This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Holomorphic curves in Exploded Torus Fibrations: Compactness

Brett Parker
parker@math.mit.edu
Abstract

The category of exploded torus fibrations is an extension of the category of smooth manifolds in which some adiabatic limits look smooth. (For example, the type of limits considered in tropical geometry appear smooth.) In this paper, we prove a compactness theorem for (pseudo)-holomorphic curves in exploded torus fibrations. In the case of smooth manifolds, this is just a version of Gromov’s compactness theorem in a topology strong enough for gluing analysis.

1 Introduction

As the subject of this paper is holomorphic curves in a new category, the reader desiring rigorous statements will be disappointed in this introduction, and should wait until terms have been defined.

Holomorphic curves have been a powerful tool in symplectic topology since Gromov’s original paper on the subject, [5]. The basic idea is that given a symplectic manifold, there is a contractible choice of almost complex structure so that the symplectic form is positive on holomorphic planes. The ‘moduli space’ of (pseudo)-holomorphic curves with a choice of one of these almost complex structures is in some sense ‘smooth’, ‘compact’ and finite dimensional, and topological invariants of this ‘moduli space’ give invariants of the underlying symplectic manifold.

The number of quotation marks required for the above statements is perhaps one indication that the category of smooth manifolds is not the ideal setting for working with holomorphic curves. In particular, in order for our moduli space to be ‘compact’, we have to include connected ‘smooth’ families of holomorphic curves which exhibit bubbling behavior, and change topology. This certainly does not fit the usual definition of a smooth family (although it is more natural from the perspective of algebraic geometry). The topology in which the moduli space is compact is also so unnatural to the smooth category that it is difficult to state simply.

A second reason to leave the smooth category is that holomorphic curves can be very difficult to find directly there. A common technique used to find holomorphic curve invariants is to consider a family of almost complex structures which degenerates somehow to a limit in which holomorphic curves become simpler, and then reconstruct holomorphic curve invariants from limiting information. In the category of exploded torus fibrations, some of these degenerations can be viewed as smooth families, and the limiting problem is a problem of the same type.

One example of such a degeneration is that considered in the algebraic case by [10]. This is a degeneration represented by a flat family of projective schemes so that the fiber at 0 has two components intersecting transversely over a smooth divisor, and all other fibers are smooth. This corresponds in the symplectic setting to the degeneration used to represent a symplectic sum, used in [7] to prove the symplectic sum formula for Gromov-Witten invariants. Similar degenerations have been used to study holomorphic curves in [9] and [2]. More generally, one might consider the case of a flat family in which the central fiber is reduced with simple normal crossing singularities, or the analogous case in the symplectic setting. This type of degeneration can be represented by a smooth family in the category of exploded torus fibrations. In the exploded category, an object 𝔅\mathfrak{B} corresponding to this central fiber will have a smooth part 𝔅\lceil\mathfrak{B}\rceil which is equal to the central fiber, and a tropical part 𝔅\lfloor\mathfrak{B}\rfloor which is a stratified integral affine or ‘tropical’ space which has a vertex for every component of the central fiber 𝔅\lceil\mathfrak{B}\rceil, an edge for every intersection, and an nn dimensional face for each complex codimension nn intersection of some number of components. Holomorphic curves in 𝔅\mathfrak{B} look like usual holomorphic curves projected to 𝔅\lceil\mathfrak{B}\rceil, and tropical curves projected to 𝔅\lfloor\mathfrak{B}\rfloor.

This tropical part 𝔅\lfloor\mathfrak{B}\rfloor is related to the large complex structure limit studied in some approaches to mirror symmetry such as found in [6], [4] or [8]. Studying the tropical curves in 𝔅\lfloor\mathfrak{B}\rfloor can sometimes give complete information about holomorphic curve invariants. One example of this is in [11].

The goal of this paper is to provide the relevant compactness theorems for holomorphic curves in exploded torus fibrations analogous to Gromov’s compactness theorem ( stated precisely on page 4.1). This is one step towards having a good holomorphic curve theory for exploded torus fibrations. It is anticipated that the moduli stack of holomorphic curves in exploded fibrations also admits a natural type of Kuranishi structure which together with these compactness results will allow the definition of holomorphic curve invariants such as Gromov Witten invariants.

2 Definitions

2.1 Tropical semiring and exploded semiring

Functions on exploded fibrations will sometimes take values in the following semirings:

The tropical semiring is a semiring 𝔱\mathfrak{t}^{\mathbb{R}} which is equal to \mathbb{R} with ‘multiplication’ being the operation of usual addition and ‘addition’ being the operation of taking a minimum. We will write elements of 𝔱\mathfrak{t}^{{\mathbb{R}}} as 𝔱x\mathfrak{t}^{x} where xx\in\mathbb{R}. Then we can write the operations as follows

𝔱x𝔱y:=𝔱x+y\mathfrak{t}^{x}\mathfrak{t}^{y}:=\mathfrak{t}^{x+y}
𝔱x+𝔱y:=𝔱min{x,y}\mathfrak{t}^{x}+\mathfrak{t}^{y}:=\mathfrak{t}^{\min\{x,y\}}

𝔱1\mathfrak{t}^{1} can be thought of as something which is infinitesimally small. With this in mind, we have the following order on 𝔱\mathfrak{t}^{\mathbb{R}}:

𝔱x>𝔱y when x<y\mathfrak{t}^{x}>\mathfrak{t}^{y}\text{ when }x<y

Given a ring RR, the exploded semiring R𝔱R\mathfrak{t}^{{\mathbb{R}}} consists of elements c𝔱xc\mathfrak{t}^{x} with cRc\in R and xx\in\mathbb{R}. Addition and multiplication are as follows:

c1𝔱xc2𝔱y=c1c2𝔱x+yc_{1}\mathfrak{t}^{x}c_{2}\mathfrak{t}^{y}=c_{1}c_{2}\mathfrak{t}^{x+y}
c1𝔱x+c2𝔱y={c1𝔱x if x<y(c1+c2)𝔱x if x=yc2𝔱y if x>yc_{1}\mathfrak{t}^{x}+c_{2}\mathfrak{t}^{y}=\begin{cases}&c_{1}\mathfrak{t}^{x}\text{ if }x<y\\ &(c_{1}+c_{2})\mathfrak{t}^{x}\text{ if }x=y\\ &c_{2}\mathfrak{t}^{y}\text{ if }x>y\end{cases}

It is easily checked that addition and multiplication are associative and obey the usual distributive rule. We will mainly be interested in 𝔱\mathbb{C}\mathfrak{t}^{\mathbb{R}} and 𝔱\mathbb{R}\mathfrak{t}^{\mathbb{R}}.

There are semiring homomorphisms

R𝜄R𝔱𝔱R\xrightarrow{\iota}R\mathfrak{t}^{\mathbb{R}}\xrightarrow{\lfloor\cdot\rfloor}\mathfrak{t}^{\mathbb{R}}

defined by

ι(c):=c𝔱0\iota(c):=c\mathfrak{t}^{0}
c𝔱x:=𝔱x\lfloor c\mathfrak{t}^{x}\rfloor:=\mathfrak{t}^{x}

The homomorphism \lfloor\cdot\rfloor is especially important. We shall call c𝔱x=𝔱x\lfloor c\mathfrak{t}^{x}\rfloor=\mathfrak{t}^{x} the tropical part of c𝔱xc\mathfrak{t}^{x}. There will be an analogous ‘tropical part’ of ‘basic’ exploded fibrations which can be thought of as the large scale.

Define the positive exploded semiring R𝔱+R\mathfrak{t}^{\mathbb{R}^{+}} to be the sub semiring of R𝔱R\mathfrak{t}^{\mathbb{R}} consisting of elements of the form c𝔱xc\mathfrak{t}^{x} where x0x\geq 0. There is a semiring homomorphism

:R𝔱+R\lceil\cdot\rceil:R\mathfrak{t}^{\mathbb{R}^{+}}\longrightarrow R

given by

c𝔱x:={c if x=00 if x>0\lceil c\mathfrak{t}^{x}\rceil:=\begin{cases}&c\text{ if }x=0\\ &0\text{ if }x>0\end{cases}

Note that the above homomorphism can be thought of as setting 𝔱1=0\mathfrak{t}^{1}=0, which is intuitive when 𝔱1\mathfrak{t}^{1} is thought of as infinitesimally small.

We shall use the following order on (0,)𝔱(0,\infty)\mathfrak{t}^{\mathbb{R}}, which again is intuitive if 𝔱1\mathfrak{t}^{1} is thought of as being infinitesimally small and positive:

x1𝔱y1<x2𝔱y2 whenever y1>y2 or y1=y2 and x1<x2x_{1}\mathfrak{t}^{y_{1}}<x_{2}\mathfrak{t}^{y_{2}}\text{ whenever }y_{1}>y_{2}\text{ or }y_{1}=y_{2}\text{ and }x_{1}<x_{2}

2.2 Exploded 𝕋\mathbb{T} fibrations

The following definition of abstract exploded fibrations is far too general, but it allows us to talk about local models for exploded torus fibrations as abstract exploded fibrations without giving too many definitions beforehand.

Definition 2.1.

An abstract exploded fibration 𝔅\mathfrak{B} consists the following:

  1. 1.

    A set of points

    {p𝔅}\{p\rightarrow\mathfrak{B}\}
  2. 2.

    A Hausdorff topological space [𝔅][\mathfrak{B}] with a map

    {p𝔅}[][𝔅]\{p\rightarrow\mathfrak{B}\}\xrightarrow{[\cdot]}[\mathfrak{B}]
  3. 3.

    A sheaf of Abelian groups on [𝔅][\mathfrak{B}], ×(𝔅)\mathcal{E}^{\times}(\mathfrak{B}) called the sheaf of exploded functions on 𝔅\mathfrak{B} so that:

    1. (a)

      Each element f×(U)f\in\mathcal{E}^{\times}(U) is a function defined on the set of points over UU,

      f:[]1(U)R𝔱f:[\cdot]^{-1}(U)\longrightarrow R^{*}\mathfrak{t}^{\mathbb{R}}

      where RR^{*} denotes the multiplicatively invertible elements of some ring RR. (For this paper, we shall use R=R=\mathbb{C}.)

    2. (b)

      ×(U)\mathcal{E}^{\times}(U) includes the constant functions if UU\neq\emptyset.

    3. (c)

      Multiplication is given by pointwise multiplication in R𝔱R^{*}\mathfrak{t}^{\mathbb{R}}.

    4. (d)

      Restriction maps are given by restriction of functions.

We shall use the terminology of an open subset UU of an abstract exploded fibration 𝔅\mathfrak{B} to mean an open subset U[𝔅]U\subset[\mathfrak{B}] along with the set of points in []1(U)[\cdot]^{-1}(U) and the sheaf ×(U)\mathcal{E}^{\times}(U). The strange notation for the set of points in 𝔅\mathfrak{B}, {p𝔅}\{p\rightarrow\mathfrak{B}\} is used because it is equal to the set of morphisms of a point into 𝔅\mathfrak{B}.

Definition 2.2.

A morphism f:𝔅f:\mathfrak{B}\longrightarrow\mathfrak{C} of abstract exploded fibrations is a map on points and a continuous map of topological spaces so that the following diagram commutes

{p𝔅}𝑓{p}[][][𝔅][f][]\begin{array}[]{ccc}\{p\rightarrow\mathfrak{B}\}&\xrightarrow{f}&\{p\rightarrow\mathfrak{C}\}\\ \downarrow[\cdot]&&\downarrow[\cdot]\\ [\mathfrak{B}]&\xrightarrow{[f]}&[\mathfrak{C}]\end{array}

and so that ff preserves ×\mathcal{E}^{\times} in the sense that if g×(U)g\in\mathcal{E}^{\times}(U), then fgf\circ g is in ×([f]1(U))\mathcal{E}^{\times}\left([f]^{-1}(U)\right)

fg:=fg×([f]1(U))f^{*}g:=f\circ g\in\mathcal{E}^{\times}\left([f]^{-1}(U)\right)

Note that the reader should not attempt to work directly with the above definition as it is too general. It simply enables us to talk about examples without giving a complete definition beforehand. The next sequence of examples will give us local models for exploded 𝕋\mathbb{T} fibrations.

Example 2.3.

Any smooth manifold MM is a smooth exploded 𝕋\mathbb{T} fibration. In this case, the underlying toplogical space [M][M] is just MM with the usual topology, the set of points {pM}\{p\rightarrow M\} the usual set of points in MM, and []:{pM}[M][\cdot]:\{p\rightarrow M\}\longrightarrow[M] the identity map. The sheaf ×(M)\mathcal{E}^{\times}(M) consists of all functions of the form f𝔱af\mathfrak{t}^{a} where fC(M,)f\in C^{\infty}(M,\mathbb{C}^{*}) and aa is a locally constant \mathbb{R} valued function.

The reader should convince themselves that this is just a strange way of encoding the usual data of a smooth manifold, and that a exploded morphism between smooth manifolds is simply a smooth map. This makes the category of smooth manifolds a full subcategory of the category of exploded 𝕋\mathbb{T} fibrations.

The above example should be considered as a ‘completely smooth’ object. At the other extreme, we have the following ‘completely tropical’ object.

Example 2.4.

The exploded fibration 𝔗n\mathfrak{T}^{n} is best described using coordinates (z~1,,z~n)(\tilde{z}_{1},\dotsc,\tilde{z}_{n}) where each z~i×(𝔗n)\tilde{z}_{i}\in\mathcal{E}^{\times}(\mathfrak{T}^{n}). The set of points in 𝔗n\mathfrak{T}^{n}, {p𝔗n}\{p\rightarrow\mathfrak{T}^{n}\} are then identified with (𝔱)n\left(\mathbb{C}^{*}\mathfrak{t}^{\mathbb{R}}\right)^{n} by prescribing the values of (z~1,,z~n)(\tilde{z}_{1},\dotsc,\tilde{z}_{n}). The underlying topological space [𝔗n][\mathfrak{T}^{n}] is 𝔱n\mathfrak{t}^{\mathbb{R}^{n}} (given the usual topology on n\mathbb{R}^{n}), and

[(z~1,,z~n)]=(z~1,,z~n) or [(c1𝔱a1,,cn𝔱an)]=(𝔱a1,,𝔱an)[(\tilde{z}_{1},\dotsc,\tilde{z}_{n})]=(\lfloor\tilde{z}_{1}\rfloor,\dotsc,\lfloor\tilde{z}_{n}\rfloor)\text{ or }[(c_{1}\mathfrak{t}^{a_{1}},\dotsc,c_{n}\mathfrak{t}^{a_{n}})]=(\mathfrak{t}^{a_{1}},\dotsc,\mathfrak{t}^{a_{n}})

Note that there is a ()n(\mathbb{C}^{*})^{n} worth of points p𝔗np\longrightarrow\mathfrak{T}^{n} over every topological point [p][𝔗n][p]\in[\mathfrak{T}^{n}].

Exploded functions in ×(𝔗n)\mathcal{E}^{\times}\left(\mathfrak{T}^{n}\right) can be written in these coordinates as

c𝔱yz~α:=c𝔱yz~iαic\mathfrak{t}^{y}\tilde{z}^{\alpha}:=c\mathfrak{t}^{y}\prod\tilde{z}_{i}^{\alpha_{i}}

where cc\in\mathbb{C}^{*}, yy\in\mathbb{R} and αn\alpha\in\mathbb{Z}^{n} are all locally constant functions.

