Their sum and product are then
a
+ fi = r  2r2  2s + rs afi = (r + r2  2s)( 3r2 +
rs)
= r(3r2 + rs) + r2(_3r2 + rs)  2s(3r2 + rs) = 3r3 + r2s  3 + r3s + 6r2s  2r3 = 3  5r3 + 7r2s + r3s.
The ring R appears in R G as the "constant" formal sums i.e., the R multiples of the identity of G (note that the definition of the addition and multiplication in RG restricted to these elements is just the addition and multiplication in R). These elements of R commute with all elements of RG. The identity of R is the identity of RG. The group G also appears in RG (the element g¡ appears as 19¡  for example, r, s E Dg are also elements of the group ring ZDg above)  multiplication in the ring RG restricted to Gis just the group operation. In particular, each element of G has a multiplicative inverse in the ring RG (namely, its inverse in G). This says that G is a subgroup ofthe group ofunits of RG. If IG I > 1 then R G always has zero divisors. For example, let g be any element of G of order m > 1. Then
(1  g)(1
+ g + ... + gml) = 1 
gm = 1  1 = O
so 1  g is a zero divisor (note that by definition of RG neither of the formal sums in the above product is zero). If S is a subring.of R then SG is a subring of RG. For instance, ZG (called the integral group ringof G) is asubring ofQG (the rational group ring of G). Furthermore, if H is a subgroup of G then R H is a subring of RG. The set of all elements of RG whose coefficients sum to zero is a subring (without identity). If IGI > 1, the set of elements with zero "constant term" (i.e., the coefficient of the identity of G is zero) is not a subring (it is not closed under multiplication). Note that the group ring IRQg is not the same ring as the Hamilton Quaternions IHI even though the latter contains acopy ofthe quaternion group Qg (under multiplication). One difference is that the unique element of order 2 in Qg (usually denoted by 1 is not ) the additive inverse of 1 in IRQg. In other words, if we temporarily denote the identity of the group Qg by gl and the unique element of order 2 by g2, then gt + g2 is not zero in IRQg, whereas 1 + (1) is zero in IHI.Furthermore, as noted above, the group ring IRQg contains zero divisors hence is not a division ringo Group rings over fields will be studied extensively in Chapter 18. EXERCISES
Let R be a commutative ring with l.
= 2x3  3x2 + 4x· 5 and let q(x) = 7x3 + 33x  4. In each ofparts (a), (b) and (e) compute p(x) + q(x) and p(x)q(x) under the assumption that the coefficients of the two given polynomials are taken from the specified ring (where the integer coefficients R , Sec. 7.2 P olyno m ial in gs.M atrixR ingsan d G ro upR in gs 237
1. Let p(x)
are taken mod n in parts (b) and (e) ): (a) R = z, (b) R = 7l./27l., (e) R = 7l./37l..
Sec. 7.2
P olyno m ial in gs.M atrixR ingsan d G ro upR in gs R ,
237
2. Let p(x)
+ao be anelementofthepolynomialring R[x]. Prove that p(x) is a zero divisor in R[x] if and only if there is a nonzero b E R such that bp(x) = O . [Letg(x) = bmxm+bm_lXml+ .. ·+bo be anonzeropolynomialofminimal degree such that g(x)p(x) = O . Show that bman = O and so ang(x) is a polynomial of degree less than m that also gives O when multiplied by p(x). Conc1ude that ang(x) = O .Apply a similar argument to show by induction on i that ani g(x) = O for i = O . 1, ...• n, and show that this implies bmP(x) = O.] +an_lXn1
= anxn
+ ... +alX
3. Define the set R[[x]] offormal power series in the indeterminate x with coefficients from
R to be all formal infinite sums
0 0
¿anXn
n=O
= ao +alx
+a2x2 +a3x3
+ ....
Define addition and multiplication of power series in the same way as for power series with real or complex coefficients i.e., extend polynomial addition and multiplication to power series as though they were "polynomials of infinite degree":
0 0
¿anXn
n=Ü
+¿
n=Ü
0 0
0 0
0 0
bnxn = Lean bn)xn
n=O
0 0
+
n=Ü
r>
0 0
x
¿ bnxn = ¿ (¿ akbnk )xn.
n=O
n=Ü
n
k=O
(The term "formal" is used here to indicate that convergence is not considered, so that formal power series need not represent functions on R.) (a) Prove that R[[x]] is a cornmutative ring with 1. (b) Show that 1  x is a unit in R[[x]] with inverse 1 + x + x2 + .... (e) Prove that anxn is a unit in R[[x]] if and only if ao is a unit in R.
L~
4. Prove that if R is an integral domain then the ring of formal power series R[[x]] is also an
integral domain. 5. Let F be a field and define the ring F «x» of formal Laurent series with coefficients from
Fby F«x»
= {¿
n~N
0 0
anxn
I an
E
F and N E Z).
(Every element of F «x» is a power series in x plus a polynomial in l/x, i.e., each e1ement of F «x» has only a finite number of terms with negative powers of x.) (a) Prove that F«x» is a fie1d. (b) Define the map
0 0
v : F«x»x
+
Z
by
v(¿
n~N
anxn) = N
where a N is the first nonzero coefficient of the series (i.e., N is the "order of zero or pole of the series at O"). Prove that v is a discrete valuation on F «x» whose discrete valuation ring is F[[x]], the ring of formal power series (cf. Exercise 26, Section 1). 6. Let S be a ring with identity 1 = 1 =O Let n E Z+ and let A be an n x n matrix with entries . from S whose i, j entry is aij. Let Eij be the element of Mn (S) whose i, j entry is 1 and whose other entries are all O.
238
Chapo7
Introduction to Rings
(a) Prove that Eij A is the matrix whose ith row equals the jth row of A and all other rows are zero. (b) Prove that AEij is the matrix whose jth column equals the ith column of A and all other columns are zero. (e) Deduce that EpqAErs is the matrix whose p, s entry is aqr and all other entries are zero. 7. Prove that thecenterofthering Mn(R) is the setofscalarmatrices (cf. Exercise 7, Section 1). [Use the preceding exercise.] 8. Let S be any ring and let n 2: 2 be an integer. Prove that if A is any strictly upper triangular matrix in M¿ (S) then A" = O (a strictly upper triangular matrix is one whose entries on and below the main diagonal are all zero). 9. Let a
= r + r2
(b) a2,

2s and {J = 3r2
+ rs be the two elements of the integral group
ring ZD8 described in this section. Compute the following elements of ZD8:
(a) {Ja, (e) a{J  {Ja, (d) (Ja{J.
10. Consider the following elements of the integral group ring ZS3:
a = 3(1 2)  5(2 3) + 14(1 2 3)
(a) a
and
(J = 6(1) (e) a2.
+ 2(2 3) 7(1
3
2) (where (1) is the identity of S3). Compute the following elements:
+ {J,
(b) 2a  3{J,
E
(e) a{J,
(d) {Ja,
11. Repeat the preceding exercise under the assumption that the coefficients of a and {J are in
Z/3Z (i.e., a, (J Z/3ZS3). 12. Let G = Isr. ... , gn} be a finite group. Prove that the element N is in the center ofthe group ring RG (cf. Exercise 7, Section 1).
= g¡ + s x + ...+ gn
13. Let K = {kl, ... , km} be a conjugacy class in the finite group G.
(a) Prove that the element K = kI + ... + km is in the center of the group ring RG (cf. Exercise 7, Section 1). [Check that g¡ Kg = K for all g E G.] (b) Let K¡, ... , K; be the conjugacy c1asses of G and for each K¡ let K¡ be the element of RG that is the sum of the members of K¡. Prove that an e1ement a E RG is in the renterof RGifandonlyifa = a¡Kl +a2K2+"'+GrKr forsomea¡, a2, ... , a; E R.
7.3 RING HOMOMORPHISMS
ANO QUOTIENT RINGS
A ring homomorphism is a map from one ring to another that respects the additive and multiplicative structures: Definition. Let R and S be rings. (1) A ring homomorphism is a map r p : R ~ S satisfying (i) rp(a + b) = rp(a) + rp(b) for a11a, b E R (so rp is a group homomor phism on the additive groups) and (ii) rp(ab) = rp(a)rp(b) for all a. b E R. (2) The kernel of the ring homomorphism rp , denoted ker rp , is the set of elements of R that map to O in S (i.e., the kernel of rp viewed as a homomorphism of
240
Chap.7
Introduction to Rings
Sec. (3) A bijective ring homomorphism is called an isomorphism.additive groups).3 Ring Homomorphismsand Quotient Rings 239 .7.
(a ) = rp (fJ) = O.is closed under multiplication by elements from R." Similarly. S 2E im q :. Let R and S be rings and let rp : R ~ S be a homomorphism.. Furthermore.. so ra . n = O.(a . ..This shows SI . Examples (1) The map e p : Z .(r) = O .S IS 2 E im q :. rz E R. The fiber above a E Q consists of the set of polynomials with a as constant termo In particular. (1) The image of q:.S 2 and q :..+ Z defined by epn(x) = nx are not in general ring horno morphisms because epn(xy) = nxy whereas epn(x)epn (y) = nxny = n2xy..+ Q be the map from the ring of polynomials in x with rational coefficients to the rationals defined by ep(p(x» = p(O) (i. Since R and S are in 240 Chap. Let q :..If the context is c1ear we shall simply use the term "homomorphism" instead of "ring homomorphism. mapping the polynomial to its constant term).. (2) For n E Z the maps epn : Z .+ Z /2Z defined by sending an even integer to O and an odd integer to 1is a ring homomorphism. (2) The kernel of q :. A rv B will always mean an isomorphism of rings unless otherwise stated . the kernel of epconsists of the polynomials with constant term O . Then q :. The kernel of e p (the fiber of tp above O E Z/2Z) is the set of even integers..is closed under subtraction and under multiplication. hence is a subring of ker rp then q :.(r) = O q :. P ro o f: (1) If S I. if a E ker e then ra and a r E ker q:. (2) If a .e. An analogous result is true for a homomorphism of rings. O.. for any r E R we have q :.for every r E R. Thus care should be exercised when dealing with rings to be sure to check that both ring operations are preserved. i.(rl . is closed under subtraction and under multiplication.(r)q :. a r E ker e.e.(r¡) and S 2 = q :.fJ ) = O and q :. is a subring of R. if A and B are rings. fJ E = = In the case of a homomorphism q :. The fiber of e p above 1 E Z/2Z is the set of odd integers. so ker q :.(a r) = q :.(r) O q :. Proposition 5.(a )q :.. so the image of q:. Hence epn is a ring homomorphism only when n2 = n. which led to the notion of a quotient group by a normal subgroup. ker q:.S 2 ..(a ) q :. and also Similarly.e. (3) Let ep : Q[x] .(a fJ ) = O. of groups we saw that the fibers of the homo morphism have the structure of a group naturally isomorphic to the image of tp. so is a subring of R. l.. The map is additive since the sum of two even or odd integers is even and the sum of an even integer and an odd integer is odd. Note that e p ois the zero homomorphism and e p I is the identity homomorphism.r2 ) = S I . i. The map is multiplicative since the product of two odd integers is odd and the product of an even integer with any integer is even.7 Introduction to Rings .(r2 ) for sorne rl. S.is a subring of S..(ra ) = q :.(rtr2 ) = S IS 2 . Then e p is a ring homomorphism since the constant term of the sum of two polynomials is the sum of their constant terms and the constant term of the product of two polynomials is the product of their constant terms. : R ~ S be a ring homomorphism with kernel l. Hence q :. Note however that epn is always a group homomorphism on the additive groups. then S I = q :.
3 Ring Homomorphismsand Quotient Rings 239 .particular additive abelian groups.is in particular a homomorphism of abelian groups Sec.7. q:.
2) As in the case for groups.. We consider when the multiplication of cosets in (2) is well defined and makes the additive abelian group RJ l into a ringo The statement that the multiplication in (2) is well defined is the statement that the multiplication is independent of the particular representatives r and s chosen. Letting r = O and letting s be arbitrary. then the fiber of (/)over a is the coset r + l of the kernel 1). Conversely. s E R and all a. Note that the additive structure of the ring RJ l is just the additive quotient group of the additive abelian group R by the (necessarily normal) subgroup l. i. that l must be closed under multiplication on the left by elements from R. we see that we must have rf3 E l for every r E R and every f3 E I.and the fibers of (/)are the additive cosets r + l of the kernel l (more precisely. the subgroup l is necessarily normal so that the quotient RJ l of cosets of l is automatically an additive abelian group. f 3 El. by letting s = O and letting r be arbitrary. Note that since R is an abelian additive group. if r is any element of R mapping to a E S.e. f3 E l. which leads to the notion of an ideal in R (the analogue for rings of a normal subgroup of a group). The question then is whether this quotient group also has a multiplicative structure induced from the multiplication in R. The answer is no in general (just as the answer is no in trying to form the quotient by an arbitrary subgroup of a group).1) l (r + l) x (s + l) = (rs) + l. we see that l must be closed under multiplication. i. we must have (r + a)(s + f3 ) + l l = rs + for all r. When l is the kernel of sorne homomorphism (/)this additive abelian quotient group also has a multiplicative structure. we can also consider whether (1) and (2) can be used to define a ring structure on the collection of cosets of an arbitrary subgroup l of R. Hence this is a necessary and sufficient condition for the multiplication in (2) to be well defined. defined by (7. which makes RJ l into a ringo As in the case for groups. This ring of cosets is called the quotient ring of R by l = ker (/)and is denoted RJ l. then X + Y is the fiber over a + b and X Y is the fiber over ab. These fibers have the structure of a ring naturally isomorphic to the image of (/): if X is the fiber over a E S and y is the fiber over b ES. . In other words. the verification that these operations define a ring structure on the collection of cosets of the kernel l ultimately rests on the corresponding ring properties of S. we see similarly that l must be closed under multiplication on the right by elements from R. In terms of cosets of the kernel l this addition and multiplication is (r + l) + (s + l) = (r + s) + (7.. Let l be an arbitrary subgroup of the additive group R. that we obtain the same coset on the right if instead we use the representatives r + a and s + f3 for any a.e. i. We shall then see that the ideals of R are exactly the kernels of the ring homomorphisms of R (the analogue for rings of the characterization of normal subgroups as the kernels of group homomorphisms). (7. Letting r = s = O . (/)(r) = a.. l must be a subring of R. defined by (2). Next. if l is closed under multiplication on the left and on the right by elements from R then the relation (*) is satisfied for all a.e.2). f 3 El.
7.3 Ring Homomorphisms and Quotient Rings 241 .Sec.
Similarly 1 is a right ideal if (i) holds and in place of (ii) one has (ii)' 1 is closed under rigbt multiplication by elements from R. Let R be a ring and let 1 be an ideal of R. a twosided ideal) of R. one of the distributive laws is verified as follows: (r + I)[(s + 1) + (t + 1)] = (r + I)[(s + t) + 1] = r(s + t) + 1 = (rs + rt) + 1 = (r s + 1) + (rt + 1) = [(r + I)(s + 1)] + [(r + I)(t + 1)]. If R has a 1 then (1)a = a so in this case 1 is an ideal if it is nonempty. s R. We emphasize that to prove a subset 1 of a ring R is an ideal it is necessary to prove that 1 is nonempty. Proposition 6. As mentioned. Note also tbat the last part of Proposition 5 proves that the kernel of any ring homomorphism is an ideal.Finally. Conversely. i. (l)rl={ralaE/} and Ir={arlaE/}. for added emphasis. We summarize the preceding discussion in the following proposition. let 1 be a subset of R and let r E R. and (ii) 1 is closed under left multiplication by elements from R. closed under addition and closed under multiplication by all the elements of R. Ir e 1 for all r E R. This shows that the quotient R /1 of the ring R by a subgroup 1 has a natural ring structure if and only if 1is also closed under multiplication on the left and on the rigbt by elements from R (so in particular must be a subring of R since it is closed under multiplication). (3) A subset 1 that is both a left ideal and a rigbt ideal is called an ideal (or. i.7 Introduction to Rings 242 .. if the multiplication of cosets defined by (2) is well defined. then this multiplication makes the additive quotient group R/I into a ringo Each ring axiom in the quotient foIlows directIy from the corresponding axiom in R. (2) A subset 1 of R is a left ideal of R if (i) 1 is a subring of R. if 1 is any subgroup such that the above operations are Chap. For commutative rings the notions of left.e.e. For example. rigbt and twosided ideal coincide. r 1 e 1 for allr E R. such subrings 1 are called the ideals of R: Definition.. Then the (additive) quotient group R/I is a ring under the binary operations: (r + 1) + (s + 1) = (r + s) + 1 +1 E and (r + 1) x (s + 1) = (r s) for all r. Let R be a ring. closed under subtraction and closed under multiplication by all the elements of R (and not just by elements of 1).
well defined.7 Introduction to Rings . then 1 is an ideal of R. 242 Chap.
add (multiply. then the map R ~ Rjl defined by r 1+ r +1 is a surjective ring homomorphism with kernel 1 (this homomorphism is called the natural projection of R onto R] 1).(s) = q. (2) It is irnmediate that n71is an ideal of 7l for any n E Z and these are the only ideals of 7l since in particular these are the only subgroups of 7l. If 1 is the kernel of s p . the image of s p is a subring of S and R] ker s p is isomorphic as a ring to fP(R). S ec. Por example. The ideal {O}is called the trivial ideal and is denoted by O. 7 TI = 77 = " 2 in 7l/1571.3 RingHomomorphismsand Quotient Rings 243 . If 1 is any ideal.(R) which respects addition and multiplication. respectively). (2) If 1 is any ideal of R. then R j 1 is a ring (in particular is an abelian group) and the map n : r t+ r + 1 is a group homomorphism with kernel l. 14. Thus every ideal is the kernel of a ring homomorphism and vice versa. 7 + TI = 18 and 18 = 3. We could also express this by writing 7 + 11 = = 3 mod 15 . (1) (The First Isomorphism Theoremfor Rings) If fP : R ~ S is a homomorphism ofrings. Tbeorem 7. An ideal 1 is proper if 1 = 1 R.1.(r)q.(r) is a bijection between the rings R] 1 and q.(r). then the cosets (under addition) of 1 are precisely the fibers of fP . the cosets r + 1.7. In particular. introduced in Chapter O. It remains to check that n is a ring homomorphism. if n = 15 then the elements of 7l/1571 are thecosets 0. hence (r + 1)(s + 1) = rs + l. Since s p is a ring homomorphism q. With this notation the addition and multiplication in the quotient ring R j 1 become simply r + s = r + s and rs = rs.. When 1 is an ideal of R the ring R j 1 with the operations in the previous proposition is called the quotient ring of R by l. . so 7 + TI = 3 in 7l/1571. s + 1 and r s + 1 are the fibers of q.(rs). The natural projection 7l ~ 7l/ n71is called reduction mod n and wiIl be discussed further at the end of these examples.Definition. over q. and take the corresponding coset containing this sum (product.. respectively. fP(s) and fP(rs). To add (or rnultiply) in thequotient. hence is a ring isomorphism. respectively) these representatives in the integers 7l. 13. Por example. Similarly. Multiplication of cosets is well defined and so 1 is an ideal and R [I is a ringo The correspondence r + 1 t+ q. 7(11) = 2 mod 15 . Proof: This is just a matter of collecting previous calculations. then the kernel of s p is an ideal of R. The associated quotient ring is 7l/n71 (which explains the choice ofnotation and which we have now proved is a ring). This is immediate from the definition of multiplication in R [I: tt : rs t+ rs +1= (r + 1)(s rr(r):rr(s). simplychooseany representatives foc the two cosets. Examples Let R be a ringo (1) The subrings R and {O}are ideals. + 1) = As with groups we shall often use the bar notation for reduction mod 1: r = r + l. .
