Vous êtes sur la page 1sur 3

MATH615 HOMEWORK 1

SOLUTIONS

Problem 1 Prove that Q/Z is not a flat Z-module.


2idQ/Z

Proof. It is clear that Z


Z is injective; but Z Z Q/Z Z Z Q/Z
is not injective (1 21 is in the kernel).

Problem 2 Let R be a commutative domain which is not a field. Prove that, if an
R-module M is both projective and injective, then M = 0.
Proof. Since M is an injective R-module, for any nonzero r R, the exact
r
r
sequence 0 R
R induces an exact sequence M = HomR (R, M )

HomR (R, M ) = M 0, i.e., rM = M for any nonzero r R.


Assume that : M R is not 0, i.e., there exists z M with (z) 6= 0.
Set r = (z). Since rM = M , there exists z 0 M with z = rz 0 . Then
r = (z) = (rz 0 ) = r(z 0 ), and hence r(1 (z 0 )) = 0 which implies
that (z 0 ) = 1 since R is domain. Therefore, any nonzero homomorphism
: M R must be surjective.
Since M is projective, there exists a free R-module F = Re such
that M F , where {e | } is a basis of F over a set . Let p :
F Re = R be the natural projection. If M 6= 0, there must exist an
such that p (M ) 6= 0 and hence the composite homomorphism M ,
p
r
F
R
R is a nonzero homomorphism for any nonzero r R, which
in turn has to be surjective. In particular, the identity 1 in the image,
i.e., there exists z M such that rp (z) = 1. Thus, R must be a field, a
contradiction. Therefore, M = 0.

Problem 3 Let (R, m) be a noetherian local ring. Prove that, if R/ m is a projective
R-module, then so is every R-module.
Proof. Since R/ m is projective, it is flat. Applying R R/ m to the short
exact sequence 0 m R R/ m 0, we have a short exact sequence
0 m R R/ m R R R/ m R/ m R R/ m 0. It is clear that
m R R/ m R R R/ m is the 0 homomorphism, thus, m R R/ m = 0.
Since m R R/ m
= m / m2 , we have m = m2 . By Nakayamas lemma, m = 0,
i.e., R is a field. Therefore every R-module is free, a fortiori projective. 
Problem 4 Let R = k[X, Y ]/(XY ), where X and Y are indeterminates over a field k.
Calculate TorR
i (R/(x), R/(y)) for all i, where x and y denote the images of
X and Y in R respectively.
1

Solution. It is clear that


y

R
R R R R/(x) 0
is a free resolution of R/(x). Applying R R/(y) and taking homology, we
have
(
k j even
TorR
j (R/(x), R/(y)) =
0 j odd

Problem 5 Let R be a commutative ring and M an R-module. Prove that, if x1 , . . . , xn
is an M -sequence, then so is xe11 , . . . , xenn for all positive integers e1 , . . . , en .
Proof. It suffices to prove that, whenever x1 , . . . , xn is an M -sequence, so
is xe11 , . . . , xn for any positive integer e1 , for, assuming this and setting
= M/xe1 M , we get xe2 , . . . , xn is M
-regular for any positive integer e2 ,
M
1
2
and so on.
Use induction on e1 to prove that xe11 , . . . , xn is also an M -sequence.
When e1 = 1, it is clear. Assume that n 2 and that xe11 1 , . . . , xn is an M sequence. We need to prove that xj is a nonzerodivisor on M/(xe11 , . . . , xj1 )M
for j 2 (since it is clear that xe11 is a nonzerodivisor on M and that
(xe11 , . . . , xn )M 6= M ). It is equivalent to proving that, if there exists zj
M such that xj zj (xe11 , . . . , xj1 )M , then zj (xe11 , . . . , xj1 )M . Since
xj zj (xe11 , . . . , xj1 )M , we can write xj zj = xe11 x1 + + xj1 zj1 for
some z1 , . . . , zj1 M . We have zj (x1e1 1 , . . . , xj1 ) since xe11 1 , . . . , xj
0
0
is an M -sequence. Write zj = x1e1 1 z10 + +xj1 zj1
with some z10 , . . . , zj1

M . Then we get
0
xe11 1 (x1 z1 xj z10 ) + x2 (z2 xj z20 ) + + xj1 (zj1 xj zj1
)=0

