ebook img

A mirror duality for families of $K3$ surfaces associated to bimodular singularities PDF

0.2 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 A mirror duality for families of $K3$ surfaces associated to bimodular singularities

K A mirror duality for families of 3 surfaces associated to bimodular singularities Makiko Mase 7 1 Key Words: K3 surfaces, toric varieties, Picard lattices 0 AMS MSC2010: 14J28 14M25 14C22 2 b e Abstract F Ebeling and Ploog [4] studied a duality of bimodular singularities 1 which is part of the Berglund–Hu¨bsch mirror symmetry. Mase and Ueda [7]showedthatthisdualityleadstoapolytopemirrorsymmetryoffami- ] G liesof K3surfaces. Wediscussin thisarticle howthissymmetryextends toa symmetry between lattices. A . h 1 Introduction t a m BimodularsingularitiesclassifiedbyArnold[1]haveadualitystudiedbyEbeling [ andPloog[4]analogoustoArnold’sstrangedualityforunimodularsingularities. 1 Namely, a pair ((B,f), (B′,f′)) of singularities B, B′ in C3 together with ap- v propriatedefiningpolynomialsf, f′ isdualifthematricesAf, Af′ ofexponents 7 of f and f′ are transpose to each other. Moreover, in some cases, such poly- 0 nomials are compactified as anticanonical members of 3-dimensional weighted 1 projectivespaceswhosegeneralmembersareGorensteinK3. Thestrangedual- 0 ityforunimodularsingularitiesisrelatedwiththepolytopemirrorsymmetryfor 0 . families of K3 surfaces that are obtained by compactifying the singularities by 2 Kobayashi [6] in a certain sense. In the study of bimodular singularities, Mase 0 7 and Ueda [7] extend the duality by Ebeling and Ploog to a polytope mirror 1 symmetry of families of K3 surfaces. More precisely, the following statement is : shown : v i X Theorem [7] Let ((B, f), (B′, f′)) be a dual pair in the sense of [4] of sin- r gularities B and B′ together with their defining polynomials f and f′ that are a respectively compactified into polynomials F and F′ as in [4]. Then, there exists a reflexive polytope ∆ such that ∆F ⊂ ∆ and ∆F′ ⊂ ∆∗. Here, ∆F and ∆F′ are respectively the Newton polytopes of F and of F′, and ∆∗ is the polar dual to ∆. (cid:4) Inthisarticle,weconsiderwhetherornotitispossibletoextendtheduality obtainedin[7]furthertothelatticemirrorsymmetryoffamiliesofK3surfaces. More precisely, our problem is stated as follows: Problem Let∆beareflexivepolytopeasin[7]. Doesthereexistgeneralmem- bers S ∈ F∆ and S′ ∈ F∆∗ such that an isometry T(S) ≃ Pic(S′)⊕U holds e e 1 ? Here, F∆ and F∆∗ are families of K3 surfaces associated to the polytopes ∆ and ∆∗, S denotes the minimal model of S, Pic(S) and T(S) are respectively the Picard and transcendental lattice of S. e e e The problem is answered in Theoreme3.2 together with an explicit descrip- tion in Proposition 3.1 of the Picard lattices Pic(∆) and Pic(∆∗) defined in section3,withranksρ(∆)andρ(∆∗),oftheminimalmodelofappropriategen- eralmembersinthefamilies. Themainresultofthisarticleissummarizedhere. In