1.

important definitions
This section is incomplete. It does not contain all important definitions introduced in my lectures.
Please see the class note.
Definition 1.1. A commutative ring R with 1 , 0 is called an integral domain if whenever ab = 0
for a, b ∈ R, then either a = 0 or b = 0.
A commutative ring R with 1 , 0 is called a field if for every nonzero element a ∈ R, there exists
an element b ∈ R such that ab = 1.
Definition 1.2. Let R be a ring.
• A nonzero element a ∈ R is called a zero divisor if there exists a nonzero element b in R
such that either ab = 0 or ba = 0.
• An element a ∈ R is called a nonzero divisor if a , 0 and a is not a zero divisor.
• An element a ∈ R is called a nilpotent element if an = 0 for some positive integer n.
• An element a ∈ R is called a unit if there exists another element b ∈ R such that ab = ba = 1.
Definition 1.3. Let R be a commutative ring with 1 , 0. A polynomial in an indeterminate x with
coefficients in R is a formal sum of the form a0 xn +a1 xn−1 +. . .+an . As per convention the polynomial
1.x is simply denoted by x. We define addition and multiplication of polynomials as follows:
(a0 xn + a1 xn−1 + . . . + an ) + (b0 xm + b1 xm−1 + . . . + bm )
(a0 + b0 )xn + (a1 + b1 )xn−1 + . . . + (an + bn ) if n = m





= a0 x + a1 xn−1 + . . . + an−m−1 xm+1 + (an−m + b0 )xm + (an−m+1 + b1 )xm−1 + . . . + (a0 + b0 ) if n > m

 n


b0 xm + b1 xm−1 + . . . + bm−n−1 xn+1 + (bm−n + a0 )xn + (bm−n+1 + a1 )xn−1 + . . . + (a0 + b0 ) if n < m



(a0 xn + a1 xn−1 + . . . + an )(b0 xm + b1 xm−1 + . . . + bm ) = an bm xm+n + (a0 b1 + a1 b0 )xm+n−1 + . . . +

X
ai b j xk + . . . + a0 b0 .
i+ j=m+n−k,0≤i≤n,0≤ j≤m

The set of polynomials is denoted by R[x]. It forms a commutative ring with identity under
addition and multiplication.
Definition 1.4. Let R be a commutative ring with 1 , 0. A power series in an indeterminate x with
coefficients in R is a formal infinite sum of the form a0 + a1 x + . . . + an xn + . . . . The addition and
multiplication of power series are defined as follows:
(a0 + a1 x + . . . an xn + . . .) + (b0 + b1 x + . . . + bn xn + . . .) = (a0 + b0 ) + (a1 + b1 )x + . . .
+ (an + bn )xn + . . .
(a0 + a1 x + . . . an xn + . . .)(b0 + b1 x + . . . + bn xn + . . .) = a0 b0 + (a0 b1 + a1 b0 )x + . . .
X
+ ai b j x n + . . .
i+ j=n,0≤i≤n,0≤ j≤n

The set of power series forms a commutative ring with identity under addition and multiplication
and it is denoted by R[|x|]. As before the element 1.x is denoted by x.
1
Definition 1.5. Let R be a ring and Mn (R) be the set of all n × n matrices with entries in R. Addition
and multiplication of matrices are defined as follows:
(ai j )n×n + (bi j )n×n = (ai j + bi j )n×n
n
X
(ai j )n×n (bi j )n×n = (ci j )n×n where ci j = aik bk j .
k=1

The set Mn (R) forms a ring under addition and multiplication of matrices.

2. Important problems
In this section, all rings are commutative and contain nonzero identities.
Question 2.1. Show that a polynomial f ∈ R[x] is a unit in R[x] if and only if f (0) is a unit in R
and all other coefficients of f are nilpotent in R.
Proof. Let f = a0 + a1 x + . . . + an xn where ai ∈ R for i = 0, 1, 2, . . . , n. First we assume that
f (0) = a0 is a unit in R and rest of the coefficients a1 , . . . , an are nilpotent elements of R. The
polynomial g = a1 x + . . . + an xn is a nilpotent element of R[x] since it is a finite sum of nilpotent
elements a1 x, a2 x, . . . , an x. This implies that f = a0 + g is a sum of a unit and a nilpotent element
of R[x]. Therefore f is a unit of R[x].
Conversely, we assume that f is a unit in R[x]. We prove the converse by induction on deg( f ). If
deg( f ) = 0, then f is a constant and there is nothing to prove. We assume that deg( f ) > 0. Since f
is a unit, there exists another polynomial g ∈ R[x] such that f g = g f = 1. Evaluating at x = 0, we
conclude that f (0), g(0) are units in R. Let g = b0 + b1 x + . . . + bm xm . We set ai = bi = 0 for i < 0
(Why is it required?). We have
X
f g = an bm xm+n + (an bm−1 + an−1 bm )xm+n−1 + . . . + ( ak bl )xi + . . . (a1 b0 + a0 b1 )x + a0 b0 .
k+l=i

Comparing coefficients of powers x for i = m + n, m + n − 1, . . . , n in both sides of f g = 1 we find

i

the following system of equations:

an bm = 0 (1)
an bm−1 + an−1 bm = 0 (2)
an bm−2 + an−1 bm−1 + an−2 bm = 0 (3)
an bm−3 + an−1 bm−2 + an−2 bm−1 + an−3 bm = 0 (4)
......
an b0 + an−1 b1 + . . . + . . . an−m bm = 0.
Multiplying both sides of (2) by an , we conclude from equations (1), (2) that a2n bm = a2n bm−1 = 0.
Now multiply both sides of (3) by a2n . It follows that a3n bm = a3n bm−1 = a3n bm−2 = 0. Continuing
in this manner we conclude that anm+1 bi = 0 for all i = 0, 1, . . . , m. But b0 = g(0) is a unit in R.
Therefore, anm+1 = 0, i.e. an is a nilpotent element.
Let h = a0 + a1 x + . . . + an−1 xn−1 = f − an xn . Since h is a difference between a unit and a
nilpotent element of R[x], it is also a unit in R[x]. We observe that deg(h) ≤ n − 1. By the induction
2
hypothesis, we conclude that a0 = h(0) is a unit in R and other coefficients a1 , . . . , an−1 are nilpotent
elements of R. Hence the converse follows. □
We proved the following in the class.
Question 2.2. Let f ∈ R[x] be a zero divisor. Show that there exists b ∈ R, b , 0 such that b f = 0.
Proof. Since f is a zero divisor, we have a nonzero polynomial g ∈ R[x] such that f g = 0. If we
can choose g ∈ R, we are through. Otherwise we choose g to be a nonzero polynomial of least
possible degree such that f g = 0 holds. Let f = a0 + a1 x + . . . + an xn and g = b0 + b1 x + . . . + bm xm .
Set ai = bi = 0 for i < 0. Comparing coefficients of various powers of x in both sides of f g = 0,
we have the following system of equations.
an bm = 0 (5)
an bm−1 + an−1 bm = 0 (6)
an bm−2 + an−1 bm−1 + an−2 bm = 0 (7)
an bm−3 + an−1 bm−2 + an−2 bm−1 + an−3 bm = 0 (8)
......
an b0 + an−1 b1 + . . . + . . . an−m bm = 0
an−1 b0 + an−2 b1 + . . . + . . . an−1−m bm = 0
......
a1 b0 + a0 b1 = 1
a0 b0 = 0
Equation 5 implies that deg(an g) < deg(g). Moreover f × (an g) = 0. From minimality of the
degree of g, we conclude that an g = 0. In particular we have an bm−1 = 0. Equation 6 gives that
an−1 bm = 0. It follows that deg(an−1 g) < deg(g). We have f × (an−1 g) = 0. Again from the
minimality of the degree of g, it follows that an−1 g = 0. Continuing in this manner we prove that
ai g = 0 for 0 ≤ i ≤ n.
The polynomial g is nonzero, so it has a nonzero coefficient, say b j . The above implies that
b j f = 0. Hence the result is proved. □
The following was discussed in the class.
Question 2.3. Let f ∈ R[|x|] be a formal power series in an indeterminate x with coefficients in R.
Then show that f is invertible in R[|x|] if and only if f (0) is invertible in R
We prove the following.
Question 2.4. A polynomial f ∈ R[x] is a nilpotent element of R[x] if and only if its all coefficients
are nilpotent elements.
Proof. If all coefficients of f are nilpotent elements, then f is a finite sum of nilpotent elements of
R[x]. This implies that f itself is a nilpotent element of R[x].
3
 We prove the converse by induction on deg( f ). We assume that f is a nilpotent element of R[x].
If deg( f ) = 0, then f is a constant and nothing remains to prove. Let f = a0 + a1 x + . . . + an xn . We
find a positive integer p such that f p = 0. Comparing leading coefficients in both sides, it follows
that anp = 0. The polynomial g = f − an xn is a nilpotent element because it is a difference between
two nilpotent elements. Moreover deg(g) < n. By induction hypothesis, all coefficients of g are
nilpotent. Therefore, the converse follows. □
Question 2.5. Let f ∈ R[|x|] be a nilpotent power series. Prove that all coefficients of f are
nilpotent elements of R. Show that the converse need not be true.
Question 2.6. Show that (x, y) ∩ (z, w) = (xz, xw, yz, yw) in the polynomial ring R[x, y, z, w] with
coefficients in a ring R.
Proof. The inclusion (xz, xw, yz, yw) ⊂ (x, y) ∩ (z, w) is easy. To prove the reverse inclusion, we
choose f ∈ (x, y) ∩ (z, w). We may write f as
f = g + h(x, y) + k(z, w)
for some polynomial g ∈ (xz, xw, yz, yw), h(x, y) ∈ (x, y) ⊂ R[x, y] and k(z, w) ∈ (z, w) ⊂ R[z, w].
Now f ∈ (x, y) implies that k(z, w) ∈ (x, y) since g, h ∈ (x, y). We write k(z, w) = xp + yq for some
polynomials p, q ∈ R[x, y, z, w]. Evaluating at x = y = 0, we conclude that k(z, w) = 0. Similarly
we have h(x, y) = 0. It follows that f ∈ (xz, xw, yz, yw). The choice of f is arbitrary. Therefore
(x, y)∩(z, w) ⊂ (xz, xw, yz, yw). Hence we establish the equality (x, y)∩(z, w) = (xz, xw, yz, yw). □
Examples 2.7. • The ring Z has infinitely many prime ideals. All nonzero prime ideals of Z
are maximal ideals.
• Let k be a field and f ∈ k[x, y]. Assume that (a, b) ∈ k×k is a zero of k, i.e. f (a, b) = 0. Then
(x − a, y − b) is a maximal ideal of k[x, y] containing f . Each zero (a, b) of f corresponds to
a maximal ideal (x − a, y − b) of the quotient ring R = k[x,y] (f)
. Thus if f has infinitely many
zeros, then R has infinitely many maximal ideals.
• Let n = pα1 1 . . . pαk k be the factorisation of n ∈ N, n > 1. Then the ring Z/(n) has exactly k
maximal ideals, viz. (p1 ), . . . , (pk ).
• Let k be a field. Every element of the power series ring k[|x|] can be written as uxk where k
is a nonnegative integer and u is a unit of k[|x|].
Question 2.8. Let R, S be rings with nonzero identities which are not necessarily commutative.
Prove that every ideal of R × S is of the form I × J where I, J are ideals of R, S respectively.
Proof. Let pr1 : R × S → R, pr2 : R × S → S be given by pr1 (r, s) = r and pr2 (r, s) = s
respectively. One easily checks that pr1 , pr2 are surjective ring homomorphisms. Let I be an ideal
of R × S . Since pr1 , pr2 are surjective ring homomorphisms, their images pr1 (I), pr2 (J) are ideals
of R, S respectively. We claim that I = pr1 (I) × pr2 (I).
The inclusion I ⊂ pr1 (I) × pr2 (I) is easy. We choose x = (r, s) ∈ pr1 (I) × pr2 (I). Then there exist
u, v ∈ I such that pr1 (u) = r and pr2 (v) = s. This implies that u, v are of the form u = (r, a) and
v = (b, s). Now we write x as
x = (r, s) = (r, a)(1R , 0) + (b, s)(0, 1S ).
4
Here 1R , 1S denote identities of R and S respectively. The RHS of the above equality belongs to I
since both (r, a), (b, s) ∈ I. Thus we obtain x ∈ I. Therefore pr1 (I) × pr2 (I) ⊂ I. Hence our claim is
established and the result follows. □
Notation : Let R be a commutative ring, r ∈ R and s a unit in R. Then rs−1 = s−1 r. We denote any
of these equal elements by r/s = rs .

3. Zorn’s Lemma
All rings in this section are nonzero, not necessarily commutative and may not contain identities.
Definition 3.1. A partial order on a nonempty set A is a relation ≺ satisfying the following :
1 x ≺ x for all x ∈ A,
2 if x ≺ y and y ≺ x, then x = y for any x, y ∈ A,
3 if x ≺ y and y ≺ z, then x ≺ z for any x, y, z ∈ A.
The set A is called a partially ordered set. It is denoted by (A, ≺).
Definition 3.2. (Maximal elements, upper bonds, chains) Let (A, ≺) be a partially ordered set and
B be a nonempty subset of A. We consider B partially ordered with the partial order ≺. Then
(1) The set B is called a chain if for any two elements x, y ∈ B, we have either x ≺ y or y ≺ x.
(2) An element u ∈ A is called an upper bound of B if x ≺ u for all x ∈ B.
(3) An element m ∈ B is called a maximal element of B if there does not exist another element
x ∈ B such that m ≺ x.
Examples 3.3. Let A = {{1, 2, 3}, {1, 2}, {2, 3}, {1, 3}, {1}, {2}, {3}, } be partially ordered by inclusion.
Let B = {{1, 2}, {2, 3}, {1, 3}, {1}, {2}, {3}}, then each of {1, 2}, {2, 3}, {1, 3} is a maximal element of B.
The element {{1, 2, 3} is an upper bound of B. The set B does not contain an upper bound.
Lemma 3.4. (Zorn’s Lemma) If A is a nonempty partially ordered set such that every chain has an
upper bound in A, then A has a maximal element.
Theorem 3.5. Let R be a ring, I be an ideal of R and S be a multiplicatively closed subset of R
such that S ∩ I = ∅. Then the following set has a maximal element.
S = {J : J is an ideal of R, I ⊂ J and J ∩ S = ∅}
If R is commutative, each maximal element is a prime ideal of R.
Proof. The set S is a partially ordered set ordered by inclusion. It is nonempty since I ∈ S. We
show that every chain in S has an upper bound in S. Then by Zorn’s lemma, we conclude that S
has a maximal element.
Let C be a chain in S. To show that C has an upper bound in S, we consider the set T = ∪ J∈C J.
Let x, y ∈ T and r ∈ R. There exist J1 , J2 ∈ C such that x ∈ J1 and y ∈ J2 . Since C is a chain,
we have either J1 ⊂ J2 or else J2 ⊂ J1 . In either cases, we obtain that x − y ∈ T . We also have
rx ∈ J1 ⊂ T . It follows that T is an ideal of R. Moreover T ∩ S = ∅ and I ⊂ T since each J ∈ C
is disjoint from S and contains I. Therefore T ∈ C. It follows that T is an upper bound of C. This
completes the first part of the proof.
5
 Now we assume that R is a commutative ring. Let M be a maximal element of S. To show that
M is a prime ideal, we choose a, b ∈ R such that ab ∈ M. Assume on the contrary that a < M and
b < M. We observe that I ⊂ M ⊂ M + (a) and I ⊂ M ⊂ M + (b). Since M is a maximal element of
S, we must have [M + (a)] ∩ S , ∅, [M + (b)] ∩ S , ∅. This implies that there exist s1 , s2 ∈ S such
that s1 = m1 + r1 a and s2 = m2 + r2 b for m1 , m2 ∈ M and r1 , r2 ∈ R. Consequently we have
s1 s2 = (m1 + r1 a)(m2 + r2 b)
= m1 m2 + (m1 r2 b + r1 am2 + r1 ar2 b)
∈ M, since R is commutative and m1 , m2 , ab ∈ M.
This is a contradiction because M ∩ S = ∅. Therefore either a ∈ M or b ∈ M. Hence M is a prime
ideal. □
Corollary 3.6. Let R be a ring with 1 , 0. Then every proper ideal of R is contained in a maximal
ideal of R. If R is a commutative ring, then maximal ideals are prime ideals.
Proof. Let S = {1}. Then S is multiplicatively closed. Let I be a proper ideal of R. Clearly
S ∩ I = ∅. In this situation the set S described in Theorem 3.5 reduces to
S = {J : J is a proper ideal of R contatining I}.
Clearly every maximal element of S is a maximal ideal of R containing I. By Theorem 3.5, S
has a maximal element. Therefore I is contained in a maximal ideal.
To show that maximal ideals are prime ideals, we take I = 0. In this case S is given by
S = {J : J is a proper ideal of R}.
Any maximal ideal of R is a maximal element of S and therefore a prime ideal of R by Theorem
3.5. □
Remark 3.7. Consider the set Q of rational numbers as a ring with usual addition and trivial
multiplication (ab = 0 for a, b ∈ Q). Then Q does not have a maximal ideal. The ring Q does not
have the identity.
Definition 3.8. Let I be an ideal of R. A prime ideal P containing I is called a minimal prime over
I if there does not exist another prime ideal Q such that I ⫋ Q ⫋ P.
Question 3.9. Let I be a proper ideal of a ring R with 1 , 0. Show that I is contained in a minimal
prime ideal over I.
Proof. We consider the set
S = {P : P is a prime ideal containing I}.
We define a relation ≺ on S as follows:
P ≺ Q for P, Q ∈ S whenever Q ⊂ P.
It is easy to check that ≺ defines a partial order on S. We claim that any maximal element M of
S is a minimal prime over I. If it were not true, we would find a prime ideal M ′ containing I such
that I ⫋ M ′ ⫋ M. This implies that M ′ ∈ S and M ≺ M ′ contradicting the maximality of M in S.
6
 We use Zorn’s lemma to show that S admits a maximal element. Since I is a proper ideal, I is
contained in a maximal ideal M0 of R. Clearly M0 ∈ S. Therefore S is a nonempty set.
Let C be a chain in S. We show that C has an upper bound contained in S. We define J = ∩P∈C P.
We have I ⊂ J since each prime in C contains I. If we can show that J is a prime ideal, then J ∈ S
and J is an upper bound of C.
Let ab ∈ J for a, b ∈ R. If possible, we assume that a, b < J. Then there exist Q, Q′ in C such
that a < Q and b < Q′ . Since C is a totally ordered set, we have either Q ⊂ Q′ or else Q′ ⊂ Q. If
Q ⊂ Q′ , we have a, b < Q. This implies that ab < Q since Q is a prime ideal. Similarly if Q′ ⊂ Q,
we conclude that ab < Q′ . Thus in either case we have ab < J which is a contradiction. Therefore
either a ∈ J or b ∈ J. It follows that J is a prime ideal.
Thus we establish that every nonempty chain of S has an upper bound in S. By Zorn’s lemma
S has an upper bound and the result follows. □

4. Rings of fractions
Let R be a nonzero commutative ring which does not necessarily contain identity. Let S be a
multiplicatively closed subset of R such that S does not contain the zero element and zero divisors.
We recall the construction of S −1 R. We define a relation ∼ on R × S by setting (r, s) ∼ (r′ , s′ )
if rs′ = sr′ . It was shown that ∼ defines an equivalence relation. As a set S −1 R is the set of
equivalence classes under ∼, i.e.
R×S
S −1 R = = {[(r, s)] : (r, s) ∈ R × S }.
∼
The set S −1 R is a commutative ring with identity [(s, s)] for s ∈ S under the following well
defined operations.
Addition : [(r, s)] + [(r′ , s′ )] = [(rs′ + sr′ , ss′ )],
Multiplication : [(r, s)][(r′ , s′ )] = [(rr′ , ss′ )].
The map i : R → S −1 R given by i(x) = [(xs, s)], s ∈ S is an injective ring homomorphism and
each element of S −1 R can be written as i(r)/i(s) = i(r)i(s)−1 for r ∈ R, s ∈ S . The map i is called
the inclusion map.

