ebook img

Algebraic Geometry: start up course PDF

72 Pages·2014·0.837 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Algebraic Geometry: start up course

algebraic geometry: start up course  A. L. G∗ ThisisGeometricintroductionintoAlgebraicgeometry. Ihopetoac- quaintthereaderswithsomebasicfiguresunderlyingthemodernal- gebraic technique and show how to translate things from infinitely rich (but quite intuitive) world of figures to restrictive (in fact, finite) butpreciselanguageofformulas. Lecturenotesaresuppliedwithex- ercises actually discussed in classes and important for understanding thesubject. Someofthemarecommentedattheendofthebook. Moscow,2014 ∗NRUHSE,ITEP,IUM,e-mail:[email protected],http://gorod.bogomolov-lab.ru/ Contents Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 §1 ProjectiveSpaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.1 Algebraicvarieties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.2 Projectivespace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.3 Projectivealgebraicvarieties . . . . . . . . . . . . . . . . . . . . . . . 7 1.4 Complementarysubspacesandprojections . . . . . . . . . . . . . . . 10 1.5 Linearprojectivetransformations . . . . . . . . . . . . . . . . . . . . 11 1.6 Cross-ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 §2 Projectiveadrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.1 Remindersfromlinearalgebra . . . . . . . . . . . . . . . . . . . . . . 18 2.2 Tangentlines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.3 Duality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.4 Conics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.5 adraticsurfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.6 Linearsubspaceslyingonasmoothquadric . . . . . . . . . . . . . . 31 2.7 Digression: orthogonalgeometryoverarbitraryfield . . . . . . . . . 32 §3 TensorGuide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 3.1 TensorproductsandSegrevarieties . . . . . . . . . . . . . . . . . . . 40 3.2 Tensoralgebraandcontractions . . . . . . . . . . . . . . . . . . . . . 43 3.3 SymmetricandGrassmannianalgebras . . . . . . . . . . . . . . . . . 46 3.4 Symmetricandskew-symmetrictensors . . . . . . . . . . . . . . . . 50 3.5 Polarisationofcommutativepolynomials . . . . . . . . . . . . . . . . 52 3.6 Polarizationofgrassmannianpolynomials . . . . . . . . . . . . . . . 57 §4 Grassmannians . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 4.1 Plücker’squadricandgrassmanianGr(2,4) . . . . . . . . . . . . . . . 60 4.2 LagrangiangrassmannianLGr(2,4) . . . . . . . . . . . . . . . . . . . 63 4.3 GrassmanniansGr(𝑘,𝑛) . . . . . . . . . . . . . . . . . . . . . . . . . . 64 4.4 Celldecomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 Commentstosomeexercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 2 §1 ProjectiveSpaces 1.1 Algebraicvarieties.Algebraicgeometrystudiesgeometricfiguresthatarelookinglocally¹ asasolutionsetofasystemofpolynomialequationsinaffinespace. Letmerecallbrieflywhat doesthelaermean. 