10.

4 Tensor Products of Modules

Jose L. Vanegas

10.4.1
Let f : R → S be a ring homomorphism from the ring R to the ring S with
f (1R ) = 1S . Verify that the relation sr = sf (r) defines a right R-action on S,
under which S is an (S, R)-bimodule.

Solution
We first show that this right action gives S an R-module structure. We verify
directly that:

(s + s′ )r = (s + s′ )f (r) = sf (r) + s′ f (r) = sr + s′ r,

and

s(r + r′ ) = sf (r + r′ ) = s(f (r) + f (r′ )) = sf (r) + sf (r′ ) = sr + sr′ .

Also,

(sr)r′ = sf (r)r′ = sf (r)f (r′ ) = sf (rr′ ) = s(rr′ ).

This proves that S has the structure of a right R-module.

Next, we verify that the left action of S is compatible with the right action
of R. This is immediate from the associativity of multiplication in S:

(s′ s)r = (s′ s)f (r) = s′ (sf (r)) = s′ (sr).

Thus, S has an (S, R)-bimodule structure.

10.4.2
Show that the element 2 ⊗ 1 is 0 in Z ⊗Z Z/2Z but is nonzero in 2Z ⊗Z Z/2Z.

Solution
This arises from the fact that 2 is divisible by 2 in Z but not in 2Z. In particular,
in Z ⊗Z Z/2Z, we have:

2 ⊗ 1 = (1 · 2) ⊗ 1 = 1 ⊗ (2 · 1) = 1 ⊗ 2 = 1 ⊗ 0 = 0.

1
To prove that 2 ⊗ 1 is nonzero in the second tensor product, consider the group
homomorphism ϕ : 2Z ⊗Z Z/2Z → Z/2Z defined by:

ϕ(2k ⊗ x) = kx,

where k ∈ Z and x ∈ Z/2Z. To see that this is a group homomorphism, it

suffices to show bilinearity in k and x. Let a, b, k, k ′ ∈ Z and x, x′ ∈ Z/2Z.
Then we compute:

ϕ(a(2k) + b(2k ′ ), x) = ϕ(2(ak + bk ′ ), x)

= (ak + bk ′ )x
= akx + bk ′ x
= aϕ(2k, x) + bϕ(2k ′ , x).

Similarly,

ϕ(2k, ax + bx′ ) = k(ax + bx′ )

= akx + bkx′
= aϕ(2k, x) + bϕ(2k, x′ ).

Thus, ϕ is linear in each coordinate, confirming it as a valid homomorphism.

Notably, ϕ(2 ⊗ 1) = 1 ̸= 0, indicating that 2 ⊗ 1 is not in the kernel of ϕ.
Hence, 2 ⊗ 1 is nonzero as desired.

10.4.3
Show that C ⊗R C and C ⊗C C are both left R-modules but are not isomorphic
as R-modules.

Solution
For this problem, we utilize the natural R-module structure obtained by consid-
ering R as a subring of C. First, we establish a general result: if R is an integral
domain contained in an integral domain S, then s ⊗ s′ ∈ S ⊗R S is nonzero if
and only if s and s′ are both nonzero. This follows by defining a bilinear map
s ⊗ s′ 7→ ss′ and observing that s ⊗ s′ is not in the kernel precisely when s and
s′ are both nonzero.
As noted, we have R ⊗R R ∼ = R for any ring R, leading to C⊗C ∼ = C, which
is a 1-dimensional vector space over C. This implies it is a 2-dimensional vector
space over R.
We will now demonstrate that C ⊗R C is a 4-dimensional vector space over
R, hence not isomorphic to C⊗C C. Consider the simple tensors 1⊗1, 1⊗i, i⊗1,
and i ⊗ i. These span C ⊗R C since any simple tensor can be expressed as:

(a + bi) ⊗ (c + di) = ac(1 ⊗ 1) + ad(1 ⊗ i) + bc(i ⊗ 1) + bd(i ⊗ i).

2
Now consider the map from C × C to R4 defined by:

(a + bi, c + di) 7→ (ac, ad, bc, bd).

This map is R-bilinear, inducing an R-module homomorphism from C ⊗R C to

R4 . The mapping is surjective, as the image of the described simple tensors
yields the standard basis for R4 . The inverse mapping can be constructed by
sending the standard basis vectors back to the respective simple tensors. Thus,
the composition of these mappings is the identity in both directions, establishing
isomorphisms.
Consequently, C ⊗R C is indeed 4-dimensional over R.

