A few years ago I sat in a lecture from Thomas Hales on, among other things, his formal abstracts project. This project is aimed at turning paper abstracts and main results into machine-readable forms that could be interepreted by theorem provers, allowing many computational applications to the body of mathematical theory.

In order to get computers to understand the meaning of mathematical statements, it seems like type theory is an indispensable tool. I’m no expert in this stuff, but basically by assigning types to symbols or expressions, the theory ensures that combinations of these objects form coherent, meaningful statements by making sure that mismatched types don’t go together. This stuff has been around for years in theoretical computer science for the construction and study of programming languages, but it’s exciting to think of it becoming part of standard practice in mathematics.

As a fan of OEIS and StackExchange, I am excited about the improvements the formal abstracts project could bring to the experience of searching for math on the web. I’m sure many of us have the experience of typing strings of mathematical notation into search engines and having them return unrelated garbage. But if the parts of these searches were “tagged” as [G: finite group] or [I: two-sided ideal], the search interpreter could use this extra layer of meaning to get better results. Especially because sometimes it seems like adding these tags ad hoc seems to make search results frequently worse. If people have tricks, please help us all.

After all, much of what humans do when reading math is to make sure expressions are build of things of the right type. What business does have mapping stuff if is just a number? A lot of how we choose notation is meant to make this sort of checking easier for us. Certain letters, fonts, and symbols are reserved for certain contexts, and then inevitably evoke them when they appear elsewhere.

I think it’s kind of fun and interesting to catalog this tacit knowledge, so I’m working on the following picture of acceptable/frequently used notation. I respect the freedom of all to alter and reinvent notations, but there are some things that we seem to agree on and I think it’s to our mutual benefit that we do. Someday, the machines will agree too.

I attend a university called Tulane in a city called New Orleans. The pre-colonial name for the place where these things are is Bulbancha. Before Paul Tulane gave money for the land on which the university now sits, the institution was the medical department of the public University of Louisiana. His endowment was motivated by at least two reasons.

Vanity (he had already tried to pay Princeton University to rename that school for him).

Perpetuating racial segregation and inequality.

This latter reason is even explicit in his letter of endowment – that the institution should serve for the education and betterment of young white persons of New Orleans.

He is also supposed to have been the largest private donor to the Confederate States of America, and was a beneficiary of organizations that constructed Confederate monuments in the post-war period.

These facts are often repeated among those that are concerned about the persistence of monuments to bigots and systems of oppression throughout our landscapes, and some of this picture can even be found on Paul Tulane’s Wikipedia page. The primary sources, though, are harder to come by, and one imagines that part of the reason is that this is an issue the modern university does not want to have to address. Indeed, on the university’s history page, the clause which founded its modern incarnation as a white institution is conveniently omitted. In the university’s timeline, there is no mention that it was in fact an exclusively white institution until 1963, when the Board succumbed to the threat of lawsuits and a move by the Ford Foundation to pull grant funding unless the school was integrated.

The university is taking baby-steps towards examining its racist past, recently removing a prominent campus bell that was “discovered” to have been used on a plantation. The bell was donated by a university board member at the time of integration. Its purpose, sources say, was to serve as a menacing reminder of the powers that be, though this history has also been white-washed. And after a resolution passed by the university’s student government calling for the renaming of an on-campus building named for stauch segregationist F. Edward Hebert, the administration refused on the grounds that they didn’t want to upset Hebert’s descendants.

While I think we should all have an opportunity to learn more about the history of Paul Tulane and the institution that bears his name before rushing to judgement and action, what I know already causes me trouble every time I have to say the name “Tulane.” It makes me feel complicit in a celebration of oppression which this man seemed to intend as his legacy. But in my personal and professional life, it is something I am forced to do just to communicate the basic facts of where I am, what I do, who I know, and how I make a living. White supremacy is baked in deep.

I knew already when I decided to attend this university that it was a symbol and bastion of inequity – a predominantly white, very expensive school for the upper-classes in a city which is predominantly black and poorer than most. But I confess I didn’t know anything about the history of the institution — an institution which, for the record, has been quite good to me. Questions of “school spirit” aside, I have learned much, and been afforded a living to pursue what I wished to during my time here. But now, on the way out, and for the good of all those that follow, I think the university needs some truth and reconciliation, and perhaps some renaming.