The reader should now be able to verify the following observations:

  1. 1.

    A morphism from 𝔗n\mathfrak{T}^{n} to a smooth manifold is simply given by a map [𝔗n]M[\mathfrak{T}^{n}]\longrightarrow M which has as its image a single point in MM.

  2. 2.

    A morphism from a connected smooth manifold MM to 𝔗n\mathfrak{T}^{n} has the information of a map [f][f] from MM to a point [p][𝔗n][p]\in[\mathfrak{T}^{n}] and a smooth map ff from MM to the ()n(\mathbb{C}^{*})^{n} worth of points over [p][p]. The map ff is determined by specifying the nn exploded functions

    f(z~i)×(M)f^{*}(\tilde{z}_{i})\in\mathcal{E}^{\times}\left(M\right)

Note in particular that ×(M)\mathcal{E}^{\times}\left(M\right) is equal to the sheaf of smooth morphisms of MM to 𝔗\mathfrak{T}. This will be true in general. A smooth morphism f:𝔅𝔗nf:\mathfrak{B}\longrightarrow\mathfrak{T}^{n} is equivalent to the choice of nn exploded functions in ×(𝔅)\mathcal{E}^{\times}\left(\mathfrak{B}\right) corresponding to f(z~i)f^{*}(\tilde{z}_{i}).

Now, a hybrid object, part ‘smooth’, part ‘tropical’.

Example 2.5.

The exploded fibration 𝔗[0,)1:=𝔗11\mathfrak{T}^{1}_{[0,\infty)}:=\mathfrak{T}^{1}_{1} is more complicated. We shall describe this using the coordinate z~×(𝔗11)\tilde{z}\in\mathcal{E}^{\times}(\mathfrak{T}^{1}_{1}). The set points {p𝔗11}\{p\rightarrow\mathfrak{T}^{1}_{1}\} is identified with 𝔱+\mathbb{C}^{*}\mathfrak{t}^{\mathbb{R}^{+}} by specifying the value of z~(p)𝔱+\tilde{z}(p)\in\mathbb{C}^{*}\mathfrak{t}^{\mathbb{R}^{+}}.

The underlying topological space is given as follows:

[𝔗11]:=𝔱+{0}={𝔱0𝔱+}:={(z,𝔱a)×𝔱+ so that za=0}\begin{split}[\mathfrak{T}^{1}_{1}]&:=\frac{\mathbb{C}\coprod\mathfrak{t}^{\mathbb{R}^{+}}}{\{0\in\mathbb{C}\}=\{\mathfrak{t}^{0}\in\mathfrak{t}^{\mathbb{R}^{+}}\}}\\ &:=\{(z,\mathfrak{t}^{a})\in\mathbb{C}\times\mathfrak{t}^{\mathbb{R}^{+}}\text{ so that }za=0\}\end{split}
\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\includegraphics{ETF1}

Our map from points to [𝔗11][\mathfrak{T}^{1}_{1}] is given by

[z~]:=(z~,z~)[\tilde{z}]:=\left(\lceil\tilde{z}\rceil,\lfloor\tilde{z}\rfloor\right)

The above expression uses the maps defined in section 2.1, c𝔱a=𝔱a\lfloor c\mathfrak{t}^{a}\rfloor=\mathfrak{t}^{a}, and c𝔱a\lceil c\mathfrak{t}^{a}\rceil is equal to cc if a=0a=0 and 0 if a>0a>0. Note that there is one point p𝔗11p\rightarrow\mathfrak{T}^{1}_{1} over points of the form (c,𝔱0)[𝔗11](c,\mathfrak{t}^{0})\in[\mathfrak{T}^{1}_{1}], a \mathbb{C}^{*} worth of points over (0,𝔱a)[𝔗11](0,\mathfrak{t}^{a})\in[\mathfrak{T}^{1}_{1}] for a>0a>0 and no points over (0,𝔱0)[𝔗11](0,\mathfrak{t}^{0})\in[\mathfrak{T}^{1}_{1}].

This object should be thought of as a copy of \mathbb{C} which we puncture at 0, and consider this puncture to be an asymptoticaly cylindrical end. Each copy of \mathbb{C}^{*} over (0,𝔱a)(0,\mathfrak{t}^{a}) should be thought of as some ‘cylinder at \infty’. Note that even though there is a (0,)(0,\infty) worth of two dimensional cylinders involved in this object, it should still be thought of being two dimensional. This strange feature is essential for the exploded category to have a good holomorphic curve theory. (Actually, we could just have easily had a +\mathbb{Q}^{+} worth of cylinders at infinity, and worked over the semiring 𝔱\mathbb{C}\mathfrak{t}^{\mathbb{Q}} instead of 𝔱\mathbb{C}\mathfrak{t}^{\mathbb{R}}, but this author prefers the real numbers.)

We shall use the notation

𝔗11:=[𝔗11]𝔱+=0=\lceil\mathfrak{T}^{1}_{1}\rceil:=\frac{[\mathfrak{T}^{1}_{1}]}{\mathfrak{t}^{\mathbb{R}^{+}}=0}=\mathbb{C}
𝔗11:=[𝔗11]=0=𝔱+\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\frac{[\mathfrak{T}^{1}_{1}]}{\mathbb{C}=0}=\mathfrak{t}^{\mathbb{R}^{+}}

We shall call 𝔗11\lceil\mathfrak{T}^{1}_{1}\rceil the smooth or topological part of 𝔗11\mathfrak{T}^{1}_{1} and 𝔗11\lfloor\mathfrak{T}^{1}_{1}\rfloor the tropical part of 𝔗11\mathfrak{T}^{1}_{1}.

We can write any exploded function h×(𝔗11)h\in\mathcal{E}^{\times}\left(\mathfrak{T}^{1}_{1}\right) as

h(z~)=f(z~)𝔱yz~α for fC(,), and y,α locally constant.h(\tilde{z})=f(\lceil\tilde{z}\rceil)\mathfrak{t}^{y}\tilde{z}^{\alpha}\text{ for }f\in C^{\infty}(\mathbb{C},\mathbb{C}^{*})\text{, and }y\in\mathbb{R},\alpha\in\mathbb{Z}\text{ locally constant.}

We can now see that 𝔗11\mathfrak{T}^{1}_{1} is a kind of hybrid of the last two examples. Restricting to an open set contained inside {z~=𝔱0}[𝔗11]\{\lfloor\tilde{z}\rfloor=\mathfrak{t}^{0}\}\subset[\mathfrak{T}^{1}_{1}] we get part of a smooth manifold. Restricting to an open set contained inside {z~=0}[𝔗11]\{\lceil\tilde{z}\rceil=0\}\subset[\mathfrak{T}^{1}_{1}], we get part of 𝔗\mathfrak{T}.

We can define ×+(𝔅){}^{+}\mathcal{E}^{\times}\left(\mathfrak{B}\right) to consist of all functions in ×(𝔅)\mathcal{E}^{\times}(\mathfrak{B}) which take values only in 𝔱+\mathbb{C}^{*}\mathfrak{t}^{\mathbb{R}^{+}}. Note that a smooth morphism f:𝔅𝔗11f:\mathfrak{B}\longrightarrow\mathfrak{T}^{1}_{1} from any exploded 𝕋\mathbb{T} fibration 𝔅\mathfrak{B} that we have described in examples so far is given by a choice of exploded function f(z~)×+(𝔅)f^{*}(\tilde{z})\in{}^{+}\mathcal{E}^{\times}\left(\mathfrak{B}\right). This will be true for exploded 𝕋\mathbb{T} fibrations in general.

We can build up ‘exploded curves’ using coordinate charts modeled on open subsets of 𝔗11\mathfrak{T}^{1}_{1}. For an example of this see the picture on page 2.18. For an example of a map of this to 𝔗n\mathfrak{T}^{n} see example 2.39 on page 2.39. Those interested in tropical geometry may like the following example 2.40. For a local example of a family of such objects, see example 2.47 on page 2.47.

We shall use the special notation 𝔗nm\mathfrak{T}^{m}_{n} to denote (𝔗11)n×𝔗(mn)(\mathfrak{T}^{1}_{1})^{n}\times\mathfrak{T}^{(m-n)}. (These spaces will be described rigorously below.) Before giving the most general local model for exploded 𝕋\mathbb{T} fibrations, we give the following definition which can be thought of as a local model for the tropical part of our exploded fibrations.

Definition 2.6.

An integral cone A𝔱mA\subset\mathfrak{t}^{\mathbb{R}^{m}} is a subset of 𝔱m\mathfrak{t}^{\mathbb{R}^{m}} given by a collection of inequalities

A:={𝔱a:=(𝔱a1,,𝔱am)𝔱m so that 𝔱aαi:=j(𝔱aj)αji𝔱+}A:=\left\{\mathfrak{t}^{a}:=(\mathfrak{t}^{a_{1}},\dotsc,\mathfrak{t}^{a_{m}})\in\mathfrak{t}^{\mathbb{R}^{m}}\text{ so that }\mathfrak{t}^{a\cdot\alpha^{i}}:=\prod_{j}(\mathfrak{t}^{a_{j}})^{\alpha_{j}^{i}}\in\mathfrak{t}^{\mathbb{R}^{+}}\right\}
In other words, A:={𝔱a so that aαi0}\text{In other words, }A:=\{\mathfrak{t}^{a}\text{ so that }a\cdot\alpha^{i}\geq 0\}

where αi\alpha^{i} are vectors in m\mathbb{Z}^{m}.

The dual cone AA^{*} to AA is defined to be

A:={αm:αa0,𝔱aA}A^{*}:=\{\alpha\in\mathbb{Z}^{m}:\alpha\cdot a\geq 0,\forall\mathfrak{t}^{a}\in A\}
\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\includegraphics{ETF2}

The following is the most general local model for exploded 𝕋\mathbb{T} fibrations.

Example 2.7.

Given any integral cone A𝔱mA\subset\mathfrak{t}^{\mathbb{R}^{m}} which generates 𝔱m\mathfrak{t}^{\mathbb{R}^{m}} as a group, we will now describe the exploded fibration n×𝔗Am\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}. We shall use coordinates

(x1,,xn,z~1,,z~m)(x_{1},\dotsc,x_{n},\tilde{z}_{1},\dotsc,\tilde{z}_{m})

The set of points {pn×𝔗Am}\{p\rightarrow\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}\} is then given by

{(x1,,xm,c1𝔱a1,,cm𝔱am)n×(𝔱)m;(𝔱a1,,𝔱am)A}\left\{(x_{1},\dotsc,x_{m},c_{1}\mathfrak{t}^{a_{1}},\dotsc,c_{m}\mathfrak{t}^{a_{m}})\in\mathbb{R}^{n}\times\left(\mathbb{C}^{*}\mathfrak{t}^{\mathbb{R}}\right)^{m};(\mathfrak{t}^{a_{1}},\dotsc,\mathfrak{t}^{a_{m}})\in A\right\}

We will describe the underlying topological space [n×𝔗Am][\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}] using some extra coordinate choices. Let {α1,,αk}\{\alpha^{1},\dotsc,\alpha^{k}\} generate the dual cone AmA^{*}\subset\mathbb{Z}^{m} as a semigroup with the relations

iri,jαi=0\sum_{i}r_{i,j}\alpha^{i}=0

Label coordinates for k\mathbb{C}^{k} as (zα1,,zαk)(z^{\alpha^{1}},\dotsc,z^{\alpha^{k}}), and use 𝔱a\mathfrak{t}^{a} to denote points in AA. We can describe the underlying topological space, [n×𝔗Am][\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}] as a subspace

[n×𝔗Am]={(x,zαi,𝔱a);i(zαi)ri,j=1,(αia)zαi=0}n×k×A[\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}]=\left\{(x,z^{\alpha^{i}},\mathfrak{t}^{a});\ \prod_{i}(z^{\alpha^{i}})^{r_{i,j}}=1,(\alpha^{i}\cdot a)z^{\alpha^{i}}=0\right\}\subset\mathbb{R}^{n}\times\mathbb{C}^{k}\times A

Again we shall use the notation

n×𝔗Am:=A, the tropical part \lfloor\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}\rfloor:=A,\text{ the tropical part }
n×𝔗Am:={i(zαi)ri,j=1}n×k, the smooth part\lceil\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}\rceil:=\left\{\prod_{i}(z^{\alpha^{i}})^{r_{i,j}}=1\right\}\subset\mathbb{R}^{n}\times\mathbb{C}^{k},\text{ the smooth part}

(This smooth part is equal to n\mathbb{R}^{n} times the toric variety corresponding to the cone AA.) Using the shorthand z~α:=iz~iαi\tilde{z}^{\alpha}:=\prod_{i}\tilde{z}_{i}^{\alpha_{i}}, and recalling that c𝔱a\lceil c\mathfrak{t}^{a}\rceil is equal to cc if a=0a=0, and equal to 0 if a>0a>0, we can write the map from the set of points down to the smooth part as

(x,z~):=(x,z~α1,,z~αk)\lceil(x,\tilde{z})\rceil:=(x,\lceil\tilde{z}^{\alpha^{1}}\rceil,\dotsc,\lceil\tilde{z}^{\alpha^{k}}\rceil)

The map from the set of points down to the tropical part is given by

(x,z~):=(z~1,,z~m) or (x,c1𝔱a1,,cm𝔱am)=(𝔱a1,,𝔱am)\lfloor(x,\tilde{z})\rfloor:=(\lfloor\tilde{z}_{1}\rfloor,\dotsc,\lfloor\tilde{z}_{m}\rfloor)\text{ or }\lfloor(x,c_{1}\mathfrak{t}^{a_{1}},\dotsc,c_{m}\mathfrak{t}^{a_{m}})\rfloor=(\mathfrak{t}^{a_{1}},\dotsc,\mathfrak{t}^{a_{m}})

The map from the set of points down to the underlying topological space is then defined by

[(x,z~)]:=((x,z~),(x,z~))[(x,\tilde{z})]:=\left(\lceil(x,\tilde{z})\rceil,\lfloor(x,\tilde{z})\rfloor\right)

(Note that we can also describe [n×𝔗Am][\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}] as the closure of the image of the above map inside n×k×A\mathbb{R}^{n}\times\mathbb{C}^{k}\times A.) Exploded functions h×(n×𝔗Am)h\in\mathcal{E}^{\times}\left(\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}\right) are all functions that can be written as

h(x,z~):=f((x,z~))z~α𝔱yh(x,\tilde{z}):=f\left(\lceil(x,\tilde{z})\rceil\right)\tilde{z}^{\alpha}\mathfrak{t}^{y}

where fC(n×k,)f\in C^{\infty}(\mathbb{R}^{n}\times\mathbb{C}^{k},\mathbb{C}^{*}), and αm\alpha\in\mathbb{Z}^{m} and yy\in\mathbb{R} are locally constant. It is important to note that two exploded functions are equal if they have the same values on points (which means that we can represent the same exploded function using different smooth functions ff.) Note also that this definition is independent of the choice of basis {αi}\{\alpha^{i}\} for AA^{*}.