The following result was proved in the class.

Theorem 4.1. Let R be a commutative ring and S be a multiplicatively closed subset of R such
that S does not contain the zero element and zero divisors. Let S −1 R be the ring of fractions and
i : R → S −1 R be the inclusion map. Let ϕ : R → Q be a ring homomorphism such that ϕ(s) is
invertible in Q for all s ∈ S . Then there exists a unique ring homomorphism ϕ̃ : S −1 R → Q such
that the following diagram commutes.

= QO
ϕ
ϕ̃

R / S −1 R
i
7
Proof. We know that each element of S −1 R can be written as i(r)/i(s) for r ∈ R, s ∈ S . We define a
map ϕ̃ : S −1 R → Q by ϕ̃(i(r)/i(s)) = ϕ(r)ϕ(s)−1 .
First we show that ϕ̃ is well defined. Assume that i(r)/i(s) = i(r′ )/i(s′ ). Then i(r)i(s′ ) = i(r′ )i(s).
This implies that i(rs′ ) = i(r′ s). Since the map i is injective, we have rs′ = r′ s. It follows that
ϕ(rs′ ) = ϕ(r′ s) which yields ϕ(r)ϕ(s′ ) = ϕ(r′ )ϕ(s). Therefore,

ϕ̃(i(r)/i(s)) = ϕ(r)/ϕ(s) = ϕ(r′ )/ϕ(s′ ) = ϕ̃(i(r′ )/i(s′ ))

which shows that ϕ̃ is well defined. For r, r′ ∈ R and s, s′ ∈ S , we have

i(r)i(s′ ) + i(r′ )i(s)
ϕ̃(i(r)/i(s) + i(r′ )/i(s′ )) = ϕ̃( )
i(s)i(s′ )
i(rs′ + r′ s)
= ϕ̃( )
i(ss′ )
= ϕ(rs′ + r′ s)ϕ(ss′ )−1
= (ϕ(r)ϕ(s′ ) + ϕ(r′ )ϕ(s))ϕ(s)−1 ϕ(s′ )−1
= ϕ(r)ϕ(s)−1 + ϕ(r′ )ϕ(s′ )−1
= ϕ̃(i(r)/i(s)) + ϕ̃(i(r′ )/i(s′ )).

One can prove similarly that

i(r) i(r′ )
ϕ̃( ) = ϕ̃(i(r)/i(s))ϕ̃(i(r′ )/i(s′ ))
i(s) i(s′ )
The above establishes that ϕ̃ is a well defined ring homomorphism.
We show that ϕ̃ is unique. If possible assume that there exists another ring homomorphism ψ
such that ψ ◦ i = ϕ. This shows that ψ(i(r)) = ϕ(r) for all r ∈ R. One also has ψ(i(s)−1 ) = ψ(i(s))−1
for all s ∈ S since ψ is a ring homomorphism. Let r ∈ R and s ∈ S . Then
i(r) i(r)
ψ( ) = ψ(i(r)i(s)−1 ) = ψ(i(r))ψ(i(s))−1 = ϕ(r)ϕ(s)−1 = ϕ̃( ).
i(s) i(s)
Therefore ϕ̃ = ψ and the uniqueness is established. □

Convention: The map i : R → S −1 R is injective. Unless there is a chance of ambiguity, we identify

i(r)
i(r) with r for r ∈ R and the element i(s) ∈ S −1 R is written as rs . Thus R becomes a subring of S −1 R.
The following lemma follows from the definition of integral domains.

Lemma 4.2. Let R be a commutative ring with 1 , 0 and S = R \ {0}. Then R is an integral domain
if and only if S is multiplicatively closed.

Definition 4.3. (Fields of fractions) Let R be an integral domain and S = R \ {0}. The set S is a
multiplicatively closed set. The ring of fractions S −1 R is a field and contains R as a subring. It is
called the field of fractions of R and is denoted by Q(R).
r
Q(R) = { : r, s ∈ R, s , 0}
s
8
Definition 4.4. (Fields of rational functions) Let R be an integral domain and P = R[x1 , . . . , xn ] be
a polynomial ring in n indeterminate x1 , . . . , xn with coefficients in R. Let S = R[x1 , . . . , xn ] \ {0}
which is multiplicatively closed since P is an integral domain. The field of fractions S −1 P is called
the field of rational functions in indeterminate x1 , . . . , xn with coefficients in R and is denoted by
R(x1 , . . . , xn ).
f (x1 , . . . , xn )
R(x1 , . . . , xn ) = { : f (x1 , . . . , xn ), g(x1 , . . . , xn ) ∈ P, g(x1 , . . . , xn ) , 0}
g(x1 , . . . , xn )
The following result was discussed in the class.
Proposition 4.5. Let R be an integral domain and K = Q(R) be the field of fractions of R. Let
R[x1 , x2 , . . . , xn ], K[x1 , x2 , . . . , xn ] be polynomial rings in n indeterminate with coefficients in R, K
respectively. Then the fields of fractions of R[x1 , x2 , . . . , xn ] and K[x1 , x2 , . . . , xn ] are isomorphic.
In other words
R(x1 , x2 , . . . , xn )  K(x1 , x2 , . . . , xn ).
Proof. The inclusion R → K induces inclusion R[x1 , . . . , xn ] ⊂ K[x1 , . . . , xn ] ⊂ K(x1 , . . . , xn ).
Let j : R[x1 , . . . , xn ] → K(x1 , . . . , xn ) denote the inclusion. The set S = R[x1 , . . . , xn ] \ {0} is
multiplicatively closed. We find that j( f ) = f is a unit in K(x1 , . . . , xn ) for all f ∈ S since 1f
exists in K(x1 , . . . , xn ). Therefore by Theorem 4.1, there exists a unique ring homomorphism ψ :
R(x1 , . . . , xn ) → K(x1 , . . . , xn ) such that the diagram below commutes, i.e. ψ ◦ i = j.

K(x , . . . , xn )
6 1 O
j
ψ

R[x1 , . . . , xn ] / R(x1 , . . . , xn )
i

The map ψ is given by ψ( gf ) = j(j(g)f ) = gf for f, g ∈ R[x1 , . . . , xn ] and g , 0. We show that ψ is an

isomorphism.
Let ψ( gf ) = 0 for f, g ∈ R[x1 , . . . , xn ], g , 0. This implies that gf = 0 in K(x1 , . . . , xn ) and
consequently f = 0 in K(x1 , . . . , xn ). This means that f = 0 in R[x1 , . . . , xn ]. It follows that gf = 0
in R(x1 , . . . , xn ). Thus we show that ker(ψ) = 0 and therefore j is an injective ring homomorphism.
Let GF ∈ K(x1 , . . . , xn ) for F, G ∈ K[x1 , . . . , xn ] and G , 0. The polynomial F is a finite sum
of terms of the form as x1i1 . . . xnin , as ∈ K. If s1 ∈ R \ {0} denotes the product of denominators
of coefficients of F, then s1 F ∈ R[x1 , . . . , xn ]. Similarly, we find s2 ∈ R \ {0} such that s2G ∈
R[x1 , . . . , xn ]. Now GF = ss11 ss22GF = ψ( ss11 ss22GF ). Therefore ψ is surjective. Hence ψ is an isomorphism.
□
Proposition 4.6. Let R = R1 × R2 × . . . × Rn be a finite product of commutative rings R1 , . . . , Rn . Let
S i be a multiplicatively closed subset of Ri which does not contain zero and zero divisors of Ri for
i = 1, . . . , n. Then the set S = S 1 × . . . × S n does not contain zero and zero divisors of R. Moreover
S −1 R = S 1−1 R1 × . . . × S n−1 Rn .
Proof. The assertion that S does not contain zero and zero divisors of R is easy to verify. We leave
it to the reader. Let ϕ : R → S 1−1 R1 × . . . × S n−1 Rn be given by ϕ(r) = (r, r, . . . , r). An element
9
s ∈ S is of the form s = (s1 , . . . , sn ), si ∈ Ri . We note that ϕ(s) is invertible in S 1−1 R1 × . . . × S n−1 Rn
with inverse (s−11 , . . . , sn ). It follows from Theorem 4.1 that there exists a ring homomorphism
−1

ψ : S −1 R → S 1−1 R1 × . . . × S n−1 Rn such that the following diagram commutes.

S 1−1 R1 × . . . × S n−1 Rn
7 O
ϕ
ψ

R / S −1 R
i

The map ψ is given by ψ( (s(r11 ,...,rn)

,...,sn )
) = ( rs11 , . . . , rsnn ). It is easy to check that ψ is surjective and
ker ψ = 0 proving that ψ is an isomorphism. □

5. The division algorithm

All rings in this section are commutative and contain nonzero identities.
Definition 5.1. Let f = a0 xn +a1 xn−1 +. . .+an ∈ R[x] be a nonzero polynomial such that a0 , 0. The
integer n is called the degree of polynomial f . The degree of f is denoted by deg( f ). If f ∈ R \ {0},
then f = a0 = an , so deg( f ) = 0.
The element a0 is called the leading coefficient of f and is denoted by l( f ). The polynomial f is
called a monic polynomial if its leading coefficient a0 is a unit in R.
The proof of the following lemma is easy and left to the reader. Please try to see why I say that
leading coefficient of f is a nonzero divisor of R[x].
Lemma 5.2. Let f, g ∈ R[x] be nonzero polynomials such that the leading coefficient of f is a
nonzero divisor in R. Then
deg( f g) = deg( f ) + deg(g).
In particular, f is a nonzero divisor in R[x].
The following result was proved in the class.
Theorem 5.3. (Division Algorithm) Let R be a commutative ring with identity 1 , 0. Let f, g be
polynomials in R[x] and f be a monic polynomial. Then there exist unique polynomials q, r ∈ R[x]
such that g = q f + r with either r = 0 or deg(r) < deg( f ).
Proof. Let f = a0 xn + a1 xn−1 + . . . + an ∈ R[x]. Since f is a monic polynomial, a0 is a unit in R. In
particular f , 0. If deg(g) < deg( f ), then we write g as g = q f + r for q = 0 and r = g. In this case
deg(r) < deg( f ), so the statement follows. Therefore we may further assume that deg(g) ≥ deg( f ).
We prove the result by induction on deg(g). If deg(g) = 0, then deg( f ) = 0, i.e. f ∈ R. Since f
is a monic, it is a unit in R. This implies that g = q f + r for q = f −1 g ∈ R[x] and r = 0.
We now turn to the case that deg(g) > 0. We consider the polynomial
g′ = g − a−1
0 l(g)x
deg(g)−deg( f )
f ∈ R[x].
Here l(g) denotes the leading coefficient of g. Clearly deg(g′ ) < deg(g). By induction hypothesis,
there exist polynomials q′ , r ∈ R[x] such that g′ = q′ f + r with r = 0 or deg(r) < deg( f ). This
10
implies that
g = q f + r, for q = q′ + a−1
0 l(g)x
deg(g)−deg( f )

Therefore existence of q, f is established.

Now we proceed to show that q, r are unique. If possible assume that g = q′ f +r′ for polynomials
q′ , r′ ∈ R[x] with either r′ = 0 or deg(r′ ) < deg( f ). Then we have f (q − q′ ) = r′ − r. If q , q′ ,
then r , r′ and deg(r′ − r) = deg( f ) + deg(q − q′ ) ≥ deg( f ) which leads to a contradiction because
deg(r − r′ ) < deg( f ). Therefore we must have q = q′ which forces r = r′ . □

The following is a trivial consequence of the above.

Lemma 5.4. Let I be an ideal of R[x] which contains a monic polynomial f of degree n. Then
every nonzero element of the quotient ring Q = R[x] I
can be written as the class of a polynomial g
such that deg(g) < n. If I = ( f ), then g is unique.

We proved the following theorem in the class.

Lemma 5.5. Let R be a commutative ring with identity 1 , 0 and a ∈ R. Then a polynomial
f ∈ R[x] in divisible by (x − a) if and only if f (a) = 0.

Proof. By division algorithm, we may write f as f = q(x)(x − a) + r where q(x) ∈ R[x] and r ∈ R.
Evaluating both sides at x = a, we have f (a) = r. This implies that f = q(x)(x − a) + f (a). The
lemma follows obviously. □

The following generalises the above lemma to the case of many variables.

Lemma 5.6. Let f ∈ R[x1 , . . . , xn ] and a1 , . . . , an ∈ R. Then f ∈ (x1 − a1 , . . . , xn − an ) if and only if

f (a1 , . . . , an ) = 0.

Proof. We recall that R[x1 , . . . , xi+1 ] = R[x1 , . . . , xi ][xi+1 ]. By repeated application of the above
lemma we have

f = qn (xn − an ) + f (x1 , x2 , . . . , xn−1 , an ), qn ∈ R[x1 , . . . , xn ]

= qn (xn − an ) + qn−1 (xn−1 − an−1 ) + f (x1 , x2 , . . . , xn−2 , an−1 , an ), qn−1 ∈ R[x1 , . . . , xn−1 ]
......
X n
= qi (xi − ai ) + f (a1 , . . . , an ), qi ∈ R[x1 , . . . , xi ].
i=1

The lemma is a trivial consequence of above. □

The following was proved in the class.

