Nonarchimedean groups and the axiom of choice
We prove several novel connections between properties of nonarchimedean groups and fragments of the axiom of choice which hold in their associated permutation model.
math.LO↗
arXiv subjects
Justin Young
Publications and source records attributed to Justin Young.
Reevaluating Causal Estimation Methods with Data from a Product Release
Recent developments in causal machine learning methods have made it easier to estimate flexible relationships between confounders, treatments and outcomes, making unconfoundedness assumptions in causal analysis more palatable. How successful are these approaches in recovering ground truth baselines? In this paper we analyze a new data sample including an experimental rollout of a new feature at a large technology company and a simultaneous sample of users who endogenously opted into the feature. We find that recovering ground truth causal effects is feasible -- but only with careful modeling choices. Our results build on the observational causal literature beginning with LaLonde (1986), offering best practices for more credible treatment effect estimation in modern, high-dimensional datasets.
econ.EM↗
Permutation Models Arising From Topological Ideals
A recent paper by Zapletal arXiv:2404.10612 discusses permutation models of set theory which arise from dynamical ideals and highlights properties of the dynamical ideal which relate to fragments of choice in the permutation model. In this paper, we provide several examples from topology which illustrate using these connections to argue that the corresponding permutation model satisfies either the axiom of countable choice or well-ordered choice.
math.LO↗
Forecasting Algorithms for Causal Inference with Panel Data
Conducting causal inference with panel data is a core challenge in social science research. We adapt a deep neural architecture for time series forecasting (the N-BEATS algorithm) to more accurately impute the counterfactual evolution of a treated unit had treatment not occurred. Across a range of settings, the resulting estimator (``SyNBEATS'') significantly outperforms commonly employed methods (synthetic controls, two-way fixed effects), and attains comparable or more accurate performance compared to recently proposed methods (synthetic difference-in-differences, matrix completion). An implementation of this estimator is available for public use. Our results highlight how advances in the forecasting literature can be harnessed to improve causal inference in panel data settings.
econ.EM↗
Brace Bar-Cobar Duality
Using Kadeishvili's formulas with appropriate signs, we show that the classical cobar construction from coalgebras to algebras Ω: CoAlg -> Alg can be enhanced to a functor from Hopf algebras to E_2 algebras (for a certain choice of E_2 operad) Ω: HopfAlg -> E_2 Alg, which, unlike its classical counterpart, is not strictly adjoint, but homotopically equivalent to the left adjoint of the enhanced bar construction B: E_2 Alg -> HopfAlg studied by Gerstenhaber--Voronov and Fresse.
math.AT↗