The earlier examples 𝔗11=𝔗𝔱[0,)1\mathfrak{T}^{1}_{1}=\mathfrak{T}^{1}_{\mathfrak{t}^{[0,\infty)}}, and 𝔗n=𝔗𝔱nn\mathfrak{T}^{n}=\mathfrak{T}^{n}_{\mathfrak{t}^{\mathbb{R}^{n}}}, and

𝔗nm=(𝔗11)n×𝔗(mn):=𝔗𝔱(+)n×(mn)m\mathfrak{T}^{m}_{n}=(\mathfrak{T}^{1}_{1})^{n}\times\mathfrak{T}^{(m-n)}:=\mathfrak{T}^{m}_{\mathfrak{t}^{(\mathbb{R}^{+})^{n}\times\mathbb{R}^{(m-n)}}}

These examples are much easier to understand than the most general model, so if the above definition was confusing, it is worthwhile reading through it again with these examples in mind. In this case, we can choose the obvious basis for AA^{*} given by the first nn standard basis vectors, so z~αi:=z~i\tilde{z}^{\alpha^{i}}:=\tilde{z}_{i}. The smooth part of 𝔗nm\mathfrak{T}^{m}_{n}, is then 𝔗nm:=n\lceil\mathfrak{T}^{m}_{n}\rceil:=\mathbb{C}^{n} with coordinates (z~1,,z~n)(\lceil\tilde{z}_{1}\rceil,\dotsc,\lceil\tilde{z}_{n}\rceil). An important example of exploded fibrations locally modeled on open subsets of 𝔗nm\mathfrak{T}^{m}_{n} is given in example 2.37 on page 2.37. We need the more confusing models 𝔗Am\mathfrak{T}^{m}_{A} because they arise naturally in the intersection theory of exploded fibrations. They also occur in moduli spaces of holomorphic curves. We need to consider families of holomorphic curves with total spaces locally modeled on these more complicated 𝔗Am\mathfrak{T}^{m}_{A}.

Definition 2.8.

A smooth exploded 𝕋\mathbb{T} fibration 𝔅\mathfrak{B} is an abstract exploded fibration locally modeled on open subsets of n×𝔗Am\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}.

In other words, each point in [𝔅][\mathfrak{B}] has a neighborhood UU which is isomorphic as an abstract exploded fibration to some open set in n×𝔗Am\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}. The model n×𝔗Am\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A} used may depend on the open set.

There are some extra sheaves of functions which are defined on any smooth exploded 𝕋\mathbb{T} fibration 𝔅\mathfrak{B} which will come in useful:

Definition 2.9.

The sheaf of smooth functions C(𝔅)C^{\infty}(\mathfrak{B}) is the sheaf of smooth morphisms of 𝔅\mathfrak{B} to \mathbb{R} considered as a smooth exploded 𝕋\mathbb{T} fibration.

Definition 2.10.

The sheaf of integral affine functions 𝒜×(𝔅)\mathcal{A}^{\times}(\mathfrak{B}) consists of functions of the form f\lfloor f\rfloor where f×(𝔅)f\in\mathcal{E}^{\times}(\mathfrak{B}).

Recall that c𝔱a:=𝔱a\lfloor c\mathfrak{t}^{a}\rfloor:=\mathfrak{t}^{a}. We shall also use the terminology that the order of c𝔱ac\mathfrak{t}^{a} is aa. The sheaf 𝒜×(𝔗Am)\mathcal{A}^{\times}(\mathfrak{T}^{m}_{A}) corresponds to functions on AA of the form 𝔱a𝔱y+αa\mathfrak{t}^{a}\mapsto\mathfrak{t}^{y+\alpha\cdot a} where αm\alpha\in\mathbb{Z}^{m} and yy\in\mathbb{R}.

Definition 2.11.

The sheaf of exploded tropical functions (𝔅)\mathcal{E}(\mathfrak{B}) is the sheaf of 𝔱\mathbb{C}\mathfrak{t}^{\mathbb{R}} valued functions which are locally equal to a finite sum of exploded functions ×\mathcal{E}^{\times}. (‘Sum’ means sum using pointwise addition in 𝔱\mathbb{C}\mathfrak{t}^{\mathbb{R}}.) The sheaf of tropical functions 𝒜(𝔅)\mathcal{A}(\mathfrak{B}) consists of functions of the form f\lfloor f\rfloor for some f(𝔅)f\in\mathcal{E}(\mathfrak{B}).

For example 𝒜(𝔗n)\mathcal{A}(\mathfrak{T}^{n}) consists of functions of the form 𝔱f\mathfrak{t}^{f} where f:nf:\mathbb{R}^{n}\longrightarrow\mathbb{R} is some continuous, concave piecewise integral affine function. Functions of this form are usually called tropical functions. We shall usually not use (𝔅)\mathcal{E}(\mathfrak{B}) because the addition is strange, and ×(𝔅)\mathcal{E}^{\times}(\mathfrak{B}) is naturally defined as the sheaf of morphisms to 𝔗\mathfrak{T}. The only place that (𝔅)\mathcal{E}(\mathfrak{B}) will be used in this paper will be in the definition of the tangent space of 𝔅\mathfrak{B}. The operation of addition is needed here to state the usual property of being a derivation. The other reason that addition was mentioned in this paper was to emphasize the links with tropical geometry.

The inclusion ι:𝔱\iota:\mathbb{C}\longrightarrow\mathbb{C}\mathfrak{t}^{\mathbb{R}} defined by ι(c)=c𝔱0\iota(c)=c\mathfrak{t}^{0} induces an inclusion of functions

ι:C(𝔅)(𝔅)\iota:C^{\infty}(\mathfrak{B})\longrightarrow\mathcal{E}(\mathfrak{B})
ι(f)(p):=ι(f(p))\iota(f)(p):=\iota(f(p))
Definition 2.12.

Using the notation of example 2.7, a CkC^{k} function on n×𝔗Am\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}, fCk(n×𝔗Am)f\in C^{k}(\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}) is a function on [n×𝔗Am][\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}] which is equal to a CkC^{k} function applied to the variables xix_{i} and zαiz^{\alpha^{i}}. A CkC^{k} exploded function fk×(n×𝔗Am)f\in\mathcal{E}^{k\times}(\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}) is a function of the form fz~α𝔱yf\tilde{z}^{\alpha}\mathfrak{t}^{y} where ff is a CkC^{k} function taking values in \mathbb{C}^{*}. We can define CkC^{k} morphisms of exploded fibrations to be morphisms of abstract exploded fibrations using the sheaf k×\mathcal{E}^{k\times} instead of ×\mathcal{E}^{\times}

We can consider an arbitrary Hausdorff topological space to be a C0C^{0} exploded fibration. A continuous morphism of 𝔅\mathfrak{B} to a Hausdorff topological space XX is given by a map f:𝔅[𝔅]Xf:\mathfrak{B}\longrightarrow[\mathfrak{B}]\longrightarrow X so that the pullback of continuous real valued functions on XX are C0C^{0} functions on 𝔅\mathfrak{B}. Note that in general, the map []:𝔅[𝔅][\cdot]:\mathfrak{B}\longrightarrow[\mathfrak{B}] is not a continuous morphism in this sense.

Definition 2.13.

The topological or smooth part of 𝔅\mathfrak{B}, 𝔅\lceil\mathfrak{B}\rceil is the unique Hausdorff topological space 𝔅\lceil\mathfrak{B}\rceil with the following universal property: There is a C0C^{0} morphism

𝔅𝔅\mathfrak{B}\xrightarrow{\lceil\cdot\rceil}\lceil\mathfrak{B}\rceil

so that any other C0C^{0} morphism from 𝔅\mathfrak{B} to a Hausdorff topological space factors uniquely through \lceil\cdot\rceil:

Given f:𝔅X\text{Given }f:\mathfrak{B}\longrightarrow X
there exists a unique h:𝔅X\text{there exists a unique }h:\lceil\mathfrak{B}\rceil\longrightarrow X
so that f=h\text{so that }\ \ \ \ \ f=h\circ\lceil\cdot\rceil

We can construct 𝔅\lceil\mathfrak{B}\rceil as the image of 𝔅\mathfrak{B} in C0(𝔅)\mathbb{R}^{C^{0}(\mathfrak{B})}. The universal property of \lceil\cdot\rceil makes it clear that taking the topological part of 𝔅\mathfrak{B} is a functor from the exploded category to the category of Hausdorff topological spaces.

Definition 2.14.

We shall use the adverb ‘topologically’ to indicate a property of 𝔅\lceil\mathfrak{B}\rceil. In particular,

  1. 1.

    A sequence of points pi𝔅p_{i}\longrightarrow\mathfrak{B} converges topologically to p𝔅p\longrightarrow\mathfrak{B} if their images converge in 𝔅\lceil\mathfrak{B}\rceil. In this case we use the notation pip\lceil p_{i}\rceil\rightarrow\lceil p\rceil or say these points converge in 𝔅\lceil\mathfrak{B}\rceil. Note that this is equivalent to f(pi)f(p_{i}) converging to f(p)f(p) for every continuous \mathbb{R} valued function fC(𝔅)f\in C(\mathfrak{B}).

  2. 2.

    The exploded fibration 𝔅\mathfrak{B} is topologically compact if 𝔅\lceil\mathfrak{B}\rceil is compact.

  3. 3.

    A map f:𝔅f:\mathfrak{B}\longrightarrow\mathfrak{C} is topologically proper if the induced map f:𝔅\lceil f\rceil:\lceil\mathfrak{B}\rceil\longrightarrow\lceil\mathfrak{C}\rceil is proper.

Note that limits in the above sense are of course not unique. A manifold is topologically compact when considered as an exploded fibration if and only if it is compact. Another example is any connected open subset of 𝔗n\mathfrak{T}^{n}. Sometimes the topological part of 𝔅\mathfrak{B} gives slightly misleading information about 𝔅\mathfrak{B}, as the following example is intended to demonstrate. We shall give a more useful analogue of ‘compactness’ in definition 2.19 shortly.

Example 2.15.

Our local model n×𝔗Am\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A} has a smooth part n×𝔗Am\lceil\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}\rceil which is equal to the product of n\mathbb{R}^{n} with the toric variety corresponding to the integral affine cone AA. The following two examples have ‘misbehaved’ smooth parts: Consider the exploded fibration formed by quotienting ×𝔗\mathbb{R}\times\mathfrak{T} by (x,z~)=(x+1,z~)(x,\tilde{z})=(x+1,\tilde{z}) and (x,z~)=(x+π,𝔱1z~)(x,\tilde{z})=(x+\pi,\mathfrak{t}^{1}\tilde{z}). The smooth part of this is a single point, even though it is made up of pieces which have smooth parts equal to a circle. Another example is given by the subset of ×𝔗\mathbb{R}\times\mathfrak{T} which is the union of the set where z~<𝔱0\lfloor\tilde{z}\rfloor<\mathfrak{t}^{0} with the set where z~>𝔱1\lfloor\tilde{z}\rfloor>\mathfrak{t}^{1} and x>0x>0. (Note that these sets overlap because 𝔱0>𝔱1\mathfrak{t}^{0}>\mathfrak{t}^{1}.) The smooth part of this is equal to \mathbb{R}.

For many of our proofs, we shall restrict to a class of exploded fibrations in which the smooth part and the tropical part of the exploded fibration give better information than the above two examples. To explain this class of ‘basic’ exploded fibrations, we shall first construct some elementary examples.

Example 2.16.

Suppose that P𝔱mP\subset\mathfrak{t}^{\mathbb{R}^{m}} is a convex integral affine polygon with nonempty interior defined by some set of inequalities:

P:={𝔱a𝔱m so that 𝔱ci+aαi𝔱0,𝔱ci+aαi<𝔱0}P:=\{\mathfrak{t}^{a}\in\mathfrak{t}^{\mathbb{R}^{m}}\text{ so that }\mathfrak{t}^{c_{i}+a\cdot\alpha^{i}}\leq\mathfrak{t}^{0},\mathfrak{t}^{c_{i^{\prime}}+a\cdot\alpha^{i^{\prime}}}<\mathfrak{t}^{0}\}

where αim\alpha^{i}\in\mathbb{Z}^{m} and cic_{i}\in\mathbb{R}. This can simply be viewed as a convex polyhedron in m\mathbb{R}^{m} which has faces with rational slopes. View PP as a stratified space in the standard way, (so the strata of PP consist of the vertices of PP, the edges of PP minus the vertices of PP, the two dimensional faces of PP minus all vertices and edges, and so on, up to the interior of PP). Note that as some of the defining inequalities may be strict, PP may be missing some faces or lower dimensional strata.

Then there is an exploded fibration 𝔗Pm\mathfrak{T}^{m}_{P} associated with P𝔱mP\subset\mathfrak{t}^{\mathbb{R}^{m}}, which is constructed by gluing together coordinate charts modeled on 𝔗Akm\mathfrak{T}^{m}_{A_{k}} where AkA_{k} are the integral affine cones corresponding to the strata of PP.

Explicitly, suppose that the point 𝔱ak𝔱m\mathfrak{t}^{a^{k}}\in\mathfrak{t}^{\mathbb{R}^{m}} is inside the kkth strata of PP, and AkA_{k} is the integral affine cone so that locally near 𝔱ak\mathfrak{t}^{a^{k}}, PP is equal to AkA_{k} shifted by 𝔱ak\mathfrak{t}^{a^{k}}.

near 𝔱ak,P={𝔱a𝔱m so that 𝔱aakAk}\text{near }\mathfrak{t}^{a^{k}},\ \ \ \ P=\{\mathfrak{t}^{a}\in\mathfrak{t}^{\mathbb{R}^{m}}\text{ so that }\mathfrak{t}^{a-a^{k}}\in A_{k}\}

There are coordinates z~1,,z~m×(𝔗Pm)\tilde{z}_{1},\dotsc,\tilde{z}_{m}\in\mathcal{E}^{\times}\left(\mathfrak{T}^{m}_{P}\right) so that the set of points in 𝔗Pm\mathfrak{T}^{m}_{P} are given by specifying values (z~1(p),,z~m(p))(𝔱)m(\tilde{z}_{1}(p),\dotsc,\tilde{z}_{m}(p))\in\left(\mathbb{C}^{*}\mathfrak{t}^{\mathbb{R}}\right)^{m} so that

(z~1(p),,z~m(p))P(\lfloor\tilde{z}_{1}(p)\rfloor,\dotsc,\lfloor\tilde{z}_{m}(p)\rfloor)\in P

Denote by PkPP_{k}\subset P the subset of PP which is the union of all strata of PP whose closure contains the kkth strata of PP. There is a corresponding subset 𝔗Pkm\mathfrak{T}^{m}_{P_{k}} of 𝔗Pm\mathfrak{T}^{m}_{P}:

𝔗Pkm:={z~ so that z~Pk}\mathfrak{T}^{m}_{P_{k}}:=\left\{\tilde{z}\text{ so that }\lfloor\tilde{z}\rfloor\in P_{k}\right\}

We put an exploded structure on 𝔗Pkm\mathfrak{T}^{m}_{P_{k}} so it is isomorphic to the corresponding subset of 𝔗Akm\mathfrak{T}^{m}_{A_{k}} with the standard coordinates given by (𝔱a1kz~1,,𝔱amkz~m)(\mathfrak{t}^{-a^{k}_{1}}\tilde{z}_{1},\dotsc,\mathfrak{t}^{-a^{k}_{m}}\tilde{z}_{m}). It is an easy exercise to check that the restriction of the exploded structure this gives to 𝔗Pkm𝔗Pkm\mathfrak{T}^{m}_{P_{k}}\cap\mathfrak{T}^{m}_{P_{k^{\prime}}} is compatible with the restriction of the one coming from 𝔗Pkm\mathfrak{T}^{m}_{P_{k^{\prime}}}, so this gives a globally well defined exploded structure.