the sequel,the names of singularitiesfollow Arnold[1], and singularitiesin a same row of Table 1 are dual to each other in the sense of [4]. Proposition3.1andTheorem3.2 Let ∆ be the reflexive polytope obtained in [7]. For the following transpose-dual pairs, the polar duality extends to a lattice mirror symmetry between the families F∆ and F∆∗, where the Picard lattices −4 1 are given in Table 1. Here we use the notation C6 := . 8 1 −2 (cid:18) (cid:19) Singularity Pic(∆) ρ(∆) ρ(∆∗) Pic(∆∗) Singularity Q U ⊕E ⊕E 16 4 U ⊕A E 12 6 8 2 18 Z U ⊕E ⊕E 17 3 U ⊕A E 1,0 7 8 1 19 E U ⊕E⊕2 18 2 U E 20 8 20 Q U ⊕A ⊕E 16 4 U ⊕C6 Z 2,0 6 8 8 17 E U ⊕E ⊕E 17 3 U ⊕A Z 25 7 8 1 19 Q U ⊕E ⊕E 16 4 U ⊕A E 18 6 8 2 30 Table 1: Picard lattices for lattice mirror symmetric pairs(cid:4) Section2istodefine thepolytope-andlattice-mirrortheoriesinsubsection 2.3, andto define the transposeduality following[4]in subsection2.4, basedon a brief introductionto lattice theory in subsection2.1, andof toric geometryin subsection 2.2, where several formulas and results are stated without proof. The main theorem of this article is stated in section 3 following auxiliary results. The facts introduced in the previous section are used in their proof. Denote by ∆ the reflexive polytope obtained in [7] for a singularity B. As B is seen in Table 1, there are isometric Picard lattices Pic(∆ ) ≃ Pic(∆ ), Q12 Q18 and Pic(∆ )≃Pic(∆ ). We consider and affirmatively answer in Proposi- Z1,0 E25 tion 4.1 the following question as an application in section 4: Problem Are the families F (resp.F ) and F (resp.F ) es- ∆Q12 ∆Z1,0 ∆Q18 ∆E25 sentially the same in the sense that general members in these families are bira- tionally equivalent ? Acknowledgement The author thanks to the referee for his helpful comments particularly about the proof of the key Proposition 3.1 in the original manuscript. 2 2 Preliminary We start with having a consensus as to Gorenstein K3 and K3 surfaces. Definition 2.1 A compact complex connected 2-dimensional algebraic variety S with at most ADE singularities is called Gorenstein K3 if (i) K ∼ 0; and S (ii) H1(S, O ) = 0. If a Gorenstein K3 surface S is nonsingular, S is simply S called a K3 surface. (cid:4) 2.1 Brief lattice theory Alattice isa non-degeneratefinitely-generatedfree Z-module witha symmetric bilinear formcalled an intersectionpairing. The discriminant group of a lattice LisdefinedbyA :=L∗/L,whichisfinitely-generatedandabelian,whereL∗ := L Hom(L, Z) is dual to L. It is known that the order |A | of the discriminant L groupis equal to the determinant of any intersection matrix of L. Let us recall a standard lattice theory by Nikulin [8]: Corollary 2.1 (Corollary 1.13.5-(1) [8]) If an even lattice L of signature (t , t ) satisfies t ≥1, t ≥1, and t +t ≥3+lengthA , then, there exists + − + − + − L a lattice T such that L≃U ⊕T, where U is the hyperbolic lattice