Topology seminars: 2023-24

Cubical sets and cubical categories are widely used in mathematics and computer science, from homotopy theory to homotopy type theory, higher-dimensional automata and, last but not least,  higher-dimensional rewriting, where our own interest in these structures lies. To formalise cubical categories with the Isabelle/HOL proof assistant along the path of least resistance, we take a single-set approach to categories, which leads to new axioms for cubical categories. Taming the large number of initial candidate axioms has relied essentially on Isabelle's proof automation. Yet we justify their correctness relative to the standard axiomatisation by Al Agl, Brown and Steiner via categorical equivalence proofs outside of Isabelle. In combination, these results present a case study in experimental mathematics with a proof assistant.  In this talk I will focus on the formalisation experience -- lights and shadows -- and conclude with some general remarks about formalised mathematics. This is joint work with Philippe Malbos and Tanguy Massacrier (Université Claude Bernard Lyon 1).

Different flavours of string diagrams arise naturally in studying algebraic structures (e.g. algebras, Hopf algebras, Frobenius algebras) in monoidal categories. In particular, closed diagrams can be realized as scalar invariants. For a structure of a given type the closed diagrams form a commutative algebra that has a richer structure of a self dual Hopf algebra. This is very similar, but not quite the same, as the positive self adjoint Hopf algebras that were introduced by Zelevinsky in studying families of representations of finite groups. In this talk I will show that the algebras of invariants admit a lattice that is a PSH-algebra. This will be done by considering maps between invariants, and realizing them as covering spaces. I will then show some applications to subgroup growth questions, and a formula that relates the Kronecker coefficients to finite index subgroups of free groups.

Polyhedral products are a topological space formed by gluing together ingredient spaces in a manner governed by a simplicial complex. They appear in many areas of study, including toric topology, combinatorics, commutative algebra, complex geometry and geometric group theory. A fundamental problem is to determine how operations on simplicial complexes change the topology of the polyhedral product. In this talk, we consider the substitution complex operation. We obtain a description of the loop space associated with some substitution complexes, and use this to build a new family of simplicial complexes such that the homotopy type of the loop space of the moment angle complex is a product of spheres and loops on spheres.

The Joker is a famous very singular example of an endotrivial module over the 8-dimension subHopf algebra of the mod 2 Steenrod algebra generated by $\operatorname{Sq}^1$ and $\operatorname{Sq}^2$. It is known that this can be realised as the cohomology of two distinct Spanier-Whitehead dual spectra. Similarly, the double and iterated double are also realisable, but then the process stops.

In the chromatic world, the double versions give rise objects whose Morava K-theory at height 2 involve endotrivial modules over the quaternion group of order 8 which lives inside the corresponding Morava stabilizer group. This gives a somewhat surprising connection between endotriviality in two different contexts.

I will explain how all this works and discuss some possible generalisations to higher chromatic heights.

Let G be a finite group and k a field of positive characteristic p dividing the order of G. An endotrivial kG-module is a finitely generated kG-module which is "invertible" in some suitable sense. Since the late 70s, these modules have been intensely studied in modular representation theory. In this talk, we review the essential background on endotrivial modules, and present some results about endotrivial modules for finite groups of Lie type, obtained jointly with Carlson, Grodal and Nakano.

Across the multitude of definitions for a higher category, a dividing line can be found between two major camps of model. On one side lives the ‘algebraic’ models where composition operations between morphisms are given, like Bénabou’s bicategories, tricategories following Gurski and the models of n-category of Batanin and Leinster, Trimble and Penon. On the other end, one finds the ‘non-algebraic’ models, where the space of possible composites is only guaranteed to be contractible. These include the models of Tamsamani and Paoli, along with quasicategories, Segal n-categories, complete n-fold Segal spaces and more. The bridges between these models remain somewhat mysterious. Progress has been made in certain instances, as seen in the work of Tamsamani, Leinster, Lack and Paoli, Cottrell, Campbell, Nikolaus and others. Nonetheless, the correspondence remains incomplete; indeed, for instance, there is no fully verified means in the literature to take an `algebraic’ homotopy n-category of any known model of (oo, n)-category for general n.

