--- title: 'Mirzakhani Volume: Weil–Petersson Moduli Volumes' url: https://www.emergentmind.com/topics/mirzakhani-volume type: topic --- # Mirzakhani Volume: Weil–Petersson Moduli Volumes Mirzakhani volume usually denotes the Weil–Petersson volume of a moduli space of hyperbolic surfaces with prescribed boundary data, together with the associated volume polynomials introduced and analyzed by Mirzakhani. For integers \(g\ge 0\), \(n\ge 1\) with \(2g-2+n>0\), the basic object is the moduli orbifold \(\mathcal M_{g,n}(L_1,\dots,L_n)\) of genus-\(g\) oriented hyperbolic surfaces with \(n\) labeled geodesic boundary components of lengths \(L_1,\dots,L_n\), equipped with the Weil–Petersson symplectic form \(\omega_{\mathrm{WP}}\). Its volume is \[ V_{g,n}(L_1,\dots,L_n) = \int_{\mathcal M_{g,n}(L_1,\dots,L_n)} \frac{\omega_{\mathrm{WP}}^{\,3g-3+n}}{(3g-3+n)!}, \] and Mirzakhani proved that this is an even symmetric polynomial in the squares \(L_1^2,\dots,L_n^2\) of total degree \(6g-6+2n\) [1108.0174][1103.4674]. In later literature, the same term also appears in extensions to cone points, analytic continuations \(L_j\mapsto i\theta_j\), and flat-geometric limits relating Weil–Petersson and Masur–Veech volumes [2212.13701][2405.10869]. ## 1. Definition and symplectic-geometric setting The moduli space \(\mathcal M_{g,n}(L_1,\dots,L_n)\) is obtained from Teichmüller space by quotienting by the mapping-class group, and in Fenchel–Nielsen coordinates \((\ell_i,\tau_i)_{i=1}^{3g-3+n}\) the Weil–Petersson form is \[ \omega_{\rm WP}=\sum_{i=1}^{3g-3+n} d\ell_i\wedge d\tau_i. \] This gives a symplectic form on the moduli orbifold and yields the volume integral above [1103.4674][1108.0174]. Mirzakhani’s framework includes both bordered surfaces and the cusp case \(L_i=0\). The notation \[ V_{g,n}:=V_{g,n}(0,\dots,0) \] is standard for the cusp-volume, and several later asymptotic results are normalized by this quantity [2011.14889][2501.06421]. The low-dimensional initial cases are \[ V_{0,3}(L_1,L_2,L_3)=1,\qquad V_{1,1}(L)=\frac{\pi^2}{12}+\frac{L^2}{48}, \] with the equivalent form \(V_{1,1}(L)=\frac{1}{48}(L^2+4\pi^2)\) also appearing in survey treatments [1509.06880][1103.4674]. A recurrent point in the literature is that the “volume” is not only a scalar invariant of \((g,n)\), but a full polynomial function of boundary data. This distinction matters because Mirzakhani’s recursion, the intersection-theoretic interpretation, and large-genus asymptotics all act on the polynomial \(V_{g,n}(L_1,\dots,L_n)\), not merely on the specialization \(V_{g,n}\) [2011.14889][2008.04458]. ## 2. Polynomial structure and intersection theory Mirzakhani showed that \(V_{g,n}(L_1,\dots,L_n)\) is a symmetric polynomial of total degree \(3g-3+n\) in the variables \(L_1^2,\dots,L_n^2\), equivalently of total degree \(6g-6+2n\) in the boundary lengths [2011.14889][1103.4674]. One expansion used in the asymptotic literature is \[ V_{g,n}(x) = \sum_{|\alpha|\le 3g-3+n} c_{g,n}(\alpha)\, \prod_{j=1}^n \frac{x_j^{2\alpha_j}}{2^{2\alpha_j}(2\alpha_j+1)!}. \] This polynomiality is one of the central structural facts about Mirzakhani volumes [2011.14889]. The coefficients encode tautological intersection numbers on \(\overline{\mathcal M}_{g,n}\). In cohomological form, one has \[ [\omega_{\rm WP}] = 2\pi^2\,\kappa_1+\sum_{k=1}^n \tfrac12 L_k^2\,\psi_k \quad\in H^2(\overline{\mathcal M}_{g,n};\mathbb R), \] and therefore \[ V_{g,n}(L_1,\dots,L_n) = \sum_{\substack{|\alpha|+m=3g-3+n}} \frac{(2\pi^2)^m}{2^{|\alpha|}\alpha!