Sei sulla pagina 1di 5

Devin Smittle Math 8611 Homework Four 16 November 2011

Assumptions/Notes
1. Every ring has a unit. 2. All modules are either left or right modules when needed to make sense in their contexte.g. when dealing with the tensor product M R N of modules M is a right R-module and N is a left R-module. 3. Everything I have written is correct.

10.4.15
Show that tensor products do not commute with direct products in general.

solution
Dene Mi := Z/2i Z for i Z>0 , A := Q Z iZ>0 Mi , and B := iZi>0 (Q Z Mi ). First, notice for any q mi Q Z Mi that q mi = (q/2i ) (2i mi ) = (q/2i ) 0 = 0, so it must be that Q Z Mi = 0 since all its simple tensors are 0. Therefore, B = 0. Now we shall show that A is not trivial. We know A = Q Z iZ>0 Mi /T or( iZ>0 Mi ) since Z is a domain and Q is the fraction eld of Z (see Theorem 4.21 in Keith Conrads notes [1]). Consider the element (m1 , . . .) iZ>0 Mi where each m1 , . . . is odd. If n(m1 , . . .) = (nm1 , . . .) = 0, then 2 | nm1 , 22 | nm2 , . . .; but since each mi is odd 2i | n for every i Z>0 which is not possible, so (m1 , . . .) is not a torsion element meaning iZ>0 Mi is not a torsion module which implies that A is non-trivial (part 2 of Theorem 4.21 [1]). Hence A = B.

10.4.25
Let R be a subring of the commutative ring S and let x be an indeterminate over S. Prove that S[x] and S R R[x] are isomorphic as S-algebras.

solution
Let Ri := (xi ) be the ideal generated by the ith power of x in R[x] and Si := (xi ) in S[x]. We know R[x] = iZ>0 Ri and S[x] = iZ>0 Si are graded rings. We need only to show S R Ri = Si as S-algebras to give a proof since this induces an S-algebra isomorphism between S[x] and S R R[x]. Let : S R Ri Si be the map (rxi ) (r)xi . The kernel is { (rxi ) | r = 0} = {r xi | r = 0} = 0, so is injective. For any xi Si we have the ber of above xi contains xi so is surjective. For R and (rxi ) S R Ri we get (( (rxi ))) = (() (rxi ) = (r)xi = ()xi = ( (rxi )), so is an S-algebra isomorphism.

10.4.26
Let S be a commutative ring containing R (with 1S = 1R ) and let x1 , . . . , xn be independent indeterminates over the ring S. Show that for every ideal I in the polynomial ring R[x1 , . . . , xn ] that S R (R[x1 , . . . , xn ]/I) = S[x1 , . . . , xn ]/IS[x1 , . . . , xn ] as S-algebras.

solution
Let I be any ideal in R[x1 , . . . , xn ]. Dene f : S (R[x1 , . . . , xn ]/I) S[x1 , . . . , xn ]/IS[x1 , . . . , xn ] by s (p + I) sp+IS[x1 , . . . , xn ]. Suppose (s1 , p1 +I) = (s2 , p2 +I) are elements of S(R[x1 , . . . , xn ]/I). Then s1 s2 = 0 and p1 p2 I, so s1 p1 s2 p2 = s1 (p1 p2 ) IS[x1 , . . . , xn ] since s1 S[x1 , . . . , xn ] and p1 p2 I. Thus f is well dened. Now note that f is R-balanced because f(s1 + s2 , p + I) = (s1 + s2 )p + IS[x1 , . . . , xn ] = (s1 p + IS[x1 , . . . , xn ]) + (s2 p + IS[x1 , . . . , xn ]) = f(s1 , p + I) + f(s2 , p + I), f(s, p1 + p2 + I) = s(p1 + p2 ) + IS[x1 , . . . , xn ] = (sp1 + IS[x1 , . . . , xn ]) + (sp2 + IS[x1 , . . . , xn ]) = f(s, p1 +I)+f(s, p2 +I), and f(s, rp+I) = srp+IS[x1 , . . . , xn ] = r(sp+IS[x1 , . . . , xn ]) = rf(s, p+I). Thus we get an induced group homomorphism : S R (R[x1 , . . . , xn ]/I) S[x1 , . . . , xn ]/IS[x1 , . . . , xn ] such that i = f where i is the natural inclusion S(R[x1 , . . . , xn ]/I) SR (R[x1 , . . . , xn ]/I). We can check on simple tensors that it actually denes an S-algebra homomorphismIm pretty sure it does anyway, but I dont want to write it because this is already suering from symbolitis.

