You are on page 1of 226

Algebraic Geometry

J.S. Milne

Version 6.01
August 23, 2015

These notes are an introduction to the theory of algebraic varieties emphasizing the similarities to the theory of manifolds. In contrast to most such accounts they study abstract
algebraic varieties, and not just subvarieties of affine and projective space. This approach
leads more naturally into scheme theory.

BibTeX information
@misc{milneAG,
author={Milne, James S.},
title={Algebraic Geometry (v6.01)},
year={2015},
note={Available at www.jmilne.org/math/},
pages={226}
}

v2.01 (August 24, 1996). First version on the web.
v3.01 (June 13, 1998).
v4.00 (October 30, 2003). Fixed errors; many minor revisions; added exercises; added two
sections/chapters; 206 pages.
v5.00 (February 20, 2005). Heavily revised; most numbering changed; 227 pages.
v5.10 (March 19, 2008). Minor fixes; TEXstyle changed, so page numbers changed; 241
pages.
v5.20 (September 14, 2009). Minor corrections; revised Chapters 1, 11, 16; 245 pages.
v5.22 (January 13, 2012). Minor fixes; 260 pages.
v6.00 (August 24, 2014). Major revision; 223 pages.
v6.01 (August 23, 2015). Minor fixes; 226 pages.
Available at www.jmilne.org/math/
Please send comments and corrections to me at the address on my web page.

The curves are a tacnode, a ramphoid cusp, and an ordinary triple point.

c 1996, 1998, 2003, 2005, 2008, 2009, 2011, 2012, 2013, 2014 J.S. Milne.
Copyright
Single paper copies for noncommercial personal use may be made without explicit permission
from the copyright holder.

Contents
Contents
1

3

Preliminaries from commutative algebra

11

a. Rings and ideals, 11 ; b. Rings of fractions, 16 ; c. Unique factorization, 21 ; d. Integral
dependence, 24; e. Tensor Products, 30 ; f. Transcendence bases, 33.

2

Algebraic Sets

35

a. Definition of an algebraic set, 35 ; b. The Hilbert basis theorem, 36; c. The Zariski
topology, 37; d. The Hilbert Nullstellensatz, 38; e. The correspondence between algebraic
sets and radical ideals, 39 ; f. Finding the radical of an ideal, 43; g. Properties of the
Zariski topology, 43; h. Decomposition of an algebraic set into irreducible algebraic sets,
44 ; i. Regular functions; the coordinate ring of an algebraic set, 47; j. Regular maps, 48; k.
Hypersurfaces; finite and quasi-finite maps, 48; l. Noether normalization theorem, 50 ; m.
Dimension, 52 .

3

Affine Algebraic Varieties

59

a. Sheaves, 59 ; b. Ringed spaces, 60; c. The ringed space structure on an algebraic set, 61
; d. Morphisms of ringed spaces, 64 ; e. Affine algebraic varieties, 65; f. The category of
affine algebraic varieties, 66; g. Explicit description of morphisms of affine varieties, 67 ;
h. Subvarieties, 70; i. Properties of the regular map Spm.˛/, 71; j. Affine space without
coordinates, 72; k. Birational equivalence, 73; l. Noether Normalization Theorem, 75; m.
Dimension, 75 .

4

Local Study

81

a. Tangent spaces to plane curves, 81 ; b. Tangent cones to plane curves, 83 ; c. The local
ring at a point on a curve, 85; d. Tangent spaces to algebraic subsets of Am , 86 ; e. The
differential of a regular map, 88; f. Tangent spaces to affine algebraic varieties, 89 ; g.
Tangent cones, 93; h. Nonsingular points; the singular locus, 94 ; i. Nonsingularity and
regularity, 96; j. Examples of tangent spaces, 97.

5

Algebraic Varieties

99

a. Algebraic prevarieties, 99; b. Regular maps, 100; c. Algebraic varieties, 101; d. Maps
from varieties to affine varieties, 103; e. Subvarieties, 103 ; f. Prevarieties obtained
by patching, 104; g. Products of varieties, 105 ; h. The separation axiom revisited,
110; i. Fibred products, 112 ; j. Dimension, 113; k. Dominant maps, 115; l. Rational
´
´
maps; birational equivalence, 115; m. Local study, 116; n. Etale
maps, 117 ; o. Etale
neighbourhoods, 120 ; p. Smooth maps, 122 ; q. Algebraic varieties as a functors, 123 ; r.
Rational and unirational varieties, 126 .

6

Projective Varieties

129

a. Algebraic subsets of Pn , 129; b. The Zariski topology on Pn , 133; c. Closed subsets of

3

An and Pn , 134 ; d. The hyperplane at infinity, 135; e. Pn is an algebraic variety, 135; f.
The homogeneous coordinate ring of a projective variety, 137; g. Regular functions on a
projective variety, 138; h. Maps from projective varieties, 139; i. Some classical maps of
projective varieties, 141
; j. Maps to projective space, 146; k. Projective space without
coordinates, 146; l. The functor defined by projective space, 146; m. Grassmann varieties,
147 ; n. Bezout’s theorem, 150; o. Hilbert polynomials (sketch), 151; p. Dimensions, 153;
q. Products, 154 .

7

Complete Varieties

157

a. Definition and basic properties, 157 ; b. Proper maps, 159; c. Projective varieties are
complete, 160 ; d. Elimination theory, 161 ; e. The rigidity theorem; abelian varieties,
165; f. Chow’s Lemma, 167 ; g. Analytic spaces; Chow’s theorem, 169; h. Nagata’s
Embedding Theorem, 170 .

8

Normal Varieties; (Quasi-)finite maps; Zariski’s Main Theorem

173

a. Normal varieties, 173 ; b. Regular functions on normal varieties, 177 ; c. Finite and
quasi-finite maps, 179 ; d. The fibres of finite maps, 184; e. Zariski’s main theorem, 186
; f. Stein factorization, 191; g. Blow-ups, 192 ; h. Resolution of singularities, 192 .

9

Regular Maps and Their Fibres

195

a. The constructibility theorem, 195; b. The fibres of morphisms, 198; c. Flat maps and
their fibres, 201 ; d. Lines on surfaces, 208; e. Bertini’s theorem, 213; f. Birational
classification, 214.

Index

223

4

namely.45)” is to this item in chapter 3. All rings will be commutative with 1.. Unadorned tensor products are over k. J. Commutative Algebra. sxnnnn) is to question nnnn on mathoverflow. We sometimes refer to the computer algebra programs CoCoA (Computations in Commutative Algebra) http://cocoa.stackexchange.. X may equal Y ). Z D ring of integers. v4. a reference “(3. The dual Homk-linear . p a prime number. Fields and Galois Theory. 2014. k/ of a finite-dimensional k-vector space E is denoted by E _ .Notations We use the standard (Bourbaki) notations: N D f0. FT: Milne. Macaulay 2 (Grayson and Stillman) http://www. C D field of complex numbers. 2. Given an equivalence relation. A reference “Section 3m” is to Section m in Chapter 3. 1. Springer.E. we often write MR for R ˝ M . For a k-algebra R and k-module M . math. i.. Fp D Z=pZ D field of p elements.ai /i 2I . and so on. or equals Y by definition.S. R D field of real numbers. Prerequisites The reader is assumed to be familiar with the basic objects of algebra.S.dima.com). J. and homomorphisms of rings are required to map 1 to 1. and Y are isomorphic. is a function i 7! ai W I ! A. We use Gothic (fraktur) letters for ideals: a b c m n p q A B C a b c m n p q A B C M N P Q M N P Q Finally X X X X def DY Y Y 'Y X X X X is defined to be Y . : : :g.net (resp. modules. Shafarevich 1994: Basic Algebraic Geometry. A family of elements of a set A indexed by a second set I . We sometimes write jSj for the number of elements in a finite set S .unige. and Y are canonically isomorphic (or there is a given or unique isomorphism). is a subset of Y (not necessarily proper. denoted . fields. Throughout.50.e. rings. CA: Milne. 2014. Hartshorne 1977: Algebraic Geometry. References Atiyah and MacDonald 1969: Introduction to Commutative Algebra. Addison-Wesley.01.it/. A reference monnnn (resp. Œ denotes the equivalence class containing .edu/Macaulay2/. Springer.math. v4. 5 . a reference “(67)” is to (displayed) equation 67 (usually given with a page reference unless it is nearby).uiuc. k is an algebraically closed field.

Soli Vishkautsan. B. Sergei Gelfand. Hossein Sabzrou.J. . Tom Savage. Nguyen Quoc Thang. 6 . . Cheuk-Man Hwang. From the notes of a lecture series that Grothendieck gave at SUNY at Buffalo in the summer of 1973 (in 167 pages. Sandeep Chellapilla. Franklin. Rankeya Datta. Dan Karliner. Sean Rostami. Lucio Guerberoff. Shalom Feigelstock. . Felipe Zaldivar. . Bhupendra Nath Tiwari. Jyoti Prakash Saha. This would cost years of study to Plato. . it seems to me that the title of the old book of Enriques is still adequate: Geometrical Theory of Equations . Florian Herzig. Tony Feng. Lars Kindler. Jasper Loy Jiabao. G ROTHENDIECK : Yes! but your “layman” should know what a system of algebraic equations is. John Miller. Q UESTION : If we try to explain to a layman what algebraic geometry is. David Rufino. Guido Helmers. Israel Vainsencher. Dubey. Umesh V. Isac Hed´en. and others.Acknowledgements I thank the following for providing corrections and comments on earlier versions of these notes: Jorge Nicol´as Caro Montoya. Joaquin Rodrigues. Darij Grinberg. Dennis Bouke Westra. Grothendieck manages to cover very little). Q UESTION : It should be nice to have a little faith that after two thousand years every good high school graduate can understand what an affine scheme is . Christian Hirsch. Daniel Gerig.

: : : . : : : . (Exercise!) 7 . (1) j D1 the starting point for algebraic geometry is the study of the solutions of systems of polynomial equations. fi .X1 . the polynomial functions form a much less rich class than the others. just as topology is the study of continuous functions and the spaces on which they are defined (topological spaces). suppose that the system (1) has coefficients aij 2 k and that K is a field containing k. Then (1) has a solution in k n if and only if it has a solution in K n . Descartes.1 but those for polynomial equations depend on whether or not k is algebraically closed and (to a lesser extent) whether k has characteristic zero. A better description of algebraic geometry is that it is the study of polynomial functions and the spaces on which they are defined (algebraic varieties). and so on: algebraic geometry regular (polynomial) functions algebraic varieties topology continuous functions topological spaces differential topology differentiable functions differentiable manifolds complex analysis analytic (power series) functions complex manifolds. Of course. i D 1. Xn / D 0. The approach adopted in this course makes plain the similarities between these different areas of mathematics. n X aij Xj D bi . fi 2 kŒX1 . but by restricting our study to polynomials we are able to do calculus over any field: we simply define X d X ai X i D i ai X i 1 : dX Moreover. Xn : One immediate difference between linear equations and polynomial equations is that theorems for linear equations don’t depend on which field k you are working over. 1 For example. m. differential topology the study of infinitely differentiable functions and the spaces on which they are defined (differentiable manifolds). : : : . calculations with polynomials are easier than with more general functions. 1619 Just as the starting point of linear algebra is the study of the solutions of systems of linear equations. i D 1. : : : . and the dimension of the space of solutions is the same for both fields. m.Introduction There is almost nothing left to discover in geometry. March 26.

Andr´e. Serre resolved the first by borrowing ideas from complex analysis and defining an algebraic variety over an algebraically closed field to be a topological space with a sheaf of functions 2 Think of S as a level surface for the function f . the inverse function theorem doesn’t hold in algebraic geometry.x. we shall learn that the equation (2) defines a surface in k 3 . y. several attempts were made to resolve these difficulties. z/ D C (2) defines a surface S in R3 . Until the mid 1940s. For example.I.4 ˘ His definition of a variety over a base field k is not intrinsic. c/ perpendicular to the gradient vector . b.. 1946. Throughout.x.X /.x a/ C . and that the tangent plane to S at a point P D . z/ with coefficients in a field k. R. Consider a nonzero polynomial f . c/ has equation2       @f @f @f . Foundations of algebraic geometry.Of /P of f at P . Then. z/. In the ensuing years. treating more advanced topics. and nor is X np 1 in characteristic p ¤ 0 — these functions can not be integrated in the field of rational functions k.8 I NTRODUCTION Consider a nonzero differentiable function f . Providence. 4 Nor did Weil use the Zariski topology in 1946.a. we learn that the equation f .x. not as submanifolds of some Euclidean space. These notes form a basic course on algebraic geometry. One other essential difference is that 1=X is not the derivative of any rational function of X . and this is one of the essential differences between algebraic geometry and the other fields. In calculus. However. we require the ground field to be algebraically closed in order to be able to concentrate on the geometry. American Mathematical Society. and note that the equation is that of a plane through . it is difficult even to make sense of a statement such as “the Gauss curvature of a surface is intrinsic to the surface but the principal curvatures are not” without the abstract notion of a surface. In 1955.e.P /. Weil was forced to develop a theory of algebraic geometry for “abstract” algebraic varieties over arbitrary fields. i. can be found on my website. specifically.z c/ D 0: (3) @x P @y P @z P The inverse function theorem says that a differentiable map ˛W S ! S 0 of surfaces is a local isomorphism at a point P 2 S if it maps the tangent plane at P isomorphically onto the tangent plane at P 0 D ˛. Additional chapters.3 but his “foundations” are unsatisfactory in two major respects: ˘ Lacking a sheaf theory. 3 Weil. algebraic geometry was concerned only with algebraic subvarieties of affine or projective space over algebraically closed fields. . y. and we shall use the equation (3) to define the tangent space at a point P on the surface. his method of patching together affine varieties to form abstract varieties is clumsy. he fixes some large “universal” algebraically closed field ˝ and defines an algebraic variety over k to be an algebraic variety over ˝ together with a k-structure.a. in order to give substance to his proof of the congruence Riemann hypothesis for curves and abelian varieties. The approach to algebraic geometry taken in these notes In differential geometry it is important to define differentiable manifolds abstractly.y b/ C . In these notes. b. y.

5 Serre. of Math. Faisceaux alg´ebriques coh´erents. commonly referred to as FAC. Jean-Pierre. . in the late 1950s Grothendieck resolved all such difficulties by developing the theory of schemes. (1955). by working only over a base field. (2) 61. we hope to provide a bridge between the intuition given by differential geometry and the abstractions of scheme theory. In this way. 197–278.5 Then. Ann. In these notes. we are able to simplify his language by considering only the closed points in the underlying topological spaces. we follow Grothendieck except that.9 that is locally affine.

.

and B is a finite1 A-algebra. : : : . A ring homomorphism A ! B is said to be of finite-type. A ring homomorphism A ! B is finite. For a ring A. in other words. a Rings and ideals Basic definitions Let A be a ring. and B is a finitely generated A-algebra if B is generated by a finite set of elements as an A-algebra. Elements x1 . then A is the zero ring. A homomorphism of A-algebras B ! C is a homomorphism of rings 'W B ! C such that '.e. if the homomorphism of A-algebras AŒX1 . An A-algebra is a ring B together with a homomorphism iB W A ! B.e. we develop the necessary commutative algebra in the context in which it is used. When 1A D 0 in a ring A. we review some basic definitions and results from commutative algebra. i. : : : . i. multiplication. then b D c. xn of an A-algebra B are said to generate it if every element of B can be expressed as a polynomial in the xi with coefficients in iB . if ab D ac and a ¤ 0. xn 2 B. xn  denote the A-subalgebra of B generated by the xi . if B is finitely generated as an A-module. and let A be a k-algebra. we let AŒx1 .A/.a// D iC . A ring is an integral domain if it is not the zero ring and if ab D 0 implies that a D 0 or b D 0. 1 The term “module-finite” is also used. Xn  ! B acting as iA on A and sending Xi to xi is surjective. : : : .. and we can identify k with its image. we can regard k as a subring of A. the map k ! A is injective. However.iB .. A is the group of elements of A with inverses (the units in the ring). When 1A ¤ 0 in A..a/ for all a 2 A. and the formation of negatives. A subring of A is a subset that contains 1A and is closed under addition.e. : : : . Let k be a field. For the most part. A D f0g.C HAPTER Preliminaries from commutative algebra Algebraic geometry and commutative algebra are closely intertwined. i. 11 1 . When A  B and x1 . in this chapter.

Note that ab  a \ b: (4) The kernel of a homomorphism A ! B is an ideal in A. An ideal m in A is maximal if it is maximal among the proper ideals of A. b 2 bg is an ideal. let z D ay C bx. i. ai bj . An ideal p is prime if p ¤ A and ab 2 p ) a 2 p or b 2 p. and that it consists of all finite sums of the form ri si with ri 2 A. Therefore. am / and b D . s2 . and if a D . for any ideal a in A. Thus m is maximal if and only if A=m is nonzero and has no proper nonzero ideals. the set of cosets of a in A forms a ring A=a. r 2 A ) ra 2 a: The ideal generated by a subset S of A is the intersection of all ideals a containing S — it is Peasy to see that this is in fact an ideal. c D a  b with a D fa j . : : : . This statement extends to finite collections of ideals. : : : . To see this. then . we write . and let a 2 a and b 2 b be such that a C b D 1. Thus p is prime if and only if A=p is nonzero and has the property that ab D 0 H) a D 0 or b D 0. then ab D .a. 0/. (b) a 2 a. : : :g.b/ is a one-to-one correspondence between the ideals of A=a and the ideals of A containing a. When S D fs1 . The ideal generated by the empty set is the zero ideal f0g. then c D c1 D ca C cb 2 ab: Hence.e. denoted by a C b. and so the canonical map A ! A=a  A=b (5) is surjective. : : :/ for the ideal it generates. which contains ab. bn /.0.s1 . Note that m maximal H) m prime. Clearly ab consists of all finite sums ai bi with ai 2 a and bi 2 b.12 1. The ideal P generated by fab j a 2 a. b/ D . P RELIMINARIES FROM COMMUTATIVE ALGEBRA Ideals Let A be a ring. 0/ 2 cg.1. 1/. si 2 S.b1 . b/ 2 c. b 2 bg is denoted by ab. b D fb j . Conversely.a. (5) is surjective with kernel ab. An ideal a in A is a subset such that (a) a is a subgroup of A regarded as a group under addition.0. b/ 2 cg: Ideals a and b in A are coprime (or relatively prime) if a C b D A.a. A=p is an integral domain.0. The map b 7! ' 1 . Let a and b be ideals in A.a. Note that if p is prime and a1    an 2 p. . and so is a field. Clearly its kernel is a \ b. The ideals of A  B are all of the form a  b with a and b ideals in A and B. Assume that a and b are coprime.a1 . For x. then z  bx  x mod a z  ay  y mod b.a1 b1 .a.. The set fa C b j a 2 a. s2 . b/ 2 c. and a 7! a C a is a homomorphism 'W A ! A=a whose kernel is a. : : : . b/ 2 c and . note that if c is an ideal in A  B and . : : : . Let c 2 a \ b. y 2 A. then at least one of the ai 2 p. 0/ D . am bn /.

ar .v. 2 We let spec. and so c D a. (a) H) (b): Let a1  a2     be an ascending chain of ideals. and so it contains a maximal element c D . For some m. there exist elements ai 2 a1 and bi 2 ai such that ai C bi D 1: The product Q i 2 . 2 2 This says the following: let R be a binary relation on a nonempty set X.a1 .2. : : : . all the ai belong to am . (c) H) (a): Let a be an ideal. : : : . A=a1    an D A=a1  . If ˙ has no maximal element. . contradicting (b). If c ¤ a. and so a1 C a2    an D A: Therefore.. then the map A ! A=a1      A=an Q T is surjective.ai C bi / lies in a1 C a2    an and equals 1. We have proved the statement for n D 2. Rings and ideals 13 T HEOREM 1. : : : .A/ denote the set of prime ideals in a ring A and spm. and suppose that. This contradicts the definition of c. an be ideals in a ring A. If ai is coprime to aj whenever i ¤ j .A/ the set of maximal ideals. for each a in X.a1 . S P ROOF.1 (C HINESE R EMAINDER T HEOREM ). then there exists a sequence . and . and let ˙ be the set of finitely generated ideals contained in a. an g of generators. i. ar /.a. with kernel ai D ai . (c) every nonempty set of ideals in A has a maximal element. (b) every ascending chain of ideals a1  a2     eventually becomes constant. and we use induction to extend it to n > 2. am D amC1 D    for some m. The following three conditions on a ring A are equivalent: (a) every ideal in A is finitely generated. For i  2. Then ai is an ideal. Then ˙ is nonempty because it contains the zero ideal. Wikipedia). Noetherian rings P ROPOSITION 1.e. and then S am D amC1 D    D ai : (b) H) (c): Let ˙ be a nonempty set of ideals in A. then the axiom of dependent choice2 implies that there exists an infinite strictly ascending chain of ideals in ˙ . then there exists an a 2 a X c. : : : . This axiom is strictly weaker than the axiom of choice (q.a2    an / ' A=a1  A=a2    an ' A=a1  A=a2      A=an by the n D 2 case by induction. (6) P ROOF. a/ will be a finitely generated ideal in a properly containing c. Let a1 .an /n2N of elements of X such that an RanC1 for all n. there exists a b such that aRb. and hence has a finite set fa1 .

14 1. 2 Now let A be a local noetherian ring with maximal ideal m. p2 . Let A be a noetherian ring. This is. it is obvious that their residues generate m=m2 . p3 .4. e2 .2). so that m D . : : : . true for every ring. as . P RELIMINARIES FROM COMMUTATIVE ALGEBRA A ring A is noetherian if it satisfies the equivalent conditions of the proposition. then M D N . n  1.a1 . Because every nonunit is contained in a maximal ideal.p/ of a prime ideal p in A is the greatest length d of a chain of distinct prime ideals p D pd  pd 1      p0 : (7) (b) The Krull dimension of A is supfht. and conversely an integral domain of Krull dimension 0 is a field. Choose a minimal set of generators fe1 . but the proof for non-noetherian rings requires Zorn’s lemma (CA 2. (a) The height ht.a1 . en generate M . the minimum number of generators for the maximal ideal is equal to the dimension of the vector space m=m2 . Conversely.pi / tends to infinity (CA. and let M be a finitely generated A-module.13). : : : .1 a1 /e1 D a2 e2 C    C an en and. : : : . . but the Krull dimension of the ring may be infinite because it may contain a sequence of prime ideals p1 . P ROPOSITION 1. : : : . (a) Suppose that M ¤ 0. Elements a1 . an of m generate m as an ideal if and only if their residues modulo m2 generate m=m2 as a vector space over k. and write e1 D a1 e1 C    C an en . an / C m2 . for M . and so M=N D 0. contradicting the minimality of the set. C OROLLARY 1. The height of every nonzero prime ideal in a principal ideal domain is 1. Thus. for a local ring A D A X m. For example. so the length of (7) is d ). On applying (c) to the set of all proper ideals containing a fixed proper ideal. P ROOF. Because A is noetherian. 2 D EFINITION 1.5. m is finitely generated. (b) The hypothesis implies that M=N D m.1 a1 / is a unit.p/ j p a prime ideal in Ag. (a) If M D mM . then M D 0: (b) If N is a submodule of M such that M D N C mM . suppose that their residues generate m=m2 . and Nakayama’s lemma shows that m D . an /. : : : such that ht.M=N /. In particular. Then . Then m is an A-module. If a1 . a field has Krull dimension 0. A ring A is said to be local if it has exactly one maximal ideal m. The height of every prime ideal in a noetherian ring is finite. en g. we see that every proper ideal in a noetherian ring is contained in a maximal ideal. an generate m. and def the action of A on m=m2 factors through k D A=m. in fact. P ROOF. and so such a ring has Krull dimension 1 (provided it is not a field). Let A be a local ring with maximal ideal m. the Krull dimension of a noetherian ring A is the supremum of the lengths of chains of prime ideals in A (the length of a chain is the number of gaps.3 (NAKAYAMA’ S L EMMA ). ai 2 m. : : : . : : : . p.

8 (K RULLTI NTERSECTION T HEOREM ). Let ST m denote the set of n homogeneous polynomials f of degree m such that S f . since these terms cancel out.a1 . ar / D gi .a1 .a. every set of generators for an ideal contains a finite generating subset. Then gi . As f and the fi are homogeneous. and let S be a set of generators for a. : : : . we may choose the gi to be homogeneous of degree deg f deg fi D d C 1 di > 0. Rings and ideals 15 D EFINITION 1. and mn consists of the functions f that vanish to order n at x D 0. the function e 1=x shows that the Krull intersection theorem fails for the ring of germs of infinitely differentiable functions at 0 (this ring is not noetherian). : : : . : : : . ar generate m. A local noetherian ring of Krull dimension d is said to be regular if its maximal ideal can be generated by d elements. then b 2 md C1 . : : : .a1 . ar / 2 m. P ROOF. Then A is a noetherian local ring with maximal ideal m D . Xn . : : : . : : : . : : : . mn consists of the elements of A of the form g. Xr  of degree n. ar / 2 n1 m . Let A be the ring of germs of analytic functions at 0 2 R (see p. Xr  generated by the Sset m Sm .a1 . there exists a finite set ff1 .6. It follows from Corollary 1. 2 A SIDE 1. L EMMA 1. Then mn consists of all finite sums X ci1 ir a1i1    arir . and so X \ b D f . Let di D deg fi .x/. Let a be an ideal in a noetherian ring A. : : : . Let b 2 n1 m . : : : . and so b D f . 2 T HEOREM 1. Xr / 2 AŒX1 . In a noetherian ring.4 that a local noetherian ring is regular if and only if its Krull dimension is equal to the dimension of the vector space m=m2 . : : : .7. : : : . The theorem says (correctly) that only 2 the zero function vanishes to all orders at 0. T mn D m  T mn .60 for the notion of a germ of a function). ar / for some homogeneous polynomial f of degree d C 1. we can omit from each gi all terms not of degree deg f deg fi . By definition. : : : . According to the lemma. f 2 Sd C1  a. Let a1 . ar /  fi . and let a be the ideal in AŒX1 .a1 . and n let d D max di .X1 . ar / 2 m  mn : i Thus. and so equals a.a1 . fT g of elements of s m Sm that generates a. Thus. then n1 mn D f0g: P ROOF. : : : . An ideal maximal among those generated by a finite subset of S must contain every element of S (otherwise it wouldn’t be maximal). By contrast. . ci1 ir 2 A: i1 CCir Dn In other words. and so f D g1 f1 C    C gs fs for some gi 2 AŒX1 . ar / for some homogeneous polynomial g. Let A be a noetherian local ring with maximal ideal m.9. and Nakayama’s lemma implies that n1 T mn D 0.a1 .

a/˛.s/ 1 : Then a c D b d H) s. and any other homomorphism A ! B with this property factors uniquely through i : A i S 1A 9Š B P ROOF.a. a. If ˇ exists. It is obviously a homomorphism.ad bc/ D 0 some s 2 S H) ˛. and so ˇ is unique. as / D ˛. s 2 S s and a ring homomorphism a 7! a1 W A ! S 1 A. Write i for the homomorphism a 7! a1 W A ! S 1 A. ab s t D ab st : It is easy to check that these do not depend on the choices of representatives for the equivalence classes.a/ H) ˇ. s/. s as D a H) ˇ.b. The pair .a/˛.S 1 A. then a 7! S 1 A is the zero ring. i / has the following universal property: every element s 2 S maps to a unit in S 1 A.s/ˇ.b/˛. but if 0 2 S .16 b 1. Define ˇ.s/ is a unit in B. as / D ˇ. and define addition and multiplication of equivalence classes in the way suggested by the notation: a s C bt D atCbs st . s/  . whose kernel is fa 2 A j sa D 0 for some s 2 S g: For example. as / D ˛.10.c/ D 0 because ˛.a. then P ROPOSITION 1. P RELIMINARIES FROM COMMUTATIVE ALGEBRA Rings of fractions A multiplicative subset of a ring A is a subset S with the property: 1 2 S. and so ˇ is well-defined.at bs/ D 0 for some u 2 S: Write as for the equivalence class containing . if A is an integral domain an 0 … S.d / ˛. and that we obtain in this way a ring na o S 1A D j a 2 A. 2 . t / ” u. a 1 is injective.s/ 1 . b 2 S H) ab 2 S: Define an equivalence relation on A  S by .a/˛.

1 hX /. Therefore AŒx has the same universal property as Ah . if m D Ap . : : :g is a multiplicative subset of A. this universal property determines the pair . If a contains an element of S .3 When A is an integral domain with field of fractions F . and we let Ah D Sh 1 A. 1 D hx. Thus each element of Ap can be written in the form ac . which contradicts a 2 p. Let S be a multiplicative subset of the ring A. Let ˛WP i such that ˛. Rings of fractions 17 As usual. for any other multiplicative subset T of A not containing 0.11. m 2 N: E XAMPLE 1. and so the two are (uniquely) isomorphic by an isomorphism that fixes elements of A and makes h 1 correspond to x. then Ah D 0.S 1 A. and we let Ap D Sp 1 A. In the ring AŒx D A ! B beP a homomorphism of rings AŒX=. which we denote F . Ap is the subring of F consisting of elements expressible in the form as . then u. and so we may assume h ¤ 0. Thus some of the ideal structure of A is lost in the passage to S 1 A. because ˛. Then Sh D f1. Thus every element of Ah can be written in the form a= hm .h/˛. and. a 2 A.e. and so ua D us … p. and so is the whole ring.1 P ROOF. c … p. a 2 A. In this case. then S 1 a contains a unit. but if 1 D as for some a 2 p and s … p. F D S 1 A is the field of fractions of A.h/ i W AŒX  ! B factors through AŒx because 1 hX 7! 1 ˛..13. E XAMPLE 1. then 1 2 m. the ring T 1 A can be identified with the subring f at 2 F j a 2 A. and a c D b d ” s. and if A is an integral domain with field of fractions F and h ¤ 0. and let S 1 A be the corresponding ring of fractions. The map p 7! S 1 p D . Ap is a local ring. Finally.ad bc/ D 0. P P ai defines an isomorL EMMA 1. this is the unique extension of ˛ to AŒx. For every ring A and h 2 A. 3 First .A/. but. h2 . i / uniquely up to a unique isomorphism. generates an ideal S 1 a in S 1 A.S 1 A/p is a bijection from the set of prime ideals of A disjoint from S to the set of prime ideals of S 1 A with inverse q 7!(inverse image of q in A). a 2 A. h. s … pg is a maximal ideal in Ap . the map ai X i 7! hi phism ' hX / ! Ah : AŒX =. Let h 2 A. s … p. then Ah is the subring of F of elements of the form a= hm . and so h is a unit. Let p be a prime ideal in A.h/ 1 D 0. t 2 S g of F . The homomorphism ai X 7! ˛. Next. much is retained.ai /˛. If h D 0.b.12. Any ideal a in A.s a/ D 0 some u … p.ahn bhm / D 0.14. P ROPOSITION 1. both rings are zero. i. We shall be especially interested in the following examples. and it is the only maximal ideal. m is maximal because every element of Ap not in m is a unit.h/ is a unit in B. some N: If h is nilpotent. check m is an ideal. and a hm D b hn ” hN . some s … p: The subset m D f as j a 2 p. Then Sp D A X p is a multiplicative subset of A. When A is an integral domain and S D A X f0g. as the next proposition shows.h/ is a unit in B. 2 Let S be a multiplicative subset of a ring A.

P ROOF. so that Am D S 1 A. and so we have to show that mm D . and is therefore an integral domain.s/ C mm D A. and so a1 D as for some a0 2 a. In every noetherian ring. as is the unique element of this ring such that s as D a. it induces isomorphisms mr =mn ! nr =nn for all r < n. a 2 mm : We now prove that the map is surjective. s 2 A X m.A=a/ where SN is the image of S in A=a.r < n/: 0 mr =mn A=mn ' 0 nr =nn Am =nn A=mr 0 ' Am =nr 0: Let S D A X m. if as 2 b. Conversely. only 0 lies in all powers of all maximal ideals. a D aec . a must be 0 in A=mm . there exist b 2 A and q 2 mm such that sb C q D 1. let bc be the inverse image of b in A. Because S contains no zero divisors. b  bce . and so the only maximal ideal in A=mm is m=mm .S 1 mm / \ A. Thus. and I’ll identify A with its image. and as sa D 0 in A=mm . and so b D b . s 2 S . Thus s 2 b . Therefore. As s is not in m=mm . . s 2 S . 2 L EMMA 1. t . the map a C mn 7! a C nn W A=mn ! Am =nn is an isomorphism. then a a c ce ce 1 2 b. let ae D . and for an ideal a of A. i. it follows that . Let m be a maximal ideal of an integral domain A. and a 2 A. Let as 2 Am . The only maximal ideal of A containing mm is m. and so ast 2 a. For an ideal b of S 1 A. Then sa 2 mm . The only maximal ideal containing mm is m (because m0  mm H) m0  m/. For an ideal b of S 1 A. and so a 2 b . An element of . 0 For an ideal a of A. P RELIMINARIES FROM COMMUTATIVE ALGEBRA P ROOF.e.18 1. certainly. if a 2 aec . since s. Moreover. a 2 A. the map a 7! a1 W A ! Am is injective.S 1 A/a be the ideal in S 1 A generated by the image of a. and let n D mAm . and so no maximal ideal contains both s and mm . it must be a unit in A=mm . then SN 1 . this implies that a 2 a: for such an ideal.16. In order to show that the map A=mn ! An =nn is injective.1 q/. we have to show that nm \ A D mm . If a is a prime ideal disjoint from S . The second statement follows from the first. We have shown that p 7! pe and q 7! qc are inverse bijections between the prime ideals of A disjoint from S and the prime ideals of S 1 A. ae is prime. Because s is invertible in Am =nm . then certainly bc is prime. then a1 2 ae .15. S 1 A=ae ' SN 1 . certainly a  aec . But nm D mn Am D S 1 mm . because of the exact commutative diagram .ba/ D a. and so sa D 0 in A=mm . s 2 S.as a0 / D 0 for some t 2 S . Thus.A=a/ is a subring of the field of fractions of A=a. the image of ba in Am also has this property and therefore equals a 2 s. For all n. For any ideal a of A. P ROPOSITION 1.S 1 mm / \ A can be written a D b=s with b 2 mm .. a 2 A. Conversely. If a is a prime ideal disjoint from S . If b is prime.

we have tˇ. we have 0 D S 1 .m / 0 0 for some m 2 M . and so is contained in some maximal ideal m. see CA 5. Rings of fractions 19 P ROOF. Let A be a ring. s. s/.15 holds for all commutative rings (ibid. n 2 M. 1.mAm / for some n (by the Krull intersection theorem 1.18.t m/ D 0. In other words. iS S 1M 9Š N: P ROPOSITION 1. s 2 S g iS 1M !S of A-modules whose kernel is fa 2 M j sa D 0 for some s 2 S g: M 1. s/  .m.ˇ ı˛/ D S 1 ˇ ıS 1 ˛. Now m s D tm ts D ˛. For more on rings of fractions.S 1 ˇ/. which implies that a … m . 5.8).8). t 2 S. The functor M S 1 M is exact.m0 / ts 2 Im.m/ 0 s D 0 and so. In particular. then fb j ba D 0g is a proper ideal. and so a n n 1 … . if the sequence of A-modules ˇ ˛ M 0 ! M ! M 00 is exact.m/ D 0. Then ˇ.19. and that we obtain in this way an S 1 A-module S and a homomorphism m 7! 1 m 1 WM M D fm s j m 2 M. If a ¤ 0. Let a be an element of a noetherian ring A. Modules of fractions Let S be a multiplicative subset of the ring A. For the reverse inclusion.S 1 ˛/: P ROPOSITION 1. t / ” u.17. Define an equivalence relation on M  S by . let m s 2 Ker. Then a1 is nonzero in Am . a 2 A: It is easily checked that these do not depend on the choices of representatives for the equivalence classes. m. am s t D am st . and so t m D ˛. The canonical map Y M ! fMm j m a maximal ideal in Ag is injective. Because ˇ ı˛ D 0.t m sn/ D 0 for some u 2 S: Write m s for the equivalence class containing .b. and let M be an A-module. and define addition and scalar multiplication by the rules: m s C nt D mtCns st . The elements of S act invertibly on S 1 M .2 N OTES . and every homomorphism from M to an A-module N with this property factors uniquely through iS . Then Ker.m. Therefore Im. then so also is the sequence of S S 1 M0 S 1˛ !S 1 A-modules 1 M S 1ˇ !S 1 M 00 : P ROOF.n.S 1 ˛/  1 ˇ/ where m 2 M and s 2 S . and let M be an A-module. for some t 2 S . 2 .S ˇ .

˛i / if S (a) M D i 2I ˛ i . there exists an s 2 A X m such that sm D 0.˛/.Mi /i 2I of A-modules together with a family .ˇ/= Im. 4 Regard 1 A ' lim A .ˇm /= Im. h ! where h runs I as a category with Hom.23. and let A be a ring.I. P ROPOSITION 1. P ROOF. Let A be a ring. / consisting of a set I and a partial ordering  on I such that for all i. Nm D Ker. P RELIMINARIES FROM COMMUTATIVE ALGEBRA P ROOF. Let . Because m maps to zero Mm . Because the functor M Mm is exact. Immediate consequence of the lemma.˛ji W Mi ! Mj /i j of j A-linear maps such that ˛ii D idMi and ˛k ı ˛ji D ˛ki all i  j  k. surjective) if and only if Mm ! Nm is injective (resp. then Nm D 0 for all m.I.18). j  k. and so N D 0 (by 1. Since this is true for all maximal ideals m. j 2 I . M ! N ! 0). A homomorphism M ! N of A-modules is injective (resp. S over the elements of S (partially ordered by division). Let m 2 M map to zero in all Mm . P ROOF. surjective) for all maximal ideals m: P ROOF. there exists a k 2 I with i. 2 C OROLLARY 1. / is a family .˛m /: If (**) is exact for all m.I.4 An A-module M together with A-linear maps ˛ i W Mi ! M such that ˛ i D ˛ j ı ˛ji for all i  j is the direct j limit of the system . The necessity is a special case of (1. a D A. A direct system of A-modules indexed by .22. 2 Direct limits A directed set is a pair . A sequence of A-modules ˇ ˛ M 0 ! M ! M 00 (*) is exact if and only if ˛m ˛m Mm0 ! Mm ! Mm00 (**) is exact for all maximal ideals m.a.20. . Then a direct system is a functor from I to the category of A-modules. in which case it contains a single element. let N D Ker. Apply the proposition to 0 ! M ! N (resp. Therefore a is not contained in m. b/ empty unless a  b.Mi . For the sufficiency.20).21.20 1.Mi /. Direct limits of A-algebras are defined similarly. The annihilator a D fa 2 A j am D 0g of m is an ideal in A. / be a directed set. 2 C OROLLARY 1. But this means that (*) is exact. 2 P ROPOSITION 1. and so it contains 1. For every multiplicative subset S of A. An A-module M D 0 if Mm D 0 for all maximal ideals m in A. Now m D 1m D 0. and (b) mi 2 Mi maps to zero in M if and only if it maps to zero in Mj for some j  i .

and so c is a unit.26. ajbc H) ajb or ajc. and these homomorphisms are compatible with the maps in the direct system. h0 D hg. Let A be an integral domain in which every nonzero nonunit element is a finite product of irreducible elements. i. 2 c Unique factorization Let A be an integral domain. Suppose that a1    am D b1    bn (8) .24..c. and write b D aq. we must have p D . and the converse holds when A is a unique factorization domain. If every irreducible element of A is prime. If A is a unique factorization domain. and q as a product of irreducible elements. P ROOF. P ROOF. As . we see that a differs from one of the irreducible factors of b or c by a unit. An integral domain A is called a unique factorization domain (or a factorial domain) if every nonzero nonunit in A can be written as a finite product of irreducible elements in exactly one way up to units and the order of the factors: P ROPOSITION 1. Therefore a is irreducible. then a is irreducible. An element a is said to be prime if . c.0/.a/ is prime and p has height 1. If a is prime. Now a D bc D aqc. and using the uniqueness of factorizations. say. If ajbc. some q 2 A: On writing each of b. the homomorphism A ! S 1 A extends uniquely to a homomorphism ha 7! ha W Ah ! S 1 A (1. there is a canonical homomorphism ha 7! ag W Ah ! h0 Ah0 .25. which implies that qc D 1 because A is an integral domain. P ROOF..a/  . 2 C OROLLARY 1. If a D bc. not a unit.e.a/ is a prime ideal.10). Now it is easy to see that S 1 A satisfies the conditions to be the direct limit of the Ah . Assume that a is prime. then A is a unique factorization domain. Therefore a divides b or c. then bc D aq. and hence an irreducible element a. and let a be an element of A that is neither zero nor a unit. a D bc H) b or c is a unit. When hjh0 .a/. then every prime ideal of height 1 is principal. Let A be an integral domain. assume that a is irreducible and that A is a unique factorization domain. For the converse. and admits only trivial factorizations. An element a of A is irreducible if it is not zero. When h 2 S. Let A be an integral domain. Unique factorization 21 P ROOF. then a divides bc and so a divides b or c. Then p contains a nonzero element. and so the rings Ah form a direct system indexed by the set S .e. We have p  . i. 2 P ROPOSITION 1. Suppose the first. Let p be a prime ideal of height 1.

As b1 is irreducible. we obtain the equality a2    am D .27 (G AUSS ’ S L EMMA ). A nonzero polynomial f D a0 C a1 X C    C am X m in AŒX  is said to be primitive if the coefficients ai have no common factor (other than units). 2 Let A be a unique factorization domain. Therefore. Every polynomial f in F ŒX  can be written f D c. and so . 2 P ROPOSITION 1. P RELIMINARIES FROM COMMUTATIVE ALGEBRA with the ai and bi irreducible elements in A. Let A be a unique factorization domain with field of fractions F . h1 . Let f D gh in F ŒX . As a1 is prime.f / 2 A. we have a factorization .28.A=. then it factors into the product of two nonconstant polynomials in AŒX . Therefore. the first coefficient of g not divisible by p. which is well-defined up to multiplication by a unit. be primitive polynomials. p doesn’t divide the . which must therefore be primitive.cd=p/f D g2 h1 in AŒX . Thus. P ROOF.i0 C j0 /th-coefficient of fg. the polynomials g1 D cg and h1 D dh have coefficients in A.ub2 /b3    bn : Continuing in this fashion. is called the content of f . it divides one of the bi . d 2 A. and so obtain a factorization of f in AŒX . i0  m. p divides all the coefficients of at least one of the polynomials g1 .p// ŒX  is an integral domain. be the first coefficient of f not divisible by p. except ai0 bj0 . Continuing in this fashion. we see that 0 D g1  h1 in .p// ŒX . If f . L EMMA 1. which we may suppose to be b1 . and P let bj0 .24. For suitable c. If an irreducible element p of A divides cd . looking modulo . which is not divisible by p.A=.X / 2 AŒX  factors into the product of two nonconstant polynomials in F ŒX .22 1. The element c. so that g1 D pg2 for some g2 2 AŒX . On cancelling a1 from both sides of (8).f /  f1 with c. and so we have a factorization cdf D g1 h1 in AŒX . 2 . P ROOF. j0  n. The product of two primitive polynomials is primitive. .f /.f / 2 F and f1 primitive. then. say g1 . Then all the terms in the sum i Cj Di0 Cj0 ai bj are divisible by p. According to Proposition 1.p/ is prime. Note that f 2 AŒX  if and only if c. we find that the two factorizations are the same up to units and the order of the factors. and let p be an irreducible element of A.p/. we can remove all the irreducible factors of cd . b1 D ua1 for some unit u. Let ai0 . We have shown that no irreducible element of A divides all the coefficients of fg. Let f D a0 C a1 X C    C am X m g D b0 C b1 X C    C bn X n .

f q/ D c.f /  f1 and g D c. We have shown that every element of AŒX  is a product of irreducible elements and that every irreducible element of AŒX  is prime.fg/ D deg. P ROOF.29). with the obvious notions of equality. h 2 AŒX . and so AŒX  is a unique factorization domain (1. 2 Polynomial rings Let k be a field. 2 P ROPOSITION 1.c. of a nonzero polynomial f is the largest total degree of a monomial occurring in f with nonzero coefficient. Xn  is irreducible if it is nonconstant and f D gh H) g or h is constant. Then c.26). g 2 F ŒX . P ROOF. the monomials form a basis for kŒX1 .fg/ D c.q/ D c. 2 C OROLLARY 1. then it divides c. If A is a unique factorization domain. : : : . aj 2 N.24). and write f D c.29. The irreducible elements in AŒX  are the irreducible elements a of A and the nonconstant primitive polynomials f such that f is irreducible in F ŒX . and write f q D g with q 2 F ŒX .f /c.f / or c. Let a be an irreducible element of A. The ring kŒX1 .26). In particular.fg/ D c. and so q 2 AŒX . it divides c. then it divides g or h in F ŒX . Then f is irreducible in F ŒX . Xn  is an integral domain and kŒX1 . P ROOF.f /  c. The degree.g/I hence every factor in AŒX  of a primitive polynomial is primitive.f /  c. and multiplication.g/ 2 A. An element f of kŒX1 . and so if f divides the product gh of g. . : : : . : : : . T HEOREM 1. Xn  are finite sums X ca1 an X1a1    Xnan .27) and (1.f / is a product of irreducible elements in A. Let A be a unique factorization domain with field of fractions F . This shows that f is a product of irreducible elements in AŒX . The elements of the polynomial ring kŒX1 . c. addition. Obvious from (1. Then c. Let f D c. Let f 2 AŒX . : : : .30.fg/ D c. Let f be an irreducible primitive polynomial in AŒX .g/  f1 g1 with f1 g1 primitive.f /f1 . We shall check that A satisfies the conditions of (1. Therefore f divides g in AŒX . Suppose the first. Xn  D k  . Then fg D c.f /c.32. deg.g/.g/  g1 with f1 and g1 primitive. Since deg. then so also is AŒX . and hence also f or g. : : : . and f1 is a product of irreducible primitive polynomials. For polynomials f.31. Xn  as a k-vector space. Unique factorization 23 P ROPOSITION 1.f /. ca1 an 2 k.g/. Xn  is a unique factorization domain.f /c. If a divides fg. kŒX1 . : : : . and so c.f / C deg.g/.q/ D c. As a is prime (1.g/.

P ROOF. : : : .C /  xj D 0.a1 /˛ n 1 C    C f . d 2 Integral dependence Let A be a subring of a ring B. 5 More generally.e.f / in kŒX1 . P RELIMINARIES FROM COMMUTATIVE ALGEBRA P ROOF. i. : : : . 6 A polynomial is monic if its leading coefficient is 1. 2 C OROLLARY 1. : : : .X/ D X n C terms of degree less than n.. : : : . Let A be a subring of a ring B. Xn  is prime if and only if f is irreducible. (9) where C is the matrix of coefficients. we shall need to apply a variant of Cramer’s rule: if x1 .33.24 1. then B is said to be integral over A. m. j D1 with coefficients in a ring A. : : : . An element ˛ of B is integral over A if and only if there exists a faithful AŒ˛-submodule M of B that is finitely generated as an A-module. j D 1. f . ai 2 A: If every element of B is integral over A.34. .A/ of B. To prove this. implies that a D 0. Xn / D a0 . In the next proof.X1 .31. ai 2 A: Thus. : : : . Xn 1 /: Since k itself is a unique factorization domain (trivially). then det. if f W A ! B is an A-algebra. Xn can be expressed uniquely as a polynomial in Xn with coefficients in kŒX1 . m. A nonzero proper principal ideal . Xn 1 . ˛ is integral over A if and only if it is integral over the subring f .24). Xn : This simply says that every polynomial f in n symbols X1 .an / D 0. : : : . Note that kŒX1 . : : : . Xn r 1 /Xn C    C ar . f . an element ˛ of B is integral over A if it satisfies an equation ˛ n C f . An A-module M is faithful if aM D 0. : : : . if it satisfies an equation ˛ n C a1 ˛ n 1 C    C an D 0. Xn 1 ŒXn  D kŒX1 . : : : .X1 .e. expand out the left hand side of 0 1 P c11 : : : c1 j 1 c1 j C1 : : : c1m i c1i xi B :: :: :: :: C D 0 det @ ::: : : : : A P cm1 : : : cm j 1 i cmi xi cm j C1 : : : cmm using standard properties of determinants. P ROPOSITION 1. a 2 A. the theorem follows by induction from Proposition 1..X1 . xm is a solution to the system of linear equations m X cij xj D 0. i D 1. i. Special case of (1. : : : . An element ˛ of B is said to be5 integral over A if it is a root of a monic6 polynomial with coefficients in A.

An A-algebra B is finite if it is generated as an A-algebra by a finite set of elements each of which is integral over A. : : : . and AŒb1 . bn . 2 C OROLLARY 1.˛    D 0: Let C be the matrix of coefficients on the left-hand side. Integral dependence 25 P ROOF. ˛m  and that ˛ini C ai1 ˛ini 1 C    C ai ni D 0.˛ a12 e2 a13 e3  D 0 a22 /e2 a23 e3  D 0 a11 /e1 a21 e1 C . we obtain an equation ˛ n C c1 ˛ n 1 C c2 ˛ n 2 C    C cn D 0. (: Immediate consequence of (1.. An A-algebra B is finite if and only if it is finitely generated and integral over A. ˛ n 1 has the property that ˛M  M . : : : . P ˛ei D aij ej . 2 P ROPOSITION 1.36. : : : . Any monomial in the ˛i divisible by some ˛ini is equal (in B) to a linear combination of monomials of lower degree. it is certainly finitely generated as an A-algebra. P ROOF. On expanding out the determinant. ai 2 A: Then the A-submodule M of B generated by 1.d. ˛. Let 2 C . . Suppose that B D AŒ˛1 . and so it is finite over A. If B is integral over A and C is integral over B.34)..35. Consider rings A  B  C . Therefore. 1  ri < ni . )W Suppose that ˛ n C a1 ˛ n 1 C    C an D 0. : : : . then C is integral over A. Then Cramer’s formula tells us that det. some aij 2 A: We can rewrite this system of equations as . Therefore is integral over A by (1. m. (W Let M be a faithful AŒ˛-submodule of B admitting a finite set fe1 . bn  is finite over A (see 1. 2 . bn Œ  is finite over AŒb1 . aij 2 A. en g of generators as an A-module. As B is finitely generated as an A-module. and it is faithful because it contains 1. : : : .35). P ROOF.34) shows that every element of B is integral over A. Then B is a faithful AŒ˛-module for all ˛ 2 B (because 1B 2 B). Then. and so (1. : : : . ): We may replace A with its image in B. As M is faithful and the ei generate M . ci 2 A: 2 P ROPOSITION 1. for each i . P ROOF.. i D 1. B is generated as an A-module by the finite set of rm monomials ˛1r1    ˛m .37. Then n C b1 n 1 C    C bn D 0 for some bi 2 B.C / D 0.35). this implies that det. Now AŒb1 .C /  ei D 0 for all i .

In fact. 2 C OROLLARY 1. P ROPOSITION 1. and let a=b be an element of its field of fractions. we obtain the equation an C a1 an 1 b C    C an b n D 0: The element p then divides every term on the left except an . am d m 2 A.26 1.35). Let A be a subring of a ring B. i.38. P ROOF.43. : : : . this shows that d˛ is integral over A.d˛/m C a1 d. ˇ is finitely generated as an A-module (1. The integral closure of A in B is the subring of B consisting of the elements integral over A.42. then b divisible by some prime element p not dividing a. then there exists a d 2 A such that d˛ is integral over A. P ROPOSITION 1. ai 2 A: On multiplying through by b n .. P ROOF. P ROOF. Then AŒ˛. and multiply through the equation by d m : . P RELIMINARIES FROM COMMUTATIVE ALGEBRA T HEOREM 1. if ˛ 2 F.a=b/n C a1 . 2 D EFINITION 1. then it satisfies an equation . Therefore (1. An integral domain A is is said to be integrally closed if it is equal to its integral closure in its field of fractions F . Unique factorization domains are integrally closed. If a=b is integral over A. Let A be an integral domain and let E be an algebraic extension of the field of fractions of A. The elements of B integral over A form an A-subalgebra of B. 2 D EFINITION 1. Then E is the field of fractions of the integral closure of A in E. If ˛ is algebraic over F . Let A be a unique factorization domain.d˛/m 1 C    C am d m D 0: As a1 d. the proposition shows that every element of E is a quotient ˇ=d with ˇ integral over A and d 2 A. P ROOF. and let ˛ be an element of some field containing F . and hence divides an .a=b/n 1 C    C an D 0. ai 2 F: Let d be a common denominator for the ai . By assumption.e. It is stable under multiplication by ˛ ˙ ˇ and ˛ˇ and it is faithful as an AŒ˛ ˙ ˇ-module and as an AŒ˛ˇ-module (because it contains 1A ). this is a contradiction. Since it doesn’t divide a.41.34) shows that ˛ ˙ ˇ and ˛ˇ are integral over A. Let A be a subring of the ring B.40. so that dai 2 A for all i . Let A be an integral domain with field of fractions F . ˛ satisfies an equation ˛ m C a1 ˛ m 1 C    C am D 0. 2 . Let ˛ and ˇ be two elements of B integral over A.39. If a=b … A. ˛ integral over A H) ˛ 2 A: A normal domain is an integrally closed integral domain.

and A is integrally closed. This proves the “only if” part of the statement. P ROOF. Let ˛ be a root of g in some extension field of F . a root of f in some splitting field of f .d. which shows that ˛ 0 is integral over A.˛/ D . 1/mn Nm.  . As ˛ is a root of f ..˛/ 2 A. Let A be a normal domain with field of fractions F . .45. Then every monic factor of f . An element of E is integral over A if and only if its minimum polynomial over F has coefficients in A. Then Nm. Let A be a normal domain.˛/ in AŒ˛.45).˛/. F Œx ' F Œ˛. 2 C OROLLARY 1.˛/ 2 A. and let ˛ 2 E be algebraic over F . and let f . x $ ˛. and let ˛ 0 be a conjugate of ˛. n 1 m 0 D am . Then NmE=F . Let ˛ 2 E be integral over A. They lie in F . 1/mn am (see FT. 2 P ROPOSITION 1. so that ˛ m C a1 ˛ m 1 C    C am D 0.X / be a monic polynomial in AŒX .˛/ in AŒ˛. i. 5. Let A  B be rings. and so Nm. and so they lie in A.38) that the coefficients of f .X / are integral over A. 2 C OROLLARY 1.X / in F ŒX  has coefficients in A. If f is the minimum polynomial of ˛.47.46. As the coefficients of f are polynomials in the conjugates of ˛. i. it is integral over A. it follows from (1. and so g has coefficients in A. S 1 A0 is the integral closure of S 1 A in S 1 B.e. and ˛ divides NmE=F . Let A  F  E be as in the proposition. there is an F -isomorphism W F Œ˛ ! F Œ˛ 0 . Integral dependence 27 Let F  E be fields. Because f .44. and let ˛ be an element of E integral over A. and let E be a finite extension of the field of fractions F of A. P ROOF.f / ! F Œ˛. The minimum polynomial of ˛ over F is the monic polynomial of smallest degree in F ŒX  having ˛ as a root. and the “if” part is obvious. ˛. Then f is also the minimum polynomial of ˛ 0 over F . and so (see above). then the homomorphism X 7! ˛W F ŒX  ! F Œ˛ defines an isomorphism F ŒX =.X / be the minimum polynomial of ˛ over F .. Let f . For any multiplicative subset S of A. Let f . and let A0 be the integral closure of A in B. Then g is the minimum polynomial of ˛. some ai 2 A. m > 0.X / D X m C a1 X m 1 C    C am n where n D ŒEW F Œ˛ be the minimum polynomial of ˛ over F . P ROPOSITION 1.e. It suffices to prove this for an irreducible monic factor g of f in F ŒX .am ˛ C    C am and so ˛ divides NmE=F .˛/ D ˛ 0 : On applying  to the above equation we obtain the equation ˛ 0m C a1 ˛ 0m 1 C    C am D 0.˛/ D 0.˛ C a1 ˛ m D 1 C    C am / n 1 m 1 n 1 am 1 / C . P ROOF.

L EMMA 1.k . and let S be a multiplicative subset of A. Then  n  n 1 b a b a C 1 C    C n D 0: s s1 s sn for some ai 2 A and si 2 S . This shows that S 1 A0 is contained in the integral closure of S 1 B. s 2 S ) be integral over S 1 A. 2 Let E=F be a finite extension of fields. we find that nb s1    sn b 2 A0 . then it is integral over each Am . P ROOF. ˇ/ 7! TrE=F . 2 P ROPOSITION 1. and therefore equals A. 2 1 sn C OROLLARY 1.Tr. Let ˇ1 .Tr.˛ˇ/W E  E ! F (10) is a symmetric bilinear form on E regarded as a vector space over F . then the trace pairing (10) is nondegenerate. Let 1 .ˇi //2 : (by 11) . :::.k .28 1.50.˛.48). If E=F is separable. Recall (FT 5. (b) Ap is integrally closed for all prime ideals p. and hence lies in each Am . and so  n  n b a1 b C s s s 1 CC an D 0: sn Therefore b=s is integral over S 1 A. and therefore that bs D sss1 s 2 S 1 A0 .ˇi ˇj // D D D D P det. (c) Am is integrally closed for all maximal ideals m.ˇi //  det.ˇj // det.Pk k .45) that TrL=K . Then b n C a1 b n 1 C    C an D 0 for some ai 2 A. k k . P ROOF. P ROOF.ˇi ˇj // det. It follows that the ideal consisting of the a 2 A such that ac 2 A is not contained in any maximal ideal m. If A is integrally closed in B. let b=s (b 2 B. For the converse. we have det.ˇj // det.ˇi /  k .k . The implication (a))(b) follows from (1. Special case of the proposition in which A0 D A.48. then S 1 A is integrally closed in S 1 B.ˇi ˇj // of the trace pairing is nonzero. ˇm be a basis for E as an F -vector space. m be the distinct F -homomorphisms of E into some large Galois extension ˝ of F . We have to show that the discriminant det. P RELIMINARIES FROM COMMUTATIVE ALGEBRA P ROOF. Hence 1  c 2 A. If c is integral over A. On multiplying this equation by s n s1    sn . It remains to prove (c))(a). and (b))(c) is obvious. Let b s 2S 1 A0 with b 2 A0 and s 2 S . Then .49.ˇ/ D 1 ˇ C    C m ˇ (11) By direct calculation. :::. Let A  B be rings. The following conditions on an integral domain A are equivalent: (a) A is integrally closed.

We shall show that 0 Aˇ1 C Aˇ2 C    C Aˇm  B  Aˇ10 C Aˇ20 C    C Aˇm : Only the second inclusion proof. d  ˇm g is still a basis for E as a vector space over F . then A is also a field. Let fˇ1 . and it is integral over A: C    C an D 0. then p is maximal if and only if q is maximal.ˇ  ˇi / 2 A (1. (b) Let p D q \ A. so also is ˇ  ˇi . As ˇi and ˇ are in B. there exists a d 2 A such that d  ˇi 2 B for all i . Then there exist c1 . If A is Noetherian.ˇi  ˇj0 / D ıij for all i. but this contradicts Dedekind’s theorem on the independence of characters (FT 5. There exist free A-submodules M and M 0 of E such that M  B  M 0. (a) For every prime ideal p of A. it follows that i ci i . Clearly fd  ˇ1 .a On multiplying through by an 1. Let a be a nonzero element of A. :::.d. Let ˇ 2 B. But X X X Tr.14/ . there is a dual basis fˇ10 . T HEOREM 1.ˇj0  ˇi / D bj  ıij D bi : j j j Hence bi 2 A.ˇ/ D 0 for all ˇ 2 E. 2 P ROPOSITION 1.53 (G OING -U P T HEOREM ).51. If K is integral over A. Because the trace 0 g of E over F with the property pairing is nondegenerate. then M 0 is a Noetherian A-module.ˇj / D 0 all j: i P By linearity. : : : .i ˇj / D 0. ˇm that Tr. ai 2 A: we find that 1 C a1 C    C an an 1 D 0. a from which it follows that a 1 1 n 1 / 1 2 K. ˇm g be a basis for E over F . and so Tr. :::. Then ˇ can be written uniquely as a Prequires 0 linear combination ˇ D bj ˇj of the ˇj0 with coefficients bj 2 F . Integral dependence 29 Suppose that det.a 1 n / C a1 . Let A  B be rings with B integral over A.44).ˇ  ˇi / D Tr. and so we may assume to begin with that each ˇi 2 B. and we have to show that each bj 2 A. Then a . (12) If A is noetherian. and let B be the integral closure of A in a separable extension E of F of degree m.52. 2 A.40). bj ˇj0  ˇi / D bj Tr. Let A be a subring of a field K. cm 2 ˝ such that X ci i . 2 L EMMA 1. According to (1. j . P ROOF. there is a prime ideal q of B such that q \ A D p. Let A be a normal domain with field of fractions F . :::. and so B is finitely generated as an A-module. P ROOF. 2 . then B is a finite A-algebra.

For the converse. n \ A D p. x 2 M.x 0 . Then there exists a prime ideal q0 of B containing q and such that q0 \ A D p0 : q0 B q  A p  p0 : P ROOF. and hence surjective. 0 0 x. in 1935.x C x 0 . N . y/ D . 2 A SIDE 1. e Tensor Products Tensor products of modules Let A be a ring. y/. Here B=n is a field. Let A be a noetherian integral domain. 366 (2014).x. finding the example cost him a year’s hard struggle).51. Let p  p0 be prime ideals of A. y/ D a. then Lemma 1. F.52 shows that p is also. then the answer is yes (1. y/. 8. x 0 2 M. if  is A-linear in each variable. A map W M  N ! P of A-modules is said to be A-bilinear if . 2 C OROLLARY 1. y /. being a linear map of a finite-dimensional vector space).52. y/. which are automatically fields (left multiplication by an element is injective. y/ C . 8. (b) The ring B=q contains A=p.x. We have A=p  B=q. there exists a prime ideal q00 in B=q such that q00 \ . AMS. . y C y / D . (a) If S is a multiplicative subset of a ring A. I shall show that for all maximal ideals n of B.3). P RELIMINARIES FROM COMMUTATIVE ALGEBRA P ROOF. y 2 N.55.n \ A/ is a field.n \ A/. If q is maximal. However. then the prime ideals of S 1 A are in one-to-one correspondence with the prime ideals of A not meeting S (see 1.ax. note that any integral domain integral over a field is a field because it is a union of integral domains finite over the field.n \ A/. which is integral over its subring A=. According to the (1.x. and that p is its unique maximal ideal. and it is integral over A=p. Trans. and P be A-modules.x.14). To complete the proof of (a). and let M . a 2 A.x. ay/ D a. and B=q is integral over A=p. where S D A p. y2N x 2 M. The inverse image q0 of q00 in B has the required properties. Consider B=n  A=. for all proper ideals b of B.53). y 0 2 N . and let B be the integral closure of A in a finite extension E of the of field of fractions F of A..30 1. x 2 M. y. y/ C .K. In this case. .x. a 2 A. Thus we may assume that A is local.e. Let A  B be rings with B integral over A. Is B always a finite A-algebra? When A is integrally closed and E is separable over F . and n \ A will be equal to p if and only if A=.x. y2N . i. b \ A  p (otherwise b  A 3 1/.54. Schmidt found another example at about the same time (see Olberding. and let q be a prime ideal of B such that q \ A D p. 4067–4095). or A is a finitely generated k-algebra.A=p/ D p0 =p. It therefore suffices to prove (a) after A and B have been replaced by S 1 A and S 1 B. This follows from Lemma 1. no. Akizuki found an example of a noetherian integral domain whose integral closure in its field of fractions is not finite (according to Matsumura 1986.

