On symplectic resolutions in the Nonabelian Hodge Correspondence
athesispresentedforthedegreeof Doctor of Philosophy of Imperial College London and the Diploma of Imperial College by Andrea Tirelli
Department of Mathematics Imperial College 180 Queen’s Gate, London SW7 2AZ
November 2018 Icertifythatthisthesis,andtheresearchtowhichitrefers,aretheproductof my own work, and that any ideas or quotations from the work of other people, published or otherwise, are fully acknowledged in accordance with the standard referencing practices of the discipline.
Signed:
ii Copyright
The copyright of this thesis rests with the author and is made available under a Cre- ative Commons Attribution Non-Commercial No Derivatives licence. Researchers are free to copy, distribute or transmit the thesis on the condition that they at- tribute it, that they do not use it for commercial purposes and that they do not alter, transform or build upon it. For any reuse or redistribution, researchers must make clear to others the licence terms of this work.
iii Thesis advisor: Dr. Travis J. Schedler Andrea Tirelli
On symplectic resolutions in the Nonabelian Hodge Correspondence
Abstract
This thesis is devoted to the study of the symplectic algebraic geometry of the mod- uli spaces of the nonabelian Hodge correspondence: namely, we consider character varieties of (open) Riemann surfaces and the moduli spaces of semistable Higgs bun- dles on a closed Riemann surface and aim at giving an answer to the following two questions: are those moduli spaces symplectic singularities? Do they admit sym- plectic resolutions? In the case of Higgs bundles on closed Riemann surfaces, we are able to completely solve the problem, using the work of Bellamy and Schedler, [BS16b], in combination with Simpson’s Isosingularity theorem, [Sim94b, §10]. For what concerns the Betti moduli spaces, we take a di↵erent perspective and realise them as (singular open subsets of) multiplicative quiver varieties,[CS06], and we study the aforementioned problem for the latter varieties. The present work repre- sents a first step of a series of investigations on the symplectic algebraic geometry involved in the nonabelian Hodge correspondence.
iv To my family.
v Acknowledgments
Firstly, I would like to express my sincere gratitude to my advisor Dr. Travis Schedler for the continuous support of my PhD study and related research, for his patience, motivation, and immense knowledge. His guidance helped me in all the time of research and writing of this thesis. I could not have imagined having a better advisor and mentor for my PhD study.
Besides my advisor, I would like to thank the rest of my thesis committee: Dr. Ben Davison and Dr. Johannes Nicaise for their insightful comments and encouragement.
I would like to thank all the members of the London School of Geometry and Number Theory for making my PhD an unforgettable experience.
Last but most importantly, I would like to thank all the members of my family, to whom this thesis is dedicated: my parents and to my brother and sister for supporting me spiritually throughout writing this thesis and my my life in general.
This work was supported by the Engineering and Physical Sciences Research Coun- cil [EP/L015234/1]. The EPSRC Centre for Doctoral Training in Geometry and Number Theory (The London School of Geometry and Number Theory), University College London and Imperial College London.
