categorification


So it’s been a while since I last posted – the end of 2013 ended up being busy with a couple of visits to Jamie Vicary in Oxford, and Roger Picken in Lisbon. In the aftermath of the two trips, I did manage to get a major revision of this paper submitted to a journal, and put this one out in public. A couple of others will be coming down the pipeline this year as well.

I’m hoping to get back to a post about motives which I planned earlier, but for the moment, I’d like to write a little about the second paper, with Roger Picken.

Global and Local Symmetry

The upshot is that it’s about categorifying the concept of symmetry. More specifically, it’s about finding the analog in the world of categories for the interplay between global and local symmetry which occurs in the world of set-based structures (sets, topological spaces, vector spaces, etc.) This distinction is discussed in a nice way by Alan Weinstein in this article from the Notices of the AMS from

The global symmetry of an object X in some category \mathbf{C} can be described in terms of its group of automorphisms: all the ways the object can be transformed which leave it “the same”. This fits our understanding of “symmetry” when the morphisms can really be interpreted as transformations of some sort. So let’s suppose the object is a set with some structure, and the morphisms are set-maps that preserve the structure: for example, the objects could be sets of vertices and edges of a graph, so that morphisms are maps of the underlying data that preserve incidence relations. So a symmetry of an object is a way of transforming it into itself – and an invertible one at that – and these automorphisms naturally form a group Aut(X). More generally, we can talk about an action of a group G on an object X, which is a map \phi : G \rightarrow Aut(X).

“Local symmetry” is different, and it makes most sense in a context where the object X is a set – or at least, where it makes sense to talk about elements of X, so that X has an underlying set of some sort.

Actually, being a set-with-structure, in a lingo I associate with Jim Dolan, means that the forgetful functor U : \mathbf{C} \rightarrow \mathbf{Sets} is faithful: you can tell morphisms in \mathbf{C} (in particular, automorphisms of X) apart by looking at what they do to the underlying set. The intuition is that the morphisms of \mathbf{C} are exactly set maps which preserve the structure which U forgets about – or, conversely, that the structure on objects of \mathbf{C} is exactly that which is forgotten by U. Certainly, knowing only this information determines \mathbf{C} up to equivalence. In any case, suppose we have an object like this: then knowing about the symmetries of X amounts to knowing about a certain group action, namely the action of Aut(X), on the underlying set U(X).

From this point of view, symmetry is about group actions on sets. The way we represent local symmetry (following Weinstein’s discussion, above) is to encode it as a groupoid – a category whose morphisms are all invertible. There is a level-slip happening here, since X is now no longer seen as an object inside a category: it is the collection of all the objects of a groupoid. What makes this a representation of “local” symmetry is that each morphism now represents, not just a transformation of the whole object X, but a relationship under some specific symmetry between one element of X and another. If there is an isomorphism between x \in X and y \in X, then x and y are “symmetric” points under some transformation. As Weinstein’s article illustrates nicely, though, there is no assumption that the given transformation actually extends to the entire object X: it may be that only part of X has, for example, a reflection symmetry, but the symmetry doesn’t extend globally.

Transformation Groupoid

The “interplay” I alluded to above, between the global and local pictures of symmetry, is to build a “transformation groupoid” (or “action groupoid“) associated to a group G acting on a set X. The result is called X // G for short. Its morphisms consist of pairs such that  (g,x) : x \rightarrow (g \rhd x) is a morphism taking x to its image under the action of g \in G. The “local” symmetry view of X // G treats each of these symmetry relations between points as a distinct bit of data, but coming from a global symmetry – that is, a group action – means that the set of morphisms comes from the product G \times X.

Indeed, the “target” map in X // G from morphisms to objects is exactly a map G \times X \rightarrow X. It is not hard to show that this map is an action in another standard sense. Namely, if we have a real action \phi : G \rightarrow Hom(X,X), then this map is just \hat{\phi} : G \times X \rightarrow X, which moves one of the arguments to the left side. If \phi was a functor, then $\hat{\phi}$ satisfies the “action” condition, namely that the following square commutes:

actionsquare

(Here, m is the multiplication in G, and this is the familiar associativity-type axiom for a group action: acting by a product of two elements in G is the same as acting by each one successively.

So the starting point for the paper with Roger Picken was to categorify this. It’s useful, before doing that, to stop and think for a moment about what makes this possible.

First, as stated, this assumed that X either is a set, or has an underlying set by way of some faithful forgetful functor: that is, every morphism in Aut(X) corresponds to a unique set map from the elements of X to itself. We needed this to describe the groupoid X // G, whose objects are exactly the elements of X. The diagram above suggests a different way to think about this. The action diagram lives in the category \mathbf{Set}: we are thinking of G as a set together with some structure maps. X and the morphism \hat{\phi} must be in the same category, \mathbf{Set}, for this characterization to make sense.

So in fact, what matters is that the category X lived in was closed: that is, it is enriched in itself, so that for any objects X,Y, there is an object Hom(X,Y), the internal hom. In this case, it’s G = Hom(X,X) which appears in the diagram. Such an internal hom is supposed to be a dual to \mathbf{Set}‘s monoidal product (which happens to be the Cartesian product \times): this is exactly what lets us talk about \hat{\phi}.

So really, this construction of a transformation groupoid will work for any closed monoidal category \mathbf{C}, producing a groupoid in \mathbf{C}. It may be easier to understand in cases like \mathbf{C}=\mathbf{Top}, the category of topological spaces, where there is indeed a faithful underlying set functor. But although talking explicitly about elements of X was useful for intuitively seeing how X//G relates global and local symmetries, it played no particular role in the construction.

Categorify Everything

In the circles I run in, a popular hobby is to “categorify everything“: there are different versions, but what we mean here is to turn ideas expressed in the world of sets into ideas in the world of categories. (Technical aside: all the categories here are assumed to be small). In principle, this is harder than just reproducing all of the above in any old closed monoidal category: the “world” of categories is \mathbf{Cat}, which is a closed monoidal 2-category, which is a more complicated notion. This means that doing all the above “strictly” is a special case: all the equalities (like the commutativity of the action square) might in principle be replaced by (natural) isomorphisms, and a good categorification involves picking these to have good properties.

(In our paper, we left this to an appendix, because the strict special case is already interesting, and in any case there are “strictification” results, such as the fact that weak 2-groups are all equivalent to strict 2-groups, which mean that the weak case isn’t as much more general as it looks. For higher n-categories, this will fail – which is why we include the appendix to suggest how the pattern might continue).

Why is this interesting to us? Bumping up the “categorical level” appeals for different reasons, but the ones matter most to me have to do with taking low-dimensional (or -codimensional) structures, and finding analogous ones at higher (co)dimension. In our case, the starting point had to do with looking at the symmetries of “higher gauge theories” – which can be used to describe the transport of higher-dimensional surfaces in a background geometry, the way gauge theories can describe the transport of point particles. But I won’t ask you to understand that example right now, as long as you can accept that “what are the global/local symmetries of a category like?” is a possibly interesting question.

So let’s categorify the discussion about symmetry above… To begin with, we can just take our (closed monoidal) category to be \mathbf{Cat}, and follow the same construction above. So our first ingredient is a 2-group \mathcal{G}. As with groups, we can think of a 2-group either as a 2-category with just one object \star, or as a 1-category with some structure – a group object in \mathbf{Cat}, which we’ll call C(\mathcal{G}) if it comes from a given 2-group. (In our paper, we keep these distinct by using the term “categorical group” for the second. The group axioms amount to saying that we have a monoidal category (\mathcal{G}, \otimes, I). Its objects are the morphisms of the 2-group, and the composition becomes the monoidal product \otimes.)

(In fact, we often use a third equivalent definition, that of crossed modules of groups, but to avoid getting into that machinery here, I’ll be changing our notation a little.)

2-Group Actions

So, again, there are two ways to talk about an action of a 2-group on some category \mathbf{C}. One is to define an action as a 2-functor \Phi : \mathcal{G} \rightarrow \mathbf{Cat}. The object being acted on, \mathbf{C} \in \mathbf{Cat}, is the unique object \Phi(\star) – so that the 2-functor amounts to a monoidal functor from the categorical group C(\mathcal{G}) into Aut(\mathbf{C}). Notice that here we’re taking advantage of the fact that \mathbf{Cat} is closed, so that the hom-”sets” are actually categories, and the automorphisms of \mathbf{C} – invertible functors from \mathbf{C} to itself – form the objects of a monoidal category, and in fact a categorical group. What’s new, though, is that there are also 2-morphisms – natural transformations between these functors.

To begin with, then, we show that there is a map \hat{\Phi} : \mathcal{G} \times \mathbf{C} \rightarrow \mathbf{C}, which corresponds to the 2-functor \Phi, and satisfies an action axiom like the square above, with \otimes playing the role of group multiplication. (Again, remember that we’re only talking about the version where this square commutes strictly here – in an appendix of the paper, we talk about the weak version of all this.) This is an intuitive generalization of the situation for groups, but it is slightly more complicated.

The action \Phi directly gives three maps. First, functors \Phi(\gamma) : \mathbf{C} \rightarrow \mathbf{C} for each 2-group morphism \gamma – each of which consists of a function between objects of \mathbf{C}, together with a function between morphisms of \mathbf{C}. Second, natural transformations \Phi(\eta) : \Phi(\gamma) \rightarrow \Phi(\gamma ') for 2-morphisms \eta : \gamma \rightarrow \gamma' in the 2-group – each of which consists of a function from objects to morphisms of \mathbf{C}.

On the other hand, \hat{\Phi} : \mathcal{G} \times \mathbf{C} \rightarrow \mathbf{C} is just a functor: it gives two maps, one taking pairs of objects to objects, the other doing the same for morphisms. Clearly, the map (\gamma,x) \mapsto x' is just given by x' = \Phi(\gamma)(x). The map taking pairs of morphisms (\eta,f) : (\gamma,x) \rightarrow (\gamma ', y) to morphisms of \mathbf{C} is less intuitively obvious. Since I already claimed \Phi and \hat{\Phi} are equivalent, it should be no surprise that we ought to be able to reconstruct the other two parts of \Phi from it as special cases. These are morphism-maps for the functors, (which give \Phi(\gamma)(f) or \Phi(\gamma ')(f)), and the natural transformation maps (which give \Phi(\eta)(x) or \Phi(\eta)(y)). In fact, there are only two sensible ways to combine these four bits of information, and the fact that \Phi(\eta) is natural means precisely that they’re the same, so:

\hat{\Phi}(\eta,f) = \Phi(\eta)(y) \circ \Phi(\gamma)(f) = \Phi(\gamma ')(f) \circ \Phi(\eta)(x)

Given the above, though, it’s not so hard to see that a 2-group action really involves two group actions: of the objects of \mathcal{G} on the objects of \mathbf{C}, and of the morphisms of \mathcal{G} on objects of \mathbf{C}. They fit together nicely because objects can be identified with their identity morphisms: furthermore, \Phi being a functor gives an action of \mathcal{G}-objects on \mathbf{C}-morphisms which fits in between them nicely.

But what of the transformation groupoid? What is the analog of the transformation groupoid, if we repeat its construction in \mathbf{Cat}?

The Transformation Double Category of a 2-Group Action

The answer is that a category (such as a groupoid) internal to \mathbf{Cat} is a double category. The compact way to describe it is as a “category in \mathbf{Cat}“, with a category of objects and a category of morphisms, each of which of course has objects and morphisms of its own. For the transformation double category, following the same construction as for sets, the object-category is just \mathbf{C}, and the morphism-category is \mathcal{G} \times \mathbf{C}, and the target functor is just the action map \hat{\Phi}. (The other structure maps that make this into a category in \mathbf{Cat} can similarly be worked out by following your nose).

This is fine, but the internal description tends to obscure an underlying symmetry in the idea of double categories, in which morphisms in the object-category and objects in the morphism-category can switch roles, and get a different description of “the same” double category, denoted the “transpose”.

A different approach considers these as two different types of morphism, “horizontal” and “vertical”: they are the morphisms of horizontal and vertical categories, built on the same set of objects (the objects of the object-category). The morphisms of the morphism-category are then called “squares”. This makes a convenient way to draw diagrams in the double category. Here’s a version of a diagram from our paper with the notation I’ve used here, showing what a square corresponding to a morphism (\chi,f) \in \mathcal{G} \times \mathbf{C} looks like:

squarepic

The square (with the boxed label) has the dashed arrows at the top and bottom for its source and target horizontal morphisms (its images under the source and target functors: the argument above about naturality means they’re well-defined). The vertical arrows connecting them are the source and target vertical morphisms (its images under the source and target maps in the morphism-category).

Horizontal and Vertical Slices of \mathbf{C} // \mathcal{G}

So by construction, the horizontal category of these squares is just the object-category \mathbf{C}.  For the same reason, the squares and vertical morphisms, make up the category \mathcal{G} \times \mathbf{C}.

On the other hand, the vertical category has the same objects as \mathbf{C}, but different morphisms: it’s not hard to see that the vertical category is just the transformation groupoid for the action of the group of \mathbf{G}-objects on the set of \mathbf{C}-objects, Ob(\mathbf{C}) // Ob(\mathcal{G}). Meanwhile, the horizontal morphisms and squares make up the transformation groupoid Mor(\mathbf{C}) // Mor(\mathcal{G}). These are the object-category and morphism-category of the transpose of the double-category we started with.

We can take this further: if squares aren’t hip enough for you – or if you’re someone who’s happy with 2-categories but finds double categories unfamiliar – the horizontal and vertical categories can be extended to make horizontal and vertical bicategories. They have the same objects and morphisms, but we add new 2-cells which correspond to squares where the boundaries have identity morphisms in the direction we’re not interested in. These two turn out to feel quite different in style.

First, the horizontal bicategory extends \mathbf{C} by adding 2-morphisms to it, corresponding to morphisms of \mathcal{G}: roughly, it makes the morphisms of \mathbf{C} into the objects of a new transformation groupoid, based on the action of the group of automorphisms of the identity in \mathcal{G} (which ensures the square has identity edges on the sides.) This last point is the only constraint, and it’s not a very strong one since Aut(1_G) and G essentially determine the entire 2-group: the constraint only relates to the structure of \mathcal{G}.

The constraint for the vertical bicategory is different in flavour because it depends more on the action \Phi. Here we are extending a transformation groupoid, Ob(\mathbf{C}) // Ob(\mathcal{G}). But, for some actions, many morphisms in \mathcal{G} might just not show up at all. For 1-morphisms (\gamma, x), the only 2-morphisms which can appear are those taking \gamma to some \gamma ' which has the same effect on x as \gamma. So, for example, this will look very different if \Phi is free (so only automorphisms show up), or a trivial action (so that all morphisms appear).

In the paper, we look at these in the special case of an adjoint action of a 2-group, so you can look there if you’d like a more concrete example of this difference.

Speculative Remarks

The starting point for this was a project (which I talked about a year ago) to do with higher gauge theory – see the last part of the linked post for more detail. The point is that, in gauge theory, one deals with connections on bundles, and morphisms between them called gauge transformations. If one builds a groupoid out of these in a natural way, it turns out to result from the action of a big symmetry group of all gauge transformations on the moduli space of connections.

In higher gauge theory, one deals with connections on gerbes (or higher gerbes – a bundle is essentially a “0-gerbe”). There are now also (2-)morphisms between gauge transformations (and, in higher cases, this continues further), which Roger Picken and I have been calling “gauge modifications”. If we try to repeat the situation for gauge theory, we can construct a 2-groupoid out of these, which expresses this local symmetry. The thing which is different for gerbes (and will continue to get even more different if we move to n-gerbes and the corresponding (n+1)-groupoids) is that this is not the same type of object as a transformation double category.

Now, in our next paper (which this one was written to make possible) we show that the 2-groupoid is actually very intimately related to the transformation double category: that is, the local picture of symmetry for a higher gauge theory is, just as in the lower-dimensional situation, intimately related to a global symmetry of an entire moduli 2-space, i.e. a category. The reason this wasn’t obvious at first is that the moduli space which includes only connections is just the space of objects of this category: the point is that there are really two special kinds of gauge transformations. One should be thought of as the morphisms in the moduli 2-space, and the other as part of the symmetries of that 2-space. The intuition that comes from ordinary gauge theory overlooks this, because the phenomenon doesn’t occur there.

Physically-motivated theories are starting to use these higher-categorical concepts more and more, and symmetry is a crucial idea in physics. What I’ve sketched here is presumably only the start of a pattern in which “symmetry” extends to higher-categorical entities. When we get to 3-groups, our simplifying assumptions that use “strictification” results won’t even be available any more, so we would expect still further new phenomena to show up – but it seems plausible that the tight relation between global and local symmetry will still exist, but in a way that is more subtle, and refines the standard understanding we have of symmetry today.

