Sei sulla pagina 1di 4

Solutions to Atiyah-Macdonald, Chapter 6

Dave Karpuk May 19, 2010

Exercise 1. (i) Let M be a Noetherian A-module and u : M M a module homomorphism. If u is surjective, then u is an isomorphism. (ii) If M is Artinian and u is injective, then again u is an isomorphism. Proof. (i) The chain . . . ker(un ) ker(un+1 ) ker(un+2 ) . . . of submodules of M must terminate by the Noetherian assumption. So pick n such that ker(un ) = ker(un+1 ). Let m ker(u). The maps uk are all surjective by the surjectivity of u. Pick x M such that un (x) = m. Applying u, we get the equation un+1 (x) = u(m) = 0 which says that x ker(un+1 ) = ker(un ). This implies that m = 0, and hence u is injective. (ii) The chain . . . im(un ) im(un+1 ) im(un+2 ) . . . of submodules must terminate by the Artinian condition. So Pick n such that im(un ) = im(un+1 ). Let m M . We can nd an x M such that un (m) = un+1 (x). Then un (m u(x)) = 0, but un is injective because u is injective. Hence m = u(x), and it follows that u is surjective and thus an isomorphism. Exercise 2. Let M be an A-module. If every non-empty set of nitely generated submodules of M has a maximal element, then M is Noetherian. Proof. Let N be a submodule of M , and let S = {Ni } be the set of all nitely generated submodules of N . By assumption, S has a maximal element, call it N . Then for any Ni S, we have Ni N . If this were not the case, then we could pick x Ni , x N . The submodule N + Ax would be a nitely generated submodule properly containing N , contradicting the maximality of N . Therefore, N= Ni N N
i

which says that N = N . We conclude that every submodule of M is nitely generated, which says that M is Noetherian. Exercise 3. Let M be an A-module and let N1 , N2 be submodules of M . If M/N1 and M/N2 are Noetherian, so is M/(N1 N2 ). Similarly with Artinian in place of Noetherian.

Proof. We may replace M/(N1 N2 ) with M and assume that N1 N2 = 0. So we must prove that for submodules N1 , N2 of M such that N1 N2 = 0, M/N1 and M/N2 Noetherian implies M Noetherian. We have an exact sequence 0 N1 M M/N1 0 of A-modules. Because N1 N2 = 0, a copy of N1 sits inside M/N2 . Thus we have another exact sequence 0 N1 M/N2 M/(N1 + N2 ) 0 which by Proposition 6.3(i) proves that N1 is Noetherian. Again using Proposition 6.3(i) and the rst exact sequence, we see that M is Noetherian. The proof for Artinian is the same, using Proposition 6.3(ii). Exercise 4. Let M be a Noetherian A-module and let a be the annihilator of M in A. Prove that A/a is a Noetherian ring. If we replace Noetherian by Artinian in this result, is it still true? Proof. Because M is Noetherian, it is nitely generated, say by m1 , . . . , mn . Let ai = Ann(mi ). By the in-text Exercise 2.2, we have a = n ai . We have an inclusion A/ai M of A-modules, i=1 induced by the map 1 mi . Thus A/ai is a Noetherian A-module for all i. By Exercise 3 and induction, we see that A/a is a Noetherian A-module and thus a Noetherian A/a-module, which proves that it is a Noetherian ring. To see that the result fails if we replace Noetherian with Artinian, consider A = Z and M = Qp /Zp . The Z-module M is Artinian by the argument of Example (3) in the text, but its annihilator is the zero ideal of Z. As Z = Z/(0) is not an Artinian ring, we see that the result fails. Exercise 5. A topological space X is said to be Noetherian if the open subsets of X satisfy the ascending chain condition (or, equivalently, the maximal condition). Since closed subsets are complements of open subsets, it comes to the same thing to say that the closed subsets of X satisfy the descending chain condition (or, equivalently, the minimal condition). Show that, if X is Noetherian, then every subspace of X is Noetherian, and that X is quasi-compact. Proof. Let Y X be a subspace, and suppose that (Un )n0 is a chain of open subsets of Y . Then Un = Vn Y for some open Vn X. We are free to replace Vn with Vn1 Vn , and so we may assume that Vn Vn+1 for all n. Because X is Noetherian, there exists N such that Vn = Vn+1 for all n N . Intersecting with Y , we see that the chain (Un ) is eventually stationary, thus Y is Noetherian. To see that X is quasi-compact, let X = i Ui be an open cover. Let be the collection of open subsets of X which are nite unions of the Ui s. The set is partially ordered with respect to inclusion, and because X is Noetherian it has a maximal element, say U = U1 Un . If U = X, pick x X U . Because the Ui s cover X, there exists Ui such that x Ui . Then U is properly contained in U Ui , contradicting the maximality of U . Therefore U = X and X is quasi-compact. Exercise 6. Prove that the following are equivalent: (i) X is Noetherian. (ii) Every open subspace of X is quasi-compact. (iii) Every subspace of X is quasi-compact. 2

