Frontispiece: let A be a ring and M an A-module...London Mathematical Society Student Texts 29 Undergraduate Commutative Algebra Miles Reid University of Warwick NIVERSITY PRESSPublished by the Press Syndicate of the University of Cambridge The Pitt Building, Trumpington Street, Cambridge CB2 1RP 40 West 20th Street, New York, NY 10011-4211, USA 10 Stamford Road, Oakleigh, Melbourne 3166, Australia © Cambridge University Press 1995 First published 1995 Printed in Great Britain at the University Press, Cambridge A catalogue record for this book is available from the British Library Library of Congress cataloging in publication data Reid, Miles (Miles A.) ‘Undergraduate commutative algebra / Miles Reid. p. cm, - (London Mathematical Society student texts; 29) Includes bibliographical references ISBN 0 521 45255 4. - ISBN 0 521 45889 7 (pbk.) 1. Commutative algebra. I. Title. II. Series. QA251.3.R45 1995 512'.24--de20 94-27644 CIP ISBN 0521452554 hardback ISBN 0521458897 paperbackContents Frontispiece: let A be a ring and M an A-module... Mlustrations Preface 0 Hello! 0.1 Where we're going 0.2 Some definitions 0.3. The elementary theory of factorisation 0.4 A first view of the bridge 0.5 The geometric side — the case of a hypersurface 0.6 Z versus k[X] 0.7 Examples 0.8 Reasons for studying commutative algebra 0.9 Discussion of contents 0.10 Who the book is for 0.11 What you’re supposed to know Exercises to Chapter 0 1 Basics 1.1 Convention 1.2 Ideals 1.3 Prime and maximal ideals, the definition of Spec A 1.4 Easy examples 1.5 Worked examples: Spec k[X,Y] and Spec Z[X] 1.6 The geometric interpretation 1.7 = Zorn’s lemma 1.8 Existence of maximal ideals vii page iv xi sf Nawwnnee iid mown se PNYONNNYYD Eee SaAAanvF Seonviii 1.9 1.10 AL 1.12 1.413 1,14 15 Contents Plenty of prime ideals Nilpotents and the nilradical Discussion of zerodivisors Radical of an ideal Local ring First examples of local rings Power series rings and local rings Exercises to Chapter 1 2 21 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10 Modules Definition of a module Harmless formalism The homomorphism and isomorphism theorems Generators of a module Examples The Cayley—Hamilton theorem The determinant trick Corollaries - Nakayama’s lemma Exact sequences Split exact sequences Exercises to Chapter 2 3 3.1 3.2 3.3 3.4 3.5 3.6 Noetherian rings The ascending chain condition Noetherian rings Examples Noetherian modules Properties of Noetherian modules The Hilbert basis theorem Exercises to Chapter 3 4 4.1 4.2 4.3 4.4 45 46 AT 48 49 Finite extensions and Noether normalisation Finite and integral A-algebras Finite versus integral Tower laws Integral closure Preview: nonsingularity and normal rings Noether normalisation Proof of Claim Another proof of Noether normalisation Field extensions 27 27 28 29 31 31 32 33 37 37 37 38 40 Al Al 43 43 45 46 49 49 50 51 52 53 54 55 58 59 60 61 61 62 63 64 65 664.10 Contents The weak Nullstellensatz Exercises to Chapter 4 5 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12 5.13 5.14 The Nullstellensatz and Spec A Weak Nullstellensatz Maximal ideals of k[X,,...,Xn] and points of k” Definition of a variety Remark on algebraically nonclosed k The correspondences V and I The Nullstellensatz Irreducible varieties The Nullstellensatz and Spec A The Zariski topology on a variety The Zariski topology on a variety is Noetherian Decomposition into irreducibles The Zariski topology on a general Spec A Spec A for a Noetherian ring Varieties versus Spec A Exercises to Chapter 5 6 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 Rings of fractions S~A and localisation The construction of S-1A Easy properties Ideals in A and S-1A Localisation Modules of fractions Exactness of S-! Localisation commutes with taking quotients Localise and localise again Exercises to Chapter 6 7 7A 7.2 7.3 74 15 76 77 78 7.9 Primary decomposition The support of a module Supp M Discussion Definition of Ass M Properties of Ass M Relation between Supp and Ass Disassembling a module The definition of primary ideal Primary ideals and Ass Primary decomposition ix 67 67 70 70 70 71 72 72 73 74 75 75 76 76 77 78 80 82 84 84 86 87 88 89 90 91 92 92 95 96 97 98 99 100 103 103 105 105x 0 Contents 7.10 Discussion: motivation and examples 7.11 Existence of primary decomposition 7.12 Primary decomposition and Ass(A/J) 7.13 Primary ideals and localisation Exercises to Chapter 7 8 DVRs and normal integral domains 8.1 Introduction 8.2 Definition of DVR 8.3 A first criterion 8.4 The Main Theorem on DVRs 8.5 General valuation rings 8.6 Examples of general valuation rings 8.7 Normal is a local condition 8.8 A normal ring is a DVR in codimension 1 8.9 Geometric picture 8.10 Intersection of DVRs 8.11 Finiteness of normalisation 8.12 Proof of Theorem 8.11 8.13 Appendix: Trace and separability Exercises to Chapter 8 9 Goodbye! 