We must examine what it means for this university to be named for this person. Many people in many places have been and continue to ask such questions, and I am grateful for their initiative and inspiration. Sometimes it has led to action, but it has always led to important discussion.

For those that may find it useful, I made the puzzle from scratch in LaTeX with tikz. There are actually dedicated LaTeX packages for making crosswords, but I was determined to homebrew.

One of our projects in number theory led us to thinking about the class number problem, which has a story too long and interesting to recount here. See the survey of Goldfeld from an old AMS Bulletin to get an overview. Briefly, the question is about the degree to which unique factorization holds in quadratic extensions of the rational numbers. In any case, one of the first important results in the program is an old theorem of G. Rabinowitsch (Rabinowitz), which gives a testable criterion for whether the number field , with a negative integer, possesses unique factorization. In modernish language (cf. Theorem 6 in the linked-to document), keeping this framework we have:

Theorem (Rabinowitsch, 1913): The field is a unique factorization domain (UFD) if and only if is prime for all , where .

The paper was published in German in Crelle’s journal over a century ago, and was somewhat hard to find on its own. We could not locate any other source where the content has been rewritten since then, so we translated and typeset the article (with the aid of various online translation tools). To preserve the spirit and style of the writing, some outdated and perhaps idiosyncratic (to author or translator, as the case may be) jargon has been allowed to survive. These designations are hopefully all made clear within the article so it is readable. Some notation has been modified for clarity.

Perhaps one day this material can be condensed and fully modernised. The article consists mostly of pleasant and clean elementary number theory, situated near the headwaters of one of the important achievements of 20th century number theory. It seems worthy of further propagation, so we post here our translation (no warranty included).

McCleary introduces the concept of a differential graded algebra in section 1.3 (Definition 1.6, p. 11). These are algebras (over a field ), which tend to be -graded, and importantly carry with them a map called a differential which is -linear, shifts the degree of elements (in the grading) up by one, and satisfies a “Leibniz rule:”

for in our algebra . This is a twisted version of what is usually called Leibniz’ rule in calculus (which is basically just product rule), which coincides with how the differential works in the algebra of differential forms.

This idea is easily extended to the notion of a differential bigraded algebra , where now the elements are graded (for the time being, later we’ll have ), but remains a total-degree 1 mapping. That is,

and still satisfies the Leibniz rule

(1)

where .

A standard construction is to form a bigraded algebra by tensoring two graded algebras together. This would work with just component-wise multiplication, but to get a working differential that satisfies our version of the Leibniz rule 1 as well, we introduce an extra sign: we mean, supposing and are differential graded algebras, then we can assign , and furthermore

(2)

Then if we define a differential on by

(3)

then satisfies the Leibniz rule 1. It is clarifying to check this, so we’ll record it here. Switching notation a bit, we will write instead of . To satisfy 1 we need

we then apply 3 to the individual terms on the right side above to get

Now applying the multiplication rule 2 and distributing, we find

(4)

To check the rule holds, we perform this computation by instead multiplying first and then applying the differential. That calculation looks like

Finally, remarking that and shows that terms of the last line above match with those of 4, so everything checks out and becomes a differential bigraded algebra.

A Chain Rule

Given the length and detail of section 1.3, surprisingly we find no glaring errors in this section, but the use of the differential becomes somewhat muddled in calculation in section 1.4. Again, perhaps as an undesirable side effect of the fact that we remain at the “informal stage,” it’s always difficult to keep track of what assumptions we’re working with in each example. Case in point, example 1.H, p. 20. The paragraph preceding definition 1.11 seems to indicate that all graded algebras are assumed to be graded commutative — at least for the rest of the section, one guesses, though the language is vague. Let’s try this here with a bit more force.

Assumption: All graded algebras are graded commutative for the rest of the post. This is to say, for all in any , we have . Now let’s have a look at the example. We suppose a spectral sequence of algebras with , converging to the graded algebra which is in degree 0 and in all others. The example asserts that if is a graded commutative polynomial algebra in one generator/variable, then is a graded commutative exterior algebra in one generator, and vice versa.