of rank 2. (cid:4) Inparticular,ifanevenlatticeLisofrkL>12,t ≥1,andt ≥1,then,there + − exists a lattice T such that L≃U ⊕T. SupposeLisasublatticeofalatticeL′ withinclusionι:L֒→L′. Denoteby L⊥ theorthogonalcomplementofLinL′. Theembedding ιiscalledprimitive, L′ and L called a primitive sublattice of L′ if the finite abelian group L′/ι(L) is torsion-free. In other words, if there is no overlattice that is an intermediate lattice between L and L′ of rank equal to the rank of L. Note that if a direct sumL ⊕L oflatticesisasublatticeofL′,then(L ⊕L )⊥ ≃(L )⊥ ⊕(L )⊥. 1 2 1 2 L′ 1 L′ 2 L′ For a K3 surface S, it is known that H2(S, Z) with the intersectionpairing isisometrictotheK3 latticeΛ thatisevenunimodularofrank22andsigna- K3 ture (3,19), being U⊕3⊕E⊕2, where E is the negative-definite even unimod- 8 8 ular lattice of rank 8. There is a standard exact sequence 0 → H1(S, O∗) →c1 S H2(S, Z) → 0 of cohomologies so H1(S, O∗) is inherited a lattice structure S from H2(S, Z). Define the Picard lattice Pic(S) of a K3 surface S as the group c (H1,1(S))∩H2(S, Z) with the lattice structure. The rank of Pic(S) is called 1 the Picard number, denoted by ρ(S). The Picard lattice is hyperbolic since a K3 surface is complex and algebraic, and is known to be a primitive sublattice of Λ under a marking H2(S, Z)→∼ Λ . K3 K3 If an evenhyperbolic lattice L of rank rkL≤20 has |A | being square-free, L L is a primitive sublattice of Λ . Indeed, if so, for any lattice L ⊂ L′′ ⊂ K3 ΛK3 the general relation |AL| = [L′′ : L]2|AL′′| implies [L′′ : L] = 1 thus L′′ ≃ L. Hence there is no overlattice of L. Moreover, by surjectivity of the period mapping [2], there exists a K3 surface S such that Pic(S) ≃ L. Let M ⊂ Λ be a hyperbolic sublattice. A K3 surface is M-polarised [3] if there K3 exists a marking φ such that all divisors in φ−1(Cpol) are ample, where Cpol M M is the positive cone in MR minus d∈∆M Hd, ∆M = {d ∈ M|d.d = −2}, and H ={x∈P(Λ )|x.d=0 for all d∈∆ }. d K3 M S Nishiyama [9] gives the orthogonal complements of primitive sublattices of type ADE of E in possible cases. 8 3 Lemma 2.1 (Lemma 4.3 [9]) There exist primitive embeddings of lattices of type ADE into E with orthogonal complements given as follows. All the nota- 8 −4 1 tion follows Bourbaki except C6 := . 8 1 −2 (cid:18) (cid:19) (A )⊥ ≃E (A )⊥ ≃E (A )⊥ ≃D (A )⊥ ≃A (A1)E⊥8 ≃A7⊕A (A2)E⊥8 ≃C66 (A3)E⊥8 ≃(−58) 4 E8 4 (D5)E⊥8 ≃D1 2 (D6)E⊥8 ≃A8 (D7)E⊥8 ≃A⊕2 (D )⊥ ≃(−4) (E4)⊥E8≃A4 (E5)⊥E8≃A3 6 E8 1 7 E8 (cid:4) 6 E8 2 7 E8 1 2.2 Brief toric geometry Here we summarize toric divisors and ∆-regularity. Let M be a rank-3 lattice with the standard basis {e , e , e }, N be its dual, and (, ) : M ×N → Z be 1 2 3 the natural pairing. From now on, a polytope means a 3-dimensional convex hull of finitely-many points in Z3 embedded into R3, namely, integral, and the origin is the only lattice point in the interior of it. LetP∆ bethetoricvarietydefinedbyapolytope∆inM⊗ZR,towhichone