > 0 such that Bry( ) Y BRy(0) precision parameter ϵ > 0 constant B 1 such that µ A B for all µ (X) a SEPor GER oracle output A sparse, ϵ-approximate solution µ (X) of (14)
Open Source Code No The paper does not contain any explicit statement about releasing its own source code, nor does it provide a link to a code repository.
Open Datasets No The paper discusses theoretical concepts and uses a simple "Bach or Stravinsky game" as an illustrative example. This is a theoretical game definition, not an empirical dataset. The paper does not mention using or making available any publicly accessible datasets for experiments.
Dataset Splits No The paper primarily presents theoretical work and uses conceptual examples rather than empirical datasets. Therefore, there is no mention of dataset splits such as training, validation, or test sets.
Hardware Specification No The paper does not provide any specific details about the hardware used for computations or simulations. It focuses on theoretical algorithms and complexity without mentioning hardware specifications like CPU/GPU models or memory.
Software Dependencies No The paper does not mention any specific software dependencies or versions (e.g., programming languages, libraries, frameworks with version numbers) used for implementing the algorithms or conducting the analyses.
Experiment Setup No The paper primarily focuses on theoretical analysis and algorithm development. It does not describe any experimental setup with specific hyperparameters, training configurations, or system-level settings for machine learning models or other empirical systems.