☆ A number of solutions are adaped from Howard Nuer's own solutions to his Technion 2019 Algebraic Geometry course. I will later go through and accredit the exercises where his approach is used.
Section 7.1: An example of a reasonable class of morphisms: Open embeddings
Exercise 7.1.A:
Verify that the class of open embeddings satisfies properties (i) and (ii) of §7.1.1.
Proof:
To see that the class of open embeddings is local on the target, first suppose that is an open embedding and is open. If then the claim is trivial. Otherwise, is nonempty and open in . By definition factors as , so that the restriction map factors as . Since is open, the restriction map is therefore an open embedding.
Simialrly, assume is a morphism, is an open cover of , and for each we have is an open immersion. Since the cover , form an open cover of . Moreover, each factor through some isomorphism (where open subset since need not be a surjection). By our usual sheaf axioms, the glue to and the glue to . However, since these glue to an open subscheme of so that is an open immersion.
To see that property (ii) holds, let and be open embeddings. Then factors through an isomorphism to an open subset . By (i), the restriction is also an open embedding and thus factors through an isomorphism ( where open subset). Since the composition of isomorphisms is an isomorphism, our map factors through the iso and is thus an open embedding.
Exercise 7.1.B:
Verify that the class of open embeddings satisfies property (iii) of §7.1.1. More specifically: suppose is an open embedding and is any morphism. Show that exists and is an open embedding. (Hint: I’ll even tell you what is: . In particular, if and are open embeddings then .
Proof:
Without loss of generality, we may assume . Following the hint, we wish to show that (\rho^{-1}(U), \OO_Y\vert_{\rho^{-1}(U)} ) indeed satisfies the universal property of the fiber product. Note that we must first come up with maps and that make the cartesian square commute in order for the diagram to make sense. Indeed the obvious map is simply , so that the top arrow in
Must simply be the inclusion . To see that the universal property is satisfied, let be another scheme along with morphisms and such that . Since we must also clearly have so that . But then the unique map must simply be itself. Therefore the fiber product exists.
The inclusion is clearly an open embedding, as is open by virtue of being continuous.
Exercise 7.1.C:
Suppose is an open embedding. Show that if is locally Noetherian, then is too. Show that if is Noetherian, then is too. However, show that if is quasicompact, need not be. (Hint: let be affine but not Noetherian, see Exercise 3.6.G(b).)
Proof:
Since an open embedding, it factors through some isomorphism . First suppose is locally Noetherian such that can be covered by where each is Noetherian. Then clearly can be covered by such sets as well, say — notice Noetherian implies Noetherian by §3.6.16. Since is an isomorphis, we may pull such a cover back to and be guaranteed that our rings are still Noetherian (since an isomorphism of schemes induces an isomorphism of sheaves). The proof showing Noetherian implies Noetherian is identical.
To see that quasicompact does not imply quasicompact, we may use the example of Exercise 3.6.G(b) almost verbatim. If we consider along with the maximal ideal , then is clearly quasicompact as it is affine. However, we showed in the linked exercise that the compliment of was not quasicompact, so we simply let denote the inclusion of into .
Exercise 7.1.D:
Show that the notion of "open embedding" is not local on the source.
Proof:
Consider the double cover given by the diagonal morphism , . Take the obvious affine cover of . Then the restriction maps and are clearly isomorphisms, so that restricted to either of its components is obviously an isomorphism (and thus an open embedding as is open in itself). However, these maps do not glue to an isomorphism as the diagonal embedding is surely not surjecive.
Algebraic interlude: Lying Over and Nakayama
Exercise 7.2.A:
Show that if is a ring morphism, in , and is integral for all , then is integral. Hint: replace by to reduce to the case where is a subring of . Suppose . Show that for some and independent of . Use a "partition of unity" argument as in the proof of Theorem 4.1.2 to show that .
Proof:
Following the hint, we may suppose without loss of generality that is a subring. Letting be arbitrary, we have for each that there exists some (the result would be trivial if there were some ), (for ) and such that
Letting be the maximum of all , we have that \( b^{t_i} a^{m_i + \beta_{i, m_i - 1} a^{m_i-1} + \dots + \beta_{i, 0} = 0 \) — in other words, . By taking to be the maximum of all and the maximum of all , we have for each . Similar to Theorem 4.1.2, we know that as , so implies . Therefore, we may find such that and thus
Exercise 7.2.B:
Show that the property of a homomorphism being integral is always preserved by localization and quotient of , and quotient of , but not localization of . More precisely: suppose is integral. Show that the induced maps , , and are integral (where is a multiplicative subset of , is an ideal of , and is an ideal of ), but need not be integral (where is a multiplicative subset of ). (Hint for the latter: show that is an integral homomorphism, but is not.)
