Ring Theory

Fall 2017

Paul Balister
University of Memphis
7261 1. Rings Fall 2017
A Ring (with 1) is a set R with two binary operations + and × such that
R1. (R, +) is an Abelian group under +.
R2. (R, ×) is a Monoid under ×, (so × is associative and has an identity 1).
R3. The distributive laws hold: a(b + c) = ab + ac, (b + c)a = ba + ca.
Many of the standard facts from algebra follow from these
∑axioms.
∑ In particular,
∑
0a = a0 = 0, a(−b) = (−a)b = −(ab), −a = (−1)a, ( i ai )( j bj ) = i,j ai bj .
The ring R is commutative if × is commutative.
An element of R is a unit if it has a (2-sided) multiplicative inverse.
The set of units R× (or U (R)) is a group under ×.
The trivial ring is the ring {0} with 0 + 0 = 0.0 = 0, and is the only ring in which 1 = 0.
A division ring or skew field is a non-trivial ring in which every non-zero element is a
unit.
A field is a commutative division ring.
An Integral Domain (ID) is a non-trivial commutative ring in which ab = 0 implies
a = 0 or b = 0. Note that any field is an ID.

Examples
1. Z, Q, R, C are all rings under the usual + and ×. Q, R, C are fields. Z is an ID.
2. Z/nZ is a ring under + and × mod n. This ring is an ID iff n is prime. In fact, if n
is prime then Z/nZ is a field.
3. If R is a ring then the set Mn (R) of n × n matrices with entries in R is a ring under
matrix addition and multiplication. Mn (R) is non-commutative in general.
4. Let (A, +) be an abelian group and let End(A) be the set of group homomorphisms
A → A. Define addition pointwise, (f + g)(a) = f (a) + g(a), and multiplication by
composition, f g(a) = f (g(a)). Then End(A) is a (usually non-commutative) ring.
∏
5. If A = i∈N Z = {(a0 , a1 , . . . ) : ai ∈ Z} then the maps R((a0 , . . . )) = (0, a0 , a1 , . . . )
and L((a0 , a1 , . . . )) = (a1 , a2 , . . . ) lie in End(A) and LR = 1 ̸= RL. Hence R has a
left, but not a right inverse. [Recall that left and right inverses must be equal if they
both exist.]
6. Let C[0, 1] be the set of continuous functions from [0, 1] to R with addition and
multiplication defined pointwise. Then C[0, 1] is a ring. It is not an ID (why?).

A subset S of R is a subring iff (S, +) is a subgroup of (R, +) and (S, ×) is a submonoid

of (R, ×). Equivalently, 1R ∈ S and a, b ∈ S implies a − b, ab ∈ S.
A subset I of R is a left ideal iff (I, +) is a subgroup of (I, +) such that for all r ∈ R,
a ∈ I, we have ra ∈ I. A subset I of R is a right ideal iff (I, +) is a subgroup of (I, +)

1
such that for all r ∈ R, a ∈ I, we have ar ∈ I. An ideal is a subset that is both a left ideal
and a right ideal. Equivalently, I ̸= ∅ and a, b ∈ I, r ∈ R, implies a − b, ra, ar ∈ I. The
sets {0} and R are ideals of R. An ideal I is proper if I ̸= R, and non-trivial if I ̸= {0}.

Examples
1. nZ is an ideal of Z but not a subring (unless n = ±1).
2. Z is a subring of R but not {(
an ideal.
) }
0 b
3. The set of the matrices I = : b, d ∈ R is a left ideal, but not a right ideal of
0 d {( ) }
a b
M2 (R). But I is a 2-sided ideal of the subring T = : a, b, d ∈ R of M2 (R).
{( ) } 0 d
α β
4. The quaternions H = : α, β ∈ C form a subring of M2 (C). Any x ∈ H
−β̄ ᾱ ( ) ( )
i 0 0 1
can be written uniquely as x = a + bi + cj + dk where i = , j= ,
( ) 0 −i −1 0
0 i
k= . Then i2 = j 2 = k 2 = −1, ij = k, ji = −k, and (a + bi + cj + dk)−1 =
i 0
(a/r) − (b/r)i − (c/r)j − (d/r)k where r = a2 + b2 + c2 + d2 . Thus H is a non-
commutative division ring.
∩
Lemma 1.1 If Sα , α ∈ A, ∩ are subrings of R then α∈A Sα is a subring of R.
If Iα are ideals of R then α∈A Iα is an ideal of R.

The ideal (S) generated by a subset

∩ S ⊆ R is the smallest ideal of R containing S. It can
be defined as the intersection J⊇S J of all ideals containing S.
An ideal I is principal if it is generated by a single element, I = (a) for some a ∈ R. An
ideal is finitely generated if it is generated by a finite set, I = (S), |S| < ∞.
We can also define the subring generated by a subset. More generally, if R is a subring
of R′ and S ⊆ R′ , then R[S] is the smallest subring of R′ containing R and S (= the
intersection of all subrings of R′ containing R and S).

Exercises
1. Show that an ideal is proper iff it does not contain a unit.
∑
2. Show that (S) = { ni=1 ri si ri′ : ri , ri′ ∈ R, si ∈ S, n ∈ N}.
3. Show that if R is commutative then the principal ideal (a) is {ra : r ∈ R}.
∑
4. Show that R[α] is the set of all polynomial expressions ni=0 ai αi with coefficients
ai ∈ R.
√ √ √
5. Deduce that Z[i] = {a+bi : a, b ∈ Z} as a subring of C and Q[ 3 2] = {a+b 3 2+c 3 4 :
a, b, c ∈ Q} as a subring of R.
6. Describe Z[1/2] as a subring of Q.
7. Let I be the set of continuous functions f ∈ C[0, 1] such that f (0.5) = 0. Show that
I is an ideal of C[0, 1] that is not principal (or even finitely generated).

2
7261 2. Ring homomorphisms Fall 2017
A (ring) homomorphism from the ring R to the ring S is a function f : R → S that is
a group homomorphism (R, +) → (S, +) and a monoid homomorphism (R, ×) → (S, ×).
Equivalently f (a + b) = f (a) + f (b), f (ab) = f (a)f (b), f (1R ) = 1S .

Examples ( ) {( ) }
a b a b
1. The map f : T → R given by f ( ) = a where T = : a, b, d ∈ R .
0 d 0 d
2. If S is a subring of R then the inclusion map i : S → R, i(r) = r, is a homomorphism.

A (ring) isomorphism is a homomorphism R → S that has a 2-sided inverse map g : S →

R which is also a homomorphism. It is sufficient for f to be a bijective homomorphism.
If I is an ideal of R then the quotient ring R/I is the quotient group (R/I, +) with
multiplication defined by (a + I)(b + I) = ab + I.

Lemma 2.1 The quotient ring R/I is indeed a ring and the projection map π : R → R/I
given by π(a) = a + I is a surjective ring homomorphism.

Example R = Z, I = (n), then R/I = Z/nZ is the integers mod n with addition and
multiplication mod n.

Theorem (1st Isomorphism Theorem) If f : R → S then Ker f = {r : f (r) = 0} is

an ideal of R, Im f = {f (r) : r ∈ R} is a subring of S and f = i ◦ f˜ ◦ π where
• π : R → R/ Ker f is the (surjective) projection homomorphism. R
f
−→ S
• f˜: R/I → Im f is a (bijective) ring isomorphism. π↓ ↑i
˜
f
• i : Im f → S is the (injective) inclusion homomorphism. R/ Ker f −→ Im f

Theorem (2nd Isomorphism Theorem) If I is an ideal of R then there is a bijection

{subgroups H of (R, +) with I ≤ H ≤ R} ↔ {subgroups of (R/I, +)},
where H corresponds to H/I. In this correspondence subrings correspond to subrings and
ideals correspond to ideals. Moreover, if J is an ideal with I ≤ J ≤ R then there is an
isomorphism R/J ∼ = (R/I)/(J/I).