In this talk, I will present my contributions to the problem of taking algebraic homotopy bicategories of non-algebraic (oo, 2)-categories. This talk also serves as an introduction to the model of (oo, 2)-category I will be using, namely complete 2-fold Segal spaces. If time permits, I will discuss how to compute the fundamental bigroupoid of a topological space with this construction.

Frobenius algebras appear in many parts of maths and have nice properties.  One can define algebra objects in any monoidal category, and there is a standard definition of when such an algebra object is Frobenius.  But this definition is not satisfied by something which we'd like to think of as an algebra object in Temperley-Lieb categories at roots of unity.  We will explore a more general definition of a Frobenius algebra object which covers this example, and will explore some of its properties.  This is joint work with Mathew Pugh.

For each r, maps which are quasi-isomorphisms on the r page provide a class of weak equivalences on the category of spectral sequences. The talk will cover homotopy theory associated with this setting. We introduce the category of extended spectral sequences and show that this is bicomplete by analysis of a certain presheaf category modelled on discs. We endow the category of extended spectral sequences with various model category structures. One of these has the property that spectral sequences is a homotopically full subcategory and so, by results of Meier, exhibits the category of spectral sequences as a fibrant object in the Barwick-Kan model structure on relative categories. We also note how the presheaf approach provides some insight into the décalage functor on spectral sequences.

This is joint work with Muriel Livernet.

In various parts of mathematics such as algebraic geometry, homotopy theory and representation theory, you can encounter situations where you have a strong monoidal functor f^* with an adjoint f_+.  One automatically gets a comparison map between f_+(a x f^*b) and f_+(a) x b where x is the monoidal product.  The projection formula is said to hold when this comparison map is an isomorphism.  Fausk, Hu and May showed that the projection formula holds under various conditions, such as f^* being a strong closed monoidal functor.  I will show how a theory of mates for parametrized adjunctions (and my graphical version of it) has helped me understand their work.

Homotopy theory allows us to study infinitesimal deformations of algebraic varieties via (partition) Lie algebras. We apply this general principle to two classical problems on Calabi-Yau varieties Z in characteristic p. First, we show that if Z has torsion-free crystalline cohomology and degenerating Hodge-de Rham spectral sequence, then its mixed characteristic deformations are unobstructed. This generalises the BTT theorem to characteristic p. If Z is ordinary, we show that it moreover admits a canonical (and algebraisable) lift to characteristic zero, thereby extending Serre-Tate theory to Calabi-Yau varieties.

This is joint work with Taelman, and generalises results of Achinger-Zdanowicz, Bogomolov-Tian-Todorov, Deligne-Nygaard, Ekedahl–Shepherd-Barron, Schröer, Serre-Tate, and Ward.

The braiding statistics of point particles in 2-dimensional topological phases are given by representations of the braid groups. One approach to the study of generalised particles in topological phases, loop particles in 3-dimensions for example, is to generalise (some of) the several different realisations of the braid group.

In this talk I will construct for each manifold M its motion groupoid $Mot_M$, whose object class is the power set of M. I will discuss several different, but equivalent, quotients on motions leading to the motion groupoid. In particular that the quotient used in the construction $Mot_M$ can be formulated entirely in terms of a level preserving isotopy relation on the trajectories of objects under flows -- worldlines (e.g. monotonic `tangles').

I will also give a construction of a mapping class groupoid $\mathrm{MCG}_M$ associated to a manifold M with the same object class. For each manifold M I will construct a functor $F \colon Mot_M \to MCG_M$, and prove that this is an isomorphism if $\pi_0$ and $\pi_1$ of the appropriate space of self-homeomorphisms of M is trivial. In particular there is an isomorphism in the physically important case $M=[0,1]^n$ with fixed boundary, for any $n\in\mathbb{N}$.

I will discuss several examples throughout.

One can think of the stabilisation of an ∞-category as the ∞-category of objects that admit infinite deloopings. For example, the ∞-category of spectra is the stabilisation of the ∞-category of homotopy types. Costabilisation is the opposite notion of stabilisation, where we are interested in objects that allow infinite desuspensions. It is easy to see that the costabilisation of the ∞-category of homotopy types is trivial. Fix a prime number p. In this talk I will show that the costablisation of the ∞-category of T(h)-local spectral Lie algebras is equivalent to the ∞-category of T(h)-local spectra, where T(h) denotes a p-local telescope spectrum of height h. A key ingredient of the proof is to relate spectral Lie algebras to (spectral) Eₙ algebras via Koszul duality.