Show that the property of being an integral extension is preserved by localization of , but not localization or quotient of . (Hint for the latter: is an integral extension, but is not.)
In fact the property of being an integral extension (as opposed to integral homomorphism) is not preserved by taking quotients of either. (Let and . Then injects into , but doesn’t inject into .) But it is in some cases. Suppose is an integral extension, and is the restriction of an ideal . (Side Remark: you can show that this holds if is prime.) Show that the induced map is an integral extension. (Hint: show that the composition is an injection.)
Proof:
Let be integral and first consider the case that is a multiplicative subset ( so that is as well ). Let be arbitrary. Since and is integral, we can find some and such that . Dividing the equation by gives us
So that integral over
Next, let be an ideal. Notice that for any , since is integral there exist and such that . This implies all iff so that integrality is well-defined and preserved on the quotient map.
Lastly, let be an ideal and denote the canonical projection map. Then for any and representative , we know that there exists some and such that . Applying our map gives the desired result.
To see, however, that integrality is not preserved by localization of we follow the example given: the identity is trivially integral, however the induced localization map given by is certainly not integral. To see this, notice that by taking , for any we have
since we cannot cancel out the first term.
This follows from part (a) since if is integral, then is integral (additionally by algebra if is injective then is injective ). Additionally, we have already seen that integrality need not hold for localization of — to see that integral extensions need not be preserved follows from the simple fact that injective does not imply is injective.
We know from part (a) again that is integral so it suffices to show it is injective. Since , factors through . Since and injections imply is an injection (easy proof by contradiction), the result follows.
Exercise 7.2.C:
Show that if and are both integral homomorphisms, then so is their composition.
Proof:
Let be arbitrary; since the map is integral, is contained in a subalgebra with is finitely generated as a -module, say by . Since the map is integral, each is contained in some subalgebra that is finitely generated as a -module. Then is also finitely generated so that must also be a finitely generated -module. Since was arbitrary the result follows.
Exercise 7.2.D:
Suppose is a ring morphism. Show that the elements of integral over form a subalgebra of .
Proof:
Suppose are integral over . By Lemma 7.2.1, we have that and for some finitely generated -modules . But then is also a finitely generated -module which contains and so the result follows.
Exercise 7.2.E:
Show that the special case where is a field translates to: if is a subring with integral over , then is a field. Prove this. (Hint: you must show that all nonzero elements in have inverses in . Here is the start: If , then , and this satisfies some integral equation over .)
Proof:
With the Lying over theorem, the exercise is immediate from the fact that a ring is a field if and only if is the only prime ideal ( the proof in the reverse direction requires the fact that maximal implies field and ). To see this, notice that if is a prime ideal then there exists some prime such that . But is a field so we must have and thus .
However, since the proof of the Lying Over theorem below relies on this exercise, we should probably avoid circular logic. Following the hint, let — we wish to show that . Since is a field, and thus we can find some and such that
Multiplying through by , we may rewrite the equation as
Thus, so that is also a field.
Exercise 7.2.F:
Suppose is an integral homomorphism (not necessarily an integral extension). Show that if is a chain of prime ideals in and is a chain of prime ideals of such that "lies over" (and then the second chain can be extended to so that this remains true. ) (Hint: reduce to the case m = 1, n = 2; reduce to the case where and ; use the Lying Over Theorem 7.2.5.)
Draw a picture of this theorem (akin to Figure 7.1).
Proof:
By induction we only consider the case that and ; that is to say, suppose that is a chain of primes in and with (we may assume without loss of generality that is a subring of ). Then the induced map is clearly integral, since for any and representative we may choose such that and reduce mod . Moreover, the map must also be injective since if in for some then so that they are indeed equivalent in . Therefore, the map is an integral extension so by the Lying Over theorem there exists some lying over . If we let denote the canonical projection, then is the desired prime lying over .
Take , and consider in and in .
Exercise 7.2.G:
Suppose is a ring, and is an ideal of contained in all maximal ideals. Suppose is a finitely generated -module, and is a submodule. If is surjective, then .
Proof:
Notice that the assumption implies that . Thus, by considering the quotient module , we have that . By Version 2 of Nagata's lemma, we have so that .
Exercise 7.2.H:
Suppose is a local ring. Suppose is a finitely generated -module, and , with (the images of) generating . Then generate . (In particular, taking , if we have generators of , they also generate .)
Proof:
Take . By assumption, is surjective ( since is surjective and the composition is surjective by assumption ) we have by the previous exercise that .