9.1 Where we've come from 9.2 Where to go from here 9.3 Tidying up some loose ends 9.4 Noetherian is not enough 9.5 Akizuki’s example 9.6 Scheme theory 9.7 Abstract versus applied algebra 9.8 Sketch history 9.9 The problem of algebra in teaching 9.10 How the book came to be written Exercises to Chapter 9 Bibliography Index 106 108 109 109 110 112 112 113 113 114 116 117 118 119 121 121 122 123 124 126 129 129 130 132 135 139 141 142 143 144 145 146 149 150Illustrations Frontispiece: let A be a ring and M an A-module... iv 0.5 Quadric cone with a line 6 0.7 The cuspidal cubic and Spec Z[V—3] 9 0.15 Smallest residue modulo Z{i] and Z[w] 7 1.6 Spec k[X, Y] and SpecZ[Y] 24 1.11 The plane curves defined by XY = 0 and X?=0 30 4.8 Projecting the hyperbola XY = 1 66 6.8 The spectrum of a local ring Spec Am 91 7.2 An A-module M gives rise to the family M = {Mp} for P & Spec A, which we can view as fibred over Spec A. 98 7.5 Maximal irreducible closed sets of Supp M are in Ass M; if A is Noetherian and M is finite, these are the irreducible components of Supp M. 101 7.10 The ideal J = (X?,XY) = (X,Y)? N(X) of functions vanishing on the Y-axis and at (0,0) with multiplicity 2. 107 9.4 Nagata’s leaf 138 xiPreface These are notes from a commutative algebra course taught at the University of Warwick sev- eral times since 1978. In addition to standard material, the book contrasts the methods and ide- ology of abstract algebra as practiced in the 20th century with its concrete applications in algebraic geometry and algebraic number theory. xiiiHello! This chapter contains a preliminary discussion of the aims and philo- sophy of the book, and is not logically part of the course. Some of the material may be harder to follow here than when it is treated more formally later in the book, so if you get stuck on something, don’t worry too much, just skip on to the next item. 0.1 Where we're going The purpose of this course is to build one of the bridges between alge- bra and geometry. Not the Erlangen program (linking geometries via transformation groups with abstract group theory) but a quite different bridge linking rings A and geometric objects X; the basic idea is that it is often possible to view a ring A as a certain ring of functions on a space X, to recover X as the set of maximal or prime ideals of A, and to derive pleasure and profit from the two-way traffic between the different worlds on each side. Algebra here means rings, always commutative with a 1, and usually closely related to a polynomial ring k[x,,.-.,2n] or Z[zi,.. -, Zn} over a field k or the integers Z, or a ring obtained from one of these by taking a quotient by an ideal, a ring of fractions, a power series completion, and so on; also their ideals and modules. In this book, A usually stands for a ring and k for a field, and I sometimes use these notations without comment. We're interested in questions such as zerodivisors (that is, 0 4 z,y € A such that zy = 0 € A), factorisation (that is, writing a € Aasa product a = b[]p? with b invertible and p; prime elements), similar questions for ideals, prime ideals, extension rings A C B, etc. The study of rings of this type includes most of algebraic number theory and a large fraction of algebraic geometry. The methods for 12 0 Hello! studying them, by and large, are either simple algebraic arguments, or depend on the link with geometry which I want to introduce here. Thus, for example, rings have a dimension theory, in which dim k[x,,...,2n] = n and dim Z[z1,...,2n] = +1 (yes, n +1 is right!), and already the language suggests a cocktail of two different subjects. The same holds for local ring, an idea at the very heart of commutative algebra. 