ScalingStacks

1.2 Toric geometry [04MB]

Original official author HTML, exact retained edition. Historical TeX conversion verdicts remain unchanged. Cited-edition alignment and mathematical self-containment are not assessed.

Complete original source context Β· Original author HTML

1.2 Toric geometry

Throughout this section, let ZZ be an rr-dimensional (normal) proper toric variety over kk, in the sense of [KKMSD73]. This means that ZZ is a normal kk-variety, containing the torus 𝕋=𝔾m,kr\mathbb{T}=\mathbb{G}^{r}_{m,k} as an open subset, and such that the torus action onto itself extends to an action on ZZ. The complement Ξ”Z=Zβˆ–π•‹\Delta_{Z}=Z\setminus\mathbb{T} is a reduced anticanonical Weil divisor in ZZ, called the toric boundary of ZZ; we write it as the sum of its irreducible components Ξ”Z=βˆ‘l∈LZl\Delta_{Z}=\sum_{l\in L}Z_{l}. We write N=Hom⁑(𝔾m,k,𝕋)N=\Hom(\mathbb{G}_{m,k},\mathbb{T}) for the free abelian group of 1-parameter subgroups of 𝕋\mathbb{T}, and Nℝ=NβŠ—β„N_{\mathbb{R}}=N\otimes\mathbb{R}.

The variety ZZ can be described by a combinatorial object, called its fan Ξ£\Sigma. The fan lives inside the finite-dimensional vector space NℝN_{\mathbb{R}}; Ξ£\Sigma is a collection of strictly convex rational polyhedral cones Ξ£={Οƒ}ΟƒβˆˆΞ£\Sigma=\{\sigma\}_{\sigma\in\Sigma} inside NℝN_{\mathbb{R}}, stable under intersection and such that each face of a cone in Ξ£\Sigma is itself in Ξ£\Sigma. The cones of Ξ£\Sigma are in inclusion-reversing bijection with the strata of Ξ”Z\Delta_{Z}; this induces in particular a bijection between the set of rays (i.e. 1-dimensional cones) of Ξ£\Sigma and the irreducible components of Ξ”Z\Delta_{Z}.
The fan Ξ£\Sigma encodes various types of algebro-geometric information about ZZ. For instance, the variety ZZ is smooth if and only if each top-dimensional cone of Ξ£\Sigma is GL⁑(N)\GL(N)-isomorphic to the standard octant ℝ⩾0rβŠ‚β„r\mathbb{R}^{r}_{\geqslant 0}\subset\mathbb{R}^{r}.

Furthermore, in the case where ZZ is smooth (which implies it has simple normal crossing boundary), the fan allows us to give a rather explicit description of the Picard group of ZZ. Indeed, each Cartier divisor can be moved via the torus action to a 𝕋\mathbb{T}-invariant Cartier divisor, which is thus supported on the boundary. This provides a canonical set of generators of Pic⁑(Z)\Pic(Z), and the kernel can be described as follows.
Write Div𝕋(Z)=βŠ•l∈Lβ„€Zl≃℀L\Div^{\mathbb{T}}(Z)=\oplus_{l\in L}\mathbb{Z}Z_{l}\simeq\mathbb{Z}^{L} the abelian group of Weil divisors supported on the boundary. The canonical map q:Div𝕋⁑(Z)⟢Pic⁑(Z)q:\Div^{\mathbb{T}}(Z)\longrightarrow\Pic(Z) sends a divisor to its class; the map p:M⟢Div𝕋⁑(Z)p:M\longrightarrow\Div^{\mathbb{T}}(Z) sends a monomial zmz^{m} to the principal divisor div​(zm)\textrm{div}(z^{m}).

Lemma 1.2.1 ([Ful93, 3.4]).

The following sequence

0⟢M→𝑝Div𝕋⁑(Z)β†’π‘žPic⁑(Z)⟢00\longrightarrow M\xrightarrow{p}\Div^{\mathbb{T}}(Z)\xrightarrow{q}\Pic(Z)\longrightarrow 0

is exact.

Let us rephrase this in term of coordinates, after fixing an isomorphism N≃℀rN\simeq\mathbb{Z}^{r} and denoting L={u1,…,us}L=\{u_{1},\ldots,u_{s}\} the primitive generators of the 1-dimensional cones of Ξ£\Sigma. Since by [Ful93, Lemma p.61], we have ordZl⁑(zm)=⟨ul,m⟩\ord_{Z_{l}}(z^{m})=\langle u_{l},m\rangle, we obtain div​(zm)=βˆ‘l∈L⟨ul,mβŸ©β€‹Zl\textrm{div}(z^{m})=\sum_{l\in L}\langle u_{l},m\rangle Z_{l} and hence p⁑(m)=(⟨ul,m⟩)l∈Lp(m)=(\langle u_{l},m\rangle)_{l\in L}. We deduce the following explicit description of Pic⁑(Z)\Pic(Z):

Corollary 1.2.2.