An alternative way to describe this structure is as follows: Consider the collection of functions w~:=𝔱cz~α\tilde{w}:=\mathfrak{t}^{c}\tilde{z}^{\alpha} so that on PP, w~𝔱0\lfloor\tilde{w}\rfloor\leq\mathfrak{t}^{0}. Choose some finite generating set {w~1,,w~n}\{\tilde{w}_{1},\dotsc,\tilde{w}_{n}\} of these functions so that any other function of this type w~\tilde{w} can be written as w~=𝔱cw~1β1w~nβn\tilde{w}=\mathfrak{t}^{c}\tilde{w}_{1}^{\beta_{1}}\dotsb\tilde{w}_{n}^{\beta_{n}} where βi\beta_{i}\in\mathbb{N} and c[0,)c\in[0,\infty). Exploded functions on 𝔗Pm\mathfrak{T}^{m}_{P} can then be described as functions of the form f(w~1,,w~n)𝔱cz~αf(\lceil\tilde{w}_{1}\rceil,\dotsc,\lceil\tilde{w}_{n}\rceil)\mathfrak{t}^{c}\tilde{z}^{\alpha}, where f:nf:\mathbb{C}^{n}\longrightarrow\mathbb{C}^{*} is smooth. The underlying topological space [𝔗Pm][\mathfrak{T}^{m}_{P}] is equal to the closure of the set

{w~1,,w~n,z~1,,z~m}n×P\{\lceil\tilde{w}_{1}\rceil,\dotsc,\lceil\tilde{w}_{n}\rceil,\lfloor\tilde{z}_{1}\rfloor,\dotsc,\lfloor\tilde{z}_{m}\rfloor\}\subset\mathbb{C}^{n}\times P

The tropical part of 𝔗Pm\mathfrak{T}^{m}_{P} is given by

𝔗Pm=P={z~1,,z~m}𝔱m\lfloor\mathfrak{T}^{m}_{P}\rfloor=P=\{\lfloor\tilde{z}_{1}\rfloor,\dotsc,\lfloor\tilde{z}_{m}\rfloor\}\subset\mathfrak{t}^{\mathbb{R}^{m}}

The smooth part of 𝔗Pm\mathfrak{T}^{m}_{P} is given by

𝔗Pm:={w~1,,w~n}n\lceil\mathfrak{T}^{m}_{P}\rceil:=\{\lceil\tilde{w}_{1}\rceil,\dotsc,\lceil\tilde{w}_{n}\rceil\}\subset\mathbb{C}^{n}

Simple examples of the above construction are 𝔗(a,b)1\mathfrak{T}^{1}_{(a,b)} which is equal to the subset of 𝔗\mathfrak{T} where 𝔱a>z~>𝔱b\mathfrak{t}^{a}>\lfloor\tilde{z}\rfloor>\mathfrak{t}^{b}, and 𝔗[0,b)1\mathfrak{T}^{1}_{[0,b)} which is equal to the subset of 𝔗11\mathfrak{T}^{1}_{1} where z~>𝔱b\lfloor\tilde{z}\rfloor>\mathfrak{t}^{b}. (Strictly speaking, we should be writing 𝔗𝔱(a,b)1\mathfrak{T}^{1}_{\mathfrak{t}^{(a,b)}}, as the above confuses the interval (a,b)(a,b)\subset\mathbb{R} with 𝔱(a,b)𝔱\mathfrak{t}^{(a,b)}\subset\mathfrak{t}^{\mathbb{R}}.)

Similarly, we can construct k×𝔗Pm\mathbb{R}^{k}\times\mathfrak{T}^{m}_{P}. This has coordinates (x,z~)k×𝔗Pm(x,\tilde{z})\in\mathbb{R}^{k}\times\mathfrak{T}^{m}_{P}, has underlying topological space [k×𝔗Pm]=k×[𝔗Pm][\mathbb{R}^{k}\times\mathfrak{T}^{m}_{P}]=\mathbb{R}^{k}\times[\mathfrak{T}^{m}_{P}], so that [(x,z~)]=(x,[z~])[(x,\tilde{z})]=(x,[\tilde{z}]), and has exploded functions equal to functions of the form f(x,w~1,,w~n)𝔱cz~αf(x,\lceil\tilde{w}_{1}\rceil,\dotsc,\lceil\tilde{w}_{n}\rceil)\mathfrak{t}^{c}\tilde{z}^{\alpha}, where f:k×nf:\mathbb{R}^{k}\times\mathbb{C}^{n}\longrightarrow\mathbb{C}^{*} is smooth.

Example 2.17.

Given a smooth manifold MM and mm complex line bundles on MM, we can construct the exploded fibration M𝔗PmM\rtimes\mathfrak{T}^{m}_{P} as follows: Denote by EE the corresponding space of mm \mathbb{C}^{*} bundles over MM. This has a smooth free ()m\left(\mathbb{C}^{*}\right)^{m} action. The exploded fibration 𝔗Pm\mathfrak{T}^{m}_{P} also has a ()m\left(\mathbb{C}^{*}\right)^{m} action given by multiplying the coordinates z~1,,z~m\tilde{z}_{1},\dotsc,\tilde{z}_{m} by the coordinates of ()m\left(\mathbb{C}^{*}\right)^{m}. M𝔗PmM\rtimes\mathfrak{T}^{m}_{P} is the exploded fibration constructed by taking the quotient of E×𝔗PmE\times\mathfrak{T}^{m}_{P} by the action of ()m\left(\mathbb{C}^{*}\right)^{m} by (c,c1)(c,c^{-1}). As the action of ()m\left(\mathbb{C}^{*}\right)^{m} is trivial on 𝔗Pm\lfloor\mathfrak{T}^{m}_{P}\rfloor, the tropical part of M𝔗PmM\rtimes\mathfrak{T}^{m}_{P} is still defined, and is equal to PP.

Alternately, choose coordinate charts on EE equal to U×()mn×nU\times\left(\mathbb{C}^{*}\right)^{m}\subset\mathbb{R}^{n}\times\mathbb{C}^{n}. The transition maps are of the form (u,z1,,zm)(ϕ(u),f1(u)z1,,fm(u)zm)(u,z_{1},\dotsc,z_{m})\mapsto(\phi(u),f_{1}(u)z_{1},\dotsc,f_{m}(u)z_{m}). Replace these coordinate charts with U×𝔗Pmn×𝔗PmU\times\mathfrak{T}^{m}_{P}\subset\mathbb{R}^{n}\times\mathfrak{T}^{m}_{P}, and replace the above transition maps with maps of the form (u,z~1,,z~m)(ϕ(u),f1(u)z~1,,fm(u)z~m)(u,\tilde{z}_{1},\dotsc,\tilde{z}_{m})\mapsto(\phi(u),f_{1}(u)\tilde{z}_{1},\dotsc,f_{m}(u)\tilde{z}_{m}). The map to the tropical part M𝔗Pm:=P\lfloor M\rtimes\mathfrak{T}^{m}_{P}\rfloor:=P in these coordinates is given by

(u,z~1,,z~n)=(z~1,,z~n)P\lfloor(u,\tilde{z}_{1},\dotsc,\tilde{z}_{n})\rfloor=(\lfloor\tilde{z}_{1}\rfloor,\dotsc,\lfloor\tilde{z}_{n}\rfloor)\in P
Definition 2.18.

An exploded fibration 𝔅\mathfrak{B} is basic if

  1. 1.

    There exists a Hausdorff topological space 𝔅\lfloor\mathfrak{B}\rfloor called the tropical part of 𝔅\mathfrak{B}, along with a map

    :{p𝔅}[𝔅]𝔅\lfloor\cdot\rfloor:\{p\rightarrow\mathfrak{B}\}\longrightarrow[\mathfrak{B}]\longrightarrow\lfloor\mathfrak{B}\rfloor
  2. 2.

    The space 𝔅\lfloor\mathfrak{B}\rfloor is a union of strata 𝔅i\lfloor\mathfrak{B}_{i}\rfloor so that

    1. (a)

      Each strata 𝔅i𝔅\lfloor\mathfrak{B}_{i}\rfloor\subset\lfloor\mathfrak{B}\rfloor is an integral affine space equal to some open convex integral affine polygon in 𝔱k\mathfrak{t}^{\mathbb{R}^{k}} of the form considered in example 2.16. (The dimension kk depends on the strata.)

    2. (b)

      The closure of 𝔅i𝔅\lfloor\mathfrak{B}_{i}\rfloor\subset\lfloor\mathfrak{B}\rfloor is a union of strata which is also equal to an integral affine polygon 𝔅i¯\overline{\lfloor\mathfrak{B}_{i}\rfloor} of the form considered in example 2.16 with some strata identified.

  3. 3.

    Use the notation 𝔅i\mathfrak{B}_{i} for inverse image of 𝔅i\lfloor\mathfrak{B}_{i}\rfloor under the map \lfloor\cdot\rfloor, and call this a strata of 𝔅\mathfrak{B}.

    1. (a)

      Each strata 𝔅i\mathfrak{B}_{i} has a smooth part 𝔅i\lceil\mathfrak{B}_{i}\rceil which is a smooth manifold.

    2. (b)

      Each strata 𝔅i\mathfrak{B}_{i} is equal to an exploded space 𝔅i𝔗𝔅ik\lceil\mathfrak{B}_{i}\rceil\rtimes\mathfrak{T}^{k}_{\lfloor\mathfrak{B}_{i}\rfloor} using the construction of example 2.17. The map \lfloor\cdot\rfloor restricted to 𝔅i\mathfrak{B}_{i} is the map :𝔅i𝔗𝔅ik𝔅i\lfloor\cdot\rfloor:\lceil\mathfrak{B}_{i}\rceil\rtimes\mathfrak{T}^{k}_{\lfloor\mathfrak{B}_{i}\rfloor}\longrightarrow\lfloor\mathfrak{B}_{i}\rfloor from example 2.17.

    3. (c)

      There is a neighborhood of each strata 𝔅i\mathfrak{B}_{i} which is open in 𝔅\lceil\mathfrak{B}\rceil which is isomorphic to some subset of 𝔅i𝔗𝔅i¯k\lceil\mathfrak{B}_{i}\rceil\rtimes\mathfrak{T}^{k}_{\overline{\lfloor\mathfrak{B}_{i}\rfloor}} which is a topologically open neighborhood of 𝔅i𝔗𝔅ik𝔅i𝔗𝔅i¯k\lceil\mathfrak{B}_{i}\rceil\rtimes\mathfrak{T}^{k}_{\lfloor\mathfrak{B}_{i}\rfloor}\subset{\lceil\mathfrak{B}_{i}\rceil\rtimes\mathfrak{T}^{k}_{\overline{\lfloor\mathfrak{B}_{i}\rfloor}}}.

It is an easy exercise to check that if 𝔅\mathfrak{B} is basic, the smooth part 𝔅\lceil\mathfrak{B}\rceil is also a stratified space, with strata equal to the manifolds 𝔅i\lceil\mathfrak{B}_{i}\rceil. The stratification of 𝔅\lceil\mathfrak{B}\rceil is in some sense dual to the stratification of 𝔅\lfloor\mathfrak{B}\rfloor in that 𝔅i\lfloor\mathfrak{B}_{i}\rfloor is in the closure of 𝔅j\lfloor\mathfrak{B}_{j}\rfloor if and only if 𝔅j\lceil\mathfrak{B}_{j}\rceil is in the closure of 𝔅i\lceil\mathfrak{B}_{i}\rceil.

All our examples up until now apart from those introduced in example 2.15 have been basic. Another example of an exploded fibration that is not basic is given by quotienting ×𝔗\mathbb{R}\times\mathfrak{T} by (x,z~)=(x+1,z~1)(x,\tilde{z})=(x+1,\tilde{z}^{-1}). A less pathological example of an exploded fibration that is not basic is given by the quotient of 𝔗\mathfrak{T} by z~=𝔱1z~\tilde{z}=\mathfrak{t}^{1}\tilde{z}. Any exploded fibration is locally basic.

To summarize the topological spaces we have involved in the case where 𝔅\mathfrak{B} is basic, we have the following topological spaces.

{p𝔅}[𝔅]𝔅𝔅\begin{split}\{p\rightarrow\mathfrak{B}\}\longrightarrow&[\mathfrak{B}]\longrightarrow\lceil\mathfrak{B}\rceil\\ &\downarrow\\ &\lfloor\mathfrak{B}\rfloor\end{split}
  1. 1.

    The smooth or topological part 𝔅\lceil\mathfrak{B}\rceil. Any continuos morphism of 𝔅\mathfrak{B} to a Hausdorff topological space factors through 𝔅\lceil\mathfrak{B}\rceil. The smooth part of a smooth manifold MM is MM. The smooth part of 𝔗n\mathfrak{T}^{n} is a single point.

  2. 2.

    The tropical part 𝔅\lfloor\mathfrak{B}\rfloor. This is a stratified integral affine space, which can be thought of as what 𝔅\mathfrak{B} looks like on the large scale. The tropical part of a connected smooth manifold is a single point. The tropical part of 𝔗n\mathfrak{T}^{n} is 𝔱n\mathfrak{t}^{\mathbb{R}^{n}}.

  3. 3.

    The underlying topological space [𝔅][\mathfrak{B}]. This strange topological space is a mixture of 𝔅\lceil\mathfrak{B}\rceil and 𝔅\lfloor\mathfrak{B}\rfloor. It is the topological space on which ×\mathcal{E}^{\times} is a sheaf, and this is the topology in which 𝔅\mathfrak{B} is locally modeled on n×𝔗Am\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}.

  4. 4.

    In the case that 𝔅\mathfrak{B} is not a manifold, the set of points {p𝔅}\{p\rightarrow\mathfrak{B}\} is not Hausdorff in any of the above topologies. We can put a topology on this set by choosing the strongest topology in which all continuous morphisms from manifolds into 𝔅\mathfrak{B} are continuous maps to {p𝔅}{\{p\rightarrow\mathfrak{B}\}}. For example {p𝔗11}\{p\rightarrow\mathfrak{T}^{1}_{1}\} is homeomorphic to the disjoint union of [0,)[0,\infty) copies of \mathbb{C}^{*}. (When we have defined tangent spaces and metrics on exploded fibrations, this would be the topology coming from any metric.)

The following is a picture of an ‘exploded curve’ built up using coordinate charts modeled on open subsets of 𝔗11\mathfrak{T}^{1}_{1}. Note that \lceil\mathfrak{C}\rceil is what we would see if we mapped \mathfrak{C} to a smooth manifold. On the other hand, we would see \lfloor\mathfrak{C}\rfloor if we mapped \mathfrak{C} to 𝔗n\mathfrak{T}^{n}.

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\includegraphics{ETF3}
Definition 2.19.

An exploded 𝕋\mathbb{T} morphism f:𝔅f:\mathfrak{B}\longrightarrow\mathfrak{C} is complete if the following conditions hold

  1. 1.

    ff is topologically proper.

  2. 2.