Since the last post, I’ve been busily attending some conferences, as well as moving to my new job at the University of Hamburg, in the Graduiertenkolleg 1670, “Mathematics Inspired by String Theory and Quantum Field Theory”.  The week before I started, I was already here in Hamburg, at the conference they were organizing “New Perspectives in Topological Quantum Field Theory“.  But since I last posted, I was also at the 20th Oporto Meeting on Geometry, Topology, and Physics, as well as the third Higher Structures in China workshop, at Jilin University in Changchun.  Right now, I’d like to say a few things about some of the highlights of that workshop.

Higher Structures in China III

So last year I had a bunch of discussions I had with Chenchang Zhu and Weiwei Pan, who at the time were both in Göttingen, about my work with Jamie Vicary, which I wrote about last time when the paper was posted to the arXiv.  In that, we showed how the Baez-Dolan groupoidification of the Heisenberg algebra can be seen as a representation of Khovanov’s categorification.  Chenchang and Weiwei and I had been talking about how these ideas might extend to other examples, in particular to give nice groupoidifications of categorified Lie algebras and quantum groups.

That is still under development, but I was invited to give a couple of talks on the subject at the workshop.  It was a long trip: from Lisbon, the farthest-west of the main cities of (continental) Eurasia all the way to one of the furthest-East.   (Not quite the furthest, but Changchun is in the northeast of China, just a few hours north of Korea, and it took just about exactly 24 hours including stopovers to get there).  It was a long way to go for a three day workshop, but as there were also three days of a big excursion to Changbai Mountain, just on the border with North Korea, for hiking and general touring around.  So that was a sort of holiday, with 11 other mathematicians.  Here is me with Dany Majard, in a national park along the way to the mountains:

Here’s me with Alex Hoffnung, on Changbai Mountain (in the background is China):

And finally, here’s me a little to the left of the previous picture, where you can see into the volcanic crater.  The lake at the bottom is cut out of the picture, but you can see the crater rim, of which this particular part is in North Korea, as seen from China:

Well, that was fun!

Anyway, the format of the workshop involved some talks from foreigners and some from locals, with a fairly big local audience including a good many graduate students from Jilin University.  So they got a chance to see some new work being done elsewhere – mostly in categorification of one kind or another.  We got a chance to see a little of what’s being done in China, although not as much as we might have. I gather that not much is being done yet that fit the theme of the workshop, which was part of the reason to organize the workshop, and especially for having a session aimed specially at the graduate students.

Categorified Algebra

This is a sort of broad term, but certainly would include my own talk.  The essential point is to show how the groupoidification of the Heisenberg algebra is a representation of Khovanov’s categorification of the same algebra, in a particular 2-category.  The emphasis here is on the fact that it’s a representation in a 2-category whose objects are groupoids, but whose morphisms aren’t just functors, but spans of functors – that is, composites of functors and co-functors.  This is a pretty conservative weakening of “representations on categories” – but it lets one build really simple combinatorial examples.  I’ve discussed this general subject in recent posts, so I won’t elaborate too much.  The lecture notes are here, if you like, though – they have more detail than my previous post, but are less technical than the paper with Jamie Vicary.

Aaron Lauda gave a nice introduction to the program of categorifying quantum groups, mainly through the example of the special case U_q(sl_2), somewhat along the same lines as in his introductory paper on the subject.  The story which gives the motivation is nice: one has knot invariants such as the Jones polynomial, based on representations of groups and quantum groups.  The Jones polynomial can be categorified to give Khovanov homology (which assigns a complex to a knot, whose graded Euler characteristic is the Jones polynomial) – but also assigns maps of complexes to cobordisms of knots.  One then wants to categorify the representation theory behind it – to describe actions of, for instance, quantum sl_2 on categories.  This starting point is nice, because it can work by just mimicking the construction of sl_2 and U_q(sl_2) representations in terms of weight spaces: one gets categories V_{-N}, \dots, V_N which correspond to the “weight spaces” (usually just vector spaces), and the E and F operators give functors between them, and so forth.

Finding examples of categories and functors with this structure, and satisfying the right relations, gives “categorified representations” of the algebra – the monoidal categories of diagrams which are the “categorifications of the algebra” then are seen as the abstraction of exactly which relations these are supposed to satisfy.  One such example involves flag varieties.  A flag, as one might eventually guess from the name, is a nested collection of subspaces in some n-dimensional space.  A simple example is the Grassmannian Gr(1,V), which is the space of all 1-dimensional subspaces of V (i.e. the projective space P(V)), which is of course an algebraic variety.  Likewise, Gr(k,V), the space of all k-dimensional subspaces of V is a variety.  The flag variety Fl(k,k+1,V) consists of all pairs W_k \subset W_{k+1}, of a k-dimensional subspace of V, inside a (k+1)-dimensional subspace (the case k=2 calls to mind the reason for the name: a plane intersecting a given line resembles a flag stuck to a flagpole).  This collection is again a variety.  One can go all the way up to the variety of “complete flags”, Fl(1,2,\dots,n,V) (where V is n-dimenisonal), any point of which picks out a subspace of each dimension, each inside the next.

The way this relates to representations is by way of geometric representation theory. One can see those flag varieties of the form Fl(k,k+1,V) as relating the Grassmanians: there are projections Fl(k,k+1,V) \rightarrow Gr(k,V) and Fl(k,k+1,V) \rightarrow Gr(k+1,V), which act by just ignoring one or the other of the two subspaces of a flag.  This pair of maps, by way of pulling-back and pushing-forward functions, gives maps between the cohomology rings of these spaces.  So one gets a sequence H_0, H_1, \dots, H_n, and maps between the adjacent ones.  This becomes a representation of the Lie algebra.  Categorifying this, one replaces the cohomology rings with derived categories of sheaves on the flag varieties – then the same sort of “pull-push” operation through (derived categories of sheaves on) the flag varieties defines functors between those categories.  So one gets a categorified representation.

Heather Russell‘s talk, based on this paper with Aaron Lauda, built on the idea that categorified algebras were motivated by Khovanov homology.  The point is that there are really two different kinds of Khovanov homology – the usual kind, and an Odd Khovanov Homology, which is mainly different in that the role played in Khovanov homology by a symmetric algebra is instead played by an exterior (antisymmetric) algebra.  The two look the same over a field of characteristic 2, but otherwise different.  The idea is then that there should be “odd” versions of various structures that show up in the categorifications of U_q(sl_2) (and other algebras) mentioned above.

One example is the fact that, in the “even” form of those categorifications, there is a natural action of the Nil Hecke algebra on composites of the generators.  This is an algebra which can be seen to act on the space of polynomials in n commuting variables, \mathbb{C}[x_1,\dots,x_n], generated by the multiplication operators x_i, and the “divided difference operators” based on the swapping of two adjacent variables.  The Hecke algebra is defined in terms of “swap” generators, which satisfy some q-deformed variation of the relations that define the symmetric group (and hence its group algebra).   The Nil Hecke algebra is so called since the “swap” (i.e. the divided difference) is nilpotent: the square of the swap is zero.  The way this acts on the objects of the diagrammatic category is reflected by morphisms drawn as crossings of strands, which are then formally forced to satisfy the relations of the Nil Hecke algebra.

The ODD Nil Hecke algebra, on the other hand, is an analogue of this, but the x_i are anti-commuting, and one has different relations satisfied by the generators (they differ by a sign, because of the anti-commutation).  This sort of “oddification” is then supposed to happen all over.  The main point of the talk was to to describe the “odd” version of the categorified representation defined using flag varieties.  Then the odd Nil Hecke algebra acts on that, analogously to the even case above.

Marco Mackaay gave a couple of talks about the sl_3 web algebra, describing the results of this paper with Weiwei Pan and Daniel Tubbenhauer.  This is the analog of the above, for U_q(sl_3), describing a diagram calculus which accounts for representations of the quantum group.  The “web algebra” was introduced by Greg Kuperberg – it’s an algebra built from diagrams which can now include some trivalent vertices, along with rules imposing relations on these.  When categorifying, one gets a calculus of “foams” between such diagrams.  Since this is obviously fairly diagram-heavy, I won’t try here to reproduce what’s in the paper – but an important part of is the correspondence between webs and Young Tableaux, since these are labels in the representation theory of the quantum group – so there is some interesting combinatorics here as well.

Algebraic Structures

Some of the talks were about structures in algebra in a more conventional sense.

Jiang-Hua Lu: On a class of iterated Poisson polynomial algebras.  The starting point of this talk was to look at Poisson brackets on certain spaces and see that they can be found in terms of “semiclassical limits” of some associative product.  That is, the associative product of two elements gives a power series in some parameter h (which one should think of as something like Planck’s constant in a quantum setting).  The “classical” limit is the constant term of the power series, and the “semiclassical” limit is the first-order term.  This gives a Poisson bracket (or rather, the commutator of the associative product does).  In the examples, the spaces where these things are defined are all spaces of polynomials (which makes a lot of explicit computer-driven calculations more convenient). The talk gives a way of constructing a big class of Poisson brackets (having some nice properties: they are “iterated Poisson brackets”) coming from quantum groups as semiclassical limits.  The construction uses words in the generating reflections for the Weyl group of a Lie group G.

Li Guo: Successors and Duplicators of Operads – first described a whole range of different algebra-like structures which have come up in various settings, from physics and dynamical systems, through quantum field theory, to Hopf algebras, combinatorics, and so on.  Each of them is some sort of set (or vector space, etc.) with some number of operations satisfying some conditions – in some cases, lots of operations, and even more conditions.  In the slides you can find several examples – pre-Lie and post-Lie algebras, dendriform algebras, quadri- and octo-algebras, etc. etc.  Taken as a big pile of definitions of complicated structures, this seems like a terrible mess.  The point of the talk is to point out that it’s less messy than it appears: first, each definition of an algebra-like structure comes from an operad, which is a formal way of summing up a collection of operations with various “arities” (number of inputs), and relations that have to hold.  The second point is that there are some operations, “successor” and “duplicator”, which take one operad and give another, and that many of these complicated structures can be generated from simple structures by just these two operations.  The “successor” operation for an operad introduces a new product related to old ones – for example, the way one can get a Lie bracket from an associative product by taking the commutator.  The “duplicator” operation takes existing products and introduces two new products, whose sum is the previous one, and which satisfy various nice relations.  Combining these two operations in various ways to various starting points yields up a plethora of apparently complicated structures.

Dany Majard gave a talk about algebraic structures which are related to double groupoids, namely double categories where all the morphisms are invertible.  The first part just defined double categories: graphically, one has horizontal and vertical 1-morphisms, and square 2-morphsims, which compose in both directions.  Then there are several special degenerate cases, in the same way that categories have as degenerate cases (a) sets, seen as categories with only identity morphisms, and (b) monoids, seen as one-object categories.  Double categories have ordinary categories (and hence monoids and sets) as degenerate cases.  Other degenerate cases are 2-categories (horizontal and vertical morphisms are the same thing), and therefore their own special cases, monoidal categories and symmetric monoids.  There is also the special degenerate case of a double monoid (and the extra-special case of a double group).  (The slides have nice pictures showing how they’re all degenerate cases).  Dany then talked about some structure of double group(oids) – and gave a list of properties for double groupoids, (such as being “slim” – having at most one 2-cell per boundary configuration – as well as two others) which ensure that they’re equivalent to the semidirect product of an abelian group with the “bicrossed product”  H \bowtie K of two groups H and K (each of which has to act on the other for this to make sense).  He gave the example of the Poincare double group, which breaks down as a triple bicrossed product by the Iwasawa decomposition:

Poinc = (SO(3) \bowtie (SO(1; 1) \bowtie N)) \ltimes \mathbb{R}_4

(N is certain group of matrices).  So there’s a unique double group which corresponds to it – it has squares labelled by \mathbb{R}_4, and the horizontial and vertical morphisms by elements of SO(3) and N respectively.  Dany finished by explaining that there are higher-dimensional analogs of all this – n-tuple categories can be defined recursively by internalization (“internal categories in (n-1)-tuple-Cat”).  There are somewhat more sophisticated versions of the same kind of structure, and finally leading up to a special class of n-tuple groups.  The analogous theorem says that a special class of them is just the same as the semidirect product of an abelian group with an n-fold iterated bicrossed product of groups.

Also in this category, Alex Hoffnung talked about deformation of formal group laws (based on this paper with various collaborators).  FGL’s are are structures with an algebraic operation which satisfies axioms similar to a group, but which can be expressed in terms of power series.  (So, in particular they have an underlying ring, for this to make sense).  In particular, the talk was about formal group algebras – essentially, parametrized deformations of group algebras – and in particular for Hecke Algebras.  Unfortunately, my notes on this talk are mangled, so I’ll just refer to the paper.

Physics

I’m using the subject-header “physics” to refer to those talks which are most directly inspired by physical ideas, though in fact the talks themselves were mathematical in nature.

Fei Han gave a series of overview talks intorducing “Equivariant Cohomology via Gauged Supersymmetric Field Theory”, explaining the Stolz-Teichner program.  There is more, using tools from differential geometry and cohomology to dig into these theories, but for now a summary will do.  Essentially, the point is that one can look at “fields” as sections of various bundles on manifolds, and these fields are related to cohomology theories.  For instance, the usual cohomology of a space X is a quotient of the space of closed forms (so the k^{th} cohomology, H^{k}(X) = \Omega^{k}, is a quotient of the space of closed k-forms – the quotient being that forms differing by a coboundary are considered the same).  There’s a similar construction for the K-theory K(X), which can be modelled as a quotient of the space of vector bundles over X.  Fei Han mentioned topological modular forms, modelled by a quotient of the space of “Fredholm bundles” – bundles of Banach spaces with a Fredholm operator around.

The first two of these examples are known to be related to certain supersymmetric topological quantum field theories.  Now, a TFT is a functor into some kind of vector spaces from a category of (n-1)-dimensional manifolds and n-dimensional cobordisms

Z : d-Bord \rightarrow Vect

Intuitively, it gives a vector space of possible fields on the given space and a linear map on a given spacetime.  A supersymmetric field theory is likewise a functor, but one changes the category of “spacetimes” to have both bosonic and fermionic dimension.  A normal smooth manifold is a ringed space (M,\mathcal{O}), since it comes equipped with a sheaf of rings (each open set has an associated ring of smooth functions, and these glue together nicely).  Supersymmetric theories work with manifolds which change this sheaf – so a d|\delta-dimensional space has the sheaf of rings where one introduces some new antisymmetric coordinate functions \theta_i, the “fermionic dimensions”:

\mathcal{O}(U) = C^{\infty}(U) \otimes \bigwedge^{\ast}[\theta_1,\dots,\theta_{\delta}]

Then a supersymmetric TFT is a functor:

E : (d|\delta)-Bord \rightarrow STV

(where STV is the category of supersymmetric topological vector spaces – defined similarly).  The connection to cohomology theories is that the classes of such field theories, up to a notion of equivalence called “concordance”, are classified by various cohomology theories.  Ordinary cohomology corresponds then to 0|1-dimensional extended TFT (that is, with 0 bosonic and 1 fermionic dimension), and K-theory to a 1|1-dimensional extended TFT.  The Stoltz-Teichner Conjecture is that the third example (topological modular forms) is related in the same way to a 2_1-dimensional extended TFT – so these are the start of a series of cohomology theories related to various-dimension TFT’s.

Last but not least, Chris Rogers spoke about his ideas on “Higher Geometric Quantization”, on which he’s written a number of papers.  This is intended as a sort of categorification of the usual ways of quantizing symplectic manifolds.  I am still trying to catch up on some of the geometry This is rooted in some ideas that have been discussed by Brylinski, for example.  Roughly, the message here is that “categorification” of a space can be thought of as a way of acting on the loop space of a space.  The point is that, if points in a space are objects and paths are morphisms, then a loop space L(X) shifts things by one categorical level: its points are loops in X, and its paths are therefore certain 2-morphisms of X.  In particular, there is a parallel to the fact that a bundle with connection on a loop space can be thought of as a gerbe on the base space.  Intuitively, one can “parallel transport” things along a path in the loop space, which is a surface given by a path of loops in the original space.  The local description of this situation says that a 1-form (which can give transport along a curve, by integration) on the loop space is associated with a 2-form (giving transport along a surface) on the original space.

Then the idea is that geometric quantization of loop spaces is a sort of higher version of quantization of the original space. This “higher” version is associated with a form of higher degree than the symplectic (2-)form used in geometric quantization of X.   The general notion of n-plectic geometry, where the usual symplectic geometry is the case n=1, involves a (n+1)-form analogous to the usual symplectic form.  Now, there’s a lot more to say here than I properly understand, much less can summarize in a couple of paragraphs.  But the main theorem of the talk gives a relation between n-plectic manifolds (i.e. ones endowed with the right kind of form) and Lie n-algebras built from the complex of forms on the manifold.  An important example (a theorem of Chris’ and John Baez) is that one has a natural example of a 2-plectic manifold in any compact simple Lie group G together with a 3-form naturally constructed from its Maurer-Cartan form.

At any rate, this workshop had a great proportion of interesting talks, and overall, including the chance to see a little more of China, was a great experience!

