ebook img

Surfaces of general type with $p_g=1$ and $q=0$ PDF

0.24 MB·
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 Surfaces of general type with $p_g=1$ and $q=0$

A CONSTRUCTION OF SURFACES OF GENERAL TYPE WITH p =1 AND q =0 g 9 0 HEESANGPARK,JONGILPARK,ANDDONGSOOSHIN 0 2 Abstract. Inthispaperwereconstructminimalcomplexsurfacesofgeneral n type with pg = 1, q = 0, and 1 ≤ K2 ≤ 2 using a Q-Gorenstein smoothing u theory. Furthermore, we also construct a new family of simply connected J surfaceswithpg=1,q=0,and3≤K2≤6usingthesamemethod. 9 2 ] G 1. Introduction A Complex surfaces of general type with p = 1 and q = 0 have been drawn g . attention because they provide counterexamples to the Torelli problems. All such h t surfaces are constructed by classical methods: Quotient and covering. Kynev [6] a constructed a surface with K2 = 1 as a quotient of the Fermat sextic in P3 by m a suitable action of a group of order 6. According to Catanese [2], all minimal [ surfaces of general type with pg = 1 and K2 = 1 are diffeomorphic and simply 1 connected. Catanese and Debarre [3] constructed surfaces with K2 = 2 by double v coverings of the projective plane P2 or smooth minimal K3 surfaces. In fact, they 5 classified such surfaces into five classes according to the degree and the image of 9 the bicanonical map. Four of them are simply connected and the other one has a 1 5 torsion Z/2Z. Todorov [13] constructed surfaces with 2 ≤ K2 ≤ 8 by considering . doublecoversofK3surfaces. His exampleswith3≤K2 ≤8havebigfundamental 6 groups. 0 9 Inthispaperweconstructminimalcomplexsurfacesofgeneraltypewithpg =1, 0 q = 0, and 1 ≤ K2 ≤ 6 by a Q-Gorenstein smoothing theory. In particular, we : constructsimplyconnected surfaceswith1≤K2 ≤6andasurfacewithK2 =2and v i H1 =Z/2Z. Weuseasimilarmethodin[8]: Weblowupanelliptic K3surfaceina X suitable set of points so that we obtain a surface with a very special configuration r of rational curves. Inside this configuration we find some disjoint chains which a can be contracted to special quotient singularities. These singularities admit a localQ-Gorensteinsmoothing, which is a smoothing whose relative canonicalclass is Q-Cartier. Finally, we prove that these local smoothings can be glued to a global Q-Gorenstein smoothing of the whole singular surface by showing that the obstructionspaceofaglobalsmoothingiszero. Thenitisnotdifficulttoshowthat a general fiber of the smoothing is the desired surface. The key ingredient of this paper is to develop a new method for proving that the obstruction space is zero because the Lee-Park’s method in [8] for a computation of the obstruction space cannot be applied to our cases; Section 3. Date:June1,2009. 2000 Mathematics Subject Classification. Primary14J29; Secondary14J10,14J17, 53D05. Keywords and phrases. Q-Gorensteinsmoothing,rationalblow-downsurgery,surfaceofgen- eraltype. 