Skip to the Main Content

Note:These pages make extensive use of the latest XHTML and CSS Standards. They ought to look great in any standards-compliant modern browser. Unfortunately, they will probably look horrible in older browsers, like Netscape 4.x and IE 4.x. Moreover, many posts use MathML, which is, currently only supported in Mozilla. My best suggestion (and you will thank me when surfing an ever-increasing number of sites on the web which have been crafted to use the new standards) is to upgrade to the latest version of your browser. If that's not possible, consider moving to the Standards-compliant and open-source Mozilla browser.

December 15, 2005

Lauda on Frobenius Algebras and Open Topological Strings

Posted by Urs Schreiber

Here are some notes on the preprint

A. Lauda
Frobenius algebras and planar open string topological field theories

The author

1) points out a relation between adjunctions and Frobenius algebras.


2) uses categorified adjunctions in order to describe categorified Frobenius algebras and topological membranes.

Turns out that point 1) relies on the same mechanism which is also responsible for the phenomenon that 2-trivializations of 2-functors give rise to Frobenius algebras.

Thanks to John Baez for making me aware of this paper by means of his TWF.

(I had a pretty long review of this paper almost done yesterday when a computer crash erased it all. Here is a second but inevitably shorter version. )


What I am talking about here is mostly the result presented first in

Aaron Lauda
Frobenius algebras and ambidextrous adjunctions

which is a strengthening of a result found in

M. Müger
From subfactors to categories and topology. I. Frobenius algebras in and Morita equivalence of tensor categories.
J. Pure Appl. Algebra, 180 (2003)

amplified by John Baez in TWF 174.


As a motivation, consider the following fact.

Every (direct sum of) matrix algebra(s) can be regarded as a Frobenius algebra. (Using Wedderburn’s theorem.) The product is ‘index contraction’, the coproduct is ‘insertion of a unit’.

Phrase this in more highbrow terms. Let V be some vector space and End(V)VV * its space of endomorphisms. Let’s regard the monoidal category Vect of vector spaces as a 2-category with a single object, whose 1-morphisms are vector spaces, 2-morphisms are linear maps and horizontal composition is the tensor product. For instance VV * looks like

(1)VV *.

‘Index contraction’ (the evaluation map of covectors on vectors) is a linear map, hence a 2-morphism, of the kind

(2)V *V e K

where K is the ground field which we are working on, regarded as a 1-D vector space. ‘Insertion of a unit’ is the map

(3)K i VV *

which sends 1K to Id VEnd(V). This is the archetypical example of what is called an adjunction in a 2-category. In fact, since Vect is symmetric, the order of V and V * in the above is not really essential and hence we really have something Aaron Lauda calls an ambidextrous adjunction. The way I have approached adjunctions here it is clear that they are related to algebras. It is a simple exercise to show that the product and coproduct coming from e and i above are such that we really have a Frobenius algebra.

Aaron Lauda turns this into a stronger and more precise theorem (his theorem 8) which essentially says that every Frobenius algebra comes from an adjunction along the lines sketched above.

Building on that, the second part of the paper is concerned with categorifying this situation. If you know about the main idea of categorification, know how Frobenius algebras are related to topological strings and know what a topological membrane is supposed to be it should not come as a surprise that there is a way to relate categorified Frobenius algebras with topological membranes. I don’t want to say any more about this part right now.

What I do want to point out is, how the above mentioned result on adjunctions and Frobenius algebras is based on the same general mechanism which relates 2-trivializations of 2-functors with Frobenius algebra.

Namely given a 2-functor (or just a 2-bundle without connection, if you prefer) which I shall call


from some 2-category of surface elements in M to some target 2-category, we can restrict it to open subset U iM and demand that there it looks like a ‘more trivial’ 2-functor tra i in some sense.

Hence we demand there to be morphisms of 2-functors

(5)tra U it itra i


(6)tra it¯ itra U i

which relate the two.

If these were 1-functors, we would define the transition between tra i and tra j as

(7)tra it¯ itra U ijt jtra j = tra ig ijtra j.

But since 2-functor live in a 2-category, we should really replace the equality here with 2-morphisms

(8)tra it¯ itra U ijt jtra j ϕ ij tra ig ijtra j


(9)tra it¯ itra U ijt jtra j ϕ¯ ij tra ig ijtra j.

This way one gets something rather similar to an adjunction. (It’s actually a little more general.) By precisely the same underlying mechanism which Aaron Lauda uses to show the relation between adjunctions and Frobenius algebras, one finds that a 2-trivialization of the above sort gives rise to something like a Frobenius algebra structure.

There would be more to say, and I did before it was erased. Right now this shall be it.

Posted at December 15, 2005 6:10 PM UTC

TrackBack URL for this Entry:

0 Comments & 1 Trackback

Read the post Bimodules, Adjunctions and the Internal Hom, Part I
Weblog: The String Coffee Table
Excerpt: Notes on certain phenomena concerning algebra objects and adjunctions in tensor categories.
Tracked: January 2, 2006 10:07 PM

Post a New Comment