Geometric interpretation of toroidal compactifications of moduli of points in the line and cubic surfaces (Q2656144)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Geometric interpretation of toroidal compactifications of moduli of points in the line and cubic surfaces |
scientific article |
Statements
Geometric interpretation of toroidal compactifications of moduli of points in the line and cubic surfaces (English)
0 references
10 March 2021
0 references
The article under review establishes that the toroidal compactifications of the moduli spaces of finitely many points in \( {\mathbb P} ^1 ({\mathbb C} ) \) or cubic surfaces are isomorphic to some geometric compactifications of these moduli spaces. Let us suppose that the rational weights \( 1 > w_1 \geq w_2 \geq \ldots \geq w_n > 0, \) \( n \in {\mathbb N} \) with \( w_1 + w_2 + \ldots + w_n = 2\) satisfy the condition \( ( 1 - w_i - w_j) ^{-1} \in {\mathbb Z} \) for \( i \neq j, \) \( w_i \neq w_j, \) \( w_i + w_j < 1 \) or \( 2 (1 - 2 w_i) ^{-1} \in {\mathbb Z} \) for \( i \neq j, \) \( w_i = w_j, \) \( w_i + w_j < 1. \) In 1986 Deligne and Mostow showed that for any such \( w = ( w_1, \ldots , w_n) \) there is an arithmetic lattice \( \Gamma _w, \) acting on the complex ball \( {\mathbb B}_{n-3} \) of dimension \( n-3 \) and a natural number \( m < n, \) so that the GIT compactification \( ( {\mathbb P} ^1 ( {\mathbb C}) ) ^n // _w \mathrm{SL}_2 ( {\mathbb C}) \times S_m \) of the moduli space of \( n \) points in \( {\mathbb P}^1 ( {\mathbb C}) \) is isomorphic to the Baily-Borel compactification \( \overline{ \Gamma _w \setminus {\mathbb B} _{n-3}} ^{\mathrm{bb}} \) of \( \Gamma _w \setminus {\mathbb B} _{n-3}. \) By a result of Hassett from 2003, the moduli space \( M _{0,n} \) of \( n \) distinct points in \( {\mathbb P}^1 ( {\mathbb C}) \) is a Zariski open subset of the smooth projective moduli space \( \overline{M}_{0,w} \) of the stable rational curves of an appropriate Deligne-Mostow weight \( w. \) For a sufficiently small \( \varepsilon \in {\mathbb Q} \cap (0,1), \) let \( w + \varepsilon = ( w_1 + \varepsilon, \ldots , w_n + \varepsilon). \) The article under review shows that for any Deligne-Mostow weight \( w \) the quotient \( \overline{M} _{0, w+ \varepsilon} / S_m \) of Hassett's moduli space \( \overline{M} _{0, w + \varepsilon} \) by the symmetric group \( S_m \) is isomorphic to the toroidal compactification \( \overline{\Gamma _w \setminus {\mathbb B} _{n-3}} ^{\mathrm{tor}} \) of \( \Gamma _w \setminus {\mathbb B}_{n-3}. \) First, it is proved for the so called ancestral cases of \( 8 \) points in \( {\mathbb P} ^1 ( {\mathbb C}), \) associated with an arithmetic lattice \( \Gamma _{\left( \frac{1}{4} \right) ^8} \) over the Gaussian integers \( {\mathbb Z} [ \sqrt{-1}] \) and of \( 12 \) points in \( {\mathbb P} ^1( {\mathbb C}), \) corresponding to an arithmetic lattice \( \Gamma _{\left( \frac{1}{6} \right) ^{12}} \) over the Eisenstein integers \( {\mathbb Z} \left[ e ^{\frac{2 \pi i}{3}} \right]. \) Briefly, Deligne-Mostow's isomorphisms with Baily-Borel compactifications provide birational morphisms, which are shown to be isomorphisms on complements of closed subsets of codimension \( \geq 2. \) A detailed account of the geometry of Hassett's moduli spaces and the toroidal compactifications allows to derive that these isomorphisms are global. For any non ancestral Deligne-Mostow's weight \( w, \) the associated configuration of points is obtained from an ancestral one by some collisions. That allows to embed the considered compactifications of \( \Gamma _w \setminus {\mathbb B} _{h-3} \) in the ones of an ancestral case and to establish the result for all Deligne-Mostow's ball quotients. The isomorphisms \( \overline{\Phi} _w : \overline{M} _{0, w + \varepsilon} / S_m \rightarrow \overline{\Gamma _w \setminus {\mathbb B} _{n-3}} ^{\mathrm{tor}} \) can be interpreted as extensions of period maps \( \Phi _w : M _{0,n} / S_m \rightarrow \Gamma _w \setminus {\mathbb B} _{n-3} \) of weight \( 1 \) Hodge structures. This is done explicitly for the ancestral case \( w = \left( \frac{1}{4} \right) ^8, \) in which \( m = 1. \) For any \(\mathrm{PGL} _2 ( {\mathbb C}) \)-orbit \( \underline{x} \) of a point \( x = ( x_1, \ldots , x_8) \in ( {\mathbb P} ^1({\mathbb C}) ) ^8 \setminus \cup \{ x_i = x_j \}, \) let \( C_{\underline{x}} \) be the Zariski closure of the curve with equation \( Y^4 = \prod _{j=1} ^8 (X - x_j) \) in the weighted projective space \( {\mathbb W} {\mathbb P} [ 1: 1:2]. \) Then \( C_{\underline{x}} \) is of genus \( 9. \) The multiplication \( Y \mapsto \sqrt{-1} Y \) by the imaginary unit \( \sqrt{-1} \in {\mathbb C}^* \) induces a \( {\mathbb C} \)-linear map \( \rho : H ^1 ( C_{\underline{x}}, {\mathbb