Exercise 7.2.I:
Recall that a -module is said to be faithful if the only element of acting on by the identity is 1 (or equivalently, if the only element of acting as the 0-map on is 0). Suppose is a subring of a ring A, and ∈ A. Suppose there is a faithful -module that is finitely generated as an -module. Show that is integral over . (Hint: change a few words in the proof of version 1 of Nakayama, Lemma 7.2.8.)
Proof:
Let be a generating set of as an -module and suppose for some . Taking this implies that
Multiplying through by implies that
Now since has entries that are polynomials in and is faithful -module, we must have that so that is integral over .
Exercise 7.2.J:
Suppose is an integral domain, and is the integral closure of in , i.e., those elements of integral over , which form a subalgebra by Exercise 7.2.D. Show that is integrally closed in .
Proof:
This is immediate from .
A gazillion finiteness conditions on morphisms
Exercise 7.3.A:
Show that the composition of two quasicompact morphisms is quasicompact. (It is also true — but not easy — that the composition of two quasiseparated morphisms is quasiseparated. This is not impossible to show directly, but will in any case follow easily once we understand it in a more sophisticated way, see Proposition 10.1.13(b).)
Proof:
Let and be quasicompact morphisms and let be affine. Then is quasicompact by definition, so by Exercise 5.1.D. Since is also quasicompact, for each we have that is quasicompact. Thus, is the finite union of quasicompact spaces and thus quasicompact (see Exercise 3.6.H).
Exercise 7.3.B:
Show that any morphism from a Noetherian scheme is quasicompact.
Show that any morphism from a quasiseparated scheme is quasiseparated. Thus by Exercise 5.3.A, any morphism from a locally Noetherian scheme is quasiseparated. Thus readers working only with locally Noetherian schemes may take quasiseparatedness as a standing hypothesis.
Proof:
Let be a morphism of schemes with Noetherian and suppose is affine. Since is continuous we have that is open. By Exercise 3.6.T since is Noetherian as a topological space we have that is open.
Let be a morphism of schemes with quasiseparated and suppose is affine. Then is open; we wish to show that it is in addition quasiseparated. Let be affine open subsets of ; then as open affines of , we know that is a finite union of open affine sets (by Exercise 5.1.F).
Exercise 7.3.C:
(quasicompactness is affine-local on the target) Show that a morphism is quasicompact if there is a cover of by affine open sets such that is quasicompact.
(quasiseparatedness is affine-local on the target) Show that a morphism is quasiseparated if there is a cover of by affine open sets such that is quasiseparated.
Proof:
Suppose that is an affine open subset; for each we have is a distinguished open of and is therefore also affine. Since is quasicompact (as a morphism), we have that is also quasicompact and thus a finite union of open affines. Taking the union over all is still a finite union of open affines so that is quasicompact.
In a similar fashion, let be an affine open subset so that is a distinguished open (and thus affine) for each . Since is quasiseparated for each , we have that is quasicompact. Thus, is the finite union of quasicompact sets and thus quasicompact.
Exercise 7.3.D:
Show that affine morphisms are quasicompact and quasiseparated. (Hint for the second: Exercise 5.1.G.)
Proof:
Let be an affine morphism and an affine open set. Then is affine, so by Exercise 3.6.G and Exercise 5.1.G it is quasicompact and quasiseparated (respectively).
Exercise 7.3.E:
What is the natural map of the Qcqs Lemma 7.3.5? (Hint: the universal property of localization, Exercise 1.3.D.)
Proof:
Recall from Exercise 1.3.D that is initial among all -algebras such that is invertible. Since is the locus where doesn't vanish, by Exercise 4.3.G(b) we know that is invertible on . Thus, by the universal property noted above we have that there is a map which we can give explicitly by
Exercise 7.3.F:
Suppose is a closed subset of an affine scheme locally cut out by one equation. (In other words, can be covered by smaller open sets, and on each such set is cut out by one equation.) Show that the complement of is affine. (This is clear if is globally cut out by one equation , even set-theoretically; then . However, is not always of this form, see §19.11.10.)
Proof:
By assumption we may cover with distinguished opens such that for some . Consider the inclusion ; for each we notice that so restricted to (the preimage of) is affine. Then by proposition 7.3.4 we have that is affine.
Exercise 7.3.G:
Show that a morphism is finite if there is a cover of by affine open sets such that is the spectrum of a finite -algebra.
Proof:
Let be an arbitrary affine open subset — if for some affine in our open cover then we may assume it is a union of distinguished open, say . Suppose ; by assumption is a finite -algebra, so we must also have is a finite -algebra for each . In other words, is clearly a finite morphism. Thus it suffices to show that if and a finite -algebra (for all ) implies is a finite -algebra.