Theorem (3rd Isomorphism Theorem) If I is an ideal of R and S is a subring of R

then S + I is a subring of R, S ∩ I is an ideal of S, and (S + I)/I ∼
= S/(S ∩ I).

Example For any ring R define f : Z → R by f (n) = n.1R (n.1R = 1R +· · ·+1R defined as
for additive groups). Then f is a ring homomorphism. The kernel is a subgroup of (Z, +)
so is nZ for some n ≥ 0. The image S = {n.1R : n ∈ Z} is called the prime subring of
R and is isomorphic to Z/nZ. The characteristic of R, char(R), is the integer n. E.g.,
char(R) = 0, char(Z/nZ) = n, char({0}) = 1.

3
A maximal ideal is a proper ideal M of R such that for any ideal I, M ⊆ I ⊆ R implies
I = M or I = R.

Example The ideal (n) is a maximal ideal of Z iff n is prime.

A non-trivial ring is simple if the only ideals of R are (0) and R. Equivalently, (0) is
maximal.

Lemma 2.2 Let R be a commutative ring. Then R is simple iff R is a field.

Proof. If R is a field and I ̸= (0) is an ideal then u ∈ I for some u ̸= 0. But u is a unit
so (ru−1 )u = r ∈ I for all r ∈ R. Thus I = R. Conversely, if a ̸= 0 and a is not a unit
then (a) = {ra : r ∈ R} is a non-trivial proper ideal of R.

Note that if R is a division ring then R is simple. However the converse fails:

Lemma 2.3 Let D be a division ring. Then Mn (D) is a simple ring for any n ≥ 1.
Proof. Let I be a non-zero ideal of Mn (D) and let A = (aij ) ∈ I, A ̸= 0. In particular
akl ̸= 0 for some k, l. Let Eij be the matrix with 1 in entry (i, j) and zeros elsewhere.
Then Eik AElj = akl Eij ∈ I. Since akl ∈ D and D is a division ring, a−1 kl ∈ D, so
a−1
kl I ∈ M (D).
n ∑ Now (a−1
kl I)(a E
kl ij ) = E ij ∈ I. But any matrix B = (b ij is a linear
)
combination (bij I)Eij , so B ∈ I and I = Mn (D).

So by the 2nd Isomorphism Theorem, for commutative R, M is maximal iff R/M is a field,

but for non-commutative R, M may be maximal without R/M being a division ring.

Exercises
1. Show that any finite ID is a field.
2. An element a of a ring is nilpotent if an = 0 for some n ∈ N. Show that if a is
nilpotent then 1 + a is a unit.
3. Show that if R is commutative then the set of nilpotent elements forms an ideal of
R. [Hint: make sure you check that a, b nilpotent implies a − b is nilpotent.]
4. Show that if r ∈ R lies in the intersection of all maximal ideals of R then 1 + r is a
unit.
5. Show that any homomorphism f : F → R from a field F to a non-trivial ring R is
injective, so in particular R contains a subring isomorphic to F .

4
7261 3. Zorn’s Lemma Fall 2017
A partial ordering on a set X is a relation ≤ satisfying the properties:
O1. ∀x : x ≤ x,
O2. ∀x, y : if x ≤ y and y ≤ x then x = y,
O3. ∀x, y, z : if x ≤ y and y ≤ z then x ≤ z.
A total ordering is a partial ordering which also satisfies:
O4. ∀x, y : either x ≤ y or y ≤ x.

Example Any collection of sets with ⊆ as the ordering forms a partially ordered set that
is not in general totally ordered.

If (X , ≤) is a partially ordered set, a chain in X is a non-empty subset C ⊆ X that is

totally ordered by ≤.
If S ⊆ X , and x ∈ X , we say x is an upper bound for S if y ≤ x for all y ∈ S. [Note
that we do not require x to be an element of S.]
A maximal element of X is an element x such that for any y ∈ X , x ≤ y implies x = y.
[Note: This does not imply that y ≤ x for all y since ≤ is only a partial order. In particular
there may be many maximal elements.]

Theorem (Zorn’s Lemma) If (X , ≤) is a non-empty partially ordered set for which

every chain has an upper bound then X has a maximal element.

This result follows from (and∏is equivalent to) the Axiom of choice, which states that if
Xi are non-empty sets then i∈I Xi is non-empty. [I will not give the proof here as it is
rather long.]
Note: If we had defined things so that ∅ were a chain, we would not need the condition that
X ̸= ∅ in Zorn’s Lemma since the existence of an upper bound for ∅ is just the condition
that an element of X exists. However, in practice it is easier to check X ̸= ∅ and then
check separately that each non-empty totally ordered subset has an upper bound.

Theorem 3.1 If I is a proper ideal of a ring R (with 1) then there exists a maximal ideal
M such that I ⊆ M .
Proof. If an ideal J contains 1 then J = R, so an ideal is proper iff it does not contain
1. Let X be the set of proper ideals J of R with I ⊆ J. The partial order on X will be
⊆. Since I ∈ X , X ̸= ∅. Now let C be a chain in X , i.e.,
∪ a set of ideals {Jα } such that for
every Jα , Jβ ∈ C either Jα ⊆ Jβ or Jβ ⊆ Jα . Let K = Jα ∈C Jα . We shall show that K is
an upper bound for C.

5
Firstly C ̸= ∅, so some ideal Jα lies in C and I ⊆ Jα ⊆ K. In particular K ̸= ∅. If
x, y ∈ K then x ∈ Jα , y ∈ Jβ , say. Since C is totally ordered, we can assume without loss
of generality that Jα ⊆ Jβ . Thus x, y ∈ Jβ , and x − y ∈ Jβ ⊆ K. If x ∈ K, r ∈ R, then
x ∈ Jα , say, so xr, rx ∈ Jα ⊆ K. Hence K is an ideal with I ⊆ K. However 1 ∈ / Jα for
each Jα ∈ C, so 1 ∈ / K. Hence K is proper. Therefore K ∈ X and is clearly an upper
bound for C.
The conditions of Zorn’s Lemma apply, so X has a maximal element M , say. Now M is a
proper ideal containing I and is maximal, since if M ⊂ J ⊂ R then J ∈ X and M would
not be maximal in X .

We now give an example from linear algebra. Let V be a vector space (possibly infinite
dimensional).
A set S ⊆ V is called linearly independent ∑n if there are no non-trivial finite linear
combinations that give 0. In other words if i=1 λi si = 0 and the si are distinct elements
of S then λi = 0 for each i.
A set S ⊆ V is called spanning if ∑
every element v ∈ V can be written as a finite linear
combinations of elements of S, v = ni=1 λi si .
A set S ⊆ V is called a basis if it is a linearly independent spanning set. Note that every
element v ∈ V can be written as a linear combination of elements of a basis in a unique
way. [Spanning implies existence, linear independence implies uniqueness.]

Theorem 3.2 Every vector space has a basis.