Lemma 5.7. Let R be a commutative ring with 1 , 0. Let s be a nonzero divisor of R and S denote
the multiplicatively closed subset {1, s, s2 , . . .}. Then S −1 R  (sx−1)
R[x]
.
11
 6. isomorphism theorems
In this section rings are not necessarily commutative and do not necessarily contain identities.
Definition 6.1. Let ϕ : R → S be a ring homomorphism. Then kernel of ϕ denoted by ker ϕ is
defined as
ker ϕ = {x ∈ R : ϕ(x) = 0}.
The following lemma is easy and was discussed in the class.
Lemma 6.2. Let ϕ : R → S be a ring homomorphism. Then ker ϕ is a two sided ideal of R. The
map ϕ is injective if and only if ker(ϕ) = 0.
Notation : Unless there is a scope of confusion, we always denote the additive coset x + I of an
ideal I of a ring R containing an element x ∈ R by x̄, i.e. throughout this note x̄ = x + I.
Theorem 6.3. (Existence of quotient rings) Let R be a ring and I be an ideal of R. Let x̄ denote the
additive coset of I containing x. Then the set of additive cosets
R/I = { x̄ : x ∈ R}
form a ring under the following operations :
Addition : x̄ + ȳ = x + y
Multiplication : x̄ȳ = xy.
Proof. We know that (R, +) is an abelian group, so I is a normal subgroup of R. It follows from
results in Group theory that R/I is also an abelian group under addition as defined above.
We show that multiplication is well defined on R/I. Let x̄ = x̄′ and ȳ = y¯′ . Then x − x′ , y − y′ ∈ I.
This implies that x = x′ + i and y = y′ + j for i, j ∈ I. Therefore we have
xy = (x′ + i)(y′ + j)
= x′ y′ + x′ j + iy′ + y j
= x′ y′ + (x′ j + iy′ + y j)
The second summand of the last line belongs to I since I is a two sided ideal of R. Consequently
xy = x′ y′ establishing that the multiplication is well defined. Checking that R/I is a ring under
above operations is straightforward and therefore left to the reader. □
Theorem 6.4. (The first isomorphism theorem) Let ϕ : R → S be a surjective ring homomorphism.
Then ker ϕ is an ideal of R and the induced map ψ : R/ ker ϕ → S given by ψ( x̄) = ϕ(x) is an
isomorphism. Here x̄ denotes the quotient class of x in R/ ker ϕ.
Proof. Let x, y ∈ ker ϕ and r ∈ R. Then ϕ(x) = ϕ(y) = 0. We have ϕ(x − y) = ϕ(x) − ϕ(y) = 0,
ϕ(rx) = ϕ(r)ϕ(x) = 0 and ϕ(xr) = ϕ(x)ϕ(r) = 0. This shows that x − y, rx, xr ∈ ker ϕ. Therefore,
ker ϕ is an ideal of R.
The map ψ : R/I → S is given by ψ( x̄) = ϕ(x). First we show that ψ is a well defined map.
Let x̄ = ȳ in R/ ker ϕ. Then x − y ∈ ker ϕ. This implies that ϕ(x − y) = 0 giving ϕ(x) = ϕ(y).
Consequently ψ( x̄) = ψ(ȳ). Therefore ψ is well defined.
12
 It is fairly easy to see that ψ is a ring homomorphism and ψ is surjective. Both properties follow
from the corresponding properties of ϕ. We leave it to the reader to work out details. To show that
ψ is injective, we prove that ker(ψ) = 0. Let x̄ ∈ ker ψ. Then ψ( x̄) = ϕ(x) = 0. This implies that
x ∈ ker ϕ. Consequently x̄ = 0 in R/ ker ϕ. Therefore ker ψ = 0. Hence ψ is an isomorphism and
the result is established. □
Theorem 6.5. (The second isomorphism theorem) Let R be a ring, S be a subring of R and I be an
ideal of R. Then the following hold true.
1 S + I = {s + i : s ∈ S , i ∈ I} is a subring of R,
2 I is an ideal of S + I,
3 S ∩ I is an ideal of S ,
4 There is an isomorphism
S S +I
 given by s̄ 7→ s + i.
S ∩I I
Proof. Statements 1, 2, 3 are fairly easy to establish. We skip details. We only prove the last
assertion. We define a map ϕ : S → S +I I
given by ϕ(x) = x̄. One easily checks that ϕ is a ring
homomorphism. Elements of S +I I
are given by classes of elements of S , so ϕ is surjective. We
observe that ϕ(x) = 0 if and only if x ∈ I proving ker ϕ = I. The isomorphism now follows from
the first isomorphism theorem. □
Theorem 6.6. (The third isomorphism theorem) Let I, J be ideals of R such that I ⊂ J ⊂ R. Then
the map R/I
J/I
→ RJ given by x̄ + JI 7→ x̄ is an isomorphism.
Proof. We define a map ϕ : R/I → R/J given by ϕ(x + I) = x + J. Since I ⊂ J, one can easily check
that ϕ is well defined. As before, it is routine to check that ϕ is a surjective ring homomorphism
with kernel J/I = { x̄ : x ∈ J}. The result follows from the first isomorphism theorem. □
Notation Let F : S → T be a map between sets S , T . Let A ⊂ S and B ⊂ T . We define
F(A) = {y ∈ T : y = F(x) for some x ∈ A} and F −1 (B) = {x ∈ S : F(x) ∈ B}. If F is a ring
homomorphism, then both are subrings of ambient rings.
Theorem 6.7. (The fourth isomorphism theorem) Let I be an ideal of a ring R. The correspondence
A ←→ AI for A being a subring of R containing I establishes an inclusion preserving bijection
between the following sets:
{A : A is a subring of R containing I} ←→ {B is a subring of R/I}.
Proof. We set S = {A : A is a subring of R containing I} and T = {B is a subring of R/I}. Let
q : R → R/I be the quotient map given by q(x) = x̄.
Given a subring A in S, we associate with it the subring q(A) = A/I in T. Conversely, given a
subring B in T, we associate with it the subring q−1 (B) in S. Here q−1 (B) is in S because it contains
q−1 (0) = I.
The proof falls naturally into three steps. Steps 1, 2 together establish that both sets are in one-
one correspondence. Step 3 shows that the correspondence is inclusion preserving.
Step - 1 : q−1 (q(A)) = A for any A ∈ S.
13
 Let x ∈ q−1 (q(A)). This implies that q(x) ∈ q(A). It follows that q(x) = q(a) for some a ∈ A.
Consequently q(x − a) = 0. Therefore x − a ∈ ker q = I ⊂ A which implies that x ∈ A. Thus we
show that q−1 (q(A)) ⊂ A. The reverse inclusion is obvious, so the assertion follows.
Step - 2 : q(q−1 (B)) = B for any B ∈ T.
The inclusion q(q−1 (B)) ⊂ B follows obviously. To prove the reverse inclusion, we choose y ∈ B.
The map q : R → R/I is surjective and B ⊂ R/I. It follows that there exists x ∈ R such that y = q(x).
Indeed x ∈ q−1 (B) since y ∈ B. This shows that y = q(x) ∈ q(q−1 (B)). Therefore B ⊂ q(q−1 (B)).
Thus the equality is established.
Step - 3 : q(A1 ) ⊂ q(A2 ) for A1 ⊂ A1 , A1 , A2 ∈ S and q−1 (B1 ) ⊂ q−1 (B2 ) for B1 ⊂ B2 , B1 , B2 ∈ T.
This step follows trivially. □
Z[x]
Question 6.8. Find all proper nonzero subrings of (2,x2 +x+1)
.
Proof. We have the following isomorphisms by the third isomorphism theorem:
Z[x] Z/2Z[x]
 2 .
(2, + x + 1) (x + x + 1)
x2
The later ring is a field containing only four elements 1, 0, x, 1 + x. Its only non-trivial proper
subring is {0, 1} which corresponds to the subset {0̄, 1̄} under the above isomorphism. Therefore
Z[x]
(2,x2 +x+1)
has only one nontrivial proper subring, viz. {0̄, 1̄}. □

7. Chinese remainder theorem

All rings in this section are commutative and contain nonzero identities.
Definition 7.1. (Products of rings) Let R, S be rings. The product of sets R and S , viz. R × S =
{(r, s) : r ∈ R, s ∈ S } forms a ring under the operations :
Addition : (r, s) + (r′ , s′ ) = (r + r′ , s + s′ ) for (r, s), (r′ , s′ ) ∈ R × S ;
Multiplication : (r, s)(r′ , s′ ) = (rr′ , ss′ ) for (r, s), (r′ , s′ ) ∈ R × S .
This ring is called the product of R and S . One similarly defines product of finitely or infinitely
many rings.
Definition 7.2. Two ideals I, J of a ring R are called co-maximal if I + J = R.
Lemma 7.3. If I, J are co-maximal ideals of a ring R, then I ∩ J = I J.
Proof. Since I + J = R, we have i ∈ I and j ∈ J such that i + j = 1. The inclusion I J ⊂ I ∩ J is
clear. To prove the reverse inclusion, we choose x ∈ I ∩ J. We have x = xi + x j. Now xi = ix is
in I J because i ∈ I and x ∈ J and so is x j. It follows that x ∈ I J. Therefore I ∩ J ⊂ I J. Hence the
lemma is proved. □
Remark 7.4. The assertion of the above lemma does not hold true if I and J are not co-maximal.
For example choose I = (x), J = (x2 , y) in the ring R[x, y] where R is an integral domain. Then we
have
I J = (x3 , xy) ⫋ (x2 , xy) = I ∩ J.
14
Lemma 7.5. Let I, J1 , J2 , . . . , Jn be ideals of R such that I and Jk are co-maximal for all k =
1, 2, . . . , n. Then I, J1 J2 , . . . Jn are co-maximal ideals, so are I, J1 ∩ . . . ∩ Jn .

Proof. We have I + J1 J2 , . . . Jn ⊂ I + J1 ∩ . . . ∩ Jn . If I, J1 J2 , . . . Jn are co-maximal ideals, then it

follows that I, J1 ∩ . . . ∩ Jn are also co-maximal ideals. Therefore we only require to prove that I,
J1 J2 , . . . Jn are co-maximal ideals. If not, we get a maximal ideal m containing I + J1 J2 , . . . Jn . Now
J1 J2 , . . . Jn ⊂ m implies that Jk ⊂ m for some k. It follows that R = I + Jk ⊂ m; a contradiction. □

Theorem 7.6. (Chinese remainder theorem) Let R be a commutative ring with 1 , 0 and I1 , . . . , In
be ideals in R. Then the map f : R → R/I1 × . . . R/In given by f (x) = (x + I1 , . . . , x + In ) is a ring
homomorphism and ker f = I1 ∩ . . . ∩ In .
Moreover if I1 , . . . , In are pairwise co-maximal ideals, then f is surjective. In particular f in-
duces an isomorphism
R/I1 ∩ . . . ∩ In → R/I1 × . . . × R/In .

Proof. It is easy to verify that f is a ring homomorphism with ker f = I1 ∩ . . . ∩ In . We do

not go into details. We only show that f is surjective when ideals are pairwise co-maximal. Let
ei = (0, 0, . . . , 1, . . . , 0), where 1 is sitting at the i-th coordinate. If we can show that each ei is in
the image of f , i.e. there exists yi ∈ R such that f (yi ) = ei , then we are through since an arbitrary
element (x1 , . . . , xn ) ∈ R/I1 × . . . R/In can be written as :
n
X
(x1 , . . . , xn ) = (xi , xi , . . . , xi )ei
i=1
n
X
= f (xi ) f (yi )
i=1
Xn
= f( xi yi ).
i=1

The ideal Ii is co-maximal to Ik for each k , i. By Lemma 7.5, the ideals Ii and I1 I2 . . . Ii−1 Ii+1 . . . In
are co-maximal ideals, so we have x ∈ Ii and y ∈ I1 I2 . . . Ii−1 Ii+1 . . . In such that x + y = 1. One easily
checks that y ≡ 0 (mod Ik ) for k , i and y ≡ 1 (mod Ii ). This shows that f (y) = ei . Therefore each
ei is in the image of f and the proof follows. □

Remark 7.7. It is well known that the class of m in Z/(n) is a unit if and only if m and n are
co-primes integers. Let n = pi11 . . . pikk be the factorisation of a positive integer n > 1 into power
of distinct primes p1 , . . . , pn . The ideals (pi11 ), (pi22 ), . . . , (pikk ) are pair wise co-maximal ideals. By
Chinese remainder theorem one has Z/(n)  Z/(pi11 ) × . . . × Z/(pikk ). For a given prime p, there are
only pl−1 (p − 1) integers which are co-prime to pl and less that pi . This shows that each Z/(pill ) has
exactly pill −1 (pl − 1) units. Therefore Z/(n) has exactly pi11 −1 . . . pikk −1 (p1 − 1) . . . (pk − 1) units.

8. Date 26th Feb, 2019

We discuss Euclidean domains, two important examples and one theorem.
15
Definition 8.1. (Euclidean domains or EDs)
An integral domain R is called a Euclidean domain if there is a function N : R \ {0} → N ∪ {0}
such that for any two elements a, b ∈ R, b , 0 there exist q, r ∈ R satisfying a = bq + r with r = 0
or N(r) < N(b).
Lemma 8.2. Show that Z is a Euclidean domain.
Proof. Let N : Z \ {0} → N ∪ {0} be given by N(x) = |x|. Consider two integers a, b such that b , 0.
Then |a| is a non-negative integer and |b| is a positive integer. Dividing |a| by |b|, we get
|a| = q′ |b| + r′ where 0 ≤ r′ < |b|.
We consider two cases.
Case - 1 : (a > 0) In this case we have a = q′ |b| + r′ where 0 ≤ r′ < |b|. If b > 0, we write
q = q′ , r = r′ otherwise we write q = −q′ , r = r′ . Thus we get a = qb + r where 0 ≤ r < |b|.
Case - 2 : (a < 0)
In this case we have a = −|a| = −q′ |b| − r′ . We consider two subcases.
Subcase - I : (a < 0, r′ = 0)
Let q = −q′ when b > 0 and q = q′ when b < 0. We have a = qb + r for r = 0.
Subcase - II : (a < 0, r′ , 0)
We have a = −a = −q′ |b| − r′ = −(q′ + 1)|b| + (|b| − r′ ). Let r = |b| − r′ . Then 0 < r < |b|. If
b > 0, we write q = −(q′ + 1) otherwise we write q = (q′ + 1). Consequently we have a = bq + r
where 0 < r < |b|.
Therefore for any a, b ∈ Z, b , 0, we find integers q, r ∈ Z such that a = bq + r with r = 0 or
N(r) < N(b). Hence Z is a Euclidean domain. □
Lemma 8.3. Let k be a field, then k is a Euclidean domain.
Proof. We define N : K \ {0} → N ∪ {0} by N(x) = 0 for all x ∈ K \ {0}. For a, b ∈ k, b , 0, we
have a = qb + r for q = ab−1 and r = 0. Therefore k is a Euclidean domain. □
Lemma 8.4. Let k be a field, then k[x] is a Euclidean domain.
Proof. For f ∈ k[x] \ {0}, we set N( f ) = deg( f ). By division algorithm (see Theorem 5.3), for any
g, f ∈ R, f , 0, there exist polynomials q, r ∈ k[x] satisfying g = q f + r with r = 0 or N(r) < N( f ).
Therefore k[x] is a Euclidean domain. □
Theorem 8.5. In a Euclidean domain all ideals are principal.
Proof. Let I be an ideal of R. If I = {0}, then we have nothing to show, so we assume that I , {0}.
Let S = {N(x) : x ∈ I \ {0}}. From the definition of N it follows that S ⊂ N ∪ {0}. Therefore we get
a y ∈ I \ {0} such that N(y) = min(S ).
We claim that I = (y). Since y ∈ I, one finds that (y) ⊂ I. We prove the reverse inclusion.
Let z ∈ I. Since R is a Euclidean domain, we get q, r ∈ R such that z = qy + r where r = 0 or
N(r) < N(y). We observe that r = z − qy ∈ I since z, y ∈ I. If r , 0, we have N(r) < N(y) which
is a contradiction because of the choice of y. Therefore r = 0 and z = qy ∈ I. Hence the result
follows. □
16
 9. Date - 28th Feb, 2019
We introduce principal ideal domains and discuss important properties.
Definition 9.1. (Principal ideal domains or PIDs) An integral domain is called a principal ideal
domain if every ideal I of R is principal, i.e. there exists an a ∈ I such that I = (a).
Remark 9.2. Theorem 8.5 shows that every Euclidean domain is a principal ideal domain.
Lemma 9.3. Let Z[i] = {a + bi : a, b ∈ Z}. Then Z[i] is a Euclidean domain.
Proof. One easily checks that Z[i] is a subring of C and contains the identity element 1. The ring
Z[i] does not have zero divisors since C is a field. Therefore Z[i] is an integral domain.
We define a function N : Z[i] → N ∪ {0} by N(α) = αα, α ∈ Z[i]. Here ᾱ denotes the complex
conjugate of α. Clearly if α = a + bi; a, b ∈ Z, then N(a + bi) = a2 + b2 . For α, β ∈ Z[i], one has
N(αβ) = αβαβ = ααββ = N(α)N(β), so N is multiplicative. For brevity, we denote the restriction
of N to Z[i] \ {0} also by N.
Let α, β ∈ Z[i] and β , 0. Then αβ = a + bi for a, b ∈ Q. Let c, d be integers nearest to a, b
respectively. We have a = c + r, b = d + s where c, d ∈ Z and r, s ∈ Q satisfying 0 ≤ |r|, |s| ≤ 12 . Let
q = c + di ∈ Z[i] and r′ = r + si. Clearly N(r′ ) = r2 + s2 ≤ 14 + 14 = 12 .
Now αβ = q + r′ which implies that α = βq + βr′ . Let r = βr′ , then r = α − βq ∈ Z[i]. We observe
that N(r) = N(βr′ ) = N(β)N(r′ ) ≤ 12 N(β) < N(β) whenever N(β) , 0 which holds if r , 0. Thus
N : Z[i] \ {0} → N ∪ {0} has the property that given α, β ∈ Z[i], β , 0, one has q, r ∈ Z[i] such that
α = βq + r with r = 0 or N(r) < N(β). Hence Z[i] is a Euclidean domain. □
Theorem 9.4. Every nonzero prime ideal of a principal ideal domain is a maximal ideal.
Proof. Let R be a PID and P be a nonzero prime ideal of the ring R. It follows from the property of
PIDs that there exists an element p ∈ R such that P = (p). Since P , {0}, we must have p , 0. Let
I be an ideal of R such that P ⊂ I ⫋ R. We prove that I = P.
Since R is a PID, we have an element x ∈ R such that I = (x). The element x is not a unit because
I is a proper ideal. We obtain (p) ⊂ (x) which implies that p = cx for c ∈ R. Now cx = p ∈ P. This
implies that either c ∈ P or x ∈ P.
We show that c < P. If possible we assume that c ∈ P. Then c = d p for d ∈ R which yields that
d px = cx = p. This implies that dx = 1 since R is an integral domain and p , 0. Therefore x is a
unit and I = (x) = R. This is a contradiction because the I is a proper ideal.
Therefore we have only one choice, viz. x ∈ P. This implies that I = (x) ⊂ P. Therefore I = P.
Thus we establish that P is not properly contained in any proper ideal of R. Hence P is a maximal
ideal. □
Remark 9.5. Let R be a PID which is not a field, e.g. R = Z. Then (0) is a prime ideal but not a
field.
Examples 9.6. The following are examples of integral domains which are not PID.
(1) The ring Z[x] is not a PID. The ideal (2, x) is not a principal ideal. First we observe that
Z[x]
(2, x) is a proper ideal since (2,x)  Z/2Z , 0. If (2, x) = ( f (x)) for some f (x) ∈ Z[x], then
17
 2 = g(x) f (x) and x = h(x) f (x) for some g(x), h(x) ∈ Z[x]. Comparing degrees of both sides
of the first equation, we get deg( f ) = 0, i.e. f ∈ Z, c.f. Lemma 5.2. Comparing the leading
coefficients, the second equation implies that f divides 1 in Z which yields f = ±1. This is
a contradiction since (2, x) is a proper ideal.
Alternatively, one observes that Z[x] (2)
= 2Z
Z
[x] which is an integral domain but not a field.
Thus (2) is a nonzero prime ideal but not a maximal ideal of Z[x]. Therefore Z[x] is not a
PID, c.f. Theorem 9.4.
(2) Let R be an integral domain. The ring R[x, y] is not a PID. We show that the ideal (x, y) is
not a principal ideal. It is a proper ideal because 1 < (x, y). If possible, we assume that
(x, y) = ( f (x, y)) for f ∈ R[x, y]. This implies that x = f g and y = f g′ for g, g′ ∈ R[x, y]. We
view R[x, y] as a polynomial ring in y with coefficients in R[x]. Comparing degrees in y of
both sides of the first equation we conclude that f is free from y, c.f. Lemma 5.2. Similarly
the second equation implies that f is free from x. Therefore f ∈ R. Now f ∈ (x, y) yields
that f = xg + yh for g, h ∈ R[x, y]. Evaluating at x = y = 0, we have f = 0 which is a
contradiction because (x, y) is a nonzero ideal.
Remark 9.7. If R is a commutative ring with 1 , 0. The ideal (x, y) is not a principal ideal of
R[x, y]. If R is not an integral domain, the above argument does not work here because Lemma 5.2
fails to hold. We use a trick.
Let m be a maximal ideal of R and R/m = k. Let f : R[x, y] → mR[x,y] R[x,y]
be a surjective ring
homomorphism. Note that mR[x,y] = k[x, y]. The image of (x, y) in R[x, y] under f is the ideal (x, y)
R[x,y]

of k[x, y]. If (x, y) is a principal ideal of R[x, y], then its image (x, y) is also a principal ideal of
k[x, y], which is not true by the argument above since k is an integral domain (actually a field).
Therefore (x, y) is not a principal ideal of R[x, y].
If R is a commutative ring with 1 , 0, one can similarly show that the polynomial ring R[x1 , x2 , . . . , xn ]
is not a PID.

