Hartshorne II.2
Posted on June 16, 20192.1. Let be the natural homomorphism given by
Observe that
induces a bijection between the ideals of
not containing
and the ideals of
via the identification
with inverse
This identification also identifies prime ideals with prime ideals and therefore defines a bijection
. Moreover, by our observations above,
and
each send closed sets to closed sets. Hence
is a homeomorphism.
It remains to show that we have a morphism Observe that for
the set
is equal to
and
Note
if
The basic open sets of
are precisely the sets
for
and
when
Thus
identifies the basic open sets of
with the basic open sets of
Moreover,
and these isomorphisms are uniquely determined from the natural isomorphism
These isomorphisms define
and thus they glue to define
for all open sets. Since each is an isomorphism,
is clearly a local homomorphism of local rings for each
2.2. Let be an affine open cover for
with
Observe that
is an open subset of
and hence
such that for each
we have
with
From proposition 2.2 we know
Therefore
and
is an affine open cover of
Therefore
is a scheme.
Lemma 1. Suppose that is a topological space,
is a presheaf on
,
is a basis for the topology on
, and
is a
-sheaf. Then
is isomorphic to the sheaf generated by
as a
-sheaf. See Eisenbud proposition I-12 and exercise I-13 for information on
-sheaves.
It suffices to observe that is an isomorphism for each set
Clearly the map
is surjective for each
by definition and it is given by mapping a section
to the function
given by
(see proposition 1.2).
It remains to show that is injective for each
Suppose that
are such that
Then
for all
Thus for each
there exists an open set
containing
such that
Since
is a basis, we may pic a basis set
containing
Therefore for each
there exists a basic open set
containing
such that
Therefore
since
is a
-sheaf.
Corollary 2. Suppose that is a topological space,
is a sheaf on
and
is a presheaf on
that forms a
-sheaf on some basis
of
If
is a map of
-sheaves (i.e. a collection of maps
for each
compatible with restrictions) such that
is an isomorphism for each
then there exists a induced map
that is an isomorphism of sheaves.
Recall that it suffices to define a map of sheaves on a basis. In our case we have a basis and a collection of compatible isomorphisms
for each
and therefore define an isomorphism of sheaves
2.3.a. Suppose that is open and
If
then
for each
Conversely, if
for some
then there exists an open neighborhood
such that
Therefore a nilpotent section over an open subset exists if and only if a nilpotent germ exists at a stalk.
2.3.b. Let us first show that for any ring we have
Let
be the nilradical of
and recall that
is the intersection of all the prime ideals of
Thus
is the unique prime ideal contained in all other prime ideals and if
then
Is not prime. So natural map
gives a bijection between the prime ideals of
By proposition 2.3c we have a map
given by
and
given by localizing
We also know that
Thus
is a homeomorphism by our observations about the relationship between the ideals of
and
For each
we have
and
Thus
Now let be the presheaf given by
The restriction maps
, that is
for
is the map induced by the map
Thus
forms a
-sheaf on the basis
of basic open sets, and in fact it is the same
-sheaf given by
above. Thus by corollary
above we have that
is isomorphic to
Suppose now that is an arbitrary scheme and
is an affine open cover of
with
By our work above we see that
for each
The sheaves
agree on basic affine open sets and thus on intersections. Therefore they glue to a sheaf
on
. Moreover, for each
we see that
is clearly equal to the sheaf
given in the problem statement. Thus
The maps of affine schemes
also agree on intersections and hence glue to a map
The map
is a homeomorphism as it is a homeomorphism on each
2.3.c. We define the morphism by
, where
is the morphism from part (b). Note that
is a homeomorphism by part (b), so
makes sense as function and is a continuous map of topological spaces.
To simplify notation let us identify the topological spaces
and
so the morphism
gives a map
We also have the map
given by the morphism
Thus to complete our construction of a morphism
with the desired properties it suffices to show that there exists a pullback map
such that
Observe that for open we have pullback morphism
.
Moreover, since
is reduced,
is reduced. So the kernel of
must contain the nilradical of
. Thus it contains the kernel of
. So there exists a unique homomorphism
such that
The diagram also commutes with restriction
for
and therefore defines a morphism
such that
With our identification of
and
we have that
Thus our construction above defines a morphism
that gives a local homomorphism of local rings and thus
define a morphism of schemes.
2.4. Let us first show the result for affine schemes (this is actually already done in Hartshorne in proposition 2.3c).
Suppose that is a ring homomorphism. Then we may define a map
by
Observe that
so
is continuous. Observe that for
we have
Moreover,
We therefore define
to be the map induced by
These glue to define a morphism
with
In order to show that defines a morphism of schemes it remains to show that
is a local homomorphism of local rings for each prime ideal
Observe that
and
is given by
Thus
So
is a local homomorphism of local rings for each
and
is a morphism of schemes.
Suppose conversely that is a morphism of schemes. The first part of our work above shows that the homomorphism
determines the morphism
Moreover, since
is a morphism of locally ringed spaces we must have
But the map
is just the localization of the global map
at
Therefore
Thus the map
is also determined by the homomorphism
Hence there is a bijection
Now suppose that is an arbitrary scheme and
is a homomorphism. Let
be an affine open cover of
The map
composed with the restriction map
is a ring homomorphism
Therefore
determines a unique morphism of affine schemes
Now let
be distinct indexes. We observe that
because for each affine open set
we have
,
, and we know that since
is affine these ring homomorphisms uniquely determine
and
Thus the
glue to a morphism of schemes
corresponding to the homomorphism
The argument above also shows that any morphism is determined by
since each morphism of affine schemes
is determined by
Thus there is a bijection