vi Contents
Introduction 1
1Preliminaryresults 7 1.1 The nonabelian Hodge correspondence ...... 7 1.1.1 Character varieties ...... 9 1.1.2 Higgs bundles ...... 10 1.2 Symplectic singularities and their resolutions ...... 13 1.3 Symplectic resolutions of quiver and character varieties ...... 15
2Symplecticresolutionsof (X, n) 20 MH 2.1 Local structure of (X, n) ...... 21 MH 2.1.1 Infinitesimal deformations of Higgs bundles ...... 22 2.1.2 The Isosingularity theorem ...... 23 2.2 Proof of the main theorem ...... 25 2.3 Extensions and future directions ...... 29
3Multiplicativequivervarieties 32 3.0.1 Summary of results on character varieties ...... 32 3.0.2 Multiplicative quiver varieties with special dimension vectors .36 3.0.3 Character varieties as (open subsets of) multiplicative quiver varieties ...... 39 3.0.4 General dimension vectors ...... 40 3.0.5 Outline of the chapter ...... 43 3.1 Multiplicative quiver varieties ...... 45 3.1.1 Multiplicative preprojective algebras ...... 45 3.1.2 Reflection functors for ⇤q(Q) ...... 48 3.1.3 Moduli of representations of ⇤q(Q) ...... 49 3.1.4 Reflection isomorphisms ...... 51
vii 3.1.5 Poisson structure on (Q, ↵) ...... 51 Mq,✓ 3.1.6 Stratification by representation type ...... 52 3.2 Punctured character varieties as multiplicative quiver varieties .... 56 3.3 Singularities of multiplicative quiver varieties ...... 63 3.3.1 Singular locus of (Q, ↵)for↵ ⌃ ...... 64 Mq,✓ 2 q,✓ 3.3.2 The q-indivisible case ...... 66 3.3.3 The q-divisible case ...... 67 3.3.4 Proof of Corollary 3.0.8 ...... 71 3.3.5 The anisotropic imaginary (p(↵),n)=(2, 2) case ...... 71 3.4 Combinatorics of multiplicative quiver varieties ...... 72 3.4.1 (2,2) cases for crab-shaped quivers ...... 73 3.5 General dimension vectors and decomposition ...... 80 3.5.1 Flat roots ...... 82
3.5.2 Fundamental and flat roots not in ⌃q,✓ ...... 85 3.5.3 Canonical decompositions ...... 87 3.5.4 Symplectic resolutions for q-indivisible flat roots ...... 90 3.5.5 Symplectic resolutions for general ↵ ...... 93 3.5.6 Classifications of symplectic resolutions of punctured charac- ter varieties ...... 94 3.5.7 Proof of Theorems 3.0.1 and 3.0.4 ...... 96
4Conclusionandfuturedirections 98 4.1 Non-emptiness of multiplicative quiver varieties ...... 99 4.2 Refined decompositions for multiplicative quiver varieties ...... 100 4.3 Symplectic resolutions and singularities ...... 102 4.4 Moduli of parabolic Higgs bundles and the Isosingularity Theorem ..103 4.5 Moduli of representations of 2-Calabi–Yau algebras ...... 107 4.6 Character varieties and Higgs bundles for arbitrary groups ...... 110 4.7 Moduli spaces of objects in 2-Calabi–Yau categories ...... 111
viii Introduction
This thesis is devoted to the study of certain moduli spaces appearing in the con- text of the nonabelian Hodge correspondence from the perspective of symplectic algebraic geometry. More precisely, we consider character varieties of Riemann sur- faces (with punctures) and the moduli spaces of semistable Higgs bundles on a smooth projective curve and tackle the problem of understanding whether they are symplectic singularities (´ala Beauville, [Bea00]) and whether they admit symplectic resolutions.
Despite the simple formulation of the problem, it turns out that its solution involves the use of many di↵erent techniques, coming from di↵erential geometry, algebra and representation theory: indeed, the nonabelian Hodge correspondence is a topic at the interface of such areas of mathematics. As such, the aim of this thesis is to initiate a series of investigations in what seems to be a very promising topic, given its multidisciplinary nature.
Higgs bundles and symplectic resolutions have become ubiquitous throughout alge- braic and di↵erential geometry, representation theory and mathematical physics. For instance, Higgs bundles, which first emerged about thirty years ago in Nigel Hitchin’s study of the self-duality equations on a Riemann surface [Hit87]andinCarlosSimp- son’s work on nonabelian Hodge theory, [Sim94a; Sim94b], play a role in many di↵er- ent areas of mathematics, including gauge theory, K¨ahler and hyperk¨ahler geometry, surface group representations, integrable systems, the Deligne–Simpson problem on products of matrices, and (more recently) mirror symmetry and Langlands dual- ity. On the other hand, the theory of (conical) symplectic resolutions has been widely studied not only in mathematics, but also in physics, and has applications and connections to representation theory, symplectic geometry, quantum cohomol- ogy, mirror symmetry, and equivariant cohomology. For a survey on some of these connections, see, e.g.,[BPW16]and[Bra+16].