10. 1st March, 2019

We discuss irreducible elements and prime elements of an integral domain R.
Definition 10.1. (prime elements, irreducible elements) Let R be an integral domain.
An element x , 0 in R is called a prime element if (x) is a prime ideal of R. Clearly x is a prime
element if and only if x is not a unit and whenever x ab for a, b ∈ R, one has either x a or x b.
An element x , 0 in R is called an irreducible element of R if x is not a unit and whenever x = ab
for a, b ∈ R, at least one of a, b must be a unit in R. An element x , 0 in R is called a reducible
element if x is not a unit and not an irreducible element.
Lemma 10.2. Every prime element of an integral domain is irreducible.
Proof. Let p be a prime element of an integral domain R. Then p , 0 and p is not a unit. Suppose
p = ab for a, b ∈ R. Since p is a prime element, p divides at least one of a, b.
If p divides a, then a = cp for c ∈ R which yields that p = ab = cpb = pcb. The ring R is an
integral domain and p , 0. Therefore 1 = cb which shows that a is a unit. If p divides b, then one
can similarly show that b is a unit. Hence p is an irreducible element. □
18
Lemma 10.3. In a PID, every irreducible element is a prime element.

Proof. Let x be an irreducible element of R. Then x is nonzero and not a unit. The ideal (x) is a
proper ideal of R. Let M be a maximal ideal of R containing I, c.f. Corollary 3.6. The ideal M
is principal since R is a PID, so we have m ∈ R such that M = (m). Indeed m is a prime element
because M is also a prime ideal.
Now we have (x) ⊂ (m). This implies that x = cm for c ∈ R. The element x is irreducible and m
is not a unit (actually a prime element), so c is a unit in R. It follows that (x) = (m) = M is a prime
ideal. Therefore x is a prime element. □
The above lemma provides us a method to check whether a given integral domain is a PID.
√
Example 10.4. The ring Z[ −3] is not a PID.
√
Proof. The ring Z[ −3] is an integral domain √ with standard addition and multiplication of complex
numbers. We define a function N : Z[ −3] → N ∪ {0} √ by N(α) = αᾱ. One easily checks
that N is multiplicative,
√ √ i.e. N(αβ) = αβ for α, β ∈ Z[ −3]. Moreover N(α) = a 2
+ 3b2
for
α = a + b −3 ∈ Z[ −3]. √ √
We consider the ring homomorphism F : Z[x] → Z[ −3] given by F Z = idZ and F(x) = −3.
Clearly F is surjective. We compute its kernel. It is easy to see that x2 +3 ∈ ker F which implies that
(x2 + 3) ⊂ ker F. Let g ∈ ker F. By division algorithm (see Theorem 5.3) we have q(x), r(x) ∈ Z[x]
such that g = q(x)(x2 +3)+r(x) with r(x) √ = 0 or deg(r(x)) < deg(x +3) = 2. We write r(x) = ax+b
2

for a, b ∈ Z. Now F(g) = F(r(x)) = a −3 + b = 0. This implies that a = b = 0 since a, b ∈ Z and

√
−3 is an irrational number. Therefore r(x) = 0 and g = q(x)(x2 + 3) ∈ (x2 + 3). Thus we √ establish
that ker F = (x + 3). By the first isomorphism theorem, we conclude that (x2 +3)  Z[ −3]. The
2 Z[x]
√
element 2̄ ∈ (xZ[x]
2 +3) corresponds to 2 ∈ Z[ −3] under this isomorphism.
We have
√ Z[x]
Z[ −3] (x2 +3)

(2) (2̄)
Z[x]
(x2 +3)
 (2,x2 +3)
(x2 +3
Z[x]
 ; by the third isomorphism theorem
+ 3, 2)
(x2
Z[x]
 2 ; since (x2 + 3, 2) = (x2 + 1, 2)
(x + 1, 2)
Z[x]
(2)
 (x2 +1,2)
; by the third isomorphism theorem
(2)
Z/2Z[x] Z[x] (x2 + 1, 2)
 2 ; since  Z/2Z[x] and corresponds to (x2 + 1)
(x + 1) (2) (2)
Z/2Z[x]

((x + 1)2 )
19
 √
The last ring is not an integral
√ domain,√so 2 is not a prime element of√ −3]. Alternatively,
Z[ √ one
sees that 2 divides 4 = (1 + −3)(1 − −3) √ but 2 divides neither 1 + −3 nor 1 − −3. This also
shows that 2 is not a prime element of √ Z[ −3].
We show that 2 is irreducible in Z[ −3]. The previous
√ computation shows that (2) is a proper
ideal, i.e. 2 is not a unit. Let 2 = αβ for α, β ∈ Z[ −3]. This implies that 4 = N(2)
√ = N(α)N(β).
√
Both N(α), N(β) are non-negative integers, so N(α) = 1, 2, 4. Let α = a + b −3 ∈ Z[ −3].
The equation N(α) = a2 + 3b2 = 2 does not have an integer √ solution, so N(α)
√ cannot be 2. If
N(α) = 1, then αᾱ = 1. This implies that α is a unit √ in Z[ −3] since ᾱ ∈ Z[ −3]. If N(α) = 4,
then N(β) = 1 which implies √ that β is a unit in Z[ −3] due √to similar reasons. Therefore 2 is an
irreducible element of Z[ −3]. By Lemma 10.3, the ring Z[ −3] is not a PID. □
√ √
Remark 10.5. The ideal (2, 1+ −3) is not a principal ideal of Z[ −3]. We leave it as an exercise.
√
Example 10.6. Show that the ring Z[ 1+ 2 −3 ] is a Euclidean domain.
√
Proof. The method of the proof is similar to that used in Lemma 9.3. Set ω = 1+ 2 −3 . We have
ωω̄ = 1. We also have (ω − 12 )2 = −3
4
which yields ω2 + ω + 1 = 0. Now Z[ω] = {a + bω : a, b ∈ Z}.
Clearly Z[ω] is closed under addition and subtraction. Let a + bω, c + dω ∈ Z[ω]. We have

(a + bω)(c + dω) = ac + (ad + bc)ω + ω2 = (ac − 1) + (ad + bc − 1)ω since ω2 = −ω − 1.

Thus Z[ω] is closed under multiplication. Therefore Z[ω] is a subring of C. Since 1 ∈ Z[ω] and C
is a field, we conclude that Z[ω] is an integral domain.
We define a function N : Z[ω] → N ∪ {0} by N(α) = αᾱ, α ∈ Z[ω]. One easily checks that N is
multiplicative, i.e. N(αβ) = N(α)N(β) for α, β ∈ Z[ω]. If α = a+bω, then N(α) = (a+bω)(a+bω̄) =
a2 + ab(ω + ω̄) + b2 ωω̄ = a2 + ab + b2 ∈ N ∪ {0}.
αβ̄
Let α, β ∈ Z[ω] and β , 0. We have αβ = N(β) = a + bω for a, b ∈ Q. We write a = c + e,
b = d + f where c, d ∈ Z and 0 ≤ |e|, | f | ≤ 2 . Let q = c + dω and r′ = e + f ω. Then q ∈ Z[ω] and
1

N(r′ ) = e2 + e f + f 2 ≤ 43 .
We have αβ = q + r′ which gives α = qβ + r where r = r′ β. If r , 0, then β , 0 and N(r) =
N(r′ )N(β) ≤ 34 N(β) < N(β). Moreover r = α − qβ ∈ Z[ω]. Thus N Z[ω]\{0} : Z[ω] \ {0} → N ∪ {0}
has the property that for any α, β ∈ Z[ω], β , 0, there exist q, r ∈ Z[ω] such that α = βq + r where
r = 0 or N(r) < N(β). Therefore Z[ω] is a Euclidean domain. □

Remark 10.7. Examples 10.4, 10.6 together show that a subring of a Euclidean domain may not
be a Euclidean domain. In fact every integral domain R is a subring or its field of fractions Q(R)
(see Definition 4.3) which is a Euclidean domain. Therefore any non-Euclidean integral domain
provides an example.
k[x,y,z]
Example 10.8. Let k be a field. Then the ring (xz−y2 )
is not a PID.

Proof. We prove that x̄ is an irreducible element but not a prime element. The result then follows
by Lemma 10.3.
20
 Let F : k[x, y, z] → k[s, t] be the ring homomorphism given by F k = idk , F(x) = s2 , F(y) = st
and F(z) = t2 . Here k[s, t] is the polynomial ring in variables s, t. Image of F is the subring
k[s2 , st, t2 ], viz. namely set of polynomials in s2 , st, t2 .
First, we show that ker F = (xz − y2 ). We observe that s2 t2 − (st)2 = 0. This implies that
xz − y2 ∈ ker F. Let g ∈ ker F. Now k[x, y, z] = k[x, z][y]. By division algorithm (see Theorem
5.3), we have q, r ∈ k[x, z][y] such that g = q(xz − y2 ) + r where r = 0 or the degree of r in y is at
most one. Therefore r = hy + j for h, j ∈ k[x, z]. Now F(g) = 0 yields F(r) = 0. It follows that
h(s2 , t2 )st + j(s2 , t2 ) = 0. The monomials appearing in h(s2 , t2 )st are of the form s2m+1 t2n+1 , m, n ∈
N ∪ {0} whereas those in j(s2 , t2 ) are of the form s2m t2n , m, n ∈ N ∪ {0}. Consequently no term
of h(s2 , t2 )st can cancel a term in j(s2 , t2 ). Therefore h(s2 , t2 )st = j(s2 , t2 ) = 0 which implies that
h(s2 , t2 ) = j(s2 , t2 ) = 0 in k[s, t]. It follows that h(x, z) = j(x, z) = 0 in k[x, z]. This gives r = 0
and consequently g = q(xz − y2 ) ∈ (xz − y2 ). Therefore ker F = (xz − y2 ). By the first isomorphism
2 )  k[s , st, t ].
k[x,y,z] 2 2
theorem, F induces an isomorphism (xz−y
2 ) . The element x̄ corresponds to s in k[s , st, t ] which
k[x,y,z] 2 2 2
We show that x̄ is irreducible in (xz−y
is not a unit. It is enough to prove that s2 is irreducible in k[s2 , st, t2 ]. Let s2 = l(s, t)m(s, t) for
l(s, t), m(s, t) ∈ k[s2 , st, t2 ]. Comparing degrees in t of both sides of the equation, we conclude
that l(s, t), m(s, t) are free from t. If none of l(s, t), m(s, t) is a unit, then both of them are linear
polynomials of the form as + b ∈ k[s], a , 0, which is a contradiction since k[s2 , st, t2 ] does not
contain a linear polynomial in s. Therefore at least one of l(s, t), m(s, t) is a unit. It follows that s2
is irreducible in k[s2 , st, t2 ]. Therefore x̄ is irreducible in (xz−y
k[x,y,z]
2) .
k[x,y,z]
But x̄ is not a prime element of (xz−y2 )
. This is true because
k[x,y,z] k[x,y,z]
(xz−y2 ) (xz−y2 ) k[x, y, z] k[x, y, z] k[y, z]
 (x,xz−y2 )
 2
 
( x̄) (x, xz − y ) (x, y2 ) (y2 )
(xz−y2 )

is not an integral domain. Hence the result follows. □

k[x,y,z]
Remark 10.9. The proof above actually shows that the ring (xz−y2 )
is not a UFD. We shall discuss
this example later.

11. 5th March, 2019

We introduce Noetherian rings. We show that every nonzero, non-unit element of a PID can be
expressed as a product of primes and the expression is unique up to associates.

Definition 11.1. Let R be a commutative ring with 1 , 0. Two elements a, b ∈ R are said to be
associate in R if there exists a unit u ∈ R such that a = ub.

Lemma 11.2. Let R be an integral domain. Two nonzero elements a, b ∈ R are associate in R if
and only if (a) = (b).

Proof. First we assume that a, b are associate in R. Then there is a unit u ∈ R such that a = ub. We
observe that a ∈ (b) which yields (a) ⊂ (b). Since u is a unit, we have b = u−1 a. This gives b ∈ (a)
and consequently (b) ⊂ (a). Therefore (a) = (b).
21
 Conversely we assume that (a) = (b). Since a ∈ (b), we have a = cb for some c ∈ R. Similarly
b ∈ (a) gives that b = c′ a for some c′ ∈ R. It follows that a = cb = cc′ a. Since R is an integral
domain and a , 0, we get 1 = cc′ which shows that c is a unit. Therefore a, b are associate in
R. □

Proposition 11.3. Let R be a ring. Then the following are equivalent.

(1) Every ascending chain of ideals I1 ⊂ I2 ⊂ I3 . . . ⊂ In ⊂ . . . stabilizes, i.e. there exists a
positive integer N such that IN = IN+i for all i ≥ 1.
(2) Every nonempty set of ideals of R has a maximal element.
(3) Every ideal of R is finitely generated.

Proof. (1) =⇒ (2) :

We assume (1). Let S be a nonempty set of ideals of R. On the contrary, we assume that S does
not have a maximal element. Let I1 ∈ S. The ideal I1 is not a maximal element of S, so there exists
an ideal I2 ∈ S such that I1 ⫋ I2 . The ideal I2 is also not a maximal element of S. This implies that
I2 ⫋ I3 for some ideal I3 ∈ S. We continue in this fashion obtaining a strictly ascending chain of
ideals :

I1 ⫋ I2 ⫋ I3 ⫋ . . . .
This contradicts our assumption (1). Therefore S has a maximal element and (2) follows.
(2) =⇒ (3) :
We assume (2). Let I be an ideal of R. We consider the set of ideals :
S = {J : J ⊂ I and J is a finitely generated ideal of R}.
We observe that S , ∅ because (0) ∈ S. By our hypothesis, S has a maximal element. Let J0 be
a maximal element of S. Since J0 ∈ S, the ideal J0 is finitely generated and J0 ⊂ I. We show that
J0 = I.
If J0 ⫋ J, we choose x ∈ I \ J0 . The ideal J0 + (x) is finitely generated because J0 is so. We also
have J0 + (x) ⊂ I. It follows that J0 + (x) ∈ S and J0 ⫋ J0 + (x). This is a contradiction since J0 is a
maximal element of S. Therefore J0 = I. Consequently I is finitely generated. Hence (3) follows.
(3) =⇒ (1) We assume (3). We consider the ascending chain of ideals : I1 ⫋ I2 ⫋ I3 ⫋ . . .. Let
J = ∪ j≥1 I j . Since the union is taken over an ascending chain of ideals, one can easily see that J
is an ideal of R. By our hypothesis J is finitely generated, i.e. there exist x1 , x2 , . . . , xr such that
J = (x1 , x2 , . . . , xr ). We choose N large enough such that x j ∈ IN for j = 1, . . . , r. Then for any
i ≥ 1, we have
J = (x1 , x2 , . . . , xr ) ⊂ IN ⊂ IN+i ⊂ J.
Therefore IN = IN+i for all i ≥ 1 and (1) follows. □

Definition 11.4. A ring R is called a Noetherian ring if R satisfies any of the equivalent statements
of Proposition 11.3.

Example 11.5. (1) Any PID is a Noetherian ring.

22
 (2) Let R be a ring with 1 , 0. We consider the sequence of indeterminate x1 , x2 , . . .. Then
R[x1 ] ⊂ R[x1 , x2 ] ⊂ . . . ⊂ R[x1 , x2 , . . . , xi ] ⊂ . . . is an ascending sequence of polynomial
rings. We define the polynomial ring in infinitely many variables x1 , x2 , . . . as

R[x1 , x2 , . . .] = ∪i≥1 R[x1 , x2 , . . . , xi ].