For each , suppose is a generating set of (as an -module) for some . Taking large enough , we may define a map such that at each we have . Then for any there exist such that . Taking , notice that since we also have
Therefore, so that is a finitely generated -algebra.
Exercise 7.3.H:
Show that if is a finite morphism, then is a finite union of points with the discrete topology, each point with residue field a finite extension of , see Figure 7.5. (An example is .) Do not just quote some fancy theorem! Possible approach: By Exercise 3.2.G, any integral domain which is a finite -algebra must be a field. If , show that every prime of is maximal. Show that the irreducible components of are closed points. Show is discrete and hence finite. Show that the residue fields K(A/p) of A are finite extensions of k. (See Exercise 7.4.D for an extension to quasifinite morphisms.)
Proof:
By definition we must have is affine such that where is a finite -algebra. Now for any we have that is an integral domain which is also necessarily a finite -algebra and thus a field so that the only ideal is --- by the correspondence theorem, this implies that must be maximal (additionally, it also follows that must be a finite extension of , possibly by using something similar to Exercise 7.2.J). By the Chinese Remainder theorem, if is a distinct set of maximal ideals in then
Since is a finite -algebra, we have that --- in particular, is a collection of no more than points. To see that has the discrete topology, it suffices to show that every point is both open and closed. Now we know that points corresponding to maximal ideals are closed, but since there are only finitely many points we may take the finite intersection of the compliment of for each .
Exercise 7.3.I:
Show that the composition of two finite morphisms is also finite.
Proof:
Let and be finite morphisms. Indeed the composition of affine morphisms is affine so we know that is indeed affine. Let , and be the affine preimages. Since is a finite -algebra, we get a surjective morphism . Similarly, as is a finite -algebra, we get a surjective morphism . Thus, we obtain a surjective map so that is a finite -algebra as desired.
Exercise 7.3.J:
If is an -algebra,define a graded ring by ,and for . (What is the multiplicative structure?\) Hint: you know how to multiply elements of together, and how to multiply elements of with elements of .) Describe an isomorphism . Show that if is a finite -algebra (finitely generated as an -module) then is a finitely generated graded ring over , and hence that is a projective -scheme (§4.5.9).
Proof:
The -grading on occurs in a natural way; acts on each as is an -algebra. Moreover, for and we declare as this is required to make this a graded -algebra (one should keep in mind that looks similar to , so as we hope to find a bijection — see Exercise 6.4.A). Notice from our -structure on , it is ambiguous how to embed any element into as ; however, as we only care about prime ideals, this allows us to construct a well-defined bijection . For each , we define by
In fact, this gives us an isomorphism in quite an obvious way. Notice that if with then must contain since for any . Thus,
Exercise 7.3.K:
Show that finite morphisms have finite fibers. (This is a useful exercise, because you will have to figure out how to get at points in a fiber of a morphism: given , and , what are the points of ? This will be easier to do once we discuss fibers in greater detail, see Remark 9.3.4, but it will be enlightening to do it now.) Hint: if and are both affine, and , then we can throw out everything in outside by modding out by ; show that the preimage is . Then you have reduced to the case where is the of an integral domain , and is the generic point. We can throw out the rest of the points of by localizing at . Show that the preimage is of localized at . Show that the condition of finiteness is preserved by the constructions you have done, and thus reduce the problem to Exercise 7.3.H.
Proof:
Fix . As fibers are a local construction, it suffices to consider so that
and . Similar to Exercise 7.3.H, we know that is a domain — thus, for any , we know that is a finite field-extension. Thus, can only have finitely many points.
Exercise 7.3.L:
Show that the open embedding has finite fibers and is affine, but is not finite.
Proof:
The open embedding \) corresponds to the ring morphism . Indeed this is affine and has finite fibre (as the preimage of any point is a point), but is certainly not a finite -algebra as cannot be expressed by for any .
Exercise 7.3.M:
Prove that integral morphisms are closed, i.e., that the image of closed subsets are closed. (Hence finite morphisms are closed. A second proof will be given in §8.2.5.) Hint: Reduce to the affine case. If is a ring map, inducing the integral morphism , then suppose cuts out a closed set of , and , then note that , and apply the Lying Over Theorem 7.2.5 here.
Proof:
Let be an integral morphism and let be closed. Since integral morphisms are affine, we may assume without loss of generality that and so that for some . If we take then clearly factors through . Since and integral, integral. By the lying over theorem, for any we have so that . Since , is clearly closed.