    Any map

    g:𝔗(0,l)1𝔅g:\mathfrak{T}^{1}_{(0,l)}\longrightarrow\mathfrak{B}

    extends to a continuous map 𝔗[0,l)1𝔅\mathfrak{T}^{1}_{[0,l)}\longrightarrow\mathfrak{B} if and only if the map fgf\circ g extends to a continuous map 𝔗[0,l)1\mathfrak{T}^{1}_{[0,l)}\longrightarrow\mathfrak{C}.

We say that an exploded fibration 𝔅\mathfrak{B} is complete if the map from 𝔅\mathfrak{B} to a point is complete.

The notation 𝔗(0,l)1\mathfrak{T}^{1}_{(0,l)} is that introduced in example 2.16. An equivalent way of stating the second condition is that the map [g]:(0,l)[𝔅][g]:(0,l)\longrightarrow[\mathfrak{B}] extends to a continuous map [0,l)[𝔅][0,l)\longrightarrow[\mathfrak{B}] if and only if [fg]:(0,l)[][f\circ g]:(0,l)\longrightarrow[\mathfrak{C}] extends to a continuous map [0,1)[][0,1)\longrightarrow[\mathfrak{C}]

Complete maps should be thought of as being analogous to proper maps. A smooth manifold is complete if and only if it is compact. An example of a non-compact exploded fibration which is complete is 𝔗n\mathfrak{T}^{n}. In the case that 𝔅\mathfrak{B} is basic, completeness means that the topological part 𝔅\lceil\mathfrak{B}\rceil is compact, and the tropical part 𝔅\lfloor\mathfrak{B}\rfloor is a complete stratified integral affine space.

An interesting example of an exploded fibration which is topologically compact but not complete is given by the following: Let M×𝔗M\times\mathfrak{T} be the product of a compact manifold with 𝔗\mathfrak{T}. This is complete, but we can construct an interesting non complete example as follows: choose an open subset UMU\subset M. Consider the subset of M×𝔗M\times\mathfrak{T} given by the union of U×𝔗(,1)1U\times\mathfrak{T}^{1}_{(-\infty,1)} with M×𝔗(0,)1M\times\mathfrak{T}^{1}_{(0,\infty)}. The smooth part of this is the compact manifold MM, but it is not complete.

An example of an interesting exploded fibration which is complete but not basic is given by the quotient of ×𝔗\mathbb{R}\times\mathfrak{T} by (x,z~)=(x+1,𝔱1z~)(x,\tilde{z})=(x+1,\mathfrak{t}^{1}\tilde{z}).

The following lemma is easy to prove:

Lemma 2.20.
  1. 1.

    if ff and gg are complete, then fgf\circ g is complete.

  2. 2.

    if fgf\circ g is complete, then gg is complete.

2.3 Tangent space

Definition 2.21.

A smooth exploded vectorfield vv on 𝔅\mathfrak{B} is determined by maps

v:C(𝔅)C(𝔅)v:C^{\infty}(\mathfrak{B})\longrightarrow C^{\infty}(\mathfrak{B})
v:(𝔅)(𝔅)v:\mathcal{E}(\mathfrak{B})\longrightarrow\mathcal{E}(\mathfrak{B})

so that

  1. 1.
    v(f+g)=v(f)+v(g)v(f+g)=v(f)+v(g)
  2. 2.
    v(fg)=v(f)g+fv(g)v(fg)=v(f)g+fv(g)
  3. 3.
    v(c𝔱yf)=c𝔱yv(f)v(c\mathfrak{t}^{y}f)=c\mathfrak{t}^{y}v(f)
  4. 4.

    The action of vv is compatible with the inclusion ι:C\iota:C^{\infty}\longrightarrow\mathcal{E} in the sense that

    v(ιf)=ιv(f)v(\iota f)=\iota v(f)

Smooth exploded vectorfields form a sheaf. The action of the restriction of vv to UU on the restriction of ff to UU is the restriction to UU of the action of vv on ff.

We can restrict a vector field to a point p𝔅p\longrightarrow\mathfrak{B} to obtain a tangent vector at that point. This is determined by maps

v:C(𝔅)v:C^{\infty}(\mathfrak{B})\longrightarrow\mathbb{R}
v:(𝔅)𝔱v:\mathcal{E}(\mathfrak{B})\longrightarrow\mathbb{C}\mathfrak{t}^{\mathbb{R}}

satisfying the above conditions with condition 2 replaced by

v(fg)=v(f)g(p)+f(p)v(g)v(fg)=v(f)g(p)+f(p)v(g)

Denote by Tp𝔅T_{p}\mathfrak{B} the vector space of tangent vectors over p𝔅p\longrightarrow\mathfrak{B}.

We can add smooth exploded vectorfields and multiply them by smooth real valued functions. We shall show below that the sheaf of smooth exploded vectorfields is equal to the sheaf of smooth sections of T𝔅T\mathfrak{B}, which is a real vector bundle over 𝔅\mathfrak{B}.

Lemma 2.22.

Differentiation does not change the order of a function in the sense that given any smooth exploded vectorfield vv and exploded function ff\in\mathcal{E},

f=vf\lfloor f\rfloor=\lfloor vf\rfloor

Proof:

v𝔱0=v(ι1)=0𝔱0v\mathfrak{t}^{0}=v(\iota 1)=0\mathfrak{t}^{0}

Now we can apply vv to the equation f=1𝔱0ff=1\mathfrak{t}^{0}f, so vf=0𝔱0f+1𝔱0vfvf=0\mathfrak{t}^{0}f+1\mathfrak{t}^{0}vf. Taking \lfloor\cdot\rfloor of this equation gives

vf=f+vf i.e. vff\lfloor vf\rfloor=\lfloor f\rfloor+\lfloor vf\rfloor\text{ i.e. }\lfloor vf\rfloor\geq\lfloor f\rfloor

(Recall that we use the order 𝔱x<𝔱y\mathfrak{t}^{x}<\mathfrak{t}^{y} if x>yx>y as we are thinking of 𝔱\mathfrak{t} as being tiny. So 𝔱x+𝔱y=𝔱x\mathfrak{t}^{x}+\mathfrak{t}^{y}=\mathfrak{t}^{x} means that 𝔱x𝔱y\mathfrak{t}^{x}\geq\mathfrak{t}^{y}.) Now,

v𝔱a=v(𝔱a𝔱0)=𝔱a0𝔱0=0𝔱av\mathfrak{t}^{a}=v(\mathfrak{t}^{a}\mathfrak{t}^{0})=\mathfrak{t}^{a}0\mathfrak{t}^{0}=0\mathfrak{t}^{a}

Now suppose that at some point vf>f\lfloor vf\rfloor>\lfloor f\rfloor. Then we can restrict to a small neighborhood of this point, and choose an aa so that vf>𝔱a>f\lfloor vf\rfloor>\mathfrak{t}^{a}>\lfloor f\rfloor. But then f+𝔱a=𝔱af+\mathfrak{t}^{a}=\mathfrak{t}^{a}, but v(f+𝔱a)=vf+𝔱a=vf\lfloor v(f+\mathfrak{t}^{a})\rfloor=\lfloor vf\rfloor+\lfloor\mathfrak{t}^{a}\rfloor=\lfloor vf\rfloor, which is a contradiction, and the lemma is proved.

\square

Lemma 2.23.

For any smooth exploded fibration 𝔅\mathfrak{B}, there exists a smooth exploded fibration T𝔅T\mathfrak{B}, the tangent space of 𝔅\mathfrak{B}, along with a canonical smooth projection π:T𝔅𝔅\pi:T\mathfrak{B}\longrightarrow\mathfrak{B} that makes T𝔅T\mathfrak{B} into a real vector bundle over 𝔅\mathfrak{B}. The sheaf of smooth exploded vectorfields is equal to the sheaf of smooth sections of this vector bundle.

In particular,

T(n×𝔗Am)=2n+2m×𝔗AmT(\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A})=\mathbb{R}^{2n+2m}\times\mathfrak{T}^{m}_{A}

Proof:

We shall first prove that T(n×𝔗Am)=2n+2m×𝔗AmT(\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A})=\mathbb{R}^{2n+2m}\times\mathfrak{T}^{m}_{A}. We shall use coordinate functions xix_{i} for n\mathbb{R}^{n} and z~i\tilde{z}_{i} for 𝔗Am\mathfrak{T}^{m}_{A}, as in example 2.7 on page 2.7. A section of 2n+2m×𝔗Amn×𝔗Am\mathbb{R}^{2n+2m}\times\mathfrak{T}^{m}_{A}\longrightarrow\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A} is given by n+2mn+2m smooth functions on n×𝔗Am\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}. To a vectorfield vv, associate the nn smooth functions v(xi)v(x_{i}), and the mm smooth complex valued functions given by v(z~i)z~i1\lceil v(\tilde{z}_{i})\tilde{z}_{i}^{-1}\rceil. (Lemma 2.22 tells us that z~i1v(z~i)=𝔱0\lfloor\tilde{z}_{i}^{-1}v(\tilde{z}_{i})\rfloor=\mathfrak{t}^{0}, so this makes sense.) This gives us n+2mn+2m smooth real valued functions, and therefore gives us a section to associate with our vectorfield.

Now we must show that the arbitrary choice of n+2mn+2m smooth functions on 𝔗Am\mathfrak{T}^{m}_{A} to equal v(xi)v(x_{i}) and v(z~i)z~i1\lceil v(\tilde{z}_{i})\tilde{z}_{i}^{-1}\rceil will uniquely determine a smooth exploded vectorfield.

First, recall that exploded functions are a sum of functions of the form f𝔱yz~αf\mathfrak{t}^{y}\tilde{z}^{\alpha}, where ff is some smooth function of xx and zαjz^{\alpha^{j}}. Use the notation

zαj:=etαj+iθαjz^{\alpha^{j}}:=e^{t_{\alpha^{j}}+i\theta_{\alpha^{j}}}

There is the following formula for v(f)v(f) which follows from the axioms:

v(f)=v(xi)fxi+i,j(v(z~i)z~i1)αijftαj+(v(z~i)z~i1)αijfθαjv(f)=\sum v(x_{i})\frac{\partial f}{\partial x_{i}}+\sum_{i,j}\Re\left(v(\tilde{z}_{i})\tilde{z}_{i}^{-1}\right)\alpha_{i}^{j}\frac{\partial f}{\partial t_{\alpha^{j}}}+\Im\left(v(\tilde{z}_{i})\tilde{z}_{i}^{-1}\right)\alpha_{i}^{j}\frac{\partial f}{\partial\theta_{\alpha^{j}}}

Note that this is well defined, despite the fact that ftαj\frac{\partial f}{\partial t_{\alpha^{j}}} and fθαj\frac{\partial f}{\partial\theta_{\alpha^{j}}} are not necessarily well defined. Note also that this is a smooth function, and is real valued if ff is real valued. The corresponding formula for an exploded function is

v(αfα𝔱yαz~α):=α(v(fα)+fαiαiv(z~i)z~i1)𝔱yαz~αv\left(\sum_{\alpha}f_{\alpha}\mathfrak{t}^{y_{\alpha}}\tilde{z}^{\alpha}\right):=\sum_{\alpha}\left(v(f_{\alpha})+f_{\alpha}\sum_{i}\alpha_{i}v(\tilde{z}_{i})\tilde{z}_{i}^{-1}\right)\mathfrak{t}^{y_{\alpha}}\tilde{z}^{\alpha}

It can be shown that vv satisfying such a formula satisfies all the axioms for being a smooth exploded vectorfield, is well defined, and is zero if and only if v(xi)=0v(x_{i})=0 and v(z~i)z~i1=0\lceil v(\tilde{z}_{i})\tilde{z}_{i}^{-1}\rceil=0.

This shows that Tn×𝔗Am=2n+2m×𝔗AmT\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}=\mathbb{R}^{2n+2m}\times\mathfrak{T}^{m}_{A}. The lemma then follows from our coordinate free definition of a smooth vectorfield, and the fact that every smooth exploded 𝕋\mathbb{T} fibration is locally modeled on exploded fibrations of this type.

\square

For any smooth exploded 𝕋\mathbb{T} fibration 𝔅\mathfrak{B}, we now have that T𝔅T\mathfrak{B} is a real vector bundle. In local coordinates {xi,z~j}\{x_{i},\tilde{z}_{j}\}, a basis for this vectorbundle is given by {xi}\{\frac{\partial}{\partial x_{i}}\} and the real and imaginary parts of {z~iz~i}\{\tilde{z}_{i}\frac{\partial}{\partial\tilde{z}_{i}}\}. The dual of this bundle T𝔅T^{*}\mathfrak{B} is the cotangent space. A basis for this is locally given by {dxi}\{dx_{i}\} and the real and imaginary parts of {z~i1dz~i}\{\tilde{z}_{i}^{-1}d\tilde{z}_{i}\}. We can take tensor powers (over smooth real valued functions) of these vector bundles, to define the usual objects found on smooth manifolds. For example, a metric on 𝔅\mathfrak{B} is a smooth, symmetric, positive definite section of T𝔅T𝔅T^{*}\mathfrak{B}\otimes T^{*}\mathfrak{B}.

Definition 2.24.

An integral vector vv at a point p𝔅p\longrightarrow\mathfrak{B} is a vector vTp𝔅v\in T_{p}\mathfrak{B} so that for any exploded function f×(𝔅)f\in\mathcal{E}^{\times}(\mathfrak{B}),

v(f)f1v(f)f^{-1}\in\mathbb{Z}

Use the notation Tp𝔅Tp𝔅{}^{\mathbb{Z}}T_{p}\mathfrak{B}\subset T_{p}\mathfrak{B} to denote the integral vectors at p𝔅p\longrightarrow\mathfrak{B}.

For example, a basis for T𝔗n{}^{\mathbb{Z}}T\mathfrak{T}^{n} is given by the real parts of z~iz~i\tilde{z}_{i}\frac{\partial}{\partial\tilde{z}_{i}}. The only integral vector on a smooth manifold is the zero vector.

Given a smooth morphism f:𝔅f:\mathfrak{B}\longrightarrow\mathfrak{C}, there is a natural smooth morphism df:T𝔅Tdf:T\mathfrak{B}\longrightarrow T\mathfrak{C} which is the differential of ff, defined as usual by

df(v)g:=v(gf)df(v)g:=v(g\circ f)

Of course, dfdf takes integral vectors to integral vectors.

2.4 Ck,δC^{k,\delta} regularity

In this section, we define the topology in which we shall prove our compactness theorem for holomorphic curves. The regularity C,δC^{\infty,\delta} is the level of regularity that the moduli stack of holomorphic curves can be expected to exhibit. A C,δC^{\infty,\delta} function is a generalization of a function on a manifold with a cylindrical end which is smooth on the interior, and which decays exponentially along with all its derivatives on the cylindrical end. The reader wishing a simple introduction to holomorphic curves in exploded 𝕋\mathbb{T} fibrations may skip this somewhat technical section on first reading.

The following is a definition of a strata of an integral affine cone, similar to the definition of strata given in definition 2.18 on page 2.18.

Definition 2.25.

The closure of a strata of an integral cone A:={𝔱a;aαi0}A:=\{\mathfrak{t}^{a}\ ;\ a\cdot\alpha^{i}\geq 0\} is a subset of AA defined by the vanishing of aαa\cdot\alpha for some αA\alpha\in A^{*}. A strata SS of AA is a subset which is equal to the closure of a strata minus the closure of all strata of smaller dimension. The zero substrata is the substrata on which aα=0a\cdot\alpha=0 for all αA\alpha\in A^{*}.

