ScalingStacks

1.3. Toric varieties and compactification of balanced polyhedra [04RI]

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 · Original author HTML

1.3. Toric varieties and compactification of balanced polyhedra

Consider the complex algebraic torus (ℂ∗)n+1(\mathbb{C}^{*})^{n+1}, where ℂ∗=ℂ∖0\mathbb{C}^{*}=\mathbb{C}\smallsetminus 0. It is a commutative Lie group under multiplication. The 2-form

(2) 12​i​∑j=1n+1d​zz∧d​z¯z¯\frac{1}{2i}\sum\limits_{j=1}^{n+1}\frac{dz}{z}\wedge\frac{d\bar{z}}{\bar{z}}

is an invariant symplectic form on (ℂ∗)n+1(\mathbb{C}^{*})^{n+1}. There is an action of the real torus Tn+1=S1×⋯×S1T^{n+1}=S^{1}\times\dots\times S^{1} on (ℂ∗)n+1(\mathbb{C}^{*})^{n+1} by coordinatewise multiplication (we treat S1⊂ℂ∗S^{1}\subset\mathbb{C}^{*} as the unit circle). The action of Tn+1T^{n+1} is Hamiltonian and thus we have a well-defined moment map (we refer to [1] for the general definition or to a textbook, e.g. [2]) Log:(ℂ∗)n+1→ℝn+1\operatorname{Log}:(\mathbb{C}^{*})^{n+1}\to\mathbb{R}^{n+1}

(3) Log⁡(z1,…,zn+1)=(log⁡|z1|,…,log⁡|zn+1|).\operatorname{Log}(z_{1},\dots,z_{n+1})=(\log|z_{1}|,\dots,\log|z_{n+1}|).

Let Δ⊂ℝn+1\Delta\subset\mathbb{R}^{n+1} be a convex polyhedron with integer (from ℤn+1\mathbb{Z}^{n+1}) vertices. Recall (see e.g. [4]) that there is a complex toric variety ℂ​TΔ⊃(ℂ∗)n+1\mathbb{C}T_{\Delta}\supset(\mathbb{C}^{*})^{n+1}. One way to construct it is to consider the Veronese embedding (ℂ∗)n+1→ℂ​ℙ#⁡(Δ∩ℤn+1)−1(\mathbb{C}^{*})^{n+1}\to{\mathbb{C}}{\mathbb{P}}^{\#(\Delta\cap\mathbb{Z}^{n+1})-1} defined by the linear system of monomials associated to Δ∩ℤn+1\Delta\cap\mathbb{Z}^{n+1}. Here we associate to a point (p1,…,pn+1)(p_{1},\dots,p_{n+1}) a monomial zp1​…​zn+1pn+1z^{p_{1}}\dots z_{n+1}^{p_{n+1}}. We define ℂ​TΔ\mathbb{C}T_{\Delta} as the closure of the image of the Veronese embedding. Note that the standard, Fubini-Study, symplectic form on the ambient space ℂ​ℙ#⁡(Δ∩ℤn+1)−1{\mathbb{C}}{\mathbb{P}}^{\#(\Delta\cap\mathbb{Z}^{n+1})-1} defines a symplectic form on ℂ​TΔ\mathbb{C}T_{\Delta} (as long as the variety ℂ​TΔ\mathbb{C}T_{\Delta} is non-singular). In particular, it gives a symplectic form ωΔ\omega_{\Delta} on (ℂ∗)n+1(\mathbb{C}^{*})^{n+1} that invariant with respect to the action of TΔT_{\Delta}. This gives us a moment map with respect to ωΔ\omega_{\Delta}

μΔ:(ℂ∗)n+1→Δ,μΔ​(z)=1∑j∈Δ∩ℤn+1|z2​j|​∑j∈Δ∩ℤn+1j​|z2​j|.\mu_{\Delta}:(\mathbb{C}^{*})^{n+1}\to\Delta,\mu_{\Delta}(z)=\frac{1}{\sum\limits_{j\in\Delta\cap\mathbb{Z}^{n+1}}|z^{2j}|}\sum\limits_{j\in\Delta\cap\mathbb{Z}^{n+1}}j|z^{2j}|.

The image of this embedding is the interior Int⁡Δ\operatorname{Int}\Delta. The map μ​Δ\mu\Delta can be compactified to the moment map μ¯Δ:ℂ​TΔ→Δ\bar{\mu}_{\Delta}:\mathbb{C}T_{\Delta}\to\Delta.

The maps Log:(ℂ∗)n+1→ℝn+1\operatorname{Log}:(\mathbb{C}^{*})^{n+1}\to\mathbb{R}^{n+1} and μΔ:(ℂ∗)n+1→Int⁡Δ\mu_{\Delta}:(\mathbb{C}^{*})^{n+1}\to\operatorname{Int}\Delta both have the orbits of Tn+1T^{n+1} as their fibers. Thus, they define a natural reparametrization

ΦΔ:ℝn+1→Int⁡Δ.\Phi_{\Delta}:\mathbb{R}^{n+1}\to\operatorname{Int}\Delta.
Definition 6.