1 2 HEESANGPARK,JONGILPARK,ANDDONGSOOSHIN InSection2wereconstructcomplexsurfaceswithp =1,q =0,andK2 =2: A g simply connected surface and a surface with H = Z/2Z by using a (generalized) 1 rational blow-down surgery. We prove in Section 3 that the obstruction space of a globalsmoothingofthesingularsurfaceconstructedinSection2iszero. Inthefinal section, we also construct various examples of simply connected complex surfaces satisfying p =1, q =0, and 1≤K2 ≤6. g Acknowledgement. The authors would like to thank Professor Yongnam Lee for somevaluablecommentsonthefirstdraftofthisarticle. JongilParkwassupported bytheKoreaResearchFoundationGrantfundedbytheKoreanGovernment(KRF- 2008-341-C00004). He alsoholdsa jointappointmentatKIASandinthe Research Institute of Mathematics, SNU. 2. A Construction of surfaces with p =1, q =0, and K2 =2 g In this section we construct a simply connected minimal surface of generaltype with p = 1, q = 0, and K2 = 2. We start with a special rational elliptic surface g E(1). Let L , L , L , and A be lines in P2 and let B be a smooth conic in P2 1 2 3 intersecting as in Figure 1(A). We consider a pencil of cubics {λ(L +L +L )+ 1 2 3 µ(A+B) | [λ : µ] ∈ P1} in P2 generated by two cubic curves L +L +L and 1 2 3 A+B, which has 4 base points, say, p, q, r and s. In order to obtain an elliptic fibration over P1 from the pencil, we blow up three times at p and r, respectively, andtwice at s, including infinitely near base-pointsat eachpoint. We performone furtherblowing-upatthebasepointq. Byblowing-uptotallyninetimes,weresolve all base points (including infinitely near base-points) of the pencil and we then get a rational elliptic surface E(1) with an I -singular fiber, an I -singular fiber, two 8 2 nodal singular fibers, and four sections; Figure 1(B). A p b B q L b L3 1 r b L sb 2 (a)Apencil (b)E(1) Figure 1: A rational elliptic surface E(1) Let Y be a double coverof the rationalelliptic surfaceE(1) branchedalongtwo general fibers. Then the surface Y is an elliptic K3 surface with two I -singular 8 fibers, two I -singular fibers, four nodal singular fibers, and four sections; Fig- 2 ure2(A). In the following constructionwe use only oneI -singularfiber, one nodal 8 singular fiber, and three sections; Figure 2(B). Let τ : V → Y be the blowing-up at the node of the nodal singular fiber and let E be the exceptional divisor of τ. Since K = 0, we have K = E. Let Y V A CONSTRUCTION OF SURFACES OF GENERAL TYPE WITH pg=1 AND q=0 3 (a)ThefibrationstructureofY (b)ApartofY Figure 2: An elliptic K3 surface Y D = D +···+D be the part of the I -singular fiber. Let S (i = 1,2,3) be the 1 6 8 i sections of the fibration V → P1 and set S = S +S +S . Let F be the proper 1 2 3 transform of the nodal fiber of the K3 surface Y; Figure 3. S1 ×2 b b D4b bD3 D2 S2 −4F D5 D1 E −1 D6 S3 2 Figure 3: A surface V =Y♯P We blow up the surface V three times totally at the three marked points • and blow up twice at the marked point . We then get a surface Z; Figure 4. There exist three disjoint linear chains of PJ1’s in Z: −3 −6 −2 −3 −2 −4 −2 −2 −3 −2 −4 ◦ − ◦ − ◦ − ◦ − ◦, ◦ − ◦ − ◦ − ◦ − ◦, ◦. Main construction. By applying Q-Gorenstein smoothing theory to the surface Z as in [8, 11, 12], we construct a complex surface of general type with p = 1, g 4 HEESANGPARK,JONGILPARK,ANDDONGSOOSHIN −3 −1 −2 −4 −−14 −1 −2 −1−3 −3 −6 −2 −2 −1 −2 2 Figure 4: A surface Z =Y♯6P q = 0, and K2 = 2. That is, we first contract the three chains of P1’s from the surface Z so that it produces a normalprojective surfaceX with three permissible singular points. In Section 3 we will show that the singular surface X has a global Q-Gorensteinsmoothing. Let X be a generalfiber ofthe Q-Gorensteinsmoothing t of X. Since X is a (singular) surface with p =1, q =0, and K2 =2, by applying g general