can associate a fan Σ whose one-dimensional cones, called one-simplices, are ∆ generatedbyprimitivelatticevectorseachofwhoseend-pointisanintersection point of N and an edge of its polar dual ∆∗ defined by ∆∗ :={y ∈N ⊗R|(x, y)≥−1 for all x∈∆}. LetP be the toricresolutionofsingularitiesinP . A toric divisor is adivisor ∆ ∆ admittingthetorusaction,identifiedwiththeclosureofthetorus-actionorbitof aonef-simplex. LetDivT(P∆)bethesetofalltoricdivisorsDi =orb(R≥0vi), i= 1,...,s on P∆, where vi is a primitive lattice vector, then, −KPf = si=1 Di. By a standard exact sequfence and a commutative diagram [10] ∆ P f 0 → M → DivT(P∆) → Pic(P∆) → 0 ↓ ↓ M → s ZfD → A (Pf) → 0 i=1 i 2 ∆ there is a system of linear equatLions among toric divisors f s (e , v )D =0, j =1,2,3. (1) j i i i=1 X Thus the solution set of the linear system is generated by (s − 3) elements corresponding to divisors which generate the Picard group Pic(P ) of P . ∆ ∆ Definition 2.2 A polytope is reflexive if its polar dual is also integral. (cid:4) f f The importance of that we consider reflexive polytopes is by the following: Theorem 2.1 c.f. [?] The following conditions are equivalent: (1) A polytope ∆ is reflexive. (2) The toric 3-fold P is Fano, in particular, general anticanonical members ∆ of P are Gorenstein K3. (cid:4) ∆ 4 Forareflexivepolytope∆,denotebyF thefamilyofhypersurfacesparametrised ∆ by the complete anticanonicallinear system of P . Note that general members ∆ in F are Gorenstein K3 due to the previous theorem so that they are bira- ∆ tionally equivalent to K3 surfaces by the existence of crepant resolution. Thus we call the family F a family of K3 surfaces. ∆ We recall from §3 of [?] the notion of ∆-regularity. Definition 2.3 Let F be a Laurent polynomial defining a hypersurface Z , F whose Newton polytope is a polytope ∆. The hypersurface Z is called ∆- F regular if for every face Γ of ∆, the corresponding affine stratum Z of Z is F,Γ F either empty or a smooth subvariety of codimension 1 in the torus T that is Γ contained in the affine variety associated to Γ. (cid:4) Itisshownin[?]that∆-regularityisageneralcondition,andsingularitiesof all ∆-regular members are simultaneously resolved by a toric desingularization of P . From now on, suppose a polytope ∆ is reflexive and S is a ∆-regular ∆ member whose minimal model S is obtained by a toric resolution. Definition 2.4 For a restriction r : P → S, let r : H1,1(P ) → H1,1(S) be e ∆ ∗ ∆ theinducedmapping. DefinealatticeL (S):=r (H1,1(P ))∩H2(S, Z)ofthe D ∗ ∆ intersection of the image of r and Hf2(S, Ze), and its orthofgonal compleement ∗ L (S):=L (S)⊥ in H2(S, Z). (cid:4) e f e 0 D H2(Se,Z) e Iet is knowne[6]that ρ(S) anderkL0(S) only depend onthe number oflattice points in edges of ∆ and ∆∗. Thus we define the Picard number ρ(∆):=ρ(S), andthe rankrkL :=rkeL (S)associaetedto ∆. Moreprecisely,denote by Γ∗ 0,∆ 0 in ∆∗ the dual