M N / =K.M N / can be expressed uniquely as a finite sum X ai .x ˝ y/ D . x 2 M. x 0 2 M.M N / generated by the following elements . xi 2 M. y/. y/ 7! x ˝ y/ has the correct universal property because any bilinear map  0 W M  N ! T 0 defines an A-linear map A. 0 x. x 2 M. y /. and gives a commutative triangle. the universal property determines the tensor product uniquely up to a unique isomorphism. Tensor Products 31 An A-module T together with an A-bilinear map W M  N ! T  M N is called the tensor product of M and N over A if it has the following universal property: every A-bilinear map T 9Š linear 0 T 0:  0W M  N ! T 0 factors uniquely through .x. Thus each element A. This might be considered a mathematical shenanigan but if you start with the state vectors of two quantum systems it exactly corresponds to the notorious notion of entanglement which so displeased Einstein.y C y 0 / D x ˝ y C x ˝ y 0 a.x.x.x.x. y/ in M ˝A N . and define M ˝A N D A. y/ 0 . y 0 2 N . ay/ a. ai 2 A.x. y C y / . y/. y/ .M  N.x. which factors through A.xi ˝ yi /.x. and let A.net.x C x 0 .M ˝A N. Every element of M ˝A N can be written as a finite sum7 X ai .ax/ ˝ y D x ˝ ay: The pair . a 2 A. a 2 A. and all relations among these symbols are generated by the following relations . As usual. y/. . 7 “An element of the tensor product of two vector spaces is not necessarily a tensor product of two vectors. y/ 7! x ˝ yW M  N ! M ˝A N is A-bilinear — we have imposed the fewest relations necessary to ensure this. y2N .” Georges Elencwajg on mathoverflow. yi 2 N: Let P be the submodule of A.M ˝A N. xi 2 M. ai 2 A. but sometimes a sum of such.x.x 0 . y. y2N . We write it M ˝A N . T / ' HomA-linear .x C x 0 / ˝ y D x ˝ y C x 0 ˝ y x ˝ . y 2 N. x 2 M.M N / ! T 0 . . y/ .xi . T /: C ONSTRUCTION Let M and N be A-modules.M N / be the free A-module with basis M  N . yi 2 N.ax. Then . yi /.e. y/ a.M N / =P: Write x ˝ y for the class of . Note that HomA-bilinear .x.

the map c ˝ b 7! cbW k ˝k B ! B is an isomorphism. XmCn . We write it A ˝k B.kŒXmC1 . has the universal property characterizing k ˝k B. XmCn  . But this map can be identified with the bijection RmCn ! Rm  Rn : In terms of the constructive definition of tensor products. E XAMPLE 1. . There is a multiplication map A ˝k B  A ˝k B ! A ˝k B for which . R/ ' Homk . together with the given map k ! B and the identity map B ! B. R/  Homk-alg . Note that Homk . and the homomorphism c 7! c.kŒX1 .A. : : : . The maps a 7! a ˝ 1W A ! C and b 7! 1 ˝ bW B ! C are homomorphisms. : : : . : : : .B. Xm  ˝k kŒXmC1 . R/ ! Homk-alg . The algebra B. E XAMPLE 1. XmCn  ! kŒX1 . : : : .kŒX1 . R/  Homk . together with the obvious inclusions kŒX1 .32 1. : : : . : : : . A k-algebra C together with homomorphisms i W A ! C and j W B ! C is called the tensor product of A and B if it has the following universal property: for every pair of homomorphisms (of k-algebras) ˛W A ! R and ˇW B ! R. : : : .a0 ˝ b 0 / D aa0 ˝ bb 0 .57. : : : . P RELIMINARIES FROM COMMUTATIVE ALGEBRA Tensor products of algebras Let A and B be k-algebras. : : : . Xm  . XmC1 .A ˝k B. and they make A ˝k B into the tensor product of A and B in the above sense.kŒXmC1 .a ˝ b/. is uniquely determined up to a unique isomorphism by this property. This makes A ˝k B into a ring. XmCn . : : : . for every k-algebra R. XmCn  is an isomorphism. the map f ˝ g 7! fgW kŒX1 . R/: C ONSTRUCTION Regard A and B as k-vector spaces. In terms of the constructive definition of tensor products. : : :. To verify this we only have to check that. : : : . The ring kŒX1 . Xm . XmCn .56. R/ induced by the inclusions is a bijection. there is a unique homomorphism W C ! R such that ı i D ˛ and ı j D ˇ: i A ˛ C 9Š j B ˇ R: If it exists.! kŒX1 . : : :. XmCn  is the tensor product of kŒX1 . and form the tensor product A ˝k B. the tensor product. Xm  and kŒXmC1 . the map Homk-alg .1 ˝ 1/ D c ˝ 1 D 1 ˝ c makes it into a k-algebra.

if A D fx1 .A/ is a transcendence basis. (a) If . (b) Let k . Xm . Then m  n. Then ˝ ˝k kŒX1 . :::. gm /.63. 1. Assume that there is a finite subset A  ˝ such that ˝ is algebraic over k. Otherwise.b˛ / is a family of generators (resp. :::. Transcendence bases 33 R EMARK 1.62. Xn : If A D kŒX1 . Xn  ' ˝Œ1 ˝ X1 . Xn / 2 kŒX1 . A transcendence basis for ˝ over k is an algebraically independent set A such that ˝ is algebraic over k. A subset B of ˝ is algebraically dependent on A if each element of B is algebraic over k. 1. : : : . Xn  such that f .x/g.A/: 1. For a subset A of ˝. basis) for B as a k-vector space.61 (F UNDAMENTAL T HEOREM ). ˛n of a k-algebra A are said to be algebraically dependent over k there exists a nonzero polynomial f . : : : .˛1 . : : : . we let k.! ˝ be fields. :::. ˇn g be two subsets of ˝. the ˛i are said to be algebraically independent over k. :::.f ˝ g/. Assume (a) A is algebraically independent (over k). :::. y/ D f . : : : .x. ˛n / D 0. Xn =.g1 .x with 1 . (b) every subset A minimal among those such that ˝ is algebraic over k.y/ realizes A ˝k B as an algebra of k-valued functions on S  T . Now let ˝ be a field containing k. Elements ˛1 . : : : . see Chapter 9 of my notes Fields and Galois Theory. then ˝ ˝k A ' ˝ŒX1 .A/.A/ consists of the quotients fg. For the proofs.xm g f. (b) A is algebraically dependent on B (over k). 1 ˝ Xn  ' ˝ŒX1 . f Transcendence bases We review the theory of transcendence bases. g 2 kŒX1 . : : : . gm /: (c) If A and B are algebras of k-valued functions on sets S and T respectively. : : : .x1 .f. tr. : : : .58.59. Xn =. Then (a) every maximal algebraically independent subset of ˝ is a transcendence basis. degk ˝.:::.g1 . of ˝ over k).X1 . Let A D f˛1 . xm g. 1. (c) all transcendence bases for ˝ over k have the same finite number of elements (called the transcendence degree.60.xm g A. : : : . then . ˛m g and B D fˇ1 . then .A/ denote the smallest subfield of ˝ containing k and . then k.A/. .1 ˝ b˛ / is a family of generators (resp. basis) for A ˝k B as an A-module. 1. The reader should note the similarity of this to the statement in linear algebra with “algebraically” replaced by “linearly”. For example.:::. :::.

34 1. Show that an f 2 kŒX1 . 3X C 6Y C 3Z. Xn  that is identically zero on k n is the zero polynomial (i. Let k  L  ˝ be fields.e.degL ˝. if A is a transcendence basis for L=k and B is a transcendence basis for ˝=L. Show that a noetherian ring is normal if and only if it is a finite product of normal integral domains.X C 2Y C 1. 1-4.degk ˝ D tr..64). 2X C 4Y C 3Z/ in kŒX. then A [ B is a transcendence basis for ˝=k. Z. Let k be an infinite field (not necessarily algebraically closed). More precisely.X C 2Y. . What standard algorithm in linear algebra will allow you to answer this question for any ideal generated by homogeneous linear polynomials? Find a minimal set of generators for the ideal .64. A ring A is said to be normal if Ap is a normal integral domain for all prime ideals p in A.degk L C tr. Then tr. 1-2. 2X C 4Y C 3Z C 3/: 1-3. Prove the statement in (1. 3X C 6Y C 3X C 2. Y. Find a minimal set of generators for the ideal . : : : . P RELIMINARIES FROM COMMUTATIVE ALGEBRA 1. has all its coefficients zero). Exercises 1-1.

a1 .1.S / D V . Recall that the ideal a generated by a set S consists of the finite sums X fi gi . ai1 X1 C    C ai n Xn D 0.S /  V . : : : .a1 . Xn . If S is a set of nonhomogeneous linear equations. Therefore the algebraic subsets of k n can also be described as the zero sets of ideals in kŒX1 . We write An for k n regarded as an algebraic set. : : : . Note that S  S 0 H) V . Alternatively. Xn : V . gi 2 S: P Such a sum fi gi is zero at every point at which the gi are all zero. : : : . If S is a set of homogeneous linear equations. m.S / as the zero set of S./ D k n . : : : . and so V . i D 1. : : : . : : : .S / D f.a/. but the reverse conclusion is also true because S  a.S / of k n is the set of common zeros of some collection S of polynomials in kŒX1 . then V . Examples 2.S / is a subspace of k n . and so V .C HAPTER Algebraic Sets a Definition of an algebraic set An algebraic subset V ..S / is either empty or is the translate of a subspace of k n . ai1 X1 C    C ai n Xn D di . an / D 0 all f 2 Sg: We refer to V . Therefore k n is an algebraic subset. 35 2 . i D 1. An empty set of polynomials imposes no conditions. then V . k n is the zero set of the zero ideal . Thus V .S /  V . fi 2 kŒX1 .0/.a/ — the zero set of S is the same as the zero set of the ideal generated by S. m.S 0 /I — more equations means fewer solutions. Xn . an / 2 k n j f . : : : .

4a3 C 27b 2 ¤ 0.36 2. because a polynomial f . the theorem states that every ideal in kŒX1 . then V .X / in one variable X has only finitely many roots.11) and (4. 1. 0/. More precisely. but the Hilbert basis theorem shows that.S / is an elliptic curve.S / is a cone.5.6. .a/ consists of the single point . Xn  can be generated by a finite set of elements. : : : . The proper algebraic subsets of k are the finite subsets.2. A nonzero constant polynomial has no zeros. Some generating sets for an ideal will be more useful than others for determining what the algebraic set is.X 2 Y 2 D X3 C 1 1/ We generally visualize algebraic sets as though the field k were R. 2. Y 2 2Y C 1.0. A LGEBRAIC S ETS 2. If S consists of the single equation Z2 D X 2 C Y 2. i.17) below. Z 2 1 C Y: The middle polynomial has (double) root 1. 2. then V . If S consists of the single equation Y 2 D X 3 C aX C b. we didn’t require the set S of polynomials to be finite. For example: Y 2 D X. However.X 2 C Y 2 C Z 2 1. b The Hilbert basis theorem In our definition of an algebraic set. in fact. 2. from which it follows that V .3. we draw the real locus of the curve. For example.. X 2 C Y 2 Y. and so the empty set is algebraic.4. every algebraic set is the zero set of a finite set of polynomials. 2. this can be misleading — see the examples (4. X Z/ can be generated by1 X Z.e. and we have already observed that a set of generators of an ideal has the same zero set as the ideal. a Gr¨obner basis for the ideal. 1 This is. the ideal a D . in fact.

: : : . L EMMA 2. c The Zariski topology Recall that.cX i C    / 2 a. j /. a0 is called the leading coefficient of f .i /.i C 1/ because.i / D a.X / D a0 X r C a1 X r 1 C    C ar . The ring kŒX1 . All one can say in general. : : : . Also.deg f /. For each i  d . In other words. is that at least n dim. a  b H) V . then linear algebra shows that it is the zero set of n dim. Xn  D kŒX1 . Then b. V .deg f / D a.55). P ROPOSITION 2. and let a.f1 / < deg. P ROOF. There are the following relations: (a) V .1/  a. Continuing in this fashion.b/I . if V is a linear subspace of k n . and a. and for each .b/. For n D 1. the ring kŒX  is a principal ideal domain. how many equations are needed to define a given algebraic set V . : : : .c. : : : . there exists an fij 2 a with leading coefficient aij . 2 A SIDE 2. : : : . f1 D g1 C f2 with g1 2 b and deg.0/ D k n . Xn / D .i / denote the set of elements of A that occur as the leading coefficient of a polynomial in a of degree i (we also include 0).a/  V .7 (H ILBERT BASIS T HEOREM ). : : : . ai 2 . As we noted in the proof of (1.i /  a. Xn 1 ŒXn : Thus an induction argument shows that the theorem follows from the next lemma. Similarly.i /     eventually becomes constant. a0 ¤ 0. then so also is AŒX .10.i / for all i .d / D a. Determining exactly how many is an area of active research — see (3.V / polynomials are needed to define V (see 3. : : : . The ideal b of AŒX  generated by the (finitely many) fij is contained in a and has the property that b. let f be a polynomial in a.V / polynomials. we find that f D g C g1 C g2 C    2 b. ai 2 A. To see this. We shall show that every ideal in AŒX  is finitely generated.i /  a. there exists a finite generating set fai1 . and if equality holds for all i . Xn  is noetherian.f /.8. If A is noetherian. and a is finitely generated. kŒX1 .2/      a. Clearly.d / contains the leading coefficient of every polynomial in a).d C 1/ D : : : (and then a. The Zariski topology 37 T HEOREM 2. there exists a g 2 b such that deg.i / is an ideal in A. Xn . the sequence a. what is not quite the same thing.. a. Because b. for ideals a and b in kŒX1 . then b D a.a/ [ V . Therefore b D a.kŒX1 . f D g C f1 with g 2 b and deg.f2 / < deg. One may ask how many elements are needed to generate a given ideal a in kŒX1 .I (b) V .f /. As A is noetherian.f1 /.i /.a \ b/ D V .i. Let a be a proper ideal in AŒX .45). Xn . but often more are required. say a. ai ni g of a. or.ab/ D V . Recall that for a polynomial f . Let b be an ideal of AŒX  contained in a.f g/ < deg. if cX i C    2 a. which means that every ideal is generated by a single element.32).9. then X.

Thus (c) is obvious.a/ ¤ 0.X1 .ab/.ai / for every family of ideals . (a) The is obvious.a/ [ V . f .X1 . by definition.a \ b/  V . observe that if a … V . Xn  more closely. m P are inconsistent if there exist fi 2 kŒX1 . fi 2 ai . The Zariski topology has many strange properties. Xn  A point P D . 2 Statements (a). b H) V . The induced topology on a subset V of An is called the Zariski topology on V .a1 ..b/: For the reverse inclusions. P (c) Recall that. and (c) show that the algebraic subsets of An satisfy the axioms to be the closed subsets for a topology on An : both the whole space and the empty set are algebraic. but it is nevertheless of great importance. The next theorem provides a converse to this. Xn  ! k. there is a topology on An for which the closed subsets are exactly the algebraic subsets — this is called the Zariski topology on An . : : : . an / in k n defines a homomorphism “evaluate at P ” kŒX1 .ab/  V . .X1 . there are no disjoint nonempty open subsets at all). A LGEBRAIC S ETS P T (c) V . 2 Nullstellensatz = zero-points-theorem. and so a … V .fg/. Xn  such that fi gi D 1. : : : . if 1 2 . : : : . : : : .a/ ¤ 0. g. Xn .e. equations gi . (b) Note that ab  a \ b  a. Xn / 7! f . when the system of equations g. ai consists of all finite sums of the form fi . : : : . : : : .11 (H ILBERT N ULLSTELLENSATZ ). g 2 b such that f . the closed subsets are just the finite sets and the whole space. Xn / D 0. : : : . butP then . and so the topology is not Hausdorff (in fact. T HEOREM 2.g1 . then there exist f 2 a. (b). i 2I ai / D i 2I V . : : : . 2 Every proper ideal a in kŒX1 . : : : .38 2. : : : . : : : . gm / or. Note that the Zariski topologies on C and C2 are much coarser (have fewer open sets) than the complex topologies.g1 . : : : . gm / D kŒX1 . We shall see in (2.ai /i 2I . i D 1. an arbitrary intersection of algebraic sets is algebraic. has a zero in k n . . Xn / D 0. For the Zariski topology on k. : : : .. i.68) below that the proper closed subsets of k 2 are finite unions of points and curves.a1 . is “consistent”. an /. but first we must answer the question of when a collection S of polynomials has a common zero. Thus.b/. g 2 S.e. i.a/ [ V . equivalently. a finite union of algebraic sets is algebraic. Obviously. d The Hilbert Nullstellensatz We wish to examine the relation between the algebraic subsets of An and the ideals of kŒX1 .a/ ¤ 0. P ROOF.

P / D 0 all P 2 W g: . if K is finitely generated as a ring over k. xr  and so. The correspondence between algebraic sets and radical ideals 39 whose kernel contains a if P 2 V . We shall also need to use that kŒX  has infinitely many distinct monic irreducible polynomials. we have to show that there exists a k-algebra homomorphism kŒX1 . In the statement of the lemma. From (1. The next lemma accomplishes this. : : : . and its field of fractions k. P D . If K is finitely generated as an algebra over k. '. Therefore c N f is integral over kŒx1  by (1.14).x1 / is a subfield of K. But this contradicts the fact that that kŒx1  has infinitely many distinct monic irreducible polynomials that can occur as the denominator of an f in k. Xn  j f . Then K D kŒX1 . To complete the proof.43). is not algebraic over k.x1 /. : : : . : : : . a 2 k. Conversely. which implies that c N f 2 kŒx1  because kŒx1  is integrally closed in k.X1 /. we need to allow k to be arbitrary in order to make the induction in the proof work (but the generality will be useful later — see Chapter 10). The case r D 0 being trivial. cxr are integral over kŒx1 . namely.12 (Z ARISKI ’ S L EMMA ). we obtain a point P in V . : : : . then p1    pr C 1 is divisible by a monic irreducible polynomial distinct from p1 . xr are algebraic over k. : : : . : : : . it suffices to prove def this for a maximal ideal m. then at least one xi .) In other words. : : : . When k is finite.x1 /. and so the induction hypothesis implies that x2 . Then f 2 K D kŒx1 .x1 /.W / for the set of polynomials that are zero on W : I. Let f 2 k. : : : . r  1: If K is not algebraic over k. L EMMA 2. : : : . then K is algebraic over k. Xn C m/.e. the polynomials X a. c N f 2 kŒx1 . not necessarily algebraically closed.'. Xn =m is a field. 2 e The correspondence between algebraic sets and radical ideals The ideal attached to a subset of k n For a subset W of k n . we see that there exists a c 2 kŒx1  such that cx2 . and it is finitely generated as an algebra over k (with generators X1 C m. : : : . pr . cx2 . xr .Xn //: Thus. for a sufficiently large N . kŒx1  is a polynomial ring in one symbol over k. When k is infinite. the minimum number of elements required to generate K as a k-algebra. then it is finitely generated as a module. Clearly K is generated as a k. We shall prove this by induction on r. to prove the theorem.x1 /-algebra by x2 . : : : .x1 / (1. we can adapt Euclid’s argument: if p1 . : : : . are distinct and irreducible. we must show K D k. Let k  K be fields. from a homomorphism 'W kŒX1 . P ROOF. (Hence K D k if k is algebraically closed. xr .a/. cxr . Xn  ! k of k-algebras whose kernel contains a.40). pr are monic irreducible polynomials in kŒX . Since every proper ideal is contained in a maximal ideal (see p.38). say x1 . Xn =a ! k.a/. we may suppose that K D kŒx1 .W / D ff 2 kŒX1 . we write I. : : : . : : : . Then.

prime ideals (a fortiori.a/ is radical (2.X1 a1 .14.a/. In particular. Xn . 2 Radicals of ideals The radical of an ideal a in a ring A is def rad.X1 a1 . some r 2 Ng: P ROPOSITION 2.P / is a proper ideal.ar /s 2 a for some s. then a \ b is radical.a/.W /.13. without nonzero nilpotent elements. : : : . It says that every nonzero polynomial in kŒX1 . but a C b need not be: consider. Then a  I. Xn =.. : : : .a/.Wi /. : : : . (b) rad. but X 2 2 a C b.15b). it must equal mP : P ROPOSITION 2.a/  V I. If a and b are radical. Xn .W / is an algebraic set containing W . for example. I. : : : .a1 . X … a C b. Let W be a subset of k n . an /.a// D rad. When we expand .V /  I. This is true with k any infinite field (see Exercise 1-1).W / is the smallest algebraic subset of k n containing W . and hence is the smallest radical ideal containing a. (See 2.. P ROOF. we find that every term has a factor ar or b s . with say ar 2 a and b s 2 a. 2 An ideal is said to be radical if it equals its radical.a/ D ff j f r 2 a. P ROOF.X 2 Y / and b D . Thus a is radical if and only if the ring A=a is reduced. E XAMPLE 2. A LGEBRAIC S ETS Clearly.a C b/rCs using the binomial theorem. (a) The radical of a is an ideal. because “evaluation at . : : : .rad.W /. (a) If a 2 rad.P /  mP . maximal ideals) are radical. i. then clearly f a 2 rad. Xn an /.a1 . it is an ideal in kŒX1 .a/ for all f 2 A. a D .15. Note that rad. an /” defines an isomorphism kŒX1 .a/. then ars D . Certainly V I. they are both prime ideals in kŒX.W / D W if W is an algebraic set.W /I (b) I. Clearly I.k n / D 0 is (perhaps) not obvious. Xn an / ! k: As I. : : : . Only the statement I.) . Then V I./ D kŒX1 . and let mP D .19 below). : : : . Y . (b) If ar 2 rad.e. Wi / D I. Xn  is nonzero at some point of k n .k n / D 0I S T (c) I. Let V D V . Alternatively.40 2.X 2 C Y /. Let P be the point . There are the following relations: (a) V  W H) I. and so lies in a. but mP is a maximal ideal.22 below. Since integral domains are reduced.a/ be another algebraic set containing W . : : : . V I. Suppose that a. and so V . Let a be an ideal in a ring A. b 2 rad. it follows from the strong Hilbert Nullstellensatz (see 2.

: : : . : : : . m 1 Y h. i D 1. : : : . the equations are inconsistent. Xn . Y  such that 1D m X fi  gi C fmC1  . Y  ! k.X1 . according to the original Nullstellensatz. and . an . : : : . : : : . .X1 .e. : : : .a1 . Xn / Y 7! h 1 to the above equality.a/ defines a one-to-one correspondence between the set of radical ideals in kŒX1 . an / 2 V .16 (S TRONG N ULLSTELLENSATZ ). Xn /.a/ D a if a is a radical ideal. I V . : : : .W / is radical for every subset W  k n . an / D 0.17. Let N be the largest of the Ni . we have to show that if a polynomial h vanishes on V . We have already noted that I V .X1 . 2 C OROLLARY 2.X1 . On applying the homomorphism  Xi 7! Xi W kŒX1 . : : : . Xn hNi for some Ni . For every ideal a in kŒX1 . h 1 /D polynomial in X1 . h.a/  rad. Xn  and the set of algebraic subsets of k n . and so.1 Y h/ i D1 (in the ring kŒX1 . gm generate a.a1 . b/ doesn’t satisfy the last equation. Xn .X1 .  gi . : : : .a/. Xn / (*) i D1 in k.a/I in particular. Xn . Xn /  gi . : : : . Xn /.a/  rad. Therefore f r is zero on the same set as f . T HEOREM 2.P / D f . For the reverse inclusion. : : : . Xn . Xn / D 0: If . an . We may assume h ¤ 0. h 1 /  gi . : : : . Xn / D 0. : : : . and it follows that the ideal I. i D1 which shows that hN 2 a. The next theorem states that these two ideals are equal.a/. Xn . Clearly fi . Therefore. consequently.X1 . The map a 7! V .a1 . : : : .X1 .a1 . In particular. I V . we obtain the identity 1D m X fi . : : : . I V . : : : . f r .a/.a/. Xn . On multiplying (*) by hN we obtain an equation hN D m X (polynomial in X1 . The correspondence between algebraic sets and radical ideals 41 The strong Nullstellensatz For a polynomial f and point P 2 k n . b/ satisfies the first m equations. : : : .P /r . Y ). then . then hN 2 a for some N > 0. its inverse is I . : : : .X1 . Let g1 .a/ D rad. there exist fi 2 kŒX1 . P ROOF. : : : . : : : . : : : . and consider the system of m C 1 equations in n C 1 symbols.

The algebraic set V .a/  m.14). this shows that the maximal ideals of kŒX1 . Xn .a/ D rad. I.a/ ma 2 Remarks 2. Xn .W \ W 0 / must be the smallest radical ideal containing both I. : : : .k n / D I V .42 2.X1 a1 . : : : . : : : .a1 . ma For each P D .0/ D rad. P ROOF.22. : : : . Because maximal ideals are radical.21. Xn  (Nullstellensatz.I.a/. 2.17 is order reversing. Let a be an ideal in kŒX1 . Therefore the maximal proper radical ideals correspond to the minimal nonempty algebraic sets.a1 . I. and that V I.0/ D 0I in other words. Let W and W 0 be algebraic sets. Xn  is equal to the intersection of the maximal ideals containing it. : : : . and the minimal nonempty algebraic sets are the one-point sets. Xn  are exactly the ideals . We have shown that \ \ rad.a/.16). I and V are inverse bijections.11). If f 2 mP for all P 2 V . : : : . P 2V .W \ W 0 / D rad. 2 C OROLLARY 2. : : : . 2. Xn an / with .13).a/  mP  m. In fact. As W \ W 0 is the largest algebraic subset contained in both W and W 0 . : : : .a/. Therefore. Because V . The radical of an ideal in kŒX1 . Xn an / (see 2. the ideal mP D . only the zero polynomial is zero on the whole of k n ..19. The one-to-one correspondence in Corollary 2. : : : . an / 2 k n .W / C I.X1 a1 .13). But the maximal proper radical ideals are simply the maximal ideals in kŒX1 . and so f 2 I V . 2.a/.18.a/. Xn . every maximal ideal containing a also contains rad. Thus mP  a ” P 2 V .P / D 0 an / is maximal in (see 2. an // D . Xn kŒX1 .a/ is empty if and only if a D kŒX1 . We know that I V .a1 . : : : .W / and I. this holds whenever k is infinite (Exercise 1-1). : : : .20. As I.X1 a1 . an / 2 k n . and so \ rad. and f 2 mP ” f . A LGEBRAIC S ETS P ROOF.0/ D k n .W 0 //: .W / D W if W is an algebraic set (2. : : : .a/ D a if a is a radical ideal (2.W 0 /: I. 2. then f is zero on V .

: : : . 1 Conversely.a.W / is the closure of W . : : : . Thus we recover Corollary 2.Q/ consists of the algebraic subsets. gs g. : : : . But the strong Nullstellensatz shows that I . .24. : : : . f Finding the radical of an ideal Typically.a. Note that W \ W 0 D f.P/ consists exactly of the radical ideals. 1 kŒX1 . 0/g.1 Y h/  .X 2 V . from the identities 1 D Y N hN C . : : : . and (by definition) V .a/ if and only if 1 2 . These algorithms have been implemented in the computer algebra systems CoCoA and Macaulay 2. Xn  because two ideals with the same radical correspond to the same algebraic subset.Q/.25. P ROPOSITION 2. an algebraic set V will be defined by a finite set of polynomials fg1 . and (2. V (X 2 − Y ) • V (X 2 + Y ) 2. Then I W P ! Q and V W Q ! P define a simple Galois correspondence between P and Q (see FT.a/ in the course of proving (2.1 C Y h C    C Y N Y h/. g Properties of the Zariski topology We now examine more closely the Zariski topology on An and on an algebraic subset of An . but when realized as the intersection of Y D X 2 and Y D X 2 . P ROOF. gs /. then 43 Y / and W 0 D I.g1 .23. Y  generated by the elements of a and 1 Y h).1 Y h/ (the ideal in Y h/ implies h 2 rad. for each subset W of An . Xn . and then we shall need to find I.17) says that there is a one-to-one correspondence between the closed subsets of An and the radical ideals of kŒX1 . Xn . We saw that 1 2 .17. V I. Under this correspondence. there is an algorithm for deciding whether or not a polynomial belongs to the ideal.16). then 1 2 a C . Xn  containing I.X. Finding the radical of an ideal For example. A polynomial h 2 rad.f. Xn . : : : . Proposition 2. 1 N 1 h / 2 Given a set of generators of an ideal. it has “multiplicity 2”.W \ W 0 / D rad. : : : . Y / (assuming characteristic ¤ 2/. 7. Y / D . The algebraic subsets of An capture only part of the ideal theory of kŒX1 . There is a finer notion of an algebraic scheme over k for which the closed algebraic subschemes of An are in one-to-one correspondence with the ideals in kŒX1 . and hence an algorithm for deciding whether or not a polynomial belongs to the radical of the ideal.P/ and V .X 2 C Y /. Xn  (see Chapter 11 on my website). A SIDE 2. It follows that I and V define a one-to-one correspondence between I . let W D V .1 we see that.V /. There are even algorithms for finding a set of generators for the radical. the closed subsets of an algebraic set V correspond to the radical ideals of kŒX1 . if hN 2 a. Y N hN / D Y N hN C .18).14 says that.0.X 2 .V / D rad. Let P be the set of subsets of k n and let Q be the set of subsets of kŒX1 . : : : .

W /.W / — we have to show that either f or g is in I. A topological space having the property (b) is said to be noetherian.W / D a \ b. Let V be an algebraic subset of An . then there exists an infinite ascending chain of sets in U (axiom of dependent choice). either f is zero or g is zero.W / is prime. Then fg 2 a \ b D I. say W \ V . 2 . the empty space is not irreducible. but Bourbaki requires a compact space to be Hausdorff).a/ or V . But then f 2 I.10). Equivalently. Let W be an algebraic set such that I. the condition means that. We conclude that b  I. Hence W D .W /.f /. each has an open neighbourhood not containing the other. (c) Given an open covering of V .V2 /  : : :. If W ¤ V . In a Hausdorff topological space.W /. Let g 2 b.W /.26. P ROPOSITION 2.W /. which eventually becomes constant because kŒX1 . any two points have disjoint open neighbourhoods. A space having property (c) is said to be quasicompact (by Bourbaki at least.a1 .b/  V .V1 /  I. every nonempty open subset is dense. it will also be quasicompact. The ideal a \ b is radical. At each point of W . A sequence V1  V2     of closed subsets of V gives rise to a sequence of radical ideals I. : : : . The proof of (c) shows that every noetherian space is quasicompact. for every pair of distinct points. Since an open subset of a noetherian space is again noetherian.a/ [ V .b/ with a and b radical ideals — we have to show that W equals V .a/ [ V .a \ b/ D V . and V . and let fg 2 I.f / [ V . An algebraic set W is irreducible if and only if I. and so g 2 I. (a) We have seen that f. then there exists an f 2 a X I. must equal W . (b) Every ascending chain of open subsets U1  U2     of V eventually becomes constant. Let W be an irreducible algebraic set. P ROOF. A LGEBRAIC S ETS P ROPOSITION 2.I. P ROOF. contradicting (b). every descending chain of closed subsets of V eventually becomes constant. h Decomposition of an algebraic set into irreducible algebraic sets A topological space is said to be irreducible if it is not the union of two proper closed subsets. hence I. (c) Every open covering of V has a finite subcovering.b/ (2.27. let U be the collection of open subsets of V that can be expressed as a finite union of sets in the covering.W / (because I. The condition is equivalent to the following: every nonempty set of closed subsets of V has a minimal element.g/.W \ V . and let W D V . and so W  V . an /g is the algebraic set defined by the ideal . others call it compact. : : : . one of these sets.a/.W / is prime). 2 A topological space whose points are closed is said to be a T1 -space.W \ V .X1 a1 . (b) We prove the second statement. Xn  is noetherian. By convention. If U does not contain V . the only irreducible Hausdorff spaces are those consisting of a single point. : : : .W // D W . (a) The points of V are closed for the Zariski topology.g//: As W is irreducible.44 2. Xn an /. and so V .b/. Therefore.W / is prime.f // [ . Equivalent conditions: every pair of nonempty open subsets has nonempty intersection.

28. ng such that Vi  Wf .f / is irreducible. Since Vi  Wf . Let f 2 kŒX1 .f / D T . : : : . 2 . there is a function gW f1. L EMMA 2. Suppose that V cannot be written as a finite union of irreducible closed subsets. and so W D W1 [ W2 . etc.W2 [ : : : [ Wr /. Vi D Vi \ Wj for some j .32) that kŒX1 .fi / V .fi / distinct irreducible algebraic sets. Thus f is irreducible H) V . : : : . the statement is the definition of “irreducible”. there will be a nonempty closed subset W of V that is minimal among those that cannot be written in this way. Suppose that V D V1 [ : : : [ Vm D W1 [ : : : [ Wn S are two irredundant decompositions of V . Then fimi   fimi distinct primary ideals . : : : . ng ! f1. because V is noetherian. and so. Thus f and g are bijections. and the decompositions differ only in the numbering of the sets.i/ D i and Vi D Wf . : : : . with W1 and W2 proper closed subsets of W . which is a contradiction.29. similarly fg D id. and so W D W1 or W D .i / .31. Xn . Consequently. then the Vi are uniquely determined up to order.f / is a prime ideal if and only if f is irreducible (1. fi distinct irreducible polynomials. : : : . Suppose that r > 2. then W D W2 or W3 [ : : : [ Wr .fi / distinct prime ideals V . P ROOF. Then Vi D j . 2 P ROPOSITION 2.33). P ROOF.fi / . suppose f factors as Y m f D fi i . Then W D W1 [ . Then. because Vi is irreducible. Xn .i / for each i .h.Vi \ Wj /.i / . : : : . V D V1 [ : : : [ Vm . Hence W is also. : : : .30. each Wi is a finite union of irreducible closed subsets. mg ! f1. Because W was minimal. we must have gf . Xn  is a unique factorization domain. if the latter.f / D S V . Similarly. Let W be an irreducible topological space. If W D W1 [ : : : [ Wr with each Wi closed. Decomposition of an algebraic set into irreducible algebraic sets 45 S UMMARY 2. We saw (1. there is a function f W f1. Then V is a finite union of irreducible closed subsets. There are one-to-one correspondences (ideals in kŒX1 .j / for each j . But W itself cannot be irreducible.i /  Vgf .W2 [ : : : [ Wr /. then W is equal to one of the Wi . and so . mg such that Wj  Vg.f / D T rad. On the other hand. algebraic subsets of An ): radical ideals $ algebraic subsets prime ideals $ irreducible algebraic subsets maximal ideals $ one-point sets: E XAMPLE 2. Let V be a noetherian topological space. If the decomposition is irredundant in the sense that there are no inclusions among the Vi . When r D 2.

fimi / noted in Example 2. so that  V D V . then the pi are uniquely determined up to order (and they are exactly the minimal prime ideals containing a). which is connected but not irreducible. S P ROOF.f / D . Write V . If there are no inclusions among the pi . as in the corollary.a// D V . V .a/ \ V .fi / are the irreducible components of V .f1 / \ : : : \ . the primary decomposition is the decomposition . a D p1 \ : : : \ pn . V .a \ b/ D V .3 In Example 2. 3 In fact.f / with f D fimi .a/ [ V . 2 Remarks 2.f / is radical if and only if f is square-free.29. this becomes Q a simpler decomposition into prime ideals. .b/: 2. A LGEBRAIC S ETS The Vi given uniquely by the proposition are called the irreducible components of V .f /.f / D .a/ D niD1 Vi .34.fr /. T In a noetherian ring. : : : . An irreducible topological space is connected. The radical of an ideal a in kŒX1 .10b V . every proper ideal a has a decomposition into primary ideals: a D qi (see CA 19).Vi /. 2.a/ D p1 \ : : : \ pn because they are both radical ideals and [ \ 2. and . f D f1 : : : fr .36. C OROLLARY 2.35.29. Xn . Then rad.rad.pi / D V . : : : . A Hausdorff space is noetherian if and only if it is finite. An algebraic subset V of An is disconnected if and only if there exist ideals a and b such that a \ b D I. : : : . For example. 2.32.b/ . because the closure of an irreducible subset is also irreducible. An algebraic set with two irreducible components.a/ as a union of its irreducible components. Xn  is a finite intersection of prime ideals. and let pi D I.V / and a C b D kŒX1 . a principal ideal . in which case f is a product of distinct irreducible polynomials.46 2. Xn .a C b/ D V . D V .a/ D V . They are exactly the maximal irreducible closed subsets of V . T For an ideal . For radical ideals. In kŒX1 . p/: i The uniqueness similarly follows from the proposition. but a connected topological space need not be irreducible. they are exactly the maximal irreducible subsets of V . in which case its irreducible components are the one-point sets.X1 X2 / is the union of the coordinate axes in A2 .33. the V .

the coordinate ring of an algebraic set Let V be an algebraic subset of An . It is reduced (because a is radical). and so kŒV  is the ring of regular functions on V . : : : . Let W D V . .28) holds for ideals in kŒV  and algebraic subsets of V .37. The quotient maps kŒX1 . each D. and so b 7! V . Xn  containing a. Then b 7!  1 . and kŒX1 . It is empty if and only if h is zero (2. In other words.b/. P ROPOSITION 2. : : : . and then to its restriction to W . Xn  ! kŒV . Xn =a.a1 .b/ is a bijection from the set of radical ideals in kŒV  to the set of algebraic sets contained in V . (b) The closed subsets of V are in one-to-one correspondence with the radical ideals of kŒV . Clearly V . Regular functions. Now (2.b/ is a bijection from the set of ideals of kŒV  to the set of ideals of kŒX1 . and kŒV  D kŒx1 . For an ideal b in kŒV .b/ D fP 2 V j f . : : : .h/ is open. The coordinate ring of V is kŒV  D kŒX1 . : : : . .e.h/.b/. because its complement is the closed set V .P / D 0. Two polynomials f. the coordinate ring of an algebraic set V is the k-algebra generated by the coordinate functions on V . An f 2 kŒX1 .a/ ¤ 0g: It is an open subset of V .b/ ' kŒV =b). This is a finitely generated k-algebra. all f 2 bg — it is a closed subset of V . set V . (c) The sets D. : : : . under which radical. : : : . Xn  define the same regular function on V if only if they define the same element of kŒV . prime. Xn = 1 .b// D V .i.V / D a. but not necessarily an integral domain. the coordinate ring of an algebraic set i 47 Regular functions. Write  for the quotient map kŒX1 . g 2 kŒX1 . i. an / 7! ai is regular. : : : . (a) The points of V are in one-to-one correspondence with the maximal ideals of kŒV . : : : .19). set D. xn .. and maximal ideals (because each of these conditions can be checked on the quotient ring. prime. are a base for the topology on V .. The coordinate function xi W V ! k. Xn  defines a function P 7! f .P /W V ! k: Functions of this form are said to be regular. Xn  ! kŒV  D kŒV  kŒX1 .h//. 1 . : : : . and let I. For h 2 kŒV . h 2 kŒV . and maximal ideals correspond to radical. Xn  ! kŒW  D a b send a regular function on k n to its restriction to V . : : : .h/ D fa 2 V j h. and every open set is a (finite) union of this form.

b/. Tn .h0 // ” hr 2 . We sometimes write Vh for D. : : : .38.48 2. : : : . Let k. finite and quasi-finite maps A hypersurface in AnC1 is the algebraic set H defined by a single nonzero nonconstant polynomial: H W f .T1 .h/. The i th component of a map 'W W ! V is 'i D xi ı '.b1 . If f1 . Thus.P //W W ! V  k n : D EFINITION 2. and the set where a continuous function is nonzero is open. k Hypersurfaces.h//  rad. ' is the map P 7! '.h/ are called the basic (or principal) open subsets of V . then the zero set of a family of continuous functions f W V ! R is closed. some g: Some of this should look familiar: if V is a topological space.h/  D.V / is a prime ideal. 'n . A LGEBRAIC S ETS P ROOF. then U D D.P /.V / be its field of fractions — k.h/  V . (a) and (b) are obvious from the above discussion. a continuous map ' is regular if and only if f ı ' is a regular function on W for every regular function f on V . . For (c).h/ is open. we have already observed that D.'1 . Let V be an irreducible algebraic set. : : : . Then I.V / is called the field of rational functions on V . which we sometimes denote '  . X / D 0. with b an ideal in kŒV . Let xi denote the i th coordinate function .P / D . Every open subset U  V is the complement S of a set of the form V .h0 / ” rad.h0 / some r ” hr D h0 g.. and so kŒV  is an integral domain. fm generate b.fi /. 2 The D. As the coordinate functions generate kŒV . A continuous map 'W W ! V of algebraic sets is regular if each of its components 'i is a regular function. j Regular maps Let W  k m and V  k n be algebraic sets.h0 / ” V . Note that D. : : : . bn / 7! bi W V ! k on V .. Thus a regular map 'W W ! V of algebraic sets defines a homomorphism f 7! f ı 'W kŒV  ! kŒW  of k-algebras.

c/ such that c is a root of the polynomial a0 . tn / 2 k: Suppose first that a0 2 k.t / D ai .t1 .t1 . Xm .e. Tm . Then the fibre over t consists of the roots of the polynomial a0 X m C a1 .t /. D EFINITION 2. tn / 2 k n is the set of points . : : : . : : : .t /.t1 .71).T. let D be the discriminant of the polynomial a0 X m C a1 X m 1 C    C am : Then D 2 kŒX1 . x/ 7! . Tn .t / is a nonzero constant independent of t. except that over the smallest subset the fibre may be a full line. and X p . and the fibre has exactly m points over the open subset D ¤ 0.t. H is a cylinder over a hypersurface in An ). Let 'W W ! V be a regular map of algebraic subsets. ignoring the possibility that D is identically zero.39. there is one point . tn . for example. in which case the fibre is a line. : : : .t /X m 1 C    C am . for each t ¤ 0. Counting multiplicities. Consider. and fewer then m points over the closed subset D D 0. More precisely. there are exactly m of these.4 We can picture it schematically as follows (m D 3): H An Now drop the condition that a0 is constant. : : : . and f is a polynomial in T1 . Then the open set where D ¤ 0 is empty. X / D TX 1. For certain t . This case occurs when the characteristic is p ¤ 0. which means that some roots have “disappeared off to infinity”.W / is dense in V . (13) in kŒX .k. def ai .t /X m C a1 . so that a0 .. tn /: We can write f in the form f D a0 X m C a1 X m 1 C    C am . (b) The map ' is quasi-finite if ' 4 I’m 1 . : : : . : : : .t1 . tn . i.P / is finite for all P 2 V . The fibre of the map H ! An over .t1 . Worse. that X occurs in f (otherwise. : : : .t /X m 1 C    C am . 1=t /. There is a nested collection of closed subsets of An such that the number of points in the fibre (counting multiplicities) drops as you pass to a smaller subset. ai 2 kŒT1 . a0 ¤ 0. and let '  W kŒV  ! kŒW  be the map f 7! ' ı f . . m 2 N: We assume that m ¤ 0. the degree of (13) may drop. Hypersurfaces. but there is no point with t D 0 (see the figure p. finite and quasi-finite maps 49 We examine the regular map H ! An defined by the projection . (a) The map ' is dominant if '. for certain t all coefficients may be zero. : : : . f .

40. : : : . the projection map . Then '  W kŒV  ! kŒW  is injective. the hyperplane H m m 1 f . Let k be an infinite field (not necessarily algebraically closed). for every maximal ideal m of kŒV  there exists a maximal ideal n of kŒW  such that m D n \ kŒV . Xn C cn T. : : : .42.43. xn0 / is finite.X1 C c1 T. then its closure Z is a proper closed subset of V . XnC1 / D 0 be a hypersurface in AnC1 . The map H ! An just considered is finite if and only if a0 is constant. : : : . after a linear change of coordinates. if the image of ' is not dense.X1 C c1 XnC1 .x1 c1 xnC1 . and so there exists a nonzero regular function f zero on Z. : : : . : : : . Let HW f . and let f 2 kŒX1 . : : : . : : : . XnC1 / D a0 XnC1 C a1 XnC1 C : The next lemma shows that the ci can be chosen so that a0 is a nonzero constant. According to the going-up theorem (1. In terms of the coordinates xi0 D xi is the zero set of cxnC1 . a fortiori as a kŒV -algebra. Then f ı ' D 0. this implies that ' is surjective.x1 . Xn . This implies that the map H ! An defined by . then f ı ' D 0 H) f D 0: On the other hand. cn 2 k such that the map H ! An defined by .53). As kŒW  is finitely generated as a k-algebra. There exist c1 . : : : . 1 C    C am . and kŒW  is integral over the image of kŒV . We show that. xnC1 / 7! . P ROOF. 2 P ROPOSITION 2. P ROPOSITION 2. cn 2 k such that f .X1 .x1 . Xn C cn XnC1 . : : : . 2 l Noether normalization theorem Let H be a hypersurface in AnC1 . Let 'W W ! V be dominant and finite.x1 . : : : .x10 . and quasi-finite if and only if the polynomials a0 . P ROOF. : : : . P ROPOSITION 2. Xn . : : : . cn 2 k. : : : .x1 . P ROOF. xn /W AnC1 ! An defines a finite map H ! An . Because of the correspondence between points and maximal ideals.36). There exist c1 . If the image of ' is dense. T / D a0 T m C a1 T m with a0 2 k  and all ai 2 kŒX1 . T . am have no common zero in k n . A regular map 'W W ! V is dominant if and only if '  W kŒV  ! kŒW  is injective. xn cn xnC1 / is finite. : : : . xnC1 / 7! . A LGEBRAIC S ETS (c) The map ' is finite if kŒW  is a finite kŒV -algebra. to say that kŒW  is a finite kŒV -algebra means that it is integral over kŒV  (1. A dominant finite map is surjective. : : : . Let c1 . xnC1 / 7! . 2 L EMMA 2. Let f 2 kŒV .50 2.41.

If the xi are algebraically independent. cn . T . xd . : : : .l. Let F be the homogeneous part of highest degree of f and let r D deg. and so there exists a nonzero f 2 kŒX1 . xn .X1 . Xn . cd / 2 k. because the polynomial F . xd C1 .45). : : : . and so A is finite over kŒx1 . There exist elements x1 . By assumption. It is not necessary to assume that A is reduced in (2. T / is still homogeneous of degree r in X1 . If n > d . the next lemma shows that A is finite over a subring B D kŒy1 .c1 . : : : . and so the coefficient of the monomial T r can be obtained by setting each Xi equal to zero in F and T to 1. as required. T occurs in f . 2 In fact. T / D f . Then F . Xd . xd . xn 1 . : : : . the polynomial def g.X1 .X1 C c1 T. : : : . Let A be a finitely generated k-algebra. : : : . : : : . 1/ is a nonzero polynomial. : : : . xd . : : : . Xd .44. L EMMA 2. T HEOREM 2. and A is finite over C . : : : . : : : . T / . : : : . xn are integral over kŒx1 . : : : . Hence x1 .X1 C c1 T. and so we can choose the ci so that F . zd algebraically independent. Xn . every algebraic set V admits a finite surjective map to Ad for some d . 1/ ¤ 0 (Exercise 1-1). there exists a finite surjective map 'W V ! Ad . xd define a map V ! Ad . xd 2 A that are algebraically independent over k. : : : . the set fx1 . Xn . although the proof we give requires it to be infinite (for the general proof. We prove the theorem by induction on n. : : : .F /. T  such that f . : : : . cn . : : : . 1/ 2 k  . Xn C cn T. : : : . Xd . xn / D 0: (14) Because fx1 . As F . and such that A is finite over kŒx1 . : : : . xn  be a finitely generated k-algebra. xd g is algebraically independent. : : : .X1 .c1 . Noether normalization theorem 51 P ROOF. nor that k is algebraically closed.X1 .1). xd g be a maximal algebraically independent subset of fx1 . Xn C cn T. 1/T r C terms of degree < r in T.46. then there exist c1 . Let A D kŒx1 . : : : . : : : . : : : . xd  ! A is finite and injective. : : : . Now f . and so f . yn 1 . : : : . cn . Xd . then (14) shows that xn is integral over kŒx1 . with F . By induction. T / D F . : : : . 1/T r C terms of degree < r in T . If a0 … k. : : : . there is nothing to prove. T / is a nonzero homogeneous polynomial.c1 .45 (N OETHER N ORMALIZATION T HEOREM ). and m > 0. Let V be an algebraic set. Xd C cd T. : : : . which is finite and surjective because kŒx1 . P ROOF. : : : . xd . T HEOREM 2. : : : . then. : : : . : : : . cd 2 k such that A is finite over kŒx1 c1 xn .X1 C c1 T.c1 . cn . for a suitable choice of . T / D F . xn 1 . and let fx1 . Otherwise. Xn C cn T. : : : . For some natural number d . If a0 2 k. : : : . : : : . This follows from the next statement applied to A D kŒV : the regular functions x1 . see CA 8.c1 . xn g is algebraically dependent. : : : .x1 . xd cd xn . F . zd  with z1 .X1 C c1 T. a0 ¤ 0. B is finite over a subring C D kŒz1 . T / D a0 T m C a1 T m 1 C    C am with ai 2 kŒX1 . : : : . Let A D kŒx1 . xn 1 . xn g.

52 2.47. xd C1 . after possibly renumbering the coordinates. (15) cd xn . : : : .W / < dim. : : : . : : : . Vm be the irreducible components of V . Now fx1 . If dim. xd g is algebraically dependent on fx1 c1 xn . xd this shows that xn is integral over kŒx1 kŒx1 c1 xn . T / D bT r C b1 T C    C br with b 2 k  (see 2.V / < 1.V / C 1. : : : . : : : .50.x1 c1j xj .43).x1 . : : : . Therefore fx1 c1 xn . xn cd xn . inductively. D EFINITION 2. Assume that V is irreducible. xn g. xn / D 0 c1 xn .61). Let V1 . then dim. V has dimension  n if and only if every proper closed subset has dimension  n 1. the above argument can be modified to prove the following more precise statement: Let x1 .49.46 shows that there exist c1 . and so A is finite over 2 Remarks 2. 2. : : : . we could drop one of its elements. : : : . and.V /.V / D max.48. xd cd xn g. xn 1 g and we can repeat the argument. A LGEBRAIC S ETS takes the form g. xd cd xn . Then every chain W0  W1     in W extends to a chain V  W0     . then there exist cij 2 k such that the map X X . : : : .dim. xd C1 . Lemma 2. : : : . : : : . . xd cd xn g is a maximal algebraically independent subset of fx1 c1 xn . xd  1 as before. it has dimension  1 if and only if every proper closed subset is a point.x1 c1 xn . xn / 7! . : : : . If the second set were not algebraically independent. but this would contradict (1. : : : . Then (obviously) dim. Xd . : : : . we may suppose that fx1 . For an irreducible algebraic subset V of An . : : : . xd g is a maximal algebraically independent subset of fx1 . xd C1 .Vi //: i 2. xn 1 . xd c1j xj /W An ! Ad d C1j n d C1j n induces a finite surjective map V ! Ad : Indeed. The dimension of V is the supremum of the lengths the chains V0  V1      Vd of distinct irreducible closed subsets (the length of the displayed chain is d ).W /  dim. and so dim. : : : . xd cd xn . xd cd xn . : : : . and let W be a proper closed subspace of V . m Dimension The dimension of a topological space Let V be a noetherian topological space whose points are closed. : : : . : : : . As g.X1 . xn be the coordinate functions on V . Thus an irreducible topological space V has dimension 0 if and only if it is a point. cn 2 k such that kŒV  is finite over kŒx1 c1 xn .

n 2 N.. and so 1 2 m. T HEOREM 2. and let Sfxg denote the multiplicative subset of A consisting of the elements of the form x n . Indeed. The dimension of an algebraic set is its dimension as a topological space (2.52. We begin with two (a) For every x 2 A and maximal ideal m  A. and an n-dimensional algebraic set is called an n-fold.54.1 ax/.Afxg /  n P ROOF.Vi /: i When the Vi all have the same dimension d . P ROPOSITION 2. and let k be a subfield of A. Dimension 53 The dimension of an algebraic set D EFINITION 2. if x 2 m. if x n .1 ax/ 2 p. Vm are the irreducible components of V . otherwise x is invertible modulo m. Indeed. a two-dimensional algebraic set is called a surface. From this. A one-dimensional algebraic set is called a curve.A/. Then dim. 1 A/ 1: ' fp 2 Spec.V / D tr.53.V /: The proof will occupy the rest of this subsection. m \ Sfxg ¤ . : : : . while statement (b) shows that a maximal chain of length n is shortened only to n 1 when x is chosen appropriately. we have p \ Sfxg D . hence 1 ax 2 m. dim. Recall (1. Let A be an arbitrary commutative ring.S observations. degk F . then dim V D max dim. which is a contradiction. then x 2 m \ Sfxg . Then tr.A/  dim. Let x 2 A.g.A/: .V / D Krull dimension of kŒV : Note that. Let A be a ring and let n 2 N. Let A be an integral domain with field of fractions F . the proposition is follows. dim. 2 P ROPOSITION 2.28). we say that V has pure dimension d .A/  n ” for all x 2 A.48). if V1 . then 1 ax 2 p (as x … p/.14) that Spec. We write dim.A/ j p \ S D . for every x 2 m X p. Statement (a) shows that every chain of prime ideals beginning with a maximal ideal is shortened when passing from A to Afxg . Because of the correspondence between the prime ideals in kŒV  and irreducible closed subsets of V (2. and let p be a prime ideal contained in m. Let V be an irreducible algebraic set.m. (b) Let m be a maximal ideal.. and so there exists an a 2 A such that 1 ax 2 m. Then dim. degk k.51. a 2 A: The boundary Afxg of A at x is defined to be the ring of fractions Sfxg1 A.A/ for the Krull dimension of A.

Monthly 112 (2005).A/ D 1. A SIDE 2. N OTES .Afxg /  n 1. : : : . P ROOF. : : : . Clearly n Dtr.A/  n. The above proof of (2. Let V be an algebraic set. 9. this implies (by induction) that dim.A/: P ROOF. A is integral over a polynomial subring kŒx1 . implies that a chain of prime ideals in kŒx1 .X1 . degk F . : : : . A SIDE 2.54) completes the proof. Xn  has Krull dimension at least n. Now (2.55. Then k. It is not true in general that the points of an algebraic set of dimension n are parametrized by n-tuples. Xn  has Krull dimension n. there is nothing to prove. for rather trivial reasons — see mo79959. xn / D n. and so Afxg D 0. 2 C OROLLARY 2.A/ D dim. then 0 D 1 x 1 x 2 Sfxg .A/ D n 1 by (1.A/ D n 2 N.64). since k.Afxg /  n 1. xn  lifts to a chain in A. degk. Let V D An .58. The polynomial ring kŒX1 . Xn over k. A LGEBRAIC S ETS P ROOF.A/  dim. and so dim. The codimension of W in V is codimV W D dim V dim W: E XAMPLE 2. The inequality in (2.V / D k. In linear algebra. Now (2.X1 . Amer. The most we can say is that there exists a finite-to-one map to k n (see 2.A/. V has dimension d if and only if there exists a finite surjective map V ! Ad . and so tr. : : : . and so we assume that tr. Xn 1/      . : : : . Xn /.45).56. : : : .x/  Afxg . We argue by induction on n. then it is transcendental over k. xn .54 2. degk F . The existence of the sequence of prime ideals .57. and so dim. Math. Examples Let W be an irreducible algebraic subset of V .54). 2 A SIDE 2.x/ F . Thus. : : : . If A is finitely generated as a k-algebra.kŒx1 . The going up theorem (1. degk F . Let x 2 A.59.X1 /  . Now (2.54) completes the proof. If x 2 k.45 is the dimension of V .A/. .V / D n.X1 .54) may be strict.2 C OROLLARY 2. which has transcendence basis X1 .53) shows that dim. We can replace k with its algebraic closure in A without changing tr. then tr. If tr. If x … k. degk F . no. degk F . : : : .60. we justify saying that a vector space V has dimension n by proving that its elements are parametrized by n-tuples. According to the Noether normalization theorem (2.0/ shows that kŒX1 . again dim. : : : . Xn /  .45). 826–829. The d in Theorem 2. Let A be an integral domain and let k be a subfield of A.55) is based on that in Coquand and Lombardi.

V / > dim. because fX1 . Xn  D kŒ`1 . : : : . Xn =c ' kŒXi1 . E XAMPLE 2. Xi1 . : : : . : : : . Xn  generated by linear forms `1 . Xin r =c ' kŒXi1 . : : : . It suffices to prove this for an irreducible hypersurface H . `r .F . xn 1 g is a transcendence basis for k.m.65. : : : . : : : .f /. and so V . Then kŒX1 . If V is irreducible and W is a proper algebraic subset of V . More specifically. . : : : . : : : .63. Such an H is the zero set of an irreducible polynomial f (see 2. Y / and G.W / D .X. Xin r : E XAMPLE 2. xi D Xi C .57. Let kŒx1 . xn  D kŒX1 . which we may assume to be linearly independent. On the other hand. : : : . let c be an ideal in kŒX1 . kŒV  is canonically isomorphic to kŒXi1 . Y / be nonconstant polynomials with no common factor. : : : . Every nonempty algebraic set contains a point. : : : . `r . `r . Dimension 55 E XAMPLE 2. Xn =.x1 . : : : .64. Let Xi1 . xn / be the field of fractions of kŒx1 .50).) E XAMPLE 2. and so no nonzero polynomial in X1 . : : : .f /.F . if kŒV  is a unique factorization domain. Therefore. use 2. Xi1 . : : : . : : : . each element of one set can be expressed as a linear combination of the elements of the other. then dim. Xin r : This is obvious if the forms are X1 . kŒX1 . xn / over k. : : : . A hypersurface in An has dimension n 1. Xn .f /. say. Xr . xn .29).X. : : : .x1 . xn 1 are algebraically independent. : : : . Xi1 . E XAMPLE 2.f / for some f 2 kŒV . and let k. `r . which is a closed irreducible subset. Xn g and f`1 . This means that x1 . it is therefore a finite set.G. : : : . If V is a linear subspace of k n (or a translate of a linear subspace). Xi1 . : : : . Xn . Therefore an irreducible algebraic set has dimension 0 if and only if it consists of a single point. : : : . : : : . and so kŒX1 . Let F . Xin r g are both bases for the linear forms. : : : . Y // must have dimension zero. As f is not the zero polynomial. : : : . Y // \ V . some Xi . : : : . Xin r g is a basis for the linear forms in X1 .61. Then Xn occurs in every nonzero multiple of f . Xin r be such that f`1 . `r .66. xn is algebraic over k. and so fx1 . P ROPOSITION 2. occurs in it.X.X. : : : . Xn =c D kŒ`1 . Let W be a closed set of codimension 1 in an algebraic set V .X. Then V . (Alternatively.64). Y // has dimension 1 by (2.62. In the general case. then I. Xin r . xn 1 /. : : : .W / (2. : : : . : : : . Xn 1 belongs to . then the dimension of V as an algebraic set is the same as its dimension in the sense of linear algebra — in fact. Xid  where the Xij are the “free” variables in the system of linear equations defining V .x1 .

V . Show that if there exists a C-algebra homomorphism C ˝Q A ! C ˝Q B.X 2 . If V has dimension 1.) 2-4.f /  V (distinct irreducible closed subsets). Thus we may suppose that W is irreducible. which contradicts the hypothesis. X Y 2 /.X a. Let A and B be (not necessarily commutative) Q-algebras of finite dimension over Q.W / < dim.a/. P ROOF. Let p D I.W / D I. and let a be an ideal in kŒX1 .a/.W /. [Hint: Choose a basis . then I. the following is a complete list of the prime ideals in kŒX.fi /. A LGEBRAIC S ETS T P ROOF. then it is a point. Xn . Identify k mn with the set of m  n matrices. then it vanishes on VK . 2-5. Let W1 .63) it can’t be a proper subset of A2 . if V has dimension zero. Y : . Y b/ with a. If V has dimension 2. Xn . it is a prime ideal. gm 2 QŒX1 .ei /i 2I for K as a k-vector space. and so if we can prove I. The closed sets of codimension 1 in An are exactly the hyperplanes.] 2-6. and let r 2 N. Xn  such that f1 g1 C    C fm gm D 1. then V D V . (Hint: linear algebra). prove that there exist g1 . then there exists a Qal -algebra homomorphism Qal ˝Q A ! Qal ˝Q B. Xn .f / is prime.64) and (2. where W D . From (1. But then (2.56 2.0/ (distinct prime ideals) and hence W D V .a/ of a in k n is dense in the zero set P VK .33) we know . so it is A2 . Finally.67.f / where f is any irreducible polynomial in I. : : : . : : : .f // < dim V . .V / (see 2.W / D dim. : : : . (Hint: The proof takes only a few lines. t 2 .X 2 .68.63) dim. and that kŒV   kŒX  (polynomial ring in one variable). then (by 2. Let k  K be algebraically closed fields. 2 E XAMPLE 2. : : : . : : : . b 2 k: Exercises 2-1. 2-2. (Assume k has characteristic zero.66). If the fi have no common zero in C.W / D . Show that V is an algebraic subset of k n . then I.0/. and let Qal be the algebraic closure of Q in C. . If . t n / j t 2 kg. Therefore it contains an irreducible polynomial f . Deduce that the zero set V . Let f1 . : : : .p/  V . : : : .f1    fr /. We classify the irreducible algebraic sets V of A2 .t. 2-3.a/ of a in K n .Wi /. and it is not zero because otherwise dim. Let V D f. Check that it is the radical of . Correspondingly.V /. then we have p  .f / with f irreducible. Show that if f 2 KŒX1 . Find I. and write f D ei fi (finite sum) with fi 2 kŒX1 .f /  . Ws be the irreducible components of W .Wi / D . Xn  vanishes on V . 2 C OROLLARY 2. : : : . fm 2 QŒX1 .f / ¤ p . : : : . Show that the set of matrices with rank  r is an algebraic subset of k mn . Combine (2. X Y 2 /.66 and its proof).) .W /.

then M and N are isomorphic A-modules. where k is an infinite field. Dimension 57 2-7.z. and let M and N be A-modules. Show that the subset f. Show that if k al ˝k M and k al ˝k N are isomorphic k al ˝k A-modules.m. e z / j z 2 Cg is not an algebraic subset of C2 . . 2-8. Let A be finite dimensional k-algebra.

.

for disjoint open subsets Ui of V .U / is a sheaf of k-algebras if the following statements hold for every open U in V : (a) OV . for every open subset U of V we have a set OV .UP /. 59 3 . Then U OV . a Sheaves Let k be a field (not necessarily algebraically closed).C HAPTER Affine Algebraic Varieties In this chapter.U / to an open subset U 0 of U is in OV . Q Note that. Then OV is a sheaf of R-algebras. Recall that a function f W U ! R on an open subset U of Rn is said to be smooth (or infinitely differentiable) if its partial derivatives of all orders exist and are continuous. and for each open subset U of V let OV .U / if and only if every point P of U has a neighbourhood UP such that f jUP lies in OV . 3. if f. This is in preparation for Chapter 5. i.1.3. (b) the restriction of an f in OV . In this way. in other words. Then OV is a sheaf of R-algebras.U / contains the constant functions and.. Let V be a topological space.U /. condition (c) says that OV .U / is a k-subalgebra of the algebra of all k-valued functions on U .U / of functions U ! k.U / is local.Ui / for all i 2 I . Let V be an open subset of Rn . Let V be a topological space.U 0 /I S (c) a function f W U ! k lies in OV . they say that the condition for f to lie in OV . and suppose that. where we define an algebraic variety to be a ringed space that is a finite union of affine algebraic varieties satisfying a natural separation axiom. let OV . g lie in OV . Conditions (b) and (c) say that a function f W U ! k lies in OV . D EFINITION 3. we are led to the notion of an affine algebraic variety — roughly speaking.U / be the set of all continuous real-valued functions on U .2.U / ' i OV .U / be the set of all smooth functions on U . we define the structure of a ringed space on an algebraic set.e. Examples 3. and for each open subset U of V .Ui /.U / if there exists an open covering U D i 2I Ui of U such that f jUi 2 OV . OV . this is an algebraic set with no preferred embedding into An . then so also do f C g and fg.

In Chapter 11. Addison-Wesley. A germ of a function at P is an equivalence class of pairs . U 0 open in U. . For each open subset U of V . Recall that a function f W U ! C on an open subset U of Cn . f / and .U / be the set of all analytic functions on U . Let . 3. 3. U0 .1c) fails.P D lim OV . let OV . f / with U an open neighbourhood of P and f 2 OV . two pairs . for each open subset U of V . and let f be the function on U1 [ U2 that takes the constant value 0 on U1 and the constant value 1 on U2 . Hermann. the smallest such sheaf).6. The set of such power series is a C-algebra. If V is reducible.U.U.60 3. then OV is not a sheaf: let U1 and U2 be disjoint open subsets of V . and.P . and. A power series n0 an .V. We often write OP for OV. and I claim that it is canonically isomorphic to the stalk OV. another holomorphic function f 0 on a 1 Cartan. Moreover.U 0 .P . In other words. Then f has a unique power series expansion f D an .z c/n in some (possibly smaller) open neighbourhood of c (Cartan 19631 .V. for each open subset U of V . For historical reasons. and let P 2 V . P Let OV be the sheaf of holomorphic functions on V D C. we sometimes write . Henri.P is a local ring with maximal ideal mP the set of germs that are zero at P . is said to be convergent if it converges on some open neighbourhood of c. Let V be a topological space.U. OV / consisting of a topological space V and a sheaf of k-algebras on V will be called a k-ringed space (or just a ringed space when the k is understood). OV.U / (direct limit over open neighbourhoods U of P ). b Ringed spaces A pair . The germs of functions at P form a k-algebra OV. By definition. OV / be a k-ringed space. let OV .5.7. Let V be an open subset of Cn . then f is not in OV .U1 [ U2 /.V. Paris.U 0 . The OV is a sheaf of k-algebras. E XAMPLE 3.U / and call its elements the sections of OV over U .U /.6). and let c 2 C. Let . Then OV is a sheaf of C-algebras. OV /. let OV . OV becomes a sheaf of k-algebras (in fact. is said to be analytic (or holomorphic) if it is described by a convergent power series in a neighbourhood of each point of U . OV / for OV . 1963.U / be the set of all functions U ! k. II 2. OV. OV / be a k-ringed space. ! In the interesting cases. let f be a holomorphic P function on a neighbourhood U of c. and so condition (3.4. When “constant” is replaced with “locally constant”. and for each open subset U of V . A FFINE A LGEBRAIC VARIETIES 3.U / be the set of all constant functions U ! k. we encounter a more general notion of a sheaf of k-algebras. Let V be a topological space. To prove this. an 2 C. all our sheaves of k-algebras are subsheaves of this one.c of OV at c. called the stalk of OV at P . the restriction OV jU of OV to U . Elementary theory of analytic functions of one or several complex variables. is a sheaf of k-algebras on U . f 0 / are equivalent if the functions f and f 0 agree on some open neighbourhood of P in U \ U 0 .z c/n .

