4.1. An embedding of into the model torus bundle
In [HZ02] we have used the GKZ machinery [GKZ94] for the
monomial estimates in the equation of to establish an embedding
of into . The same estimates can be
used to find bounds on the discrepancy of this embedding from being
holomorphic and isometric. Establishing these bounds will occupy the
rest of the section.
To define the fibration we assume that is
sufficiently far in the interior of , so that
is small in the -scale.
Lemma 4.1.
The amoeba lies outside of
. In particular, the foliation
of induces
the fibration by projection along the
leaves.
Proof.
Note that for any , if
, then . Hence the equation
cannot have solutions for
.
∎
From now on we will use the bi-PIKAS to identify
with and fix the vector field and the
foliation in . With this identification,
given a subset we denote by the closure of the set
in the toric variety
(cf. [HZ02]).
Then the smooth part of the hypersurface is defined as:
|
|
|
We define the map over the charts as the
restriction to of the quotient map:
|
|
|
Note that if is in a boundary toric divisor ,
then and are not well defined, but and
are. Hence, the map is well defined over
. The meaning of this map is the choice of local
coordinates for near (cf. [HZ02, Lemma 3.9]). Hence it
is holomorphic.
Before defining the map over the charts let us first make
some estimates in the spirit of Lemma 4.1. For an
element we will write if the (standard
Euclidean) distance from to 0 is less than . This
inequality is vacuous for .
Lemma 4.2.
If a point of lies in , then
|
|
|
|
|
|
where as .
Proof.
First of all note that since , both and are well
defined. Also by (2) of Lemma 3.2 the values of
the vector field are in . Hence, for any
the ray is in and the standard estimates on values
of the monomials at
apply:
|
|
|
Or putting them all together we have
|
|
|
Hence,
|
|
|
where as . Writing the equation of
in as
|
|
|
or, equivalently,
|
|
|
will give the claimed estimates.
∎
Now, identifying the tangent spaces of the torus fibers
with , we can pull back the vector field
to get a (constant) vector field on . Then the map
for the points in
will be defined as:
|
|
|
Here if is large enough, i.e. ,
then according to the Lemma 4.2 there is a preferred
continuous lift of to (the
neighborhood of 0 in) . We use this lift to first define the value
for in and then project it back to
.
Since for the definition of the map
over the charts is consistent with the above
definition on possible overlaps . Thus, we have a well defined map
which is an embedding [HZ02].