First, we observe that xn+1 < (x1 , . . . , xn ). This is true because otherwise we have xn+1 =
i=1 xi fi for some fi ∈ R[x1 , x2 , . . .] which leads to 1 = 0, a contradiction by putting x1 =
Pn
. . . = xn = 0 and xn+1 = 1. Therefore we get a strictly ascending chain of ideals :

(x1 ) ⫋ (x1 , x2 ) ⫋ (x1 , x2 , x3 ) ⫋ . . . ⊊ (x1 , x2 , . . . , xn ) ⊊ . . . .

It follows that R[x1 , x2 , . . .] is not a Noetherian ring.

We show that the ideal I = (x1 , x2 , . . . , xn , . . .) is not finitely generated. If not, we
have polynomials f1 , f2 , . . . , fr ∈ R[x1 , x2 , . . .] such that I = ( f1 , f2 , . . . , fr ). Each fi ∈
(x1 , x2 , . . . , xn , . . .). Let N be an integer large enough such that fi ∈ (x1 , x2 , . . . , xN ) for
i = 1, 2, . . . , r. This implies that

I = ( f1 , f2 , . . . , fr ) ⊂ (x1 , x2 , . . . , xN ) ⊂ I.

Therefore, I = (x1 , x2 , . . . , xN ) which is a contradiction since xN+1 ∈ I but xN+1 < (x1 , x2 , . . . , xN ).
Hence, I = (x1 , x2 , . . . , xn , . . .) is not a finitely generated ideal.

Lemma 11.6. Let R be a PID and x ∈ R not a unit. Then x is divisible by a prime element.

Proof. We consider the ideal I = (x). Since x is not a unit, ideal I is a proper ideal of R. It follows
that I is contained in a maximal ideal, say M, c.f. Corollary 3.6. The ideal M is a principal ideal
because R is a PID. Therefore M = (p) for some p ∈ R. The ideal M is also a prime ideal, so p is a
prime element. Now x ∈ I ⊂ (p) yields x = cp for c ∈ R. This completes the proof. □

Proposition 11.7. Let R be a PID. Every nonzero, non-unit element of R can be written as a product
of finitely many prime elements of R.

Proof. We begin with an element x ∈ R which is nonzero and not a unit. By Lemma 11.6, we find a
prime element p1 ∈ R such that x = p1 x1 for x1 ∈ R. Since x , 0, we have x1 , 0. We get (x) ⫋ (x1 )
because p1 is not a unit.
If x1 is a unit, then (x) = (p1 ) is a prime ideal, i.e. x is a prime element. We are done. Otherwise
by Lemma 11.6, we find a prime element p2 ∈ R such that x1 = p2 x2 for x2 ∈ R. The inequality
x1 , 0 yields x2 , 0. We observe that (x1 ) ⫋ (x2 ) because p2 is not a unit.
If x2 is a unit, then (x1 ) = (p2 ) which implies that x1 is a prime element. It follows that x = p1 x1
is a prime factorization. In this case the proof completes here. If x2 is not a unit, we get a prime
element p3 ∈ R such that x2 = p3 x3 by Lemma 11.6. Since p3 is not a unit, we have (x2 ) ⫋ (x3 ).
23
 Now we check whether x3 is a unit and repeat the argument as before. The following diagram
explains the steps in the argument.

(x)

{ #
(p1 ) (x1 )

{ "
(p2 ) (x2 )

|

(p3 )

We must have xN to be a unit for some N since otherwise we would obtain a strictly ascending
chain of ideals
(x) ⫋ (x1 ) ⫋ (x2 ) ⫋ . . .
which contradicts that R is a Noetherian ring. It follows that xN−1 = pN xN is a prime element.
Therefore x = p1 p2 . . . pN−1 xN−1 is a factorization of x into prime elements. This completes the
proof. □

Lemma 11.8. Let R be an integral domain and p1 , . . . , pn , q1 , . . . , qm be prime elements of R such

that
p1 p2 . . . pn = uq1 q2 . . . qm
for some unit u in R. Then m = n and there exists a pertmutation σ ∈ S n such that pi and qσ(i) are
associate in R for i = 1, . . . , n.

Proof. It is easy to see that n = 0 if and only if m = 0. Without loss of generality we may assume
that 1 ≤ n ≤ m. We have u−1 p1 p2 . . . pn = q1 q2 . . . qm , so p1 divides q1 q2 . . . qm . Since p1 is a prime
element, p1 must divide some qi1 . This implies that qi1 = p1 u1 for u1 ∈ R. Now qi1 is irreducible
and p1 is not a unit. Therefore u1 is a unit in R. Thus we have p1 p2 . . . pn = uu1 p1 j,i1 q j . Since
Q
R is an integral domain, we may cancel out p1 from both sides. We get p2 . . . pn = uu1 j,i1 q j .
Q
Repeating this argument and cancelling out p2 , . . . , pn one by one, we obtain
n
Y Y
1=u uj qj
i=1 j∈I

where u1 , u2 , . . . , un are units and I = {1, . . . , m} \ {i1 , . . . , in }. If n < m, then I , ∅. We get a

contradiction because we have a prime element q j , j ∈ I which divides 1. Therefore n = m. Let
σ ∈ S n be given by σ( j) = i j for 1 ≤ j ≤ n. Our argument shows that qi j = p j u j , i.e. p j and qσ( j)
are associate in R for j = 1, . . . , n. □

A careful analysis of the proof of the lemma above actually gives the following.
24
Lemma 11.9. Let R be an integral domain. Let p1 , . . . , pn be prime elements and q1 , . . . , qm be
irreducible elements of R such that
p1 p2 . . . pn = uq1 q2 . . . qm
for some unit u in R. Then m = n and there exists a pertmutation σ ∈ S n such that pi and qσ(i) are
associate in R for i = 1, . . . , n.
Can you prove it yourself? You cannot say n ≤ m without loss of generality in your proof.

12. 7th March, 2019

We introduce UFD, g. c. d. and discuss examples.
Definition 12.1. An integral domain R is called a Unique Factorization domain (UFD) if every
nonzero, non-unit r ∈ R has the following two properties :
(1) r can be written as a finite product of irreducible elements, i.e. r = p1 p2 . . . pn for some
irreducible elements p1 , . . . , pn .
(2) the decomposition is unique up to assciates, i.e. if r = q1 q2 . . . qm is another factorization of r
into irreducible elements, then m = n and there exists a permutation σ ∈ S n such that pi and
qσ(i) are associate in R for i = 1, 2, . . . , n.
Example 12.2. (1) Any field is vacuously a UFD.
(2) Proposition 11.7 and Lemma 11.8 together show that any PID is a UFD.
Lemma 12.3. In a UFD, every irreducible element is a prime element.
Proof. Let x be an irreducible element of R. The element x is not a unit. Let x divide ab for some
a, b ∈ R. We want to show that x divides one of a, b. It is enough to assume that both a, b are
nonzero elements.
Now ab = cx for some c ∈ R. Since a, b , 0, we get c , 0. If one of a, b is a unit, then x divides
the other element. Therefore we may assume that both a, b are nonzero and not units. It follows
that the element c is not a unit because otherwise we would get x = c−1 ab, a contradiction as x is
irreducible and both c−1 a, b are not units.
Since R is a UFD, each of the elements a, b, c can be factorised into a finite product of irreducible
elements as follows:
a = p1 p2 . . . pm ,
b = pm+1 . . . pn ,
c = q1 . . . qk .
Now one can factorise ab into irreducible elements in two different ways, viz.
ab = p1 p2 . . . pn = q1 . . . qk x.
By uniqueness of decomposition in a UFD, we get n = k + 1. Moreover x is associated to some
pi , 1 ≤ i ≤ n, i.e. pi = ux for some unit u in R. If 1 ≤ i ≤ m, then x divides a, otherwise x divides
b. Hence x is a prime element and the lemma follows. □
25
 √
Example 12.4. 1) The ring Z[ −3] is not a UFD since 2 is an irreducible element but not a
prime element.
k[x,y,z]
2) The ring (xz−y 2 ) is not a UFD since x̄ is an irreducible element but not a prime element.

Definition 12.5. (Greatest common divisors(g.c.d.s)) Let a, b be elements of an integral domain R.

A nonzero element d of R is called a greatest common divisor (g.c.d.) of a and b if it satisfies the
following:
1) d divides both a and b,
2) if c divides a and b, then c divides d.

Remark 12.6. In an integral domain R, g.c.d. of two elements a, b is unique upto associate. Let
d, d′ both be g.c.d. of a, b. Then d divides d′ and d′ divides d. This implies that (d) = (d′ ). Therefore
d and d′ are associate in R (see Lemma 11.2). The g.c.d. of a and b is denoted by g. c. d.(a, b).

Proposition 12.7. Let a, b be elements of an integral domain R such that at least one of a, b is
nonzero. Then the g.c.d. of a, b exists if and only if the ideal
I = ∩(a,b)⊂(x) (x)
is a principal ideal. Moreover if I = (d), then d is the g.c.d. of a and b.

Proof. First we assume that g. c. d.(a, b) exists and equals d. Clearly (a, b) ⊂ (d). If (a, b) ⊂ (x),
then x a and x b. It follows that x d which gives (d) ⊂ (x). Therefore I = ∩(a,b)⊂(x) (x) = (d).
Conversely we assume that I is principal and I = (d) for d ∈ R. Since I , 0, the element d , 0.
We have (a, b) ⊂ I = (d). Therefore d a and d b. If x ∈ R\{0} such that x a and x b, then (a, b) ⊂ (x).
This implies that (d) = I ⊂ (x) and consequently x d. Therefore d is the g.c.d. of a and b. □

Example 12.8. 1) If R is a PID, then the g.c.d. of any two elements of R exists. Let a, b ∈ R.
The ideal (a, b) is a principal ideal. If (a, b) = (d), we have ∩(a,b)⊂(x) (x) = (d). Therefore d is
the g.c.d. of a and b.
2) Let R = Z[x]. Then g. c. d.(4, 2x) = 2, but (4, 2x) ⫋ (2). Thus in general the principal ideal
generated by g. c. d.(a, b) may not be equal to the ideal generated by a and b.

The following result is easy and can be proved using Lemmas 11.9, 12.3.

Theorem 12.9. Let R be an integral domain. Then R is a UFD if and only if every nonzero, non-unit
can be expressed as a finite product of prime elements.

13. 8th March, 2019

We begin with the following crucial observations.
1) A proper subring R of an integral domain S may be isomorphic to S , e.g. S = R[x, y] and
R = R[x2 , y2 ].
2) Any integral domain is a subring of its field of fractions. A field is a Euclidean domain.
Therefore any integral domain which is not a UFD shows that a subring of a Euclidean
domain (PID or UFD) may not be a Euclidean domain( PID or UFD).
26
 √
3) Consider the ring extension√ Z ⊂ Z[ −1]. The integer 2 is irreducible (prime) in Z but not
irreducible (prime) in Z[ −1]. Therefore irreducible and prime properties are not preserved
under ring extensions.
4) Consider the ring extension k[x, y] ⊂ k(x)[y]. The element xy is an irreducible (prime) el-
ement of k(x)[y] but xy is reducible in k[x, y]. Thus irreducible and prime properties of an
element are not preserved when the element is viewed as an element of a proper subring.
5) If all nonzero prime ideals√of an integral domain are maximal ideals, then the integral domain
may not be a PID, e.g. Z[ −3]. √
6) The g.c.d. √of two elements in an integral√domain may not exist. Let R = Z[ −5], a = 6 and
b = 2(1 + −5). √ Let the function
√ N : Z[ −5] → N ∪ {0} be given by N(α)√= αᾱ = a2 + 5b2
for α =√ a + b −5 ∈ Z[ −5]. Then N(αβ) = N(α)N(β) for α, β ∈ Z[ −5]. Moreover
α ∈ Z[ −5] is a unit if and only if N(α) = 1.
If possible, we assume that g. c. d.(a, b) exists and is equal to d. We observe that N(a) = 36
and N(b) = 24. Since d a and d b, we obtain N(d) 36 and N(d) 24. It follows that N(d) 12.
√ √
Now 2 6 and 2 {2(1 + −5)} together imply 2 d and therefore 4 N(d). Similarly (1 + −5) a
√ √
and (1 + −5) b together imply that (1 + −5) d. Consequently 6 N(d). Therefore 12 N(d).
Thus we conclude that N(d) = 12 which is a contradiction since a2 + 5b2 = 12 has no integer
solution. Hence g. c. d.(a, b) does not exist.
13.1. Computation of g.c.d. in a Euclidean domain. Let R be a Euclidean domain and a, b be
two nonzero elements of R. Let N : R \ {0} → N ∪ {0} be the Euclidean function. We compute
g.c.d. of a and b.
We have a = q1 b + r1 where q1 , r1 ∈ R and r1 = 0 or N(r1 ) < N(b). If r1 = 0, we stop here. If
r1 , 0, we have q2 , r2 ∈ R such that b = q2 r1 + r2 where r2 = 0 or N(r2 ) < N(r1 ). If r2 = 0 we stop.
If r2 , 0, we have q3 , r3 ∈ R such that r1 = q3 r2 + r3 where r3 = 0 or N(r3 ) < N(r2 ). If r3 = 0, we
stop. Otherwise we argue as before.
We must get rN = 0 after a finite number of steps since otherwise we would have a strictly
decreasing sequence of non-negative integers, viz. N(r1 ) > N(r2 ) > . . . which is absurd.
Thus we obtain a sequence of equations a = q1 b + r1 , b = q2 r1 + r2 , r1 = q3 r2 + r3 , . . . ,
rN−2 = qN rN−1 . The first equation gives (a, b) = (b, r1 ). The second equation gives (b, r1 ) = (r1 , r2 )
and so on. Therefore we have (a, b) = (b, r1 ) = (r1 , r2 ) = . . . = (rN−1 ). Hence g. c. d.(a, b) = rN−1 .
Example 13.1. Let R = Z[i], a = 1 + 3i and b = 1 + 7i. We recall the proof of Lemma 9.3 showing
that Z[i] is a Euclidean domain.
We have ab = 1+7i
1+3i
= 501 (1 + 3i)(1 − 7i) = 501 (22 − 4i) = 0.44 − 0.08i. Integers nearest to 0.44, 0.08
are 0, 0 respectively. Then q1 = 0 + 0i = 0 and r1 = a − bq1 a = a
Now rb1 = ba = 1+3i
1+7i
= 101 (1 + 7i)(1 − 3i) = 101
(22 + 4i) = 2.2 + 0.4i. Integers nearest to 2.2, 0.4 are
2, 0 respectively. Therefore q2 = 2 + 0i = 2 and r2 = b − r1 q2 = (1 + 7i) − 2(1 + 3i) = −1 + i.
Now rr12 = −1+i
1+3i
= 12 (1 + 3i)(−1 − i) = 21 (2 − 4i) = 1 − 2i ∈ Z[i]. It follows that q3 = 1 − 2i and
r3 = r1 − q3 r2 = 0. Therefore (a, b) = (r2 ) and g. c. d.(a, b) = r2 = −1 + i.
Lemma 13.2. Let R be a UFD. Let a = upa11 . . . pann where u is a unit, ai ∈ N and p1 , p2 , . . . , pn are
prime elements of R such that no two different pi ’s are associate in R. If pb1 a, then b ≤ a1 .
27
Proof. It is given that pb1 a, so a = upa11 . . . pann = pb1 c for some c ∈ R. If possible we assume that
b > a1 .
The ring R is an integral domain, so we may cancel out pa11 on both sides of the above equality.
We obtain upa22 . . . pann = pb−a
1
1
c. Multiplying both sides by u−1 , we get pa22 . . . pann = u−1 pb−a
1
1
c.
Since b − a1 > 0, we have p1 p2 . . . pn . It follows that p1 pi for 2 ≤ i ≤ n. Since pi is irreducible,
a2 an

we conclude that p1 and pi are associate in R which contradicts the hypothesis. Therefore, our
assumption is wrong and consequently b ≤ a1 . □

Theorem 13.3. Let R be a UFD, then the g.c.d. of any two nonzero elements exists.

Proof. Let a, b be nonzero elements of R. Now each of a, b can be factorized into a finite product
of irreducible elements as follows :
a = upa11 . . . pann qb11 qb22 . . . qbmm ,
a′ a′
b = u′ p11 . . . pnn r1c1 . . . rkck ,
Where u, u′ are units in R, no two different prime elements in each of the sets {p1 , . . . , pn , q1 , . . . , qm },
{p1 , . . . , pn , r1 , . . . , rk }, {q1 , q2 , . . . , qm , r1 , . . . , rk } are associate in R and ai , bi , a′i , ci ∈ N. We see at
min{a ,a′ } min{a ,a′ }
once that d = p1 1 1 . . . pn n n divides both a and b.
Let c a and c b. Any prime factor p of c divides both of a and b. This implies that p is associate
to one of the prime factors of a as well as b. If p is associated to qi , then p is not associate to any
prime factor of b since qi is so. Therefore p is associated to some pi . In other words, each prime
factor of c is associate to one of {p1 , . . . , pn }. Therefore c can be written as c = u′′ pd11 . . . pdnn for
some unit u′′ in R and nonnegative integers d1 , . . . , dn . Each pdi i divides both a and b. By Lemma
13.2, di ≤ min{ai , a′i } for i = 1, . . . , n . Therefore c divides d. Hence g. c. d.(a, b) exists and is equal
to d. □

The following lemma follows from the formula of g.c.d. of two elements in a UFD as described
in the proof of the above lemma.

Lemma 13.4. Let R be a UFD and a, b, c be nonzero elements of R. Then

g. c. d.(ac, bc) = uc g. c. d.(a, b)
for some unit u in R.

14. 12th March, 2019

We show that R[x] is a UFD if and only if R is so. We begin with the following Lemma.

Lemma 14.1. Let R be an integral domain and r ∈ R. Then r is an irreducible (prime or unit)
element of R if and only if r is an irreducible (prime or unit) element of R[x].