0 hh0 and so is regular at P .P / such that f 0 and f 0 agree with gh and gh0 respectively on a neighbourhood U 0 of P . The ringed space structure on an algebraic set 61 neighbourhood U 0 of c defines the same power series if and only if f and f 0 agree on some neighbourhood of c contained in U \ U 0 (ibid. We have to check the conditions of (3.P / ¤ 0 such that f .D.U / denote the set of regular functions on an open subset U of V . More formally: D EFINITION 3.P /m is well-defined and injective.h/.h/. h 2 kŒV  with h ¤ 0. there exist g. Let U be an open subset of V . A function f W U ! k is regular if it is regular at every P 2 U . The map U OV .c. h 2 kŒV : A pair g. P ROPOSITION 3.Q/= h. c The ringed space structure on an algebraic set Let V be an algebraic subset of k n .U / when U is a basic open subset of V .1). Recall that the basic open subsets of V are those of the form D.8. (b. a constant function is regular.Q/ ¤ 0g. If g= hm is zero on D. if gh D 0. Therefore gh is zero in kŒV . Thus we have a well-defined injective map from the ring of germs of holomorphic functions at c to the ring of convergent power series.3).Q/ for all Q in some neighbourhood of P . Then f C f 0 agrees 0 Cg 0 h 0 with ghhh on U 0 .h/.P / D 0 for all P 2 D.h/ ! k is zero if and only if and only if gh D 0 in kŒV . P ROOF. Suppose f and f 0 are regular on U . g 0 . and so g. then g.Q/ D g. then gh is zero on V because h is zero on the complement of D. h 2 kŒV  with h.P /= h.P /m W D. h 2 kŒV  with h ¤ 0 defines a function Q 7! g. Let g. and let P 2 U .P / ¤ 0 ¤ h0 . and so f C f 0 is regular at P .h/ ! k. By assumption. h0 2 kŒV . Conversely.11). L EMMA 3.P / D 0 for all P 2 V . 2 We next determine OV .h/ D fQ j h. ff 0 agrees with gg on U 0 . h. I 4. A function f W U ! k is regular at P 2 U if there exist g. P ROOF. h.P /= h. with h. Similarly.U / is a sheaf of k-algebras on V . which is obviously surjective. (a) Clearly.Q/ W D. 2 Let kŒV h denote the ring kŒV  with h inverted (see 1. Let OV . . The function P 7! g.h// sending g= hm to the regular function P 7! g.9.c) The condition for f to be regular is local. The lemma shows that the map kŒV h ! OV .P /h.Q/ A function on an open subset of V is regular if it is locally of this form.10.

aiN / D D.. and so h lies in rad.hj /..h21 . (*) V . the basis provided by the basic open subsets. after replacing gi with gi gi0 and hi with aiN . S We now have that D. By assumption D.h/ defined by ahiNgi hi : Let P be a point of D.hj /: hj2 m X ai gi hi D f hj2 hN : i D1 Since hj2 is never zero on D.ai /. OP D lim ! h.h/. For every P 2 V .e.h21 . to find that.hi / D D.h/ arises S from an element of kŒV h .h// is an isomorphism of k-algebras.ai /  D. i.D.hj /. Vi D D.hi /. On D.h2i /.h// D V . : : : . I claim that f is the function on D.f.. as claimed.D. we know that there is an open covering D.h/ is quasicomi g pact. h2m //.h/ D Vi and elements gi . it suffices to consider pairs .hj / D .hj /. By definition. In the definition of the germs of a sheaf at P . P ROOF. for example.ai / for some ai 2 kŒV ./ ai gi hi D i D1 m X . A FFINE A LGEBRAIC VARIETIES P ROPOSITION 3. hi hj2 gi D h2i hj gj : S S Because D. where mP is the maximal ideal I. and so aiN D hi gi0 for some N 2 N and gi0 2 kŒV  (see p. We have the following equalities in kŒV : hj2 m X . P ROOF. Therefore we have the following equality of functions on D. : : : .hi hj /.h/ D D.10) that hi hj . As ghi D hj on D.h/.h/ D D.hi / \ D. U / with U lying in a some basis for the neighbourhoods of P . we can assume that the covering is finite. P we can cancel on D. say D.hi /. i D1 gj hj But f jD.gi hj gj hi / D 0.hi / and that f jD. The above map kŒV h ! OV .hj / D D. i j we have (by Lemma 3. we can assume that Vi D D. Therefore.48). there is a canonical isomorphism OP ! kŒV mP .23) lim kŒV h ' kŒV mP : ! h…mP 2 .11) . Because D. Therefore.hj / equals that defined by ai gi hi . OV / ' (1. hi 2 kŒV  with hi nowhere zero on Vi such that f jVi D ghi .12.. It remains to show that every regular function f on D. h2m /: there exist ai 2 kŒV  such that hN D m X ai h2i : (**) i D1 P for some N . We may assume that each i set Vi is basic. the function f hN 2 C OROLLARY 3.hi / D ghi . it./ ai gj h2i hj D gj hj hN .11.P /¤0 (3. i.62 3. say. f D gi gi0 gi gi0 gi : D D hi hi gi0 aiN Note that D.hi /. Then P will be in one of the D.e.P /.ai /. f hj and gj agree as functions on D.

: : : . OV / purely in terms of A: ˘ V is the set of maximal ideals in A. N 2 N ˚ OP D gh 2 k. Following tradition.h/ is n o g 2 k. ˘ For each f 2 A.V / j h. all the rings attached to it can be identified with subrings of the field k. Xn / j h.V / defines a function on some dense open subset of V .P / D 0: 2 The space.f / D fm j f … mg.:::. and its maximal ideal consists of those g= h with g. OV / if U D D.U.X1 . the ring of regular functions on D. the local ring at P is ˚ OP D gh 2 k. As OV.P / ¤ 0 \ . in which case the statement follows from (3.P is an integral domain if P lies on a single irreducible component of V . and let A D kŒV . ˘ For f 2 Ah and m 2 D. For example. 3.V / j g 2 kŒV .D. Proposition 1. we call the elements of k.0/ can’t be prime. terminology is similar to that of “meromorphic function”. Xn /. : : : . OV / D Ah for all h 2 A. : : : .m/ denote the image of f in Ah =mAh ' k. OV / D hgN 2 k. On the other hand. Let V be an algebraic set and let P be a point on V . let D.14. The ring of regular functions on An is kŒX1 . Let V be an algebraic subset of k n .V.V / rational functions on V . 3. and OV is the unique sheaf of k-valued functions on V such that . Xn .Xn an / . let f . For a nonzero polynomial h.f / form a base. which is also not a function on the whole . OV / D OP \P 2U [ D .X1 . if P lies on more than one irreducible component of V . an /. n o . : : : . this says that there is a one-to-one correspondence between the irreducible closed subsets of V passing through P and the prime ideals in OP .14 shows that there is a one-to-one correspondence between the prime ideals of kŒV  contained in mP and the prime ideals of OP . N 2 N : 1 n 1 n N h For a point P D . Xn . 3.V /.P depends only on .X . we may suppose that V itself is irreducible.13). When V is irreducible. The ideal p corresponding to an irreducible closed subset Z consists of the elements of OP represented by a pair .D.13.h/.U.2 Examples 3.h/ ! k.hi /. In geometric terms. Propositions 2.17.P / ¤ 0 D kŒX1 .37 and 3. the topology on V is that for which the sets D.12). : : : . : : : . and so the ideal . in this way Ah becomes identified with a k-algebra of functions D. OV jU / for U an open neighbourhood of P . f / with f jZ\U D 0. : : : .X1 a1 .15.a1 . X .c. The local ring OV. then OP contains more than one minimal prime ideal (3.16.h/.U.h/.D.hi /: Note that every element of k. X / j g 2 kŒX .11 allow us to describe . The ringed space structure on an algebraic set 63 Remarks 3.

U / D kŒV : d Morphisms of ringed spaces A morphism of k-ringed spaces .64 3. it is a homeomorphism).V. Then. and these homomorphisms are compatible with restriction to smaller open subsets. a regular function f on U can be expressed f D g.W.X. Y /Y M D h. Y /).P for each P 2 V . it can’t divide the right hand side either. More precisely. Therefore. Y / h.18. If U is an open subset of V .0. which sends the germ represented by .U /. f / with f .V. Let U D A2 X f.P / into mP : it is a local homomorphism of local rings.U. because its complement f. Y /X N : Because X doesn’t divide the left hand side. the ring of regular functions on U is OU .X / [ D. OV / ! .f / for f ı '. Y . OV / ! . Y / D : XN YM We may assume that X − g and Y − h.U 0 /  U .P / D 0. for every pair of open subsets U  W and U 0  V with '. and so f 2 kŒX.P / ! OV.'.P / ! OV. Thus.X.X. and therefore can’t be of the form V .U. we find that g. A morphism of ringed spaces maps germs of functions to germs of functions. A FFINE A LGEBRAIC VARIETIES 3.' 1 .W.U /. we get a homomorphism of k-algebras f 7! f ı 'W OW . Since U D D. OV jV / ! . In the interesting cases.X. M D 0.U. We have shown that every regular function on U extends uniquely to a regular function on A2 : OA2 .64). On multiplying through by X N Y M . Sometimes we write '  .Y /. f / to the germ represented by .P is a local ring with maximal ideal mP consisting of the germs represented by pairs .P defined by ' maps m'. f ı '/.! V is a morphism of k-ringed spaces . It is an open subset of A2 .X.' 1 U / for all open subsets U of W .'.V. Similarly.U / D kŒX. and so N D 0. Y Y (intersection inside k. then the inclusion U . . a morphism 'W . OW / is a continuous map 'W V ! W such that f 2 OW .. 0/g has dimension 0. A morphism of ringed spaces is an isomorphism if it is bijective and its inverse is also a morphism of ringed spaces (in particular.U / H) f ı ' 2 OV . but it is not a basic open subset.U 0 / ! OV . Y X \ kŒX.f // (see 2. the homomorphism OW . 0/g.0. OV. OV /. OW / induces a k-algebra homomorphism OW.

Xn  ! A is a radical ideal. With these definitions.f / form a base. defines a function m 7! g. for every maximal ideal m  A. In particular. xn . the map k ! A ! A=m is an isomorphism. there is a topology on V for which the D. A pair of elements g. but replace R with C and “smooth” with “analytic”. 3. : : : . Recall from advanced calculus that a map 'W V ! W  Rm def is said to be smooth if each of its component functions 'i D xi ı 'W V ! R is smooth. We usually regard An as an affine variety.V.h/ ! k: h.18) implies that the intersection of the maximal ideals in A is 0.U / to be the set of functions f W U ! k that are of this form in some neighbourhood of each point of U . A map 'W V ! W is regular (or a morphism of affine algebraic varieties) if it is a morphism of k-ringed spaces. Since a similar statement is true for functions f on open subsets of W . For f 2 A. there exist xi 2 A such that A D kŒx1 .21.e. an / 7! ai W V ! R. Affine algebraic varieties 65 Examples 3.20). let D. and let xi be the coordinate function . Therefore (2. (2.V. We usually denote an affine algebraic variety .f / \ D. i. : : : . e Affine algebraic varieties We have just seen that every algebraic set V  k n gives rise to a k-ringed space .m/ D f (mod m/. we write f .m/ for the image of f in A=m D k.V. OV / to A by letting V be the set of maximal ideals in A. OV / and .m/ For U an open subset of V . h ¤ 0. then f ı ' is smooth for every smooth function f W W ! R.3).a1 .V. Let V and W be open subsets of Rn and Rm respectively. We usually shorten “affine algebraic variety” to “affine variety”. Moreover.19.f / D fm j f . h 2 A.V. OW /. An affine k-algebra is a reduced finitely generated k-algebra. Let V and W be topological spaces endowed with their sheaves OV and OW of continuous real valued functions (3.fg/ D D.m/ W D. OW / be affine algebraic varieties. f .12) implies that.. For f 2 A. the regular functions define the structure of an affine variety on every algebraic set. The affine varieties we have constructed so far have all been embedded in An . Let . Every continuous map 'W V ! W is a morphism of ringed structures . If ' is smooth. the affine algebraic varieties become a category. Same as (3. We attach a ringed space . OV / by V . . and the kernel of the homomorphism Xi 7! xi W kŒX1 . We now explain how to construct affine varieties with no preferred embedding.W. OV / ! . we see that a continuous map 'W V ! W is smooth if and only if it is a morphism of the associated ringed spaces (3.W. OV /. A k-ringed space isomorphic to one of this form is called an affine algebraic variety over k. 3. we define OV . Thus we can identify A=m with k. : : : .20.e.m/ ¤ 0g D fm j f … mg: Since D. For such an algebra A.g/.2).

Vol 11. A FFINE A LGEBRAIC VARIETIES P ROPOSITION 3.h/ : m h ˛. then S and T are homeomorphic. OV / is an affine algebraic variety with Ah for each h 2 A X f0g.22.A/ for the k-ringed space . Doklady 22. Xn =a D kŒx1 . 1939–1989..e. Represent A as a quotient kŒX1 . f The category of affine algebraic varieties For each affine k-algebra A. f . We now make this correspondence into an equivalence of categories. 3. .V.A/ for the topological space V .23. and so the homomorphism A ! B ! B˛. for such a space S .n/. For m D ˛ 1 . and conversely. It may seem strange to be describing a topological space in terms of maximal ideals in a ring.D.n/ '. 11-15.V.f /. For every h 2 A.h/ .g/ 7! W Ah ! B˛.h/ is invertible in B˛. 2 We write spm. Banach . xn . OV /. A SIDE 3.V.n/. the map def P 7! mP D ff W S ! R j f . the commutativity of the diagram means that. m D ˛ 1 . we have an affine variety Spm.V.P / D 0g is one-to-one correspondence between the points in the space and maximal ideals in the ring. f ı ' D ˛: 3 On rings of continuous functions on topological spaces. OV / is isomorphic to the k-ringed space attached to the algebraic set V .A/. Then . : : : .h/ extends to a homomorphism g ˛. OV /. we have an affine k-algebra kŒV  D . Thus ˛ defines a map 'W spm B ! spm A. (*) See also Allen Shields. Therefore. OV /. : : : . Intelligencer.n/ D m: D '.'. ˛.n/ D ˛ 1 . i. and Spm. we have a commutative diagram: A ˛ B A=m ' B=n: Recall that the image of an element f of A in A=m ' k is denoted f . for each affine variety . Math.n// D ˛. no. but the analysts have been doing this for more than 70 years.V. The pair .66 3.n/ is maximal in A because A=m ! B=n D k is an injective map of k-algebras which implies that A=m D k. p15. OV / ' P ROOF. Gel’fand and Kolmogorov in 19393 proved that if S and T are compact topological spaces.h/. We have attached to an affine k-algebra A an affine variety whose underlying topological space is the set of maximal ideals in A. for f 2 A.h/m For every maximal ideal n of B. Let ˛W A ! B be a homomorphism of affine k-algebras. The proof begins by showing that.m/.15). and the rings of real-valued continuous functions on S and T are isomorphic (just as rings). Algebras.a/ (see 3.

(c) f 2 kŒW  H) f ı ' 2 kŒV . it follows from (*) that ' 1 . OW / of affine algebraic varieties defines a homomorphism of the associated affine k-algebras kŒW  ! kŒV . the functor . It follows that f 7! f ı ' sends regular functions on any open subset of spm.˛. it is regular.W.h// defined by ˛. The functor A category of affine k-algebras to the category of affine algebraic varieties over k. Explicit description of morphisms of affine varieties Since ' 1 D. from (*) we see that f ı ' is the function on D. we have shown: P ROPOSITION 3.Spm.˛. OW /. (b) the components '1 . bn / 7! bi W W ! k: Hence this implication follows directly from the definition of a regular map. and (b) says that the map sends the coordinate functions yi on W into kŒV . Let f be a regular function on D.V. a morphism of 'W . Spm.h/ to regular functions on D. : : : .f ı '/ (obviously).A. Since these maps are inverse. ' Mor.A/ to regular functions on the inverse image of the open subset.˛. Let V  k m and W  k n be algebraic subsets.f // D D. (b) H) (c). kŒV /: In terms of categories. OV / ! . .A/. Thus ˛ defines a morphism of ringed spaces Spm. by definition.f 67 / D D.h//. Conversely. OV / is a quasi-inverse.h/. and so ' is continuous. For all affine algebras A and B.g. Proposition 3.f //. 'm of ' are regular functions on V . ' Homk-alg . g Explicit description of morphisms of affine varieties P ROPOSITION 3. By definition 'i D yi ı ' where yi is the coordinate function . : : : . (a) H) (b).kŒW .A//I for all affine varieties V and W .D. In particular. B/ ! Mor.V. Then.h/m .W. Since the yi generate kŒW  as a k-algebra.24.B/. and write f D g= hm .g/=˛. OV / . g 2 A. P ROOF. The map f 7! f ı ' is a k-algebra homomorphism from the ring of all functions W ! k to the ring of all functions V ! k.V. The following conditions on a continuous map 'W V ! W are equivalent: (a) ' is a morphism of ringed spaces . W / ! Homk-alg . this implies that it sends kŒW  into kŒV .V.b1 . OV / ! . and so f 7! f ı ' maps regular functions on D.B/ ! Spm.24 can be restated as: Spm A is a (contravariant) equivalence from the P ROPOSITION 3.26.25.V.

'. : : :/ D g.b/. they define a regular map 'W Am ! An .P // D ˛. ˛.a//  V .a1 . . : : : . Conversely.kŒW /. that sending Yi 7! fi . because ˛.b/  a. A FFINE A LGEBRAIC VARIETIES (c) H) (a). : : : .kŒW / with W .P / D 0. : : : . I claim that ' maps V . fn . they define a homomorphism ˛W kŒY1 .b/ if and only if ˛.P //: Now consider closed subsets V . When these conditions hold. f .68 3. we see that the regular maps V . : : : .P /.'.X1 .a/  Am and V . Then.Q/ D 0 ” f 2 mQ : Therefore ˛ 1 .f / 2 mP ” f . : : : . Indeed. suppose ˛. ' is the morphism of affine varieties V . ˛. : : : .b/. Xm : In particular. : : : . for f 2 kŒW . and it remains to show that this coincides with ' when we identify spm . they all extend to regular maps Am ! An . The map f 7! f ı ' is a homomorphism ˛W kŒW  ! kŒV . namely.P / D g. : : : . and let g 2 b.a/ ! V . am //: On the other hand. fn . Yn =b ! kŒX1 .f1 . Xm / ::: Yn D fn . Thus.f1 . Yn  ! kŒX1 . namely.P /. for f 2 a.b/ corresponding to the homomorphism kŒY1 . fi .kŒV / with V and spm .b/  An with a and b radical ideals. and let P 2 V .V .: : : .P /. Now consider equations Y1 D f1 . : : : .b/ are all of the form P 7! .a1 .X1 . 2 The equivalence of (a) and (b) means that 'W V ! W is a regular map of algebraic sets (in the sense of Chapter 2) if and only if it is a regular map of the associated affine algebraic varieties. which is what we needed to show.g/.P // D 0 ” f .b/  a. Let P 2 V . .f / 2 I V .f /.a/. Xn  of kalgebras.X1 . am / 7! .m / P D mQ . Xm =a defined by ˛.a/ into V .b/ D b. : : : . suppose '. : : : .'. fi 2 kŒX1 . for Q 2 V . : : : . am /. Xn /: This map coincides with g 7! g ı '. and so ˛.a/ ! V . let Q D '.Q// D 0. It therefore defines a map spm . Xm /: On the one hand.'. and let mP and mQ be the ideals of elements of kŒV  and kŒW  that are zero at P and Q respectively.g/.kŒV / ! spm .P //. and so '.a/.Q/ D g. : : : .a1 .P / 2 V .

g. Explicit description of morphisms of affine varieties

69

Examples of regular maps
3.27. Let R be a k-algebra. To give a k-algebra homomorphism kŒX  ! R is the same as
giving an element (the image of X under the homomorphism):
Homk-alg .kŒX ; R/ ' R
Therefore
3.24

Mor.V; A1 / ' Homk-alg .kŒX ; kŒV / ' kŒV :
In other words, the regular maps V ! A1 are simply the regular functions on V (as we
would hope).
3.28. Let A0 D Spm k. Then A0 consists of a single point and .A0 ; OA0 / D k. Every
map A0 ! V from A0 to an affine variety, and so Mor.A0 ; V / ' V . Alternatively,
Mor.A0 ; V / ' Homk-alg .kŒV ; k/ ' V
where the last map sends ˛W kŒV  ! k to the point corresponding to the maximal ideal
Ker.˛/.
3.29. Consider the regular map t 7! .t 2 ; t 3 /W A1 ! A2 . This is
bijective onto its image,
VW

Y 2 D X 3;

but it is not an isomorphism onto its image because the inverse
map is not regular. In view of (3.25), to prove this it suffices to
show that t 7! .t 2 ; t 3 / does not induce an isomorphism on the
rings of regular functions. We have kŒA1  D kŒT  and kŒV  D
kŒX; Y =.Y 2 X 3 / D kŒx; y. The map on rings is
x 7! T 2 ;

y 7! T 3 ;

kŒx; y ! kŒT ;

which is injective, but its image is kŒT 2 ; T 3  ¤ kŒT . In fact,
kŒx; y is not integrally closed: .y=x/2 x D 0, and so .y=x/
is integral over kŒx; y, but y=x … kŒx; y (it maps to T under the
inclusion k.x; y/ ,! k.T //.
3.30. Let k have characteristic p ¤ 0, and consider the regular map
p

.a1 ; : : : ; an / 7! .a1 ; : : : ; anp /W An ! An :
This is a bijection, but it is not an isomorphism because the corresponding map on rings,
p

Xi 7! Xi W kŒX1 ; : : : ; Xn  ! kŒX1 ; : : : ; Xn ;
is not surjective.
This is the famous Frobenius map. Take k to be the algebraic closure of Fp , and write F
for the map. Recall that for each m  1 there is a unique subfield Fpm of k of degree m over
m
Fp , and that its elements are the solutions of X p D X (FT 4.21). The fixed points of F m

70

3. A FFINE A LGEBRAIC VARIETIES

are precisely the points of An with coordinates in Fpm . Let f .X1 ; : : : ; Xn / be a polynomial
with coefficients in Fpm , say,
X
f D
ci1 in X1i1    Xnin ; ci1 in 2 Fpm :
If f .a1 ; : : : ; an / D 0, then
0D
pm 

X

c˛ a1i1    anin 

p m

D

X

p m i1

c˛ a1 

  anp

mi

n

;

pm

and so f .a1 ; : : : ; an / D 0. Here we have used that the binomial theorem takes the simple
m
m
m
form .X C Y /p D X p C Y p in characteristic p. Thus F m maps V .f / into itself, and
its fixed points are the solutions of
f .X1 ; : : : ; Xn / D 0
in Fpm .
A SIDE 3.31. In one of the most beautiful pieces of mathematics of the second half of the twentieth
century, Grothendieck defined a cohomology theory (´etale cohomology) and proved a fixed point
formula that allowed him to express the number of solutions of a system of polynomial equations
with coordinates in Fpm as an alternating sum of traces of operators on finite-dimensional vector
spaces, and Deligne used this to obtain very precise estimates for the number of solutions. See my
article The Riemann hypothesis over finite fields: from Weil to the present day and my notes Lectures
´
on Etale
Cohomology.

h

Subvarieties

Let A be an affine k-algebra. For any ideal a in A, we define
V .a/ D fP 2 spm.A/ j f .P / D 0 all f 2 ag
D fm maximal ideal in A j a  mg:
This is a closed subset of spm.A/, and every closed subset is of this form.
Now let a be a radical ideal in A, so that A=a is again reduced. Corresponding to the
homomorphism A ! A=a, we get a regular map
Spm.A=a/ ! Spm.A/
The image is V .a/, and spm.A=a/ ! V .a/ is a homeomorphism. Thus every closed subset
of spm.A/ has a natural ringed structure making it into an affine algebraic variety. We call
V .a/ with this structure a closed subvariety of V:
P ROPOSITION 3.32. Let .V; OV / be an affine variety and let h be a nonzero element of
kŒV . Then
.D.h/; OV jD.h// ' Spm.Ah /I
in particular, it is an affine variety.
P ROOF. The map A ! Ah defines a morphism spm.Ah / ! spm.A/. The image is D.h/,
and it is routine (using (1.13)) to verify the rest of the statement.
2

i. Properties of the regular map Spm.˛/

71

If V D V .a/  An , then
.a1 ; : : : ; an / 7! .a1 ; : : : ; an ; h.a1 ; : : : ; an /
defines an isomorphism of D.h/ onto V .a; 1
phism of affine varieties

1

/W D.h/ ! AnC1 ;

hXnC1 /. For example, there is an isomor-

a 7! .a; 1=a/W A1 X f0g ! V  A2 ;
with V equal to the subvariety X Y D 1 of A2 .

XY = 1

R EMARK 3.33. We have seen that all closed subsets and all basic open subsets of an affine
variety V are again affine varieties with their natural ringed structure, but this is not true for
all open subsets of V . For an open affine subset U , the natural map U ! spm .U; OU / is
a bijection. However, for
U D A2 X .0; 0/ D D.X / [ D.Y /  A2 ;
we know that .U; OA2 / D kŒX; Y  (see 3.18), but U ! spm kŒX; Y  is not a bijection,
because the ideal .X; Y / is not in the image. Clearly U is a union of affine algebraic
varieties, and in Chapter 5 we shall recognize it as a (nonaffine) algebraic variety.

i

Properties of the regular map Spm.˛/

P ROPOSITION 3.34. Let ˛W A ! B be a homomorphism of affine k-algebras, and let
'W Spm.B/ ! Spm.A/
be the corresponding morphism of affine varieties.
(a) The image of ' is dense for the Zariski topology if and only if ˛ is injective.
(b) The morphism ' is an isomorphism from Spm.B/ onto a closed subvariety of Spm.A/
if and only if ˛ is surjective.

72

3. A FFINE A LGEBRAIC VARIETIES

P ROOF. (a) Let f 2 A. If the image of ' is dense, then
f ı ' D 0 H) f D 0:
On the other hand, if the image of ' is not dense, then the closure of its image is a proper
closed subset of Spm.A/, and so there is a nonzero function f 2 A that is zero on it. Then
f ı ' D 0. (See 2.40.)
(b) If ˛ is surjective, then it defines an isomorphism A=a ! B where a is the kernel of ˛.
This induces an isomorphism of Spm.B/ with its image in Spm.A/. The converse follows
from the description of the closed subvarieties of Spm.A/ in the last section.
2
A regular map 'W V ! W of affine algebraic varieties is said to be a dominant if its
image is dense in W . The proposition then says that:
' is dominant ” f 7! f ı 'W .W; OW / !

.V; OV / is injective.

A regular map 'W V ! W of affine algebraic varieties is said to be a closed immersion
if it is an isomorphism of V onto a closed subvariety of W . The proposition then says that
' is a closed immersion ” f 7! f ı 'W .W; OW / !

j

.V; OV / is surjective.

Affine space without coordinates

Let E be a vector space over k of dimension n. The set A.E/ of points of E has a natural
structure of an algebraic variety: the choice of a basis for E defines a bijection A.E/ ! An ,
and the inherited structure of an affine algebraic variety on A.E/ is independent of the choice
of the basis (because the bijections defined by two different bases differ by an automorphism
of An ).
We now give an intrinsic definition of the affine variety A.E/. Let V be a finitedimensional vector space over a field k. The tensor algebra of V is
M
T V D
V ˝i D k ˚ V ˚ .V ˝ V / ˚ .V ˝ V ˝ V / ˚   
i 0

with multiplication defined by
.v1 ˝    ˝ vi /  .v10 ˝    ˝ vj0 / D v1 ˝    ˝ vi ˝ v10 ˝    ˝ vj0 :
It is a noncommutative k-algebra, and the choice of a basis e1 ; : : : ; en for V defines an
isomorphism
e1    ei 7! e1 ˝    ˝ ei W kfe1 ; : : : ; en g ! T  .V /
to T  V from the k-algebra of noncommuting polynomials in the symbols e1 ; : : : ; en .
The symmetric algebra S  .V / of V is defined to be the quotient of T  V by the twosided ideal generated by the elements
v˝w

w ˝ v;

v; w 2 V:

This algebra is generated as a k-algebra by commuting elements (namely, the elements
of V D V ˝1 ), and so is commutative. The choice of a basis e1 ; : : : ; en for V defines an
isomorphism
e1    ei 7! e1 ˝    ˝ ei W kŒe1 ; : : : ; en  ! S  .V /

Birational equivalence 73 to S  . en . This allows us to suppose that A and B have a common field of fractions K.E _ /.S  . If U is an open affine subvariety of V . : : : . there exists a d 2 A such that dxi 2 A for all i . fn  ! S  .3. Irreducible affine varieties V and W are birationally equivalent if and only if there exist open affine subvarieties UV and UW of V and W respectively such that UV  UW .E .S  .k.E _ /. P ROOF. P ROPOSITION 3.W /. : : : . then B  Ad . i / consisting of S  .V /.E/ to be Spm.E/ ! An whose map on the underlying sets is the isomorphism E ! k n defined by the basis of E.V / for its field of fractions.V / is an affine k-algebra.V / has the following universal property: every k-linear map V ! A from V into a k-algebra A extends uniquely to a k-algebra homomorphism S  . k Birational equivalence Recall that if V is irreducible. A/ _ . .V /  k. i / uniquely up to a unique isomorphism.36. : : : .A/.V / and the natural k-linear map i W V ! S  . xn generate B as k-algebra.E// ' Homk-algebra . then kŒV  is an integral domain.S  . A. : : : . Two irreducible affine algebraic varieties V and W over k are birationally equivalent if k. For an affine k-algebra A.E/. We use the isomorphism to identify k. this universal property determines the pair . and let A D kŒV  and B D kŒW .V /. Let V and W be birationally equivalent irreducible affine varieties. en for E determines a (dual) basis f1 .V / ! A: i V k-linear S  . A/ . Let x1 . This shows that S  .V / from the commutative polynomial ring in the symbols e1 . A. Now B  Ad  Be H) Be  Ade  . We now define A. and we write k. The map of algebraic varieties defined by this homomorphism is the isomorphism A. : : : . D EFINITION 3. As K is the field of fractions of A.Spm.16/ ' E ˝k A .V / (16) 9Š k-algebra A: As usual.35.V / is also the field of fractions of kŒU . fn of E _ .Be /e D Be .S  .linear algebra/: In particular. and so k.V / and k. and hence an isomorphism of k-algebras kŒf1 .V /.W /. then kŒV   kŒU   k.24/ ' Homk-linear . the choice of a basis e1 . Mor.E _ // where E is the dual vector space.k/ ' E: Moreover. The same argument shows that there exists an e 2 B such Ad  Be . The pair .

zd C1 will be separably algebraic over k. hence a transcendence basis (1. every p p polynomial in X1 . If F has characteristic zero. xd C1 / D 0 for some nonzero irreducible polynomial f . xd C2 / D k. : : : . : : : . as in the first paragraph of the proof. xn /. : : : . Let E be an algebraic extension of F . xd . : : : . and the “if” part is obvious. zd /.x1 .x1 . : : : . y. Xd C1 .74 3. xd /). if gcd. According to (3.x1 . after renumbering. xd g will be a transcendence basis for ˝ over k and xd C1 will be separable over k. Because k is perfect.X1 . : : : . Now ˝ D k. Xd C1  be an irreducible polynomial satisfied by the xi . there exist rational functions x1 .38. Thus. xd C1 /. : : : .63). xn /.z1 . xd are algebraically independent. xd C1 g. A FFINE A LGEBRAIC VARIETIES and so Ade D Be . we may suppose that @f =@Xd C1 ¤ 0. we may suppose that x1 . : : : . Let ˝ be a finitely generated field extension of k of transcendence degree d . zd g is a transcendence basis for ˝ over k (1. we may assume that k has characteristic p ¤ 0. : : : . : : : .f. : : : .x1 .x1 . xn are separable over k.1) shows that there exists a y 2 ˝ for which ˝ D k. and let H be the hypersurface f D 0. xd . : : : . After renumbering fx1 .de/  V and D. equivalently. y/. We have proved the “only if” part.e/  W are isomorphic.V /  k. and hence a pth power.x1 .1) shows that there exists a y 2 ˝ for which k.x1 . 2 Let F be a field. xd . Let ˝ D k. Let V be an irreducible variety of dimension d . P ROOF. : : : . After renumbering.63).37. y/. : : : . An element ˛ of E is separable over F if its minimum polynomial over F is separable. and so the primitive element theorem (FT. : : : . After renumbering. : : : . then xd C1 . Not all polynomials @f =@Xi are zero. 5. : : : . xd . xd / and xd C2 is algebraic over k. fx1 .x1 . then there exist x1 . xd /. : : : xd /. xd C1 / (hence over k.x1 . : : : . A polynomial f 2 F ŒX  is separable if it has distinct roots in any splitting field.x1 . 2 T HEOREM 3. Xd C1 / with coefficients in k. A finite extension E of F is separable over F if HomF . This shows that the open subvarieties D. and this implies that fz1 . Now xd C1 is separably algebraic over k. for otherwise p p f would be a polynomial in X1 . : : : . xd C3 . Let ˝ D k.x1 . : : : . xn /. and so the primitive element theorem (FT 5. zd C1 / for some zi 2 ˝. xd C1 on V such that k. : : : . f 0 / D 1. P ROOF. : : : . : : : . : : : . Then k. Every irreducible affine algebraic variety of dimension d is birationally equivalent to a hypersurface in Ad C1 . The field E is separable over F if every element of E is separable over F . xd C1 /.V / D k. contradicting the minimality of n. ˝/ has ŒEW F  elements for every suitably large extension ˝ of F . : : : . Xn with coefficients in k is a pth power in kŒX1 .H / and so V and H are birationally equivalent.x1 .38) below. Then f .z1 . We shall assume that n > d C 1 and obtain a contradiction. : : : .x1 .E. : : : . P ROPOSITION 3. 2 . We have shown that ˝ D k. xd C1 2 ˝ such that ˝ D k. : : : . or if E D F Œ˛ with ˛ separable over F . Xn : X i p ai1 in X11 : : : Xnin p D X 1 aip1 in X1i1 : : : Xnin p : (17) Let n be the minimum order of a generating set for ˝ over k. The argument in the last paragraph shows that.x1 . If k is perfect. Let f 2 kŒX1 .

As V . and let f be a nonzero regular function on V . Immediate consequence of (2. Because the Zj are closed.! kŒV  therefore induces an inclusion kŒAd = rad. there exists an open affine neighbourhood U of P in V not meeting any Zj with j ¤ i . D EFINITION 3.V /=k. the dimension d of an affine variety V is the maximum length of a chain V0  V1     of distinct closed irreducible affine subvarieties. Let V be an irreducible affine variety.f /. and so g m D f h for some h 2 kŒV . The Noether normalization theorem allows us to deduce this from the special case V D An .f / is a prime ideal p in kŒV . and so rad.45). In this section. The inclusion kŒAd  . If f has a zero in V . we find that g me D Nm. P ROOF.! kŒV  realizing kŒV  as a finite kŒAd -algebra. rad. Noether Normalization Theorem l 75 Noether Normalization Theorem D EFINITION 3.f jU / D Zi \ U . We have to show that dim Zi D dim V 1 for each i . m 2 Dimension By definition.Ad / f . the radical of . which is irreducible. T HEOREM 3. Let V be an affine algebraic variety of dimension n. According to the Noether normalization theorem (2. The dimension of an affine algebraic variety is the dimension of the underlying topological space (2. Let f0 D Nmk. we may assume that V . then its zero set is of pure codimension 1. there exists an inclusion kŒAd  .42.f /  p.45).f0 /  p \ kŒAd . A regular map 'W W ! V of affine algebraic varieties is finite if the map '  W kŒV  ! kŒW  makes kŒV  a finite kŒW -algebra. we prove that it is the length of every maximal chain of such subvarieties.! kŒV =p: 4 This proof was found by John Tate.41. Then g 2 p D rad.V / W k.39.40. T HEOREM 3.f0 / .f / is irreducible. m 2 N. There exists a point P 2 Zi not contained in any other Zj . 4 Let Z1 . def Let g 2 p \ kŒAd . Zn be the irreducible components of V . Hence f0 2 . P ROOF.f /. Then f0 2 kŒAd  and f divides f0 in kŒV  (see 1. proved in (2. as claimed.f0 /. on replacing V with U . . We claim that. : : : . Then there exists a finite map V ! An . Taking norms.h/ 2 .f / is irreducible. in fact.l. Therefore.45). Now V .64). and so g 2 rad.f h/ D f0  Nm.48). where e D Œk.f0 / D p \ kŒAd .An /.f0 /.

f / for otherwise it wouldn’t be maximal in V . C OROLLARY 3.F / D .45. : : : . P ROOF. 1 2 We can restate Theorem 3.76 3. then Z D Z 0 and codim. Also Z is an irreducible component of Z 0 \ V .f / be the zero set of f in V .Z 0 / C 1  r. : : : .Z 0 /  r 1.f1 .Z 0 /  r 1.42 shows that dim Z D dim V 1. 2 C OROLLARY 3. Because Z is a irreducible closed subset of V . it is contained in some irreducible component Z 0 of V . Then dim. Because Z is a proper closed subset of V .f0 /. fr / and Z is a maximal irreducible closed subset of V .V / 1. Every irreducible component Z of V . and f0 2 p ) f0 is nonconstant. fr /. We use induction on r. so that V .f0 /: Clearly f ¤ 0 ) f0 ¤ 0. the theorem shows that Z has codimension 1 in Z 0 . Let V be an irreducible affine variety. Thus Theorem 3. A FFINE A LGEBRAIC VARIETIES This makes kŒV =p into a finite algebra over kŒAd = rad. and so.Z/  r: For example. otherwise. there exists a nonzero regular function f on V vanishing on Z.Z/ D codim. if the fi have no common zero on V . codim. Let V .f1 .f1 .fr / because Z  Z 0 \ V . : pure codimension 1 otherwise. Let V be an irreducible affine variety.f /  V . from 2 C OROLLARY 3.f0 / D d by (2. and Z must be an irreducible component of V .42 as follows: let V be a closed irreducible subvariety of An and let F 2 kŒX1 .f1 . Every maximal (i.43. and let f1 . and codim. : : : . : : : fr 1 /. then there are no irreducible components. P ROOF. : : : fr 1 /. Then Z  V . If fr vanishes identically on Z 0 . then 8 if F is identically zero on V < V if F has no zeros on V V \ V . and so the fields of fractions of these two k-algebras have the same transcendence degree: dim V . and let Z be a maximal proper irreducible closed subset of V .44. we find that dim V0 D dim V1 C 1 D dim V2 C 2 D    D dim Vd C d D d: 1. Xn . By induction. Therefore dim V .Z/ D dim.43). Let V be an irreducible affine variety. : : : fr / has codimension at most r: codim.fr /  V .p/ D dim V . and the statement is vacuously true. The last set Vd must be a point and each Vi must be maximal in Vi (3.e.f1 ..V /.Z/ D codim. 2 . P ROOF. fr be regular functions on V .f1 . fr / is empty. nonrefinable) chain V0  V1      Vd (18) of distinct irreducible closed subsets of V has length d D dim. : : : .64). : : : .

Ym .45). . If an ideal a of a ring A is not contained in any of the prime ideals p1 . The element riD1 hi is therefore in a but not in any pi . fs 1 have been chosen.0. and Zs will be an irreducible component of Y1 \ V . every closed subset Z of codimension 1 is of the form V . then it is not contained in their union.46. When V is an affine variety whose coordinate ring is a unique factorization. . and so I.Y  the prime S i /.Zs / is not contained in any of the ideals I. /g [ f. Then there exist regular functions f1 .47.f1 .f1 .f / for some f 2 kŒV  (see 2. the components of V . fr 2 kŒV  such that.X1 . : : : . /g: it still has codimension 1 in V . 2 E XAMPLE 3. and it is possible for all of them to have codimension < r without any of the fi being redundant. fr / and all irreducible components of V . 0. Then V .X2 / \ V D f. and let Y1 . : : : . fr / need not all have the same dimension. Then hi D jr D1 fj lies in each pj with j ¤ i and a. pr . : : : . fs 1 /. because then we could omit it. V . f1 ¤ 0.f1 .48 (P RIME AVOIDANCE L EMMA ). and this is the function we want. In the setting of (3. numbered so that Zs 1 D Y1 .f1 . P ROOF. We prove this by induction on s. Dimension 77 E XAMPLE 3. : : : . P ROPOSITION 3. . X2 / \ V consists of a single plane f. Similarly. fs / have codimension s. and apply Theorem 3. choose an fi 2 a X pi and. : : : . We may assume that none of the prime ideals pi is contained in a second. : : : . P ROOF.Zs / X i I. For example. 0. fr on V such that Z is an irreducible component of V .X1 / \ V is the union of two planes: V .66). Zs is an irreducible component of V . Let Z be an irreducible closed subvariety of codimension r in an affine variety V . : : : . for each j ¤ i .f1 .X1 / \ V D f. 0. V . but not in pi (here we P use that pi is prime).Z1 /. : : : . : : : . /g: Both of these have codimension 1 in V (as required by (3. let V be the cone X1 X4 X2 X3 D 0 in A4 .0. We shall show that there exist f1 . 2 L EMMA 3.X2 / \ V is the union of two planes. We seek an element fs that is identically zero on Zs but is not identically zero on any Yi —for such an fs .f1 . for all s  r. . For a fixed i . . We know that there exists a chain of irreducible closed subsets V  Z1      Zr D Z with codim Zi D i . choose an def Q fj 2 pj X pi .49. fs / and all irreducible components of V .m. : : : . but it requires both X1 and X2 to define it.fs / will have codimension s. Now avoidance lemma (see below) tells us that there exist an fs 2 I. fr / have codimension r. But no Yi is contained in Zs because Zs has smaller dimension than Yi .42.0. For s D 1.42)). 0/g: However V . take any f1 2 I. 0.. be the irreducible components of V .fs /. all irreducible components of Yi \ V . . /g [ f. Suppose f1 . : : : .0. 0. Y2 .Yi / .

If p has height r.A/. fr 2 A such that p is minimal among the prime ideals containing . 3. : : : .f1 .f /  . see 1.51) may fail. fr /. but the ideal .42. If Z D V . In order to prove this. Theorem 3. 0. p has height 1.53. Restatement in terms of affine algebras We restate some of these results in terms of affine algebras.50. and . f D gq for some q 2 A. According to (3. In particular. which has codimension 2 in Z 0 . a noetherian ring may have finite Krull dimension d without all of its maximal ideals having height d . Because f is irreducible. then V . and so kŒX1 . q is a unit.42 says the following: let A be an affine k-algebra. Let p be minimal among the prime ideals containing . and let p be a prime ideal of height r in A.50). X  by inverting the elements of kŒt  X . every maximal ideal in A has height dim.47 says the following: let A be an affine k-algebra.53) are true for all noetherian rings (CA 21. : : : . X /  . 0. X3 . .0. A SIDE 3. For example.f jZ 0 / D f. Moreover.15 a noetherian ring may have infinite Krull dimension.3.tX 1/ is maximal (because A=. if A is an integral domain and f 2 A is neither zero nor a unit. . (3. as we noted on p. Thus Z is not of the form V . Proposition 3. 3. let A D RŒX  where R D kŒt .0/ is a maximal chain of prime ideals. /g: Z D f. 3.44 says the following: let A be an affine k-algebra.f / D . X2 . X4 =. The Krull dimension of A is 2. Statements (3. 0/g. 0/g. V W X1 X4 X2 X3 D 0 in A4 and let Z and Z 0 be the planes Z 0 D f.8). For example. and so . A FFINE A LGEBRAIC VARIETIES The condition that kŒV  be a unique factorization domain is definitely needed.13) and of height 1 (because it is in kŒt.f / has height 1 (principal ideal theorem).g/ D p — the element f is prime.f / for some regular function f on V .0.26). Let A be an affine k-algebra.t.X1 X4 X2 X3 / is not a unique factorization domain. and so it is principal.54.f /. if A is integral domain.. 0.g/.f /.52. then every prime ideal p minimal among those containing . which has codimension 2.0. 3. Corollary 3. say p D . if A is an integral domain and every prime ideal of height 1 in A is principal.t /  .50) and 3.tX 1/ ' R t . then every maximal chain pd  pd 1      p0 of distinct prime ideals has length equal to the Krull dimension of A. 0.g/.78 3. As . in violation of 3.t / is a discrete valuation ring with maximal ideal . 21. then there exist elements f1 .51. 0/g Then Z \ Z 0 D f. However. Again consider the cone. it suffices to show that every irreducible element f of A is prime (1. then A is a unique factorization domain. X  and A is obtained from kŒt. . : : : .t /.t /).

I. for each P 2 Z. there exists a germ of a function . p136. Obviously there are many ideal-theoretic complete intersections. Find the image of the regular map . f2 / for some f1 .x.m. no abelian variety of dimension > 1 is an ideal-theoretic complete intersection (being an ideal-theoretic complete intersection imposes constraints on the cohomology of the variety. Finally. 3-2. : : : . Pn / is said to be a set-theoretic complete intersection if there exist r polynomials fi 2 kŒX1 . fr on U such that U \ V D V .V \ U / can be generated by r regular functions on U .17). Thus U \ V is a set-theoretic complete intersection in U . p140).U 0 . 5 In . For example. f3 / for suitable polynomials f1 . it is not known whether two polynomials always suffice to define a curve in A3 — see Kunz 1985. Macaulay (the man.47 shows that a curve C in A3 is an irreducible component of V . f2 .f1 . and let Z D Spm. : : : . homogeneous polynomials fi 2 kŒX0 . 4. f 0 / at P on V such that f 0 jU 0 \ Z D f jU 0 \ Z. In 1891 Vahlen asserted that the result was best possible by exhibiting a curve in P3 which he claimed was not the zero locus of 3 equations. f2 2 kŒX.f1 . Exercise 2.f1 . a closed variety V of codimension r in An (resp.A=a/  Spm. thus refuting Vahlen’s claim which had been accepted for half a century.A/. : : : . that Perron gave 3 equations defining Vahlen’s curve. not the program) showed that for every r  1. but it is unknown whether all connected curves in P3 can be defined by two polynomials. fr /. : : : . Proposition 3. there is a curve C in A3 such that I. Chapter V of Kunz’s book is concerned with the question of when a variety is a complete intersection. Exercise 8. fr /: Such a variety is said to be an ideal-theoretic complete intersection if the fi can be chosen so that I.f1 . but most of the varieties one happens to be interested in turn out not to be. Note that ideal-theoretic complete intersection ) local complete intersection at all p: It is not difficult to show that a variety is a local complete intersection at every nonsingular point (cf. Show that a map between affine varieties can be continuous for the Zariski topology without being regular. One says that V is a local complete intersection at P 2 V if there is an open affine neighbourhood U of P in An such that the ideal I. Xn  (resp.C / requires at least r generators (see the same exercise in Hartshorne for a curve whose ideal can’t be generated by 2 elements).5 In general.V / D . : : : . Dimension 79 A SIDE 3. Z. The union of two skew lines in P3 can’t be defined by two polynomials (ibid.55.47) shows that there is a neighbourhood U of P in An and functions f1 . y/ 7! . Then (3. in 1973 Eisenbud and Evans proved that n equations always suffice to describe (set-theoretically) an algebraic subset of Pn (mo35468 Georges Elencwajg). Let V D Spm. Y. fr / (zero set in U /. Show that a function f on an open subset U of Z is regular if and only if. 3-3. which are not fulfilled in the case of abelian varieties). in 1941. Let P be a point on an irreducible variety V  An .A/. and f3 — this is an exercise in Shafarevich 1994 (I 6. 1882 Kronecker proved that every algebraic subset in Pn can be cut out by n C 1 polynomial equations. f2 . Xn / such that V D V . In fact C D V .f1 . It was only 50 years later.36). : : : . Apparently. xy/W A2 ! A2 and verify that it is neither open nor closed.x. Exercises 3-1. see also Hartshorne 1977.

A FFINE A LGEBRAIC VARIETIES 3-4.t 2 Is ' an isomorphism? 1. 3-5.t 2 1//W A1 ! C: . Let C be the curve Y 2 D X 2 C X 3 . but that neither is isomorphic to A1 .80 3. t . and let ' be the regular map t 7! . Show that the circle X 2 C Y 2 D 1 is isomorphic (as an affine variety) to the hyperbola XY D 1.

F .Fi / expresses V as a union of its irreducible components (see 2.X. b/ @X 81 and @F @Y b/ D 0: (19) .) Descartes In this chapter.X @X a/ C This is the equation of a line unless both the equation of a plane.a. We begin with the case of a plane curve. and then V D V . Moreover. y is an integral domain.V / over k. Y / D Q Fi .a. then x is transcendental over k and y is algebraic over k. then y is a transcendence basis for k.X.X. Y / D 0 We assume that F . Y //. b/. a Tangent spaces to plane curves Consider the curve V in the plane defined by a nonconstant polynomial F . @F .C HAPTER Local Study Geometry is the art of drawing correct conclusions from incorrect figures.V / over k. b/ is defined by the equation @F . and so x is a transcendence basis for k.64) and so V has pure dimension 1.X. if F ¤ Y c. Y // is a radical ideal and I. V W F . (La g´eom´etrie est l’art de raisonner juste sur des figures fausses. Y // D kŒx. b/ be a point on V . If we were doing calculus. we examine the structure of an affine algebraic variety near a point. so that .a. if F ¤ X c. If F .F . Y /. Let .V / D . we would say that the tangent space at P D .F .X. b/ are zero.Fi / has dimension 1 (by 2. We canS factor F into a product of irreducible polynomials. Y =.F .X. Y / itself is irreducible.a. Each component V .Y @Y @F .X. Y / has no multiple factors.x/. b/. Y /. in which case it is 4 .a.a. then kŒV  D kŒX. since the ideas in the general case are the same but the proofs are more complicated. Similarly.X.29).X.

Here again only . TP .a.a.1.F G/ @G @F DF  C  G. The tangent space at .0.a. 0/ is singular. we see that V is singular if and only if X 3 C aX C b has a multiple root.Y b/ D 0: All points on the curve are nonsingular unless the characteristic of k divides m. and we say that P is singular or multiple.6. In (2. 4. 2Y.a. 3.G/. b/ is given by the equation mam 1 a/ C mb m . c/ with c a common zero of X 3 C aX C b.82 4. The curve V is said to be nonsingular or smooth if all its points are nonsingular.F G/ @G @F DF  C  G: @Y @Y @Y . Otherwise.F / \ V . Y 2 D X 3 (sketched in 4. 4.X m0 C Y m0 1/p : 4.G/. 4. Xm C Y m 1 D X m0 p C Y m0 p 1 D . @F If @X .V / is a line through .X a/ C 2b.a.G/. 3X 2 C a. 1 . Then V D V . which consist of the points . in which case X m C Y m 1 has multiple factors. 4.Y b/ D 0: The only singular point is . Y 2 D X 2 . The tangent space at . The singular points of the curve are the common zeros of the polynomials Y2 X3 aX b. The tangent space TP V to V at P D . and we @Y say that P is a nonsingular or smooth point of V .V / has dimension 2. but we can define @X and @Y by  X  X @ X @ X aij X i Y j D aij X i Y j D i aij X i 1 Y j .12 below). @X @X @X @. X m C Y m D 1. the reader is invited to sketch the curve.2) we sketch two nonsingular examples of such curves. 3X 2 C a.11) we sketch two singular example. and a point . a singular point of V . then TP . This follows immediately from the product rule: @. 0/.F G/ where F G has no multiple factors (so F and G are coprime).2.3.X C 1/ (sketched in 4. b/ and @F .a. As 3X 2 C a is the derivative of X 3 C aX C b. L OCAL S TUDY @ @ We are not doing calculus. Examples For each of the following examples. b/ is given by the equation 3a2 .5. Y 2 D X 3 C aX C b. or a point of V . V D V .F / [ V . and in (4.4. b/.X 1 .10) and (4.0. Assume that char. b/ are not both zero. jaij X i Y j @X @Y and then the same definition applies. b/ is singular if and only if it is a singular point of V . D EFINITION 4.0.k/ ¤ 2. b/ is the algebraic subset defined by equation (19).F /.10 below).a.

this is impossible unless they are both zero. then F0 D 0 and (20) becomes F D aX C bY C higher degree terms. and so we may assume that V (hence F ) is irreducible. F /.7. It suffices to show that the nonsingular points form a dense open subset of each irreducible component of V . translate to the origin.V .X. If P D . The term F1 will be denoted F` and called the linear form of F . 0/ 2 V .X. In common usage. To obtain the tangent cone at any other point. A similar remark applies to @F=@Y . @X @F . The nonsingular points of a plane curve form a dense open subset of the curve. and hence both multiples of F . but. 0/ is the zero set of F . b Tangent cones to plane curves A polynomial F . Thus if @F=@X and @F=@Y are both zero. The tangent cone is the pair . Y / be a polynomial without square factors. and the equation of the tangent space is aX C bY D 0: D EFINITION 4. Y / can be written (uniquely) as a finite sum F D F0 C F1 C F2 C    C Fm C    (20) with each Fm a homogeneous polynomial of degree m. “singular” means uncommon or extraordinary as in “he spoke with singular shrewdness”. Y p . @F . We have to show that the set of singular points is a proper closed subset. Thus the proposition says that singular points (mathematical sense) are singular (usual sense). and let V be the curve defined by F . A SIDE 4. @Y it is obviously closed.0. called the singular locus.8. .0. If . then the geometric tangent cone to V at . as they have lower degree than F . and the first nonzero term on the right of (20) (the homogeneous summand of F of least degree) will be denoted F and called the leading form of F .0.9.F / where F is a nonconstant polynomial in kŒX. and then translate back. Let V D V . 2 Thus the singular points form a proper closed subset. Tangent cones to plane curves 83 The singular locus P ROPOSITION 4. Since it is the set of common zeros of the polynomials F. P ROOF. Y  without multiple factors.b. 0/ is on the curve V defined by F . Let F . then F is constant (characteristic zero) or a polynomial in X p . It will be proper unless @F=@X and @F=@Y are both identically zero on V .F /. Clearly @F=@X D 0 if and only if F is a polynomial in Y (k of characteristic zero) or is a polynomial in X p and Y (k of characteristic p/. These are contrary to our assumptions. and hence a pth power (characteristic p/.

Examples The following examples are adapted from Walker.X. Here the origin is a cusp.X.X. . Princeton Mathematical Series. The origin is again a double point. The origin is again a double point.11. In this case. multP .. the tangent cone also gives you information on the natural of a singularity.14. especially an old book. Robert J.. i. The origin is an isolated point of the real locus. F . F . 4.X. Y / D X 3 C X 2 Y 2 . Y / D X 3 Y 2 .10.e. Princeton. While the tangent space tells you whether a point is nonsingular or not. J. The tangent cone at . Y / D 2X 4 3X 2 Y C Y 2 2Y 3 C Y 4 . but this time it is a ramphoid cusp. The tangent cone is again defined by Y 2 .Y ai X /ri : P Then deg F D ri is called the multiplicity of the singularity. but the tangent cone is defined by Y 2 C X 2 . but this time it is a tacnode. 0/ is defined by Y 2 X 2 .12. Y / D X 3 X 2 Y 2 .0). the real locus of the tangent cone is just the point (0. 4.X. which is the pair of lines Y D ˙iX .13. 4. 4. Y / D X 4 C X 2 Y 2 2X 2 Y X Y 2 Y 2 .84 4. The tangent cone is again defined by Y 2 . N. vol.. F . 4.V /. 1950 (reprinted by Dover 1962). It is again a node.X. Algebraic Curves. In general we can factor F as Y F . F . ri D 1 all i . An ordinary double point is called a node. It is the pair of lines Y D ˙X . Princeton University Press. The tangent cone is defined by Y 2 .0. 13. Y / D X r0 . and the singularity is a node. F . There are many names for special types of singularities — see any book. A multiple point is ordinary if its tangents are nonmultiple. which is the X -axis (doubled). on algebraic curves. L OCAL S TUDY Note that the geometric tangent cone at a point on a curve always has dimension 1.

Y / D .X. Y /=. Y /=. A SIDE 4. and dimk . one translates the point to the origin. 4. and m=m2 D .64). 4. 1968. The local ring at a point on a curve 85 4. The tangent cone is defined by 4X 2 Y 2 . this proves the proposition in this case. b/ except that one uses the variables X 0 D X a and Y 0 D Y b. for its maximal ideal to be principal means .X.x 2 .x.c. Clearly dimk . and otherwise dimk .3) shows that the last condition is equivalent to n being a principal ideal. which is the triple of lines Y D 0. Then m D .X 2 C Y 2 /3 4X 2 Y 2 . y 2 /. in essence. Y //: In this quotient. Y // D kŒx. but its real locus is the image of R under the analytic map t 7! .0.17) is smooth even though the curve itself is singular.m=m2 / D 2.F . Let P be a point on a plane curve V .m=m2 / D 1 means for the local ring OP D kŒV m .t 3 C 2t. c The local ring at a point on a curve P ROPOSITION 4. which consists of a triple line Y 3 D 0 and a pair of lines Y D ˙iX . Another example of such a curve is Y 3 C 2X 2 Y X 4 D 0. xy.16.X. or mo98366 (Elencwajg). The origin has multiplicity 4.18. F . every element is represented by a linear polynomial cx C dy. 2 We explain what the condition dimk . The map m ! n induces an isomorphism m=m2 ! n=n2 (see 1.m2 C F . which is the union of the X and Y axes. F . Let n be the maximal ideal mkŒV m of this local ring.X. As OP has Krull dimension one (2.15). The tangent cone is defined byp3X 2 Y Y 3 . and hence an embedding into R2 with closed image.. t . y/ in kŒV  D kŒX.X.19.m=m2 / D 1 if F` ¤ 0. The origin is an ordinary triple point. The tangent cone is defined by X 2 Y 3 C Y 5 .15. and immersive.0.t 3 C 2//. Note that the real locus of the algebraic curve in (4. Y D ˙ 3X. The same argument works for an arbitrary point . Y 2 . Singular points of complex hypersurfaces. Assume first that P D . X Y.m=m2 / D 1. Y / D X 6 X 2 Y 3 Y 5 .X. F .a. proper. See Milnor. Since F` D 0 is the equation of the tangent space. Y / D . PUP.17. then dimk . and so we have P nonsingular ” dimk m=m2 D 1 ” dimk n=n2 D 1: Nakayama’s lemma (1. If P is nonsingular.m=m2 / D 2 otherwise. Y =.x. F .X 2 C Y 2 /2 C 3X 2 Y Y 3 . 0/. Y // D . y/ D 0. each doubled. and let m be the corresponding maximal ideal in kŒV . P ROOF.X. which is injective. y. This is singular at . J. Note that m2 D .X.X 2 . 0/. and the only relation is F` .

r0 mr D . : : : . By assumption. Thus X1 . Therefore . and so it is a unit. it is not nilpotent in A=p. s 2 N such that a 2 mr X mrC1 and b 2 ms X msC1 . Then a D u r and b D v s with u and v units. and so a D .0/. As a is finitely generated. Let a be a proper nonzero ideal in A. P ROOF. T and let m D . there exists an r 2 N such that a  mr but a 6 mrC1 .21. and ab D uv rCs ¤ 0. The statement summarizes the above discussion. A ring satisfying these conditions is called a discrete valuation ring.20.aij / be an n  m matrix. let Xi be the i th coordinate function a 7! ai .8). The second condition implies that c … m. (b) A has exactly one prime element up to associates. there exists an a D c r 2 a such that a … mrC1 . r /. We have shown that all ideals in A are principal. Therefore. r / D . for a point P on a curve. P ROOF.86 4. T HEOREM 4. L INEAR ALGEBRA For a vector space k m . Then A=p is an integral domain. L OCAL S TUDY that it is a regular local ring of dimension 1 (see 1. in which case it is a discrete valuation ring. there exists a prime ideal p properly contained in m. Thus. (c) A is local and is not a field.6).k m . It defines a linear map ˛W k m ! k n . Every regular local ring of dimension one is a principal ideal domain. A point P on a plane algebraic curve is nonsingular if and only if OP is regular. A linear form ai Xi can be regarded as an element of the dual vector space . d 2 Tangent spaces to algebraic subsets of Am Before defining tangent spaces at points of an algebraic subset of Am we review some terminology from linear algebra (which should be familiar from advanced calculus). P nonsingular ” OP regular. k/. Let A be such a ring. Let a and b be nonzero elements of A. According to the Krull intersection theorem (1. As  … p.k m /_ D Hom. P ROPOSITION 4. Xm is the P m dual basis to the standard basis for k . and hence not nilpotent in A. r / D mr . Hence A is an integral domain./ be its maximal ideal. There exist r. by 1 0 1 0 1 0Pm a1 a1 j D1 a1j aj B :: C B : C B C :: @ : A 7! A @ :: A D @ A: : Pm am am j D1 anj aj . 2 It follows from the elementary theory of principal ideal domains that the following conditions on a principal ideal domain A are equivalent: (a) A has exactly one nonzero prime ideal.a/  a  . Let A D .

Then m X Yi ı ˛ D aij Xj : j D1 This says that the i th coordinate of ˛.a/. so that Fj . otherwise it is singular (or multiple). .d.a/  . A point a on an algebraic set V is nonsingular (or smooth) if the dimension of the tangent space at a is equal to the maximum dimension of an irreducible component of V passing through a. it suffices to take theP F to lie in a generating subset for a. then the . Xm  at a by the equation: ˇ m X @F ˇˇ .dF /a D 0. . : : : .dF /a D .V /. Yn for the coordinate functions on k n .dXi /a for . am / of V is the subspace of the vector space with origin a cut out by the linear equations ˇ m X @F ˇˇ .Xi ai / D 0. and let a D I. : : : .dF1 /a . Tangent spaces to algebraic subsets of Am 87 Write X1 .dG/a D Hj . In terms of differentials.22. in (21) and (22).dFj /a C Fj .a/  .V / is the subspace of Ta . : : : .Am / defined by the equations: . D EFINITION 4.a/ D 0 for all j .dG/a D X Hj . The tangent space Ta .dFj /a : j Thus .Am / defined by the equations (21).dXi /a : @X ˇ i D1 i a It is again a linear form on Ta .a/ is m X aij .dXi /a form a basis for the dual vector space Ta .Am / — in fact.a1 . then . Let V  k m be an algebraic subset of k m . : : : . Ta . we define the differential of a polynomial F 2 kŒX1 .Am / is the vector space of dimension m with origin a.a/  .Am /. (21) @Xi ˇa iD1 In other words. As in advanced calculus. Ta . and Ta .Am /_ to Ta . Write .23. they are the coordinate functions on Ta . Fr generate a and a 2 V .V / is the subspace of Ta . F 2 a: (22) I claim that. The product rule for differentiation shows that if G D j Hj Fj .V / to V at a point a D . Xm for the coordinate functions on k m and Y1 .dFr /a generate the k-vector space f. : : : .dHj /a : j If F1 .dF /a j F 2 ag.Xj a/ D j D1 m X aij aj : j D1 TANGENT SPACES D EFINITION 4.Xi ai /.Am /_ . : : : . then this equation becomes X . F 2 a.

Let a 2 Am .a/ D . : : : . Pn . because it is the set where certain determinants vanish.F /. . : : : . and so the nonsingular points of V form a nonempty open subset of V . Xm /. : : : . @X1 @Xm dim. : : : . fa 2 V j rank J. @X @X1 m Then the equations defining Ta . L OCAL S TUDY Thus. .a/.d Yi /b ı . a 7! . For each r. relative to the standard bases.37) we shall show that the maximum value of rank J.F1 .a/  rg is closed in V .X1 .a/. Xm //. am /.a/ is m dim V .X1 . and is nonsingular on it. e The differential of a regular map Consider a regular map 'W Am ! An . Let a D . : : : .a1 .e. linear algebra shows that dimk Ta . i D 1. if V is a hypersurface. a point on V is nonsingular if and only if it lies on a single irreducible component of V . then   @F @F Jac. and the rank jumps on closed subsets. a point a on an irreducible algebraic set V is nonsingular if and only if dim Ta . Fr /.d'/a D @Xj ˇa i. Therefore.a/. : : : . Define . : : : n. : : : .a/ m B @X1: : :: :: Jac. Yn  ! kŒX1 .P1 .a/.d'/a W Ta .   @Xm B @X:1 @Fi :: C C :: DB J D Jac. @Xm .X1 . and so a is nonsingular if and only if the rank of Jac.Am / have matrix J. there is an open subset U of V on which rank J. : : : .F . Xm /.88 4.a/.a/ attains its maximum value. say I.F1 . : : : n: It corresponds to the map of rings '  W kŒY1 .V /. Therefore.. Pn /. Fr / D : A: @ @Xj @Fr @Fr .V / D .P1 . .a/ is equal to m For example.d'/a is the map with matrix 0 @P1 @P1 . : : : . and let 0 1 @F1 @F1 . Xm  sending Yi to Pi .V / D dim V .Am / ! Tb .a/. : : : .F1 . : : : . Later (4. : : : . @F and a is nonsingular if and only if not all of the partial derivatives @X vanish at a. @X . Fr /.dXj /a . As in the case of plane curves.V / D m rank J. and let b D '.V / as a subspace of Ta . : : : .a1 .An / to be the map such that X @Pi ˇˇ ˇ . : : : . : : : .a/ . i D 1. : : : .a/ D B @ @Pn @Pn .a/ @X1 1 C C: A . i We can regard J as a matrix of regular functions on V . : : : . am //: We think of ' as being given by the equations Yi D Pi .

