You are on page 1of 2

Solutions to Homework 10

52. (Dummit-Foote 10.3 #16) The kernel of the map M → M/A1 M × · · · × M/Ak M is the set of
all m ∈ M such that m ∈ Ai M for all i. But this is exactly A1 M ∩ · · · ∩ Ak M .

(Dummit-Foote 10.3 #17) The same proof that works for the ring R in the text works for a
module M just by making the obvious substitutions. Here is another proof, using tensor products.
Since the ideals Ai are all comaximal, we have by the Chinese Remainder Theorem for the ring
R that
R/A1 · · · Ak ' R/A1 × · · · × R/Ak . (1)
Now for any ideal I of R there is an isomorphism (R/I) ⊗R M ' M/IM of R-modules. As tensor
products commute with finite direct products, we can apply − ⊗R M to the isomorphism (1), to
obtain an isomorphism

M/A1 · · · Ak M ' M/A1 M × · · · × M/Ak M

of R-modules.

(Dummit-Foote 10.3 #18) As (a)M = 0, we have M/(a)M = M . The ideals pαi i are all
comaximal, so we can apply the Chinese Remainder Theorem to obtain an isomorphism

M ' M/(pα1 1 )M × · · · × M/(pαk k )M.

Now let m = (m1 , . . . , mk ) be an element of the right hand side, and suppose that pαi i m = 0. Then
α α
pαi i mj = 0 for all j. If j 6= i, then pj j mj = 0 as well, and since (pαi i ) and (pj j ) are comaximal, this
implies that mj = 0. Thus the pαi i -primary component of the right hand side is M/(pαi i M ).
The isomorphism obtained from the Chinese Remainder Theorem thus gives a direct sum (=
direct product, since the number of factors is finite) decomposition of M into its pαi i -primary
components. But by definition, the pαi i -primary component of M is Mi . Thus

M = M1 ⊕ · · · ⊕ Mk .

53. We have an exact sequence


f
0 → ker(f ) → V → im(f ) → 0

of R-modules. The map f : im(f ) → V defines a section of f : V → im(f ). To see this, note that if
f (x) ∈ im(f ), then f (f (f (x))) = f (f (x)) = f (x), because f 2 = f . Thus the exact sequence splits.

54. (a) Consider first an n × n matrix X = (Xij ) of variables Xij , where Xij ∈ Z[Xij ]. We must
prove the relations Xadj(X) = det(X)In and adj(X)X = det(X)In . Now, det(X) and the entries
of adj(X)X and Xadj(X) are all polynomials in the Xij , with coefficients only ±1, thus they live
in Z[Xij ]. However, this polynomial ring embeds in its field of fractions, where we are assuming
Cramer’s Rule to hold. Hence the polynomial relations hold for an n × n matrix of indeterminates.
Now if X = (xij ) is an n × n matrix with entries in R, simply substitute xij for Xij in the
equations we obtained from X · adj(X) = det(X)In and adj(X) = det(X)In . This proves Cramer’s

1
Rule for R.

(b) This is a similar argument to part (a). First consider matrices X = (Xij ) and Y = (Yij ).
Proving the equation det(XY ) = det(X) det(Y ) amounts to proving that n2 equations in the
polynomial ring Z[Xij , Yij ] are true. But this polynomial ring embeds in Q(Xij , Yij ), so the result
holds for matrices of indeterminates.
Now if X = (xij ) and Y = (yij ) are elements of Mn (R), substitute the xij for the Xij , and
the yij for the Yij , into the equations proving the result for Z[Xij , Yij ]. This yields the formula in
question for arbitrary R.

(c) Suppose that A ∈ Mn (R) has an inverse B, so that AB = I. Then 1 = det(AB) =


det(A) det(B) by part (b), so det(A) ∈ R× .
Conversely, if det(A) ∈ R× , then by Cramer’s Rule we have A · (det(A)−1 adj(A)) = I.

56. Let S = { Ann(x) | x ∈ M }. Then S is non-empty, because if x ∈ M is non-zero, then


0 ∈ Ann(x) and 1 ∈ / Ann(x), so Ann(x) is a non-empty proper ideal of R. By the Noetherian
hypothesis, S has a maximal element, call it Ann(x0 ). We will prove that Ann(x0 ) is prime.
Let a, b ∈ R, and suppose that ab ∈ Ann(x0 ) and b ∈
/ Ann(x0 ) (so that bx0 6= 0). Then abx0 = 0,
so a ∈ Ann(bx0 ). But anything that annihilates x0 annihilates bx0 , hence Ann(x0 ) ⊆ Ann(bx0 ).
By maximality, Ann(x0 ) = Ann(bx0 ), so a ∈ Ann(x0 ), and it follows that the ideal is prime.

You might also like