The first confusion appears in a restatement of the Leibniz rule near the bottom of page 20, except this time there are tensors involved. This appears to be a mixed use/abuse of notation, which was slightly different in the first edition of the book, but not more consistent. The idea is as follows. and embed into under the maps and . Then one can also write an element (mind the inexplicable inconsistent choice of letters) as

(5)

since the degree of 1 is zero in each graded algebra. Note that this also allows us to regard as graded commutative with the tensor product as multiplication between pure and pure elements, writing

One can apply Leibniz rule to the product in 5 so that if comes with a differential , we get

The thing is we really need not write the tensor product ; it is just as correct to write on it’s own, as we often do with polynomial algebras and so on. Then the above can be written instead as

as McCleary does near the bottom of page 20. What makes this confusing is that up to this point we had only seen differentials acting on tensors by defining the bigraded differential from tensoring two differential graded algebras together, seen above. In this context, the differential of the bigraded algebra must act on an element of the algebra coming from , it cannot act on just one side of the tensor. What’s different here is that the tensor product is actually the multiplication operation on each page of the spectral sequence. Thus, the restatement of the familiar rule with new notation.

Nevertheless, the next equality is also a bit confounding at first, partly because McCleary, goes back to writing the extra in the tensor, suggesting that we need to pay attention to its effect. He says that if , then

(6)

which looks sort of reasonable as it resembles something like a chain rule, . It is presented as if it should follow immediately from the Leibniz rule stated before. But this seems weird when the degree of is odd. To be totally transparent about this, let’s illustrate the case where , suppressing the subscript on the differential again, but maintaining the tensorial notation.

where the last line follows since has total degree , so the sign inside the sum there has exponent which is even. We see that if has odd degree then, these terms cancel and we get 0. So you say “wait a minute, that’s not right, I want my chain rule looking thing” until you eventually realize that if has odd degree, since it’s sitting in a graded commutative algebra, is actually zero! And the same goes for all higher powers of . Then, makes complete sense. Meanwhile, if has even degree, the terms will pile up with positive sign and we get the chain rule looking thing that was claimed. So the statement 6 is in fact true, though it really breaks down into two distinct cases.

Going forward in the example, McCleary only really seems to use the chain rule (liberally mixing in the described sort of abuse of notation) on terms of even degree, so it’s tempting to think that it only applies there, but it is sort of “vacuously true” in odd degree as well. Oh well. Onwards.

The next thing to address in McCleary is an apparent mistake on p. 9 of section 1.2. Here we again assume a first quadrant spectral sequence converging to a graded vector space . This is mentioned at the beginning of the section, but it’s easy to forget that when a bold-faced and titled example (1.D) seems to be presenting a reset of assumptions, rather than building upon prior discussion. Furthermore, in this example, McCleary seems to be working again with the assumption from example 1.A that for . On the other hand, this can be seen as a consequence of the fact that our spectral sequence is limited to the first quadrant, provided the filtration is finite in the sense of Weibel’s Homological Algebra, p. 123 ( for some ). But then it would be unclear why McCleary took this as an additional assumption rather than as a consequence of prior assumptions in the first case. : /

The new part of this example is the assumption that unless or , so all terms of the spectral sequence are to be found just in two horizontal stripes. In particular is only possibly non-zero in these stripes, and since these correspond to filtration quotients, the filtration takes a special form.

First, we might look at the filtration on where . Note that the spectral sequence terms that give information about are those along the diagonal line where . Since , the only place where anything interesting might happen is when this line crosses the -axis, i. e. when . This forces , so the only possible nonzero filtration quotient is

working with the assumption that . So on the one hand, we get no interesting filtration of for , but on the other hand we can see exactly what it is from the spectral sequence limit.

Now we treat the case of , where . I find this awkward notation again, preferring to reserve for a pure arbitrary spectral sequence index, but since we are trying to address the mistake in this notation, we should keep it for now. The filtration of this vector space/cohomology is interesting when and , where the quotients are given by

Every where else, successive quotients are 0, meaning the filtration looks like…

In the filtration on page 9, McCleary puts one of the (possibly) non-trivial quotients at instead of at where it should be. That’s all I’m saying.

This situation is modeled on a spectral sequence for sphere bundles i.e. bundles where the fibers are spheres of a given dimension. The stripes coincide with the fact that a sphere has nontrivial cohomology only at and . This sort of computation is famous enough that it has a name: the Thom-Gysin sequence (or just Gysin sequence).