7 Introduction to Rings . i. Similarly. Two polynomials p(x). homomorphism from R to another ring Sto obtain a ring homomorphism from R[x] to S. Let 1 be the collection of polynomials whose terms are of degree at least 2 (i. the set of a11polynomials with integer coefficients whose constant tenn is even. The kernel of E¿ is given by {f E R I f(e) = O}(the set of functions from X to A that vanish at e).•having no terms of degree O or degree 1) together with the zero polynomial. More generally. follows easily that a complete set of representatives for the quotient R] 1 is given by the polynomials a + bx of degree at most 1. For each e E [O. Also. For example.e.. . Then 1 is an ideal: the sum of two such polynomials again has terms of degree at least 2 and the product of a polynomial whose terms are of degree at least 2 with any polynomial again only has terms of degree at least 2. For example. yo).. (6) Fix sorne n E Z with n ~ 2 and consider the noncornmutative ring Mn(R). E¿ is surjective: given anya E A the constant function f (x) = a maps to a under evaluation ate. If J is any ideal of R then Mn(J). the quotient ring is Zj2Z.e. Addition and multiplication in the quotient are again performed by representatives. The other fiber of this homomorphism is the coset of polynomials whose constant tenn is odd.x4 are in the same coset of l It .7x + 15x2) = 4 so that (4) Let A be a ring. For example. as we detennined earlier.1] in IR and let R be the ring of a11 continuous real valued functions on [0. let X be the closed interval [0. q (x) are in the same coset of 1 if and only if they differ by a polynomial whose terms are of degree at least 2. (1 + 3x) + ( 4 + 5x) = 3 + and ( 1+ 3x )( 4 8x = Z[x] + 5x) = (4 7x .The kernel of E¿ is the ideal of a11continuous functions whose graph crosses the x axis at e. let R = Z and consider the homomorphism Z[x] ~ Zj2Z defined by the composition p(x) t+ p(O) t+ p(O) mod 2 E Zj2Z.1]. Thus Rjker Ec""'" A. = O . 1].for example. For each fixed e E X the map E¿ : R ~ A Note that in this quotient ring Rj 1 we have x x = x2 Rj 1 has zero divisors. the polynomials 3 + 5x + x3 + x5 and 3 + 5x . i. the fiber of E¿ aboye the real number Yo is the set of a11continuous functions that pass through the point (e. defined by Ec(f) = f(e) (ca11edevaluation at e) is a ring homomorphism because the operations in R are pointwise addition and multiplication of functions. is a 244 Chap. let X be any nonempty set and let R be the ring of a11functions from X to A. i. p (O) = O We can compose this homomorphism with any . if and only if p(x) and q (x) have the same constant and first degree terms. (5) Themap fromthepolynomialring R[x] to R defined by p(x) t+ p(O) (evaluationatO) is aring homomorphism whose kernel is the set of a11polynomials whose constant term is zero. evaluation at e is a surjective ring homomorphism (since R contains the constant functions) and so R] ker E¿ """IR.e .e. even though R = Z[x] does not. the n x n matrices whose entries come from J. Since the homomorphism is clearly surjective. The kernel ofthis composite map is given by {p(x) E Z[x] I p(O) E 2Z}.(3) Let R be the ring of polynomials in x with integer coefficients..
For instance..7 Introduction to Rings . the 3 x 3 matrices whose entries are a11even is the twosided ideal M3(2Z) 244 Chap..+ Mn(Rj J) which reduces each entry of a matrix mod J.e. which maps each entry aij to aij (here bar denotes passage to R] J).. when n = 3 and R = Z.twosided ideal of Mn(R). i. This ideal is the kernel of the surjective homomorphism Mn(R) .
The map from the group ring RG to R defined by L ? = l gi 1+ L ? = l is easily seen to a. The choice n = 4 is particularly efficacious. Moreover. Then Elj E L¡ but EljEji = El. 1 (mod 4). Reading the aboye equation mod 4 (that is. L¡ if i = f ..gj is an element of the augmentation ideal for all i. considering this equation in the quotient ring Zj4Z).. 2. Cheeking the few possibilities shows that we must take the O each time. To gn). Example: (Tbe Reduction Homomorpbism) The canonical projeetion map from Z to ZjnZ obtained by faetoring out by the ideal nZ of Z is usualIy referred lo as "reducing modulo n:" The faet that this is a ring homomorphism has important consequences for elementary number theory.. This equation simply states a relation between these elements in the ring Z. . y and z (such problems are frequent1y referred to as Diophantine equations after Diophantus. who was one of the first to systematically examine the existence of integer solutions of equations).. This shows L¡ is a left ideal of M. the quotient ring is isomorpbic to R. for example. for the following reason: the squares mod 4 are just 02. n}). since otherwise we could divide through this equation by the square of this common factor and obtain another set of integer solutions smaller than the initial ones. the augmentation ideal. so Lj is not closed under right multiplication by arbitrary ring elements.. this relation must hold in ZjnZ for any integer n. . . g¡ .. As such. (R). is the set of elements of RG whose coefficients sum to O . the formal sums whose coeffi cients are all equal (equivalent1y. For example.e. We exhibit sorne onesided ideals in the ring M. L¡ is not a right ideal (hence is not a twosided ideal). Suppose such integers existo Observe first that we may assume x. l Another ideal in RG is {L ~ =ag ¡ I a E R}. .O and let n E Z with n ~ 2. For example. q E {l. y and z have no faetors in common. 12•22. we must have { ~} + { ~} = 3 { ~ } = 4) {~} (mod where the { ~ }. + (8) Let R be a commutative ring with identity 1 = f . i. see this. For each j E {l. all Rmultiples of the element gl + . In particular. It is clear that L¡ is closed under subtraction. (R). The kernel of the augmentation map.. gn} be a finite group. i. then R¡ is a right ideal which is not a left ideal. suppose we are trying to solve the equation X2 + y2 3z2 = in integers x. the same relation must also hold in any quotient ring as well. Since the augmentation map is surjective. let E pq be the matrix with 1 in the pth row and q th column and zeros elsewhere (p. fj.• n} let Lj be the set of all n x n matrices in Mn (R) with arbitrary entries in the jth column and zeros in all other columns..of M3{Z) and the quotient M3{Z)jM3{2Z) is isomorpbic to M3{Zj2Z). O . (7) Let R be a commutative ring with L and let G = {gl.. . indicates that either a O or a 1 may be taken. i.i. It follows direet1y from the definition of matrix multiplication that for any matrix T E M¿ (R) and any A E L¡ the product TAhas zero entries in the ith column for all i = f . Ifthe ring R has an identity then the exercises below show that every twosided ideal of Mn{R) is of the form M¿ (J) for sorne twosided ideal J of R. 32.e.. An analogous argument shows that if R¡ is the set of all n x n matrices in Mn(R) with arbitrary entries in the jth row and zems in all other rows. These onesided ideals will play an important role in Part VI. This means that each . a¡ be a homomorphism.i. called the augmentation map .
3 Ring Homomorphism sand Quotient Rings 245 .7.Sec.
. b E Bl is a subring of R.e. We also observe that this technique has a number of limitations. Tbeorem 8. (3) (The Fourth or Lattice Isomorphism Theoremfor Rings) Let 1be an ideal of R. Togetherwith the Chinese RernainderTheorem (describedin Section 6).but which do not have integer solutions. in the case of the Fourth Isomorphism Theorem) and then check that this group iso morphism (or correspondence. y and z must be evenintegers (squaresof the odd integersgaveus 1 mod 4). Let R be a ringo (1) (The Second 1somorphism Theoremfor Rings) Let A be a subring and let B be an ideal of R.7 Introduction to Rings .for example.there are equationswhich have solutions modulo every integer. and rirz + 1H riri + J = (rl + J)(r2 + J) by the definition ofthe multiplication in the quotient ring R/ J. This map is multiplicative since (rI + 1)(r2 + 1) = riri + 1by the definition ofthe multiplication in the quotient ring R/I. this technique gives informationabout the possible residues ofthe solutionsmod n (since we couldjust as well have examined the possibilities mod n as mod 4) and note that for each choice of n we have only a finite problem to solve because there are only finitely many residue c1assesmod n. we mention that the map from the ring Z[x] of polynomialswith integercoefficientsto the ring Zj pZ[x] of polynomialswith coefficientsin Zj pZ for a prime p given by reducing the coefficients modulo p is a ring homomorphism.But this contradictsthe assumptionof no common factors for these integers. and shows that this equationhas no solutions in nonzero integers.An easy example (but extremelyhard to verify that it does indeed have this property) is the equation of x. Note that even had solutions existed. In each case the verification is irnmediate from the definition of multiplication in quotient rings. Por example. we can then determinethe possible solutionsmoduloverylarge integers. A n B is an ideal of A and (A + B)/ B "'" A/(A n B). The proofs for the other parts of the theorem are similar. respectively) is a multiplicative map. and so defines a ring isomorphism.This exampleof reductionwiIlbe used in Chapter9 in trying to determinewhetherpolynomialscan be factored The following theorem gives the remaining Isomorphism Theorems for rings. Then A + B = {a + b I a E A. (2) (The Third Isomorphism Theorem for Rings) Let 1 and J be ideals of R with 1e J.whichgreat1yassists in finding them numerica1ly(when they exist). As a fina] example of this technique. the map that gives the isomorphism in (2) below is defined by lp : r + 1 t+ r + J. lp(r¡r2) = lp(r¡)lp(r2). Each of these may be proved as follows: first use the corresponding theorem from group theory to obtain an isomorphism of additive groups (or correspondence of groups. i. The correspondence A ~ A/I is an inc1usion preserving bijection between the 246 Chap. Then J /1 is an ideal of R/I and (R/ 1)/(1/1) "'"R/ J.
246 Chap. Furthermore.7 Introduction to Rings .set of subrings A of R that contain 1 and the set of subrings of R / l. A (a subring containing 1) is an ideal of R if and only if A / 1 is an ideal of R / 1.
. the exercises). b E J} consisting just of products of elements from I and J is in general not closed under addition. that x2 + 4 cannot be written as a single product p(x)q(x) oftwo elements of l. Prove that the rings 2Z and 3Z are not isomorphic. EXERCISES Let R be a ring with identity 1 = f . Prove that the rings Z[x] and Q[x] are not isomorphic. 4Z/12Z. On the other hand. cf. Then 1+ J consists of a11integers of the form 6x + 10y with x. 2 = 6(2) + 10(1) shows that the ideal I + J contains the ideal2Z. 6Z and 12Z = I of R containing 1. The set {ab I a El. (2) Let I be the ideal in Z[x] consisting of the polynomials with integer coefficients whose constant term is even (cf. 3Z/12Z. respectively. .Example Let R = Z and let I be the ideal12Z. Since every such integer is divisible by 2. hence is not in general an ideal. (1) Define the sum of I and J by 1+ J = {a + b ¡ a E 1. If I and J are ideal s in the ring R then the set of sums a + b with a E I and b E J is not only a subring of R (as in the Second Isomorphism Theorem for Rings). as is their sum x2 + 4. In is defined inductively by defining [1 = 1. Examples (1) Let I = 6Z and J = lOZ in Z. 2 and x2 = x . 2Z. Itis easy to check.3. denoted by I J. (2) Define the product of I and J. an with a¡ E [ for all i. y E Z. 6Z/12Z. where d is the greatest common divisor of m and n.. The product I J consists of a11finite sums of elements of the form (6x)(10y) with x. mZ + nZ = dZ. Example 5). so both 4 = 2 . . to be the set consisting of all finite sums of elements of the fonn al a2 . but is an ideal in R (the set is clearly closed under sums and r{a + b) = ra + rb E 1+ J since ra E I and r b E J). (3) For any n > 1. and In = I r:' for n = 2. to be the set of all finite sums of elements of the fonn ab with a E I and b EJ. 4Z. The quotient ring R = R/ I = Z/12Z has ideal s R. The two polynomials 2 and x are contained in 1. the ideal I + J is contained in 2Z. denoted by In. . 2. Note also that the elements of the product ideal I J are finite sums of products of elements ab from I and J. however. 1. x are elements of the product ideal 12 = 11. It is easy to see that the sum I + J of the ideals I and J is the smallest ideal of R containing both I and J and that the product [ J is an ideal contained in I n J (but may be strictly smaller. . Let I and J be ideals of R. define the n th power of 1.O . so that 6Z + 10Z = 2Z. 3Z.. Equivalently. We can also define the product of two ideals: Definition. and O = 12Z/12Z corresponding to the ideals R = Z. which c1early gives the ideal60Z. y E Z. In general. b E J}. 2Z/12Z.
3.3 Ring Homomorphismsand Quotient Rings 247 .7. Sec. Find a11homomorphic images of Z.
+ a 1 .+ m defined by (/)(f) = . bE Z} (e) If D is squarefree.7 Introduction to Rings .. 10. E 1 . O) I a E Z} (d) {(a..e. Decide which ofthe following are ideals ofthe ring Z[x]: (a) the set of all polynomials whose constant term is a multiple of 3 (b) the set of all polynomials whose coefficient of x2 is a multiple of 3 (e) the set of all polynomials whose constant term. Decide which of the following are ring homomorphisms from M2 (Z) to Z: (a) (. Prove that the map (/) : R .+ Z x Z defined by (/): (~ ~) H  (a. la..[i5] . d) is a surjective homomorphism and describe its kernel. b b) E Z}. b ! )la.+ S defined by (/)(a + b. ¡¿ f(t)dt is a homomorphism of 12.1 entry) (b) (.be (the trace of the matrix) (the detenninant of the matrix). 11. (b) If D is not a perfect square in Z prove that the map (/) : Z[. (a) Prove that S is a subring of M2(Z). additive groups but not a ring homomorphism.1] to m ... In each case describe the kernel and the image.. In each case describe the kernel and the image. 6.[i5) = (. 9.+ a (projection onto the 1. a) I a E Z} (b) {(2a. 5. Decide which of the following are ideals of the ring Z x Z: (a) {(a.. Prove that the map (/) : R .. a) I a E Z}. the polynomials in which only even powers of x appear) (e) the set ofpolynomials whose coefficients sum to zero (1) thesetofpolynomials p(x) suchthatp'(O) = O. (e) (. bE Z} (e) {(2a.where p'(x) is the usual first derivative of p(x) with respect to x. Let R be the ring of all continuous real valued functions on the closed interval [O.+ ad 7.b = 1 mod4 !) is a ring isomorphísm. ~) !) +d . 248 Chap. Find all ring homomorphisms from Z to Z/30Z... Let D be an integer that is not a perfect square in Z and let S = {( : . ~) 1 .. coefficient of x and coefficient of x2 arezero (d) Z[x2] (i. Decide which of the sets in Exercise 6 of Section 1 are ideals of the ring of all functions from [0. d Z} be the subring of M2(Z) ofupper triangular matrices. .2b) la.. b. Let R = {( ~ !) la. prove that the set {( (D _a1)b/4 a! is a subring of M2(Z) and is isomorphic to the quadratic integer ring O. 8. 1].4. Describe all ring homomorphisms from the ring Z x Z to Z.
