plwonks

Polymorphic Bottom-up Weighted Relational Programming

Wed, 15 Apr 2026 00:15:00 +0000

Time and Location

Date: Wednesday, April 15
Time: 3:00PM
Location: Ballantine Hall (BH) 142

Abstract

Logic programming languages usually choose one of two approaches: top-down search (Prolog, miniKanren, etc.), or bottom-up fact collection (Datalog, SQL, etc.). Polymorphism in programming languages means operating on unknown types of data. While polymorphism usually comes for free in top-down languages, it is at odds with the fact-collection approach of bottom-up languages. How can we collect facts like “x is related to y” when we don’t know what domains x and y are drawn from?

This talk introduces semiringKanren, a bottom-up variant of miniKanren with weights, algebraic data types, and polymorphic relations. In semiringKanren, weighted relations denote finite multidimensional arrays. Particularly, even polymorphic relations are stored as a single array; there is no need to generate separate versions for each type.

The Story of Arjun Guha, or: the Arc of a Research Project

Wed, 25 Mar 2026 00:16:00 +0000

Time and Location

Date: Wednesday, March 25
Time: 4:00PM-5:30PM
Location: Luddy Hall (BLIF) 4012

Abstract

There are many ways to do research successfully, so you should take all advice with a pinch of salt. That said, I will try to describe some of the attributes that have worked well for my students and me. To make it concrete, I will use my former student, Arjun Guha, as a case study. I will strip away most of the technical detail while trying to retain the essential ideas, so people from any area of computer science will be able to follow almost all of it. Questions are welcome at any time. The talk will be followed by plenty of time for Q&A.

The Cognitive and Human Factors of Formal Methods

Wed, 25 Mar 2026 00:10:00 +0000

Time and Location

Date: Wednesday, March 25
Time: 10:00-11:30AM
Location: Luddy Hall (BLIF) 4012

Abstract

As formal methods improve in expressiveness and power, they create new opportunities for non-expert adoption. In principle, formal tools are now powerful enough to enable developers to scalably validate realistic systems artifacts without extensive formal training. However, realizing this potential for adoption requires attention to not only the technical but also the human side—which has received extraordinarily little attention from formal-methods research.

This talk presents some of our efforts to address this paucity. We apply ideas from cognitive science, human-factors research, and education theory to improve the usability of formal methods. Along the way, we find misconceptions suffered by users, how technically appealing designs that experts may value may fail to help, and how our tools may even mislead users.

How to Design Loops

Fri, 31 Oct 2025 14:00:00 +0000

Time and Location

Date: Friday, October 31
Time: 2:00-3:00PM
Location: Luddy Center for Artificial Intelligence (BLLU) 1001

Abstract

In this talk I describe a step-by-step translation process for loop construction that beginning programming students can follow. Loops in programming languages are difficult for beginning students to correctly construct. At their core, the idea is simple: repeat a given task until a condition is falsified. Loops, however, have multiple possible points of failure and opportunities for mistakes. Via a series of correctness-preserving transformations, students engage with an fully-mechanical translation pipeline that tries to removes many instances of failure and frustration when designing loops. We performed two evaluative studies on students from the CS1 and CS2 population at a large R1 midwestern university to test our methodology against the traditional approach to teaching loops in a second-semester computer science course. The first study emphasized translating tail recursive functions into loops. The second study, which builds upon the results from the first, moves students from a problem statement, a recursive function, then to its loop counterpart. The results from our first study indicate that our translation pipeline method improves students’ success in translating tail recursive functions to equivalent loops. The results of our second study indicate that more students appear to enjoy the presentation of the translation pipeline versus others’ attitudes toward the traditional means of learning loops. Combining these facts together can potentially imply that students who use the translation pipeline perform better at writing loops than those trained by traditional lectures.