: : : . b D '. whose differential will be a linear map .W /.cij /.0. then .e. f D f ..f.d'/a .a/: f Tangent spaces to affine algebraic varieties The definition (4. we obtain an identification of the differential of F (F regarded as a regular map) with the differential of F (F regarded as a regular function).Am / D k m and Tb . i D 1.22) of the tangent space at a point on an algebraic set uses the embedding of the algebraic set into An . L EMMA 4.V /: The chain rule holds in our situation: n X @f @Yj @f D .a/: When we identify Tb .V / into Tb . : : : . @Xi @Yj @Xi Yj D Pj . 0/ and b D . : : : .d'/a . which is zero if f 2 b because then f ı ' 2 a.b/  k n . and the map on tangent spaces is given by the matrix . b D '.W /. Thus .cij /t.X1 .V / ! Tb . Xm . then .24. b D '.t/ 2 Tb . Yn /: j D1 If ' is the map given by the equations Yj D Pj .f ı '/a D .df /b ı . In this section. If ' maps V D V .f ı '/a . We can regard F as a regular map Am ! A1 .W /. : : : .Y1 .An / D k n .t/. The usual rules from advanced calculus hold. 2 We therefore get a map .d'/a is zero on Ta . and have to prove that f 2 b ) . so that Ta . We are given that f 2 b ) f ı ' 2 a. For example. b D F . Let 'W Am ! An be as at the start of this subsection. suppose a D . Tangent spaces to affine algebraic varieties 89 For example.d /b ı .a/. 0/.Am / ! Tb . P ROOF.d'/a D d.0. : : : . : : : . Xm /. ı '/a . we give an intrinsic definition of the tangent space at a point of an affine algebraic variety that makes clear that it depends only on the local ring at the point. Xm /.a/: Let t 2 Ta . then the chain rule implies d. : : : . Let F 2 kŒX1 .X1 . and m X Pi D cij Xj C .dF /a W Ta .a/  k m into W D V . n.df /b ı .A1 / with k. it is simply t 7! .V /. .t/ D d.d'/a .d'/a W Ta .df /b ı .d'/a maps Ta . j D 1.d'/a D j cij Xj .A1 /. i. n: j D1 P Then Yi ı .higher terms). : : : . .

25).k/ equipped with a tangent vector. an C "bn / D " bi : @Xi ˇa Thus.kŒ"/. The tangent space to V at P is TP .a1 C "b1 .kŒV . Let V be an affine algebraic variety.V / (as defined in 4. : : : .25) if and only if . an / D 0 all f 2 ag: A homomorphism R ! S of k-algebras defines a map V . ˇ X @F ˇ ˇ . TP . a0 .R/ D f. To give a morphism Spm. and so mP D ˛ 1 . an / C @Xi ˇa with R a finite sum of products of at least two terms .kŒ"/ if and only if . : : : . According to (4.a1 .V /.a1 C "b1 .b1 .22).b1 . we define V . and .V / is a homomorphism of k-algebras ˛W kŒV  ! kŒ" whose "7!0 composite with kŒ" ! k is the point P .a C b"/.a1 .kŒ"/ is an element of TP . an /. Choose an embedding V .Xi ai / C R F . R/. . Let P 2 V . : : : . D EFINITION 4. Geometrically.a1 .S / of sets. then V . Thus kŒ" D k ˚ k" as a k-vector space. bn /" of An . On rewriting a polynomial F .a1 . and let V 0 D . Then the point . we obtain: ˇ  X @F ˇˇ F .V.kŒ"/ ! V amounts to giving a point P of V (the image of Spm.a1 . b.X 2 / where " D X C . . On setting Xi equal to ai C "bi in the above formula.V / (definition 4.! An .V / (definition 4.25).25.R/ ! V .V /. For example. : : : .kŒ"/ j P 0 7! P under V . an / C . : : : . b 0 2 k: Note that there is a k-algebra homomorphism " 7! 0W kŒ" ! k.k/g: Thus an element of TP .V 0 / (as defined in 4. bn / is an element of TP . 2 We can restate this as follows."//.a1 .ab 0 C a0 b/". OV / be V equipped with its canonical structure of an affine algebraic variety. an /.Xi ai /.kŒ"/ as being the point Spm. : : : . we can think of Spm. : : : .a0 C b 0 "/ D aa0 C . : : : .kŒ"/ ! V . Xn / in terms of the variables Xi ai . : : : . bn / 2 Ta ..b1 .R/ to be Homk-alg .26. Then TP . we obtain the (trivial Taylor) formula.V / D a and let P D .X1 . Let F 2 a. : : : . P ROPOSITION 4.X 2 /. The ring of dual numbers is kŒ" D kŒX =. and let P 2 V .V / D fP 0 2 V . and let P map to . an / 2 Rn j f . Xn / D F .k/) together with a tangent vector at P . P ROOF. Let I. : : : . if V  An and a D I. : : : .V 0 / consists of the elements a C "b of kŒ"n D k n ˚ k n " lying in V . To say that kŒV  ! k is the point P means that its kernel is mP .90 4.22) ' TP . : : : . a. Let P be a point on an affine algebraic variety V over k. Let V be an algebraic subset of An . an C "bn / lies in V .X1 . L OCAL S TUDY Dual numbers For an affine algebraic variety V and a k-algebra R (not necessarily an affine k-algebra).

b0 C D˛ .29.f .m/ A D k ˚ m.A. A k-derivation is a map DW A ! M such that (a) D. (c) D. 0 D . ! k is the identity map. m/ be as above.a/" .m//: Let ˛W A ! kŒ" be a local homomorphism of k-algebras.m/ C D. let f . 2 Derivations D EFINITION 4.28. if DW A ! k is a k-derivation. Let A be a k-algebra and M an A-module.OP . and so.a0 C D˛ . and ˛.f.g g.a/ D a0 C D˛ . and let P 2 V .b/ C b0 D˛ . and the map def f 7! df D f f .fg/ D f  Dg C g  Df (Leibniz’s rule).a0 D˛ . There is a canonical isomorphism TP . The composite k ! A k-vector space.ab/".a/˛. Because ˛ is a homomorphism of k-algebras.ab/0 C D˛ . and assume that A=m D k. We write Derk .a/"/. Then f f .f f . (b) D. k/ ! Homk-linear . There are canonical isomorphisms Homlocal k-algebra . By definition.f / C D. and write ˛. Tangent spaces to affine algebraic varieties 91 P ROPOSITION 4.m/ 2 m.m/g. We have ˛.V / ' Hom. Let .m/ C f  .fg/ C f  dg C g  df: P ROPOSITION 4.a/". P ROOF. and so ˛ extends (uniquely) to a homomorphism ˛ 0 W OP ! kŒ". For f 2 A.V / is a homomorphism ˛W kŒV  ! kŒ" such that ˛ 1 . kŒ"/ ! Derk .kŒ" X . when regarded as f $ .c/ D 0 for all c 2 k. an element of TP .b/ D . mod m2 .m//.m// D d. f f . then ˛W a 7! a.A. Therefore ˛ maps elements of kŒV  X mP into .a//": On comparing these expressions.m/.m// C g. A decomposes into f 7!f .m/.m// D fg C f .A. Conversely. k/ c7!c P ROOF.b/"/ D a0 b0 C .g/.g g.m/ mod m2 is a k-derivation A ! m=m2 because. and therefore is a k-derivation OP ! k.27. Let V be an affine variety.m=m2 .f f ."// D mP . Note that the conditions imply that D is k-linear (but not A-linear). By construction. and clearly every such homomorphism arises in this way from an element of TP . kŒ"/ (local homomorphisms of local k-algebras). For example."// D kŒ" ..m/ denote the image of f in A=m. let A be a local k-algebra with maximal ideal m. ˛ 0 is a local homomorphism of local k-algebras.f C g/ D D.ab/ D . M / for the k-vector space of all k-derivations A ! M . a0 D a.V /.A. we see that D˛ satisfies Leibniz’s rule.

A regular map 'W V ! W defines a map '. Let V be an affine algebraic variety. d' is induced by '  W OQ ! OP .m/ ˚ D˛ . To give an element of TP .kŒ"/ "7!0 "7!0 This map of tangent spaces is called the differential of ' at ' V . (a) By definition.89."// D mP . If Q D '.kŒ"/W V .OP .kŒV . k/ 2 Homk-linear .V / is the fibre of V . 4. then ' maps the fibre over P to the fibre over Q.kŒ"/ d' ' TQ . TP . kŒV  acts on k through kŒV  ! kŒV =mP ' k. It therefore defines a k-linear map m=m2 ! k. f 2 kŒV . kŒ"/ ! Hom. f ..W /: TP .W / W .OP .f / D f . (a) When V and W are embedded as closed subvarieties of An . and on the bottom row OP acts on k through OP ! OP =nP ' k. kŒ"/. it defines a map d'W TP . k/ 2 . Conversely.OP . 4.31.kŒ"/.V / Derk .f / 2 k: 2 The map D˛ is a k-derivation kŒV  ! k. There are canonical isomorphisms TP . k/: In the middle term on the top row. k/ Homk-linear . Similarly.V / V .nP =nP .e.P /.k/ W .mP / 2 k.V / amounts to giving a homomorphism ˛W kŒV  ! kŒ" such that ˛ 1 .V / ! TQ . L OCAL S TUDY is a local homomorphism of k-algebras A ! kŒ". (b) The homomorphism ˛ in (a) can be decomposed. d' has the description in p. a k-derivation kŒV  ! k extends uniquely to a k-derivation OP ! k. . A derivation DW A ! k is zero on k and on m2 (by Leibniz’s rule).15).. D˛ . (c) The homomorphism ˛W kŒV  ! kŒ" in (a) extends uniquely to a local homomorphism OP ! kŒ". a k-linear map m=m2 ! k defines a derivation by composition A f 7!df ! m=m2 ! k: 2 Tangent spaces and differentials We now summarize the above discussion in the context of affine algebraic varieties. (b) As a map Hom. kŒ"/ Derk .92 4. and let P be a point on V .30. and all such homomorphisms arise in this way.k/ P. i.kŒ"/ ! V .P in the local ring at P .OQ .k/ over P . and D˛ induces a k-linear map mP =mP ! k.f /". The maps have the following descriptions. ˛. Write mP for the corresponding maximal ideal in kŒV  and nP for the maximal ideal mP OV. 2 2 (d) The two right hand groups are related through the isomorphism mP =mP ! nP =nP of (1.mP =mP (23) Homlocal k-algebra .kŒ"/ ! W .

V /.X1 . : : : . : : : .a /. Xn /i =.V . it lies in a . Xn /i =. 0/ 2 V . : : : . The geometric tangent cone at P . : : : . Assume that P D . k/ ! Hom. : : : . and let a D I. Y Z. Fr /. the subspace of a consisting of homogeneous polynomials of degree i ).A/ D L mi =mi C1 (because of (1.Y 2 Z 2 // Z  . : : : .0.F /.32. and hence is radical.Y 2 Z 2 /: Let A be a local ring with maximal ideal n. The associated graded ring is M gr. The map kŒX1 . Xn =a ! gr.X1 .34. Xn =a /. : : : .OP / is an isomorphism.F /. Xn /i C1 C a \ .A. : : : .Y 2 Z 2 / D Y  . then a D . kŒX1 . Then X gr. E XAMPLE 4.X Y / lies in a and is homogeneous. XZ. d' is induced by the map mQ =m2Q ! 2 mP =mP defined by '  W kŒW  ! kŒV . Fr /. Xn =a corresponding to P .X Y.83).Y 2 Z 2 //. Then To .33. XZ C Z. Let E be a finite dimensional vector space over k. but if a D . Xn /i C1 C ai where ai is the homogeneous piece of a of degree i (that is. Xn /iC1 C ai D i th homogeneous piece of kŒX1 .X1 . CP . As the polynomial Y Z.V /. : : : . Tangent cones 93 2 (c) As a map Hom. Thus. P ROPOSITION 4. Xn =a to the class of Xi in gr. Let m be the maximal ideal in kŒX1 . : : : . : : : .mP =mP . P ROOF. then it need not be true that a D .E// ' E: A SIDE 4. CP .OP / sending the class of Xi in kŒX1 .g. k/.mQ =m2Q .X1 .OP / D mi =mi C1 X D .15)). XZ.F1 . Xn /i =.F1 .X1 . If a is principal. Define a to be the ideal generated by the polynomials F for F 2 a.kŒ"/ ! V should be thought of as a curve in V but with only the first infinitesimal structure retained. where F is the leading form of F (see p.XZ C Z. A map Spm.A/ D ni =ni C1 : i 0 Note that if A D Bm and n D mA. a is the ideal generated by X Y. then gr.V / is V . Obviously. and the tangent cone is the pair .a /. Xn /i X D . Consider for example a D . In fact.X1 . Xn =a : 2 . : : : . the descriptions of the tangent space provided by the terms in the top row of (23) correspond to the three standard descriptions of the tangent space in differential geometry (Wikipedia TANGENT SPACE). But .X1 . g Tangent cones Let V be an algebraic subset of k m .V /  TP . say a D . One can show that this is an intersection of prime ideals. : : : . : : : . : : : . but it is not in the ideal generated by X Y .

on an irreducible variety V of dimension d . : : : . fd are all defined on some open affine neighbourhood U of P . A point P on an affine algebraic variety V is said to be nonsingular or smooth if it lies on a single irreducible component W of V .e. : : : . i. : : : . where gr. The set of singular points of a variety is called its singular locus. fd in U . As in the case of a curve. Xn =a defined by a map of algebraic varieties is not the obvious one.14) that .f1 . If not. gr. there will be some irreducible component Z ¤ P of V . 2 P is nonsingular ” dimk TP . we obtain an open neighbourhood of P with the required property.OP /. we have . we define the geometric tangent cone CP .f1 .V / D TP . h Nonsingular points. : : : . Let V be an irreducible variety of dimension d . On removing the remaining irreducible components of V .34. : : : .OP //. and I claim that P is an irreducible component of the zero set V .V / ! CQ . it is necessary to use Proposition 4. Thus.f1 . fd / of f1 .f1 . P ROPOSITION 4.OP /red /. and let P be a nonsingular point on V . we find (using 1.OP / is a polynomial ring in dim V variables if and only if OP is regular.OP /red is the quotient of gr.p/ with p a prime ideal in kŒU . P is nonsingular (see below) if and only if gr. Let f1 .OP /. fd defined in an open neighbourhood U of P such that P is the only common zero of the fi on U .. fd / passing through P . A regular map 'W V ! W sending P to Q induces a homomorphism gr. otherwise the point is said to be singular. 2 More precisely. . Write Z D V . : : : fd on U such that . Because V . Therefore. : : : .V / D dim W . Moreover. fd /  pOP ¤ nP (ideals in OP /: This contradicts the assumption that the fi generate nP .V / D d ” dimk . fd / is the ideal of all regular functions on U zero at P .e. fd / and because Z contains P and is not equal to it. : : : .f1 . Suppose that P is nonsingular.gr. fd / from U . To see what it is.V /.f1 . : : : .V /.f1 . the dimension of the geometric tangent cone at P is the same as the dimension of V (because the Krull dimension of noetherian local ring is equal to that of its graded ring). Then f1 . a point P on V is nonsingular if and only if there exists an open affine neighbourhood U of P and functions f1 ..OP / is a polynomial ring in d variables. gr. fd /  p ¤ mP (ideals in kŒU /: On passing to the local ring OP D kŒU mP .V / of the geometric tangent cones.35.36. : : : . : : : . fd generate the maximal ideal nP in OP . : : : . the singular locus D EFINITION 4.  The map on the rings kŒX1 . P ROOF. it is not necessarily induced by the same map on polynomial rings as the original map.p/  V . : : : .94 4. Hence P is an irreducible component of V .f1 .OP / by its nilradical. and dim TP . in which case CP . Then there exist d regular functions f1 . fd /.OQ / ! gr.CP . i.V / of V at P to be Spm.nP =nP / D d. A variety is nonsingular if all of its points are nonsingular. and we define the tangent cone to be . and hence a map CP . : : : . L OCAL S TUDY For an affine algebraic variety V and P 2 V . it is necessary to work with the rings gr.

Xd C1 /. The set of nonsingular points of an affine algebraic variety is dense and open. Xi C1 . say. the singular locus 95 T HEOREM 4. then dim V D min dim TP . @F @F . : : : .F /. Let V be an irreducible component of the variety. Let P1 . The constant dimension is the dimension of V . : : : .n d /  .a/ D B : : @ A @Pr @Pr . We now show that the singular locus is not equal to V . if @F=@Xi is identically zero on V . with equality if and only if P is nonsingular.37. Pr generate a. and so F is a polynomial in X1 . Then the singular locus is the zero set of the ideal generated by the . then it is the zero polynomial. We may suppose that V D V . and so all points are nonsingular. Xd C1 (characteristic p). By definition dim TP .38. The group must act by isomorphisms.V /. An irreducible algebraic variety is nonsingular if and only if the tangent spaces TP .V /  dim V . According to (3. P ROOF.64).V /. if the singular locus equals V . : : : .a/ @X1 @Xm which is closed.P1 .:::. 2 In particular.X1 . every group variety is nonsingular. : : : . The singular locus is the set of common zeros of the polynomials F. 2 C OROLLARY 4. @X1 @Xd C1 and so it will be proper unless the polynomials @F=@Xi are identically zero on V . and so the tangent spaces have constant dimension. Pr /.2 C OROLLARY 4. equal to the zero set of the nonconstant irreducible polynomial F . 2 This formula can be useful in computing the dimension of a variety. If V is irreducible. As in the proof of (4. dim V D d . : : : . As there exists a nonsingular P .a/ : : : @X . P 2 V . Nonsingular points. then F is constant (characteristic 0) or a pth power (characteristic p). : : : Xd C1 (characteristic zero) or in p X1 .7). .36) and (3. P ROOF. : : : . C OROLLARY 4. By (2. dim V is the minimum value of dim TP . .n d / minors of the matrix 0 1 @P1 @P1 .37) some nonempty open subset of V is isomorphic to a nonempty open subset of an irreducible hypersurface in Ad C1 . We first show that it is closed.39. P 2V P ROOF. which contradicts the hypothesis. Xi .a/ @X1 m B C :: :: C. have constant dimension.V /.h. Therefore.40.a/ : : : . We have to show that the singular locus of V is a proper closed subset. Jac.a/  An . P ROOF. Every variety on which a group acts transitively by regular maps is nonsingular. Xi 1 . and so we may suppose that V itself is an irreducible hypersurface in Ad C1 .

0/ has equation X Y D 0. the singular locus is the line X D 0 D Z (and the singularity at .V / has dimension 3. the corollary holds also for reducible varieties.X Y. Every generating set for the maximal ideal nP of OP has at least d elements.W /. then the proof of (4. 0/ is very bad. a regular local ring is an integral domain. For the surface V W Z 3 D X 2 Y . 2 C OROLLARY 4. P ROOF. and so To .45. Therefore (3. XZ/. 0/. The intersection of the surface with the surface Y D c is the cuspidal curve X 2 D Z 3 =c: 0. If P lies on two irreducible components of a V .V / isomorphically onto To . A point P on an irreducible variety is nonsingular if and only if OP is regular.96 4. The tangent cone at .0. L OCAL S TUDY Examples 4.45) shows that 0  d r.f1 .X Y / in A2 .43. and so r  d . For the surface Z 3 D X Y . fr / in some open neighbourhood U of P . and so OP is not regular.44.14). : : : . 2 According to CA 22. whereas To W has dimension 2. An isomorphism V ! W would have to send the singular point o to the singular point o and map To . and let W be the zero set of XY. fr generate nP . 4. : : : . Let V be the union of the coordinate axes in A3 . Y Z.41.0. Are they isomorphic as algebraic varieties? Obviously. V and W are not isomorphic.42. i Nonsingularity and regularity T HEOREM 4. Each of V and W is a union of three lines meeting at the origin. the only singular point is .3. the origin o is the only singular point on V or W .0. . If f1 . The rest of the statement has already been noted. But V D V . and so it is the union of two planes intersecting in the z-axis. 0.5 1 2 y 4. Therefore. then OP is not an integral domain (3. 0.1 0. Let P be a point on an irreducible variety V . Therefore. and there exists a generating set with d elements if and only if P is nonsingular. This is a restatement of the second part of the theorem. P ROOF.36) shows that P is an irreducible component of V .

Examples of tangent spaces j 97 Examples of tangent spaces The description of the tangent space in terms of dual numbers is particularly convenient when our variety is given to us in terms of its points functor.kŒ"/. whose kernel is defined to be the special orthogonal group SOn . SLn is semisimple.kŒ"/.G/ ! Lie. Te .G 0 /. and we call two semisimple groups G and G 0 locally isomorphic if G=Z. For I C "A 2 Mn . which we denote Lie. and SLn =n D PSLn . The determinant defines a surjective regular homomorphism detW On ! f˙1g. This is the group of matrices preserving the quadratic form X12 C  CXn2 .k/ j A is skew-symmetricg: A SIDE 4.GLn / into a Lie algebra.kŒ"/ j A is skew-symmetricg ' fA 2 Mn . /. Since det.I C "A/tr  . there is a bracket operation def ŒM. and G 7! Lie. . Te . for example.GLn / D fI C "A j A 2 Mn g ' Mn . Write e for I when it is to be regarded as the identity element of GLn . 4.A/ D 0g: 4.G/ is independent of the embedding of G into GLn (in fact. For any closed algebraic subgroup G of GLn . : : : . and let On be orthogonal group: On D fA 2 GLn j Atr  A D I g: (Atr denotes the transpose of A).SLn / D fI C "A j trace.G/. On the tangent space Te .G/ is a functor from the category of linear group varieties to that of Lie algebras. and so lies in GLn . For example.. A matrix I C "A has inverse I "A in Mn .49. it has an intrinsic definition in terms of left invariant derivations). One can prove that G is semisimple ” Lie. N  D MN NM which makes Te .On / D Te . Such a group G may have a finite nontrivial centre Z. For example.G/.47.46.SOn / D fI C "A 2 Mn . and let I be the identity matrix.48. Assume that k has characteristic zero. every e´ tale homomorphism G ! G 0 defines an isomorphism Lie. let Mn be the set of n  n matrices.A/ (using that "2 D 0).G/ is stable under the bracket operation on Te .k/: 4. Therefore.GLn / ' Mn of GLn .I C "A/ D I C "Atr C "A. the set of diagonal matrices diag.G/  G 0 =Z. and so Te . This functor is not fully faithful.G/ is semisimple. Assume that the characteristic ¤ 2.k/ j trace.kŒ"/.j. A Lie algebra is semisimple if it has no commutative ideal (except f0g). but it is nevertheless very useful. The Lie algebra structure on Lie.  n D 1. .GLn / and is a sub-Lie-algebra of Mn .A/ D 0g ' fA 2 Mn .I C "A/ D I C "trace. Te . A connected algebraic group G is said to be semisimple if it has no closed connected solvable normal subgroup (except feg). with centre n .G 0 /.

See p. 4-4.98 4. 1966. and let RW D k ˚W endowed with the k-algebra structure for which W 2 D 0. 4-9. and the tangent cones at the singular points.174.P.G/ defines a one-to-one correspondence between the set of local isomorphism classes of semisimple algebraic groups and the set of isomorphism classes of Lie algebras. and let J be the matrix I 0 group Spn is the group of 2n  2n matrices A with determinant 1 such that Atr  J  A D J .RW / D Homk-algebra . .P / . Show that def the elements of V .V / ˚ TQ . Lectures on curves . 4-7.V /. Ta0  Ta . for each of (a) Y 3 Y 2 C X 3 X 2 C 3Y 2 X C 3X 2 Y C 2X Y I (b) X 4 C Y 4 X 2 Y 2 (assume that the characteristic is not 2). Exercises 4-1. and also the dimension of Spn . Show that P is a nonsingular point on each irreducible component of V \ H on which it lies.P. but need they always be different? 4-10. Suppose that a ¤ I. let Ta0 be the subspace of Ta . Show that T. f 2 a.df /a D 0. The classification of semisimple algebraic groups can be deduced from that of semisimple Lie algebras and a study of the finite coverings of semisimple algebraic groups — this is quite similar to the relation between Lie groups and Lie algebras. The symplectic 4-6. show that there is a curve C and a point P on C such that the tangent space to C at P has dimension n (hence C can’t be embedded in An 1 ). t / with P 2 V and t 2 W ˝ TP .V / (cf. Let V D V . Let W be a finite-dimensional k-vector space. . even though the origin is singular (characteristic ¤ 2). Let H be a hyperplane in An (i.V  W / D TP .V /.Q/ . L OCAL S TUDY and the map G 7! Lie. Mumford. and for a 2 V . Let V  An be an irreducible affine variety. (Each irreducible component has codimension 1 in V — you may assume this. Must P be singular on V \ H if H  TP . and let P be a nonsingular Ppoint on V .V / ! C˛.kŒV .a/  An .V /. RW / are in natural one-to-one correspondence with the pairs . Show that the cone X 2 C Y 2 D Z 2 is a normal variety. Given a smooth point on a variety and a tangent vector at the point.e. (It is the group of matrices fixing a nondegenerate skew-symmetric form. 4-8. . Let P and Q be points on varieties V and W .. Find the singular points. Let V be an affine algebraic variety over k. Find a regular map ˛W V ! W which induces an isomorphism on the geometric tangent cones CP .W /: 4-5. .) Find the tangent space to Spn at its identity element.V /? 4-3. p25).) Give an example with H  TP . For each n.An / defined by the equations .   0 I . 4-2.V / and P singular on V \ H . Let I be the n  n identity matrix. show that there is a smooth curve passing through the point with the given vector as its tangent vector (see mo111467).W / but is not e´ tale at P . the subvariety defined by a linear equation ai Xi D d with not all ai zero) passing through P but not containing TP . Clearly.

for example.U. (a) A topological manifold of dimension n is a ringed space . OV / is an algebraic prevariety over k if there exists a finite open covering V D Vi such that . An open subset U of an algebraic prevariety V such that .V. which is frequently not the case. Because V is a finite union of open 1 Provided the latter are stated correctly. 3.V. D EFINITION 5.U.V.U .U. OV jU / is isomorphic to the ringed space of smooth functions on an open subset of Rn (cf.2)). These definitions are easily seen to be equivalent to the more classical definitions in terms of charts and atlases. OV jU / is isomorphic to the ringed space holomorphic functions on an open subset of Cn (cf.C HAPTER Algebraic Varieties An algebraic variety is a ringed space that is locally isomorphic to an affine algebraic variety.2. OV jVi / is an affine algebraic variety over k for all i . 3.1. OV jU / is isomorphic to the ringed space of regular functions on an algebraic set over k. both are required to satisfy a separation axiom. a ringed S space .U. Thus.3).1 Often one imposes additional conditions on V . (c) A complex manifold of dimension n is a ringed space such that V is Hausdorff and every point of V has an open neighbourhood U for which . An algebraic variety will be defined to be an algebraic prevariety satisfying a certain separation condition.Vi . just as a topological manifold is a ringed space that is locally isomorphic to an open subset of Rn . OV / such that V is Hausdorff and every point of V has an open neighbourhood U for which . that it be connected or that it have a countable base of open subsets.4). (b) A differentiable manifold of dimension n is a ringed space such that V is Hausdorff and every point of V has an open neighbourhood U for which . recall the following definitions. 3. OV / such that V is quasicompact and every point of V has an open neighbourhood U for which . a Algebraic prevarieties As motivation. An algebraic prevariety over k is a k-ringed space . OV jU / is isomorphic to the ringed space of continuous functions on an open subset of Rn (cf. 99 5 . D EFINITION 5. OV jU / is an affine algebraic variety is called an open affine (subvariety) in V .

This motivates the following definition.a0 W : : : W an / 7!  a a0 an bi ai .a0 .a0 . : : : .V. OV / be an algebraic prevariety. cbn / some c 2 k  : Thus the equivalence classes are the lines through the origin in k nC1 (with the origin omitted).100 5. : : : .3. : : : . j . bn / ” . In Chapter 6 we shall show that there is a unique structure of a (separated) algebraic variety on Pn for which each Ui is an open affine subvariety of Pn and each map ui is an isomorphism of algebraic varieties.U. Note that we have three categories: (affine varieties)  (algebraic prevarieties)  (ringed spaces). : : : . the morphisms Mor. b Regular maps In each of the examples (5.U /. a continuous map 'W V ! W is regular if f 7! f ı ' sends a regular function on an open subset U of W to a regular function on ' 1 . then f W U ! k is regular if and only if f jUi \ U is regular for all i (by 3. Thus understanding the regular functions on open subsets of V amounts to understanding the regular functions on the open affine subvarieties and how these subvarieties fit together to form V .b0 . OW / be algebraic prevarieties. smooth map.Wj / by open affines..1a. Note that if . an /  .V. The functions f W U ! k lying in . OW / be S prevarieties.Wj / D Vj i be a covering of ' 1 . . let Ui D f. A LGEBRAIC VARIETIES affines. P ROPOSITION 5.a0 W : : : W ai W : : : W an / 2 Pn j ai ¤ 0g: Then Pn D S Ui .Ui / is an open covering of V by affine varieties. A composite of regular maps is again regular (this is a general fact about morphisms of ringed spaces). E XAMPLE 5. A map 'W V ! W is said to be regular if it is a morphism of k-ringed spaces. and the map . Let .e. : : : . Let . OV / and . and let 'W V ! W be a continuous map (of topological spaces). holomorphic map respectively) is just a morphism of ringed spaces. (Projective space). and let U be an open subset of V .4. S and let ' 1 . Then ' is regular if and only if its restrictions 'jVj i W Vj i ! Wj are regular for all i. an /. For each i . OV / are called regular. : : : .W. : : : . OV / and . i. a morphism of manifolds (continuous map.c). W / are the same in the three categories.W.a0 .b. Let W D Wj be a covering of W by open affines.a0 W : : : W an / for the equivalence class containing . ai . Let . an / D .V. ai  ui W Ui ! A n (the term ai =ai is omitted) is a bijection. Each subcategory is full. it follows that the open affines form a base for the topology on V . Write . In other words. and in each open affine the open affines (in fact the basic open subsets) form a base for the topology.V.1(c)). Let Pn denote k nC1 X foriging modulo the equivalence relation .cb0 .

Jean-Pierre. The subset of Z on which '1 and '2 agree is closed. the set fz 2 Z j '1 . of Math. : : : .z/ D '2 . Thus a continuous map into the space should be determined by its values on a dense subset. This is not true for algebraic varieties. 2 These are sometimes called “algebraic varieties in the sense of FAC” (Serre. 2 A SIDE 5. Let '1 . and prove that it is regular. for two reasons: (a) We are not assuming our varieties are nonsingular (see Chapter 5 below). In Grothendieck’s language.10. because even affine algebraic varieties are not Hausdorff. '2 W Z  V regular maps of affine algebraic varieties. Intuitively. 10. except that we omit the nonclosed points.55) below. or quasi-projective — usually because their foundations do not allow for a more flexible definition.c.1(b)). and the set where '1 and '2 agree is 2 i D1 V . Let '1 and '2 be regular maps Z  V from an algebraic prevariety Z to a separated prevariety V . Thus ' is regular. c Algebraic varieties In the study of topological manifolds.P /. .7. the Hausdorff condition eliminates such bizarre possibilities as the line with the origin doubled where a sequence tending to the origin has two limits (see 5. '2 W Z  V with Z an affine algebraic variety.P // identifies V with a closed subset of An (take the xi to be any set of generators for kŒV  as a k-algebra). Faisceaux alg´ebriques coh´erents. and so f ı 'j' 1 . Equivalently.6. see (5. the significance of this condition is that it prevents a sequence in the space having more than one limit. Ann.8. i (this is our assumption). Some authors use a more restrictive definition — they may require a variety to be connected. It is not immediately obvious how to impose a separation axiom on our algebraic varieties. The subset of Z on which '1 and '2 agree is closed.e. D EFINITION 5. However. irreducible. (1955). A differentiable manifold of dimension n is locally isomorphic to an open subset of Rn . There are regular functions xi on V such that P 7! . EGA IV. (b) The inverse function theorem fails in our context: a regular map that induces an isomorphism on the tangent space at a point P need not induce an isomorphism in a neighbourhood of P . Let f be a regular function on an open subset U of W . An algebraic prevariety V is said to be separated if it satisfies the following additional condition: Separation axiom: for every pair of regular maps '1 . Now Tn xi ı '1 and xi ı '2 are regular functions on Z. 197–278.10 below). In particular. The key is to restate the Hausdorff condition. xn . The converse is even easier. L EMMA 5.U / (sheaf condition 3. if '1 and '2 are continuous maps Z ! U that agree on a dense subset of Z then they should agree on the whole of Z. provided '1 and '2 are required to be regular maps. 34).U / \ Vj i is regular for each j. It follows that f ı ' is regular on ' 1 . (2) 61. '2 W Z  U agree should be closed. Then f jU \ Wj is regular for each Wj (sheaf condition 3. all manifolds of the same dimension are locally isomorphic. cf. We assume that ' satisfies this condition. i.. affine varieties have this property.xi ı '1 xi ı '2 /.2 P ROPOSITION 5. Algebraic varieties 101 P ROOF. P ROOF.z/g is closed in Z. An algebraic variety over k is a separated algebraic prevariety over k. the set where two continuous maps '1 .5.x1 . Surprisingly. they are separated and reduced schemes of finite type over k (assumed to be algebraically closed). which is closed.1(c)).

For example. 5. and give it the obvious topology. A LGEBRAIC VARIETIES P ROOF. 1=n converges to both a and b. Let V  D V1 t V2 (disjoint union).! V  . OP D fg= h 2 k. Then X is not Hausdorff because a and b cannot be separated by disjoint open sets.11. 0/g is quasi-affine but not affine. W \ U is the subset of U on which '1 jU and '2 jU agree. This implies that W is closed because Z is a finite union of open affines. Define a function on an open subset to be regular if its restriction to each Vi is regular.102 5.10. and so W \ U is closed. Let W be the set on which '1 and '2 agree. and defines an equivalence of the first category with the subcategory of the second whose objects are the affine algebraic varieties. Let R denote the real line with the origin removed but with two points a ¤ b added. where u < 0 < v. Let Vark denote the category of algebraic varieties over k and regular maps. for example.P / ¤ 0g \ OV . 0/ [ fbg [ . A subvariety of an affine variety is said to be quasi-affine. A2 X f.33).0. When V is irreducible. agree exactly on A1 A1 D V2 . 3 This . which is not closed in A1 .U / D fOV .V /. all the rings attached to it can be identified with subrings of the field k.u.0. but not a variety: it fails the separation axiom because the two maps A1 D V1 . V D V1 [ V2 .! V  f0g. Every sequence that converges to a also converges to b. 2 E XAMPLE 5. The open subspace U D A2 X f.9.u.0.0.)3 Let V1 and V2 be copies of A1 . v/ and . The topology is generated by the open intervals in R together with the sets of the form . and V1 \ V2 D A1 f0g.V / j h. For example. E XAMPLE 5. v/. The functor A Spm . 0/ [ fag [ . (The affine line with the origin doubled. This makes V into a prevariety. 3. U 0 open affineg \ D fOP j P 2 U g: is the algebraic analogue of the standard example of a non Hausdorff topological space. For any open affine U of Z. Define an equivalence relation on V  by x (in V1 /  y (in V2 / ” x D y and x ¤ 0: Let V be the quotient space V D V  = with the quotient topology (a set is open if and only if its inverse image in V  is open):   Then V1 and V2 are open subspaces of V .A/ is a fully faithful contravariant functor Affk ! Vark . 0/g of A2 becomes an algebraic variety when endowed with the sheaf OA2 jU (cf.U 0 / j U 0  U .

V. Spm. A function f on an open subset U of Z is regular if. we get a map 'W V ! spm. This defines a ringed structure OZ on Z.V. OV //: Let V be an algebraic variety such that . The proposition shows that the regular map 'W V ! Spm. we get a k-algebra homomorphism f 7! f ı 'W A ! .OV / has the following universal property: every regular map from V to an affine algebraic variety U factors uniquely through ': V ' Spm. OV // defined by id . A regular map 'W W ! V .V.V.V. OZ / is a variety it suffices to check that. OV /.A// ' Homk-algebra . For any P 2 V .V. k-algebra. To show that . called an open subvariety of V . Then U is a union of open affines. OV / be a homomorphism from an affine k-algebra A to the k-algebra of regular functions on V . for every P 2 U . Closed subvarieties Let Z be an closed subset of V . OV / be an algebraic variety over k. the ringed space .W / of varieties. OV / need not be finitely generated as a Subvarieties Let .A/ which is easily seen to be regular.U. f 7! ˛. f 0 / of a regular function at P on V such that f 0 jU 0 \ Z D f jU 0 \ U . for every open affine U  V .12. Such a pair .U 0 . e .Z. Open subvarieties Let U be an open subset of V . For a nonaffine algebraic variety V .P / is a k-algebra homomorphism A ! k. from a regular map 'W V ! Spm. OV / be an algebraic variety. A regular map 'W W ! V is an open immersion if '. . OV jU / is a variety.W / is open in V and ' defines an isomorphism W ! '.d. For an algebraic variety V and an affine k-algebra A. there is a canonical one-to-one correspondence Mor. but this is only an exercise (Exercise 3-2 to be precise). . and it follows that .U \ Z. Maps from varieties to affine varieties d 103 Maps from varieties to affine varieties Let .Z. OZ jU \ Z/ is affine algebraic variety. OZ / is called a closed subvariety of V . OV // 9Š U: A SIDE 5. and let ˛W A ! .A. . Conversely. we have proved the following result.13.V. In this way. and so its kernel '.V. there exists a germ . P ROPOSITION 5. Since these maps are inverse.V. OV / is an affine k-algebra.V.A/.V.P / is a maximal ideal in A.f /.

One checks easily that the subsets U  V such that U \ Vi is open for all i are the open subsets for a topology on V satisfying (a). Z is a variety. This structure on W has the following characterization: the inclusion map W . and (b) for each i 2 I . Vi \ Vj is open in both Vi and Vj and OVi jVi \ Vj D OVj jVi \ Vj . Every immersion is the composite of an open immersion with a closed immersion (in both orders). Because they agree on a dense subset of SN they agree on the whole of SN . '2 W Z  V . Conversely. If V is separated. A LGEBRAIC VARIETIES is a closed immersion if '. P ROOF. and a map 'W V 0 ! W with V 0 a variety is regular if and only if it is regular when regarded as a map into V . '2 W Z  V agree is closed.W / is closed in V and ' defines an isomorphism W ! '. Assume that the following “patching” condition holds: for all i. If every Vi is an algebraic prevariety. Let SN be the closure of S in Z. SN has the structure of a closed prevariety of Z and the maps '1 jSN and '2 jSN are regular. According to the above discussion. j . and so must be the whole of the Z. A prevariety V is separated if and only if two regular maps from a prevariety to V agree on the whole prevariety whenever they agree on a dense subset of it.! V is a homeomorphism of Vi onto an open set. and to give a regular map from V to a prevariety W amounts to giving a family of regular maps 'i W Vi ! W such that 'i jVi \ Vj D 'j jVi \ Vj : P ROOF. A locally closed subset W of a variety V acquires a natural structure as a variety: write it as the intersection W D U \ Z of an open and a closed subset. and that this is the only topology to satisfy . Application P ROPOSITION 5. and W (being open in Z/ therefore acquires the structure of a variety. Suppose that the set V is a finite union V D i 2I Vi of subsets Vi and that each Vi is equipped with ringed space structure. We assume that V has the property in the statement of the proposition. A regular map 'W W ! V is an immersion if it induces an isomorphism of W onto a subvariety of V . and let S be the subset of Z on which they agree. Then there is a unique structure of a ringed space on V for which (a) each inclusion Vi .! V is regular. then so also is V . W is called a subvariety of V . With this structure. Equivalent conditions: W is the intersection of an open and a closed subset of V .14. then the set on which a pair of regular maps '1 .104 5. and so S D SN is closed. OV jVi D OVi . W is open in its closure. and show that S is closed. consider a pair of maps '1 . 2 f Prevarieties obtained by patching S P ROPOSITION 5. Subvarieties A subset W of a topological space V is said to be locally closed if every point P in W has an open neighbourhood U in V such that W \ U is closed in U .W / of varieties.15.

V /  Hom. .A0 . pW V  W ! V. define on the set V  W the structure of a prevariety such that (a) the projection maps p. and (b) a map 'W T ! V  W of sets (with T an algebraic prevariety) is regular if its components p ı '. the Vi may be separated without V being separated (see. from the definition of products we have (underlying set of V  W / ' Mor.Z. A triple . for the moment. that V  W exists. In other words. V  W / ' Mor. Moreover. V  W / ! Hom.Z. For any prevariety Z. the map A0 ! Z with image z is regular. Suppose. q ı ' are regular. and these are all the regular maps (cf. 3. and that it is the only such sheaf. The product.U / to be the set of functions f W U ! k such that f jU \ Vi 2 OVi . qW V  W ! W / is said to be the product of V and W if it has the following universal property: for every pair of morphisms Z ! V .V  W. so also is . Again.10). Z/ is the underlying set of Z. q ı '/W Hom. the triple is a product if the map ' 7! . Products of varieties 105 (a).A0 . and the product of two topological spaces (in the category of topological spaces) is the product of the underlying sets endowed with the product topology.Z. for any z 2 Z.p ı '.Vi . V /  Mor. W are regular.28). There can be at most one such structure on the set V  W . and ' is regular if and only 'jVi is regular for each i . there exists a unique morphism Z ! V  W making the diagram Z 9Š V p V W q W commute. Z ! W in C. more precisely. OVi / is a finite union of open affines. For example. for example.g.U \ Vi / for all i. For the final statement. to give a map 'W V ! W amounts to giving a family of maps 'i W Vi ! W such that 'i jVi \ Vj D 'j jVi \ Vj (obviously).29) below. our problem can be restated as follows: given two prevarieties V and W .V. We shall show that products exist in the category of algebraic varieties. g Products of varieties Let V and W be objects in a category C. W / ' (underlying set of V /  (underlying set of W /: Hence. the product of two sets (in the category of sets) is the usual cartesian product of the sets. if it exists.A0 . Thus. qW V  W  V. one checks easily that OV is a sheaf of k-algebras satisfying (b). we give a condition on an open affine covering of a prevariety sufficient to ensure that the prevariety is separated. 2 Clearly. In (5. 5. Define OV . if each . W / is a bijection. Mor. is uniquely determined up to a unique isomorphism by its universal property.A0 . OV /.

A ˝k B/ will be the product of Spm A and Spm B in the category of affine algebraic varieties once we have shown that A ˝k B is an affine k-algebra. V . : : : . R/ to Homk-alg . and suppose that ˛ is nilpotent. b/ Again this comes down to checking that the natural map from Homk-alg . but with the arrows reversed. Xm  and kŒXmC1 . Xm =a. : : : . and we have shown that A ˝k B is reduced. (b) If A and B are integral domains.25). the bi are linearly independent over k. As the bi are linearly independent over k. b/ in RmCn . XmCn =b. the ai lie in all maximal ideals m of A.16. b/ be the ideal in kŒX1 . Then there is an isomorphism f ˝ g 7! fgW kŒX1 . A LGEBRAIC VARIETIES Products of affine varieties E XAMPLE 5. b/. then so also is A ˝k B. But the three sets are respectively V .a. Because of the category antiequivalence (3.ai C ci an / ˝ bi : i D1 Thus we can suppose that in the original expression of ˛. R/ is a bijection. Thus. If one of the bi P is a linear combination of the remaining b . using the bilinearity of ˝.kŒX1 . say. P P ROOF. this shows that Spm. and so this is obvious. Then ˛ D niD1 ai ˝ bi . b/ D zero set of .kŒX1 .17. : : : . : : : .18). Xm  kŒXmC1 . Hence ˛ D 0. P ROPOSITION 5. then. Let ˛ 2 A ˝k B. bn D ni D11 ci bi .kŒXmC1 . R/  Homk-alg . From a 7! aW N A ! A=m D k we obtain homomorphisms ' a ˝ b 7! aN ˝ b 7! abW N A ˝k B ! k ˝k B ! B P The image aN i bi of ˛ under this homomorphism is a nilpotent element of B. Let m be a maximal ideal of A. Now assume A and B to be reduced. Let A and B be k-algebras with A finitely generated. : : : . : : : .b/ D zero set of b in Rn . bi 2 B. : : : .a. we find that ˛D n X1 i D1 ai ˝ bi C n X1 i D1 ci an ˝ bi D n X1 . ci 2 k.106 5. XmCn =. (a) If A and B are reduced. XmCn  ˝k ! : a b . Let a and b be ideals in kŒX1 .a/ D zero set of a in Rm . V . : : : . XmCn  generated by the elements of a and b. and hence is zero (because B is reduced). XmCn  respectively. and so are zero (see 2. then so also is A ˝k B. some ai 2 A. The tensor product of two k-algebras A and B has the universal property to be a product in the category of k-algebras. this means that the aN i are all zero. and let .a. : : : .a. .a. XmCn  kŒX1 .

aN i bi / D 0 or . For example. because then A=m may be a finite extension of k over which the bi become linearly dependent. The following examples show that the statement of (5. Let k 0 D kŒ˛.4 This shows that spm. : : : g and fb 0 . : : :g each linearly independent over P fb1 .X / D i .V  W.) . D EFINITION 5.f .kŒV  ˝k kŒW / 4 Here is where we use that k is algebraically closed. i 2 over R/: The fact that k is algebraically closed forces A=m D k. we can write ˛ D ai ˝bi and ˛ 0 D ai0 ˝bi0 with the sets fb1 . : : : . which is a priori only an algebraic field extension of k. (1.A/ D V . K D kŒ˛ D kŒX =. We know that fb .17) also fails in this case. j  p 1.a1 . : : : .X ˛i / with ˛i 2 ˝. The product of the affine varieties V and W is .a ˝ 1 1 ˝ a/ D . C ˝R C ' CŒX =. aN i bi / D 0. For each maximal P 0 0ideal m of A. aN i bi / D 0 in B. but . so we can apply the linear independence condition. b . The proof of (5.a1 . we know 0 0 .g. (From sx599391.X /. so that there exists an element ˛ in an algebraic closure of k such that ˛ … k but ˛ p 2 k.A/ is irreducible (see 2. Thus either all the ai 2 m or all the ai0 2 m.because a 2 k/ D 0: Thus k 0 ˝k k 0 is not reduced. b2P P k. am / [ V . an0 /: As spm. 2 E XAMPLE 5. the elements ˛ i ˝ ˛ j .˛ ˝ 1 1 ˝ ˛/ ¤ 0 in k 0 ˝k k 0 (in fact. f . the ˛i are distinct. : : : .27). and in the second ˛ 0 D 0. : : :g 0 0 and .X //. b 0 . and let ˛ p D a. Assume that ˝ is large enough to split f .18. am / or V . In the first case ˛ D 0. (a) Suppose that k is nonperfect of characteristic p.1). ˛ 0 2 A ˝k B be such that P P ˛˛ 0 D 0. The elements aN i and aN i0 live in A=m.A/ equals either V . Then . (b) Let K be a finite separable extension of k and let ˝ be a second field containing k. : : : . 0  i.X // Y ' ˝ŒX =. By the primitive element theorem (FT 5. for some ˛ 2 K and Q its minimum polynomial f . b2 . and so either .17) fails when k is not algebraically closed.X C i / ' C  C: The proposition allows us to make the following definition. form a basis for k 0 ˝k k 0 as a k-vector space). As before.a10 . it follows that spm. aN i bi /. OV W / D Spm.1) which is not an integral domain. Because K=k is separable.19. and so ˝ ˝k K ' ˝ŒX =. It is possible for elements of a k-algebra to be linearly independent over k but not over some extension (consider 1. say. : : : g are linearly 1 2 1 2 independent over k (emphasis on k).X ˛i /. even though k 0 is a field. an0 /.a10 .X i /  CŒX =.58(b)) (1.1 ˝ a 1 ˝ a/ .f . and let ˛. Products of varieties 107 Now assume that A and B are integral domains.˛ ˝ 1 1 ˝ ˛/p D .

2 The corollary shows that V  W is the product of V and W in the category of prevarieties (hence also in the categories of varieties). the topology on V  W is strictly finer than product topology .68). then so also is V  W . q.b/. : : : .a1 . (a) As noted at the start of the subsection.17b) and (2.20. and note that V can be regarded as the variety obtained by patching the .108 5. in particular. OVi Wj /. Then [ V W D Vi  Wj and the . whereas A2 has many more closed subsets (see 2.a1 . the set V  W is the product of the sets V and W and p and q are the projection maps. OW / in the category of affine algebraic varieties.17a). Let V and W be affine varieties. 2 C OROLLARY 5.27).b/ is the product of V . (b) If V and W are irreducible. we can define . (a) The variety . the first statement follows from (5. amCn / D .a1 .V  W. OV / and . (b) This follows from (5. OVi /.20) implies that ' is regular when restricted to any open affine of T . OV W / to be the variety obtained by patching the . : : : .V  W. For example. which implies that it is regular on T .15). P ROOF.a/ and V . Let V and W be affine varieties. and the second statement then follows by the argument on p.22. E XAMPLE 5.a. a map 'W T ! V  W is regular if p ı ' and q ı ' are regular.105. b/ ! V . for the product topology on A2 D A1  A1 . Products in general We now define the product of two algebraic prevarieties V and W . Similarly. (b) It follows from (5. : : : . W defined by the homomorphisms f 7! f ˝ 1W kŒV  ! kŒV  ˝k kŒW  g 7! 1 ˝ gW kŒW  ! kŒV  ˝k kŒW : P ROPOSITION 5. then (5.21.Vi  Wj .Vi .a/ p q V . : : : . amCn /. .V.  When V and W have dimension > 0.W. A LGEBRAIC VARIETIES with the projection maps p. For every prevariety T . Therefore. write W as a union of open affines W D Wj . (a) It follows from (1. If p ı ' and q ı ' are regular. P ROOF. OV W / is the product of . OVi Wj / satisfy the patching condition. this coveringSsatisfies the patching condition (5. every proper closed subset is contained in a finite union of vertical and horizontal lines.57) that AmCn endowed with the projection maps  p. qW V  W ! V. is the product of Am and An . S Write V as a union of open affines V D Vi . am / m p mCn q n A A !A . amCn / D .Vi  Wj .16) that V . in particular.amC1 .

! V  W: Similarly. The set where '1 . Since v0 and w0 were arbitrary. As these open sets cover T . For v0 2 V . P ROOF.Xij / SLn .kŒX11 .k/ j det M D 1g 1// becomes a group variety when endowed with its usual group structures. Let '1 . Products of varieties 109 P ROPOSITION 5. 1 . P ROOF. which shows that . Matrix multiplication P . If V and W are separated. 2 P ROPOSITION 5. P ROOF.v0 .k/ D fM 2 Mn . : : : . and Cramer’s rule gives an explicit expression of the entries of A as polynomials in the entries of A. Every finite group N can be made into a group variety by setting N D Spm. q ı '2 agree.25. X12 . '2 agree is the intersection of the sets where p ı '1 . w/ and and . then so also is V  W . we have continuous maps closed W ' v0  W . In particular. which is closed. eW A0 ! G.23. V  W becomes the product of V and W in the category of prevarieties.g. 2 P ROPOSITION 5. S S the structure of prevariety on V  W defined by the coverings V D Vi and W D Wj are independent of the coverings. and let 'W T ! V  W be a map of sets such that p ı ' and q ı ' are regular. X 1  and Ga D Spm kŒX .! V  W: The images of V and W in V  W intersect in . invW G ! G.aij /  .v.bij / D . is given by polynomials. cij D nlD1 ai l blj . then so also is V  W .v0 . w0 / and are connected.  SLn D Spm. Then (5.cij /. If V and W are connected. The only affine group varieties of dimension 1 are Gm D Spm kŒX. Xnn =. Let T be a prevariety. With the sheaf of k-algebras OV W just defined. this shows that ' is regular. '2 be two regular maps U ! V  W . 2 Group varieties A group variety is an algebraic variety G together with a group structure defined by regular maps mW G  G ! G.24. w0 / lie in the same connected component of V  W for all v 2 V and w 2 W . p ı '2 and q ı '1 . this shows that any two points lie in the same connected component. for w0 2 W .21) implies that the restriction of ' to ' 1 . For example.Vi  Wj / is regular.det. we have continuous maps closed V ' V  w0 .A/ with A the k-algebra of all maps f W N ! k.

V / is closed. P ROOF. and so it is closed if V is separated. Conversely. the reverse argument shows that is closed. which is closed in U  U because U is separated (5. For any morphism 'W V ! W of prevarieties. We shall use the criterion (5. '. 2 C OROLLARY 5.v//W V ! 5 Recall '. this proves the first statement.x.! V  W ! V is an inverse to v 7! .8): V is separated if and only if. the subset of Z on which '1 and '2 agree is closed. In particular.27.6).x. '. The graph ' of a regular map 'W V ! W is defined to be Γϕ ϕ(v) (v. if V is Hausdorff. consider a topological space V and the diagonal   V  V . and V \ .z// is regular because its components '1 and '2 are regular (see p.'. '2 /W Z ! V  V.105). .v. the diagonal is a locally closed subset of V  V . For a variety V . Let P 2 V . the graph ' of ' is locally closed in V  W .26.U  U / D U .'1 . The second statement follows from the fact that the regular map p ' . 2 W Thus V is always a subvariety of V  V .5 P ROOF. P ROOF. z 7! .v// 2 V  W j v 2 V g: v V C OROLLARY 5. v/ j v 2 V g of V  V . Thus the statement does not contradict the fact that V is not Hausdorff. V is the set on which the two projection maps V  V ! V agree. Thus V is Hausdorff.v// is an isomorphism of V onto ' (as algebraic prevarieties). P ROPOSITION 5. ϕ(v)) f. The map . for every pair of regular maps '1 . then there are open neighbourhoods U and U 0 of x and y respectively such that U \ U 0 D ¿. '2 . In other words. and so . then every pair of points . A LGEBRAIC VARIETIES The separation axiom revisited def By way of motivation. but this is exactly the subset on which '1 and '2 agree. In particular.v.v. and let U be an open affine neighbourhood of P .'1 .28.110 h 5. we let  D V (the diagonal) be the subset f. For any prevariety V . If  is closed (for the product topology).v/. and as ' is the inverse image of W under this map. '2 W Z  V .v. The map . y/ …  has an open neighbourhood U  U 0 such that U  U 0 \  D ¿. w/ 7! .  D f. Then U  U is an open neighbourhood of . '. The map v 7! . it is continuous. Conversely. '2 / 1 . x/ j x 2 V g. Assume that V is closed. An algebraic prevariety V is separated if and only if V is closed. 2 that the topology on V  V is not the product topology.v. and it is closed if and only if V is separated. and it is closed if W is separated. it is continuous. if x and y are distinct points in V . P / in V  V .'1 .z/. w/W V  W ! W  W is regular because its composites with the projections are ' and idW which are regular.P.

Thus the sets U0 and U1 satisfy the condition in (b). Assume that (b) holds for the sets in an open affine covering .U  U 0 / \ V is closed in U  U 0 ” v 7! . (b) for every pair of open affines U and U 0 in V . X 1  1 . (c) Let V be A2 with the origin doubled.P / with f and g regular functions on U and U 0 . P ROOF. then these statements imply (b). Let U and U 0 be open affines in V .U \ U 0 /  V: If V is closed. and let U and U 0 be the upper and lower copies of A1 in V . We shall prove that (i) if  is closed then U \ U 0 affine. the statements (i) and (ii) imply the theorem. X 1  X 7! XW kŒX  ! kŒX. but the maps on rings corresponding to the inclusions U0 \ U1 . Then U \ U 0 is affine.P /g. Then . and the map f ˝ g 7! f jU \U 0  gjU \U 0 W kŒU  ˝k kŒU 0  ! kŒU \ U 0  is surjective.X 1 /W kŒX  ! kŒX. 2 In more down-to-earth terms.X / 7! f . X f . and let U0 and U1 be the standard open subsets (see 5.U  U 0 / \ V is a closed subvariety of an affine variety. Now (5. Thus the sets U0 and U1 fail the condition in (b). which implies (a).Ui /i 2I of V . and hence is affine.! Ui are X 7! XW kŒX  ! kŒX.! Ui are f . and the maps on rings corresponding to the inclusions U0 \ U1 . then .U  U 0 / \  of . (a) Let V D P1 . X 1 .28) implies that U \ U 0 is affine.Ui  Uj /. Since kŒU  U 0  D kŒU  ˝k kŒU 0 .! V is the subset . condition (b) says that U \ U 0 is affine and every regular function on U \ U 0 is a sum of functions of the form P 7! f .30.X /W kŒX  ! kŒX.i. j /.3).i. (b) Let V be A1 with the origin doubled (see 5. and let U and U 0 be the upper and lower copies of A2 in V . . . (ii) when U \ U 0 is affine. v/W U \ U 0 ! U  U 0 is a closed immersion ” kŒU  U 0  ! kŒU \ U 0  is surjective (3. The following three conditions on a prevariety V are equivalent: (a) V is separated. Proof of (i): The graph of the inclusion U \ U 0 . and V \ .j /2I I is an open affine covering of V  V . Then U \ U 0 is not affine (see 3. The separation axiom revisited 111 T HEOREM 5. (c) the condition in (b) holds for the sets in some open affine covering of V .U  U 0 / \  is closed ” kŒU  ˝k kŒU 0  ! kŒU \ U 0  is surjective: Assume (a).29. Thus.Ui  Uj / is closed in Ui  Uj for each pair .34). U \ U 0 is an open affine.33).10). this completes the proof of (ii). E XAMPLE 5.v. Then U0 \ U1 D A1 X f0g. Then .h. Proof of (ii): Assume that U \ U 0 is affine.X / 7! f .

w/ 7! wW V S W ! W and 0 for the map .B/ D Spm. Notes 5.v. If f W V ! S is a regular map and U is a subvariety of S . ˇW T ! W such that ' ˛ D ˇ.˛. one might hope that ‹‹ Spm. t 7! .v.˛. 0 / has the following universal property: for any regular maps ˛W T ! V . Note that if S consists of a single point. because it is the set where ' ı p and ı q agree. and so it has a canonical structure of an algebraic variety (see p.v. Since a tensor product of rings A ˝R B has the opposite universal property to that of a fibred product. there is a unique such map of sets. namely. then V S U is the inverse image of U in V . w/ 2 V  W j '.T.V S W.103).t /.32. then V S W D V  W . W /: Indeed. which is called the fibre of V over P .T. ˇ/W T ! V S W such that the following diagram T ˇ .w/g is a closed in V  W. For any point P 2 S. Hom.T.˛. w/ 7! vW V S W ! V . The set def V S W D f.R/ Spm. The map ' 0 in the above diagrams is called the base change of ' with respect to . which is regular because it is as a map into V  W .112 i 5. Write ' 0 for the map .S / Hom. ˇ. The algebraic variety V S W is called the fibred product of V and W over S. ' 0 .31.P / ! P induced by '.A ˝R B/: . the base change of 'W V ! S with respect to P . ˇ / ˛ '0 V S W W 0 ' V S commutes.A/ Spm.t //. A LGEBRAIC VARIETIES Fibred products Let 'W V ! S and W W ! S be regular maps of algebraic varieties. V / Hom.v/ D . V S W / ' Hom. there is a unique regular map . We then have a commutative diagram: '0 V S W W 0 ' V S: The system . In other words.! S is the map ' 1 .T. E XAMPLE 5.

U / is the field of fractions of kŒU . Spm. in an irreducible topological space.33.R// For the third isomorphism. OT .0/ D Spm.A.A ˝R B=N// ' Hom. and so k.kŒX =.X 2 //.B/ ' Spm.T. the scheme-theoretic fibre is Spec. Fibred products may differ depending on whether we are working in the category of algebraic varieties or algebraic schemes. Mor. but in general it may have nonzero nilpotent elements.B// . j Dimension Recall that.A ˝R B. Spm.T // ' Hom.T / is reduced) Hom. Mor.Spm.5. Then U \ U 0 is also a nonempty open affine (5.V /.29). Therefore kŒU   kŒU \ U 0   k. For example.5. OT . we used the universal property of A ˝R B. 5.T // ' Hom. and consider the R-algebras  kŒX  ! k. OT .34.T //  Hom.j.12). Spec.U / is also the field of fractions of kŒU \ U 0  and of kŒU 0 . Thus. The fibre ' 1 .R/ Spm.T // ' Mor.R/ Spec. Fibred products exist also for prevarieties.A/ Spm.k/. attached to V there is a field k. To prove this. called the field of .U / ! OV .a/ consists of two points if a ¤ 0. Consider the map x 7! x 2 W A1 ! A1 (see 5. which is dense in U . Thus ' 1 . X 7! X p : Then A ˝R B ' k ˝kŒX p  kŒX  ' kŒX =. Dimension 113 This is true if A ˝R B is an affine k-algebra. and one point if a D 0. and so the restriction map OV .T. note that for any algebraic variety T . V S W is only locally closed in V  W .kŒX =. Spm.A ˝R B/ ' in the category of schemes. Spm.B/ ' Spec.A ˝R B=N.A//  . For example. X 7! a kŒX  ! kŒX .X p which contains the nilpotent element x The correct statement is a/.X //.B.OT . and let U and U 0 be nonempty open affines in V .T. However.12) (the ring OT .T // Mor. every nonempty open subset is dense and irreducible.49). 5.U 0 / is injective. Let V be an irreducible algebraic variety V . 1 a p if p D char. OT .A ˝R B=N/ (24) where N is the ideal of nilpotent elements in A ˝R B. let R D kŒX . In this case.R.U / where k. which reflects the fact that 0 is “doubled” in the fibre over 0.A/ Spec.T.

it is not true that if M 0 and N 0 are R-submodules of M and N .W / D e. Consequently S (see 2.35. : : : .V  W /.29). xm  kŒW  D kŒy1 .V  W / is d C e. : : : . Then dim. then dim. : : : . for any (nonempty) irreducible component Z of V \ W . Then6 def kŒV  W  D kŒV  ˝k kŒW   kŒx1 . 1  j  n. Let V and W be irreducible varieties. xd g and fy1 . xd  ˝k kŒy1 . : : : . codim. : : : .W /: P ROOF. xd . then M 0 ˝ N 0 is an RR submodule of M ˝R N . ye g are maximal algebraically independent sets of elements of kŒV  and kŒW . We may suppose V and W to be affine.U / for any open subset U of V . The dimension of V is defined to be the transcendence degree of k. Thus the statement follows from (3. if Z is a proper closed subvariety of V . . An algebraic variety is a finite union of noetherian topological spaces. ye : Thus the transcendence degree of k. It follows that some of the results in 2 carry over — for example. : : : .V / C dim.V / D max dim. When all the irreducible components of V have dimension n.V / over k.Vi /. and this is defined by the n equations Xi D Yi in V  W . V is said to be pure of dimension n (or to be of pure dimension n). dim. : : : .V / D dim. and all of them are algebraic over kŒx1 . y1 . 1 ˝ ye g will be algebraically independent in kŒV  ˝k kŒW . 1 ˝ yj .Z/ < dim.V / C codim. 1  i  m. : : : .U / for U an open affine in V . and so is noetherian. which is the field of fractions of kŒU  for every open affine U in V .Z/  dim. 1 ˝ y1 . However. we saw that V \ W is isomorphic to  \ . dim.V / C dim. : : : . ye  ' kŒx1 .Z/  codim.36. P ROPOSITION 5.31). ye g are transcendence bases of k. Let V and W be closed subvarieties of An . xd  ˝k kŒy1 .45). Then fx1 . Note the dim.114 5. 2 We extend the definition of dimension to an arbitrary variety V as follows. : : : .V / D d and dim.V / D dim. this is true if R is a field. xd g and fy1 . A LGEBRAIC VARIETIES rational functions on V .V /. and so dim.W /: P ROOF. because then M 0 and N 0 will be direct summands of M and N . : : : . : : : . : : : .V / and k. xd ˝ 1. yn  where the x and y have been chosen so that fx1 . and we define dim. ye : Therefore fx1 ˝ 1. and tensor products preserve direct summands. P ROPOSITION 5. 2 6 In general. In the course of the proof of (5. In particular. Obviously kŒV  W  is generated as a k-algebra by the elements xi ˝ 1.12).W / (see FT 9. : : : .V  W / D dim.W / nI that is. Write kŒV  D kŒx1 . V is a finite union V D Vi of its irreducible components.

k. l Rational maps. If V 0 and W 0 are open affine subsets of V and W such that '. and a birational map is a rational map admitting a rational inverse.U.36 becomes false if An is replaced by an arbitrary affine variety. Suppose V and W are irreducible. Note however. '1 / 2 ' — clearly. Dominant maps R EMARK 5. 'U 0 / are said to be equivalent if 'U and 'U 0 agree on U \ U 0 . and there is a regular map '1 W U1 ! W such that . (b) Proposition 5. Proposition 5.34) implies that the map f 7! f ı 'W kŒW 0  ! kŒV 0  is injective. which is the pair of lines Y D ˙iX if k D C. Two such pairs .0. in the equivalence class. . 'U / and . Z W X2 D 0 D X4 I Z D f. Because V is a hypersurface in A4 . 0.V /. 0.V 0 /  W 0 . An equivalence class of pairs is called a rational map 'W V Ü W . 'U / 2 '.'U /2' U and we can define '1 to be the regular map such that '1 jU D 'U for all . 0/g.0. /g and Z \ Z 0 D f. and consider pairs . a regular map 'W V ! W is said to be dominant if the image of ' is dense in W . 0.k. Hence.36 becomes false if one looks only at real points (and the pictures we draw can mislead). 'U / with U largest (and U is called “the open subvariety on which ' is defined”). . and each of Z and Z 0 has dimension 2. k Dominant maps As in the affine case.36) fails because the diagonal in V  V cannot be defined by 3 equations (it takes the same 4 that define the diagonal in A4 ) — the diagonal is not a set-theoretic complete intersection. 'U / 2 '. Thus codim Z \ Z 0 D 3 — 1 C 1 D codim Z C codim Z 0 : The proof of (5.U. . and this map is independent of the choice of V 0 and W 0 . A rational map ' is said to be defined at a point v of V if v 2 U for some . there is always a pair . 'U / where U is a dense open subset of V and 'U is a regular map U ! W . The set U1 of v at which ' is definedS is open.W / ! k.U. (a) The subvariety  115 X 2 C Y 2 D Z2 Z D0 of A3 is the curve X 2 C Y 2 D 0.U. it has dimension 3.U1 .U 0 . which has codimension 3.U. Therefore it extends to a map on the fields of fractions. the codimension is 2. then (3. birational equivalence Loosely speaking. 0. that real locus is f. Let V and W be varieties over k. 0/g. 0/g Z 0 W X1 D 0 D X3 I Z 0 D f.0. a rational map from a variety V to a variety W is a regular map from a dense open subset of V to W .. Thus. U1 D .U. Consider for example the affine cone V X1 X4 X2 X3 D 0: It contains the planes. in particular.37.