1.1.1 Polynomials.Let𝑉 be𝑛-dimensionalvectorspaceoveranarbitraryfield𝕜. Itsdual space 𝑉∗ isthespaceofall𝕜-linearmaps𝑉 → 𝕜. Wewrite⟨𝜑, 𝑣⟩ = 𝜑(𝑣) ∈ 𝕜forthevalueof linearform𝜑 ∈ 𝑉∗onvector𝑣 ∈ 𝑉. Givenabasis𝑒 ,𝑒 ,…,𝑒 ∈ 𝑉,itsdualbasis𝑥 ,𝑥 ,…,𝑥 ∈ (cid:2869) (cid:2870) (cid:3041) (cid:2869) (cid:2870) (cid:3041) 𝑉∗ consistsofthecoordinateformsdefinedbyprescriptions 1 if 𝑖 = 𝑗 (cid:3564)𝑥(cid:3036), 𝑒(cid:3037)(cid:3565) = (cid:3696)0 otherwise. Let us write 𝑆𝑉∗ = 𝕜[𝑥 ,𝑥 ,…,𝑥 ] for the algebra of polynomials in 𝑥 ’s with coefficients in (cid:2869) (cid:2870) (cid:3041) (cid:3036) 𝕜. Another choice of a basis in 𝑉∗ leads to isomorphic algebra obtained from the initial one by an invertible linear change of variables. We write 𝑆(cid:3031)𝑉∗ ⊂ 𝑆𝑉∗ for the subspace of homo- geneous polynomials of degree 𝑑. It is stable under linear changes of variables and has a basis (cid:3040) (cid:3040) (cid:3040) 𝑥 (cid:3117)𝑥 (cid:3118)…𝑥 (cid:3289) numbered by all collections 𝑚 = (𝑚 ,𝑚 ,…,𝑚 ) of integers 0 ⩽ 𝑚 ⩽ 𝑑 with (cid:2869) (cid:2870) (cid:3041) (cid:2869) (cid:2870) (cid:3041) (cid:3036) ∑𝑚 = 𝑑. (cid:3036) E1.1. Showthatdim𝑆(cid:3031)𝑉∗ = (cid:3512)(cid:3041)+(cid:3031)−(cid:2869)(cid:3513)assoondim𝑉 = 𝑛. (cid:3031) In fact, symmetric powers 𝑆(cid:3031)𝑉∗ and symmetric algebra 𝑆𝑉∗ of vector space 𝑉∗ admit intrinsic coordinate-freedefinitionbutwepostponeitforn∘3.3.1onp.46below. Notethat 𝑆𝑉∗ = (cid:3026)𝑆(cid:3031)𝑉∗ and 𝑆(cid:3038)𝑉∗⋅𝑆(cid:3040)𝑉∗ ⊂ 𝑆(cid:3038)+(cid:3040)𝑉∗. (cid:3031)⩾(cid:2868) 1.1.2 Polynomial functions. Each polynomial 𝑓 = ∑𝑎 𝑥(cid:3040)(cid:3117)…𝑥(cid:3040)(cid:3289) ∈ 𝑆𝑉∗ produces a (cid:3040) (cid:2869) (cid:3041) (cid:3040) polynomialfunction𝑉 → 𝕜thattakes (cid:3040) (cid:3040) 𝑣 ↦ (cid:3037)𝑎(cid:3040)(cid:3564)𝑥(cid:2869), 𝑣(cid:3565) (cid:3117)…(cid:3564)𝑥(cid:3041), 𝑣(cid:3565) (cid:3289) (1-1) (cid:3040) (evaluationof𝑓 atthecoordinatesof𝑣). Wegetahomomorphismofalgebras 𝑆𝑉∗ → {functions𝑉 → 𝕜}. (1-2) that takes polynomial 𝑓 to function (1-1), which we will denote by the same leer 𝑓 in spite of thenextclaimsayingthatthisnotationisnotquitecorrectforfinitefields. P1.1 Homomorphism(1-2)isinjectiveifanonlyifthegroundfield𝕜isinfinite. P. If𝕜consistsof𝑞elements,thenthespaceofallfunctions𝑉 → 𝕜consistsof𝑞(cid:3044)(cid:3289) elements whereasthepolynomialalgebra𝕜[𝑥 ,𝑥 ,…,𝑥 ]isinfinite. Hence,homomorphism(1-2)cannot (cid:2869) (cid:2870) (cid:3041) be injective. Now let 𝕜 be infinite. For 𝑛 = 1 each non zero polynomial 𝑓 ∈ 𝕜[𝑥 ] vanishes (cid:2869) ¹i.e. inaneighboorofeachpoint 3 4 §1ProjectiveSpaces in at most deg𝑓 pints of 𝑉 ≃ 𝕜. Hence, the polynomial function 𝑓 ∶ 𝑉 → 𝕜 is not the zero function. For 𝑛 > 1 we proceed inductively. Expand 𝑓 ∈ 𝕜[𝑥 ,𝑥 ,…,𝑥 ] as polynomial in 𝑥 (cid:2869) (cid:2870) (cid:3041) (cid:3041) withcoefficientsin𝕜[𝑥 ,𝑥 ,…,𝑥 ]: (cid:2869) (cid:2870) (cid:3041)−(cid:2869) 𝑓 = 𝑓(𝑥 ,𝑥 ,…,𝑥 ; 𝑥 ) = (cid:3037)𝑓 (𝑥 ,𝑥 ,…,𝑥 )⋅𝑥(cid:3092). (cid:2869) (cid:2870) (cid:3041)−(cid:2869) (cid:3041) (cid:3092) (cid:2869) (cid:2870) (cid:3041)−(cid:2869) (cid:3041) (cid:3092) Letthepolynomialfunction𝑓 ∶ 𝑉 → 𝕜vanishidenticallyon𝕜(cid:3041). Evaluatingthecoefficients𝑓 (cid:3092) atany𝑤 ∈ 𝕜(cid:3041)−(cid:2869),wegetpolynomial𝑓(𝑤;𝑥 ) ∈ 𝕜[𝑥 ]thatproducesidenticallyzerofunctionof (cid:3041) (cid:3041) 𝑥 . Hence, 𝑓(𝑤;𝑥 ) = 0 in 𝕜[𝑥 ]. us, all coefficients 𝑓 (𝑤) are identically zero functions of (cid:3041) (cid:3041) (cid:3041) (cid:3092) 𝑤 ∈ 𝕜(cid:3041)−(cid:2869). Byinduction,theyarezeropolynomials. (cid:3) E1.2. Let𝑝beaprimenumber,𝔽 = ℤ∕(𝑝)betheresiduefieldmod𝑝. Giveanexplicit (cid:3043) exampleofnon-zeropolynomial𝑓 ∈ 𝔽 [𝑥]thatproducesidenticallyzerofunction𝔽 → 𝔽 . (cid:3043) (cid:3043) (cid:3043) 1.1.3 Affinespaceandaffinevarieties.Associatedwith𝑛-dimensionalvectorspace𝑉isan affine space 𝔸(cid:3041) = 𝔸(𝑉) of dimension 𝑛 also called an affinization of 𝑉. By the