Performance Between Universes: Type-Informed Optimisation for Structural Abstraction and Environment-Specific Data

Abstract

We propose an implementation process for type systems to permit subdivision of types into distinct universes based on individual type theories each presenting their own unique properties. Multi-universe systems can interface between smooth and discrete definitions of similar values at the type level. We then present a potential resolution framework for these values to be incrementally optimised by a preprocessor for both self-defined primitive types and high-dimensional abstractions.