A One-Saddle 1/5 Approximation Algorithm for Common-Kernel Bimatrix Games: Recognition, Exact Segment Optimization, Sharp Selector Bounds, and Certified Robustness

Davit Gondauri

Home/Archive/arXiv

arXiv · 2609.33471

A One-Saddle 1/5 Approximation Algorithm for Common-Kernel Bimatrix Games: Recognition, Exact Segment Optimization, Sharp Selector Bounds, and Certified Robustness

Davit Gondauri

View original source ↗Read source document ↗Export metadata ↓

Abstract

We study a fixed-normalization class of symmetric bimatrix games generated by a common kernel and show that it admits efficient approximation despite exact symmetric-equilibrium computation remaining PPAD-hard. For every rational game in the class, a single auxiliary zero-sum saddle computation yields a rational symmetric $1/5$-approximate Nash equilibrium in polynomial time. We also give exact recognition and unique kernel recovery, and an exact polynomial-time post-processing algorithm that minimizes regret along the segment joining the selected saddle strategies. For arbitrary rational square games, we formulate the nearest common-kernel projection as a linear program with an explicit dual certificate, obtaining a certified $(1/5+2η^*)$-approximation guarantee. A separate selector result proves a sharp regret-to-uniformity constant for the full-subset construction. The results provide a tractable and certifiable structured-game regime and clarify which common-kernel reduction architectures cannot support certain fine-grained approximation-hardness objectives.

Explore related subjects

econ.EM

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Davit Gondauri. 2026-09-27. A One-Saddle 1/5 Approximation Algorithm for Common-Kernel Bimatrix Games: Recognition, Exact Segment Optimization, Sharp Selector Bounds, and Certified Robustness. https://arxiv.org/abs/2609.33471

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Identification and Estimation in a Nonseparable Instrumental Variable Model with Multivariate Outcomes

This paper proposes a novel nonseparable instrumental variable model to address endogeneity in settings with multivariate outcomes. While traditional approaches typically estimate separate one-dimensional quantile models for each outcome, the inherent correlation among different dimensions frequently invalidates their fundamental assumptions. To overcome this limitation, our framework jointly models the distribution of multiple outcomes. By assuming that the structural function is cyclically monotone, the model accommodates flexible dependence across outcome dimensions. We establish the local identification of these structural functions in the presence of discrete treatments and instruments under a completeness-type condition, which reduces to the well-known full-rank condition in one dimension. Furthermore, we introduce a computationally tractable estimator and prove its uniform consistency. Simulation studies demonstrate the estimator's stable performance in finite samples.

econ.EM↗

arXiv2026

Difference-in-Differences Meets Synthetic Control: Doubly Robust Identification and Estimation

Difference-in-Differences (DiD) and Synthetic Control (SC) are widely used methods for causal inference in panel data, each with distinct strengths and limitations. We propose a novel method for short-panel causal inference that integrates the advantages of both approaches. Our method delivers a doubly robust identification strategy for the average treatment effect on the treated (ATT) under either of two non-nested assumptions: parallel trends or a group-level SC condition. Building on this identification result, we develop a unified semiparametric framework for estimating the ATT. Notably, the identification-robust moment function satisfies Neyman orthogonality under the parallel trends assumption but not under the SC assumption, leading to different asymptotic variances across the two identification strategies. To ensure valid inference, we propose a multiplier bootstrap method that consistently approximates the asymptotic distribution under either assumption. Furthermore, we extend our methodology to accommodate repeated cross-sectional data and staggered treatment designs. As an empirical application, we evaluate the impact of the 2003 minimum-wage increase in Alaska on family income. Finally, in simulation studies based on empirically calibrated data-generating processes, we demonstrate that the proposed estimation and inference methods perform well in finite samples under either identification assumption.

econ.EM↗

arXiv2026

Simple Specification-Free Causal Inference in Large Panels

We propose a simple and transparent approach to causal inference using panel data with a large number of units and periods. Instead of imposing a particular parametric specification on the untreated potential outcomes, we assume they follow a nonparametric nonseparable factor model. We use long pre-treatment outcome histories to construct a measure of similarity between the latent factors of different units. Using this measure, we impute missing counterfactual means and propensity scores through kernel smoothing. Under weak smoothness conditions, these estimates can attain the optimal nonparametric convergence rate up to logarithmic factors, provided that the pre-treatment history grows sufficiently quickly with the sample size. We construct a doubly robust estimator of the period-specific average treatment effect on the treated (ATT) and provide conditions under which it is $\sqrt{N}$-consistent and asymptotically normal, centered at the true ATT. Simulations demonstrate accurate inference for a wide range of linear and nonlinear data-generating processes.

econ.EM↗