8 0. OverviewNote that the space <strong>of</strong> metric ribb<strong>on</strong> graphs MRG Γ (x) possesses a tractable system <strong>of</strong>coordinates, provided by the edge lengths. In particular, with respect to these coordinates,Ω is a c<strong>on</strong>stant 2-form.Furthermore, MRG Γ (x) has a simple descripti<strong>on</strong> — it is thequotient <strong>of</strong> a polytope by the acti<strong>on</strong> <strong>of</strong> the finite group Aut Γ. Therefore, modulo someanalytical details, we have the equalitylimN→∞V Γ (Nx)N 6g−6+2n = ∫MRG Γ (x)Ω 3g−3+n(3g − 3 + n)! .From the previous discussi<strong>on</strong>, this is essentially the integral <strong>of</strong> a c<strong>on</strong>stant volume formover a polytope. The integrati<strong>on</strong> is most easily evaluated after taking the Laplace transform,which results in the desired product over edges <strong>of</strong> Γ. Furthermore, the acti<strong>on</strong> <strong>of</strong> the1finite group Aut Γ <strong>on</strong> this polytope naturally introduces the factor <strong>of</strong> which appears|Aut Γ|<strong>on</strong> the right hand side <strong>of</strong> K<strong>on</strong>tsevich’s combinatorial formula.Part 3: Where does the combinatorial c<strong>on</strong>stant come from?Interestingly, the remaining factor <strong>of</strong> 2 2g−2+n <strong>on</strong> the right hand side <strong>of</strong> K<strong>on</strong>tsevich’s combinatorialformula is no simple matter to explain. In fact, its appearance boils down to thefollowing statement.Theorem 3.22. Let Γ be a trivalent ribb<strong>on</strong> graph <strong>of</strong> type (g, n) with n edges coloured white and theremaining 6g − 6 + 2n edges coloured black. Let A be the n × n adjacency matrix formed betweenthe faces and the white edges. Let B be the (6g − 6 + 2n) × (6g − 6 + 2n) oriented adjacencymatrix formed between the black edges. Thendet B = 2 2g−2 (det A) 2 .Perhaps surprisingly, there does not exist a purely combinatorial pro<strong>of</strong> <strong>of</strong> this statementin the literature. Instead, we essentially follow the argument from Appendix C <strong>of</strong> K<strong>on</strong>tsevich’spaper [26], which uses the torsi<strong>on</strong> <strong>of</strong> an acyclic chain complex associated to atrivalent ribb<strong>on</strong> graph. However, our expositi<strong>on</strong> is greatly expanded to make the pro<strong>of</strong>both more thorough and more elementary.Of course, K<strong>on</strong>tsevich’s combinatorial formula in itself is not a new result. What is novel, in thispart <strong>of</strong> the thesis, is the hyperbolic geometric approach and the explicit c<strong>on</strong>necti<strong>on</strong> between thework <strong>of</strong> K<strong>on</strong>tsevich and Mirzakhani. We believe that this pro<strong>of</strong> <strong>of</strong> K<strong>on</strong>tsevich’s combinatorialformula is rather intuitive in nature, avoids the technical difficulties which are inherent in theoriginal pro<strong>of</strong>, and may lead to further insights. Indeed, as a part <strong>of</strong> joint work with Safnuk [10],we have extended these ideas to produce a recursive formula à la Mirzakhani which computesthe asymptotics <strong>of</strong> V g,n (L). The differential versi<strong>on</strong> <strong>of</strong> this formula is the Virasoro c<strong>on</strong>straintc<strong>on</strong>diti<strong>on</strong>, thereby providing a new path to Witten’s c<strong>on</strong>jecture. We also believe that it will notbe difficult to extend these ideas to integrati<strong>on</strong> over the combinatorially defined Witten cycles.
Chapter 1A gentle introducti<strong>on</strong> to <strong>moduli</strong><strong>spaces</strong> <strong>of</strong> <strong>curves</strong>In this chapter, we introduce the main characters <strong>of</strong> our story, <strong>moduli</strong> <strong>spaces</strong> <strong>of</strong> <strong>curves</strong>. The aimis to provide a c<strong>on</strong>cise expositi<strong>on</strong> <strong>of</strong> the important results and ideas which form the backgroundto this thesis. Newcomers to the area will hopefully find this chapter a suitable point <strong>of</strong> entry tothe now vast body <strong>of</strong> knowledge c<strong>on</strong>cerning <strong>moduli</strong> <strong>spaces</strong> <strong>of</strong> <strong>curves</strong>. However, the selecti<strong>on</strong><strong>of</strong> material presented here is necessarily <strong>on</strong>ly a small subset, chosen to suit our specific needsand reflect our particular point <strong>of</strong> view. For example, a great deal <strong>of</strong> attenti<strong>on</strong> has been paidto intersecti<strong>on</strong> <str<strong>on</strong>g>theory</str<strong>on</strong>g> <strong>on</strong> <strong>moduli</strong> <strong>spaces</strong> <strong>of</strong> <strong>curves</strong>, to the interacti<strong>on</strong> between algebraic andhyperbolic geometry, and to the recent results <strong>of</strong> Mirzakhani. Throughout the chapter, detailsand pro<strong>of</strong>s have <strong>of</strong>ten been omitted for the sake <strong>of</strong> clarity and space. For those interested infurther informati<strong>on</strong>, there are numerous references to the relevant sources in the literature. 11.1 Moduli <strong>spaces</strong> <strong>of</strong> <strong>curves</strong>First principlesInformally, the points <strong>of</strong> a <strong>moduli</strong> space classify objects <strong>of</strong> a certain type, while its geometryreflects the way in which these objects can vary in families. For example, c<strong>on</strong>sider the <strong>moduli</strong>spaceM g = {C | C is a smooth algebraic curve <strong>of</strong> genus g} / ∼1 In particular, we start by menti<strong>on</strong>ing the articles [43, 55, 56] which have influenced our expositi<strong>on</strong> and which aresuitable for those wishing to discover this remarkable area <strong>of</strong> mathematics for the first time.9