definition, points of 𝔸(𝑉) are the vectors of 𝑉. A point corresponding to thezerovectoriscalledanoriginanddenotedby𝑂. All otherpointscanbeimaginedasthe«ends»ofnonzero vectors«drawn»fromtheorigin. Eachpolynomial𝑓 ∈ 𝑆𝑉∗on𝑉producespolynomial function 𝑓 ∶ 𝔸(𝑉) → 𝕜. e set of its zeros is denoted (cid:24)) n by 𝑉(𝑓) ≝ {𝑝 ∈ 𝔸(𝑉)| 𝑓(𝑝) = 0} and is called an affine n A algebraic hypersurface. An intersection of (any set¹ o) P( O affine hypersurfaces is called an affine algebraic variety. y t In other words, an algebraic variety is a figure 𝑋 ⊂ 𝔸(cid:3041) fini n definedbyanarbitrarysystemofpolynomialequations. i e simplest hypersurface is an affine hyperplane given by affine linear equation 𝜑(𝑣) = 𝑐, where 𝜑 ∈ 𝑉∗ is non-zero linear form and 𝑐 ∈ 𝕜. Such a hyperplane passes through the origin iff 𝑐 = 0. In this case it coin- cideswiththeaffinespace𝔸(Ann𝜑)associatedwiththe affine chart U (cid:24) vectorsubspaceAnn(𝜑) = {𝑣 ∈ 𝑉| 𝜑(𝑣) = 0}. Ingeneral Fig.1⋄1.Projectiveword. case,affinehyperplane𝜑(𝑣) = 𝑐isashiof𝔸(Ann𝜑)by anyvector𝑢suchthat𝜑(𝑢) = 𝑐. 1.2 Projective space. Associated with (𝑛 +1)-dimensional vector space 𝑉 is a projective space ℙ = ℙ(𝑉)ofdimension𝑛alsocalledaprojectivizationof𝑉. Bythedefinition,pointsofℙ(𝑉)are (cid:3041) 1-dimensional vector subspaces in 𝑉, i.e. the lines in 𝔸(cid:3041)+(cid:2869) = 𝔸(𝑉) passing through the origin. To see them as «usual» points we have to use a screen — an affine hyperplane 𝑈 ⊂ 𝔸(𝑉) that (cid:3093) doesnotcontaintheorigin,likeonfig. 1⋄1. Itisgivenbyaffinelinearequation𝜉(𝑣) = 1,where 𝜉 ∈ 𝑉∗ ∖ 0 is uniquely determined by 𝑈 , and is called an affine chart. us, affine charts on (cid:3093) ℙ(𝑉)stayinbijectionwithnon-zero𝜉 ∈ 𝑉∗. Noaffinechartdoescoverthewholeofℙ(𝑉). e differenceℙ ⧵𝑈 = ℙ(Ann𝜉) ≃ ℙ consistsofalllineslyingintheparallelcopyof𝑈 drawn (cid:3041) (cid:3093) (cid:3041)−(cid:2869) (cid:3093) through𝑂. Itiscalledaninfinity ofchart𝑈 . Wegetdecompositionℙ = 𝔸(cid:3041)⊔ℙ . Repeating (cid:3093) (cid:3041) (cid:3041)−(cid:2869) ¹maybeaninfiniteset 1.2.Projectivespace 5 itforℙ andfurther,wesplitℙ intodisjointunionofaffinespaces: (cid:3041)−(cid:2869) (cid:3041) ℙ = 𝔸(cid:3041) ⊔𝔸(cid:3041)−(cid:2869)⊔ℙ = ⋯ = 𝔸(cid:3041) ⊔𝔸(cid:3041)−(cid:2869)⊔ … ⊔𝔸(cid:2868) (1-3) (cid:3041) (cid:3041)−(cid:2870) (notethat𝔸(cid:2868) = ℙ isonepointset). (cid:2868) E1.3. Considerdecomposition(1-3)overfinitefield𝔽 of𝑞 elements,computecardi- (cid:3044) nalitiesofbothsidesindependently,andlookatanidentityon𝑞 youwillget. 1.2.1 Homogeneouscoordinates.Achoiceofbasis𝜉 ,𝜉 ,…,𝜉 ∈ 𝑉∗identifies𝑉with𝕜(cid:3041)+(cid:2869) (cid:2868) (cid:2869) (cid:3041) bysending𝑣 ∈ 𝑉to(𝜉 (𝑣),𝜉 (𝑣),… , 𝜉 (𝑣)) ∈ 𝕜(cid:3041)+(cid:2869). Twonon-zerocoordinaterows𝑥,𝑦 ∈ 𝕜(cid:3041)+(cid:2869) (cid:2868) (cid:2869) (cid:3041) representthesamepoint𝑝 ∈ ℙ(𝑉)ifftheyareproportional,i.e. 𝑥 ∶ 𝑥 = 𝑦 ∶ 𝑦 forall 0 ⩽ 𝜇 ≠ 𝜈 ⩽ 𝑛 (cid:3091) (cid:3092) (cid:3091) (cid:3092) (wheretheidentities0 ∶ 𝑥 = 0 ∶ 𝑦and𝑥 ∶ 0 = 𝑦 ∶ 0areallowedaswell). us,points𝑝 ∈ ℙ(𝑉) stayinbijectionwithcollectionsofratios(𝑥 ∶ 𝑥 ∶ … ∶ 𝑥 )calledhomogeneouscoordinateson (cid:2868) (cid:2869) (cid:3041) ℙ(𝑉)w.r.t. thechosenbasis. 