Proof. (i) (iii) If Y is a subspace of X, then Y is Noetherian by the previous exercise, and thus quasi-compact (also by the previous exercise). (iii) (ii) Obvious. (ii) (i) Let (Un )n0 be a chain of open subsets of X, and let U = i Ui . The Ui form an open cover of the quasi-compact open set U , thus there is a nite subcover. Hence i Ui = N Ui i=0 for some N and thus the chain is eventually stationary, proving that X is Noetherian. Exercise 7. A Noetherian space is a nite union of irreducible closed subsets. Hence the set of irreducible components of a Noetherian space is nite. Proof. Let be the set of non-empty closed subsets of X which are not nite unions of irreducible closed subspaces. Suppose that = . Because X is Noetherian, has a minimal element C. Then C itself is not irreducible, hence C = D1 D2 for some proper non-empty closed subsets Di of C. By minimality of C, we can express each Di as a nite union of irreducible subspaces of X, but then C is a nite union of irreducible subspaces of X, which is a contradiction. Since = , we have that X itself is a nite union of irreducible closed subsets. Each of these irreducible closed subsets is contained in a unique irreducible component of X. Conversely, every irreducible component contains one of these irreducible closed subsets, thus the number of irreducible components of X is nite. Exercise 8. If A is a Noetherian ring then Spec(A) is a Noetherian topological space. Is the converse true? Proof. Let V (a0 ) V (a1 ) be a descreasing chain of closed subsets of Spec(A). Replacing each ai with its radical does not change the sets V (ai ), thus we may assume that all of the ai are radical ideals. Thus they induce a sequence a0 a1 of ideals of A which must terminate because A is Noetherian. Thus an = an+1 for suciently large n and because the ai are radical, this implies that V (an ) = V (an+1 ) for suciently large n. Therefore Spec(A) is Noetherian. The converse is false. Let k be a eld and consider the ring A = k[x1 , x2 , . . .]/(x2 , x2 , . . .). Let p 1 2 be a prime ideal in this ring. Then p corresponds to some prime ideal q of the ring k[x1 , x2 , . . .] which contains x2 for all i. Because q is prime, we must have xi q for all i and thus q = (x1 , x2 , . . .). i Therefore p is the prime ideal of A generated by the images of all of the xi . The quotient A/p = k is a eld and thus p is maximal. Therefore p is the only prime ideal of A and Spec(A) is the one-point topological space which is clearly Noetherian. However, A is not a Noetherian ring. It contains the strictly increasing chain (x1 ) (x1 , x2 ) (x1 , x2 , x3 ) . Exercise 9. Deduce from Exercise 8 that the set of minimal prime ideals in a Noetherian ring is nite. Proof. By Exercise 8, Spec(A) is a Noetherian topological space. By Exercise 7, Spec(A) is thus a union of nitely many irreducible components. By Chapter 1, Exercise 20(iv), the irreducible components of Spec(A) correspond bijectively to the minimal prime ideals of A. Thus there are only nitely many minimal prime ideals of A. Exercise 10. If M is a Noetherian module (over an arbitrary ring A) then Supp(M ) is a closed Noetherian subspace of Spec(A). 3

Proof. By Chapter 3, Exercise 19, we have Supp(M ) = V (Ann(M )) = Spec(A/Ann(M )). By Exercise 4, A/Ann(M ) is a Noetherian ring and so by Exercise 8 Spec(A/Ann(M )) is a Noetherian topological space. Exercise 11. Let f : A B be a ring homomorphism and suppose that Spec(B) is a Noetherian space (Exercise 5). Prove that f : Spec(B) Spec(A) is a closed mapping if and only if f has the going-up property. Proof. The fact that f closed implies that f has the going-up property is contained in Exercise 10 of Chapter 5. To prove the converse, we must prove that f (Spec(B/b)) is closed in Spec(A) for all b B. Replacing B by B/b (whose spectrum is still Noetherian) and f by the composite A B B/b (which still has the going-up property), we are reduced to proving that f (Spec(B)) is closed in Spec(A). As Spec(B) is Noetherian, we can write Spec(B) = n Spec(B/qi ) as the nite union of i=1 irreducible closed subspaces Spec(B/qi ). So it suces to prove that f (Spec(B/qi )) is closed for all i. That is, we may assume that B is a domain and that Spec(B) is irreducible. Thus f (Spec(B)) = f (V (0)), so by Chapter 1, Exercise 21(iii) we have f (Spec(B)) = V (ker(f )). f (Spec(B)), As B is a domain, we have ker(f ) = p for some prime p of A. To prove f (Spec(B)) = we must prove that every prime ideal containing p is contracted. But if p p , then there exists a prime q of B such that q c = p by the fact that f has the going-up property. Therefore f (Spec(B)) is equal to its closure and hence closed.

Exercise 12. Let A be a ring such that Spec(A) is a Noetherian space. Show that the set of prime ideals of A satises the ascending chain condition. Is the converse true? Proof. Let p1 p2 be a chain of prime ideals of A. Applying V () we obtain a descending chain V (p1 ) V (p2 ) of closed subsets of Spec(A), which must terminate because Spec(A) is Noetherian by assumption. Thus there exists n such that V (pn ) = V (pn+1 ) = , but for prime ideals we have V (pn ) = V (pn+1 ) if and only if pn = pn+1 . Therefore the chain of prime ideals terminates.

Potrebbero piacerti anche