2. The hybrid space associated to an snc model [0153]
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
2. The hybrid space associated to an snc model
In this section, we show how to perform a topological surgery in a complex manifold, replacing a simple normal crossing divisor with its dual complex. Our construction is similar to the one used by Morgan-Shalen in [MS84, §I.3], and can even be traced back to the pioneering work of Bergman [Berg71].
2.1. The dual complex
Let be an effective divisor with simple normal crossing (snc) support in a complex manifold . By definition, with and a finite family of smooth irreducible divisors such that
is either empty or smooth of codimension (with finitely many connected components) for each . A connected component of a non-empty is called a stratum. Together with , the locally closed submanifolds define a partition of .
The dual complex is the simplicial complex44 4 This is understood in the slightly generalized sense that the intersection of two faces is a union of common faces. defined as follows: to each stratum corresponds a simplex
and is a face of if and only if . This description equips with an integral affine structure, by which we mean a compatible choice of integral affine structures on each simplex . This further induces a -PA structure on .
We write for the stratum of a face . Each point belongs to for a unique stratum , obtained as the connected component of containing , with . We denote by the corresponding face of .
2.2. The hybrid topology
Next we define a natural topology on the disjoint union
Consider a connected open set meeting and local coordinates on . We say that the pair is adapted (to ) if the following conditions hold:
- (i)
if are the irreducible components of intersecting , then we have for a component of ;
- (ii)
is an equation of with , .
We call the stratum of , and denote by
the corresponding face of . The function is an equation of in , with , and we get a continuous map by setting
For any two adapted coordinate charts , , with the same stratum , we have with nonvanishing on , for (after a possible reindexing); it follows that
| (2.1) |
locally uniformly on . We next show how to globalize this construction.
Proposition 2.1.
There exists an open neighborhood of and a continuous map such that for each adapted coordinate chart with we have and
| (2.2) |
uniformly on compact subsets of .
This will be accomplished by means of a partition of unity, using the following elementary special case of [Cle77, Theorem 5.7].
Lemma 2.2.
There exists a family of adapted coordinate charts, such that forms a locally finite covering of and such that the strata of the satisfy
| (2.3) |
for every finite .
Proof of Proposition 2.1.
Pick an open cover as in Lemma 2.2, and denote by the corresponding maps. Set , and pick a partition of unity subordinate to . We claim that for each there exists an open neighborhood of and a face of such that
for any . Indeed, using (2.3) it is easy to see that
satisfies this property. By convexity of , it follows that is well-defined on , and hence yields a continuous map . The last property is a direct consequence of (2.1). ∎
We extend the previous map as
by setting on .
Definition 2.3.
The hybrid topology on is defined as the coarsest topology such that:
- (i)
is an open embedding;
- (ii)
For every open neighborhood of in , the set is open in ;
- (iii)
is continuous.
Using (2.2), this definition is easily seen to be independent of the choice of map . If is compact and is a compact neighborhood of , then one easily checks that the corresponding subset is compact (Hausdorff). When has only one irreducible component, is simply the Tychonoff one-point compactification of .
Example 2.4.
Set and the union of the coordinate axes, with coordinates . Then is itself an adapted coordinate chart. In these coordinates, becomes the map sending to . As a consequence, given and , the closure in of the closed subset
is given by , where . Further, the sets , for form a basis of closed neighborhoods of the point in . See Figure 1.