Proof. Let X be the set of all linearly independent sets in V partially ordered
∪ by ⊆. Since
∅ is linearly independent, X ̸= ∅. Let C be a chain in X and let S = Sα ∈C Sα . We shall
show that S is linearly independent.
∑
Suppose ni=1 λi si = 0 and si ∈ Sαi ∈ C (the si are distinct but the αi need not be).
Then by total ordering of the S∑ αi , there must be one Sαj that contains all the others
(use induction on n). But then ni=1 λi si = 0 is a linear relation in Sαj which is linearly
independent. Thus λi = 0 for all i. Hence S is linearly independent, so S ∈ X and is an
upper bound for C.
Now apply Zorn’s Lemma to give a maximal linearly independent set M . We shall show
that M spans V and so is a basis. Clearly any element of M is a linear combination
of elements of M , so pick any v ∈ / M and consider M ∪ {v}. By maximality ∑n of M this
cannot be linearly independent. Hence there is a linear combination λv + i=1 λi si = 0,
si ∈ M , with not all the λ’s zero. If λ = 0 this gives
∑na linear relation in M , contradicting
linear independence of M . Hence λ ̸= 0 and v = i=1 (−λi /λ)si is a linear combination
of elements of M .

6
7261 4. Miscellaneous topics Fall 2017
Anti-isomorphisms
An anti-homomorphism is a map f : R → S such that f (a + b) = f (a) + f (b), f (1) = 1,
and f (ab) = f (b)f (a). An anti-isomorphism is an invertible anti-homomorphism.

Examples The transpose map T : Mn (R) → Mn (R).

The map H → H given by f (a + bi + cj + dk) = a − bi − cj − dk.

The opposite ring Ro of R is the ring R with multiplication defined by a×Ro b = b×R a.
Note that Roo = R.

Lemma 4.1 A map f : R → S is an anti-homomorphism iff it is a homomorphism viewed

as a map R → S o (or Ro → S).

Example Mn (R)o is isomorphic to Mn (R), one isomorphism being the transpose map T .

Rngs (Rings without 1s)

A Rng (or “ring which does not necessarily have a 1”) is a set R with + and × defined
so that (R, +) is an abelian group, (R, ×) is a semigroup (× is associative), and the
distributive laws hold. However, R need not contain a multiplicative identity.
Subrngs, rng-homomorphisms etc., can be defined without the conditions involving 1. The
definition of an ideal is the same, and an ideal is a special case of a subrng. The theory
of rngs is similar to that of rings, although they are more awkward to deal with later on.
The following lemma shows that we can regard a rng as an ideal of a bigger ring.

Lemma 4.2 Let R be a rng and define R1 = Z × R with addition (n, r) + (m, s) =
(n + m, r + s) and multiplication (n, r)(m, s) = (nm, n.s + m.r + rs), where n.s = s + · · · + s
etc.. Then R1 is a ring containing an ideal {0} × R isomorphic to R.

Direct sums and the Chinese Remainder Theorem

If R1 and R2 are rings, define the ring R1 ⊕ R2 as the set R1 × R2 with addition (a1 , a2 ) +
(b1 , b2 ) = (a1 + b1 , a2 + b2 ) and multiplication (a1 , a2 )(b1 , b2 ) = (a1 b1 , a2 b2 ). The identity is
(1, 1). The direct sum R1 ⊕ · · · ⊕ Rn is defined similarly. Note that even if R1 and R2 are
IDs, R1 ⊕ R2 will not be since (1, 0)(0, 1) = (0, 0).
If R is a ring and I and J are ideals of R, we can define the following ideals.
• I + J = {a + b : a ∈ I, b ∈ J}
• I ∩ J = {c : c ∈ I, c ∈ J}
∑
• IJ = { ni=1 ai bi : ai ∈ I, bi ∈ J, n ∈ N}

7
It is easily checked that each of these is indeed an ideal. Note that in general IJ ̸= {ab :
a ∈ I, b ∈ J}, but IJ is the ideal generated by all the products ab, a ∈ I, b ∈ J.

Example For R = Z, I = (x) = {ax : a ∈ Z}, J = (y) = {by : b ∈ Z}

1. I + J = (gcd(x, y)).
Note gcd(x, y) = ax + by for some a, b ∈ Z, so gcd(x, y) ∈ I + J. Conversely
I + J = {ax + by : a, b ∈ Z} and ax + by is always a multiple of gcd(x, y).
2. I ∩ J = (lcm(x, y)).
m ∈ I ⇐⇒ x | m and m ∈ J ⇐⇒ y | m. Hence if m ∈ I ∩ J then m must
be a common multiple of x and y. Thus m ∈ (lcm(x, y)) Conversely lcm(x, y) is a
common multiple of x and y so lies in I ∩ J. Hence I ∩ J = (lcm(x, y)).
3. IJ = (xy).
∑
IJ = { ai xbi y : ai , bi ∈ Z} ⊆ (xy). Conversely xy ∈ IJ, so (xy) ⊆ IJ.

Ideals I and J are relatively prime if I + J = R. Equivalently ∃a ∈ I, b ∈ J : a + b = 1

(recall that an ideal equals R iff it contains 1).

Lemma 4.3 IJ ⊆ I ∩ J. Moreover, if R is commutative and I + J = R then IJ = I ∩ J.

∑ ∑
Proof. If ai ∈ I then ai bi ∈ I. If bi ∈ J then ai bi ∈ J. Hence IJ ⊆ I ∩ J.
Now let I + J = R so that a + b = 1 for some a ∈ I, b ∈ J. Then if c ∈ I ∩ J, ac + cb ∈ IJ.
But ac + cb = c(a + b) = c, so c ∈ IJ. Thus I ∩ J ⊆ IJ and so IJ = I ∩ J.

Theorem (Chinese Remainder Theorem) If I and J are ideals of a commutative

ring R and I + J = R then R/IJ ∼
= R/I ⊕ R/J.
Proof. Let f : R → R/I ⊕ R/J be defined by f (r) = (r + I, r + J). Then f (r + s) =
(r + s + I, r + s + J) = (r + I, r + J) + (s + I, s + J) = f (r) + f (s), f (rs) = (rs + I, rs + J) =
(r +I, r +J)(s+I, s+J) = f (r)f (s), and f (1) = (1+I, 1+J) is the identity in R/I ⊕R/J.
Now Ker f = {r : r+I = I, r+J = J} = I ∩J so Ker f = IJ by Lemma 4.3. For the image
of f , write 1 = a+b with a ∈ I, b ∈ J. Then f (sa+rb) = (sa+r(1−a)+I, s(1−b)+rb+J) =
(r + I, s + J). Thus f is surjective. Hence R/IJ ∼ = R/I ⊕ R/J.

Example If gcd(n, m) = 1 then Z/nmZ ∼

= Z/nZ ⊕ Z/mZ.

Exercises
1. Show that composing two anti-homomorphisms gives a homomorphism and compos-
ing an anti-homomorphism with a homomorphism gives an anti-homomorphism.
2. Define ϕ(n) = |(Z/nZ)× |. Show that if gcd(n, m) = 1 then ϕ(nm) ∏= aϕ(n)ϕ(m). If
i −1
n = p1 . . . pr is the prime factorization of n, deduce that ϕ(n) = i pi (pi − 1).
a1 ar

3. Generalize the CRT: if I1 , . . . , In are ideals of a commutative ring R and for each i
and j, Ii + Ij = R, show that R/I1 I2 . . . In ∼ = I1 ⊕ I2 ⊕ · · · ⊕ In .

8
7261 5. Primes and Localization Fall 2017
Throughout this section we shall assume R is a commutative ring.
Recall: An Integral Domain (ID) is a non-trivial ring in which ab = 0 implies either
a = 0 or b = 0.
A prime ideal of a commutative ring R is a proper ideal such that ab ∈ P implies either
a ∈ P or b ∈ P .

Lemma 5.1 An ideal P is prime iff R/P is an ID.