0.2 Some definitions Before describing briefly the geometric side and some aspects of the bridge, I recall a few very elementary algebraic topics and introduce some definitions, which I hope are mostly already familiar. Let A be a ring, commutative with a 1. Zerodivisors of A are nonzero elements x, y € A such that zy = 0. If A has no zerodivisors and A # 0, it is an integral domain; note that 0 # 1 is part of the definition of integral domain. An integral domain is contained in a unique field K such that every element of K is a fraction a/b with a,b € A and b # 0; this is the field of fractions of A, sometimes written K = Frac A, and I assume you understand its construction. An element x € A is invertible or a unit of A if it has an inverse in A, that is, there exists y € A such that zy = 1. An element x € A is nilpotent if x" = 0 for some n. Prove for yourself that x nilpotent implies that 1 —z is invertible in A, [Hint: write out (1 —)~? as a power series.] Prove also that x and y nilpotent implies that az + by is nilpotent for all a,b € A, so that the set of nilpotent elements of A is an ideal, the nilradical nilrad A. [Hint: use the binomial theorem.] An element x € A is idempotent if x? = x. Obviously if x is idempotent then so is x’ = 1— 2, and then x + 2’ = 1 and zz’ = 0 (please check all this for yourself), so that z and x’ are complementary orthogonal idempotents, now by writing a = az + az’ for any a € A, you see that A is a direct sum of rings A = A, @ Ap, where A, = Az and Ag = Ar’. 0.3 The elementary theory of factorisation Suppose that A is an integral domain. A nonzero element x € A is irreducible if x itself is not invertible, and x = yz with y,z € A implies that either y or z is invertible. x € A is a prime element if it is not a unit, and x | yz implies either x | y or z | z. It is trivial to see that prime implies irreducible, but the other way round is false in general.0.4 A first view of the bridge 3 Aisa UFD (unique factorisation domain) if (i) every element zx factors as a product of finitely many irreducibles « = [| x; with 2; irreducible, and (ii) irreducible implies prime. Proposition In a UFD A, the expression of x = b]|p}* as a product of irreducibles with p; { p; is unique (up to invertible elements). T assume you know this. Otherwise, see any textbook on algebra, for example, [C], [H & H], Chapter 4 or [W]. In the following sections, I need to assume that the polynomial ring k[,...,2n] is a UFD; the proof of this is discussed in Exs. 0.8-9 below. 0.4 A first view of the bridge For simplicity, and to be able to describe in a few intuitive words a rep- resentative case of the geometric side, suppose that k is an algebraically closed field, for example k = C. Then the polynomial ring k[z1,...,2n] is a ring of functions on k”, because a polynomial g € k[z,...,£p] is a function g = g(z1,...,Zp) Of 21,...,2n. Moreover, evaluating a poly- nomial at a point P = (a),...,@n) € k” determines a homomorphism k[zi,..-,2n] — & defined by g +> 9(P), whose kernel is the maximal ideal mp = (x — 41,...,&n —@n) (see Ex. 1.15). This is the correspon- dence between a ring A and a space X in ideal form: A = k[x,...,2n] is the ring of polynomial functions on X = k", and the points of X correspond to maximal ideals of A. 0.5 The geometric side — the case of a hypersurface Suppose that 0 4 F € k[x,..., 2p]; then the locus X =V(F) = {P= (a1,...,an) | F(P) = 0} ck” is a hypersurface. It is (n — 1)-dimensional because you can (almost always) use the equation F = 0 to solve for a, in terms of a2,...,Gn. Now consider the quotient ring A = k[x1,.-.-,2n]/(F), that is, the ring of residue classes modulo the ideal generated by F. Then an element g € A defines a k-valued function on X: indeed, if g is the class in A of a polynomial g € k[z1,...,2n] then for x € X the value g(x) = 9(z) does not depend on the choice of 9. Now we can see (fairly light-weight) traffic passing over the bridge. First, to what extent can A be viewed as a ring of functions on X?0 Hello! (i) If F has no multiple factors, say F = [] fi with fi { f,, then it can be shown that F generates the ideal of all functions vanishing on X. (You can try the proof as an exercise after Chapter 5, see Ex. 5.5.) It follows from this that an element g € A is uniquely determined by the corresponding function g: X — k, so that A is contained in the ring of k-valued functions on X. On the other hand, if F has a multiple factor, say F = f*g with k > 2 then also fg = 0 everywhere on X, and hence F does not generate the ideal of all functions vanishing on X. At the same time, A has nonzero nilpotent elements (because z = im fg € A satisfies z* = 0). In this case, it is not reasonable to try to view the nilpotent element z as a function on X, because it is zero everywhere on X. Thus F has a multiple factor <=> more functions vanish on X than (F) <=> A has nilpotent