Let ul=(ul,1,…,ul,r)u_{l}=(u_{l,1},\ldots,u_{l,r}) for l∈{1,…,s}l\in\{1,\ldots,s\}. Then Pic⁑(Z)\Pic(Z) is generated by the line bundles π’ͺZ​(Zl)\mathcal{O}_{Z}(Z_{l}), with the rr relations:

π’ͺZ​(βˆ‘l=1mul,1​Zl)=…=π’ͺZ​(βˆ‘l=1mul,r​Zl)=0.\mathcal{O}_{Z}(\sum_{l=1}^{m}u_{l,1}Z_{l})=\ldots=\mathcal{O}_{Z}(\sum_{l=1}^{m}u_{l,r}Z_{l})=0.

In particular, the divisors in the rr-tuple

Ξ”=βˆ‘l∈LulβŠ—Zl∈NβŠ—Div𝕋⁑(Z)≃(Div𝕋⁑(Z))r\Delta=\sum_{l\in L}u_{l}\otimes Z_{l}\,\in\,N\otimes\Div^{\mathbb{T}}(Z)\simeq(\Div^{\mathbb{T}}(Z))^{r}

are principal.

We now want to describe how the fan Ξ£\Sigma encodes the intersection theory on ZZ. Each 11-cycle in ZZ being numerically equivalent to a sum of toric strata, it is enough to study the intersection numbers (Cβ‹…Zl)(C\cdot Z_{l}), where ZlZ_{l} is a boundary component of ZZ and CC is a 1-dimensional toric stratum, which is isomorphic to β„™1\mathbb{P}^{1} by properness. The stratum CC is thus a rational curve with two marked points pp and qq, which are the intersection points of CC with two components of Ξ”Z\Delta_{Z}, denoted here by ZpZ_{p} and ZqZ_{q}, with corresponding rays ρp\rho_{p} and ρq\rho_{q}. The curve CC corresponds to a (rβˆ’1)(r-1)-dimensional cone ΟƒC\sigma_{C} of Ξ£\Sigma, while the points pp and qq correspond to the maximal cones generated by <ΟƒC,ρp><\sigma_{C},\rho_{p}> and <ΟƒC,ρq><\sigma_{C},\rho_{q}>.

Lemma 1.2.3 ([Ful93, p. 99]).

The primitive generators of the rays of the fan satisfy the following relation:

up+uq=βˆ’βˆ‘ulβˆˆΟƒC(Cβ‹…Zl)ul.u_{p}+u_{q}=-\sum_{u_{l}\in\sigma_{C}}(C\cdot Z_{l})u_{l}.

Observing that we have (Cβ‹…Zp)=(Cβ‹…Zq)=1(C\cdot Z_{p})=(C\cdot Z_{q})=1, and (Cβ‹…Zl)=0(C\cdot Z_{l})=0 for any other ll, this may be rewritten in a more synthetic way:

(1.2.4) βˆ‘l∈L(Cβ‹…Zl)​ul=0.\sum_{l\in L}(C\cdot Z_{l})u_{l}=0.

Note that this lemma holds even if ZZ is not proper.
Finally, we have the following vanishing theorem for nef divisors on a proper toric variety.

Proposition 1.2.5.

Let DD be a nef Cartier divisor on a proper toric variety ZZ. Then Hi​(Z,π’ͺZ​(D))=0H^{i}(Z,\mathcal{O}_{Z}(D))=0 for i>0i>0.

Proof.

The divisor DD being nef is equivalent to it being globally generated, by [Mus02, Theorem 3.1]. Thus, the result follows directly from [Ful93, p. 74]. ∎

Following [NXY19], we will say that an RR-scheme of finite type 𝒡\mathscr{Z} is toric if there exists a toric kk-scheme of finite type 𝒡\mathcal{Z}, together with a toric morphism t:π’΅βŸΆπ”Έk1t:\mathcal{Z}\longrightarrow\mathbb{A}^{1}_{k}, such that 𝒡≃𝒡×𝔸1R\mathscr{Z}\simeq\mathcal{Z}\times_{\mathbb{A}^{1}}R. Writing N^\widehat{N} for the lattice of 1-parameter subgroups of the torus of 𝒡\mathcal{Z}, such a scheme is described by a fan Ξ£^\widehat{\Sigma} in N^ℝ\widehat{N}_{\mathbb{R}}, together with a linear map ord⁑(t):|Ξ£^|βŸΆβ„β‰₯0\ord(t):\lvert\widehat{\Sigma}\rvert\longrightarrow\mathbb{R}_{\geq 0}, defined by ord⁑(t)​(n)=ord0⁑(t∘n)\ord(t)(n)=\ord_{0}(t\circ n) for a 1-parameter subgroup n:𝔾m→𝒡n:\mathbb{G}_{m}\rightarrow\mathcal{Z}. Note that the map ord⁑(t)\ord(t) recovers the function tt uniquely, since it is a monomial.

Definition 1.2.6.

Let 𝒳\mathscr{X} be a normal RR-scheme of finite type, and YY be a stratum of 𝒳k\mathscr{X}_{k}. We say that 𝒳\mathscr{X} is toric along Y if there exists a toric RR-scheme 𝒡\mathscr{Z}, a stratum WW of 𝒡k\mathscr{Z}_{k} and a formal isomorphism over RR

𝒳/Y^≃𝒡/W^.\widehat{\mathscr{X}_{/Y}}\simeq\widehat{\mathscr{Z}_{/W}}.

Original mathematics by the credited authors. Source-backed reader collection; mathematical self-containment is not assessed.