Proof. Assume P is prime. Then R/P is non-trivial since P is proper. If (a + P )(b + P ) =
0 + P then ab + P = P and so ab ∈ P . Thus either a ∈ P or b ∈ P , so either a + P = P or
b + P = P . Thus R/P is an ID. Conversely, if R/P is an ID then P is proper since R/P
is non-trivial. If a, b ∈
/ P , then a + P, b + P ̸= 0 + P , so (a + P )(b + P ) = ab + P ̸= 0 + P ,
so ab ∈
/ P . Thus P is a prime ideal.

Corollary 5.2 Any maximal ideal of a commutative ring is also a prime ideal.
Proof. M maximal ⇒ R/M is a field ⇒ R/M is an ID ⇒ M is prime.

The converse does not hold: (0) is prime but not maximal in Z.
Examples of prime ideals: (p) in Z, (0) in any ID. The ideal (X) in the ring Z[X] of
polynomials in X with coefficients in Z. This last example is also not maximal.
Every field is an ID. Furthermore, every subring of a field is an ID (e.g., Z ⊆ Q). We shall
show that conversely, every ID can be embedded as a subring of a field.
Assume R is a commutative ring and S ⊆ R is a submonoid of (R, ×). In other words,
1 ∈ S and a, b ∈ S implies ab ∈ S. For example, set S = R \ P for any prime P . One
particularly important case is when R is an ID and S = R \ {0}.
Define S −1 R as (R × S)/∼, where (r, s) ∼ (r′ , s′ ) iff ∃u ∈ S : urs′ = ur′ s. We write r/s
for the equivalence class (r, s) ∈ S −1 R.
Note: if S contains no zero-divisors then (r, s) ∼ (r′ , s′ ) iff rs′ = r′ s.

Lemma 5.3 The relation ∼ defined above is an equivalence relation and S −1 R can be
made into a ring so that the map i : R → S −1 R, i(r) = r/1 is a homomorphism. Also
i(S) ⊆ (S −1 R)× and the map i is injective iff S contains no zero-divisors.
Proof. Reflexivity and symmetry of ∼ are immediate. For transitivity, if (r, s) ∼ (r′ , s′ ) ∼
(r′′ , s′′ ) then ∃u, u′ : urs′ = ur′ s, u′ r′ s′′ = u′ r′′ s′ . Hence (uu′ s′ )(rs′′ ) = u′ s′′ us′ r =
u′ s′′ usr′ = usu′ r′ s′′ = usu′ r′′ s′ = (uu′ s′ )(r′′ s). But uu′ s′ ∈ S, so (r, s) ∼ (r′′ , s′′ ).
Define addition by r1 /s1 +r2 /s2 = (r1 s2 +r2 s1 )/(s1 s2 ) and multiplication by (r1 /s1 )(r2 /s2 ) =
(r1 r2 )/(s1 s2 ). A long and rather tedious check shows that under these operations S −1 R
becomes a commutative ring with identity 1/1.

9
The map i(r) = r/1 is a ring homomorphism since i(r) + i(r′ ) = r/1 + r′ /1 = (r + r′ )/1 =
i(r + r′ ), i(r)i(r′ ) = (r/1)(r′ /1) = (rr′ )/1 = i(rr′ ), and i(1) = 1/1.
The element 1/s ∈ S −1 R is the inverse of i(s) = s/1, so i(S) ⊆ (S −1 R)× .
The kernel of i is {r ∈ R : r/1 = 0/1} = {r ∈ R : ∃u ∈ S : ur = 0}. Thus Ker i = {0} iff
S contains no zero-divisors.

Lemma 5.4 S −1 R satisfies the following universal property: If f : R →

f
R → R′
R′ is a homomorphism with f (S) ⊆ (R′ )× then f factors uniquely as i↓ ↗h
f = h ◦ i where h : S −1 R → R′ is a homomorphism. −1
S R

Proof. Any such f˜ must satisfy f˜(r/s)f˜(s/1) = f˜(r/1) and f˜(t/1) = f (t). Hence
f˜(r/s)f (s) = f (r) and f˜(r/s) = f (r)f (s)−1 . Conversely, defining f˜(r/s) = f (r)f (s)−1
gives a homomorphism S −1 R → R′ (check this!).

Notation: If S = R \ P for some prime ideal P , we also write S −1 R as RP and call it the
localization of R at P .

Lemma 5.5 If R is an ID then (R \ {0})−1 R = R(0) is a field containing a subring

isomorphic to R.
Proof. Let S = R \ {0}. If r/s ̸= 0/1 then r ̸= 0, so s/r ∈ S −1 R and (s/r)(r/s) = 1/1.
Hence any non-zero element of S −1 R is invertible. The map i is injective, so Im i is a
subring of S −1 R isomorphic to R.

In this case we call R(0) = S −1 R the field of fractions of R, or Frac R. For example
Frac(Z) = Q.

Exercises
1. Show that the units of RP consists of the elements r/s where r ∈ / P and there is a
unique maximal ideal of RP consisting of all the non-unit elements. [Rings that have
a unique maximal ideal are called local rings.]

2. Show that if R is an ID, then for any prime ideal P , RP is isomorphic to a subring
of Frac R.

3. Describe Z(2) explicitly as a subring of Q.

4. What is the field of fractions of a field?

5. What is the field of fractions of the ring of entire functions (holomorphic functions
f : C → C)?
∑
6. What is the field of fraction of the ring of polynomial functions C[X] = { ni=0 ai X i :
ai ∈ C, n ∈ N}?

10
7261 6. Polynomial rings Fall 2017
Assume that R is a commutative ring. We wish to construct the ring R[X] of polynomials
in X with coefficients in R.
Define R[X] as the set of sequences (a0 , a1 , . . . ) with the property that all but finitely ⊕many
of the ai s are zero. Define (a0 , . . . ) + (b0 , . . . ) = (a0 + b0 , a1 + b1 , . . . ) (so R[X]
∑ = i∈N R
as group under +) and define (a0 , . . . )(b0 , . . . ) = (c0 , c1 , . . . ) where ci = 0≤j≤i aj bi−j .
We call R[X] the ring of polynomials in X over R. Let i : R → R[X] be defined by
i(a) = (a, 0, 0, . . . ) and let X ∈ R[X] be the element X = (0, 1, 0, 0, . . . ). Note that
X(a0 , a1 , . . . ) = (0, a0 , a1 , . . . ) and i(a)(a0 , a1 , . . . ) = (aa0 , aa1 , . . . ).

Lemma 6.1 R[X] is a ring, i : R →∑ R[X] is an injective ring homomorphism, and if

ai = 0 for all i > n then (a0 , a1 , . . . ) = ni=0 i(ai )X i

∑n shall inormally identify i(a) with a and write polynomials f (X) ∈ R[X] in the form
We
i=0 ai X . The degree deg f (X) of a polynomial is the largest n such that an ̸= 0, (or
−∞ if f = 0). The leading coefficient of f (X) is an where n = deg f , (or 0 if f = 0). A
polynomial is monic if the leading coefficient is 1.

Lemma 6.2 If f, g ∈ R[X] then

1. deg(f + g) ≤ max{deg f, deg g},
2. deg(f g) ≤ deg f + deg g, with equality holding if R is an ID.

Lemma 6.3 If R is an ID then R[X] is an ID and (R[X])× = R× .

Proof. If f, g ∈ R[X] and f, g ̸= 0 then deg(f g) = deg f + deg g ≥ 0, so f g ̸= 0. If f g = 1
then 0 = deg(f g) = deg f + deg g so deg f = deg g = 0 and f, g ∈ R. Hence f ∈ (R[X])×
implies f ∈ R× . Conversely f ∈ R× clearly implies f ∈ (R[X])× .