Exercise 7.3.N:
Suppose is integral. Show that for any ring homomorphism , the induced map is integral. (Hint: We wish to show that any is integral over . Use the fact that each of the finitely many are integral over , and then Exercise 7.2.D.) Once we know what "base change" is, this will imply that the property of integrality of a morphism is preserved by base change, Exercise 9.4.B(e).
Proof:
Let and be our ring morphisms giving and the structure of -algebras. Recall that comes with the -algebra structure induced by the commutative diagram
In particular, for every we have that . Now let be a pure tensor; as is integral over , there exist and such that . Tensoring with , this tells us
Since each lies in the image of the natural map , we have that is integral over .
Exercise 7.3.O:
Show that a morphism is locally of finite type if there is a cover of by affine open sets such that is locally finite type over
Proof:
Suppose is an affine open neighborhood. Then is an affine open cover of by distinguished opens. As is quasicompact, we may choose some finite affine open cover of (so that as an algebra). If we let , then we get maps for each — by an argument similar to Exercise 7.3.G, this implies that is a finitely generated -algebra as desired.
Exercise 7.3.P:
(easier) Show that finite morphisms are of finite type.
Show that a morphism is finite if and only if it is integral and of finite type.
Proof:
This is trivial by definition since if is a finite morphism and , then with a finite -algebra and thus finitely generated -algebra.
The forward direction follows from the previous part (and finite algebras are integral). Conversely, suppose is integral and of finite-type. Since integral maps are affine, we may restrict to the case and so that is integral and finite type over . To see that is a finite -algebra, let be the generators of (as a -algebra). Now is integral over so is a -subalgebra finite over by Lemma 7.2.1. Similarly, is integral over so by induction the result follows.
Exercise 7.3.Q:
Show that every open embedding is locally of finite type, and hence that every quasicompact open embedding is of finite type. Show that every open embedding into a locally Noetherian scheme is of finite type.
Show that the composition of two morphisms locally of finite type is locally of finite type. (Hence as the composition of two quasicompact morphisms is quasicompact, Easy Exercise 7.3.A, the composition of two morphisms of finite type is of finite type.)
Suppose is locally of finite type, and is locally Noetherian. Show that is also locally Noetherian. If is a morphism of finite type, and is Noetherian, show that is Noetherian.
Proof:
This follows since if we restrict to and with then each is finitely generated by over . If is locally Noetherian, then we may modify the above assumption to the case that is a Noetherian ring. By Exercise 3.6.T, every open subset of a Noetherian topological space is quasicompact; thus, is a quasicompact morphism and therefore of finite type.
Suppose and are locally of finite type and let be an affine open. Using the equivalent definition of locally of finite type, may be covered by affine opens such that each is a finitely generated -algebra. Similarly for each , may be covered by affine opens such that is a finitely generated -algebra. Since each is a finitely generated -algebra, we get that is a finitely generated -algebra.
Since is locally Noetherian, it may be covered by open affines where each is Noetherian. Since is locally of finite type, may be covered by where is a finitely generated -algebra. By Exercise 3.6.X, since is finitely-generated as a module and is Noetherian, must be Noetherian as well. By running over our affine cover , we may cover by with Noetherian so that is locally Noetherian.
Now if is of finite type, then it is locally of finite type and quasicompact. Assuming is Noetherian, we may cover by finitely many with Noetherian. Since is quasicompact, is quasicompact so we can cover by finitely many of the as above (each a finitely generated -algebra and thus Noetherian). Since there were originally finitely many , we can thus cover by a finite number of so that is also Noetherian.
Exercise 7.3.R:
Suppose is prime and . Let , and . Define by for each , and let be the map of schemes corresponding to .
Show that is the identity on the level of sets, but is not the identity morphism.
Show that is a bijection, but is not an isomorphism of schemes.
If , show that the morphism of -schemes corresponding to is a bijection, but no power of is the identity on the level of sets.
Proof:
Recall that as a map of sets is simply the map . Now if then (since is prime) so that — therefore, the map of sets is indeed a bijection.
Now by construction, is the splitting field of over so that every element satisfies . However, this need not hold for elements (for example, is free so ).
On the level of sets, is clearly a bijection since . However, cannot be an isomorphism since is not surjective (pick with coprime to )
If we consider to be the set of all polynomials of the form for some then is the splitting field of (a common Galois theory exercise). To see that is a bijection requires a little more work than before, as it need not hold that is a power of for any . However, if we write for some multi-index , then for each there exists some such that and thus . Taking , we have that . Now any prime is finitely-generated (as is obviously Noetherian) so that we may apply this argument to a finite set of generators of giving us some with . Thus, our map must be a bijection. However, our integer depends on each point so that no works universally. That is to say if we choose some arbitrary and consider the map then we could simply choose such that is not a root of .
