WG211/M25Steuwer
E-Graphs are a core data structure used in automatic reasoning and program optimization. They space-efficiently represent a set of equivalent terms.
However, representing programs with bindings in e-graphs using existing techniques quickly defeats the efficiency purpose in many practical applications, as the e-graph represents alpha-equivalent terms many times.
In this talk, I am going to present our work on Slotted E-Graphs that augment E-Graphs with a build-in notion of binders. By taking inspirations from the nominal approach, we are able to build a data-structure that is capable of space-efficiently representing programs with binders and the guarantee that alpha-equivalent terms are represented uniquely.
This talk presents work that has been published at PLDI 2025.