Suppose maps the simple tensor s(p+I) to 0. Then sp IS[x1 , . . . , xn ]. This implies p I, so s(p+I) = s 0 = 0, so ker is trivial. Now let f+IS[x1 , . . . , xn ] S[x1 , . . . , xn ]/IS[x1 , . . . , xn ]. Write f+IS[x1 , . . . , xn ] = i si Mi +IS[x1 , . . . , xn ] where Mi is a monic monomial. Then is surjective because it maps i (si Mi + I) to f + IS[x1 , . . . , xn ]. Thus is an isomorphism, which proves the assertion.

10.5.5
Let A1 and A2 be R-modules. Prove that A1 A2 is a at R-module if and only if both A1 and A2 are at More generally, prove that an arbitrary direct sum Ai of R-modules is at if and only if each Ai is at.

solution
Let (Ai ) be an arbitrary collection of R-modules. Suppose each Ai is at. Then for any exact sequence 0 M N we have that 0 Ai R M Ai R N is exact. Thus 1 R f is injective, which implies that i (1 R f) is injectiveif (a1 m1 , . . .) R maps to 0, then each 1R f(ai mi ) must map to 0 which means the each (ai mi ) = 0 since each 1R f is injective. Thus the (1R f) sequence 0 i (Ai R M) i i (Ai R N) is exact. Note we have isomorphisms i (Ai R M) = ( i Ai ) R M ( and (Ai R N) = Ai ) R N denoted by and respectively. We can check that (( 1) f) = ( (1 f))
i i f

1 f

which implies that (

) f must be injective. Thus the sequence


f

i i A i ) R M (

1)R f

i Ai ) R N is exact,

hence i Ai is at. Now assume i Ai is at and let 0 M N be any exact sequence. Then we have the induced homomorphism ( i 1) f is injective. To show each Ai is exact, we need to show each 1 f induced by tensoring the exact sequence with Ai is injective. In a similar way as before, we must have that i (1 f) is injective. Suppose 1 f maps ai mi to 0. Then (0, . . . , ai mi , 0, . . .) is mapped to 0 under i (1 f) so ai mi = 0. Hence each 1 f is injective,
ith coordinate

implying that each Ai is at. Here is a diagram to help see some of this:
0

i (Ai

R M)

i (1R f)

i (Ai

R N)

/(

 ( i A i ) R M

1)R f

/(

 A i ) R N

10.5.7
Let A be a non-zero nite abelian group. (a) Prove that A is not a projective Z-module. (b) Prove that A is not an injective Z-module.

solution
(a) Suppose A is projective. Then there exists an Z-module N such that N A is free. Since A is nite, for any 0 = (0, a) N A there is an non-zero integer m such that m(0, a) = (0, ma) = (0, 0). Thus we have ma = 0 but Z is a domain so either m or a is 0 which is a contradiction. (b) Suppose A is injective. Then, for any n Z, nA = A since Z is a PID. The fundamental theorem of nitely generated abelian groups says A has the form of some nite product i Z/ni Z. Set n := i ni . Then nA = 0 = A, a contradiction.