5.V / with a subfield of k. we may suppose that kŒU   kŒU 0 .V / D dim W .V 0 / be a k-algebra homomorphism.1 below. and so there exists a nonzero d 2 kŒU 0  such that dxi 2 kŒU 0  for all i . : : : . P ROOF.V 0 /.V / ! k.V 0 /. : : : .! k.. This proves the “if” part. Xd C1  be an irreducible polynomial satisfied by the xi . which is the field of fractions of kŒU 0 . U 0 ) be a open affine subset of V (resp. There are canonical isomorphisms 2 TP . In this case. the inclusion k. extends mutatis mutandis. Let V and V 0 be irreducible varieties over k. According to Proposition 9. xd .kŒ"/ ! V .V / at a point P on an algebraic variety V is the fibre of V . After inverting d .43.V / D k. A regular map 'W U 0 ! U from an open subset U 0 of V 0 onto an open subset U of V defines a k-algebra homomorphism k. U 0 ). 5. P ROOF. and let H be the hypersurface f D 0. Assume that k.V /  k.U / 0 ' ! U0 is the required map. According to (3. called the singular locus. xd C1 2 k. the image of ' contains an open subset U0 of U . being local.V 0 /.V /  k.V / . 2 A rational (or regular) map 'W V Ü W is birational if there exists a rational map ' 0 W W Ü V such that ' 0 ı ' D idV and ' ı ' 0 D idW as rational maps. k/ ' Homk-linear .V / such that k.38). The first part of the statement is obvious. Every irreducible algebraic variety of dimension d is birationally equivalent to a hypersurface in Ad C1 . The tangent space TP .V 0 /. Two irreducible varieties V and V 0 are birationally equivalent if and only if k. so let k. : : : . Let V be an irreducible variety of dimension d .! k.36).V / ' Derk . xm .V 0 / is induced by a dominant regular map 'W U 0 ! U . Let U (resp.V 0 / (as k-algebras).H /. A variety is nonsingular (or smooth) if every point is nonsingular. A point P on nonsingular (or smooth) if it lies on a single irreducible component W and dim TP . Each xi 2 k. Now ' 1 . xd C1 /. to general algebraic varieties. A LGEBRAIC VARIETIES P ROPOSITION 5. Then k. replacing U 0 with basic open subset.38.OP .40. 2 m Local study Everything in Chapter 4.116 5. and the “only if” part is obvious.e. P ROOF. A point P is nonsingular if and only if the local ring OP is regular.39. Let f 2 kŒX1 . The singular points form a proper closed subvariety.nP =nP .41. P ROPOSITION 5. in which case the existence of U  U 0 is proved in (3.x1 . xd . We may suppose that V and W are affine.k/ over P .42. . : : : . there exist x1 . Two varieties V and V 0 are birationally equivalent if there exists a birational map from one to the other.V / . k/ where nP is the maximal ideal of OP . Let kŒU  D kŒx1 . there exist dense open subsets U and U 0 of V and V 0 respectively such that U  U 0 .V /  k. We identify k. i. and every such homomorphism arises in this way. Thus. 2 P ROPOSITION 5. 5.

write A D Spm kŒX1 . : : : .a/ ¤ 0 j P 5. But P @f d. Pn /.b/ .f1 .P1 . I claim that the map W ! V is e´ tale at .a.a/ : : : .a1 . To see this. Xn =.a/ : : : @X .W / ! Ta .f . : : : .d'/P W TP . and let f D ci X i 2 AŒX  be such that AŒX=.a/ : : : . : : : . f /: The tangent spaces to W and V at .A/ be an affine variety. Equivalent condition: det @X .a. b/ if and only if b is a simple root of ci .f . because the map  on the tangent @Pi spaces has matrix Jac.a/ @f @f @f @X1 @Xm . ' is e´ tale if it is e´ tale at all points of V .44. : : : . i ci . @X dX P which is zero if and only if b is a multiple root of i ci . and consider the map W ! V corresponding to the inclusion A . Let V D Spm. This map is an isomorphism if and only if @X solution of the smaller set of equations extends uniquely to a solution of the larger set.f1 .! AŒX =. b/¤ 0.V / ! T'.a/ 0 @f1 @f1 @X1 m .V / is induced by the projection map k nC1 ! k n omitting the @f .a. A regular map 'W An ! An . an // is e´ tale at a if and only if rank Jac.a/ 0 @ @X1 A @fn @fn @Xm . b/ D . a D .f / AŒX  A W V  A1 V ThePpoints of W lying over a point a 2 V are the pairs .a/X i .b/. Xn =a. Examples 5.a.X // is reduced. so that AŒX =. Pn /.a/ : : : @X . because then every last coordinate.f /. b/ and a respectively are the null spaces of the matrices 0 @f1 1 @f1 0 1 .a/X i .46. : : : .a/ @X . Etale maps n 117 ´ Etale maps D EFINITION 5.´ n. which is ramified exactly at the points where two or more sheets cross. A regular map 'W V ! W of smooth varieties is e´ tale at a point P of V if the map . Let W D Spm. fr .P1 . : : : .a/ : : : @X .AŒX =.a/ D n. a 7! .a/ B C @X1 : : m :: :: B C B C :: :: B C B C : : B @fn C @ A @fn .W / is an isomorphism. : : : .a/X i / .P / . .a/X i .X //. fr /. The intuitive picture is that W ! V is a finite covering with deg.a. : : : .a. b/ 2 V  A1 such that b is a root of P ci .a/). Thus AŒX =.P1 . b/ @X1 m and the map T. : : : .a1 .45.a.f / D kŒX1 .f / sheets. an /. Pn .

j 5. and assume V to normal. The example in (b) is typical. As a point on the surface moves around O. : : : .50. it shifts from one sheet to the other.46) with f . Then ' is 2 W 1 except over 0. y/ 7! x. where T is the coordinate function on A1 . The failure of the inverse function theorem for the Zariski topology 5. because the Jacobian matrix is .X //. there exist open neighbourhoods U and U 0 of a and '. More precisely.49. Pn / (same number of polynomials asvariables). To get a section to '. let 'W W ! V be e´ tale at P 2 W . Yn =. I claim that there do not exist nonempty open subsets U and U 0 of A1 f0g such that ' defines an isomorphism U ! U 0 . If there did. Let V be the plane curve Y 2 D X and ' the map V ! A1 . the inverse function theorem says that a map ' that is e´ tale at a point a is a local isomorphism there. which has rank one for all X ¤ 0 (alternatively. and so we may view it schematically as V ϕ A1 | 0 However. it is of the form (5.48. A LGEBRAIC VARIETIES 5. and a commutative diagram 0 W U10  W ?  U ?1  ? ? ? ? ? ? 0 y' y y y' 0 V  U2  U2  V 0 with the U all open subvarieties and P 2 U1 . and X 2 c has distinct roots for c ¤ 0). In advanced calculus (or differential topology.A1 / ! k. : : : .X /. A similar argument to the above shows that ' is e´ tale  @Pi if and only if det @X .2X /. when viewed as a Riemann surface. Consider a dominant map 'W W ! V of smooth affine varieties. in fact every e´ tale map is locally of this form. it is necessary to remove a line in C from 0 to infinity. which is not closed for the Zariski topology.x. Thus the true picture is more complicated. corresponding to a map A ! B of rings. 5.C/ consists of two sheets joined at a single point O.e.a/ such that ' induces an isomorphism U ! U 0 . provided V is normal (in the sense defined below p.173). or complex analysis).118 5. But on the fields of fractions. at least not for the Zariski topology: a map can be e´ tale at a point without being a local isomorphism. which is not an isomorphism. ' defines the map k.a/ is never zero. Consider for example the map 'W A1 X f0g ! A1 X f0g. then ' would define an isomorphism kŒU 0  ! kŒU  and hence an isomorphism on the fields of fractions k.47. i..P1 . V . X 7! X 2 . Suppose B can be written B D AŒY1 . then there exist a map ' 0 W W 0 ! V 0 with kŒW 0  D kŒV 0 ŒX =.f .X / D X 2 T .X / ! k. . This is not true in algebraic geometry. Nevertheless.A1 /. a 7! a2 : This is e´ tale if the characteristic is ¤ 2. .

Carl. : : : .P / / ! gr. We can rewrite this as follows: let t1 . and the coefficients of P the Taylor expansion n0 an .OP /. Zeinzu/. Seite 162 168.23). For f 2 OO P . ´ n.X /= h. a regular map ˛W V ! W induces a homomorphism gr. P ROPOSITION 5.X // where f . We say that ˛ is e´ tale at P if this is an isomorphism. : : : .hathitrust.V / ! C˛.n/ . Now the statement follows from (5. OP consists of quotients f .46). : : : . : : : . then there is a canonical isomorphism OO P ! kŒŒt1 . Vorlesungen u¨ ber Riemann’s theorie der Abel’schen integrale. Etale maps 119 Generated for jmilne (University of Michigan) on 2014-06-05 00:08 GMT / http://hdl.f . the image of f 2 kŒŒt1 . : : : . Tergl.X / can be computed as in elementary calculus courses: an D f . 10.a/ ¤ 0.W /. Roughly speaking. For example.W / is a finite separable extension of k.handle.V /. to be e´ tale at P . Let 'W W ! V be a map of irreducible affine varieties. 213-214ml 218-2 21 LiAdhsti\MSvyer. It is a fairly elementary result that a local homomorphism of local rings ˛W A ! B induces an isomorphism on the graded rings if and only if it induces an isomorphism on the completions (Atiyah-MacDonald 1969. Obviously. and let P be the point a. Thus ˛W V ! W is e´ tale at P if and only if the map OO ˛.a/=nŠ. ´ Etale maps of singular varieties Using tangent cones. P ROOF.P / . Then t D X a is a local parameter at a. Hence (5.X / with h.V /. td . After passing to open subvarieties. If k.net/2027/ucm. td be a local system of parameters at a nonsingular point P . 2 .Die lliemaim'sche Windungsfläche erster Ordnung.53) shows that the choice of a local system of parameters f1 . we need the map on geometric tangent cones to be an isomorphism and to preserve the “multiplicities” of the components. Google-digitized / http://www. we can extend the notion of an e´ tale morphism to singular varieties.O˛. Xd .X / D g. td  can be regarded as the Taylor series of f . let V D A1 .530650220x Public Domain. but this map may be an isomorphism without ˛ being e´ tale at P . then ' is e´ tale on a nonempty open subvariety of W .51.X a/n of f .P / ! OO P is an isomorphism. Leipzig : Teubner. we may assume that W and V are nonsingular. fd at a nonsingular point P determines an isomorphism OO P ! kŒŒX1 . 1865.org/access_use#pd-google It is not possible to fit the graph of the complex curve Y 2 D X into 3-space. and that kŒW  D kŒV ŒX =. Note that then there is an isomorphism of the geometric tangent cones CP .X / is separable when considered as a polynomial in k. but the picture at right is an early depiction of it (from Neumann.

d˛/a is . W U ! V / with  an e´ tale map from a nonsingular variety U to V and Q a point of U such that .d fQi /a . fd are represented by pairs . An e´ tale neighbourhood of a point P of V is pair . The hypotheses imply that W and V have the same dimension d . P ROOF. o ´ Etale neighbourhoods Recall that a regular map ˛W W ! V is said to be e´ tale at a nonsingular point P of W if the map .An /. and so it fails on a closed subset of W . A LGEBRAIC VARIETIES A SIDE 5. fr generate I. : : :  . say W  Am and V  An .a/ . : : : .53. : : : . Let ff1 . Equivalent conditions: the images of f1 .fQ1 .V /. f2 . U /. implies that det is a nonzero constant (by the Nullj  @Xj @Pi stellensatz 2. has not been settled even for k D C and n D 2. : : : . Pn /. which.a/ @X j @Pi @Xj . fd g of germs of regular functions at P generating the maximal 2 ideal nP  OP . : : : . we can choose to be nonsingular.fQ1 . : : : . then it is e´ tale at all points in an open neighbourhood of P .X1 . Obviously.P1 . fd in nP =nP generate it as a k-vector space (see 1. If we write ' D .d˛/a W Ta .a/ . .52. 2 Let V be a nonsingular variety. : : : .Q/ D P . : : : . which doesn’t contain P . .df1 /P .X1 .fQ1 . Let P be a nonsingular point on a variety V of dimension d .V / of Ta . the fi are represented by regular functions fQi defined on a single open neighbourhood U 0 of P . Let f1 . This is a polynomial condition on a. : : : . then ' has a inverse.4). : :  : . Let W and V be nonsingular varieties.54.W /.37).A / is a linear map with matrix @X .W / ! T˛. or . This conjecture. fQd /W U ! Ad is e´ tale. det @X never zero. Then ˛ is not e´ tale at a if and only if the matrix ! @fi .a/ The condition.dfd /P is a basis for dual space to TP . The next lemma then shows that ˛ is e´ tale on an open neighbourhood U of P . because the dual map to . . See the Wikipedia JACOBIAN CONJECTURE. and let P 2 V . Xm /.Q. There is an old conjecture that every e´ tale map 'W An ! An is an isomorphism. P ROOF.a/ is never zero (for a 2 k n ). 2 L EMMA 5. and that their tangent spaces all have dimension d .P / . fd g be a local system of parameters at a nonsingular point P of V . The map ˛ D .d˛/P W TP . It has caused many mathematicians a good deal of grief. If ˛W W ! V is e´ tale at P .A / ! T˛. : : : . Pn . and that ˛ is given by polynomials P1 . because of (4.a/ has rank less than m. We may assume W and V to be affine.V / is an isomorphism.11 applied to the ideal generated by det @X ). A local system of parameters at P is a family ff1 . despite the existence of several published proofs and innumerable announced proofs. U / and the map .120 5.fQd . and ˛ is not e´ tale at a j if and only if the kernel of this map contains a nonzero vector in the subspace Ta . Xm /. P ROPOSITION 5. @Xj     @Pi @Pi . Then there is a nonsingular open neighbourhood U of P such that f1 . then this becomes the statement:   @Pi if det . It is probably harder than it is interesting.dXi /o 7! . : : : . : : : . which is known as the j Jacobian conjecture. Then @Pi m n . fQd /W U 0 ! Ad is e´ tale at P .

:::.57 (I NVERSE F UNCTION T HEOREM ).:::.56. Note the analogy with the definition of a differentiable manifold: every point P on nonsingular variety of dimension d has an open neighbourhood that is also a “neighbourhood” of the origin in Ad .TP .P / to be the e´ tale neighbourhood 'jU W U ! W of '.0. any two nonsingular varieties are locally isomorphic (this is not true for the Zariski topology). According to (5.x1 . P ROOF. To get the above diagram. 2 The rank theorem For vector spaces. The inverse function theorem (for the e´ tale topology) T HEOREM 5. T HEOREM 5.:::/ kn A similar result holds locally for differentiable manifolds. Let 'W V ! W be a regular map of nonsingular varieties of dimensions m and n respectively. Etale neighbourhoods 121 C OROLLARY 5. then there exists a commutative diagram open V ' W UP  '0 e´ tale U'. If a regular map of nonsingular varieties 'W V ! W is e´ tale at P 2 V . there is a commutative diagram 0 0 ˛ V W  km  . and ' 0 an isomorphism.xm /7!. 0/ 2 Ad . : : : . There is an open Zariski neighbourhood U of P and a map W U ! Ad realizing .P /. then there exist bases for V and W relative to which ˛ has matrix Ir 0 . there is the following weaker analogue.xr . and ' 0 to be the identity map. U'. and let P 2 V . In algebraic geometry. there exists an open neighbourhood U of P such that the restriction 'jU of ' to U is e´ tale. U / as an e´ tale neighbourhood of .0. P ROOF. Relative to this “topology”.55.58 (R ANK T HEOREM ). Uf .xm /7!.P / an e´ tale neighbourhood '.P / with UP an open neighbourhood U of P .:::. and let P 2 V . we can take UP D U .x1 . There is a “topology” on algebraic varieties for which the “open neighbourhoods” of a point are the e´ tale neighbourhoods. the rank theorem says the following: let ˛W V ! W be a linear map of k-vector  spaces of rank r.xn / An . This is a restatement of the Proposition.:::.x1 .x1 . The “topology” is called the e´ tale ´ topology — see my notes Lectures on Etale Cohomology.P /.P.´ o. If rank. In other words. 2 A SIDE 5.'// D n. then there exists a commutative diagram UP 'jUP e´ tale e´ tale Am W .54). Let V be a nonsingular variety of dimension d .

fm /W UP ! Am . dfm is a basis for TP . : : : . Then df1 . gn /W U'. There is a description of the tangent vectors at a point P on a smooth algebraic variety V as certain equivalence classes of regular maps from an e´ tale neighbourhood U of 0 in A1 to V . if 'jUP factors in this way.54). : : : . if ' is smooth at P . and the second is the projection of Am n U '.53). and there exist fnC1 . p Smooth maps D EFINITION 5. there exist open neighbourhoods UP of P and U'.V /.P / ! U'. fm is a local system of parameters at P . In the general case there is a map from an open neighbourhood of the point P in X onto affine space sending P to 0 and inducing an isomorphism from tangent space at P to that at 0 (5. and let f1 D g1 ı'.P / . : : : . : : : .62. 2 A SIDE 5. then it is smooth on an open neighbourhood of V . : : : . Certainly.P / ! An are e´ tale. A regular map 'W V ! W of nonsingular varieties is smooth at a point P of V if . Let V and W be nonsingular varieties. According to (5.d'/P W TP .V / ! T'.V /_ . For V D An . Conversely.g1 .P / ! U'.P / . Unfortunately.59. ' is smooth if it is smooth at all points of V . dfn are linearly independent forms on TP . T HEOREM 5.f1 . the maps from U  A1 to An need not lift to X. 2 C OROLLARY 5.P / onto U'. : : : .P / respectively such that 'jUP factors into UP e´ tale !Adim V dim W q  U'. P ROOF.P / respectively and the vertical maps are e´ tale. and so it is necessary to allow maps from smooth curves into X (pull-backs of the covering X ! An by the maps from U into An ). there is a similar description with a curve taken to be a regular map from an open neighbourhood U of 0 in A1 to V .122 5. If 'W V ! W is smooth at P . then we get a diagram as in the rank theorem. A map 'W V ! W is smooth at P 2 V if and only if there exist open neighbourhoods UP and U'. fm such df1 .P / such that the maps . Tangent vectors at a point P on a smooth manifold V can be defined to be certain equivalence classes of curves through P (Wikipedia TANGENT SPACE). A LGEBRAIC VARIETIES in which UP and U'. In fact. : : : .P /. 2 .P / of P and '.61.P / are open neighbourhoods of P and '. : : : .60. P ROOF. They give the vertical maps in the above diagram. it is smooth.P / : P ROOF.W / is surjective. gn at '.P / of '. it is smooth on the neighbourhood UP in the theorem. Choose a local system of parameters g1 . Then f1 .P / : The first is e´ tale. fn D gn ı '. From it we get maps UP ! Am An U'.

P ROOF.a/ 2 R g [ fa 2 Rn j f .˛/ V . T HEOREM 5.R/ D f.65. : : : . V D V1 [ V2 . (b) Conversely if ' is dominant and separable.W / is separably generated over k.64.ˇ / '.a/  k n .R/. then ' is dominant and separable.a1 .a1 . but Rn ¤ fa 2 Rn j f .63. Let V be the union of two subvarieties. Algebraic varieties as a functors 123 Separable maps D EFINITION 5.R/W V . but not necessarily otherwise. T =. An D Df [ V .a/ D 0g in general. one for each affine k-algebra R.V /.66.R/ D V1 . We define a point of V with coordinates in R to be a regular map Spm. In particular V . '. 2 q Algebraic varieties as a functors Let R be an affine k-algebra. if V D V . : : : .S / W . Note that . and let V be an algebraic variety. (a) If there exists a nonsingular point P of W such that 'P is nonsingular and . then the set of P 2 W satisfying (a) is open and dense. . which is what you should expect.1 Tf // and V . then V . A regular map 'W V ! W of algebraic varieties defines a family of maps of sets.V  W /.S / (*) W . If V1 and V2 are both open. for any polynomial f .q..k/ D V (as a set). R ˛ S V .R/ D V . Let 'W W ! V be a map of irreducible varieties.R/ [ V2 .R/. : : : .X1 .f / is the zero set of f . then V .R/ (property of a product). A dominant map 'W W ! V of irreducible algebraic varieties is separable if k.kŒX1 .R/ '.R/ V .d'/P is surjective. such that for every homomorphism ˛W R ! S of affine k-algebras. Every family of maps with this property arises from a unique morphism of algebraic varieties. R EMARK 5.f / where Df ' Spm. an / 2 Rn j f . V (as a set) can be identified with the set of points of V with coordinates in k.e. For example.R/ ! V .R/ V . Then apply the rank theorem. : : : . T HEOREM 5. an / D 0 all f 2 ag.R/  W .R/ ! W . Xn /. Xn .S / commutes. Replace W and V with their open subsets of nonsingular points. For example. i.

Each intersection Ui \ Uj is affine (5.Affk . Let . we see that they arise from a regular map V  V ! V . Let ' be a morphism haff V ! hV 0 .x1 / D ˛2 . For example: SLn W R fM 2 Mn . F2 . As hV  hV ' hV V (by definition of V  V ).R/ j det. 2 It is not unusual for a variety to be most naturally defined in terms of its points functor. Sets/ aff is fully faithful.x2 /g: When F1 .R/. ˛2 W F2 ! F3 of sets is the set F1 F3 F2 D f. and so ' gives rise to a commutative diagram Y Y 0 hV . A LGEBRAIC VARIETIES For a variety V . C/: We now describe the essential image of h 7! hV W Vark ! Fun. Xn . : : : .M / D 1g GLn W R fM 2 Mn . G is isomorphic to the functor sending R to the set of zeros of S in Rn .R/ . T HEOREM 5.v. 2 C OROLLARY 5. there is a functor F D F1 F3 F2 such that . The multiplication maps G.Ui \ Uj / i 0 hV 0 .Ui / i Y hV 0 .Ui \ Uj / i.R. Similarly. X2 . x2 / j ˛1 . The Yoneda lemma (q.j in which the pairs of maps are defined by the inclusions Ui \ Uj .T / hV . P ROOF.R/ D F1 .x1 .R/ F3 . Sets/.R/ j det.124 5.66 follows. Certainly an affine group variety defines such a functor.T / Y i. ˛2 .68. As the rows are exact (5. Conversely. the inverse map and the identity-element map are regular.R/ give a morphism of functors hV hV ! hV . Sets/ if fully faithful.R/ F2 . To give an affine group variety is the same as giving a functor GW Affk ! Gp such that for some n and some finite set S of polynomials in kŒX1 . P ROOF.Vark .Ui / hV . F3 are functors and ˛1 .67.Affk .! Ui . We can restate as Theorem 5.15).29).F1 F3 F2 /.j hV 0 .R/ ! G. Wikipedia) shows that the functor V hV W Vark ! Fun. let haff V be the functor sending an affine k-algebra R to V .M / 2 R g Ga W R . The functor V haff V W Vark ! Fun. the conditions imply that G D hV for an affine algebraic variety V (unique up to a unique isomorphism). which (by the Yoneda lemma) arises from a unique regular map V ! V 0 .R/G. Uj .Ui /i 2I be a finite affine covering of T . and let T be a variety. ˛3 are morphisms of functors. The fibred product of two maps ˛1 W F1 ! F3 . this shows that 'V extends uniquely to a functor hV ! hV 0 .

It is again an affine variety (Exercise 5-2). m: Note that. V . : : : .V / (i.k/ is an equivalence relation on U.70.U / of hA is defined by an open subscheme of Spm.Ui /S i 2I of open subfunctors of X is an open covering of X if each Ui is open in X and X D Ui .e.kŒV .72) is R R ˝k E.kŒX1 . the subfunctor ' 1 . A family . To simplify the statement of the next proposition. and let U D Ui .K/ for every field K. The functor R is hU 0 where U 0 is the disjoint union of the varieties Ui \ Uj .X1 ..26). and we define V . By a functor we mean a functor from the category of affine k-algebras to sets. if V is affine. then V .kŒX1 . a radical ideal) and not just an ideal a such that V . : : :=rad. and so Homk . and the map U.k/. : : : . The functor defined by A. we write U for hU when U is an affine variety. it is important to choose the fi to generate I. it is closed (5. R/: More explicitly. . Choose a finite open affine covering S.f1 . As U 0 is the inverse image of V in U  U .R/ D Homk-algebra . A criterion for a functor to arise from an algebraic prevariety 5. for all maps 'W hA ! X . (b) the set R.˛i /i 2I 2 i 2I Vi . P ROPOSITION 5. j 2 I .A/.E/ (see p. This proves (a).R/ such that ˛i agrees with ˛j on Vi \ Vj for all i.a/.a/.a/ D VS. 2 R EMARK 5. 7 Let a be an ideal in kŒX .k/. A subfunctor U of a functor X is open if. If A has no nonzero nilpotent elements. P ROOF an algebraic variety. we write V as a finite Q union of open affines V D i Vi . fm /.69. : : : ! A that is zero on a is also zero on rad. A/: This is not true if A has nonzero nilpotents. Algebraic varieties as a functors 125 for all affine k-algebras R. Again.R/ the points of V with coordinates in R.29).R/ to be the set of families . A/ ' Homk . and agrees with the previous definition when V is affine.R/ is the set of solutions in Rn of the system equations fi .V / D . if V  k n and I. then every k-algebra homomorphism 1 kŒX1 .71. and so U 0 is affine.k/ ! F . For example.k/ as the quotient of U. i D 1. A functor F W Affk ! Sets is in the essential image of Vark if and only if there exists an affine scheme U and a morphism U ! F such that def (a) the functor R D U F U is a closed affine subvariety of U  U and the maps R  U defined by the projections are open immersions. The converse is omitted for the present. This is independent of the choice of the covering. : : :=a. we call the elements of V . Let F D hV for V F V D Ui of V . : : :.k/ realizes F . when we allow R to have nilpotent elements. Xn / D 0.7 For a general variety V .k/ by R. These are affine (5. A variety V defines a functor R V .q. and (b) is obvious.R/ from the category of all k-algebras to Sets. : : : .

see: Arnaud Beauville. X1 . and it is unirational if k. an algebraic prevariety over k is defined to be a functor satisfying this criterion.13. and more discussion of the L¨uroth problem in characteristic zero. for example. Fano attempted to find counter-examples to the L¨uroth problem in dimension 3 among the so-called Fano varieties. Castelnuovo and Severi proved that all unirational surfaces are rational. For a proof over any field. see. A1 / — it is a k-algebra. but initially he was not able to construct this as a projective variety. In characteristic zero. I. arXiv:1507.8 A functor U is affine if O. For a functor U .U. while obviously unirational. groupes commutatifs.Rfi fj / i i.X. of Demazure and Gabriel. A little history In his first proof of the Riemann hypothesis for curves over finite fields. In 1876 (over C).72. Let V be an algebraic variety over k. This led him to introduce “abstract” algebraic varieties.U / is an isomorphism.126 5. The L¨uroth problem asks whether every unirational variety is rational. but in characteristic p ¤ 0. Let A1 denote the functor sending a k-algebra R to its underlying set. r Rational and unirational varieties D EFINITION 5. three examples of nonrational unirational three-folds were found. g´en´eralit´es. Zariski showed that some surfaces of the form Z p D f . Y /. Surfaces of this form are now called Zariski surfaces. neither affine nor projective one needs to be more careful to ensure that O. but none of his attempted proofs satisfies modern standards. For a description of them.Rfi /  X. L¨uroth proved that every unirational curve is rational. for example.11. restrict U and A1 to the category of k-algebras of the form kŒX0 .U / D Hom. A local functor admitting a finite covering by open affines is representable by an algebraic variety over k. are not rational. for example.V / is contained in such an extension of k. V is rational if k.Xi / indexed by N.R/ ! X.02476. p. FT. let O.V / is a pure transcendental extension of k. : : :=a for a fixed family of symbols . (a) V is unirational if there exists a dominant rational map Pn Ü V . 3. (b) V is rational if there exists a birational map Pn Ü V: In more down-to-earth terms. this is a difficult problem. Weil made use of the Jacobian variety of the curve. A LGEBRAIC VARIETIES A functor X is local if. Already for surfaces. 1. 1970. for all k-algebras R and all finite families . The L¨uroth problem. the sequence of sets Y Y X. In the functorial approach to algebraic geometry. Groupes alg´ebriques: g´eom´etrie alg´ebrique.U / is a set.fi /i of elements of A generating the ideal A.U / is an affine k-algebra and the canonical map U ! hO. See. .j is exact. In 1971-72. 8 Actually. Chapter 9.

] 5-2. 5-3. this proves the statement in this case. which is also the unique connected component of G containing e. Show that V is affine if and only if each Vi is affine. y/ 7! . Weil first made use of the Zariski topology when he introduced fibre spaces into algebraic geometry (in 1949). Show: (a) The neutral element e of G is contained in a unique irreducible component G ı of G.x.G.r. This set has an obvious topology and ringed space structure for which it is an algebraic variety. Rational and unirational varieties 127 (in 1946).71).x. 5-5. Prove the final statement in (5. see my article: The Riemann hypothesis over finite fields: from Weil to the present day. (b) The subvariety G ı is a normal subgroup of G of finite index. Exercises 5-1. Vn . For more on this.k/ is a group. When k D C. The general case is easier. Show that a prevariety V is separated if and only if it satisfies the following condition: a regular map U X fP g ! V with U a curve and P a nonsingular point on U extends in at most one way to a regular map U ! V . Show that . 5-4. and one knows from complex analysis that the only holomorphic functions on the Riemann sphere are constant. Since regular functions are holomorphic. 5-6. and every algebraic subgroup of G of finite index contains G ı . Show that every subgroup variety of a group variety is closed. m/ is a group variety. and assume that G. Let G be an algebraic group. . xy/W G  G ! G  G is an isomorphism of algebraic varieties. Let G be a smooth monoid variety over field of characteristic zero. [Thus P1 is not affine. : : : . and deduced that that inv is regular. 5-7. Show that the only regular functions on P1 are the constant functions. P1 is the Riemann sphere (as a set). A monoid variety is an algebraic variety G together with a monoid structure defined by regular maps m and e. Let V be the disjoint union of algebraic varieties V1 . hence .

.

F . : : : . an /. Every affine variety is quasiprojective.C HAPTER Projective Varieties Recall (5. and let ui be the bijection   ui . Xn / 129 6 . bn / for some c 2 k  : Let .X0 .3) that we defined Pn to be the set of equivalence classes in k nC1 X foriging for the relation . equivalently. but not all algebraic varieties are quasiprojective. : : : . : : : . : : : . even when we would like to ignore it. : : : . bn / ” .a0 . 0/g ! Pn :  Let Ui be the set of . a Algebraic subsets of Pn A polynomial F .b0 . one has only an inequality.in X0i0    Xnin with i0 C    C in D d . and the “part at infinity” often plays a role. and a variety isomorphic to a locally closed subvariety of Pn is called a quasiprojective variety.0. We study morphisms between quasiprojective varieties. a famous theorem of Bezout (see 6.tX0 . an / D c. : : : .b0 . an /  . : : : .a0 . tXn / D t d F . and let  denote the map k nC1 X f. we show that Pn has a unique structure of an algebraic variety for which these maps become isomorphisms of affine algebraic varieties. In this chapter. A variety isomorphic to a closed subvariety of Pn is called a projective variety. : : : .a0 W : : : W an / 7! aa0i .a0 W : : : W an / denote the equivalence class of .37 below) says that a curve of degree m in the projective plane intersects a curve of degree n in exactly mn points (counting multiplicities). aabii . For affine curves. : : : .X0 . : : : . : : : .a0 W : : : W an / 2 Pn such that ai ¤ 0. Xn / is said to be homogeneous of degree d if it is a sum of terms ai0 .a0 . For example. Projective varieties are important for the same reason compact manifolds are important: results are often simpler when stated for projective varieties. aani W Ui ! An ( aaii omitted).:::.

and M kŒX0 . : : : . Xn . Xn / at P .X0 .0W 1W 0/ corresponds to the point . Let P D . and that . Xn  D kŒX0 . : : : .a0 .ca0 . . Consider the projective algebraic subset of P2 defined by the homogeneous equation E W Y 2 Z D X 3 C aXZ 2 C bZ 3 . Note that E \ U1 is the affine curve Z D X 3 C aXZ 2 C bZ 3 . P ROJECTIVE VARIETIES for all t 2 k.0. 0/ is nonsingular on E \ U1 . if F is homogeneous. : : : .x W y W 1/ on the affine curve E \ U2 Y 2 D X 3 C aX C b (see 2.ca0 W : : : W can / for any c 2 k  . A nonsingular curve of the form (25) is called an elliptic curve. an /.5) that E is nonsingular unless X 3 C aX C b has a multiple root. every polynomial F can be written uniquely as a sum F D Fd with Fd homogeneous of degree d . : : : . However. The polynomials homogeneous of degree d form a subspace kŒX0 .1. can / D c d F . (25) It consists of the points . then F . An algebraic set in Pn (or projective algebraic set) is the set of common zeros in Pn of some set of homogeneous polynomials. 0/ on E \ U1 : Z D X 3 C XZ 2 C Z 3 As . and so it does make sense to say that F is zero or not zero at P . Xn d of kŒX0 .2) together with the point “at infinity” . : : : .130 6. : : : . : : : .a0 W : : : W an / 2 Pn . E XAMPLE 6.0 W 1 W 0/. Then P also equals . we deduce from (4. and so we can’t speak of the value of a polynomial F . Xn d I d 0 P in other words.0.

: : : . 0/g :  The sets V . Fr are homogeneous generators for a. If F1 . we let V . For a graded ideal a.a/ X f. if F 2 a H) Fd 2 a.a/. product.Q/.0. It is easy to compute the torsion subgroup of E.a/  V . we can speak of the zeros of (25) with coordinates in Q. Xn  is said to be graded or homogeneous if it contains with any polynomial F all the homogeneous components of F .e. i. More precisely.Q/.. Algebraic subsets of Pn 131 An elliptic curve has a unique structure of a group variety for which the point at infinity is the zero: Q P P CQ When a. i.a/ D . then V .a/ is also the set of common zeros of the Fi . It is a cone in k nC1 . : : : . Clearly a  b H) V . Clearly every polynomial in a is zero on every representative of a point in V . there is an “algorithm” which works in practice. but which has not been proved to always terminate after a finite amount of time. V . all d: It is straightforward to check that ˘ an ideal is graded if and only if it is generated by (a finite set of) homogeneous polynomials. We write V aff .a. but there is at present no known algorithm for computing the rank of E.a/ denote the set of common zeros of the homogeneous polynomials in a. : : : . ˘ an intersection..e. which Mordell showed to be finitely generated. or sum of graded ideals is graded. ˘ the radical of a graded ideal is graded.Q/. whose origins can be traced back almost 1.a/ in Pn have similar properties to their namesakes in An .) An ideal a  kŒX0 . (See my book on Elliptic Curves for all of this. b 2 Q. and V aff . together with any point P it contains the line through P and the origin.a/ for the set of common zeros of a in k nC1 .800 years to Diophantus. They also form a group E. There is a very beautiful theory surrounding elliptic curves over Q and other number fields.b/.

2. Xn .a0 . we define the affine cone over S in k nC1 to be C D 1 . Xn /I (b) V . note that V . 0/g ” rad. : : : . Pn is irreducible. Obviously. ” V aff . There are the following relations: (a) V . ai / D V .ai /.132 6. Note that we have bijections falgebraic subsets of Pn g V S 7! C fnonempty closed cones in k nC1 g I fproper graded radical ideals in kŒX0 .a/  . V is irreducible if and only if the closure of  1 .S / D . The remaining statements can be proved directly.ab/ D V .a0 . Xn : if F .3.. which proves the first statement because the composite of the top map with I is I in the sense of projective geometry. then C is the closure of  1 .V / is a prime ideal.a/  f.ca0 .ca0 . This topology is called the Zariski topology on Pn .V / is prime. and the maps V and I are in the sense of projective geometry and affine geometry respectively. 2 . The maps V and I define inverse bijections between the set of algebraic subsets of Pn and the set of proper graded radical ideals of kŒX0 .S / [ foriging and we set I. If C is a cone in k nC1 . The composite of any three of these maps is the identity map.X0 . : : : . P ROJECTIVE VARIETIES P ROPOSITION 6.a/.a/ D ¿ ” rad. as in (2.a/ and V aff .0/ D Pn I V .0. in particular.S / D I.a/ [ V . Xn g Here the top map sends S to the affine cone over S . an /  c d D F . : : : . : : : .C / is a graded ideal in kŒX0 .10). and that I. then I. : : : . : : : . Xn / (strong Nullstellensatz 2.V / is irreducible.X0 . which is true if and only if I. P ROOF.a/  . P ROPOSITION 6.a \ b/ D V . An algebraic set V in Pn is irreducible if and only if I. and so Fd . can / D 0 for all c 2 k  .2 shows that the projective algebraic sets are the closed sets for a topology on Pn . an /X d is the zero polynomial. can / D 0. For a subset S of Pn . d P for infinitely many c. : : : .16).a/ D . Note that if S is nonempty and closed. For the second statement in (a).S / is spanned by the homogeneous polynomials in kŒX0 . Xn  that are zero on S. 2 Proposition 6. or by using the relation between V . then X Fd .b/I P T (c) V . : : : . : : : .C /. : : : . : : : . P ROOF.

b The Zariski topology on Pn For a graded polynomial F . X1 . P ROOF. : : : . : : : . Clearly. The Zariski topology on Pn 133 Note that the graded ideals . D V . : : : . : : : . : : : . : : : . : : : . : : : .X0 /. : : : .f / D D. the bijection Ui $ An . : : : .P / ¤ 0g: Then. A SIDE 6. It suffices to prove this with i D 0. : : : . D. D. . Xn // \ U0 D D. To each polynomial f . 11. X0 . Xn /. X1 . let D.X0 . : : : .X0 . 2 . just as in the affine case. an / becomes a homeomorphism.X0 . . As in the opening paragraph of this chapter.kŒX0 . and when we endow it with the induced topology. . : : : .a0 W : : : W 1 W : : : W an / $ . D.Sd //. Xn  are both radical. Alg. II.X0 . Xn / and so the correspondence between irreducible subsets of Pn and radical graded ideals is not quite one-to-one. we attach the homogeneous polynomial of the same degree   deg. but V .5. this means that it is a graded submodule of . Xn // D D. and so it is a basic open subset in Pn . for any homogeneous polynomial F 2 kŒX0 .F / and. Each subset Ui of Pn is open in the Zariski topology on Pn .F .F / is open and the sets of this type form a base for the topology of Pn . Xn /: P ROPOSITION 6. : : : .X0 . : : : . A graded ring is a pair . all d.Sd /d 2N / consisting of a ring S and a family of additive subgroups Sd such that ( M SD Sd d 2N Sd Se  Sd Ce .X1 .a \ Sd /.f  / \ U0 : Thus. which proves that the bijection is a homeomorphism. 0 and to each homogeneous polynomial F .F / D fP 2 Pn j F . for any polynomial f 2 kŒX1 . e 2 N: An ideal a in S is graded if and only if aD M d 2N . In English “homogeneous ideal” is more common than “graded ideal”. Xn / D F .1.Xi /.S. Xn /. The set U0 D D.4. the basic open subsets of An correspond to the intersections with Ui of the basic open subsets of Pn . we attach the polynomial F .1.f / X1 Xn f  . Xn / and kŒX0 . : : : . ai 1 . Xn / D . we let Ui D D. Xn . but we follow Bourbaki. Xn . The quotient of a graded ring S by a graded L ideal a is a graded ring S=a D d Sd =.a \ Sd /.S. under the bijection U0 $ An . ai C1 . Xn / D X0 f X .F .a0 .b.X1 .

: : : .1/ and . For V W Y 2 D X 3 C aX C b.a / with a D I. Closed subsets of An and Pn c We identify An with U0 . It follows that its closure is all of Pn .X1 . X12 C X2 / D f. 0/gI then V . ¤ V . : : : . For some i . and .V /. and so must meet Ui \ Uj .  and . and examine the closures in Pn of closed subsets of An .V / D V . When V is a closed subset of Pn . in this case a D . X1 . U \ Ui is nonempty.8.a / with a D I.6. For a closed subset V of An .f1 . Therefore U meets every Uj . 1 Of . and if no irreducible component of V lies in H1 or contains H1 . We apply the criterion that a space is irreducible if and only if every nonempty open subset is dense (see p. X2 /. course. (a) Let S V be a closed subset of A .f1 .1W 0W 0/g. X2 /. Let V D V .V  / D V . For V D H1 D V . (b) Let V be a closed subset of Pn . let V D V . Xn . fm / not contained in H1 .0W 0W 1/ (which is contained in H1 ). Then V is the closure of V in P . P ROJECTIVE VARIETIES R EMARK 6.X1 .V / D . and V  D f. : : : . P ROOF. Straightforward.44). fm /.X0 X1 . : : : .10. and . we have V D . a D .V /. With each graded ideal a in kŒX0 . we have V  W Y 2 Z D X 3 C aXZ 2 C bZ 3 . X12 C X0 X2 / consists of the two points .X0 /. Xn ]. we set V D V .X1 . set V  D V . then V is a proper subset of An . For example. and so is dense in every Uj . Xn  generated by ff  j f 2 ag.a0 W : : : W 1 W : : : W aj W : : : W an / with aj ¤ 0/. H1 D V .1 6. then V  D Vi is the decomposition of V  into its irreducible components. 6.9. Note that Pn D An t H1 .1W 0W 0/ (the closure of V ) and . Xn  generated by fF j F 2 ag. and that Ui \ Uj is open and dense in each of Ui and Uj (as a subset of Ui .0. Note that each Ui is irreducible.X0 /: With each ideal a in kŒX1 . It is possible to use this to give a different proof that Pn is irreducible. Let U be a nonempty open subset of Pn . we associate the ideal a in kŒX1 . 2 Examples 6. it is the set of points . D V . then U \ Ui is open in Ui . we associate the graded ideal a in kŒX0 . : : : . If V D Vi is the decomposition of V into its irreducible components. Then V D V \ An .V / D S V . and so this example doesn’t contradict the proposition.134 6. n  n P ROPOSITION 6. then the closure of V in Pn is the union of the irreducible components of V . : : : .7.

X . L EMMA 6. P ROOF.gjUij / with f 2 .0 W b W a/. The hyperplane at infinity d 135 The hyperplane at infinity It is often convenient to think of Pn as being An D U0 with a hyperplane added “at infinity”. Xn /.0. j / D .Uij . Xn / generated by the quotients — it is the polynomial ring in the n symbols X1 n kŒ X . : : : . Write Uij D Ui \ Uj . with H1 consisting of a single point.d. X X0  defines a map 0 X1 Xn X0 . which equals . Note that a=b is the slope of the line 1 C aX1 C bX2 D 0. Similarly. It suffices to prove this for . the complement of U0 in Pn is H1 D f.X0 . then Oi jUij D Oj jUij . When endowed with this sheaf.i. as in Example 6. aa02 . aan0 / 2 An : Xn 1 X2 Let kŒ X X0 . g 2 . e Pn is an algebraic variety For each i.X0 / at the point . Recall that U0 D f. .a0 W a1 W : : : W an / $ .0 W 1 W a=b/ when b ¤ 0. Oi /.:::. X1 . Note that in this example the point at infinity on the elliptic curve Y 2 D X 3 C aX C b is the intersection of the closure of any vertical line with H1 .Uj . Consider the line 1 C aX1 C bX2 D 0 in A2 .f jUij /. and P2 D A2 [ H1 with H1 a projective line. X0 . we can think of Pn as being An with one point added at infinity for each direction in An — being parallel is an equivalence relation on the lines in An . aa10 . Xi X0 X1 n An element f .0 W a1 W : : : W an /  Pn g. : : : .a0 W a1 W : : : W an / j a0 ¤ 0g. We can replace U0 with Un in the above discussion. : : : . Oj /. write Oi for the sheaf on Ui  Pn defined by the homeomorphism ui W Ui ! An . X0 . and write Pn D Un t H1 with H1 D f. aa10 .:::. .Ui . X X0 / 2 0 . P1 D A1 t H1 (disjoint union). 1/. : : : .a0 W a1 W : : : W an / 7! f .11. Its closure in P2 is the line X0 C aX1 C bX2 D 0: This line intersects the line H1 D V . For example. aan0 /W U0 ! k. X0  be the subring of k.a0 W : : : W an 1 W 0/. which can be identified with Pn 1 . . and there is one point at infinity for each equivalence class of lines. We can think of the projective plane P2 as being the affine plane A2 with one point added at infinity for each “direction” in A2 . we identify the set U0 with An . moreover. and so the point at which a line intersects H1 depends only on the slope of the line: parallel lines meet in one point at infinity. : : : . Oi / is generated as a k-algebra by the functions . X1 .1. Uij is an affine algebraic variety.X0 . All rings occurring in the proof will be identified with subrings of the field k. More precisely. : : : .

a 1 /.x W y W z/ 7! . X0 . : : : .! Pn is a homeomorphism onto an open subset of Pn and OPn jUi D OUi . : : : .  X1 X0 X1 . and the third is obvious: kŒ X X0 . and is therefore an affine subvariety of . Therefore.'a . X1  and kŒ X1 . X1 / of kŒ X0 . X1 . : : : . X1  is generated by X0 X2 Xn Xn 1 its subrings kŒ X X0 . U1 D f. aan1 . : : : .a0 W : : : W an / 7! f . aa01 . a1 ¤ 0g: 1 It is D. 2 E XAMPLE 6. X1 / in kŒ X1 .a0 W : : : W an / 7! f .29c) to be separated. X0 .! kŒ X1 . aa01 . X1 . the lemma shows that Pn satisfies the condition (5. : : : . O0 /.X0 . : : : .   X0 X0 Xn Xn Xn 0 X2 and we identify U1 with Spm kŒ X .P. there is an automorphism 'a . : : : . . O1 /. X1  defines the function .P /. P 2 C . and the inclusion U01 . X X0  . This Xn X0 1 proves the first two assertions. aa01 . and an element of this ring defines the same function on U01 regardless of which of the two rings it is considered an element.! U0 X1 X1 Xn X0 n corresponds to the inclusion of rings kŒ X .k/ ¤ 2. . : : : . a/ with . X0  of k. X1  . whether we regard U01 as a subvariety of U0 or of U1 it inherits the same structure as an affine algebraic variety (3.12.15). : : : . and is therefore an affine subvariety of . and U0 with Spm kŒ X X0 .a0 W : : : W an / 7! f . a0 .:::.U1 .136 6. Next consider the open subset of U0 . : : : .a0 W : : : W an / j a0 ¤ 0. X0 . Similarly.! U1 corresponds to the inclusion of rings X0 X0 X0 Xn Xn X1 Xn X1 Xn X1 0 kŒ X X1 .a0 W a1 W : : : W an / $ . and the lemma shows that this covering satisfies the patching condition (5. An element 0 X1 a1 Xn X0 Xn X0 an a0 1 f .13.ax W y W a3 z/W C ! C: Patch two copies of C  A1 together along C  . this structure makes Pn into an algebraic prevariety. A polynomial f . X0  becomes  Xn 1 U0 . aa10 / on U01 .a0 W a1 W : : : W an / j a1 ¤ 0g. Xn / are equal. X0  and kŒ X1 . 2 P ROPOSITION 6. : : : . : : : . P ROJECTIVE VARIETIES Xn 1 X2 and in this way kŒ X identified with the ring of regular functions on X0 . X0 . X0  . The inclusion U01 . and so Pn has a unique structure of a ringed space for which Ui . X1 . because each Ui is an algebraic variety. One obtains in this way a singular surface that is not . 0 When regarded as an open subset of U1 .A1 f0g/ by identifying . Finally. U01 D f. a1 / on U01 . X0 . X1  defines the map . : : : . : : : . X0  defines the function . For each a 2 k  . a0 . : : : . X X1 /. X0 / of kŒ X1 . X1 . Then Pn D Ui .15). X0 . Endow each Ui withS the structure of an affine algebraic variety for which ui is an isomorphism. X1 . Moreover. X1 . An element f . : : : . a 2 A1 X f0g. : : : . : : : . X1 . : : : . : : : . : : : . P ROOF. X1 .X X0 . X X0 /.U0 . X1 . Let C be the plane projective curve C W Y 2Z D X 3 and assume that char. aan1 /W U1 ! k. : : : . There is a unique structure of a (separated) algebraic variety on Pn for which each Ui is an open affine subvariety of Pn and each map ui is an isomorphism of algebraic varieties. aan1 / 2 An . X0 . X1 . X0 Xn X0 Xn X1 1 The two subrings kŒ X X0 .! kŒ X0 . U01 D D.

an / is a well-defined function. We now extend this discussion to any irreducible projective variety V . whereas khom Œ. It is known that every irreducible separated curve is quasiprojective. X1 . We say that a nonzero f 2 khom ŒV  is homogeneous of degree d if it can be represented by a homogeneous polynomial F of degree d in kŒX0 . The homogeneous coordinate ring of a projective variety 137 quasiprojective (see Hartshorne 1977. and wish to identify the field of fractions of kŒU0  as a subfield of k. Xn /. : : : . and it follows that the field of fractions of kŒU0  is ˇ   G. : : : . Xn / deg.X0 . : : : .a0 . : : : .V / is the field of fractions of kŒU  for any open affine U  V .X0 . Xn /0 for this field (the subscript 0 is short for “subfield of elements of G degree 0”).X0 . It is even complete — see below — and so if it were quasiprojective. but khom ŒP1  D kŒX0 .p/ with p a graded radical ideal in kŒX0 . : : : . Xn / ˇˇ k.Pn / D k. Xn /0 .X0 . which is obviously regular (look at its restriction to Ui /.X0 .F /. and we say that 0 is homogeneous of degree 0. X0 . Xn /. Note that khom ŒV  is the ring of regular functions on the affine cone over V . For example.13). it would be projective.e.X0 . We regard this as a subring of k. . Xn . Every nonzero F 2 kŒU0  can be written Xn 1 F. : : : .X0 .a0 .F / X0 with F  homogeneous of degree deg. but on the embedding of V into Pn . Such a V can be written V D V . X1 . there is a field k.X0 . and so these examples are minimal. which is the affine coordinate ring of the smooth variety A2 . Recall that kŒU0  D kŒ X X0 . it is not intrinsic to V . : : : .f. : : : .V / with the property that k.H / ! k. Xn . Exercise 7.114) that attached to each irreducible variety V . : : : . : : : . not only on V . : : : . VI 2. : : : . and every nonsingular complete surface is projective.a02 W a0 a1 W a12 /W P1 ! P2 is an isomorphism from P1 onto its image . f The homogeneous coordinate ring of a projective variety Recall (p.a0 W a1 / 7! . . so that k. which is the affine coordinate ring of the singular variety X0 X2 X12 . : : : . We now Xn 1 describe this field in the case that V D Pn . In Shafarevich 1994. X0 / D F  .. : : : . Note that for F D H in k. an / W D. Xn /0 .  .X0 X2 X12 /. and we define the homogeneous coordinate ring of V (with its given embedding) to be khom ŒV  D kŒX0 . therefore its dimension is dim. there is an example of a nonsingular complete variety of dimension 3 that is not projective.3.U0 / D G. i. H homogeneous of the same degree [ f0g: H.V / C 1: It depends.X X0 .P1 / D kŒX0 .P1 /W X0 X2 D X12 (see 6. Xn / ˇ Write k. H.23 below).a0 W : : : W an / 7! G. X2 =. Xn =p. : : : .

: : : . an /  h0 . an / D g 0 . then gh0 g 0 h (in khom ŒV ) and so g.a0 . an / is replaced by . h . we can identify kŒV0  with a subring of khom ŒV . : : : . Let F represent f .P /. As in the case of Pn . : : : .16.P / ¤ 0.2 but if f D g g0 D 0 h h (in k. an /: Thus. when read modulo p. there is an open subset U of V where f . For any P D . For such an element h.a0 W : : : W an / with h.a0 .V /. and let f 2 k.V /. def f .V /0 ). every regular function on an open subset of V arises from a unique element of k. : : : . Define ˇ ng o ˇ khom .P / ¤ 0. Since khom ŒV  is an integral domain. Suppose f also has a decomposition f D gi .g/ D c deg. and P 7! f .F G/i 2 p. this gives P a decomposition of f of the required type.V / D khom .V /0 D 2 khom . def P ROOF. 2 g Regular functions on a projective variety Let V be an irreducible projective variety. 2 It therefore makes sense to speak of homogeneous elements of kŒV . By definition. Then F G 2 p.h/ .a0 . an /  h. : : : .a0 . can /.V / ˇ g and h homogeneous of the same degree [ f0g: h def P ROPOSITION 6. : : : .P / is a regular function on U . For each f 2 k.a0 . if h0 . P ROJECTIVE VARIETIES L EMMA 6. we define D. we can write f D gh with g and h homogeneous of the same degree in khom ŒV  and h ¤ 0.V / D khom . then both the numerator and denominator are multiplied by c deg. : : : . Each element of khom ŒV  can be written uniquely in the form f D f0 C    C fd with fi homogeneous of degree i .V /0 .138 6. We can write f in the form gh in many different ways.V /0 . and the homogeneity of p implies that Fi Gi D .14.V /. we can form its field of fractions khom . Therefore fi D gi . 2 Unless khom ŒV  is a unique factorization domain. P ROOF. then F can be written F D F0 C    C Fd with Fi homogeneous of degree i .ca0 . and then the field of fractions of kŒV0  becomes identified with khom . there will be no preferred representation f D g . Consider V0 D U0 \ V .15.P / ¤ 0g. an / is well-defined: if . def P ROPOSITION 6.V /0 . The field of rational functions on V is k.h/ D fP 2 V j h.P / D g.a0 . with gi represented by the homogeneous polynomial Gi of degree i . : : : . an / h. the two representations give the same value for f .P / is defined.a0 .

From the above discussion.18. The ring of regular functions on it is kŒD.17.) It now follows from (5.V \ Ui / D k. A map ˛W Pn ! V with V a prevariety is regular if and only if ˛ ı  is regular. and let P 2 U . if U is an open nonaffine subset of V . the meromorphic functions on the Riemann sphere are the rational functions in z.] (b) We shall see presently (6. 2 R EMARK 6. an / 7! . The map W AnC1 X foriging ! Pn . However.V /0 should be regarded as the algebraic analogues of meromorphic functions on a complex manifold. This shows that f is the function defined by fP on the whole of U .V /. then the ring . and the meromorphic functions on the manifold are precisely the rational functions on the variety. i. Conversely. : : : .e. which is the regular map of affine varieties corresponding to the map of k-algebras h i Xn 1 0 k X . OV / of regular functions can be almost anything — it needn’t even be a finitely generated k-algebra! h Maps from projective varieties We describe the morphisms from a projective variety to another variety. let f be a regular function on an open subset U of V .V / D khom . when k D C. D.Xi / ai /W k ! Ui .Xi / is the projection . the regular functions on an open subset U of V are the “meromorphic functions without poles” on U .h/ where h runs over the denominators of expressions f D gh with g and h homogeneous of the same degree in khom ŒV .h. For example.V / on an open neighbourhood of P . any regular function on an irreducible (connected will do) projective variety is constant. Xj Xi is to be thought of as a single symbol.24) that. [In fact. : : : .a0 . The restriction of  to D. P ROOF. Maps from projective varieties 139 S P ROOF. then fP and fQ define the same function on some open affine U 0 . an / 7! . and so fP D fQ as elements of kŒU 0   k.h/ D fg= hm j g homogeneous of degree m deg.V / in a neighbourhood of a second point Q of U . Then P lies in the open affine subvariety V \ Ui for some i . (a) The elements of k.a0 W : : : W an / is an open morphism of algebraic varieties.4) that . Xn ŒXi : (In the first algebra  is regular. we see that f defines a regular function on U D D. Xi Xi ! kŒX0 . for any nonzero homogeneous h 2 khom ŒV .h/ is an open affine subset of V .U. If f coincides with the function defined by fQ 2 k. aa0i W : : : W an nC1 X V .a0 . and so f coincides with the function defined by some fP 2 k. P ROPOSITION 6.. : : : . : : : . .h/g [ f0g: We shall also see that the ring of regular functions on V itself is just k. this is more than an analogy: a nonsingular projective algebraic variety over C defines a complex manifold.

More precisely. Its restriction to any subvariety V of P will also be regular.F0 . It may be possible to extend the map to a larger set by representing it by different polynomials. Fn . define W C ! P1  by . after a change of variables. homogeneous of the same degree. : : : . because U D cU . bm / W : : : W Fn .b W c/ if a ¤ 0 : if b ¤ 0 Note that.a0 W : : : W am / 7! .X0 .Fi / D P X V . U D  .b0 . Thus  sends open sets to open sets. 2 E XAMPLE 6.F0 . Xm /. that is. Define 'W P1 ! C . This equation can be rewritten . Straightforward. A map 'W V ! W is regular if and only if.X C iY /. am / W : : : W Fn .. : : : . Xm / of the same degree. : : : .U /. bm // for all points .a W b/ .21.F1 . and x 7! cx is an automorphism of k 0 nC1 The complement Z of U in k X foriging is a closed cone. Then U is again open 0 nC1 X foriging. : : : . m is.a2 W ab W b 2 /: For the inverse. : : : . a ¤ 0 ¤ b. : : : . and let U 0 be the union of all the lines 0 1 0 through the origin that meet in k nC1 X S U . the regular maps Pn ! V are just the regular maps AnC1 X foriging ! V factoring through Pn (as maps of sets). there exist polynomials F0 .20. : : : . We prove that the circle X 2 CY 2 D Z 2 is isomorphic to P1 .b0 W : : : W bn // D . there is the following result. c b D b a and so the two maps agree on the set where they are both defined. on the set D. ac D b 2 H) are . Xm /.X iY / D Z 2 .b/  Pn . and one checks directly that they are inverse. c 2 k . R EMARK 6. P ROJECTIVE VARIETIES Let U be an open subset of k nC1 X foriging. the equation of the circle becomes C W XZ D Y 2 .a W b W c/ 7! .b0 . am // n m obviously defines that S a regular nmap to P on the open subset of P where not all Fi vanish.X0 . The rest of the proof is straightforward.b0 W : : : W bm / in some neighbourhood of P in V . for every P 2 V . : : : . Consider polynomials F0 .a W b/ 7! .a/. : : : . The map .a/  Pm and W D V .U / is the complement of . both ' and regular.Z/.a0 .19. : : : . Let V D V . every such map arises in this way.X0 . Xm /. Fn /.a W b W c/ 7! . P ROPOSITION 6.140 6. Conversely. 2 Thus.3) shows that its image is closed in Pn . such that ' . Fn . and so. Clearly.a0 .X0 . and the proof of (6. but . : : : . at least locally. . P ROOF. foriging.

with inverse . : : : . : : : . Therefore. X1 .X0 .L/. S is contained in an open affine subset D. Xn / D 1 of AnC1 . Moreover. Some classical maps of projective varieties i 141 Some classical maps of projective varieties We list some of the classic maps.:::.i0 .a0 . HYPERSURFACE SECTIONS 6. each quotient P c jX is to be thought of as a single ci Xi i i P i Xi symbol. For a fixed P D . Then the map   a0 an . As V . PXc nX : In this ring.a/ is a bijection of D.a/ L. Let I D f.a0 W : : : W an / 2 Pn .P / D ci ai ¤ 0 is a nonempty open subset of Pn (n > 0). if S  V where V is a closed subvariety of Pn . and its ring of regular functions i i X 0 is kŒ PX . for any finite set S of points of Pn .Lc /g is a nonempty open subset of Pn (because Pn is irreducible). any one symbol i i PXj ci Xi for which cj ¤ 0 can be omitted.23.a0 W : : : W an / 7! . the set of c D . fc 2 Pn j S  D. L. the components of the first map are the regular X functions P c jX .22. T HE V ERONESE MAP .i. In particular.a0 W : : : W an /: Both maps are regular — for example.c0 W : : : W cn / such that X def Lc .Lc /: any finite set of points of a projective variety is contained in an open affine subvariety. and cj P c jX D 1. an / 7! .L/  Pn onto the hyperplane L. so also is D. Let L D P ci Xi be a nonzero linear form in n C 1 variables. : : : . : : : . thus it is a polynomial ring in n symbols. in / 2 NnC1 j X ij D mg: . H YPERPLANE SECTIONS AND COMPLEMENTS 6.L 1/ is affine. then S  V \ D.Lc / of Pn .

all h: 3 This can be proved by induction on m C n. which is correct.X0 . : : : . the Veronese map is P1 ! Pm . .X C 1/mCn and this proves the induction.Pn /.m can be written .0 W b1. . X1 .X0 . Note that I indexes the monomials of degree m in n C 1 variables. the Veronese mapping sends an n C 1-tuple .a0m W a0m 1 a1 W : : : W a1m /: I claim that. (*) . Example 6. A general F . an / ¤ . We shall see in the next chapter that the image of any projective variety under a regular map is closed. and the map  .b1.m . thus a point of Pn.a0 W a1 / 7! .a0 .2 / if b0.0.a0 W a1 / 7! . bi0 :::in . : : : . As L ¤ 0 ) Lm ¤ 0.Pn / is defined by the system of equations: bi0 :::in bj0 :::jn D bk0 :::kn b`0 :::`n . But    mCn D mCn 1 C mCn 1 n m m 1 because they are the coefficients of X m in . The Veronese mapping is defined to be bi0 :::in D a0i0 : : : anin : vW Pn ! Pn. If m D 0 D n.X C 1/. 1 . . .P1 / W X0 X2 D X12 . 0/ ) .a0 W : : : W an / to the set of monomials in the ai of degree m. For example. ih C jh D kh C `h .1 / if b2.a0 W : : : W an / 7! . Xn / C X0 F2 . L1 ¤ cL2 ) Lm 1 ¤ cL2 . It has m  Write n.X1 .2 ¤ 0: is an inverse . that is.: : : W bi0 :::in W : : :/. because kŒX0 .X C 1/mCn D .1 W b0.: : : . then the coordinates of . : : : .142 6. the map  is defined on the whole of Pn .b2. In other words. the image of  is a closed subset of Pn.0 ¤ 1 . P ROJECTIVE VARIETIES  mCn elements3 . Xn / D F1 . : : :/ ¤ . : : : .: : : W bi0 :::in W : : :/.m D mCn 1.a02 W a0 a1 W a12 /: Its image is the curve .P / are the coefficients of the homogeneous polynomial Lm with the binomial coefficients omitted. (Cf.P1 / ! P1 . : : : . but in this case we can prove directly that . the Veronese map is P1 ! P2 .b2. First note that the map has the following interpretation: if we P regard the coordinates ai of a point P of Pn as being the coefficients of a linear form L D ai Xi (well-defined up to multiplication by nonzero scalar). It is clear from its definition that  is regular.0 W b1. : : : .m whose coordinates are m indexed by I .0.2 / 7! . and so  is injective. X1 . in the general case. : : : . and consider the projective space Pn. 0/: m Moreover. then homogeneous polynomial of degree m can be written uniquely as 0 0  D 1. Xn / with F1 homogeneous of degree m and F2 homogeneous of degree m 1. when n D 1 and m D 2.) When n D 1 and m is general.1 W b0.22. Xn  is a unique factorization domain.m and that  defines an isomorphism of projective varieties W Pn ! .

i. Some classical maps of projective varieties

143

Obviously Pn maps into the algebraic set defined by these equations. Conversely, let
Vi D f.: : : : W bi0 :::in W : : :/ j b0:::0m0:::0 ¤ 0g:
Then .Ui /  Vi and  1 .Vi / D Ui . It is possible to write down a regular map Vi ! Ui
inverse to jUi : for example, define V0 ! Pn to be
.: : : W bi0 :::in W : : :/ 7! .bm;0;:::;0 W bm 1;1;0;:::;0 W bm 1;0;1;0;:::;0 W : : : W bm 1;0;:::;0;1 /:
S
Finally, one checks that .Pn /  Vi .
For any closed variety W  Pn , jW is an isomorphism of W onto a closed subvariety 
.W / of .Pn /  Pn;m .
6.24. The Veronese mapping has a very important property. If F is a nonzero homogeneous
form of degree m  1, then V .F /  Pn is called a hypersurface of degree m and V .F / \ W
is called a hypersurface section of the projective variety W . When m D 1, “surface” is
replaced by “plane”.
Now let H be the hypersurface in Pn of degree m
X
ai0 :::in X0i0    Xnin D 0,
and let L be the hyperplane in Pn;m defined by
X
ai0 :::in Xi0 :::in :
Then .H / D .Pn / \ L, i.e.,
H.a/ D 0 ” L..a// D 0:
Thus for any closed subvariety W of Pn ,  defines an isomorphism of the hypersurface
section W \ H of V onto the hyperplane section .W / \ L of .W /. This observation often
allows one to reduce questions about hypersurface sections to questions about hyperplane
sections.
As one example of this, note that  maps the complement of a hypersurface section of
W isomorphically onto the complement of a hyperplane section of .W /, which we know
to be affine. Thus the complement of any hypersurface section of a projective variety is an
affine variety.