As a final remark on section 1.2, McCleary says that the sequence in example 1.C is the Gysin sequence. Example 1.C doesn’t exist, we mean example 1.D : )

A goose chase through homological algebra etc. has led us to start reading McCleary’s A User’s Guide to Spectral Sequences. The book seems like a nice introduction for those that know their way around graduate topology and geometry, but haven’t yet encountered cause to pull out this extra machinery to compute (co)homology. First published by Spivak’s Publish or Perish Press in the 80’s, a second edition was released by Cambridge University Press in 2000. Though it sounds like the second edition is rather improved, there seem to be a number of mistakes remaining which may frustrate those trying to learn a notoriously complicated subject for the first time. Pending an official list of errata, we may as well collect some of them here.

Section 1.1 – Notation

The first comment worth making is regarding some confusing notation, largely an overuse of the letter . The first use comes on p. 4 (Section 1.1), given a graded vector space and a filtration of , by defining

Here, the symbol seems to designate an endofunctor on (graded) vector spaces — it eats one and gives back another; transporting morphisms through the filtration and quotient shouldn’t be a problem either. It isn’t really clear what the subscript is supposed to indicate at this point, but the reader sits tight expecting the truth to be revealed.

However, on the very same page, McCleary twists things by making the assignment

(1)

where , the graded piece of the filtration. Now, with the extra index , is a vector space on it’s own. The notation doesn’t indicate reference to , though in this case it really depends on . For instance, McCleary indicates that we should write something like

The definition immediately afterwards (Definition 1.1) indicates is to be used to designate a vector space in a spectral sequence which is irrespective of any for all . The typical way to relate a spectral sequence , to a graded vector space is the situation of convergence (Definition 1.2, p. 5) where instead

The right hand side above has nothing to do with the spectral sequence (since we take in our definition), it is just an instance of the definition from equation 1… but with distinct use of notation… oh. So on the one hand, should be a standalone vector space, like the other ‘s, but also it needs to come from an so one should really write as in Definition 1.2. Wha? Shoot. Couldn’t we have used like an instead or something?

Perhaps there is good reasoning for all of this to be discovered once we get further in. Also, it seems so far that initial terms are usually . Why not ? And why don’t we allow -pages? In these cases the differentials would be vertical and horizontal (resp.) instead of diagonal, which feels less interesting somehow, though this doesn’t seem like it would be totally frivolous… TBD.

Splicing Short Exact Sequences

Finishing out the first section, we address what seems to be a typo in example 1.A (p. 6). McCleary’s expository style consists of many statements which are not obvious, though usually not difficult to work out. This is perhaps for the best, as the community seems to indicate that the only real way to learn spectral sequences (make that: all math?) is by working them out. Nevertheless, it is a bit discouraging to find yourself at odds with the author at the first example…

We have assumed a first quadrant spectral sequence with initial term converging to with a filtration satisfying for all . Then we have a filtration on in particular, given by

since, by the assumption, etc., and by definition. By convergence, then,

so is a submodule of . But also because lies on the -axis (depicted as what is usually the -axis) and our spectral sequence has only first quadrant terms, must be the zero map for all . Furthermore, is too close to the -axis to get hit by any differential , thus survives as the kernel of every , mod the image of a from a zero vector space in the second quadrant. This is all to say

We then have part of the short exact sequence McCleary gives for as

How can we describe the third term using the spectral sequence? Well, from our definitions, . The book seems to be indicating that but this is not necessarily the case! It also doesn’t make sense with how the short exact sequences are spliced later on.

Let’s address the first claim first. Because lies on the -axis, and the differentials point “southeast” towards the empty fourth quadrant, is the zero map for any , but it can’t be hit by anything so we have now

The denominator is the image of a map from a zero vector space, so it is zero, and thus is a subspace of , but this latter space can be larger! This is all to say, the short exact sequence for is misprinted, and should go

(2)

One can confirm this by examining the SES given just below, where we see injecting into :

(3)

This is a standard decomposition of the map in the middle: for any morphism (in an abelian category at least, we suppose) there is a SES

It remains to see that . Because of where sits on the -axis, it is again the kernel of for all . Further, it can only possibly be hit by , so in fact survives through all further terms to give the desired equality

To splice all this together, we recall that we can connect

as

where . We maintain exactness since and .

Performing this surgery on sequences 2 and 3 yields the main exact sequence claimed by the example, namely

(4)

Stay tuned for more clarifications from Chapter 1.