However, it is easy to check that, whenever there are an M -sequence


y1 , , yl and an equation y1 g1 + + yl gl = 0 with g1 , . . . , gl M , one
has gi (y1 , . . . , yl )M for all gi . Hence, x1 z1 xj z10 (xe11 1 , . . . , xj1 )M .
Thus, xj z10 (x1 , . . . , xj1 )M which implies that z10 (x1 , . . . , xj1 )M
since x, . . . , xj is an M -sequence. Consequently, zj (xe11 , . . . , xj1 )M .
This finishes the proof.

f

Problem 6 Let R be a commutative ring and let 0 L


M
N 0 be an
exact sequence of R-modules. Suppose that a sequence x1 , . . . , xn is both
L-regular and N -regular. Prove that x1 , . . . , xn is also M -regular.
Proof. Use induction on n.
When n = 1, we need to prove that, (i) x1 is a nonzerodivisor on M and
(ii) x1 M 6= M . If x1 z = 0 for some z M , then x1 g(z) = g(x1 z) = 0. Since
x1 is N -regular, g(z) = 0. Then there exists y L such that z = f (y).
Hence f (x1 y) = x1 f (y) = x1 z = 0, which implies that x1 y = 0 since f
is injective. But x1 is L-regular, y has to be 0 and hence z = 0. This
proves (i). Since we always have a surjection M/x1 M N/x1 N and x1 is
N -regular, M/x1 M 6= 0 i.e., M 6= x1 M . This proves (ii).
2

= L/(x1 , . . . , xn1 )L,


Assume that n 2 and x1 , . . . , xn1 is M -regular. Set L


M = M/(x1 , . . . , xn1 )M , and N = N/(x1 , . . . , xn1 )N . Then 0 L

M N 0 is also an exact sequence (by Corollary 4 in 1.1 in your

-regular; therefore, by our proof of


notes) and xn is both L-regular
and N

the case when n = 1, xn is M -regular. This finishes the proof.



Problem 7 Let (R, m) be a noetherian local ring and M a finite R-module. Prove that,
if x1 , . . . , xn is an M -sequence, then
dim(M ) = dim(M/(x1 , . . . , xn )M ) + n.
Proof. Use induction on n.
When n = 1, it is clear that x1 is not contained in any associated prime
of M , in particular x1
/ P for any P Min(AssR (M )) = Min(SuppR (M )).
Consequently,
dim(M/x1 M ) = dim(SuppR (M/x1 M )) dim(Supp(M )) 1 = dim(M ) 1.
On the other hand, Krulls Height Theorem (page 121 in Math614 notes)
tells us that dim(M/x1 M ) dim(M ) 1. Therefore,
dim(M/x1 M ) = dim(M ) 1.
Assume that n 2 and dim(M/(x1 , . . . , xn1 )M ) = dim(M ) (n 1).
= M/(x1 , . . . , xn1 )M
Applying our proof of the case when n = 1 to M
finishes the proof.

Problem 8 Let R be a commutative ring. Suppose that x1 , . . . , xn is an R-sequence
and let I = (x1 , . . . , xn ). Prove that
gr (R)
= R/I[Y1 , . . . , Yn ],
where grI (R) =

i
i=0 I /I i+1

and Y1 , . . . , Yn are indeterminates over R/I.

i
i+1
Proof. Let : R[Y1 , . . . , Yn ] grI (R) =
be defined by Yj 7
i=0 I /I
2
x
j I/I (this is a well-defined ring homomorphism since I i /I i+1 is generated by monomials in x
1 , . . . , x
n of degree i). It is clear that is a
surjection. It remains to prove that ker() = I[Y1 , . . . , Yn ].
It is clear that I[Y1 , . . . , Yn ] ker() (since II i = I i+1 ). We need to
show ker() I[Y1 , . . . , Yn ], i.e., if a polynomial P is contained in ker(),
then all coefficients of P are in I. Since grI (R) is a direct sum, each
homogeneous piece of P is also contained in ker(). Thus, we may assume
that P is homogeneous of degree d. It is clear that (P ) = 0 implies
P (x1 , . . . , xn ) I d+1 . Since x1 , . . . , xn is an R-sequence, Theorem 7 in
1.1 in your notes implies that all coefficients of P are in I. This finishes
the proof.


Vous aimerez peut-être aussi