This blog has been on hiatus for a while, as I’ve been doing various other things, including spending some time in Hamburg getting set up for the move there. Another of these things has been working with Jamie Vicary on our project on the groupoidified Quantum Harmonic Oscillator (QHO for short). We’ve now put the first of two papers on the arXiv – this one is a relatively nonrigorous look at how this relates to categorification of the Heisenberg Algebra. Since John Baez is a high-speed blogging machine, he’s already beaten me to an overview of what the paper says, and there’s been some interesting discussion already. So I’ll try to say some different things about what it means, and let you take a look over there, or read the paper, for details.

I’ve given some talks about this project, but as we’ve been writing it up, it’s expanded considerably, including a lot of category-theoretic details which are going to be in the second paper in this series. But the basic point of this current paper is essentially visual and, in my opinion, fairly simple. The groupoidification of the QHO has a nice visual description, since it is all about the combinatorics of finite sets. This was described originally by Baez and Dolan, and in more detail in my very first paper. The other visual part here is the relation to Khovanov’s categorification of the Heisenberg algebra using a graphical calculus. (I wrote about this back when I first became aware of it.)

As a Representation

The scenario here actually has some common features with my last post. First, we have a monoidal category with duals, let’s say C presented in terms of some generators and relations. Then, we find some concrete model of this abstractly-presented monoidal category with duals in a specific setting, namely Span(Gpd).

Calling this “concrete” just refers to the fact that the objects in Span(Gpd) have some particular structure in terms of underlying sets and so on. By a “model” I just mean a functor C \rightarrow Span(Gpd) (“model” and “representation” mean essentially the same thing in this context). In fact, for this to make sense, I think of C as a 2-category with one object. Then a model is just some particular choices: a groupoid to represent the unique object, spans of groupoids to represent the generating morphisms, spans of spans to represent the generating 2-morphisms, all chosen so that the defining relations hold.

In my previous post, C was a category of cobordisms, but in this case, it’s essentially Khovanov’s monoidal category H' whose objects are (oriented) dots and whose morphisms are certain classes of diagrams. The nice fact about the particular model we get is that the reasons these relations hold are easy to see in terms of a the combinatorics of sets. This is why our title describes what we got as “a combinatorial representation” Khovanov’s category H' of diagrams, for which the ring of isomorphism classes of objects is the integral form of the algebra. This uses that Span(Gpd) is not just a monoidal category: it can be a monoidal 2-category. What’s more, the monoidal category H' “is” also a 2-category – with one object. The objects of H' are really the morphisms of this 2-category.

So H' is in some sense a universal theory (because it’s defined freely in terms of generators and relations) of what a categorification of the Heisenberg algebra must look like. Baez-Dolan groupoidification of the QHO then turns out to be a representation or model of it. In fact, the model is faithful, so that we can even say that it provides a combinatorial interpretation of that category.

The Combinatorial Model

Between the links above, you can find a good summary of the situation, so I’ll be a bit cursory. The model is described in terms of structures on finite sets. This is why our title calls this a “combinatorial representation” of Khovanov’s categorification.

This means that the one object of H (as a 2-category) is taken to the groupoid FinSet_0 of finite sets and bijections (which we just called S in the paper for brevity). This is the “Fock space” object. For simplicity, we can take an equivalent groupoid, which has just one n-element set for each n.

Now, a groupoid represents a system, whose possible configurations are the objects and whose symmetries are the morphisms. In this case, the possible configurations are the different numbers of “quanta”, and the symmetries (all set-bijections) show that all the quanta are interchangeable. I imagine a box containing some number of ping-pong balls.

A span of groupoids represents a process. It has a groupoid whose objects are histories (and morphisms are symmetries of histories). This groupoid has a pair of maps: to the system the process starts in, and to the system it ends in. In our model, the most important processes (which generate everything else) are the creation and annihilation operators, a^{\dagger} and a – and their categorified equivalents, A and A^{\dagger}. The spans that represent them are very simple: they are processes which put a new ball into the box, or take one out, respectively. (Algebraically, they’re just a way to organize all the inclusions of symmetric groups S_n \subset S_{n+1}.)

The “canonical commutation relation“, which we write without subtraction thus:

A A^{\dagger} = A^{\dagger} A + 1

is already understood in the Baez-Dolan story: it says that there is one more way to remove a ball from a box after putting a new one into it (one more history for the process A A^{\dagger}) than to remove a ball and then add a new one (histories for a^{\dagger} a). This is fairly obvious: in the first instance, you have one more to choose from when removing the ball.

But the original Baez-Dolan story has no interesting 2-morphisms (the actual diagrams which are the 1-morphisms in H), whereas these are absolutely the whole point of a categorification in the sense Khovanov gets one, since the 1-morphisms of H' determine what the isomorphism classes of objects even are.

So this means that we need to figure out what the 2-morphisms in Span(Gpd) need to be – first in general, and second in our particular representation of H.

In general, a 2-morphism in Span(Gpd) is a span of span-maps. You’ll find other people who take it to be a span-map. This would be a functor between the groupoids of histories: roughly, a map which assigns a history in the source span to a history in the target span (and likewise for symmetries), in a way that respects how they’re histories. But we don’t want just a map: we want a process which has histories of its own. We want to describe a “movie of processes” which change one process into another. These can have many histories of their own.

In fact, they’re not too complicated. Here’s one of Khovanov’s relation in H' which forms part of how the commutation relation is expressed (shuffled to get rid of negatives, which we constantly need to do in the combinatorial model since we have no negative sets):

We read an upward arrow as “add a ball to the box”, and a downward arrow as “remove a ball”, and read right-to-left.  Both processes begin and end with“add then remove”. The right-hand side just leaves this process alone: it’s the identity.

The left-hand side shows a process-movie whose histories have two different cases. Suppose we begin with a history for which we add x and then remove y. The first case is that x = y: we remove the same ball we put in. This amounts to doing nothing, so the first part of the movie eliminates all the adding and removing. The second part puts the add-remove pair back in.

The second case ensures that x \neq y, since it takes the initial history to the history (of a different process!) in which we remove y and then add x (impossible if y = x, since we can’t remove this ball before adding it). This in turn is taken to the history (of the original process!) where we add x and then remove y; so this relates every history to itself, except for the case that x = y. Overall the sum of these relations give the identity on histories, which is the right hand side.

This picture includes several of the new 2-morphisms that we need to add to the Baez-Dolan picture: swapping the order of two generators, and adding or removing a pair of add/remove operations. Finding spans of spans which accomplish this (and showing they satisfy the right relations) is all that’s needed to finish up the combinatorial model.  So, for instance, the span of spans which adds a “remove-then-add” pair is this one:

If this isn’t clear, well, it’s explained in more detail in the paper.  (Do notice, though, that this is a diagram in groupoids: we need to specify that there are identity 2-cells in the span, rather than some other 2-cells.)

So this is basically how the combinatorial model works.

Adjointness

But in fact this description is (as often happens) chronologically backwards: what actually happened was that we had worked out what the 2-morphisms should be for different reasons. While trying to to understand what kind of structure this produced, we realized (thanks to Marco Mackaay) that the result was related to H, which in turn shed more light on the 2-morphisms we’d found.

So far so good. But what makes it possible to represent the kind of monoidal category we’re talking about in this setting is adjointness. This is another way of saying what I meant up at the top by saying we start with a monoidal category with duals.  This means morphisms each have a partner – a dual, or adjoint – going in the opposite direction.  The representations of the raising and lowering operators of the Heisenberg algebra on the Hilbert space for the QHO are linear adjoints. Their categorifications also need to be adjoints in the sense of adjoint 1-morphisms in a 2-category.

This is an abstraction of what it means for two functors F and G to be adjoint. In particular, it means there have to be certain 2-cells such as the unit \eta : Id \Rightarrow G \circ F and counit \epsilon : F \circ G \Rightarrow Id satisfying some nice relations. In fact, this only makes F a left adjoint and G a right adjoint – in this situation, we also have another pair which makes F a right adjoint and G a left one. That is, they should be “ambidextrous adjoints”, or “ambiadjoints” for short. This is crucial if they’re going to represent any graphical calculus of the kind that’s involved here (see the first part of this paper by Aaron Lauda, for instance).

So one of the theorems in the longer paper will show concretely that any 1-morphism in Span(Gpd) has an ambiadjoint – which happens to look like the same span, but thought of as going in the reverse direction. This is somewhat like how the adjoint of a real linear map, expressed as a matrix relative to well-chosen bases, is just the transpose of the same matrix. In particular, A and A^{\dagger} are adjoints in just this way. The span-of-span-maps I showed above is exactly the unit for one side of this ambi-adjunction – but it is just a special case of something that will work for any span and its adjoint.

Finally, there’s something a little funny here. Since the morphisms of Span(Gpd) aren’t functors or maps, this combinatorial model is not exactly what people often mean by a “categorified representation”. That would be an action on a category in terms of functors and natural transformations. We do talk about how to get one of these on a 2-vector space out of our groupoidal representation toward the end.

In particular, this amounts to a functor into 2Vect – the objects of 2Vect being categories of a particular kind, and the morphisms being functors that preserve all the structure of those categories. As it turns out, the thing about this setting which is good for this purpose is that all those functors have ambiadjoints. The “2-linearization” that takes Span(Gpd) into 2Vect is a 2-functor, and this means that all the 2-cells and equations that make two morphisms ambiadjoints carry over. In 2Vect, it’s very easy for this to happen, since all those ambiadjoints are already present. So getting representations of categorified algebras that are made using these monoidal categories of diagrams on 2-vector spaces is fairly natural – and it agrees with the usual intuition about what “representation” means.

Anything I start to say about this is in danger of ballooning, but since we’re already some 40 pages into the second paper, I’ll save the elaboration for that…

So I’ve been travelling a lot in the last month, spending more than half of it outside Portugal. I was in Ottawa, Canada for a Fields Institute workshop, “Categorical Methods in Representation Theory“. Then a little later I was in Erlangen, Germany for one called “Categorical and Representation-Theoretic Methods in Quantum Geometry and CFT“. Despite the similar-sounding titles, these were on fairly different themes, though Marco Mackaay was at both, talking about categorifying the q-Schur algebra by diagrams.  I’ll describe the meetings, but for now I’ll start with the first.  Next post will be a summary of the second.

The Ottawa meeting was organized by Alistair Savage, and Alex Hoffnung (like me, a former student of John Baez). Alistair gave a talk here at IST over the summer about a q-deformation of Khovanov’s categorification of the Heisenberg Algebra I discussed in an earlier entry. A lot of the discussion at the workshop was based on the Khovanov-Lauda program, which began with categorifying quantum version of the classical Lie groups, and is now making lots of progress in the categorification of algebras, representation theory, and so on.

The point of this program is to describe “categorifications” of particular algebras. This means finding monoidal categories with the property that when you take the Grothendieck ring (the ring of isomorphism classes, with a multiplication given by the monoidal structure), you get back the integral form of some algebra. (And then recover the original by taking the tensor over \mathbb{Z} with \mathbb{C}). The key thing is how to represent the algebra by generators and relations. Since free monoidal categories with various sorts of structures can be presented as categories of string diagrams, it shouldn’t be surprising that the categories used tend to have objects that are sequences (i.e. monoidal products) of dots with various sorts of labelling data, and morphisms which are string diagrams that carry those labels on strands (actually, usually they’re linear combinations of such diagrams, so everything is enriched in vector spaces). Then one imposes relations on the “free” data given this way, by saying that the diagrams are considered the same morphism if they agree up to some local moves. The whole problem then is to find the right generators (labelling data) and relations (local moves). The result will be a categorification of a given presentation of the algebra you want.

So for instance, I was interested in Sabin Cautis and Anthony Licata‘s talks connected with this paper, “Heisenberg Categorification And Hilbert Schemes”. This is connected with a generalization of Khovanov’s categorification linked above, to include a variety of other algebras which are given a similar name. The point is that there’s such a “Heisenberg algebra” associated to different subgroups \Gamma \subset SL(2,\mathbf{k}), which in turn are classified by Dynkin diagrams. The vertices of these Dynkin diagrams correspond to some generators of the Heisenberg algebra, and one can modify Khovanov’s categorification by having strands in the diagram calculus be labelled by these vertices. Rules for local moves involving strands with different labels will be governed by the edges of the Dynkin diagram. Their paper goes on to describe how to represent these categorifications on certain categories of Hilbert schemes.

Along the same lines, Aaron Lauda gave a talk on the categorification of the NilHecke algebra. This is defined as a subalgebra of endomorphisms of P_a = \mathbb{Z}[x_1,\dots,x_a], generated by multiplications (by the x_i) and the divided difference operators \partial_i. There are different from the usual derivative operators: in place of the differences between values of a single variable, they measure how a function behaves under the operation s_i which switches variables x_i and x_{i+1} (that is, the reflection in the hyperplane where x_i = x_{i+1}). The point is that just like differentiation, this operator – together with multiplication – generates an algebra in End(\mathbb{Z}[x_1,\dots,x_a]. Aaron described how to categorify this presentation of the NilHecke algebra with a string-diagram calculus.

So anyway, there were a number of talks about the explosion of work within this general program – for instance, Marco Mackaay’s which I mentioned, as well as that of Pedro Vaz about the same project. One aspect of this program is that the relatively free “string diagram categories” are sometimes replaced with categories where the objects are bimodules and morphisms are bimodule homomorphisms. Making the relationship precise is then a matter of proving these satisfy exactly the relations on a “free” category which one wants, but sometimes they’re a good setting to prove one has a nice categorification. Thus, Ben Elias and Geordie Williamson gave two parts of one talk about “Soergel Bimodules and Kazhdan-Lusztig Theory” (see a blog post by Ben Webster which gives a brief intro to this notion, including pointing out that Soergel bimodules give a categorification of the Hecke algebra).

One of the reasons for doing this sort of thing is that one gets invariants for manifolds from algebras – in particular, things like the Jones polynomial, which is related to the Temperley-Lieb algebra. A categorification of it is Khovanov homology (which gives, for a manifold, a complex, with the property that the graded Euler characteristic of the complex is the Jones polynomial). The point here is that categorifying the algebra lets you raise the dimension of the kind of manifold your invariants are defined on.

So, for instance, Scott Morrison described “Invariants of 4-Manifolds from Khonanov Homology“.  This was based on a generalization of the relationship between TQFT’s and planar algebras.  The point is, planar algebras are described by the composition of diagrams of the following form: a big circle, containing some number of small circles.  The boundaries of each circle are labelled by some number of marked points, and the space between carries curves which connect these marked points in some way.  One composes these diagrams by gluing big circles into smaller circles (there’s some further discussion here including a picture, and much more in this book here).  Scott Morrison described these diagrams as “spaghetti and meatball” diagrams.  Planar algebras show up by associating a vector spaces to “the” circle with n marked points, and linear maps to each way (up to isotopy) of filling in edges between such circles.  One can think of the circles and marked-disks as a marked-cobordism category, and so a functorial way of making these assignments is something like a TQFT.  It also gives lots of vector spaces and lots of linear maps that fit together in a particular way described by this category of marked cobordisms, which is what a “planar algebra” actually consists of.  Clearly, these planar algebras can be used to get some manifold invariants – namely the “TQFT” that corresponds to them.

Scott Morrison’s talk described how to get invariants of 4-dimensional manifolds in a similar way by boosting (almost) everything in this story by 2 dimensions.  You start with a 4-ball, whose boundary is a 3-sphere, and excise some number of 4-balls (with 3-sphere boundaries) from the interior.  Then let these 3D boundaries be “marked” with 1-D embedded links (think “knots” if you like).  These 3-spheres with embedded links are the objects in a category.  The morphisms are 4-balls which connect them, containing 2D knotted surfaces which happen to intersect the boundaries exactly at their embedded links.  By analogy with the image of “spaghetti and meatballs”, where the spaghetti is a collection of 1D marked curves, Morrison calls these 4-manifolds with embedded 2D surfaces “lasagna diagrams” (which generalizes to the less evocative case of “(n,k) pasta diagrams”, where we’ve just mentioned the (2,1) and (4,2) cases, with k-dimensional “pasta” embedded in n-dimensional balls).  Then the point is that one can compose these pasta diagrams by gluing the 4-balls along these marked boundaries.  One then gets manifold invariants from these sorts of diagrams by using Khovanov homology, which assigns to

Ben Webster talked about categorification of Lie algebra representations, in a talk called “Categorification, Lie Algebras and Topology“. This is also part of categorifying manifold invariants, since the Reshitikhin-Turaev Invariants are based on some monoidal category, which in this case is the category of representations of some algebra.  Categorifying this to a 2-category gives higher-dimensional equivalents of the RT invariants.  The idea (which you can check out in those slides) is that this comes down to describing the analog of the “highest-weight” representations for some Lie algebra you’ve already categorified.