Definition 2.26.

Given any C0C^{0} function ff defined on a strata SAS\subset A in n×𝔗Am\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}, define

eS(f)(x,z~):=f(x,z~𝔱a)e_{S}(f)(x,\tilde{z}):=f(x,\tilde{z}\mathfrak{t}^{a})

where 𝔱a:=(𝔱a1,,𝔱am)\mathfrak{t}^{a}:=(\mathfrak{t}^{a_{1}},\dotsc,\mathfrak{t}^{a_{m}}) is any point in SS, and z~𝔱a\tilde{z}\mathfrak{t}^{a} means (z~1𝔱a1,,z~m𝔱am)(\tilde{z}_{1}\mathfrak{t}^{a_{1}},\dotsc,\tilde{z}_{m}\mathfrak{t}^{a_{m}}).

For example, 𝔗22:=𝔗[0,)22\mathfrak{T}^{2}_{2}:=\mathfrak{T}^{2}_{[0,\infty)^{2}} has two one dimensional strata

S1:={z~2=𝔱0,z~1𝔱0}S2:={z~1=𝔱0,z~2𝔱0}S_{1}:=\{\lfloor\tilde{z}_{2}\rfloor=\mathfrak{t}^{0},\lfloor\tilde{z}_{1}\rfloor\neq\mathfrak{t}^{0}\}\ \ \ \ \ S_{2}:=\left\{\lfloor\tilde{z}_{1}\rfloor=\mathfrak{t}^{0},\lfloor\tilde{z}_{2}\rfloor\neq\mathfrak{t}^{0}\right\}

If we have a function fC0(𝔗22)f\in C^{0}(\mathfrak{T}^{2}_{2}), then

eS1f(z1,z2)=f(0,z2)eS2f(z1,z2)=f(z1,0)e_{S_{1}}f(z_{1},z_{2})=f(0,z_{2})\ \ \ \ \ \ \ e_{S_{2}}f(z_{1},z_{2})=f(z_{1},0)

Note that the operations eSie_{S_{i}} commute and eSieSi=eSie_{S_{i}}e_{S_{i}}=e_{S_{i}}.

Definition 2.27.

If II denotes any collection of strata {S1,,Sn}\{S_{1},\dotsc,S_{n}\}, we shall use the notation

eIf:=eS1(eS2(eSnf))e_{I}f:=e_{S_{1}}\left(e_{S_{2}}\left(\dotsb e_{S_{n}}f\right)\right)
ΔIf:=(SiI(ideSi))f\Delta_{I}f:=\left(\prod_{S_{i}\in I}(\operatorname{id}-e_{S_{i}})\right)f

For example,

ΔS1,S2f(z1,z2):=(1eS1)(1eS2)(f)(z1,z2):=f(z1,z2)f(0,z2)f(z1,0)+f(0,0)\begin{split}\Delta_{S_{1},S_{2}}f(z_{1},z_{2})&:=(1-e_{S_{1}})(1-e_{S_{2}})(f)(z_{1},z_{2})\\ &:=f(z_{1},z_{2})-f(0,z_{2})-f(z_{1},0)+f(0,0)\end{split}

Note that if SIS\in I, eSΔI=0e_{S}\Delta_{I}=0. In the above example, this corresponds to ΔS1,S2f(z1,0)=0\Delta_{S_{1},S_{2}}f(z_{1},0)=0 and ΔS1,S2f(0,z2)=0\Delta_{S_{1},S_{2}}f(0,z_{2})=0.

We shall need a weight function wIw_{I} for every collection of nonzero strata II. This will have the property that if ff is any smooth function wI1ΔIfw_{I}^{-1}\Delta_{I}f will be bounded on any topologically compact subset of n×𝔗Am\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A} (in other words, a subset with a compact image in the topological part n×𝔗Am\lceil\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}\rceil). Consider the ideal of functions of the form z~α\tilde{z}^{\alpha} so that ΔIzα=zα\Delta_{I}\lceil z^{\alpha}\rceil=\lceil z^{\alpha}\rceil (In other words, eSzα=0e_{S}\lceil z^{\alpha}\rceil=0 for all SIS\in I.) Choose some finite set of generators {z~αi}\{\tilde{z}^{\alpha^{i}}\} for this ideal. Then define

wI:=|z~αi|w_{I}:=\sum\left\lvert\lceil\tilde{z}^{\alpha^{i}}\rceil\right\rvert

Continuing the example started above, we can choose wS1=|z1|w_{S_{1}}=\left\lvert z_{1}\right\rvert, wS2=|z2|w_{S_{2}}=\left\lvert z_{2}\right\rvert, wS1,S2=|z1z2|w_{S_{1},S_{2}}=\left\lvert z_{1}z_{2}\right\rvert, and wS1S2=|z1|+|z2|w_{S_{1}\cap S_{2}}=\left\lvert z_{1}\right\rvert+\left\lvert z_{2}\right\rvert.

Note that given any other choice of generators, the resulting wIw^{\prime}_{I} is bounded by a constant times wIw_{I} on any topologically compact subset of n×𝔗Am\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}. Note also that wIwIw_{I}w_{I^{\prime}} is bounded by a constant times wIIw_{I\cup I^{\prime}} on any topologically compact subset of n×𝔗Am\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}.

We shall now define Ck,δC^{k,\delta} for any 0<δ<10<\delta<1:

Definition 2.28.

Define C0,δC^{0,\delta} to be the same as C0C^{0}. A sequence of smooth functions fiC(n×𝔗Am)f_{i}\in C^{\infty}(\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}) converge to a continuous function ff in Ck,δ(n×𝔗Am)C^{k,\delta}(\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}) if the following conditions hold:

  1. 1.

    Given any collection of at most kk nonzero strata II,

    |wIδΔI(fif)|\left\lvert w_{I}^{-\delta}\Delta_{I}(f_{i}-f)\right\rvert

    converges to 0 uniformly on compact subsets of n×𝔗Am\lceil\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}\rceil as ii\rightarrow\infty. (This includes the case where our collection of strata is empty and fiff_{i}\rightarrow f uniformly on compact subsets.)

  2. 2.

    For any smooth exploded vectorfield vv, v(fi)v(f_{i}) converges to some function vfvf in Ck1,δC^{k-1,\delta}.

Define Ck,δ(n×𝔗Am)C^{k,\delta}(\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}) to be the closure of CC^{\infty} in C0C^{0} with this topology. Define C,δC^{\infty,\delta} to be the intersection of Ck,δC^{k,\delta} for all kk.

In particular, CkCk,δC^{k}\subset C^{k,\delta} for 0<δ<10<\delta<1. Functions in Ck,δC^{k,\delta} can be thought of as functions which converge a little slower than CkC^{k} functions when they approach different strata. Thinking of a single strata as being analogous to a cylindrical end, this is similar to requiring exponential convergence (with exponent δ\delta) on the cylindrical end. We shall now start showing that we can replace smooth functions with C,δC^{\infty,\delta} functions in the definition of exploded torus fibrations to create a category of C,δC^{\infty,\delta} exploded torus fibrations.

Definition 2.29.

A Ck,δC^{k,\delta} exploded function fk,δ,×(n×𝔗Am)f\in\mathcal{E}^{k,\delta,\times}(\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}) is a function of the form

f(x,z~):=g(x,z~)z~α𝔱a where gCk,δ(n×𝔗Am,),αm,𝔱a𝔱f(x,\tilde{z}):=g(x,\tilde{z})\tilde{z}^{\alpha}\mathfrak{t}^{a}\text{ where }g\in C^{k,\delta}\left(\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A},\mathbb{C}^{*}\right),\ \alpha\in\mathbb{Z}^{m},\ \mathfrak{t}^{a}\in\mathfrak{t}^{\mathbb{R}}

A sequence of exploded functions giz~α𝔱aig^{i}\tilde{z}^{\alpha}\mathfrak{t}^{a_{i}} converge in C,δC^{\infty,\delta} if the sequence of functions gig^{i} does, and the sequence aia_{i} is eventually constant.

A Ck,δC^{k,\delta} exploded fibration is an abstract exploded fibration locally modeled on open subsets of n×𝔗Am\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A} with the sheaf k,δ,×\mathcal{E}^{k,\delta,\times}.

The following observations are easy to prove:

Lemma 2.30.

Ck,δC^{k,\delta} is an algebra over CC^{\infty} for any 0<δ<10<\delta<1.

Lemma 2.31.

Given any ‘linear’ map

α:n×𝔗Amn×𝔗Bm\alpha:\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}\longrightarrow\mathbb{R}^{n^{\prime}}\times\mathfrak{T}^{m^{\prime}}_{B}
α(x,z~):=(Mx,z~α1,,zαm)\alpha(x,\tilde{z}):=\left(Mx,\tilde{z}^{\alpha^{1}},\dotsc,z^{\alpha^{m^{\prime}}}\right)

where MM is a nn by nn^{\prime} matrix and αji\alpha^{i}_{j} is a mm by mm^{\prime} matrix with \mathbb{Z} entries, α\alpha preserves C,δC^{\infty,\delta} in the sense that given any function fC,δ(n×𝔗Bm)f\in C^{\infty,\delta}(\mathbb{R}^{n^{\prime}}\times\mathfrak{T}^{m^{\prime}}_{B}),

αfC,δ(n×𝔗Am)\alpha\circ f\in C^{\infty,\delta}\left(\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}\right)

More difficult is the following:

Lemma 2.32.

Given any Ck,δC^{k,\delta} section of T(n×𝔗Am)T\left(\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A}\right), we can define a map of the form

exp(f)(x,z~1,,z~m):=(x+fn(x,z~),ef1(x,z~)z~1,,efm(x,z~)z~m)\exp(f)(x,\tilde{z}_{1},\dotsc,\tilde{z}_{m}):=\left(x+f_{\mathbb{R}^{n}}(x,\tilde{z}),e^{f_{1}(x,\tilde{z})}\tilde{z}_{1},\dotsc,e^{f_{m}(x,\tilde{z})}\tilde{z}_{m}\right)

If hh is in Ck,δC^{k,\delta}, then hexpfh\circ\exp f is.

Proof:

We must show that expressions of the following form decay appropriately.

ΔI(hexpf)\Delta_{I}(h\circ\exp f)

Note that

eS(hexpf)=(eSh)exp(eSf)e_{S}(h\circ\exp f)=(e_{S}h)\circ\exp(e_{S}f)

A little thought shows that we can rewrite our above expression in the following form

ΔI(hexpf)=II′′=I,II′′=((eIΔI′′h)expΔI)f\Delta_{I}(h\circ\exp f)=\sum_{I^{\prime}\cup I^{\prime\prime}=I,I^{\prime}\cap I^{\prime\prime}=\emptyset}\left(\left(e_{I^{\prime}}\Delta_{I^{\prime\prime}}h\right)\circ\exp\circ\Delta_{I^{\prime}}\right)f

As an example for interpreting the notation above, we write

(hexpΔS)f:=hexpfhexp(eSf)\left(h\circ\exp\circ\Delta_{S}\right)f:=h\circ\exp f-h\circ\exp(e_{S}f)

as opposed to

hexp(ΔSf):=hexp(feSf)h\circ\exp\left(\Delta_{S}f\right):=h\circ\exp(f-e_{S}f)

We have that because ff is bounded on compact sets,

(ΔIh)f decays appropriately.(\Delta_{I}h)\circ f\text{ decays appropriately.}

We can bound expressions of the form

((ΔIh)expΔI′′)f\left(\left(\Delta_{I^{\prime}}h\right)\circ\exp\circ\Delta_{I^{\prime\prime}}\right)f

by the size of the first |I′′|\left\lvert I^{\prime\prime}\right\rvert derivatives of ΔIh\Delta_{I^{\prime}}h times the sum of all products of the form |ΔIif|\prod\left\lvert\Delta_{I_{i}}f\right\rvert where I′′I^{\prime\prime} is the disjoint union of {Ii}\{I_{i}\}. We therefore get that all of the terms in the above expression decay appropriately. The appropriate decay of the derivatives of hexpfh\circ\exp f follows from the fact that C,δC^{\infty,\delta} is closed under multiplication, and a similar argument to the above one.

\square

Consider a map for which the pull back of exploded coordinate functions is in k,δ,×\mathcal{E}^{k,\delta,\times} and the pullback of real coordinate functions is Ck,δC^{k,\delta}. Any map of this form factors as a composition of maps in the form of Lemma 2.31 and Lemma 2.32; (first a map in the form of Lemma 2.31 to the product of the domain and target, then a map of the form of Lemma 2.32, then a projection to the target, which is of the form of Lemma 2.31.) We therefore have the following:

Corollary 2.33.

A morphism is Ck,δC^{k,\delta} if and only if the pull back of exploded coordinate functions are k,δ,×\mathcal{E}^{k,\delta,\times} functions, and the pull back of real coordinate functions are Ck,δC^{k,\delta} functions.

We shall need the following definition of convergence.

Definition 2.34.

A sequence of Ck,δC^{k,\delta} exploded maps fi:𝔄𝔅f^{i}:\mathfrak{A}\longrightarrow\mathfrak{B} converges to f:𝔄𝔅f:\mathfrak{A}\longrightarrow\mathfrak{B} in Ck,δC^{k,\delta} if the pullback under fif^{i} of any local coordinate function on 𝔅\mathfrak{B} converges in Ck,δC^{k,\delta} to the pullback under ff.

2.5 Almost complex structures

Definition 2.35.

An almost complex structure JJ on a smooth exploded 𝕋\mathbb{T} fibration 𝔅\mathfrak{B} is an endomorphism of T𝔅T\mathfrak{B} given by a smooth section of T𝔅T𝔅T\mathfrak{B}\otimes T^{*}\mathfrak{B} which squares to become multiplication by 1-1, so that for any coordinate chart modeled on an open subset of n×𝔗m\mathbb{R}^{n}\times\mathfrak{T}^{m} JJ of the real part of z~iz~i\tilde{z}_{i}\frac{\partial}{\partial\tilde{z}_{i}} is the imaginary part.

An almost complex structure JJ is a complex structure if there exist local coordinates z~𝔗Am\tilde{z}\in\mathfrak{T}^{m}_{A} so that for all vectorfields vv, iv(z~j)=(Jv)(z~j)iv(\tilde{z}_{j})=(Jv)(\tilde{z}_{j}).

This differs from the usual definition of an almost complex structure only in the extra requirement that in standard coordinates, (Jv)z~j=ivz~j(Jv)\tilde{z}_{j}=iv\tilde{z}_{j} which is only required to hold when z~j\tilde{z}_{j} is locally a coordinate on 𝔗\mathfrak{T}. For instance, on 𝔗11\mathfrak{T}^{1}_{1}, this requirement is only valid when z~<𝔱0\lfloor\tilde{z}\rfloor<\mathfrak{t}^{0}. This extra requirement makes holomorphic curves C,δC^{\infty,\delta} exploded 𝕋\mathbb{T} morphisms, and makes the tropical part of holomorphic curves piecewise linear one complexes. If it did not hold, then we would need to use a different version of the exploded category using \mathbb{R}^{*} instead of \mathbb{C}^{*} to explore holomorphic curve theory. The analysis involved would be significantly more difficult.