elements <=> Ahas nonzero elements that are 0 as functions on X. (ii) If F has a factorisation F = f, fo, where fy, fe are polynomials with no common factors, then A has zerodivisors (because x = im fi, Z2 = im fz satisfy 2; # 0,22 4 0 but x22 = 0); this corresponds to a decomposition X = X, U X2 of X as a union of two smaller hypersurfaces X; given by f; = 0 fori = 1,2. Thus A has zerodivisors (not nilpotents) <=> X is reducible: X = X1 U Xo. That is, something in algebra equals something in geometry. (iii) I mentioned complementary orthogonal idempotents and direct sums of rings in 0.2; you can’t get much more abstract algebraic than that. However, it is easy to see that A has nontrivial idem- potents if and only if X is a disjoint union of two hypersurfaces, X = X,UX2. If k =C, this just means that X C C” is a dis- connected topological space; you can’t get much more geometric than that. The ring of functions (say, continuous) on a discon- nected space X = XUX~ is a direct sum of the rings of functions on X and X92. (iv) We will see that there is a close relation between ideals Jc A and subvarieties of X; we can already see that if I C A is an ideal then it defines a subvariety V(I) C X, the subset of P € X where0.6 Z versus k[X] 5 f(P) =0 for all the functions f € J. But this is quite a long story that I defer until later. For the time being, I state without proof the following result (a special case of the weak Nullstellensatz, see Theorem 4.10 and 5.1). Proposition Marimal ideals of A are in one-to-one correspondence with points P € X. That is, P=(ai,.--,4n)€X @ mp = (1 -401,.-.,2n— Gn) C A. To repeat my refrain, something in algebra (the maximal ideals of A) equals something in geometry (the points of X). (v) Assume that F € k[x1,..-,2n] is irreducible, so that A is an integral domain. When is A = k{,,...,2n]/(F) a UFD? For example, if F = xz — y? € k{z,y, 2] then zz = y? holds in the quotient ring A, whereas it is not hard to check that x, y, z are irreducible; therefore A is not a UFD. Now draw the picture of the locus X : (zz = y*), which is the ordinary quadric cone (see Figure 0.5). I will come back to this picture several times later in the book. Observe that X is a cone, and so contains lots of lines, for example, the lines L, C X defined by x = ?z,y = Az; these are codimension 1 subvarieties of X. I take A = 0 to simplify the notation, so consider the line L = Lo C X defined by z = y = 0. The special feature of X is that the ideal J, C A of functions vanishing on L is not generated by one element. In fact I; = (z,y), but y also vanishes along a second line y = z = 0, whereas x vanishes along L with multiplicity 2. Geometrically, this corresponds to the fact that the plane z = 0 is everywhere tangent to X along L; or, to put it another way, at any point where z # 0, I have x = y?/z. In this sense, L is not locally defined by one equation. In other words, the geometric question of codimension 1 sub- varieties and how to define them by equations is closely related to the algebraic question of unique factorisation in the ring A. 0.6 Z versus k[X] The comparison between the ring of integers Z and the polynomial ring k[X] in a single variable over a field k is one of the central points to be made at the outset in a commutative algebra course. From the algebraic6 0 Hello! 2 Xixz=y Figure 0.5. Quadric cone with a line point of view, these two rings are very similar in many formal respects, and yet they are very different in substance. (Compare also Exs. 0.10-12 and the worked example 1.5.) Points of similarity Recall that Z and k[X] are both Euclidean rings, that is, integral domains satisfying “division with remainder” or the “Buclidean algorithm”: for any a, b, there exists an expression @ = bg+r with r less than b. The ideal theory of the two rings proceeds in parallel from this fact: from division with remainder it follows easily that every ideal is principal (generated by one element), either (0) or (f) for an element f. I assume you know all this (see Exs. 0.1-9), including how you deduce the familiar af + bg = h property of the highest common factor h = hef(f,g), unique factorisation, etc.; if not, see, for example, (C] or [H & H], Chapter 4. Points of difference Obviously Z and k[X] are as different as chalk and cheese, but it is worthwhile trying to pin down the difference. I give two illustrations. First an algebraic statement: k[X] contains a field, whereas Z doesn’t. As you know, for any ring