Exercise 7.3.S:
Suppose is a scheme over . Explain how to define (without choice) an endomorphism such that for each affine open subset , corresponds to the map given by for all .(The morphism is called the absolute Frobenius morphism.)
Proof:
If we let be any affine cover, then by definition each is a -algebra so we get a map given by , which by the previous exercise induces the identity on points of . Then these maps clearly glue together since our open sets glue.
Exercise 7.3.T:
Show that the notion of "locally of finite presentation" is affine-local on the target.
Proof:
This is somewhat immediate from the definition. Let be a morphism locally of finite presentation and ; we first wish to show that is locally of finite presentation. If is an affine open of , then it is also an affine open of so by virtue of being locally of finite type, can be covered by open affines with each a finitely presented -algebra.
For the converse, suppose is an affine cover for with each locally of finite presentation, and let be an arbitrary open affine. Then whenever is non-empty, we may assume it is of the form of some distingished open for some . As this is also a distinguished open of (for some different ), we have with each a finitely-presented -algebra (and thus -algebra). Thus, it suffices to show that if with each a finitely-presented module, then is a finitely presented -module. Similar to before, we know that since finitely generated over each and there exist a surjective map ; by looking at the cokernel of this map, it must be the case that since it vanishes at each localization. Similarly, the kernel must be finitely generated, as it is finitely generated at each localization.
Exercise 7.3.U:
Show that if is locally of finite presentation, then for any affine open subscheme of and any affine open subscheme of with , is a finitely presented -algebra. In particular, the notion of "locally of finite presentation" is affine-local on the source.
Proof:
This ammounts to showing that if and a finitely-presented -algebra for each , then is a finitely-presented -algebra. I will outsource this proof to [Stacks 00EO].
Exercise 7.3.V:
Show that open embeddings are locally finitely presented.
Proof:
Let be our open embedding. Indeed we may reduce to the case and consider the identity morphism . Taking , we simply apply the previous exercise.
Exercise 7.3.W:
Show that the composition of two locally finitely presented morphisms is locally finitely presented. (Then once we show that the composition of two quasiseparated morphisms is quasiseparated in Proposition 10.1.13(b), we will know that the composition of two finitely presented morphisms is finitely presented — recall that the composition of two quasicompact morphisms is quasicompact, by Easy Exercise 7.3.A.)
Proof:
It suffices to reduce to the affine case, so that we have ring maps and where is a finitely presented -algebra and is a finitely presented algebra. Thus, there exist integers and surjections and such that and are finitely-generated ideals of and respectively. Then may be extended to a morphism in the natural way, so that is finitely generated in . By taking compositions, we see that is clearly a surjection, so that is a finitely-generated -algebra. To see that it is finitely presented, we must show that is finitely generated.
Suppose , or in other words . Since is assumed to be finitely generated in (say by some ) we have that for . Since surjects onto , we can find some with . By linearity, this implies or . Since this is finitely generated by and , we must have that is generated by and .
Images of morphisms: Chevalley’s Theorem and elimination theory
Exercise 7.4.A:
Recall that a subset of a topological space X is locally closed if it is the intersection of an open subset and a closed subset. (Equivalently, it is an open subset of a closed subset, or a closed subset of an open subset. We will later have trouble extending this to open and closed and locally closed subschemes, see Exercise 8.1.M.) Show that a subset of a Noetherian topological space is constructible if and only if it is the finite disjoint union of locally closed subsets. As a consequence, if is a continuous map of Noetherian topological spaces, then the preimage of a constructible set is a constructible set. (Important remark: the only reason for the hypothesis of the topological space in question being Noetherian is because this is the only setting in which we have defined constructible sets. An extension of the notion of constructibility to more general topological spaces is mentioned in Exercise 9.3.I.)
Proof:
For the forward direction, suppose is a finite disjoint union of locally closed subsets; that is to say with closed and open. In fact if we just assume the sets are disjoint we may write . Now since every open set is in the family of constructible sets and is closed under taking compliments, all closed sets are in . Similarly, as finite intersections and complements are in we must also have finite unions are in . Thus each and consequently is in .