Prove that if 11 e 12 e . Let a be an element of the ring R. Prove that the image of the center of R is contained in the center of S (cf. [Use Exercise 6(c) of Section 2 to show first that the set of entries of matrices in an ideal of Mn(R) form an ideal in R.A Prove that the map (XA is called the characteristic function 01 A with values in Zj2Z). I where the bar denotes passage to Rj1.ls then (/}(1R) is a zero divisor in S. Prove that the Binomial Theorem (a +b ) n= i: ( : ) a k b n k k=ü Sec. P (X) + R defined by A 1 * XA is a ring isomorphism. (b) Prove that if (/)is surjective and 1 is an ideal of R then (/}(l) is an ideal of S. 14. (b) Prove that if (/}(1R) = ls then (/}(u) is a unit in S and that (/}(u1) = (/}(u)1 for each unit u of R. 18. (a) If 1 and J are ideals of R prove that their intersection 1 n J is also an ideal of R. (b) Prove that the intersection of an arbitrary nonempty collection of ideals is again an ideal (do not assume the collection is countable).] 22. Let R be the ring of all functions from X into 7l. 23.3 Ring Homomorphismsand Quotient Rings 249 .1(1) is an ideal of R. (a) Prove that {x E R I ax = O}is aright ideal and {y E R I ya = O}is a left ideal (called respectively the right and left annihilators of a in R). (a) Prove that if (/}(1R) = f : . (R) is equal to M. Por each A E P(X) define the function XA : X + Zj2Z by XA(X) = { O 1 ifXEA if x r ¡ ..7. Prove that 1 n S is an ideal of S. Let (/): R + S be a ring homomorphism. 17. 24. are ideals of R then U~11n is an ideal of R. Let 1 be an ideal of R and let S be a subring of R.Prove that the ring M2 (IR) contains a subring that is isomorphic to C. Deduce that if S is an integral domain then every ring homomorphism from R to S sends the identity of R to the identity of S. 16. (a) Prove that if J is an ideal of S then (/}. Let X be a nonempty set and let P(X) be the Boolean ring of all subsets of X defined in Exercise 21 of Section 1. Prove that if S n = O then S rv S. Give an example where this fails if (/)is not surjective. Let S be a subring of R and let 1 be an ideal of R. 21. Let (/): R + S be a nonzero homomorphism of rings.j2Z. 15. Prove that every (twosided) ideal of M. 13. 25. (1) for sorne (twosided) ideal J of R. Let R and S be nonzero rings with identity and denote their respective identities by IR and ls. IH I. 19. Assume R is a commutative ring with 1. Apply this to the special case when R is a subring of S and (/)is the inclusion homomorphism lo deduce that if J is an ideal of S then J n R is an ideal of R.. Prove that the ring M4(IR) contains a subring that is isomorphic to the real Hamilton Quaternions. Exercise 7 of Section 1). Show by example that not every ideal of a subring S of a ring R need be of the form 1 n S for sorne ideal 1 of R. 20. (b) Prove that if L is a left ideal of R then {x E R I xa = Ofor all a EL} is a twosided ideal (called the left annihilator of L in R). Let (/): R + S be a surjective homomorphism of rings.
is a ring of characteristic O.+ 1 = O (n times) in R.[x]. prove that 91(Rj91(R» = O. +alX +ao bean element ofthepolynomialring R[x].l n J. Section 1) that an element x E R is nilpotent if x" = O for some n E Z+. + 1 (k times) ifk > O ifk kr+ { O 11..j n'll. 250 Chap. where the binomial coefficient ( Z )is interpreted in R as the sum 1 + 1 + .. al. is a ring of characteristic n for each positive integer n and 'll. 31. b e R. a2. [See Exercise 14 of Section 1. 26.] (b) Prove that p(x) is nilpotent in R[x] if and only if ll{}. (b) Determine the characteristics ofthe rings « 1 .e. Prove that a nonzero Boolean ring has characteristic 2 (cf. For examp1e.. (e) Give an example where l J = f . (b) Prove that 1J is an ideal contained in l n J.. Exercise 15. Let R be a commutative ringo Recall (cf. Prove that if R is a commutative ring and 91(R) is its nilradical (cf. where p is either a prime or O (cf. [Use the Binomial Theorem to show 91(R) is closed under addition. 33. . The characteristic of a ring R is the smallest positive integer n such that 1 + 1 +. ..'ll. Exercise 26).[x]. Prove that an integral domain has characteristic p. (a) Prove that l + J is the smallest ideal of R containing both l and J. Let p(x) = ax" +an_1Xn1 + . then (a + b)P = aP + b" for a11a. 29. (a) Prove that p(x) is a unit in R[x] if and only if ao is a unit and al. Prove that if x is a nilpotent element of R then c p (x ) is nilpotent in S. Zjn'll. .7 Introductionto Rings . + 1 of the identity 1 in R taken (~) times. Deduce that the set of nilpotent elements in the noncommutative ring M2(Z) is not an ideal.. Let l and J be ideals of R. the preceding exercise) then zero is the only nilpotent element of RjíJt(R) i.. an are nilpotent in R. Exercise 13.. 27.holds in R. an are nilpotent elements of R. 28. where n is the characteristic of R (this explains the use of the terminology "characteristic O" instead of the archaic phrase "characteristic 00" for rings in which no sum of l's is zero). Section 1). Assume R is commutative. . 'll.] 30.. 34. Prove that the elements (~ ~) and (~ ~) are nilpotent elements of M2 (Z) whose sum is not nilpotent (note that these two matrices do not commute).. 1 (k times) =O ifk < O is a ring homomorphism whose kernel is nZ. (a) Prove that the map Z + R defined by 1 + 1 + . (e) Prove that if p is a prime and if R is a commutative ring of characteristic p.... if no such integer exists the characteristic of R is said to be O. Prove that the set of nilpotent elements form an idealcalled the nilradical of R and denoted by 91(R).. 32.. Let c p : R + S be a homomorphism of rings.
(d) 35. = l n J.7. J. K be ideals of R. Let (a) (b) Prove that if R is commutative and if l + J = R then l J 1. Sec.4 Propertiesof Ideals 251 . + J K. Prove that l (J + K) = l J + 1K and (I + J)K = 1K Prove that if J e l then l n (J + K) = J + (I n K).
. a¡ E A. RA = {rlal + r2a2 + . n E Z+}. Thus RA is precisely the left ideal generated by A. Definition. This left ideal is obtained from A by closing A under all the operations that define a left ideal. . Conversely.e.. 1 anideal A~l i. we shall drop the set brackets and simply write (a). AR is the right ideal generated by A and RAR is the (twosided) ideal generated by A. n E Z+} (where the convention is RA = Oif A = 0). (4) An ideal generated by a finite set is called afinitely generated ideal.. Since the intersection of any nonempty collection of ideals of R is also an ideal (cf. Since R has an identity. (2) Let RA denote the set of all finite sums of elements of the form ra with r E R and a E A i.. etc. In particular.. .. .4 PROPERTIES OF IDEALS Throughout this section R is a ring with identity 1 = f :O. }. + rnanr~ I ri. a¡ E A. a2. The left ideal generated by A is the intersection of allleft ideals of R that contain A. 37. Thus RA is a left ideal of R which contains A..... m ~ O}is the set of polynomials whose first nonzero tenn has degree at least n. Similarly. AR = {alrl + azri + .. Show that if I is the ideal of all polynomials in Z[x] with zero constant tenn then In = {anxn + an+IXn+1+ . Prove that the ideal pZ/ o" Z is a nilpotent ideal in the ring Z/ p" Z. called the ideal generated byA. Section 3) and A is always contained in at least one ideal (namely R). respectively. Exercise 18. Let A be any subset of the ring R. rf E R. An ideal N is called nilpotent if N" is the zero ideal for sorne n ~ 1.4). When A = {a} or {al. n E Z+} and RAR = {rlalr~ + r2a2r~+ . 7. (A) is the intersection of all ideals of R that contain the set A. r E R and a E A and so must contain RA.e. we have (A ) = nI.. + rnan I r¡ E R.. (1) Let (A) denote the smallest ideal of R containing A. Similarly. a2. any left ideal which contains A must contain all finite sums of elements of the form ra.. It is irnmediate from the definition that R A is closed under addition and under left multiplication by any ring element. + an+mxn+mI a¡ E Z. The notion of ideals generated by subsets of a ring is analogous to that of subgroups generated by subsets of a group (Section 2. ) for (A). RA contains A.. a¡ E A. (al. + anrn I r¡ E R.36.. (3) An ideal generated by a single element is called a principal ideal.
the principal ideal (a) generated by a is just the set of all Rmultiples of a. the set .if R is commutative then RA = AR = RAR = (A). If R is not cornmutative. however. When R is a cornmutative ring and a E R.
Also.{ras I r. x) is not principal. The formation of principal ideals in a commutative ring is a particularly simple way of creating ideals. d: (n. The notation for (n. nZ e mZ if and only if m divides n in Z.e.e. integers. p(x) E {±l. n and m are relatively prime if and only if (n. Note that the symbol (A) is ambiguous if the ring is not specified: the ideal generated by 2 and x in Q[x] is the entire ring (1) since it contains the elernent » We sha11see in Chapter 9 that for any field F. the ideal generated by two nonzero integers n and m is the principal ideal generated by their greatest common divisor.we could make it unique by choosing a nonnegative generator). put another way. The only possibility is a(x) = ±2. As noted in the preceding section. Observe that (2.. Let g(x) be the function whichis !and 1 at a11other points. Thus containment relations between ideals. Since 2 is a prime number. s E R).1] into IRlet M be the ideal {f !2= 1. this is a proper ideal. m) = (d). x) = {2p(x) + xq(x) I p(x). clearly impossible. (3) We show that the ideal (2. a11ideals of F[ x] are principal.in particular. Assume by way of contradiction that (2. x) = (a(x» for sorne a(x) E Z[x]. But now x E (a (x = (2) = (2) and so x = 2q(x) for sorne polynomial q(x) with integer eoefficients. these are a11the ideals of Z so every ideal of Z is principal. i. hence bothp(x) anda(x) mustbeconstant polynomials. Examples (1) The trivial ideal O and the ideal R are both principal: O = (O) and R = (1). ±2}. is seen lo capture sorne of the arithmetic of general commutative rings. If a(x) were ±1 then every polynomial would be a multiple of a(x). so the lattice of ideals containing nZ is the same as the lattice of divisors of n. whicb consists of allfinite sums of elernents of the form ras. Thus our notation for aR is consistent with the definition of n7l. a divides b in R. i. (4) If R is the ring of a11functions from the closed interval [0. similar to generating cyclic subgroups of a group..we have been using. x) generated by 2 and x in Z[x] is not a principal ideal. b E (a) if and only if (b) e (a). I f(!) at x = = O } (the kernel of evaluation at ~). m) = (1). Since 2 E (a(x» there must be sorne p(x) such that 2 = p(x)a(x). if and only if bis a multiple of a or. Then f = f g for a11f E M so M zero is a . q(x) E Z[x]} and so this ideal consists precisely of the polynomials with integer coefficients whose constant term is even (as discussed in Example 5 in the preceding seetion) . m) as the greatest cornmon divisor of n and m is thus consistent with the same notation for the ideal generated by n and m (although a principal generator for the ideal generated by n and m is determined only up to a ± sign . contrary to (a(x» being a proper ideal. Commutative rings in which all ideals are principal are among the easiest to study and these wiIl play an important role in Chapters 8 and 9. The degree of p(x)a(x) equalsdegree p(x) + degreea(x). in particular between principal ideals. Notice that the element b E R belongs lo the ideal (a) if and only if b = ra for sorne r E R. s E R} is not necessarily the twosided ideal generated by a since it need not be closed under addition (in tbis case the ideal generated by a is the ideal RaR. This contradiction proves that (2. (2) In Z we have nZ = Zn = (n) = (n) for a11integers n. r. a(x). Furthermore. In particular. For positive integers n and m.
any function which is zero at and nonzero at a11other points is another generator for the same ideal M. if R is the ring of a11continuous functions frorn [0. In faet.principal ideal with generator g. ! . On the other band.1] to IRthen {f I f(!) = O } is no! principal nor is it even finitely generated (cf. the exercises).
1 I g E G}.if O and R are the only ideals of R let u be any nonzero element of R. (2) The ring R is a field if and only if every nonzero element is a unit. Corollary 10. the only (left. Conversely. which gives an analogue of Proposition 9(2) if R is not commutative (see the exercises). Rings whose only twosided ideals are O and the whole ring (which are called simple rings) will be studied in Chapter 18. although this need not be a minimal set of generators.For example. Every nonzero element of R is therefore a unit and so R is a field. then for any n > 2 the only twosided ideals in the matrix ring Mn(F) are O and Mn(F). The kernel of a nonzero homomorphism is a proper ideal hence is O by the proposition. andR. However. (2) Assume R is commutative. If D is a ring with identity 1 = 1 O in which the only left ideals and the only right ideals are O and D. If G is a finite group and R is a commutativering with 1then the Proposition 9.e. u is a unit. left and right ideals: cf. Let I be an ideal of R. if F is a field.(5) augmentationideal is generatedby the set {g . which shows that Proposition 9(2) does not hold for noncommutative rings. Then R is a field if and only if its only ideals are O Proof: (1) If I = R then I contains the unit 1. Thus there is sorne v E R such that 1 = vu. thenthe augmentation ideal is a principal ideal with generatora .if Gis a cyclicgroupwith generato ra. Converse1y. These results show that the ideal structure of fields is trivial.l. so by the first part R is the only nonzero ideal. (1) I = R if and only if I contains a unit. Section 3). if u is a unit in I with inverse v. even though this is not a division ring (it does have proper. Proof: The kernel of a ring homomorphism is an ideal. If R is a field every nonzero ideal contains a unit. OUTapproach to studying an algebraic structure through its homomorphisms will still playa fundamental role in fie1dtheory (Part IV) when we study injective homomorphisms (embeddings) of one field into another and automorphisms of fields (isomorphisms of a field to itself).. i. By hypothesis (u) = R and so 1 E (u). 1 = r(vu) = (rv)u E I hence R = l. then D is a division ringo Conversely. One important class of ideals are those which are not contained in any other proper ideal: Definition. then for any r E R r = r . nontrivial. An ideal M in an arbitrary ring S is called a maximal ideal if M = 1 S and  . right or two sided) ideals in a division ring D are O and D. If R is a field then any nonzero ring homomorphism from R into another ring is an injection.
the only ideals containing M are M and S o .
AeC We first show that J is an ideal. In a ring with identity every proper ideal is contained in a maximal ideal. 254 Chap. In tbis case. take any abelian group which has no maximal subgroups (for example. hence any result involving maxirnal ideals forces a ring to be nonzero..1) and Proof: Let R be a ring with identity and let 1 be a proper ideal (so R cannot be the zero ring. b e is nonernpty e closed under left and right rnultiplication by elernents of R. independent of Zorn's Lemma. By Zom's Lemma S has a maxirnal element which is therefore a maximal (proper) ideal containing l. so J is closed under subtraction. b. By Proposition 9(2) we see that M is maximal if and only ir R/ M is a field. For commutative rings the next result characterizes maxirnal ideals by the structure of their quotient rings.cf. however. Let S be the set of a11proper ideal s of R which contain l. This proves that each chain has an upper bound in S. then there are ideals Proof: This follows frorn the Lattice Isornorphism Theorem together with Proposi tion 9(2).e.b E J.7 Introduction to Rings . In many specific rings. Since each A E is e E J. By the Lattice Isomorphism Theorem the ideals of R containing M correspond bijec tively with the ideals of R/ M. so is J.rnake it into a trivial ring by defining ab = Ofor a11a. by definition of J we must have 1 E A for sorne A E C. If I define J to be the union of a11ideals in e e: J = UA.. B E such that a E A and b E B. The zero ring has no maximal ideals. i. A general ring need not have maximal ideals. The ideal M is rnaxirnal if and only if there are no ideals 1 with M e 1 e R. Assume R is commutative. If a. Q . The next proposition shows that rings with an identíty 1 = f : . By definition of a chain either A e B or B e A. The ideal M is a rnaximal ideal if and only if the quotient ring R / M is a field. Proposition 11. Exercise 16. Section 6. Then is a chain in S. the result that a finitely generated group has maximal subgroups or that every vector space has a basis) the proof relies on Zom's Lemma (see Appendix 1). 1 = f : . Proposition 12. This is a contradiction because each A is a proper ideal (A E e e S). S is nonernpty ( E S) and is partialIy ordered by inclusion. Certainly J is nonernpty because specifically.O). A. O E J since O is in every ideal A. the presence of maximal ideals is often obvious.O always possess maximal ideals. In such a ring the ideals are simply the subgroups and so there are no rnaximal ideals. Like many such general existence theorems (e. If J is not a proper ideal then 1 EJ.g. In either case a . so M is maximal if and only if the only ideals of R/ M are O and R/ M. This proves J is an ideal. For example.
Sec.4 Propertiesof Ideals 255 . 7.The proposition aboye indicates how to construct sorne fields: take the quotient of any commutative ring R with identity by a maximal ideal in R. We shall use tbis in Part IV to construct a11finite fields by taking quotients of the ring Z[x] by maximal ideals.
. Put another way. at least one of a.I. Since evaluationis a surjectivehomomorphismfrom R to IR. Assume R is commutative. Let n be a nonnegative integer. 1 (to ensure that the ideal is proper) and provided every time the product ab of two integers is an element of nZ. b one of a and b is an element of P. This is not true in general.. An ideal P is called a prime ideal if P :j:.. however.7 Introduction to Rings . Example 7 at tbe end of tbe preceding section).. Since íZ is not a field. Proposition 13. The augmentationideal is tbe kernel of tbe augmentationmap which is a surjectivehomo morphismonto tbe field F (i. (4) Let R be tbe ring of all functionsfrom [0. Assume R is commutative.. a natural generalization of the notion of a "prime" in the integers Z. b is an element of nZ. x) is a maxirnalideal in íZ[x] because its quotientring is tbe field íZI2íZ . According to the aboye definition the ideal nZ is a prime ideal provided n :j:. It is. R and whenever the product ab of two elements a. Pirst we translate the notion of prime ideals into properties of quotient rings as we did for maximal ideals in Proposition 12. We saw in Section 3 tbat tbis is tbe case if and only if n is a prime number. The ideal níZ of íZ is a maxirnaIideal if and onIy if íZlníZ is a field. but we shall see shortly that every maximal ideal is a prime ideal.l]let M¿ be tbe kernel of evaluationat a. tben tbe augmentationideal 1 is a maximal ideal of tbe group ring FG (cf. E R is an element of P. O that has no zero divisors.tbe kernel of evaluationat any fixed point is a maximal ideal in tbe ring of continuousreal valued functionson [O.e. we see again tbat (x) is not a maximal ideal in íZ[x].tbe implication in Proposition12 that 1 is a maximal ideal if RII is a fieldholds for arbitraryrings. a field). F GIl . (5) If F is a field and G is a finite group. Similarly.cf. Por the integers Z there is no difference between the maximal ideals and the nonzero prime ideals. This is equivalent to the usual definition that n is a prime number. x) e íZ[x]. The quotient ring íZ[x]1 (x) is isomorphicto íZ (tbe ideal (x) in íZ[x] is tbe kernel of tbe swjective ring homomorphismfrom íZ[x] to íZ given by evaluationat O). then at least The notion of a maximal ideal is fairly intuitive but the definition of a prime ideal may seem a little strange. (2) The ideal (2. Thus the prime ideals of Z are just the ideals pZ of Z : generated by prime numbers p together with the ideal O. (3) The ideal (x) in íZ[x] is not a maximalidealbecause (x) e (2.we see that RIMa . Note that Proposition 12 does not apply directIy since FG need not be commutative.1].1]to IRand for each a E [O. Example 3 aboye and Example 5 at tbe end of Section 3.Examples (1) Let n be a nonnegativeinteger.however. F. n must divide a or divide b. Then the ideal P is a prime ideal in R if 256 Chap.Rand hence M¿ is a maximal ideal. Definition. Recall that an integral domain is a commutative ring with identity 1:j:... if n :j:. O. it must have the property that whenever n divides ab.. Tbis also follows directly from tbe containmentof ideals of íZ describedin Example 2 aboye..