Let Π⊂ℝn+1\Pi\subset\mathbb{R}^{n+1} be an nn-dimensional balanced polyhedral complex. By Proposition 1.4 there is a convex lattice polyhedron Δ\Delta dual to Π\Pi. We define Π¯⊂Δ\bar{\Pi}\subset\Delta, the compactification of Π\Pi, by taking the closure of ΦΔ​(Π)\Phi_{\Delta}(\Pi) in Δ\Delta. We call Π¯∖ΦΔ​(Π)\bar{\Pi}\smallsetminus\Phi_{\Delta}(\Pi) the boundary of Π¯\bar{\Pi}. For convenience from now on we identify Π\Pi and ΦΔ​(Π)\Phi_{\Delta}(\Pi).

Proposition 1.12.

Let Π\Pi be a dual Δ\Delta-complex and let Δ′⊂Δ\Delta^{\prime}\subset\Delta be a (k+1)(k+1)-dimensional face. Then the intersection Π¯∩Δ′\bar{\Pi}\cap\Delta^{\prime} is a compactification of a dual Δ′\Delta^{\prime}-complex Π′\Pi^{\prime}. If Π\Pi is maximal then Π′\Pi^{\prime} is also maximal.

We prove this proposition simultaneously with the following proposition describing the behavior of Π\Pi near infinity. Recall that a supporting vector v→\stackrel{{\scriptstyle\to}}{{v}} at a face Δ′⊂Δ\Delta^{\prime}\subset\Delta is a vector such that pv→|Δp_{\stackrel{{\scriptstyle\to}}{{v}}}|_{\Delta} reaches its maximum precisely over Δ′\Delta^{\prime}, where pv→p_{\stackrel{{\scriptstyle\to}}{{v}}} is the orthogonal projection in the direction of v→\stackrel{{\scriptstyle\to}}{{v}}.

Proposition 1.13.

The complex Π′\Pi^{\prime} from Proposition 1.12 can be obtained in the following way. Let L⊂ℝn+1L\subset\mathbb{R}^{n+1} be the linear (k+1)(k+1)-subspace parallel to the face Δ′\Delta^{\prime}. Let v→\stackrel{{\scriptstyle\to}}{{v}} be a supporting vector at Δ′\Delta^{\prime}. For a sufficiently large R>0R>0 we have Π′=(Π−Rv→)∩L\Pi^{\prime}=(\Pi-R\hskip-5.0pt\stackrel{{\scriptstyle\to}}{{v}})\cap L.

Proof.

From the finiteness condition in Definition 1 we have that the complex Π′=(Π−Rv→)∩L⊂L\Pi^{\prime}=(\Pi-R\hskip-5.0pt\stackrel{{\scriptstyle\to}}{{v}})\cap L\subset L does not depend on the choice of R>0R>0 and v→\stackrel{{\scriptstyle\to}}{{v}} as long as v→\stackrel{{\scriptstyle\to}}{{v}} is supporting and RR is sufficiently large. The proof of Proposition 1.4 ensures that Π′\Pi^{\prime} is a dual Δ′\Delta^{\prime}-complex. If Π\Pi is maximal then it is dual to a triangulation of Δ\Delta into simplices of minimal volume. Such a triangulation induces a triangulation into simplices of minimal volume on the faces Δ′\Delta^{\prime} and thus Π′\Pi^{\prime} is also maximal. ∎

If Π\Pi is a maximal dual Δ\Delta-complex then it is generic everywhere except at the points of its boundary ∂Π\partial\Pi. The following proposition describes the local topology of Π¯\bar{\Pi} near the boundary. It is a corollary of Proposition 1.12.

Proposition 1.14.

Suppose that Π\Pi is a maximal dual Δ\Delta-complex. A point xx in Π¯\bar{\Pi} has a neighborhood of one of the following (n+1)​(n+2)2\frac{(n+1)(n+2)}{2} types: ℝk×Σl−k×[0,+∞)n−l\mathbb{R}^{k}\times\Sigma^{l-k}\times[0,+\infty)^{n-l}, where k≤l≤nk\leq l\leq n. Here kk is the dimension of the open cell of Π¯\bar{\Pi} which contains xx while l+1l+1 is the dimension of the open face of Δ\Delta which contains xx.

We call a point with such a neighborhood a (k,l)(k,l)-point of Π¯\bar{\Pi}.

Remark 1.15.

The concept of generic polyhedron is closely related to that of special spine in Topology. We remind its definition. Let MM be a compact (n+1)(n+1)-manifold with boundary and Π¯⊂M\bar{\Pi}\subset M be an nn-dimensional CW-complex such that its every open cell is smoothly embedded to MM. The complex Π¯\bar{\Pi} is called a spine of MM if Π¯\bar{\Pi} is a deformational retract of MM. The spine Π¯\bar{\Pi} is called special if for any point x∈Π¯∖∂Mx\in\bar{\Pi}\smallsetminus\partial M from an open kk-cell there exists a neighborhood isomorphic to ℝk×Σn−k\mathbb{R}^{k}\times\Sigma^{n-k}.

Note that if Int⁡Δ∩ℤn+1=∅\operatorname{Int}\Delta\cap\mathbb{Z}^{n+1}=\emptyset then all the triangulation vertices of a dual Δ\Delta-polyhedron Π\Pi are from ∂Δ\partial\Delta then Π¯\bar{\Pi} is a spine of Δ\Delta. In general, Π¯\bar{\Pi} is a spine of the polyhedron Δ\Delta minus a small neighborhood of the interior lattice points. Note that Π¯\bar{\Pi} can be treated as a special spine of Δ\Delta if we treat Δ\Delta as a manifold with boundary and corners.

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