The Lie theory point here, you might remember, is that representations of Lie algebras \mathfrak{g} can be analyzed by decomposing them into “weight spaces” V_{\lambda}, associated to weights \lambda : \mathfrak{g} \rightarrow \mathbf{k} (where \mathbf{k} is the base field, which we can generally assume is \mathbb{C}).  Weights turn Lie algebra elements into scalars, then.  So weight spaces generalize eigenspaces, in that acting by any element g \in \mathfrak{g} on a “weight vector” v \in V_{\lambda} amounts to multiplying by \lambda{g}.  (So that v is an eigenvector for each g, but the eigenvalue depends on g, and is given by the weight.)  A weight can be the “highest” with respect to a natural order that can be put on weights (\lambda \geq \mu if the difference is a nonnegative combination of simple weights).  Then a “highest weight representation” is one which is generated under the action of \mathfrak{g} by a single weight vector v, the “highest weight vector”.

The point of the categorification is to describe the representation in the same terms.  First, we introduce a special strand (which Ben Webster draws as a red strand) which represents the highest weight vector.  Then we say that the category that stands in for the highest weight representation is just what we get by starting with this red strand, and putting all the various string diagrams of the categorification of \mathfrak{g} next to it.  One can then go on to talk about tensor products of these representations, where objects are found by amalgamating several such diagrams (with several red strands) together.  And so on.  These categorified representations are then supposed to be usable to give higher-dimensional manifold invariants.

Now, the flip side of higher categories that reproduce ordinary representation theory would be the representation theory of higher categories in their natural habitat, so to speak. Presumably there should be a fairly uniform picture where categorifications of normal representation theory will be special cases of this. Vlodymyr Mazorchuk gave an interesting talk called 2-representations of finitary 2-categories.  He gave an example of one of the 2-categories that shows up a lot in these Khovanov-Lauda categorifications, the 2-category of Soergel Bimodules mentioned above.  This has one object, which we can think of as a category of modules over the algebra \mathbb{C}[x_1, \dots, x_n]/I (where I  is some ideal of homogeneous symmetric polynomials).  The morphisms are endofunctors of this category, which all amount to tensoring with certain bimodules – the irreducible ones being the Soergel bimodules.  The point of the talk was to explain the representations of 2-categories \mathcal{C} – that is, 2-functors from \mathcal{C} into some “classical” 2-category.  Examples would be 2-categories like “2-vector spaces”, or variants on it.  The examples he gave: (1) [small fully additive \mathbf{k}-linear categories], (2) the full subcategory of it with finitely many indecomposible elements, (3) [categories equivalent to module categories of finite dimensional associative \mathbf{k}-algebras].  All of these have some claim to be a 2-categorical analog of [vector spaces].  In general, Mazorchuk allowed representations of “FIAT” categories: Finitary (Two-)categories with Involutions and Adjunctions.

Part of the process involved getting a “multisemigroup” from such categories: a set S with an operation which takes pairs of elements, and returns a subset of S, satisfying some natural associativity condition.  (Semigroups are the case where the subset contains just one element – groups are the case where furthermore the operation is invertible).  The idea is that FIAT categories have some set of generators – indecomposable 1-morphisms – and that the multisemigroup describes which indecomposables show up in a composite.  (If we think of the 2-category as a monoidal category, this is like talking about a decomposition of a tensor product of objects).  So, for instance, for the 2-category that comes from the monoidal category of \mathfrak{sl}(2) modules, we get the semigroup of nonnegative integers.  For the Soergel bimodule 2-category, we get the symmetric group.  This sort of thing helps characterize when two objects are equivalent, and in turn helps describe 2-representations up to some equivalence.  (You can find much more detail behind the link above.)

On the more classical representation-theoretic side of things, Joel Kamnitzer gave a talk called “Spiders and Buildings”, which was concerned with some geometric and combinatorial constructions in representation theory.  These involved certain trivalent planar graphs, called “webs”, whose edges carry labels between 1 and (n-1).  They’re embedded in a disk, and the outgoing edges, with labels (k_1, \dots, k_m) determine a representation space for a group G, say G = SL_n, namely the tensor product of a bunch of wedge products, \otimes_j \wedge^{k_j} \mathbb{C}^n, where SL_n acts on \mathbb{C}^n as usual.  Then a web determines an invariant vector in this space.  This comes about by having invariant vectors for each vertex (the basic case where m =3), and tensoring them together.  But the point is to interpret this construction geometrically.  This was a bit outside my grasp, since it involves the Langlands program and the geometric Satake correspondence, neither of which I know much of anything about, but which give geometric/topological ways of constructing representation categories.  One thing I did pick up is that it uses the “Langlands dual group” \check{G} of G to get a certain metric space called Gn_{\check{G}}.  Then there’s a correspondence between the category of representations of G and the category of (perverse, constructible) sheaves on this space.  This correspondence can be used to describe the vectors that come out of these webs.

Jim Dolan gave a couple of talks while I was there, which actually fit together as two parts of a bigger picture – one was during the workshop itself, and one at the logic seminar on the following Monday. It helped a lot to see both in order to appreciate the overall point, so I’ll mix them a bit indiscriminately. The first was called “Dimensional Analysis is Algebraic Geometry”, and the second “Toposes of Quasicoherent Sheaves on Toric Varieties”. For the purposes of the logic seminar, he gave the slogan of the second talk as “Algebraic Geometry is a branch of Categorical Logic”. Jim’s basic idea was inspired by Bill Lawvere’s concept of a “theory”, which is supposed to extend both “algebraic theories” (such as the “theory of groups”) and theories in the sense of physics.  Any given theory is some structured category, and “models” of the theory are functors into some other category to represent it – it thus has a functor category called its “moduli stack of models”.  A physical theory (essentially, models which depict some contents of the universe) has some parameters.  The “theory of elastic scattering”, for instance, has the masses, and initial and final momenta, of two objects which collide and “scatter” off each other.  The moduli space for this theory amounts to assignments of values to these parameters, which must satisfy some algebraic equations – conservation of energy and momentum (for example, \sum_i m_i v_i^{in} = \sum_i m_i v_i^{out}, where i \in 1, 2).  So the moduli space is some projective algebraic variety.  Jim explained how “dimensional analysis” in physics is the study of line bundles over such varieties (“dimensions” are just such line bundles, since a “dimension” is a 1-dimensional sort of thing, and “quantities” in those dimensions are sections of the line bundles).  Then there’s a category of such bundles, which are organized into a special sort of symmetric monoidal category – in fact, it’s contrained so much it’s just a graded commutative algebra.

In his second talk, he generalized this to talk about categories of sheaves on some varieties – and, since he was talking in the categorical logic seminar, he proposed a point of view for looking at algebraic geometry in the context of logic.  This view could be summarized as: Every (generalized) space studied by algebraic geometry “is” the moduli space of models for some theory in some doctrine.  The term “doctrine” is Bill Lawvere’s, and specifies what kind of structured category the theory and the target of its models are supposed to be (and of course what kind of functors are allowed as models).  Thus, for instance, toposes (as generalized spaces) are supposed to be thought of as “geometric theories”.  He explained that his “dimensional analysis doctrine” is a special case of this.  As usual when talking to Jim, I came away with the sense that there’s a very large program of ideas lurking behind everything he said, of which only the tip of the iceberg actually made it into the talks.

Next post, when I have time, will talk about the meeting at Erlangen…

Now for a more sketchy bunch of summaries of some talks presented at the HGTQGR workshop.  I’ll organize this into a few themes which appeared repeatedly and which roughly line up with the topics in the title: in this post, variations on TQFT, plus 2-group and higher forms of gauge theory; in the next post, gerbes and cohomology, plus talks on discrete models of quantum gravity and suchlike physics.

TQFT and Variations

I start here for no better reason than the personal one that it lets me put my talk first, so I’m on familiar ground to start with, for which reason also I’ll probably give more details here than later on.  So: a TQFT is a linear representation of the category of cobordisms – that is, a (symmetric monoidal) functor nCob \rightarrow Vect, in the notation I mentioned in the first school post.  An Extended TQFT is a higher functor nCob_k \rightarrow k-Vect, representing a category of cobordisms with corners into a higher category of k-Vector spaces (for some definition of same).  The essential point of my talk is that there’s a universal construction that can be used to build one of these at k=2, which relies on some way of representing nCob_2 into Span(Gpd), whose objects are groupoids, and whose morphisms in Hom(A,B) are pairs of groupoid homomorphisms A \leftarrow X \rightarrow B.  The 2-morphisms have an analogous structure.  The point is that there’s a 2-functor \Lambda : Span(Gpd) \rightarrow 2Vect which is takes representations of groupoids, at the level of objects; for morphisms, there is a “pull-push” operation that just uses the restricted and induced representation functors to move a representation across a span; the non-trivial (but still universal) bit is the 2-morphism map, which uses the fact that the restriction and induction functors are bi-ajdoint, so there are units and counits to use.  A construction using gauge theory gives groupoids of connections and gauge transformations for each manifold or cobordism.  This recovers a form of the Dijkgraaf-Witten model.  In principle, though, any way of getting a groupoid (really, a stack) associated to a space functorially will give an ETQFT this way.  I finished up by suggesting what would need to be done to extend this up to higher codimension.  To go to codimension 3, one would assign an object (codimension-3 manifold) a 3-vector space which is a representation 2-category of 2-groupoids of connections valued in 2-groups, and so on.  There are some theorems about representations of n-groupoids which would need to be proved to make this work.

The fact that different constructions can give groupoids for spaces was used by the next speaker, Thomas Nicklaus, whose talk described another construction that uses the \Lambda I mentioned above.  This one produces “Equivariant Dijkgraaf-Witten Theory”.  The point is that one gets groupoids for spaces in a new way.  Before, we had, for a space M a groupoid \mathcal{A}_G(M) whose objects are G-connections (or, put another way, bundles-with-connection) and whose morphisms are gauge transformations.  Now we suppose that there’s some group J which acts weakly (i.e. an action defined up to isomorphism) on \mathcal{A}_G(M).  We think of this as describing “twisted bundles” over M.  This is described by a quotient stack \mathcal{A}_G // J (which, as a groupoid, gets some extra isomorphisms showing where two objects are related by the J-action).  So this gives a new map nCob \rightarrow Span(Gpd), and applying \Lambda gives a TQFT.  The generating objects for the resulting 2-vector space are “twisted sectors” of the equivariant DW model.  There was some more to the talk, including a description of how the DW model can be further mutated using a cocycle in the group cohomology of G, but I’ll let you look at the slides for that.

Next up was Jamie Vicary, who was talking about “(1,2,3)-TQFT”, which is another term for what I called “Extended” TQFT above, but specifying that the objects are 1-manifolds, the morphisms 2-manifolds, and the 2-morphisms are 3-manifolds.  He was talking about a theorem that identifies oriented TQFT’s of this sort with “anomaly-free modular tensor categories” – which is widely believed, but in fact harder than commonly thought.  It’s easy enough that such a TQFT Z corresponds to a MTC – it’s the category Z(S^1) assigned to the circle.  What’s harder is showing that the TQFT’s are equivalent functors iff the categories are equivalent.  This boils down, historically, to the difficulty of showing the category is rigid.  Jamie was talking about a project with Bruce Bartlett and Chris Schommer-Pries, whose presentation of the cobordism category (described in the school post) was the basis of their proof.

Part of it amounts to giving a description of the TQFT in terms of certain string diagrams.  Jamie kindly credited me with describing this point of view to him: that the codimension-2 manifolds in a TQFT can be thought of as “boundaries in space” – codimension-1 manifolds are either time-evolving boundaries, or else slices of space in which the boundaries live; top-dimension cobordisms are then time-evolving slices of space-with-boundary.  (This should be only a heuristic way of thinking – certainly a generic TQFT has no literal notion of “time-evolution”, though in that (2+1) quantum gravity can be seen as a TQFT, there’s at least one case where this picture could be taken literally.)  Then part of their proof involves showing that the cobordisms can be characterized by taking vector spaces on the source and target manifolds spanned by the generating objects, and finding the functors assigned to cobordisms in terms of sums over all “string diagrams” (particle worldlines, if you like) bounded by the evolving boundaries.  Jamie described this as a “topological path integral”.  Then one has to describe the string diagram calculus – ridigidy follows from the “yanking” rule, for instance, and this follows from Morse theory as in Chris’ presentation of the cobordism category.

There was a little more discussion about what the various properties (proved in a similar way) imply.  One is “cloaking” – the fact that a 2-morphism which “creates a handle” is invisible to the string diagrams in the sense that it introduces a sum over all diagrams with a string “looped” around the new handle, but this sum gives a result that’s equal to the original map (in any “pivotal” tensor category, as here).

Chronologically before all these, one of the first talks on such a topic was by Rafael Diaz, on Homological Quantum Field Theory, or HLQFT for short, which is a rather different sort of construction.  Remember that Homotopy QFT, as described in my summary of Tim Porter’s school sessions, is about linear representations of what I’ll for now call Cob(d,B), whose morphisms are d-dimensional cobordisms equipped with maps into a space B up to homotopy.  HLQFT instead considers cobordisms equipped with maps taken up to homology.

Specifically, there’s some space M, say a manifold, with some distinguished submanifolds (possibly boundary components; possibly just embedded submanifolds; possibly even all of M for a degenerate case).  Then we define Cob_d^M to have objects which are (d-1)-manifolds equipped with maps into M which land on the distinguished submanifolds (to make composition work nicely, we in fact assume they map to a single point).  Morphisms in Cob_d^M are trickier, and look like (N,\alpha, \xi): a cobordism N in this category is likewise equipped with a map \alpha from its boundary into M which recovers the maps on its objects.  That \xi is a homology class of maps from N to M, which agrees with \alpha.  This forms a monoidal category as with standard cobordisms.  Then HLQFT is about representations of this category.  One simple case Rafael described is the dimension-1 case, where objects are (ordered sets of) points equipped with maps that pick out chosen submanifolds of M, and morphisms are just braids equipped with homology classes of “paths” joining up the source and target submanifolds.  Then a representation might, e.g., describe how to evolve a homology class on the starting manifold to one on the target by transporting along such a path-up-to-homology.  In higher dimensions, the evolution is naturally more complicated.

A slightly looser fit to this section is the talk by Thomas Krajewski, “Quasi-Quantum Groups from Strings” (see this) – he was talking about how certain algebraic structures arise from “string worldsheets”, which are another way to describe cobordisms.  This does somewhat resemble the way an algebraic structure (Frobenius algebra) is related to a 2D TQFT, but here the string worldsheets are interacting with 3-form field, H (the curvature of that 2-form field B of string theory) and things needn’t be topological, so the result is somewhat different.

Part of the point is that quantizing such a thing gives a higher version of what happens for quantizing a moving particle in a gauge field.  In the particle case, one comes up with a line bundle (of which sections form the Hilbert space) and in the string case one comes up with a gerbe; for the particle, this involves associated 2-cocycle, and for the string a 3-cocycle; for the particle, one ends up producing a twisted group algebra, and for the string, this is where one gets a “quasi-quantum group”.  The algebraic structures, as in the TQFT situation, come from, for instance, the “pants” cobordism which gives a multiplication and a comultiplication (by giving maps H \otimes H \rightarrow H or the reverse, where H is the object assigned to a circle).

There is some machinery along the way which I won’t describe in detail, except that it involves a tricomplex of forms – the gradings being form degree, the degree of a cocycle for group cohomology, and the number of overlaps.  As observed before, gerbes and their higher versions have transition functions on higher numbers of overlapping local neighborhoods than mere bundles.  (See the paper above for more)

Higher Gauge Theory

The talks I’ll summarize here touch on various aspects of higher-categorical connections or 2-groups (though at least one I’ll put off until later).  The division between this and the section on gerbes is a little arbitrary, since of course they’re deeply connected, but I’m making some judgements about emphasis or P.O.V. here.

Apart from giving lectures in the school sessions, John Huerta also spoke on “Higher Supergroups for String Theory”, which brings “super” (i.e. \mathbb{Z}_2-graded) objects into higher gauge theory.  There are “super” versions of vector spaces and manifolds, which decompose into “even” and “odd” graded parts (a.k.a. “bosonic” and “fermionic” parts).  Thus there are “super” variants of Lie algebras and Lie groups, which are like the usual versions, except commutation properties have to take signs into account (e.g. a Lie superalgebra’s bracket is commutative if the product of the grades of two vectors is odd, anticommutative if it’s even).  Then there are Lie 2-algebras and 2-groups as well – categories internal to this setting.  The initial question has to do with whether one can integrate some Lie 2-algebra structures to Lie 2-group structures on a spacetime, which depends on the existence of some globally smooth cocycles.  The point is that when spacetime is of certain special dimensions, this can work, namely dimensions 3, 4, 6, and 10.  These are all 2 more than the real dimensions of the four real division algebras, \mathbb{R}, \mathbb{C}, \mathbb{H} and \mathbb{O}.  It’s in these dimensions that Lie 2-superalgebras can be integrated to Lie 2-supergroups.  The essential reason is that a certain cocycle condition will hold because of the properties of a form on the Clifford algebras that are associated to the division algebras.  (John has some related material here and here, though not about the 2-group case.)