10.4.4
Show that Q ⊗Z Q and Q ⊗Q Q are isomorphic left Q-modules. [Show they are
both 1-dimensional vector spaces over Q.]

Solution
Similar to the previous problem, we know that Q ⊗Q Q = ∼ Q. In particular, all
simple tensors a ⊗ b can be written as 1 ⊗ ab, so all tensors are simple and of the
form 1 ⊗ q for some q ∈ Q. Thus, 1 ⊗ 1 forms a basis for Q ⊗Q Q as a Q-vector
space.
We claim that 1 ⊗ 1 is also a basis for Q ⊗Z Q. First, we show that every
element in Q ⊗Z Q can be expressed as 1 ⊗ q for q ∈ Q. Given an arbitrary
simple tensor ab ⊗ dc , we have:
a c a c
⊗ = ⊗b·
b d b db
ab c
= ⊗
b db
c
=a⊗
db
ac
=1⊗ .
db
Thus, every simple tensor can be written in the form 1 ⊗ q.
Next, we argue that 1 ⊗ q is nonzero exactly when q is nonzero. To show
this, we define a map ϕ : 1 ⊗ q 7→ q from the tensor product to Q. This map is
Z-linear, and it follows that 1 ⊗ q is in the kernel only if q = 0. Therefore, the
only element of Q ⊗Z Q that is zero is 1 ⊗ 0. Hence, the result follows.

10.4.5
Let A be a finite abelian group of order n and let pk be the largest power of
the prime p dividing n. Prove that Z/pk Z ⊗Z A is isomorphic to the Sylow
p-subgroup of A.

3
Solution
Since A is a finite abelian group, we can write it as a direct sum of cyclic groups
A ∼= C1 ⊕ C2 ⊕ · · · ⊕ Cm , where each Ci is a cyclic group of order ai . By
properties of the tensor product, we have:
m
!
∼ Z/p Z ⊗Z
M
k k
Z/p Z ⊗Z A = Ci
i=1
m
∼
M
Z/pk Z ⊗Z Ci


=
i=1
m
∼
M
Z/pk Z ⊗Z Z/ai Z .


=
i=1

But Z/mZ ⊗Z Z/nZ is isomorphic to Z/(m, n)Z. Therefore, terms in the

direct sum for which ai is not divisible by p vanish. We conclude that:
m
Z/pk Z ⊗Z A ∼
M
= Z/pαi Z,
i=1

where αi is the largest power of p dividing ai for all i. This is precisely the
direct sum of the Sylow p-subgroup for each Ci , which is the same as the Sylow
p-subgroup for A overall.

10.4.6
If R is any integral domain with a quotient field Q, prove that (Q/R)⊗R (Q/R) =
0.

Solution
We will show that all simple tensors are zero. Let ab ⊗ dc be an arbitrary element
of the tensor product, where a, b, c, d ∈ R. Then we have:
a c a c
⊗ = ⊗b·
b d b db
ab c
= ⊗
b db
c
=a⊗
db
c
=0⊗
db
= 0.

Hence, all simple tensors vanish, and we are done.

4
10.4.7
If R is any integral domain with quotient field Q and N is a left R-module,
prove that every element of the tensor product Q ⊗R N can be written as a
simple tensor of the form (1/d) ⊗ n for some nonzero d ∈ R and some n ∈ N .

Solution
Note that simple tensors clearly have this form since we can write ad ⊗n = d1 ⊗an.
Every element of Q⊗R N is a finite sum of simple tensors, which may be written
as
k
X 1
⊗ ni .
d
i=1 i
Q
Letting d be the product of all di and defining ai = j̸=i di , we have that:
k k
X 1 X ai
⊗ ni = ⊗ ni
d
i=1 i i=1
d
k
X 1
= ⊗ ai ni
i=1
d
k
1 X
= ⊗ ai ni .
d i=1

This proves the desired result.