Proof. We recall the following isomorphism of rings :

R[x]
 R/(r)[x].
(r)
28
 The ring R[x]
(r)
= 0 if and only if R/(r) = 0, so r is a unit in R[x] if and only if r is a unit in R. The
R[x]
ring (r) is an integral domain if and only if R/(r) is so. Therefore r is a prime element of R[x] if
and only r is a prime element of R.
To prove the remaining part of the lemma, we show that r is reducible in R if and only if r is so
in R[x]. First, we assume that r is reducible in R. Then r = ab where a, b are nonzero, non-unit
elements of R. Since a, b are non-units in R, they are not-units in R[x]. Therefore, r is reducible in
R[x].
Conversely, we assume that r is reducible in R[x]. Then r = f g for nonzero, non-unit polynomi-
als f, g ∈ R[x]. Since R is an integral domain, comparing degrees of both sides we conclude that
deg( f ) = deg(g) = 0, i.e. f, g ∈ R (see Lemma 5.2). The elements f, g are not units in R. Therefore
f is reducible in R. □
The following is an immediate consequence of the above.

Lemma 14.2. Let R[x] be a UFD, then R is a UFD.

If a1 , . . . , an are nonzero elements of a UFD R, then we define

g. c. d.(a1 , a2 , . . . , an ) = g. c. d.(a1 , g. c. d.(a2 , g. c. d.(a3 , . . .) . . .).
The order in which the parentheses appear in the RHS does not really matter. This is so be-
cause g. c. d.(a, g. c. d.(b, c)) = g. c. d.(g. c. d.(a, b), c) upto associates. In fact either side of the later
equality equals an element d such that (d) = ∩{a,b,c}⊂(x) (x).

Definition 14.3. Let R be a UFD and f ∈ R[x]. The content of f is defined to be the g.c.d. of the
coefficients of f . It is denoted by c( f ).

The following is a trivial consequence of Lemma 13.4.

Lemma 14.4. Let R be a UFD, f ∈ R[x] and d ∈ R. Then c(d f ) = ud c( f ) for some unit u in R.
Moreover f = c( f ) f1 where f1 ∈ R[x] such that c( f1 ) = 1.

Proposition 14.5. Let R be a UFD and f, g ∈ R[x], then c( f g) = u c( f ) c(g) for some unit u in R.

Proof. First, we show that c( f g) c( f ) c(g). We prove this by induction on the number n of prime
factors of c( f g). If n = 0, then c( f g) is a unit, so c( f g) c( f ) c(g) trivially. We assume n ≥ 1. Let p
be a prime factor of c( f g). It follows that p f g. Since p is a prime element of R, it is also a prime
element of R[x]. Therefore either p f or p g. Without loss of generality we assume that p f . Then
f = p f ′ for some f ′ ∈ R[x]. Clearly c( f ) = pu c( f ′ ) and c( f g) = c(p f ′ g) = u′ p c( f ′ g) for some
units u, u′ ∈ R. One observes that the number of prime factors of c( f ′ g) is less than that of c( f g).
Therefore by induction hypothesis, we get c( f ′ g) c( f ′ ) c(g). This implies that p c( f ′ g) p c( f ′ ) c(g).
Therefore c( f g) c( f ) c(g).
Any coefficient of f g is of the form ai b j where ai is a coefficient of f and b j is that of g. Now
c( f ) ai and c(g) b j together imply that c( f ) c(g) ai b j . Consequently c( f ) c(g) c( f g).
Since R is an integral domain, c( f g) c( f ) c(g) and c( f ) c(g) c( f g) together imply that c( f g) =
u c( f ) c(g) for some unit u in R. □
29
Lemma 14.6. (Gauss’ Lemma) Let R be a UFD and F be the field of fraction of R. Assume that
p(x) ∈ R[x]. If p(x) is reducible in F[x], then p(x) is reducible in R[x]. More precisely if p(x) = f g
for f, g ∈ F[x] \ F, then p = FG for some polynomials F, G ∈ R[x] \ R such that F = r f and G = sg
for some nonzero r, s ∈ F.
Proof. We are given that p(x) = f g for f, g ∈ F[x] \ F. We find nonzero r0 , s0 ∈ R such that
F0 = r0 f ∈ R[x] and G0 = s0 g ∈ R[x]. Then r0 s0 p(x) = F0G0 . Comparing contents of both sides
we get that r0 s0 c(p) = u c(F0 ) c(G0 ) for some unit u in R.
Now F0 = c(F0 )F1 and G0 = c(G0 )G1 for some polynomials F1 , G1 ∈ R[x] such that c(F1 ) =
c(G1 ) = 1. It follows that r0 s0 p(x) = F0G0 = c(F0 ) c(G0 )F1G1 = u−1 r0 s0 c(p)F1G1 . Now R is an
integral domain and r0 , s0 are nonzero elements of R, so cancelling out r0 s0 from both sides we get
p(x) = u−1 c(p)F1G1 = FG,
where F = u−1 c(p)F1 and G = G1 .
Clearly both F, G ∈ R[x]. One observes that

F = u−1 c(p)F1 = u−1 c(p) c(F0 )−1 F0 = r0 u−1 c(p) c(F0 )−1 f and
G = G1 = c(G0 )−1G0 = s0 c(G0 )−1 g.
Taking r = r0 u−1 c(p) c(F0 )−1 and s = s0 c(G0 )−1 , we get F = r f and G = sg. Moreover r, s are
nonzero elements of F, therefore the result follows. □
Lemma 14.7. Let R be a UFD and c( f ) = 1. If f is reducible in R[x], then f is also reducible in
F[x] where F is the field of fraction of R.
Proof. Since f is reducible in R[x], we get nonzero, non-unit polynomials g, h ∈ R[x] such that
f = gh. If either one of g, h were a constant, we would find that a nonzero, non-unit element (viz.
g or h) of R divides c( f ), a contradiction since c( f ) = 1. Therefore deg(g), deg(h) ≥ 1. Hence f is
reducible in F[x]. □
Remark 14.8. Let R and F be as in Lemma 14.6. It shows that if f is irreducible in R[x], then f is
so in F[x]. Lemma 14.7 shows that the converse is true when c( f ) = 1. The assumption c( f ) = 1 is
necessary in Lemma 14.7, e.g. 2x is reducible in Z[x] but not reducible in Q[x] (Why?).
Lemma 14.9. Let R be a UFD and f, g ∈ R[x] be such that c( f ) = c(g) = 1. Let F be the field of
fraction of R. Then f, g are associate in R[x] if and only if they are so in F[x].
Proof. Let f , g are associate in R[x]. Then f = ug for some unit u in R[x]. Since R is an integral
domain u is actually a unit in R and therefore in F. Hence f, g are associate in F[x].
Conversely we assume that f and g are associate in F[x], i.e. f = ug for some unit u in F[x].
Since F is a field, u is actually a nonzero element of F. Let u = ba . It follows that b f = ag. Since
c( f ) = c(g) = 1, we have c(b f ) = b and c(ag) = a. Therefore a = u′ b for some unit u′ of R. Thus
u = u′ and so u is a unit in R. Hence f, g are associate in R[x]. □
Theorem 14.10. If R is a UFD, then R[x] is a UFD.
30
Proof. Let f be a nonzero, non-unit element of R[x]. We show that f can be written as a finite
product of irreducible elements and the decomposition is unique upto associates. Let F be the field
of fraction of R. We consider two cases.
Case - 1 : c( f ) = 1. We view f as an element of F[x]. Since F[x] is a PID, it is a UFD. Therefore
f can be written as f = p1 , . . . , pm for some irreducible polynomials p1 , . . . , pm ∈ F[x]. By Gauss’
lemma (see Lemma 14.6) we have f = q1 , . . . qm where each qi is a polynomial in R[x] and is
associated to pi in F[x]. Since pi is irreducible in F[x], the polynomial qi is irreducible in F[x].
By Gauss’ Lemma 14.6, qi is also irreducible in R[x]. Thus we find a decomposition of f into a
product of finitely many irreducible polynomials in R[x].
Now we assume that f = q1 q2 . . . qm , f = r1 r2 . . . rn are decompositions of f into products of two
different sets {q1 , . . . , qm } and {r1 , r2 , . . . , rn } of irreducible polynomials in R[x]. Since c( f ) = 1,
we conclude by Proposition 14.5 that c(qi ) = c(r j ) = 1 for all i and j. Since each qi , r j are
irreducible polynomials in R[x], they are so in F[x]. Now F[x] is a PID and therefore is a UFD.
It follows that m = n and after suitable reordering of r j ’s, the polynomial qi is associated to ri in
F[x] for i = 1, . . . , n. By Lemma 14.9, the polynomial qi is associated to ri in R[x] for i = 1, . . . , n.
Therefore decomposition of f is unique upto associates.
Case - 2 : c( f ) is not a unit.
We write f = c( f ) f1 for some polynomial f1 in R[x] such that c( f1 ) = 1. Now c( f ) is an element
of R and R is a UFD, so it is a finite product of irreducible elements of R. By Case - 1, f1 is also
a finite product of irreducible elements of R[x]. Each irreducible element in R is also irreducible
in R[x] (see Lemma 14.1). Thus we find a decomposition f into a finite product of irreducible
elements of R[x].
Now we assume that f = q1 q2 . . . qm , f = r1 r2 . . . rn are two different decompositions into finite
products of irreducible elements in R[x]. Let q1 , . . . , q s , r1 , . . . , rt be elements in R and the rest do
not belong to R. The content of a polynomial divides itself. Each of q s+1 , . . . , qm , rt+1 , . . . , rn are
irreducible elements of R[x] of degree at least one, so each of them has content 1. Therefore we get
c( f ) = q1 , . . . , q s and also c( f ) = r1 r2 . . . rn . It follows that

q1 , . . . , q s = ur1 r2 . . . rt

for some unit u of R. This implies that

uq s+1 , . . . , qm = rt+1 , . . . , rn

since q1 q2 . . . qm = r1 r2 . . . rn .
The ring R is a UFD and q1 , . . . , q s , r1 , . . . , rt are irreducible elements in R. Therefore the first
equation gives that s = t and qi is associated to ri in R (so in R[x]) for i = 1, . . . , s after some
reordering of ri ’s. The product in either sides of the second equation has content 1. By Case - 1,
m − s = n − t which yields m = n and after some reordering of r j , s + 1 ≤ j ≤ m the polynomial q j
is associated to r j for j = s + 1, . . . , m. Therefore decomposition of f into a product of irreducible
elements is unique upto associates. □

Corollary 14.11. Let R be a UFD, then R[x1 , x2 , . . . , xn ] is a UFD.

31
 Let f ∈ R[x]. If f (r) = 0 for some r ∈ R, then r is called a root of f . The following two results
are easy to establish. We skip details.
Lemma 14.12. Let F be a field and let p(x) ∈ F[x]. Then p(x) has a factor of degree one if and
only if p(x) has a root in F.
Lemma 14.13. A polynomial of degree at most three over a field F is reducible if and only if it has
a root in F.
Theorem 14.14. (Eisenstein’s Criterion) Let P be a prime ideal of an integral domain R and f (x) =
xn + a1 xn−1 + . . . + an be a polynomial in R[x] such that a1 , . . . , an ∈ P but an < P2 . Then f (x) is
irreducible in R[x].
Proof. If possible we assume that f is not irreducible. Then f = gh for some nonzero g, h ∈ R[x]
which are not units in R[x]. Since f is a monic, if one of g, h were a constant polynomial, it would
be a unit in R. Therefore deg(g), deg(h) ≥ 1. Also both g, h are monic polynomials since f is a
monic polynomial and R is an integral domain.
There is a natural surjective ring homomorphism F : R[x] → RP [x] given by F(a0 xn + a1 xn−1 +
. . . + an ) = a0 xn + a1 xn−1 + . . . + an . Here ai denotes the image of ai in R/P. Applying F on both
sides of f = gh, we get xn = F(g)F(h). Since g, h are monic polynomials of degree at least one, it
follows that F(g), F(h) are of the form uxi , 1 ≤ i ≤ n − 1 for some unit u of R/P. Therefore constant
terms of g, h, viz g(0), h(0) are in P. This implies that an = g(0)h(0) ∈ P2 , a contradiction. Hence
f is an irreducible polynomial. □
As an application of the above result, following polynomials can be shown to be irreducible in
the respective rings, they are contained in.
(1) x4 + 10x + 5 ∈ Z[x],
(2) xn − p ∈ Z[x] where p is a prime number,
(3) zn + xy ∈ Z[x, y, z], n ≥ 1,
(4) x1n + x2n + . . . + xmn − 1 ∈ Z[x1 , x2 , . . . , xm ] where m ≥ 2 and n ≥ 1.

15. 14th March, 2019

In the next few sections we discuss the theory of modules over rings. Our final goal is to prove
the structure theorem of finitely generated modules over principal ideal domains. In the literature
there are two notions of modules, viz. left modules and right modules. To make the exposition
simple and avoid unnecessary details, we shall only consider left modules and call them modules.
Thus in a true sense, all modules considered in this note are actually left modules. We assume that
all rings contain identities. We do not assume rings to be commutative in the next few sections.
Definition 15.1. Let R be a ring with 1. A module over R is a set M together with two binary
operations, viz. addition + : M×M → M and multiplication × : R×M → M given by (a, b) 7→ a+b
and (a, b) 7→ ab respectively such that the following hold :
(1) (M, +) is an abelian group,
(2) (rs)m = r(sm) for r, s ∈ R and m ∈ M,
32
 (3) (r + s)m = rm + sm for r, s ∈ R and m ∈ M,
(4) r(m + n) = rm + rn for r ∈ R and m, n ∈ M,
(5) 1m = m for m ∈ M.
Example 15.2. (1) Let R be a ring with 1 , 0. Then any ideal I of R is an R-module under
addition and left-multiplication by elements of R.
(2) Let Rn = R × . . . × R, n times. Then Rn is a module over R under componentwise addition and
multiplication.
(3) Any abelian group G is an Z module. Upto isomorphism, we may assume that the operation
on G is additive. If + denotes the additive operation on G, then the left-multiplication of
a ∈ G by an integer n is given by
na = a + a + . . . + a, n times if n > 0
= (−a) + (−a) + . . . + (−a), (−n) times if n < 0
= 0 if n = 0.
(4) Let V be a vector space over a field k, then V is a module over k.
(5) Let V be a vector space over a field k and T : V → V be a linear transformation. Then
V has a module structure over the polynomial ring k[X] under additive operation on V and
left-multiplication by elements of k[X] given by
(a0 X n + a1 X n−1 + . . . + an−1 X + an )v = a0 T n (v) + a1 T n−1 (v) + . . . + an−1 T (v) + an v.
Moreover any k[X]-module M can be obtained in this manner from a k-linear transformation
T : M → M given by T (v) = Xv.
Definition 15.3. Let M be an R-module. An additive subgroup N of M is called a submodule of M
if it is closed under multiplication by elements of R.
Example 15.4. • Let R be a ring. Then any submodule of R is a left ideal of R.
• Let T : V → V be a linear transformation on a vector space V over k. We consider V as a
k[X] module as described above. Then a k[X] submodule of V is a subspace W of V such that
T (W) ⊂ W.
The following is easy.
Lemma 15.5. Let R be a ring with 1 and M be an R-module. A nonempty subset N of M is a
submodule of M if and only if x + ry ∈ N for all x, y ∈ N and r ∈ R.
Quotient modules : Let M be a module over a ring R and N be a submodule of M. The additive
quotient group M/N has a module structure over R under addition and multiplication respectively
given by
m̄ + m̄′ = m + m′ , for m, m′ ∈ M,
rm̄ = rm, for r ∈ R and m ∈ M.
Here m̄ denotes the additive coset of N containing m. The reader should verify that the multipli-
cation by elements of R is well defined and M/N satisfies all axioms of modules.
33
Definition 15.6. Let M be a module over a ring R. Then we define annihilator of R as
ann M = {r ∈ R : rm = 0 for all m ∈ M}
Lemma 15.7. Let M be a module over a ring R. Then ann M is an ideal of R
Proof. Let x, y ∈ ann M and r ∈ R. For any m ∈ M we observe the following :
(x − y)m = xm − ym = 0 − 0 = 0,
(xr)m = x(rm) = 0,
(rx)m = r(xm) = 0.
Therefore x − y, rx, xr ∈ ann M. Hence ann M is an ideal of R. □
Lemma 15.8. Let M be a module over R and I be an ideal of R contained in ann M. Then M has
a module structure over R/I under addition and multiplication given by r̄m = rm for r ∈ R and
m ∈ M.
Proof. We shall only check that the action of R/I on M is well defined and leave it to the reader
to verify that M satisfies all axioms to be a module over R/I under addition and multiplication by
elements of R/I. Let r̄ = s̄ and m ∈ M. Then r − s ∈ I ⊂ ann M which gives (r − s)m = 0. Therefore
rm = sm and the action of R/I on M is well defined. □