Since we’re talking about higher versions of Lie groups/algebras, an important bunch of concepts to categorify are those in representation theory.  Derek Wise spoke on “2-Group Representations and Geometry”, based on work with Baez, Baratin and Freidel, most fully developed here, but summarized here.  The point is to describe the representation theory of Lie 2-groups, in particular geometrically.  They’re to be represented on (in general, infinite-dimensional) 2-vector spaces of some sort, which is chosen to be a category of measurable fields of Hilbert spaces on some measure space, which is called H^X (intended to resemble, but not exactly be the same as, Hilb^X, the space of “functors into Hilb from the space X, the way Kapranov-Voevodsky 2-vector spaces can be described as Vect^k).  The first work on this was by Crane and Sheppeard, and also Yetter.  One point is that for 2-groups, we have not only representations and intertwiners between them, but 2-intertwiners between these.  One can describe these geometrically – part of which is a choice of that measure space (X,\mu).

This done, we can say that a representation of a 2-group is a 2-functor \mathcal{G} \rightarrow H^X, where \mathcal{G} is seen as a one-object 2-category.  Thinking about this geometrically, if we concretely describe \mathcal{G} by the crossed module (G,H,\rhd,\partial), defines an action of G on X, and a map X \rightarrow H^* into the character group, which thereby becomes a G-equivariant bundle.  One consequence of this description is that it becomes possible to distinguish not only irreducible representations (bundles over a single orbit) and indecomposible ones (where the fibres are particularly simple homogeneous spaces), but an intermediate notion called “irretractible” (though it’s not clear how much this provides).  An intertwining operator between reps over X and Y can be described in terms of a bundle of Hilbert spaces – which is itself defined over the pullback of X and Y seen as G-bundles over H^*.  A 2-intertwiner is a fibre-wise map between two such things.  This geometric picture specializes in various ways for particular examples of 2-groups.  A physically interesting one, which Crane and Sheppeard, and expanded on in that paper of [BBFW] up above, deals with the Poincaré 2-group, and where irreducible representations live over mass-shells in Minkowski space (or rather, the dual of H \cong \mathbb{R}^{3,1}).

Moving on from 2-group stuff, there were a few talks related to 3-groups and 3-groupoids.  There are some new complexities that enter here, because while (weak) 2-categories are all (bi)equivalent to strict 2-categories (where things like associativity and the interchange law for composing 2-cells hold exactly), this isn’t true for 3-categories.  The best strictification result is that any 3-category is (tri)equivalent to a Gray category – where all those properties hold exactly, except for the interchange law (\alpha \circ \beta) \cdot (\alpha ' \circ \beta ') = (\alpha \cdot \alpha ') \circ (\beta \circ \beta ') for horizontal and vertical compositions of 2-cells, which is replaced by an “interchanger” isomorphism with some coherence properties.  John Barrett gave an introduction to this idea and spoke about “Diagrams for Gray Categories”, describing how to represent morphisms, 2-morphisms, and 3-morphisms in terms of higher versions of “string” diagrams involving (piecewise linear) surfaces satisfying some properties.  He also carefully explained how to reduce the dimensions in order to make them both clearer and easier to draw.  Bjorn Gohla spoke on “Mapping Spaces for Gray Categories”, but since it was essentially a shorter version of a talk I’ve already posted about, I’ll leave that for now, except to point out that it linked to the talk by Joao Faria Martins, “3D Holonomy” (though see also this paper with Roger Picken).

The point in Joao’s talk starts with the fact that we can describe holonomies for 3-connections on 3-bundles valued in Gray-groups (i.e. the maximally strict form of a general 3-group) in terms of Gray-functors hol: \Pi_3(M) \rightarrow \mathcal{G}.  Here, \Pi_3(M) is the fundamental 3-groupoid of M, which turns points, paths, homotopies of paths, and homotopies of homotopies into a Gray groupoid (modulo some technicalities about “thin” or “laminated”  homotopies) and \mathcal{G} is a gauge Gray-group.  Just as a 2-group can be represented by a crossed module, a Gray (3-)group can be represented by a “2-crossed module” (yes, the level shift in the terminology is occasionally confusing).  This is a chain of groups L \stackrel{\delta}{\rightarrow} E \stackrel{\partial}{\rightarrow} G, where G acts on the other groups, together with some structure maps (for instance, the Peiffer commutator for a crossed module becomes a lifting \{ ,\} : E \times E \rightarrow L) which all fit together nicely.  Then a tri-connection can be given locally by forms valued in the Lie algebras of these groups: (\omega , m ,\theta) in  \Omega^1 (M,\mathfrak{g} ) \times \Omega^2 (M,\mathfrak{e}) \times \Omega^3(M,\mathfrak{l}).  Relating the global description in terms of hol and local description in terms of (\omega, m, \theta) is a matter of integrating forms over paths, surfaces, or 3-volumes that give the various j-morphisms of \Pi_3(M).  This sort of construction of parallel transport as functor has been developed in detail by Waldorf and Schreiber (viz. these slides, or the full paper), some time ago, which is why, thematically, they’re the next two speakers I’ll summarize.

Konrad Waldorf spoke about “Abelian Gauge Theories on Loop Spaces and their Regression”.  (For more, see two papers by Konrad on this)  The point here is that there is a relation between two kinds of theories – string theory (with B-field) on a manifold M, and ordinary U(1) gauge theory on its loop space LM.  The relation between them goes by the name “regression” (passing from gauge theory on LM to string theory on M), or “transgression”, going the other way.  This amounts to showing an equivalence of categories between [principal U(1)-bundles with connection on LM] and [U(1)-gerbes with connection on M].  This nicely gives a way of seeing how gerbes “categorify” bundles, since passing to the loop space – whose points are maps S^1 \rightarrow M means a holonomy functor is now looking at objects (points in LM) which would be morphisms in the fundamental groupoid of M, and morphisms which are paths of loops (surfaces in M which trace out homotopies).  So things are shifted by one level.  Anyway, Konrad explained how this works in more detail, and how it should be interpreted as relating connections on loop space to the B-field in string theory.

Urs Schreiber kicked the whole categorification program up a notch by talking about \infty-Connections and their Chern-Simons Functionals .  So now we’re getting up into \infty-categories, and particularly \infty-toposes (see Jacob Lurie’s paper, or even book if so inclined to find out what these are), and in particular a “cohesive topos”, where derived geometry can be developed (Urs suggested people look here, where a bunch of background is collected). The point is that \infty-topoi are good for talking about homotopy theory.  We want a setting which allows all that structure, but also allows us to do differential geometry and derived geometry.  So there’s a “cohesive” \infty-topos called Smooth\infty Gpds, of “sheaves” (in the \infty-topos sense) of \infty-groupoids on smooth manifolds.  This setting is the minimal common generalization of homotopy theory and differential geometry.

This is about a higher analog of this setup: since there’s a smooth classifying space (in fact, a Lie groupoid) for G-bundles, BG, there’s also an equivalence between categories G-Bund of G-principal bundles, and SmoothGpd(X,BG) (of functors into BG).  Moreover, there’s a similar setup with BG_{conn} for bundles with connection.  This can be described topologically, or there’s also a “differential refinement” to talk about the smooth situation.  This equivalence lives within a category of (smooth) sheaves of groupoids.  For higher gauge theory, we want a higher version as in Smooth \infty Gpds described above.  Then we should get an equivalence – in this cohesive topos – of hom(X,B^n U(1)) and a category of U(1)-(n-1)-gerbes.

Then the part about the  “Chern-Simons functionals” refers to the fact that CS theory for a manifold (which is a kind of TQFT) is built using an action functional that is found as an integral of the forms that describe some U(1)-connection over the manifold.  (Then one does a path-integral of this functional over all connections to find partition functions etc.)  So the idea is that for these higher U(1)-gerbes, whose classifying spaces we’ve just described, there should be corresponding functionals.  This is why, as Urs remarked in wrapping up, this whole picture has an explicit presentation in terms of forms.  Actually, in terms of Cech-cocycles (due to the fact we’re talking about gerbes), whose coefficients are taken in sheaves of complexes (this is the derived geometry part) of differential forms whose coefficients are in L_\infty-algebroids (the \infty-groupoid version of Lie algebras, since in general we’re talking about a theory with gauge \infty-groupoids now).

Whew!  Okay, that’s enough for this post.  Next time, wrapping up blogging the workshop, finally.

Continuing from the previous post, there are a few more lecture series from the school to talk about.

Higher Gauge Theory

The next was John Huerta’s series on Higher Gauge Theory from the point of view of 2-groups.  John set this in the context of “categorification”, a slightly vague program of replacing set-based mathematical ideas with category-based mathematical ideas.  The general reason for this is to get an extra layer of “maps between things”, or “relations between relations”, etc. which tend to be expressed by natural transformations.  There are various ways to go about this, but one is internalization: given some sort of structure, the relevant example in this case being “groups”, one has a category {Groups}, and can define a 2-group as a “category internal to {Groups}“.  So a 2-group has a group of objects, a group of morphisms, and all the usual maps (source and target for morphisms, composition, etc.) which all have to be group homomorphisms.  It should be said that this all produces a “strict 2-group”, since the objects G necessarily form a group here.  In particular, m : G \times G \rightarrow G satisfies group axioms “on the nose” – which is the only way to satisfy them for a group made of the elements of a set, but for a group made of the elements of a category, one might require only that it commute up to isomorphism.  A weak 2-group might then be described as a “weak model” of the theory of groups in Cat, but this whole approach is much less well-understood than the strict version as one goes to general n-groups.

Now, as mentioned in the previous post, there is a 1-1 correspondence between 2-groups and crossed modules (up to equivalence): given a crossed module (G,H,\partial,\rhd), there’s a 2-group \mathcal{G} whose objects are G and whose morphisms are G \ltimes H; given a 2-group \mathcal{G} with objects G, there’s a crossed module (G, Aut(1_G),1,m).  (The action m acts on a morphism in such as way as to act by multiplication on its source and target).  Then, for instance, the Peiffer identity for crossed modules (see previous post) is a consequence of the fact that composition of morphisms is supposed to be a group homomorphism.

Looking at internal categories in [your favourite setting here] isn’t the only way to do categorification, but it does produce some interesting examples.  Baez-Crans 2-vector spaces are defined this way (in Vect), and built using these are Lie 2-algebras.  Looking for a way to integrate Lie 2-algebras up to Lie 2-groups (which are internal categories in Lie groups) brings us back to the current main point.  This is the use of 2-groups to do higher gauge theory.  This requires the use of “2-bundles”.  To explain these, we can say first of all that a “2-space” is an internal category in Spaces (whether that be manifolds, or topological spaces, or what-have-you), and that a (locally trivial) 2-bundle should have a total 2-space E, a base 2-space M, and a (functorial) projection map p : E \rightarrow M, such that there’s some open cover of M by neighborhoods U_i where locally the bundle “looks like” \pi_i : U_i \times F \rightarrow U_i, where F is the fibre of the bundle.  In the bundle setting, “looks like” means “is isomorphic to” by means of isomorphisms f_i : E_{U_i} \rightarrow U_i \times F.  With 2-bundles, it’s interpreted as “is equivalent to” in the categorical sense, likewise by maps f_i.

Actually making this precise is a lot of work when M is a general 2-space – even defining open covers and setting up all the machinery properly is quite hard.  This has been done, by Toby Bartels in his thesis, but to keep things simple, John restricted his talk to the case where M is just an ordinary manifold (thought of as a 2-space which has only identity morphisms).   Then a key point is that there’s an analog to how (principal) G-bundles (where F \cong G as a G-set) are classified up to isomorphism by the first Cech cohomology of the manifold, \check{H}^1(M,G).  This works because one can define transition functions on double overlaps U_{ij} := U_i \cap U_j, by g_{ij} = f_i f_j^{-1}.  Then these g_{ij} will automatically satisfy the 1-cocycle condidion (g_{ij} g_{jk} = g_{ik} on the triple overlap U_{ijk}) which means they represent a cohomology class [g] = \in \check{H}^1(M,G).

A comparable thing can be said for the “transition functors” for a 2-bundle – they’re defined superficially just as above, except that being functors, we can now say there’s a natural isomorphism h_{ijk} : g_{ij}g_{jk} \rightarrow g_{ik}, and it’s these h_{ijk}, defined on triple overlaps, which satisfy a 2-cocycle condition on 4-fold intersections (essentially, the two ways to compose them to collapse g_{ij} g_{jk} g_{kl} into g_{il} agree).  That is, we have g_{ij} : U_{ij} \rightarrow Ob(\mathcal{G}) and h_{ijk} : U_{ijk} \rightarrow Mor(\mathcal{G}) which fit together nicely.  In particular, we have an element [h] \in \check{H}^2(M,G) of the second Cech cohomology of M: “principal \mathcal{G}-bundles are classified by second Cech cohomology of M“.  This sort of thing ties in to an ongoing theme of the later talks, the relationship between gerbes and higher cohomology – a 2-bundle corresponds to a “gerbe”, or rather a “1-gerbe”.  (The consistent terminology would have called a bundle a “0-gerbe”, but as usual, terminology got settled before the general pattern was understood).

Finally, having defined bundles, one usually defines connections, and so we do the same with 2-bundles.  A connection on a bundle gives a parallel transport operation for paths \gamma in M, telling how to identify the fibres at points along \gamma by means of a functor hol : P_1(M) \rightarrow G, thinking of G as a category with one object, and where P_1(M) is the path groupoid whose objects are points in M and whose morphisms are paths (up to “thin” homotopy). At least, it does so once we trivialize the bundle around \gamma, anyway, to think of it as M \times G locally – in general we need to get the transition functions involved when we pass into some other local neighborhood.  A connection on a 2-bundle is similar, but tells how to parallel transport fibres not only along paths, but along homotopies of paths, by means of hol : P_2(M) \rightarrow \mathcal{G}, where \mathcal{G} is seen as a 2-category with one object, and P_2(M) now has 2-morphisms which are (essentially) homotopies of paths.

Just as connections can be described by 1-forms A valued in Lie(G), which give hol by integrating, a similar story exists for 2-connections: now we need a 1-form A valued in Lie(G) and a 2-form B valued in Lie(H).  These need to satisfy some relations, essentially that the curvature of A has to be controlled by B.   Moreover, that B is related to the B-field of string theory, as I mentioned in the post on the pre-school… But really, this is telling us about the Lie 2-algebra associated to \mathcal{G}, and how to integrate it up to the group!

Infinite Dimensional Lie Theory and Higher Gauge Theory

This series of talks by Christoph Wockel returns us to the question of “integrating up” to a Lie group G from a Lie algebra \mathfrak{g} = Lie(G), which is seen as the tangent space of G at the identity.  This is a well-understood, well-behaved phenomenon when the Lie algebras happen to be finite dimensional.  Indeed the classification theorem for the classical Lie groups can be got at in just this way: a combinatorial way to characterize Lie algebras using Dynkin diagrams (which describe the structure of some weight lattice), followed by a correspondence between Lie algebras and Lie groups.  But when the Lie algebras are infinite dimensional, this just doesn’t have to work.  It may be impossible to integrate a Lie algebra up to a full Lie group: instead, one can only get a little neighborhood of the identity.  The point of such infinite-dimensional groups, and ultimately their representation theory, is to deal with string groups that have to do with motions of extended objects.  Christoph Wockel was describing a result which says that, going to 2-groups, this problem can be overcome.  (See the relevant paper here.)

The first lecture in the series presented some background on a setting for infinite dimensional manifolds.  There are various approaches, a popular one being Frechet manifolds, but in this context, the somewhat weaker notion of locally convex spaces is sufficient.  These are “locally modelled” by (infinite dimensional) locally convex vector spaces, the way finite dimensonal manifolds are locally modelled by Euclidean space.  Being locally convex is enough to allow them to support a lot of differential calculus: one can find straight-line paths, locally, to define a notion of directional derivative in the direction of a general vector.  Using this, one can build up definitions of differentiable and smooth functions, derivatives, and integrals, just by looking at the restrictions to all such directions.  Then there’s a fundamental theorem of calculus, a chain rule, and so on.

At this point, one has plenty of differential calculus, and it becomes interesting to bring in Lie theory.  A Lie group is defined as a group object in the category of manifolds and smooth maps, just as in the finite-dimensional case.  Some infinite-dimensional Lie groups of interest would include: G = Diff(M), the group of diffeomorphisms of some compact manifold M; and the group of smooth functions G = C^{\infty}(M,K) from M into some (finite-dimensional) Lie group K (perhaps just \mathbb{R}), with the usual pointwise multiplication.  These are certainly groups, and one handy fact about such groups is that, if they have a manifold structure near the identity, on some subset that generates G as a group in a nice way, you can extend the manifold structure to the whole group.  And indeed, that happens in these examples.

Well, next we’d like to know if we can, given an infinite dimensional Lie algebra X, “integrate up” to a Lie group – that is, find a Lie group G for which X \cong T_eG is the “infinitesimal” version of G.  One way this arises is from central extensions.  A central extension of Lie group G by Z is an exact sequence Z \hookrightarrow \hat{G} \twoheadrightarrow G where (the image of) Z is in the centre of \hat{G}.  The point here is that \hat{G} extends G.  This setup makes \hat{G} is a principal Z-bundle over G.