AUTOMORPHISMS OF Pn
6.25. An element A D .aij / of GLnC1 defines an automorphism of Pn :
P
.x0 W : : : W xn / 7! .: : : W aij xj W : : :/I
clearly it is a regular map, and the inverse matrix gives the inverse map. Scalar matrices act
as the identity map.
Let PGLnC1 D GLnC1 =k  I , where I is the identity matrix, that is, PGLnC1 is the
2
quotient of GLnC1 by its centre. Then PGLnC1 is the complement in P.nC1/ 1 of the
hypersurface det.Xij / D 0, and so it is an affine variety with ring of regular functions
kŒPGLnC1  D fF .: : : ; Xij ; : : :/= det.Xij /m j deg.F / D m  .n C 1/g [ f0g:

144

6. P ROJECTIVE VARIETIES

It is an affine group variety.
The homomorphism PGLnC1 ! Aut.Pn / is obviously injective. We sketch a proof that
it is surjective.4 Consider a hypersurface
H W F .X0 ; : : : ; Xn / D 0
in Pn and a line
L D f.t a0 W : : : W t an / j t 2 kg
in Pn . The points of H \ L are given by the solutions of
F .t a0 ; : : : ; t an / D 0,
which is a polynomial of degree  deg.F / in t unless L  H . Therefore, H \ L contains 
deg.F / points, and it is not hard to show that for a fixed H and most L it will contain
exactly deg.F / points. Thus, the hyperplanes are exactly the closed subvarieties H of Pn
such that
(a) dim.H / D n

1;

(b) #.H \ L/ D 1 for all lines L not contained in H .
These are geometric conditions, and so any automorphism of Pn must map hyperplanes to
hyperplanes. But on an open subset of Pn , such an automorphism takes the form
.b0 W : : : W bn / 7! .F0 .b0 ; : : : ; bn / W : : : W Fn .b0 ; : : : ; bn //
where the Fi are homogeneous of the same degree d (see 6.20). Such a map will take
hyperplanes to hyperplanes if only if d D 1.

T HE S EGRE MAP
6.26. This is the mapping
..a0 W : : : W am /; .b0 W : : : W bn // 7! ..: : : W ai bj W : : ://W Pm  Pn ! PmnCmCn :
The index set for PmnCmCn is f.i; j / j 0  i  m; 0  j  ng. Note that if P
we interpret
the tuples
on
the
left
as
being
the
coefficients
of
two
linear
forms
L
D
ai Xi and
1
P
L2 D bj Yj , then the image of the pair is the set of coefficients of the homogeneous
form of degree 2, L1 L2 . From this observation, it is obvious that the map is defined on
the whole of Pm  Pn .L1 ¤ 0 ¤ L2 ) L1 L2 ¤ 0/ and is injective. On any subset of the
form Ui  Uj it is defined by polynomials, and so it is regular. Again one can show that it
is an isomorphism onto its image, which is the closed subset of PmnCmCn defined by the
equations
wij wkl wi l wkj D 0
– see Shafarevich 1994, I 5.1. For example, the map
..a0 W a1 /; .b0 W b1 // 7! .a0 b0 W a0 b1 W a1 b0 W a1 b1 /W P1  P1 ! P3
has image the hypersurface
HW
4 This

W Z D X Y:

is related to the fundamental theorem of projective geometry — see E. Artin, Geometric Algebra,
Interscience, 1957, Theorem 2.26.

i. Some classical maps of projective varieties

145

The map
.w W x W y W z/ 7! ..w W y/; .w W x//
is an inverse on the set where it is defined. [Incidentally, P1  P1 is not isomorphic to
P2 , because in the first variety there are closed curves, e.g., two vertical lines, that don’t
intersect.]
If V and W are closed subvarieties of Pm and Pn , then the Segre map sends V  W
isomorphically onto a closed subvariety of PmnCmCn . Thus products of projective varieties
are projective.
The product P1  Pn contains many disjoint copies of Pn as closed subvarieties. Therefore a finite disjoint union of copies of Pn is projective, which shows that a finite disjoint
union of projective varieties is projective.
There is an explicit description of the topology on Pm  Pn W the closed sets are the sets
of common solutions of families of equations
F .X0 ; : : : ; Xm I Y0 ; : : : ; Yn / D 0
with F separately homogeneous in the X and in the Y .

P ROJECTIONS WITH GIVEN CENTRE
6.27. Let L1 ; : : : ; Ln d be linearly independent linear forms in n C 1 variables. Their
zero set E in k nC1 has dimension d C 1, and so their zero set in Pn is a d -dimensional
linear space. Define W Pn E ! Pn d 1 by .a/ D .L1 .a/ W : : : W Ln d .a//; such a map
is called a projection with centre E. If V is a closed subvariety disjoint from E, then 
defines a regular map V ! Pn d 1 . More generally, if F1 ; : : : ; Fr are homogeneous forms
of the same degree, and Z D V .F1 ; : : : ; Fr /, then a 7! .F1 .a/ W : : : W Fr .a// is a morphism
Pn Z ! Pr 1 .
By carefully choosing the centre E, it is possible to linearly project any smooth curve in
n
P isomorphically onto a curve in P3 , and nonisomorphically (but bijectively on an open
subset) onto a curve in P2 with only nodes as singularities.5 For example, suppose we have
a nonsingular curve C in P3 . To project to P2 we need three linear forms L0 , L1 , L2 and
the centre of the projection is the point P0 where all forms are zero. We can think of the
map as projecting from the centre P0 onto some (projective) plane by sending the point P
to the point where P0 P intersects the plane. To project C to a curve with only ordinary
nodes as singularities, one needs to choose P0 so that it doesn’t lie on any tangent to C , any
trisecant (line crossing the curve in 3 points), or any chord at whose extremities the tangents
are coplanar. See for example Samuel, P., Lectures on Old and New Results on Algebraic
Curves, Tata Notes, 1966.
Projecting a nonsingular variety in Pn to a lower dimensional projective space usually
introduces singularities. Hironaka proved that every singular variety arises in this way in
characteristic zero. See Chapter 8.

A PPLICATION
P ROPOSITION 6.28. Every finite set S of points of a quasiprojective variety V is contained
in an open affine subset of V .
nonsingular curve of degree d in P2 has genus
genus g can’t be realized as a nonsingular curve in P2 .
5A

.d 1/.d 2/
.
2

Thus, if g is not of this form, a curve of

146

6. P ROJECTIVE VARIETIES

P ROOF. Regard V as a subvariety of Pn , let VN be the closure of V in Pn , and let Z D VN X V .
Because S \ Z D ;, for each P 2 S there exists a homogeneous polynomial FP 2 I.Z/
such that FP .P / ¤ 0. We may suppose that the FP have the same degree. An elementary
argument shows that some linear combination F of the FP , P 2 S , is nonzero at each P .
Then F is zero on Z, and so VN \ D.F / is an open affine of V , but F is nonzero at each P ,
and so VN \ D.F / contains S .
2

j

Maps to projective space

Under construction.

k

Projective space without coordinates

Let E be a vector space over k of dimension n. The set P.E/ of lines through zero in E has
a natural structure of an algebraic variety: the choice of a basis for E defines a bijection
P.E/ ! Pn , and the inherited structure of an algebraic variety on P.E/ is independent of
the choice of the basis (because the bijections defined by two different bases differ by an
automorphism of Pn ). Note that in contrast to Pn , which has n C 1 distinguished hyperplanes,
namely, X0 D 0; : : : ; Xn D 0, no hyperplane in P.E/ is distinguished.

l

The functor defined by projective space

Let R be a k-algebra. A submodule M of an R-module N is said to be a direct summand of
N if there exists another submodule M 0 of M (a complement of M ) such that N D M ˚ M 0 .
Let M be a direct summand of a finitely generated projective R-module N . Then M is also
finitely generated and projective, and so Mm is a free Rm -module of finite rank for every
maximal ideal m in R. If Mm is of constant rank r, then we say that M has rank r. See CA
12.
Let
P n .R/ D fdirect summands of rank 1 of RnC1 g.
Then P n is a functor from k-algebras to sets. When K is a field, every K-subspace of K nC1
is a direct summand, and so Pn .K/ consists of the lines through the origin in K nC1 .
Let Hi be the hyperplane Xi D 0 in k nC1 , and let
Pi .R/ D fL 2 P n .R/ j L ˚ HiR D RnC1 g:
Let L 2 Pi .R/; then
ei D ` C

X

aj ej .

j ¤i

Now
L 7! .aj /j ¤i W Pi .R/ ! Ui .R/ ' Rn

R/ 0i n Y Pn . When d D 0 or n. 9.s0 . Changing the basis for S multiplies A. Let L be a line in P3 . G2 . and so from now on we assume that 0 < d < n. so that . Fix an S0 2 Gd .m.30.S / on the left by an invertible d  d matrix. E XAMPLE 6. Thus.E/ ! closed subset of P n d 1   n 1 P d is injective. The map realizes Gd . R EMARK 6.R/ \ Uj .41. These combine to give an isomorphism P n . P ROPOSITION 6.R/: 0i.34) we shall give a “coordinate-free description” of the map. and let S 2 Gd . The affine cone over a line in P3 is a two-dimensional subspace of k 4 .S / in P n d 1 . Thus.S /W Gd . Gd . yi yj ˇ depends only on L.R/ \ Pj . pij D ˇ .p01 W p02 W p03 W p12 W p13 W p23 / 2 P .E/ has a single element.R/ Pi .   The map S 7! P . The choice of a basis for S then determines a d  n matrix A.j n Y Ui . and let x D . m Grassmann varieties Let E be a vector space over k of dimension n.29.R/ 0i. with image a . see (9. The maps P defined by different bases of E differ by an n 1 automorphism of P d .R/ ! Pn .j n More generally.31.S / is determined up to multiplication by a nonzero constant. .y0 W y1 W y2 W y3 / be distinct points on L. : : : .L/ D .R/ 0i n Ui . and let Gd .k 4 / onto the quadric ˘ W X01 X23 X02 X13 C X03 X12 D 0 in P5 . and let Gd .S / whose rows are the coordinates of the basis elements.  and so defines a point P . sn are sections of L that generate it. sn // where L is an invertible sheaf on V and s0 . For a direct elementary proof of this.E/ as a projective algebraic variety called the Grassmann variety of d -dimensional subspaces of E. the family of d  d minors of A.E/S 0 .E/ be the set of d -dimensional subspaces of E. The map L 7! P .R/ Pi . : : : . to give a regular map from a variety V to Pn is the same as giving an isomorphism class of pairs .R/: Y Y P n .42) below. and so the statement is independent of the choice of the basis — later (6.k 4 / can be identified with the set of lines in P3 .E/. Then ˇ ˇ xj ˇˇ def ˇ xi 5 ˇ P .L. Grassmann varieties 147 is a bijection.x0 W x1 W x2 W x3 / and y D . We give the proof  below.L/ is a bijection from G2 . Fix a basis for E.V / such that S \ S 0 D f0g. Let S 0 be a subspace of E of complementary dimension n d .E/S 0 be the set of S 2 Gd .

148 6. and so S is the graph of a homomorphism S0 ! S 0 : s 7! s 0 ” . the graph of any homomorphism S0 ! S 0 lies in Gd . In particular.E=S 0 .S0 . if e1 . ed i and S 0 D hed C1 .S0 .E/ is nonsingular (4.S / D . Therefore. and let S0 D he1 . ed of S .V /S 0  Hom. R EMARK 6.72).a0 W : : : W aij W : : : W : : :/ where a0 is the left-most d  d minor of A.32. the tangent space to Gd . The elements of d E are called (exterior) exterior algebra of E is a finite-dimensional graded algebra over k V d -vectors:The V with 0 E D k. B has a natural structure as an irreducible algebraic variety. this shows that Gd . is the minor obtained from the left-most d  d minor by replacing the ith column with the j th column.39).34.35 below). Gd .S1 0 .E// ' Hom. : : : . : : : . en / 7! he1 .E=S 0 . The bijection (26) identifies Gd . S 0/. E/ (28) which the choice of a basis for E identifies with S 7! P . S 0 // defined by the vector space Hom.S0 . : : : .E/ ! P. : : : .S /.Gd .29. there is a well-defined map ^d ^d S 7! S W Gd .V /S 0 .33.S 0 .E/S 0 . n E V has dimension 1. 1 E D E. : :: .Hom.S0 .E/ is also irreducible. 1  i  d . S 0 /  Hom.E/ is a surjective regular map. and so Gd . 6. d S is the one-dimensional V subspace of d E spanned by e1 ^ : : : ^ ed for any basis e1 . The choice of a basis for E identifies 2 B with GLn . e n of E. TS0 . e 2 E.S /. S 0 /: (26) The isomorphism Gd . Thus. S 0 / depends on the choice of S0 — it is the element of Gd .e1 . p.V /S 0 . S 0 / ' Hom. ed iW B ! Gd . In particular.V /S 0  Hom. which is the principal open subset of An where det ¤ 0.S0 . Fix a basis e1 . en form an ordered basis for V .E/ at S0 . d < j  n. For a subspace S of E of dimension d .S 0 / 01 acts simply transitively on Gd . Let B be the set of all bases of E.S 0 / 0 . . S 0 / (cf. P ROJECTIVE VARIETIES E D S0 ˚ S 0 .E=S 0 . and the bijections (26) show that 0 End. Order the coordinates in P n d 1 so that P . Note that the subspace spanned by e1 .S .S /   the group Hom. s 0 / 2 S: Conversely. F IRST PROOF OF P ROPOSITION 6. : : : . The map .s.E/S 0 with the affine variety A. then the dn wedge V V products ei1 ^ : : : ^ eid (i1 <    < id ) form an ordered basis for d E. The decomposition E D S0 ˚ S 0 End. V L V R EMARK 6. and aij .S0 / gives a decomposition End. : : : .E/ D Hom.V /S 0 corresponding to 0 2 Hom. The exterior algebra E D d 0 d E of E is the quotient of the tensor V algebra by the ideal generated by all vectors e ˝ e. en can be recovered from the line through e1 ^ : : : ^ ed as the space of vectors v such that v ^ e1 ^ : : : ^ ed D 0 (cf. en i. : : : . R EMARK 6.S0 / Hom. Thus. E=S0 /: (27) Since the dimension of this space doesn’t depend on the choice of S0 . For any S 2 Gd . the projection S ! S0 given by this decomposition is an isomorphism.

: : : . Therefore.28). ai1 :::id 2 k. It follows that m  d . 2 For a nonzero d -vector w. then the basis element corresponding to the zero row lies in S 0 \ S. L EMMA 6. Write X wD ai1 :::id ei1 ^ : : : ^ eid .E/S 0 . : : :/ 7! .aij / 2 k d. each nonzero term in the sum is of the form ae1 ^ : : : ^ em ^ : : :. the image of P W Gd . P . the map ^d ^d C1 w 7! . en of V . let P ei0 D ei C d <j n aij ej (29) denote the unique element of S projecting to ei .n d / (this is a restatement of (26) in terms of matrices). S ECOND PROOF OF P ROPOSITION 6. of the lines through the pure d -vectors. For S 2 Gd . : : : . E/ .aij /W : : :/ where fk . Conversely.E/ in P. S $ . Now. Let e1 .29. Thus. ed0 is a basis for S.E/S 0 $ k d.n d / .w/  d .aij / gives a one-to-one correspondence Gd . the image of Gd . em .S/ determines . 1  i  d . Clearly. 7 6 If e 2 S 0 \ S is nonzero. P ROOF. and m D d if and only if w D ae1 ^ : : : ^ ed with a ¤ 0. 7 In more detail. we may choose it to be part of the basis for S .S / 2 U0 if and only if S 2 Gd .: : : . for any . An exterior d -vector v is said to be pure (or decomposable) if there exist vectors Vd e1 . if S $ . let Œw denote the line through w.E/S 0 ! U0 is the graph of the regular map . Conversely. then ej ^ w ¤ 0. fk . Grassmann varieties 149   n 1 P d Let U0 be the (“typical”) standard open subset of consisting of the points with nonzero zeroth coordinate.w/. aij . The lemma shows that V Œw 2 Gd .S/ has a row of zeros. Therefore.aij / is a polynomial in the aij whose coefficients are independent of S . the projection S ! S0 is bijective. the ei0 defined by (29) span an S 2 Gd . Then e10 . For each i . : : :/W A d.E/ is defined by the vanishing of the minors of order n d C 1 of this map. : : : .aij /. Let w be a nonzero d -vector and let M.E/ if and only if the linear map v 7! v ^ wW E 7! d C1 E has rank  n d (in which case the rank is n d ). We shall prove the proposition by showing that P W Gd . According to (6. 1i1 <:::<id If there is a nonzero term in this sum in which ej does not occur. if the left-most d  d submatrix is singular.w/ D fv 2 E j v ^ w D 0gI then dimk M.E/S 0 and project to the ei .n d / !   n d. and then the left-most d  d submatrix of A. ed 2 E/ consists V such that v D e1 ^ : : : ^ ed .v 7! v ^ w/W E ! Homk . Thus Gd . which is closed (5. : : : .1 W : : : W aij W : : : W : : : W fk .35.aij /.E/S 0 .aij / and hence also S . em be a basis of M. : : : . and extend it to a basis e1 .n d / 1 A d .m.S / D . then P . we can change the basis for S so that it has a row of zeros.E.E/S 0 ! U0 is injective with closed image. : : : .6 P . with equality if and only if w is pure.34). Moreover.

(31) that are compatible in the sense that ' i jE i C1  ' i C1 mod E i C1 : A SIDE 6.F .E/ ' i W E i ! V =E i . I.150 6. in algebraic geometry.V / is defined by the vanishing of many minors. : : : . then it follows from (6. called a flag variety: The tangent space to Gd . Xn // with F a homogenous polynomial without repeated factors.F0 / is a parabolic subgroup of GL.E/ D P. en for E is adapted to the flag F if it contains a basis e1 . if u is a pure d -vector and u i i i C1 is a pure di C1 -vector.F0 / ! Gd . and most famous. d C1 E// (once a basis has been chosen V for E. Because GL. Thus. and the basis then determines the flag. it acts transitively on Gd .35) that M. P ROJECTIVE VARIETIES Flag varieties The discussion in the last subsection extends easily to chains of subspaces. Let d D . E//: V The condition rank  n d defines a closed subset W of P. Because Gd .E/ acts transitively on the set of bases for E.E/ is projective. di E/ Q realizes Gd .33).uiC1 / if and only if the map v 7! . . the subgroup P . For such a variety. We define the degree of V to be the degree of F . Vd E/ \ W: 8 For example.E/  i P. and so it is a projective variety.E/.E/ as a closed subset8 i Gdi . As in (6. every flag admits such a basis.V /. : : : . : : : . and let Gd .Homk . and the map g 7! gF0 W GL.E/=P . E/ . dr / be a sequence of integers with 0 < d1 <    < dr < n.Homk . 1  i  r.F / stabilizing F is an algebraic subgroup of GL.E/ at the flag F consists of the families of homomorphisms Gd . n Bezout’s theorem Let V be a hypersurface in Pn (that is. The map F 7!.E. this shows that P .V / D .v ^ ui .36.E. : : : .E/. eji for each E i . and so defines an injective regular map ^d ^d C1 P.E/ be the set of flags F W E  E1      Er  0 (30) with E i a subspace of E of dimension di . Gd .V i / Q Q V ! i Gdi .ui /  M.X0 . the condition becomes the vanishing of the minors of order n d C 1 of a linear map E ! d C1 E). The next theorem is one of the oldest.d1 . For a flag F . Clearly. v ^ uiC1 /W V ! has rank  n di (in which case it has rank n ^di C1 V˚ ^di C1 C1 V di ).E/.E/ is an isomorphism of algebraic varieties.! P. is injective and linear. A basis e1 . and Gd .E/ is irreducible. a closed subvariety of dimension n 1). this implies that Gd .

in contradiction with our assumption. xn . If the latter occurs. Clearly sm ¤ 0 ¤ tn . Z/ be the polynomials defining C and D. then they intersect in exactly mn points.X. Z/ and G. Decompose C and D into their irreducible components. Y / is a homogeneous polynomial of degree r not divisible by X . and so there is a point . Let P be an isolated point of C \ D. : : : . We can therefore assume that C and D are themselves irreducible.ai W bi W ci / is a point on C \ D.a W b W c/ is a point on C \ D (so a ¤ 0/. which contradicts the finiteness of C \ D. g/ D m — and so the multiplicity is one. Let R be the resultant of F and G.28) this means that the polynomials F .a. Xn =b D kŒx0 .a W b W c/ 2 C \ D.a.T /. bi . Thus we see in this case. P ROOF.T / is a Y polynomial of degree  mn in T D X . and let ˛1 . for otherwise F and G would have Z as a common factor. Clearly it suffices to prove the theorem for each irreducible component of C and each irreducible component of D. : : : . bi / D 0. we can assume that a ¤ 0 for all points . Thus R Y is a nonzero polynomial of degree mn. g/ is finite-dimensional over k.o. It follows that OP =. that C \ D has precisely mn points.f. and we define the multiplicity of P in C \ D to be dimk . regarded as polynomials in Z. then ab is a root of R . g/ D m. b/ 2 k 2 .X. Thus . There will be an affine neighbourhood U of P and regular functions f and g on U such that C \ U D V . It is a homogeneous polynomial of degree mn in X and Y . or else it is identically zero.f. Y / D X mn R . o Hilbert polynomials (sketch) Recall that for a projective variety V  Pn . then for every .f / and D \ U D V . and write F D s0 Z m C s1 Z m 1 C    C sm .OP =. Suppose first that deg R D mn. Z/ and G. For example. 2 R EMARK 6. b. The above proof has the defect that the notion of multiplicity has been too obviously chosen to make the theorem come out right. Then R. then f and g will form a system of local parameters at P — . According to (7. If C and D have no irreducible component in common. Obviously R. Z/ and G. and clearly rad. provided we take the multiplicity of . if . Hilbert polynomials (sketch) 151 T HEOREM 6.f.f. G D t0 Z n C t1 Z n 1 C    C tn with si and tj polynomials in X and Y of degrees i and j respectively.g/. if C and D cross transversely at P . . F .0. Each such root can be written ˛i D abii . Now suppose that R has degree r < mn. and conversely. ˛mn be the roots of R (some of them may be multiple).ai .X.a W b W c/ to be the multiplicity of ab as a root of R .X. Z/ have a common zero. We know from (2.37. X / where R . The attempt to find good notions of multiplicities in very general situations motivated much of the most interesting work in commutative algebra in the second half of the twentieth century.X. After a change of variables. Y.ai .a.ai . : : : .63) that C \ D is of dimension zero. Y / D X mn r P . g//. Let F . and so is finite. b. khom ŒV  D kŒX0 . bi .38. the maximal ideal in OP . counted with appropriate multiplicities. We can regard f and g as elements of the local ring OP . Y.0 W 1 W c/ in C \ D. Let C and D be curves in P2 of degrees m and n respectively. Y / where P . Z/ have a common root ci .X. and R. It is possible to show that the theorem holds with the following more natural definition of multiplicity. 1/ D 0. Write R.

Atiyah and Macdonald 1969. : : : . m/ D dimk kŒV m for all m sufficiently large. Despite the notation. T / D T Cn n  D . We observed that b is graded.a0 W a1 / 7! .a0d W a0d 1 a1 W : : : W a1d /W P1 ! Pd : Then khom ŒV m can be identified with the set of homogeneous polynomials of degree m  d in two variables (look at the map A2 ! Ad C1 given by the same equations).V. Macaulay knows how to compute Hilbert polynomials. R EFERENCES : Hartshorne 1977. T HEOREM 6.152 6. Lecture 13. m0 where khom ŒV m is the subspace generated by the monomials in the xi of degree m.V.V. khom ŒV  D kŒX0 . T / D d T C 1: Thus V has dimension 1 (which we certainly knew) and degree d . T / such that P . .42. P ROJECTIVE VARIETIES where b D I. Chapter 11. Clearly khom ŒV m is a finite-dimensional k-vector space. Omitted.Pn .40.V /. T HEOREM 6. which is a space of dimension d m C 1. and so n nŠ P . There is a unique polynomial P . T / in the theorem is called the Hilbert polynomial of V . dim khom ŒV m D mCn D .mC1/ . Let V be a projective variety of dimension d and degree ı. dŠ P ROOF.41. T / D ı d T C terms of lower degree. E XAMPLE 6.mCn/.V. then P . Harris 1992. P ROOF.T C 1/ : nŠ The polynomial P .T C n/    .7. and therefore khom ŒV  is a graded ring: M khom ŒV  D khom ŒV m . Omitted.V. 2 The degree of a projective variety is the number of points in the intersection of the variety and of a general linear variety of complementary dimension (see later). Xn . 2 E XAMPLE 6. it depends not just on V but also on its embedding in projective space. For V D Pn . and (see the footnote on page 141). Let V be the image of the Veronese map . I.39. and so P .

Then V aff . 2 P ROPOSITION 6.p. and every irreducible component of it has codim. : : : . then V \ V . Since the dimension of a variety is equal to the dimension of any dense open affine subset. 0/). then V \ V . : : : fr / is nonempty..47. V 0 \ W 0 has dimension  1.f1 .W /. the only part that doesn’t follow immediately from (3. 2 P ROPOSITION 6.b/. P ROOF. : : : .    . Xn . Follows from the affine case.W /  n.. and consider the affine cones V 0 D V . Let V be a closed subvariety of Pn of dimension r < n. Let V D V .a/. Let Z be a irreducible closed subvariety of V . Xn . Clearly V aff .a/ has dimension  2. Let V aff . and … a. nonconstant. I. and let Z be an irreducible component of V \ V . the affine cone over V /. then there is a linear projective variety E of dimension n r 1 (that is. Use the same argument as in the proof (3. 2 C OROLLARY 6.0. Then dim. P ROOF. then V \ W ¤ .45.W 0 / D dim. The rest of the statement follows from the affine case..V /  r.. then V \W ¤ ..Z/ D . i.W / C 1  n C 2: As V 0 \ W 0 ¤ . Let V and W be closed subvarieties of Pn . This implies that the polynomials in a have a zero in common with f other than the origin. P ROOF. Dimensions p 153 Dimensions The results for affine varieties extend to projective varieties with one important simplification: if V and W are projective varieties of dimensions r and s in Pn and r C s  n. and so V . : : : fr /. : : : .. and so it contains a point other than the origin.b/ over them.V /Ccodim.f / is nonempty.43. 2 P ROPOSITION 6. 2 C OROLLARY 6.V / C 1 C dim.46. Therefore V \ W ¤ .48.. : : : . fr in kŒX0 . then there exist homogeneous polynomials f1 . and if dim.f1 .a/ be the zero set of a in AnC1 (that is.a/ \ V . Write V D V . if dim. if codim.f / is nonempty and of pure codimension 1.f / has dimension  1.a/ and W 0 D V . P ROOF.f1 .a/ \ V aff . and so it has codimension 1 in V aff .Z/  r.a/ \ V aff . : : : .a/  Pn be a projective variety of dimension  1. T HEOREM 6. Induction on r. and so V aff . fr be homogeneous nonconstant elements of kŒX0 .Z/ D r. as before. E is defined by r C 1 independent linear forms) such that E \ V D . fr /.Z/ codim. P ROOF. Let f1 . and let f 2 kŒX0 .44.e.a/ and W D V . .V 0 / C dim.f / for some f homogeneous element of kŒX0 .f / ¤ . : : : .V / C dim.42) is the fact that V \ V . : : : . Xn  be homogeneous. Every pure closed subvariety Z of Pn of codimension one is principal.47). Xn  such that Z is an irreducible component of V \ V .f / is nonempty (it contains . Then codim.

with coefficients not all zero. then V is a finite set. they all have dimension  r.a// defines a map V ! Pr . 2 L EMMA 6.a//: If m < n and the Fi are nonconstant.1 and II 6. The problem becomes that of showing V is not a finite union of proper subspaces Ei_ . If E is defined by the linear forms L0 . : : : . if m < n.F0 . then ˛ is constant. Take E D E 0 \ H . and there exist polynomials F0 .a0 W : : : W an / D . P ROOF. 2 q Products It is useful to have an explicit description of the topology on some product varieties. can not be identically zero on k n (Exercise 1-1). Hartshorne II 7.a//: As ˛ ı  factors through Pn .2).50.a/ W    W Lr . All line bundles on An X foriging are trivial (see. Er be a finite set of nonzero subspaces of W . Pass to the dual space V of W . then (6. : : : . P ROJECTIVE VARIETIES P ROOF. from which it follows that all regular maps AnC1 X foriging ! Pm are given by a family of homogeneous polynomials. Suppose r > 0.43) shows they have a common zero and so ˛ is not defined on all of Pn . We have to show that Lj is not identically zero on V . C OROLLARY 6. Let W AnC1 foriging ! Pn be the map .154 6. : : : . H contains none of the irreducible components Vi of V . But this follows from the statement that a polynomial in n variables. and let E1 . and so it can be regarded as a projective version of the Noether normalization theorem. : : : . Lr then the projection a 7! . The lemma below shows that there is a hyperplane H containing none of the nonzero Ei . Replace each Ei_ by a hyperplane HiQcontaining it. . Fm 2 kŒX0 . there is an linear subvariety E 0 not meeting V \ H . Then Hi is defined by a nonzero linear form Li . Then there is a hyperplane H in W containing none of the Ei . In general. : : : . it is possible to prove the following result. the Fi must be homogeneous of the same degree.49.a/ W : : : W Fm . Then ˛ ı  is regular. an / 7! . By assumption.a0 . Induction on r. Note that ˛.a0 W : : : W an /.a0 . and let V1 .L0 . and so each Vi \ H is a pure variety of dimension  r 1 (or is empty). By induction. Vs be the irreducible components of V . : : : . Let W be a vector space of dimension d over an infinite field k. The intersection Ei of all the linear projective varieties containing Vi is the smallest such variety.48.F0 .a/ W : : : W Fm . Xn  such that ˛ ı  is the map . a regular map from a variety V to Pn corresponds to a line bundle on V and a set of global sections of the line bundle. Assuming this. If r D 0. for example. consequently. P ROOF. We shall see later that this map is finite. Let ˛W Pn ! Pm be regular. : : : . Hence the Fi must be constant. and the lemma below shows that there is a hyperplane in k nC1 not meeting V . 2 Let V and E be as in Proposition 6. an / 7! .

Xm I Y0 . : : : .X0 . then the closure of its inverse image in V  AnC1 is also irreducible. It makes sense to speak of the zero set V .b/ is closed. : : : . Xn . and let B D AŒX0 . and so I V . (c) If V is irreducible. each of which is separately homogeneous in the X and Y . i 2 I . then V D V . and so I aff . X1 . (a) For each graded ideal b  B.b/  V  Pn of such an ideal. aB is graded. is zero on V  foriging if and only if its constant term is zero. a 2 A. For example. Suppose we have a collection of polynomials Fi . For any ideal a  A. (a) In the case that A D k. For (c). Xm I Y0 . and every closed subset of Pm  Pn arises in this way from a (finite) set of polynomials. with each Fi homogeneous in the Y . Yn /.b/ is closed. 2 .b/ is prime. : : : . But X ai0 :::in X0i0 : : : Xnin . We look more closely at the topology on V  Pn in terms of ideals. if XiN 2 b for all i.b/ \ Ui is closed for all i. Xn /. P ROOF. : : : . cover Pn with the standard open affines Ui and show that V . then obviously V . Note that B D A ˝k kŒX0 . and similar arguments apply in the present more general situation. The topology on V  Pn Let V be an irreducible affine algebraic variety. Xn /: Thus. i 2 I.b/ D . Let A D kŒV . to see that V .b/ is empty. The topology on Am  Pn The closed subsets of Am  Pn are exactly those defined by sets of equations Fi .v.b/ for some graded prime ideal b.1) and (6. : : : .V  foriging/ D . ai0 :::in 2 kŒV . Conversely. and V .b/ is empty if and only if the cone V aff .b/  V  AnC1 defined by b is contained in V  foriging. : : : . : : : . : : : . Xn . Xn . Xn /. ) rad.b/  . Then the equations Fi .q. Yn / D 0. Yn / D 0.X0 .X0 .2).a/W V  AnC1 ! k: P The ring B has an obvious grading — a monomial aX0i0 : : : Xnin .X1 . Xm I Y0 . L EMMA 6.aB/ D V .b/ D . note that if V . i 2 I. and every closed subset of V  Pn is of this form. : : : .X0 . : : : .a/  Pn . f ˝ g is the function . the Nullstellensatz shows that V . : : : . we proved this in (6. a/ 7! f .v/  g.b/ is empty if and only if rad.X0 . Products 155 The topology on Pm  Pn . define a closed subset of Pm  Pn .b/ is irreducible. The set V . has degree ij — and so we have the notion of a graded ideal b  B. (b) The set V . the set V . : : : .51. and so we can view it as the ring of regular functions on V  AnC1 : for f 2 A and g 2 kŒX0 .

1/ D 1: Show that ˛ is the identity map. Z. P ROJECTIVE VARIETIES Exercises 6-1. Verify that Y 2 Z D X 3 C aXZ 2 C bZ 3 is nonsingular if X 3 C aX C b has no repeated root.0W 1/. P1 . Write 0. even if a is radical. Pr be points in Pn .1/ D P2 : (c) Deduce that Aut..V / D . and it is the tangent at that point. : : : . and .@F=@Y /P Y C . and find the tangent line at the point at infinity on the curve. 6-8.e.Z X 3 / /. Show that there is a hyperplane H in Pn passing through P0 but not passing through any of P1 .156 6.k/. Show that the functor R P n . 6-2.@F=@X /P X C . and @F=@Z are all zero at P . (a) Let ˛ be an automorphism of P1 such that ˛. Show that the image of the Segre map Pm Pn ! PmnCmCn (see 6.1/ D P1 . it does not follow that F1 .R/ D fdirect summands of rank 1 of RnC1 g satisfies the criterion (5. Z X 3 /  A3 . 6-7.0/ D 0.0/ D P0 . if F1 . Is the subset f. : : : .X.V /  kŒW. but ZW X Y … . b ¤ 0g [ f. (b) ZW X Y 2 I.Y X 2..1W 1/. Let V D V .1 W 0 W 0/g of P2 locally closed? 6-6. Prove X 3 /.P1 / ' PGL2 . Show that a point P on a projective curve F . 1.26) is not contained in any hyperplane of PmnCmCn . or (b) L intersects C in exactly 1 point. show that the tangent line at P has the (homogeneous) equation . (Thus. If P is nonsingular. Y.) . Show that either (a) L intersects C in exactly 2 points. ˛. X 2. Let P0 . Pr . Let L be a line in P2 and let C be a nonsingular conic in P2 (i. . : : : .1/ D 1. Y. .a W b W c/ j a ¤ 0. Fr generate a . (b) Let P0 . ˛.Y X 2 / . a curve in P2 defined by a homogeneous polynomial of degree 2). X. Show that there exists an ˛ 2 PGL2 . 1 for the points . Z 6-3. 6-5.Y (a) I.1W 0/ on P1 . (This gives an alternative definition of Pn . @F=@Y . ˛.71) to arise from an algebraic prevariety.@F=@Z/P Z D 0. P2 be distinct points on P1 . : : : . Fr generate a.k/ such that ˛. ˛. Z/ D 0 is singular if and only if @F=@X.) 6-4.

its projection becomes the whole of A1 . N.1. i.3. for example. An algebraic variety V is complete if for all algebraic varieties T .xW 0/.xW z/.2. I. and so the restriction of the closed map qW V  T ! T to Z  T is also closed. the variety V W X Y D 1 is closed in A2 but its image in A1 omits the origin. Let Z be a closed subvariety of a complete variety V .2. Corollary 1 to Theorem 1). Closed subvarieties of complete varieties are complete. a Hausdorff space V is compact if and only if. for all topological spaces T .22) that projective varieties are complete. and so must be its closure in P1  A1 . when we replace V with its closure in P1  A1 .. Properties 7. y/ 2 P1  A1 j xy D z 2 g contains V as an open dense subset. Z  T is closed in V  T . Note that a complete variety is required to be separated — we really mean it to be a variety and not a prevariety. Moreover.C HAPTER Complete Varieties Complete varieties are the analogues in the category of algebraic varieties of compact topological spaces in the category of Hausdorff topological spaces. the projection map qW V  T ! T is closed. 0/ of VN maps to 0. the projection map qW V  T ! T is closed. maps closed sets to closed sets (see Bourbaki. E XAMPLE 7.. 10.x.e. However. y/ 7! yW A1  A1 ! A1 This is not closed. and hence is closed if the image space is Hausdorff.. 157 7 . To see this. a Definition and basic properties Definition D EFINITION 7. The point . Recall that the image of a compact space under a continuous map is compact. note that VN D f. For any variety T . General Topology. Consider the projection map .. We shall see (7.

7. then '.28). (Alternatively. Vn be complete varieties. If W is complete.Zi / is closed in T . In order to show that a variety V is complete. C OMPLETE VARIETIES 7.5. and its image in T equals that of Z. The only regular functions on a complete connected variety are the constant functions.10. and write ' D . .'1 . In particular. In particular. The projection is the composite of the projections Q i  Vi  T ! T V1      Vn  T ! V2      Vn  T !    ! Vn  T ! T.8). Let V1 . Conversely. A regular map 'W V ! W from a complete connected variety to an affine variety has image equal to a point. it must either be a single point (and ' is constant) or P1 (and ' is onto). it suffices to check that qW V  T ! T is a closed mapping when T is affine (or even an affine space An ). Therefore '. 7. Each 'i is a regular function on V . A regular map V ! P1 from a complete connected variety V is either constant or surjective. 7. and hence is constant.w. The only proper closed subsets of P1 are the finite sets. Products of complete varieties are complete. 7.8.W / is the projection of ' into V . Therefore. and let T be a variety. and such a set is connected if and only if it consists of a single point. every complete subvariety of a variety is closed. suppose that the irreducible components Vi of a variety V are complete. and the second follows from the first applied to the identity map.4. to which we can apply (7.Vi  T / is closed in Vi  T . and (7. Each irreducible component is closed.Zi / is also closed. 7.W / is a complete closed subvariety of V .w//g  W  V be the graph of '. Then Z 0 is closed. and hence complete if the variety is complete (7.12. The second statement follows from the first applied to the identity map. 'n / where 'i is the composite of ' with the coordinate function xi W An ! A1 . If 'W W ! V is surjective and W is complete.6. apply 5.158 7.) This proves the first statement. Because '. A regular function on a variety V is a regular map f W V ! A1  P1 . Let Z 0 be the inverse image of Z in W  T .11. then V is complete.3). It is a closed subset of W  V (because V is a variety. Embed W as a closed subvariety of An . every complete connected affine variety is a point. q.Z/ D q. all of which are closed. and so q.W / is closed. and '.6) shows that it is complete.7. see 5. A variety is complete if and only if its irreducible components are complete. Let ' D f. If Z is def closed in V  T . 7. '. then S Zi D Z \ . Let 'W W ! V be a regular map of varieties. 7. Let T be a variety. : : : .V / is connected and closed. and let Z be a closed subset of V  T .9. : : : .

But T can be realized as a closed subvariety of An .18. The statement that a complete variety V is closed in every larger variety W perhaps explains the name: if V is complete. P ROOF. Let V3 ! V2 ! V1 be proper maps.19. this says that all Cauchy sequences have limits in V and that the limits are unique. then the fibre ' 1 .14. if for all regular maps T ! S .15.13. For example.b.Zi / is closed in Ti for all i . Proper maps 159 S Every variety T can be written as a finite union of open affine subvarieties T D Ti .17. ' 7. this shows that q. and W is a closed subvariety of V . A regular map 'W V ! S of varieties is said to be proper if it is “universally closed”. If V ! S is proper and S is complete. Here is another criterion: a variety V is complete if and only if every regular map C X fP g ! V extends uniquely to a regular map C ! V . .12. Remarks 7. q. that is. and hence so also is the outer square. If Z def is closed in V  T . a variety V is complete if and only if the map V ! fpointg is proper. and dim V D dim W . In particular. it is clear that the base change of a proper map is proper.V2 V1 T / ' V3 V1 T closed V2 V2 V1 T closed V1 T: Both smaller squares are cartesian. Consider the diagram V3 V3 V2 . and let T be a variety. From its very definition.V  Ti / is closed in V  Ti . 7. then V is complete. W is connected. As q. The statement is now obvious from the fact that a composite of closed maps is closed. P ROPOSITION 7.Zi / D q. A composite of proper maps is proper. the base change ' 0 W V S T ! T of ' is closed.Z/ is closed. Therefore. then V D W . Intuitively. 2 C OROLLARY 7. b Proper maps D EFINITION 7. Contrast An  Pn . then W ! S is proper.16. here P is a nonsingular point on a curve C . If 'W V ! S is proper. This shows that it suffices to check that V  T ! T is closed for all affine varieties T . 7.Z/ \ Ti . and then V  T ! T is closed if V  An ! An is closed. (b) if 'W V ! S is proper. then V  S ! S is proper. (a) if V is complete. 7.P / over a point P of S is complete. then Zi D Z \ .

. After (7. BN  mB C b H) BN D mBN C BN \ b: .51b). and V S Z ' ' 1 . This allows us to assume that p. P ROOF. then Z is contained in W 0  Pn . tn / 2 An is the subvariety of P1 defined by the polynomial f . and so we can assume it to be nonempty.20.mB C b/ D V . P ROOF. and let Z be a complete subvariety of S. tn . 2 C OROLLARY 7. If Z is empty.X0 . Tn . and so mB C b contains the set BN of homogeneous polynomials of degree N . Special case of the proposition. Write p for the projection W  Pn ! W . then mB C b  . and we now have to show that p. : : : . : : : . The fibre over a point . 2 E XAMPLE 7. then p.Z/ is dense in W .mB C b/ D .Z/ is contained in some closed subvariety W 0 of W .15). : : : .b/ D . ai 2 k: Assume that not all ai are zero.160 7. But if V . Because mB and b are graded ideals. Let w 2 W : we shall show that if w … p.11).mB/ \ V . Xn  (6. the image of the cone V aff . Y / D a0 X m C a1 X m 1 Y C    C am Y m . X. : : : . X. and V . and so (see 3. it suffices to prove the Theorem for projective space Pn itself. The points that “disappeared off to infinity” when P1 was taken to be A1 (see p. The projection map An  P1 ! An defines a regular map H ! An .21. 7.b/ under the projection W  AnC1 ! W is also dense in W .Zi / is closed.Z/ is dense in W .. A projective variety is complete.22. i.T1 . Then Z is a finite union of irreducible closed subsets Zi of W  Pn . this is true.w  Pn / \ V . which is a contradiction.e. Xn /N for some N (by 6.w. Thus we may assume that Z is irreducible. thus we have to prove that the projection map Pn  W ! W is a closed mapping in the case that W is an irreducible affine variety (7. which is proper (7. Let f 2 kŒT1 .34a) the map A ! B=b is injective.22.51). Because p. c Projective varieties are complete The reader may skip this section since the main theorem is given a more explicit proof in Theorem 7. Let m  A be the maximal ideal corresponding to w.Z/.b/ with b a graded prime ideal in B D AŒX0 . We have to show that Z closed in W  Pn implies that p.t1 . and it suffices to show that each p.3). Then mB C b is a graded ideal. Then V S Z ! Z is proper.Z/. and so w will be in the image of Z unless V . and let H be the subvariety of An  P1 defined by f . : : : . and we can replace W with W 0 . Tn .Z/ D W . C OMPLETE VARIETIES P ROOF. if there does not exist a P 2 Pn such that . and hence that Z D V . Y  be homogeneous of degree m in X and Y . X. : : : .Z/ closed in W . P / 2 Z.Z/ is empty.49) have literally become the point at infinity on P1 .. T HEOREM 7. The inverse image of a complete variety under a proper map is complete.31 below. If p.t1 . Y / D 0.mB C b/ ¤ . Then this is a homogeneous of degree m and so the fibre always has m points counting multiplicities.b/. Let 'W V ! S be proper.

proved by Kleiman1 . : : : . bi 2 B. and so XiN 2 b for all i . regarded as an A-module.a1 . am .1 C m/M D 0 for some m 2 m.3) shows that . The argument in the proof of Nakayama’s lemma (1. .1 C m/M D 0 implies that M D 0. an m-tuple . we find P P that fN D gN C hN .a1 . b0 . Ample subvarieties of algebraic varieties. : : : .24. Steven L. : : : . of Math. and so the equation . Xm I Y0 . In 1 Kleiman. Xm I Y0 . and go back to find the solutions of the original system. Xm / D 0 with the following property. (2) 84 1966 293–344. Together these show f 2 mBN C BN \ b. : : : . Ann. : : : . which contradicts the assumption that Z D V . Let M D BN =BN \ b. bn / such that .X1 . 1970. : : : . 7../ : ˆ : Rs .X1 . I 9 p45. Toward a numerical theory of ampleness. Every complete curve is projective. The displayed equation says that M D mM .23. Elimination theory 161 In detail: the first inclusion says that an f 2 BN can be written f D g C h with g 2 mB and h 2 b. Because A ! B=b is injective. Lecture Notes in Mathematics. : : : .26. Hironaka). 156 Springer.25. 7.X1 .b/ is nonempty.b0 . which is an integral domain. : : : . On equating homogeneous components. bn / is a solution of (*). There exist nonsingular complete three-dimensional varieties that are not projective (Nagata.d. Vol. Every nonsingular complete surface is projective (Zariski). Robin. we then find the solutions of the reduced system. mi 2 m. : : : . the image of 1 C m in B=b is nonzero. then gN D mi biN . see 6. 2 Remarks 7. if g D mi bi . But M D BN =BN \ b  B=b. Yn / D 0 such that each Pi is homogeneous in the Y . Xm / D 0 :: . there exists a system of polynomial equations 8 ˆ < R1 . Elimination theory does this more systematically. : : : . A nonsingular complete irreducible variety V is projective if and only if every finite set of points of V is contained in an open affine subset of V (Conjecture of Chevalley. Moreover: fN D f . 7.22 for the necessity). Yn / D 0 :: . but there exist singular complete surfaces that are not projective (Nagata)./ : ˆ : Pr . we first use some of the equations to eliminate some of the variables. Hence BN  b. Note that the fact that Pn is complete has the following explicit restatement: for each system of polynomial equations 8 ˆ < P1 . See also. am / is a solution of (**) if and only if there exists a nonzero n-tuple .X1 . d Elimination theory When given a system of polynomial equations to solve. and hN 2 b because b is homogeneous. Hartshorne.

or (b) the two polynomials have a common root. The purpose of elimination theory is to provide an algorithm for doing this. Yn / have a common zero if and only if Rj . am / D 0 for all j . Q.m C n/. gives no indication of how to construct (**) from (*). : : : . so that there exist polynomials P1 and Q1 of degrees m 1 and n 1 respectively such that P . which implies that Res. Q/ D 0 if and only if (a) both s0 and t0 are zero.162 7.27. while short and elegant. we find that the coefficients of P1 and Q1 are the solutions of a system of m C n linear equations in m C n unknowns.P. X. The resultant of P and Q is defined to be the determinant ˇ ˇ ˇ s0 s1 : : : sm ˇ ˇ ˇ n rows ˇ ˇ s : : : s 0 m ˇ ˇ ˇ ::: : : : ˇˇ ˇ ˇ t0 t1 : : : tn ˇ ˇ ˇ ˇ ˇ t0 : : : tn ˇ ˇ m rows ˇ ::: ::: ˇ There are n rows of s’s and m rows of t ’s. : : : . P ROOF. 1 in (32) to zero.X ˛/Q1 .X /.X /P1 .X / Q. all blank spaces are to be filled with zeros. The resultant Res.X / D . Q/ D 0 because the first column is zero.a1 . C OMPLETE VARIETIES other words.X /Q1 .P. am I Y0 . then Res. The matrix of coefficients of the system is the transpose of the matrix 0 1 s0 s1 : : : sm B C s0 : : : sm B C B C : : : : : : B C B t0 t1 : : : tn C B C @ A t0 : : : tn ::: ::: The existence of the solution shows that this matrix has determinant zero.X ˛/P1 . P ROPOSITION 7. Unfortunately. so that the matrix is . : : : .X / D . : : : . the proof we gave of the completeness of Pn . Q/ D 0. If (a) holds. we find that P . The resultant is a polynomial in the coefficients of P and Q.a1 .X /: Using these equalities.X / D 0: (32) On equating the coefficients of X mCn 1 .P. Suppose that ˛ is a common root of P and Q.m C n/  . the polynomials Pi . . The polynomials Rj are said to have been obtained from the polynomials Pi by elimination of the variables Yi . Elimination theory: special case Let P D s0 X m C s1 X m 1 C    C sm and Q D t0 X n C t1 X n 1 C    C tn be polynomials.

Xn .P. Throughout. 3-4.T1 .) Maple can find the resultant of two polynomials in one variable: for example.x C b/5 .22 shows only that the Rj exist. not only have we shown that the projection of V .22. we give a proof of Theorem 7. homogeneous in the Xi . P ROOF. Y /.. (7. Y / D s0 X m C s1 X m 1 Y C    C sm Y m . For any graded ideal a in kŒX0 . exactly one of the following statements is true: 2 Cartier. Q/ are parametrized by the space AmC1  AnC1 D AmCnC2 . Q/ in AmCnC2  P1 .P. The pairs of polynomials .a W 1/ with a a root of P . : : : .28. T HEOREM 7. Q/ D 0 but neither s0 nor t0 is zero.P.X. A simple proof of the main theorem of elimination theory in algebraic geometry. : : : . Because the above matrix has determinant zero. The zeros of P . we can solve the linear equations to find polynomials P1 and Q1 satisfying (32). Xn / have a common zero if and only if Rj . following Cartier and Tate 19782 .. (b) .X ˛/. Macaulay doesn’t seem to know how to do more. (Theorem 7. J. : : : . which is a more explicit than that given above. Explanation: the polynomials have a common root if and only if a D b.29. Math. . cancel X ˛ from the left hand side of (32). Q/ D 0. exactly as in the nonhomogeneous case.28) is a restatement of (7. Elimination theory 163 Conversely. P. Q.27).T1 . Thus. Since a similar statement is true for Q. and this can happen in 25 ways.d. Elimination theory: general case In this subsection. no. We define the resultant of two homogeneous polynomials P . Xn /. : : : . Q/ D 0 if and only if P and Q have a common zero in P1 . P ROPOSITION 7. am I X0 .P. this argument eventually leads to a root of P that is not a root of P1 . A root ˛ of P must be also be a root of P1 or of Q.. : : : . k is a field (not necessarily algebraically closed) and K is an algebraically closed field containing k. 311–317. Enseign. suppose that Res. : : : . a C b/25 .1 W 0/ in the case that s0 D 0. .X. (2) 24 (1978). Y / D t0 X n C    C tn Y n . Elimination theory does this in general.a1 . x/” gives the answer . elimination theory gives an algorithm for finding polynomials Rj . and consider a root ˇ of P1 =. but we have given an algorithm for passing from the polynomials defining the closed set to those defining its projection. 2 Now regard the coefficients of P and Q as indeterminates.x C a/5 . Consider the closed subset V .X.X. Q/ is closed. As deg P1 < deg P .P. an / D 0 for all j . : : : . Tm I X0 . Y / in P1 are of the form: (a) .P. and so must be a root of Q.a1 . Tn / such that the Pi .X. If the former. Tate. The proposition shows that its projection on AmCnC2 is the set defined by Res. entering “resultant. 1/. The resultant Res. Given a family of polynomials Pi . 2 The proposition can be restated in projective terms.

2 L T HEOREM 7. Following Cartier and Tate. : : : .30) is a restatement of (7. Let R D d 2N Rd be a graded k-algebra such that R0 D k.32 We shall prove (7. Xn /. Tm  with the following property: there exists an integer N  1 such that f X0N .P1 . The projection of V into An . The hypotheses on R say that it is a quotient of kŒX0 . (b) the ideal a has a nontrivial zero in K nC1 . Tm I X0 . For a direct proof of it. Xn  by a graded ideal. Therefore (7.2c).K/. : : : .32. Let V be the zero set of J in An . : : : .30. R is generated as a k-algebra by R1 . : : : . Then exactly one of the following statements is true: (a) there exists an integer d0  0 such that Rd D 0 for all d  d0 .K/  Pn . Pr be polynomials in kŒT1 . we leave it to reader to check that (7. Xn .31). : : : . A ring homomorphism 'W A0 ! K extends to a ring homomorphism def L W A ! K not annihilating the ideal AC D d 1 Ad if and only if '. Tm I X0 . Then B is a graded B0 -algebra with Bd the B0 -submodule generated by the monomials of L degree d in X0 . Let J be the ideal . T HEOREM 7. Ad is finitely generated as an A0 -module.32) is equivalent to (7. : : : . K is any algebraically closed field. Xn .S/ D 0. Consider the ring B D kŒT1 . and let A be the ideal of polynomials f in kŒT1 . (b) no Rd D 0. Statement (a) says that the radical of a contains . Let A D A d 2N d L be the quotient graded ring B=J D d 2N Bd =. P ROOF. Xn  with Pj homogeneous of degree dj in the variables X0 . : : : .Bd \ J /. : : : .X0 .K/ is the zero set of A. P ROOF. (b) for every d  0. Pr / in kŒT1 . and Rd is finite-dimensional for all d . : : : . and J is a homogeneous (graded) ideal in B. T HEOREM 7. Proof of Theorem 7. . C OMPLETE VARIETIES (a) there exists an integer d0  0 such that a contains every homogeneous polynomial of degree d  d0 . : : : . see the article of Cartier and Tate. Let S be the ideal of elements a of A0 such that aAd D 0 for all sufficiently large d . In the statement of the theorem. Xn  and its subring B0 D kŒT1 . and so the theorem is a restatement of (6. : : : . f XnN all lie in J .164 7. : : : . A is generated by A1 . Tm I X0 . The proof proceeds by replacing A with other graded rings with the properties (a) and (b) and also having the property that no Ad is zero. which we deduced from the strong Nullstellensatz. 2 Let P1 .32) for any graded ring A D conditions: L d 0 Ad satisfying the following two (a) as an A0 -algebra.31. : : : .29). and there exists a k-algebra homomorphism R ! K whose kernel is not def L equal to RC D d 1 Rd . : : : . Xn . Tm . : : : .

Let R be the quotient of A00 by the graded ideal P00  A00 . finally eliminates from algebraic geometry the last traces of elimination-theory.30. e The rigidity theorem. which. Let A00 be the ring of fractions of A0 whose denominators are in ˙ D A00 X P. there exists a k-algebra homomorphism f W R ! K such that f . First an observation: for any point w 2 W . 1992. hence in P. Then P is a prime ideal of A0 containing S. Let A00d be the set of fractions with numerator in A0d and denominator in ˙. As A00d is a nonzero finitely generated module over the local ring A000 .v. w/W V ! V  W (this map is regular because its components are). and A0 is a local ring with maximal def ideal P00 D P0  A00 . At this point R satisfies the hypotheses of Theorem 7. to start the chant “eliminate the eliminators of elimination theory”. The ideal J is graded. Now. it has become fashionable again. and T is separated. Section 5.” . by Theorem 7. and let P D Ker. Step 4. and the quotient ring A0 D A=J has the required properties.” Demazure 2012 quotes Dieudonn´e as saying: “Il faut e´ liminer la th´eorie de l’´elimination. Let J be the ideal of elements a of A for which there exists an s 2 A0 X P such that sa D 0. Nakayama’s lemma shows that A00d ¤ P00 A00d . The composite map D f ı " has the required properties. the projection map V  W ! V defines an isomorphism V  fwg ! V with inverse v 7! . The rigidity theorem. T HEOREM 7. this is nothing but the natural map from A0 to k with kernel P. w0 / is independent of v for one w0 2 W . for more on elimination theory. Let 'W V  W ! T be a regular map.v. 2 See Cox.S/ D 0. def Step 2. def Therefore R is graded ring with the required properties. For every d  0.3 provoking Abhyankar. and algorithms. Ideals.e. who prefers equations to abstractions. In degree 0. With the rise of computers.31: “The device that follows. As ' has the same kernel. and assume that V is complete.22 eliminated elimination theory from mathematics.33. Elimination theory became unfashionable several decades ago—one prominent algebraic geometer went so far as to announce that Theorem 7. Step 3. making K into an algebraically closed extension of k. then '. If '. p. and so J \ Ad ¤ Ad .RC / ¤ 0. Chevalley’s Princeton lectures.v. Chapter 8.w/ with g a regular map gW W ! T . w/ D g. and O’Shea. Then A00 D L 00 00 d 0 Ad is a graded ring with the required properties. abelian varieties 165 Let 'W A0 ! K be a homomorphism such that '. abelian varieties The paucity of maps between complete varieties has some interesting consequences. the annihilator of the A0 -module Ad is contained in S. it may be hoped.30. is borrowed from C. A SIDE 7. V and W are irreducible.34 (R IGIDITY T HEOREM ). varieties. Step 1.'/. it factors through "0 . V W q ' W g T 3 Weil 1946/1962. and k D R0 D A000 =P00 is a field. Little. Let " be the composite of the natural maps A ! A0 ! A00 ! R.

then. Commutative groups are distinguished among all groups by the fact that the map taking an element to its inverse is a homomorphism: if . t0 2 T such that '.a0 /: Then '. and consider the regular map gW W ! T. With g as in the proof of the theorem. ' 1 . which must therefore be “rigid”. that V and W are irreducible. 2 C OROLLARY 7. C OMPLETE VARIETIES P ROOF. '. After composing ˛ with a translation.V  W / D ft0 g.V  fwg/ D '. Therefore W X C is nonempty.a C a0 / ˛. and def C D q.36. This means that ˛ is a homomorphism. T HEOREM 7. Let U be an open affine neighbourhood U of f .fv0 g  W /. a 7! aW A ! A. Let 'W V  W ! T be a regular map. a0 / D ˛.v0 . Consider the map 'W A  A ! B.38.V.0/ D 0. then '. w/ … U for some v 2 V .w/ D '. The group law on an abelian variety is commutative.35. w0 / D f . w/ D t0 : 2 In more colloquial terms.166 7. in fact '.v. C consists of the w 2 W such that '. we find that gh D hg.0/ D 0 is a homomorphism. we see that w0 2 W X C . As V  fwg is complete and U is affine. w 7! '. P ROOF. P ROOF.v0 . 2 C OROLLARY 7.V  fw0 g/ D ft0 g D '. '. the preceding corollary shows that it is a homomorphism. P ROOF. a regular map ˛W A ! B such that ˛. '. w0 /. takes the identity element to itself.0  A/ and so ' D 0.T X U / is closed in V  W . By definition. the corollary says that if ' collapses a vertical and a horizontal slice to a point. w0 /.10). w/: We shall show that ' D g ı q. on taking inverses.gh/ 1 D g 1 h 1 .a/ ˛.' 1 .37.W X C / of V  W . then it collapses the whole of V  W to a point. and assume that V is complete.A  0/ D 0 D '. and so it is dense in W . If there exist points v0 2 V .T X U // is closed in W . w/ D g.v0 .v0 . and so W X C D fw 2 W j '. we may suppose that ˛. Since the negative map. D EFINITION 7. w/ D g. w0 2 W . then T X U is closed in T . the projection map qW V  W ! W is closed. 2 . Because V is complete.w/: We have shown that ' and g ı q agree on the dense subset V  . An abelian variety is a complete connected group variety.V  fwg/  U g: As f . and that T is separated. Every regular map ˛W A ! B of abelian varieties is the composite of a homomorphism with a translation. in particular.v. Choose a v0 2 V .a. and therefore on the whole of V  W .V  fwg/ must be a point whenever w 2 W X C (see 7.v0 .

U is a dense in V . is projective).U / ! U (in particular.U / and V 0 equal to the closure of U 0 in V P1   Pn .U / ! U. it suffices to find S g open subsets Vi  P such that q 1 . Thus.V 0 / D p. 'i ) denote the inclusion of U into V (resp. 'n /W U ! V  P1      Pn . into Pi ).U /: The map f is dominant. V D U1 [: : :[Un . be the diagonal map.f.'1 .'0 . which is closed because P is complete.Vi /  V 0 and each map V 0 \ q 1 . : : : .Vi / ! Vi is an immersion.26). Because V is irreducible. Let '0 (resp. : : : . We shall construct an algebraic variety V 0 and regular maps f W V 0 ! V and gW V 0 ! P such that (a) f is surjective and induces an isomorphism f (b) g is a closed immersion (hence V0 1 . and let U D Ui . T Write V as a finite union of nonempty open affines.V /. f is birational). Realize each Ui as a dense open subset of a def Q projective variety Pi . The projection maps pW V  P ! V and qW V  P ! P restrict to regular maps f W V 0 ! V and gW V 0 ! P . The set U 0 is the graph of the map . and f . For every complete irreducible variety V . there exists a surjective regular map f W V 0 ! V from a projective algebraic variety V 0 to V such that. It shows that a complete variety is not far from a projective variety.39 (C HOW ’ S L EMMA ). we have a commutative diagram V '0 f U ' V0 p V P (33) q g P P ROOF OF (a) In the upper-left triangle of the diagram (33). As this is a local condition. and so U 0 D V 0 \ . Hence f is surjective. Therefore f restricts to an isomorphism U 0 ! U .U  P / D f 1 .28). and let ' D . 'n /W U ! P . . which is closed in U  P (5. f induces an isomorphism f 1 . P ROOF OF (b) We first show that g is an immersion. for some dense open subset U of V . Then P D i Pi is a projective variety (6. the maps ' and '0 are isomorphisms from U onto its images U 0 and U . '1 . We set U 0 D '. T HEOREM 7. Chow’s Lemma f 167 Chow’s Lemma The next theorem is a useful tool in extending results from projective varieties to complete varieties.

i is closed in q 1 .Ui / replaced by U 0 . Hence g. : : : .168 7. hence the sets f 1 .Ui / for all i . and let V1 .Vi / D V  Ui  P i  V  P: denote the graph of the map   pi Ui  P i ! Ui .21). Let V be a complete variety. and so the projection map q 1 . We next show that g V 0 \ q 1 .! V :   Being a graph. and so V 0 is complete (7.Vi / ! Vi restricts to an immersion V 0 \ q 1 .Vi /. We first show that the sets q 1 . Consider the diagrams q 1 .Vi /. But U 0 is dense in V 0 . .Ui / cover V 0 .U / i f 'i V P pi ıq pi ıq '0 ' 'i V P Pi U Ui V P Pi : Ui 'i pi ıq Pi The diagram at left is cartesian.21). hence in f 1 . it contains V 0 \ q 1 . Now Vi0 is projective (6.Vi / and contains U 0 . Finally. i is closed in V  Ui  P i and the projection map V  Ui  P i ! Ui  P i restricts to an isomorphism i ! Ui  P i .Vi / ! Vi is an immersion for each i . and so it suffices to show that the middle diagram commutes. and the composite G G Vi0 ! Vi ! V is surjective and birational.V / is closed (7.40. where P i D Y j ¤i Pj : and so q Let i 1 . Vs be the irreducible components of V .Vi / cover V 0 . In other words.V i/ Ui f 1 . C OMPLETE VARIETIES We set Vi D pi 1 . But then it becomes the diagram at right.Ui / D P1      Ui      Pn where pi is the projection map P ! Pi . which obviously commutes. The sets Ui cover V .Vi / ! Vi .26). Recall that V i D Ui  P i . Notes 7.Vi / ! Vi restricts to an isomorphism i ! Vi . V  P is complete because V and P are. and the projection map q 1 . As i is closed in q 1 . and so it suffices to show that 1 q 1 . Each Vi is complete (7. and so it suffices to prove that the middle diagram commutes with f 1 .4). and so there exists a surjective birational regular map Vi0 ! Vi F with Vi0 projective (7.Ui /.Vi /  f .39). and so g is a closed immersion.

3.4 7. Then the functor F 7! F an is an equivalence from the category of coherent OV an -modules to the category of coherent OV modules. Let V be a projective variety over C. . Serre 1955-56.42.12.. Cf. J. 1–42.V. The holomorphic functions on open subsets of V define on V the structure of a C-ringed space. P ROOF. Proposition 6. Inst. Ann. An algebraic variety V is complete if and only if V . . Grenoble 6 (1955–1956). Let V 0 be an open subset of an analytic set V . Lemma 1)5 proved essentially the statement (7. the subset U equals V 0 if and only if V is complete. In particular. OV an / ' . Lefschetz.6. A symposium in honor of S. under which locally free modules correspond. An analytic space is a C-ringed space . There is the following more general statement: For every algebraic variety V . there exists a projective algebraic variety V 0 and a birational regular map ' from an open dense subset U of V 0 onto V whose graph is closed in V 0  V . N. A subset V of Cn is analytic if every v 2 V admits an open neighbourhood U in Cn such that V \ U is the zero set of a finite collection of holomorphic functions on U .V. Algebraic geometry and topology. See also EGA II.C/ is compact in for the complex topology.19. g Analytic spaces. Princeton University Press. G´eom´etrie alg´ebrique et g´eom´etrie analytique.V an . He used the lemma to prove that all homogeneous spaces are quasiprojective. D EFINITION 7.41. OV /. for every v 2 V 0 . for each i .41) by essentially the above argument.44. pp. 5.12). 2 There is a natural functor V V an from algebraic varieties over C to complex analytic spaces (ibid. 1957.45. Chow’s theorem 169 7.1. We omit the definition of a coherent sheaf of OV -modules. the C-ringed space . (b) the topological space V is Hausdorff. The proof uses Chow’s lemma (ibid. Fourier. On the projective embedding of homogeneous varieties. This summarizes the main results of Serre 1955-56 (Th´eor´eme 2. An analytic subset is locally closed. there exists an open neighbourhood U of v in Cn and a holomorphic function h on U such that f D hjV 0 \ U .20). Princeton. A function f W V 0 ! C is holomorphic if. Chow’s theorem We summarize a little of Serre 1955-56.Vi . p.43. OV jVi / is isomorphic to an analytic set equipped with its sheaf of holomorphic functions. OV / satisfying the following two conditions: S (a) there exists an open covering V D Vi of V such that. P ROPOSITION 7. p. 2). 5 Chow. 4 Serre. 2 Jean-Pierre.g. 122–128. T HEOREM 7. Analytic spaces. P ROOF. p. Chow (1956. p. Wei-Liang.

