The Stacks project

Generalize section on infinitesimal deformations

We want to use the definitions even when the base scheme is not
necessarily Noetherian.

We also removed the requirement in (RS*) that the affine schemes in
question map into an affine open of the base scheme. This is never
needed when one actually verifies a given stack has (RS*) and in
fact this more general notion is always a consequence of the
original restricted one.

Excise finite type conditions

We need to have the functors T_x and Inf_x for all modules not just
for finite type ones. This actually simplifies the discussion a
tiny bit as we don't have to carry along the finite type conditions
everywhere.

Thanks to David Rydh for pointing this out (see
commit 2ccbbe3087e4dc2b1df2193c81ede7486931424c for more
information).

Fix error in artin.tex pointed out by David Rydh

The proof of Lemma Tag 07YM was wrong in two ways:

-- We were using the condition (RS*) in some cases where it didn't
   apply, namely for an extension A' of a Noetherian ring A by a
   module isomorphic to the residue field of a finite type point.
-- The second mistake was in some sense the same mistake thinking that
   the ring A' was Noetherian.

The solution, as pointed out by David Rydh, is twofold:

-- Strengthen assumption (RS*) to cover the first snafu
-- Only prove openess of versality at closed points

As far as I can see, this is harmless in all applications.

There are still some things left to fix. In particular, some of
the material concerning T_x(-) has to be fixed to deal with
non-finitely generated modules.

Tags: Added new tags

Some material about obstruction theories

	This includes the notion of a naive obstruction theory and a
	proof that such a thing is good enough to give openness of
	versality. This was introduced on the blog

	http://math.columbia.edu/~dejong/wordpress/?p=2472

	It is just working out what you would get if you actually had an
	algebraic stack. It seems likely that Artin was motivated in
	formulating his axioms by the idea that such a thing should
	exist.

	Hopefully we can use this to prove openness of versality for
	some nice cases such as families of proper flat algebraic
	spaces... We'll see.

Infinitesimal defos + auts + obs

	Just an introduction.