Theorem (Universal property of polynomial rings) If ϕ : R → R′ is a ring homo-

morphism and α ∈ R′ then there exists a unique homomorphism evϕ,α : R[X] → R′ such
that evϕ,α (a) [= evϕ,α (i(a))] = ϕ(a) for all a ∈ R and evϕ,α (X) = α.

If R is a subring of R′ and ϕ is the inclusion map we write f (α) for evϕ,α (f ). More generally,
if just R is a subring of R′ we write ϕ(f )(α) for evϕ,α (f ).

Lemma 6.4 If R is a subring of R′ and α ∈ R′ then R[α] is isomorphic to a quotient

R[X]/I where I is an ideal of R[X] containing no non-zero constants: I ∩ R = {0}.
Proof. Apply 1st Isomorphism Theorem to evα : R[X] → R′ .

We say α ∈ R′ is transcendental over R ⊆ R′ if the map evα is injective. In other words,

if f (α) = 0 implies f (X) = 0. Otherwise we say that α is algebraic over R.

11
Examples
√ √ The element π ∈ R is transcendental over Z, so Z[π] ∼ = Z[X]. The elements
i, 2, 3 ∈ C are all algebraic over Z. However π is algebraic over R (since it is a root of
4

X − π ∈ R[X]).

Theorem (Division Algorithm) If f, g ∈ R[X] and the leading coefficient of g is a

unit in R, then there exist unique q, r ∈ R[X] such that f = qg + r and deg r < deg g (or
r = 0).

If a, b ∈ R, we say a divides b, a | b, if there exists c ∈ R such that b = ca.

Examples In any ring, u | 1 iff u ∈ R× , a | 0 for all a. In Z, 7 | 21. In Q, 21 | 7.

Lemma 6.5 If α ∈ R and f ∈ R[X] then f (X) = (X −α)q(X)+f (α) for some q ∈ R[X].
In particular, X − α | f iff f (α) = 0.

Lemma 6.6 If R is an ID and f ∈ R[X], f ̸= 0, then |{α ∈ R : f (α) = 0}| ≤ deg f .

Lemma 6.7 If R is an ID and G is a finite subgroup of R× then G is cyclic.

Proof. G is a finite abelian group, so G ∼ = Cd1 × · · · × Cdr . But then xd1 = 1 for all
x ∈ G. Thus the polynomial X d1 − 1 has |G| zeros. Thus |G| = d1 d2 . . . dr ≤ d1 , so
d2 = · · · = dr = 1 and G ∼
= Cd1 is cyclic.

We can generalize polynomial rings to polynomials in many variables. If {Xi }i∈I is a set
(possibly infinite) of indeterminates, define a term t to be a function I → N which is
non-zero
∏ for only finitely many i ∈ I. We think of t as corresponding to a finite product
t(i)
i∈I X i . Let T be the set of terms. Now define the ring
⊕
R[{Xi }i∈I ] = R = {(at )t∈T | at = 0 for all but finitely many t},
t∈T

with addition of coefficients componentwise

∑ (at )+(bt ) = (at +bt ) and multiplication defined
by (at )(bt ) = (ct ) where ct = r+s=t ar bs (note that this is a finite sum). As for R[X] we
can identify R as a subring of R[{Xi }i∈I ] and define elements Xi so that (at )t∈T is equal
∑ ∏ t(i)
to the (finite) sum t∈T at i∈I Xi .

Theorem (Universal property of polynomial rings) If ϕ : R → R′ is a ring

homomorphism and αi ∈ R′ for all i ∈ I then there exists a unique homomorphism
evϕ,(αi ) : R[{Xi }i∈I ] → R′ such that evϕ,(αi ) (a) = ϕ(a) for all a ∈ R and evϕ,(αi ) (Xi ) = αi
for all i ∈ I.

If I is finite then we can also identify R[X1 , . . . , Xn ] with R[X1 , . . . , Xn−1 ][Xn ] (use uni-
versal properties to define the isomorphism).

12
7261 7. Euclidean Domains and PIDs Fall 2017
A Euclidean Domain is an ID for which there is a function d : R \ {0} → N such that if
a, b ∈ R, b ̸= 0 then there exists q, r ∈ R such that a = qb + r with either d(r) < d(b) or
r = 0.

Examples
1. Z with d(a) = |a|.
2. F [X], where F is a field, d(f ) = deg f .
3. F , where F is a field, d(a) = 0.
4. Z[i], with d(a + ib) = |a + ib|2 = a2 + b2 . [Write a/b = x + iy and let q = x′ + iy ′ with
|x−x′ |, |y−y ′ | ≤ 12 . Then d(r) = |qb−a|2 = |q−a/b|2 |b|2 = ((x−x′ )2 +(y−y ′ )2 )d(b) ≤
1
2
d(b).]

A Principal Ideal Domain (PID) is an ID in which every ideal I is principal, i.e., I = (a)
for some a ∈ R.

Theorem 7.1 Every Euclidean Domain is a PID.

Proof. If R is Euclidean then R is an ID, so it is enough to show that any ideal I is
principal. Let I be an ideal of R and assume I ̸= (0). Pick b ∈ I \ {0} with minimal value
of d(b) (by well ordering of N). If a ∈ I then a = qb + r with d(r) < d(b) or r = 0. But
r = a − qb ∈ I, so by choice of b we must have r = 0. Thus a = qb ∈ (b). Thus I ⊆ (b).
But b ∈ I, so (b) ⊆ I. Thus I = (b) is principal.

Note: PID ̸⇒ Euclidean.

If I = (a) is a principal ideal then b ∈ I implies there exists a c ∈ R with b = ca. Thus
b ∈ I is equivalent to a | b. In particular (b) ⊆ (a) ⇐⇒ a | b. If (a) = (b) then b = ua
and a = vb. Thus either a = b = 0 or uv = 1 and u, v ∈ R× . Conversely, if a = ub with
u ∈ R× then (a) = (b).
The elements a, b ∈ R are called associates if b = ua for some u ∈ R× . Equivalently, a | b
and b | a both hold, or (a) = (b). Write a ∼ b if a and b are associates.
A greatest common divisor (gcd) of a set of elements S ⊆ R is an element d ∈ R such
that
G1. d | a for all a ∈ S, and
G2. if c | a for all a ∈ S then c | d.
Greatest common divisors are unique up to multiplication by units. To see this, let d, d′
be two gcds. Then condition G2 with c = d′ and G1 with d = d′ imply d′ | d. Similarly
d | d′ , so d′ = ud for some unit u ∈ R× .

13
Lemma 7.2 If R is a PID then gcds of any S ⊆ R ∑ exist. Indeed, if (S) = (d) then d is
a gcd of S and hence can be written in the form d = ni=1 ci ai , for some ai ∈ S, ci ∈ R.
Proof. Since R is a PID, (S) = (d) for some d. If a ∈ S then a ∈ (S) = (d), so d | a. If
c | a for all a ∈ S, then a ∈ (c) for all a ∈ S, so (S) = (d) ⊆ (c). Hence c | d. Thus d is a
gcd of S.