10.4.8
Suppose R is an integral domain with quotient field Q and let N be any R-
module. Let U = R× be the set of nonzero elements in R and define U −1 N to
be the set of equivalence classes of ordered pairs of elements (u, n) with u ∈ U
and n ∈ N under the equivalence relation (u, n) ∼ (u′ , n) if and only if u′ n = un′
in N .
(a) Prove that U −1 N is an abelian group under the addition defined by
(u1 , n1 ) + (u2 , n2 ) = (u1 u2 , u2 n1 + u1 n2 ).
Prove that r(u, n) = (u, rn) defines an action of R on U −1 N , making it
into an R-module. [This is an example of localization considered in general
in Section 4 of Chapter 15; cf. also Section 6 in Chapter 7.]
(b) Show that the map from Q × N to U −1 N defined by sending (a/b, n) to
(b, an) for a ∈ R, b ∈ U , n ∈ N , is an R-balanced map, so it induces a
homomorphism f from Q ⊗R N to U −1 N . Show that the map g from
U −1 N to Q ⊗R N defined by g((u, n)) = (1/u) ⊗ n is well-defined and
is an inverse homomorphism to f . Conclude that Q ⊗R N ∼ = U −1 N as
R-modules.

5
 (c) Conclude from (b) that (1/d) ⊗ n is 0 in Q ⊗R N if and only if rn = 0 for
some nonzero r ∈ R.
(d) If A is an abelian group, show that ⊗A = 0 if and only if A is a torsion
abelian group (i.e., every element of A has finite order).

Solution
xxx

10.4.9
Suppose R is an integral domain with the quotient field Q and let N be any R-
module. Let Q⊗R N be the module obtained from N by extension of scalars from
R to Q. Prove that the kernel of the R-module homomorphism ι : N → Q ⊗R N
is the torsion submodule of N (cf. Exercise 8 in Section 1). [Use the previous
exercise.]

Solution
xxx

10.4.10
Suppose R is commutative and N ∼ = Rn is a free R-module of rank n with
R-module basis e1 , . . . , en .

(a) For any nonzero R-module MP , show that every element of M ⊗ N can be
n
written
Pn uniquely in the form i=1 mi ⊗ ei where mi ∈ M . Deduce that if
i=1 m i ⊗ e i = 0 in M ⊗ N , then mi = 0 for i = 1, . . . , n.
P
(b) Show that if mi ⊗ ni = 0 in M ⊗ N where ni are merely assumed to
be R-linearly independent, then it is not necessarily true that all mi are
0. [Consider R =, n = 1, M = /2 and the element 1 ⊗ 2.]

Solution
xxx

10.4.11
Let {e1 , e2 } be a basis of V =2 . Show that the element e1 ⊗ e2 + e2 ⊗ e1 in V ⊗V
cannot be written as a simple tensor v ⊗ w for any v, w ∈2 .

Solution
xxx

6
10.4.12
Let V be a vector space over the field F and let v, v ′ be nonzero elements of V .
Prove that v ⊗ v ′ = v ′ ⊗ v in V ⊗F V if and only if v = av ′ for some a ∈ F .

Solution
xxx

10.4.13
Prove that the usual dot product of vectors defined by letting (a1 , . . . , an ) · · · (b1 , . . . , bn )
be a1 b1 + · · · + an bn is a bilinear map from Rn × Rn to .

Solution
xxx

10.4.14
Let I be an arbitrary nonempty index set and for each i ∈ I, let Ni be left
R-modules. Let M be a right R-module. Prove the group isomorphism:
!
Ni ∼
M M
M⊗ = (M ⊗ Ni ),
i∈I i∈I

where the direct sum of an arbitrary collection of modules is defined in Exercise

20, Section 3. [Use the same argument as for the direct sum of two modules,
taking care to note where the direct sum hypothesis is needed — cf. the next
exercise.]

Solution
xxx

10.4.15
Show that tensor products do not commute with direct products in general.
[Consider the extension of scalars from Z to Q of the direct product of the
modules Mi = /2i , i = 1, 2, . . .].

Solution
xxx

7
10.4.16
Suppose R is commutative and let I and J be ideals of R, so R/I and R/J are
naturally R-modules.

(a) Prove that every element of R/I ⊗R R/J can be written as a simple tensor
of the form (1 mod I) ⊗ (r mod J).
(b) Prove that there is an R-module isomorphism R/I ⊗R R/J ∼
= R/(I + J)
mapping (r mod I) ⊗ (r′ mod J) to rr′ mod (I + J).

