Date of Award

2022

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Graduate Group

Computer and Information Science

First Advisor

Stephanie Weirich

Abstract

Embedding describes the process of encoding a program's syntax and/or semantics in another language---typically a theorem prover in the context of mechanized reasoning. Among different embedding styles, deep embeddings are generally preferred as they enable the most faithful modeling of the original language. However, deep embeddings are also the most complex, and working with them requires additional effort. In light of that, this dissertation aims to draw more attention to alternative styles, namely shallow and mixed embeddings, by studying their use in mechanized reasoning about programs' properties that are related to "how". More specifically, I present a simple shallow embedding for reasoning about computation costs of lazy programs, and a class of mixed embeddings that are useful for reasoning about properties of general computation patterns in effectful programs. I show the usefulness of these embedding styles with examples based on real-world applications.

Share

COinS