R and whenever Sec.4 Propertiesof Ideals 257 . : . Proof: This proof is simply a matter of translating the definition of a prime ideal into the language of quotients.and only if the quotient ring R / P is an integral domain. 7. The ideal P is prime if and only if P f .
] (b) Let G = {gl. Let L¡ be the left ideal of Mn(R) consisting of arbitrary entries in the jth column and zero in all other entries and let Eij be the element of M¿ (R) whose i. This ideal is not a maximal ideal. (a) Let p be a prime and let G be an abelian group of order pn. 1] and let 1 be the collection offunctions f(x) in R with f(1/3) = f(1/2) = O. Section 3). 1. 8. b « R. + gn) = O. Assume R is commutative.r= ah E P. Prove that the augmentation ideal in the group ring RG is generated by {g . . Prove that the nilradical of the group ring 1FpG is the augmentation ideal (cf. Let R be an integral domain. Note that r E P if and only if the element r is zero in the quotient ring R/ P.then either a = O or b = O . 5.1. Proof: If M is a maximal ideal then R/ M is a field by Proposition 12. Let R be the ring of a11continuous functions on [O. Examples (1) The principal ideals generated by primes in Z are both prime and maximal ideals. + . [Use the preceding exercise. Prove that the principal ideal generated by x in the polynomial ring R[x] is a prime ideal if and only if R is an integral domain..1 I g E G}.e. Use the bar notation for elements of R/ P: r + P. Prove that if G = {a } is cyclic then the augmentation ideal is generated by a . The ideal Ois a prime ideal in Z[x]. then either a E P or b E P. A field is an integral domain so the corollary follows from Proposition 13. . [See Éxercise 6. Prove that R is a field Üand only if Ois a rnaximal ideal. Corollary 14.] 2. Prove that if M is an ideal such that R/M is a field then M is a maximal ideal (do not assume R is commutative). 3. but is not a maximal ideal. Prove that L¡ = Mn(R)Eij for any i. Prove that (a) = (b) for sorne elements a. 9. Prove that (x) is a maximal ideal if and only if R is a field. Exercise 29. if and only if a = ub for sorne unit u of R. R/ P is an integral domain. 258 C hapo7 Introduction to Rings . Let R be a commutative ring with 1.. 6. Section 2. Prove that R is a division ring if and only if its only left ideals are (O)and R... Every maximal ideal of R is a prime ideal. The zero ideal in Z is prime but not maximal. It follows in particular that a commutative ring with identity is an integral domain if and only if Ois a prime ideal. (2) The ideal (x) is a prime ideal in Z[x] since Z[x ]/(x) rv Z.i. EXERCISES Let R be a ring with identity ] = 1 O. Assume R is commutative.] 4. Prove that if r is any eIernent of the augmentation ideal of RG then rts. Assume R is commutative. gn} be a finite group and assume R is commutative.. Thus in the tenninology of quotients P is a prime ideal if and only if R # O and whenever ah = ab = O . j entry is ] and whose other entries are a11O. Prove that 1 is an ideal of R but is not a prime ideal. (The analogous result holds when "left' is replaced by "right.") 7. [Use the preceding exercise.
then p(x) = f : q(x).] 15. . monic polynomial f(x). Prove either 1 or J is contained in P.16).. i. Assume R is commutative. (a) Prove that if Pis a prime ideal of S then either c p l (P) = R or c p l (P) is a prime ideal of R.. . Let 1 and J be ideals of R and assume P is a prime ideal of R that contains 1J (for example. an) and J = (bl.. (e) Let p be a prime.n. Prove that if P is a prime ideal of R and P contains no zero divisors then R is an integral domain. [Use Exercise 26(c) of Section 3.q(x) is an R[x]multiple of the . Let x4 . (a) Show that every e1ement of R[xJf(f(x» p(x) E R[x] of degree less than n. + an_IXn1 I a o .. Give an example to show that this need not be the case if c p is not surjective. .10. m. (b) Prove that if M is a maximal ideal of S and c p is surjective then c p l (M) is a maximal ideal of R. . prove that x is nilpotent in R[x]f(f(x».] + . [Otherwise p(x) . 13. aZ. . (b) Prove that x ..Ilx9 + 5x5 .a for sorne nilpotent element a E R. (a) Find a polynomial of degree ::::3: that is congruent to 7x13 . + bo then xn = (bn_lXn1 this to reduce powers of x in the quotient ring. (a) Prove that E has 4 elements: 0. Assume R is commutative. (b) Write out the 4 x 4 addition table for E and deduce that the additive group E is isomorphic to the KIein 4group. Assume R is commutative... Let x3 . Let x be an mdeterminete. 11.16). LetxZ + x + 1 be an element ofthe polynomial ring E = 1Fz[x]and use the barnotation to denote passage to the quotient ring 1Fz[x]f(xz + x + 1). prove thata(x) is a zero divisor in R[x]f(f(x».. Assume R is commutative and suppose 1 = (al. Apply this to the special case when R is a subring of S and c p is the inclusion homomorphism to deduce that if P is a prime ideal of S then P n R is either R or a prime ideal of R. Letp(x) = 2x7 7x5+4x39x+l 16. if P contains 1 n J). • anl E R}. let f(x) be a monic polynomial in R[x] of degree n ~ 1 and use the bar notation to denote passage to the quotient ring R[x]f(f(x».2 and x + 2 are zero divisors in E. .2x3 + 3 modulo (x4 . 1..al.16 be an element of the polynomial ring E = Z[x] and use the bar notation to denote passage to the quotient ring Z[x]f(x4 . + bol.e. x and x + l. Deduce that E is a field. 17. 12. ... . (d) If f(x) = x" .a for sorne a E 1Fp•Prove that x .. bz •. assume R = 1Fpand f(x) = xP . (e) Write out the 4 x 4 multiplication table for E and prove that EX is isomorphic to the cyclic group of order 3. 14...2x + 1 be an e1ement of the polynomial ring E = Z[x] and use the bar notation to denotepassagetothequotientringZ[x]f(x32x+l).. Let c p : R .. bm) are two finitely generated ideals in R.+ S be a homomorphism of commutative rings. R[x]f(f(x» [If f(x) is of the form p(x) for sorne polynomial = {ao + alx + ..] (e) If f(x) = a(x)b(x) where both a(x) and b(x) have degree less than n. = x" + bn_lXn1 + .a is nilpotent in R[x]f(f(x». and j = 1. Use (b) Prove that if p(x) and q(x) are distinct polynomials in R[x] which are both of degree less than n. Prove that the product ideal 1J is finitely generated by the elements aibj for i = 1.2...2 •.
.1)4.and let q(x) = (x .
Exercise 15.e. (b) Prove that E is not an integral dornain. (b) Prove that the lFp Quatemions contain zero divisors (and so they cannot be a division ringla [Use the preceding exercise. Prove that in a Boolean ring every finitely generated ideal is principal. a pgroup). Section 1). Section 2). Let R be a finite commutative ring with identity. c. Let p E Z+ be a prime and let the lFp Quatemions be defined by a + bi + cj + dk ZjpZ a. where addition is cornponentwise and rnultiplication is defined using the same relations on i.. Exercise 29. Prove that a nonzero finite commutative ring that has no zero divisors is a field (if the ring has an identity. Prove that every prime ideal of R is a maximal ideal.(a) Express each of the following elernents of E in the fonu f (x) for sorne polynomial f(x) of degree ::: 2: p(x). 22. Prove that the augmentation ideal in the group ring Zj pZG is a nilpotent ideal. . i. Prove that if R is an integral domain and R[[x]] is the ring of formal power series in the indeterminate x then the principal ideal generated by x is a prime ideal (cf. 26. Section 3).. (a) Prove that the lFp Quatemions are a hornornorphic image of the integral Quatemions (cf Section 1). (e) Prove that x is a unit in E. Prove that a finite ring with identity 1 i= O that has no zero divisors is a field (you may quote Wedderbum's Theorern)..] Prove that in a Boolean ring (cf.ab is a unitfor allb E R. so do not assume the ring has a 1). 27. 25. b.) [Use Exercise 2. Prove that every prime ideal of R is a maximal ideal. 24. Exercise 3. am) where each a¡ is a nilpotent elernent. then N is a nilpotent ideal (cf Exercise 37. k as for the real Quatemions. 19. j. Exercise 13. the nilradical ofthe quotient ring Rj 1. Section 3). Section 1) every prime ideal is a maximal ideal.e. (Note that this ring rnay be noncommutative. (rad I)j l = fJt(Rj 1) (cf Exercise 29. 21. (It is shown in Section 15.2 that the nilradical of R is equal to the intersection of a11prime ideals of R. 29. this is Corollary 3. Deduce that if the nilradical of R is finitely generated then it is a nilpotent ideal Let p be a prime and let G be a finite group of order a power of p (i.] Let l be an ideal of the commutative ring R and define rad l = {r E R I r" E 1 for sorne n E Z+} called the radical of l Prove that rad 1 is an ideal containing l and that (rad 1) jI is . d e 23. 18. Prove that if a is a nilpotent elernent of R then 1. Prove that a prime ideal in a commutative ring R contains every nilpotent elernent (cf. Assume R is commutative and for each a E R there is an integer n > 1 (depending on a) such that a" = a.) Let R be a commutative ring with 1 i= O. oz. Prove that if R is a commutative ring and N = (al. . Deduce that the nilradical of R (cf. 28. Prove that the principal ideal generated by x is a maximal ideal if and only if R is a field. 30. .. Section 3) is contained in the intersection of a11the prime ideals of R. 20. p(x) + q(x) and p(x)q(x). q(x).
An ideal l of the commutative ring R is called a radical ideal if rad l = l.31. .
Prove that O is a radical ideal in 7L. . Let 1 be an ideal of the commutative ring R and define Jac 1 to be the intersection of all maximal ideal s of R that contain 1 where the convention is that Jac R = R.(a) Prove that every prime ideal of R is a radical ideal.jn7Li. more generally.. Prove that M = 1 =Me for any e E J R (cf. 34. [Use Zom's Lemma. Deduce that (n) is a radical ideal of 7 L if and only if n is a product of distinct primes in 7 L . Prove that 1 is an ideal of R that is not a prime ideal.in terms of the prime factorization of n.+ Me is an injection from X to the set of maximal ideals of R. n is square free). 38. (b) Let n > 1 be an integer. f3(X) is a compact space under this topology and is called the StoneCedi eompaetifieation of X. Jac O is called the Jaeobson radical of the ring R. 1] such thatM (b) Prove that if b and e are distinct points in [0. a2. e 32.. For any subset X of J R or. A commutative ring R is called a local ring if it has a unique maximal ideal.. an) be a nonzero finitely generated ideal of R. (a) Let 1 be the collection offunctions f(x) in R with eompaet support (i. the map e . 33. Describe Jac n7L.. Let A = (al. by (a». (e) Prove that Me is not equal to the principal ideal generated by xc. so Jac 1 is the preimage in R of the Jacobson radical of R j l. (If 1 is the zero ideal. Assume R is commutative. = Me..] 36. for any completely regular topological space X. .) (a) Prove that Jac 1 is an ideal of R containing l.1] then Mb = 1 =Me. (e) Let n > 1 be an integer.M is a unit. (a) Prove that if Mis any maximal ideal of R then there is a real number e E [O . Let R be the ring of al! continuous functions from the closed interval [0. (d) Prove that Me is not a finitely generated ideal. Let R be the ring of all continuous functions from J R to J R and for each e E J R let Me be the maximal ideal {f E R I f(e) = O}.e. [Use Zom's Lemma.1] and the set of maximal ideals in the ring R of al! of continuous functions on [0.1] given by e ~ Me. (b) Prove that rad 1 e Jac 1. Let f3(X) be the set of maximal ideals of R. Prove that there is an ideal B which is maximal with respect to the property that it does not contain A. then R is a local ring with unique maximal ideal M.. Prove that the ring of all rational numbers whose denominators is odd is a local ring whose .] 37.e. (b) Let M be a maximal ideal of R containing 1 (properly.where rad 1 is the radical of 1 defined in Exercise 30. Prove conversely that if R is a commutative ring with 1 in which the set of nonunits forms an ideal M. the preceding exercise). . 1] let Me = {f E R I f(e) = O } (recall thet Me was shown to be a maximal ideal of R). One can put a topology on f3(X) in such a way that if we identify X with its image in f3(X) then X (in its given topology) becomes a subspace of f3(X).1] to J R and for each e E [O . Moreover. 35. . Prove that the set of prime ideals in R has a minimal element with respect to inclusion (possibly the zero ideal). Prove that if R is a local ring with maximal ideal M then every element of R .. f(x) = O for Ixl sufficiently large). The preceding exercise shows that there is a bijection between the points of the closed interval [0.f and only if n is a product of distinct primes to the first power (i. where R is the ring of all bounded continuous real valued functions on X and Me is the maximal ideal of functions that vanish at e.
unique maximal ideal is the principal ideal generated by 2. Following the notation of Exercise 26 in Section 1. 39. let K be a field. let v be a discrete .
Prove that the following are equivalent: (see also Exercises 13 and 14 in Section 1) (i) R has exaetly one prime ideal (ii) every elernent of R is either nilpotent or a unit (iii) R/1](R) is a field (cf. if a is a zero divisor we cannot always canee] the a's in the equation ab = ae to obtain b = e (take e = O for example). Thus a nonzero element that ís not a zero divisor enjoys sorne of the properties of a unit without necessarily possessing a multiplicative inverse in R.= . (e) An ideal Q of R is primary if and only if every zero divisor in R / Q is a nilpotent element of R / Q. For each integer k > O define Ak = {r E R I v(r) ~ k} U {O }. 7. Q then b" E Q for sorne positive integer n. (a) The primary ideals of íZ are Oand (pn).. A proper ideal Q of the commutative ring R is called primary if whenever ab E Q and a r . The aim of tbis section is to prove that a cornmutative ring R is always a subring of a larger ring Q in which every nonzero element of R that is not a zero divisor is a unit in Q. by definition. in which case tbis ring Q will be a field . Assurne R is commutative. j . where p is a prime and n is a positive integer. Indeed. are related b y 246 a b e d if and only if ad = be.valuation on K and let R be the valuation ring of v. Exercise 30)..). On the other hand. 1 = 23=Th. (a) Prove that Ak is a principal ideal and that Ao = > Al 2 A2 2 . b) . In order to see the essential features of the construction of the field Q from the integral domain Z we review the basic properties of fractions. Each rational number may be represented in many different ways as the quotient of two integers (for example. Exercise 29. we saw in Section 1 that a zero divisor a cannot be a unit in R and. the fraction a b is the equivalence class of ordered pairs (a. (b) Every prime ideal of R is a primary ideal.. ese representatíons .. etc. (b) Prove that if 1 1 S any nonzero ideal of R.5 RINGS OF FRACTIONS Throughout tbis section R is a cornmutative ringo Proposition 2 shows that if a ís not zero nor a zero divisor and ab = ae in R then b = e. The principal application of this will be to integral dornains. then 1 = Ak for sorne k ~ O.. Deduce that R is a local ring with unique rnaximal ideal A l. 40. then a positive power of a and a positive power of b both lie in Q. (d) If Q is a primary ideal then rad(Q) is a prime ideal (cf. the paradigrn for the construction of Q from R is the one offered by the construction of the field of rational numbers from the integral dornain Z. Section 3).. (Note that the symmetry between a and b in this definition implies that if Q is a primary ideal and ab E Q with neither a nor b in Q. In more precise terms.called its field of fraetions or quotient field.) Establish the following faets about primary ideals. 41.
d) if and only if . b) '" (e. (a.of integers with b = f : O under the equivalen ce relation: .
.e.. Then there is an injective homomorphism 4 > : Q + S such that 4 > IR = (/J. Theorem 15.5 Ringsof Fractions Ir E R. then we should expect to have d bd O d=~===O 1 b b in the "ring of fractions" (where. ab E D for all a. d) Sec. Remark: In Section 15. Then there is a commutative ring Q with 1 such that Q contains R as a subring and every element of D is a unit in Q. say bd = O. Readers wishing greater generality may read the proof below and the beginning of Section 15. Thus if we allow zero or zero divisors as denominators there must be sorne collapsing in the sense that we cannot expect R to appear naturally as a subring of this "ring of fractions. allowing arbitrary denominators. in the following sense. we have assumed R has al). a field). and if we allow b as a denominator. (TIa E Z) of Q and every nonzero integer II I a has an verse ~l'n '\e' . however.4 in concert..4 a more general construction is given. Let S be any commutative ring with identity and let (/J : R + S be any injective ring homomorphism such that (/J(d ) is a unit in S for every d E D .does not contain any zero divisors and is closed under multiplication (i. The main result of this section shows that these two restrictions are sufficient to construct a ring of fractions for R. Note that this theorem ineludes the construction of Q from Z as a special case. any ring containing an isomorphic copy of R in which all the elements of D become units must also contain an isomorphic copy of Q.7. on :F by 5 . If. Proof: Let:F = {(r. Let R be a commutative ringo Let D be any nonempty subset of R that does not contain O.{O}then Q is a field. In other words.e. then bd must also be a denominator. if D = R . m a It seems reasonable to attempt to follow the same steps for any commutative ring R.ad The arithmetic operations on fractions are given by a e ad + be a e ae b + d = bd and b x d = bd These are weIl defined (independent of choice of representatives of the equivalence classes) and make the set offractions into a commutative ring (in fact. (2) (uniqueness of Q) The ring Q is the "smallest" ring containing R in which al] elements of D become units. for convenience. The integers Z are identified with the subring = be. The ring Q has the following additional properties. The proof of the general result is more technical but relies on the same basic rationale and steps as the proof of Theorem 15. In particular. the set of denominators must be closed under multiplication in R. i. b E D).. (1) every element of Q is ofthe form rd:' for sorne r E R andd E D." A second restriction is more obviously imposed by the laws of addition and multiplication: if ring elements b and d are allowed as denominators. d E D} and define the relation . bis zero or a zero divisor in R. Q.
