Work / Topos School

The Topos School provides an occasion for Topos researchers to keep each other up-to-date with foundational material relevant to our ongoing research programmes. While designed for an audience of Topos researchers, in the open spirit of Topos, this page shares the course notes and videos with the wider public.

Berkeley 2026

A guided tour of double categories

Speaker: Nelson Niu

Abstract: If category theory is the language of formal mathematics, then double category theory is the language of formal category theory. Not only are double categories interesting mathematical objects in their own right, providing a glimpse into higher-dimensional category theory while remaining firmly tractable; double categories are a natural setting for categorical logic, categorical systems theory, and any other field where something must be thought of as both an object and a morphism.

In this course, we aim to review the foundations of double category theory in a way that is useful for both novices and experts, theorists and practitioners alike, but assuming only 1-categorical fluency. Rather than regurgitating definitions and theorems that we could all read, we will focus on mapping out the big picture, spotlighting tricky technicalities, and clearing up misconceptions. As someone who is still learning myself, I will provide opportunities for the working double category theorists in the audience to share their own insights as well.

Possible topics include coherence and strictification of double categories; companions, conjoints, and proarrow equipments; virtual double categories; double limits; double adjunctions; and double categorical Yoneda theory.

Presenting categorical structures: a survey

Speaker: Evan Patterson

Abstract: A major program at Topos Institute is to create new technologies based on cutting-edge mathematical research in category theory and related fields. Sometimes we speak quickly of “implementing mathematics”. But what does it mean to implement a piece of mathematics? Mathematical structures, especially the very general ones in category theory, can often be concretized in many different ways. In between mathematics and implementation is something that might be called “presentation.” The word evokes familiar ideas like group presentations in abstract algebra but is construed broadly to encompass the range of formal languages, data structures, and algorithms needed to operationalize category-theoretic structures on a computer.

Presentation of categorical structures cannot yet be said to be a subject. It draws on ideas from disparate branches of mathematics and computer science, and there is no clear agreement on what are its most fundamental elements. Nevertheless, it is an essential and growing component of our work at Topos. It is a subject that we hope to help create by finding the connections between siloed fields.

This short course will offer a series of vignettes on presenting categorical structures, highlighting important techniques without attempting a unified theory. Topics may include: Algebraic approaches: free constructions in category theory, quotients, unbiasedness Combinatorial approaches: data-oriented category theory, categorical databases Diagrammatic approaches: category-theoretic diagrams, wiring diagrams Linguistic approaches: type theory, internal languages

Thinking Like a Type Theorist

Speaker: Mitchell Riley

Abstract: After working in type theory for a while, one gains a sixth sense for which systems of rules are likely to work and which are going to encounter problems. The process of inventing a type system to capture a given setting can seem mysterious, but there is a set of patterns and heuristics that explain why the rules are set up the way they are. The same thinking applies across a variety of type systems; simple/dependent, linear/non-linear, homotopical, modal, and so on.

The purpose of this course is to teach these tricks, so that attendees will have the same instincts when faced with a new type system in a new setting. Though we will use categorical ideas as inspiration, the focus will be on the systems of rules themselves rather than on categorical semantics.

Questions I hope to answer:

  • What makes a system of rules “a type theory”?
  • Where is linearity/non-linearity hiding in the rules?
  • How do we fit dependency into this picture?
  • If time permits, why does Owen keep going on about “SOGATs”?

Articulating forms that matter: The mechanics of working together

Speaker: David Spivak

Abstract: Topos Institute was designed to be a self-sustaining organization that improves the language by which humanity can communicate, hence make better collective sense of its condition, and hence more effectively direct material changes in the world. But what is the relationship between language, sensemaking, and directed material change? I do not have a complete answer, but I propose that the question is well worth pursuing. The purpose of this course is both to make this question clear as a possibility for mathematical research and to explain a certain mathematical approach to it.

The mathematics presented in this course is a compositional grammar, an operad of “responsive arrangements”, in which classical physics—heat, pressure, springs, waves, electricity, etc.—and modern learning theory are both present. The idea is that each part in an arrangement sends values forward and receives some amount of pushback, to which it responds by changing its state. For example, heat flow along a rod can be modeled as a chain of learners, each adjusting its temperature to reduce its squared difference from its neighbors. Ordinary gradient descent gives discrete heat flow; retaining momentum without damping gives discrete wave motion. In the appropriate continuum limits, these become the heat and wave equations.

The mechanisms of learning can be described within the same compositional mathematics as heat, motion, and electricity, suggesting that the physical basis of learning need not originate with organisms. The attention mechanism, which is the conceptual heart of large language models, is found within the same operad of responsive arrangements, where it appears as a technique by which parts change their communication patterns in order to better solve problems together.

This returns us to our opening question. Words do not themselves push matter; they select among what the listener is already poised to do. If this course is successful, it will merely engage a curiosity and learning ability that are already self-sustaining in the participants. My goal is to engage you in the question of how to mathematically account for where those came from.