Motivic stuff

Cohomology, homotopy theory, and arithmetic geometry

Archive for the ‘Cohomology breadcrumb trail’ Category

The Chow ring

Posted by Andreas Holmstrom on March 24, 2009

Matt DeLand recently posted a very nice introduction to the Chow ring and Chern classes on Rigorous Trivialities. We will need this material when talking more about Weil cohomology and pure motives, and I will simply refer to his post, and to this introduction by Gillet, for all things related to Chow groups/rings.

Posted in Cohomology breadcrumb trail | Tagged: , , , | Leave a Comment »

Weil cohomology

Posted by Andreas Holmstrom on March 18, 2009

The most useful and natural notion of cohomology for smooth projective varieties is given by the set of axioms known as Weil cohomology. When studying more general varieties (see this post), we need different notions of cohomology – will come back to this in the future. For the basic definitions and properties of Weil cohomology, I will simply refer to this excellent short note by de Jong. There are a few slightly different ways to define Weil cohomology, but I don’t want to get into a lot of details of this at the moment. The most significant difference between de Jong’s note and some other references is that some authors omit the notion of Tate twist. However, from the point of number theory this is  a bad thing to do.

Looking at varieties over a field k say, we have a range of Weil cohomology theories, each with its own coefficient field. The term “coefficient field” simply means the field K over which our cohomology groups are vector spaces. We list here the most well-known Weil cohomology theories, for various fields k.

For every prime number \ell different from the characteristic of k, we have the \ell-adic cohomology, sometimes referred to as \ell-adic étale cohomology, or just étale cohomology. The coefficient field for \ell-adic cohomology is the field \mathbf{Q}_{\ell} of \ell-adic numbers, so the cohomology groups are vector spaces over this field.

If char(k) = 0, we have algebraic de Rham cohomology, with coefficient field k itself.

If \sigma: k \to \mathbf{C} is an embedding of k into the field of complex numbers, we have the so called Betti cohomology associated to \sigma, which is just the singular cohomology of the variety viewed as a complex variety by means of the embedding \sigma. The singular cohomology here is taken with rational coefficients, so the coefficient field of Betti cohomology is the field \mathbf{Q} of rational numbers.

If k is the field \mathbf{Q}_{p} of p-adic numbers for some prime number p, we have p-adic étale cohomology. (This can also be defined for more general fields similar to \mathbf{Q}_{p} – more about this in a future post.)

If k is a perfect field of characteristic $p > 0$, we have crystalline cohomology. The coefficient field in this case is the fraction field of the ring of Witt vectors of k.

All of these cohomologies come with rich extra structure (in addition to being K-vector spaces), such as for example Galois action or Hodge structure. In coming posts we will try to look at each of these cohomology theories in some detail, with their extra structure, and also define all of the terms left undefined above. We will also look at Grothendieck’s idea of pure motives as a “universal Weil cohomology theory”.

Posted in Cohomology breadcrumb trail | Tagged: , , , , , , | 6 Comments »


Posted by Andreas Holmstrom on March 5, 2009

The most basic class of geometric object encountered in algebraic geometry is of course varieties. Before talking about cohomology of varieties, it seems sensible to say a few words about different types of varieties. I will assume that you know what a variety is – if not, look at Hartshorne or any other introductory book on algebraic geometry, or online notes of Dolgachev, MilneDebarre, VakilGathmann and other people.

A cohomology theory for varieties will typically be a functor from some category Var of varieties to the category of abelian groups or vector spaces. When reading about cohomology for some class of varieties, there are three key questions to ask about the category of varieties considered. 

Question 1: Are the varieties required to be complete/proper/projective? Although these words don’t mean exactly the same thing, they are morally and for most practical purposes the same. 

Question 2: Are the varieties required to be smooth/nonsingular?

Question 3: What is the base field?

The possible answers to Q1 and Q2 give us four possible classes of varieties: Smooth proper varieties, smooth varieties not necessarily proper, proper varieties not necessarily smooth, and general varieties. The third of these seems to be less common, so excluding it leaves us with the three most important classes of varieties, in increasing complexity:

  • A: Smooth proper varieties (really nice and well-behaved)
  • B: Smooth varieties (a bit more complicated, but still nice)
  • C: Arbitrary varieties (nasty things, very hard to understand)

When seeing a category of varieties being introduced, it is often useful to make an internal note of which of the three situations we are in. For example, someone talking about “quasiprojective nonsingular varieties” would be in class B, someone talking about “smooth projective varieties” would be in class A, and someone talking about “integral separated schemes of finite type over the base field” is in class C. In most texts, the author states in the very beginning what he means by “variety”, and it is often one of the first two. We will see later that the right notion of cohomology depends on which situation we are in.

Question 3 also has a big impact on the study of cohomology theories for the varieties in question. Different cohomology theories are defined for different base fields. The most common base fields are: Finite fields, global fields and local fields, algebraic closures of these fields, and the fields \mathbf{R} and \mathbf{C}.

We will soon start looking at cohomology theories for smooth projective varieties, i.e. Weil cohomology theories.

Posted in Cohomology breadcrumb trail | Tagged: , , | 2 Comments »

Cohomology breadcrumb trail, the beginning

Posted by Andreas Holmstrom on February 25, 2009

On of the main reasons for setting up this blog is that I would like to write a reasonably coherent set of notes, giving an overview of cohomology theories in algebraic geometry. Actually “overview” might not be the right word, I am rather thinking of a “fil d’Ariane”, or “breadcrumb trail”, which would allow a serious student to obtain some kind of overview if she so wished. The notes I would like to write is the kind of thing I wish someone had given me when I started my graduate studies. At that time, I tried to think about various problems in number theory, but found that I always ran into trouble with various kinds of cohomology, and I could not make any sense or see much pattern in them. I asked five different mathematicians what a cohomology theory actually is, in algebraic geometry, and I got five different answers. When I a few months into my graduate studies listened to a talk by Guido Kings, and he used “rigid syntomic cohomology” as if nothing could have been more basic or natural, I decided I would start writing down notes and collecting facts and references with the aim of one day in the distant future becoming fluent in cohomological language. That day is still rather far away, but at least I hope I have come to the point where writing down a set of rough notes would help my own thought processes. So this is what I will try to do, and if anyone else gets any benefit from this, that would be an extra bonus. I will use the tag “Cohomology breadcrumb trail” for posts which belong to these notes.

One of the things that makes algebraic geometry difficult and interesting, is that there are lots of different kinds of geometric objects. Examples include various classes of varieties, various kinds of more general schemes (for example over arithmetic rings), different kinds of stacks, algebraic spaces, motives, and simplicial sheaves. There are also notions such as log geometry, rigid geometry, derived algebraic geometry, and various forms of noncommutative geometry. For each of these types of geometry, there is a a number of different cohomology theories which can be used to define invariants of the geometric objects.  

The multitude of cohomology theories is frequently a source of confusion. To mention just one single example, people often talk about the “universal cohomology theory”. However, “universal” can mean different things, and depending on what you mean, the universal cohomology theory can be Grothendieck’s Chow motives, Voevodsky’s motivic cohomology, or the algebraic cobordism theory of Levine and Morel. I hope to be able to clarify this and many other similar things, and to give a short introduction to all kinds of cohomology in algebraic geometry. This might of course be too ambitious a goal, but there is no harm in trying…

Posted in Cohomology breadcrumb trail | Tagged: , , | 3 Comments »