.(r. (s. d) . e) if and only if re = sd..7 Introduction to Rings . 262 Chap.
.. Checking the details in the other parts of (1) lo (4) involves even easier manipulations and so is left as an exercise. b) I a r E r R.e. e e'.(i. to check that addition is well defined assume ~ = ~ (i. Denote the equivalence class of (r. (2) Q is an abelian group under addition.sd = O and sf . Note that r = re in Q for all e d de a e ae x d = bd' E Let Q be the set of equivalence classes under r.td)e = O. ed = e d). do not depend on the choice of representatives for the equivalence classes).te = O . .= . rd = re . We now define an additive and multiplicative structure on Q: a b+ e d = ad bd + be and b In order to prove that Q is a cornmutative ring with identity there are a number of tbings to check: (1) these operations are well defined (i. distributive and cornmutative. and (4) Q has an identity (= ~ for any a « D). 1).td = O .e.e. Hence addition of fractions is well defined. Since e E D is neither zero nor a zero divisor we must have rl . b E D and rb = ad}. (r. The left hand side of this equation is ab' dd' + cd' bb' substituting a' b for ab' and e'd for ed' gives a' bdd' + e'd bb'. These are all completely straightforward calculations involving only arithmetic in R and the definition of r: Again we need D to be closed under multiplication for addition and multiplication to be defined. . i. and . where the additive identity is d for any d E D O and the additive inverse of : is da.e. For example. d) '" (t. which is the right hand side.. hence an equivalence relation. d) "' (s. e) rv (t. Next we embed R into Q by defining L : R~ Q by L : r ~ d rd where d is any element of D. e) and (s. (3) multiplication is associative. . Multiplying the first of these equations by I and the second by d and adding them gives (r I .. Then re . We must show that d d' bd b'd' (ad + be)(b' d') = (a'd' + b'e') (bd). . Suppose (r. ad + be = a' d' b' e' l..e. since D is closed under multiplication. This proves "'is transitive. ab' = a b) ' b b' + . 1).It is irnmediate that this relation is reflexive and syrnmetric.  D. . d) by d: d = Ha.
L(r) does not depend on the choice of d E D. Since d e D is closed under multiplication. . one checks directly that L is a ring homomorphism.Since for all d. e E D.
Corollary 16. Extend <pto a map tP : Q ~ S by defining tP(rdI) = <p(r)<p(d)l for aIl r E R. s p : R ~ F is an injective homomorphism from R into the field F. since <pis injective this forces r and hence also rd1 to be zero.the details are left as an exercise. = Jt is straightforward to check that tP is a ring homomorphism . Let R be an integral dornain and let Q be the field of fractions of R. In particular.{O }. any subfield of F containing A contains the subfield generated by A). (2) If R is an integral domain and D = R . D and Q be as in Theorem 15. if A is a subring of F). The subring teR) of Q is therefore isomorphic to R. if d is represented by the fraction de then its multiplicative inverse is !_ One then sees that .Furthermore. We henceforth identify each r E R with t (r) and so consider R as a subring of Q.every nonzero element of Q has a multiplicative inverse and Q is a field. Finally. e ~ every element of Q may be written as r . Let tP : Q ~ F be the extension of <pto Q as in the theorem. t is injective because t(r) = o { } . Assume <p : R ~ S is an injectivering homomorphism such that <p(d) is a unit in S for all d E D. d E D. If A is a subset of a field F (for example. By Theorem 15. The next corollary shows that the smallest field containing an integral dornain R is its field of fractions.= . Let R. Next note that each d E D has a multiplicative inverse in Q: namely.{O }. . tP is injective. (1) The ring Q is called the ring offractions of D with respect to R and is denoted DIR. tP is injective because rd1 E ker tP implies r E ker tP n R = ker e. then the intersection of all the subfields of F containing A is a subfield of F and is called the subfield generated by A. It remains to establish the uniqueness property of Q. so <p(r)<p(e) <p (s)<p (d ). This map is well defined. and then = = tP(rd1) = <p(r)<p(d)l = <p(s)<p(e)l tP e (s 1). This subfield is the smallest subfield of F containing A (namely. dl for sorne r E R and sorne d E D. If a field F contains a subring R' isomorphic to R then the subfield of F generated by R' is isomorphic to Q. since rd1 se1 implies re = sd. This completes the proof. if D = R .{} rdz = o { } r = o d d rd o because d (hence also~) is neither zero nor a zero divisor. Definition. In particular. Proof: Let <p : R ::::R::' e F be a (ring) isomorphism of R to R'.Q is called thefield offractions or quotient field of R.
Now. any subfield of F containing R' = <p(R) contains the elements <p(rt)<p(rz)l = <p(rlr¡l) for all rl. rz E R.so tP (Q) is an isomorphic copy of Q in F containing <p(R) = R'. Since .
EXERCISES Let R be a commutative ring with identity 1 = 1 = o. b E R and b # : O.every element of Q is oftheform rlr¡l for some rj . q(x} are polynomials in Q[x] then for sorne integer N..ent o f two po 1ynomi ilth m. 1. Let F be a field. . so that 4> (Q ) is the subfield of F generated by R'. p(x) and q(x) may both be constant polynomials. so F = Q. Then p(x) = can b e wnritten as th e quotí. Fill in a11the details in the proof ofTheorem 15. (4) If R is any integral dornain. which explains the notation. 2.. In this way any nonzero element of R that is not a zero divisor can be inverted in a larger ring containing R. Nq(x) have integer coefficients (let N be a eommon denominator for all the coefficients in p(x) and q(x). Prove that the ring of fractions D 1R is isomorphic to a subring of the quotient field of R (hence is also an integral dornain). Np(x}. d3. . 3. we 8ha11denote the field ofrational funetions by F(x). suppose R = Z. Prove that F contains a unique smallest subfield Fo and that Fo is isomorphie to either Q or ZjpZ for sorne prime p (Fo is called the prime subfield of F). d2. (2) The integers Z are an integral dornain whose field of fraetions is the field Q of rationa] numbers. The elements of this field are of the form . For example. so the field of fraetions of Q[x] is the same as the field of fractions of Z[x]. itfollows that any subfield of F containing R' contains the field 4> (Q ). (5) If R is any commutative ring with identity and d is neither zero nor a zero divisor in R we may form the ring R[ljd] by setting D = {l. Let R be an integral domain and let D be a nonempty subset of R that is closed under multiplication. Examples (1) If R is a field then its field of fraetions is just R itself. Note how an identity "appears" in the field of fraetions. so the field of rational functions contains the field of fractions of R: elements of the form ~ sueh that a. (3) The subring 2Z of Z also has no zero divisors (but has no identity). The quadratic integer ring O of Seetion 1 is an integral dornain whose field of fraetions is the quadratic field Q( ID). namely F(x). Note that the elements of R[1jd] look like polynomials in Ijd with coefficients in R. proving tbe corollary. d. then the polynomial ring R [x] is also an integral dornain. The associated field of fractions is the field of rational functions in the variable x over R. In particular. r z E R. p(x) q(x) where p(x) and q(x) are polynomials with coefficients in R with q (x) not the zero polynomial. If F is a field. Note that R is the subring of elements of the form  q(x) Nq(x) i. Its field of fraetions is also Q.7 lntroduction to Rings . teger coNp(x} ials efficients. 8 Chap. } and defining R[ljd] to be the ring of fractions D 1R. for example). If p(x}. Thus if F is the field of fraetions of the integral dornain R then the field of rational functions over R is the same as the field of rational funetions over F.
Prove that any subfield of lR must contain Q.] 4. Seetion 3.7 lntroduction to Rings .[See Exercise 26. 9 Chap.
= F j the ideals A. SO k} with i R/(AIA2··· Ak) = R/(At n A2 n··· n Ak)""" R/AI x R/Az x .. This in tum is equivalent to nZ + mZ = Z as ideals (in general. we shall denote by RI x R2 tbeir direct product (as rings)..) i 7.. and R2 are two rings.A. More generally. A B. Exercise 34. + B = Recall that tbe product. (Chinese Remainder Theorem) Let Al. .] 6. r2) with rl E R.. . their (ring) direct product is defined to be their direct product as (abelian) groups made into a ring by defining multiplication componentwise. then AB = (ah). if R.6 THE CHINESE REMAINDER THEOREM Throughout this section we shall as sume unless otherwise stated that all rings are com mutative with an identity 1 = FO . Show the field offractions ofthe power series ring Z[[x]] is properly contained in the field of Laurent series Q((x». . contain a subring A with 1 E A and A maximal (under inc1usion) with respect to the property that f / .. r2)(SI. If for each i. IR. A2. j E {1. Prove that the real numbers.. then Al . If A = (a) and B = (h). X R/Ak. . Theorem 17. ak). Section 3). and Aj are comaximal. . S 2) = (r) + SI. x E A and y E B (cf. 2. prove that the field of fractions of F[[x]] (the ring of formal power series in the indeterminate x with coefficients in F) is the ring F ((x» of formal Laurent series (cf. A2.. The ideals A and B of the ring R are said to be comaximal if A R. r2S2). Ak. .. the set of ordered pairs (rl. n Ak = AIA2 . of the ideals A and B of R is the ideal consisting of all finite sums of elements of the form xy.. tbat is. n ) Z). This motivates tbe following definition: Definition. then this map is surjective and Al n A2 n . If Ai = (ai).3 shows IRis the quotient field of A. Given an arbitrary collection of rings (not necessarily satisfying the conventions above).] (Exercise 13 in Section 15. and ri E R2 where addition and multiplication are performed componentwise: (rl. r2) + (SI. nZ + mZ = (m.... Xk such that Xi E A¡ for all i. The map R~ R/AI xR/Azx···xR/Ak defined by is a ring homomorphism with kernel Al n A2 n . Exercises 3 and 5 of Section 2). In particular.. Ak is the ideal of all finite sums of elements of the form XIXZ . [Use Zom's Lernma.. S2) = (nSI. r z + S2) and (rl. [Consider the series for eX..... . so IRis the quotient field of a proper subring. Ak be ideals in R. Ak = (al· . In Z this is equivalent to being able to solve the equation nx + my = 1 in integers x and y (tbis fact was stated in Cbapter O and will be proved in Chapter 8).5. n Ak. tbe product of tbe ideals Al. We note that a map c p from a ring R into a direct product ring is a homomorphísm if and only if the induced maps into each of the components are homomorphisms. If F is a field. There is a generalization to arbitrary rings of the notion in Z of two integers n and m being relatively prime (even to rings wbere the notion of greatest cornmon divisor is not defined)..
6 The ChineseRemainderTheorem 265 .7.Sec.
... for eacb i E {2. r2 mod B)(O. If A and B are cornaximal and x and y are as above./n 7 l.) / . then the element rix + rlY maps to this element since cp(r2x + rIY) = cp(r2)cp(X) + cp(rl)cp(y) + (rl = (r2 mod A.m 7 l. Ak once we show that Al and A2 . 1) + (rl mod A. O) = (r¡ mod A. This equation shows that cp(x) = (O. Chinese Remainder Theorem gives the m the following isomorphism on the groups of units: 10 Chap. Since A + B = R. it follows that 1 = (X2 + Y2) .. . relates to simultaneously solving two congruences modulo relatively prime integers (and states that such congruences can always be solved.. i.. This establishes the reverse inclusion A n B e AB and completes the proof when k = 2. e = el = ex + ey E AB.7 Introduction to Rings . Since the isomorphism in the Chinese Remainder Theorem is an isornorphism of rings.. O) = (O. rI mod B)(1.. 1) and cp(y) = (1. hence the name. This theorem obtained its name from the special case 7 l .. cpis surjective and A n B = AB. Consider the map cp : R ~ R/ A x R/ B defined by = = cp(r) = (r mod A. This shows that tp is indeed surjective. where mod A means the class in R/ A containing r (that is. O)since.. The kernel of cp consists of alI the elements r E R that are in A and in B. k} there are elements x¡ E Al and Yi E A¡ such that xi + Yi = 1. A n B.. x is an element of A and x = 1 .. By hypothesis. r2 mod B) mod A. for example.Proof: We first prove this for k = 2. the ideal AB is always contained in A n B. 3. This map is a ring homomorphism because cpis just the natural projection + of R into R/ A and R/ B for the two components. r mod B). .7 l.. rz mod B)./ n 7 l.. Since x¡ + Yi = Yi mod Al.e..m n 7 l... In the case of 7 l . and uniquely). The general case follows easily by induction from the case of two ideals using A = Al and B = A2 . It is easy to see that the units in any direct product of rings are the elements tbat have units in each of the coordinates. then for any e E A n B.. the general case will follow by induction. This completes the proof. FinalIy. Sorne examples are provided in the exercises.. Ak). phrased in numbertheoretic terms. / x (7 l..) rings when m and n are relatively prime integers. _(. there are elements x E A and y E B such that x + y = 1. We proved this as isomorphismjust for the additive groups earlier. r A). (Xk + Yk) is an element in Al + (A2 . This isomorphism. ri mod B) is an arbitrary element in R/ A x R/ B. in particular the groups of units on both sides must be isomorphic. Such problems were considered by the ancient Chinese.y E 1 + B. To complete the proof in this case it remains to show that when A and B are comaximal. .. Ak are comaximal. Let A Al and B A2. Ifnow (rI mod A.
7 Introduction to Rings . 11 Chap.More generally we have the following result.
which is relatively prime to ni byassumption.x ..7 x . n~) = 1 gives ti = b) and that these then quickly give the solutions to the systern of congruences aboye for any choice of al. and that the solution x is unique rnod n = nln2 . (b) Let n~ = nfn¡ be the quotient of n by ni.5.jn 'lL .). we obtain the formula for the Euler 91function.jP k a k 'lL .. • ak E 7 L there ..) X r. x = =al mod x¡ .j a z 'lL . Prove that the solution x in (a) is given by x = al tI ní + a2t2n~ + . Show that e and 1e are identities for the subrings Re and R(1e) respectively. is a solution x E 7 L to the simultaneous congruences .../2 7 L [Use the preceding exercise. Note that the elernents ti can be quickly found by the EuclideanAlgorithm as described in Section 2 of the Preliminaries chapter (writing an¡ + bn~ = (ni._ L . n2. EXERCISES Let R be a ring with identity 1 = 1 =O. x 7 L .. Let ti be the inverse of n~ rnod ni.. as rings. Letnj .. Exercise 15 of Section 1).··.. . Prove that ir R and S are nonzero rings then R x S is never a field.j 'lL . X p2 .] . 4. Prove that Re and R(1 .jP I al 'lL ). 'lL ). P k ak be its factorization into powers of distinct primes. Then 'lL . . 2.. Prove that R . An elernent e E R is called an idempotent if e2 = e. Let n be a positive integer and let P I al P 2 az ••.j « a k P If we compare orders on the two sides of this last isomorphism.. . nk beintegerswhicharerelativelyprimeinpairs: (ni. 5.) n x r v ('lL . nk../2 7 L . l.e). The value of 91on prime powers pa is easi1y seen to be 9 1 (p a ) = p a 1 (p 1) (cf. Chapter O).jn 'lLv ('lL ..) into a direct product of cyclic groups. 3.jP la l'lL .. namely that 91(ab) = 91(a)91(b) whenever a and b are relatively prime positive integers.jP 2 a z 'lLx ). as.. The complete structure x is derived at the end of Section 9.. . nj) = 1 for all i '# i. Let R be a finite Boolean ring with identity 1 = 1 =O (cf. X ('lL .e) are twosided ideal s of R and that R .. so in particular we have the following isomorphism of multiplicative groups: ('lL . Let R and S be rings with identities....x X ('lL . . Assurne e is an idernpotent in R and er = re for all r E R.Corollary 18.) ('lL .R. . a2. X • • • x ('lL . From this and the multiplicativity of 91we obtain its value on al1positive integers.e x R(1.... Prove that every ideal of R x S is of the form 1 x J where 1 is an ideal of R and J is an ideal of S. Corollary 18 is also a step toward a determination of the decomposition of the abelian group ('lL . (a) Show that the Chinese Rernainder Theorern implies that for any al. This in tum implies that 91is what in elementary number theory is termed a multiplicative function.. + aktkn~ rnod n.