Note: In an arbitrary ID, gcds may not exist, and even if they do, they may not be a
linear combination of elements of S. For example the elements 2 and X in Z[X] have
1 as a gcd, but 1 is not of the form 2c1√+ Xc2 , c1 , c2 ∈ Z[X]. For an example where
the gcd does not exist, consider R = Z[ −5]. If a √ ∈ R then |a|2√∈ Z. Hence√ if a | b
in R then |a|
√ | |b| in Z.√ Now let√x = −3(3 − −5)
2 2
√ = (1 + 2√ −5)(1 + −5) and
y = −7(1 + −5) = (1 − 2 −5)(3 − −5). Then 1 + −5 and 3 − −5 are two common
factors of x and y. If d is a gcd of x and y, then |d|2 must be a factor√of |x|2 = 2.32 .7
and √|y|2 = 2.3.72 . On the other hand, |d|2 must be a multiple√of |1 + −5|2 = 2.3 and
|3 − −5|2 = 2.7. Thus |d|2 = 2.3.7 = 42. However, if d = α + β −5 then |d|2 = α2 + 5β 2 ,
which is never equal to 42.

The Euclidean Algorithm

We can turn Lemma 1 into an algorithm in the case when R is Euclidean. Assume we
need to find the gcd of a0 = a and a1 = b. Inductively define an+1 for n ≥ 1 and an ̸= 0 by
an−1 = qn an + an+1 , d(an+1 ) < d(an ) or an+1 = 0
Since the d(an ) are a sequence of decreasing non-negative integers, eventually an+1 = 0.
However ai+1 ∈ (ai , ai−1 ) and ai−1 ∈ (ai , ai+1 ) imply the two ideals (ai−1 , ai ) and (ai , ai+1 )
are equal. Hence (a0 , a1 ) = (an , an+1 ) = (an ) and an is a gcd of a0 and a1 .
This algorithm is called the Euclidean Algorithm. For more than two elements, one can
calculate the gcd inductively by using gcd(c1 , c2 , . . . , cr ) = gcd(c1 , gcd(c2 , . . . , cr )).

Exercises
1. Prove that gcd(c1 , . . . , cr ) = gcd(c1 , gcd(c2 , . . . , cr )) provided the gcds on the RHS
exist. What is gcd(∅)?
√
2. Let R = Z[ω] where ω = 12 (1 + −3). Show that R = {a + bω : a, b ∈ Z} and that
R is Euclidean.
3. Use the Euclidean algorithm to find the gcd of 7 − 3i and 5 + 3i in Z[i].
4. Determine ((Z/nZ)[X])× . [Hint: Consider the case n = pr first.]
5. Solve the congruences
x ≡ i mod 1 + i x ≡ 1 mod 2 − i
in Z[i] (use Chinese Remainder Theorem).

14
7261 8. Unique Factorization Fall 2017
/ R× , and a = bc implies b ∈ R× or c ∈ R× .
An element a ∈ R is irreducible if a ̸= 0, a ∈
×
An element a ∈ R is a prime if a ̸= 0, a ∈
/ R and a | bc implies a | b or a | c.

Lemma 8.1 Let R be an ID, and a ∈ R. Then

1. a is a prime element iff (a) is a non-zero prime ideal,
2. a is irreducible iff (a) is maximal among proper principal ideals
(i.e., (a) ⊆ (b) implies (b) = (a) or (b) = R),
3. if a is prime then a is irreducible,
4. if a is irreducible and R is a PID then a is prime.
Proof.
1. If a is prime and bc ∈ (a) then a | bc. Hence a | b or a | c, so either b ∈ (a) or c ∈ (a).
Also, a ̸= 0, a ∈ / R× implies (a) ̸= (0), R. Conversely, if (a) is a prime ideal and a | bc,
then bc ∈ (a), so either b ∈ (a) or c ∈ (a), so either a | b or a | c and (a) ̸= (0), R implies
a ̸= 0, a ∈/ R× .
2. If a ∈ R be irreducible and (a) ⊆ (b) then a = bc, so either c ∈ R× and (b) = (a) or
b ∈ R× and (b) = R. Conversely if (a) is maximal among all proper principal ideals and
a = bc then (a) ⊆ (b), so either (a) = (b) and c is a unit or (b) = R and b is a unit.
3. If a is a prime and a = bc then a | bc. Thus either a | b and c ∈ R× , or a | c and b ∈ R× .
4. By part 2, (a) is a maximal ideal. Hence (a) is prime and so a is prime.

A ring R is a Unique Factorization Domain (UFD) if R is an ID such that

U1. Every a ∈ R \ {0} can be written in the form a = up1 . . . pr where u ∈ R× and the
pi are irreducible.
U2. Any two such factorizations are unique in the sense that if up1 . . . pr = vq1 . . . qs then
r = s and there is a permutation π ∈ Sr such that pi ∼ qπ(i) for all i.

Lemma 8.2 R is a UFD iff R is an ID satisfying

A. there is no infinite sequence (ai )i∈N with ai+1 | ai and ai+1 ̸∼ ai , and
B. every irreducible is prime.
Proof.
A ⇒ U1. Suppose a1 ∈ R has no such factorization. Then a1 is neither a unit nor
/ R× , and either b or c also has no factorization into irreducibles.
irreducible, so a1 = bc, b, c ∈
Assume b has no factorization into irreducibles and set a2 = b. Repeating this process we
get a sequence ai with ai+1 | ai and ai+1 ̸∼ ai .
B ⇒ U2. Since p1 is prime and p1 | vq1 . . . qs , we must have p1 | qi for some i. But qi is
irreducible, so p1 ∼ qi . Cancelling a factor of p1 from both sides (R is an ID) and using
induction on r gives the result.
U1 and U2 ⇒ A and B is clear.

15
A ring is Noetherian if every sequence of ideals Ii with Ii ⊆ Ii+1 is eventually constant,
In = In+1 = . . . , for some n.

Lemma 8.3 R is Noetherian iff every ideal is finitely generated.

Proof. ⇐: Let I = ∪In . Then I is an ideal, so I = (d1 , . . . , dr ) for some di ∈ R. But then
there is an ni with di ∈ In . Let n = max ni , so that I = (d1 , . . . , dr ) ⊆ In ⊆ In+1 ⊆ · · · ⊆ I,
and so In = In+1 = . . . .
⇒: Assume I is not finitely generated. Then (using Axiom of choice), pick inductively
dn ∈ I \ (d1 , . . . , dn−1 ). Then In = (d1 , . . . dn ) is a strictly increasing sequence of ideals.

Theorem Every PID is a UFD.

Proof. Every ideal in a PID is finitely generated (by one element), so PID ⇒ Noethe-
rian. By considering the ideals (ai ), Noetherian rings satisfy condition A of Lemma 8.2.
Lemma 8.1 part 4 implies condition B of Lemma 8.2, so PID ⇒ UFD.

GCDs and factorizations

Lemma 8.4 If R is a UFD and S ⊆ R then a gcd of S exists.
Proof. The relation ∼ is an equivalence relation on the set of irreducibles in R. So by
choosing a representative irreducible from each equivalence class we can construct a set P
of∏pairwise non-associate∏irreducible elements of R. We can write any element a ∈ R as
u p∈P pnp and if∏b = v p∈S pmp then U2∏implies a | b iff np ≤ mp for all p. Write each
ai ∈ S as ai = ui p∈P pni,p . If we let d = p∈P pmp with mp = minai ∈S ni,p then it is clear
that d is a gcd for S.

A partial converse to Lemma 8.4 is true.

Lemma 8.5 If R is an ID in which the gcd of any pair of elements exists then every
irreducible is prime.
Proof. First we prove that if gcds exist then gcd(ab, ac) ∼ a gcd(b, c). Let e = gcd(ab, ac)
and d = gcd(b, c). Then d | b, c, so ad | ab, ac, so ad | e. Writing e = adu then e | ab, ac, so
du | b, c, so du | d. Thus u ∈ R× and e ∼ ad (or d = 0 = e).
Now let p be an irreducible and assume p ̸ | a, b. Then gcd(p, b) ∼ 1 since the gcd must
be a factor or p and p ̸ | b. Hence gcd(p, ab) | gcd(ap, ab) ∼ a. But gcd(p, ab) | p, so
gcd(p, ab) | gcd(a, p) ∼ 1. Hence gcd(p, ab) ∼ 1 and p̸ | ab. Hence p is prime.