C}) \rightarrow H ^1 ( C_{\underline{x}}, {\mathbb C}). \) The period map \( \Phi _{\left( \frac{1}{4} \right) ^8 } \) associates to \( \underline{x} \in M _{0,8} \) the weight \( 1 \) Hodge structure on the \( (-1) \)-eigenspace of the involution \( \rho ^2 : H ^1 ( C_{\underline{x}}, {\mathbb C}) \rightarrow H^1 ( C_{\underline{x}}, {\mathbb C}). \) Let \( \mathcal{S} \) be a connected component of the boundary of \( \overline{M} _{0, \left( \frac{1}{4} \right) ^8 + \varepsilon}, \) over which the curve \( C_{\underline{x}} \) degenerates to a pair of genus \( 3 \) curves. The article shows that the restriction of \( \overline{\Phi} _{\left( \frac{1}{4} \right) ^8 } \) to \( \mathcal{S} \) computes the corresponding limiting mixed Hodge structure. As an application of the isomorphism \( \overline{\Phi} _{\left( \frac{1}{4} \right) ^8} : \overline{M} _{0, \left( \frac{1}{4} \right) ^8 + \varepsilon} \rightarrow \overline{ \Gamma _{\left( \frac{1}{4} \right) ^8 } \setminus {\mathbb B}_5} ^{\mathrm{tor}}, \) the authors observe that Kollár-Shepherd-Barron-Alexeev's compactification \( \overline{K} \) of the moduli space of \( K3 \) surfaces with a purely non-symplectic automorphism of order \( 4 \) and \( U(2) \oplus D_4 ^{\oplus 2} \) lattice polarization is isomorphic to the quotient of the toroidal compactification \( \overline{\Gamma _{\left( \frac{1}{4} \right)^8 } \setminus {\mathbb B}_5 } ^{\mathrm{tor}} \) by \( (S_4 \times S_4) \rtimes S_2. \) An arbitrary smooth cubic surface \( S \subset {\mathbb P} ^3 ( {\mathbb C}) \) is a blow up of \( {\mathbb P}^2 ( {\mathbb C}) \) at \( 6 \) points in a general linear position, which do not belong to a conic. Any labelling of these \( 6 \) points induces a labelling of the \( 27 \) lines on \( S \) and is called a marking of \( S. \) Let \( Y \) be the moduli space of the marked smooth cubic surfaces. In 1982 Naruki constructs a smooth compactification \( \overline{N} \) of \( Y \) by a divisor with normal crossings. In 2005 Dolgachev, van Geemen and Kondo show that the field \( {\mathbb C} (Y) \) of the rational functions of \( Y \) is an extension of the field of the rational functions of \( {\mathbb P} ( H^0 ( {\mathbb P} ^3, \mathcal{O} (3))) /\!/ _{\mathcal{O}(1)}\mathrm{SL} _4({\mathbb C}), \) whose Galois group is the Weyl group \( W(E_6). \) The normalization \( \overline{Y} _{\mathrm{GIT}} \) of \( {\mathbb P} ( H^0 ( {\mathbb P} ^3, \mathcal{O}(3))) /\!/ _{\mathcal{O}(1)}\mathrm{SL}_4({\mathbb C}) \) in \( {\mathbb C} (Y) \) admits a morphism \( \overline{N} \rightarrow \overline{Y} _{\mathrm{GIT}}, \) which blows down \( 40 \) of the \( 76 \) components of the boundary of \( \overline{N}. \) An article of Allcock, Carlson and Toledo from 2002 establishes that \( \overline{Y} _{\mathrm{GIT}}\) is isomorphic to Baily-Borel compactification \( \overline{\Gamma _c \setminus {\mathbb B} _4} ^{\mathrm{bb}} \) of an appropriate quotient \( \Gamma _c \setminus {\mathbb B} _4 \) of the complex \( 4 \)-ball. They conjecture that Naruki's compactification \( \overline{N} \) of \( Y \) is isomorphic to the toroidal compactification \( \overline{\Gamma _c \setminus {\mathbb B}_4} ^{\mathrm{tor}}. \) The article under review proves the aforementioned conjecture. For any smooth cubic surface \( S \subset {\mathbb P} ^3 ( {\mathbb C}), \) for the divisor \( B \) of the \( 27 \) lines on \( S \) and for any \( c \in {\mathbb Q} \cap \left( \frac{1}{9}, 1 \right), \) the pair \( (S, cB) \) is stable, i.e., log canonical with ample \( K_S + cB. \) In 2009 Hacking, Keel and Tevelev have studied the Kollár-Shepherd-Barron-Alexeev's compactification of \( Y \) by stable pairs and showed that Naruki's compactification \( \overline{N} \) is isomorphic to the log canonical model of \( Y. \) They conjectured that \( \overline{N} \) parameterizes the stable pairs \( \left( S, \left( \frac{1}{9} + \varepsilon \right) B \right) \) and their degenerations. The article under review proves this conjecture and describes exlicitly the stable pairs \( \left( S_o, \left( \frac{1}{9} + \varepsilon \right) B_o \right), \) parameterized by the boundary of \( \overline{N}. \) It turns out that either \( S_o \) has \( 1 \leq k \leq 4 \) singularities of type \( A_1 \) or \( S_o \) is a normal crossing union of three planes in \( {\mathbb P} ^3 ( {\mathbb C}). \) In each case, the authors compute the multiplicities of the components of \( B_o \) and describe their mutual position with respect to the singularities of \( S_o. \)
0 references
moduli space
0 references
compactification
0 references
pointed line
0 references
cubic surface
0 references
Hodge theory
0 references
stable pairpair
0 references
0 references
0 references
0 references
0 references
0 references