Scribd
Upload
14 views4 pages

tensor product problems

The document discusses various mathematical claims and proofs related to modules, tensor products, and localization in algebra. It establishes that Q/Z ⊗ Q/Z is isomorphic to 0, and demonstrates isomorphisms for products of fields and modules. Additionally, it explores properties of flat and projective modules, localization, and the behavior of ideals under localization.

Uploaded by

Chudamani Satnami
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
14 views4 pages

tensor product problems

The document discusses various mathematical claims and proofs related to modules, tensor products, and localization in algebra. It establishes that Q/Z ⊗ Q/Z is isomorphic to 0, and demonstrates isomorphisms for products of fields and modules. Additionally, it explores properties of flat and projective modules, localization, and the behavior of ideals under localization.

Uploaded by

Chudamani Satnami
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

Problem 1.

We claim that Q/Z ⊗ Q/Z ≃ 0.


Indeed, Q/Z ⊗Z Q/Z is generated by the tensors α ⊗ β, where α, β ∈ Q/Z.
Write pq and m
n
with integer p, q, m, n for representatives of α, β in Q. We have
p m np m p m p m p
⊗ = ⊗ =n ⊗ = ⊗n = ⊗ m.
q n nq n nq n nq n nq
p
But m = 0 in Q/Z hence nq ⊗ m = 0, therefore any generator of Q/Z ⊗ Q/Z
is 0 hence the module is 0.
We claim that the product map
ϕ : Q ⊗Z Q → Q, α ⊗ β 7→ αβ
is an isomorphism. Indeed, similarly to the previous computation we have
p p
q
⊗m n
= nq ⊗ m = pm
nq
⊗ 1 i.e. any bilinear map Q/Z × Q/Z → M factors
through ϕ.
We have
ϕ
C ⊗R C ≃ R[x]/(x2 + 1) ⊗R C →
− C[x]/(x2 + 1) ≃ C/((x + i)(x − i)) ≃ C ⊕ C,
the last isomorphism following from the Chinese remainder theorem and the
map ϕ given by [f (x)] ⊗ α 7→ [αf (x)]. The latter is correctly defined and as
x2 + 1|f ⇔ x2 + 1|αf and has an inverse given in R-basis by
[1] 7→ [1] ⊗ 1; [i] 7→ [1] ⊗ i; [x] 7→ [x] ⊗ 1; [ix] 7→ [x] ⊗ i.
Therefore ϕ is an isomorphism, hence C ⊗R C ≃ C ⊕ C.
We claim that ϕ : Q[x] ⊗Q C → C[x] given by f (x) ⊗ α 7→ αf (x) is an
isomorphism. Indeed, the inverse is given by αxi 7→ xi ⊗ α.
Problem 2.1.
L
Given a short exact sequence 0 → P → Q → S → 0, tensoring with i∈I Mi
we obtain the direct sum of sequences 0 → P ⊗ Mi → Q ⊗ Mi → S ⊗ Mi → 0,
whose exactness is equivalent to exactness of each of the summand, therefore
the assertion follows.
Problem 2.2. Applying the previous problem to the free rank 1 modules
Mi = R we deduce that any free module is flat. Applying the problem again
we deduce that any direct summand of a free module is flat, i.e. any projective
module is flat.
Problem 2.3.
Consider R = Z and take N = Q.
Recall that for any Z-module V we can explicitely describe V ⊗Z Q as a
localization, namely V ⊗ Q = VQ := { vr |v ∈ V, r ∈ Z}/ ∼, where ∼ is the
1
2