At this point in the project, we begin to understand why so many books and papers are written in dry, terse, often difficult mathematical prose. Discursiveness which is beneficial is risky, time consuming and hard to pull off! As we continue to hurtle preliminary posts, maybe a bit of motivation is in order.

At the highest level, far-out, hardcore algebraic geometry is finding application in strange places these days. We will just cite the three instances which are close to the heart of mathlab: Hodge Theory in Combinatorics, Geometric Complexity Theory, and the Macdonald Positivity Conjecture where Hilbert schemes played a starring role. Understanding this last one is what we are heading for in these posts. Moving on…

So let’s just do this the way where you put a loaded definition out front and then try to tease it apart, following Nakajima‘s exposition, repeated in seminar.

Definition Let be a projective scheme over an algebraically closed field . The contravariant Hilbert functor is the functor that associates to a scheme the set of closed subschemes such that the natural morphism from to is flat.

From the top, we are choosing an algebraically closed field so the “projective” here is with respect to , as described in the last post. We would like to take this opportunity to express our regret for not really describing or the fiber product in these posts even though we earlier gestured towards thoroughness. Hopefully the references serve.

Moving on, an arbitrary scheme is said to be projective (over ) if it is of the form for a finitely generated graded -algebra. A subset of a scheme is said to be a closed subscheme iff:

The inclusion map is an affine morphism, meaning that for each open affine subset , the inverse image for some ring . “The inverse image of affines is affine.”

Additionally, for each of these pairs, the map is surjective.

\underline{Note}: In case is not a subset of but we still have a map which satisfies the above properties, that map is said to be a closed embedding (or, less tastefully, a closed immersion) of schemes.

Back in the definition, the map from to is the composition of the inclusion and the projection map . Remember, this isn’t set-wise projection, our product is in the category of schemes, where we can also think of it as the fibered product over the terminal object, . It is guaranteed to exist as in section 9.1 of The Rising Sea.

Finally we come to flatness, which is sort of exciting because this is the first time I’ve come across it in the wild since studying for my algebra qual, but still sort of mysterious. Apparently Mumford agrees, even as he reassures us that it is “technically the answer to many prayers,” in The Red Book (III.10). Remember that an module is said to be flat if tensoring with preserves short exact sequences. More loosely, flat modules are pretty close to (i.e. are direct limits of) free modules.

That the map is flat means that for all , the induced map on the stalks makes into a flat -module. For motivation on why flatness is the “right” condition in these sorts of circumstances, see [EH] II.3.4.

An important theorem of Grothendieck says that with as above is representable by a scheme which is “locally of finite type” called a *gasp* HILBERT SCHEME! A version of the proof with additional discussion and background is here. It is also explained in section 9 there that we can weaken the requirements so that need only be quasi-projective. In fact, one of the most interesting cases for us will be where is ( in particular). Further on, we will refine this by introducing Hilbert polynomials. Yay!

Having established that to each scheme we can associate its functor of points , we continue our introduction by mentioning some eminent examples.

First the projective space . This is sort of the basic example of a moduli space, which is also what were after with Hilbert Schemes. While (over your ring of choice) has a few different constructions, it’s points correspond to other geometric objects: lines in the affine space . This is the essence of moduli, and their value for “counting” objects of a given type.

We also have a description of the functor of points for projective space, , recalling that for a given scheme . See Vakil, Theorem 16.4.1, or also Eisenbud and Harris [EH], Theorem III-37

Theorem Morphisms are in bijection with the data (up to isomorphism) of where is a line bundle (invertible sheaf) on and the are global sections which do not vanish simultaneously.

By “up to isomorphism” here we mean that the value of the section at a given point on is depends on the trivialization used, but it will differ according to multiplication by a scalar which multiplies all simultaneously. Thus, the ensemble of all sections together effectively gives a point in (or a fiber of the tautological bundle, if you prefer), provided they are not all zero. Ranging over all of we get a morphism of schemes.

You might wonder which projective space we are working with here, i.e. over which field/ring. As will usually be the case, it doesn’t much matter because maps behave nicely under base change, but if we want to be as general as possible we should probably use . This is because is the initial object in the category , which makes it the final object in the category of affine schemes. Initialness comes from the fact that there is a unique morphism from to any other commutative ring with identity which just takes 1 to itself.

