1. The
-modules are the (additive) abelian groups, so the simple ones are the simple abelian groups. These are the prime cycles
(in additive notation) where
p is a prime. Hence the semisimple
-modules are the the direct sums of copies of these
for various primes
p. The homogeneous semisimple
-modules are the direct sums of copies of
for a fixed prime
p.
2. a. Since
RR is complemented by Theorem 1, we have
R =
L ⊕
M for some left ideal
M. Write 1 =
e +
f where
e L and
f M. Hence
Re ⊆
L, and we claim this is equality. If
x L then
x −
xe =
xf L ∩
M = 0 because
x −
xe L and
xf M. Hence
x =
xe Re, so
L ⊆
Re, as required.
3. a. Let
α,
β E. Then (
α +
β) ·
x = (
α +
β)(
x) =
α(
x) +
β(
x) =
α ·
x +
β ·
x so Axiom M2 holds (see Section 11.1). Similarly,
proves Axiom M3. The other axioms are routinely verified.
5. Let
N1,
N2, . . .,
Nm be maximal submodules of
M. Define
by
for all
x M. Then
α is
R-linear and
ker(
α) = {
x M x +
Ni = 0 for each
i} = ∩
iNi. Since each
is simple,
M/(∩
iNi) ≅
α(
M) is semisimple by Corollary 1 of Theorem 2.
7. If
M is a finitely generated semisimple module then (Lemma 4)
M is a finite direct sum of simple submodules. By Theorem 2
M =
H1 ⊕
⊕
Hn where the
Hi are the homogeneous components of
M. By the preceding exercise,
end(
M) ≅
end(
H1) ×
×
end(
Hn), so it suffices to show that
end(
H) is semisimple for any finitely generated, homogeneous, semisimple module
H. But then
H is a finite direct sum of isomorphic simple modules (using Lemma 4), and so
H ≅
Kn for some simple module
K. Now Lemma 3 §11.1 gives
end(
H) ≅
Mn(
end K) . Since end
K is a division ring by Schur's lemma, we are done.
9. Let
K be a simple left ideal of
R. If
R is a domain and then
K2 ≠ 0 so Brauer's lemma shows that
K =
Re where
e2 =
e. But
e(1 −
e) = 0 so, again since
R is a domain,
e = 0 or
e = 1 . Since
K ≠ 0 we must have
e = 1, whence
R =
Re is simple as a left
R-module. But then, given 0 ≠
a R, we have
Ra =
R, say
ba = 1 . Again,
b ≠ 0 so
Rb =
R, say
cb = 1 . Now compute:
Hence ab = cb = 1 and we have proved that a is a unit with inverse b. Since a ≠ 0 was arbitrary, this shows that R is a division ring.
11. If
A is an ideal of
R, we must show that every left ideal
of the ring
R/
A is semisimple as an
R/
A-module. But
is an
R-submodule of the left
R-module
R/
A because (
r +
A)(
s +
A) =
rs +
A =
r(
s +
A) . Hence
is semisimple as a left
R -module by Corollary 1 of Theorem 1 (since
RR is semisimple). Since the
R-action and the
R/
A-action on
are the same, it follows that
is semisimple as a left
R/
A-submodule. This is what we wanted.
13. (2) ⇒ (1). Assume that
eRe is a division ring, and that
R is semiprime. Choose 0≠
x Re ; we must show that
Rx =
Re, that is
e Rx. Since
R is semiprime we have
xRx ≠ 0, say
xax ≠ 0,
a R. Since
x Re, we have
xe =
x, so
xe(
ax) ≠ 0 . This means
eax ≠ 0 so
eaxe ≠ 0 (again since
x =
xe) . As
eRe is a division ring, there exists
t eRe such that
t(
eax) =
e. Thus
e Rx, as required.
15. Let
RR be semisimple. Then Lemma 4 implies that
R =
K1 ⊕
K2 ⊕
⊕
Kn where the
Ki are simple left ideals. Hence
R is left noetherian by the Corollary to Lemma 2 §11.1. Since
RR is also semisimple by Theorem 5, the same argument shows that
R is right noetherian. [Note that this also shows that
R is left and right artinian.]
16. Let
R be a semiprime ring.
a. If
LM = 0 where
L and
M are left ideals, then
Since ML is a left ideal and R is semiprime, this gives ML = 0.
c. If rA = 0 then (Ar)2 = ArAr = A0r = 0, Hence Ar = 0 because Ar is a left ideal and R is semiprime. The converse is similar.
17. Write
R =
R1 ×
R2 ×
×
Rn. By Exercise 4, every ideal
A of has the form
A =
A1 ×
A2 ×
×
An where each
Ai an ideal of
Ri. Hence
A2 = 0 if and only if
for each
i. The result follows.
19. If
R is semiprime, let
be an ideal of
Mn(
R) . By Lemma 3 §3.3,
has the form
for some ideal
A of
R. Hence if
then
A2 = 0, whence
A = 0, so
This shows that
Mn(
R) is semiprime for any
n ≥ 1 . Conversely, assume that
Mk(
R) is semiprime for some
k. If
A2 = 0 where
A is an ideal of
R, then
Mk(
A) is an ideal of
Mk(
R) and
Mk(
A)
2 = 0 . Hence
Mk(
A) = 0, whence
A = 0 . This shows that
R is semiprime.
20. a. If
ab = 0 in
R then (
Ra)(
Rb) = 0 because
R is commutative. Hence, if
R is prime either
Ra = 0 or
Rb = 0, that is
a = 0 or
b = 0 . Thus
R is a domain. Conversely, if
R is a domain and
AB = 0 then, if
A ≠ 0 ≠
B, choose 0 ≠
a A and 0 ≠
b B. Then
ab AB = 0, a contradiction. So either
A = 0 or
B = 0 ; that is
R is a prime ring.
21. Write
P =
P1 ⊕
P2 ⊕
⊕
Pn, and view
P as an internal direct sum. Then each
Pi is isomorphic to a direct summand of a free module
Fi ; by Lemma 9 we may assume that
Fi =
Pi ⊕
Qi. Then
F =
F1 ⊕
F2 ⊕
⊕
Fn is also free, and
F = (
P1 ⊕
P2 ⊕
⊕
Pn) ⊕ (
Q1 ⊕
Q2 ⊕
⊕
Qn) . Hence
P1 ⊕
P2 ⊕
⊕
Pn is projective by Theorem 3.
22.
a. If α : M → N is R-linear and K ⊆ M is simple then either α(K) = 0 or γ(K) ≅ K by Schur's lemma. Either way, α(K) ⊆ soc N. Since soc M is a sum of simple submodules of M, it follows that α(soc M) ⊆ soc N.
c. If
M =
N1 ⊕
N1 it is clear that
soc(
N1)⊕
soc(
N2) ⊆
soc M because
soc(
Ni) is semisimple for each
i. For the other inclusion, define the projection
π1 :
M →
N1 by
π1(
n1 +
n2) =
n1, and define
π2 :
M →
N2 analogously. If
K ⊆
M is any simple module then
πi(
K) ⊆
socNi for each
i [
πi(
K) is either 0 or isomorphic to
K by Schur's lemma]. Hence if
k K then
k =
π1(
k)+
π2(
k)
soc(
Ni)⊕
soc(
Ni) . It follows that