Multiplicative preprojective algebras were first defined by Crawley-Boevey and Shaw
1 in [CS06], with the aim of better understanding Katz’s middle convolution opera- tion for rigid local systems, [Kat96]. Another important application contained in the seminal paper [CS06]isthesolutionofthemultiplicativeDeligne-Simpsonprob- lem in terms of the root data of a certain star-shaped quiver. Moduli spaces of representations of these algebras, in the sense of King, [Kin94], give rise to the so called multiplicative quiver varieties,whichcanbethoughtofasamultiplicative version of Nakajima’s quiver varieties, [Nak94]. A number of authors have stud- ied multiplicative quiver varieties since their definition: among others, Jordan, in [Jor14], considered quantizations of such varieties from a representation theoretic point of view by constructing flat q-deformations of the algebra of di↵erential oper- ators on certain a ne spaces; a more geometric approach was used by Yamakawa in [Yam08], where a symplectic structure on these moduli spaces was defined and studied – some of these results will be recalled in the next chapters. More recently, (derived) multiplicative preprojective algebras appeared in the study of wrapped Fukaya categories of certain Weinstein 4-manifolds, see [EL17]. In the recent work of Chalykh and Fairon [CF17], multiplicative quiver varieties are used to construct a new integrable system generalising the Ruijsenaars–Schneider system, which plays a central role in supersymmetric gauge theory and cyclotomic DAHAs. Moreover, we have been informed of the on-going work of Gammage and McBreen [Gam18], related to [BK16, §7], on mirror symmetry for multiplicative quiver varieties, which, at least in the case of ALE spaces (i.e. Dynkin quivers), are believed to be self- mirrors. Given their appearance in so many di↵erent contexts, it seems natural to perform a careful analysis on multiplicative quiver varieties from the point of view of symplectic algebraic geometry.
The subject of symplectic resolutions and symplectic singularities has recently gained importance in many areas of mathematics and physics. Their quantisations subsume many of the important examples of algebras appearing in representation theory (Cherednik and symplectic reflection algebras, D-modules on flag varieties and rep- resentations of Lie algebras, quantised hypertoric and quiver algebras, etc.). More- over, there is a growing theory of symplectic duality, or three-dimensional physical mirror symmetry ([Nak16; CHZ14; Bra+16]andmanyothers),betweenpairsof these varieties. Pioneering work of Braverman, Maulik and Okounkov shows that their quantum cohomology is also deeply tied to connections arising in representa- tion theory, which is expected to relate to derived autoequivalences of these varieties.
2 Since, as mentioned before, quiver varieties play such an important role here, it is expected that multiplicative quiver varieties will as well. Moreover, the varieties in question are instances of moduli spaces parametrizing geometric objects. The study of such spaces and their singularities is, in general, important in algebraic geometry.
In this thesis, we shall see that, for the moduli spaces we consider, the existence of symplectic resolutions is intrinsically linked to certain numerical and combinatorial data associated to such varieties. As an instance of such a connection, in Chapter 3 we study a combinatorial problem related to multiplicative quiver varieties. This aspect represents yet another reason for considering the topic very interesting.
We now outline the structure of the thesis and point out the main results contained in this work: we shall only state the most important theorems as all other ancillary results will be discussed in depth in the relevant chapters. Chapter 1 is devoted to recalling some preliminary results which will be extensively used throughout this thesis as well as to defining and giving the main properties of the varieties appearing in the nonabelian Hodge correspondence, which play a key role in this work. As such, we outline the main statements of the nonabelian Hodge correspondence and then we focus on the Betti and Dolbeault moduli spaces. More- over, in order to make this thesis as self-contained as possible, we recall the definition of symplectic singularities and resolutions and present some important properties that will be extensively used in Chapters 2 and 3. Note that, for sake of brevity, we limited ourselves to treating such topics only from the perspective of symplectic algebraic geometry: as such, we leave out from our analysis some other important aspects, more closely related to mathematical physics and di↵erential geometry – such as the gauge-theoretic point of view of the nonabelian Hodge correspondence: nonetheless, for convenience of the reader, we carefully point out the relevant refer- ences in the existing literature. Finally, in the last part of the chapter, we summarise the main results of the work of Bellamy and Schedler, [BS16b], which inspired the theorems and conjectures outlined in this thesis. Indeed, in [BS16b], the authors carry out a detailed study of the symplectic algebraic geometry of Nakajima’s quiver varieties and character varieties of compact orientable surfaces and, in order to do so, they establish a general procedure which can be adapted to other contexts and, in particular, to the cases treated in this thesis.