Now, finding central extensions of Lie algebras is comparatively easy, and given a central extension of Lie groups, one always falls out of the induced maps.  There will be an exact sequence of Lie algebras, and now the special condition is that there must exist a continuous section of the second map.  The question is to go the other way: given one of these, get back to an extension of Lie groups.  The problem of finding extensions of G by Z, in particular as a problem of finding a bundle with connection having specified curvature, which brings us back to gauge theory.  One type of extension is the universal cover of G, which appears as \pi_1(G) \hookrightarrow \hat{G} \twoheadrightarrow G, so that the fibre is \pi_1(G).

In general, whether an extension can exist comes down to a question about a cocycle: that is, if there’s a function f : G \times G \rightarrow Z which is locally smooth (i.e. in some neighborhood in G), and is a cocyle (so that f(g,h) + f(gh,k) = f(g,hk) + f(h,k)), by the same sorts of arguments we’ve already seen a bit of.  For this reason, central extensions are classified by the cohomology group H^2(G,Z).  The cocycle enables a “twisting” of the multiplication associated to a nontrivial loop in G, and is used to construct \hat{G} (by specifying how multiplication on G lifts to \hat{G}).  Given a  2-cocycle \omega at the Lie algebra level (easier to do), one would like to lift that up the Lie group.  It turns out this is possible if the period homomorphism per_{\omega} : \Pi_2(G) \rightarrow Z – which takes a chain [\sigma] (with \sigma : S^2 \rightarrow G) to the integral of the original cocycle on it, \int_{\sigma} \omega – lands in a discrete subgroup of Z. A popular example of this is when Z is just \mathbb{R}, and the discrete subgroup is \mathbb{Z} (or, similarly, U(1) and 1 respectively).  This business of requiring a cocycle to be integral in this way is sometimes called a “prequantization” problem.

So suppose we wanted to make the “2-connected cover” \pi_2(G) \hookrightarrow \pi_2(G) \times_{\gamma} G \twoheadrightarrow G as a central extension: since \pi_2(G) will be abelian, this is conceivable.  If the dimension of G is finite, this is trivial (since \pi_2(G) = 0 in finite dimensions), which is why we need some theory  of infinite-dimensional manifolds.  Moreover, though, this may not work in the context of groups: the \gamma in the extension \pi_2(G) \times_{\gamma} G above needs to be a “twisting” of associativity, not multiplication, being lifted from G.  Such twistings come from the THIRD cohomology of G (see here, e.g.), and describe the structure of 2-groups (or crossed modules, whichever you like).  In fact, the solution (go read the paper for more if you like) to define a notion of central extension for 2-groups (essentially the same as the usual definition, but with maps of 2-groups, or crossed modules, everywhere).  Since a group is a trivial kind of 2-group (with only trivial automorphisms of any element), the usual notion of central extension turns out to be a special case.  Then by thinking of \pi_2(G) and G as crossed modules, one can find a central extension which is like the 2-connected cover we wanted – though it doesn’t work as an extension of groups because we think of G as the base group of the crossed module, and \pi_2(G) as the second group in the tower.

The pattern of moving to higher group-like structures, higher cohomology, and obstructions to various constructions ran all through the workshop, and carried on in the next school session…

Higher Spin Structures in String Theory

Hisham Sati gave just one school-lecture in addition to his workshop talk, but it was packed with a lot of material.  This is essentially about cohomology and the structures on manifolds to which cohomology groups describe the obstructions.  The background part of the lecture referenced this book by Fridrich, and the newer parts were describing some of Sati’s own work, in particular a couple of papers with Schreiber and Stasheff (also see this one).

The basic point here is that, for physical reasons, we’re often interested in putting some sort of structure on a manifold, which is really best described in terms of a bundle.  For instance, a connection or spin connection on spacetime lets us transport vectors or spinors, respectively, along paths, which in turn lets us define derivatives.  These two structures really belong on vector bundles or spinor bundles.  Now, if these bundles are trivial, then one can make the connections on them trivial as well by gauge transformation.  So having nontrivial bundles really makes this all more interesting.  However, this isn’t always possible, and so one wants to the obstruction to being able to do it.  This is typically a class in one of the cohomology groups of the manifold – a characteristic class.  There are various examples: Chern classes, Pontrjagin classes, Steifel-Whitney classes, and so on, each of which comes in various degrees i.  Each one corresponds to a different coefficient group for the cohomology groups – in these examples, the groups U and O which are the limits of the unitary and orthogonal groups (such as O := O(\infty) \supset \dots \supset O(2) \supset O(1))

The point is that these classes are obstructions to building certain structures on the manifold X – which amounts to finding sections of a bundle.  So for instance, the first Steifel-Whitney classes, w_1(E) of a bundle E are related to orientations, coming from cohomology with coefficients in O(n).  Orientations for the manifold X can be described in terms of its tangent bundle, which is an O(n)-bundle (tangent spaces carry an action of the rotation group).  Consider X = S^1, where we have actually O(1) \simeq \mathbb{Z}_2.  The group H^1(S^1, \mathbb{Z}_2) has two elements, and there are two types of line bundle on the circle S^1: ones with a nowhere-zero section, like the trivial bundle; and ones without, like the Moebius strip.  The circle is orientable, because its tangent bundle is of the first sort.

Generally, an orientation can be put on X if the tangent bundle, as a map f : X \rightarrow B(O(n)), can be lifted to a map \tilde{f} : X \rightarrow B(SO(n)) – that is, it’s “secretly” an SO(n)-bundle – the special orthogonal group respects orientation, which is what the determinant measures.  Its two values, \pm 1, are what’s behind the two classes of bundles.  (In short, this story relates to the exact sequence 1 \rightarrow SO(n) \rightarrow O(n) \stackrel{det}{\rightarrow} O(1) = \mathbb{Z}_2 \rightarrow 1; in just the same way we have big groups SO, Spin, and so forth.)

So spin structures have a story much like the above, but where the exact sequence 1 \rightarrow \mathbb{Z}_2 \rightarrow Spin(n) \rightarrow SO(n) \rightarrow 1 plays a role – the spin groups are the universal covers (which are all double-sheeted covers) of the special rotation groups.  A spin structure on some SO(n) bundle E, let’s say represented by f : X \rightarrow B(SO(n)) is thus, again, a lifting to \tilde{f} : X \rightarrow B(Spin(n)).  The obstruction to doing this (the thing which must be zero for the lifting to exist) is the second Stiefel-Whitney class, w_2(E).  Hisham Sati also explained the example of “generalized” spin structures in these terms.  But the main theme is an analogous, but much more general, story for other cohomology groups as obstructions to liftings of some sort of structures on manifolds.  These may be bundles, for the lower-degree cohomology, or they may be gerbes or n-bundles, for higher-degree, but the setup is roughly the same.

The title’s term “higher spin structures” comes from the fact that we’ve so far had a tower of classifying spaces (or groups), B(O) \leftarrow B(SO) \leftarrow B(Spin), and so on.  Then the problem of putting various sorts of structures on X has been turned into the problem of lifting a map f : X \rightarrow S(O) up this tower.  At each point, the obstruction to lifting is some cohomology class with coefficients in the groups (O, SO, etc.)  So when are these structures interesting?

This turns out to bring up another theme, which is that of special dimensions – it’s just not true that the same phenomena happen in every dimension.  In this case, this has to do with the homotopy groups  – of O and its cousins.  So it turns out that the homotopy group \pi_k(O) (which is the same as \pi_k(O_n) as long as n is bigger than k) follows a pattern, where \pi_k(O) = \mathbb{Z}_2 if k = 0,1 (mod 8), and \pi_k(O) = \mathbb{Z} if k = 3,7 (mod 8).  The fact that this pattern repeats mod-8 is one form of the (real) Bott Periodicity theorem.  These homotopy groups reflect that, wherever there’s nontrivial homotopy in some dimension, there’s an obstruction to contracting maps into O from such a sphere.

All of this plays into the question of what kinds of nontrivial structures can be put on orthogonal bundles on manifolds of various dimensions.  In the dimensions where these homotopy groups are non-trivial, there’s an obstruction to the lifting, and therefore some interesting structure one can put on X which may or may not exist.  Hisham Sati spoke of “killing” various homotopy groups – meaning, as far as I can tell, imposing conditions which get past these obstructions.  In string theory, his application of interest, one talks of “anomaly cancellation” – an anomaly being the obstruction to making these structures.  The first part of the punchline is that, since these are related to nontrivial cohomology groups, we can think of them in terms of defining structures on n-bundles or gerbes.  These structures are, essentially, connections – they tell us how to parallel-transport objects of various dimensions.  It turns out that the \pi_k homotopy group is related to parallel transport along (k-1)-dimensional surfaces in X, which can be thought of as the world-sheets of (k-2)-dimensional “particles” (or rather, “branes”).

So, for instance, the fact that \pi_1(O) is nontrivial means there’s an obstruction to a lifting in the form of a class in H^2(X,\mathbb{Z}), which has to do with spin structure – as above.  “Cancelling” this “anomaly” means that for a theory involving such a spin structure to be well-defined, then this characteristic class for X must be zero.  The fact that \pi_3(O) = \mathbb{Z} is nontrivial means there’s an obstruction to a lifting in the form of a class in H^4(X, \mathbb{Z}).  This has to do with “string bundles”, where the string group is a higher analog of Spin in exactly the sense we’ve just described.  If such a lifting exists, then there’s a “string-structure” on X which is compatible with the spin structure we lifted (and with the orientation a level below that).  Similarly, \pi_7(O) = \mathbb{Z} being nontrivial, by way of an obstruction in H^8, means there’s an interesting notion of “five-brane” structure, and a Fivebrane group, and so on.  Personally, I think of these as giving a geometric interpretation for what the higher cohomology groups actually mean.

A slight refinement of the above, and actually more directly related to “cancellation” of the anomalies, is that these structures can be defined in a “twisted” way: given a cocycle in the appropriate cohomology group, we can ask that a lifting exist, not on the nose, but as a diagram commuting only up to a higher cell, which is exactly given by the cocycle.  I mentioned, in the previous section, a situation where the cocycle gives an associator, so that instead of being exactly associative, a structure has a “twisted” associativity.  This is similar, except we’re twisting the condition that makes a spin structure (or higher spin structure) well-defined.  So if X has the wrong characteristic class, we can only define one of these twisted structures at that level.

This theme of higher cohomology and gerbes, and their geometric interpretation, was another one that turned up throughout the talks in the workshop…

And speaking of that: coming up soon, some descriptions of the actual workshop.

Marco Mackaay recently pointed me at a paper by Mikhail Khovanov, which describes a categorification of the Heisenberg algebra H (or anyway its integral form H_{\mathbb{Z}}) in terms of a diagrammatic calculus.  This is very much in the spirit of the Khovanov-Lauda program of categorifying Lie algebras, quantum groups, and the like.  (There’s also another one by Sabin Cautis and Anthony Licata, following up on it, which I fully intend to read but haven’t done so yet. I may post about it later.)

Now, as alluded to in some of the slides I’ve from recent talks, Jamie Vicary and I have been looking at a slightly different way to answer this question, so before I talk about the Khovanov paper, I’ll say a tiny bit about why I was interested.

Groupoidification

The Weyl algebra (or the Heisenberg algebra – the difference being whether the commutation relations that define it give real or imaginary values) is interesting for physics-related reasons, being the algebra of operators associated to the quantum harmonic oscillator.  The particular approach to categorifying it that I’ve worked with goes back to something that I wrote up here, and as far as I know, originally was suggested by Baez and Dolan here.  This categorification is based on “stuff types” (Jim Dolan’s term, based on “structure types”, a.k.a. Joyal’s “species”).  It’s an example of the groupoidification program, the point of which is to categorify parts of linear algebra using the category Span(Gpd).  This has objects which are groupoids, and morphisms which are spans of groupoids: pairs of maps G_1 \leftarrow X \rightarrow G_2.  Since I’ve already discussed the backgroup here before (e.g. here and to a lesser extent here), and the papers I just mentioned give plenty more detail (as does “Groupoidification Made Easy“, by Baez, Hoffnung and Walker), I’ll just mention that this is actually more naturally a 2-category (maps between spans are maps X \rightarrow X' making everything commute).  It’s got a monoidal structure, is additive in a fairly natural way, has duals for morphisms (by reversing the orientation of spans), and more.  Jamie Vicary and I are both interested in the quantum harmonic oscillator – he did this paper a while ago describing how to construct one in a general symmetric dagger-monoidal category.  We’ve been interested in how the stuff type picture fits into that framework, and also in trying to examine it in more detail using 2-linearization (which I explain here).

Anyway, stuff types provide a possible categorification of the Weyl/Heisenberg algebra in terms of spans and groupoids.  They aren’t the only way to approach the question, though – Khovanov’s paper gives a different (though, unsurprisingly, related) point of view.  There are some nice aspects to the groupoidification approach: for one thing, it gives a nice set of pictures for the morphisms in its categorified algebra (they look like groupoids whose objects are Feynman diagrams).  Two great features of this Khovanov-Lauda program: the diagrammatic calculus gives a great visual representation of the 2-morphisms; and by dealing with generators and relations directly, it describes, in some sense1, the universal answer to the question “What is a categorification of the algebra with these generators and relations”.  Here’s how it works…

Heisenberg Algebra

One way to represent the Weyl/Heisenberg algebra (the two terms refer to different presentations of isomorphic algebras) uses a polynomial algebra P_n = \mathbb{C}[x_1,\dots,x_n].  In fact, there’s a version of this algebra for each natural number n (the stuff-type references above only treat n=1, though extending it to “n-sorted stuff types” isn’t particularly hard).  In particular, it’s the algebra of operators on P_n generated by the “raising” operators a_k(p) = x_k \cdot p and the “lowering” operators b_k(p) = \frac{\partial p}{\partial x_k}.  The point is that this is characterized by some commutation relations.  For j \neq k, we have:

[a_j,a_k] = [b_j,b_k] = [a_j,b_k] = 0

but on the other hand

[a_k,b_k] = 1

So the algebra could be seen as just a free thing generated by symbols \{a_j,b_k\} with these relations.  These can be understood to be the “raising and lowering” operators for an n-dimensional harmonic oscillator.  This isn’t the only presentation of this algebra.  There’s another one where [p_k,q_k] = i (as in i = \sqrt{-1}) has a slightly different interpretation, where the p and q operators are the position and momentum operators for the same system.  Finally, a third one – which is the one that Khovanov actually categorifies – is skewed a bit, in that it replaces the a_j with a different set of \hat{a}_j so that the commutation relation actually looks like

[\hat{a}_j,b_k] = b_{k-1}\hat{a}_{j-1}

It’s not instantly obvious that this produces the same result – but the \hat{a}_j can be rewritten in terms of the a_j, and they generate the same algebra.  (Note that for the one-dimensional version, these are in any case the same, taking a_0 = b_0 = 1.)

Diagrammatic Calculus

To categorify this, in Khovanov’s sense (though see note below1), means to find a category \mathcal{H} whose isomorphism classes of objects correspond to (integer-) linear combinations of products of the generators of H.  Now, in the Span(Gpd) setup, we can say that the groupoid FinSet_0, or equvialently \mathcal{S} = \coprod_n  \mathcal{S}_n, represents Fock space.  Groupoidification turns this into the free vector space on the set of isomorphism classes of objects.  This has some extra structure which we don’t need right now, so it makes the most sense to describe it as \mathbb{C}[[t]], the space of power series (where t^n corresponds to the object [n]).  The algebra itself is an algebra of endomorphisms of this space.  It’s this algebra Khovanov is looking at, so the monoidal category in question could really be considered a bicategory with one object, where the monoidal product comes from composition, and the object stands in formally for the space it acts on.  But this space doesn’t enter into the description, so we’ll just think of \mathcal{H} as a monoidal category.  We’ll build it in two steps: the first is to define a category \mathcal{H}'.

The objects of \mathcal{H}' are defined by two generators, called Q_+ and Q_-, and the fact that it’s monoidal (these objects will be the categorifications of a and b).  Thus, there are objects Q_+ \otimes Q_- \otimes Q_+ and so forth.  In general, if \epsilon is some word on the alphabet \{+,-\}, there’s an object Q_{\epsilon} = Q_{\epsilon_1} \otimes \dots \otimes Q_{\epsilon_m}.

