Honors Algebra 4, MATH 371 Winter 2010

Solutions 2
1. Let R be a ring.

(a) Let I be an ideal of R and denote by π : R → R/I the natural ring homomorphism
defined by π(x) := x mod I (= x + I using coset notation). Show that an arbitrary ring
homomorphism φ : R → S can be factored as φ = ψ ◦ π for some ring homomorphism
ψ : R/I → S if and only if I ⊆ ker(φ), in which case ψ is unique.
(b) Suppose that R is commutative with 1. An R-algebra is a ring S with identity equipped
with a ring homomorphism φ : R → S mapping 1R to 1S such that im(φ) is contained
in the center of S (i.e. the set

c(S) := {z ∈ S | zs = sz for all s ∈ S}

of all elements of S that commute with every other element). If (S, φ) and (S 0 , φ0 ) are
two R-algebras then a ring homomorphism f : S → S 0 is called a homomorphism of
R-algebras if f (1S ) = 1S 0 and f ◦ φ = φ0 . For an R-algebra (S, φ) we will frequently
simply write rx for φ(r)x whenever r ∈ R and x ∈ S.
Prove that the polynomial ring R[X] in one variable is naturally an R-algebra, and that
if S is an R-algebra then for any s ∈ S there exists a unique R-algebra homomorphism
f : R[X] → S such that f (X) = s. In other words, mapping R[X] to S is the “same” as
choosing an element s of S.

Solution:

(a) One direction is obvious. For the other direction, assume that I ⊆ ker(φ) and define
ψ : R/I → S by the rule
ψ(r + I) := φ(r).
Note that this is well-defined since it doesn’t depend on the choice of coset representative
as φ(I) = 0. Clearly φ = ψ◦π and if ψ 0 : R/I → S is another ring map with this property
then we must have ψ = ψ 0 as π is surjective. Hence ψ is unique.
(b) That R[X] is an R-algebra via the map R → R[X] sending r ∈ R to the constant
polynomial r ∈ R[X] is obvious. If S is any R-algebra and s ∈ S, we define f : R[X] → S
as
f (a0 + a1 X + a2 X 2 + · · · + an X n ) := a0 + a1 s + · · · an sn .
It is easy to check that f is an R-algebra homomorphism. On the other hand, if f :
R[X] → S is any homomorphism of R-algebras with f (X) = s then we must have
f (X n ) = f (X)n = sn and hence

f (a0 + a1 X + a2 X 2 + · · · + an X n ) = f (a0 ) + f (a1 )s + · · · + f (an )sn = a0 + a1 s + · · · an sn .

 We conclude that f exists and is uniquely determined by the requirement that f (X) = s.
2. Let R be a ring with 1.
(a) Prove that there is a unique map of rings fR : Z → R. Conclude that every ring with 1
is a Z-algebra in a unique way.
(b) For a ring R with 1, the kernel of the ring homomorphism fR as in (2a) is an ideal of
Z so it has the form c(R)Z for a unique c(R) ∈ Z satisfying c(R) ≥ 0. By definition,
the characteristic of R is this integer c(R). Convince yourself that when c(R) > 0, this
number is the least number of times we have to add 1 ∈ R to itself to get 0 ∈ R. Now
prove that if R is a ring with 1 that is an integral domain, then the characteristic of R
is either 0 or a prime number.
(c) Prove that for g : R → S a homomorphism of rings with 1 taking 1R to 1S the charac-
teristic of S divides the characteristic of R.
(d) Let g : R → S be a homomorphism of rings with 1 taking 1R to 1S . If g is injective,
prove that c(R) = c(S). Give an example with g not injective where c(R) 6= c(S).
Solution:
(a) In general, one wants maps of rings with 1 to take 1 to 1, but I should have explicitly
demanded this. In this situation, for n > 0
f (n) = f (1) + f (n − 1) = 1 + f (n − 1)
and it follows by induction that f (n) for n > 0 is uniquely determined. Using the
existence of additive inverses in R, we must have f (0) = 0 as f (0) = f (0 + 0) =
f (0) + f (0). We conclude that for n > 0 we have
0 = f (0) = f (n + (−n)) = f (n) + f (−n)
and hence that f (−n) = −f (n) is again uniquely determined. Thus, there is a unique
map of rings Z → R (provided we require 1 maps to 1).
(b) In any case, we have an injective homomorphism of rings
Z/c(R)Z ,→ R.
If R is a domain then so is Z/c(R)Z since any subring of a domain is a domain and
it follows that (c(R)) must be a prime ideal. Hence either c(R) = 0 or it is a prime
number.
(c) The composite homomorphism
fR
Z /R /S

coincides with fS by uniqueness and hence ker(fR ) ⊆ ker(fS ) as desired.

 (d) When g : R → S is injective, the composite

 
Z/c(R)Z  / R
fR
/S

is also injective and we deduce that c(S) := ker(fS ) = c(R). As a counterexample to

this equality when g fails to be injective, consider the quotient map Z → Z/pZ.