In Chapter 2 we deal with the first of the two main topics treated in this thesis:
3 Higgs bundles on smooth projective curves. More precisely, we consider the moduli space (X, n) of semistable Higgs bundles of rank n and degree 0 on a compact MH Riemann surface X of positive genus g and give an answer to the two following questions: is (X, n) a symplectic singularity? Does (X, n)admitasym- MH MH plectic resolution? As we have already pointed out, we approach these problems essentially using the techniques developed in [BS16b]and,inparticular,theworkof the authors on symplectic resolutions of character varieties. Another fundamental ingredient in the proof of the main results of Chapter 2 is the Isosingularity theorem, proved by Simpson in [Sim94b, §10], which establishes an ´etale-local isomorphism at corresponding points between the Betti, the de Rham and the Dolbeault moduli spaces. Combining these results, we manage to prove the following
Theorem. (Theorems 2.2.3, 2.2.4, 2.2.5) The following holds:
1. The moduli space (X, n) is a symplectic singularity; MH 2. (X, n) admits a symplectic resolution if and only if g(X), the genus of X, MH is 1 or (n, g(X)) = (2, 2) or n =1.
As we shall see in Chapter 2,inthecasewhen(g(X),n) =(2, 2) and g(X),n>1, we 6 show that (X, n)isfactorialandhasterminalsingularities,whichimpliesthat MH symplectic resolutions can not exist. We limit ourselves to consider Higgs bundles, instead of the more general G-Higgs bundles: nonetheless, it seems that some of the ideas used for (X, n)couldbeextendedtothemodulispacesofsemistable MH G-Higgs bundles and we give an account of this aspect at the end of the chapter.
In Chapter 3, we focus on the Betti side of the nonabelian Hodge correspondence in the noncompact case: namely, we consider character varieties of Riemann sur- faces with punctures. In order to study such varieties, we consider moduli spaces of representations of multiplicative preprojective algebras – a class of quiver algebras defined by Crawley-Boevey and Shaw in [CS06]. Considering such algebras and their representations enables us to study the symplectic algebraic geometry of character varieties of open surfaces and is of interest in its own sake. We begin the chap- ter by introducing, following [CS06], multiplicative preprojective algebras,denoted
q Q0 by ⇤ (Q), where Q is a quiver and q (C⇤) aparameter;wethenrecalltheir 2 most important properties, from the perspective of both algebraic geometry and representation theory. We outline the construction of the varieties parametrising
4 their (semistable) representations and we study the singularities of such varieties. We detect the singular locus and, in certain cases, prove that the singularities are symplectic. Note that one of the most critical aspects in the study of multiplica- tive quiver varieties is related to the fact that it is still not known when they are nonempty. In Chapter 3,wediscussinmoredepththisissueandthecasesinwhich one is able to give a positive answer. Nonetheless, all the theorems contained in the chapter have to take into account this subtlety. This is shown, e.g. in one of the main results, which is reported below. In the following result, we will assume that the dimension vector defining the multiplicative variety lies in a special region,
Q0 denoted by ⌃q,✓,ofthesetofalldimensionvectors,N ,whereQ0 denotes the set of vertices fo the quiver Q.Thedefinitionofsuchasubsetanditspropertiesare given in full details in Chapter 3.