1.2.2 Local affine coordinates. Pick up an affine chart 𝑈 = {𝑣 ∈ 𝑉| 𝜉(𝑣) = 1} on (cid:3093) ℙ = ℙ(𝑉). Any 𝑛 covectors 𝜉 ,𝜉 ,…,𝜉 ∈ 𝑉∗ such that 𝜉, 𝜉 ,𝜉 ,…,𝜉 form a basis of 𝑉∗ (cid:3041) (cid:2869) (cid:2870) (cid:3041) (cid:2869) (cid:2870) (cid:3041) provide 𝑈 with local affine coordinates. Namely, consider the basis 𝑒 ,𝑒 ,…,𝑒 ∈ 𝑉 dual to (cid:3093) (cid:2868) (cid:2869) (cid:3040) 𝜉, 𝜉 ,𝜉 ,…,𝜉 andlet𝑒 ∈ 𝑈 betheoriginofaffinecoordinatesystemand𝑒 ,𝑒 ,…,𝑒 ∈ Ann𝜉 (cid:2869) (cid:2870) (cid:3041) (cid:2868) (cid:3093) (cid:2869) (cid:2870) (cid:3041) be its axes. Given a point 𝑝 ∈ ℙ with homogeneous coordinates (𝑥 ∶ 𝑥 ∶ … ∶ 𝑥 ), its local (cid:3041) (cid:2868) (cid:2869) (cid:3041) affinecoordinatesinoursystemarecomputedasfollows: rescale𝑝togetvector𝑢 = 𝑝∕𝜉(𝑝) ∈ 𝑈 (cid:3043) (cid:3093) and evaluate 𝑛 covectors 𝜉(cid:3092) at 𝑢(cid:3043) to get an 𝑛-tiple 𝑡(𝑝) = (cid:3512)𝑡(cid:2869)(𝑝), 𝑡(cid:2870)(𝑝), … , 𝑡(cid:3041)(𝑝)(cid:3513) in which 𝑡 (𝑝) ≝ 𝜉 (𝑢 ) = 𝜉 (𝑝)∕𝜉(𝑝). Notethatlocalaffinecoordinates(𝑡 ,𝑡 ,…,𝑡 )arenon-linear func- (cid:3036) (cid:3036) (cid:3043) (cid:3036) (cid:2869) (cid:2870) (cid:3041) tionsofhomogeneouscoordinates(𝑥 ∶ 𝑥 ∶ … ∶ 𝑥 ). (cid:2868) (cid:2869) (cid:3041) x1 (p0:p1)=(1:t)=(s:1) U1:x1=1 s=p0/p1 (0,1) t=p1/p0 (1,0) O x0 1 = 0 x : 0 U Fig.1⋄2.estandardchartsonℙ (cid:2869) E1.1() Projectivelineℙ = ℙ(𝕜(cid:2870))iscoveredbytwoaffinecharts𝑈 = 𝑈 and𝑈 = 𝑈 ,whicharethe (cid:2869) (cid:2868) (cid:3051) (cid:2869) (cid:3051) (cid:3116) (cid:3117) linesin𝔸(cid:2870) = 𝔸(𝕜(cid:2870))givenbyequations𝑥 = 1and𝑥 = 1(see.fig. 1⋄2). echart𝑈 coversthe (cid:2868) (cid:2869) (cid:2868) 6 §1ProjectiveSpaces wholeofℙ exceptforonepoint (0 ∶ 1)correspondingtotheverticalcoordinateaxisin𝕜(cid:2870). A (cid:2869) (cid:3051) point(𝑥 ∶ 𝑥 )with𝑥 ≠ 0isvisiblein𝑈 as 1 ∶ (cid:3117) . Function𝑡 = 𝑥 | = 𝑥 ∕𝑥 canbetaken (cid:2868) (cid:2869) (cid:2868) (cid:2868) (cid:3585) (cid:3051)(cid:3116)(cid:3586) (cid:2869) (cid:3022)(cid:3116) (cid:2869) (cid:2868) (cid:3051) as local affine coordinate in 𝑈 . Similarly, the chart 𝑈 covers all points (𝑥 ∶ 𝑥 ) = (cid:3116) ∶ 1 (cid:2868) (cid:2869) (cid:2868) (cid:2869) (cid:3585)(cid:3051) (cid:3586) (cid:3117) with𝑥 ≠ 0and𝑠 = 𝑥 | = 𝑥 ∕𝑥 canbeusedaslocalaffinecoordinatein𝑈 . einfinitepoint (cid:2869) (cid:2868) (cid:3022) (cid:2868) (cid:2869) (cid:2869) (cid:3117) of𝑈 is(1 ∶ 0)correspondingtothehorizontalaxisin𝕜(cid:2870). Assoonasapoint(𝑥 ∶ 𝑥 ) ∈ ℙ is (cid:2869) (cid:2868) (cid:2869) (cid:2869) visibleinthebothcharts,itslocalaffinecoordinates𝑠and𝑡satisfytherelation𝑠 = 1∕𝑡. E1.4. Checkit. N t=1/s p 1 ∅ S s=1/t Fig.1⋄3.ℙ (ℝ) ≃ 𝑆(cid:2869) (cid:2869) usℙ isobtainedbygluingtwodistinctcopiesof𝔸(cid:2869) alongthecomplementstotheoriginby (cid:2869) the following rule: point 𝑠 of the first 𝔸(cid:2869) is glued with point 1∕𝑠 of the second. Over 𝕜 = ℝ wegetinthiswayacircleofdiameter1gluedfromtwooppositetangentlines(see.fig. 1⋄3)via thecentralprojectionofeachtangentlineonthecirclefromthetangencypointoftheopposite tangentline. N 1 t=1/s U0≃C i p i 1 S s=1/t U1≃C Fig.1⋄4.ℙ (ℂ) ≃ 𝑆(cid:2870) (cid:2869) Similargluingover𝕜 = ℂalsocanberealizedbymeansofcentralprojectionsoftwotangent planes drown through south and nord poles of the sphere of diameter 1 onto the sphere from the poles opposite to the tangency poles, see fig. 1⋄4. If we identify each tangent plane with ℂ respectingtheirorientations¹likeonfig. 1⋄4,thenthecomplexnumbers𝑠,𝑡layingondifferent ¹oneℂshouldbeobtainedfromtheotherbycontinuousmovealongthesphere 1.3.Projectivealgebraicvarieties 7 planes are projected to the same point of sphere iff they have opposite arguments and inverse absolutevalues¹,i.e. 