Lemma ∑ 8.6 If R is an ID in which every set S has a gcd which can be written in the
form ri ai for some ai ∈ S, ri ∈ R, then R is a PID.
∑
Proof. Let I be an ideal and write I = (S) for some S (e.g., S = I). Let d = ri ai be
a gcd∑of S. Then d | a for all a ∈ S. Hence a ∈ (d), so S ⊆ (d). Thus I ⊆ (d). However
d= ri ai ∈ I. Then (d) ⊆ I. Hence I = (d) is principal.

16
7261 9. Factorization of Polynomials Fall 2017
Assume throughout this section that R is a UFD.
∑
Let f (X) = ni=0 ai X i ∈ R[X]. Define the content of f (X) to be c(f ) = gcd{a0 , a1 , . . . , an }.
Note that if f ̸= 0 then c(f ) ̸= 0. We call f primitive iff c(f ) ∼ 1.
Note that monic polynomials are primitive, but not conversely, e.g. 2X + 3 ∈ Z[X].

Lemma (Gauss) If R is a UFD and f, g ∈ R[X] are primitive, then so is f g.

Proof. Assume otherwise and let p be a prime dividing c(f g). Reducing the polynomials
mod p we get f¯, ḡ ∈ (R/(p))[X] with f¯, ḡ ̸= 0, but f¯ḡ = f g = 0 (the map f 7→ f¯
R[X] → (R/(p))[X] is a special case of the evaluation homomorphism evπ,X where X is
sent to X and evπ,X acts as the projection map π : R → R/(p) on constants). Now p
is prime, so (p) is a prime ideal and R/(p) is an ID. Hence f¯, ḡ ̸= 0 implies f¯ḡ ̸= 0, a
contradiction.

Corollary 9.1 If R is a UFD then c(f g) ∼ c(f )c(g).

Proof. The result clearly holds if f or g is zero, so assume f, g ̸= 0 and hence c(f ) ̸=
0. Since gcd{aai } ∼ a gcd{ai }, c(af ) ∼ ac(f ) for all a ∈ R. But f /c(f ) ∈ R[X], so
c(f )c(f /c(f )) = c(f ) and so f /c(f ) is primitive. Now f g/(c(f )c(g)) = (f /c(f ))(g/c(g)) is
primitive. Hence c(f g) ∼ c(f )c(g)c(f g/c(f )c(g)) ∼ c(f )c(g).

Lemma 9.2 If deg f > 0 and f is irreducible in R[X] then f is irreducible in F [X],
where F = Frac R is the field of fractions of R.
Proof. Suppose f = gh in F [X]. By multiplying by denominators, there exist non-zero
a, b ∈ R with ag, bh ∈ R[X]. Thus abf = (ag)(bh) ∈ R[X] and c(abf ) ∼ c(ag)c(bh). But
/ (R[X])× =
f = c(f )(f /c(f )) is a factorization of f in R[X] and if deg f > 0, f /c(f ) ∈
R . Thus c(f ) ∈ R and so c(abf ) ∼ ab. Now ab/c(ag)c(bh) = u ∈ R× and f =
× ×

(u−1 ag/c(ag))(bh/c(bh)) is a factorization of f in R[X]. Hence either deg g = 0 or deg f = 0

and so g or h is a unit in F [X].

Lemma 9.3 If R is a UFD then f ∈ R[X] is irreducible iff either

(a) f ∈ R is an irreducible in R, or (b) f is primitive in R[X] and irreducible in F [X].
Proof. Assume first that deg f = 0. If f = ab in R, f = ab in R[X]. Conversely, if f = gh
in R[X] then deg g = deg h = 0, so f = gh in R. Since R× = (R[X])× , irreducibility in
R[X] is equivalent to irreducibility in R. Assume now that deg f > 0. If f is irreducible
in R[X] then by the previous lemma, f is irreducible in F [X]. Also, f = c(f )(f /c(f )), so
c(f ) ∈ (R[X])× = R× and f is primitive. Conversely, if f is primitive and irreducible in
F [X] and f = gh in R[X], then f = gh in F [X], so wlog g ∈ (F [X])× ∩ R[X] = R. But
then g | c(f ) in R, so g ∈ R× = (R[X])× . Thus f is irreducible in R[X].

17
Theorem 9.4 If R is a UFD then R[X] is a UFD.
Proof. Write f = c(f )f ′ where f ′ is primitive. Now c(f ) = up1 . . . pr where u ∈ R× =
(R[X])× and pi are irreducible in R. If f ′ = gh with g, h ∈ / (R[X])× = R× then
c(g)c(h) ∼ 1, so g, h are primitive and deg g, deg h > 0 (since otherwise either g or h
would lie in R× ). ∏By induction on the degree, f ′ is the product of irreducible primitive
polynomials f ′ = fi . Hence f has a factorization into irreducibles.
Now assume f = up1 . . . pr f1 . . . ft = vq1 . . . qs g1 . . . gu where u, v ∈ R× , p1 , qj are irre-
ducible in R and fi , gj are primitive and irreducible in F [X]. The ring F [X] is a PID,
so is a UFD. The elements up1 . . . pr and vq1 . . . qs are units in F [X], so t = u and wlog
fi = γi gi for some γi ∈ (F [X])× = F \ {0}. Write γi = ai /bi with ai , bi ∈ R. Now
bi fi = ai gi , so bi ∼ c(bi fi ) = c(ai gi ) ∼ ai . Thus γi ∈ R× and fi ∼ gi in R[X]. Now
c(f ) ∼ up1 . . . pr ∼ vq1 . . . qs , so by unique factorization in R, r = s and wlog pi ∼ qi in R
and hence in R[X]. Hence the factorization of f is unique in R[X].

Factorization methods
Evaluation method: If g | f in R[X] then g(c) | f (c) in R for all c ∈ R.
Example: If f = X 3 − 4X + 1 ∈ Z[X], then f (±2) = 1. If f = gh then we can assume wlog
that g is linear. But then g(±2) = ±1. The only linear polynomials with this property are
±X/2 which do not lie in Z[X]. Hence f is irreducible in Z[X] (and hence also in Q[X]).
Reduction mod p: If f = gh in R[X] and p is a prime then f¯ = ḡ h̄ in (R/(p))[X].
Example: If f = X 4 − X 2 + 4X + 3 ∈ Z[X], then if p = 2, f¯ = X 4 + X 2 + 1 =
(X 2 +X +1)(X 2 +X +1) in (Z/2Z)[X] and if p = 3 then f¯ = X 4 −X 2 +X = X(X 3 −X +1)
in (Z/3Z)[X]. In Z[X], f cannot factor as a product of two quadratics (since there is no
quadratic factor mod 3), nor can it have a linear factor (no linear factor mod 2), hence f
is irreducible in Z[X].
∑
Lemma (Eisenstein’s irreducibility criterion) Assume R is a UFD, f = ni=0 an X n ∈
R[X], is primitive, and p is a prime such that p ̸ | an , p | ai for i < n and p2 ̸ | a0 . Then f
is irreducible in R[X].
Proof. Suppose f = gh. Then ḡ h̄ = an X n in (R/(p))[X]. Thus ḡ = aX i and h̄ = bX j
for some a, b ∈ R/(p) and i + j = n. But deg g + deg h = n and i ≤ deg g, j ≤ deg h.
Hence i = deg g and j = deg h. If g and h are not units in R[X] and f is primitive then
deg g, deg h > 0. Hence ḡ(0) = h̄(0) = 0, so p | g(0), h(0). Thus p2 | g(0)h(0) = f (0) = a0 ,
a contradiction. Hence f is irreducible.