face to a face Γ of ∆, and l∗(Γ) is the number of lattice poinets in the interior of Γ, and ∆[1] ethe set of edges in ∆. Let s be the number of one-simplices of Σ . Then ∆ rkL0,∆ = l∗(Γ)l∗(Γ∗)=rkL0,∆∗, (2) Γ∈X∆[1] ρ(∆) = s−3+rkL , (3) 0,∆ ρ(∆)+ρ(∆∗) = 20+rkL . (4) 0,∆ If l∗(Γ∗)=n and l∗(Γ)=m for an edge Γ of ∆, there is a singularityof type Γ Γ A with multiplicity m +1 on an affine variety associated to Γ. nΓ+1 Γ As we will see later, we only need formulas when rkL = 0 for the inter- 0,∆ section numbers of the divisors {r D } in H2(S, Z) given as follows. ∗ i r D2 =r D .r D = 2l∗(vi∗)−2 eif vi is a vertex of ∆∗, (5) ∗ i ∗ i ∗ i (−2 otherwise. If end-points of v and v are on the edge Γ∗ of ∆∗, then i j ij 1 if v and v are next to each other, i j r∗Di.r∗Dj = l∗(Γij)+1 if l∗(Γ∗ij)=0, and vi, vj are both vertices,(6) 0 otherwise. ThePicardlatticesofthe minimalmodels ofany∆-regularmembers,which are generated by components of restricted toric divisors, are isometric. Define 5 the Picard lattice Pic(∆) of the family F as the Picard lattice of the minimal ∆ model of a ∆-regular member with rank ρ(∆). The orthogonal complement T(∆)=Pic(∆)⊥ is called the transcendental lattice of F . ΛK3 ∆ 2.3 Mirrors We define the polytope- and lattice-mirror theories. 2.3.1 Polytope Mirror We focus on polytopes that “represent” the anticanonical members in a toric variety as is seen in subsection 2.2. Definition 2.5 A pair (∆ , ∆ ) of reflexive polytopes or a pair (F , F ) 1 2 ∆1 ∆2 of families of K3 surfaces associated to ∆ and ∆ is called polytope mirror 1 2 symmetric if an isometry ∆ ≃∆∗ holds. (cid:4) 1 2 2.3.2 Lattice Mirror For a K3 surface S, Pic(S) is the Picard lattice, and T(S) = Pic(S)⊥ is the ΛK3 transcendentallattice. AmirrorforfamilyofM-polarisedK3surfacesisdefined whenM isasublattice ofΛ ingeneral[3]. Here,wedealwiththemoststrict K3 case, namely, mirror for K3 surfaces with Picard lattice as their polarisation. Definition 2.6 (1) A pair (S, S′) of K3 surfaces is called lattice mirror sym- metric if an isometry T(S)≃Pic(S′)⊕U holds. (2) A pair (F, F′) of families whose general members are Gorenstein K3 sur- faces is lattice mirror symmetric if there exist general members S ∈ F and S′ ∈F′ pair of whose minimal models is lattice mirror symmetric. (cid:4) Note thata lattice mirrorpair (S, S′)of K3surfacessatisfies,by definition, rkPic(S′)+2=rkT(S)=22−rkPic(S), thus ρ(S′)+ρ(S)=20. (7) 2.4 Bimodular singularities and the transpose duality BeingclassifiedbyArnold[1]in1970’s,bimodularsingularitieshavetwospecific classes: quadrilateralandexceptional. Quadrilateralbimodularsingularitiesare 6 in number with exceptional divisor of type I∗, whilst exceptional are 14 in 0 numberwithexceptionaldivisoroftypeII∗, III∗ orIV∗ inKodaira’snotation. Anon-degeneratepolynomialf inthreevariablesiscalledinvertibleiff has three terms f = 3j=1xa1jya2jza3j such that its matrix Af := (aij)1≤i,j≤3 of exponents is invertible in GL (Q). 