The following assumption allows us to use standard (pseudo)holomorphic curve results.

Definition 2.36.

An almost complex structure JJ on 𝔅\mathfrak{B} is civilized if it induces a smooth almost complex structure on the smooth part of 𝔅\mathfrak{B}. This means that given any local coordinate chart modeled on 𝔗Am\mathfrak{T}^{m}_{A}, if we consider 𝔗Am\mathfrak{T}^{m}_{A} as a subset of 𝔗nm\mathfrak{T}^{m^{\prime}}_{n} defined by setting some monomials equal to 11, there exists an almost complex structure J^\hat{J} on 𝔗nm\mathfrak{T}^{m^{\prime}}_{n} which induces a smooth almost complex structure on 𝔗nm=n\lceil\mathfrak{T}^{m^{\prime}}_{n}\rceil=\mathbb{C}^{n} so that the subset corresponding to 𝔗Am\mathfrak{T}^{m}_{A} is holomorphic, and the restriction of J^\hat{J} is JJ.

The word civilized should suggest that our almost complex structure is well behaved in a slightly unnatural way. Any complex structure is automatically civilized, and there are no obstructions to modifying an almost complex structure to civilize it. If we assume our almost complex structure is civilized, then holomorphic curves are smooth maps. Otherwise, they will just be C,δC^{\infty,\delta} for any δ<1\delta<1.

Example 2.37.

Suppose that we have a complex manifold MM along with a collection of immersed complex submanifolds NiN_{i} so that NiN_{i} intersect themselves and each other transversely. Then there is a smooth complex exploded fibration Expl(M)\operatorname{Expl}(M) called the explosion of MM. (There is actually a functor Expl\operatorname{Expl} from complex manifolds with a certain type of log structure which is defined by the submanifolds NiN_{i} to the category of exploded fibrations.)

We define Expl(M)\operatorname{Expl}(M) as follows: choose holomorphic coordinate charts on MM which are equal to balls inside n\mathbb{C}^{n}, so that the image of the submanifolds NiN_{i} are equal to the submanifolds {zi=0}\{z_{i}=0\}. Then replace a coordinate chart UnU\subset\mathbb{C}^{n} by a coordinate chart U~\tilde{U} in 𝔗nn\mathfrak{T}^{n}_{n} with coordinates z~i\tilde{z}_{i} so that

U~:={z~ so that z~U}\tilde{U}:=\{\tilde{z}\text{ so that }\lceil\tilde{z}\rceil\in U\}

Define transition functions as follows: the old transition functions are all of the form fi(z)=zjg(z)f_{i}(z)=z_{j}g(z) where gg is holomorphic and non vanishing. Replace this with f~i(z~)=z~jg(z~)\tilde{f}_{i}(\tilde{z})=\tilde{z}_{j}g(\lceil\tilde{z}\rceil), which is then a smooth exploded function. This defines transition functions which define Expl(M)\operatorname{Expl}(M) as a smooth complex exploded fibration.

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\includegraphics{ETF4}
Definition 2.38.

An exploded 𝕋\mathbb{T} curve is a 22 real dimensional, complete exploded 𝕋\mathbb{T} fibration with a complex structure jj.

A holomorphic curve is a holomorphic map of an exploded 𝕋\mathbb{T} curve to an almost complex exploded fibration.

A smooth exploded curve is a smooth map of an exploded 𝕋\mathbb{T} curve to a exploded 𝕋\mathbb{T} fibration.

By a smooth component of a holomorphic curve \mathfrak{C}, we shall mean a connected component of the set of points in \mathfrak{C} which have a neighborhood modeled on an open subset of \mathbb{C}. By an edge of \mathfrak{C} we shall mean connected component in [][\mathfrak{C}] minus the image of all smooth components. We shall call edges that have only one end attached to a smooth component ‘punctures’. Each smooth component is a punctured Riemann surface with punctures corresponding to where it is connected to edges. The information in a holomorphic curve \mathfrak{C} is equal to the information of a nodal Riemann surface plus gluing information for each node parametrized by 𝔱(0,)\mathbb{C}^{*}\mathfrak{t}^{(0,\infty)} (for more details of this gluing information see example 2.47 on page 2.47).

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\includegraphics{ETF7}
Example 2.39.

Consider a smooth curve f:𝔗nf:\mathfrak{C}\longrightarrow\mathfrak{T}^{n}. This is given by nn exploded functions f(z~1),,f(z~n)×()f^{*}(\tilde{z}_{1}),\dotsc,f^{*}(\tilde{z}_{n})\in\mathcal{E}^{\times}(\mathfrak{C}). Each smooth component of \mathfrak{C} is sent to the ()n\left(\mathbb{C}^{*}\right)^{n} worth of points over a particular point in the tropical part 𝔗n\lfloor\mathfrak{T}^{n}\rfloor. We can choose local holomorphic coordinates w~𝔗11\tilde{w}\in\mathfrak{T}^{1}_{1} on \mathfrak{C}. In these coordinates,

f(w~)=(g1(w~)𝔱a1w~α1,,gn(w~)𝔱anw~αn)giC(,),αnf(\tilde{w})=(g_{1}(\lceil\tilde{w}\rceil)\mathfrak{t}^{a_{1}}\tilde{w}^{\alpha_{1}},\dotsc,g_{n}(\lceil\tilde{w}\rceil)\mathfrak{t}^{a_{n}}\tilde{w}^{\alpha_{n}})\ \ g_{i}\in C^{\infty}(\mathbb{C},\mathbb{C}^{*}),\ \alpha\in\mathbb{Z}^{n}

(Our smooth curve is holomorphic if and only if the functions gig_{i} in our local coordinate representation above are holomorphic.) In particular, ff restricted to a smooth component gives a smooth map of the corresponding punctured Riemann surface to ()n\left(\mathbb{C}^{*}\right)^{n}, and the homology class in H1(()n)H_{1}\left(\left(\mathbb{C}^{*}\right)^{n}\right) of a loop around the puncture corresponding in the above coordinates to w~=0\lceil\tilde{w}\rceil=0 is given by α\alpha. Of course, the sum of all such homology classes from punctures of a smooth component is zero. As α\alpha also determines the derivative of ff on edges, this can be viewed as some kind of conservation of momentum condition for the tropical part of our curve, f\lfloor f\rfloor. In tropical geometry, this is called the balancing condition.

Example 2.40.

One way to consider the image of some holomorphic curves in 𝔗n\mathfrak{T}^{n} is as the ‘locus of noninvertablity’ of some set of polynomials

Pi(z~):=c~i,αz~αi=1,,n1P_{i}(\tilde{z}):=\sum\tilde{c}_{i,\alpha}\tilde{z}^{\alpha}\ i=1,\dotsc,n-1

We can consider the set

Z{Pi}:={z~ so that Pi(z~)0𝔱i}Z_{\{P_{i}\}}:=\left\{\tilde{z}\text{ so that }P_{i}(\tilde{z})\in 0\mathfrak{t}^{\mathbb{R}}\ \forall i\right\}

Suppose that for all points pZ{Pi}p\longrightarrow Z_{\{P_{i}\}}, the differentials {dPi}\{dP_{i}\} at pp are linearly independent. Then Z{Pi}Z_{\{P_{i}\}} is the image of some holomorphic curve.

Let us examine the set Z{Pi}Z_{\{P_{i}\}} more closely. For any point z~0\tilde{z}_{0}, denote by Si,z~0S_{i,\tilde{z}_{0}} the set of exponents α\alpha so that Pi(z~0)=c~i,αz~α\lfloor P_{i}(\tilde{z}_{0})\rfloor=\lfloor\tilde{c}_{i,\alpha}\tilde{z}^{\alpha}\rfloor. Then there exists some neighborhood of z~0\lfloor\tilde{z}_{0}\rfloor in 𝔗n\lfloor\mathfrak{T}^{n}\rfloor so that

Pi=αSi,z~0c~i,αz~αP_{i}=\sum_{\alpha\in S_{i,\tilde{z}_{0}}}\tilde{c}_{i,\alpha}\tilde{z}^{\alpha}

The points inside Z{Pi}Z_{\{P_{i}\}} over z~0\lfloor\tilde{z}_{0}\rfloor are then given by solutions of the equations

αSi,z~0ci,αzα=0 where c~i,α=ci,αc~i,α and z~=zz~\sum_{\alpha\in S_{i,\tilde{z}_{0}}}c_{i,\alpha}z^{\alpha}=0\text{ where }\tilde{c}_{i,\alpha}=c_{i,\alpha}\lfloor\tilde{c}_{i,\alpha}\rfloor\text{ and }\tilde{z}=z\lfloor\tilde{z}\rfloor

Note that the above equation has solutions for z()nz\in(\mathbb{C}^{*})^{n} if and only if Si,z~0S_{i,\tilde{z}_{0}} has more than 11 element. This corresponds to the tropical function Pi\lfloor P_{i}\rfloor (which is continuous, piecewise integral affine, and convex) not being smooth at z~0\lfloor\tilde{z}_{0}\rfloor. We therefore have that Z{Pi}\lfloor Z_{\{P_{i}\}}\rfloor is contained in the intersection of the non-smooth locus of the tropical polynomials Pi\lfloor P_{i}\rfloor.

2.6 Fiber product, refinements

Definition 2.41.

Two smooth (or Ck,δC^{k,\delta}) exploded morphisms

𝔄𝑓𝑔𝔅\mathfrak{A}\xrightarrow{f}\mathfrak{C}\xleftarrow{g}\mathfrak{B}

are transverse if for every pair of points p1𝔄p_{1}\longrightarrow\mathfrak{A} and p2𝔅p_{2}\longrightarrow\mathfrak{B} so that f(p1)=g(p2)f(p_{1})=g(p_{2}), df(Tp1𝔄)df(T_{p_{1}}\mathfrak{A}) and dg(Tp2𝔅)dg(T_{p_{2}}\mathfrak{B}) span Tf(p1)T_{f(p_{1})}\mathfrak{C}.

Definition 2.42.

If ff and gg are transverse smooth (or Ck,δC^{k,\delta}) exploded morphisms,

𝔄𝑓𝑔𝔅\mathfrak{A}\xrightarrow{f}\mathfrak{C}\xleftarrow{g}\mathfrak{B}

The fiber product 𝔄×gf𝔅\mathfrak{A}{}_{\hskip 3.0ptf\hskip-2.0pt}\times_{g}\mathfrak{B} is the unique smooth (or Ck,δC^{k,\delta}) exploded 𝕋\mathbb{T} fibration with maps to 𝔄\mathfrak{A} and 𝔅\mathfrak{B} so that the following diagram commutes

𝔄×gf𝔅𝔄𝔅\begin{array}[]{cll}\mathfrak{A}{}_{\hskip 3.0ptf\hskip-2.0pt}\times_{g}\mathfrak{B}&\longrightarrow&\mathfrak{A}\\ \downarrow&&\downarrow\\ \mathfrak{B}&\longrightarrow&\mathfrak{C}\end{array}

and with the usual universal property that given any commutative diagram

𝔇𝔄𝔅\begin{array}[]{lll}\mathfrak{D}&\longrightarrow&\mathfrak{A}\\ \downarrow&&\downarrow\\ \mathfrak{B}&\longrightarrow&\mathfrak{C}\end{array}

there exists a unique morphism 𝔇𝔄×gf𝔅\mathfrak{D}\longrightarrow\mathfrak{A}{}_{\hskip 3.0ptf\hskip-2.0pt}\times_{g}\mathfrak{B} so that the following diagram commutes

𝔇𝔄𝔅𝔄×gf𝔅\begin{array}[]{llc}\mathfrak{D}&\rightarrow&\mathfrak{A}\\ \downarrow&\searrow&\uparrow\\ \mathfrak{B}&\leftarrow&\mathfrak{A}{}_{\hskip 3.0ptf\hskip-2.0pt}\times_{g}\mathfrak{B}\end{array}

The existence of the fiber products defined above is left as an exercise. The existence in some special cases was proved in [12].

Example 2.43.

Consider the map f:𝔗Am𝔗nf:\mathfrak{T}^{m}_{A}\longrightarrow\mathfrak{T}^{n} given by

f(z~)=(z~α1,,z~αn)f(\tilde{z})=(\tilde{z}^{\alpha^{1}},\dotsc,\tilde{z}^{\alpha^{n}})

Denote by α\alpha the m×nm\times n matrix with entrys αji\alpha^{i}_{j}. The derivative of ff is surjective if α:mn\alpha:\mathbb{R}^{m}\longrightarrow\mathbb{R}^{n} is. Denote by |α|\left\lvert\alpha\right\rvert\in\mathbb{N} the size of α1αnn(m)\alpha^{1}\wedge\dotsb\wedge\alpha^{n}\in\bigwedge^{n}(\mathbb{Z}^{m}). (In other words, |α|\left\lvert\alpha\right\rvert\in\mathbb{N} is the smallest number so that the above wedge is at most |α|\left\lvert\alpha\right\rvert times any nonzero element of n(m)\bigwedge^{n}(\mathbb{Z}^{m}).) The fiber product of ff with the point (1,,1)(1,\dotsc,1) corresponds to the points in 𝔗Am\mathfrak{T}^{m}_{A} so that z~αi=1\tilde{z}^{\alpha^{i}}=1 for all ii. This is then equal to |α|\left\lvert\alpha\right\rvert copies of 𝔗𝔱kerαAmn\mathfrak{T}^{m-n}_{\mathfrak{t}^{\ker\alpha}\cap A} where we identify 𝔱mn=𝔱kerα\mathfrak{t}^{\mathbb{R}^{m-n}}=\mathfrak{t}^{\ker\alpha}.

The following should not be difficult to prove.

Conjecture 2.44.

If ff and gg are complete and transverse

𝔄𝑓𝑔𝔅\mathfrak{A}\xrightarrow{f}\mathfrak{C}\xleftarrow{g}\mathfrak{B}

then the fiber product 𝔄×gf𝔅\mathfrak{A}{}_{\hskip 3.0ptf\hskip-2.0pt}\times_{g}\mathfrak{B}\longrightarrow\mathfrak{C} is complete.

Definition 2.45.

A family of exploded 𝕋\mathbb{T} fibrations over 𝔉\mathfrak{F} is a map f:𝔉f:\mathfrak{C}\longrightarrow\mathfrak{F} so that:

  1. 1.

    ff is complete

  2. 2.

    for every point pp\longrightarrow\mathfrak{C},

    df:TpTf(p)𝔉 is surjective df:T_{p}\mathfrak{C}\longrightarrow T_{f(p)}\mathfrak{F}\text{ is surjective }
     and df:TpTp𝔉 is surjective \text{ and }df:{}^{\mathbb{Z}}T_{p}\mathfrak{C}\longrightarrow{}^{\mathbb{Z}}T_{p}\mathfrak{F}\text{ is surjective }

(The definition of complete is found on page 2.19. Recall also that integer vectors in Tp{}^{\mathbb{Z}}T_{p}\mathfrak{C} are the vectors vv so that for any exploded function, vfvf is an integer times ff. For example on 𝔗\mathfrak{T} the integer vectors are given by integer multiples of the real part of z~z~\tilde{z}\frac{\partial}{\partial\tilde{z}}, so the map 𝔗𝔗\mathfrak{T}\longrightarrow\mathfrak{T} given by z~2\tilde{z}^{2} is not a family as it does not obey the last condition above.)