16. 15th March 2019

All rings contain identities and are not necessarily commutative.
Definition 16.1. Let M, N be modules over a ring R. A map f : M → N is called an R-module
homomorphism if it satisfies the following :
(1) f (m + m′ ) = f (m) + f (m′ ) for any m, m′ ∈ M,
(2) f (rm) = r f (m) for r ∈ R and m ∈ M.
A bijective homomorphism is called an isomorphism.
Definition 16.2. Let f : M → N be an R-module homomorphism. Then we define kernel of f as
ker f = {x ∈ M : f (x) = 0} and image of f as Im f = { f (x) : x ∈ M}.
The following are easy to prove.
(1) Let f : M → N be an R-module homomorphism. If ker f = 0, then f is injective.
(2) Let M, N be modules over a commutative ring R. Let Hom(M, N) denote the set of R-
module homomorphisms from M to N. Then Hom(M, N) forms an R-module under addition
and multiplication respectively given by
( f + g)(x) = f (x) + g(x) for f, g ∈ Hom(M, N) and x ∈ M,
(r f )(x) = r f (x) for f ∈ Hom(M, N) and r ∈ R.
Definition 16.3. Let N1 , N2 be submodules of an R-module M. Then we define
N1 + N2 = {x + y : x ∈ N1 , y ∈ N2 }.
34
 It is easy to check that N1 + N2 is a submodule of M. We state isomorphism theorems for modules
below. The proofs are verbatim as that of their counterparts for rings. We leave it to the reader to
work out the detail.
Theorem 16.4. (The first isomorphism theorem) Let f : M → N be a surjective R-module homo-
morphism. Then ker f is a submodule of M. The map ϕ̃ : M → N given by ϕ̃( x̄) = ϕ(x) is an
isomorphism. Here x̄ denotes the class of an element x in M.
Theorem 16.5. (The second isomorphism theorem) Let N1 , N2 be submodules of an R-module M.
Then N1N+N
2
2
 N1N∩N
1
2
.
Theorem 16.6. (The third isomorphism theorem) Let N1 , N2 be submodules of an R-module M
such that N1 ⊂ N2 ⊂ M. Then NM/N 1
2 /N1
 NM2 .
Theorem 16.7. (The fourth isomorphism theorem) Let N be a submodule of an R-module M. Then
the correspondence P ←→ NP establishes an inclusion preserving bijection between the following
sets:
{P : P is a submodule of M containing N} ←→ {B is a submodule of M/N}.
The centre of a ring R is the set of all elements x ∈ R which commute with every element of R,
i.e. xr = rx for all r ∈ R.
Definition 16.8. Let R be a commutative ring with identity 1R , 0. An R-algebra is a ring A (not
necessarily commutative) with identity 1A which admits a ring homomorphism iR : R → A such
that the subring f (R) is contained in the centre of A and iR (1R ) = 1A .
Example 16.9. Let R be a commutative ring with 1R , 0.
(1) R[x1 , . . . , xn ] is an R-algebra. Here iR (r) = r for r ∈ R.
(2) Mn (R) is an R algebra. If In denotes the identity matrix of size n, then iR (r) = rIn for r ∈ R .
Note that any R-algebra A is a module over R under addition and multiplication given by ra =
iR (r)a for r ∈ R, a ∈ A.
Definition 16.10. Let R be a commutative ring with 1R , 0 and A, B be algebras over R. Let
i : R → A, j : R → B be structural ring homomorphisms. An R-algebra homomorphism is a ring
homomorphism f : A → B such that the following diagram commutes
A^ f / B.
@

i j
R
If we regard R-algebras A, B as modules over R as explained above, then an R-algebra homomor-
phism f : A → B is a ring homomorphism sending 1A to 1B , which is an R-module homomorphism
as well.
Remark 16.11. Let R be a commutative ring and A be an algebra over R. Then an R-algebra
homomorphism f : R[x1 , . . . , xn ] → A is uniquely determined by the images of x1 , . . . , xn under f .
35
 17. 19th March 2019
Let M be a module over a ring R and X be a nonempty subset of M. We define
RX = {r1 x1 + r2 x2 + . . . + rn xn : ri ∈ R, xi ∈ X for i = 1, 2, . . . , n and n ∈ N}.
One can check that RX is a submodule of M. The module M is said to be generated by X if
M = RX. M is called a finitely generated module if M = RX for some finite subset X of M.

Exercise 1. Let R = k[x, y] and m = (x, y). Find a generating set of mn containing n + 1 elements.
Show that mn cannot be generated by less than n + 1 elements.

The following is a generalisation of Definition 16.3.

Definition 17.1. Let Mα , α ∈ I be a family of submodules of a module N. Then we define sum of

the modules
X
Mα = {xα1 + xα2 + . . . + xαk : xαi ∈ Mαi , αi ∈ I for 1 ≤ i ≤ k and k ∈ N}
α∈I

One readily checks that α∈I Mα is a submodule of M. If I = {1, 2, . . . , n}, then ni=1 Mi is the set
P P
of all elements of the form ni=1 xi where xi ∈ Mi .
P

Definition 17.2. Let Mα , α ∈ I be a family of modules over a ring R. We define the product of
modules as
Y
Mα = {(xα ) : xα ∈ Mα for α ∈ I}.
α∈I

Here (xα ) denotes a set map I → ⊔α∈I Mα (disjoint union) such that α 7→ xα ∈ Mα . Clearly
Πα∈I Mα is a module over R under component wise addition and multiplication by elements of R. If
I = {1, 2, . . . , n}, then we have
M1 × . . . × Mn = {(x1 , . . . , xn ) : xi ∈ Mi for i = 1, 2, . . . , n}.

Definition 17.3. Let Mα , α ∈ I be a family of modules over a ring R. The direct sum of modules is
the submodule of Πα∈I Mα defined as
M Y
Mα = {(xα ) ∈ Mα : xα = 0 for all but finitely many α ∈ I}.
α∈I α∈I

If I = N, then M = {(x
Li ) : xi = 0 for all i ≥ N for some N ∈ N}. It is easy to see that if I
L
Qα∈I α
is a finite set, then α∈I Mα = α∈I
Mα .

Proposition 17.4. Let Ni , i ∈ I be a family of submodules of a module M over a ring R. Then the
following are equivalent.
(1) The map π : ⊕i∈I Ni → i∈I Ni given by π({ni }i∈I ) = i∈I ni is an isomorphism.
P P
(2) Ni ∩ j∈I\{i} N j = 0 for all i ∈ I.
P
(3) Every element x ∈ i∈I Ni can be uniquely written as x = xi1 + . . . + xik where i1 , . . . , ik are k
P
distinct elements of I and xi j ∈ Ni j , j = 1, . . . , k.
36
Proof. (1) =⇒ (2) : Let x ∈ Ni ∩ j∈I\{i} N j . Then we get x = ni = lk=1 n jk where ni ∈ Ni ,
P P
n jk ∈ N jk for k = 1, . . . , l and { j1 , j2 , . . . , jl } are l distinct elements of I \ {i}. We define an element
(xα ) ∈ ⊕i∈I Ni as follows :
ni if α = i,





xα =  −n jk if α = jk , k = 1, . . . , l,





0 otherwise.


It easy to see that π(xα ) = 0. Since π is injective, it follows that xα = 0. Therefore x = ni = 0.

Hence (2) follows.
(2) =⇒ (3) : If possible we assume that
x = xi1 + . . . + xik = x j1 + . . . + x jl
where {i1 , . . . , ik } and { j1 , . . . , jl } are subsets of I containing k, l distinct elements respectively. We
further assume that all xis , x jt are nonzero elements. We show that {i1 , . . . , ik } = { j1 , . . . , jl } and
xis = x jt whenever i s = jt .
If i s < { j1 , . . . , jl }, then xis = x j1 + . . . + x jl − xi1 − . . . − xis−1 − xis+1 − xik ∈ Nis ∩ j∈I\{is } N j = 0
P
which is a contradiction. Therefore i s ∈ { j1 , . . . , jl }. We obtain similarly that each jt ∈ {i1 , . . . , ik }.
Therefore {i1 , . . . , ik } = { j1 , . . . , jl } which implies that k = l. Now we assume that i s = jt . We have
X
xis − x jt = x j1 + . . . x jt−1 + x jt+1 + . . . + x jl − xi1 − . . . − xis−1 − xis+1 . . . − xik ∈ Nis ∩ N j = 0.
j∈I\{i s }

Therefore (3) follows.

(3) =⇒ (1) : It is clear that π is a surjective R-module homomorphism. We only need to show
that π is an injective R-module homomorphism. Let π((xi )) = 0. We have a finite set {i1 , . . . , ik } ⊂ I
such that xi = 0 whenever i < {i1 , . . . , ik }. It follows that xi1 + . . . + xik = 0. By uniqueness of
presentation, we get that xi = 0 for all i ∈ {i1 , . . . , ik }. Therefore (xi ) = 0. Hence π is an injective
R-module homomorphism. □

Definition 17.5. Let M be an R-module and {Nα : α ∈ I} be a family of submodules of M. We say

that M is the direct sum of Nα , α ∈ I if the following hold :
(1) M = α∈I Nα ,
P
(2) Nα , α ∈ I satisfy one of the equivalent conditions of Proposition 17.4.
In this case the module M is written as M =
L
N .
α∈I α

Example 17.6. Let M be an R-module and N1 , N2 be submodules of M. Then M = N1

L
N2 means
that either of the following equivalent conditions hold :
N2 → M given by (n1 , n2 ) 7→ n1 + n2 is an R-module
L
(1) the R-module homomorphism N1
isomorphism.
(2) M = N1 + N2 and N1 ∩ N2 = {0},
(3) each x ∈ M can be uniquely written as x = n1 + n2 for some n1 ∈ N1 and n2 ∈ N2 .

Exercise 2. Let M be an R-module and N1 , N2 be submodules of M such that M = N1

L
N2 . Then
M/N1  N2 and M/N2  N1 .
37
Remark 17.7. Assume that N1 , N2 are submodules of an R-module M such that N1 ∩ N2 = 0. Then
by Definition 17.3, we have N1 ⊕ N2 = N1 × N2 . On the other hand, by Definition 17.5, we have
N1 ⊕ N2 = N1 + N2 ⊂ M. The two modules are isomorphic, so we are not getting a contradiction.
Here is the convention. Let Nα , α ∈ I be a family of modules over a ring R. If the modules Nα ’s
are embedded as a submodule
L in an R-module M and satisfy one of the equivalent conditions of
P
Proposition 17.4, then N should be
α∈I α L understood to be the submodule α∈I Nα of M. If the
module M is not available in the context, N is understood to be as in Definition 17.3.
α∈I α

Definition 17.8. Let X be a non empty subset of an LR-module M. Then M is called free on X, if M
is the direct sum of submodules Rx, x ∈ X, i.e. M = x∈X
Rx.

If M is free on X, we often say in short that M is a free R-module. In this case the set X is called
a basis of M. Every element of M is uniquely written as a finite linear combination of elements of
X with coefficients in R.

Lemma 17.9. Let R be a ring and X be a non empty set. Then there exists a module free on X.

Proof. Let x ∈ X. First, we construct a free module on {x}. We consider the set of symbols

Rx = {rx : r ∈ R}.

The set Rx forms an R-module under addition and multiplication by elements of R given by

r1 x + r1 x = (r1 + r2 )x, r1 , r2 ∈ R; (9)

r(r1 x) = (rr1 )x, r, r1 ∈ R. (10)

We identify x with 1x. Then x ∈ Rx and Rx is free on {x} since every element of Rx is uniquely
written as rx = r(1x).
We consider the R-module
M
Rx.
x∈X
 L
We identify x ∈ X with the map (xα ) : X → ⊔ x∈X Rx L ∈ x∈X
Rx whose value
L is 1x at x and
zero otherwise. Thus we L have an inclusion map X → x∈X
Rx. The module x∈X
Rx is free on
X, since every element in x∈X
Rx has a unique presentation of the form r1 x1 + . . . + rk xk for ri ∈ R
and xi ∈ X. □

Example 17.10. (1) Let X = {x1 , x2 }. Then the free module on X is the module

Rx1 ⊕ Rx2 = {(rx1 , sx2 ) : r, s ∈ R}.

We identify x1 with (1x1 , 0) and x2 with (0, 1x1 ). Thus X ⊂ Rx1 ⊕ Rx2 .

If an R-module M is free on a non empty set X then any R-module homomorphism from M to
another R-module N is uniquely determined by images of elements of X. This is because each ele-
ment of M can be uniquely written as a finite linear combination of elements of X with coefficients
in R. In fancy language, this fact is presented in the form of the following theorem.
38
Theorem 17.11. Let X be a non empty set and M be an R-module free on X. Let i : X → M denote
the inclusion map. Let N be another R-module and j : X → N be a set map. Then there exists a
unique R-module homomorphism ϕ : M → N such that the following diagram commutes.

> NO
j
ϕ

X
i / M

18. 21st March 2019

The following lemma shows that a finitely generated module has a finite basis.
Lemma 18.1. Let M be a finitely generated free module over a ring R. Then any basis has only
finitely many elements.
Proof. Let X be a basis of M and M be generated by g1 , . . . , gm . Each generator gi is an R-linear
combination of finitely many elements of X. Therefore we get a finite subset {x1 , . . . , xn } ⊂ X such
that
Xn
gi = ri j x j for i = 1, . . . , m and j = 1, 2, . . . , n
j=1
for some ri j ∈ R. This shows that M is generated by x1 , . . . , xn . If X \ {x1 , . . . , xn } , ∅, then any
x ∈ X \ {x1 , . . . , xn } is an R-linear combination of x1 , . . . , xn , which contradicts that X is a basis.
Hence X = {x1 , . . . , xn }. □
Lemma 18.2. Let R be a commutative ring with 1 , 0, m be a maximal ideal of R and k = R/m
denote the residue field. Let M be a finitely generated free module over R. Assume that M is free
on X. Then
dimk M/mM = #X.
Proof. Since M is finitely generated, the basis X is a finite set by Lemma 18.1. Let X = {x1 , . . . , xn }.
Now M/mM is a vector space over the field k = R/m, c.f. Lemma 15.8. We claim that the residue
classes x¯1 , . . . , x¯n form a basis of the vector space M/mM.
It is clear that x¯1 , . . . , x¯n generate M as a k-vector space. Let c̄1 x̄1 + . . . + c̄n x̄n = 0̄ where c̄i ∈ k.
It follows that c1 x1 + . . . + cn xn ∈ mM. Elements in mM are linear combination of x1 , . . . , xn
with coefficients in m. Therefore c1 x1 + . . . + cn xn = m1 x1 + . . . + mn xn for mi ∈ m. We obtain
(c1 − m1 )x1 + . . . + (cn − mn )xn = 0. Since X is a basis and xi ∈ X, we get ci − mi = 0 for i = 1, . . . , n.
Thus we obtain c̄i = 0 for i = 1, . . . , n. Therefore x̄1 , . . . , x̄n are linearly independent. Hence our
claim follows and dimk M/mM = n = #X. □
The following is a trivial consequence of the above Lemma.
Theorem 18.3. Let M be a finitely generated free module over a commutative ring R. Then any
two bases of M have the same number of elements.
Definition 18.4. Let M be a finitely generated R-module which is free on a set X. Then the rank of
M is defined to be the number of elements in X. It is denoted by rank(M).
39
Exercise 3. Let R be a ring and I be an ideal of R. Let M be a module over R. Define
I M = {i1 m1 + . . . + in mn : ik ∈ I, mk ∈ M for k = 1, . . . , n and n ∈ N}.
Show that M/I M is a module over R/I. Let N1 , N2 be submodules of M such that M = N1 ⊕ N2 .
Show that M/I M  N1 /IN1 ⊕ N2 /IN2 as an R/I module.
Ln
Ln In the setting of Lemma 18.2, we have M = i=1
Rxi . The above exercise shows that M/mM 
i=1
kxi as a vector space over k. The RHS is a vector space of dimension n. Thus we get an
alternative proof of Lemma 18.2.
Lemma 18.5. Let R be a commutative ring and I be a nonzero ideal of R. Then I is a free R-module
if and only if I = (d) for some nonzero divisor d in R.
Proof. If I is a principal ideal generated by a nonzero divisor d ∈ I, then any element of I can be
uniquely written as rd for some r ∈ R. Therefore I is free on {d}.
Now we assume that I is free on a nonempty subset X ⊂ I. If X contains at least two elements,
we choose two distinct elements x1 , x2 ∈ X. We observe that x2 x1 + (−x1 )x2 = 0x1 + 0x2 = 0. It
follows that x1 = −x2 = 0 which is a contradiction. Therefore X contains only one element. Let
X = {d}. Then I = Rd. If rd = 0 for some r ∈ R, then we have rd = 0d = 0. Since X = {d} is a
basis, we obtain r = 0. Therefore d is a nonzero divisor. Hence I is a principal ideal generated by
a nonzero divisor d ∈ R. □