Analyzing Control Flow More Like a Human

Wed, 06 Aug 2025 13:00:00 +0000

Time and Location

Date: Wednesday, August 6
Time: 1:00-2:00PM
Location: Luddy Hall 1104

Abstract

Control-flow analysis has been around for over four decades. Over this time period, the typical formulation has remained essentially an exhaustive analysis over whole programs—an analyzer ingests an entire higher-order program and produces a flow fact for each point therein. An alternative approach is demand-driven. Implementing this approach, an analyzer allows the user to indicate a point within a higher-order program and subsequently determines its flow fact, determining as few subsidiary facts as possible. The analyzer keeps the set of subsidiary facts relatively small by exploring control flow both forwards and backwards, similar to how a human might reason. This talk introduces the demand-driven approach, discusses aspects that make it particularly appealing for today’s language infrastructure and tooling, and outlines possible applications and enhancements.

Lightning talks

Fri, 28 Mar 2025 12:15:00 +0000

Time and Location

Date: Friday, March 28
Time: 12:15-1:45 PM
Location: Luddy Hall 1106

Agenda

During the afternoon of Prof. Dan Friedman’s retirement party, some current and past Wonks held a session of lightning talks.

Ron Garcia (UBC): Dan’s Mexico talk, and dynamic program slicing
Darshal Shetty (IU): Parallel deep learning with nanopasses
Lindsey Kuper (UC Santa Cruz): Communicating chorrectly with a choreography
Paulette Koronkevich (UBC): Type universes and heaps
Will Byrd (UAB): An update on quines and twines in miniKanren

Formalizing locally nameless syntax with cofinite quantification

Fri, 21 Feb 2025 15:00:00 +0000

Time and Location

Date: Friday, February 21
Time: 3:00-4:00 PM
Location: Luddy Hall 0119

Abstract

This talk explores the challenges of representing and manipulating variable bindings in formal verification and programming language theory, focusing on the formalization of the simply typed lambda calculus (STLC) in Lean using the locally nameless representation with cofinite quantification. This approach combines the readability of named representations with the automation-friendly properties of de Bruijn indices while simplifying the handling of bound variables, particularly in inductive proofs. We discuss the benefits of cofinite quantification, which enables strong induction principles. The talk covers the implementation of confluence and strong normalization proofs, highlighting the necessity of simultaneous and multivariable substitution. The talk is aimed at those interested in proof assistants, type theory, and automated reasoning, offering insights into the advantages and practical applications of this approach in Lean.

Indexed Families in Category Theory, Part III

Fri, 01 Nov 2024 14:00:00 +0000

Time and Location

Date: Friday, November 1
Time: 2:00-3:00 PM
Location: Luddy Hall 4111

Abstract

This is the third in a series of lectures introducing how the language of category theory captures “dependency.” In this lecture, we will continue our discussion of pullbacks, analyzing their relationship to fibers and base change.

In this lecture series we will start with set-indexed families of sets, generalize from Set to arbitrary categories, and conclude with category-indexed families of categories. Topics covered will include bundles and sections, slice categories, pullbacks, base change, and Grothendieck fibrations.

The categorical prerequisites are minimal: I will assume you know about categories, functors, terminal objects, and binary products. Knowledge of dependent type theory is not necessary but may provide additional motivation. If you already know what a Grothendieck fibration is, you probably won’t learn anything new.

By the end of this series, you will be prepared to understand the categorical semantics of dependent type theory (such as comprehension categories, display map categories, and categories with families / natural models), the categorical gluing approach to logical relations, and why everyone loves pullbacks so much.

Indexed Families in Category Theory, Part II

Fri, 25 Oct 2024 14:00:00 +0000

Time and Location

Date: Friday, October 25
Time: 2:00-3:00 PM
Location: Luddy Hall 4111

Abstract

This is the second in a series of lectures introducing how the language of category theory captures “dependency.” In this lecture, we will discuss slice categories and their structure. (The previous lecture was mostly devoted to background and examples, so you can still catch up if you missed last week.)

Indexed Families in Category Theory

Fri, 18 Oct 2024 14:00:00 +0000

Time and Location

Date: Friday, October 18
Time: 2:00-3:00 PM
Location: Luddy Hall 4063

Abstract

This is the first in a series of lectures introducing how the language of category theory captures “dependency.” We will start with set-indexed families of sets, generalize from Set to arbitrary categories, and conclude with category-indexed families of categories. Topics covered will include bundles and sections, slice categories, pullbacks, base change, and Grothendieck fibrations.