7.Sec.6 The ChineseRemainderTheorem 267 .
jm7l. Suppose for every pair of índices i. Fora E Ai.47 mod 81. then so is Pi for a11i (so we may then identify each A¡ as a subset of A). and y . !z(x) .] 7.(e) Solve the simultaneous system of eongruenees x=lmod8..b E Aj showthattheoperation a + b = Pik(a) + pjk(b) where k is any index with l..e. . then A may likewise be given the strueture of a eommutative ring with 1 sueh that a11Pi are ring homomorphisms.. Appendix 1). .is an equivalenee relation on B.. Deduce that the maps Pi in (b) are group homomorphisms from A¡ to A. is well defined and makes A into an abelian group. . . i with i : : s j there is a map Pij : A¡ + Aj sueh that : the following hold: i. In Exercise 8 assume also that lis a direeted set: for every i. the eoefficients of f (x) agree with the eoefficients of f¡ (x) mod ni. nz. Let f¡(x). (a) Show that r. f(x) = !2(x) ns mod nj . (d) Show thar if a11Ai are commutative rings with 1 and a11Pij are ring homomorphisms that send 1 to 1.. + ZjnZ is also surjective on the units: (7l. . /k(x) be polynomials with integer coefficiems of the same degree d. In these exercises l is a nonempty index set with a partial order : : s (ef. (The set of equivalenee classes is ealled the direct or induetive limit of the directed system {A¡ }. and : : . : for a E A¡ and b E Aj. For : each i E l let A¡ be an additive abelian group. Pjk o Pij = Pik whenever i : : s j : : s k. j E 1 there is sorne k E l with i : : s k and j : : s k. Show that if a11 the f¡ (x) are monie.7 Introduction to Rings . Use the Chinese Remainder Theorem to prove there exists a polynomial f(x) with integer eoeffieients and of degree d with f(x) .)X • The next four exereises develop the eoneept of direet limits and the "dual" notion of inverse limits. Pi¡ = 1 fora11i E l Let B be the disjoint union of a11the A¡. and x . Prove that the natural surjective ring projection 7l. and is denoted !!w A¡. Let ni.)X + (7l. . (e) Assume all o.jn7l. ii. 12 Chap. nj) = 1 for a11i = 1j). Define a relation "" on B by a " V b if and only ifthere exists k with i. ..e. j ~k. [Apply the Chinese Remainder Theorem in Z to eaeh of the eoeffieients separately. y x = 2 mod 25. (n¡. f(x) = /k(x) mod i. : : 8. then f (x) may a1sobe ehosen monie.) (b) Let x denote the class of x in A and define Pi : A¡ + A by Pi (a) = a. Show that if each Pij is injective. and the simultaneous system y=5mod8.f¡(~m) od nI . .. . . Inthe remaining parts of this exercise let A = !!w A¡..3 mod 81 = 12 mod 25. are grouphomomorphisms. nk be integers which are relatively prime in pairs (i. j : : s k and Pik(a) = pjk(b). 6. Let m and n be positive integers with n dividing m.jm7l.
then there is a unique j homomorphism lp : A + e sueh that lp o Pi = lpi for all i. 13 Chap.7 Introduction to Rings .(e) Under the hypotheses in (e) prove that the direct limit has the following universal property: if e is any abelian group sueh that for each r E l there is a homomorphism lpi : Ai + e with lpi = lp o p¡j whenever i ~ i.
and let 1 = z+ (usual ordering). (e) Prove that bo + bs p + b2p2 + h 3 + . and A¡ is a group for all i El. and is denoted by 7lp.let A¡ = 'll/ pi'll and let Ilji be the natural projection maps pi). with each é .9. Exercise 15. as in (a) is a unit in 'llp if and only if bo = 1 p3 O.+a(mod The inverse limit ijm 'll/ pi 'll is called the ring of padic integers.1 distinct (p ... 10. Ilji : a (mod pj) !.) (b) Prove that 'llp is an integral domain that contains a copy of the integers. Prove that every ideal of 7lp is ofthe form pn'llp for sorne integer n :::O . then so is Il¡ for all i (so each Ai is a quotient group of P).1)8t roots of 1.. Ilii = 1for all i E l. then A may likewise be given the structure of a cornmutative ring with 1 such that all Ili are ring homomorphisms... For each U let Au be the ring of continuous real valued functions on U. In the notation of the preceding exercise. = IljiOJij wheneveri ::::j:. (e) Show that if all A¡ are cornmutative rings with 1 and all Ilj¡ are ring homomorphisms that send 1 to 1. and is denoted ~ A¡ . Let p be a prime let 1 = 'll+. Section 5. [Write a least residue in each 7l/ pi 'll in its base p expansion and then describe the maps Ilji'] (Note in particular that 'llp is uncountable.. = . We now develop the notion of inverse limits. Deduce that 'llp contains p .l .. (b) Assume the hypotheses in (a). Describetherulesfor adding and multiplying such formal sums corresponding to addition and multiplication in the ring 'llp. (d) Under the hypotheses in (a) prove that the inverse limit has the following universal property: if D is any group such that for each i E 1 there is a homomorphism n¡ : D + Ai withzr. (a) Show that every element of 'llp may be written uniquely as an infinite formal sum bo+blP+b2p2+~p3+ .) (a) Assume all Ilji are group homomorphisms. j with r :::::j there is a map Ilji : Aj + Ai such that thc following hold: i. pl}. For each i E 1 let Il¡ : P + Ai be the projection of Ponto its ith component. Continue to assume 1is a partially ordered set (but not necessarily directed). The set Pis called the invers eorprojective limitofthe system {A¡}.1). 11. E {O. LetPbethesubsetofelements(a¡)ieI in the directproductj'[j . (d) Prove that p'llp is the unique maximal ideal of 'llp and 'llp/p'llp '" 'll/p'll (where p = 0+ lp + Op2 + Op3 + . show that the maps ou : Au + A are not iniective but are all surjective (A is called the ring of germs 01continuous functions at p). b) on the realline containing a fixed real numberp. A¡ suchthatllj¡(aj) = a¡ whenever i ~ j (here ai and aj are the ith and jth components respectively ofthe element in thedirectproduct).:a:nd Ü. For V e U define the restriction maps ouv : Au + A v by 1 ~ 1 lv. Suppose for every pair of índices i.. ). Let 1 be the collection of open intervals U = (a. thenthereisauniquehomomorphism Ji : D + P such that Il¡ o Ji = Ji¡ for all i. Orderthesebyreverseinclusion: U::::: VifV ~ U (notethatlisadirectedset).. Show that if each Ilji is surjective. Ilji o Ilkj = Ilki whenever i :::j: ::k:. Let A = !!w Au be the direct limit. Show that P is a subgroup of the direct product group (cf. (e) Show that if al i= O (mod p) then there is an element a = (a¡) in the direct limit 'llp satisfying af 1 (mod pj) and Iljl(aj) = al for all j. the usual restriction of a function on U to a function on the subset V (which is easily seen 10 be a ring homomorphism)...
7.Sec.6 The ChineseRemainderTheorem 269 .
1 EUClIDEAN DOMAINS We first define the notion of a nonn on an integral domain R. Definition. We prove here all the theorems on the integers Z stated in tbe Preliminaries chapter as special cases of results valid for more general rings.CHAPTER8 Euclidean Domains. The principal examples of such rings are the ring Z of integers and polynomial rings F[x] witb coefficients in sorne field F. Definition. Those considered in this chapter are rings witb a division algorithm (Euclidean Domains). 270 . rings in which every ideal is principal (Principal Ideal Domains) and rings in which elements have factorizations into primes (Unique Factorization Domains). The element q is called the quotient and the element r the remainder of the division. and Unique Factorization Domains There are a number of classes of rings witb more algebraic structure tban generic rings. This is essentially no more than a measure of "size" in R. These results will be applied to the special case of the ring F[x] in the next chapter. 8. Principal Ideal Domains. AH rings in this chapter are cornmutative. We observe that this notion of a nonn is fairly weak and that it is possible for the same integral domain R to possess several different norms. If N (a) > Ofor a = 1 O define N to be e positive nonn. The integral domain R is said to be a Euclidean Domain (or possess a Division Algorithm) if there is a nonn N on R such that for any two elements a and b of R with b = 1 Othere exist elements q and r in R with a =qb+r with r = Oor N(r) < N(b). Any function N : R ~ Z+ U {O}with N (O) = Ois called a nonn on the integral domain R.
.balso satisfy the conditions ofthe DivisionAlgorithm applied to a. r are as aboye with r > O . If for exarnple b > O and q. (k+ l)b). then the polynomial ring F[x] is a Euclidean Domain with norm given by N(p(x» = the degree of p(x).The importance of tbe existence of a Division Algoritbm on an integral domain R is tbat it allows a Euclidean Algorithm for two elements a and b of R: by successive "divisions" (tbese aetually are divisions in tbe field of fraetions of R) we can write a = qob + ro (O) (1) (2) b = qlro +rl rn2 = qnrnl Tn + (n+ (n) 1) rnl = qn+lrn where N(rl) > . Note also that tbere is no guarantee tbat these elements are unique. The quotient and remainder are unique if we require the remainder to be nonnegative. then a = q'b + r' with q' = q + 1 and r' = r .g. then a = (q)b + r satisfies the requirements ofthe DivisionAlgorithm for a and b. If b < O (so b > O ). This argurnent can be made more formal by using induction on la l. where q = abl. n E Z partition the realline and so a is in one of them. The existence of a DivisionAlgorithm in Z (the familiar "long division" of elementary arithmetic) is verified as follows. (n+l)b ). The half open intervals [nb.8. 2 . b.1 are the two ways of applying the Division Algorithm in Z to a = 5 and b = 2. 2 + 1 = 3 . The proof is very similar to that for Z and is given in the next chapter Sec. r « is tbe last nonzero remainder.. Thus 5 = 2 . by what we have just seen there is an integer q such that a = q( b) + r with either r = O or Irl < I . (2) If F is a field. r: the proof aboye always produced a positive remainder r.. The DivisionAlgorithm for polynomials is sirnply "long división" of polynomials which may be familiar for polynomials with real coefficients.qb = r E [O . (1) The integers Z are a Euclidean Domain with nOnTIgiven by N(a) = lal. saya E [kb. b with b = 1 = we have a = qb + O . > N(rn) is adecreasing sequence ofnonnegativeintegersiftberemainders are nonzero. Such an rn exists since N(b) > N(ro) > Examples (O) Fields are trivial examples of Euclidean Dornains where any nOnTIwill satisfy the defining condition (e. the usual absolute value. Let a and b be two nonzero integers and suppose first that b > O . N (a) = O for all a).bl. For q = k we have a . lb!) as needed. Note that if a is not a multiple of b there are always two possibilities for the pair q. and such a sequence cannot continue indefinitely. This is because for every a.1 EuclideanDomains O 15 .
and Unique FactorizationDomains . (3) The quadratic integer rings O in Section 7.1 are integral domaíns with a norm defined by the absolute value of the field nOnTI(to ensure the values taken are nonnegative.(although for polynomials the quotient and remainder are shown to be unique). In order for a polynomial ring to be a Euclidean Domain the coefficients must come from a field since the division algorithm ultimately rests on being able to divide arbitrary nonzero coefficients.8 Euclidean. 272 Chap. We shall prove in Section 2 that R[x] is not a Euclidean Domain if R is not a field.Principal Ideal.
Let K be a field.1) that a diserete valuation ring is obtained as follows. Let p be an integer closest to the rational number r and let q be an integer closest to the rational number s . (ii) v is surjective.q)i and set y = f3e.(p + qi)f3 is easily computed. however.J=I9) /2] is not a Euclidean Domain with respect to any nonn appears at the end of this section.1) for D = 2.when D < o the field nonn is itself a nonn). Note also that the quotient need not be unique: if r (or s) is half of an odd integer then there are two choices for p (or for q. y E KX withx + y = 1o .p)2 + (s . but in general O is not Euclidean with respect to this nonn (or any othernonn). v(y)} for a11x.p) + (s . and a proof that Z[ (1 + .q I are at most 1/2. Since N(e) = (r . The Division Algorithm follows irrunediately once we show a = (p with + q i)f3 + y for sorne y E Z[i] N(y) 1 s "2N(f3) which is even stronger than necessary. Let a = a + bi. Exercise 26 in Section 7. Then y = a. A discrete valuation ring is easily seen to be a Euclidean Domain with respect to the nonn defined by N (O) = O and N = v on the nonzero elements of R. sothat y E Z[i] is a Gaussian integer and a = (p+qi)f3+y. and then the remainder y = a . This is because for a. A diserete valuation on K is a function v : K x ~ Z satisfying (i) v(ab) = v(a) + v(b) (i. so that both Ir .pi and Is . b E R with b = 1O (a) if N(a) < N(b) then a = O· b + a. respectively). 11 (cf.{p+qi)f3.J'=5] is not Euclidean with respect to any norm. Let e = (r . (4) Recall (cf. 3. and (iii) v(x + y) 2: min{v(x). Then in the field a Q(i) we have f3 = r +si where r = (ae +bd)/(e2 +d2) and s = (be .q)2 is at most 1/4 + 1/4 = 1/2. v is a homomorphism from the multiplicative group of nonzero elements of K to Z). An integral domain R is called a discrete valuation ring if there is a valuation v on its field of fractions such that R is the valuation ring of v. For example the ring R of a11rational numbers whose denominators are relatively prime to the fixed prime pEZ is a discrete valuation ring contained in Q. the multiplicativity of the nonn N implies that N{y) = N(e)N(f3) ~ ~N(f3) as claimed. and . We sha11see shortly that Z[. the exercises).e. also the end of Section 3). Note that the algorithm is quite explicit since a quotient p + qi is quickIy deter mined from the rational numbers r and s.ad)/(e2 +d2) are rational numbers. f3 = e + di be two elements of Z[i] with f3 = 1o . The Gaussian integers Z[i] (where D = 1 ).. are a Euclidean Domain with respect to the nonn N (a + bi) = a2 + b2. The set {x E KX I v(x) 2: O} U {O}is a subring of K called the valuation ring of v. as we now show (cf. 7. This proof that Z[i] is a Euclidean Domain can also be used to show that O is a Euclidean Domain (with respect to the field nonn defined in Section 7.
The first implication of a Division Algorithm for the integral domain R is that it forces every ideal of R to be principal.(b) if N(a) 2: N(b) then it follows from property (i) of a discrete valuation that q = ab1 E R. so a = qb + O. .
we see that r must be D. since the results in Section 2. then a + b.. . Suppose 1 = (a + b.. even though the ring Q[x] of polynomials with rational coefficients is a Euclidean Dornain.... This fundamental property of Z was previously determined (in Section 7. Proposition 1 shows that every ideal of Z is principal. let N be the associated field norm N(a+b. a.=5) for sorne a. By the minimality of the norm of d. (2) Let R be the quadratic integer ring Z[.=5) and 2+. Note that these are really the same proof. Thenr = a qd and both a and q d are in 1..=5)a = 1 for sorne y. a E R. If the value is 9 then N (a) = 1 and a = ± 1. One of the fundamenta] consequences of the Euclidean Algorithm in Z is that it produces a greatest common divisor of two nonzero elements. Proof: If 1 is the zero ideal.=5 = ± 1 and the ideal 1 would be the entire ring R..=5].3 ultimately relied on the Euclidean Algorithm in Z.. so r is also an element of l. so a + b.Proposition 1... Taking norms in the first equation gives 9 = N(a)(a 2 + 5b2) and since a2 + 5b2 is a positive integer it rnust be 1. The value cannotbe 3 since there areno integersolutions toa2 +5b2 = 3.=5 would then imply that 2. Proposition 1 can also be used to prove that sorne integral domains R are not Euclidean Domains (with respect to any norm) by proving the existence of ideals of R that are not principal.. b E Z. were principal.. x) is not principal (cf. To show the reverse inclusion let a be any element of 1 and use the DivisionAlgorithmtowritea = qd+r withr = O or N(r) < N(d). Ifthe value is 1. Thus a = qd E (d) showing 1 = (d).=5 = (J(a + b. It follows that 1 is not a principal ideal and so R is not a Euclidean Dornain (with respect lo any norm).... But then 1 would be an elernent of 1.=5 is a multiple of 3 in R.4).. there is nothing to proveo Otherwise let d be any nonzero element of 1 of minimum norm (such a d exists since the set {N(a) I a E 1} has a minimum element by the Well Ordering of Z).=5). More precisely.e. it follows that the ring Z[x] ofpolynomials with irueger coefficients is not a Euc1idean Domain (for any choice of norm).3) from the (additive) group structure of Z... Example 3 at the beginning of Section 7.. which is impossible by the second equation since the coefficients of 2+. Clearly (d) e 1 since d is an element of l. Examples (1) Let R . Every ideal in a Euclidean Domain is principal..=5) generatedby3 and 2+.2+.=5 are not divisible by 3..3. so 3y + (2+...3 or 9.. if 1 is any nonzero ideal in the Euclidean Domain R then 1 = (d). i.. This is true in any . (J E R.=5.=5 = ±3. using the classification of the subgroups of cyclic groups in Section 2.' I = Z[xl: Since the ideal (2... a contradiction...=5) = a2+5b2 andconsidertheideal 1 = (3. 3 = a (a + b. where d is any nonzero element of 1 of minimum norm. Multiplying both sides by 2.
The notion of a greatest common divisor of two elements (if it exists) can be made precise in general rings.Euclidean Domain. .
This explains why the symbol (a. There are rings in which greatest common divisors do not existo This discussion immediately gives the following sufficient condition for the exis tence of a greatest cornmon divisor. Thus R = (1) is the unique principal ideal containing both 2 and x. and (ii) if d' a and d' b then d' A greatest cornmon divisor of a and b will be denoted by g.d. then d is a greatest cornmon divisor of a andbif (i) 1 is contained in the principal ideal (d).  (1) a is said to be a multiple of b if there exists an element x E R with a = bx. b) generated by two elements is principal is called a Bezout Domain. The defining properties (i) and (ii) of a greatest cornmon divisor of a and b translated into the language of ideals therefore become (respectively): if 1 is the ideal of R generated by a and b. In tbis case b is said to divide a or be a divisor of a. We shall see other examples along these lines in Section 3. so 1 is a greatest cornmon divisor of 2 and X. b) is often used to denote both the ideal generated by a and b and a greatest cornmon divisor of a and b. (a. i. if d is any divisor of both a and b then (d) must contain both a and b and hence must contain the ideal generated by a and b. The exercises in this and subsequent sections explore these rings and show that there are Bezout Domains containing nonprincipal (necessarily infinitely generated) ideals. Note that the condition in Proposition 2 is not a necessary condition.1 EudideanDomains 275 .e. (d) = (d'). nonprincipal ideal (cf. the examples in Section 7. If a and b are nonzero elements in the cornmutative ring R such that the ideal generated by a and b is a principal ideal (d). written b a. If two elements d and d' of R generate the same principal ideal.4). In particular.8. Sec. Before returning to Euc1idean Domains we examine the uniqueness of greatest common divisors. or (abusing the notation) simply (a. b E R with b = 1 O. and I (ii) if (d') is any principal ideal containing 1 then (d) e (d').c. Proposition 3. Proposition 2. then d' = ud for sorne unit u in R. : Thus a greatest cornmon divisor of a and b (if such exists) is a generator for the uniqhe smallest principal ideal containing a and b.. b). then d is a greatest cornmon divisor of a and b. in the ring R = Z[x] the elements 2 and x generate a maximal.Definition. Let R be an integral domain. An integral domain in which every ideal (a. Let R be a cornmutative ring and let a. (2) A greatest common divisor of a and b is a nonzero element d such that (i) d a and d b. I I I I I Id . For example. Note that b a in a ring R if and only if a E (b) if and only if (a) e (b). In particular. b).