19. 22nd March 2019

We introduce short exact sequence and discuss related results. We begin with the following
crucial observations.
Observations :
(1) All modules are not free. Let R = k[x, y] and I = (x, y). The ideal I is a finitely generated
module over R. The ideal I is not principal, so I is not a free R module.
(2) A maximal independent set of a module may not be a basis of the module. In the above
example {x} is a maximal independent set of the module I, but it does not generate I.
(3) A minimal generating set of a module may not be a basis of the module. In the above example,
{x, y} is a minimal generating set of I, but it is not a basis.
α β
Definition 19.1. A pair of R-module homomorphisms X → − Z is said to be exact if ker β = Im α.
− Y→
A sequence of R-module homomorphisms :
∂n+1 ∂n ∂n−1
. . . −−−→ Xn −→ Xn−1 −−−→ . . .
is called exact if Im ∂n+1 = ker ∂n for all n ∈ Z. Clearly ∂n ◦ ∂n+1 = 0 for all n ∈ Z.
A short exact sequence(s.e.s) of R modules is an exact sequence of R-modules of the form
ψ ϕ
0→A→
− B→
− C → 0.
Thus the above short exact sequence has the following properties :
(1) ψ is an injective R-module homomorphism,
(2) Im ψ = ker ϕ,
40
 (3) ϕ is a surjective R-module homomorphism
Example 19.2. Let M be an R-module and N be a submodule of R. Let i : N → M denote the
inclusion map and q : M → M/N denote the quotient map. Then the following is a short exact
sequence.
i q
0 → N→ − M→ − M/N → 0.
ψ ϕ
Definition 19.3. A short exact sequence 0 → A → − B →− C → 0 is called a split short exact
sequence if there is an R-module homomorphism τ : C → B such that ϕ ◦ τ = id.
Example 19.4. Let A, C be modules over a ring R and B = A ⊕ C. Let i : A → B be defined by
i(a) = (a, 0), a ∈ A and pr : B → C be defined by pr(a, c) = c, (a, c) ∈ A ⊕ C.
Then the sequence
i pr
0 → A→ − B −→ C → 0
is a split short exact sequence. The sequence is exact since i is injective, pr is surjective and
Im i = ker pr = {(a, 0) : a ∈ A}. We define a homomorphism τ : C → B by τ(c) = (0, c), c ∈ C.
Clearly pr ◦ τ = id. Therefore the above is a split short exact sequence.
i q
Proposition 19.5. Let E : 0 → A → − B→ − C → 0 be a short exact sequence of R-modules. Then the
following are equivalent :
(1) the sequence E is a split short exact sequence,
(2) there is a submodule D of B such that B = i(A) ⊕ D.
Proof. We begin with the observation that q ◦ i = 0.
(1) =⇒ (2) : We assume that E is a split short exact sequence. We have an R-module homomor-
phism τ : C → B such that q ◦ τ = id. We prove that B = i(A) ⊕ τ(C).
Let x ∈ B. We observe that q(x − τq(x)) = q(x) − qτq(x) = q(x) − q(x) = 0. This implies that
x − τq(x) ∈ ker q = Im i = i(A). Now x = (x − τq(x)) + τq(x) ∈ i(A) + τ(C). Therefore we get
B = i(A) + τ(C).
Let x ∈ i(A) ∩ τ(C). Then x = i(a) = τ(c) for a ∈ A and c ∈ C. We observe that c = qτ(c) =
qi(a) = 0 which gives x = τ(c) = 0. It follows that i(A) ∩ τ(C) = 0. Therefore B = i(A) ⊕ τ(C). We
take D = τ(C), so (2) follows.
(2) =⇒ (1) :
Let q0 denote the restriction of q to D. We show that q0 : D → C is an isomorphism. Let c ∈ C.
Since q is a surjective homomorphism, we get b ∈ B such that q(b) = c. Since B = i(A) ⊕ D, we
may write b as b = i(a) + d for some a ∈ A and d ∈ D. Now c = q(b) = q(i(a) + d) = qi(a) + q(d) =
q(d) = q0 (d). Therefore q0 is a surjective homomorphism.
Let d ∈ D be such that q0 (d) = q(d) = 0. Then d ∈ ker q ∩ D = i(A) ∩ D = 0. Therefore
q0 : D → C is an injective R-module homomorphism and hence an isomorphism. We define
τ : C → B to be j ◦ q−1
0 where j : D → B is the inclusion. Then q ◦ τ = q ◦ j ◦ q0 = q0 ◦ q0 = id.
−1 −1

Hence (1) follows. □

i q
Remark 19.6. The proof of the above proposition shows that if 0 → A → − B→ − C → 0 is a split
short exact sequence and τ : C → B satisfies q ◦ τ = id, then B = i(A) ⊕ τ(C).
41
 The reader is encouraged to prove the following.
i q
Proposition 19.7. Let E : 0 → A →
− B→
− C → 0 be a short exact sequence of R-modules. Then the
following are equivalent :
(1) there is an R-module homomorphism γ : B → A such that γ ◦ i = id,
(2) there is a submodule D of B such that B = i(A) ⊕ D,
(3) E is a split short exact sequence.
i q
Exercise 4. Consider the short exact sequence of Z modules : 0 → Z/2Z → − Z/4Z →− Z/2Z → 0
where i(n̄) = 2n̄ for n̄ ∈ Z/2Z; q(n̄) = 1 if n is odd and zero otherwise for n̄ ∈ Z/4Z. Show that it is
not a split short exact sequence.
i q
Lemma 19.8. Let 0 → A → − B→ − C → 0 be a short exact sequence such that C is a free R-module.
Then it is a split short exact sequence.
Proof. Let C be free on X. Since q is a surjective map, for each x ∈ X, we get a b x ∈ B such
that q(b x ) = x. The set map X → B given by x 7→ b x extends to an R-module homomorphism
τ : C → B, c.f. Theorem 17.11. We observe that q ◦ τ : C → C is an R-module homomorphism and
it is identity on the basis X of C. Therefore q ◦ τ = id which proves that the short exact sequence is
a split short exact sequence. □
Theorem 19.9. Let M be a finitely generated free module over a PID R and N be a nonzero
submodule of M. Then N is a free module and rank(N) ≤ rank(M).
Proof. We prove the result by induction on rank(M). If rank(M) = 1, then M is isomorphic to R
as an R-module. Since R is a PID, all ideals are principal. Nonzero submodules of R are nonzero
ideals generated by nonzero divisors. It follows that all nonzero submodules of R are free of rank
one, c.f. Lemma 18.5. Therefore all nonzero submodules of M are also free of rank one.
Now we assume that rank(M) = n. Let M be free on X = {x1 , . . . , xn }. Let M ′ = Rx1 +. . .+Rxn−1 .
Then M ′ is free on {x1 , x2 , . . . , xn−1 }. We have the following split short exact sequence
i pr
0 → M′ →
− M −→ R → 0.
Here i : M ′ → M is the inclusion map and pr : M → R is defined by pr(r1 x1 + . . . + rn xn ) = rn ,
ri ∈ R.
The module M ′ is a free module of rank n − 1. If N ⊂ M ′ , then by induction hypothesis N is a
free module and rank(N) ≤ rank(M ′ ) < rank(M). We assume that N 1 M ′ . Then pr(N) , 0. We
have the following short exact sequence
pr
′ i N
− N −−−→ pr(N) → 0.
0→N∩M →
Now pr(N) is a nonzero ideal of R. Since R is a PID, it is a free R module. Therefore the above
is a split short exact sequence. Let pr(N) = Rd and y ∈ N, y , 0 be such that pr(y) = d. Then
N = i(N ∩ M ′ ) ⊕ Ry. By induction hypothesis N ∩ M ′ is a free module of rank at most n − 1.
Therefore N is a free module and rank(N) = rank(N ∩ M ′ ) + 1 ≤ n. □
42
 20. 25th March 2019
All rings in this section are principal ideal domains unless otherwise specified. The following
theorem strengthens Theorem 19.9.

Theorem 20.1. Let R be a PID and M = Re1 ⊕. . . Ren be a free module of rank n. Let N be a nonzero
submodule of M. Then N is a free module and there exist a basis { f1 , . . . , fn } of M and nonzero
elements a1 , a2 , . . . , am ∈ R satisfying (a1 ) ⊃ (a2 ) ⊃ . . . ⊃ (am ) such that {a1 f1 , a2 f2 , . . . , am fm } is a
basis of N.

Proof. We prove the result by induction on rank(M). If rank(M) = 1, the proof is clear. Let
πi : M → R be the R-module homomorphism defined by πi (ei ) = 1 and πi (e j ) = 0 for i , j. We
consider the set of ideals
S = {ϕ(N) : ϕ ∈ Hom(M, R)}.

The ring R is Noetherian, so the set S has a maximal element. Let ψ(N) be a maximal element of
S for ψ ∈ Hom(M, R).
Claim - 1 : ψ(N) , 0
Since ψ(N) is a maximal element of S, no ideal belonging to S properly contains ψ(N). The zero
ideal is properly contained in every nonzero ideal. Therefore if we show that S contains a nonzero
ideal, then ψ(N) , 0. It is given that N is a nonzero module. Let x = c1 e1 + . . . + cn en be a nonzero
element of N. Now at least one coefficient ci must be nonzero. We have ci = πi (x) ∈ πi (N). This
shows that πi (N) is a nonzero ideal contained in S. Therefore ψ(N) , 0.
Since the ring R is a PID, we have a nonzero element a1 ∈ R such that ψ(N) = (a1 ). We choose
f ∈ N such that ψ( f ) = a1 . Clearly f , 0. Let f = b1 e1 + . . . + bn en .
Claim - 2 : The element a1 divides each coefficient bi .
Let (a1 , bi ) = (a′1 ), a′1 ∈ R. Then there exist p, q ∈ R such that a′1 = pa1 + qbi . We consider
the homomorphism pψ + qπi ∈ Hom(M, R). Then (pψ + qπi )( f ) = pa1 + qbi = a′1 . Therefore
a′1 ∈ (pψ + qπi )(N). It follows that ψ(N) = (a1 ) ⊂ (a′1 ) ⊂ (pψ + qπi )(N). The maximality of ψ(N)
forces all inclusions to be equality. Therefore (a1 ) = (a′1 ) = (a1 , bi ) which implies that a1 |bi .
Since each coefficient bi of f is divisible by a1 , we write bi = a1 hi , hi ∈ R for i = 1, . . . , n. Let
f1 = h1 e1 + . . . + hn en . Then f1 ∈ M and f = a1 f1 . We have a1 ψ( f1 ) = ψ(a1 f1 ) = ψ( f ) = a1 . Since
a1 , 0 and R is an integral domain, we get ψ( f1 ) = 1.
We have the following short exact sequence
i ψ
0 → ker ψ →
− M→
− R → 0.

The module R is a free on {1} and ψ( f1 ) = 1. Therefore the sequence is a split short exact sequence.
The R-module homomorphism τ : R → M defined by τ(r) = r f1 satisfies ψ ◦ τ = id. By Remark
19.6 we obtain M = ker ψ ⊕ R f1 . By Theorem 19.9, ker ψ is a free module of rank (n − 1).
We also have the following short exact sequence
i ψ
0 → ker ψ ∩ N →
− N→
− Ra1 → 0.
43
 The module Ra1 is free on {a1 } and ψ( f ) = ψ(a1 f1 ) = a1 . Therefore this sequence is also a split
short exact sequence and N = (ker ψ ∩ N) ⊕ Ra1 f1 (see Remark 19.6). If ker ψ ∩ N = 0, then
N = Ra1 f1 . Any basis of ker ψ together with f1 give a basis of M. The proof completes here.
We assume that ker ψ ∩ N , 0. The module (ker ψ ∩ N) is a nonzero submodule of the free
module ker ψ and rank(ker ψ) = n−1. By induction hypothesis we get a basis f2 , . . . , fn and nonzero
elements a2 , . . . , am satisfying (a2 ) ⊃ . . . ⊃ (am ) such that {a2 f2 , . . . , am fm } is a basis of ker ψ ∩ N. It
follows that M = ker ψ ⊕ R f1 = R f1 ⊕ . . . ⊕ R fn and N = (ker ψ ∩ N) ⊕ Ra1 f1 = Ra1 f1 ⊕ . . . ⊕ Ram fm .
We already have (a2 ) ⊃ . . . ⊃ (am ). We require to show that (a1 ) ⊃ (a2 ).
Let γ : M → R be defined by γ( fi ) = 1 for i = 1, 2 and 0 otherwise. Then γ(a1 f1 ) = a1 and
γ(a2 f2 ) = a2 . Since a1 f1 , a2 f2 ∈ N, we obtain (a1 , a2 ) ⊂ γ(N). It follows that ψ(N) = (a1 ) ⊂
(a1 , a2 ) ⊂ γ(N). The maximality of ψ(N) in S forces ψ(N) = γ(N), so (a1 ) = (a1 , a2 ). Therefore
(a1 ) ⊃ (a2 ). The proof completes here. □

Exercise 5. Let M be a module over a ring R and P, Q be submodules of M such that M = P ⊕ Q.

Let P′ , Q′ be submodules of P, Q respectively. Let M ′ = P′ + Q′ . Then show that M ′ = P′ ⊕ Q′ and
M/M ′  P/P′ ⊕ Q/Q′ .

Theorem 20.2. (Invariant factor form) Let M be a finitely generated module over a PID R. Then
M  Rr ⊕ R/(b1 ) ⊕ . . . ⊕ R/(bl )
such that R ⫌ (b1 ) ⊃ . . . ⊃ (bl ) ⫌ (0).

Proof. Since M is finitely generated module over R, we get a surjective R-module homomorphism
ϕ : Rn → M. Let K = ker ϕ. Then M  Rn /K. It follows from Theorem 20.1 that Rn = R f1 ⊕. . .⊕R fn
and K = Ra1 f1 ⊕ . . . ⊕ Ram fn such that (a1 ) ⊃ (a2 ) ⊃ . . . ⊃ (am ). Then
M  Rn /K  R/(a1 ) ⊕ . . . ⊕ R/(am ) ⊕ Rn−m .
If (ai ) = R, then R/(ai ) = 0 and if (ai ) = 0, then R/(ai ) = R. Let (a1 ) = . . . = (ar ) = R and
(ar+l+1 ) = . . . = (am ) = 0. Then we get
M = R/(ar+1 ) ⊕ . . . ⊕ R/(ar+l ) ⊕ Rm−r−l ⊕ Rn−m
= R/(ar+1 ) ⊕ . . . ⊕ R/(ar+l ) ⊕ Rn−r−l
= Rn−r−l ⊕ R/(ar+1 ) ⊕ . . . ⊕ R/(ar+l )
We observe that R ⫌ (ar+1 ) ⊃ . . . ⊃ (ar+l ) ⫌ (0). Therefore the result follows. □
Let p1 , p2 , . . . , pk be prime elements in a PID R such that no two distinct pi ’s are associate in R.
Let a = pl11 pl22 . . . plkk . By Chinese remainder theorem, it follows that as rings
R/(a)  R/(pl11 ) × . . . × R/(plkk ).
The above isomorphism is also an R-module homomorphism. Therefore we obtain an R-module
isomorphism
R/(a)  R/(pl11 ) ⊕ . . . ⊕ R/(plkk ).
The following result follows from Theorem 20.2 by the above observation.
44
Theorem 20.3. (Elementary divisor form) Let R be a PID and M be a finitely generated R-module.
Then there are prime elements p1 , . . . , pt such that
M  Rr ⊕ R/(pα1 ) ⊕ . . . ⊕ R/(pαt )
1 t

for some positive integers α1 , . . . , αt .

Definition 20.4. Let M be a module over an integral domain R. Then torsion of M is defined to be
the set
Tor(M) = {m ∈ M : rm = 0 for some r , 0}.
The module M is called torsion free if Tor(M) = 0. On the other hand M is called a torsion module
if M = Tor(M).
Show that in the setting of the above definition, Tor(M) is a submodule of M.
Exercise 6. Let M be a module over an integral domain R and N1 , N2 be submodules of M such
that M = N1 ⊕ N2 . Then show that Tor(M) = Tor(N1 ) ⊕ Tor(N2 ) and M/ Tor(M) is a torsion free
module.
Now further assume that M is a free R-module and I is an ideal of R. Prove that Tor(M/I M) =
M/I M if I , 0 and zero otherwise.
Corollary 20.5. Let M be a finitely generated torsion free module over a PID R. Then M is a free
R module.

21. 28th March 2019

We prove that decompositions in Theorems 20.2, 20.3. We begin with the following lemma.
Lemma 21.1. Let R be a PID and p be a prime element of R. Let
R/(pa1 ) ⊕ . . . R/(pam )  R/(pb1 ) ⊕ . . . ⊕ R/(pbn ).
Then m = n and ai = bσ(i) for i = 1, . . . , n and some permutation σ ∈ S n .
I do not have time to complete this section because I need to prepare the question paper. Please
read pages 466 - 468 of the book, i.e. Theorem 9, Corollary 10, Corollary 11. You must have a look
at solutions to all problems discussed in the classes. Do remember that this note does not contain
course material of the first few classes of this semester. Good luck.

45
 22. assignments
22.1. Assignment I. Please solve the following problems from the book : Abstract Algebra, Dum-
mit and Foote.
(1) Page no 230. Problems 3, 5, 6, 7, 13, 14, 17, 20, 26, 27.
(2) Page no 237. Problems 2, 3, 4, 5.
(3) Page no 247. Problems 1, 2, 3, 4, 17, 18, 24, 25, 26, 28, 29, 30, 32, 33, 34, 35, 36.
(4) Page no 256. Problems 2, 3, 5, 7, 8, 10, , 12, 13, 14, 19, 21, 25, 28, 33, 34, 36, 40.
(5) Page no 264. Problems 3, 4, 5.
(6) Page no 267. Problems 1, 3, 4, 5, 6.
22.2. Assignment II. Please solve the following problems from the book : Abstract Algebra,
Dummit and Foote.
(1) Page no 277. Problems 1(b), 2(a), 3, 4(a), 10, 11
(2) Page no 282. Problems 1, 2, 3, 5, 6, 8.
Problem 8 is slightly more difficult. One first shows that if I is an ideal of R, then
a
D−1 I = { : a ∈ I, b ∈ D}
b
is an ideal of D R. Then, one shows that every ideal of D−1 R is of the form D−1 I for some
−1

ideal I of R. From this one concludes that all ideals of D−1 R are principal.
(3) Page no 292. Problems 2, 4, 5, 8
(4) Page no 298. Problems 1, 3, 4, 5, 6, 7, 8, 9, 12, 13, 14, 15, 16, 17
In problem 8, show that the ideal (x, x2 y, x3 y2 , . . . , xn yn−1 , . . .) is not finitely generated.
In problem 13, show that (yF[x,y] F[x,y]
2 −x) is a PID but (y2 −x2 ) is a product of two fields.

(5) Page 301. 1, 2, 3, 4, 5, 7, 8,

(6) Page 306. Problems 1, 2, 3, 4.
In problem no 3, the ring is F[x2 , x3 ].
(7) Page no 311. Problems 1, 2, 3, 4, 7, 9, 10, 11, 13,16, 17.
22.3. Assignment III.
(1) Page no 343. Problems 1, 3, 5, 6, 7, 8, 9, 13, 14, 15, 18, 19, 20, 23
(2) Page no 350. Problems 1, 4, 6, 8, 9, 11, 12, 13, 14
(3) Page no 356. Problems 1, 2, 4, 5, 6, 7, 8, 12, 13, 14, 15, 16, 17, 18, 20, 21, 22, 23, 24*(diffi-
cult).
(4) Page no 468. Problems 7, 8, 9, 11, 12, 13, 16.
Email address: [email protected]

46