𝑡 = 1∕𝑠. us,complexprojectivelineisnothingbutasphere. E 1.5. Make sure that ) real projective plane ℙ (ℝ) is the Möbius tape glued with (cid:2870) disc along the boundary circle² ) real projective 3D-space ℙ (ℝ) coincides with the Lie (cid:2871) groupSO (ℝ)ofrotationsofEuclideanspaceℝ(cid:2871) abouttheorigin. (cid:2871) E1.2(ℙ ) (cid:3041) Acollectionof(𝑛+1)affinecharts𝑈 = 𝑈 givenin𝕜(cid:3041)+(cid:2869)byaffinelinearequations{𝑥 = 1}is (cid:3092) (cid:3051) (cid:3092) (cid:3340) calledastandardaffinecoveringofℙ = ℙ(𝕜(cid:3041)+(cid:2869)). Foreach𝜈 = 0, 1, … , 𝑛wetakethefunctions (cid:3041) 𝑥 𝑡((cid:3092)) = 𝑥 | = (cid:3036) , where0 ⩽ 𝑖 ⩽ 𝑛, 𝑖 ≠ 𝜈, (cid:3036) (cid:3036) (cid:3022)(cid:3340) 𝑥 (cid:3092) as 𝑛 standard local affine coordinates inside 𝑈 . One can think of ℙ as the result of gluing (cid:3092) (cid:3041) (𝑛 + 1) distinct copies 𝑈 ,𝑈 ,…,𝑈 of affine space 𝔸(cid:3041) along their actual intersections inside (cid:2868) (cid:2869) (cid:3041) ℙ . In terms of homogeneous coordinates 𝑥 = (𝑥 ∶ 𝑥 ∶ … ∶ 𝑥 ) on ℙ intersection 𝑈 ∩𝑈 (cid:3041) (cid:2868) (cid:2869) (cid:3041) (cid:3041) (cid:3091) (cid:3092) consistsofall𝑥with𝑥 ≠ 0and𝑥 ≠ 0. Intermsoflocalaffinecoordinatesinside𝑈 and𝑈 this (cid:3091) (cid:3092) (cid:3091) (cid:3092) intersectionisgivenbyinequalities𝑡((cid:3091)) ≠ 0and𝑡((cid:3092)) ≠ 0respectively. Twopoints𝑡((cid:3091)) ∈ 𝑈 and (cid:3092) (cid:3091) (cid:3091) 𝑡((cid:3092)) ∈ 𝑈 aregluedwitheachotherinℙ iff𝑡((cid:3091)) = 1∕𝑡((cid:3092)) and𝑡((cid:3091)) = 𝑡((cid:3092))∕𝑡((cid:3092)) for𝑖 ≠ 𝜇,𝜈. RHSof (cid:3092) (cid:3041) (cid:3092) (cid:3091) (cid:3036) (cid:3036) (cid:3091) theserelationsarecalledtransitionfunctionsfromlocalcoordinates𝑡((cid:3092)) tolocalcoordinates𝑡((cid:3091)). 1.3 Projective algebraic varieties. If a basis 𝑥 ,𝑥 ,…,𝑥 ∈ 𝑉∗ is chosen, non-constant poly- (cid:2868) (cid:2869) (cid:3041) nomials in 𝑥 ’s do not produce the functions on ℙ(𝑉) any more, because the values 𝑓(𝑣) and (cid:3036) 𝑓(𝜆𝑣) are different in general. However for any homogeneous polynomial 𝑓 ∈ 𝑆(cid:3031)𝑉∗ its zero set 𝑉(𝑓) ≝ {𝑣 ∈ 𝑉| 𝑓(𝑣) = 0}isstillwelldefinedasafigureinℙ(𝑉),because 𝑓(𝑣) = 0 ⟺ 𝑓(𝜆𝑣) = 𝜆(cid:3031)𝑓(𝑣) = 0. In other words, affine hypersurface 𝑉(𝑓) ⊂ 𝔸(𝑉) defined by homogeneous 𝑓 is a cone ruled by lines passing through the origin. e set of these lines is denoted by 𝑉(𝑓) ⊂ ℙ(𝑉) as well and is called a projective hypersurface of degree 𝑑. Intersections of such hypersurfaces³ are called projectivealgebraicvarieties. esimplestexamplesofprojectivevarietiesareprojectivesubspacesℙ(𝑈) ⊂ ℙ(𝑉)associated withvectorsubspaces𝑈 ⊂ 𝑉 andgivenbysystemsoflinearhomogeneousequations𝜑(𝑣) = 0, where𝜑runsthroughAnn𝑈 ⊂ 𝑉∗. Say,aline(𝑎𝑏)isassociatedwiththelinearspanof𝑎,𝑏and consistsofpoints𝜆𝑎+𝜇𝑏. Itcouldbegivenbylinearequations𝜉(𝑥) = 0with𝜉runningthrough Ann(𝑎)∩Ann(𝑏)oranybasisofthisspace. Ratio(𝜆 ∶ 𝜇)betweencoefficientsin𝜆𝑎+𝜇𝑏 ∈ (𝑎,𝑏) canbetakenasinternalhomogeneouscoordinateon(𝑎𝑏). E 1.6. Show that 𝐾 ∩ 𝐿 ≠ ∅ for any two projective subspaces 𝐾,𝐿 ⊂ ℙ such that (cid:3041) dim𝐾+dim𝐿 ⩾ 𝑛. Forexample,anytwolinesonℙ havenonemptyintersection. (cid:2870) ¹wehaveseenthisonfig.1⋄3before ²theboundaryoftheMöbiustapeisacircleaswellastheboundaryofthedisc;thisallowstogluethe disktothetapealongthiscircle ³maybeinfinitecollectionsofhypersurfacesofdifferentdegrees 8 §1ProjectiveSpaces E1.3() Ontherealprojectiveplaneℙ(ℝ(cid:2871))letusconsideracurvegivenbyhomogeneousequation 𝑥(cid:2870)+𝑥(cid:2870) = 𝑥(cid:2870) (1-4) (cid:2868) (cid:2869) (cid:2870) andlookatitsimprintsinseveralaffinecharts. Inthestandardchart𝑈 ,where𝑥 = 1,inlocal (cid:3051) (cid:2869) (cid:3117) affinecoordinates𝑡 = 𝑥 | = 𝑥 ∕𝑥 ,𝑡 = 𝑥 | = 𝑥 ∕𝑥 equation(1-4)turnstohyperbola (cid:2868) (cid:2868) (cid:3022) (cid:2868) (cid:2869) (cid:2870) (cid:2870) (cid:3022) (cid:2870) (cid:2869) (cid:3299)(cid:3117) (cid:3299)(cid:3117) 𝑡(cid:2870)−𝑡(cid:2870) = 1. (cid:2870) (cid:2868) Inotherstandardchart𝑈 , where𝑥 = 1, inco- (cid:3051) (cid:2870) (cid:3118) ordinates 𝑡 = 𝑥 | = 𝑥 ∕𝑥 (cid:2868) (cid:2868) (cid:3022) (cid:2868) (cid:2870) (cid:3299)(cid:3118) 𝑡 = 𝑥 | = 𝑥 ∕𝑥 (cid:2869) (cid:2869) (cid:3022) (cid:2869) (cid:2870) (cid:3299)(cid:3118) itturnstocircle𝑡(cid:2870)+𝑡(cid:2870) = 1. Inslantedchart𝑈 , (cid:2868) (cid:2869) (cid:3051)(cid:3117)+(cid:3051)(cid:3118) where𝑥 +𝑥 = 1,inlocalcoordinates (cid:2869) (cid:2870) 𝑡 = 𝑥 | = 𝑥 ∕(𝑥 +𝑥 ) (cid:2868) (cid:3022) (cid:2868) (cid:2869) (cid:2870) (cid:3299)(cid:3117)+(cid:3299)(cid:3118) 𝑢 = (𝑥 −𝑥 )| = (𝑥 −𝑥 )∕(𝑥 +𝑥 ) (cid:2870) (cid:2869) (cid:3022) (cid:2870) (cid:2869) (cid:2870) (cid:2869) (cid:3299)(cid:3117)+(cid:3299)(cid:3118) we get parabola¹ 𝑡(cid:2870) = 𝑢. us, affine ellipse, hy- perbolaandparabolaarejusttheseveralpiecesof one projective curve 𝐶 visible in different affine Fig.1⋄5.Realprojectiveconic. charts. Howdoes𝐶looklikeinagivenchart𝑈 ⊂ (cid:3093) ℙ dependsonpositionalrelationshipbetween𝐶andtheinfinitelineℓ = ℙ(Ann𝜉)ofthechart: (cid:2870) ∞ ellipse,hyperbolaandparabolaappearwhenℓ doesnotintersect𝐶,doestouch𝐶 atonepoint ∞ ordoesintersect𝐶 intwodistinctpointsrespectively(see.fig. 1⋄5). 1.3.1 Projectiveclosureofanaffinevariety.Eachaffinealgebraichypersurface 𝑆 = 𝑉(𝑓) ⊂ 𝔸(cid:3041) givenbynon-homogeneouspolynomial𝑓(𝑥 ,𝑥 ,…,𝑥 )ofdegree𝑑iscanonicallyextenttopro- (cid:2869) (cid:2870) (cid:3041) jectivehypersurface𝑆 = 𝑉(𝑓) ⊂ ℙ givenbyhomogeneouspolynomial𝑓(𝑥 ,𝑥 , …, 𝑥 ) ∈ 𝑆(cid:3031)𝑉∗ (cid:3041) (cid:2868) (cid:2869) (cid:3041) ofthesamedegree𝑑andsuchthat𝑆∩𝑈 = 𝑆,where𝑈 = 𝑈 isthestandardaffinechartonℙ . (cid:2868) (cid:2868) (cid:3051) (cid:3041) (cid:3116) If𝑓(𝑥 ,𝑥 ,…,𝑥 ) = 𝑓 +𝑓 (𝑥 ,𝑥 ,…,𝑥 )+𝑓 (𝑥 ,𝑥 ,…,𝑥 )+ ⋯ +𝑓 (𝑥 ,𝑥 ,…,𝑥 )where𝑓 (cid:2869) (cid:2870) (cid:3041) (cid:2868) (cid:2869) (cid:2869) (cid:2870) (cid:3041) (cid:2870) (cid:2869) (cid:2870) (cid:3041) (cid:3031) (cid:2869) (cid:2870) (cid:3041) (cid:3036) ishomogeneousofdegree𝑖,then 𝑓(𝑥 ,𝑥 , …, 𝑥 ) = 𝑓 ⋅𝑥(cid:3031) +𝑓 (𝑥 ,𝑥 ,…,𝑥 )⋅𝑥(cid:3031)−(cid:2869)+ ⋯ +𝑓 (𝑥 ,𝑥 ,…,𝑥 ), (cid:2868) (cid:2869) (cid:3041) (cid:2868) (cid:2868) (cid:2869) (cid:2869) (cid:2870) (cid:3041) (cid:2868) (cid:3031) (cid:2869) (cid:2870) (cid:3041) whichturnsto𝑓 as𝑥 = 1. ecomplement𝑆∖𝑆 = 𝑆∩𝑈(∞),thatistheintersectionof𝑆 with (cid:2868) (cid:2868) the infinite hyperplane 𝑥 = 0, is given in homogeneous coordinates (𝑥 ∶ 𝑥 ∶ ⋯ ∶ 𝑥 ) on (cid:2868) (cid:2869) (cid:2870) (cid:3041) the infinite hyperplane by equation 𝑓 (𝑥 ,𝑥 ,…,𝑥 ) = 0, that is by vanishing the top degree (cid:3031) (cid:2869) (cid:2870) (cid:3041) component of 𝑓. us, infinite points of 𝑆 are nothing else than asymptotic directions of affine hypersurface𝑆. For example, projective closure of affine cubic curve 𝑥 = 𝑥(cid:2871) is projective cubic 𝑥(cid:2870)𝑥 = 𝑥(cid:2871) (cid:2869) (cid:2870) (cid:2868) (cid:2869) (cid:2870) that has exactlyone infinite point 𝑝 = (0 ∶ 1 ∶ 0). Note that in the standard chart 𝑈 , which ∞ (cid:2869) containsthispoint,𝐶 lookslikesemi-cubicparabola𝑥(cid:2870) = 𝑥(cid:2871) withacuspat𝑝 . (cid:2868) (cid:2870) ∞ ¹move𝑥(cid:2870)toR.H.S.of (1-4)anddividethebothsidesby𝑥 +𝑥 (cid:2869) (cid:2870) (cid:2869) 1.3.Projectivealgebraicvarieties 9 1.3.2 Space of hypersurfaces. Since proportional polynomials define the same hypersur- faces𝑉(𝑓) = 𝑉(𝜆𝑓),projectivehypersurfacesoffixeddegree𝑑arethepointsofprojectivespace 𝒮 = 𝒮 (𝑉) ≝ ℙ(𝑆(cid:3031)𝑉∗)calledaspaceofdegree𝑑hypersufacesinℙ(𝑉). (cid:3031) (cid:3031) E1.7. Finddim𝒮 (𝑉)assumingdim𝑉 = 𝑛+1. (cid:3031) Since,. Projective subspaces of 𝒮 are called linear systems¹ of hypersurfaces. For example, all (cid:3031) degree 𝑑 hypersurfaces passing through a given point form a a linear system of codimension 1,i.e. ahyperplanein𝒮 ,becauseequation𝑓(𝑝) = 0islinearin𝑓 ∈ 𝑆(cid:3031)𝑉∗foranyfixed𝑝 ∈ ℙ(𝑉). (cid:3031) Eachhypersurfacelayinginalinearsystemspannedby𝑉(𝑓 ), 𝑉(𝑓 ), … , 𝑉(𝑓 ),isgivenbyan (cid:2869) (cid:2870) (cid:3040) equationoftheform𝜆 𝑓 +𝜆 𝑓 +⋯+𝜆 𝑓 = 0,where𝜆 ,𝜆 ,…,𝜆 ∈ 𝕜aresomeconstants. (cid:2869) (cid:2869) (cid:2870) (cid:2870) (cid:3040) (cid:3040) (cid:2869) (cid:2870) (cid:3040) In particular, any such a hypersurface contains the intersection 𝑉(𝑓 ) ∩ 𝑉(𝑓 ) ∩ … ∩ 𝑉(𝑓 ). (cid:2869) (cid:2870) (cid:3040) Traditionally,linearsystemsofdimensions1,2and3arecalledpencils,netsandwebs. E1.8. Showthateachpencilofhypersurfacescontainsahypersurfacepassingthrough anyprescribedpoint(overanarbitraryfield𝕜). 1.3.3 Workingexample: collectionsofpointsonℙ .Let𝑈 = 𝕜(cid:2870) withthestandardcoordi- 𝟏 nates𝑥 ,𝑥 . Eachfinitesetofpoints²𝑝 ,𝑝 ,…,𝑝 ∈ ℙ = ℙ(𝑈)isthesetofzerosforaunique (cid:2868) (cid:2869) (cid:2869) (cid:2870) (cid:3031) (cid:2869) upascalarfactorhomogeneouspolynomialofdegree𝑑 (cid:3031) (cid:3031) 𝑓(𝑥 ,𝑥 ) = (cid:3038)det(𝑥,𝑝 ) = (cid:3038)(𝑝 𝑥 −𝑝 𝑥 ) , where𝑝 = (𝑝 ∶ 𝑝 ). (1-5) (cid:2868) (cid:2869) (cid:3092) (cid:3092),(cid:2869) (cid:2868) (cid:3092),(cid:2868) (cid:2869) (cid:3092) (cid:3092),(cid:2868) (cid:3092),(cid:2869) (cid:3092)=(cid:2869) (cid:3092)=(cid:2869) We will say that the points 𝑝 are the roots of 𝑓. Each non-zero homogeneous polynomial of (cid:3036) degree 𝑑 has at most 𝑑 distinct roots on ℙ . If the ground field 𝕜 is algebraically closed, the (cid:2869) number of roots³ equals 𝑑 precisely and there is a bijection between the points of ℙ(𝑆(cid:3031)𝑈∗) and non-orderedcollectionsof𝑑 pointsonℙ . (cid:2869) Overanarbitraryfield𝕜thosecollectionswhereall𝑑 pointscoincideformacurve 𝐶 ⊂ ℙ = ℙ(𝑆(cid:3031)𝑈∗) (cid:3031) (cid:3031) calledVeronesecurve⁴ofdegree𝑑. ItcoincideswithanimageoftheVeroneseembedding (cid:3101)↦(cid:3101)(cid:3279) 𝑣(cid:3031) ∶ ℙ×(cid:2869) = ℙ(cid:3512)𝑈∗(cid:3513) −−−−−→ ℙ(cid:3031) = ℙ(cid:3512)𝑆(cid:3031)𝑈∗(cid:3513) (1-6) that takes a linear polynomial 𝜑 ∈ 𝑈∗, whose zero set is some point 𝑝 ∈ ℙ(𝑈), to 𝑑th power 𝜑(cid:3031) ∈ 𝑆(cid:3031)(𝑈∗),whosezerosetis𝑑-tiplepoint𝑝. Letuswritepolynomials𝜑 ∈ 𝑈∗ and𝑓 ∈ 𝑆(cid:3031)(𝑈∗)as 𝑑 𝜑(𝑥) = 𝛼 𝑥 +𝛼 𝑥 and 𝑓(𝑥) = (cid:3037)𝑎 ⋅ 𝑥(cid:3031)−(cid:3092)𝑥(cid:3092) (cid:2868) (cid:2868) (cid:2869) (cid:2869) (cid:3092) (cid:3585)𝜈(cid:3586) (cid:2868) (cid:2869) (cid:3092) and use (𝛼 ∶ 𝛼 ) and (𝑎 ∶ 𝑎 ∶ … ∶ 𝑎 ) as homogeneous coordinates in ℙ× = ℙ(𝑈∗) and in (cid:2868) (cid:2869) (cid:2868) (cid:2869) (cid:3031) (cid:2869) ℙ = ℙ(𝑆(cid:3031)𝑈∗)respectively. entheVeronesecurvecomeswiththeparametrization (cid:3031) (𝛼(cid:2868) ∶ 𝛼(cid:2869)) ↦ (𝑎(cid:2868)∶ 𝑎(cid:2869)∶ … ∶ 𝑎(cid:3031)) = (cid:3512)𝛼(cid:2868)(cid:3031) ∶ 𝛼(cid:2868)(cid:3031)−(cid:2869)𝛼(cid:2869) ∶ 𝛼(cid:2868)(cid:3031)−(cid:2870)𝛼(cid:2869)(cid:2870) ∶ ⋯ ∶ 