8 Euclidean. 274 Chap. and Unique Fadorization Domains . then d' = ud for sorne unit U.if d and d' are both greatest common divisors of a and b. Principal Ideal.
once we show that (i) d a and d b (so (a. and (2) the principal ideal (d) is the ideal generated by a and b. Much of the material above may be familiar from elementary arithmetic in the case Sec. ro) e (a.qk+lrk E (rkl. b) e I I (d» (ii) d is an Rlinear combination of a and b (so (d) e (a.qlro E (b. rkl = qk+lrk + rk+l.e. b). b). divides a. It follows from equation (O) that ro E (a. rnl' Clearly rn rn· By induction (proceeding from index n downwards to I I index O) assume rn divides rk+l and ri. Then (1) d is a greatest common divisor of a and b. Then by the (k+l)st equation rk+l = rkl . To prove that r. Proof: By Proposition 1. both x and y are units. Starting from the (n+ 1)st equation. By the (k+l)st equation. xy = 1.e. Both parts of the theorem will follow therefore once we show d = r « generates this ideal. b». n In we see that r. Since d' E (d) there is sorne . and since rn divides both terms on the right hand side we see that rn also divides rkl. b) generated by a and b proceed similarly by induction proceeding from equation (O) to equation (n). Let d = r « be the last nonzero remainder in the Euclidean Algorithm for a and b described at the beginning of this chapter. b). i. b do have a greatest common divisor.xy) = O . Let R be a Euclidean Domain and let a and b be nonzero elements of R. One of the most important properties ofEucIidean Domains is that greatest common divisors always exist and can be computed algorithmically. i. This induction shows rn E (a. is in the ideal (a. the ideal generated by a and b is principal so a.1 EudideanDomains 275 . Thus d = xyd and so d(1.8.. that is. Since d E (d') there is sorne x E R such that d = xd'. rk) e (a. rnl = q +r . namely any element which generates the (principal) ideal (a. The second assertion follows from the first since any two greatest common divisors of a and b generate the same principal ideal (they divide each other). b).. Theorem 4. b) and by equation (1) that rI = b . To prove that d divides both a and b simply keep track of the divisibilities in the Euclidean Algorithm. which completes the proof. there are elements x and y in R such that d=ax+by. By induction assume rkl. In particular. y E R such that d' = yd. d can be written as an Rlinear combination of a and b. rk E (a.Proof: This is cIear if either d or d' is zero so we may assume d and d' are nonzero. b). This proves the first assertion. Since d = f :O . From the 1stequation in the Euclidean AIgorithm we obtain that rn divides b and then from the O thequation we get that r. Thus (i) holds.
For example. since 13 is the greatest common divisor of 2210 and 1131. Principal Ideal. This faet follows quickly from the Euclidean AIgorithm: 2210 = 1 . 1131 + 1079 274 Chap.of the integers Z. if a = 2210 and b = 1131 then the smallest ideal of Z that contains both a and b (the ideal generated by a and b) is 13Z. except possibly for the translation into the language of ideals.8 Euclidean. and Unique Fadorization Domains .
Indeed.1131 = 1·1079+52 1079 = 20 . It is a theorern tbat the nurnber of steps required to determine the greatest cornmon divisor of two integers a and b is at worst 5 times the number of digits of the srnaller of the two nurnbers. tbis algorithm is logarithmic in the size of the integers. Mag. Put another way. Wilson in A principal ideal ring that is not a Euclidean ring. b) for sorne integer m (positive or negative). and use a simplification by Kenneth S. b) = ax' + by'.. 1131) is the last nonzero rernainder. This latter theorem (a proof of which is outlined in the exercises) provides a com plete solution of the First Order Diophantine Equation ax + by = N provided we know there is at least one solution to this equation. Using the procedure of Theorern 4 we can also write 13 as a linear combination of 2210 and 1131 by first solving the next to last equation aboye for 13 = 52 . 39. . There is no uniqueness statement for the integers x and y in (a. 46(1973). i. Math. N is divisible by d. b) = ax + by. We end tbis section with another criterion that can sornetimes be used to prove that a given integral dornain is not a Euclidean Domain.. Williams in Note on nonEuclidean Principal Ideal Domains. this is the same as saying N E (d). 176177. ofideas ofTh. etc. we would have expected at worst only 500 divisions. notice that for the example aboye we would have expected at worst 5 ·4 = 20 divisions (the example required far fewer). The answer in tbis case is 13 = (22) ·2210 + 43·1131.m (a.e.d. the principal ideal generated by the greatest cornmon divisor d of a and b.e. 52 + 39 52 = 1· 39 + 13 39 = 3·13 so that 13 = (2210. The Euclidean AIgorithm in tbe integers Z is extrernely fast. then using previous equations to solve for 39 and 52. of a and b (and then the result quoted aboye gives a full set of solutions of tbis equation).c. x' = x + b and y' = Y . If we had started with integers on the order of 10100 (large numbers by physical standard s).a satisfy (a. Mag. pp.one can prove that if Xo and Yo are solutions to the equation ax + by = N. then any other solutions x and y to tbis equation are of the form b x = Xo + m (a..1 . To obtain an appreciation of the speed implied here. But the equation ax + by = N is simply another way of stating that N is an element of the ideal generated by a and b.. Math. Motzkin. Hence. finally writing 13 entirely in tenns of 2210 and 1131. the equation ax + by = N is solvable in integers x and y if and only if N is divisible by the g. 3438. This is essentially the only possibility . Since we know tbis ideal is just (d). 48(1975). pp. b) a y = Yo .! For any integral dornain let lThe material here and in sorne of the following section follows the exposition by l.
8 Euclidean. and Unique FactorizationDomains .276 Chap.Principal Ideal.
Anelernentu E RR is called a universal side divisor if for every x E R there is sorne z E R such that u divides x . b = 779086434385541 (the Euclidean Algorithm requires only 7 steps for these integers). (a) a = 13. EXERCISES 1. Example We can use Proposition 5 to prove that the quadratic integer ring R = Z[ ( + . equals ±l. In either case the minirnality of u irnplies r E R . n = 3797. We have already determined that ±1 are the only units in R and so R = {O. Hence the only divisors of2 in R are {±1. ±1}.z in R. = 20.O or 2 ± 1 in R.J=19) /2 it is easy to check that none of x. Pro!!j: Suppose R is Euclidean with respect to sorne norm N and let u be an elernent of R . Note that if a. b = 60808. so the technique in the examples following Proposition 1 do not 3'Ply to this ring). For any x E R.e. If 2 = af3 then 4 = N(a)N(f3) and by the remark above it follows that one of a or f3 has norm 1. If R is a Euc1irlean Dornain then there are universal sirle divisors in R. For each of the following five pairs of integers a and b.. so none of these is a universal side divisor. so the only possible values for u are ±2 or ±3. Suppose u E R is a universal side divisor and let N(a + b(1 + .. (e) a = 1891.. b = 372. For each of the following pairs of integers a and n. (e) a = 91442056588823.e. u is a nonunit divisor of 2 or 3 in R. there is a type of "division algorithrn" for u: every x may be written x = q u + z where z is either zero or a unit.R = RXU{O}denotethecollectionofunitsof RtogetherwithO. The existence of universal side divisors is a weakening of the Euclidean condition: Proposition 5. the only divisors of 3 in R are {±l. i. . Let R be an integral dornain that is not a field. Taking x = ( + . Taking x = 2 in the definition of a universal side divisor it follows that u must divide one of 2 . show that a is relatively prime to n and determine the inverse of a mod n (cf. b E Z and b '# O then a2 + ab + 5b2 = (a + b/2)2 + 19/4b2 ::: 5 and so the smallest nonzero values of N on R are 1 (for the units ±1) and 4 (for ±2). x ± 1 are l divisible by ±2 or ±3 in R. i. ±2}. determine their greatest common divisor d and write d as a linear combination ax (a) a (b) a (e) a + by of a and b. = 69. n = 20. write x = qu + r where r is either Oor N(r) < N(u). b = 13. Similarly. a = 507885. 2.R (which is nonernpty since R is not a field) ofminirnal norm. i. ±3}. n = 89. (d) = 11391. (b) a = 69.Hence u is a universal side divisor in R.1.J=19) /2] l is not a Euclidean Domain with respect to any norm by showing that R contains no universal side divisors (we shall see in the next section that a11of the ideals in R are principal.J=19)/2) = a2 + ab + 5~ denote the field norm on R as in Section 7. b = 5673. Section 3 of the Preliminaries chapter).e.
Sec.1 EuclideanDomains 277 .8.
Find a generator for the ideal (85. Let F = Q(J i5 )be a quadratic field with associated quadratic integer ring C J and field norm N as in Section 7. Plotting l the points of C Jin e may be helpful. 1+13i) in Z[i].e. More generally. Prove that the full set of solutions to this equation is given by x=xo+m. 7 or 11. The results of tbe next exercise determine precisely which quadratic integer rings with D < O are Euc1idean. Chap.b can be obtained using only stamps in denominations a and b). then a divides e. 2. Determine al! integer solutions of the following equations: (a) 2x +4y = 5 (b) 17x + 29y = (e) 85x + 145y = 505. Y is a solution to ax + by = N. show that if a divides be with nonzero a.1. 6.(d) a steps for these integers). Y O is a solution: axo + byo = N . b) = 1 and a divides be.b is a positive linear combination of a and b (so every "postage" greater tban ab . 67. show = b(yo  y) and use (a).xo) b (a. (a) Suppose D is 1. thereis aninteger No such that for all N ~ No the equation ax + by = N can be solved with both x and Y nonnegative integers. b then _a_ divides e. Prove that every nonzero element of R of norm m is a unit. by the Euclidean AIgorithm. b are nonzero. Prove that C J is a Euclidean Domain with respect to N. which is less than 1 for these values of D.8 or 163.a . in the terminology of the next section). and 163 are the only negative values of D for which every ideal in C Jis principal (i. Le.. Prove that C Jis not a Euclidean Domain witb 278 Euclidean.. and Unique FactorizationDomains . b) = 600372285 7n. Let R be a Euclidean Domain. 3.I. 7.Principal Ideal. 8.] (b) Suppose that D 31 = 43. 7.a .a . Letm be the minimuminteger in the set of norms of nonzero elements of R. Prove in faet that the integer ab . b) [ti x. 7. 2. as m ranges over the integers. 11 prove that every element of F differs from an element in C Jby an element whose norm is at most ( + IDI) 2j(16IDI). i. Let R be a Euclidean Dornain. Deduce that a nonzero element of norm zero (if such an element exists) is a unit.b cannot be written as a positive linear combination of a and b but that every integer greater than ab . that a(x .. = 776952369 7(3the EuclideanAIgorithm requires only 3 y=yom a (a. 4. b) (a.] 5. [Modify the prooffor Z[i] (D = 1) in the texto For D = 3.13i. (a) Prove that if (a. It is known (but not so easy to prove) that D = 1. 3. 43. 11. Do the same for the ideal (47. 67. b and N are integers and a. 3. 53 + 56i). C Jis a P.e. (The Postage Stamp Problem) Let a and b be two relatively prime positive integers. (b) Consider the Diophantine Equation ax + by = N where a. Prove that every sufficiently large positive integer N can be written as a linear combination ax+by of a andbwherex andy are bothnonnegative. Suppose xo. a greatest common divisor for 85 and 1+ 13i. 19.D.
Ji) is a Euc1ideanDomain with respeet to tbe norm given by the absolute value ofthe field norm N in Seetion 7. Prove that the quotient ring Z[i]/I is finite for any nonzero ideal 1 of Z[i].8 Euclidean..] Prove that the ring of integers C J in tbe quadratic integer ring Q(. and Unique FactorizationDomains . 10.respeet lo any norm. = 19 in the text.1. [Use the faet 278 Chap. [Apply the same proof as for D 9.Principal Ideal.
(b) Deduce that any two nonzero elements in a Euclidean Domain have a least common multiple which is unique up to multiplication by a unit. b) is the greatest common divisor of a and b.I. = Remark: This result is the basis for a standardPublic Key Codeo Suppose N = pq is the product oftwo distinct large primes (each on the order of 100 digits.D. To deeode the message it seems necessary to determine d'...D.I.2 Principal Ideal Domains(P. _ !!:!:_where .2 PRINCIPAL IDEAL DOMAINS (P. whieh requires the determination of tbe value cp(N) = cp(pq} = (p . Let M be an integer relatively prime to N and let d be an integer relatively prime to cp(N). however).D.that 1 = (a) for sorne nonzero a and then use the DivisionAlgorithm in this Euclidean Domain to see that every coset of 1 is represented by an element of norm less than N (a). Proposition 1 proved that every Euclidean Domain is a Principal Ideal Domain so that every result about Principal Ideal Domains automatically holds for Euclidean Domains. which can be unscrambled (decoded) by computing Mf' (mod N) (these powers can be computed quite easily even for large values of M and N by successive squarings). the most naive method of eheeking a11 faetors up to .¡¡.D. Let R be a commutative ring with 1 and let a and b be nonzero elements of R.) is an integral domain in which every ideal is principal. (a) Prove that a least common multiple of a and b (if such exists) is a generator for the unique largest principal ideal contained in (a) n (b). and of course one can always increase the size of p and q).] 11. and (ii) if a e' and b e' then e e'. or approximately 300 ¡ years even at 10 billion computations per second.4 that the polynomial ring Z[x] contains nonprincipal ideals.I.. for example). If M is a message. then MI = Md (mod N) is a scrambled (encoded) version of M. The success of this method as acode rests on the neeessity of determining the fuctorization of N into primes. Definition.1) (no one has as yet proved that there is no other decoding scheme. b) 12.s) 279 . The values of N and d (but not p and q) are made public1y known (hence the name) and then anyone with a message M can send their encoded message Md (mod N). Examples (1) As mentioned after Proposition 1.I. hence is not a P. (2) Example 2 following Proposition 1 showed that the quadratic integer ring Z[ ~] is not a P. (A Public Key Code) Let N be a positive integer. Prove that if MI = =Md (mod N) then M = =Mf (mod N) where d' is the inverse of d mod cp(N): dd' 1 (mod cp(N)}. w ouldhere require on the order of 10100 computations. in faet the ideal (3.I.s) A Principal Ideal Domain (P. 1 + ~) is a nonprincipal ideal. A least common multiple of a and b is an element e of R such that (i) a e and b e. where cpdenotes Euler's cpfunction. the integers Z are a P.8. for which no sufficient1y efficient algorithm exists (for example. It is possible Sec. 8. I I I I I (e) Prove that in a Euclidean Domain the least common multiple of a and b is (a.lD. (a.D. We saw in Seetion 7.l)(q .
From an idealtheoretic point of view Principal Ideal Domains are a natural class of rings to study beyond rings which are fields (where the ideals are just the trivial ones: (O) and (1».8 Euclidean. and Unique Fadorization Domains 280 .e. Proposition 7. however. This useful fact is true in an arbitrary Principal Ideal Domain. i. however. Proposition 6. is that although greatest common divisors exist in both settings. concrete applications ofthese results) are more effectively carried out using a Euclidean Algorithm (if one is available). Recall that maximal ideals are always prime ideals but the converse is not true in general.on the other hand.. in Euclidean Domains one has an algorithm for computing them.4. We shall prove below that the quadratic integer ring Z[ (1 + . 1 +~) and (3. 1 + ~)(3. If. so sm = 1 (recall that R is an integral domain) and m is a unit so 1 = R.for examplethe ideals (3. Then (1) d is a greatest common divisor of a and b (2) d can be written as an Rlinear combination of a and b. (3. It is not true that every Principal Ideal Domain is a Euc1idean Domain. Proof: Let (p) be a nonzero prime ideal in the Principal Ideal Domain R and let 1 = (m) be any ideal containing (p). We must show that 1 = (p) or 1 = R. either r or m must lie in (p). . Chap. Let d be a generator for the principal ideal generated by a and b. there are elements x and y in R with d =ax +by (3) d is unique up to multiplication by a unit of R.e. In tbis case p = rm = psm. i. 1 . which was shown not to be a Euclidean Domain in the previous section. Principal Ideal. Thus (as we shall see in Chapter 12 in particular) results which depend on the existence of greatest common divisors may often be proved in the larger class of Principal Ideal Domains although computation of examples (i. We collect sorne facts about greatest common divisors proved in the preceding section.~) = (3) (cf. Now P E (m) so p = rm for sorne r E R. 1 . Proof: This is just Propositions 2 and 3..~) are both nonprincipaland their product is the principalideal generatedby 3.for the product 1J of two nonprincipalideals 1and J to be principal. Every nonzero prime ideal in a Principal Ideal Domain is a maximal ideal. If m E (p) then (p) = (m) = l. Since (p) is a prime ideal and rm E (p). Many of the properties enjoyed by Euclidean Domains are also satisfied by Principal Ideal Domains.D. Exercise5 and the exampleprecedingProposition 12 below). A significant advantage of Euclidean Domains over Principal Ideal Domains.I. We observed in Section 7.J=19) /2]. r E (p) write r = ps. Let R be a Principal Ideal Domain and let a and b be nonzero elements of R.e. that every nonzero prime ideal of Z is a maximal ideal.. nevertheless is a P. as the following proposition shows.