Theorem. (Theorem 3.0.7) Let ↵ ⌃ and assume that ↵ =2 for N and 2 q,✓ 6 2 q,✓ p( )=2. Then, assuming it is non-empty, (Q, ↵) satisfies the following: Mq,✓
• its normalisation is a symplectic singularity;
• if ↵ is q-indivisible, then for suitable generic ✓0, it admits a symplectic resolu- tion of the form (Q, ↵) (Q, ↵); Mq,✓0 !Mq,✓
• if ↵ = m for ⌃q,✓ and m 2, and q,✓(Q, ) = , then q,✓(Q, ↵) 2 M 6 ; M does not admit a symplectic resolution. Moreover, for suitable generic ✓0, (Q, ↵) is a singular factorial terminalisation. In fact, (Q, ↵) itself Mq,✓0 Mq,✓ contains a singular, factorial, terminal open subset.
In order to apply the above result to the nonabelian Hodge correspondence, we also construct a connection between multiplicative quiver varieties and punctured character varieties by proving that, with a clever choice of the quiver and the dimen- sion vector, one can embed any punctured character variety – by which we mean character varieties of Riemann surfaces with punctures – inside an appropriate mul- tiplicative quiver variety as a singular open subset, see [CS06], which is where the inspiration for this result came from.
We then proceed to proving some generalisations to the above result, to multiplica- tive quiver varieties whose dimension vector does not lie in ⌃q,✓ and, to achieve this goal, we consider a decomposition of multiplicative quiver varieties which is analo- gous to Crawley-Boevey’s decomposition of (ordinary) quiver varieties, see [Cra02;
5 BS16b]. Moreover, we perform some combinatorial computations to classify all the possible (2, 2) cases, for which we conjecture that there exists a symplectic resolu- tion. Analogously to Chapter 2,thetechniquesweusetostudymultiplicativequiver varieties are taken from the analysis carried out in [BS16b] for Nakajima’s quiver varieties.
The last chapter contains a number of conjectures that arise naturally from the work carried out in Chapters 2 and 3.Moreprecisely,fortheDolbeaultmodulispaces, we conjecture that the results proved in Chapter 2 could be extended to the case of parabolic (G-)Higgs bundles; on the other hand, motivated by [CS06, §3], [EE07] and the work of Bocklandt, Galluzzi and Vaccarino of moduli spaces of represen- tations of 2-Calabi–Yau algebras, [BGV16], we conjecture that, in the non-Dynkin case, multiplicative preprojective algebras are indeed 2-Calabi–Yau: we give some arguments in favour of such a conjecture and outline some possible applications. Finally, in the last part of the chapter, we discuss some more general ideas for how to give a unified proof of the results contained in Chapters 2 and 3.Inorderto do that, we consider the setting of 2-Calabi–Yau categories and make conjectures on the geometric properties of the associated moduli spaces of (polystable) objects. If such conjectures were true, they would represent general theorems, of which the results on the Betti and Dolbeault moduli spaces are particular cases. We hope to come back to these very interesting topics in future research.
6 The di culty lies not so much in developing new ideas as in escaping from old ones. John Maynard Keynes 1 Preliminary results
As outlined in the Introduction, in this chapter we recollect the main results and fix the notation which will be used throughout the following chapters. More precisely, the first section is devoted to an introduction to the nonabelian Hodge correspon- dence and the associated moduli spaces. Then, we provide the reader with the necessary notions of birational geometry needed to understand the content of the following chapters and lastly, we give a brief account on the main results of [BS16b], whose techniques are used extensively throughout the thesis.
1.1 The nonabelian Hodge correspondence
As suggested by the name, the nonabelian Hodge correspondence can be thought of as a nonabelian version of the well known Hodge Theorem, which gives an isomor- n phism between the de Rham cohomology HdR(X, C) and the Dolbeault cohomology Hq(X, ⌦p)ofacompactK¨ahlermanifoldX.Puttingthisresulttogether p+q=n with the classical de Rham Theorem, one obtains isomorphisms:
n n q p HB(X, C) ⇠= HdR(X, C) ⇠= H (X, ⌦ ), p+q=n M
7 n where HB(X, C)isthesingularcohomologyofthemanifoldX with coe cients in the field C. In the nonabelian Hodge correspondence, C is substituted by a complex algebraic group G and the above cohomology spaces are replaced by the so-called Betti, de Rham and Higgs (or Dolbeault) moduli spaces respectively, which all have a much richer geometric structure than their abelian counterparts. More explicitly, in the case of G = GL(n, C), one considers the following spaces.
