6
$\begingroup$

There is a notion of relative cohomology in both algebraic topology and algebraic geometry and the flexibility granted by this relative viewpoint is quite powerful in both cases. However, the two notions of relativeness are different:

In algebraic topology, we consider cohomology (or homology or homotopy) of the form $H^k(X,A;R)$ where $A \subset X$ is a pair of spaces and $R$ is a ring. This roughly corresponds to the cohomology of $X/A$ where we identify $A$ with a point.

On the other hand, in algebraic geometry, we consider relative cohomology for a pair of schemes of the form $\pi: X\to A$ and the "relative" cohomology corresponds to gluing together the cohomology of the fibers of $\pi$ into one object.

Why is there this different between the two notions of relative? Are they equivalent (maybe formally or even just in the way they are applied) and if not, are there attempts to introduce the other form of relative into algebraic topology/geometry?

Primarily, I am thinking of introducing the algebraic geometry defn of relative cohomology into algebraic topology because the notion of $X/A$ is not so well defined for schemes $A \subset X$.

More generally, in algebraic topology we seem to be interested in the comma category with an initial object while in algebraic geometry, we are interested in the comma category with a terminal object. Why this difference?

$\endgroup$
5
  • 12
    $\begingroup$ I think in both fields people are interested in both versions. In algebraic geometry people also consider cohomology of pairs, and in algebraic topology fibrations $X \to A$ play an important role. The two versions of "relative" that you describe are not equivalent; that is just an unfortunate coincidence in terminology. $\endgroup$
    – jmc
    Commented Jul 2, 2018 at 11:09
  • 1
    $\begingroup$ What is the definition of a cohomology of a pair in algebraic geometry? Very preliminary searching turns up this link to another question: mathoverflow.net/questions/168551/… but without any answer. $\endgroup$
    – Asvin
    Commented Jul 2, 2018 at 11:29
  • 7
    $\begingroup$ I agree with jmc. The map from language to mathematics is many to one, so I wouldn't read too much into different uses of the same word. In any case, algebraic geometers think cohomology with support and topologists think about fibrations, and these correspond to "relative" on the other side. $\endgroup$ Commented Jul 2, 2018 at 11:54
  • 2
    $\begingroup$ One thing is that points have interesting automorphism groups in algebraic geometry but not in topology, this might explain the initial vs terminal issue. $\endgroup$ Commented Jul 2, 2018 at 14:44
  • $\begingroup$ OT: also "regular", "normal" are terms abused in maths... $\endgroup$ Commented Jul 5, 2018 at 5:53

0

You must log in to answer this question.