Lifting E-Graphs

This article discusses the concept of 'Lifting E-Graphs' (previously called 'Thinning E-Graphs') as a solution to problems with explicit variable names in equality saturation (EqSat). The author identifies three key issues: generative processes requiring fresh names leading to redundancy, missed sharing between terms with different variable names, and accidental over-sharing that conflates distinct mathematical objects (e.g., sin(x) as a function of one vs. two variables). The core insight is that context should be considered part of what a term is, not just where it appears. The article is technical, focused on programming language theory, e-graphs, and semantic models, making it relevant to IT/technology.