10.5.9
Assume that R is a commutative ring with 1. (a) Prove that the tensor product of two free R-modules is free. (b) Use (a) to prove that the tensor product of two projective R-modules is projective.

solution
(a) Let A = R and B = l R be two free R-modules. Then AR B = iI R and any direct sum of this formof Ris free.
iI iI

R =

iI

R R =

(b) Let M, N be two projective R-modules. Then there exists R-modules A and B such that A M and B N are free modules. Thus (A M) (B N) = (A B) (M B) (A N) (M N) is free, so M N is a direct summand of a free module and hence is projective.

10.5.12
Let A be an R-module, let I be any non-empty index set and for each i I let Bi be an R-module. Prove the following isomorphisms of abelian groups; when R is commutative prove also that these are R-module isomorphisms. (a) HomR ( (b) HomR (A,
iI

Bi , A) = Bi ) =

iI iI

HomR (Bi , A) HomR (A, Bi ).

iI

solution
(a) Dene the map : HomR ( iI Bi , A) iI HomR (Bi , A) by (f : i Bi A) i (fi : Bi A) where fi : bi f(0, . . . , bi , 0, . . .). We shall show is an R-module homomorphism. For f, g HomR ( iI Bi , A)
ith coordinate

and r R we have (f + g) = i ((fi + gi ) : Bi A) which we can easily see to be i (fi : Bi A) + i (gi : Bi A) = (f) + (g). Similarly, (rf) = i (rfi : Bi A) = r i (fi : Bi A) = r(f). Thus is an R-module homomorphism. Now let i (fi : Bi A) i HomR (Bi , A). We always have a map f HomR ( i Bi , A) such that fi (b) = f(0, . . . , b, 0, . . .) so is surjective. Now suppose f ker . Then each fi in the image of f is 0. For b i Bi write it as the nite sum bi . Then f(b) = f(bi ) = fi (bi ) = 0, so f 0 implying is injective. We have now shown that is an isomorphism. (b) Dene pii : i Bi Bi to be the projection to the ith coordinate, and dene : HomR (A, i Bi ) i HomeR (A, Bi ) by (f : A i Bi ) i (i f). For f, g HomR (A, i Bi ) we have (f+g) = i i (f+g) = i ((i f) + (i g)) i (i f) + i (i g) = (f) + (g). Also, for r R, (rf) = i (i (rf)) = r i (i f) = r(f). Thus is an R-module homomorphism. Like before, we clearly have that is surjective. If f ker then each i f = 0, so it follows directly that f = (1 f, . . .) must also be 0. Thus is an isomorphism.

10.5.16
This exercise proves Theorem 38 that every left R-module M is contained in an injective left R-module. (a) Show that M is contained in an injective Z-module Q. (b) Show that HomR (R, M) HomZ (R, M) HomZ (R, Q). (c) Use the R-module isomorphism M = HomR (R, M) and the exercise 10.5.15 to conclude that M is contained in an injective R-module.

solution
(a) Since M is an abelian group, it is a Z-module. By Corollary 37, it is contained in an injective Z-module Q. (b) Since 1 R any R-module homomorphism : R M is a Z-module homomorphism because (mr) = ((m1)r) = (m1)(r) = m(r). Thus we have HomR (R, M) HomZ (R, M). The inclusion HomZ (R, M) HomZ (R, Q) is true since M Q. (c) By 10.5.15 HomZ (R, Q) is an injective R-module. Since M = HomR (R, M) HomZ (R, Q) we have M is contained in an injective R-module.

10.5.27
Let M, A, and B be R-modules. (a) Suppose f : A M and g : B M are R-module homomorphisms. Prove that X = {(a, b) | a A, b B, f(a) = g(b)} is an R-submodule of the direct sum A B and that there is a commutative diagram X
1 2

/B  /M
g

 A

where 1 and 2 are the natural projections onto the rst and second components. (b) Suppose f : M A and g : M B are R-module homomorphisms. Prove that the quotients Y of A B by {(f (m), g (m)) | m M} is an R-module and that there is a commutative diagram M
f g

/B
2

 A

 /Y

where 1 and 2 are the natural maps to the quotient induced by the maps into the rst and second components.