On the other hand. Apply the preceding theorem to the graph of f . and let Z be a compact analytic subset. J.Z V 0 /. Every variety V admits an open immersion V . Let V and W be algebraic varieties over C. no. Let 0 D \. An important theorem of Nagata says that this condition is also sufficient. By Chow’s lemma (7. See: Nagata. Now Z D f . 2 T HEOREM 7.! Pn . Manuscripta Math. 3 1963 89–102. His personal notes are available in three versions. a dense open subset U of V 0 . The proof in the general case is quite difficult. it is a Zariski closed subset of V 0 (hence an algebraic variety). and so there exists a coherent algebraic sheaf F on V such that F an D OZ an .49. and take W to be the closure of V in Pn . 0 is compact. C OMPLETE VARIETIES T HEOREM 7. The support of F is Zariski closed. On compactification of schemes.41).V 00 / is constructible (Zariski sense. Let V be a projective space.47. V 00 D f 1 . 1. .! W into a complete variety W . h Nagata’s Embedding Theorem A necessary condition for a prevariety to be an open subvariety of a complete variety is that it be separated. Kyoto Univ. which shows that it is an analytic subset of U . Every compact analytic subset of an algebraic variety is algebraic. A generalization of the imbedding problem of an abstract variety in a complete variety. 80 (1993). 2 1962 1–10. Kyoto Univ. see: L¨utkebohmert. Every closed analytic subset of a projective variety is algebraic. Deligne translated Nagata’s work into the language of schemes. and let Z be a closed analytic subset of V an . If V is affine. P ROOF. As Z and V 0 are compact and is closed.29). T HEOREM 7. W. 2 C OROLLARY 7. then every holomorphic map f W V an ! W an is algebraic.! An . The graph of z 7! e z W C ! C  C is closed in C  C but it is not Zariski closed.50.46 (C HOW ’ S T HEOREM ). 2 E XAMPLE 7. p.170 7. and a surjective regular map 'W U ! V whose graph is closed in V V 0 . Math. P ROOF. and therefore also of V 0 . see 9. Math. If V is complete. According to Chow’s theorem. P ROOF. Masayoshi. A theorem of Henri Cartan states that OZ an is a coherent analytic sheaf on V an . In the 1970s.7 below).48. For a modern exposition. and therefore its Zariski closure coincides with its closure for the complex topology. 95–111.Z/. but (by assumption) it is closed. Let V be an algebraic variety. J. and so its projection V 00 on V 0 is also compact. then one can embed V . there exists a projective variety V 0 . Imbedding of an abstract variety in a complete variety. and equals Z (ibid.

1–42. 0  i. Israel J. Deligne’s notes on Nagata compactifications. Y / D aij X i Y j . j  m.P. See also: Temkin. J. Q/ D 0 can be written h D f ı p C g ı q where f W V ! A and gW W ! A are regular maps carrying P and Q to 0 and p and q are the projections V  W ! V. Identify the set of homogeneous polynomials F . P.. Nagata’s Embedding Theorem 171 Deligne. and let A be an abelian variety. B.. arXiv:0706. Exercises P 7-1. Kyoto J. Relative Riemann-Zariski spaces. 185 (2011). W . 205–257. with an affine space. Number 4 (2010). P. 22 (2007). Conrad. Show that the subset of reducible polynomials is closed. 50. 2007. Le th´eor`eme de plongement de Nagata. Math. no. Soc. Vojta. Michael. Let V and W be complete irreducible varieties.h. Math.X. Nagata’s embedding theorem. 661-670. Let P and Q be points of V and W .1907. . 7-2.. Ramanujan Math. 19pp. Show that any regular map hW V  W ! A such that h.. 3. A little history Weil introduced the term “complete variety” to denote the algebraic geometers analogue of a compact manifold when he defined abstract algebraic varieties.

.

U / is a normal domain. In its original form. Moreover.V / satisfy f n C a1 f n 1 C    C an D 0.C HAPTER Normal Varieties. These varieties have some of the good properties of nonsingular varieties.14). which are therefore its connected components. (b) For all open affine subsets U of V . (c) For all open subsets U of V . the ring OV . P ROPOSITION 8. the theorem says that a birational regular map to a normal algebraic variety fails to be a local isomorphism only at points where the fibre has dimension > 0. P ROOF.49).1.P is a normal domain. Zariski’s Main Theorem We begin by studying normal varieties.49). a rational function on V that satisfies a monic polynomial equation on U whose coefficients are regular on U is itself regular on U . The equivalence of (a) and (b) follows from (1. A point P on an algebraic variety V is normal if OV. 8 . and it is easy to show that every variety is birationally equivalent to a normal variety.U /. An algebraic variety is said to be normal if all of its points are normal. (a) H) (c). After studying finite and quasi-finite maps. a Normal varieties Recall (1. a normal variety is a disjoint union of its irreducible components. 173 ai 2 OV .42) that a normal domain is an integral domain that is integrally closed in its field of fractions. (a) The variety V is normal. The following conditions on an irreducible variety V are equivalent. and let f 2 k. Since the local ring at a point lying on two irreducible components can’t be an integral domain (see 3.2. which says that every quasi-finite map of algebraic varieties can be obtained from a finite map by removing a closed subset from the source variety. we discuss the celebrated Zariski’s Main Theorem (ZMT). D EFINITION 8. (Quasi-)finite maps. that an integral domain A is normal if and only if Am is normal for every maximal ideal m in A (see 1. Let U be an open subset of V .

Then ai 2 OV . and let E be a finite field extension of its field of fractions F . is normal over F (and is called the normal closure of F in F al ). Let F al denote an algebraic closure of F containing E.E=F / (FT.E=F / is purely inseparable over F (because HomF . 0 so is its submodule A (because A is noetherian).E=F / . and Q If AQ is finitely generated as an A-module. then E is Galois over E Aut.U / is integrally closed in k. k has characteristic p ¤ 0.V /).E=F /. a normal surface need not be nonsingular: the cone X2 C Y 2 Z2 D 0 is normal. : : : . Z ARISKI ’ S M AIN T HEOREM (equality in k. a normal surface can only have isolated singularities — the singular locus can’t contain a curve.51).X1 . : : : . According to the Noether normalization theorem 2. P ROPOSITION 8. 3. The composite in F al of the fields E. Let F al be an algebraic closure of F containing E. For example. The singular locus of a normal variety V must have dimension  dim V 2 (see 8.A0 / and A0 is the integral closure of A0 in E. The normalization of an algebraic variety Let E  F be a finite extension of fields. and normal curves are nonsingular. and so f 2 OP for all P 2 U . there is a unique Yi 2 F al such that Yi D Xi .10).3. In this case. and so we only need to consider the case that A is a polynomial ring kŒX1 .42) is not normal. (Q UASI -) FINITE MAPS .174 8.  2 Aut. According to the above discussion.V /. If E is normal over F . (c) H) (b). As E is finitely generated over F . and E Aut. Yd / . F al / consists of a single element). Conversely. Thus nonsingular varieties are normal. The extension E=F is said to be normal if the minimum polynomial of every element of E splits in E. for each x 2 E. Then the integral closure A0 of A in E is a finite A-algebra (hence a finitely generated k-algebra). there is a single power q of p such that x q 2 F for all x 2 E. Now F D k.12 below). This implies that f 2 OV . N ORMAL VARIETIES . Let A1 denote the integral closure of A in E1 .x/ 2 F . Now E is a finite extension of F .U /  OP for all P 2 U . and. In particular. 2 A regular local noetherian ring is normal — this is a difficult result that we don’t prove here (see CA 22. P ROOF. a normal local domain of dimension one is regular. E  E1  F with E Galois over E1 and E1 purely inseparable over F . For each q i .11). there is a power q. A contains a polynomial subalgebra A0 and is finite over A0 .E Aut. Then A0 is a finite A1 -algebra (1. However. Let A be a finitely generated k-algebra.5 for references). Xd . Therefore we only need to consider the case that E is normal over F .U / (5. Let EQ denote the normal closure of E in some algebraic closure of F containing E. then let AQ denote the integral closure of A in E. Assume that A is an integral domain. but it is singular at the origin — the tangent space at the origin is k 3 . Therefore we only need to consider the case that E is purely inseparable over F . the surface Z 3 D X 2 Y (see 4. : : : . Xd /  E  k.Y1 . The condition applied to an open affine subset U of V implies that OV . and so it suffices to show that A1 is a finite A-algebra.45.x/ of p such that x q.

Let A0 be the integral closure of A in its field of fractions. as before. t 3 t /. : : : . : : : .W. then the integral closure A0 of A in a finite extension E of F .32.6. Corollary 8. (a) The normalization of the cuspidal cubic V W Y 2 D X 3 in k. Td .W / D K and a regular map 'W W ! V such that. A routine argument shows that . P ROOF.V / is the map A1 ! V . P ROOF. 2 C OROLLARY 8. : : : . 2 The proposition shows that if A is an integral domain finitely generated over k.38). P ROPOSITION 8.' 1 . for all open affines U in V .t 2 .v/.47 for some fi 2 A0 . According to (3.A0 / is an irreducible algebraic variety.2 E XAMPLE 8. Let W D v2V W . let W . : : : . and this implies. ' 1 .43).U / ' spm.U / is affine and kŒ' 1 .U // ' Spm.V /. : : : . Let A be as in (8. : : : . and so there exists an h 2 A X m such that. m 2 NI . Obviously kŒY1 .Am /0 D Am .V / is the map A1 ! V . V is birationally equivalent to a hypersurface H in Ad C1 . then Ah is normal for some h 2 A X m. t 7! .29).5. a0 ¤ 0. hfi 2 A.U / is the integral closure of kŒU  in K. P ROPOSITION 8. Xd   A0  kŒY1 . Yd  because kŒY1 . and let K be a finite field extension of k. d D dim V .a.37.4.v/ be the set of maximal ideals in the integral closure of Ov in K. that A0 is a finite A-algebra. Now . If Am is normal for some maximal ideal m in A. and so Ah is normal. The map ' (or just W ) is called the normalization of V in K. t 3 / (see 3. ai 2 kŒT1 .4 shows that the set of normal points is open. called the normalization of Spm. Spm.10) in k. (b) The normalization of the nodal cubic V W Y 2 D X 3 C X 2 (4. fm  1. Normal varieties 175 and A D kŒX1 . Let V be an irreducible algebraic variety. OW j' 1 . Therefore. P ROOF.t 2 1. for all i .A/ has the same properties. For each Fv 2 V .3).A0 /m D . and let 'W W ! V be the map sending the points of W . Yd  contains A and is integrally closed (1. This construction extends without difficulty to nonaffine varieties.kŒU 0 / where kŒU 0 is the integral closure of kŒU  in K. We endow W with the k-ringed space structure for which .U /. Then there exists an irreducible algebraic variety W with k. ' 1 .7.A/ in E. and it remains to show that it is nonempty. Yd  is a finite A-algebra. For an open affine subset U of V . HW a0 X m C a1 X m 1 C    C am .kŒU 0 /. Then A0 D AŒf1 . Now A0h D Ah .v/ to v. 1. OW / is an algebraic variety with the required properties. t 7! . Let V be an irreducible algebraic variety. The normal points in an irreducible algebraic variety form a dense open subset. 3.

OV 0 / D Spm. j D 0.176 8. : : : .kŒU 0 /: In particular. then V 0 D Spm. m 1: By Cramer’s rule (p.U /.34) of my notes Algebraic Number Theory. ' induces an isomorphism ' 1 . for some dense open subset U of V.x 1Cj / C    C cm 1 Tr. we get a system of linear equations c0  Tr.x m 1Cj / D Tr.U /. N ORMAL VARIETIES .44).' 1 . : : : .7 shows that the normalization of V in k.Tr. Td .V / over k.W /: The integral closure kŒU 0 of . Z ARISKI ’ S M AIN T HEOREM moreover.x j / C c1  Tr. . OW / is a finite kŒU -algebra (because it is a kŒU -submodule of the integral closure of kŒU  in k. P ROOF. which implies that V does also.24). Hence kŒH  becomes normal once we invert the nonzero element D. Td ).T1 . 2 P ROPOSITION 8. and hence is integral over A (cf. The normalization of V in W is a regular map ' 0 W V 0 ! V such that. then kŒH  D AŒX =.kŒV 0 / where kŒV 0 is the integral closure of kŒV  in kŒW . Let A D kŒT1 . T1 .V / has this property. there exists a normalization of V in W .A/ .yx j / is a sum of conjugates of yx j . .U /. Let y D c0 C    C cm 1x m 1 . For each open affine U in V we have kŒU   . There is a commutative triangle j W ' V0 '0 V: 1 See. for every open affine U in V . there exists a surjective regular map 'W V 0 ! V from a normal algebraic variety V 0 to V such that.U. For each j 2 N. the proof of 1.H / integral over A. it is an element of A. (2.8.yx j /.H /=F . 2 8. det. OV / in .a0 X m C    C am / D AŒx. OW /  k.A/.Tr. : : : .' 0 1 .x i Cj // D D. ' 0 is an affine map. Trk. (Q UASI -) FINITE MAPS . More generally. Proposition 8.1 and so cl 2 AŒD 1 . m 1: But det. l D 0. if W and V are affine.W /). For example. (34) be an element of k. Td /. On multiplying (34) with x j and taking traces. : : : .9.' 1 . : : : . ci 2 k. As it lies in F . : : : . For every irreducible algebraic variety V .U / ! U (in particular ' is birational). Td can be chosen to be a separating transcendence basis for k. for example. We have shown that H contains a dense open normal subvariety.x i Cj //  cl 2 A. for a dominant map 'W W ! V of irreducible algebraic varieties. Therefore the discriminant D of the polynomial a0 X m C    C am is nonzero (it is an element of kŒT1 .

Y ). For an algebraic variety V . Let . Y 3  is kŒX.37). Y 3 / is a surface in A4 with exactly one singular point. ˘ Every singular curve has singular locus of codimension 1 (hence fails all conditions). 3.Z \ U / D .11. Let Z be a prime divisor on V .24. the origin. Let V be a normal algebraic variety of dimension d . : : : . ni 2 Z. 2 P ROPOSITION 8. nonsingular H) factorial H) normal H) singular locus has codimension  2.39 below).b. and suppose that its singular locus has an irreducible component W of codimension 1. P ROPOSITION 8. then Z is locally principal at P because then we can choose U so that Z \ U D . Z EROS AND POLES OF RATIONAL FUNCTIONS ON NORMAL VARIETIES Let V be a normal irreducible variety. There exists a nonsingular point P on W (4. Recall that an integral domain is factorial if and only if every prime ideal of height 1 is principal (1.13. An irreducible variety V is factorial at a point P if and only if every prime divisor on V is locally principal at P . P ROOF. After replacing V with an open subvariety. 2 S UMMARY 8. Y 2 . Zi a prime divisor on V: The support jDj of D is the union of the Zi corresponding to nonzero ni .U. X Y.11). it does not follow that OV . P ROOF. we may suppose that it is affine and that W is principal. Thus a divisor D can be written uniquely as a finite (formal) sum X DD ni Zi . fd 1 /.f /. say. 4. A divisor on V is an element of the free abelian group Div. This contradicts the definition of W . . ˘ The variety Spm. and let P 2 V . If P … Z.kŒX.f / (see 8. the regular function f is then called a local equation for Z at P .36).U. An algebraic variety V is factorial at a point P if OP is a factorial domain. f / generate the maximal ideal in OV. The variety V is factorial if it is factorial at all points P . : : : .U. but the variety is not normal (the normalization kŒX. A prime divisor Z on a variety V is a closed irreducible subvariety of codimension 1. X Y.U.Z \ U / D .P . . f1 /.P (cf. f1 /.52). fd 1 / be germs of functions at P (on V ) whose restrictions to W generate the maximal ideal in OW. A divisor is said to be effective (or positive) if ni  0 for all i . namely. and I. Regular functions on normal varieties b 177 Regular functions on normal varieties D EFINITION 8.1/. ˘ The variety X12 C    C X52 is factorial but singular. When V is factorial. The codimension of the singular locus in a normal variety is at least 2. Its singular locus has codimension 2.V / generated by the prime divisors. Y 2 ..U / is factorial for all open affines U in V .12. We get a partial ordering on the divisors by defining D  D 0 to mean D D 0  0: .10.U. we say that Z is locally principal at P if there exists an open affine neighbourhood U of P and an f 2 kŒU  such that I. Then . and so P is nonsingular on V . . ˘ The cone Z 2 D X Y in A3 is normal but not factorial (see 9. W D .

V D Zi (see 3.f / is said to be principal. hence irreducible.14.g/ div.p/D1 Ap . (The definition of the Picard group of a general algebraic variety agrees with this definition only for nonsingular varieties.15.Z/  2.R/. each of which extends to V .f / either is empty or is a finite union of prime divisors. by choosing an open affine subvariety U of V such that U \ Z ¤ . it may differ for normal varieties. h 2 R. then f is regular on U . More intrinsically we can define OZ to be the set of rational functions on V that are defined an open subset U of V meeting Z. we can assume fS2 R.V / is the field of fractions of R. Z ARISKI ’ S M AIN T HEOREM Because V is normal. O/I so Rp is a normal domain with exactly one nonzero prime ideal pR — it is therefore a discrete valuation ring (4. The map V X Z ! W .V / of V is defined to be the group of divisors on V modulo principal divisors.178 8. Then k. P ROOF.f /  Z: The sum is over all the prime divisors of V . denoted D  D 0 . Now apply the following statement (proof omitted): a noetherian domain is normal if and only T if Ap is a discrete valuation ring for all prime ideals p of height 1 and A D ht. the Picard group Pic. and div.a/ : The divisor of a nonzero element f of k. and two divisors are said to be linearly equivalent.42) and ordZ .V / ! Div. but in fact ordZ . we can assume that V is affine (because it is a finite union of affines). (Q UASI -) FINITE MAPS . regular outside a subset of codimension  2.) T HEOREM 8.V / ! Z with valuation ring OZ . which is defined to be OZ . A rational function on a normal variety. We may suppose that W is affine. and so the ideal p corresponding to it is minimal among the nonzero ideals of R D . The zero set of f . 2 C OROLLARY 8.f / D ordZ . if they differ by a principal divisor. and so we can write f D g= h with g.f / D div. there is associated with every prime divisor Z on V a discrete valuation ring OZ ..U.h/.f / D 0 unless Z is one of the Zi . is regular everywhere. When V is nonsingular. P ROOF. and embed it as a closed subvariety of An . The map f 7! div.V / is defined to be X div. We may assume that V is connected. then U \ Z is a maximal proper closed subset of U . onto Let ordZ be the valuation k.V / is a homomorphism. and let f be rational function on V . Therefore V X Z ! An extends to An .! An is given by n regular functions on V X Z. for example. C OROLLARY 8. Therefore. and let 'W V X Z ! W be a regular map defined on the complement of a closed subset Z of V . 2 . If f has no zeros or poles on an open subset U of V .20).16. A divisor of the form div. and its image is contained in W .f /W k.f / D 0 for all but finitely many Z . N ORMAL VARIETIES . Let V and W be affine varieties with V normal. This is a restatement of the theorem. then ' extends to a regular map on the whole of V . Let V be a normal variety. if  is a prime element of OZ . then a D unit   ordZ . thus. say V D Spm. In proving this. V . This can be defined. If codim. P ROOF.

c. fn / D A.18.fi / with fi 2 A. : : : . After multiplying through by a power of fi .19. the set ' 1 . say. we shall assume in the proof that V and W are irreducible . P ROOF.a11 b11 C    C a1m1 b1m1 / C    C hn . A S regular map 'W W ! V of algebraic varieties is finite if there exists a finite covering V D i Ui of V by open affines such that. Then b=1 2 Bfi . The necessity being obvious. bi mi g generate Bfi as an Afi -module.5) and (8.f1 . and so bD ai1 ai mi bi mi .f / \ Ui D D.f1r1 . Finite and quasi-finite maps c 179 Finite and quasi-finite maps Finite maps D EFINITION 8. some aij 2 A and ri 2 N: ri bi1 C    C fi firi The ideal .Ui / is a finite kŒUi -algebra.Ui / is a finite kŒUi -algebra.' 1 . The next lemma shows that. some hi 2 A: Now b D b  1 D h1  bf1r1 C    C hn  bfnrn D h1 . We shall show that the family of all bij generate B as an A-module. as required. Each Ui is a finite union of basic open subsets of V .Ui / is affine and kŒ' 1 . 1 D h1 f1r1 C    C hn fnrn .f1r1 .W. For simplicity. Ui D D. Therefore. : : : .f1 . for maps of affine algebraic varieties.f jUi /.Ui / is affine and kŒ' 1 . Then ' is finite.Ui /i be a finite family of open affines covering V and such that. We have to show that B is a finite A-algebra. : : : .20.an1 bn1 C    C anmn bnmn /. 2 L EMMA 8. fn / D A and that Bfi is a finite Afi -algebra for each i .17. we may assume that the bij lie in B.V /. This follows from the definition (8.U /. : : : . Let . Let V be an irreducible algebraic variety. Let A D kŒV  and B D kŒW . OW /: . These are also basic open subsets of Ui . Let fbi1 . Let b 2 B. the above definition agrees with the definition (2. : : : . because D. for each i . we prove the sufficiency. E XAMPLE 8. and so we may assume that the original Ui are basic open subsets of V . L EMMA 8. the set ' 1 . fnrn / would have to contain .3). There is a canonical isomorphism of k-algebras . A regular map 'W W ! V of affine algebraic varieties is finite if and only if kŒW  is a finite kŒV -algebra.39). and let 'W W ! V be the normalization of V in a finite extension of k. fnrn / D A because any maximal ideal containing . Let 'W W ! V be a regular map with V affine. and let U be an open affine in V . We are given that . OW / ˝kŒV  kŒU  ! . for each i.

and consider the commutative diagram: . Let Vi be an open affine covering of V (which we may suppose to be finite) such def that Wi D ' 1 .Wi . We have a morphism of varieties over V U0 canonical Spm. OW // (35) V which we shall show to be an isomorphism. Therefore the canonical morphism is an isomorphism over each Vi . O W /.j . for every open affine U in V . But (8.U / is affine and kŒ' 1 . Z ARISKI ’ S M AIN T HEOREM P ROOF. sequence 0 ! M 0 ! M ! M 00 is exact if and only if 0 ! Am ˝A M 0 ! Am ˝A M ! Am ˝A M 00 is exact for all maximal ideals m of A (1. and so it is an affine k-algebra finite over kŒU .U 0 . OW / 0 Q .U / is a finite kŒU -algebra. then.W.U 0 .U 0 .180 8. Let 'W W ! V be a regular map.U 0 .W. ' 1 . .2 The varieties Wi and Wi \ Wj are all affine. and that a submodule of a noetherian module is noetherian. Therefore. The bottom row is exact because OW is a sheaf. g/ 7! .U /. 3 Recall that a module over a noetherian ring is noetherian if and only if it is finitely generated. 2 S UMMARY 8. 5.22.21. N ORMAL VARIETIES . and so it is an isomorphism.U 0 \ Wi .21) shows that Spm. 2 P ROPOSITION 8. g ı 'jU 0 /W .31. Let W D Wi be a finite open affine covering of W . Let U 0 D ' 1 . OW /  kŒU  ! .21). This implies that the first is also an isomorphism.U .U \ Wi . 2A . . a submodule of a finitely generated module is finitely generated.U \ Wij . OW //. (Q UASI -) FINITE MAPS . it is an isomorphism (see 5. and let U D ' . .Wij . OW // is an isomorphism.U / is affine and kŒ' 1 .P ' OV. OW /: When W is also S affine.P ' kŒV mP for all P 2 U . 0 1 0 Let Q U be0 an open affine in V . The map is defined by the kŒV -bilinear pairing . P ROOF.f jU 0 .f. and the top row is exact because OW is a sheaf and kŒU  is flat over kŒV . OW / ˝kŒV  kŒU  i Q i Q i. Then . OW / ˝kŒV  kŒU  . If ' is finite. . OW / ˝kŒV  kŒU  0 .U 0 \ Wi .Vi / is an affine subvariety of W for all i and kŒWi  is a finite kŒVi -algebra. Let 'W W ! V be a regular map of algebraic varieties. We know that each of the maps U 0 \ Wi ! Spm.32).U 0 \ Wi . and consider the following condition on an open affine subset U of V : (*) ' 1 . O3W / is a subalgebra of i . OW // is the inverse image of Vi in Spm. This implies the claim because kŒU mP ' OU.U / is a finite over kŒU . OW / Here Wij D Wi \ Wj .U /. and so the two vertical arrows at right are products of isomorphisms. OW / .

This is a consequence of the going-up theorem (1. and fbj0 g is a set of generators for B 0 as an A-module. the fbi ˝ bj0 g is a set of generators for B ˝A B 0 as an A-module. it suffices to prove (a) and (b) for maps of affine varieties. The base change of a finite map is finite. and let Z be a closed subset of W . and the map Z \ U ! U corresponds to a map A ! A=a of rings.25. T 1  is contained in T n kŒT  for some n).23a and b). and so it remains to show that the first map is surjective. note that if fbi g is a set of generators for B as an A-module. The map corresponds to a finite map of rings A ! B. . The restriction of ' to Z is finite (by 8. in which case (*) holds for all open affines of V .! A1 is not finite because the ring kŒT. For example.A=a/ with the closed subvariety V . (a) Closed immersions are finite. Finite maps of algebraic varieties are closed. (a) Let Z be a closed subvariety of a variety V . which is obviously finite. P ROOF. To see this. the inclusion A1 X f0g . (c) If B and B 0 are respectively finite A and A0 -algebras. open immersions are rarely finite. P ROOF. It is therefore affine.B/ ! Spm. Then the statements become statements in commutative algebra. T 1  is not finitely generated as a kŒT module (any finitely generated kŒT -submodule of kŒT.24. and so we can replace W with Z.c.'/ is closed. P ROPOSITION 8. Then Z \ U is a closed subvariety of U . 2 The base change of a finite map Recall that the base change of a regular map 'W V ! S is the map ' 0 in the diagram: V S W '0 W 0 V ' S: P ROPOSITION 8.! B. As to be finite is a local condition.23. Let 'W W ! V be a finite map of affine varieties. It suffices to prove this for affine varieties. then B ˝k B 0 is a finite A ˝k A0 -algebra. This proves (a). and let U be an open affine subvariety of V . T HEOREM 8. then C is a finite A-algebra. (b) The composite of two finite morphisms is finite. Finite and quasi-finite maps 181 The map ' is finite if (*) holds for the open affines in some covering of V . from which we obtain maps Spm. and fcj g is a set of generators for C as a B-module.A/: The second map identifies Spm. note that if fbi g is a set of generators for B as an A-module. To see this. we then have to show that Im.53). (b) If B is a finite A-algebra and C is a finite B-algebra. This will factors as A ! A=a .! Spm.a/ of Spm. (c) The product of two finite morphisms is finite.A/. 2 By way of contrast.A=a/ . then fbi cj g is a set of generators for C as an A-module.

then so also is V . we wish to show ' 1 . if S is complete. As for affine varieties (2. . is surjective. 2 Quasi-finite maps Recall that the fibres of a regular map 'W W ! V are the closed subvarieties ' 1 . (Q UASI -) FINITE MAPS . mn be maximal ideals in A. They are obviously coprime in pairs. A finite k-algebra A has only finitely many maximal ideals. t 3 /W A1 ! V . Let m1 . The map ' then corresponds to a map ˛W A ! B of affine k-algebras. The base change of a finite map is finite. 2 Finite and quasi-finite maps of prevarieties are defined as for varieties. P ROPOSITION 8.71) is quasi-finite but not finite — its image is not closed in A1 . After replacing V with an affine neighbourhood of P . Examples 8. and f1.2 P ROPOSITION 8. Then the statement becomes: if A is a finite R-algebra.t 2 . 2 C OROLLARY 8.30. and then W will be affine also. N ORMAL VARIETIES .P / is finite.28. Clearly B=˛.m/B. P ROOF. : : : . T g is a set of generators for kŒT  as a kŒT 2 . The map t 7! .39). 8. we can suppose that it is affine. then A ˝R B=N is a finite B-algebra. and so the points of W mapping to P are in one-to-one correspondence with the maximal ideals of B=˛. Z ARISKI ’ S M AIN T HEOREM P ROOF. which is obvious.29.mQ / D mP . T 3 -module (see 3. Finite maps of algebraic varieties are proper.29). ai mod mi .A=mi /  n — here dimk means dimension as a k-vector space. But this holds if and only if mQ  ˛.182 8.27.31. and so it is a finite k-algebra. and hence closed. We may assume that all the varieties concerned are affine. a 7! . It follows that dimk A  X dimk .P / of W for P 2 V . : : :/. we say that a regular map of algebraic varieties is quasi-finite if all of its fibres are finite.mP /. T 3 .26. and a point Q of W maps to P if and only ˛ 1 . The next lemma shows that it has only finitely many maximal ideals. P ROOF.m/B is generated as a k-vector space by the image of any generating set for B as an A-module.26). Y  in kŒT  is kŒT 2 . P ROOF.1) shows that the map A ! A=m1      A=mn . Combine (7.: : : . Let 'W V ! S be finite. Let P 2 V . and so the Chinese Remainder Theorem (1. X 1  is not finite over kŒX .Y 2 X 3 /  A2 from the line to the cuspidal cubic is finite because the image of kŒX.19) and (8. The projection from the curve X Y D 1 onto the X axis (see p. and kŒX. A finite map 'W W ! V is quasi-finite. P ROOF. 2 L EMMA 8.

0. 8. 8. The minimum polynomial of Y over kŒA. B. : : : . an . X =. and so it has exactly n points. : : : . by multiplying the equation by A. and consider the projection map . The map A2 X foriging . an / 2 An is the set of solutions of X n C a1 X n 1 C    C an D 0. Y / and let f be the map defined on the ring level by X 7! X D A Y 7! X Y 2 C Y C 1 D B: Then f is (obviously) quasi-finite. tn / 7! .32. B is AY 2 C Y C 1 B D 0.49.36. Y  is not integral over its subring kŒA. one can show directly that Y can never satisfy an equation Y s C g1 .37. 8.t1 . an /W V ! An : The fibre over a point . counted with multiplicities.A.34. The obvious map . B. : : : . 0/ is bijective but not finite (here fOg D Spm. 8. B/Y s 1 C    C gs . because the inverse image of A2 is not affine (see 3. B/ D 0. B (see 1.A. : : : .kŒX. Let V be the hyperplane X n C T1 X n 1 C    C Tn D 0 in AnC1 .35. Tn . B/ 2 kŒA. Tn  ! kŒT1 . Let V D A2 D Spec.31). The map A2 X f.! A2 is quasi-finite but not finite. a 7! am is finite. For this we have to show that kŒX. 1 C    C Tn /: .a1 . but it is not finite.k/ D A0 ). gi .0.X n C T1 X n See also the more general example p. 8. : : : . The map (8.44). which shows that it is not integral over kŒA. and the Frobenius map p . kŒT1 . it is also finite because it corresponds to the finite map of k-algebras.c.a1 . Finite and quasi-finite maps 183 8.33.33). are examples of finite bijective regular maps that are not isomorphisms. : : : . x/ 7! .A1 with the origin doubled / ! A1 is quasi-finite but not finite (the inverse image of A1 is not affine). The map is certainly quasi-finite. tnp /W An ! An in characteristic p ¤ 0. Alternatively.A. The map A1 ! A1 .a1 . : : : .t1 . 0/g t fOg ! A2 sending O to .

a0 .41. y .184 8.! kŒT . t . Z ARISKI ’ S M AIN T HEOREM 8. 0. It is one-to-one except that the points t D ˙1 both map to 0.W / D dim.W / is separable over k.t 2 1. T HEOREM 8.P /ˇ  deg. Y / D T0 X n C T1 X n 1 Y C    C Tn Y n .) See also the more general example p. : : : . : : : . N ORMAL VARIETIES . / some of the roots “vanish off to 1”. it corresponds to the inclusion  x 7! T 2 1 kŒx. The projection map ' . : : : . and so k. 8. an /W V ! AnC1 has finite fibres except for the fibre above o D . 0/. the projection map V ! AnC1 X fog is finite.30) is a special case of this. Let V be the hyperplane T0 X n C T1 X n 1 C    C Tn D 0 in AnC2 .'/.AnC1 X fog/. but not finite.a1 . Let P . .W / to be separable over k. ˇ' 1 . Above points of the form . The map ' is said to be separable if the field k. : : : . : : : .V / for the second part of (b).39.T 2 1/ .38. an . and that we need k.V /. y 7! T .P /ˇ D deg. Its restriction to V X ' 1 . . Recall that jSj denotes the number of elements in a finite set S . and consider the map t 7! . Its degree is called the degree of the map '.40. and let V be its zero set in P1  .X. E XAMPLE 8. In this case. ˇ ˇ (b) The set of points P of V such that ˇ' 1 . Before proving the theorem. we give examples to show that we need W to be separated and V to be normal in (a). (b) Let C be the curve Y 2 D X 3 C X 2 .0.49.o/ is quasi-finite. which is A1 . and assume that V is normal.W / is a finite field extension of k. (Q UASI -) FINITE MAPS .'/ is an open subset of V . (Example (8. (a) The map fA1 with origin doubled g ! A1 has degree one and is one-to-one except over the origin where it is two-to-one. Then dim.0. x/ 7! .V /. and it is nonempty if ' is separable.V /. d The fibres of finite maps Let 'W W ! V be a finite dominant morphism of irreducible varieties.t 2 1//W A1 ! C . On coordinate rings. ˇ ˇ (a) For all P 2 V . Let 'W W ! V be a finite surjective regular map of irreducible varieties.

Consider the discriminant disc F of F . we may assume that V and W are affine. : : : . Because (*) has distinct roots. y in its field of fractions k. The factorization T 7!f kŒV  ! kŒV ŒT =.X1 .F /. i. and so m D r. This proves that ' 1 . Then the polynomial T m C a1 .X1 .'/ points. (c) The Frobenius map p .Qi // D 0. and part (a) of the theorem then implies that it has exactly deg.P 0 / has at least deg. The field extension corresponding to the map is p k. : : : . In order to prove the first part of (b). and it has coefficients in kŒV  because V is normal (see 1. .F // ! V: Each point P 0 2 U has exactly m inverse images under the second map.e. and so the map corresponds to a finite map of k-algebras. y/ D k. According to the lemma. The fibres of finite maps 185 and so is of degree one. Qr be distinct points on an affine variety V . but has degree p n .d. L EMMA 8.P / ¤ 0.Qi / are all roots of a single polynomial of degree m. The ring kŒx. 2 P ROOF ( OF 8. and therefore surjective (recall that a finite map is closed). we show that. P ROOF.'/ elements. and so disc. Let ' 1 .F / is nonzero on an open neighbourhood U of P .V / D deg '.P /  f . disc.40). then the same is true for all points in an open neighbourhood of P . : : : . kŒV  ! kŒW .P / D 0: Therefore the f . Then there is a regular function f on V taking distinct values at the Qi . anp /W An ! An in characteristic p ¤ 0 is bijective on points. y is not integrally closed — in fact kŒT  is the integral closure of kŒx. and the first map is finite and dominant. : : : . : : : .42. an / 7! .a1 .W / W k.F / ! kŒW  gives a factorization W ! Spm.P / D fQ1 . Now F . Xnp / which is purely inseparable..a1 . there exists an f 2 kŒW  taking distinct values at the Qi .kŒV ŒT =.x.Qi /m 1 C    C am .V /. We can embed V as closed subvariety of An . : : : . Qr g. Choose f as in the last paragraph corresponding to such a P .T / D T m C a1 T m 1 C    C am be the minimum polynomial of f over k. Let F .f / D 0 implies F . It has degree m  Œk. and then it suffices to prove the statement with V D An — almost any linear form will do. f .44).P / has deg. if there is a point P 2 V such that ' 1 .T /.f . In proving (a) of the theorem.P /  T m 1 C    C am . Xn /  k.'/. Let Q1 .P / D 0 (*) has r D deg ' distinct roots.'/ points for P 0 2 U .Qi /m C a1 . and so r  m  deg.

(b) For all i .V /. this was proved in (8. Texts and Readings in Mathematics. To prove the general case. and assume that V is normal.44.V /Œf  D k. Statement and proof One obvious way to construct a nonfinite quasi-finite map is to take a finite map W ! V and remove a closed subset of W .W / over k. 20.9). Zariski’s Main Theorem (ZMT) shows that. with equality holding on a dense open subset U. factor ' as the composite of a purely inseparable map with a separable map. P ROOF. the statement is true with ' 0 W V 0 ! V equal to the normalization of V in W (in the sense of 8.45 (Z ARISKI ’ S M AIN T HEOREM ). N ORMAL VARIETIES . C.q / ! V factors through '.W / is separable over k. T HEOREM 8.12 of Musili.43.F // ! V shows that #' 1 . Hindustan Book Agency. . Algebraic geometry for beginners. If ' is separable. we explain a fundamental theorem of Zariski.'/ and its discriminant is a nonzero element of kŒV . (Q UASI -) FINITE MAPS .P / has deg ' elements.'/.! V 0 ! V with ' 0 finite and j an open immersion: W open immersion quasi-finite V0 finite V When ' is a dominant map of irreducible varieties. Every quasi-finite map of algebraic vari'0 j eties 'W W ! V factors into W . For an elementary proof. ˇ ˇ Vi D fP 2 V j ˇ' 1 .40). the Frobenius map V . If ' is purely inseparable. and the separable degree of E over F is defined to be the degree of F sep over F . Its minimum polynomial F has degree deg. for some q. The diagram W ! Spm.P / ¤ 0.P /ˇ  i g is closed in V . see Theorem 63. 2001. ˇ' 1 . e Zariski’s main theorem In this section.V /.P /ˇ  sepdeg. Let 'W W ! V be a finite surjective regular map of irreducible varieties. T HEOREM 8. New Delhi. Z ARISKI ’ S M AIN T HEOREM We now show that if the field extension is separable. A finite map from a variety onto a normal variety is open (hence both open and closed). The elements of E separable over F form a subfield F sep of E. 2 Let E  F be a finite extension of fields.W /. then there exists a point such that ' 1 . The separable degree of a finite surjective map 'W W ! V of irreducible varieties is the separable degree of k. 2 A SIDE 8. there exists a f 2 kŒW  such that k. Because k.P /  deg. every quasi-finite map arises in this way.f /. ˇ ˇ (a) For all P 2 V .AŒT =.186 8. then ' 1 F is one-to-one because. for algebraic varieties.'/ for P a point such that disc.

Bfi / form an open covering of W and Af0 ' Bfi i for all i .46 ( LOCAL VERSION OF ZMT). Because ' is quasi-finite. 1 . . each Ai is finite over A (see 1.p/-algebra. As W quasi-compact. finitely many sets spm. Zariski’s main theorem 187 The key result needed to prove (8. The argument in the proof of (8. 2 Recall (Exercise 8-3) that a regular map 'W W ! V is affine if ' U is an open affine subset of V . here m D ' 1 . P ROOF. Let 'W W ! V be a regular map of algebraic varieties. The isolated points v of an algebraic variety V are those such that fvg is both open and closed. we can take V 0 D Spm. then n 2 spm. Theorem 8. and let i W A ! B be a finitely generated A-algebra.Bfi / suffice to cover W . Thus they are the irreducible components of V of dimension 0.B/ is isolated in its fibre if and only if Bn =mBn is a finite k-algebra. finite over A. Every quasi-finite map of affine algebraic varieties 'W W ! V factors j '0 into W ! V 0 ! V with j a dominant open immersion and ' 0 finite. let A0 denote the integral closure of A in B. P ROOF. Let Ww and Vv be open affine neighbourhoods of w and v D '. As Spm.e. and spm. The proof is quite elementary.B/ ! spm.A00 / is an open immersion because it is when restricted to Spm. If Bq =pBq is a finite .A/ D V is finite.46 shows that there exist fi 2 A0 such that the sets spm.B/ ! Spm.w// of W .47) shows that there exists an A-subalgebra A00 of A0 .n/. P ROPOSITION 8. Theorem 8. then there exists an f 2 A0 not in q such that the map Af0 ! Bf is an isomorphism. fn and is such that Bfi ' Af00 for all i . but intricate — see 17 of my notes CA. T HEOREM 8.A00 /.w/ D 0g — it is the maximal ideal in B corresponding to w.Bfi / for each i .47. For a ring A and a prime ideal p in A.'/W spm. Bf D Aif for all sufficiently large Ai .'. Now the map i W D Spm. and let A D kŒVv  and B D kŒWw . Let A0 be the integral closure of A in B. Let q be a prime ideal of B. 2 Recall that a point v in a topological space V is isolated if fvg is an open subset of V . Let A be a commutative ring. The set W 0 of points of W isolated in their fibres is open in W .46 shows that there exists an f 2 A0 not in m such that Af0 ' Bf .U / is affine whenever .A00 / ! Spm. We say that w 2 W is isolated in its fibre if it is isolated in the subspace ' 1 .Ww /  Vv .48. Bf is quasi-finite over A.A/. Let A D kŒV  and B D kŒW .p/ denotes the field of fractions of A=p. which contains f1 . and consider spm. Finally. Let 'W A ! B be a homomorphism of finitely generated k-algebras.q/. Write A0 as the union of the finitely generated A-subalgebras Ai of A0 containing f : [ A0 D Ai : i Because A0 is integral over A. Let w 2 W 0 . Let n D ff 2 B j f . and let p D i 1 . As the Ai are finite over A. P ROOF. Let 'W W ! V be a continuous map of topological spaces. 2 P ROPOSITION 8.45) is the following statement from commutative algebra.w/ such that '. : : : .Bf / is an open neighbourhood of w consisting of quasi-finite points.35). We have [ Bf ' Af0 D Aif : i Because Bf is a finitely generated A-algebra.

W / is closed.w. T.U / cover W . N ORMAL VARIETIES .3)). the map ' is separated if V is separated. Every quasi-finite regular map 'W W ! V of algebraic varieties with W complete is finite. Because ' is proper.! W 0 ! W with ˛ finite and j an open immersion. Let 'W W ! V be a quasi-finite map of algebraic varieties. Z ARISKI ’ S M AIN T HEOREM P ROPOSITION 8. 8.! V 0 in (8.53.51. The map j W W .. Then ' 0 is open (8.28.45 from (8. in the proof of (8.W / is finite. To deduce the general case of Theorem 8.U /.A0 / is an open immersion.44) requires an additional argument. this proves Theorem 8. If W is complete (7. Theorem 8.w. Assume that V is normal in (8. and Wedhorn. this implies that j W W ! V 0 is an open immersion.jw/ ! W V W 0 .A0 / is an open subvariety of V 0 and the sets ' 1 . and so ' is open. and then ' 1 .U / ! Spm. which is finite over A (8. Let A D kŒU  and B D kŒ' 1 . every quasi-finite map from an algebraic variety to a normal algebraic variety is open. the second map is closed. 2010.83 of G¨ortz.2 P ROPOSITION 8.54. As Spm.45 is false for prevarieties (see 8. there is a factorization ' D ' 0 ı j with ' 0 finite and j a dominant open immersion. it is true for separated maps of prevarieties. P ROPOSITION 8. Wiesbaden. P ROOF. Factor j into W w7!.33). Applications to finite maps Zariski’s main theorem allows us to give a geometric criteria for a regular map to be finite.52. In this case. N OTES 8.9) is an open immersion. 2 As V 0 ! V is finite. Factor ' into W . Vieweg + Teubner.44).W / in V 0 . v/W V ! V S V is closed.45 in the case that ' is an affine map of irreducible varieties. (Q UASI -) FINITE MAPS .45). U. Hence j is an open immersion with closed image.48) we can take A00 D A0 . we may replace V 0 with the closure of the image of j . the normalization A0 of A in B is finite over A (because it is contained in the normalization of A in k. See Theorem 12. and so the restriction of ' 0 to j.49. and that W is isomorphic to that connected component.188 8. Algebraic geometry I. Thus. Let 'W W ! V be an affine map of irreducible algebraic varieties.50. W 0 is separated because it is finite over a variety — exercise). 2 . Let U be an open affine in V . P ROOF.45) is an isomorphism of W onto its image j. Then the map j W W ! V 0 from W into the normalization V 0 of V in W (8.W /.21). then j.w 0 /7!w 0 ! W 0: The image of the first map is j . Thus. In (8. j ˛ P ROOF. (A regular map 'W V ! S of algebraic prevarieties is separated if the image V =S of the map v 7! .45). However. which is closed because W 0 is a variety (see 5. Every proper quasi-finite map 'W W ! V of algebraic varieties is finite.v. Thus. It follows that its image is a connected component of W 0 .) 8.

54. if we require the target variety to be normal (thereby excluding example (b)). If V is normal and the map ' is quasi-finite. a bijective regular map of algebraic varieties need not be an isomorphism.t 2 . : : : . Both rings have k. (a) The inclusion of an open subset into a variety is birational. Applications to birational maps Recall (p. It follows that V 0 D V . See the section on blow-ups below.V /. xn / 7! . When W and V are curves. In topology. 2 8. sending each a 2 A1 X f0g to a and O 2 A0 to 0. Consider. .V / (because ' is birational). kŒ' 0 1 . One may ask how a birational regular map 'W W ! V can fail to be an isomorphism.V / ! k. But k. Thus it is easy to give examples of closed surjective quasi-finite. Similarly. but it is if W is compact and V is Hausdorff.55. For each open affine subset U of V . Here C is the cubic Y 2 D X 3 .56. a continuous bijective map 'W W ! V need not be a homeomorphism. then we are left with (a).P / is an isomorphism onto V X P .U / (as varieties). and we require the map to be quasi-finite (thereby excluding example (c)). The next result says that. Here are three examples. (b) The map (8. 8.x1 . for example. t 3 /.57. there is a regular birational map 'W V 0 ! V such that the restriction of ' to V 0 X ' 1 . but nonfinite.45. the Frobenius map p . but ' 1 .P / is the projective space attached to the vector space TP . but it is not an isomorphism even though An is normal.116) that a regular map 'W W ! V of irreducible varieties is said to be birational if it induces an isomorphism k. (c) For any smooth variety V and point P 2 V . and kŒU  is integrally closed in k.T / as their fields of fractions. and so U D ' 0 1 .x1 . and the map kŒC  ! kŒA1  D kŒT  identifies kŒC  with the subring kŒT 2 .V / (because V is normal). every surjective map W ! V is closed. is birational.W / D k. then ' is an isomorphism from W onto an open subvariety of V . maps.e. Factor ' as in the Theorem 8. xnp /W An ! An is bijective and regular. : : : . the map  A1 X f0g t A0 ! A1 . because the map is only closed.U / is the integral closure of kŒU  in k. Let 'W W ! V be a birational regular map of irreducible varieties. T 3  of kŒT . A1 ! C.31) from A1 to the cuspidal cubic. P ROOF.W /.W / on the fields of rational functions. This doesn’t violate the Proposition 8.58. Zariski’s main theorem 189 N OTES 8. Here are three examples: (a) In characteristic p. P ROPOSITION 8. not universally closed. t 7! .

 is an isomorphism if the varieties are irreducible. Its graph is the union of . The projection  from to W is a bijective regular map. If V is normal and ' is quasi-finite. even in characteristic zero. The hypotheses imply that ' is an isomorphism of W onto an open subset of V (8. N ORMAL VARIETIES . dim. if 'W W ! V is a map (in the set-theoretic sense) of algebraic varieties V. and we may apply (8. lists the following variants of ZMT. P ROPOSITION 8. (c) Consider the regular map A1 ! A1 sending x to 1=x for x ¤ 0 and 0 to 0. see (8. the map . Harvard notes.V / (see Theorem 9.45). then ' is an isomorphism. 0/ and the hyperbola xy D 1. Because ' is bijective.w.V / is 1 (apply 8.59). 1966. if a linear map 'W W ! V def between two Banach spaces has a closed graph D f. t 3 / ! tW C ! A1 inverse to that in (8. Hence ' is birational. Reprinted as “The Red Book of Varieties and Schemes” . Let 'W W ! V be a bijective regular map of irreducible varieties. If the map ' is birational and V is normal. 9. One can ask (cf. Let 'W W ! V be a bijective regular map of irreducible algebraic varieties. Wikipedia). (Q UASI -) FINITE MAPS . 2 8. mo113858) whether a similar statement is true in algebraic geometry.x. t 3 / from A1 to the cuspidal cubic (see 8.57). the inverse of the Frobenius map x 7! x p provides another counterexample.60. 'w/ j w 2 W g. and  is generically separable. 2 In fact.4 III. P ROOF. Original form (8. then ' is continuous (q.59. According to (8. Variants of Zariski’s main theorem Mumford.9 below) and the separable degree of k.t 2 .W / D dim. but it is not an isomorphism even though A1 is normal.56b) has closed graph but is not regular.61. If we require the map to be birational (thereby excluding example (a)). regular. a map between irreducible normal algebraic varieties in characteristic zero is regular if its graph is closed. example (a) can be excluded by requiring that ' be generically separable (instead of birational). In particular. then ' is an isomorphism of W onto an open subvariety of V . P ROPOSITION 8. If V is normal and k. W whose graph is closed (for the Zariski topology). For a third counterexample. (with the introduction of misprints) by Springer 1999. and the varieties to be irreducible (thereby excluding example (c)). For example. W is normal. which is a closed subvariety of A1  A1 .40 to the variety V 0 in 8. then ' is an isomorphism. birational map. Because ' is bijective. P ROOF. and so ' will be regular if  is an isomorphism. Z ARISKI ’ S M AIN T HEOREM (b) The map t 7! . but not an isomorphism. then the map is an isomorphism.V /.190 8.t 2 . then is ' a regular map? The answer is no in general. y/ 7! xW ! A1 is a bijective. The projection .57) Let 'W W ! V be a birational regular map of irreducible varieties.0. In characteristic p.v. the closed graph theorem states that. 4 Introduction to Algebraic Geometry.56b) is bijective.58c).W / over k. Specifically. the open subset must be the whole of V .60). V to be normal (thereby excluding example (b)).W / is separably generated over k. In functional analysis.

W // ! V: The first major step in the proof of the theorem is to show that ' OW is a coherent sheaf on V . is the map attached by Spm to the map of k-algebras kŒU  ! OW .8. and let v be a (closed) normal point of V .A. Let S be the singular locus of V . Then the complex neighbourhoods U of v such that U X U \ S is connected form a base for the system of complex neighbourhoods of v.' 1 . This. Grothendieck’s form (8. 1998. T. That its fibres are connected is a consequence of the following extension of Zariski’s connectedness theorem to non birational maps..Z be the local ring attached to an irreducible closed subset of V (cf. Trans. Introduction to Algebraic Geometry. The original form of the theorem (8. for Zariski’s original statement and proof of this theorem.45) Every quasi-finite map of algebraic varieties factors as the composite of an open immersion with a finite map.62 (S TEIN FACTORIZATION ). See Springer.. Math. V 2. over every open affine subset U of V . Foundations of a general theory of birational correspondences. 490–542. See the discussion in Mumford 1966. Power series form Let V be a normal variety.' OW / ! V: By construction. .177). (1943). N OTES . The ' 1 . p. The last two forms are much deeper and were proved by Grothendieck. it is the sheaf of OU -algebras defined by a finite kŒU -algebra. T HEOREM 8. and let v 2 V .f. who showed that it implied the original form. f Stein factorization The following important theorem shows that the fibres of a proper map are disconnected only because the fibres of finite maps are disconnected. When V is affine. then so also is its completion. The original form of the theorem was proved by Zariski using a fairly direct argument whose method doesn’t seem to generalize. 5 See Lang. 53.U //: To say that ' OW is coherent means that. Here ' OW is the sheaf of OV -algebras on V ..v/ is a connected set (in the Zariski topology).' 1 . S.57) is the “Main theorem” of Zariski. on every open affine subset U of V .' OW / ! V that. this is the factorization W ! Spm. U OW . Soc. Linear Algebraic Groups. means that there exists a regular map '2 W Spm.OW . Amer. Connectedness theorem Let 'W W ! V be a proper birational map. Stein factorization 191 Topological form Let V be a normal variety over C. Every proper map 'W W ! V of algebraic va'1 '2 rieties factors into W ! W 0 ! V with '1 proper with connected fibres and '2 finite. If OV. in turn. '2 is finite and '1 W W ! W 0 has the property that OW 0 ! '1 OW is an isomorphism. 1958. and let OV. 5. for a direct proof of (8. O.2.5 The power series form was also proved by Zariski.59).Z is an integrally closed integral domain.U //: The Stein factorization is then '1 def '2 W ! W 0 D Spm.

: : : .39 then shows that it suffices to prove resolution of singularities for projective varieties.V /: In other words. the blow-up at P replaces P with P. Let  be the graph of .V / is the geometric tangent cone at P . . and let W An X fOg ! Pn 1 be the map . and Chow’s lemma 7.V / be the tangent space at P . the normalization of V (see 8. More generally. When a variety admits a desingularization. N OTES . Wikipedia).64. N ORMAL VARIETIES .V // ! V X Sing.CP . we shall consider only projective varieties.192 8. 11. then we say that resolution of singularities holds for V . the nonsingular variety W is the same as V except over the singular locus of V . if V is projective. Then the fibres of ' are connected. The Stein factorization was originally proved by Stein for complex spaces (q. Z ARISKI ’ S M AIN T HEOREM T HEOREM 8. The nodal cubic 8. A desingularization of V is birational regular map W W ! V such that W is nonsingular and  is proper. Note that with “nonsingular” replaced by “normalization”. The blow-up of V at P is a regular map VQ ! V that replaces P with the projective space P. Blowing up a point on a variety Examples 8. and let Tp . Let 'W W ! V be a proper map such that the map OV ! ' OW is an isomorphism.a1 W : : : W an /. See Hartshorne 1977.5) provides such a map (resolution of abnormalities). Let P be a nonsingular point on an algebraic variety V . and let A n n f ! A defined by the projection map An  Pn 1 ! An is the blow-up of An The map W A at O.65. III.50 shows that it suffices to prove resolution of singularities for complete varieties.v. and  should induce an isomorphism W X 1 . (Q UASI -) FINITE MAPS . g Blow-ups Under construction. then W should also be projective. From now on.a1 . Nagata’s embedding theorem 7. .Sing.63. The cuspidal cubic h Resolution of singularities Let V be an algebraic variety.V // where CP . an / 7! fn be the closure of  in An  Pn 1 .TP . Blowing up the origin in An Let O be the origin in An .V //.

W / is allowed to be a finite extension of k. de Jong proved a weaker result in which. Show that if V is separated. is a complete rational equivalence class). There he noted that the singular locus of a normal variety has codimension at least 2 and that the system of hyperplane sections of a normal variety relative to a projective embedding is complete (i. One criterion is that a variety is normal if and only if every surjective finite birational map onto it is an isomorphism (8. and observed that the normalization W VQ ! V of a curve V in k. def Vi D fP 2 V j ' 1 . 1939.. Prove that a finite map is an isomorphism if and only if it is bijective and e´ tale.12). find a finite map 'W W ! V with the following property: for all 1  i  n. It is not easy to give a geometric intuition for “normal”.57). The resolution of singularities for three-folds in characteristic zero is much more difficult. For every n  1. J. His result was extended to nonzero characteristic by his student Abhyankar and to all varieties in characteristic zero by his student Hironaka. but the first to be accepted as rigorous is that of Walker (patching Jung’s local arguments. 14.V /. normalization gives a surface with only point singularities (8. The resolution of singularities for higher dimensional varieties in nonzero characteristic is one of the most important outstanding problems in algebraic geometry. Zariski’s introduction of the notion of a normal variety and of the normalization of a variety was an important insertion of commutative algebra into algebraic geometry. .V / is a desingularization of V . Zariski showed that the desingularization of a surface in characteristic zero can be obtained by alternating normalizations and blow-ups.P / has  i pointsg is a nonempty closed subvariety of dimension i . Let 'W V ! W be a regular map with the property that ' 1 .h. 8-3. A little history Normal varieties were introduced by Zariski in a paper. Algebraic Curves).) 8-2. then so also is W . of Math. See mo109395 for a discussion of this question. p. (Cf. Zariski introduced the notion of the normalization of a variety. Amer. instead of the map  being birational.e. and was first achieved by Zariski (Ann. 1944). 8-4. 1935). Exercises 8-1. 61.249–194. There were several proofs of resolution of singularities for surfaces over C. In 1996. Give an example of a surjective quasi-finite regular map that is not finite (different from any in the notes). For a surface V . Resolution of singularities 193 Resolution of singularities for curves was first obtained using blow-ups (see Chapter 7 of Fulton’s book. Harris 1992. k. which can then be blown up. Math.U / is an open affine subset of W whenever U is an open affine subset of V (such a map is said to be affine).9.

.

However.4.A2 X fy-axisg/ [ f.a. In particular. y/ 7! . As c runs over the elements of k.a/ ¤ 0 extends to a homomorphism ˇW kŒW  ! k with ˇ. According to Lemma 9. 0/ ' . b=a/ 1 if . y-plane except for the y-axis. Then '. in fact. b/ D Y -axis : . P ROOF.W / contains an open dense subset of V .a. it turns out that almost everything that can go wrong already goes wrong for in this example. . the homomorphism g 7! g. 2 Before beginning the proof of Lemma 9. we should look at an example.0. (b) the dimensions of the fibres can jump only over closed subsets.0. The line X D c maps to the line Y D cX . b ¤ 0: From this unpromising example. this line sweeps out the whole x. Let 'W W ! V be a dominant regular map of irreducible affine algebraic varieties. xy/ (Exercise 3-3). which is neither open nor closed.4 below.x. a The constructibility theorem T HEOREM 9. 0/g.a. the target variety is normal. and so the image of ' is C D .1. and.a/. it would appear that it is not possible to say anything about the image of a regular map or its fibres.1/ ¤ 0. provided the map is finite. and k has characteristic zero. b/ D .P /W kŒV  ! k extends to a nonzero homomorphism ˇW kŒW  ! k.C HAPTER Regular Maps and Their Fibres Consider again the regular map 'W A2 ! A2 . (c) the number of elements (if finite) in the fibres can drop only on closed subsets.x. We shall show: (a) the image of a regular map is a finite union of locally closed sets. 195 9 . The fibre 8 if a ¤ 0 < point . is not even locally closed. the map f 7! f ı 'W kŒV  ! kŒW  is injective. for P 2 D. The kernel of ˇ is a maximal ideal of kŒW  whose zero set is a point Q of W such that '. Because ' is dominant.Q/ D P . there exists a nonzero a 2 kŒV  such that every homomorphism ˛W kŒV  ! k such that ˛. if a D 0.

am / D 0.T / D bn T n C    C b0 . Let A  B be finitely generated k-algebras.r/: Q Because ˛.3.T /. Therefore. Because deg g 0 < m. let h0 . deg r < m: On applying ˛Q to this equation.g Q i T //.f Q / C ˛.am /d ˛. r 2 AŒT .b0 / D ˛.b0 / ¤ 0: . h0 . 2 . Extend ˛ to a homomorphism ˛W Q AŒT  ! kŒT T to T . If a D 0.f Q / has degree m > 0.f / with f D am T m C    C a0 .cr 1 /c C    C c0 c r D 0: On taking h D g. B D AŒt  ' AŒT =a.T /. Let A be an affine k-algebra. we may assume n < m.cr / C ˛.196 9. If a0 does not contain nonzero constants.T / can’t exist.T / with r.a/ Q contains a nonzero constant. R EGULAR M APS AND T HEIR F IBRES E XAMPLE 9. and so ˛.2.T /f . the induction hypothesis implies that ˛ extends to a homomorphism B 0 ! k.am / T m C    C ˛.i > 0/. When does a homomorphism ˛W A ! k extend to B? The extensions of ˛ correspond to roots of the polynomial ˛.c/ D ˛. In particular.r/ Q is a nonzero constant.a0 / in k.g Q i / D ci ˛. and that B is generated by a single element. and on taking h D f .b0 /¤ 0.a/ Q is an ideal (because T  ci ˛. ˛. Then every homomorphism ˛W A ! k such that ˛.T /. and so there exists an extension unless this is a nonzero constant polynomial. Then the Amodule generated by the polynomials T s h0 . Let f D am T m C    C a0 be a nonzero polynomial of minimum degree in a such that ˛.T / C r. ˛ extends if ˛. s  0. The homomorphism ˛Q AŒT  ! kŒT  h7!h. we see that c D 0.T / D q. P ROOF. Then a0 contains the polynomial g 0 D bn C    C b0 T n . Moreover. which implies t D 0 and A D B (since B is an integral domain). is an ideal a0 in AŒT . In the contrary case. On dividing f .T / D cr T r C cr 1 T r 1 C    C c0 2 a. it generates a proper ideal in kŒT . and so we may suppose m > 1 and (by induction) that the lemma holds for smaller values of m. The k-submodule of P by sendingP kŒT  generated by ˛. we have that m  1. Unless ˛. there is a c 2 k such that.am / ¤ 0. say. q. Thus we may assume that a ¤ 0. T 7! T 7! c then factors through AŒT =a D B and extends ˛. and ˛.T / into g.T / D cr T r C cr 1 T r 1 C    C c0 .c/ ¤ 0 extends to a homomorphism B ! k.bi / D 0 ˛. we must have ˛.q/ Q ˛. we obtain the contradiction ˛. and let B D AŒ T =.11). If m D 1. such a g.c/ ! k. then set B 0 D AŒT =a0 D AŒt 0 . for all h.q/ Q D 0. then every homomorphism ˛ extends. L EMMA 9. d 2 N.T / D cr C    C c0 T r . we obtain ˛. For h. a contains a polynomial g. Let c  A be the set of leading coefficients of the polynomials in a. a0 contains a nonzero constant if and only if a contains a nonzero polynomial cT r .T /. we find that d am g. Assume that A and B are integral domains. which will have a zero c in k (2. After replacing g. h 2 a.am / ¤ 0. Because B ¤ 0. Note that c is an ideal in A.

Let W be a topological space. Contrast Q. ai 2 A. Case (ii).b/ ¤ 0. P then there P exists a c 2 k such that f . xn 1  ! k such that ˛.c/ ¤ 0. it is defined by a finite number of statements of the form f .b/ ¤ 0. Assume that A and B are integral domains. statements of the form f D 0 combined using “and”. “or”. we need the notion of a constructible set. it and h are coprime over that field. Let h. we obtain an element a 2 A with the required property.0/. AŒX ! AŒx. v 2 AŒT  and c 2 A X f0g such that uh C vf D c: It follows now that cam satisfies our requirements.    .bn 1 / ¤ 0 extends to a homomorphism ˇW B ! k such that ˇ. The constructibility theorem 197 L EMMA 9. Let A  B be finitely generated k-algebras.T / 2 AŒT  represent b. : : : . It is also possible to deduce Theorem 9. Because f is irreducible over the field of fractions of A. and “not”). but does not contain an open subset of R. and so ˇ. xn .b/ ¤ 0. As b is nonzero. which is dense in R (real topology).a.1) to arbitrary maps of arbitrary varieties. and let b be a nonzero element of B. P ROOF Suppose that we know the proposition in the case that B is generated by a single element. or any infinite subset of A1 that omits an infinite set.4. Xn / ¤ 0 combined using only “and” and “or” (or. h … a. if C is constructible and V  W . xn 2  etc. A constructible set in An is definable by a finite number of polynomials. Hence there exist u. Thus we may assume B D AŒx. Write b D f .0/. am ¤ 0. Then there exists a bn 2 2 AŒx1 . If ˛W A ! k is such that ˛. Then there exists a nonzero a 2 A with the following property: every homomorphism ˛W A ! k from A into k such that ˛. Let a be the kernel of the homomorphism X 7! x. P ROPOSITION 9. better. Then there exists an element bn 1 2 AŒx1 .    . In order to generalize (9.X1 . Let f . The ideal a D . and ˇ. Let C be a constructible set whose closure CN is irreducible. Obviously.a0 / ¤ 0. : : : .X1 .u. Case (i). be an element of a of minimum degree. and write B D AŒx1 . then ˛ can be extended to ˇW B ! k by the preceding lemma. g.di /c i .x/  b/ D ˇ. Xn / D 0. and take a D a0 . The next proposition shows that a constructible set C that is dense in an irreducible variety V must contain a nonempty open subset of V . more precisely.x/ D a0 x n C a1 x n 1 C    C an .11). then C \ V is constructible. Then C contains a nonempty open subset of CN .c/ ¤ 0. : : : .1 from the generic freeness theorem (CA 21. The ideal a ¤ .cam / ¤ 0.a/ ¤ 0 extends to a homomorphism ˇW B ! k such that ˇ.5. . and we can take ˇ to be the homomorphism di x i 7! ˛. Continuing in this fashion.T / D am T m C    a0 . for if ˛.6. 2 A SIDE 9. xn 1  with the following property: every homomorphism ˛W AŒx1 . A subset C of W is said to constructible if it is a finite union of sets of the form U \ Z with U open and Z closed. : : : .