\,m!} \, \bigl\langle \psi_1^{\alpha_1}\cdots\psi_n^{\alpha_n}\kappa_1^m \bigr\rangle \, L_1^{2\alpha_1}\cdots L_n^{2\alpha_n}. \] The top-degree part \(m=0\) recovers the \(\psi\)-class intersection numbers \[ \langle\tau_{d_1}\cdots\tau_{d_n}\rangle = \int_{\overline{\mathcal M}_{g,n}} \psi_1^{d_1}\cdots\psi_n^{d_n}, \qquad \sum d_i=3g-3+n \] [1103.4674]. A closely related formulation appears in expositions of Mirzakhani’s proof of Witten’s conjecture: the coefficient of \(L_1^{2a_1}\cdots L_n^{2a_n}\) in \(V_{g,n}(L)\) is expressed as \[ \frac{1}{2^{|a|}a_1!\cdots a_n!(3g-3+n-|a|)!} \int_{\overline{\mathcal M}_{g,n}} \psi_1^{a_1}\cdots\psi_n^{a_n}\,\omega^{3g-3+n-|a|}, \] where \(\omega=c_1(\kappa_1)\) is the Weil–Petersson class [1509.06880]. This establishes the volume polynomial as a generating object for mixed \(\psi\)- and \(\kappa_1\)-intersections. ## 3. Mirzakhani’s recursion The defining dynamical feature of Mirzakhani volume is the recursive identity obtained by integrating generalized McShane identities over moduli space. In one standard form, for \(2g+n>3\), \[ \begin{aligned} 2\,\frac{\partial}{\partial L_1}\bigl(L_1V_{g,n}(L)\bigr) &= \int_0^\infty\!\!\int_0^\infty x\,y\,H(x+y,L_1)\, V_{g-1,n+1}(x,y,L_2,\dots,L_n)\,dx\,dy \\ &\quad+ \sum_{\substack{g_1+g_2=g\\ I\sqcup J=\{2,\dots,n\}}} \int_0^\infty\!\!\int_0^\infty x\,y\,H(x+y,L_1)\, V_{g_1,|I|+1}(x,L_I)\, V_{g_2,|J|+1}(y,L_J)\,dx\,dy \\ &\quad+ \sum_{k=2}^n \int_0^\infty x\Bigl[ H(x,L_1+L_k)+H(x,L_1-L_k) \Bigr]\, V_{g,n-1}(x,L_2,\dots,\widehat{L_k},\dots,L_n)\,dx, \end{aligned} \] where \[ H(x,y)=\frac{1}{1+e^{(x+y)/2}}+\frac{1}{1+e^{(x-y)/2}}. \] This is the version emphasized in survey expositions [1103.4674][1509.06880]. Geometrically, the recursion arises by fixing a distinguished boundary, decomposing orthogeodesic rays into cases, and using the resulting McShane–Mirzakhani identity. Each term corresponds to cutting off a pair of pants and either lowering genus, disconnecting the surface, or merging two boundary components [1509.06880][1108.0174]. The method depends on an unfolding argument over intermediate moduli spaces decorated by simple closed geodesics, together with the splitting of the Weil–Petersson form under cutting and gluing [1509.06880]. A later reformulation replaces the transcendental kernel by a “simple recursion” \[ \begin{aligned} &\frac{1}{4\pi i}\Bigl[(b+2\pi i)V_{g,n}(b+2\pi i,\mathbf b)-(b-2\pi i)V_{g,n}(b-2\pi i,\mathbf b)\Bigr] \\ &\qquad= \tfrac12\!\!\iint_{\substack{b',b''\ge 0\\ b'+b''\le b}} b'b''\,V_{g-1,n+1}(b',b'',\mathbf b)\,db'\,db'' +\tfrac12 \sum_{\substack{g_1+g_2=g\\ \mathbf b_1\sqcup\mathbf b_2=\mathbf b}} \iint_{\substack{b',b''\ge 0\\ b'+b''\le b}} b'b''\,V_{g_1,n_1}(b',\mathbf b_1)\,V_{g_2,n_2}(b'',\mathbf b_2)\,db'\,db'' \\ &\qquad\quad +\tfrac12\sum_{j=1}^{n-1}\left(\int_0^{b+b_j}+\int_0^{|b-b_j|}\right) b'V_{g,n-1}(b',\mathbf b\backslash b_j)\,db', \end{aligned} \] making polynomiality manifest by induction [2008.04458]. This formulation also recovers both the DVV identity from the top-degree part and a tautological push-forward formula from the lowest-degree part [2008.04458]. A common