Now suppose we let denote the collection of subsets of that can be written as a finite disjoint union of locally closed subsets; by the previous paragraph we have . To show that , it suffices to show that also satisfies properties (i), (ii), and (iii). Clearly contains all open sets since and is closed. To see that is closed under intersections, notice that if , then the intersection is which is also clearly an element of . To prove that is closed under compliments, we let denote the collection of subsets of which can be expressed as a disjoint union of exactly locally-closed sets. First notice that for any we have . Then which is clearly in . Proceeding inductively, suppose that compliments in are contained in for . If we let then we can write for some and . Then ; by inductive hypothesis both and and we have shown is closed under intersection. Since , this completes the proof.
Exercise 7.4.B:
Show that the generic point of does not form a constructible subset of (where is a field).
Proof:
Let denote the generic point; as is dense, every open subset intersects so that for all . Therefore, by the previous theorem we would have is constructible if and only if is closed. This would hold if and only if which is impossible by Exercise 3.2.D as is infinite.
Exercise 7.4.C:
Show that a constructible subset of a Noetherian scheme is closed if and only if it is "stable under specialization". More precisely, if is a constructible subset of a Noetherian scheme , then is closed if and only if for every pair of points and with , if , then . Hint for the "if" implication: show that can be written as where is open and is closed. Show that can be written as (with possibly different , , ) where each is irreducible and meets . Now use &qupt;stability under specialization" and the generic point of to show that for all , so .
Show that a constructible subset of a Noetherian scheme is open if and only if it is "stable under generization". (Hint: this follows in one line from (a).)
Proof:
For the "only if" direction, suppose that is closed and let with in . Since is closed, so that as well. Conversely, suppose is constructible and stable under specialization; as is constructible, we may write by Exercise 7.4.A with each open and closed. As is Noetherian, we know that each may be written as for irreducible components of (by Proposition 3.6.15). Now by using an inductive argument, if then we may write — however, now may not be reducible in the Noetherian subspace , so we must again break it into finitely many irreducible components (as a helpful example, take to be an open neighborhood of the origin and consider — by removing we have split into two irreducible components). After finitely many steps, this allows us to write where each is irreducible. Let denote the unique generic point of - since nontrivially intersects any open set intersecting , we know so that . Thus, for any we know that for some ; as we just showed this implies that since is stable under specialization. In other words, . However since for all we get the reverse containment so that is closed.
Much like the previous part, it is immediate that open sets are stable under generization: if is open and with , then by definition every open neighborhood of nontrivially intersects . Since is such an open neighborhood, . For the reverse direction, if is a constructible set stable under generization, then is a constructible set stable under specialization and therefore closed. Thus is open.
Exercise 7.4.D:
Suppose is a quasifinite morphism. Show that is finite. (Hint: deal first with the affine case, , where is finitely generated over . Suppose contains an element that is not algebraic over , i.e., we have an inclusion . Exercise 7.3.H may help.)
Proof:
By definition, is a finite set; moreover, we may write by quasicompactness. As the morphism is locally of finite type, we know that each is a finitely generated -algebra (say by ). By the same argument to that in the proof of the Nullstellensatz, each must in fact then be a finite -algebra, so that has support which is a point. Thus is discrete and thus the spectrum of the product of our so that is affine.
Exercise 7.4.E:
Assume Chevalley’s Theorem 7.4.2. Show that a morphism of affine -varieties is surjective if and only if it is surjective on closed points (i.e., if every closed point of is the image of a closed point of ). (Once we define varieties in general, in Definition 10.1.7, you will see that your argument works without change with the adjective "affine" removed.)
Proof:
For the forward direction, if is surjective it is certainly surjective on closed points. Conversely, suppose that is surjective on closed points. By Chevalley's theorem, the image of is constructible so is constructible. We wish to show ; if we suppose to the contrary that there exists some then it must be the case that is not a closed point (as all closed points are in the image by assumption). Now as is constructible, by Exercise 7.4.A we may write so that . We may think of as an irreducible affine subvariety, so that is a non-empty open subset. But then must contain a closed point by Exercise 3.6.J as the closed points are dense, which must then also be a closed point of — a contradiction.
Exercise 7.4.F:
Show that itself satisfies .
Proof:
Let be a finitely-generated -module with generators . Then for any multiplicatively closed subset , is generated by (as a -module). By taking , since is an integral domain we have that is a field and is a -vector space. Thus, we can extract a basis for some ; then for each we can write
for . Then by taking , each relation above holds in so has a linearly independent basis and is thus free.
Exercise 7.4.G:
Reduce the proof of Lemma 7.4.4 to the following statement: if is a finitely-generated -algebra satisfying , then does too. (Hint: induct on the number of generators of as a -algebra.)
Proof:
If is a finitely generated -algebra, then it is isomorphic to for some and ideal . Thus, a finitely generated -module is a finitely-generated -module with . By assuming the statement in the exercise, one may use induction to show that as a -module must have some corresponding such that is a free -module.