equivalence relation
v v′
∼ ′ ⇔ ∃n ∈ Z : n(r′ v − rv ′ ) = 0.
r r
Now let us firstly verify that N is flat. Given an injective map ϕ : M ,→ M ′
of Z-modules, suppose that the induced map ϕQ : M ⊗ Q → M ′ ⊗ Q is not
injective. Take a non-zero element mq ∈ Ker(ϕQ ). As mq is nonzero, then
m ∈ M is non-torsion, otherwise if rm = 0 then m ⊗ 1q = rm ⊗ rq1 = 0. By
injectivity of ϕ it follows that ϕ(m) ∈ M ′ is non-torsion as well. But then ϕ(m)
1
is a non-zero element of MQ′ , thus m ⊗ 1 and hence mq cannot belong to the
kernel of ϕQ , a contradiction.
We are left to verifying that N is not projective. Indeed, otherwise it would
be a direct summand of Z⊕I but this module does not contain any non-trivial
divisible elements.
Problem 3.
To extend α set α( rs ) := α(s)−1 α(r), it follows from the construction of the
localization that this is correctly defined. Now to verify uniqueness of the
extension use that r = s · rs ∈ S −1 R, therefore it should satisfy
r
α(r) = α(s)α( ) ∈ R′ .
s
′
As α(s) is invertible in R , multiplying by its inverse we have
r
α( ) = α(s)−1 α(r).
s
Problem 4a.
Pick some elements mi ∈ Mi . We have
m1 + m2 m1 m2
= + ,
s s s
hence S −1 (M1 + M2 ) ⊂ S −1 M1 + S −1 M2 . Also we have
m1 m2 s2 m1 + s1 m2
+ = ,
s1 s2 s1 s2
hence S −1 M1 + S −1 M2 ⊂ S −1 (M1 + M2 ).
Problem 4b.
If m ∈ M1 ∩ M2 then ms lies in both S −1 M1 and S −1 M2 , hence
S −1 (M1 ∩ M2 ) ⊂ S −1 M1 ∩ S −1 M2
and this inclusion holds for arbitrary intersections. Now consider
m1 m2
v= = ∈ S −1 M1 ∩ S −1 M2 .
s1 s2
3

By the definition of localization there is s ∈ S such that


s(s2 m1 − s1 m2 ) = ss2 m1 − ss1 m2 = 0 ∈ M,
in particular ss2 m1 ∈ M1 ∩ M2 and hence v = ss 2 m1
ss2 s1
∈ S −1 (M1 ∩ M2 ) which
proves the converse inclusion.
For a counterexample consider Mi = (i) ⊂ Z as Z-modules and S = Z \ {0}.
We have ∩i Mi = {0}, however S −1 Mi = Q = S −1 Z and ∩i S −1 Mi = Q.
Problem 4c.
√ n
Pick an element a ∈ I. We have an ∈ I for some n, hence ( as )n = asn ∈ S −1 I,
√ √
therefore S −1 I ⊂ S −1 √I. n
Pick an element sb ∈ S −1 I. We have ( sb )n ∈ S −1 I for some n, i.e. sbn =
a
r
∈ S −1 R for some a ∈ I, r ∈ S. Therefore
t(rbn − sn a) = 0 ∈ R
√
for some t ∈ S, in particular trbn ∈ I and sb = trs trb
∈ S −1 I as (trb)n =
(tr)n−1 trbn ∈ I. This proves the converse inclusion.
Problem 5a.
Suppose I ⊂ J, i.e. there is an injective map I ,→ J ⊂ R of R-modules. As
localization is exact, the map IP → JP is injective as well.
Now suppose for any maximal ideal P ⊂ R the ideal IP is contained in JP .
We need to verify that I ⊂ J i.e. that the composite I ,→ R → R/J vanishes.
Consider the following diagram:

I R/J
(0.1)
Q 0 Q
P IP P (R/J)P

where the lower map is 0 being the composite of


Y Y Y
IP → RP → RP /JP = (R/J)P .

The last equality is the natural identification following from exactness of lo-
calization. Q
Now we use that for any module M the natural map M → P MP is
injective. In particular so is the right vertical arrow of the diagram 0.1 implying
that the upper horizontal arrow is 0.
Problem 5b.
Applying Problem 4c to the zero ideal we obtain that N il(R)P = N il(RP ).
Now apply the previous problem to the nilradical of R.
4

Problem 5c.
Consider the ring Q × Q. It is not an integral domain as (0, 1) · (1, 0) = 0,
but the localizations at maximal ideals (Q, 0) and (0, Q) are integral domains
being Q. Geometrically localization detects local properties at the point and
does not know about connected components which do not contain this point.

You might also like