results of complex surface theory and Q-Gorenstein smoothing theory, one may conclude that a general fiber X is a complex surface of general type with t p = 1, q = 0, and K2 = 2. Furthermore, it is not difficult to show that a general g fiberX isminimalandsimplyconnectedbyusingasimilartechniquein[8,11,12]. t Remark. CataneseandDebarre[3]provedthatsurfacesofgeneraltypewithp =1, g q =0,andK2 =2aredividedintofivetypesaccordingtothedegreeandtheimage of the bicanonical map. Four of them are simply connected and the other one has atorsionZ/2Z. It is aninteresting problemto determine inwhichclass thesimply connected example constructed in this section is contained. We leave this question for the future research. 2.1. An example with K2 = 2 and H = Z/2Z. We construct a minimal com- 1 plex surface of general type with p = 1, q = 0, K2 = 2, and H = Z/2Z. Let A, g 1 L (i = 1,2,3) be lines on the projective plane P2 and B a nonsingular conic on i P2 which intersect as in Figure 5(A). Consider a pencil of cubics generated by the two cubics A+B and L +L +L . By resolving the base points of the pencil 1 2 3 of cubics including infinitely near base-points, we obtain a rational elliptic surface E(1) with an I -singular fiber, an I -singular fiber, an cusp singular fiber, a nodal 7 2 singular fiber, and five sections as in Figure 5(B). Let Y be a double cover of the rational elliptic surface E(1) branched along two general fibers near the cusp singular fiber. Then Y is an elliptic K3 surface with two I -singular fibers, two I -singular fibers, two cusp singular fibers, two 7 2 nodal singular fibers, and five sections. We use only two I -singular fibers, two 7 I -singularfibers,andtwosections; Figure6(A). We blow upeighttimes totally at 2 the eight marked points •. We then get a surface Z; Figure 6(B). There exist six disjoint linear chains of P1’s in Z: −5 −3 −2 −2 −5 −3 −2 −2 −3 −2 −3 −3 −2 −3 −4 −4 ◦ − ◦ − ◦ − ◦, ◦ − ◦ − ◦ − ◦, ◦ − ◦ − ◦, ◦ − ◦ − ◦, ◦, ◦ We now briefly explain how to prove that the complex surface obtained by the above configuration has H = Z/2Z. Let L (i = 1,...,6), M (i = 1,...,6), N 1 i j k (k =1,2) be the parts of the I -singular fibers and the I -singular fibers of the K3 7 2 A CONSTRUCTION OF SURFACES OF GENERAL TYPE WITH pg=1 AND q=0 5 A B L L 3 1 L 2 (a)Apencil (b)E(1) Figure 5: A rational elliptic surface E(1) for K2 =2 and H =Z/2Z 1 S1 b b N1 N2 b L6 bM3 L5 M4 M2 L4 M5 M1 L1 L3 b M6 L2 b S2 S4 b b S3 (a)Y −1 −1 −5 −1 −3 −3 −2 −−31 −2 −2 −3 −2 −4 −4 −1 −2 −2 −3 −3 −1 −5 −1 −1 (b)Z=Y♯8P2 Figure 6: An example with H =Z/2Z 1 surface Y and let S (l =1,...,4) the four sections of Y; Figure 6(A). The matrix l whose entries are the intersection numbers of L , M , N , and S is invertible. i j k l Therefore the Picard number of the K3 surface Y is at least 18. Furthermore, during the fiber sum surgery, four new homology elements, say T ,...,T , are 1 4 constructed. Twoelementsamongthefourelementsarerepresentedbytoriandthe other two elements are represented by spheres. Therefore the homology H (Y,Z) 2 aregeneratedbytheeighteen(−2)-curvesL ,M ,N ,S ,andthefournewelements i j k l T . Hence the homology H (Z,Z) are generated by the proper transforms of L , m 2 i 6 HEESANGPARK,JONGILPARK,ANDDONGSOOSHIN M ,N ,S ,T ,andthe eightexceptionalcurves. Then,byusingasimilarmethod j k l m in[7, 10],it is