Section3 3

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

3.

3 Factor Rings
Suppose that R is a ring and that I is a (two-sided) ideal of R. Then we can use R and I to create a new ring, called the factor ring of R modulo I. This ring is denoted R/I (read R mod I), and its elements are certain subsets of R associated to I. The most well known examples are the rings Z/nZ, created from the ring Z of integers and its ideals. Denition 3.3.1 Let R be a ring and let I be a (two-sided) ideal of R. If a R, the coset of I in R determined by a is dened by a + I = {a + r : r I}. Thus a + I is a subset of R; it consists of all those elements of R that differ from a by an element of I. Note that a + I does not generally have algebraic structure in its own right, it is typically not closed under the addition or multiplication of R. We will show that the set of cosets of I in R is itself a ring, with addition and multiplication dened in terms of the operations of R. NOTES 1. a + I is a coset of the subgroup (I, +) of the additive group of R. 2. Suppose R = Z and I = 5 = 5Z. Then 2 + I = {2 + 5n, n Z} = {. . . , 3, 2, 7, 12, . . . }. This is the congruence class of 2 modulo 5. So in Z, the cosets of nZ in Z are the congruence classes modulo n - there is a nite number n of them and each has exactly one representative in the range 0, . . . , n 1 (this is guaranteed by the division algorithm in Z). 3. Let F be a eld and let I be an ideal in F[x]. Then I = f(x) for some polynomial f(x), by Lemma 3.2.3. If g(x) F[x] then the coset g(x) + I contains all those polynomials that differ from g(x) by a multiple of f(x). If F is innite then the number of cosets of I in F[x] is innite but each has exactly one representative of degree less than that of f(x). QUESTION
FOR THE

SEMINAR : Why is this?

If F is nite (e.g. F = Z/pZ for some prime p), then the number of cosets of I in F[x]is nite. Lemma 3.3.2 Let a and b be elements of a ring R in which I is a two-sided ideal. Then (i) If a b I, a + I = b + I. 30

(ii) If a b I, the cosets a + I and b + I are disjoint subsets of R. Proof: (i): Suppose a b I and let x a + I. Then x = a + m for some m I and we can write x = a b + b + m = b + (a b) + m. Since a b I and m I this means (a b) + m I and so x b + I. Thus a + I b + I. Now a b belongs to I and so b a = (a b) does also. It then follows from the above argument that b + I a + I. Thus a + I = b + I. (ii) Suppose a b I and let c (a + I) (b + I). Then c = a + m1 = b + m2 where m1 , m2 I. It follows that a b = m2 m1 which is a contradiction since a b I. Lemma 3.3.2 shows that the different cosets of I in R are disjoint subsets of R. We note that their union is all of R since every element a of R belongs to some coset of I in R : a a+ I. The set of cosets of I in R is denoted R/I. We can dene addition and multiplication in R/I as follows. Let a + I, b + I be cosets of I in R. We dene their sum by (a + I) + (b + I) = (a + b) + I. Claim: This addition is well-dened. QUESTION FOR THE SEMINAR: What is this claim saying? Why is there doubt about the denition of addition given above? What the claim is concerned with is the following : if a + I = a1 + I and b + I = b1 + I, how do we know that (a + b) + I = (a1 + b1 ) + I? How do we know that the coset sum (a + I) + (b + I) as dened above does not depend on the choice a and b of representatives of these cosets to be added in R? PROOF
OF

CLAIM : Suppose a + I = a1 + I and b + I = b1 + I

for elements a1 , b1 of R. Then a a1 I and b b1 I, by Lemma 3.3.2. Hence (a a1 ) + (b b1 ) = (a + b) (a1 + b1 ) belongs to I. Thus (a + b) + I = (a1 + b1 ) + I, by Lemma 3.3.2 again. 31

Multiplication in R/I is dened by (a + I)(b + I) = ab + I for cosets a + I and b + I of I in R. Claim: Multiplication is well-dened in R/I (i.e. the coset ab + I does not depend on the choice of representatives of a+ I and b + I). PROOF
OF