Imagine we have a metric space whose points we think of as warehouses, and whose distances give the cost of moving a unit of stock. Now imagine we have two probability distributions that tell us how much stock is in each warehouse. A classical problem from optimal transport theory asks: how we might rearrange one distribution into another with minimal cost? The 'minimal cost' in this scenario defines a metric on the space of all probability measures, this metric is called earth-mover's distance.

Now instead of a metric space imagine we have a category enriched over the extended non-negative reals. As Lawvere points out, these enriched categories can be thought of as generalised metric spaces. We show that from this perspective, probability measures might be thought of as functors and the natural transformation object that exists between them is actually equal to the earth-mover's distance.

What's more, we show that, when we take consider sub-probability measures - that is, measures with total mass less than one - the natural transformation object improves on the earth-mover's distance and can be intuited as the 'minimal cost of meeting demand'.

A usual test for a suitable semi-strict notion of n-category is that in its degenerate cases, it produces particular lower-dimensional monoidal structures as predicted by Baez and Dolan's Stabilisation Hypothesis. These structures are of interest in topology in that they produce algebraic homotopy n-types which are not equivalent to a fully strict notion of n-category. We are concerned with doubly-degenerate tricategories, which should produce a structure equivalent to a braided monoidal category. Gordon, Power, and Street show that in the case of Gray-categories, where interchange of 2-cells is weak but all other composition is strict, this is certainly the case. Joyal and Kock show further that the weakness, like a bump under a carpet, can be pushed solely into the horizontal units for 2-cells, and that this notion also matches braided monoidal categories in the doubly-degenerate case.

In this talk I will introduce a notion of tricategory in which only the vertical composition of 2-cells is weak. These will be identified with categories strictly enriched in the category of bicategories and strict 2-functors with cartesian monoidal product, which, although constituting an unusual mix of weakness and strictness allows a very straightforward algebraic characterisation of weak vertical tricategories using the theory of 2-monads and 2-distributive laws. Thus far only object-level correspondences have been considered, but we show that with special consideration given to icon-like higher cells, we can form a 2-categorical totality of these degenerate structures, along with their weak maps and transformations, allowing us to give a full comparison with the 2-category of braided monoidal categories.

Hermitian K-theory of a commutative ring R is the algebraic K-theory of finitely generated projective R-modules equipped with a non-degenerate symmetric/symplectic/quadratic form. The algebra generated in degree (1,1) modulo the Steinberg relation in degree (2,2) is called Milnor-Witt K-theory and plays an important role in A1-homotopy theory.  Multiplicativity of Hermitian K-theory defines a graded ring homomorphism from Milnor-Witt K-theory to Hermitian K-theory. We prove a homology stability result for symplectic groups and use this to construct a map from Hermitian K-theory of a local ring to Milnor-Witt K-theory in degrees 2,3 mod 4. Finally, we compute the composition of the maps from Milnor-Witt to Hermitian and back to Milnor-Witt K-theory as multiplication with a particular integral form.

By a relative category we mean a category $\mathcal{C}$ equipped with a class $\text{we}$ of weak equivalences.  Given such a thing, one can construct a simplicial set $N\mathcal{C}$, called the relative nerve.  (In the case where $\text{we}$ is just the class of identity morphisms, this is just the usual nerve of $\mathcal{C}$.)  Under mild conditions on $\mathcal{C}$, one can show that $N\mathcal{C}$ is a quasicategory (as defined by Joyal and studied by Lurie), and that the homotopy category of $N\mathcal{C}$ is the category of fractions $\mathcal{C}[\text{we}^{-1}]$.  Lennart Meier gave a proof of this, but it depended on quoting a large body of theory related to model categories in the sense of Quillen.  I will explain a different approach which instead uses more concrete combinatorial constructions with various specific finite posets.

I'll give definitions of some generalizations of rook-Brauer algebras (and their subalgebras) by introducing equivariance and braiding. I'll discuss how we can identify the homology of some of these algebras with the group homology of braid groups and certain semi-direct product groups. I'll also discuss how we can deduce homological stability results and discuss some ideas for future work.