• The Betti moduli space B(X, n): the space of representations ⇡1(X) M ! GL(n, C)modulotheconjugationactionofGL(n, C), which is also known as the character variety X(g, n), recalled below, in Section 1.1;
• The de Rham moduli space dR(X, n): the moduli space of rank n holomor- M phic vector bundles on X equipped with a flat connection;
• The Higgs (or Dolbeault) moduli space H (X, n): the moduli space of semistable M Higgs bundles of degree 0 and rank n,whichwerecallbelow,inSection1.1.
From the work of Hitchin [Hit87], Donaldson [Don87], Corlette [Cor88]andSimpson [Sim92], we know that one can construct natural bijections
(X, n) = (X, n) = (X, n). (?) MB ⇠ MdR ⇠ MH
Moreover, the nonabelian Hodge correspondence states that much more is true: indeed, as a consequence of results in [Sim94b], we have the following:
Theorem 1.1.1. Denote by : (X, n) (X, n) and : (X, n) MB !MdR MdR ! (X, n) the natural bijections mentioned in (?). Then, one has that MH
(1) in an isomorphism of the associated complex analytic spaces;
(2) is a homeomorphism of topological spaces.
The fact that Theorem 1.1.1.(1)isnotjustaset-theoreticbijection,butaniso- morphism of complex analytic spaces, is a key ingredient in the proof of the main result of Chapter 2,asitenablesustotransfertheformalisomorphismgivenby the Isosingularity theorem to another formal isomorphism between di↵erent spaces. This will become clear in Section 2.1 of the next chapter.
8 In general, the moduli spaces above are analytic varieties: let sm(X, n)denote M⇤ their smooth locus, for = B,dR and H.Thesearecomplexmanifoldswiththe ⇤ same real manifold structure. Indeed, we have the following
Theorem 1.1.2. [Sim94b] The complex manifolds sm(X, n), sm(X, n) and MB MdR sm(X, n) are isomorphic as real di↵erentiable manifolds. MH Remark 1.1.3. Even though they have the same real di↵erentiable structure, the above manifolds are not isomorphic as complex manifolds.
Remark 1.1.4. Even though we stated the above theorem in the special case when G = GL(n, C), it is worth pointing out that, by the work of Simpson, [Sim94b], the result holds for an arbitrary algebraic group G.Weshallcomebacktothispoint and explain the implications of such a general version of the above theorem at the end of the next chapter, in Section 2.3.
In the next two subsections, we study in more depth the properties of the Betti and Dolbeault moduli spaces.
1.1.1 Character varieties
Let X be a smooth complex projective curve of genus g,andletG be a reductive al- gebraic group over C.ConsiderthespaceYG = Hom(⇡1(X),G)ofhomomorphisms from the fundamental group of X to the group G. Using the presentation by gener- ators and relations of ⇡1(X), we can give YG the structure of a ne variety: indeed, we know that ⇡ (X) = a ,...,a ,b ,...,b /R, 1 ⇠ h 1 g 1 gi g where R is the relation R = i=1[ai,bi]and[a, b]denotesthecommutator[a, b]:= 1 1 2g aba b .Then,wecanembedQ YG into G as follows:
Y G2g,⇢ (⇢(a ),...,⇢(b )), G ! 7! 1 g
2g which is equivalent to considering YG as the subvariety of G cut out by the equation
g
[Ai,Bi]=1, i=1 Y
9 for A ,B G, i =1,...,g.ThegroupG acts by conjugation on G and one may i i 2 define the G-character variety of X as the categorical quotient (see [MFK02])
X(g, G)=YG G.
Algebraically, this is just
G X(g, G)=Spec(C[YG] ),
the spectrum of the ring of G-invariant functions on YG. Remark 1.1.5. In the notation X(g, G)weomittedX since the character variety depends only on the topology of X,i.e.onitsgenus,andnotonthecomplex structure.
