arXiv:1003.5217v3 [hep-th] 28 Oct 2010 g lse over classes ogy INTRODUCTION I. h asessetu sgvnb eti cohomology certain by given is dimensions, space-time spectrum four massless to the ten from string heterotic theory. compact- of string classes of various the ifications for in problem from motivation this originates of Our however, appearance problem, own. this its approaching of question mathematical CONTENTS I.SmayadConclusions and Summary III. d) c) b) a) Algorithm Computational A Bundles: Line of Cohomology I loih o iebnl cohomology bundle line for Algorithm II. .Mteaia Background Mathematical A. [email protected] [email protected] [email protected] [email protected] .Introduction I. ti la httecmuaino iebnl cohomol- bundle line of computation the that clear is It is fal nsprymti opcictoso the of compactifications supersymmetric in all, of First .Frhreape 4 examples Further D. .Peiiais2 Preliminaries A. .Apeiiayagrtmcnetr 4 conjecture algorithm preliminary A C. B. .Tefia loih ojcue6 conjecture algorithm final The E. .Sefchmlg 8 3. cohomology Sheaf 2. Sheaves 1. .Idxterm 9 9 theorems divisors Index and 5. bundles line Holomorphic 4. 2) in,weetemnflso neetaecmlt nescin o heterotic intersections complete both are bundles. in interest vector modes of additional manifolds massless the computing c where cohomology for tions, valued point bundle line starting computing basic for algorithm an present We 2010) October 29 (Dated: USA ap Blumenhagen, Ralph 1) ehchmlg 8 cohomology Cech cnwegmn 7 Acknowledgement ehchmlg for cohomology Cech ˇ ˇ de:Pofo h loih 7 algorithm the of Proof Added: e ez- ufc 5 4 surface Pezzo-3 del surface Pezzo-1 del al nttt o hoeia hsc,Kh al CB S UCSB, Hall, Kohn Physics, Theoretical for Institute Kavli a-lnkIsiu ¨rPyi,Forne ig6 80805 6, F¨ohringer Ring f¨ur Physik, Max-Planck-Institut D dmninlvreisi ninteresting an is varieties -dimensional ,2, 1, P 2 a) ejmnJurke, Benjamin ,2, 1, b) 7 7 3 2 1 7 hrtnRahn, Thorsten loih yacul fnntiileape ae on based examples the non-trivial of of computations couple direct a by algorithm it find we non- it. form, after reveal generic eventually now, most to its that appropriate in to so used it economical, algorithm extending and trivially the easier that much us is to appears it However, dles pcsmgtlo ieipiil perdi h papers the toric in general appeared and implicitly to like theorem look Bott’s might spaces of generalization a the such of one instanton. case three-brane in Euclidean a modes is zero matter seven-branes instanton charged and the modes both zero count classes di- manifold. cohomology internal wrapped compact Such seven-brane three-dimensional of a on intersection visors the which to com- curves, due with over arise classes instance cohomology for bundle confronted line is puting one Calabi-Yau Here on B-type fourfolds. compactifications with F-theory in orientifolds or IIB D-branes Type like compactification a manifold in any tricks. classes invoking dependent without cohomology tool i.e. bundle a way, have line straightforward to these important determine be would to it that obvious pretty h rbe a ytmtclyapoce n solved and that approached literature revealed. systematically mathematics was completely the problem in really the was never it recently, was More papers those in heflsi ie ycmlt nescin vrtoric over intersections complete varieties by ambient given is threefolds pcrm ..o etrbnl ooooycass boils cohomology massless classes, the the cohomology determine to of bundle down vector computation of the i.e. How- cases, spectrum, three bundles. vector all holomorphic in stable ever, Calabi- provides of construction fibered class cover large elliptically a spectral for the hand, manifolds, other bun- Yau line the of On sums direct of dles. sequences bundle (exact) vector monad The via the hand, defined which is one for the literature. constructions, on extension the are, and ones in prominent discussed most three been have structions a three-fold Yau loih sfruae sacnetr wihi so is (which conjecture a as formulated is algorithm classes nti etrw ilpoedb oiaigthe motivating by proceed will we letter this In ntesrn hoyltrtr,sm niain how indications some literature, theory string the In string of kinds other in appears also problem same The arXiv:1003.5217 M¨unchen, Germany 2 oee oorkoldeteagrtmused algorithm the knowledge our to however , L naBraa A93106, CA Barbara, anta vrtrcabetvarieties ambient toric over H i 1, ( Y c) ; yesrae ntrcvreissupporting varieties toric in hypersurfaces f V n emtRoschy Helmut and Y ftevco bundle vector the of ) h ags ls fsc Calabi-Yau such of class largest The . X n yeIBFter compactifica- IIB/F-theory Type and o h etrbnlsvroscon- various bundles vector the For . assoe oi aite.Ti sthe is This varieties. toric over lasses ehchmlg.Te the Then cohomology. Cech ˇ 1, d) H X i ( hrfr,i is it Therefore, . X V ; vraCalabi- a over L fln bun- line of ) 1,2 4 1 3 . 2 simple that it fits less than a single page). We postpone with the extracted set the computation of vector bundle cohomologies and N many further string theoretic applications to a more F = {xα = xα = ... = xα =0} . (4) 5 1 2 |Sα| extensive paper . α =1 The information contained in F is nothing else than the The high-performance C/C++ implementation Stanley-Reisner ideal SR of the toric variety X cohomCalg of the algorithm is available under SR = S , S ,..., S . (5) http://wwwth.mppmu.mpg.de/members/... 1 2 N ...blumenha/cohomcalg/. Note that the more physical definition of a toric variety given here is equivalent to the one in terms of cones, fans and triangulations. A choice of a phase corresponds to II. ALGORITHM FOR LINE BUNDLE COHOMOLOGY the choice of a triangulation of the polytope defined via D the I vertices νi ∈ Z . These vertices satisfy the R linear In this section we will present an algorithm for the relations computation of line bundle cohomologies Hi(X; O(D)) I (r) over a toric variety X. We will motivate the algorithm Qi νi =0 for r =1,...,R. (6) following the original development. The final conjecture i=1 is given in section II E. For further introductions to toric geometry consult7–11 or for a more mathematical treatment12,13. A. Preliminaries Given such a toric space, the hypersurface {xi = 0} naturally defines a divisor Di of X. By the relation of divisors and line bundles summarized in appendix A 4, To define the framework we are working in, let us this also defines a line bundle Li = OX (Di), which we briefly summarize some facts about toric varieties re- also denote as spectively, their physical counter part, the gauged linear (1) (R) sigma models (GLSM). Li = OX Qi ,...,Qi . (7) A toric variety is a generalization of projective spaces, The tangent bundle T of the toric variety is defined as which contains a set of homogeneous coordinates H := X the cokernel of the map β in the sequence {xi : i =1,...,I} equipped with a number R of equiva- I lence relations β ⊕R α ։ (r) (r) 0 −→ OX ֒−→ OX (Di) − TX −→ 0 . (8) Q1 QI (x1,...,xI ) ∼ (λr x1,...,λr xI ), (1) i=1
(r) In order to compute, for instance, the vector bundle co- for r =1,...,R with the weights Q ∈ Z and λ ∈ C∗ = i r homology Hq(X; T ), one needs as input only the line C−{0}. We will always assume that the weights are cho- X bundle cohomology classes Hq(X; O (D )) with q = sen such that one really gets a bona fide toric space. A X i 0,...,D. powerful method to study such spaces is the GLSM6, in The determination of these cohomology classes for gen- which the homogeneous coordinates are chiral superfields (r) eral line bundles L = O (k ,...,k ) is the problem we and the Q are charges under R Abelian U(1) gauge X 1 R i will address in this letter. Due to the Riemann-Roch- symmetries in a two-dimensional N = (2, 2) supersym- Hirzebruch theorem, the cohomology classes Hq(X; L) metric gauge theory. The Fayet-Iliopoulos parameters ξ r satisfy an index theorem (see appendix A 5) of these U(1) gauge symmetries, can be interpreted as the K¨ahler parameters of the geometric space. D The vanishing of the D-terms of a GLSM then splits χ(X; L)= (−1)q hq(X; L)= ch(L) Td(X) , (9) the space of Fayet-Iliopoulos parameters ξ ∈ RR into q=0 X R-dimensional cones, in which the D-flatness conditions where the right hand side can be computed easily, once can be solved. These cones are also called phases and the intersection form on X is known. correspond geometrically to the K¨ahler cones. In each The simplest toric varieties are the projective spaces such cone, for the D-terms to be solvable one finds sets n P , for which F = {x1 = x2 = ... = xn+1 = 0}. For of collections of coordinates these spaces, the line bundle cohomology is known due to Bott’s theorem Sα = xα1 , xα2 ,...,xα|Sα| with α =1,...,N, (2) n+k for q =0, k ≥ 0 which are not allowed to simultaneously vanish. The toric n q Pn −k− 1 variety of dimension D = I − R in this phase is then h ( ; O(k)) = n for q = n, k< −n . (10) defined as 0 else I C − F The goal is now to generalize this result to toric ambient X = (3) (C∗)R spaces. 3
B. Cechˇ cohomology for P2
To motivate the algorithm, we consider a simple ex- ample for which the Cechˇ cohomology can really be com- puted with pencil and paper. We want to compute the Cechˇ cohomology of P2. First we need a suitable covering by open sets, which for this projective space is provided by
2 Ui := [x1 : x2 : x3] ∈ P : xi =0 . (11) Then the Cechˇ cochains for the covering UP2 = {U1,U2,U3} take the form
0 Cˇ (UP2 ; F )= F (U1) ⊕ F (U2) ⊕ F (U3) 1 Cˇ (UP2 ; F )= F (U12) ⊕ F (U23) ⊕ F (U13) (12) 2 Cˇ (UP2 ; F )= F (U123), where we used the abbreviation Uij...k = Ui∩Uj ∩ ∩Uk. Now we consider a line bundle F = O(−p) with p > 0 and compute the local sections on all open sets, where the FIG. 1. The region subdivision of the lattice. Note, that it is homogeneous coordinates which are non-vanishing can meant that we have the inclusions Ri ∪ Rj ⊂ Rij for i < j appear with a negative exponent. This leads to and Si Ri ∪ Si m m −p−m −m x 1 x 2 x 1 2 , 2 1 2 3 by itself restricts to Cˇ (UP2 ; F ). Thus we get a sequence m ,m ≥ 0 for U 0 → 1 → 2 → 1 → 0 with trivial cohomology. An 1 2 3 (13) with m ≥ 0, m ≤−p − m for U . element (m1,m2) ∈ Rij \ Ri ∪ Rj first contributes to 1 2 1 2 1 2 Cˇ (UP2 ; F ) and restricts to Cˇ (UP2 ; F ). The induced m ≤−p − m , m ≥ 0 for U 1 2 2 1 sequence 0 → 0 → 1 → 1 → 0 again has trivial cohomol- These monomials can be represented by points ly- ogy. Finally, if ing in different regions Ri in the (m1,m2)-plane, cf. figure 1. (m1,m2) ∈ R123 \ Rij ∪ Ri , (16) i