CLAIM : Suppose that a + I = a1 + I and b + I = b1 + I

for elements a1 , b1 of R. Then a a1 I and b b1 I, by Lemma 3.3.2. We need to show that ab + I = a1 b1 + I. By Lemma 3.3.2, this means showing that ab a1 b1 I. To see this observe that ab a1 b1 = ab a1 b + a1 b a1 b1 = (a a1 )b + a1 (b b1 ). Now since I is a two-sided ideal we know that (a a1 )b I and a(b b1 ) I. Thus (a a1 )b + a1 (b b1 ) = ab a1 b1 I, and this proves the claim. That addition and multiplication in R/I satisfy the ring axioms follows easily from the fact that these axioms are satised in R. The ring R/I, with addition and multiplication dened as above, is called the factor ring R modulo I. NOTES: 1. The zero element of R/I is the coset 0R + I = I. 2. It is clear that R/I has some properties in common with R. For example R/I is commutative if R is commutative. If R contains an identity element 1R for multiplication, then 1R + I is an identity element for multiplication in R/I If u is a unit in R with inverse u1 , then u + I is a unit in R/I, with inverse u1 + I. 3. However, R/I can be structurally quite different from R. For example, R/I can contain zero-divisors, even if R does not. It is also possible for R/I to be a eld if R is not. QUESTION FOR THE SEMINAR : Find examples of both of these phenomena. 32

In the next section we will look at conditions on I under which R/I is an integral domain or a eld, for a commutative ring R. Our nal goal in this section is to prove the Fundamental Homomorphism Theorem for rings, which states that if : R S is a ring homomorphism, then the image of is basically a copy of the factor ring R/ ker . Denition 3.3.3 Let : R S be a ring homomorphism. Then is called an isomorphism if 1. is surjective (onto); i.e. Im = S, and 2. is injective (one-to-one) i.e. (r1 ) = (r2 ) whenever r1 = r2 in R. NOTE: is injective if and only if ker is the zero ideal of R. To see this rst suppose is injective. Then ker = {0R }, otherwise if r ker for some r = 0 we would have (r) = (0R ), contrary to the injectivity of . On the other hand suppose ker = {0R }. Then if there exist elements r1 and r2 of R with (r1 ) = (r2 ) we must have (r1 r2 ) = (r1 ) (r2 ) = 0S . This means r1 r2 ker , so r1 r2 = 0R and is injective. The characterisation of injectivity in the above note can be very useful. If : R S is an isomorphism, then S is an exact copy of R. This means that S and R are structurally identical, and only differ in the way their elements are labelled. We say that R and S are isomorphic and write R = S. Theorem 3.3.4 (The Fundamental Homomorphism Theorem) Let : R S be a homomorphism of rings. Then the image of is isomorphic to the factor ring R/ ker . Proof: Let I denote the kernel of , so I is a two-sided ideal of R. Dene a function : R/I Im by (a + I) = (a) for a R. 1. is well-dened (i.e. the image of a + I does not depend on a choice of coset representative). Suppose that a + I = a1 + I for some a, a1 R. Then a a1 I by Lemma 3.3.2. Hence (a a1 ) = 0S = (a) (a1 ). Thus (a) = (a1 ) as required. 2. is a ring homomorphism. Suppose a + I, b + I are elements of R/I. Then ((a + I) + (b + I)) = = = = 33 ((a + b) + I) (a + b) (a) + (b) (a + I) + (b + I).

So is additive. Also ((a + I)(b + I)) = = = = (ab + I) (ab) (a)(b) (a + I)(b + I).

So is multiplicative - is a ring homomorphism. 3. is injective. Suppose a + I ker . Then (a + I) = 0S so (a) = 0S . This means a ker , so a I. Then a + I = I = 0R + I, a + I is the zero element of R/I. Thus ker contains only the zero element of R/I. 4. is surjective. Let s Im. Then s = (r) for some r R. Thus s = (r + I) and every element of Im is the image under of some coset of I in R. Thus : R/ ker Im is a ring isomorphism, and Im is isomorphic to the factor ring R/ ker .

34

You might also like