Terminal singularities of the moduli space of curves on low degree hypersurfaces and the circle method

Jakob Glas, Matthew Hase-Liu · 2024

Read Terminal singularities of the moduli space of curves on low degree hypersurfaces and the circle method by Jakob Glas, Matthew Hase-Liu (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-14923.txt.

Introduction

Jakob Glas and Matthew Hase-Liu study rough spots in a space of curves in a 2024 mathematics paper. The space collects all maps of a set degree from smooth curves onto a smooth surface of low degree. Geometers need to know how bad the space's own kinks can get.

For map degree large next to the curve's genus, the authors prove the kinks are at worst terminal. That class of kink is mild enough for many uses in geometry. Their path runs through jet schemes, which record how curves sit to high order. For those, they build a form of the circle method, a classic counting tool from number theory. The work ties two far-apart fields in a new way.

Catalog record

CREATORJakob Glas, Matthew Hase-Liu
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE116 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-14923

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Terminal singularities of the moduli space of curves on low degree hypersurfaces and the circle method Jakob Glas, Matthew Hase-Liu Abstract We study the singularities of the moduli space of degree e e maps from smooth genus g g curves to an arbitrary smooth hypersurface of low degree. For e e large compared to g g , we show that these moduli spaces have at worst terminal singularities. Our main approach is to study the jet schemes of these moduli spaces by developing a suitable form of the circle method. 1 Introduction The interplay between geometry and number theory has had a long and rich history. Examples like Deligne’s resolution of the Weil conjectures [ 8 , 9 ] , which serves as a powerful tool for estimating exponential sums and makes crucial use of the heavy machinery of algebraic geometry, shows the impact of geometry on number theory should not be underestimated. Occasionally, this flow of information can be reversed and tools from analytic number theory can be used to investigate objects of geometric interest. For instance, building on ideas of Ellenberg and Venkatesh, Browning and Vishe [ 6 ] used the circle method over function fields to deduce crude geometric properties of the moduli space \mathcal{M}_{0,0}(X,e) of rational curves of degree e on a smooth hypersurface X of low degree. Our aim in this paper is to enrich this flow of information by developing a suitable form of the circle method to show the moduli space \mathcal{M}_{g,0}(X,e) of smooth projective genus g curves of degree e on X\subset\mathbb{P}^{n}_{\mathbb{C}} has at worst terminal singularities for n large compared to d and e large compared to d and g . Let \bar{\mu}=\underset{\dim H^{0}(\deg e\text{ line bundle on genus }g\text{ curve})^{n+1}}{\underbrace{\left(n+1\right)\left(e-g+1\right)}}-\underset{\dim H^{0}(\deg de\text{ line bundle})}{\underbrace{\left(de-g+1\right)}}+\underset{\dim\Pic^{e}\left(\text{genus }g\text{ curve}\right)}{\underbrace{g}}-\underset{\text{scaling}}{\underbrace{1}}+\underset{\dim\mathcal{M}_{g}}{\underbrace{\left(3g-3\right)}}. (1.1) A naive heuristic based on Riemann–Roch and deformation theory suggests that \dim\mathcal{M}_{g,0}(X,e)=\bar{\mu} . For g=0 , n\geq d+2 , and X general in moduli, Riedl and Yang [ 25 ] have shown this is indeed the case, in addition to establishing irreducibility of \mathcal{M}_{0,0}(X,e) ; in fact, they even demonstrate these results for the Kontsevich compactification \overline{\mathcal{M}}_{0,0}(X,e) . Methods based on analytic number theory from Browning and Vishe [ 6 ] allow one to deal with any smooth hypersurface at the cost of requiring n\geq(5d-4)2^{d-1} , which was later refined to n\geq(2d-1)2^{d-1} by Browning and Sawin [ 5 ] . Bilu, Browning, and Sawin [ 4 , 2 ] have also developed cohomological and motivic analogs of this strategy. For the higher genus setting, the second author [ 14 ] extended the above results on irreducibility and expected dimension to \mathcal{M}_{g,0}(X,e) by re-interpreting the analytic strategy geometrically. Recent work of Sawin [ 28 ] also addresses the case of X a Fermat hypersurface, in which case the lower bound on n can be improved to depend only linearly on d . Perhaps the next natural geometric question is to ask how singular \mathcal{M}_{g,0}(X,e) is. For d=1,2 , smoothness of \mathcal{M}_{0,0}(X,e) follows from work of Kontsevich [ 18 ] , where he works in the more general context of convex varieties X . For higher genus, the second author and Amal Mattoo [ 15 ] proved that \mathcal{M}_{g,0}(X,e) is essentially never smooth for any smooth Fano hypersurface X of degree at least 2. For g=0 , Starr [ 30 ] investigated the singularities and Kodaira dimension of \overline{\mathcal{M}}_{0,0}(X,e) , proving that for e\geq 2 and d\geq 3 , the spaces \overline{\mathcal{M}}_{0,0}(X,e) , as well as their coarse spaces (by inversion–of–adjunction) have at worst canonical singularities if n\geq d+e and X is general . He also computed the canonical divisor of \overline{\mathcal{M}}_{0,0}(X,e) , proving that it is big in many cases. Essentially nothing else seems to be known about the quality of the singularities of \mathcal{M}_{0,0}(X,e) , certainly much less those of \mathcal{M}_{g,0}(X,e) when g\geq 1 . Our main result is that for n large compared to d , the singularities of the moduli spaces \mathcal{M}_{g,0}(X,e) are as mild as possible, at least from the perspective of the minimal model program. For an integer g\geq 0 , define f(g)=\begin{cases}0&\text{if }g\in\{0,1\},\\ (g+1)/2&\text{if }g\geq 2.\end{cases} Theorem 1 . Let X\subset\mathbb{P}^{n}_{\mathbb{C}} be a smooth hypersurface of degree d\geq 2 and g\geq 0 be an integer. Then \mathcal{M}_{g,0}(X,e) has at worst terminal singularities when n+1>\begin{cases}2^{d-2}d(2d+1)&\text{if }g=0,e=1\text{, and }d\geq 4,\\ 2^{d-1}(d-1)\left(1+\frac{d-1}{2d-1}+\frac{2(d-1)(de+1)}{d}\right)&\text{if }g=0,2\leq e\leq d-1\text{, and }d\geq 3,\\ 2^{d-2}(d-1)(4d^{2}-4d+3)-1&\text{if }g=0,d\leq e\text{, and }d\geq 3,\\ 2^{d-2}(d-1)(4d^{2}-4d+3)&\text{if }g\geq 1\text{ and }d\geq 2,\end{cases} provided that e>e_{0} , whose value is given in Table 1 . Table 1: The value of e_{0} (for terminal singularities) g d e_{0} =0 \geq 3 =0 \geq 1 \geq 3 =2^{d-2}\left(d-1\right)^{2}\left(4d^{2}g+\left(4d^{3}-8d^{2}+7d-3\right)f(g)+4d(g-2)-g+4\right) \qquad-1+(d+1)g+(d-1)^{2}f(g) \geq 1 =2 =29g+14f(g)-5 By precisely the same methods, we have the following similar result for canonical singularities, where we can slightly improve the range of admissible number of variables. Theorem 2 . Let X\subset\mathbb{P}^{n}_{\mathbb{C}} be a smooth hypersurface of degree d\geq 2 and g\geq 0 be an integer. Then \mathcal{M}_{g,0}(X,e) has at worst canonical singularities when n+1>\begin{cases}2^{d-2}d(2d+1)&\text{if }g=0,e=1\text{, and }d\geq 4,\\ 2^{d-1}(d-1)(de+2)&\text{if }g=0,2\leq e\leq d-2\text{, and }d\geq 3,\\ 2^{d-1}(d-1)(d^{2}-d+1)-1&\text{if }g=0,d-1\leq e\text{, and }d\geq 3,\\ 2^{d-1}(d-1)(d^{2}-d+1)&\text{if }g\geq 1\text{ and }d\geq 2,\end{cases} provided e>e_{0} , whose value is given in Table 2 . Table 2: The value of e_{0} (for canonical singularities) g d e_{0} =0 \geq 3 =0 \geq 1 \geq 3 =(g-1)(2^{d-1}(d-1)^{2}(2d-1)+2) \qquad+(d-1)(g+(d-1)f(g))(2^{d-1}(d-1)(d^{2}-d+1)+1) \geq 1 =2 =19g+7f(g)-3 For the particular case e=1 , the space F_{1}(X)=\mathcal{M}_{0,0}({X},{1}) is the Fano variety of lines of X . It is a classical result due to Altman and Kleiman [ 1 ] that the Fano variety of lines of any smooth cubic hypersurface X\subset\mathbb{P}^{n} is smooth for n\geq 4 . If X\subset\mathbb{P}^{n} is a hypersurface of degree d\geq 4 , then F_{1}(X) is still known to be smooth, if one assumes that X is general and d\leq 2n-3 [ 16 , Theorem V.4.3] . In particular, Theorem 1 gives new information about F_{1}(X) for smooth X as soon as d\geq 4 and n+1>2^{d-2}d(2d+1) . Under these circumstances, Theorem 1 even gives new insight into the Hodge numbers of F_{1}(X) . Recall that for a singular projective variety Y over \mathbb{C} , the cohomology groups H^{p+q}(X,\mathbb{Q}) admit a mixed Hodge structure with an increasing Hodge filtration F_{\bullet} . Then, the Hodge numbers h^{p,q}(Y) are defined as the \mathbb{C} -dimensions of the graded pieces \Gr^{F}_{-p}H^{p+q}(X,\mathbb{C}) . Combining our main results with those of Park and Popa [ 24 ] on the relationship between symmetries of the Hodge diamond of Y and higher rational singularities of Y and those of Bilu and Browning [ 2 ] on the Hodge–Deligne polynomial of the Fano variety of lines gives the following corollary. Corollary 3 . Let X\subset\mathbb{P}^{n}_{\mathbb{C}} be a smooth hypersurface of degree d\geq 4 with n+1>2^{d-2}d(2d+1) . Denote by N=2n-d-3 the dimension of F_{1}(X) and h^{p,q}=h^{p,q}(F_{1}(X)) for 0\leq p,q\leq N . Then, h^{0,q}=h^{q,0}=h^{N,N-q}=h^{N-q,N} for 0\leq q\leq N , which is moreover 0 for q>4n+2d-6-2^{2-d}(n+1) . Also, h^{1,1}=h^{N-1,N-1}=1 . Proof. By Theorem 1 , F_{1}(X) is a normal complete intersection with at worst terminal singularities and hence at worst rational singularities. The claim h^{0,q}=h^{q,0}=h^{N,N-q}=h^{N-q,N} for 0\leq q\leq N then follows immediately from Theorem C of [ 24 ] . Corollary 1.6 of [ 2 ]

Read the complete text