Dear Categorists - Sorry to take a while to respond. People at UCR have been unable to receive posts on the category theory mailing list, due to problems with our internet connection. I'd asked for some nice examples of an object G in a rig category equipped with an isomorphism from G to G^2 + 1. Steve Schanuel replied:
I was unable to understand John Baez' golden object problem, nor his description of the solutions. He refuses to tell us what 'nice' means, [...]
The problem was deliberately open-ended, but you seem to have understood it perfectly, since you've given a nice solution, including a precise specification of what you consider "nice". Let me repeat the two solutions given by Robin Houston: 1) The first solution works in any rig category having an object H equipped with an isomorphism to H^2 + 4H + 1. The solution is to take G = H + 2. I described how Houston uses the isomorphism H -> H^2 + 4H + 1 to construct an isomorphism G^2 -> G + 1. What's nice about this is that it reduces a problem that's not obviously of fixed-point form to one that is. 2) Houston's second solution works in any monoidal cocomplete category, tensor product distributing over colimits, that contains an object X equipped with an isomorphism to 2X + 1. The solution is to let G be the object of "binary planar rooted trees with X-labelled leaves", i.e. G = X + X^2 + 2X^3 + 5X^4 + 14X^5 + 42X^6 + ... where the coefficients are Catalan numbers. He uses the obvious isomorphism G -> G^2 + X to construct an isomorphism G^2 -> G + 1. What's nice about this is that it shows Propp's originally proposed golden object really is one: just take the category of sigma-polytopes with its cartesian product, and let X be the open interval! And, it makes precise the sense in which the alternating sum of Catalan numbers equals the golden ratio. Steve writes:
I don't know whether there is an extensive category with N[X]/(X^2=X+1) as its full Burnside rig; perhaps one or both of the examples John mentioned would do, if I knew what they were.
I think example 1) does the job if we take the free distributive category on an object H equipped with an isomorphism to H^2 + 4H + 1. Right? Steve also writes:
He is very generous, allowing one to use a category with both plus and times as extra monoidal structures. (Does anyone know an example of interest in which the plus is not coproduct?) This freedom is unnecessary [...]
It's unnecessary, but handy: I think there's also an golden object in the rig category of reps of quantum SU(2) at a suitable value of q. Here the tensor product is not cartesian. In the lingo of quantum group theory, this object has "quantum dimension" equal to the golden number. It's interesting how such nonintegral but algebraic "dimensions" show up naturally in quantum group theory, just as nonintegral but algebraic "cardinalities" show up in the theory of distributive categories. I don't know any golden objects in rig categories where the plus is not coproduct, and I agree that such rig categories arise less often than those where times is not product. But, if you use the obvious way of making the groupoid of finite sets into a rig category, + isn't coproduct, nor is x product.
While I'm airing my confusions, can anyone tell me what 'categorification' means? I don't know any such process; the simplest exanple, 'categorifying' natural numbers to get finite sets, seems to me rather 'remembering the finite sets and maps which gave rise to natural numbers by the abstraction of passing to isomorphism classes'.
You're right: categorification is not a systematic process! I explained this idea back in "week121", and also in my paper "Categorification", at http://www.arXiv.org/abs/math.QA/9802029. Here's what I said. If one studies categorification one soon discovers an amazing fact: many deep-sounding results in mathematics are just categorifications of facts we learned in high school! There is a good reason for this. All along, we have been unwittingly `decategorifying' mathematics by pretending that categories are just sets. We `decategorify' a category by forgetting about the morphisms and pretending that isomorphic objects are equal. We are left with a mere set: the set of isomorphism classes of objects. To understand this, the following parable may be useful. Long ago, when shepherds wanted to see if two herds of sheep were isomorphic, they would look for an explicit isomorphism. In other words, they would line up both herds and try to match each sheep in one herd with a sheep in the other. But one day, along came a shepherd who invented decategorification. She realized one could take each herd and `count' it, setting up an isomorphism between it and some set of `numbers', which were nonsense words like `one, two, three, ...' specially designed for this purpose. By comparing the resulting numbers, she could show that two herds were isomorphic without explicitly establishing an isomorphism! In short, by decategorifying the category of finite sets, the set of natural numbers was invented. According to this parable, decategorification started out as a stroke of mathematical genius. Only later did it become a matter of dumb habit, which we are now struggling to overcome by means of categorification. While the historical reality is far more complicated, categorification really has led to tremendous progress in mathematics during the 20th century. For example, Noether revolutionized algebraic topology by emphasizing the importance of homology groups. Previous work had focused on Betti numbers, which are just the dimensions of the rational homology groups. As with taking the cardinality of a set, taking the dimension of a vector space is a process of decategorification, since two vector spaces are isomorphic if and only if they have the same dimension. Noether noted that if we work with homology groups rather than Betti numbers, we can solve more problems, because we obtain invariants not only of spaces, but also of maps. In modern language, the nth rational homology is a *functor* defined on the *category* of topological spaces, while the nth Betti number is a mere *function*, defined on the *set* of isomorphism classes of topological spaces. Of course, this way of stating Noether's insight is anachronistic, since it came before category theory. Indeed, it was in Eilenberg and Mac Lane's subsequent work on homology that category theory was born! Decategorification is a straightforward process which typically destroys information about the situation at hand. Categorification, being an attempt to recover this lost information, is inevitably fraught with difficulties.
Finally, a note to John: While you're trying to give your audience some feeling for the virtues of n-categories, couldn't you give them a little help with n=1, by being a little more precise about objects and maps?
I hope it's clearer now. Best, jb