Then, is also final in because we have this ring morphism (which is an morphism in reverse) into all of the open affine covering neighborhoods of a given scheme, and these glue up nicely. So, whether we construct from a graded ring as or as the gluing of a bunch of , we again have a canonical scheme morphism . This together with lets us construct the fiber product ; see [EH] III.2.5 for a more thorough treatment of all of this. I’m still essentially just recording the things I had to remind myself of in jumping into this so it’s maybe more garbled than pedagogical.

[EH] here also gives the description of as subsheafs of which are locally of rank , which we think of as the kernels of the surjection from the direct sum sheaf , surjection being guaranteed by the non-vanishing of at least one of the sections above (at every point). This leads naturally into our other key example, the Grassmannian.

According to the Yoneda philosophy (which is often attributed to Grothendieck as well) we should actually think of the Grassmannian (and the special case of projective space) not just as the “space of -planes in -space,” but as a moduli space for certain bundles (locally free sheaves in AG-speak) and their sections. Really, it is a functor which can be applied to other schemes, and returns the collection of morphism which define the scheme that represents the functor. : S

The change with the Grassmann functor is that now we look at surjections from where is locally free of rank . At neighborhoods in a locally trivializing cover, we identify with and we can treat as a matrix. The Grassmannian will be covered by open sets which are the nonvanishings of the determinant of the maximal minors. This leaves the other entries of the matrix free, so they can vary and together determine a map in the way the sections gave a map to before. Thus, we can cover by a bunch of affine spaces which glue together. In the parlance, these are open subfunctors represented by the schemes .

It’s worth noting that the full Grassman functor is represented by a bona fide scheme of the form where denotes a subset of of size , which indexes the variables, an is a particular homogeneous ideal given by the Plücker equations. So we have a more traditional geometric object to think of as well. We may come back to this in more detail later if we need to. For now it suffices to say that applying the exterior power functor to gives a surjection from a locally free sheaf of rank onto a line bundle. So by the prior discussion, we have a map from the Grassmannian into which is the Plücker embedding.

Phew, okay I think in the next post we might actually define a Hilbert Scheme.

Next, we introduce the Hilbert scheme functorially following notes by Maclagan. See also Bertram’s course notes, who points out that we (as in Grothendieck) are trying to develop a category-theoretic framework for counting objects with a given property. First we should say a little bit about morphisms of schemes.

To do this, one should indicate that schemes (like manifolds) actually carry the additional structure of a locally ringed space. This means that when you take a kind of (filtered co-)limit on the structure sheaf over all the open neighborhoods containing some point , then we get a local ring . In this case of an affine scheme it works out that is actually the localization , as is a prime ideal of , and localization satisfies exactly the universal property prescribed by the colimit.

First, remember that schemes are ringed spaces, and that a morphism of ringed spaces is a continuous map of topological spaces together with a morphism of sheaves from to the push-forward . This induces a map on stalks

where and supposing sends the maximal ideal of one to the other, then we have a morphism of locally ringed spaces, and it is THIS, my friends, which will be a scheme morphism when the locally ringed spaces happen to be schemes… oh hell, this can’t be useful for anyone as exposition; just seriously go skim the first six chapters of The Rising Sea to clear things up. It’s only 200 pages.

Now we have objects and morphisms so we can talk about the category and functors therefrom. We are interested in the confusingly named functor of points for a given scheme , which is really the one that comes up in many versions of the Yoneda Lemma. The idea is that if category is locally small, meaning the collection of morphisms (arrows) between any two objects forms a set, then we can understand an object by the morphisms going into (or out of) it. Bearing in mind that a contravariant functor between categories is the same as a covariant functor , we have the following.

Definition Suppose is an object of , the category of schemes. Then there is a contravariant functor such that on objects we have , and for a given morphism we have

In general, functors that come from such ‘s are said to be representable. Elements of the set are called the Y-valued points of X, which is weird, because on sight they are just maps. Further, by laziness, in case for some ring (or field) A, then we will call theses maps the “-valued points of X.”

Endnote: for an interesting discussion on whether scheme theory can be taught using just , the maximal ideals of a ring, instead of the primes, that then digresses into thinking of schemes as their functors of points instead of the whole locally ringed space shebang, as well as other AG/Berkeley cultural inside jokes by a bunch of dudes, see the post and extensive comments here. Other interesting background can be found on nLab.