𝛼(cid:2869)(cid:3031)(cid:3513) (1-7) ¹orlinearseriesinoldterminology ²someofpointsmaycoincide ³countedwithmultiplicities,whereamultiplicityofaroot𝑝isdefinedasmaximal𝑘suchthatdet(cid:3038)(𝑥,𝑝) divides𝑓 ⁴thereareseveralothernames: rationalnormalcurve,twistedrationalcurveofdegree𝑑etc 10 §1ProjectiveSpaces by the points of ℙ . It follows from (1-7) that 𝐶 consists of all (𝑎 ∶ 𝑎 ∶ … ∶ 𝑎 ) ∈ ℙ that (cid:2869) (cid:3031) (cid:2868) (cid:2869) (cid:3031) (cid:3031) formageometricprogression,i.e. suchthattherowsofmatrix 𝑎 𝑎 𝑎 … 𝑎 𝑎 𝐴 = (cid:2868) (cid:2869) (cid:2870) (cid:3031)−(cid:2870) (cid:3031)−(cid:2869) (cid:3638)𝑎 𝑎 𝑎 … 𝑎 𝑎 (cid:3639) (cid:2869) (cid:2870) (cid:2871) (cid:3031)−(cid:2869) (cid:3031) areproportional. econditionrk𝐴 = 1isequivalenttovanishingofall2×2-minorsof𝐴. us, 𝐶 ⊂ ℙ isgivenbyasystemofquadraticequations. (cid:3031) (cid:3031) Anintersectionof𝐶 withanarbitraryhyperplanegivenbyequation (cid:3031) 𝐴 𝑎 +𝐴 𝑎 +⋯+𝐴 𝑎 = 0, (cid:2868) (cid:2868) (cid:2869) (cid:2869) (cid:3031) (cid:3031) consistsoftheVeronese-imagesoftheroots(𝛼 ∶ 𝛼 ) ∈ ℙ ofhomogeneouspolynomial (cid:2868) (cid:2869) (cid:2869) (cid:3037)𝐴 ⋅𝛼(cid:3031)−(cid:3092)𝛼(cid:3092) (cid:3092) (cid:2868) (cid:2869) (cid:3092) ofdegree𝑑. Sinceithasatmost𝑑 roots,any𝑑+1distinctpointsontheVeronesecurvedonot lieinahyperplane. isimpliesthatany𝑚pointsof𝐶 spanasubspaceofdimension𝑚+1and (cid:3031) donotlieinacommonsubspaceofdimension(𝑚−2)assoon2 ⩽ 𝑚 ⩽ 𝑑+1. If𝕜isalgebraicallyclosed,𝐶 intersectsanyhyperplaneinprecisely𝑑points(someofwhich (cid:3031) maycoincide). isexplainswhywedidsaythat𝐶 hasdegree𝑑. (cid:3031) E1.4(V) e Veronese conic 𝐶 ⊂ ℙ consists of quadratic trinomials 𝑎 𝑥(cid:2870) + 2𝑎 𝑥 𝑥 + 𝑎 𝑥(cid:2870) that are (cid:2870) (cid:2870) (cid:2868) (cid:2868) (cid:2869) (cid:2868) (cid:2869) (cid:2870) (cid:2869) perfectsquaresoflinearforms. Itisgivenbywellknownequation 𝑎 𝑎 𝐷∕4 = −det (cid:2868) (cid:2869) = 𝑎(cid:2870)−𝑎 𝑎 = 0 (1-8) (cid:3638)𝑎 𝑎 (cid:3639) (cid:2869) (cid:2868) (cid:2870) (cid:2869) (cid:2870) andcomeswithrationalparametrization 𝑎 = 𝛼(cid:2870) , 𝑎 = 𝛼 𝛼 , 𝑎 = 𝛼(cid:2870) . (1-9) (cid:2868) (cid:2868) (cid:2869) (cid:2868) (cid:2869) (cid:2870) (cid:2869) 1.4 Complementarysubspacesandprojections.Projectivesubspaces𝐾 = ℙ(𝑈)and𝐿 = ℙ(𝑊) in ℙ = ℙ(𝑉) are called complementary, if 𝐾 ∩𝐿 = ∅ and dim𝐾 +dim𝐿 = 𝑛−1. For example, (cid:3041) any two non-intersecting lines in ℙ are complementary. In terms of linear algebra, the vector (cid:2871) subspaces𝑈,𝑊 ⊂ 𝑉 havezerointersection𝑈∩𝑉 = {0}and dim𝑈+dim𝑊 = dim𝐾+1+dim𝐿+1 = (𝑛+1) = dim𝑉. us,𝑉 = 𝑈⊕𝑊andeach𝑣 ∈ 𝑉hasauniquedecomposition𝑣 = 𝑢+𝑤where𝑢∈𝑈and𝑤∈𝑊. If 𝑣 lies neither in 𝑈 nor in 𝑊, both components 𝑢, 𝑤 are non zero vectors. us, each point 𝑝 ∉ 𝐾⊔𝐿liesonauniquelineintersectingbothsubspaces𝐾,𝐿. E1.9. Makeitsure. Givenapairofcomplementarysubspaces𝐾,𝐿 ⊂ ℙ ,aprojectionfrom𝐾to𝐿isamap (cid:3041) 𝜋(cid:3012) ∶ (ℙ ⧵𝐾) → 𝐿 (cid:3013) (cid:3041) that sends each point 𝑝 ∈ ℙ ⧵(𝐾 ⊔𝐿) to a unique point 𝑏 ∈ 𝐿 such that line 𝑝𝑏 intersects 𝐾 (cid:3041) and sends each point of 𝐿 to itself. In homogeneous coordinates (𝑥 ∶ 𝑥 ∶ … ∶ 𝑥 ) such that (cid:2868) (cid:2869) (cid:3041) (𝑥 ∶ 𝑥 ∶ … ∶ 𝑥 ) are the coordinates in 𝐾 and (𝑥 ∶ 𝑥 ∶ … ∶ 𝑥 ) are the coordinates (cid:2868) (cid:2869) (cid:3040) (cid:3040)+(cid:2869) (cid:3040)+(cid:2870) (cid:3041) in𝐿,projection𝜋(cid:3012) removesthefirst(𝑚+1)coordinates𝑥 ,0 ⩽ 𝜈 ⩽ 𝑚. (cid:3013) (cid:3092)

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.