A, there is a unique0.7 Examples 7 homomorphism t: Z — A: after taking 1 + 1,, the rest is forced. If A contains a field then ¢ factors through the prime subfield, either F, or Q. Ifa ring A contains a field k as a subring, then A and any A-module are k-vector spaces. In the case A = Z, there is obviously no way of embedding either F, or Q into A. The same holds for A = Z/ (p?), with pa prime number: the additive group Z/(p) is not a vector space over any field; see Ex. 0.10. For this reason, one sometimes says that a ring containing a field k has equal characteristic char k (either 0 or p), whereas ating like Z has unequal characteristic, in the sense that the ring itself has characteristic 0, but it has residue fields F, of characteristic p. Here is another difference: k[X] contains variables. To put it alge- braically, a typical maximal ideal of k[X] is (X), and it makes sense to differentiate with respect to X; that is, there is a k-linear map “a k[X] > k[X], defined by 2+ nz"? for all n, with the properties and applications that you know about. By contrast, the maximal ideals of Z are (2), (3), (5), ete., and ddd @2’ 3’ db" are of course completely meaningless. There is no nonzero derivation of Z to anything. To put it another way, multiplication by a natural number n is the additive operation a + a+----+ a (with n summands), and therefore this operation, and the ideal (n) generated by n, are already determined by the additive structure. 0.7 Examples Recall that we intend to study extension rings of Z and of k, and that the distinctions in 0.6 will carry over to these. I continue the theme of 0.6 with two slightly more substantial examples from algebraic geometry and algebraic number theory illustrating this and other points. Example 1 Suppose that k is an algebraically closed field of charac- teristic # 2, and let A be the ring A = k[X,Y]/(¥? - X°). By the correspondence of 0.5, A is a ring of functions on the plane curve Ccr given by Y? = X°. See Figure 0.7(a). Now A can be viewed as the extension ring k[X][VX®] obtained by adjoining the square root of X® to k[X]. If (X — a) is a maximal ideal8 0 Hello! of k[X] with a € k, it is contained in maximal ideals (X — a, ¥ — §); corresponding to the square roots 6 = +Vo3; there are obviously two of: these if a # 0, and one otherwise. These ideals require two generators, Moreover, it is easy to see that the elements X and Y are irreducible in A, but not prime: indeed, Y? = X° in A, so that X | ¥?, but X + Y. At this point, observe that I am doing something fairly silly by taking the square root of X°. It is clearly much more sensible to take the square root of X instead. Let A’ = k{t] where t = Y/X = VX; this is a slightly bigger ring that A. Then in it, X = ¢? and Y = ¢® € A’ = K{d], so that A’ is just a polynomial ring, so of course it is a UFD, and every ideal is principal. Example 2 Now consider B = Z[/—3], the extension ring obtained by adjoining /—3 to Z. What are its maximal ideals? If P is a nonzero prime ideal of B then P/Z is a nonzero prime ideal of Z, so of the form (p); we say that P lies over p. I check first that every prime number p # 2,3 either splits as a product p= f, f— of two prime elements of B, or remains a prime element of B, and, in particular, any prime ideal of B not lying over 2 is principal. Indeed, any p = 1 mod6 can be written as p = 3a? +b? with a,b € Z; this was proved in 1760 by Euler, and is worked out in Ex. 0.14. Thus p factors in B as a product of two irreducible elements p = f, f-, with f4 = b+ a/—3; for example, 7 = (2+ V—3) (2-v-3). It is easy to check that B/(f1) & F,, so that f, are two prime elements of B lying over p. If p = 5mod6 then X? + 3 is irreducible in F,[X] by quadratic reciprocity (see Ex. 0.13), and B/(p) © F,[X]/(X? +3) = Fpz is a field, so that p is a prime element of B. It is also not hard to check that (/—3) = 3Z @Zy-3 C B, so that B/(V—3) = Fs, and again /-3 is a prime element of B. However, 2 is bad in B: you can easily prove it is irreducible, because (a+ b¥-3)(a — b¥—3) =a? +30? 