As in other categorifications in the Khovanov-Lauda vein, we define the morphisms of \mathcal{H}' to be linear combinations of certain planar diagrams, modulo some local relations.  (This type of formalism comes out of knot theory – see e.g. this intro by Louis Kauffman).  In particular, we draw the objects as sequences of dots labelled + or -, and connect two such sequences by a bunch of oriented strands (embeddings of the interval, or circle, in the plane).  Each + dot is the endpoint of a strand oriented up, and each - dot is the endpoint of a strand oriented down.  The local relations mean that we can take these diagrams up to isotopy (moving the strands around), as well as various other relations that define changes you can make to a diagram and still represent the same morphism.  These relations include things like:

which seems visually obvious (imagine tugging hard on the ends on the left hand side to straighten the strands), and the less-obvious:

and a bunch of others.  The main ingredients are cups, caps, and crossings, with various orientations.  Other diagrams can be made by pasting these together.  The point, then, is that any morphism is some \mathbf{k}-linear combination of these.  (I prefer to assume \mathbf{k} = \mathbb{C} most of the time, since I’m interested in quantum mechanics, but this isn’t strictly necessary.)

The second diagram, by the way, are an important part of categorifying the commutation relations.  This would say that Q_- \otimes Q_+ \cong Q_+ \otimes Q_- \oplus 1 (the commutation relation has become a decomposition of a certain tensor product).  The point is that the left hand sides show the composition of two crossings Q_- \otimes Q_+ \rightarrow Q_+ \otimes Q_- and Q_+ \otimes Q_- \rightarrow Q_- \otimes Q_+ in two different orders.  One can use this, plus isotopy, to show the decomposition.

That diagrams are invariant under isotopy means, among other things, that the yanking rule holds:

(and similar rules for up-oriented strands, and zig zags on the other side).  These conditions amount to saying that the functors - \otimes Q_+ and - \otimes Q_- are two-sided adjoints.  The two cups and caps (with each possible orientation) give the units and counits for the two adjunctions.  So, for instance, in the zig-zag diagram above, there’s a cup which gives a unit map \mathbf{k} \rightarrow Q_- \otimes Q_+ (reading upward), all tensored on the right by Q_-.  This is followed by a cap giving a counit map Q_+ \otimes Q_- \rightarrow \mathbf{k} (all tensored on the left by Q_-).  So the yanking rule essentially just gives one of the identities required for an adjunction.  There are four of them, so in fact there are two adjunctions: one where Q_+ is the left adjoint, and one where it’s the right adjoint.

Karoubi Envelope

Now, so far this has explained where a category \mathcal{H}' comes from – the one with the objects Q_{\epsilon} described above.  This isn’t quite enough to get a categorification of H_{\mathbb{Z}}: it would be enough to get the version with just one a and one b element, and their powers, but not all the a_j and b_k.  To get all the elements of the (integral form of) the Heisenberg algebras, and in particular to get generators that satisfy the right commutation relations, we need to introduce some new objects.  There’s a convenient way to do this, though, which is to take the Karoubi envelope of \mathcal{H}'.

The Karoubi envelope of any category \mathcal{C} is a universal way to find a category Kar(\mathcal{C}) that contains \mathcal{C} and for which all idempotents split (i.e. have corresponding subobjects).  Think of vector spaces, for example: a map p \in End(V) such that p^2 = p is a projection.  That projection corresponds to a subspace W \subset V, and W is actually another object in Vect, so that p splits (factors) as V \rightarrow W subset V.  This might not happen in any general \mathcal{C}, but it will in Kar(\mathcal{C}).  This has, for objects, all the pairs (C,p) where p : C \rightarrow C is idempotent (so \mathcal{C} is contained in Kar(\mathcal{C}) as the cases where p=1).  The morphisms f : (C,p) \rightarrow (C',p') are just maps f : C \rightarrow C' with the compatibility condition that p' f = p f = f (essentially, maps between the new subobjects).

So which new subobjects are the relevant ones?  They’ll be subobjects of tensor powers of our Q_{\pm}.  First, consider Q_{+^n} = Q_+^{\otimes n}.  Obviously, there’s an action of the symmetric group \mathcal{S}_n on this, so in fact (since we want a \mathbf{k}-linear category), its endomorphisms contain a copy of \mathbf{k}[\mathcal{S}_n], the corresponding group algebra.  This has a number of different projections, but the relevant ones here are the symmetrizer,:

e_n = \frac{1}{n!} \sum_{\sigma \in \mathcal{S}_n} \sigma

which wants to be a “projection onto the symmetric subspace” and the antisymmetrizer:

e'_n = \frac{1}{n!} \sum_{\sigma \in \mathcal{S}_n} sign(\sigma) \sigma

which wants to be a “projection onto the antisymmetric subspace” (if it were in a category with the right sub-objects). The diagrammatic way to depict this is with horizontal bars: so the new object S^n_+ = (Q_{+^n}, e) (the symmetrized subobject of Q_+^{\oplus n}) is a hollow rectangle, labelled by n.  The projection from Q_+^{\otimes n} is drawn with n arrows heading into that box:

The antisymmetrized subobject \Lambda^n_+ = (Q_{+^n},e') is drawn with a black box instead.  There are also S^n_- and \Lambda^n_- defined in the same way (and drawn with downward-pointing arrows).

The basic fact – which can be shown by various diagram manipulations, is that S^n_- \otimes \Lambda^m_+ \cong (\Lambda^m_+ \otimes S^n_-) \oplus (\Lambda_+^{m-1} \otimes S^{n-1}_-).  The key thing is that there are maps from the left hand side into each of the terms on the right, and the sum can be shown to be an isomorphism using all the previous relations.  The map into the second term involves a cap that uses up one of the strands from each term on the left.

There are other idempotents as well – for every partition \lambda of n, there’s a notion of \lambda-symmetric things – but ultimately these boil down to symmetrizing the various parts of the partition.  The main point is that we now have objects in \mathcal{H} = Kar(\mathcal{H}') corresponding to all the elements of H_{\mathbb{Z}}.  The right choice is that the \hat{a}_j  (the new generators in this presentation that came from the lowering operators) correspond to the S^j_- (symmetrized products of “lowering” strands), and the b_k correspond to the \Lambda^k_+ (antisymmetrized products of “raising” strands).  We also have isomorphisms (i.e. diagrams that are invertible, using the local moves we’re allowed) for all the relations.  This is a categorification of H_{\mathbb{Z}}.

Some Generalities

This diagrammatic calculus is universal enough to be applied to all sorts of settings where there are functors which are two-sided adjoints of one another (by labelling strands with functors, and the regions of the plane with categories they go between).  I like this a lot, since biadjointness of certain functors is essential to the 2-linearization functor \Lambda (see my link above).  In particular, \Lambda uses biadjointness of restriction and induction functors between representation categories of groupoids associated to a groupoid homomorphism (and uses these unit and counit maps to deal with 2-morphisms).  That example comes from the fact that a (finite-dimensional) representation of a finite group(oid) is a functor into Vect, and a group(oid) homomorphism is also just a functor F : H \rightarrow G.  Given such an F, there’s an easy “restriction” F^* : Fun(G,Vect) \rightarrow Fun(H,Vect), that just works by composing with F.  Then in principle there might be two different adjoints Fun(H,Vect) \rightarrow Fun(G,Vect), given by the left and right Kan extension along F.  But these are defined by colimits and limits, which are the same for (finite-dimensional) vector spaces.  So in fact the adjoint is two-sided.

Khovanov’s paper describes and uses exactly this example of biadjointness in a very nice way, albeit in the classical case where we’re just talking about inclusions of finite groups.  That is, given a subgroup H < G, we get a functors Res_G^H : Rep(G) \rightarrow Rep(H), which just considers the obvious action of H act on any representation space of G.  It has a biadjoint Ind^G_H : Rep(H) \rightarrow Rep(G), which takes a representation V of H to \mathbf{k}[G] \otimes_{\mathbf{k}[H]} V, which is a special case of the formula for a Kan extension.  (This formula suggests why it’s also natural to see these as functors between module categories \mathbf{k}[G]-mod and \mathbf{k}[H]-mod).  To talk about the Heisenberg algebra in particular, Khovanov considers these functors for all the symmetric group inclusions \mathcal{S}_n < \mathcal{S}_{n+1}.

Except for having to break apart the symmetric groupoid as S = \coprod_n \mathcal{S}_n, this is all you need to categorify the Heisenberg algebra.  In the Span(Gpd) categorification, we pick out the interesting operators as those generated by the - \sqcup \{\star\} map from FinSet_0 to itself, but “really” (i.e. up to equivalence) this is just all the inclusions \mathcal{S}_n < \mathcal{S}_{n+1} taken at once.  However, Khovanov’s approach is nice, because it separates out a lot of what’s going on abstractly and uses a general diagrammatic way to depict all these 2-morphisms (this is explained in the first few pages of Aaron Lauda’s paper on ambidextrous adjoints, too).  The case of restriction and induction is just one example where this calculus applies.

There’s a fair bit more in the paper, but this is probably sufficient to say here.


1 There are two distinct but related senses of “categorification” of an algebra A here, by the way.  To simplify the point, say we’re talking about a ring R.  The first sense of a categorification of R is a (monoidal, additive) category C with a “valuation” in R that takes \otimes to \times and \oplus to +.  This is described, with plenty of examples, in this paper by Rafael Diaz and Eddy Pariguan.  The other, typical of the Khovanov program, says it is a (monoidal, additive) category C whose Grothendieck ring is K_0(C) = R.  Of course, the second definition implies the first, but not conversely.  The objects of the Grothendieck ring are isomorphism classes in C.  A valuation may identify objects which aren’t isomorphic (or, as in groupoidification, morphisms which aren’t 2-isomorphic).

So a categorification of the first sort could be factored into two steps: first take the Grothendieck ring, then take a quotient to further identify things with the same valuation.  If we’re lucky, there’s a commutative square here: we could first take the category C, find some surjection C \rightarrow C', and then find that K_0(C') = R.  This seems to be the relation between Khovanov’s categorification of H_{\mathbb{Z}} and the one in Span(Gpd). This is the sense in which it seems to be the “universal” answer to the problem.

A more substantial post is upcoming, but I wanted to get out this announcement for a conference I’m helping to organise, along with Roger Picken, João Faria Martins, and Aleksandr Mikovic.  Its website: https://sites.google.com/site/hgtqgr/home has more details, and will have more as we finalise them, but here are some of them:

Workshop and School on Higher Gauge Theory, TQFT and Quantum Gravity

Lisbon, 10-13 February, 2011 (Workshop), 7-13 February, 2011 (School)

Description from the website:

Higher gauge theory is a fascinating generalization of ordinary abelian and non-abelian gauge theory, involving (at the first level) connection 2-forms, curvature 3-forms and parallel transport along surfaces. This ladder can be continued to connection forms of higher degree and transport along extended objects of the corresponding dimension. On the mathematical side, higher gauge theory is closely tied to higher algebraic structures, such as 2-categories, 2-groups etc., and higher geometrical structures, known as gerbes or n-gerbes with connection. Thus higher gauge theory is an example of the categorification phenomenon which has been very influential in mathematics recently.

There have been a number of suggestions that higher gauge theory could be related to (4D) quantum gravity, e.g. by Baez-Huerta (in the QG^2 Corfu school lectures), and Baez-Baratin-Freidel-Wise in the context of state-sums. A pivotal role is played by TQFTs in these approaches, in particular BF theories and variants thereof, as well as extended TQFTs, constructed from suitable geometric or algebraic data. Another route between higher gauge theory and quantum gravity is via string theory, where higher gauge theory provides a setting for n-form fields, worldsheets for strings and branes, and higher spin structures (i.e. string structures and generalizations, as studied e.g. by Sati-Schreiber-Stasheff). Moving away from point particles to higher-dimensional extended objects is a feature both of loop quantum gravity and string theory, so higher gauge theory should play an important role in both approaches, and may allow us to probe a deeper level of symmetry, going beyond normal gauge symmetry.

Thus the moment seems ripe to bring together a group of researchers who could shed some light on these issues. Apart from the courses and lectures given by the invited speakers, we plan to incorporate discussion sessions in the afternoon throughout the week, for students to ask questions and to stimulate dialogue between participants from different backgrounds.

Provisional list of speakers:

  • Paolo Aschieri (Alessandria)
  • Benjamin Bahr (Cambridge)
  • Aristide Baratin (Paris-Orsay)
  • John Barrett (Nottingham)
  • Rafael Diaz (Bogotá)
  • Bianca Dittrich (Potsdam)
  • Laurent Freidel (Perimeter)
  • John Huerta (California)
  • Branislav Jurco (Prague)
  • Thomas Krajewski (Marseille)
  • Tim Porter (Bangor)
  • Hisham Sati (Maryland)
  • Christopher Schommer-Pries (MIT)
  • Urs Schreiber (Utrecht)
  • Jamie Vicary (Oxford)
  • Konrad Waldorf (Regensburg)
  • Derek Wise (Erlangen)
  • Christoph Wockel (Hamburg)

The workshop portion will have talks by the speakers above (those who can make it), and any contributed talks.  The “school” portion is, roughly, aimed at graduate students in a field related to the topics, but not necessarily directly in them.  You don’t need to be a student to attend the school, of course, but they are the target audience.  The only course that has been officially announced so far will be given by Christopher Schommer-Pries, on TQFT.  We hope/expect to also have minicourses on Higher Gauge Theory, and Quantum Gravity as well, but details aren’t settled yet.

If you’re interested, the deadline to register is Jan 8 (hence the rush to announce).  Some funding is available for those who need it.

In the most recent TQFT Club seminar, we had a couple of talks – one was the second in a series of three by Marco Mackaay, which as promised previously I’ll write up together after the third one.

The other was by Björn Gohla, a student of João Faria Martins, giving an overview on the subject of “Tricategories and Trifunctors”, a mostly expository talk explaining some definitions.  Actually, this was a bit more specific than a general introduction – the point of it was to describe a certain kind of mapping space.  I’ve talked here before about representing the “configuration space” of a gauge theory as a groupoid: the objects are (optionally, flat) connections on a manifold M, and the morphisms are gauge transformations taking one connection to another.  The reason for the things Björn was talking about is analogous, except that in this case, the goal is to describe the configuration space of a higher gauge theory.

There are at least two ways I know of to talk about higher gauge theory.  One is in terms of categorical (or n-categorical) groups – which makes it a “categorification” of gauge theory in the sense of reproducing in \mathbf{Cat} (or \mathbf{nCat}) an analog of a sturcture, gauge theory, originally formulated in \mathbf{Set}.  Among other outlines, you might look at this one by John Baez and John Huerta for an introduction.  Another uses the lingo of crossed modules or crossed complexes.  In either case, the essential point is the same: there is some collection of groups (or groupoids, but let’s say groups to keep everything clear) which play the role of the single gauge group in ordinary gauge theory.

In the first language, we can speak of a “2-group”, or “categorical group” – a group internal to \mathbf{Cat}, or what is equivalent, a category internal to \mathbf{Grp}, which would have a group of objects and a group of morphisms (and, in higher settings still, groups of 2-morphisms, 3-morphisms, and so on).  The structure maps of the category (source, target, composition, etc.) have to live in the category of groups.

A crossed complex of groups (again, we could generalize to groupoids, but I won’t) is a nonabelian variation on a chain complex: a sequence of groups with maps from one to the next.  There are also a bunch more structures, which ultimately serve to reproduce all the kind of composition, source, and target maps in the n-categorical groups: some groups act on others, there are “bracket” operations on one group valued in another, and so forth.  This paper by Brown and Higgins explains how the two concepts are related when most of the groups are abelian, and there’s a lot more about crossed complexes and related stuff in Tim Porter’s “Crossed Menagerie“.

The point of all this right now is that these things play the role of the gauge group in higher gauge theory.  The idea is that in gauge theory, you have a connection.  Typically this is described in terms of a form valued in the Lie algebra of the gauge group.  Then a (thin) homotopy classes of curves, gets a holonomy valued in the group by integrating that form.  Alternatively, you can just think of the path groupoid of a manifold \mathcal{P}_1(M), where those classes of curves form the morphisms between the objects, which are just points of M.  Then a connection defines a functor \Gamma : \mathcal{P}_1(M) \rightarrow G, where G is the gauge group thought of as a category (groupoid in fact) with one object.  Or, you can just define a connection that way in the first place.  In higher gauge theory, a similar principle exists: begin with the n-path groupoid \mathcal{P}_n(M) where the morphisms are (thin homotopy classes of) paths, the 2-morphisms are surfaces (really homotopy classes of homotopies of paths), and so on, so the k-morphisms are k-dimensional bits of M.  Then you could define an n-connection as a n-functor into an n-group as defined above.  OR, you could define it in terms of a tower of differential k-forms valued in the crossed complex of Lie algebras associated to the crossed complex of Lie groups that replaces the gauge group.  You can then use an integral to get an element of the group at level k of the complex for any given k-morphism in \mathcal{P}_n(M), which (via the equivalence I mentioned) amounts to the same thing as the other definition of connection.