Solution
xxx

10.4.17
Let I = (2, x) be the ideal generated by 2 and x in the ring R = [x]. The ring
/2 = R/I is naturally an R-module annihilated by both 2 and x.
(a) Show that the map ϕ : I × I → /2 defined by
a0
ϕ(a0 + a1 x + · · · + an xn , b0 + b1 x + · · · + bm xm ) = b1 mod 2
2
is R-bilinear.

(b) Show that there is an R-module homomorphism from I⊗R I → /2 mapping

p(x) ⊗ q(x) to p(0) ′ ′
2 q (0) where q denotes the usual polynomial derivative
of q.
(c) Show that 2 ⊗ x ̸= x ⊗ 2 in I ⊗R I.

Solution
xxx

10.4.18
Suppose I is a principal ideal in the integral domain R. Prove that the R-
modules I ⊗R I have no nonzero torsion elements (i.e., if rm = 0 with 0 ̸= r ∈ R
and m ∈ I ⊗R I, then m = 0).

Solution
xxx

8
10.4.19
Let I = (2, x) be the ideal generated by 2 and x in the ring R = [x] as in
Exercise 17. Show that the nonzero element 2 ⊗ x − x ⊗ 2 in I ⊗R I is a torsion
element. Show in fact that 2 ⊗ x − x ⊗ 2 is annihilated by both 2 and x, and
that the submodule of I ⊗R I generated by 2 ⊗ x − x ⊗ 2 is isomorphic to R/I.

Solution
xxx

10.4.20
Let I = (2, x) be the ideal generated by 2 and x in the ring R = [x]. Show that
the element 2 ⊗ 2 + x ⊗ x in I ⊗R I is not a simple tensor, i.e., cannot be written
as a ⊗ b for some a, b ∈ I.

Solution
TODO

10.4.21
Suppose R is commutative and let I and J be ideals of R.

(a) Show that there is a surjective R-module homomorphism from I ⊗R J to

the product ideal IJ mapping I ⊗ J to the element ij.
(b) Give an example to show that the map in (a) need not be injective (cf.
Exercise 17).

Solution
xxx

10.4.22
Suppose that m is a left and a right R-module such that rm = mr for all r ∈ R
and m ∈ M . Show that the elements 1 r2 and r2 r1 act the same on M for every
r1 , r2 ∈ R. (This explains why the assumption that R is commutative in the
definition of an R-algebra is a fairly natural one.)

Solution
xxx

9
10.4.23
Verify the details that the multiplication in Proposition 19 makes A ⊗R B into
an R-algebra.

Solution
xxx

10.4.24
Prove that the extension of scalars from to the Gaussian integers [i] of the ring
is isomorphic to as a ring: [i]⊗∼
= as rings.

Solution
xxx

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
xxx

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 ].

Solution
xxx

10.4.27
The next exercise shows the ring C⊗R introduced at the end of this section is
isomorphic to ×. One may also prove this via Exercise 26 and Proposition 16
in Section 9.5, since ∼
= [x]/(x2 + 1). The ring C× is also discussed in Exercise
23 of Section 1.

10
 (a) Write down a formula for the multiplication of two elements

a · 1 + b · e2 + c · e3 + d · e4

and
a′ · 1 + b′ · e2 + c′ · e3 + d′ · e4
in the example A = ⊗ following Proposition 21 (where 1 = 1 ⊗ 1 is the
identity of A).
(b) Let ϵ1 = 21 (1 ⊗ 1 + i ⊗ i) and ϵ2 = 12 (1 ⊗ 1 − i ⊗ i). Show that ϵ1 ϵ2 = 0,
ϵ1 + ϵ2 = 1, and ϵ2j = ϵj for j = 1, 2 (where ϵ1 and ϵ2 are called orthogonal
idempotents in A). Deduce that A is isomorphic as a ring to the direct
product of two principal ideals: A ∼ = Aϵ1 × Aϵ2 (cf. Exercise 1, Section
7.6).

(c) Prove that the map ϕ : × → × defined by ϕ(z1 , z2 ) = (z1 z2 , z1 z2 ), where

z2 denotes the complex conjugate of z2 , is an -bilinear map.
(d) Let Φ be the -module homomorphism from A to × obtained from ϕ in (c).
Show that Φ(ϵ1 ) = (0, 1) and Φ(ϵ2 ) = (1, 0). Show also that Φ is -linear,
where the action of is on the left tensor factor in A and on both factors in
×. Deduce that Φ is surjective. Show that Φ is a -algebra isomorphism.

Solution
xxx

11