42 for any a,beEZ, but it is not a prime element since 2? = (1+ /—3)(1 - V—3). Thus the prime ideal over 2 in B is (2,1-+/—3), which needs 2 generators. At this point, anyone who knows algebraic number theory will see that I am doing something fairly silly by taking the square root of —3. It is clearly much more sensible to take w = (-1+/—3)/2 instead; note that w = exp(2mi/3) is a primitive cube root of 1, satisfying w+wt1=0. Let B’ = Z[w], a slightly bigger ring than B. The analysis of its prime0.7 Examples 9 ~ cusp y? = 23 (@) (0, 0) (b) (2+1-3) °° ° ° <> a di Spec Z[V-3] ° @2-V-3) jo | ° ° ° Spec Z O° ° O° O° ° °° (0 —_- 2 @ © ) (11) ) Figure 0.7. The cuspidal cubic and Spec 2(v-3]10 0 Hello! ideals is exactly as for B except above 2, which is a prime element of BY, because B’/(2) = F2[w] = Fy, a quadratic extension field of Fz. In fact you can prove that B’ is a UFD (see Ex. 0.16). Figure 0.7(b) draws the prime ideals of B = Z[/—3] in schematic form. I draw two points in a bubble over the primes p = 5 mod 6 to rep- resent the single prime p € B, because I have in mind the two conjugate points X = +,/—3 of the X-line defined over F,2. 0.8 Reasons for studying commutative algebra Commutative algebra is the crossroads between algebraic number theory, algebraic geometry and abstract algebra. Although much of the material of this book develops techniques of algebra, it should be clear that my main interest is the applications of these ideas to geometry and number theory. a. Algebraic number theory Galois theory studies field extensions, with motivation coming from the study of polynomials and their roots; thus, corresponding to a polynomial F(X) = anX” + an-1X" 1 +--+ + a9 € KX] with coefficients in a field k, one knows how to build a field extension k C K over which f has a root, or splits into linear factors. It often happens that k contains a subring A C k of interest, for example Z C Q, such that the coefficients of f are contained in A; we might then want to study a subring of K corresponding to A, for example the subring B generated by A and the roots of f. As a famous example, let « = exp(2mi/n) = */1 € C; over the ring B = Ze], a number of the form az” — 2" with x and z coprime factorises into n factors: a” — 2” -TI (a — ez). i=0 Suppose we happened to know that B = Z[e] is a UFD; then comparing two factorisations into primes coming from the left and right sides of ma (x —e'z) = y" would obviously impose very strong restrictions on integer solutions of x” + y" = 2", and in fact it is known that this can be used to prove that there are only the trivial solutions with x or y or z= 0; see, for example, (B& Sh], Chapter III, 1.1. Unfortunately, Z[e]0.8 Reasons for studying commutative algebra ll js not usually a UFD for large n, so you have to work harder if you want to prove this statement for all n and claim the due reward. Rings like Z[e] (or B’ = Z[w] in 0.7) are called rings of integers of number fields, and the study of their ideals in the 19th century, in the context of Kummer’s study of Fermat’s last theorem (now Wiles’ theo- rem), marks the start of commutative algebra. b. Algebraic geometry Quite generally, an algebraic variety VC k” has a coordinate ring kV] = k[X,...,Xnl/Ix = {polynomial functions y: V + k}, and the study of k[V] gives lots of information about V. From the point of view of ring theory, every integral domain that is a finitely generated k-algebra occurs as k[V] for some variety V. For example, if you study a plane curve such as C: (y? = 2°) c k?, you will rapidly come to the conclusion that its properties are closely related to those of the ring A = k[X,Y]/(Y? — X°). Thus, as we saw in 0.7, Example 1, A can be embedded in a bigger ring A’ = k[é] with X = #?,Y = #3, which corresponds to the parametrisation x = y= of the curve C. The fact that the origin (0,0) is a singular point of C is reflected in the algebra, as we saw in 0.7, and will see in several places below. c. Abstract algebra Commutative rings were studied extensively at the turn of the century: as I have just sketched, they occurred in the 19th century as the rings of integers of number fields, and as the rings k[V] associated with algebraic varieties V C k”. In either case, these are quo- tients of polynomial rings A = Z[z1,...,¢n]/I or A = k(x1,...,2n)/J; in the early 20th century, notably in the work of Hilbert, Emmy Noether, Krull, Emil Artin and others, they occurred as abstract structures sat- isfying the ring axioms, and whatever other axioms were needed to get reasonable results, notably the a.c.c. (ascending chain condition, see Chapter 3). A very important motivating idea for the development of algebra in the 1910s and 1920s was the fact that the abstract approach is often simpler and more general, and yields many of the results for the concrete quotients of polynomial rings with considerably less effort; see, for example, Remark 1.5. In this vein, this course discusses a number of results which are showcases of the methods of abstract algebra; at the same time, I point out some problems where the abstract approach has its limitations.12 0 Hello! In conclusion, rings of the form Z[X1,..., Xn]/J or k[X1,...,Xn]/I for suitable ideals J are important in both algebraic geometry and alge- braic number theory, and have also been central to the development of algebra. 