The last result in this section will be used to prove tbat not every P. pp. Jour.. If R is any commutative ring such tbat tbe polynomial ring R[x] is a Principal Ideal Domain (or a Euclidean Domain). Über eindeutige Zerlegung in Primelemente oder in Primhauptideale in Integritlitsbereichen. Then tbe DedekindHasse condition on N and tbe minimality of b implies tbat a E (b). so tbis is indeed a weakening of tbe Euclidean condition. Principal Ideal Domains are almost Euclidean." Proof: Let 1 be any nonzero ideal in R and let b be a nonzero element of 1 witb N (b) minimal.8.4. hence also a Principal Ideal Domain (this will be proved in tbe next chapter). hence tbe quotient R is a field by Proposition 12 in Section 7. pp. Math. so 1 = (b) is principal. Math. if 1 is an ideal in R (such as tbe ideal (2) in Z) tben tbe ideal (1. (and is equivalent when R is a ring of algebraic integers) is the c1assical Criterion of Dedekind and Hasse. The converse to tbis is also true. b) of norm strictly smaller tban tbe norm of b (i.lD. Corollary 8. The observation that the converse holds generally is more recent and due to John Greene. 159(1928). so that tbe ideal (a. The integral domain R is a P. für die Reine und Angew.I.I. 2That a DedekindHasse norm on R implies that R is a P. hence in general is not principal. Suppose a is any nonzero element in 1. Monthly.D.s) 281 .D.2 Principal Ideal Dornains(P. eitber b divides a in R or tbere exist s.tb) < N(b».I. tER witb O < N(sa . Amer. if F is a field. tben tbe polynomial ring F[x] is a Euclidean Domain. Note tbat R is Euclidean witb respect to a positive norm N if it is always possible to satisfy the DedekindHasse condition witb s = 1. 312. x) in R[x] (such as tbe ideal (2. Define N to be a DedekindHasse nonn if N is a positive norm and for every nonzero a. b E R eitber a is an element of tbe ideal (b) or tbere is a nonzero element in tbe ideal (a. Sec. if and only if R has a DedekindHasse norm. then R is necessarily a field. The converse will be proved in tbe next section (Corollary 16). Definition. Proposition 9.e. Intuitively. Since R is a subring of R[x] tben R must be an integral domain (recall tbat R[x] has an identity if and only if R does). b) is contained in l.. 154156. is a Euclidean Domain and relates tbe principal ideal property witb anotber weakening oftbe Euclidean condition.As we have already mentioned. (x) is a maximal ideal. ef. x) in Z[x]) requires one more generator tban does 1. 104(1997). By Proposition 7.D. Proof: Assume R[x] is a Principal Ideal Domain. The ideal (x) is a nonzero prime ideal in R[x] because R[x]/(x) is isomorphic to tbe integral domain R.
Example
Let R = Z (1+.J=19) [ /2] be thequadratic integerring considered at theend ofthe previous section. We show that the positive field norm N{a + b{l + .J=19)/2) = a2 + ab + 5b2 defined on R is a DedekindHasse norm, which by Proposition 9 and the results of the previous section will prove that R is a P.I,D, but not a Euclidean Domain. Suppose a. f3 are nonzero elernents of R and a/f3 lI . R. We must show that there are elements s, tER with O < N (sa  tf3) < N (f3), which by the multiplicativity of the field norm is equivalent to 0< N{f3S t) Wn'te a
= a+b.J=19
a
< 1. mtegers a, b g no cornmon d"ivisor an d
E 1~f 1 I [ r117:\:119]wíitth 'V
•e h avm.
e with e > 1 (since f3 is assumed not to divide a). Since a. b, e have no cornmon divisor there are integers x, y, z with ax + by + cz = 1. Write ay  19bx = cq + r for sorne quotient q and rernainder r with Irl ~ c/2 and let s = y + x.J=19 and t = q  z.J=19. Then a quick computation shows that a (ay  19bx  cq)2 + 19{ax + by + CZ)2 1 19 2 O < N{s  t) = < + f3 c2 c  4 and so (*) is satisfied with this s and t provided e ~ 5. Suppose that e = 2. Then one of a, bis even and the other is odd (otherwise a/ f3 E R), . (a  1) + b.J=19 and then a quick check shows that s = 1 and t = 2 are elernents of R satisfying (e). Suppose that e = 3. The integer a2 + 19b2 is not divisible by 3 {modulo 3 this is a2 + b2 which is easily seen to be Omodulo 3 if and only if a and b are both Omodulo 3; but then a, b, e have a cornmon factor}. Write a2 + 19b2 = 3q + r with r = 1 or 2. Then again a quick check shows that s = a  b.J=19, t = q are elements of R satisfying
(*).
fJ
Finally, suppose that e = 4, so a and b are not both even. If one of a, b is even and the other odd, then a2 + 19b2 is odd, so we can write a2 + 19b2 = 4q + r for sorne q, r EZ and O < r < 4. Then s = a  b.J=19 and t = q satisfy {*}. If a and b are both odd, then a2 + 19b2 = 1 + 3 rnod 8, so we can write a2 + 19b2 = 8q + 4 for sorne q E Z. Then a  b.J=19 s= 2 and t = q are elernents of R that satisfy (*).
EXERCISES
(a) and (b) are comaximal (cf. Section 7,6) if and only if a greatest cornmon divisor of a and b is 1 (in which case a and b are said to be coprime or relatively prime). 2. Prove that any two nonzero elements of a P.I.D. have a least cornmon multiple (cf. Exercise 11, Section 1).
1. Prove that in a Principal Ideal Dornain two ideals
3. Prove that a quotient of a P.I.D. by a prime ideal is again a P.I.D. 4. Let R be an integral domain. Prove that if the following two conditions hold then R is a Principal Ideal Domain: (i) any two nonzero elements a and b in R have a greatest cornmon divisor which can be
written in the form ra
+ sb for sorne r, s E R, and
282
Chapo8 Eudidean, Principal Ideal, and Unique FactorizationDom ains
(ii) if al,
are nonzero elements of R such that ai +1 I ai for a11i, then there is a positive integer N such that an is a unit times an for a11n ::: N.
oz. a3, . ..
5. Let R be the quadratic integer ring íZ[.J=5]. Define the ideals I: = (2,1+ .J=5), 13 = (3, 2 + .J=5), and I~ = (3, 2  .J=5). (a) Prove that h 13, and I~ are nonprincipal ideals in R. [Note that Example 2 following . Proposition 1 proves tbis for 13.] (b) Prove that the product of two nonprincipal ideals can be principal by showing that is the principal ideal generated by 2, i.e., = (2).
li
II
(e) Prove similarlythat E = (1.J=5)andh/~ rj = (l+.J=5) that the principal ideal (6) is the product of 4 idea1s: (6) =
li 13I!~.
are principal. Conc1ude
6. Let R be an integral domain and suppose that every prime ideal in R is principal. Tbis exercise proves that every ideal of R is principa1. i.e., R is a P.I.D. (a) Assume that the set of idea1s of R that are not principal is nonempty and prove that tbis set has a maximal e1ement under inc1usion (which, by hypothesis, is not prime). [Use Zorn's Lemma.] (b) Let I be an ideal wbich is maximal with respect to being nonprincipal, and let a, b E R with ab E I but a fj. I and b fj. l. Let la = (1, a) be the ideal generated by I and a, let lb = (1, b) be the ideal generated by I and b, and define J = {r E R I r I¿ el}. Prove that la = (a) and J = (f3) are principal ideals in R witb I ~ lb ~ J and laJ = (af3) e l. (e) If x E I show that x = sa for sorne s EJ. Deduce that I = la J is principal, a contradiction, and conc1ude that R is a P.I.D. 7. An integral domain R in whicb every ideal generated by two elements is principal (i.e., for every a, b E R, (a, b) = (d) for sorne d E R) is called a Bezout Domain. [cf. a1so Exercise 11 in Section 3.J (a) Prove that the integral domain R is a Bezout Domain if and on1y if every pair of elements a, b of R has a g.c.d. d in R that can be written as an Rlinear combination of a and b, i.e., d = ax + by for sorne x, y E R. (b) Prove that every finitely generated ideal of a Bezout Domain is principal. [cf. the exercises in Sections 9.2 and 9.3 for Bezout Domains in which not every ideal is principal.] (e) Let F be the fraction field ofthe Bezout Domain R. Prove that every element of F can be written in the form afb with a. b E R and a and b relatively prime (cf. Exercise 1). 8. Prove tbat if R is a Principal Ideal Domain and D is a multiplicatively closed subset of R, then D1 R is a1so a P.I.D. (cf. Section 7.5).
8.3 UNIQUE FACTORIZATION DOMAINS (U.ED.s )
In the case of the integers Z, there is another method for determining the greatest common divisor of two elements a and b familiar from elementary arithmetic, namely the notion of "factorization into primes" for a and b, from which the greatest common divisor can easily be detennined. This can also be extended to a larger class of rings called Unique Factorization Domains (U.F.D.s)  these will be defined shortly. We shall then prove that
every Principal Ideal Domain is a Unique Factorization Domain
Sec.8.3 Unique FactorizationDomains(U.F.D.s)
283
so that every result about Unique Factorization Domains wiU automatically hold for both Euclidean Dornains and Principal Ideal Dornains. We first introduce sorne tenninology. Definition. Let R be an integral dornain. (1) Suppose r E R is nonzero and is not a unit. Then r is called irreducible in R if whenever r = ab with a, b E R, at least one of a or b must be a unit in R. Otherwise r is said to be reducible. (2) The nonzero element p E R is called prime in R if the ideal (p) generated by
p is a prime ideal. In other words, a nonzero element p is a prime if it is not a unit and whenever p ab for any a, b e R, then either p la or p b. (3) Two e1ernentsa and b of R differing by a unit are said to be associate in R (i.e., a = ub for sorne unit u in R).
I
I
Proposition 10. In an integral domain a prime elernent is always irreducible. Proof: Suppose (p) is a nonzero prime ideal and p = abo Then ab = p E (p), so by definition of prime ideal one of a or b, saya, is in (p). Thus a = pr for sorne r. This implies p = ab = prb so rb = 1 and b is a unit. This shows that p is irreducible. It is not true in genera] that an irreducible elernent is necessarily prime. For example, consider the e1ernent 3 in the quadratic integer ring R íl.[ ~]. The cornputations in Section 1 show that 3 is irreducible in R, but 3 is not a prime since (2+ .J= 5")(2.J= 5") 32 is divisible by 3, but neither 2+ .J= 5" 2.J= 5" is divis = nor ible by 3 in R. If R is a Principal Ideal Domain however, the notions of prime and irreducible elements are the same. In particular these notions coincide in íl. and in F[x] (where F is a field).
=
Proposition 11. In a Principal Ideal Domain a nonzero element is a prime if and only if it is irreducible. Proof: We have shown aboye that prime implies irreducible. We must show con versely that if p is irreducible, then p is a prime, i.e., the ideal (p) is a prime ideal. If M is any ideal containing (p) then by hypothesis M = (m) is a principal ideal. Since p E (m), p = rm for sorne r. But p is irreducible so by definition either r or m is a unit. This means either (p) = (m) or (m) = (1), respectively. Thus the only ideal s containing (p) are (p) or (1), i.e., (p) is a maximal ideal. Since maximal ideals are prime ideals. the proof is complete. Example
Proposition 11 gives another proof tbat the quadratic integer ring Z[ ~] since 3 is irreducible but not prime in this ringo is not a P.I.D.
284
Chap.8
Euclidean.Principal Ideal. and Unique FactorizationDomains
The irreducible elements in the integers Z are the prime numbers (and their nega ves) familiar from elementary arithmetic. qm is another faetorization of r into irreducibles.ED. = eñnítíen. we find 1131 = 3· 13·29. . Z and F[x] where F is a fíeld are both Unique Factorization Domains).. The integer 2 is a prime. General rings with the analogous property are given a name.s) 285 . where i2 = 1.. then it in tum can be faetored into two (smaller) integers. xamples (1) A field F is trivially a Unique Factorization Domain since every nonzero element is a unit. If they are oth primes. (3) We shall also prove in the next chapter that the ring R [x] of polynomials is a Unique Factorization Domain whenever R itself is a Unique Factorization Domain (in contrast to the properties ofbeing a Principal Ideal Domain or being a EuclideanDomain. n. The restriction to positive integers is to avoid considering le faetorizations (3)(5) and (3)(5) of 15 as essentially distinct. Similarly. . 2. i. This result together with the preceding example will show that Z[x] is a Unique Factorization Domain. Herethealgorithm rminates. as follows.D. if n = 2210. we shall prove shortly that every Principal Ideal Domain is a Unique Factorization Domain (so. since both 13 and 17 are primes. If n is not itself a prime then by definition it is possible to write = nirn for two other integers nI and n2 neither of which is a unit. we bave already written n as a product of primes. and so we have written n as a product of primes. but of course this need not be the case generally. then m = n and there is sorne renumbering of the faetors so that Pi is associate to q¡ for i = 1. b E Z}. (2) As indicated above... is an integral domain but not a Unique Factorization Domain (rings of this nature were íntroduced in Exercise 23 of Section 7.1). 1105. In these examples eaeh ime occurs only to the first power.) is an integral domain R in which rery nonzero element r E R which is not a unit has the following two properties: (i) r can be written as a finite product of irreducibles Pi of R (not necessarily distinct): r = PIP2 .F. but in fact this decomposition is unique in the sense that any two prime fae rizations of the same positive integer n differ only in the order in which the positive ime faetors are written...3 Unique FactorizationDornains(U. Theinteger5isprime. since we can write n = 2 . the algorithm above proceeds as follows: n is not self prime. and two integers a and b are associates of le another if and only if a = ±b. This uniquefac vization property of Z (which we shall prove very shortly) is extremely useful for the ithmetic of the integers. 8. . A Unique Factorization Domain (U. The elements 2 and 2i are e. but 1105 is not: L05 = 5·221. we must at sorne point be left only with prime reger faetors.e. This gives the prime factoriuuion of 2210 i 2210 = 2·5· 13· 17. Since integers cannot ecrease in absolute value indefinitely. In the ring Z not only is it true that every integer n can be written as a product of imes. (4) The subring of the Gaussian integers R = Z[2i] = {a + 2bi I a. if r = qlq2 .but221isnot: 221 13·17.. For example. In the integers Z any integer n can be written as a product of primes (not necessarily . which do not carry over from a ring R to the polynomial ring R[x ]). If one of nI or n a is not ime. P» and (ii) the decomposition in (i) is unique up to associates: namely.stinct). so there are no elements for which properties (i) and (ii) must be verified. in particular. neither of hich is ± 1. Both n) and na must be smaller in absolute value tban n itself.
we must show tbat p divides either a or b.... While the elements of the quadratic integer ring O need not have unique factor ization.[. But tben P divides a...[.Principal Ideal.. Pn for u a unit and sorne (possibly empty set of) irreducibles P2. One migbt tbink tbat we could deduce Proposition 11 from tbis proposition togetber with tbe previously mentioned tbeorem (tbat we sball prove shortly) tbat every Principal Ideal Domain is a Unique Factorization Domain.. The principal ideal (6) in 71. . We may assume tbat p is associate to one of tbe irreducibles in tbe factorization of a..r:::s) with P2P3 = (1 .. and Unique FactorizationDomains . Proof: Let R be a Unique Factorization Domain. To say tbat p divides ab is to sayab = pe for sorne e in R. Since by Proposition 10.. completing the proof. Pn.r:::s] be interpreted as arising can from two rearrangements of this product of ideal s into products of principal ideals: the product of = (2) with P3P3 = (3). Z[i].... (5) The quadratic integer ring 71. . I 286 Chap. (2i) has two distinct factorizations in R. i.8 Euclidean. One may also check directly that 2i is irreducible but not prime in R (since RI(2i) '" 71. [.R. and the product of P2P3 = (1 + .. Proposition 12. primes of R are irreducible it remains to prove tbat each irreducible element is a prime.. It was the failure to have unique factorization into irreducibles for elements in algebraic integer rings in number theory that originally led to the definition of an ideal. Writing a and b as a product of irreducibles..)I.r:::s) gives two distinct factorizations of 6 into irreducibles... since a = pd witb d = U P 2 . and 4 = 2·2 = (2i) ..3 (one in which we do not have to check that 2 and 2i are irreducibles). b E R.. n the larger ring of Gaussian integers.e..16) that every ideal in O can be written uniquely as a product of prime ideals.1471.r:::s] can be written as a product of 4 nonprincipal prime ideals: (6) = p } P3P~ and the two distinct factorizations of the element 6 in 71.[2i] is not a Unique Factorization Domain at the end of Section 9.r:::s] another example of an integral domain that is is not a Unique Factorization Domain. In a Unique Factorization Dornain a nonzero element is a prime if and only if it is irreducible..r:::s) (cf.irreducibles which are not associates in R since i f J . 3 = (1 + .r:::s)(1 . pi The first property of irreducible elements in a Unique Factorization Domain is tbat tbey are also primes. however Proposition 11 will be used in tbe proof of the latter theorem.. it is a theorem (Corollary 16. that a can be written as a product a = (up) P2 ..... The unique factorization of ideals into the product of prime ideals holds in general for rings of integers of algebraic number fields (exam ples of which are the quadratic integer rings) and leads to the notion of a Dedekind Domain considered in Chapter 16. we see from tbis last equation and from tbe uniqueness of tbe decomposition into irreducibles of ab tbat tbe irreducible elernent p must be associate to one of tbe irreducibles occurring eitber in tbe factorization of a or in tbe factorization of b. (which is a Unique Factorization Domain) 2 and 2i are associates since i is a unit in this larger ringo We shall give a slightly different proof that 71.. P n . Exercise 8).. The resulting uniqueness of the decomposition into prime ideals in these rings gave the elements of the ideals an "ideal" (in the sense of "perfect" or "desirable") behavior that is the basis for the choice of terminology for these (now fundamental) algebraic objects.. Let P be an irreducible in R and assume p ab for sorne a. since 6 = 2 .