solution
(a) Let X be the pullback of f and g. We have (0, 0) X, so it is non-empty. Let (a, b), (c, d) X and r R. Then f(a + rc) = f(a) + rf(c) = g(b) + rg(d) = g(b + rd) so (a + rc, b + rd) = (a, b) + r(c, d) X. Be proposition 1, X is an R-submodule. To show the diagram commutes, just see (g 2 )(a, b) = g(b) = f(a) = f 1 (a, b). (b) Let Y be the pushout of f and g . If H = {(f (m), g (m)) | m M} is a submodule of A B then Y is a module since it is the quotient modules. Clearly, (0, 0) H. Let (a, b), (c, d) H and r R. Then (a, b) + r(c, d) = (a + rc, b + rd) = (f (m1 ) + rf (m2 ), g (m1 ) + r(g (m2 ))) = (f(m1 + rm2 ), g(m1 + rm2 )) H. By Proposition 1, H is a subgroup, so Y is a module. Now lets show the diagram commutes; let m M. Then (2 g )(m) = (0, g (m)) + H = (0, g (m)) + (f (m), g (m)) + H = (f (m), 0) + H = (1 f )(m)), so the diagram commutes.

10.5.28
(a) If 0 K P M and 0 K P M are exact sequences of R-modules where P and P are projective, prove P K = P K as R-modules. (b) If 0 M Q L 0 and 0 M Q L injective, prove Q L = Q L as R-modules.

0

are exact sequences of R-modules where Q and Q are

solution
(a) Let X be the pullback for and . Consider the exact sequence 0 ker 1 X 1 P 0 where 1 (a, b) = a; 1 is surjective since, for any a, (a) = (b) for some b since is surjective meaning (a, b) X. Note ker 1 = {(0, b) | (b) = (0) = 0} = ker = K . Since P is projective, the exact sequence splits so X = P K as R-modules. Now consider the exact sequence 0 ker 2 X 2 P 0 where 2 (a, b) = b (2 is surjective for similar reasons as above). In the same way we see ker 2 = ker = K. Since P is projective, X = P K as R-modules. It P K as R-modules. follows P K = (b) The diagram at the end helps make this easier to understand. Let 1 map q Q to (q, 0) and 2 map q Q to (0, q ). Let Y be the pushout of and (see problem statement). Then we have the following sequences
0

inclusion

inclusion

Q 1 Y Y/im1 0 and 0 Q 2 Y Y/im2 0. To see that these are exact, we need to show 1 and 2 are injective. Suppose 1 (q ) = (0, 0) + H (H is dened in my solution of 10.5.27.b). Then (q , 0) = ((m), (m)) for some m. Then (m) = 0 implying m = 0 since is injective. Thus ker 1 is trivial so it is injective. Similarly, 2 is injective. Now let us show that Y/im1 = L. For l L there exists a q Q such that (q) = m. For each l L x a ql such that (ql ) = l. Then dene the map : L Y/im1 by l p (2 (ql )). Now we show is an isomorphism. Suppose (l) = 0 + im1 . Then 2 (ql ) = ((m), (m)) = (0, ql ) for some

p , projection

p, projection

m M. This only possible if m = 0 so ql = 0. Since (ql ) = (0) = 0 we must have had l = 0. Now notice p 2 = . Since the left-hand-side maps onto X/im1 , is surjective, and is injective, we conclude that is an isomorphism, so Y/im1 = L. Similarly, Y/im2 = L . Since Q and Q are injective, our sequences split and we Y = Q L as R-modules. have Q L =
0

 /M

/Q  /Y
2

/L

 Q  L 
0

/ Y/im 1

 Y/im2 
0

References
[1] Keith Conrad, Tensor products, Keith Conrads notes on tensor products, available at http://www.math.uconn.edu/ kconrad/blurbs/linmultialg/tensorprod.pdf.

Potrebbero piacerti anche