0.9 Discussion of contents The course can be viewed as a continuation of an algebra course on, rings and modules, as taught in the second or third year in many British universities. The book covers roughly the same material as Atiyah and: Macdonald [A & M], Chaps. 1-8, but is cheaper, has more pictures, and is considerably more opinionated. However, rather than talking about abstract algebra for its own sake, my main aim is to discuss and exploit the idea that a commutative ring A can be thought of as the ring of functions on a space X = Spec A. Here are some points I will try to emphasise along these lines: (1) Prime spectrum Spec A The attempt to re-establish the space X as the spectrum or prime spectrum, the set of prime ideals of A. In important cases, the prime ideals are controlled by maximal ideals. (2) Geometric rings and the Nullstellensatz A ring that can be written as the coordinate ring A = k[V] of a variety V is by definition a ring of functions on V, and the correspondence between geometry and algebra is especially close in this case. (3) Localisation A ring of fractions A[1/f] corresponds to restricting functions on X to the open subset X; = {x € X | f(z) #0}. (4) Primary decomposition A module can be pictured as a geometric object living over a subset of X (see the Frontispiece). (5) Integral extensions and normalisation (6) Discrete valuation rings Discrete valuation rings (DVRs) are the best kind of UFDs, having only one prime. The idea of a discrete valuation corresponds closely to the idea of measuring the power of a prime p dividing an integer, or the order of zeros and poles of a function on a normal algebraic variety or complex analytic space. A Noetherian normal ring is an intersection of DVRs. This re- sult was somehow omitted from [A & M], but is very important. The statement and its proof is an exemplary chain of abstract al- gebraic reasoning. However, the result includes (i) the statement 30.10 Who the book is for 13 that a meromorphic function on a complex manifold having no poles in codimension one is holomorphic, and (ii) the characteri- sation of algebraic integers among all algebraic numbers in terms of p-adic valuations. Finiteness of normalisation An algebraic result that provides the ring of integers of a number field, and the resolution of singu- larities of algebraic curves. At the same time, it gives a salutary reminder: finiteness of normalisation holds for practically all rings of importance in the world, but not for all Noetherian rings. (8) 0.10 Who the book is for Although this course will give the student intending to study algebraic geometry or complex analytic geometry an introduction to basic mate- rial in commutative algebra, an equally important objective is to make accessible to algebraists and number theorists the benefits of geometric intuition in studying commutative rings. However uncomfortably they fit into the framework of axioms and abstract arguments, the pictures in this book are (in my opinion) what commutative algebra is all about. If you are the kind of student of algebra who suffers from vertigo when more than one or two logical steps above the axioms, you may convince yourself that all the material is entirely rigorous, and, if so desired, treat the geometric stuff as fanciful digressions. Or maybe my approach is just not for you; by all means study the excellent book of Atiyah and Macdonald [A & M] instead. 