Let be a ring homomorphism. We say that is a finitely-generated as an -algebra if there exists a surjection
We say that is finitely-presented if it is finitely-generated and we can choose to be a finitely-generated ideal.
Remark
If there exists a such that is finitely-generated, then for all other , is finitely-generated due to the lemmalemma below. It is the analogue of Schanuel’s lemmaSchanuel’s lemma for algebras.
Example
has a non-finitely-generated kernel.
Lemma
Let be a ring and be a finitely-presented -algebra. If is any surjection of -algebras, then is a finitely-generated ideal.
Proof
Let be a presentation, where is a finitely-generated ideal. Form the following diagram with exact rows:
Because is surjective, we can find a lift of to
which gives us the middle vertical map. By the Snake Lemma, we get a surjection
Since is a finitely-generated -module, it is certainly a finitely-generated -module via the horizontal map
Therefore is a finitely-generated -module. From vertical exact sequence on the left, we get a short exact sequence
and so is a finitely-generated -module by 3.3. of this propositionthis proposition; note here that is a -module, and becomes a -module via the vertical map
But because this map is surjective, is a finitely-generated ideal of .
Tensor product of algebras
Defining the algebra structure
Consider -algebras and in the following diagram:
Viewing and as just -modules, we can take their tensor product , and there are -linear maps
Since the ring multiplication in and are -bilinear, we get two induced maps out of the tensor products
Tensoring these two maps together and reshuffling the tensor factors,
we get an -bilinear multiplication on . This makes into an -algebra. It satisfies the following universal property: for all -algebra maps , there exists a unique -algebra map such that the following diagram commutes
This is the coproduct in the category of -algebras.
Remark
Compare this to the coproduct in the category of sets:
Note that the nature of the coproduct for -algebras is quite different.
vs
Taking , we get
Recall that we can take the fibre product as sets. In this case, we get a unique map which fits in the following diagram:
However, and the fibre product are quite different.
Examples
Consider and . As sets, the following are all isomorphic (i.e., equinumerous)
since and are both just one-point spaces. But if we instead consider ,
we have
by the Chinese Remainder Theorem. Thus .
The prime ideals of are for prime.
. Where and , we have
However, for example, but does not come from either or .
Base change (for algebras)
Let and be -algebras. Then is a -algebra and every -algebra map , where is a -algebra, admits a unique factorisation through , i.e.,
Chain conditions
Definition (Noetherian module)
Let be a ring. Let be an -module. We say that is Noetherian if every ascending chain of -submodules
eventually stabilises.
Proposition (Noetherian things)
Let be a ring and be an -module.
1.
Consider an exact sequence of -modules
Then:
If is Noetherian, then so are and .
If and are Noetherian then is.
2.
is Noetherian if and only if every -submodule is finitely-generated.
3.
If is Noetherian over itself, then the following are equivalent: