Research Collection
Doctoral Thesis
Non-Abelian Orbifold Theory and Twisted Modules for Vertex Operator Algebras
Author(s): Gemünden, Thomas
Publication Date: 2020
Permanent Link: https://doi.org/10.3929/ethz-b-000446462
Rights / License: In Copyright - Non-Commercial Use Permitted
This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.
ETH Library diss. eth no. 26903
NON-ABELIANORBIFOLDTHEORYANDTWISTEDMODULES FORVERTEXOPERATORALGEBRAS
A thesis submitted to attain the degree of doctor of sciences of eth zurich¨ (Dr. sc. ETH Z¨urich)
presented by thomas gemunden¨ Master of Mathematics, University of Cambridge Master of Arts, University of Cambridge born on 05.02.1994 citizen of Germany
accepted on the recommendation of Prof. Dr. G. Felder, Referent Prof. Dr. C. A. Keller, Korreferent Prof. Dr. N. Scheithauer, Korreferent
2020 Thomas Gem¨unden, Non-Abelian Orbifold Theory and Twisted Modules for Vertex Operator Algebras Abstract
The topic of this thesis is the theory of holomorphic extensions of orbifolds of holomorphic vertex operator algebras. We develop an orbifold theory for non-abelian groups of the form Zq o Zp and give an explicit formula for the characters of the holomorphic extensions. Additionally, we show how groups of the form Zq o Zp can be realised as automorphism groups of lattice vertex operator algebras. Furthermore, we construct the action of automorphisms on twisted modules for lattice vertex operator algebras and give explicit expressions for the corresponding graded trace functions. Using these results, we conduct a survey of orbifolds by cyclic groups and groups of the form Zq o Zp of lattice vertex operator algebras associated to extremal lattices in dimensions 48 and 72. This is the first systematic survey of holomorphic vertex operator algebras at higher central charge, as well as the first systematic survey of non-abelian orbifold theories. We construct around 200 new holomorphic vertex operator algebras at central charge c = 48 and c = 72. These include those with the lowest number of low-weight states currently known among holomorphic vertex operator algebras of the respective central charges. These allow us to prove that there exist holomorphic vertex operator algebras that cannot be constructed as abelian orbifolds of lattice vertex operator algebras. Finally, we investigate the notion of ‘large-central charge limit’ of a family of vertex operator algebras. We define a vertex algebra that can be considered such a large-central charge limit for a family of permutation orbifolds and give a sufficient condition for its existence.
3 Zusammenfassung
Das Thema dieser Abhandlung ist die Theorie der holomorphen Erweiterungen der Orbifaltigkeiten von holomorphen Vertexoperatoralgebren. Wir entwickeln eine Theorie der Orbifaltigkeiten f¨urnicht-abelsche Gruppen der Form Zq o Zp und bestimmen einen expliziten Ausdruck f¨urdie Charaktere der holomorphen Erweiterungen. Zus¨atzlich, zeigen wir, wie Gruppen der Form Zq o Zp als Automorphismusgruppen von Gittervertexoperatoralgebren realisiert werden k¨onnen. Weiterhin konstruieren wir die Wirkung von Automorphismen auf die verdrehten Moduln von Gitter- vertexoperatoralgebren und bestimmen die dazugeh¨origengradierten Spurfuntionen. Mithilfe dieser Resultate unternehmen wir eine Untersuchung der Orbifaltigkeiten von zyklischen Gruppen und Gruppen der Form Zq o Zp f¨urGittervertexoperatoralgebren von extremalen Gittern in den Dimensionen 48 und 72. Dies ist die erste systematische Untersuchung von holomorphen Vertexoper- atoralgebren gr¨ossererzentraler Ladung, sowie die erste systematische Untersuchung von nicht-abelschen Orbifaltigkeiten. Wir konstruieren etwa 200 neue holomorphe Vertexoperatoralgebren der zentralen Ladungen 48 und 72. Darunter sind die holomorphen Vertexoperatoralgebren mit der jeweils kleinsten bekannten Anzahl von leichten Zust¨andenunter holomorphen Vertexoperatoralgebren ihrer zentralen Ladung. Diese erlauben uns zu beweisen, dass holomorphe Vertexoperatoralgebren existieren, die nicht als abelsche Orbifaltigkeiten von Gittervertexoperatoralgebren konstruiert werden k¨onnen. Zuletzt untersuchen wir das Konzept des ‘Grenzwertes zur grossen zentralen Ladung‘ einer Familie von Vertexoperatoralgebren. Wir definieren eine Vertexalgebra, die als Grenzwert zur grossen zentralen Ladung einer Familie von Permutationsorbifaltigkeiten angesehen werden kann und geben eine hinreichende Bedingung f¨urihre Existenz.
4 Acknowledgements
I would like to express my gratitude to Prof. Dr. Christoph Keller for his excellent supervision and his patience and encouragement throughout a remarkably eventful doctorate. I would like to thank Prof. Dr. Giovanni Felder for valuable discussions and advise, helpful comments on this text and finally for taking me in as his student. I am grateful to Prof. Dr. Nils Scheithauer for invaluable advise and helpful comments on this text. I would like to thank Prof. Dr. Gerald H¨ohnfor his kind hospitality during my stay in Manhattan. Finally, I would like to thank my colleague Simon Brun, for his fantastic work as a group organiser. My work is supported by the Swiss National Science Foundation Project Grant 175494.
5 Contents
Introduction 8
1 Preliminaries and Definitions 13 1.1 Formal Calculus ...... 13 1.2 Vertex Operator Algebras ...... 13 1.3 Automorphisms of Vertex Operator Algebras ...... 14 1.4 Modules and Twisted Modules ...... 14 1.5 Regularity and Rationality ...... 16 1.6 Holomorphic Vertex Operator Algebras ...... 17 1.7 Homomorphisms, Schur’s Lemma and Action of Automorphisms on Twisted Modules . . 17 1.8 Characters, Graded Trace Functions and Modular Invariance ...... 18 1.9 Intertwining operators, Fusion and Vertex Tensor Categories ...... 19
2 Orbifold Theory 22 2.1 Holomorphic Extensions of Vertex Operator Algebras ...... 22 2.2 Cyclic Orbifolds ...... 25 2.3 Non-Abelian Orbifolds ...... 27 2.3.1 Groups with Periodic Cohomology ...... 28 2.3.2 Semidirect Products of Cyclic Groups Zq oφ Zp ...... 29 2.3.3 Characters of Holomorphic Extensions for Zq oφ Zp ...... 30 3 Lattice Vertex Operator Algebras 33 3.1 Lattices ...... 33 3.1.1 Integral Lattices ...... 33 3.1.2 Lattice theta-functions and Modularity ...... 34 3.1.3 Lattice Automorphisms ...... 35 3.1.4 Smith Normal Form and Lattice Quotients ...... 36 3.2 Construction of Lattice Vertex Operator Algebras ...... 37 3.3 Automorphisms of Lattice Vertex Operator Algebras ...... 39 3.4 Graded Trace Functions for Lattice Vertex Operator Algebras ...... 40 3.5 Finite Automorphism Groups for Lattice Vertex Operator Algebras ...... 41 3.5.1 Lifting Semidirect Products of Cyclic Groups Zq oφ Zp ...... 43 4 Twisted Modules for Lattice Vertex Operator Algebras 47 4.1 Construction of Twisted Modules ...... 47 4.2 The Vacuum Representation ...... 49 4.3 A Basis for VL(ˆg)...... 52 4.4 Action of Automorphisms on Twisted Modules ...... 53 4.5 Graded Trace Functions for Twisted Modules ...... 53 4.6 Examples: Orbifolds of VE8 ...... 55 4.6.1 An S3-Orbifold ...... 55 4.6.2 A Z2 × Z2-Orbifold ...... 55 4.6.3 A Z3 × Z3-Orbifold ...... 56
5 Cyclic Orbifolds of Lattice Vertex Operator Algebras at c = 48 and c = 72 57 5.1 Computational Aspects — Characters of Holomorphic Extensions for Cyclic Groups . . . 57 5.1.1 Eta-Quotients and Modular Transformations of Lattice Theta-Functions ...... 57 5.1.2 Computing Characters of Cyclic Orbifolds ...... 60 5.2 The Orbifold Algorithm ...... 61 5.3 Cyclic Orbifolds of Extremal Lattices in d =48...... 62 5.3.1 Cyclic Orbifolds for the Lattice P48m ...... 63 5.3.2 Cyclic Orbifolds for the Lattice P48n ...... 63
6 5.3.3 Cyclic Orbifolds for the Lattice P48p ...... 65 5.3.4 Cyclic Orbifolds for the Lattice P48q ...... 65 5.4 Cyclic Orbifolds for the Lattice Γ72 ...... 66
6 Non-Abelian Orbifolds of Lattice Vertex Operator Algebras at c = 48 and c = 72 68 6.1 Zq oφ Zp-Orbifolds for d =48...... 68 6.1.1 Zq oφ Zp-Orbifolds for the lattice P48p ...... 68 6.1.2 Zq oφ Zp-Orbifolds for the lattice P48q ...... 68 6.1.3 Zq oφ Zp-Orbifolds for the lattice P48m ...... 69 6.1.4 Zq oφ Zp-Orbifolds for the lattice P48n ...... 69 6.2 Zq oφ Zp-Orbifolds for the Lattice Γ72 ...... 70 6.3 Bounds for Abelian Orbifolds ...... 70
7 The Large-N-Limit for Permutation Orbifolds 72 7.1 Oligomorphic Families of Permutation Groups ...... 72 7.1.1 Unitary Vertex Operator Algebras ...... 72 7.1.2 Oligomorphic Orbifolds ...... 73 7.1.3 Space of states ...... 74 7.1.4 Structure Constants ...... 75 7.1.5 An Example: SN ...... 77 7.2 The Limit of GL(N, q)-Orbifolds ...... 78
Bibliography 81
7 Preface
The topic of this text are certain algebraic structures, called vertex operator algebras. They can be used to construct full realisations of conformal field theories, special models in particle physics that play a central role in string theory. In informal terms, our goal will be to construct new vertex operator algebras from old ones and their symmetries and hence construct new conformal field theories. The theory of vertex operator algebras is a relatively new field of study with deep roots in different areas of mathematics and physics that appear entirely unrelated at first sight. First introduced in the 1980s by Richard Borcherds [Bor86] vertex operator algebras played a crucial role in his proof of the “monstrous moonshine” conjectures — a remarkable set of conjectures by Conway and Norton [CN79] revolving around an unexpected connection between the Monster finite simple group M and Klein’s j-function that was first established with the construction of the famous moonshine module V \ by Frenkel, Lepowsky and Meurman [FLM88]. This lead to various important developments regarding connections between algebraic structures and the theory of modular forms as well as in physics.
Introduction
Let V be a vertex operator algebra. A (2-dimensional) chiral conformal field theory can, roughly speaking, be considered a suitable category of V -modules together with corresponding intertwining operators satisfying certain modular invariance and convergence properties. The problem of constructing a conformal field theory from a given vertex operator algebra is hard because the representation theory of vertex operator algebras is in general poorly understood. A central problem in the field of conformal field theory is the problem of construction. One approach to construction that has been very successful in the past is orbifold theory. Let V be a vertex operator algebra and G an automorphism group of V . Then we denote the subalgebra of G-invariant elements of V by V G. The fundamental objective of orbifold theory is to construct and classify the conformal field theories associated to the vertex operator algebra V G. In the most general case, this is very much an open problem. The first challenge is to construct and study modules of V G. Since it has proved difficult to infer properties of V G-modules from those of V directly, a more fruitful approach is to study so-called g-twisted V -modules for g ∈ G that, in a sense, are considered a generalisation of V G-modules. The most successful construction method for g-twisted modules currently available associates a g-twisted V -module to a module for a certain associative algebra Ag(V ), called twisted Zhu’s algebra [DLM98, HY19]. Problems with this approach are that, in this framework, it is hard to study properties of general g-twisted modules and that Ag(V ) and its modules are, in general, hard to construct and classify themselves. Promising advances were recently made by Huang [Hua19a], who constructed a universal lower-bounded g-twisted [g] V -module McB that can be considered the vertex operator algebraic analogue to a Verma module for a finite-dimensional Lie algebra. These objects have the potential to facilitate the construction of twisted modules as well as systematic research into their properties. On the problem of constructing and studying twisted intertwining operators, much less progress has been made. Twisted intertwining operators among twisted modules associated to non-commuting automorphisms were, in fact, only introduced recently [Hua18]. The central questions in orbifold conformal field theory are the convergence of products, iterates and q-traces, the operator product expansion (also commonly referred to as associativity) and the modular invariance of twisted intertwining operators, especially in the nonabelian case. All these properties were proved by Huang in the case of untwisted intertwining operators for strongly rational vertex operator algebras [Hua05a, Hua05b]. A particularly interesting class of vertex operator algebras are the so-called holomorphic vertex operator algebras, containing among others the moonshine module V \. The defining property of a holomorphic vertex operator algebra is that it is its own unique irreducible module. The problem of constructing a conformal field theory from a holomorphic vertex operator algebra V then becomes trivial, as V is a conformal field theory by itself. The most prominent method to construct such vertex operator algebras is the lattice construction, where a holomorphic vertex operator algebra is associated to an even, self-dual lattice. In the very restricted setting where V is a holomorphic vertex operator algebra and G a finite group of automorphisms, much more is known about orbifold theory. Specifically, significant progress has recently been made on the problem of classifying extensions of the fixed-point vertex operator algebras V G by a suitable category of modules to a new holomorphic vertex operator algebra. While there has been
8 significant attention in recent years, a full theory of holomorphic extensions is only available for individual families of finite groups. A recent breakthrough in this area was the theory of cyclic orbifolds by van Ekeren, M¨ollerand Scheithauer [vEMS17]. They were able to give a necessary and sufficient condition for the existence of holomorphic extensions for the case where G is cyclic and showed that the characters of the extensions can be calculated by making use of the modular transformation properties of graded trace functions on V without having to construct the irreducible V G-modules explicitly. In the past years, this program of constructing holomorphic vertex operator algebras using orbifold theory has been extremely successful. A recent breakthrough was the almost complete classification and construction of all holomorphic vertex operator algebras at central charge c = 24, a research endeavour of over two decades [FLM88, DGM90b, Don93,DGM96,Lam11, LS12,Lam11,Miy13,vEMS17,M¨o16,SS16,LS16a, LS16b, LL16]. The remaining piece in this classification and one of the big open problems in the theory of vertex operator algebras is a proof of the uniqueness of the moonshine module V \. Recent attempts to find a proof were unsuccessful, and no promising pathways towards finding a proof in the near future have been proposed. The theory of cyclic orbifolds was, furthermore, used in Scott Carnahan’s recent proof of the Generalized Moonshine Conjecture [Car12]. It was also shown [MS19] that at c = 24 all 70 possible holomorphic vertex operator algebras with non-trivial weight-1 space can be constructed as extensions of cyclic orbifolds of lattice vertex operator algebras. Terry Gannon and David Evans recently made significant progress on the problem of orbifolding by general finite groups in [EG18]. Building on the fact that if V is holomorphic and G finite, the module G category V − mod is equivalent to the module category of a twisted Drinfeld double Dω(G) − mod, with ω ∈ Z3(G, C∗), they were able to fully classify all possible holomorphic extensions. The cohomological twist ω acts, in a certain sense, as an obstruction to orbifolding, so that it is central for us to understand if ω is in the trivial class for a given V G. In general, this is an open problem. In particular, ω can generally not be inferred from modular data alone [MS17]. To construct orbifold vertex operator algebras, we first need to construct the irreducible V G-modules. When G is finite, they can be obtained as submodules of g-twisted V -modules. Specifically, the centraliser G CG(g) acts projectively on the irreducible g-twisted module V (g) and the irreducible V -modules arise as the multiplicities of the corresponding irreducible projective representations of CG(g); a property referred to as the Schur-Weyl-Type Duality [DLM00, DRX17]. The critical step in finding these submodules is to understand how elements of CG(g) act on the irreducible g-twisted module. Twisted modules for lattice vertex operator algebras were first explicitly described by Lepowsky [Lep85] and later again by Bakalov and Kac [BK04] so that in this case substantial calculations are possible. Another topic of current interest is the structure of holomorphic vertex operator algebras at a central charge larger than 24, where very little is known. Attempts to find a full classification at any higher central charge are widely considered hopeless because the corresponding result for even self-dual lattices already appears to be forever out of reach: The number of even self-dual lattices in higher dimensions is known to be inconceivably large, and no useful method to construct them all seems to be anywhere in sight [CS99]. However, a question that is still worth asking is whether anything can be said about which holomorphic vertex operator algebras can be constructed from a given one using the orbifold construction at higher central charge. It is, for example, known that, in principle, any even, self-dual lattice can be constructed from any other one of the same dimension using Kneser’s neighbour method [Kne57] in a finite number of steps. Using the theory of cyclic orbifolds, it is easy to show that this result can, in fact, be extended to lattice vertex operator algebras by orbifolding by inner automorphisms. Whether it might be possible to construct any holomorphic vertex operator algebra as an orbifold of a lattice vertex operator algebra, is a completely open problem. A further important application of vertex operator algebras and conformal field theory is the AdS/CFT correspondence that conjecturally maps theories of quantum gravity to certain families of vertex operator algebras that admit a large-central charge limit [HR15, BKM15, BKM16, KM19], but as of yet no mathematically precise definition of such a limit has been given. Families of particular interest are those with a small number of light states that are expected to correspond to theories of pure gravity [Wit07]. However, very few suitable vertex operator algebras are currently known, and vertex operator algebra with a minimal number of weight-2 states have been conjectured not to exist by physicists. The existence of conformal field theories with a smaller number of low-weight states has recently also been studied by Gerald H¨ohn[H¨o19]. As orbifold theory provides good control of the low-weight subspaces, it is well-suited as a tool to construct such vertex operator algebras.
9 Structure and Results
This text is structured as follows: In the first chapter, we give a review of basic definitions and results on vertex operators algebras, their representation theory, and their automorphisms. The second chapter first reviews the theory of holomorphic extensions of orbifolds of holomorphic vertex operator algebras and the theory of cyclic orbifolds. We show that the finite groups with periodic cohomology are exactly the groups for which the character of the holomorphic extension can be computed without the construction of twisted modules by using the modular invariance of graded trace functions. Then we discuss the orbifold theory of non-abelian groups with periodic cohomology of the form Zq o Zp. In particular, we show that:
3 ∗ Theorem. Let Zq o Zp be a finite group with periodic cohomology. Then H (Zq o Zp, C ) is cyclic, and 3 ∗ [ω] ∈ H (G, C ) is trivial if and only if the type 0 condition is satisfied for the cyclic subgroups Zp and Zq. Furthermore, we give an explicit expression for the character of the holomorphic extension in terms of characters of orbifolds by cyclic subgroups. In the third chapter, we discuss the construction of lattice vertex operator algebras, their automorphism groups, and graded trace functions. Then we study the properties of finite automorphism groups of lattice vertex operator algebras. In particular, we construct groups of the form Zq oφ Zp as automorphism groups of lattice vertex operator algebras. In the fourth chapter, we review the construction of twisted modules for lattice vertex operator algebras. We explicitly construct the lowest weight-space of an irreducible twisted module as an irreducible projective representation of a certain finite abelian group. Building on this, we construct the action of automorphisms on twisted modules for lattice vertex operator algebras and give explicit expressions for the graded trace functions on twisted modules. Finally, we construct irreducible modules for the fixed-point subalgebras for three examples of non-cyclic finite groups acting on the E8 lattice vertex operator algebra and verify that their characters exhibit the expected modular transformation properties. In the fifth chapter, we conduct a survey of cyclic orbifolds of lattice vertex operator algebras associated to extremal lattices in dimensions 48 and 72. This is the first systematic survey of holomorphic vertex operator algebras at higher central charge in the literature, and we construct around 150 new holomorphic vertex operator algebras at central charge c = 48 and c = 72, including ones with a small number of low-weight states. In particular, we prove the following theorem: Theorem. There exists a strongly rational, holomorphic vertex operator algebra with central charge c = 48, such that dim V(1) = 0 , dim V(2) = 48 , and a strongly rational, holomorphic vertex operator algebra with central charge c = 72, such that
dim V(1) = 0 , dim V(2) = 36 dim V(3) = 408 . In the sixth chapter, we systematically construct orbifolds for groups with periodic cohomology of the form Zq o Zp of lattice vertex operator algebras associated to the same lattices. This is the first systematic survey of holomorphic extensions for non-abelian orbifolds in the literature, and we construct around 50 new holomorphic vertex operator algebras. In particular, we prove that Theorem. A strongly rational, holomorphic vertex operator algebra with central charge c = 48, such that
dim V(1) = 0 , dim V(2) = 48 can be constructed as an orbifold of any lattice vertex operator algebra associated to one of the three extremal lattices in d = 48 with automorphisms of large order [Neb13], P48n,P48p and P48q. Furthermore, we prove that Theorem. There exists a strongly rational, holomorphic vertex operator algebra with central charge c = 72 and dim V(1) = 0 , dim V(2) = 12 , dim V(3) = 200 . By the obtaining a lower bound for the dimension of the weight-2-space for abelian orbifolds of lattice vertex operator algebras we can additionally show that:
10 Corollary. There exists a strongly rational, holomorphic vertex operator algebra V at central charge −3 −1 d = 72 with character χV (τ) = q + 12q + 200 + O(q), that cannot be constructed as a cyclic orbifold of a lattice vertex operator algebra by a lifted lattice automorphism.
N The AdS/CFT correspondence maps a given theory of quantum gravity to a family {V }N∈N of vertex operator algebras, whose central charges are parametrized by N. Physicists are most interested in the ‘large central charge limit’, that is the limit N → ∞ of this family. These notions of families of vertex operator algebras and their limits have not been defined mathematically, and much less investigated systematically. In the seventh chapter, we investigate the notion of the large-central charge limit for families of permutation orbifolds. We define a vertex algebra that can be considered such a large-central charge limit and give a sufficient condition for its existence depending only on the permutation group involved. Furthermore, we investigate a family of permutation orbifolds by general linear groups GL(N, q). In particular, we show that for such orbifolds the conformal weights of all twisted modules diverge in the large-central charge limit. This thesis is based on the three papers [GK19c,GK19b] and [GK19a].
Outlook
Due to recent advances, many interesting and important open problems in the field of orbifold conformal field theory are now accessible [Hua20]. Some problems that will be addressed in future work are the following:
• Grading restricted quotients of universal twisted modules: In the past year, Huang has made enormous advances in the theory of twisted modules and solved some long-standing open problems [Hua19a, Hua19b, Hua19c]. Particularly with his construction of the universal twisted modules, many further open problems and conjectures on twisted modules can be solved. A very fundamental open problem that needs to be addressed is the following: As the orbifold-theoretic analogue to Verma modules, Huang’s universal twisted modules have, by construction, infinite-dimensional homogeneous subspaces. An open problem is to construct quotients of such universal twisted modules whose homogeneous subspaces are finite-dimensional and to find out under which conditions such quotients exist. This problem is an orbifold-theoretic analogue of the well-known result in the theory of finite-dimensional Lie algebras that Verma modules generated by highest weights have finite-dimensional irreducible quotients if the highest weights are dominant and integral. Apart from the fundamental importance of this question, improved construction methods for twisted modules would enable us to study orbifold theories associated to further classes of vertex operator algebras. Currently, substantial calculations in orbifold theory are limited largely to lattice vertex operator algebras. The first step towards this objective is to establish the connection between twisted Zhu’s algebras and universal twisted modules. • Convergence and modular invariance of twisted intertwining operators for lattice orbifolds: The central questions in orbifold conformal field theory are the convergence of products, iterates and q-traces, the operator product expansion (also commonly referred to as associativity) and the modular invariance of twisted intertwining operators, especially in the non-abelian case. In particular, while current construction methods for orbifold conformal field theories are, for the most part, restricted to their existence and character, solving this question would enable the construction and study of full orbifold conformal field theories. We want to approach this question by generalising Huang’s methods for (untwisted) intertwining operators [Hua05a, Hua05b] for some explicit examples of non-abelian orbifold vertex operator algebras constructed in this thesis. • The structure of lattice orbifolds: Greg Moore asked a very fundamental question: Can every holomorphic vertex operator algebra be constructed as an orbifold theory of a lattice vertex operator algebra? This is an extremely difficult problem and, as of yet, there is very little intuition for how it might best be approached. In fact, there is no evidence that the answer to this question should be ”yes”. Even for holomorphic vertex operator algebras of central charge c = 24, where, at first sight, it appears to be the case, that all theories can be constructed as orbifold theories of lattice
11 vertex operator algebras, in fact, a proof of the uniqueness of the moonshine module is still missing. On the other hand, if it were possible to prove that the answer to this problem is yes, then it would open up a potential pathway to prove the uniqueness of the moonshine module. We can reformulate this problem as follows: What types of holomorphic vertex operator algebras are orbifold theories constructed from a lattice vertex operator algebra? To approach this question, we will construct and classify further orbifold theories from lattice vertex operator algebras in order to make deductions about their general structure. Making use of our experience with orbifold theories constructed from lattice vertex operator algebras gained in previous work and the new construction method for twisted modules, we will be able to construct interesting examples that will give us some fundamental insights into this difficult open problem. • More holomorphic extensions: One of the large obstacles to the construction of further holomorphic extensions is to understand if the cohomological twist associated to a holomorphic, strongly rational vertex operator algebra and a finite automorphism group vanishes. As of yet, this problem has only been solved for a few very simple families of finite groups. In future work, we will aim to find conditions for the vanishing of the cohomological twist for further groups. A good first target will be the generalised quaternion groups and, by extension, the finite groups with periodic cohomology. • Conformal field theory with large mass-gap: A related problem is that of conformal packing [H¨o19].We would like to construct vertex operator algebras, whose minimal conformal weight, that is the smallest positive conformal weight of a Virasoro highest-weight vector, is as large as possible. Examples of holomorphic vertex operator algebras with minimal conformal weight 2 are known, but none have been constructed for higher minimal conformal weights, and physicist have conjectured that they may not exist. A further research goal would be to construct a vertex operator algebra with large minimal conformal weight using techniques developed in this thesis and possibly disprove this conjecture. In a large survey of lattices and their automorphism groups, we could only find two lattice automorphism groups whose orbifolds could realise such a holomorphic vertex operator algebra. In future work, we will construct and classify the lifts of these groups and the corresponding orbifolds.
12 1 Preliminaries and Definitions
1.1 Formal Calculus
Let W be a C-vector space. We will denote the various spaces of formal series in the formal variable x as:
−1 P k • W x, x = { k∈Z akx : ak ∈ W }, the space of (doubly-infinite) formal Laurent series J K P k • W x = { k∈Z akx : ak ∈ W }, the space of formal power series J K ≥0 −1 P k • W [x, x ] = { k∈Z akx : ak ∈ W, all but finitely many ak = 0}, the space of formal Laurent polynomials • W [x] = {P a xk : a ∈ W, all but finitely many a = 0}, the space of formal polynomials k∈Z≥0 k k k
P k • W ((x)) = { k∈Z akx : ak ∈ W, ak = 0 for sufficiently negative k}, the space of truncated formal Laurent series
P k • W {x} = { k∈C akx : ak ∈ W }, the space formal series with complex exponents For an introduction to the theory of formal power series, we refer the reader to the standard texts on vertex operator algebras such as [LL04] and [FHL93].
1.2 Vertex Operator Algebras
The following definition of a vertex operator algebra was first given in [FLM88] and is based on Borcherd’s definition of vertex algebras [Bor86]. In particular, the central defining property is the Jacobi identity. An equivalent definition in terms of the weak commutativity of vertex operators is given in [Kac98]. Definition 1.2.1. A vertex operator algebra of central charge c consists of a graded vector space (space of states) M V = V(n) with dimV(n) < ∞ for all n ∈ Z≥0 (1.2.1) n∈Z together with a linear map (vertex operator map)
−1 X −n−1 Y (·, x): V → (EndV ) x, x , v 7→ Y (v, x) = vnx (1.2.2) J K n∈Z
and two distinguished elements |0i ∈ V(0) (vacuum vector) and ω ∈ V(2) (conformal vector) such that the following axioms are satisfied for all u, v ∈ V :
• (truncation) umv = 0 for m 0,
• (vacuum) Y (|0i, x) = 1V , • (creation) Y (v, x)|0i = v + O(x), • (Jacobi identity) x − x x − x x−1δ 1 2 Y (u, x )Y (v, x ) − x−1δ 2 1 Y (v, x )Y (u, x ) 0 x 1 2 0 −x 2 1 0 0 (1.2.3) −1 x1 − x0 = x2 δ Y (Y (u, x0)v, x2), x2
13 Let X W −n−2 Y (ω, x) = Ln x . (1.2.4) n∈Z W Then the modes Ln satisfy the following relations: • (Virasoro) The relations of the Virasoro algebra at central charge c: c [L ,L ] = (m − n)L + δ (m3 − m)1 , (1.2.5) m n m+n n+m,0 12 V
• (translation) Y (L−1v, x) = ∂xY (v, x),
• (grading) L0|V(n) = n1V(n) and V(n) = 0 for n sufficiently negativ. We can also rewrite the Jacobi identity as a condition for the modes. Lemma 1.2.1 ( [Kac98]). Let V be a vertex operator algebra. Then for any a, b, c ∈ V the modes satisfy the Borcherds identity
∞ ∞ ∞ X m X n X n ((a b) c = (−1)j a b c − (−1)j+n b a c. (1.2.6) j n+j m+k−j j m+n−j k+j j n+k−j m+j j=0 j=0 j=0
Definition 1.2.2. We say a vertex operator algebra V is of CFT-Type if
V(n) = 0, for all n < 0. (1.2.7)
1.3 Automorphisms of Vertex Operator Algebras
An automorphism of a vertex operator algebra is a linear bijection that preserves the vertex operators, vacuum vector and conformal vector. They will play a central role in this text.
Definition 1.3.1. Let V be a vertex operator algebra and g a bijective, linear map V → V . Then g is called a V -automorphism if for any u, v ∈ V
gY (u, x)v = Y (gu, x)gv. (1.3.1)
The group of V -automorphisms will be denoted Aut(V ).
Furthermore, we have the following notion of inner automorphisms of vertex operator algebras:
Definition 1.3.2. Let V be a vertex operator algebra and a ∈ V(1) a homogenous element of weight 1. Then the map σa = exp(2πia0) is an automorphism of V . Such automorphisms are called inner automorphisms of V . The inner automorphism group will be denoted by Inn(V ).
The group-theoretic structure of automorphism groups of vertex operator algebras has been investigated in [DG02]. For finitely-generated vertex operator algebras (cf. [FHL93]), the automorphism group is isomorphic to an algebraic group, and for strongly rational vertex operator algebras, it has been conjectured that the connected subgroup is given by Inn(V ).
1.4 Modules and Twisted Modules
Here we will review the basic definitions of modules and twisted modules for vertex operator algebras. For a very readable introduction to the theory of twisted modules refer to [GG09].
Definition 1.4.1. [DLM00][g-twisted V -module] Let V be a vertex operator algebra of central charge c ∈ C and let g be an automorphism of V of order n. Then a g-twisted V -module W is determined by the following data:
14 • (space of states) a C-graded vector space
W = ⊕λ∈CWλ
with weight wt(w) = λ for w ∈ Wλ extended linearly, such that the homogeneous subspaces are finite dimensional dim(Wλ) < ∞ and the grading is bounded from below
Wλ+r/n = {0} for r sufficiently small and fixed λ ∈ C .
• (vertex operators) a linear map
1/n −1/n YW (·, x): V → EndC(W )[[x , x ]] X W −k−1 (1.4.1) a → YW (a, x) = ak x k∈(1/n) Z
W where for each w ∈ W, ak w = 0 for sufficiently large k, subject to the following axioms:
• (twist compatibility) For each a ∈ V r
X W −k−1 YW (a, x) = ak x (1.4.2) k=−r/n+Z
• (left vacuum axiom) YW (1, x) = idW .
• (translation axiom) For any a ∈ V, [T,YW (a, x)] = ∂xYW (a, x). • (twisted Jacobi identity) For a ∈ V r and b ∈ V , x − x x − x x−1δ 1 2 Y (a, x )Y (b, x ) − x−1δ 2 1 Y (b, x )Y (a, x ) = 0 x W 1 W 2 0 x W 2 W 1 0 0 (1.4.3) −1 x1 − x0 r/n x1 − x0 x2 δ YW (Y (a, x0)b, x2) x2 x2
• (Virasoro relations) Let X W −n−2 YW (ω, x) = Ln x . (1.4.4) n∈Z W Then the modes Ln satisfy the relations of the Virasoro algebra at central charge c: m3 − m [LW ,LW ] = (m − n)LW + δ id c (1.4.5) m n m+n 12 m+n W W for m, n ∈ Z and L−1 = T and L0 w = wt(w)w for homogenous w ∈ W . Note that a 1-twisted V -module is commonly referred to as a V -module or, to emphasise the distinction, an untwisted V -module. In order to distinguish them from the other, closely related types of modules defined below, g-twisted V -modules are sometimes referred to as ordinary g-twisted V -modules. Moreover, the following alternate notions of (twisted) modules for vertex operator algebras were defined in [DLM97] and [DLM98]. Definition 1.4.2 (Weak g-twisted module). A weak g-twisted V -module is a not necessarily graded vector space M, together with a vertex operator map
1/n −1/n YW (·, x): V → EndC(W )[[x , x ]] X W −k−1 (1.4.6) a → YW (a, x) = ak x k∈(1/n) Z
such that the twisted Jacobi identity and the Virasoro relations are satisfied.
15 Definition 1.4.3 (Admissible g-twisted module). An admissible g-twisted V -module is a weak g-twisted L V -module W that admits a grading W = 1 W (k) such that for all homogenous v ∈ V k∈ n Z vW (m)W (k) ⊆ W (k + wt(v) − m − 1). (1.4.7)
We have the usual notions of reducibility of twisted V -modules and irreducible twisted V -modules. Note that the compatibility of the vertex operators with the grading implies the following: Definition 1.4.4. If W is an irreducible twisted V -module then the grading will be of the form M W = Wρ+k/n, (1.4.8) k∈N where we call ρ the conformal weight of W . A vertex operator algebra V has itself the structure of a V -module. As such it is often called the adjoint module. Furthermore, there is a natural notion of contragredient (read dual) twisted modules that was introduced in [FHL93, Xu00, Xu01].
0 Definition 1.4.5 (Contragredient twisted module). Let (W, YW ) be a g-twisted V -module. Let W be 0 0 graded dual of W . Then W together with the vertex operator (YW ) defined by
0 0 0 xL(1) −2 L(0) −1 h(YW ) (v, x)w , wi = hw ,YW (e (−x ) v, x )wi (1.4.9) is a g−1-twisted V -module called the contragredient module.
1.5 Regularity and Rationality
The representation theory of vertex operator algebras is, in general, not well understood. Extensive general results are only available under some rather restrictive regularity conditions that we will discuss in this section. The following definition of rationality is due to [DLM97]. An alternate, but equivalent definition was given in [Zhu96]. Definition 1.5.1 (Rationality). A vertex operator algebra V is called rational if every admissible V -module is isomorphic to a direct sum of irreducible admissible V -modules. Note that by [DLM98], if V is a rational vertex operator algebra, then up to isomorphism there are only finitely many irreducible V-modules and also only finitely many irreducible g-twisted V -modules for any automorphism g. The following condition has been defined by Zhu [Zhu96]:
Definition 1.5.2 (C2-Cofiniteness). Let V be a vertex operator algebra. Then is said to be C2-cofinite if the space C2 := spanC(a−2b | a, b ∈ V ) (1.5.1) has finite codimension in V .
By the results of [Miy04] and [DLM98], C2-cofiniteness is equivalent to the semi-simplicity and finite- dimensionality of Zhu’s algebra A(V ), an associative algebra, whose irreducible modules are in 1 to 1 correspondence with irreducible V -modules. It is conjectured that rationality implies C2-cofiniteness. By [DLM00], a consequence of rationality and C2-cofiniteness of a vertex operator algebra is the rationality of its central charge and the conformal weights of its irreducible modules. Note that for vertex operator algebras the notions of simplicity and irreduciblity coincide. Definition 1.5.3. A vertex operator algebra V is called simple if it is irreducible as a V -module. Definition 1.5.4. A vertex operator algebra V is called self-contragredient if it is isomorphic to V 0 as a V -module. By [Li94], self-contragredience implies, in particular, that V can be equipped with an invariant, symmetric, bilinear form, the vertex operator algebraic analogue, to the Killing form in the theory of finite-dimensional Lie algebras.
16 Definition 1.5.5 (Regularity). A vertex operator algebra V is called regular if every weak V -module is a direct sum of irreducible V -modules.
Note that by [ABD04], if V is of CFT-Type, regularity is equivalent to rationality and C2-cofiniteness. By [DY12], regularity is equivalent to the semi-simplicity of the weak module category. In this text, we will mostly be interested in vertex operator algebras that satisfy all of these conditions. We define: Definition 1.5.6. A vertex operator algebra is strongly rational if it is simple, self-contragredient, rational, C2-cofinite and of CFT-Type. A central open conjecture in the orbifold theory of strongly rational vertex operator algebras is the following: Conjecture 1.5.1. If V is strongly rational and G ∈ Aut(V ) a finite group, then the fixed-point vertex operator algebra V G is also strongly rational. In this text, we will assume that any vertex operator algebra is strongly rational unless indicated otherwise. Note that this condition is rather restrictive. Not in particular that there are many important instances of vertex operator algebras that are not rational or C2-cofinite, such as affine vertex operator algebras at admissible level [AM09] and the bosonic ghost vertex algebra [RW15a].
1.6 Holomorphic Vertex Operator Algebras
So-called holomorphic vertex operator algebras have the defining property that their adjoint module is their unique irreducible module. Definition 1.6.1 (Holomorphicity). A rational vertex operator algebra V is called holomorphic if the adjoint module V is the only irreducible V -module up to isomorphism.
That, in particular, makes them the simplest conformal field theories with only one module and one intertwining operator. As such, they have been of interest to both the mathematics [DM94, H¨oh96]and physics [God89,Sch93a,Sch93b,Mon94] communities for a long time. The most famous example is, of course, the moonshine module V \. So far, work has mostly focussed on holomorphic vertex operator algebras at central charge c = 24. Let us finally discuss some consequences of holomorphicity that were shown by Dong, Li and Mason [DLM00].
Lemma 1.6.1. Let V be a C2-cofinite, holomorphic vertex operator algebra. Then 1. its central charge c is a multiple of 8. 2. there exists a unique irreducible g-twisted V -module for every automorphism g of finite order.
1.7 Homomorphisms, Schur’s Lemma and Action of Automorphisms on Twisted Modules
In this section, we will define homomorphisms for twisted modules and present the appropriate manifesta- tion of Schur’s lemma. As a consequence, we will see that for two commuting automorphisms g and h, h induces a homomorphism of g-twisted modules (and vice versa).
Definition 1.7.1. [LL04][V -homomorphism] Let V be a vertex operator algebra and W1 and W2 (g-twisted) V -modules. Then a V -homomorphism from W1 to W2 is a linear map ψ such that
ψ(YW1 (v, x)w) = YW2 (v, x)ψ(w), (1.7.1)
for all v ∈ V and w ∈ W1. The space of V -homomorphisms from W1 to W2 will be denoted by HomV (W1,W2). If W1 = W2 = W we call ψ a V -endomorphism and the space of such maps will be denoted EndV (W ).
17 Theorem 1.7.1. [LL04][Schur’s Lemma] Let V be a vertex operator algebra and W1 and W2 irreducible (g-twisted) V -modules. Then
( ∼ ∼ C, if W1 = W2 HomV (W1,W2) = (1.7.2) 0, if W1 W2 so that in particular for any irreducible V -module W ∼ EndV (W ) = C . (1.7.3)
Now we can create new twisted modules by inserting an automorphism into the corresponding vertex operators.
Lemma 1.7.2. Let V be a vertex operator algebra, g and h automorphisms of V and (W, YW (·, x)) a −1 g-twisted V -module. Then (W, YW (h·, x)) is a h gh-twisted V -module. Hence we find that if g and h commute, h induces a homomorphism of g-twisted modules. We are particularly interested in the case where V is holomorphic. Let V be a holomorphic vertex operator algebra, g and h automorphisms of V such that gh = hg and (W, YW (·, x)) an irreducible g-twisted V -module. Then by lemma 1.6.1 (W, YW (·, x)) and (W, YW (h·, x)) are isomorphic irreducible g-twisted V -modules.
Definition 1.7.2. Hence there exists a linear map φ(h): W → W such that
−1 φg(h)YW (v, x)φg(h) = YW (hv, x). (1.7.4)
Theorem 1.7.3. φg : CAut(V )(g) → Hom(W, W ) is a projective representation of CAut(V )(g) on W .
0 Proof. First we show that the map φg(h) is unique up to multiplication by a scalar. Let φg(h) be another −1 0 map satisfying Equation 1.7.4. Then φg (h)φg(h) ∈ EndV (W ) is a homomorphism of twisted modules as
−1 0 0 −1 φg (h)φg(h)YW (v, x)φg(h) φg(h) = YW (v, x) (1.7.5)
−1 0 and hence by Schur’s Lemma we find that φg (h)φg(h) ∈ C. Furthermore for two automorphisms h, k ∈ CAut(V )(g) we find that
−1 0 −1 −1 φg(hk) φg(h)φg(k)YW (v, x)φg(k) φg(h) φg(hk) = YW (v, x) (1.7.6)
−1 and hence that φg(hk) φg(h)φg(k) ∈ EndV (W ). Then there exists a cocycle cg(h, k) such that
φg(h)φg(k) = cg(h, k)φg(hk). (1.7.7)
Hence φg is indeed a projective representation.
1.8 Characters, Graded Trace Functions and Modular Invariance
In this section, we will discuss the characters and graded trace functions associated to (twisted) modules of vertex operator algebras. They are relatively simple invariants of a vertex operator algebra that encode important information about its representation theory that have been studied since the early days of conformal field theory [MS89] and orbifold theory [DHVW85,DHVW86,DGM90a]. The proof of the modular invariance of characters by Zhu [Zhu96] was a milestone in our field. Dong, Li and Mason [DLM00] later proved the modular invariance of graded trace functions. Famously, the Verlinde Formula that is going to be discussed in section 1.9 relates the modular transformation properties of characters for the modules of a vertex operator algebra to the tensor category structure of its module category. In sections 2.2 and 2.3.2, we will use the graded trace functions to make deductions about the module category of the fixed-point subalgebra V G for certain families of finite groups G. Using the action of commuting automorphisms on twisted modules we can define the following graded trace functions:
18 Definition 1.8.1 (Graded trace function). Let V be a C2-cofinite, holomorphic vertex operator algebra, g and h be commuting automorphisms and W the unique irreducible g-twisted module. Then graded trace functions are defined as LM −c/24 T (g, h, τ) = tr|W φg(h)q 0 , (1.8.1) where q = e2πiτ , τ ∈ H. Note that graded trace functions can also be defined for vertex operator algebras that are not holomorphic, when the g-twisted V -module W is h-stable, but in this text we are not interested in such cases. Special cases of these trace functions are the characters of twisted modules. Definition 1.8.2. Let W be a twisted V -module. Then its character is given by