notdifficult toshow thatthe complex surfaceobtainedbythe above configuration has H =Z/2Z. 1 3. The Existence of Q-Gorenstein smoothings This section is devoted to the proof of the following theorem. Theorem 3.1. The singular surface X constructed in the main construction in Section 2 has a global Q-Gorenstein smoothing. For this, we first briefly review some basic facts on Q-Gorenstein smoothing theoryfornormalprojectivesurfaceswithspecialquotientsingularities(referto[8] for details). Definition. Let X be a normal projective surface with quotient singularities. Let X → ∆ (or X/∆) be a flat family of projective surfaces over a small disk ∆. The one-parameter family of surfaces X →∆ is called a Q-Gorenstein smoothing of X if it satisfies the following three conditions; (i) the general fiber X is a smooth projective surface, t (ii) the central fiber X is X, 0 (iii) the canonical divisor KX/∆ is Q-Cartier. AQ-Gorensteinsmoothingforagermofaquotientsingularity(X ,0)isdefined 0 similarly. A quotient singularity which admits a Q-Gorenstein smoothing is called a singularity of class T. Proposition 3.2 ([5, 9]). Let (X ,0) be a germ of two dimensional quotient sin- 0 gularity. If (X ,0) admits a Q-Gorenstein smoothing over the disk, then (X ,0) is 0 0 either a rational double point or a cyclic quotient singularity of type 1 (1,dna−1) dn2 for some integers a,n,d with a and n relatively prime. −4 −3 −2 −2 −2 −3 Proposition 3.3 ([5, 9]). (1) The singularities ◦ and ◦ − ◦ − ◦ −···− ◦ − ◦ are of class T. (2) If the singularity −◦b1−···−−◦br is of class T, then so are −◦2−−◦b1−···−−b◦r−1−−b◦r−1 and −b◦1−1−−◦b2−···−−◦br−−◦2. (3) Every singularity of class T that is not a rational double point can be obtained by starting with one of the singularities described in (1) and iterating the steps described in (2). Let X be a normal projective surface with singularities of class T. Due to the resultofKoll´arandShepherd-Barron[5],thereisaQ-Gorensteinsmoothinglocally foreachsingularityofclassT onX. The naturalquestionariseswhether this local Q-GorensteinsmoothingcanbeextendedovertheglobalsurfaceX ornot. Roughly geometric interpretationis the following: Let ∪ V be an open coveringof X such α α that each V has at most one singularity of class T. By the existence of a local Q- α Gorensteinsmoothing,thereisaQ-GorensteinsmoothingV /∆. Thequestionisif α these families glueto a globalone. The answercanbe obtainedby figuringoutthe obstructionmapofthe sheavesofdeformationTi =Exti (Ω ,O )for i=0,1,2. X X X X For example, if X is a smooth surface, then T0 is the usual holomorphic tangent X A CONSTRUCTION OF SURFACES OF GENERAL TYPE WITH pg=1 AND q=0 7 sheaf T and T1 =T2 =0. By applying the standard result of deformations to a X X X normal projective surface with quotient singularities, we get the following Proposition 3.4 ([14, §4]). Let X be a normal projective surface with quotient singularities. Then (1) The first order deformation space of X is represented by the global Ext 1-group T1 =Ext1 (Ω ,O ). X X X X (2) The obstruction lies in the global Ext 2-group T2 =Ext2 (Ω ,O ). X X X X Furthermore, by applying the general result of local-global spectral sequence of ext sheaves to deformation theory of surfaces with quotient singularities so that Ep,q =Hp(Tq)⇒Tp+q, and by Hj(Ti )=0 for i,j ≥1, we also get 2 X X X Proposition 3.5 ([9, 14]). Let X be a normal projective surface with quotient singularities. Then (1) We have the exact sequence 0→H1(T0)→T1 →ker[H0(T1)→H2(T0)]→0 X X X X where H1(T0) represents the first order deformations of X for which the sin- X gularities remain locally a product. (2) If H2(T0)=0, every local deformation of the singularities may be globalized. X Asmentionedabove,thereisalocalQ-Gorensteinsmoothingforeachsingularity ofX duetotheresultofKoll´arandShepherd-Barron[5]. Henceitremainstoshow that every local deformation of the singularities can be globalized. Note that the following proposition tells us a sufficient condition for the existence of a global Q-Gorenstein smoothing of X. Proposition 3.6 ([8]). Let X be a normal projective surface with singularities of class T. Let π : V → X be the minimal resolution and let A be the reduced exceptional divisor. Suppose that H2(T (−logA)) = 0. Then H2(T0) = 0 and V X there is a Q-Gorenstein smoothing of X. Furthermore, the proposition above can be easily generalized to any log resolu- tionofX bykeepingthevanishingofcohomologiesunderblowingupatthepoints. It is obtained by the following well-known result. Proposition 3.7 ([4, §1]). Let V be a nonsingular surface and let A be a simple ′ normal crossing divisor in V. Let f :V →V be a blowing up of V at a point p of A. Set A′ =f−1(A)red. Then h2(TV′(−logA′))=h2(TV(−logA)). Therefore Theorem 3.1 follows from Proposition 3.6 and the following proposi- tion: Proposition 3.8. H2(T (−log(D+S+F)))=H0(Ω (log(D+S+F))(E))=0. V V The idea of the proof is as follows: There is an exact sequence of locally free sheaves 6 3 0→Ω →Ω (log(D+S+F))→ O ⊕ O ⊕O →0. V V Di Si F Mi=1 Mi=1 By tensoring O (E), we have an exact sequence V 6 3 0→Ω (E)→Ω (log(D+S+F))(E)→ O ⊕ O ⊕O (E)→0. V V Di Si F Mi=1 Mi=1 8 HEESANGPARK,JONGILPARK,ANDDONGSOOSHIN Since H0(Ω (E)) = 0, the proof of Proposition 3.8 is done if we show that the V connecting homomorphism 6 3 H0(O )⊕ H0(O )⊕H0(O (E))→H1(Ω (E)) Di Si F V Mi=1 Mi=1 is injective. The proof of Proposition 3.8 begins with the following Lemma. We have a commutative diagram 6 3 H0(O )⊕ H0(O )⊕H0(O ) c1 // H1(Ω ) (3.1) Di Si F V Mi=1 Mi=1 β1 (cid:15)(cid:15) 6 3 (cid:15)(cid:15) H0(O )⊕ H0(O )⊕H0(O (E)) d1 //H1(Ω (E)) Di Si F V Mi=1 Mi=1 Lemma 3.9. The composition map 6 3 β ◦c : H0(O )⊕ H0(O )⊕H0(O )→H0(Ω (E)) 1 1 Di Si F V Mi=1 Mi=1 is injective. Proof. Note that the map c is the first Chern class map. Since the intersection 1 matrix, whose entries are the intersection numbers of D (i = 1,...,6), S (j = i j 1,2,3), and F, is invertible, their images by the map c are linearly independent. 1 Therefore the map c is injective. 1 Consider the commutative diagram 0 (cid:15)(cid:15) H0(Ω ⊗O (E)) V E δ 6 3 (cid:15)(cid:15) H0(O )⊕ H0(O )⊕H0(O ) c1 //H1(Ω ) Di Si F V Mi=1 Mi=1 β1 (cid:15)(cid:15) 6 3 (cid:15)(cid:15) H0(O )⊕ H0(O )⊕H0(O (E)) d1 // H1(Ω (E)) Di Si F V Mi=1 Mi=1 where the vertical sequence is induced from the exact sequence 0→H0(Ω ⊗O (E))−→δ H1(Ω )−β→1 H1(Ω (E))−→γ H1(Ω ⊗O (E))→0. V E V V V E A CONSTRUCTION OF SURFACES OF GENERAL TYPE WITH pg=1 AND q=0 9 Claim: Theconnectinghomomorphismδ :H0(Ω ⊗O (E))→H1(Ω )isthefirst V E V Chern class map of O (E): Since H0(Ω ⊗O (E))=0, we have an isomorphism V E E ∼ H0(O (−E)⊗O (E))−=→H0(Ω ⊗O (E)). E E V E Here the above map is given by z ⊗ 1 7→ dzα , where z is a local equation of α zα n zα o α E. Therefore the connecting homomorphism δ : H0(Ω ⊗O (E)) → H1(Ω ) is V E V given by dz dz dz z z α α β α α 7→ − = d / , (cid:26) z (cid:27) (cid:26) z z (cid:27) (cid:26) (cid:18)z (cid:19) (cid:18)z (cid:19)(cid:27) α α β β β which is the first Chern class map of O (E). This proves the claim. V Since E is the exceptional divisor, it is independent of the other divisors in H1(Ω ). Thereforeimc ∩imδ =0;hencethe compositionmapβ ◦c isinjective. V 1 1 1 (cid:3) ′ We now concentrate on the following restriction map d of the map d in (3.1): 1 1 d′ :H0(O (E))→H1(Ω (E)), 1 F V which is also regarded as a connecting homomorphism induced from the exact sequence 0→Ω →Ω (logF)→O →0 V V F tensored by O (E). V Lemma 3.10. We have H0(Ω (logF)(E))=0. V Proof. Let C be a general fiber of the elliptic fibration φ : Y → P1. By the projection formula we have H0(Ω (logF)(E))⊆H0(Ω (F +E))=H0(Ω (τ∗C−E)) V V V ⊆H0(Ω (τ∗C))=H0(Y,Ω (C)). V Y OntheotherhanditisnotdifficulttoshowthatH0(Y,ΩY(C))=H0(P1,ΩP1(1))= 0 by a similar method in [8, Lemma 2]. Therefore the assertion follows. (cid:3) By the above lemma, we have the following commutative diagram of exact se- quences: 0 0 (3.2) (cid:15)(cid:15) (cid:15)(cid:15) H0(ΩV ⊗OE(E)) H0(ΩV(logF)⊗OE(E)) δ (cid:15)(cid:15) (cid:15)(cid:15) 0 //H0(OF) // H1(ΩV) α1 //H1(ΩV(logF)) //0 β1 β2 0 //H0(OF(cid:15)(cid:15) (E)P)PPPγP◦dPd′1P′1PPPP// HP((1(ΩV(cid:15)(cid:15)(cid:15)(cid:15)γ(E)) α2 // H1(ΩV(lo(cid:15)(cid:15)(cid:15)(cid:15)gF)(E)) //0 H1(ΩV ⊗OE(E)) H1(ΩV(logF)⊗OE(E)) (cid:15)(cid:15) (cid:15)(cid:15) 0 0 10 HEESANGPARK,JONGILPARK,ANDDONGSOOSHIN Lemma 3.11. h0(Ω (logF)⊗O (E))=1. V E Proof. Let r :Ω (logF)⊗O →Ω ⊗O (F) V E E E be the restriction to the subsheaf Ω (logF) of the restriction map r⊗id:Ω | ⊗ V V E O (F)→Ω ⊗O (F). E E E Claim. kerr =O (−E): Let z , z be the local equations of F, E, respectively. E 1 2 Note that the restriction map r :Ω | →Ω is defined by V E E r :adz +bdz =adz . 1 2 1 Hence the map r is defined by dz dz 1 1 r a +bdz =a . 2 (cid:18) z (cid:19) z 1 1 It follows that kerr =kerr =O (−E). E By the claim we have ker[r⊗id:Ω (logF)⊗O (E)→Ω ⊗O (F +E)]=O (−E)⊗O (E)=O . V E E E E E E Hence we have the exact sequence 0→H0(O )→H0(Ω (logF)⊗O (E))→H0(im(r⊗id)). E V E Since H0(im(r⊗id))⊂H0(Ω ⊗O (F +E))=0, the result follows. (cid:3) E E Lemma 3.12. The composition map γ◦d′ :H0(O (E))→H1(Ω ⊗O (E)) 1 F V E is surjective. Proof. Onthesecondcolumnin(3.2),wehaveh0(Ω ⊗O (E))=1,h1(Ω )=21, V E V h1(Ω ⊗O (E))=2. Henceh1(Ω (E))=22. Sinceh0(O )=1andh0(O (E))= V E V F F 3, it follows from the first and the second columns that h1(Ω (logF)) = 20 and V h1(Ω (logF)(E)) = 19. Therefore, on the third column, we have h1(Ω (logF)⊗ V V O (E)) = 0 because h0(Ω (logF)⊗O (E)) = 1 by Lemma 3.11. Hence β is E V E 2 surjective. Note that H0(O ) ⊆ β (H1(Ω ))∩kerα in H1(Ω (E)). On the other hand, F 1 V 2 V since α and β are surjective, we have 1 2 (α ◦β )(H1(Ω ))=(β ◦α )(H1(Ω ))=H1(Ω (logF)(E)). 2 1 V 2 1 V V Therefore dim[β (H1(Ω ))∩kerα ]=1; hence H0(O )=β (H1(Ω ))∩kerα in 1 V 2 F 1 V 2 H1(Ω (E)). Thus V dim[β (H1(Ω ))∩H0(O (E))]=1. 1 V F Since kerγ = imβ , we have dim[kerγ∩H0(O (E))] = 1; hence the composition 1 F γ◦d′ is surjective because h0(O (E))=3 but h1(Ω ⊗O (E))=2. (cid:3) 1 F V E

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.