3 P Definition 2.7 c.f. [4] Let B = (0,(f = 0)) and B′ = (0,(f′ = 0)) be germs of singularities in C3. A pair (B,B′) of singularities is called transpose dual if the following three conditions are satisfied. (1) Defining polynomials f and f′ are invertible. (2) MatricesAf andAf′ ofexponentsoff andf′ aretransposetoeachother. 6 (3) f (resp. f′) is compactified to a four-term polynomial F (resp. F′) in |−KP(a)| (resp. |−KP(b)|), where P(a) (resp. P(b)) is the 3-dimensional weightedprojectivespacewhosegeneralmembersareGorensteinK3with weight a (resp. b) out of the list of 95 weights classified by [12][5][11]. (cid:4) Condition (1) and (2) is said that they are Berglund–Hu¨bsch mirror sym- metric. Ebeling-Ploog [4] show that there are 16 transpose-dual pairs among quadrilateral and exceptional bimodular singularities, exceptional unimodular singularities, and the singularities E , E , X , and Z . 25 30 2,0 2,0 3 The transpose dual and the lattice mirror For a transpose-dual pair (B,B′) of defining polynomial f (resp. f′) being compactified to a polynomial F (resp. F′), consider the Newton polytope ∆ F (resp. ∆F′) of F (resp. F′) all of whose corresponding monomials are fixed by an automorphism action on (F =0) (resp. (F′ =0)). Remark 1 AcompactifiedmemberF tof doesnotalwaysdefineaGorenstein K3 surface because ∆ may not be reflexive. F However,the Newton polytopes are extended to be reflexive and dual. Theorem 3.1 [7] For each transpose-dual pair (B,B′), there exists a reflexive polytope ∆ such that ∆F ⊂∆ and ∆F′ ⊂∆∗. (cid:4) Computing Picardlattices is generally difficult, but it seems possible for ∆- regular members by subsection 2.2. Let us reformulate our problem. Problem Forapolytope∆obtainedin[7],isapair(S, S′)ofminimalmodels of ∆-regular S ∈F∆ and ∆∗-regular S′ ∈F∆∗ lattice mirror symmetric ? e e First we study the rank rkL . 0,∆ Lemma 3.1 The list of rkL for the reflexive polytope ∆ obtained in [7] is 0,∆ given in Table 2. transpose-dual pair rkL transpose-dual pair rkL 0,∆ 0,∆ (Q , E ) 0 (Z , Q ) 2 12 18 17 2,0 (Z , E ) 0 (U , U ) 2 1,0 19 1,0 1,0 (E , E ) 0 (U , U ) 2 20 20 16 16 (Q , Z ) 0 (Q , Z ) 2 2,0 17 17 2,0 (E , Z ) 0 (W , W ) 3 25 19 1,0 1,0 (Q , E ) 0 (W , S ) 5 18 30 17 1,0 (Z , Z ) 1 (W , W ) 6 1,0 1,0 18 18 (Z , J ) 2 (S , X ) 6 13 3,0 17 2,0 Table 2: rkL 0,∆ Proof. The assertion follows from direct and case-by-case computation by formula (2) in subsection 2.2. 7 1. (Q , E ) The polytope ∆ is given in subsection 4.7 of [7]. There is 12 18 no contribution to rkL since l∗(Γ) or l∗(Γ∗) is zero for any edge Γ. Thus, 0,∆ rkL =0 by formula (2). 0,∆ Similarforthecases(Z , E ), (E , E ), (Q , Z ), (E , Z ), (Q , E ). 1,0 19 20 20 2,0 17 25 19 18 30 2. (Z , J ) The polytope ∆ is given in subsection 4.1 of [7]. The only con- 13 3,0 tributiontorkL isbyanedgeΓbetweenvertices(0, 0, 1)and(−2, −6, −9), 0,∆ whosedualΓ∗isbetween(8, −1, −1)and(−1, 2, −1)sol∗(Γ)=1andl∗(Γ∗)= 2. Thus, rkL =1×2=2 by formula (2). 0,∆ Similarforthecases(Z ,Z ), (Z ,Q ), (U ,U ), (U ,U ), (Q ,Z ), 1,0 1,0 17 2,0 1,0 1,0 16 16 17 2,0 (W ,W ), (W ,S ), (W ,W ), (S ,X ). (cid:3) 1,0 1,0 17 1,0 18 18 17 2,0 Corollary 3.1 No∆-and∆∗-regularmembersfortranspose-dualpairs(Z ,J ), 13 3,0 (Z ,Z ), (Z ,Q ), (U ,U ), (U ,U ), (Q ,Z ), (W ,W ), (W ,S ), 1,0 1,0 17 2,0 1,0 1,0 16 16 17 2,0 1,0 1,0 17 1,0 (W ,W ), (S ,X ) admit a lattice mirror symmetry. 18 18 17 2,0 Proof. For each ∆ associated to the presented pairs, by formula (4), ρ(∆)+ρ(∆∗)=20+rkL >20 0,∆ since rkL >0 by Lemma 3.1. Thus, the equation (7) does not hold. There- 0,∆ fore, ∆- and ∆∗-regular members do not admit a lattice mirror symmetry. (cid:3) Corollary 3.2 The restriction mapping r : Pic(P ) → Pic(S), for ∆-regular ∗ ∆ S ∈F issurjectiveforthetranspose-dualpairs(Q ,E ), (Z ,E ), (E ,E ), ∆ 12 18 1,0 19 20 20 (Q2,0,Z17), (E25,Z19), (Q18,E30). f e Proof. By Lemma 3.1, rkL = 0 for each case. By definition, rkL is 0,∆ 0,∆ equal to the rank of the orthogonal complement of r (H1,1(P )) in H2(S, Z). ∗ ∆ Thus, that rkL =0 means that r is surjective. (cid:3) 0,∆ ∗ f e By Corollary 3.1, we may only focus on the transpose-dual pairs appearing in Corollary 3.2, whose statement means moreoverthat Pic(∆) is generated by restricted toric divisors generating Pic(P ), and analogous to ∆∗. ∆ Let AL denote the discriminant grofup, qL the quadratic form, and discrL the discriminantofalattice L. Ifp=discrLisprime,thenA ≃Z/pZ. Before L stating our main results, note a fact in Proposition 1.6.1 in [8]. Suppose that latticesS andT areprimitivelyembeddedintotheK3latticeΛ . IfA ≃A K3 S T and q =−q , then, it is determined that the orthogonal complement S⊥ in S T ΛK3 Λ is T. And q =−q if and only if discrS =−discrT. K3 S T Proposition 3.1 The Picard lattice Pic(∆∗) for ∆ in Corollary 3.2 is as in Table 3, where singularities in a row are transpose-dual. In each case, one gets discrPic(∆)=−discr(U ⊕Pic(∆∗)), and APic(∆) ≃AU⊕Pic(∆∗). 8 −4 1 Denote C6 := . 8 1 −2 (cid:18) (cid:19) Singularity ρ(∆∗) Pic(∆∗) Singularity Q 4 U ⊕A E 12 2 18 Z 3 U ⊕A E 1,0 1 19 E 2 U E 20 20 Q 4 U ⊕C6 Z 2,0 8 17 E 3 U ⊕A Z 25 1 19 Q 4 U ⊕A E 18 2 30 Table 3: Pic(∆∗) for ∆ in Corollary 3.2 Proof. 1.Q and E The polytope ∆ is given in subsection 4.7 of [7] and the asso- 12 18 ciated toric 3-fold has 19 toric divisors D corresponding to the one-simplices i generated by vectors m =(1, 2, 2) m =(0, 1, 1) m =(−8, −11, −9) 1 2 3 m =(1, −2, 0) m =(1, 1, 0) m =(−4, −5, −4) 4 5 6 m =(−7, −10, −8) m =(−6, −9, −7) m =(−5, −8, −6) 7 8 9 m =(−4, −7, −5) m =(−3, −6, −4) m =(−2, −5, −3) 10 11 12 m =(−1, −4, −2) m =(0, −3, −1) m =(1, 0, 1) 13 14 15 m =(−5, −7, −6) m =(−2, −3, −3) m =(1, −1, 0) 16 17 18 m =(1, 0, 0) 19 and by solving the linear system (1) : 19 (e , m )D = 0 (j = 1,2,3), we i=1 j i i get linear relations among toric divisors P D ∼ −9D +3D −D −2D −3D −4D −5D −6D −7D −8D 1 4 5 7 8 9 10 11 12 13 14 −5D +D +2D −5D −D , 15 16 17 18 19 D ∼ D +10D −2D +2D +3D +4D +5D +6D +7D +8D 2 3 4 5 7 8 9 10 11 12 13 +9D +5D −D +6D +2D , 14 15 17 18 19 D ∼ −2D −2D +D −2D −2D −2D −2D −2D −2D −2D 6 3 4 5 7 8 9 10 11 12 13 −2D −D −D −D . 