3. Let I and J be ideals of a ring R. We define

(a) I + J := {a + b | a ∈ I, b ∈ J}
(b) IJ := {a1 b1 + · · · + as bs | a ∈ I, b ∈ J}

Prove that I +J is the smallest ideal of R containing I and J and that IJ is an ideal contained
in the intersection I ∩ J. Convince yourself that I ∩ J is an ideal of R, and show that if R is
commutative and I + J = R then IJ = I ∩ J. Show by giving examples that IJ 6= I ∩ J in
general, and that I ∪ J (set-theoretic union) need not be an ideal.
Solution: It is easy to see that I + J is an ideal of R. If K is any ideal of R containing I
and J then it contains a for all a ∈ I and b for all b ∈ J and hence a + b. Thus, K contains
I + J.
We obviously have IJ ⊆ I ∩ J. To get the reverse inclusion, we have to require that 1 ∈ R
(this should have been stated as an assumption in the problem). Suppose that r ∈ I ∩ J and
write 1 = i + j for i ∈ I and j ∈ J. Then r = ri + rj lies in IJ. As for counterexamples,
consider the ring R = 2Z which does not have an identity and the ideals I = 6Z and J = 8Z.
These ideals clearly satisfy I + J = R. We have I ∩ J = 24Z but IJ = 48Z. Now consider
2Z and 3Z as ideals of Z. Their set-theoretic union contains 2 and 3 but not 2 + 3 = 5 since
5 isn’t a Z-multiple of either 2 or 3.

4. Let R be a commutative ring and I, J ideals of R. If P is a prime ideal of R containing IJ,

prove that P contains I or P contains J.
Solution: Suppose that P does not contain I and let j ∈ J be arbitrary. Since P does not
contain I, there exists i ∈ I with i 6∈ P . But ij ∈ P whence j ∈ P as P is prime. Hence P
contains J.

5. Let R be a commutative ring.

(a) Show that the set of all nilpotent elements of R ( called the nilradical of R) is an ideal.
Hint: this is basically 1(b) from assignment 1, but be careful about showing that this
set is really an abelian group under addition.
(b) Prove that the nilradical of R is contained in the intersection of all prime ideals of R.
 (c) Let G := Z/pZ as a group under addition (it is cyclic of order p). Let Fp := Z/pZ as a
ring, and note that this is a field with p elements. Let R be the group ring R := Fp G.
What is the nilradical of R?

Solution:

(a) Using assignment 1, it remains to show that if x is nilpotent then so is −x. Note that
for any r ∈ R we have

0 = 0 · r = (x + (−x))r = xr + (−x)r

so (−x)r = −xr. We deduce that

(
xn n ∈ 2Z
(−x)n =
−xn else

and hence that −x is nilpotent of x is. Note that we don’t need to assume that R has
an identity.
(b) If x ∈ R satisfies xn = 0 for n > 1 and P is a prime ideal then xn = x · xn−1 ∈ P so by
induction x ∈ P . It follows that x lies in the intersection of all prime ideals.
(c) Arguing as in assignment 1, we have an isomorphism of rings

Fp [X]/(xp − 1) = Fp G.

But as polynomials over Fp we have xp − 1 = (x − 1)p so our task is to find the nilradical
of Fp [X]/(x − 1)p . In other words, we seek to find all f ∈ Fp [X] such that f k ∈ (x − 1)p
for some k. Since (x − 1) is a prime ideal of Fp [X], we conclude that we must have
f ∈ (x − 1)i for some i ≥ 1 and hence the nilradical is precisely the principal ideal
generated by (x − 1).

6. Let R be a commutative ring. Prove that the set of prime ideals in R has minimal elements
with respect to inclusion. Such minimal elements are called minimal primes.
Solution: This exercise should require R to have an identity 1 6= 0. Let S be the set of prime
ideals of R, ordered by inclusion. Since R is not the zero ring, R has at least one maximal
(hence prime) ideal so S is nonempty. Suppose that I is any totally ordered set and that
{Pi }i∈I is a chain in S. We claim that
\
P := Pi
i∈I

is a prime ideal of R. It is clearly an ideal, so suppose that ab ∈ P . Then for all i, either
a ∈ Pi or b ∈ Pi . If a 6∈ Pi for some i ∈ I, then a 6∈ Pj for all j ≤ i as Pj ⊆ Pi and hence
 b ∈ Pj for all j ≤ i. As we must also then have b ∈ Pj for all j ≥ i we deduce that b ∈ P and
P is prime. Thus, every chain in S is bounded below and we conclude by Zorn’s Lemma (in
the form with minimal elements) that S has minimal elements, as desired.

7. Let R be a finite (as a set) commutative ring with 1. Prove that every prime ideal of R is
maximal.
Solution: Let P be a prime ideal of R. Then R/P is a domain with finitely many elements,
and is hence a field. (Indeed, if x ∈ R/P is nonzero then the powers of x can not all be
distinct by finiteness so xj = xj for some 0 < i < j and we conclude that xj−i (xi − 1) = 0 so
since R/P is a domain and x 6= 0 we conclude that xi = 1 for some i ≥ 1 whence x is a unit.)
We conclude that P is maximal, as desired.

8. Let ϕ : R → S be a homomorphism of commutative rings and I an ideal of S. Prove that

ϕ−1 (I) (set-theoretic inverse image) is an ideal of R that is prime whenever I is a prime ideal
of S. Show that this holds with “prime” replaced by “maximal” provided we assume that ϕ
is surjective. Give a counterexample to this if we drop the surjectivity requirement.
Solution: The map ϕ induces an injective homomorphism of rings

R/ϕ−1 (I) ,→ S/I

so if the target is a domain, so is the source as any subring of a domain is a domain. In the
case that ϕ is surjective, this induced map is an isomorphism so if I is maximal both target
and source are fields and ϕ−1 (I) must be maximal as well. As a counterexample, consider the
map Z ,→ Q given by inclusion. The zero ideal of Q is maximal as Q is a field, but clearly
its inverse image—the zero ideal of Z—is not maximal.
Suppose that ab ∈ ϕ−1 (I). Then ϕ(a)ϕ(b) ∈ I so if I is prime one of ϕ(a), ϕ(b) lies in I and
hence one of a, b lies in ϕ−1 (I). If ϕ is surjective and I is maximal