Exercise 7.4.H:
Show that multiplication by induces a surjection
Proof:
By inductive construction of our , we can simply express our -submodules as
Then for any , we have that , from which it is obvious that multiplication by gives a surjection .
Exercise 7.4.I:
Show that for , is an isomorphism. Hint: use the ascending chain condition on .
Proof:
The hint is quite literally the answer. As is a finitely generated algebra over a Noetherian ring, it is Noetherian. Since the form an ascending chain, they eventually terminate so will eventually be trivial in which case will also be trivial (and thus an isomorphism).
Exercise 7.4.J:
Show that there is a nonzero such that is free as a -module, for all . Hint: as varies, passes through only finitely many isomorphism classes.
Proof:
Since we are assuming satisfies , each is a finitely-generated -module so that there exists some nonzero such that is a free -module. Since there are only finitely many isomorphism classes by the previous exercise, we may take for sufficiently large so that the claim follows.
Exercise 7.4.K:
Suppose is a -module that is an increasing union of submodules , with , and that every is free. Show that is free. Hint: first construct compatible isomorphisms by induction on . Then show that the colimit is an isomorphism. Side Remark: More generally, your argument will show that if the are all projective modules (to be defined in §23.2.1), then is (non-naturally) isomorphic to their direct sum.
Proof:
We first wish to construct the compatible isomorphisms . The base case is obvious since . Now suppose is an isomorphism for , and consider the short exact sequence
Since is free, the sequence splits so that we get a (non-natural) isomorphism . By inductive hypothesis, we have an isomorphism — composing in the first component of gives the desired isomorphism. Since is a functor, it takes isomorphisms ( over a filtered category, so in our case the entire directed system of isomorphisms ) to isomorphisms. Since the submodules all have directed arrows to , it is fairly easy to see that ; thus we obtain the desired isomorphism
Exercise 7.4.L:
Suppose is a finite type morphism of Noetherian schemes, and is irreducible. Show that there is a dense open subset of such that the image of either contains or else does not meet . (Hint: suppose is such a morphism. Then by the Generic Freeness Lemma 7.4.4, there is a nonzero such that is a free -module. It must have zero rank or positive rank. In the first case, show that the image of does not meet . In the second case, show that the image of contains .)
Proof:
Since is finite type, it suffices to check on our finite affine cover; that is we restrict to and as noted in the hint. Moreover, Generic Freeness Lemma implies there is some nonzero with a free -module. In the case that is rank zero it is necessarily torsion — that is to say we can find some such that . Since no prime ideal may contain a zero-divisor and becomes a zero-divisor in , we have that the composition is zero.
Otherwise when has positive rank we get a surjective morphism , so that composing with the surjective morphism we get that is actually in the image of . Since is assumed to be irreducible, we get that is open and dense ( Exercise 3.6.B ).
Exercise 7.4.M:
Show that to prove Chevalley’s Theorem, it suffices to prove that if is a finite type morphism of Noetherian schemes, the image of is constructible.
Proof:
We assume the statement above (image of finite type + noetherian is constructible) and wish to show that this implies that for any constructible, is constructible. Let for each open; since the finite disjoint union of constructible sets is constructible, it suffices to show that each is constructible. But the inclusion is finite type by Exercise 7.3.Q so that composing with gives us the restriction which must then also be a finite type morphism of Noetherian schemes. By assumption, this tells us is constructible so that is as well.
Exercise 7.4.N:
Reduce further to the case where is affine, say . Reduce further to the case where is affine
Proof:
Since is a Noetherian scheme, it is quasicompact by Exercise 3.6.T; thus we may take some finite affine cover for . But then
Since the finite union of constructible subsets is constructible, it suffices to consider --- in other words, simply consider .
Since is finite type, it is locally finite type and quasicompact. Since we are assuming , must be quasicompact and covered by finitely many open affines. Since the finite union of constructible sets is constructible, we may reduce to .
Exercise 7.4.O:
Complete the proof of Chevalley’s Theorem 7.4.2, by making the above argument precise.
Exercise 7.4.P:
Fix an algebraically closed field . Suppose
are given. Show that there is a (Zariski)-constructible subset of such that
and
has a solution if and only if
Proof:
Following the hint, let so that is finite type, as is finitely-generated over . Now for each , since is a finitely-generated -algebra, is of finite type. Since there are finitely many , we get that the morphism is of finite type and thus the composition. Thus, the map is of finite type so by Chevalley's theorem, it has a constructible image. Since is finitely generated over , the morphism is of finite-type. Applying the Nullstellensatz cleans things up a bit.