14 15 16 18 So the set {D |i 6= 1,2,6} of toric divisors is linearly independent. Let L be i the lattice generated by the set {r D |i 6= 1,2,6} of their restrictions to a ∆- ∗ i regular member. We shall check that L is primitively embedded into the K3 latticetoshowthatLisindeedthePicardlatticeofthefamilyF . Bycomputer ∆ calculationwithformulas(5)and(6),thedeterminantofanintersectionmatrix 9 of L is −3 since this matrix is given by −2 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0 −2 0 0 0 0 0 0 0 0 1 1 0 0 1 0   0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1  1 0 0 −2 1 0 0 0 0 0 0 0 0 0 0 0     0 0 0 1 −2 1 0 0 0 0 0 0 0 0 0 0     0 0 0 0 1 −2 1 0 0 0 0 0 0 0 0 0     0 0 0 0 0 1 −2 1 0 0 0 0 0 0 0 0     0 0 0 0 0 0 1 −2 1 0 0 0 0 0 0 0   ,  0 0 0 0 0 0 0 1 −2 1 0 0 0 0 0 0     0 0 0 0 0 0 0 0 1 −2 1 0 0 0 0 0     0 1 0 0 0 0 0 0 0 1 −2 0 0 0 0 0     0 1 0 0 0 0 0 0 0 0 0 −2 0 0 0 0     1 0 0 0 0 0 0 0 0 0 0 0 −2 1 0 0     0 0 1 0 0 0 0 0 0 0 0 0 1 −2 0 0     0 1 0 0 0 0 0 0 0 0 0 0 0 0 −2 1     0 0 1 0 0 0 0 0 0 0 0 0 0 0 1 −2      Since the discriminant of L is −3 that is square-free,there exists no overlattice of L ; indeed, if H ⊂ Λ were an overlattice of L, then, by the standard K3 relation −3 = [H : L]2discrH, we get [H : L] = 1 and discrH = −3 so that L≃H. Hence, LisprimitivelyembeddedintotheK3lattice. Byconstruction, L is indeed the Picard lattice of the family F . ∆ The dual polytope ∆∗ associates a toric 3-fold with 7 toric divisors D′ cor- i responding to the one-simplices generated by vectors v =(1, 1, −2) v =(1, −2, 1) v =(2, −3, 2) v =(−1, 0, 1) 1 2 3 4 v =(−1, 0, 0) v =(1, 0, −1) v =(1, −1, 0) 5 6 7 and by solving the linear system (1) : 7 (e , m )D′ = 0 (j = 1,2,3), we i=1 j i i get linear relations among toric divisors P D′ ∼2D′ +3D′ +D′, D′ ∼3D′ +4D′ +D′ +2D′, D′ ∼D′. 1 2 3 7 4 2 3 6 7 5 3 Thus the set {D′|i 6= 1,4,5} of toric divisors is linearly independent. Let L′ i be the lattice generated by the set {r D′|i 6= 1,4,5} of their restrictions to a ∗ i ∆∗-regularmember. By formulas(5) and(6), anintersectionmatrix associated to L′ is given by −2 1 0 1 1 0 0 0   0 0 −2 1  1 0 1 −2      that is equivalent to U ⊕A by re-taking the generators as 2 {r D′ +r D′, r D′, r D′, r D′ +r D′}. ∗ 2 ∗ 3 ∗ 3 ∗ 6 ∗ 3 ∗ 7 Since the discriminantofL′ is −3thatis square-free,there existsno overlattice of L′ ; indeed, if H′ ⊂ Λ were an overlattice of L′, then, by the standard K3 relation −3 = [H′ : L′]2discrH′, we get [H′ : L′] = 1 and discrH′ = −3 so that H′ ≃ L′. Hence, L′ is primitively embedded into the K3 lattice. By 10

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.