0.11 What you’re supposed to know As already mentioned, this course assumes some prior knowledge of fields and rings, for example, polynomial rings, the elementary theory of fac- torisation, division with remainder and its application to proving that Z and k[X] are PIDs (principal ideal domains), therefore UFDs. This stuff is treated in any number of books, for example [C], [H & H], Chap- ter 4 or [W]. The first few (very easy!) exercises to this section sketch some standard material I will assume later, for example the fact that the polynomial rings Z(z1,...,2n] and k[z,...,2n] are UFDs. I also use some basic material in set theory, notably the definition of partially ordered set, but I give a discussion of Zorn’s lemma and the a.c.c. from first principles. This course is intended for advanced undergraduates or beginning14 0 Hello! graduate students, and in addition to background from algebra it as sumes basic facts from analysis such as radius of convergence of a power series and the definition of a topological space. On the other hand, I develop modules from scratch and (except for Appendix 8.13 to Chap- ter 8) do not assume any material from Galois theory beyond the simple lemma on primitive field extensions sketched in Ex. 0.5. The relation between commutative algebra, algebraic geometry and al- gebraic number theory is symbiotic, and the student will derive valuable motivation from an interest in one or other of these subjects. However, I have tried to keep this course reasonably self-contained, sometimes at the expense of repeating material that is covered perfectly well elsewhere. Exercises to Chapter 0 Exercises 1-9 are intended to sketch out some material which you are supposed to know. 0.1 Let A be any ring, and consider the polynomial ring A[T]. Prove that T is not a zerodivisor in A[Z']. Generalise the argument to prove that a monic polynomial fs T™ t+ anaT™! +---+09 (a polynomial with leading coefficient 1) is not a zerodivisor in A(T]. 0.2 Let A be any ring, a € A and f € A[T]. Prove that there exists an expression f = (T’—a)q+r with q € A[I] andr € A. (Hint: subtract off a suitable multiple of (T’— a) to cancel the leading term, then use induction on deg f.]| By substituting T = a, show that r = f(a). (This result is often called the remainder theorem in algebra textbooks.) 0.3 Remind yourself that (i) Z and k[T] are Euclidean rings, that is, they have division with remainder (see [C], [H & H], Chapter 4 or [W)); (ii) a ring having division with remainder is a PID; and (iii) a PID is a UFD. 0.4 Let A[T] be the polynomial ring over a ring A, and B any ring. Suppose that y: A — B is a given ring homomorphism; show that ring homomorphisms 7: A[T] > B extending ¢ are in one- to-one correspondence with elements of B. 0.5 Remind yourself that (i) a nonzero prime ideal m C k[T] is of the form (f) with f an irreducible polynomial, and the quotient ring k[T]/m is a finite algebraic extension field of k in which f0.6 0.7 0.8 0.9 0.10 Exercises 15 has a root; and (ii) if A is a ring containing a field k and ae A any element then the subring k[a] C A generated by k and a is isomorphic to k[T' or is a field extension of the form k[T]/(f) with f irreducible. Let B = k[T] with k a field; a k-automorphism of B is a ring homomorphism y: B — B that is the identity on k and is an automorphism of B (that is, one-to-one and onto). (i) Describe the group Aut; B of k-automorphisms of B; (ii) now do the same for the field IX = k(T) of rational functions. [Hint: use Ex. 0.4. The answer to (ii) is in terms of 2 x 2 matrixes.] Let A be a UFD, K its field of fractions and f € A[T] a monic polynomial (defined in Ex. 0.1 above). Prove that if f has a root a € K then in fact a € A. [Hint: you can probably remember this for Z C Q; write a = p/q where p,g have no common factors, and consider the equation f(p/q) = 0.] Let A be a UFD and K its field of fractions. A polynomial g = baT™ $+ bgrT™ | +++ + bo € A[T] is primitive if its coefficients b; have no common factors in-A (other than units). Prove Gauss’ lemma: the product of two primitive polynomials is primitive. A polynomial f € K[T] has a reduced expression f = afo where a € K, and fy € A is primitive, unique up to multiplying by a unit. (This just amounts to clearing denominators and taking out any common factor.) The point of Gauss’ lemma is that if f = afo and g = bgo are reduced expressions for f and g then fg = (ab)(fogo) is a reduced expression for fg. Prove that A a UFD implies A[T] a UFD. [Hint: you know that K[T] is a UFD, where K = Frac A, Use Gauss’ lemma to compare factorisation in K(Z'] and A[T].] Deduce that the polynomial rings Z[z1,...,2n] and k[z,,...,%n] are UFDs. Let k = F, © Z/(p) be the finite field with p elements. Compare the two rings k[T]/(T”) and Z/(p") for n 2 2. Show that their elements can be written respectively in the form Qo taiT +--+ +an-1T"! or ag +aipt+--++ @n—1p"—" with a; € {0,1,...,p—1} fori =0,...,n—1, determine the ad- dition and multiplication of these power series in the two rings, and note that they differ only by the p-adic “carry”.