Exercises
1. Show that for p a prime in Z, f (X) = 1 + X + . . . X p−1 = (X p − 1)/(X − 1) is
irreducible in Q[X] [Hint: consider f (X + 1) and use Eisenstein’s criterion].
2. Let f = X 3 − X + 1. Show that (Z/3Z)[X]/(f ) is a field with 27 elements.

18
7261 10. Symmetric Polynomials Fall 2017
A polynomial f (X1 , . . . , Xn ) ∈ R[X1 , . . . , Xn ] is called symmetric if
f (X1 , . . . , Xn ) = f (Xπ(1) , . . . , Xπ(n) )
for any permutation π ∈ Sn .

Examples X12 +X22 +X32 and X1 X2 +X2 X3 +X3 X1 are symmetric polynomials in the ring
Z[X1 , X2 , X3 ], however X12 X2 + X22 X3 + X32 X1 is not symmetric (consider the permutation
π = (12)).

The
∑ elementary symmetric ∑ polynomials
∏ σr ∈ R[X1 , . . . , Xn ] are defined by σr =
i1 <i2 <···<ir Xi1 . . . Xir = |S|=r i∈S Xi where in the second expression the sum is over
all subsets S of {1, . . . , n} of size r.

Examples For n = 3, σ0 = 1, σ1 = X1 + X2 + X3 , σ2 = X1 X2 + X2 X3 + X3 X1 ,
σ3 = X1 X2 X3 .

Note: (X + X1 )(X + X2 ) . . . (X + Xn ) = X n + σ1 X n−1 + σ2 X n−2 + · · · + σn .

Define the degree of cX1a1 . . . Xnan ∈ R[X1∑ , . . . , Xn ], c ̸= 0, as the n-tuple (a1 , . . . , an ).
a1 an
More generally define the degree of f = a1 ,...,an ca1 ,...,an X1 . . . Xn as the maximum
value of (a1 , . . . , an ) over all ca1 ,...,an ̸= 0, where n-tuples are ordered lexicographically:
(a1 , . . . , an ) < (b1 , . . . , bn ) iff there exists an i such that ai < bi and aj = bj for all j < i.

Example In R[X1 , X2 , X3 ], deg(X12 X29 + X17 X3 ) = (7, 0, 1).

In R[X1 , . . . , Xn ], deg σr = (1, . . . , 1, 0, . . . , 0), where there are r ones and n − r zeros.

Lemma 10.1 The lexicographic ordering on Nn is a well ordering: Nn is totally ordered

and every non-empty S ⊆ Nn has a minimal element.
Proof. To prove every S ̸= ∅ has a minimal element, inductively construct sets Si with
S0 = S and Si equal to the set of elements (a1 , . . . , an ) of Si−1 for which ai is minimal. It
̸ ∅ and the (unique) element of Sn is the minimal element of S.
is clear that Si =

Lemma 10.2 If f ∈ R[X1 , . . . , Xn ] is symmetric and deg f = (a1 , . . . , an ) then a1 ≥

a2 ≥ · · · ≥ an .
Proof. Assume otherwise and let ai < aj with i > j. Then if π = (ij), f (X1 , . . . , Xn ) =
f (Xπ(1) , . . . , Xπ(n) ) has a term with degree (aπ(1) , . . . , aπ(n) ) which is larger than (a1 , . . . , an ),
contradicting the definition of the degree.

Lemma 10.3 If f, g ∈ R[X1 , . . . , Xn ] and f, g are monic (the term with degree equal to
deg f or deg g has coefficient 1) then deg f g = deg f + deg g where addition of degrees is
performed componentwise: (a1 , . . . , an ) + (b1 , . . . , bn ) = (a1 + b1 , . . . , an + bn ).

19
Proof. Prove that in the lexicographical ordering, a < b and c ≤ d imply a + c < b + d.
The rest of the proof is the same as for the one variable case.

Theorem 10.1 The polynomial f ∈ R[X1 , . . . , Xn ] is symmetric iff f ∈ R[σ1 , . . . , σn ].

Clearly σi is symmetric, and the set of symmetric polynomials forms a subring of the ring
R[X1 , . . . , Xn ]. Hence every element of R[σ1 , . . . , σn ] is symmetric. We now need to show
every symmetric polynomial can be written as a polynomial in σ1 , . . . , σn . We use induction
on deg f . Let f be a counterexample with minimal deg f (using Lemma 1). Let deg f =
(a1 , . . . , an ) and let the leading term have coefficient c ∈ R. Then g = cσ1a1 −a2 σ2a2 −a3 . . . σnan
has deg g = (a1 , . . . , an ) = deg f (by Lemma 3) and the same leading coefficient c. Thus
deg(f − g) < deg f . Now g is symmetric, so f − g is symmetric. By induction on deg f ,
f − g ∈ R[σ1 , . . . , σn ]. But g ∈ R[σ1 , . . . , σn ]. Hence f ∈ R[σ1 , . . . , σn ], contradicting the
choice of f .
If α ∈ R′ and R is a subring of R′ , we call α algebraic over R if the map evα : R[X] →
R′ is not injective, i.e., there exists a non-zero f (X) ∈ R[X] with f (α) = 0. More
generally we say α1 , . . . , αn are algebraically dependent if evα1 ,...,αn : R[X1 , . . . , Xn ] →
R′ is not injective, or equivalently there exists a non-zero polynomial f ∈ R[X1 , . . . , Xn ]
with f (α1 , . . . , αn ) = 0. We say α1 , . . . , αn are algebraically independent over R if they
are not algebraically dependent.

Theorem 10.2 The elements σ1 , . . . , σn are algebraically independent over R. The ele-
ments Xi are algebraic over R[σ1 , . . . , σn ].
∑
Proof. Assume ca1 ,...,an σ1a1 . . . σnan = 0 in R[X1 , . . . , Xn ]. Among the (finite set of)
(b1 , . . . , bn ) such that cb1 ,...,bn ̸= 0, pick one such that (b1 + · · · + bn , b2 + · · · + bn , . . . , bn )
is maximal in the lexicographical ordering. The map sending (a1 , . . . , an ) to (a1 + · · · +
an , a2 + · · · + an ,∑ . . . , an ) is an injection Nd to Nd , so this (b1 , . . . , bn ) is uniquely deter-
mined. Now deg ca1 ,...,an σ1a1 . . . σnan = (b1 + · · · + bn , b2 + · · · + bn , . . . , bn ) contradicting
∑
ca1 ,...,an σ1a1 . . . σnan = 0. Thus σ1 , . . . , σn are algebraically independent. The elements Xi
are algebraic over R[σ1 , . . . , σn ] since they are roots of Xn − σ1 X n−1 + · · · ± σn = 0.

As a consequence of Theorem 2, any symmetric polynomial f ∈ R[X1 , . . . , Xn ] can be

written as g(σ1 , . . . , σn ) with g a unique element of R[X1 , . . . , Xn ]. For example, X12 +
X22 + X32 = σ12 − 2σ2 .

Exercises
∏
1. Let δ = i<j (Xi − Xj ) ∈ Z[X1 , . . . , Xn ]. Show that δ 2 is symmetric and for n = 3
express δ 2 in terms of σ1 , σ2 , σ3 .

2. Let f (X) = X 3 − 3X + 5 have complex roots α1 , α2 , α3 . Find a polynomial with

complex roots α12 , α22 , α32 .

20