Despite their simple definition, character varieties have a very rich geometry and have been the subject of a large body of literature. For the purposes of this thesis, we will be interested in the case where G = GL(n, C)andwewillusethenotation X(g, n) for the character variety X(g, GL(n, C)), where X is a compact Riemann surface of genus g.
1.1.2 Higgs bundles
In what follows we shall give a brief overview of Higgs bundles and their moduli spaces. Our main references are [GR15] and the seminal papers of Hitchin [Hit87] and Simpson [Sim92; Sim94a; Sim94b].
Definition 1.1.6. A Higgs bundle on X is a pair (E, ), where E is a holomorphic vector bundle on X and , the Higgs field,isanEnd(E)-valued 1-form on X,i.e. H0(End(E) ⌦1 ). 2 ⌦ X
In order to have a moduli space with a meaningful geometric structure, one has to consider bundles of fixed rank and degree which satisfy the stability condition defined below. Without such a stability condition, the moduli problem would not admit a coarse moduli space in the category of schemes and one would have to consider (X, n) as an algebraic stack, which we want to avoid. Nonetheless, the MH interested reader may consult [CW18].
10 Even though we will limit ourselves to studying Higgs bundles on Riemann surfaces, it is important to point out that the definition we give below extends to higher dimensional projective manifolds, see [Sim94b]. For the sake of completeness, we report it here.
Definition 1.1.7. Let M be an d-dimensional complex K¨ahler manifold. A Higgs bundle on E is given by a pair (E, ), where E is a holomorphic vector bundle on M and is a section of the sheaf End(E) ⌦1 satisfying the following integrability ⌦ M condition: =0. ^ Remark 1.1.8. Clearly, the definition of Higgs bundle on a curve is a particular case of the above definition, since, when d =1,theintegrabilityconditionaboveis automatically satisfied.
Definition 1.1.9. A Higgs bundle (E, ) on X is semistable if for any nontrivial subbundle F of E such that (F ) F ⌦1 ,onehas ⇢ ⌦ X
µ(F ) µ(E), where µ(E):=deg(E)/rank(E) is the slope of a vector bundle. The Higgs bundle (E, )issaidtobestable if the strict inequality holds for all strinct subbundles. Moreover, (E, )issaidtobepolystable if it is either stable or a direct sum of stable Higgs bundles with the same slope.
Remark 1.1.10. Explaining the origin of the notion of Higgs bundles and the use of such a terminology, first introduced by Hitchin, is beyond the scope of this thesis, and thus we would like only to highlight that Higgs bundles were defined in the context of the study of certain self-duality equations on a Riemann surface. The interested reader may wish to refer to the original paper [Hit87]andtothereferences mentioned in the introduction for further details.
As mentioned before, imposing the (semi)stability condition makes it possible to have a moduli space that has su ciently nice properties. The construction of such a moduli space was firstly carried out by Hitchin, in the rank 2 case [Hit87], and generalized to arbitrary rank by Nitsure [Nit91], via the use of Geometric Invariant Theory. We summarise this in the following theorem, which describes the structure of the moduli space in the case when the g(X) 2. Before stating this result, we 11 need to give a definition of S-equivalence between Higgs bundles. This definition goes as follows: given a semistable Higgs bundle (E, ), it is possible to show that it admits a Jordan-H¨older filtration
E E, 0 ⇢···⇢ of -invariant subbundles such that Ei/Ei 1 is a stable Higgs bundle of the same slope for every i.Theisomorphismclassoftheassociatedgradedbundle iEi/Ei 1 is unique, and semistable Higgs bundles are said to be S-equivalent if their associated graded bundles are isomorphic.
We now define the moduli functor for the problem of parametrizing S-equivalence classes of semistable Higgs bundles and report the fundamental result that states that such a moduli functor has a coarse moduli space which is a quasi-projective variety.
Definition 1.1.11. Let X be a smooth projective curve of positive genus g, S a C-scheme and fix d, n two integers with n 1. A family of semistable Higgs bundles on X of degree d and rank n parametrized by S is an S-flat family of vector bundles of rank n and degree d on S X, X S,togheterwithamorphism ⇥ E! ⇥