V U / is constructible in ' 1 .C \ Wi /. If W1 . Let 'W W ! V be a dominant regular map of irreducible varieties. the set '. it is therefore constructible if the '. We may therefore assume that CN D W . Then (a) dim.C \ ' 1 .W1 / [ : : : [ '.CN / < dim. we see that it suffices to prove this in the case that both W and V are affine. We may also replace V with '. then '.W /. b The fibres of morphisms We wish to examine the fibres of a regular map 'W W ! V . : : : . R EGULAR M APS AND T HEIR F IBRES S P ROOF. Replace V be the closure of '. C is a finite union of its irreducible components. Moreover.7.V U / is a proper closed subset of W (the complement of V U is dense in V and ' is dominant). then the closure of X for the Zariski topology is equal to its closure for the complex topology.W / . Wr are the irreducible components of W . A regular map 'W W ! V sends constructible sets to constructible sets.C / is constructible (by induction). Clearly C  Zi . If dim. Then C \ Wi is constructible in Wi . then the statement is obvious because W is a point.W / D '.V U //: Then ' 1 . 2 A SIDE 9. T HEOREM 9. We may therefore assume that C also is irreducible.C / D U [ '.V U /.V /. We are in the situation of (9.Ui \ Zi / D Ui \ Zi : Thus Ui \ Zi D Ui \ CN is a nonempty open subset of CN contained in C . if U is a nonempty open subset of W . If CN ¤ W . By considering suitable open affine coverings of W and V .W / D 0.8. it must be contained in one of the Zi .W /  dim. and so assume that both W and V are irreducible.V U // is constructible in V by induction. But then CN contains a nonempty open subset of W .W / in V . For this i C  Ui \ Zi  Ui \ CN  Ui \ C  Ui \ . As C \ ' 1 . We are given that C D . CN is then an irreducible closed subvariety of W .C / contains an nonempty open subset U of its closure.C /.C \ ' 1 . which completes the proof. and so it suffices to prove the statement in the case that W is irreducible.1).Ui \ Zi / with each Ui open and each Zi closed. We can replace V by the closure of '. 2 T HEOREM 9.U / contains a nonempty open subset of its closure in V . In particular. we see that '. and write '. If X is constructible for the Zariski topology on Cn .198 9.9. Hence we may assume that W is irreducible. and '. and because C is constructible in CN . . Let X be a subset of Cn . and so the case just proved shows that '.W / is the union of the '. '. Let Wi be the irreducible components of W .Wr / . and these are closed in C . then the closure of '.C \ Wi / are. S We N may assume that each set Ui \ Zi in this decomposition is nonempty.W / in V and so assume that ' is dominant. We now prove the theorem. they are therefore constructible. then dim. and as CN is irreducible. P ROOF We first prove the “in particular” statement of the Theorem. We shall prove the theorem by induction on the dimension of W .

(a) Because the map is dominant. and so (if nonempty) has codimension  m in W (see 3. Then k.V / . fP 2 V j dim ' 1 . In particular. kŒV  ! kŒW  is injective.W / dim. : : : . and obviously tr degk k.P /  r 0 g is closed in V .P / 2 k.V /  k.P //. Let m be the dimension of V . : : : .P //  i g are closed in '.aij .45). i D 1.b. yN . : : : .V / remains algebraically independent in k.47) we know that there exist regular functions f1 .aij . Since the rank of the matrix . E XAMPLE 9. (b) In proving the first part of (b). Consider the subvariety W  V  Am defined by r linear equations m X aij Xj D 0. fP 2 V j rank. and it may jump further on closed subsets.V /: In proving the second part of (b). we can suppose that P D V .W /).W / (an algebraically independent subset of k. xM  and kŒW  D kŒy1 . Then ' 1 .W /.W / dim. Before proving the theorem.W /. the dimension of the fibre is > dim.f1 .P //  dim.V / for every P 2 V . fm /. then dim.P / is the zero set of the regular functions f1 ı '. aij . r.W /. Write kŒV  D kŒx1 . j D1 and let ' be the projection W ! V .P /  dim W m D dim. we may replace V by any open neighbourhood of P . From (3. hence. i D 1. : : : . there is a homomorphism k.V /  tr degk k. j D1 and so its dimension is m rank. we can replace both W and V with open affine subsets.W / dim. for each r 2 N. : : : . For P 2 V .aij . and we may regard it as an inclusion (we identify a function x on V with x ı ' on W /. the fibre ' 1 . In other words.V /.P / is the set of solutions of system of equations m X aij . : : : .' 1 .W / 199 dim. r. (c) The sets Vi D fP 2 V j dim. we should look at an example. : : : . with equality holding exactly on a nonempty open subset U of V .P // drops on closed subsets. fm /. the dimension of the fibre jumps on closed subsets.W /. fm ı '. : : : . Since ' is dominant. and suppose V and W have dimensions m and n . fm such that P is an irreducible component of V .P //  rg is a closed subset of V (see Exercise 2-2). P ROOF.P / has the expected dimension dim. we can assume V to be affine. for each r 0 2 N. ' 1 . aij 2 kŒV . Hence dim ' 1 .f1 .10.! k.P /Xj D 0. for P on a dense open subset U of V . More precisely. On the closed complement of U (possibly empty). The fibres of morphisms (b) if P 2 '.' 1 . After replacing V by a smaller neighbourhood of P .V /.

and so for P 2 U .7).P /. yi / D 0. There are three possibilities. Finally.9) shows that there is a nonempty open subset U of V such that for P 2 U . Let U D V X Vi —it is a nonempty open subset of V .P / dim. yNn m . We know from (b) that there is an open subset U of V such that 1 dim ' .P / D ' 1 .P /. if for a particular point P . we obtain the result.x1 . : : : . : : : . then we can modify the above argument to show that the same is true for all points in an open neighbourhood of P .Z/ < n C dim. Let 'W W ! V be a regular surjective closed map of varieties. There are therefore relations Fi .V / < n: is not contained in Z.P / \ Z. yNi / D 0. xm . R EGULAR M APS AND T HEIR F IBRES respectively. it has fibres .e. Xm . yn with Fi . When we restrict them to ' 1 .P / D n m ” P 2 U: Let Z be the complement of U in V . Yi / is identically zero.Zj / ! Zj for each j .'jZ/ 1 . dim. then W is irreducible of dimension dim. 2 Recall that a regular map 'W W ! V of algebraic varieties is closed if. : : : .P /. and we may suppose that y1 . y1 . : : : . and let n 2 N.200 9.P /. Zr be the irreducible components of Z. Y1 . dim. : : : . xm . xm .' 1 Thus. Then '. xm . y1 . the dimension of ' 1 .Z/ is a proper closed subset of V . (a) '. yn m are algebraically independent over kŒx1 .V /. yn m . yNN / over k will be  n m. : : : .P /.P /.V / C n. : : : .X1 .x1 . P ROPOSITION 9.x1 . : : : . Then m . i Dn m C 1.. if none of the polynomials Fi . xm . N: If these are nontrivial algebraic relations. Yi / is identically zero.x1 .11.P / D n m. : : : . On applying the induction to the restriction of ' to the map ' 1 . : : : . If P 2 U . the fibre ' . : : : . yNN : The equations (36) give an algebraic relation among the functions x1 . and hence D n m by (a). Let Vi be the closed subvariety of V defined by the simultaneous vanishingSof the coefficients of this polynomial—it is a proper closed subset of V . Let Z1 . xm .yN1 .V /. : : : .P / is  n m. Yi / as a polynomial in the Y ’s with coefficients polynomials in the x’s. Y1 . for example. : : : . : : : . : : : . : : : . Yn of yi to ' 1 . (c) We prove this by induction on the dimension of V —it is obviously true if dim V D 0.W / has transcendence degree n m over k. : : : . : : : . Yn m . for P 2 U . dim ' 1 . Yn m . regard Fi .Z/ D V . i Dn m C 1. Thus. We write yNi for the restriction . If V is irreducible and all fibres ' 1 . and consider the map 'jZW Z ! V . yi on W . they become equations: Fi . and that the remaining yi are algebraic over kŒx1 . : : : .P /. Y1 .P / \ Z/ D dim. Then (b) of (9.x1 . Y1 .Z/ ¤ V . then the transcendence degree of k. yN1 . (b) '. Let Z be a irreducible closed subset of W . Then k. then none of the polynomials Fi . N: (36) a nonzero polynomial. . thus Z D Vn mC1 . : : : . Yn m .P /. Yi / 1 kŒ' m . i. xm . W is complete (7.P / of ' are irreducible of dimension n. P ROOF.P / D kŒyN1 .Z/ 1 . : : : . : : : .

in addition. It is faithfully flat if.X / ! 0 with k. then n D ˛.P / \ Zi /.b ˝ n2 / if ˇ. there will be an open subset U of V such that forSP in U .' 1 . It therefore defines an A-linear map gW B ˝A N 00 ! Q.n/ — both are obvious.Z/ dim.n0 /. thus ' 1 . Let b 2 B.Z/ D V . Let B ˝A N ! Q be the cokernel of 1 ˝ ˛. dim.ˇ ı ˛/ D 0.P / is irreducible and ' 1 . Zr be the irreducible components of W .n2 /.b ˝ n/ factors through B  N 00 . and let B be an A-algebra.b ˝ n/ and that f ı g D idB˝A N 00 on elements of the form b ˝ ˇ. . We shall construct an inverse g to f . I claim that (c) holds for at least one of the Zi .9) shows that dim. It follows by linearity that . If ˇ.n0 / for some n0 2 N 0 . and let n 2 N . : : : .V /.1 ˝ ˇ/ ı . Now let Z1 .b ˝ n1 / D . when we tensor the exact sequence of kŒX -modules 0 ! kŒX  f 7!X f ! kŒX  f 7!f mod .Z/  n C dim.12.P / 1 .V /  n  Z for all P 2 V .n1 / D ˇ. . An A-algebra B is flat if M ! N injective H) B ˝A M ! B ˝A N injective. To show that f and g are inverse. n/ 7! .P / is contained in none of the Zi . moreover dim Z D n. Flat maps and their fibres 201 (c) '. and so Z D W .P / \ Z/  dim.n/ D 0.c. Then (b) of (9. ' 1 .P / D . it suffices to check that g ı f D idQ on elements of the form . there is a unique A-linear map f W Q ! B ˝A N 00 such that f ı  D 1 ˝ ˇ.X / ! kŒX =. then the sequence of B-modules 1˝ˇ 1˝˛ B ˝A N 0 ! B ˝A N ! B ˝A N 00 ! 0 is exact.b ˝ n/ D 0. and so this is impossible. For example. we get the sequence 0 id k ! k ! k ! 0: D EFINITION 9.1 ˝ ˛/ D 1 ˝ .b. Because . and so . hence b ˝ n D b ˝ ˛. B ˝A M D 0 H) M D 0: 1 The  surjectivity of 1 ˝ ˇ is obvious. and so the A-bilinear map B  N ! Q.' for all P . Otherwise. but ' 1 . 2 c Flat maps and their fibres Flat maps Let A be a ring. If the sequence of A-modules ˇ ˛ 0 ! N 0 ! N ! N 00 ! 0 is exact.1 but B˝A N 0 ! B˝A N need not be injective.

13). (b) Every open immersion is flat (obvious). because A ! B is flat.B ˝A A0 / ˝A0 M ' B ˝A . B ˝A M 0 is a submodule of B ˝A M . B ˝A M from A- E XAMPLE 9. we see that '. and so the statement follows from the proposition. Let x 2 M . Therefore '. Hence a D B. Let A ! A0 be a homomorphism of rings. and let m D ˛ 1 . Let n be a prime ideal of B.a/  B  n ¤ B. the map A˛ 1 .N 0 ! N / is injective (9. 2 P ROPOSITION 9. Then a is an ideal in A.13.17. and M 0 D Ax ' A=a.13c).m/ is contained in a maximal ideal n of B. an A-algebra B is flat if and only if the functor M modules to B-modules is exact.m/B: As B ˝A M ¤ 0. Then S 1 A is a flat A-algebra (1. and so the composite Am ! Bn is flat (9. nc D m. Now ' 1 . (a) Let S be a multiplicative subset of A. and hence equals m. B ˝A M 0 ' B='.14. x maps to zero in Mn . Because a is proper. For the converse. Let m and n be the (unique) maximal ideals of A and B. P ROOF. P ROOF.a/  '. and let M D A=m.2 C OROLLARY 9. Moreover. let N 0 ! N be an injective homomorphism of A-modules.n/ — it is a prime ideal in A. Hence '.15. P ROOF.n/ for some maximal ideal n of B.18). then so also is A0 ! B ˝A A0 . By hypothesis. A flat homomorphism 'W A ! B is faithfully flat if and only if every maximal ideal m of A is of the form ' 1 . .x/ D fa 2 A j ax D 0g. and let n be a maximal ideal of B.A0 ˝A0 M / ' B ˝A M: In other words.n/ is a proper ideal in A containing m. If A ! B is flat. ): Let m be a maximal ideal of A. for all maximal ideals n in B. Let x be a nonzero element of M .Am ˝A .N 0 ! N / has the property that Mn D 0.202 9. R EGULAR M APS AND T HEIR F IBRES Therefore. and let a D def def ann.n/ ! Bn is flat. and let a D fb 2 B j bx D 0g.14). it is contained in a maximal ideal m of A.N 0 ! N / ' Bn ˝Am .a/  B and. The map Sm 1 B ! Sn 1 B D Bn is flat (9. the map Bn ˝A . For any A0 -module M . (c) The composite of two flat maps is flat (obvious). If A ! B is flat. Then Am ˝A . (: Let M be a nonzero A-module. then B ˝A M ' B='. A homomorphism ˛W A ! B of rings is flat if and only if. A flat local homomorphism A ! B of local rings is faithfully flat. P ROOF.13a). and therefore '.N 0 ! N // is injective. and so x D 0.m/B ¤ B.m/  n for some maximal ideal n of A. 2 P ROPOSITION 9. For each maximal ideal n of B. and so B ˝A M  B ˝A M 0 ¤ 0. and so a contains an element not in n. and so the kernel M of B ˝A . Therefore it preserves exact sequences. 2 .16. P ROPOSITION 9. then so is Am ! Am ˝A B ' Sm 1 B (9. tensoring an A0 -module M with B ˝A A0 is the same as tensoring M (regarded as an A-module) with B. Therefore.

Because A ! B is flat.P is flat.T / is a dense subset of S. The ring B ˝A .22.p/ D S 1 .qBq /c . such that qc D p. Then q contains a prime ideal q0 such that q0c D p0 : q0 B q  A p  p0 : P ROOF. Let B be a faithfully flat A-algebra. and after the second step only with those such qc \ S D .e. we are left with the prime ideals q of B such that qc  p. and let T be a closed irreducible subset of W such that '.p/ is nonzero.18.16). P ROOF. The ideal p0 Ap is prime (1.p/ ¤ 0 and A ! B is faithfully flat. for all P 2 W . A regular map 'W W ! V of affine algebraic varieties is flat (resp. P ROPOSITION 9. 2 P ROPOSITION 9. 2 D EFINITION 9. and because pAp D . and so there exists a prime ideal of Bq lying over p0 Ap (by 9.p/ from B by first passing to B=pB and then making the elements of A not in p act invertibly. 2 P ROPOSITION 9. and let p be a prime ideal of A.. Every prime ideal p of A is of the form qc for some prime ideal q of B. Let A ! B be a flat homomorphism. Let B be an A-algebra. The contraction of this ideal to B is contained in q and contracts to p0 in A. it is faithfully flat (9. After the first step.19. because . Let 'W W ! V be a flat map of affine algebraic varieties. Then . Let S D A X p.'/W spec.A/ includes all maximal ideals of A. Flat maps and their fibres 203 Properties of flat maps L EMMA 9. A flat homomorphism 'W A ! B is faithfully flat if the image of spec. P ROPOSITION 9.B/ ! spec. 2 S UMMARY 9. . Therefore we obtain B ˝A . The prime ideals of B contracting to p are in natural one-to-one correspondence with the prime ideals of B ˝A . Then there exists a closed irreducible subset T 0 of W containing T and such that '. and so it has a prime (even maximal) ideal q. the homomorphism Ap ! Bq is flat.21 (G OING .14). Let S  S 0 be closed irreducible subsets of V .24.A=p/.20. faithfully flat) if and only if the map f 7! f ı 'W kŒV  ! kŒW  is flat (resp. P ROOF. in which case it includes all prime ideals of A.p/..16).DOWN THEOREM FOR FLAT MAPS ). and let q be a prime ideal in B such that qc D p. and it is faithfully flat if it is flat and surjective. P ROOF. Apply (9. qc D p.T 0 / is a dense subset of S 0 . the map OV. A regular map 'W W ! V of algebraic varieties is flat if.15) and (9. faithfully flat). Let p  p0 be prime ideals in A. i.19).23. For this ideal.'Q ! OW.c.

V D Spm. Moreover ' qc D p because T ! S is dominant and so the map kŒS  D kŒV =p ! kŒT =q is injective. : : : . In other words. : : : .A/ and W D Spm. xn q is an integral domain). : : : . : : : . Then b1 . xm  does not contain the polynomial q because the coefficient a of q is invertible in Aa . Let a be some nonzero coefficient of the polynomial q. : : : . 2 T HEOREM 9.204 9. br generate Bq as an AŒx1 . B injective finite F ˝A B E ˝AŒx1 . For every regular map 'W W ! V of irreducible algebraic varieties. : : : . and q D I. 2 . xm q and so is zero (because AŒx1 . We may assume that W and V are affine. Then each bi is integral over Aa . According to (9. Each bi satisfies a monic polynomial equation with coefficients in F Œx1 . Let b1 .T /. This completes the proof. xm q ! Baq : The first and third arrows realize their targets as nonzero free modules over their sources. p0 D I. : : : .13). which implies that each element of its kernel is killed by a nonzero element of AŒx1 . and so Bq is free of finite rank over AŒx1 . Let a 2 A be a common denominator for the coefficients of these polynomials. bn generate B as an Aalgebra. we may suppose that B is a finite AŒx1 . : : : . and let b1 . say. xm q . xm q . the Noether normalization theorem (2.21). : : : . We regard B as a subring of F ˝A B. we may suppose that it lies in B.S 0 /.x1 . : : : . : : : . 1 . Let F be the field of fractions of A.x1 . xm . The middle arrow is flat by (9. and so the map Aa ! Aa Œx1 . xm  A F: def E D F . xm of F ˝A B such that F Œx1 . : : : . : : : .xm  B finite finite AŒx1 . : : : . : : : . xm . : : : . : : : . Then p  p0 because S  S 0 . xm -algebra (1. : : : . Let q be a common denominator for the coefficients arising from a set of generators for B as an AŒx1 . Let m be a maximal ideal in Aa . this shows that Ba is a finite Aa Œx1 . there exists a nonempty open subset U of V such that ' faithfully flat.xm  B as an E-vector space. xm q is faithfully flat (apply 9. As the bi generate Ba as an Aa -algebra. Then mAa Œx1 . xm -module. xm rq ! Bq (*) is surjective. Hence the map (*) is an isomorphism.36). xm  F Œx1 . xm q -module. As F ˝A B is a finitely generated F -algebra. Therefore. Each element of B can be expressed as a linear combination of the bi with coefficients in E. : : : .:::. : : : . : : : .S /. This map becomes an isomorphism when tensored with E over AŒx1 .:::. : : : . xm -algebra. : : : . : : : . xm  is a polynomial ring over F and F ˝A B is a finite F Œx1 . R EGULAR M APS AND T HEIR F IBRES P ROOF.q0 / has the required properties. after replacing A with Aa and B with Ba .B/.16). xm q is a proper ideal of Aa Œx1 . xn / Let E D F . After multiplying each xi by an element of A. : : : . and consider the maps Aa ! Aa Œx1 . there exists a prime ideal q0 in kŒW  contained in q and such that q0c D p0 . : : : . Let p D I. xm  ! Aa Œx1 . : : : . br be elements of B that form a basis for E ˝AŒx1 .45) shows that there exist elements x1 . xm / be the field of fractions of AŒx1 . : : : . xm q . cr / 7! ci bi W AŒx1 . xm -algebra.U / ' ! U is P ROOF. Now V .25 (G ENERIC FLATNESS ).c1 . and so are faithfully flat. Hence mAa Œx1 . the map P .

P ROOF.V /.W / dim. Let P 2 '. Let 'W W ! V be a regular map of irreducible algebraic varieties. '.9).'U 0 /.29.Z/ D dim. P ROOF. Then '.6).Z/ D fP g.25). P ROPOSITION 9.c.28.P /  dim.27).U 0 / is flat. we may suppose that both V and W are affine. Let V be an algebraic variety.'U /. P ROOF. Let Z be an irreducible component of CN .' 1 . Let Z  Z1      Zn . U 0 D ' 1 .V /. Therefore V X '. A constructible subset C of V is closed if it has the following property: let Z be a closed irreducible subset of V .U // is open in V (9. Let 'W W ! V be a dominant map of irreducible algebraic varieties. Then Z \ C is constructible and it is dense in Z. Then there exists a dense open subset U of W such that '. then we shall prove (more precisely) that.U / ! U is flat. Note that ' 1 .Wi / is a dense subset of Vi .27. U D ' 1 .P /.U 0 / is open. if Z \ C contains a dense open subset of Z. According to (9.V /). 2 T HEOREM 9. A flat map 'W W ! V of algebraic varieties is open.U / is flat.U / is closed.U / is open.U /. Hence Z  C .W / dim. and so it contains a nonempty open subset U of Z (9. and ' U ! '.24) there exists a chain of irreducible closed subsets W  W1      Wm D Z such that '. and so (see 9. ' and U 0 ! '. Let U 0 D ' 1 .V /: After replacing V with an open neighbourhood of P and W with an open subset intersecting V . In particular.U /.21) implies that V X '. Now '. Let U be an open subset of W . '. there exists a dense open subset U of V such that ' 1 . Then  dim ' 1 .'. Then U 0 is a dense open subset of W .U / is constructible (9. Let V  V1      Vm D fP g be a maximal chain of distinct irreducible closed subsets of V (so m D dim.7) and the goingdown theorem (9.26. if Z is an irreducible component of ' 1 . (37) and equality holds if ' is flat.' 1 .U // D ' 1 . Flat maps and their fibres 205 L EMMA 9. ' P ROOF.U / satisfies the hypotheses of the lemma. The inequality was proved in (9. 2 Fibres and flatness The notion of flatness allows us to sharpen our earlier results. then dim. 2 C OROLLARY 9. If ' is flat. then Z  C .

2 P ROPOSITION 9. Hochschild 1981.V /.W / dim. 2 P ROPOSITION 9. The inequality can be proved by a similar argument to that in (9. then dim.'U /. The equality can be deduced by the same argument as in (9.T / D dim.W / dim. Springer.32. 1981.206 9.U / is open in V . Let P 2 '. Let 'W W ! V be a dominant map of irreducible algebraic varieties.S / such that '. and let T be an irreducible component of ' 1 . 2 P ROPOSITION 9. this implies that we have equality. Let V be an irreducible algebraic variety.U / is flat.V / C dim. Gerhard P. There exists ' a nonempty open subset U of W such that '.Z/: Together with (37).W /  m C n D dim. Let S be a closed irreducible subset of V .S / C dim. P ROOF.9) — see. 1 .2 X. and U ! '. Then dim.S / C dim. Let U be an open subset of W as in (9. It suffices to prove this with V affine.29). Let 'W W ! V be a dominant map of irreducible varieties.28).T / is dense in S .'. Let 'W W ! V be a dominant map of irreducible varieties. for example. and T is an irreducible component of ' 1 .33.V /: There exists a dense open subset U of W such that '.28). . U D ' and equality holds for all P 2 '. Theorem 2.V /: P ROOF.5 (equivalence of (d) and (e)). and equality holds if ' is flat. 2 F INITE MAPS P ROPOSITION 9.31.T /  dim. Basic theory of algebraic groups and Lie algebras. Let U be an open subset of W as in (9. The existence of the chain W  W1      Wm  Z1      Zn shows that dim.W / dim. Then  dim ' 1 . A finite map 'W W ! V is flat if and only if X dimk OQ =mP OQ Q7!P is independent of P 2 W .U / is open. P ROOF.S / meeting U . 2 2 Hochschild. in which case it follows from CA 12..30.U //.V /.U /.P /  dim. R EGULAR M APS AND T HEIR F IBRES be a maximal chain of distinct irreducible closed subsets of V (so n D dim. U D ' 1 .Z/). P ROOF.U /.1. If S is a closed irreducible subset of V meeting '.

134). B is Cohen-Macaulay. P Therefore Q7!P dimk OQ =mP OQ D m for every P . p. The theorem says that a finite map is flat if and only if the number of points in each fibre (counting multiplicities) is constant. Let 'W A ! B be a local homomorphism of noetherian local rings. P ROOF. Tn  and let 'W V ! An be the projection map (see p. P ROPOSITION 9. c/ with c a root of the polynomial X m C a1 . CUP 1986. and let m be the maximal ideal of A. : : : . but a prime ideal can equal a finite intersection of ideals only if it equals one of the ideals. c/ in its fibre is the multiplicity of c as a root of the polynomial. and dim.B/ D dim. p. then ' is flat. (b) Zero-dimensional and reduced one-dimensional noetherian rings are Cohen-Macaulay (ibid. Then B is Cohen-Macaulay if and only if both A and B=mB are Cohen-Macaulay (ibid.18). P ROOF. If A is regular.1. Conversely. 23. The fibre over a point P of An is the set of points . for every maximal ideal m of A (9.181). and that ht.n/ D ht. and so the map ' is flat. but merely list some of its properties.P. For example. H.36. Note that dim.P / D 0: The multiplicity of . Flat maps and their fibres 207 The integer dimk OQ =mP OQ is the multiplicity of Q in its fibre.P /X m 1 C    C am . p. Let 'W A ! B be a finite homomorphism noetherian rings with A regular.49). then ' is flat by (9.35d). 2 9.3 hence Cohen-Macaulay. then B is Cohen-Macaulay by (9. and let m be the maximal ideal of A. if B is Cohen-Macaulay. p. let V be the subvariety of AnC1 defined by an equation X m C a1 X m 1 C    C am D 0.137). Therefore it has only finitely many maximal ideals.B=mB/ is zero-dimensional.34). If ' is flat.. Commutative Ring Theory.nc / for every maximal ideal n of B. We don’t define notion of being Cohen-Macaulay here (see ibid. Every prime ideal in C is an intersection of maximal ideals (2.139). Matsumura.34. (c) Regular noetherian rings are Cohen-Macaulay (ibid. . Then ' is flat if and only if B is Cohen-Macaulay.P.B=mB/. (a) A noetherian ring A is Cohen-Macaulay if and only if Am is Cohen-Macaulay for every maximal ideal m of A (this is part of the definition).35b).A/ C dim. 2 3 Note def that C D B=mB D B ˝A A=m is a finite k-algebra. Criteria for flatness T HEOREM 9.c. (d) Let 'W A ! B be a flat local homomorphism of noetherian local rings. ai 2 kŒT1 .35.

CA 23. Rather than using exterior algebras. the lines in P3 should form a four-dimensional set. y/ 7! .P / D dim W dim V (38) for all P 2 V . Immediate consequence of (9. 2 def A SIDE 9.39. that is. pij D ˇ . Moreover ' is finite. The quotient map 'W W ! V is .x. Not every point in P5 should be of the form p.L/ D . : : : . to each line L in P3 . d Lines on surfaces As an application of some of the above results.L/ in P5 .33). and V is normal because kŒV  D kŒX. We first need a way of describing lines in P3 . but it is not flat because  X 3 if P D . 0/ dimk OQ =mP OQ D 2 otherwise Q7!P (see 9. An algebraic variety V is said to be Cohen-Macaulay if OV. 2.) We shall show that there is natural one-to-one correspondence between the set of lines in P3 and the set of points on a certain hyperspace ˘  P5 .12). we can attach point p. Then ˇ ˇ xj ˇˇ def ˇ xi 5 ˇ p. (Fix two planes in P3 . y/ 7! . Let V and W be algebraic varieties with V nonsingular and W CohenMacaulay. The theorem fails with “nonsingular” weakened to “normal”. P ROOF.x. giving a line in P3 corresponds to choosing a point on each of the planes. and so its fibres have constant dimension 0.L/—heuristically. Let A be a finite kŒX1 . x. we consider the problem of describing the set of lines on a surface of degree m in P3 .45). Let L be a line in P3 and let x D . we assume for the rest of this chapter that k has characteristic zero.34). .x 2 . The quotient of W by this action is the quadric cone V  A3 defined by T V D U 2 . Let Z=2Z act on W D A2 by . Xn  ! A is flat if and only if A is Cohen-Macaulay. 0. The variety W is nonsingular. T HEOREM 9.38.t. y/. The map kŒX1 . after Pl¨ucker (18011868). To such a subspace W  k 4 . See mo117043. xy.0. The pij are called the Pl¨ucker coordinates of L. R EGULAR M APS AND T HEIR F IBRES E XAMPLE 9.208 9.35c). An affine algebraic variety V is Cohen-Macaulay if and only if kŒV  is Cohen-Macaulay (9. we can attach a one-dimensional subspace V2 V W in 2 k 4  k 6 . A nonsingular variety is Cohen-Macaulay (9.37.p01 W p02 W p03 W p12 W p13 W p23 / 2 P . v/ D . u. and lines in P3 as being two-dimensional subspaces in k 4 .35a). A regular map 'W W ! V is flat if and only if dim ' 1 . Y G (cf. This allows us to think of points in P3 as being one-dimensional subspaces in k 4 .y0 W y1 W y2 W y3 / be distinct points on L. To avoid possible problems. Xn -algebra (cf. : : : . Recall that we can associate with each projective variety V  Pn an affine cone over VQ in k nC1 .x0 W x1 W x2 W x3 / and y D . yi yj ˇ depends only on L. y 2 /. I shall usually give the old-fashioned proofs.P is Cohen-Macaulay for all P 2 V .

L/ lies on the quadric ˘  P5 defined by the equation X01 X23 X02 X13 C X03 X12 D 0: P ROOF.X0 .41. x2 .a0 W a1 W a2 W a3 / 2 P3 g: Q be the cone over L in k 4 —it is a two-dimensional subspace of k 4 —and let P ROOF.P e2 . aj p2j .42. 0  i. X2 . write e0 . X F . We define pij for all i.L/ for a unique line L.p01 W p02 W p03 W p12 W p13 W p23 /: A similar construction works when one of the other coordinates is nonzero. y1 . Every point of ˘ is of the form p.L/ as follows: P P P P L D f. aj p0j . e1 . x3 / and y D . j  3 by the same formula — thus pij D pj i . 2 Thus we have a canonical one-to-one correspondence flines in P3 g $ fpoints on ˘ gI that is.0 W p01 W p02 W p03 / and . y2 . so that x. e3 for theV canonical basis for k 4 . We may now use the methods of algebraic geometry to study the set. L EMMA 9. Let L Q Then x D . x1 . X1 . Assume p03 ¤ 0. P ROOF. Q D ff . then space with basis ei ^ ej .p01 p23 p02 p13 C p03 p12 / ˇ 0Dˇ ˇ ˇ x0 x1 x2 x3 ˇ ˇ y0 y1 y2 y3 ˇ (expansion in terms of 2  2 minors). j aj p0j W j aj p1j W j aj p2j W j aj p3j / j . and x^ y D pij ei ^ ej with pij given by the above formula.) We next consider the set of homogeneous polynomials of degree m in 4 variables.40. P 2 4 k is a 6-dimensional vector regarded as a point of k 4 is xi ei . 0  i < j  3. 2 L EMMA 9. and y D yi eP i . This can be verified by direct calculation.x/y j f W k 4 ! k linearg: aj Xj .y/x P P P P f . or by using that ˇ ˇ ˇ x0 x1 x2 x3 ˇ ˇ ˇ ˇ y0 y1 y2 y3 ˇ ˇ D 2. Lines on surfaces 209 In terms of exterior algebras.d.x0 . and this way we get inverse maps.x/y D . then the line through the points . aj p3j /: 2 L EMMA 9.y0 . aj p1j . The point p. p01 p03 W p02 p03 W p03 W p01 p23 p02 p13 W p03 p13 W p03 p23 / „ ƒ‚ … p03 p12 D . The line L can be recovered from p. y3 / be two linearly independent vectors in L. X3 / D ai0 i1 i2 i3 X0i0 : : : X3i3 : i0 Ci1 Ci2 Ci3 Dm . j . then f . we have identified the set of lines in P3 with the points of an algebraic variety.p03 W p13 W p23 W 0/ has Pl¨ucker coordinates 2 . (This is a special case of the Grassmannians discussed in 6.y/x L Write f D P f .

44. The theorem says that the dimension of 1 is 5. : : : I p01 . We first show that m is closed. See the footnote p.45. the discussion in 6.F /.: : : . FD ai0 i1 i2 i3 X0i0 X1i1 X2i2 X3i3 : The map is not quite injective—for example. bj p0j W bj p1j W bj p2j W bj p3j / D 0. R EGULAR M APS AND T HEIR F IBRES L EMMA 9. X1 . We shall apply Proposition 9. 6 E XAMPLE 9. X2 . X1 .mC1/. Then we have a surjective map .mC3/ m 6 attached to the vector space of homogeneous polynomials of degree m in 4 variables (p.43. It remains to compute the dimension of m .: : : W ai0 i1 i2 i3 P ! fsurfaces of degree m in P3 g. X3 / D 0 if and only if P P P P F .mC1/. X 2 Y and X Y 2 define the same surface— but nevertheless. X2 .L. F / m  ˘  P ' L ˘ .40) we see that L lies on the surface F .L/ D . and so the set is closed in ˘  P (cf. we can (somewhat loosely) think of the points of P as being (possibly degenerate) surfaces of degree m in P3 .141.X0 . F / consisting of a line L in P3 lying on the surface F . For m D 1. Since there are 13 planes in P3 . P ROOF. The set ofhomogeneous polynomials of degree m in 4 variables is a vector space of dimension 3Cm m P ROOF.210 9. T HEOREM 9. it is therefore a projective variety. The set m is a irreducible closed subset of ˘  P . 2  1. : : : . and regard P as the projective space Let  D 3Cm 1 D . The dimension of m is m.L.146). all . Then F .mC2/. X3 / D 0. Let m  ˘  P  P5  P be the set of pairs . X W : : :/ 7! V .51). m is the set of pairs consisting of a plane in P3 and a line on the plane. ai0 i1 i2 i3 . b3 / 2 k 4 : Expand this out as a polynomial in the bj ’s with coefficients polynomials in the ai0 i1 i2 i3 and pij ’s.mC5/ C 3. Let p.11 to the projection map .X0 . p02 W : : :/ with P homogeneous separately in the a’s and p’s. this seems to be correct.b0 . But each coefficient is of the form P . and each has 12 lines on it.p01 W p02 W : : :/ FD X ai0 i1 i2 i3 X0i0    X3i3 : From (9.:::/ D 0 for all b 2 k 4 if and only if the coefficients of the polynomial are all zero.

We now compute its dimension. A quadric is said to be nondegenerate if it is defined by an irreducible polynomial of degree 2.m C 1/.m C 5/ 6 1: is irreducible and that D dim. Then L lies on F . 6 as claimed. F 2' 1 . When m > 3. the surfaces of degree m containing no line correspond to an open subset of P . if m > 3.9) tells us that dim fact of course.47. then dim P . the lines on any plane form a 2-dimensional family.F /  2. and so ' 1 .L/ is a linear subspace of P .. m . ! P3 is 1 .m C 1/. m/ . i. Recall that F has  C 1 coefficients altogether. and  D 9. and so all F . the number with i0 D 0 D i1 is m C 1.m C 1/ D m.X0 . X2 .m C 3/ C1 D 4 6 < dim P . P ROOF. We have dim m dim P D m.' 1 . and (9. we can assume that L is the line  X0 D 0 X1 D 0. In 1 . Here dim m D 10. .L// D m.L/ ” ai0 i1 i2 i3 D 0 whenever i0 D 0 D i1 : Thus ' 1 . This proves the claim. L D f. X1 . Let m D 1.m C 2/. Then  D 3 and dim 1 D 5.d.L/ has dimension . which suggests that should be surjective and that its fibres should all have dimension  1. After a change of variables.m C 5/ C3 6 Therefore.m C 1/. After a change of coordinates.m C 1/: is a proper closed subvariety of 2 We now look at the case m D 2. ' 1 .F / (taken up to nonzero scalars). and so .m C 1/. 2 Now consider the other projection By definition 1 . 0.F /g: E XAMPLE 9. Lines on surfaces 211 For L 2 ˘ .e.m C 3/ 6 We can now deduce from (9.˘ / C dim. i. X3 / D 0 if and only if X0 or X1 occurs in each nonzero monomial term in F . We shall see that this is correct.0.11) that dim.e. it is irreducible. m/ 1 m . /g. The projection surjective (every plane contains at least one line).m C 1/.L/ consists of the homogeneous polynomials of degree m such that L  V ..46.m C 5/ C 3. any nondegenerate quadric will be defined by an equation X W D Y Z: .F / D fL j L lies on V . in particular.m C 2/.F / D 2 for W 1 T HEOREM 9.

a0 W a1 /.b0 W b1 / 7! . There are the following lines on the surface.9) tells us that all the fibres have dimension  1. (b) Consider the surface X1 X2 X3 D X03 : . one defines a regulus to be a nondegenerate quadric surface together with a choice of a pencil of lines.S / 2 # R # P9 D pairs. in a natural way. the horizontal family and the vertical family. We now look at the case m D 3. They map to two families of lines on the quadric:   t0 X D t1 Z t0 X D t1 Y and t0 Y D t1 W t0 Z D t1 W: Since a degenerate quadric is a surface or a union of two surfaces. over the set of nondegenerate quadrics. for a total of 27 lines. One can show that the set of reguli is. 0  i. The set of cubic surfaces containing exactly 27 lines corresponds to an open subset of P19 . an algebraic variety R.49. and the set where the dimension is > 1 is a proper closed subset.212 9. E XAMPLE 9. we see that every quadric surface contains a line. j  2:    X0 C  i X1 D 0 X0 C  i X2 D 0 X0 C  i X3 D 0 j j X2 C  X3 D 0 X1 C  X3 D 0 X1 C  j X2 D 0 There are three sets. each with nine lines. 1 . Classically. then 1 1 1 morphic to the disjoint union of two lines. Here dim 3 D 19. the remaining surfaces either contain an infinite number of lines or a nonzero finite number  27.F / is isoIt follows from the above discussion that if F is nondegenerate.38) below).26) .F. . namely.48. that is. factors into the composite of two regular maps: 1 .F /  P [ P . (a) Consider the Fermat surface X03 C X13 C X23 C X33 D 0: Let  be a primitive cube root of one. S D set of nondegenerate quadrics. R EGULAR M APS AND T HEIR F IBRES This is just the image of the Segre mapping (see 6. Thus (9. .  D 19 W we have a map W 3 ! P19 : T HEOREM 9.a0 b0 W a0 b1 W a1 b0 W a1 b1 / W P1  P1 ! P3 : There are two obvious families of lines on P1  P1 . which we know to be closed (this was a homework problem in the original course). L/ with L on F I D set of reguli. that W 2 ! P9 is surjective. Factorizations of this type occur quite generally (see the Stein factorization theorem (9. and that. and the second map is finite and two-to-one. and of dimension 1 (they are all isomorphic to P1 /. . and so is called a pencil of lines. The fibres of the top map are connected. each is parametrized by P1 . In fact the dimension of the fibre is > 1 exactly on the set of reducible F ’s.

in fact. The set of hyperplanes H not containing X and 4 According contains a line. it is not hard to show that there is an open subset U 0 where the cubics have exactly 27 lines (see Reid. Then “in general” a line joining a point on L to a point on L0 will meet the surface in exactly one further point. In this way one obtains an invertible regular map from an open subset of P1  P1 to an open subset of V ./ is isomorphic to the Weyl group of the root system of a simple Lie algebra of type E6 . and have been much studied—see A. which contradicts two of the above examples. pp106–110).40) to see that all cubics in U to U 0 have fewer than 27 lines. R EMARK 9. the fibres have dimension > 0. The Twenty-Seven Lines Upon the Cubic Surface. Let L and L0 be two skew lines. X1 D 0. Y D t. The twenty-seven lines on a cubic surface were discovered in 1849 by Salmon and Cayley. and hence V is birationally equivalent to P2 . the surface contains exactly three lines. to Miles Reid (1988. which contains the line X D t. The intersection of the surface with this plane is given by X1 X2 X3 D 0 (homogeneous coordinates). If W 3 ! P19 were not surjective. (c) Consider the surface X13 C X23 D 0: Here there is a pencil of lines:  t0 X1 D t1 X0 t0 X2 D t1 X0 : (In the affine space where X0 ¤ 0. it is known that the group of permutations of the set of 27 lines preserving intersections (that is. Henderson. Now look where X0 D 0. U 0 can be taken to be the set of nonsingular cubics. and hence has 25920 elements. and the nonempty fibres would all have dimension  1 (by 9. such that L \ L0 ¤ . and so we can apply (8. there are exactly three lines. It is known that there is a set of 6 skew lines on a nonsingular cubic surface V . outside U . X2 D 0.9).48). Bertini’s theorem 213 In this case. For example. in the plane at infinity. look first in the affine space where X0 ¤ 0—here we can take the equation to be X1 X2 X3 D 1. X3 D 0: Therefore. but a direct inspection shows that no such line lies on the surface. Cambridge University Press.26). all t:/ We now discuss the proof of Theorem 9. A line in A3 can be written in parametric form Xi D ai t C bi . To see this. and there is an open subset U of P19 where the fibres have dimension 0.e. p126) every adult algebraic geometer knows the proof that every cubic .L/ \  . which is the union of three lines.50.L0 / ¤ . 1911. According to (8. the equation is X 3 C Y 3 D 0. namely. Given that every cubic surface has at least one line. e Bertini’s theorem Let X  Pn be a nonsingular projective variety. that is. then . 1988. the restriction of 1 . ”  . 3 / would be a proper closed subvariety of P19 . The hyperplanes H in Pn form a projective space Pn_ (the “dual” projective space).U / is finite. Therefore the map is surjective4 .

or (b) a surface of the form C  P1 for a unique nonsingular projective curve C . A nonsingular projective surface is said to be minimal if it can not be obtained from another such surface by blowing up. the functor V k. strictly lower dimension. Exercises 9-1. even projective and nonsingular.W /. According to the theory of Jordan canonical forms.g 0 v/ for all g. will have the same function field. and that there exists at least one closed orbit. Let G D GL2 D V . repeated roots. If dim. This means that the varieties themselves become isomorphic once a proper closed subset has been removed from each (3. The main theorem for surfaces (Enriques 1914. the problem becomes very involved.V / is a contravariant equivalence from the category of nonsingular projective algebraic curves over k and dominant maps to the category of fields finitely generated and of transcendence degree 1 over k. f Birational classification Recall that two varieties V and W are birationally equivalent if k. although much progress has been made — see Wikipedia M INIMAL MODEL PROGRAM . minimal polynomial the same. . the orbits are of three types: (a) Characteristic polynomial X 2 C aX C b. R EGULAR M APS AND T HEIR F IBRES such that X \ H is nonsingular. For surfaces. For each type. the closure of the orbit. the equations defining it (as a subvariety of V ). every blow-up of a point on a surface produces a birationally equivalent surface. and which other orbits are contained in the closure. Deduce that each orbit is a nonsingular subvariety of V . Let G be a connected group variety. 9-2. Show that N and that ON X O is a union of orbits of each orbit O D Gv of G is open in its closure O. i. form an open subset of Pn_ .gg 0 /v D g.e. Kodaira 1966) says that a birational equivalence class contains either (a) a unique minimal surface.36). and let G act on V by conjugation. then the intersections X \ H are connected.X /  2. More precisely. (b) Characteristic polynomial X 2 C aX C b. g 0 2 G and v 2 V . (c) Characteristic polynomial X 2 C aX C b D . For example. For curves this is easy: in each birational equivalence class there is exactly one nonsingular projective curve (up to isomorphism)..V /  k. distinct roots. In higher dimensions. a regular map G  V ! V such that . the problem is already much more difficult because many surfaces. The main problem of birational algebraic geometry is to classify algebraic varieties up to birational equivalence by finding a particularly good representative in each equivalence class. minimal polynomial X ˛. and consider an action of G on a variety V .X ˛/2 .214 9. find the dimension of the orbit.

) [The enthusiasts may wish to carry out the analysis for GLn . if you wish. and let 'W W ! V be G-equivariant regular map satisfying the hypotheses in (a). . d  3. Show that there are open subsets U 0 and U of W and V such that '. 9-4.] 9-3. p the characteristic. Find 3d 2 lines on the Fermat projective surface X0d C X1d C X2d C X3d D 0.p. you may assume the following (fairly difficult) result: for any closed subgroup H of an group variety G. that the characteristic is zero.U 0 /  U and 'W U 0 ! U is finite. and a map G=H ! V is regular if the composite G ! G=H ! V is regular. Birational classification 215 (You may assume. Then ' is finite. Also. . (a) Let 'W W ! V be a quasi-finite dominant regular map of irreducible varieties. G=H has a natural structure of an algebraic variety with the following properties: G ! G=H is regular. d / D 1. dim G=H D dim G dim H . (b) Let G be a group variety acting transitively on irreducible varieties W and V .f. and hence proper.

.

0  dim Z  2.Z/ D . : : : . by the degree-one case. then dim Z D 0 by (2. unless the characteristic of k is 2. Xn  ! kŒT  induces an isomorphism kŒV  ! A1 . : : : . Z C 1/. 2-3 Clearly V D V .0/.X. : : : .] 2-4 We use that the prime ideals are in one-to-one correspondence with the irreducible closed subsets Z of A2 .Solutions to the exercises 1-1 Use induction on n. can be interpreted as saying that a certain set of polynomials has a zero in k.r C 1/  .X 2 .a. : : : does not contain 1.X C 2Y. 2-7 Regard HomA . X Y 2 /  . and so I. for example).X // D . then the ideal they generate in CŒX1 . Case dim Z D 2. Now P follows i suppose n > 1 and write f D gi Xn with each gi 2 kŒX1 . an 1 .Xn X1n . Then Z ¤ A2 . Xn 1 . by induction. Now. then some gi is not the zero polynomial.X / D rad. Z/. If f is not the zero polynomial. A fortiori the ideal they generate in QŒX1 . and the set of matrices with rank  r is the set where all . : : : .. Case dim Z D 1.a1 . Hence I.t. .f /. . X Y 2 /. there exists a b such that f .X 2 .X /.X 2 . Gaussian elimination (to reduce the matrix of coefficients to row echelon form).64. b/ (see 2. [Hence t 7! . The map Xi 7! T i W kŒX1 . Y /. the elements not isomorphisms are the zeros of a polynomial.W / D . B ˝Q k/ ¤ . 1-2 . If the polynomials have a common zero in C.a1 . and so I. X2 X12 /.63).1/.Z/ contains a nonzero polynomial f .11) implies that the polynomials have a common zero in k.X C 1. the polynomial is not identically zero. Xn / is not the zero polynomial. we find that .Z/ D . there exist . t n / is an isomorphism of affine varieties A1 ! V . an 1 .A ˝Q k.a1 . and so I. an 1 / 2 k n 1 such that f . and so the Nullstellensatz (2. Then Z is a point . For such a set. 2-1 W D Y -axis.X /. Then Z D A2 . Case dim Z D 0.Z/ ¤ . in which case the ideal is . X Y 2 / and rad. use that a nonzero polynomial in one variable has only finitely many roots (which from unique factorization. : : : . 2. If I.r C 1/ minors are zero. : : : does not contain 1. 217 . 2-2 The d  d minors of a matrix are polynomials in the entries of the matrix.62). therefore it has a nonzero in k (Exercise 1-1). 2-6 The statement Homk algebras . On taking radicals. : : : .f /. : : : . and the corresponding ideal is . N / as an affine space over k. For n D 1. b/ ¤ 0.X a. Clearly. Y b/.X /  rad. Therefore. because M and N become isomorphic over k al .M.

218

9. R EGULAR M APS AND T HEIR F IBRES

3-1 A map ˛W A1 ! A1 is continuous for the Zariski topology if the inverse images of finite
sets are finite, whereas it is regular only if it is given by a polynomial P 2 kŒT , so it is easy
to give examples, e.g., any map ˛ such that ˛ 1 .point/ is finite but arbitrarily large.
3-2 The argument in the text shows that, for any f 2 S ,
q

f .a1 ; : : : ; an / D 0 H) f .a1 ; : : : ; anq / D 0:
This implies that ' maps V into itself, and it is obviously regular because it is defined by
polynomials.
3-3 The image omits the points on the Y -axis except for the origin. The complement of the
image is not dense, and so it is not open, but any polynomial zero on it is also zero at .0; 0/,
and so it not closed.
3-5 No, because both C1 and 1 map to .0; 0/. The map on rings is
kŒx; y ! kŒT ;

x 7! T 2

1;

y 7! T .T 2

1/;

which is not surjective (T is not in the image).
5-1 Let f be regular on P1 . Then f jU0 D P .X / 2 kŒX , where X is the regular function
.a0 W a1 / 7! a1 =a0 W U0 ! k, and f jU1 D Q.Y / 2 kŒY , where Y is .a0 W a1 / 7! a0 =a1 . On
U0 \ U1 , X and Y are reciprocal functions. Thus P .X / and Q.1=X / define the same
function on U0 \ U1 D A1 X f0g. This implies that they are equal in k.X /, and must both be
constant.
F
Q
5-2 Note that .V; OV / D
.Vi ; OVi / — to give a regular function on Vi is the same
as to give a regular function on each
Thus, if
Q Vi (this is the “obvious” ringed space structure).
F
V is affine, it must equal Specm. Ai /, where Ai D .Vi ; OVi /, and so V D Specm.Ai /
(use the description of the ideals in A  B on in Section 1a). Etc..
5-5 Let H be an algebraic subgroup of G. By definition, H is locally closed, i.e., open in
its Zariski closure HN . Assume first that H is connected. Then HN is a connected algebraic
group, and it is a disjoint union of the cosets of H . It follows that H D HN . In the general
case, H is a finite disjoint union of its connected components; as one component is closed,
they all are.
4-1 (b) The singular points are the common solutions to
8
H) X D 0 or Y 2 D 2X 2
< 4X 3 2X Y 2 D 0
3
2
4Y
2X Y D 0
H) Y D 0 or X 2 D 2Y 2
: 4
X C Y 4 X 2 Y 2 D 0:
Thus, only .0; 0/ is singular, and the variety is its own tangent cone.
4-2 Directly from the definition of the tangent space, we have that
Ta .V \ H /  Ta .V / \ Ta .H /.
As
dim Ta .V \ H /  dim V \ H D dim V

1 D dim Ta .V / \ Ta .H /;

we must have equalities everywhere, which proves that a is nonsingular on V \ H . (In
particular, it can’t lie on more than one irreducible component.)

f. Birational classification

219

The surface Y 2 D X 2 C Z is smooth, but its intersection with the X-Y plane is singular.
No, P needn’t be singular on V \ H if H  TP .V / — for example, we could have
H  V or H could be the tangent line to a curve.
4-4 We can assume V and W to affine, say
I.V / D a  kŒX1 ; : : : ; Xm 
I.W / D b  kŒXmC1 ; : : : ; XmCn :
If a D .f1 ; : : : ; fr / and b D .g1 ; : : : ; gs /, then I.V  W / D .f1 ; : : : ; fr ; g1 ; : : : ; gs /. Thus,
T.a;b/ .V  W / is defined by the equations
.df1 /a D 0; : : : ; .dfr /a D 0; .dg1 /b D 0; : : : ; .dgs /b D 0;
which can obviously be identified with Ta .V /  Tb .W /.
4-5 Take C to be the union of the coordinate axes in An . (Of course, if you want C to be
irreducible, then this is more difficult. . . )
4-6 A matrix A satisfies the equations
.I C "A/tr  J  .I C "A/ D I
if and only if
Atr  J C J  A D 0: 

Such an A is of the form

M
P 

N
with M; N; P; Q n  n-matrices satisfying
Q

N tr D N;

P tr D P;

M tr D Q.

The dimension of the space of A’s is therefore
n.n C 1/
n.n C 1/
(for N ) C
(for P ) C n2 (for M; Q) D 2n2 C n:
2
2
4-7 Let C be the curve Y 2 D X 3 , and consider the map A1 ! C , t 7! .t 2 ; t 3 /. The
corresponding map on rings kŒX; Y =.Y 2 / ! kŒT  is not an isomorphism, but the map on
the geometric tangent cones is an isomorphism.
4-8 The singular locus Vsing has codimension  2 in V , and this implies that V is normal.
[Idea of the proof: let f 2 k.V / be integral over kŒV , f … kŒV , f D g= h, g; h 2 kŒV ;
for any P 2 V .h/ X V .g/, OP is not integrally closed, and so P is singular.]
4-9 No! Let a D .X 2 Y /. Then V .a/ is the union of the X and Y axes, and I V .a/ D .X Y /.
For a D .a; b/,
.dX 2 Y /a D 2ab.X
.dX Y /a D b.X

a/ C a2 .Y
a/ C a.Y

If a ¤ 0 and b D 0, then the equations
.dX 2 Y /a D a2 Y D 0
.dX Y /a D aY D 0

b/.

b/

220

9. R EGULAR M APS AND T HEIR F IBRES

have the same solutions.
6-1 Let P D .a W b W c/, and assume c ¤ 0. Then the tangent line at P D . ac W bc W 1/ is   

  

     
@F
@F
@F
a
@F
b
XC
Y
C
Z D 0:
@X P
@Y P
@X P c
@Y P c
Now use that, because F is homogeneous,   

 

@F
@F
@F
F .a; b; c/ D 0 H)
aC
C
c D 0.
@X P
@Y P
@Z P
(This just says that the tangent plane at .a; b; c/ to the affine cone F .X; Y; Z/ D 0 passes
through the origin.) The point at 1 is .0 W 1 W 0/, and the tangent line is Z D 0, the line at 1.
[The line at 1 meets the cubic curve at only one point instead of the expected 3, and so the
line at 1 “touches” the curve, and the point at 1 is a point of inflexion.]
6-2 The equation defining the conic must be irreducible (otherwise the conic is singular).
After a linear change of variables, the equation will be of the form X 2 C Y 2 D Z 2 (this is
proved in calculus courses). The equation of the line in aX C bY D cZ, and the rest is easy.
[Note that this is a special case of Bezout’s theorem (6.37) because the multiplicity is 2 in
case (b).]
6-3 (a) The ring
kŒX; Y; Z=.Y

X 2; Z

X 3 / D kŒx; y; z D kŒx ' kŒX ;

which is an integral domain. Therefore, .Y X 2 ; Z X 3 / is a radical ideal.
(b) The polynomial F D Z X Y D .Z X 3 / X.Y X 2 / 2 I.V / and F  D ZW
XY . If
ZW X Y D .Y W X 2 /f C .ZW 2 X 3 /g;
then, on equating terms of degree 2, we would find
ZW

X Y D a.Y W

X 2 /;

which is false.
6-4 Let P D .a0 W : : : W anP
/ and Q D .b0 W : : : W bn / be two points of Pn , n  2. The condition
that the hyperplane Lc W ci Xi D 0 pass through P and not through Q is that
P
P
ai ci D 0;
bi ci ¤ 0:
The .n C 1/-tuples P
.c0 ; : : : ; cn / satisfying these conditions form a nonempty open subset of
the hyperplane H W ai Xi D 0 in AnC1 . On applying this remark to the pairs .P0 ; Pi /, we
find that the .n C 1/-tuples c D .c0 ; : : : ; cn / such that P0 lies on the hyperplane Lc but not
P1 ; : : : ; Pr form a nonempty open subset of H .
6-5 The subset
C D f.a W b W c/ j a ¤ 0;

b ¤ 0g [ f.1 W 0 W 0/g

P2

of
is not locally closed. Let P D .1 W 0 W 0/. If the set C were locally closed, then P
would have an open neighbourhood U in P2 such that U \ C is closed. When we look in
U0 , P becomes the origin, and
C \ U0 D .A2 X fX -axisg/ [ foriging.

f. Birational classification

221

The open neighbourhoods U of P are obtained by removing from A2 a finite number of
curves not passing through P . It is not possible to do this in such a way that U \ C is closed
in U (U \ C has dimension 2, and so it can’t be a proper closed subset of U ; we can’t have
U \ C D U because any curve containing all nonzero points on X-axis also contains the
origin).
P
6-6 Let cij Xij D 0 be a hyperplane containing the image of the Segre map. We then have
P
cij ai bj D 0
for all a D .a0 ; : : : ; am / 2 k mC1 and b D .b0 ; : : : ; bn / 2 k nC1 . In other words,
aC bt D 0
for all a 2 k mC1 and b 2 k nC1 , where C is the matrix .cij /. This equation shows that
aC D 0 for all a, and this implies that C D 0.
7-2 Define f .v/ D h.v; Q/ and g.w/ D h.P; w/, and let ' D h .f ı p C g ı q/. Then
'.v; Q/ D 0 D '.P; w/, and so the rigidity theorem (7.35) implies that ' is identically zero.
8-2 For example, consider
x7!x n

.A1 X f1g/ ! A1 ! A1
for n > 1 an integer prime to the characteristic. The map is obviously quasi-finite, but it is
not finite because it corresponds to the map of k-algebras
X 7! X n W kŒX  ! kŒX; .X
which is not finite (the elements 1=.X
so also over kŒX n ).

1/

1



1/i , i  1, are linearly independent over kŒX , and

8-3 Assume that V is separated, and consider two regular maps f; gW Z  W . We have
to show that the set on which f and g agree is closed in Z. The set where ' ı f and
' ı g agree is closed in Z, and it contains the set where f and g agree. Replace Z
with the set where ' ı f and ' ı g agree. Let U be an open affine subset of V , and let
Z 0 D .' ı f / 1 .U / D .' ı g/ 1 .U /. Then f .Z 0 / and g.Z 0 / are contained in ' 1 .U /,
which is an open affine subset of W , and is therefore separated. Hence, the subset of Z 0 on
which f and g agree is closed. This proves the result.
[Note that the problem implies the following statement: if 'W W ! V is a finite regular
map and V is separated, then W is separated.]
8-4 Let V D An , and let W be the subvariety of An  A1 defined by the polynomial
Qn
Ti / D 0:
i D1 .X
Q
The fibre over .t1 ; : : : ; tn / 2 An is the set of roots of .X ti /. Thus, Vn D An ; Vn 1 is the
union of the linear subspaces defined by the equations
Ti D Tj ;
Vn

2

1  i; j  n;

i ¤ jI

is the union of the linear subspaces defined by the equations
Ti D Tj D Tk ;

1  i; j; k  n;

i; j; k distinct,

222

9. R EGULAR M APS AND T HEIR F IBRES

and so on.
9-1 Consider an orbit O D Gv. The map g 7! gvW G ! O is regular, and so O contains an
open subset U of ON (9.7). If u 2 U , then gu 2 gU , and gU is also a subset of O which is
open in ON (because P 7! gP W V ! V is an isomorphism). Thus O, regarded as a topological
N contains an open neighbourhood of each of its points, and so must be open
subspace of O,
N
in O.
We have shown that O is locally closed in V , and so has the structure of a subvariety.
From (4.37), we know that it contains at least one nonsingular point P . But then gP is
nonsingular, and every point of O is of this form.
From set theory, it is clear that ON X O is a union of orbits. Since ON X O is a proper
N all of its subvarieties must have dimension < dim ON D dim O.
closed subset of O,
N
Let O be an orbit of lowest dimension. The last statement implies that O D O.
9-2 An orbit of type (a) is closed, because it is defined by the equations
Tr.A/ D a;

det.A/ D b;  

˛ 0
(as a subvariety of V ). It is of dimension 2, because the centralizer of
, ˛ ¤ ˇ, is
0 ˇ   

0
, which has dimension 2.
0 
An orbit of type (b) is of dimension 2, but is not closed: it is defined by the equations  

˛ 0
Tr.A/ D a; det.A/ D b; A ¤
; ˛ D root of X 2 C aX C b.
0 ˛  

˛ 0
An orbit of type (c) is closed of dimension 0: it is defined by the equation A D
.
0 ˛
An orbit of type (b) contains an orbit of type (c) in its closure.
9-3 Let  be a primitive d th root of 1. Then, for each i; j , 1  i; j  d , the following
equations define lines on the surface   

X0 C  i X1 D 0
X0 C  i X2 D 0
X0 C  i X3 D 0
X2 C  j X3 D 0
X1 C  j X3 D 0
X1 C  j X2 D 0:
There are three sets of lines, each with d 2 lines, for a total of 3d 2 lines.
9-4 (a) Compare the proof of Theorem 9.9.
(b) Use the transitivity, and apply Proposition 8.26.

Index
dimension, 75, 114
of a topological space, 52
of an algebraic set, 53
pure, 53, 114
direct limit, 20
direct system, 20
directed set, 20
discrete valuation ring, 86
divisor, 177
effective, 177
locally principal, 177
positive, 177
prime, 177
principal, 178
support of, 177
domain
factorial, 21
normal, 26
unique factorization, 21

algebra
affine, 65
finite, 11
finitely generated, 11
algebraically dependent, 33
algebraically independent, 33
An , 35
analytic space, 169
axiom
separation, 101
base change, 112
basis
transcendence, 33
birationally equivalent, 73, 116
boundary, 53
codimension, 54
complete intersection
ideal-theoretic, 79
local, 79
set-theoretic, 79
component
of a function, 48
cone, 131
affine over a set, 132
content of a polynomial, 22
convergent, 60
Cramer’s rule, 24
curve, 53
elliptic, 36, 130, 135

element
integral over a ring, 24
irreducible, 21
prime, 21
F .A/, 17
faithfully flat, 201
fibre, 112
field of rational functions, 48, 113
flat, 201, 203
form
leading, 83
function
holomorphic, 169
rational, 63
regular, 47, 61, 100

degree
of a hypersurface, 150
of a map, 184
of a projective variety, 152
derivation, 91
desingularization, 192
differential, 87

generate, 11
germ
223

224
of a function, 60
graph
of a regular map, 110
group
symplectic, 98
group variety, 109
homogeneous, 137
homomorphism
finite, 11
of algebras, 11
hypersurface, 48, 143
hypersurface section, 143
ideal, 12
generated by a subset, 12
graded, 131, 133
homogeneous, 131
maximal, 12
prime, 12
radical, 40
immersion, 104
closed, 72, 104
open, 103
integral closure, 26
integral domain, 11
integrally closed, 26
irreducible components, 46
isolated in its fibre, 187
isomorphic
locally, 97 
.p/, 187
lemma
Gauss’s, 22
Nakayama’s, 14
prime avoidance, 77
Zariski’s, 39
linearly equivalent, 178
local equation, 177
local ring
regular, 15
local system of parameters, 120
manifold
complex, 99
differentiable, 99
topological, 99

I NDEX
map
affine, 193
bilinear, 30
birational, 116, 189
dominant, 49, 72, 115
e´ tale, 117, 119
finite, 49, 75, 179, 182
Frobenius, 69
proper, 159
quasi-finite, 49, 182
rational, 115
regular, 48
Segre, 144
separable, 123, 184
Veronese, 141
minimal surface, 214
morphism
of affine algebraic varieties, 65
of ringed spaces, 64
mP , 40
multiplicity, 207
of a point, 84
n-fold, 53
neighbourhood
e´ tale, 120
nilpotent, 40
node, 84
nondegenerate quadric, 211
normalization, 175, 176
open subset
basic, 48
principal, 48
pencil of lines, 212
Picard group, 178
point
factorial, 177
multiple, 87
nonsingular, 82, 87
normal, 173
ordinary multiple, 84
singular, 87
smooth, 82, 87
with coordinates in a ring, 123
polynomial
Hilbert, 152
homogeneous, 129

112 of algebraic varieties. 169 multiplicative. 40 rational map. 115 real locus. 97 separable degree. 212 resolution of singularities. 60 subring. 83 Spm. 44 noetherian. 132 variety abelian. 99 surface. 59 singular locus. 48 regulus. 87 tensor product of modules. 37 Hilbert Nullstellensatz. 53 T1 space. 157 factorial. 93 coordinate. 66 spm. 31 theorem Bezout’s . 99 separated. 105 tensor. 90 reduced. 13 going-up. 108 of objects. 51 Stein factorization. 129 quasi-affine. 38 Noether normalization. 101 product fibred. 197 sheaf of algebras. 35 analytic. 121 Zariski. 109 monoid. 14 noetherian. 145 radical of an ideal. 166 affine algebraic. 65 of algebraic sets. 47 graded. 60 semisimple group. 150 Grassmann. 83. 70 open affine. 192 resultant. 130 constructible. 93 geometric. 44 tangent cone. 186 topological space irreducible . 177 flag. 83.A/. 162 ring associated graded. 82. 97 Lie algebra. 44 topology e´ tale. 22 prevariety algebraic. 32 projection with centre. 14 normal. 101 Cohen-Macaulay. 100 of affine algebraic varieties. 127 normal. 102 . 186 set (projective) algebraic. 93. 208 complete. 147 group. 104 subvariety. 11 subset 225 algebraic. 104 closed. 66 stalk. 60 section of a sheaf. 65 algebraic. 44 quasicompact.Index primitive. 29 Hilbert basis. 36 regular map. 16 subspace locally closed. 41 Zariski’s main. 34 of dual numbers. 173 projective. 191 strong Hilbert Nullstellensatz. 151 Chinese Remainder. 38. 94 tangent space. 40 ringed space. 133 local.A/.

35 .226 I NDEX quasi-projective. 129 rational. 126 unirational. 126 zero set.