misconception is that polynomiality is merely a byproduct of explicit low-genus calculations. The recursion shows instead that it is a structural theorem: the right-hand side is built from integrals of lower-degree polynomials over polygonal regions and intervals, hence the polynomial property propagates inductively [2008.04458]. ## 4. Relations to Witten–Kontsevich, Laplace transform, and matrix models Mirzakhani’s recursion is not only a volume-computation device; it is equivalent to the Virasoro/KdV constraints for the generating function of \(\psi\)-class intersections. In one standard formulation, \[ F(t_0,t_1,t_2,\dots) = \sum_{g,n}\frac1{n!} \bigl\langle\tau_{d_1}\cdots\tau_{d_n}\bigr\rangle_g \,t_{d_1}\cdots t_{d_n}, \] and the Witten–Kontsevich theorem is \[ L_m(e^F)=0\qquad(m\ge -1). \] Survey expositions explain that extracting the top-degree terms of Mirzakhani’s recursion reproduces the string, dilaton, and higher Virasoro relations [1509.06880][1108.0174]. There is also a Laplace-transform formulation. One defines \[ W_{g,n}(z_1,\dots,z_n) = \int_0^\infty\!\cdots\!\int_0^\infty \Bigl(\prod_{i=1}^n L_i\,dL_i\Bigr) V_{g,n}(L_1,\dots,L_n)\, e^{-\sum_i L_i z_i}. \] Eynard–Orantin show that the Laplace-transformed Mirzakhani recursion becomes the standard topological recursion for the spectral curve \[ x(z)=z^2,\qquad y(z)=\frac{\sin(2\pi z)}{2\pi}, \qquad dE_u(z)=\frac12\left(\frac{1}{z-u}-\frac{1}{z+u}\right)du, \] and that this is precisely the matrix-model loop equation for the Kontsevich curve [0705.3600]. The matrix-model perspective further identifies a special choice of times in the Kontsevich integral, \[ t_{2k+3}=\frac{1}{(2k+1)!}\,2(-1)^k\,\zeta(2k+2) = \frac{B_{2k+2}}{(2k+2)!}(2\pi)^{2k+2}, \] for which the corresponding genus expansion reproduces the Laplace transforms of Weil–Petersson volumes [0705.3600]. This gives a second route from hyperbolic geometry to the Witten–Kontsevich theorem. A plausible implication is that “Mirzakhani volume” occupies an intermediate position between hyperbolic geometry and integrable hierarchies: the same object can be read as a Weil–Petersson symplectic volume, as a polynomial encoding tautological intersections, and as a solution of topological-recursion or loop-equation formalisms [0705.3600][2008.04458]. ## 5. Large-genus asymptotics For fixed \(n\), the large-\(g\) behavior of \(V_{g,n}(x_1,\dots,x_n)\) admits a full asymptotic expansion after normalization by the cusp-volume \(V_{g,n}\): \[ \frac{V_{g,n}(x)}{V_{g,n}} = \sum_{k=0}^N \frac{f_n^{(k)}(x)}{g^k} + O_{n,N}\!\left( \frac{\langle x\rangle^{\,3N+1}}{g^{N+1}e^{x_1+\cdots+x_n}} \right), \] where \(\langle x\rangle=\sqrt{1+|x|^2}\), and each \(f_n^{(k)}(x)\) is an explicitly described combination of \(\sinhc(x_i)=\sinh(x_i)/x_i\) and \(\cosh(x_i)\) [2011.14889]. The leading term is \[ f_n^{(0)}(x)=\prod_{j=1}^n \sinhc(x_j), \] equivalently \[ V_{g,n}(x)\sim V_{g,n}\,\prod_{j=1}^n\frac{\sinh(x_j)}{x_j} \qquad (g\to\infty), \] with error \(O(|x|\,g^{-1}e^{\sum x_j})\) [2011.14889]. The second coefficient \(f_n^{(1)}(x)\) is also known explicitly; in the notation \(sc(x)=\sinh(x)/x\), \(c(x)=\cosh(x)\), \[ \begin{aligned} f_n^{(1)}(x) &= \frac1{\pi^2}\sum_{i=1}^n \Bigl[c(x_i)+1-\bigl(\tfrac{x_i^2}{16}+2\bigr)\,sc(x_i)\Bigr] \prod_{k\neq i} sc(x_k) \\ &\quad -\frac1{2\pi^2}\sum_{1\le i