João Martins has done some work on this sort of thing when n is dimension 2 (with Tim Porter) and 3 (with Roger Picken), which I guess is how Björn came to work on this question.  The question is, roughly, how to describe the moduli space of these connections.  The gist of the answer is that it’s a functor n-category [\mathcal{P}_n(M),\mathcal{G}], where \mathcal{G} is the n-group.  A little more generally, the question is how to describe mapping spaces for higher categories.  In particular, he was talking about the case n=3, which is where certain tricky issues start to show up.  In particular every bicategory (the weakest form of 2-category) is (bi)equivalent to a strict 2-category, so there’s no real need to worry about weakening things like associativity so that they only work up to isomorphism – these are all equalities.  With 3-categories, this fails: the weakest kind of 3-category is a tricategory (introduced by Gordon, Power and Street, though also see the references beyond that link).  These are always tri-equivalent to something stricter than general, but not completely strict: Gray-categories.  The only equation from 2-categories which has to be weakened to an isomorphism here is the interchange law: given a square of four morphisms, we can either compose vertically first, and then horizontally, or vice versa.  In a Gray-category, there’s an “interchanger” isomorphism

I_{\alpha,\alpha ',\beta,\beta'} : (\alpha \circ \beta) \cdot (\alpha ' \circ \beta ') \Rightarrow (\alpha \cdot \alpha ') \circ (\beta \cdot \beta ')

where \cdot is vertical composition of 2-cells, and \circ is horizontal (i.e. the same direction as 1-cells).  This is supposed to satisfy a compatibility condition.  It’s essentially the only one you can come up with starting with (\alpha \cdot \alpha ') \circ \beta (and composing it in different orders by throwing in identities in various places).

There’s another way to look at things, as Björn explained, in terms of enriched category theory.  If you have a monoidal category (\mathcal{V},\otimes), then a (\mathcal{V},\otimes)-enriched category \mathbb{G} is one in which, for any two objects x,y, there is an object \mathbb{G}(x,y) \in \mathcal{V} of morphisms, and composition gives morphisms \circ_{x,y,z} : \mathbb{G}(y,z) \otimes \mathbb{G}(x,y) \rightarrow \mathbb{G}(x,z).  A strict 3-category is enriched in \mathbf{2Cat}, with its usual tensor product, dual to its internal hom [-,-] (which gives the mapping 2-category of functors, natural transformations, and modifications, between any two 2-categories).  A Gray category is similar, except that it is enriched in \mathbf{Gray}, a version of \mathbf{2Cat} with a different tensor product, dual to the hom functor [-,-]' which gives the mapping 2-category with pseudonatural transformations (the weak version of the concept, where the naturality square only has to commute up to a specified 2-cell) as morphisms.  These are not the same, which is where the unavoidability of weakening 3-categories “really” comes from.   The upshot of this is as above: it matters which order we compose things in.

Having defined Gray-categories, let’s say A and B (which, in the applications I mentioned above, tend to actually be Gray-groupoids, though this doesn’t change the theory substantially), the point is to talk about “mapping spaces” – that is, Gray-categories of Gray-functors (etc.) from A to B.

Since they’ve been defined terms of enriched category theory, one wants to use the general theory of enriched functors, transformations, and so forth – which is a lot easier than trying to work out the correct definitions from scratch using a low-level description.  So then a Gray-functor F : A \rightarrow B has an object map F_0 : A_0 \rightarrow B_0, mapping objects of A to objects of B, and then for each x,y \in A_0, a morphism in \mathbf{Gray} (which is our \mathcal{V}), namely F_{x,y} : A(x,y) \rightarrow B(F(x),F(y)).  There are a bunch of compatibility conditions, which can be expressed for any monoidal category \mathcal{V} (since they involve diagrams with the map \circ_{x,y,z} for any triple, and the like).  Similar comments apply to defining \mathcal{V}-natural transformations.

There is a slight problem here, which is that in this case, \mathcal{V} = \mathbf{Gray} is a 2-category, so we really need to use a form of weakly enriched categories…  All the compatibility diagrams should have 2-cells in them, and so forth.  This, too, gets complicated.  So Björn explained is a shortcut from drawing n-dimensional diagrams for these mapping n-categories in terms of the arrow category \vec{B}. This is the category whose objects are the morphisms of B, and whose morphisms are commuting squares, or when B is a 2-category, squares with a 2-cell, so a morphism in \vec{B} from f: x \rightarrow y to f' : x' \rightarrow y' is a triple g = (g_x,g_y,g_f) like so:

Morphism in arrow category

The 2-morphisms in \vec{B} are commuting “pillows”, where the front and back face are morphisms like the above. So \beta : g \Rightarrow g' is \beta = (\beta_x,\beta_y), where \beta_x : g_x \Rightarrow g_{x'} is a 2-cell, and the whole “pillow” commutes.  When B is a tricategory, then we need to go further – these 2-morphsims should be triples including a 3-cell \beta_f filling the “pillow”, and then 3-morphisms are commuting structures between these. These diagrams get hard to draw pretty quickly. This is the point of having an ordinary 2D diagram with at most 1-dimensional cells: pushing all the nasty diagrams into these arrow categories, we can replace a 2-cell representing a natural transformation with a diagram involving the arrow category.

This uses that there are source and target maps (which are Gray-functors, of course) which we’ll call d_0, d_1: \vec{B} \rightarrow B. So then here (in one diagram) we have two ways of depicting a natural transformation \alpha :  F \rightarrow G between functors F,G : A \Rightarrow B:

One is the 2-cell, and the other is the functor into \vec{B}, such that d_0 \circ \alpha = F and d_1 \circ \alpha = G.
To depict a modification between natural transformations (a 3-cell between 2-cells) just involves building the arrow category of \vec{B}, say \vec{\vec{B}}, and drawing an arrow from A into it. And so on: in principle, there is a tower above B built by iterating the arrow category construction, and all the different levels of “functor”, “natural transformation”, “modification”, and all the higher equivalents are just functors into different levels of this tower.  (The generic term for the k^{th} level of maps-between-maps-etc between n-categories is “(n,k)-transfor“, a handy term coined here.)
The advantage here is that at least the general idea can be extended pretty readily to higher values of n than 3.  Naturally, no matter which way one decides to do it, things will get complicated – either there’s a combinatorial explosion of things to consider, or one has to draw higher-dimensional diagrams, or whatever.  This exploding complexity of n-categories (in this case, globular ones) is one of the reasons why simplicial appreaches – quasicategories or \infty-categories – are good.  They allow you to avoid talking about those problems, or at least fold them into fairly well-understood aspects of simplicial sets.  A lot of things – limits, colimits, mapping spaces, etc. are pretty well understood in that case (see, for instance, the first chapter of Joshua Nicholls-Barrer’s thesis for the basics, or Jacob Lurie’s humongous book for something more comprehensive).  But sometimes, as in this case, they just don’t happen to be the things you want for your application.  So here we have some tools for talking about mapping spaces in the world of globular n-categories – and as the work by Martins/Porter/Picken show, it’s motivated by some fairly specific work about invariants of manifolds, differential geometry, and so on.

Last week I spoke in Montreal at a session of the Philosophy of Science Association meeting.  Here are some notes for it.  Later on I’ll do a post about the other talks at the meeting.

Right now, though, the meeting slowed me down from describing a recent talk in the seminar here at IST.  This was Gonçalo Rodrigues’ talk on categorifying measure theory.  It was based on this paper here, which is pretty long and goes into some (but not all) of the details.  Apparently an updated version that fills in some of what’s not there is in the works.

In any case, Gonçalo takes as the starting point for categorifying ideas in analysis the paper “Measurable Categories” by David Yetter, which is the same point where I started on this topic, although he then concludes that there are problems with that approach.  Part of the reason for saying this has to do with the fact that the category of Hilbert spaces has many bad properties – or rather, fails to have many of the good ones that it should to play the role one might expect in categorifying ideas from analysis.

Yetter’s idea can be described, very roughly, as follows: we would like to categorify the concept of a function-space on a measure space (X,\mu).  That is, spaces like L^2(X,\mu) or L^{\infty}(X,\mu).  The reason for this is that the 2-vector-spaces of Kapranov and Voevodsky are very elegant, but intrinsically finite-dimensional, categorifications of “vector space”.  An infinite-dimensional version would be important for representation theory, particularly of noncompact Lie groups or 2-groups, but even just infinite ones, since there are relatively few endomorphisms of KV 2-vector spaces.  Yetter’s paper constructs analogs to the space of measurable functions \mathcal{M}(X), where “functions” take values in Hilbert spaces.

A measurable field of Hilbert spaces is, roughly, a family of Hilbert spaces indexed by points of X, together with a nice space of “measurable sections”.  This is supposed to be an infinite-dimensional, measure-theoretic counterpart to an object in a KV 2-vector space, which always looks like \mathbf{Vect}^k for some natural number k, which is now being replaced by (X,\mu).  One of the key tools in Yetter’s paper is the direct integral of a field of Hilbert spaces, which is similarly the counterpart to the direct sum \bigoplus in the discrete world.  It just gives the space of measurable sections (taken up to almost-everywhere equivalence, as usual).  This was the main focus of Gonçalo’s talk.

The direct integral has one major problem, compared to the (finite) direct sum it is supposed to generalize – namely, the direct sum is a categorical coproduct, in \mathbf{Vect} or any other KV 2-vector space.  Actually, it is both a product and a coproduct (\mathbf{Vect} is abelian), so it is defined by a nice universal property.  The direct integral, on the other hand, is not.  It doesn’t have any similarly nice universal property.  (In the infinite-dimensional case, colimits and limits would be expected to become different in any case, but the direct integral is neither).  This means that many proofs in analysis will be hard to reproduce in the categorified setting – universal properties mean one doesn’t have to do nearly as much work to do this, among their other good qualities.  This is related to the issue that the category \mathbf{Hilb} does not have all limits and colimits

Gonçalo’s paper and talk outline a program where one can categorify a lot of the proofs in analysis, by using a slightly different framework which uses a bigger category than \mathbf{Hilb}, namely Ban_C, whose objects are Banach spaces and whose maps are (linear) contractions.  A Banach Category is a category enriched in Ban_C.  Now, Banach spaces have a norm, but not necessarily an inner product, and this small weakening makes them much worse than Hilbert spaces as objects.  Many intuitions from Hilbert spaces, like the one that says any subspace has a complement, just fail: the corresponding notion for Banach spaces is the quasicomplement (X and Y are quasicomplements if they intersect only at zero, and their sum is dense in the whole space), and it’s quite possible to have subspaces which don’t have one.  Other unpleasant properties abound.

Yet Ban_C is a much nicer category than Hilb.  (So we follow the general dictum that it’s better to have a nice category with bad objects than a bad category with nice objects – the same motivation behind “smooth spaces” instead of manifolds, and the like.)  It’s complete and cocomplete (i.e. has all limits and colimits), as well as monoidal closed – for Banach spaces A and B, the space Hom(A,B) is also in Ban_C.  None of these facts holds for Hilb.  On the other hand, the space of bounded maps between Hilbert spaces is a Banach space (with the operator norm), but not necessarily a Hilbert space.  So even Hilb is already a Banach category.

It also turns out that, unlike in Hilb, limits and colimits (where those exist in Hilb) are not necessarily isomorphic.  In particular, in Ban_C, the coproduct and product A + B and A \times B both have the same underlying vector space A \oplus B, but the norms are different.  For Hilbert spaces, the inner product comes from the Pythagorean formula in either case, but for Banach spaces, the coproduct gets the sum of the two norms, and the product gets the supremum.  It turns out that coproducts are the more important concept, and this is where the direct integral comes in.

First, we can talk about Banach 2-spaces (the analogs of 2-vector spaces): these are just Banach categories which are cocomplete (have all weighted colimits).  Maps between them are cocontinuous functors – that is, colimit-preserving ones.  (Though properly, functors between Banach categories ought to be contractions on Hom-spaces).  Then there are categorified analogs of all sorts of Banach space structure in a familiar way – the direct sum (coproduct) is the analog of vector addition, the category Ban_C is the analog of the base field (say, \mathbb{R}), and so on.

This all gives the setting for categorified measure theory.  Part of the point of choosing Ban_C is that you can now reason out at least some of how it works by analogy.  To start with, one needs to fix a Boolean algebra \Omega – this is to be the \sigma-algebra of measurable sets for some measure space, though it’s important that it needn’t have any actual points (this is a notion of measure space akin to the notion of locale in the topological world).  This part of the theory isn’t categorified (arguably a limitation of this approach, but not one that’s any different from Yetter’s).  Instead, we categorify the definition of measure itself.

A measure is a function \mu : \Omega \mapsto \mathbb{R} – it assigns a number to each measurable set.  The pair (\Omega,\mu) is a measure algebra, and relates to a measure space the way a locale relates to a topological space.  So a categorified measure \nu should be a functor from \Omega (seen now as a category) into Ban_C.  (We can generalize this: the measure could be valued in some vector space over \mathbb{R}, and a categorified measure could be a functor into some other Banach 2-space.)  Since we’re thinking of \Omega as a lattice of subsets, it makes some sense to call \nu a presheaf, or rather co-presheaf.  What’s more, just as a measure is additive (\mu(A + B) = \mu(A) + \mu(B), for disjoint sets, where + is the union), so also the categorical measure \nu should be (finitely) additive up to isomorphism.  So we’re assigning Banach spaces to all the measurable sets.  This is a “co”-presheaf – which is to say, a covariant functor, so the spaces “nest”: when for measurable sets, we have A \subset B, then \nu(A) \leq \nu(B) also.

An intuition for how this works comes from a special case (not at all exhaustive), where we start with an actual, uncategorified, measure space (X,\mu).  Then one categorified measure will arise by taking \nu(E) = L_1(E,\mu): the Banach space associated to a measurable set E is the space of integrable functions.  We can take any “scalar” multiple of this, too: given a fixed Banach space B, let \nu(E) = L_1(E,\mu) \otimes B.  But there are lots of examples that aren’t like this.

All this is fine, but the point here is to define integration.  The usual way to go about this when you learn analysis is to start with characteristic functions of measurable sets, then define a sequence through simple functions, measurable functions, and so forth.  Eventually one can define L^p spaces based on the convergence of various integrals.  Something similar happens here.

The analog of a function here is a sheaf: a (compatible) assignment of Banach spaces to measurable sets.  (Technically, to get to sheaves, we need an idea of “cover” by measurable sets, but it’s pretty much the obvious one, modulo the subtlety that we should only allow countable covers.) The idea will be to start with characteristic sheaves for measurable sets, then take some kind of completion of the category of all of these as a definition of “measurable sheaf”.  Then the point will be that we can extend the measure from characteristic sheaves to all measurable sheaves using a limit (actually, a colimit), analogous to the way we define a Lebesgue integral as a limit of simple functions approximating a measurable one.

A characteristic sheaf \chi(E) for a measurable set E \in \Omega might be easiest to visualize in terms of a characteristic bundle, which just puts a copy of the base field (we’ve been assuming it’s \mathbb{R}) at each point of E, and the zero vector space everywhere else.  (This is a bundle in the measurable sense, not the topological one – assuming X has a topology other than \Omega itself.)  Very intuitively, to turn this into a sheaf, one can just use brute force and take a set A the product of all the spaces lying in A.  A little less crudely, one should take a space of sections with decent properties – so that \chi(E) assigns to A a space of functions on E \cap A.  In particular, the functor \chi : \Omega \rightarrow L_{\infty}(\Omega) which picks out all the (measurable) bounded sections is a universal way to do this.

Now the point is that the algebra of measurable sets, \Omega, thought of as a category, embeds into the category of presheaves on it by \chi : \Omega \rightarrow \mathbf{PShv}(\Omega), taking a set to its characteristic sheaf.  Given a measure valued in some Banach category, \nu : \Omega \rightarrow \mathcal{B}, we can find the left Kan extension \int_X d\nu : \mathbf{PShv}(\Omega) \rightarrow \mathcal{B}, such that \nu = \int_X d\nu \circ \chi.  The Kan extension is a universal way to extend \nu to all of \mathbf{PShv}(\Omega) so that this is true, and it can be calculated as a colimit.

The essential fact here is that the characteristic sheaves are dense in \mathbf{PShv}(\Omega): any presheaf can be found as a colimit of the characteristic ones.  This is analogous to how any function can be approximated by linear combinations of characteristic functions.  This means that the integral defined above will actually give interesting results for all the sheaves one might expect.

I’m glossing over some points here, of course – for example, the distinction between sheaves and presheaves, the role of sheafification, etc.  If you want to get a more accurate picture, check out the paper I linked to up above.

All of this granted, however, many of the classical theorems of measure theory have analogs that are proved in essentially the same way as the standard versions.  One can see the presheaf category as a categorified analog of L_1(X,\nu), and get the Fubini theorem, for instance: there is a canonical equivalence (no longer isomorphism) between (a suitable) tensor product of \mathbf{PShv}(X) and \mathbf{PShv}(Y) on one hand, and on the other \mathbf{PShv}(X \times Y).  Doing integration, one can then do all the usual things – exchange order of integration between X and Y, say – in analogous conditions.  The use of universal properties to define integrals etc. means that one doesn’t need to fuss about too much with coherence laws, and so the proofs of the categorified facts are much the same as the original proofs.

Next Page »

Follow

Get every new post delivered to your Inbox.

Join 43 other followers