The local normal form for coordinate charts on a smooth family is a map 𝔗Am𝔗Bk\mathfrak{T}^{m}_{A}\longrightarrow\mathfrak{T}^{k}_{B} given by (z~,w~)z~(\tilde{z},\tilde{w})\mapsto\tilde{z}, where the cone BB is given by the projection of AA to the first kk coordinates, and the projection of every strata of AA is a strata of BB. This may differ from a product because the cone AA may not be a product of BB with something.

Example 2.46.

We can represent the usual compactified moduli space of stable curves ¯g,n\bar{\mathcal{M}}_{g,n} as a complex orbifold. There exist local holomorphic coordinates so that the boundary of ¯g,n\bar{\mathcal{M}}_{g,n} in these coordinates looks like {zi=0}\{z_{i}=0\}. As in example 2.37 on page 2.37, we can replace these coordinates ziz_{i} with z~i\tilde{z}_{i} to obtain a complex orbifold exploded fibration Expl(¯g,n)\operatorname{Expl}\left(\bar{\mathcal{M}}_{g,n}\right). The forgetful map π:¯g,n+1¯g,n\pi:\bar{\mathcal{M}}_{g,n+1}\longrightarrow\bar{\mathcal{M}}_{g,n} induces an map

π:Expl(¯g,n+1)Expl(¯g,n)\pi:\operatorname{Expl}\left(\bar{\mathcal{M}}_{g,n+1}\right)\longrightarrow\operatorname{Expl}\left(\bar{\mathcal{M}}_{g,n}\right)

This is a family, and each stable exploded curve with genus gg and nn marked points corresponds to the fiber over some point pExpl(¯g,n)p\longrightarrow\operatorname{Expl}\left(\bar{\mathcal{M}}_{g,n}\right). This example is considered in a little more detail in [12]. (Obviously, it would take a little more work to prove that Expl¯g,n\operatorname{Expl}\bar{\mathcal{M}}_{g,n} actually represents the moduli stack of stable exploded curves, but that is not the subject of this paper.)

Example 2.47.

The following example contains all interesting local behavior of the above example. It is not a family only because it fails to be complete (it is not topologically proper). Consider the map

π:𝔗22𝔗11 given by πz~=w~1w~2\pi:\mathfrak{T}^{2}_{2}\longrightarrow\mathfrak{T}^{1}_{1}\text{ given by }\pi^{*}\tilde{z}=\tilde{w}_{1}\tilde{w}_{2}

The derivative is surjective, as can be seen by the equation

π(z~1dz~)=w~11dw~1+w~21dw~2\pi^{*}(\tilde{z}^{-1}d\tilde{z})=\tilde{w}_{1}^{-1}d\tilde{w}_{1}+\tilde{w}_{2}^{-1}d\tilde{w}_{2}

The fibers of this map over smooth points z~=c𝔱0\tilde{z}=c\mathfrak{t}^{0} are smooth manifolds equal to \mathbb{C}^{*} considered just as a smooth manifold with coordinates w1w_{1} and w2w_{2}\in\mathbb{C}^{*} related by w1w2=cw_{1}w_{2}=c. (Note that there is no point with z~=0𝔱0\tilde{z}=0\mathfrak{t}^{0}.)

In contrast, the fibers of this map over points z~=c𝔱x\tilde{z}=c\mathfrak{t}^{x} with x>0x>0 are not smooth manifolds. They can be described using coordinates w~1,w~2𝔗[0,x)1𝔗11\tilde{w}_{1},\tilde{w}_{2}\in\mathfrak{T}^{1}_{[0,x)}\subset\mathfrak{T}^{1}_{1}, and transition maps give by w~2w~2=c𝔱x\tilde{w}_{2}\tilde{w}_{2}=c\mathfrak{t}^{x} on their intersection, which is 𝔗(0,x)1\mathfrak{T}^{1}_{(0,x)}. The picture below shows some of the fibers of this map

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\includegraphics{ETF5}
Definition 2.48.

A refinement of 𝔅\mathfrak{B} is an exploded 𝕋\mathbb{T} fibration 𝔅\mathfrak{B}^{\prime} with a map f:𝔅𝔅f:\mathfrak{B}^{\prime}\longrightarrow\mathfrak{B} so that

  1. 1.

    ff is complete

  2. 2.

    ff gives a bijection between points in 𝔅\mathfrak{B}^{\prime} and 𝔅\mathfrak{B}

  3. 3.

    dfdf is surjective

Example 2.49.

All refinements are locally of the following form: Suppose that the integral affine cone A𝔱mA\subset\mathfrak{t}^{\mathbb{R}^{m}} is equal to the union of a collection of cones {A^i}\{\hat{A}_{i}\} so that the intersection of any two of these A^i\hat{A}_{i} has codimension at least 11. Then using the standard coordinates (x,z~)(x,\tilde{z}) from example 2.7 (on page 2.7) on n×𝔗A^im\mathbb{R}^{n}\times\mathfrak{T}^{m}_{\hat{A}_{i}} for each A^i\hat{A}_{i}, we can piece the coordinate charts n×𝔗A^im\mathbb{R}^{n}\times\mathfrak{T}^{m}_{\hat{A}_{i}} together using the identity map as transition coordinates. In these coordinates, the map down to n×𝔗Am\mathbb{R}^{n}\times\mathfrak{T}^{m}_{A} is again the the identity map in these coordinates. (The fact that all refinements are locally of this form can be proved using the observation that the pull back of coordinate functions on 𝔅\mathfrak{B} must locally be coordinate functions on 𝔅\mathfrak{B}^{\prime}.) So in the case that 𝔅\mathfrak{B} is basic, a refinement is simply determined by a subdivision of the tropical part 𝔅\lfloor\mathfrak{B}\rfloor. The effect on the smooth part 𝔅\lceil\mathfrak{B}\rceil should remind the reader of a toric blowup.

The following lemma follows from the above standard local form for refinements.

Lemma 2.50.

Given a refinement of 𝔅\mathfrak{B},

f:𝔄𝔅f:\mathfrak{A}\longrightarrow\mathfrak{B}

Any smooth exploded vector field vv on 𝔅\mathfrak{B} lifts uniquely to a smooth exploded vector field v~\tilde{v} on 𝔄\mathfrak{A} so that df(v~)=vdf(\tilde{v})=v.

The above lemma tells us that any smooth tensor field (such as an almost complex structure or metric) lifts uniquely to a smooth tensor field on any refinement. The above normal form implies that given any morphism 𝔅\mathfrak{C}\longrightarrow\mathfrak{B} and a refinement 𝔅𝔅\mathfrak{B}^{\prime}\longrightarrow\mathfrak{B}, there exists a refinement \mathfrak{C}^{\prime}\longrightarrow\mathfrak{C} and a morphism 𝔅\mathfrak{C}^{\prime}\longrightarrow\mathfrak{B}^{\prime} lifting the above map.

Definition 2.51.

Call a holomorphic curve stable if it has a finite number of automorphisms, and is not a nontrivial refinement of another holomorphic curve.

If 𝔅\mathfrak{B} has an almost complex structure, there is a bijection between stable holomorphic curves in 𝔅\mathfrak{B} and stable holomorphic curves in any refinement 𝔅\mathfrak{B}^{\prime}. In fact, when the moduli space of stable holomorphic curves in 𝔅\mathfrak{B} is smooth, the moduli space of stable holomorphic curves in 𝔅\mathfrak{B}^{\prime} is a refinement of this moduli space.

2.7 Symplectic structures

Our definition of symplectic structures below should be regarded as provisional. Our definitions are enough to tame holomorphic curves, which is all that is required for this paper, but there is probably a better theory of symplectic structures on exploded fibrations.

Definition 2.52.

A symplectic exploded 𝕋\mathbb{T} fibration (𝔅,ω)(\mathfrak{B},\omega) is an even dimensional exploded 𝕋\mathbb{T} fibration 𝔅\mathfrak{B} with a two form ω\omega so that each point has a neighborhood equal to an open set in 2n×𝔗Am\mathbb{R}^{2n}\times\mathfrak{T}^{m}_{A} with a two form defined as follows

ω:=ω2nijdzαjdz¯αj\omega:=\omega_{\mathbb{R}^{2n}}-i\sum_{j}d{z^{\alpha^{j}}}\wedge d\bar{z}^{\alpha^{j}}

Here ω2n\omega_{\mathbb{R}^{2n}} is the standard form on 2n\mathbb{R}^{2n} and {αj}\{\alpha^{j}\} is some finite set generating AA^{*}.

The second part of this symplectic form can be written as

id(αidθi)\sum_{i}d(\alpha_{i}d\theta_{i})
αi=jαij|zαj|2 or α:=j|zαj|2αj\alpha_{i}=\sum_{j}\alpha_{i}^{j}\left\lvert z^{\alpha^{j}}\right\rvert^{2}\text{ or }\alpha:=\sum_{j}\left\lvert z^{\alpha^{j}}\right\rvert^{2}\alpha^{j}

By dθid\theta_{i}, we mean the imaginary part of z~1dz~\tilde{z}^{-1}d\tilde{z}. The Hamiltonian torus action on this standard model given by the vectorfields θi\frac{\partial}{\partial\theta_{i}} has moment map given by α:=(α1,,αn)n\alpha:=(\alpha_{1},\dotsc,\alpha_{n})\subset\mathbb{R}^{n}. The image of this moment map is the dual cone to AA defined by

{α so that αa0 for all aA}\{\alpha\text{ so that }\alpha\cdot a\geq 0\text{ for all }a\in A\}
Example 2.53.

Suppose that we have a compact symplectic manifold MM with some collection of embedded symplectic submanifolds NiN_{i} which intersect each other orthogonally. (In other words, if xNiNjx\in N_{i}\cap N_{j}, the vectors vTxMv\in T_{x}M so that ω(v,)\omega(v,\cdot) vanishes on TxNiT_{x}N_{i} are contained inside TxNjT_{x}N_{j}.) Then a standard Moser type argument gives that there exists some collection of neighborhoods UiU_{i} of NiN_{i} so that:

  1. 1.

    UiU_{i} is identified with some neighborhood of the zero section in a complex line bundle over NiN_{i} (with 𝕋\mathbb{T} action given by multiplication by eiθe^{i\theta}), and a 𝕋\mathbb{T} invariant connection one form αi\alpha_{i} so that ω=ωNi+d(ri2αi)\omega=\omega_{N_{i}}+d(r_{i}^{2}\alpha_{i}), where rir_{i} indicates some radial coordinate on the line bundle, NiN_{i} is identified with the zero section, and ωNi\omega_{N_{i}} indicates the pullback of the form ω\omega under the projection to NiN_{i}.

  2. 2.

    These identifications are compatible in the sense that on iIUi\bigcap_{i\in I}U_{i}, all the different projections to NiN_{i} and the 𝕋\mathbb{T} actions commute, and

    ω=ωNi+d(ri2αi)\omega=\omega_{\bigcap N_{i}}+\sum d(r_{i}^{2}\alpha_{i})

    where ωNi\omega_{\bigcap N_{i}} is the pullback of ω\omega under the composition of projection maps to Ni\bigcap N_{i}.

We can associate an exploded 𝕋\mathbb{T} fibration to the above as follows:

  1. 1.

    Choose coordinate charts VV on MM which are identified with open balls in n\mathbb{C}^{n} so that the intersection of VV with the submanifolds NiN_{i} is equal to the union of the planes zk=0z_{k}=0. If NiN_{i} intersects this coordinate chart, it does so where some coordinate zki=0z_{k_{i}}=0. Choose these charts so that the fibers of the complex line bundle over NiN_{i} are identified with the slices where all other coordinates are constant, and the 𝕋\mathbb{T} action is multiplication of zkiz_{k_{i}} by eiθe^{i\theta}, and ri=|zki|r_{i}=\left\lvert z_{k_{i}}\right\rvert.

  2. 2.

    Replace VnV\subset\mathbb{C}^{n} with the corresponding subset of 𝔗nn\mathfrak{T}^{n}_{n}, where each coordinate ziz_{i} is replaced by the standard coordinate z~i\tilde{z}_{i}. Replacing ziz_{i} with z~i\tilde{z}_{i} in transition functions between these coordinates gives transition functions which are smooth exploded isomorphisms, which defines an exploded fibration 𝔐\mathfrak{M}. A standard Moser type argument then gives that the two form on 𝔐\mathfrak{M} corresponding to ω\omega is in the form defined above.

The following is a picture of the result of this procedure on a toric symplectic manifold, where for our symplectic submanifolds we use the fixed loci of circle actions. We shall represent the smooth part 𝔐\lceil\mathfrak{M}\rceil by its moment map:

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\includegraphics{ETF6}
Definition 2.54.

An almost complex structure JJ on a symplectic exploded 𝕋\mathbb{T} fibration (𝔅,ω)(\mathfrak{B},\omega) is tamed by ω\omega if ω(v,Jv)\omega(v,Jv) is positive for any vector vv so that vf0vf\neq 0 for some smooth function ff.

The above definition of taming is clearly inadequate for taming holomorphic curves in 𝔗n\mathfrak{T}^{n} where it is satisfied trivially by a two form that is identically 0 because all smooth functions on 𝔗n\mathfrak{T}^{n} are locally constant. We remedy this as follows: (Note that we shall use freely the fact that any holomorphic curve in 𝔅\mathfrak{B} lifts to a holomorphic curve (with a refined domain) in 𝔅\mathfrak{B}^{\prime} for any refinement 𝔅𝔅\mathfrak{B}^{\prime}\longrightarrow\mathfrak{B}.)

Definition 2.55.

A strict taming of a complete almost complex exploded 𝕋\mathbb{T} fibration 𝔅\mathfrak{B} is a set Ω\Omega of closed two forms on refinements of 𝔅\mathfrak{B} containing at least one symplectic form. Each form in Ω\Omega must be nonnegative on JJ holomorphic planes, and there must exist a metric gg on 𝔅\mathfrak{B} and a positive number c>0c>0 smaller than the injectivity radius of gg so that

  1. 1.

    Given any smooth exploded 𝕋\mathbb{T} curve f:𝔅f:\mathfrak{C}\longrightarrow\mathfrak{B}, the ω\omega energy, fω\int_{\mathfrak{C}}f^{*}\omega is a constant independent of ωΩ\omega\in\Omega

  2. 2.

    Given any point p𝔅p\longrightarrow\mathfrak{B}, there exists a taming form ωpΩ\omega_{p}\in\Omega so that on the ball of radius cc around pp,

    ωp(v,Jv)g(v,v)\omega_{p}(v,Jv)\geq g(v,v)

The Ω\Omega energy of a map f:𝔅f:\mathfrak{C}\longrightarrow\mathfrak{B} where \mathfrak{C} is two dimensional and oriented is given by

EΩ(f):=supωΩfωE_{\Omega}(f):=\sup_{\omega\in\Omega}\int f^{*}\omega

For example, any compact almost complex manifold tamed by a symplectic form is strictly tamed by that symplectic form. A strict taming is what is needed to get local area bounds on holomorphic curves.

The following example is very important for applications of the theory of holomorphic curves in exploded 𝕋\mathbb{T} fibrations.

Lemma 2.56.

Suppose that a symplectic exploded fibration (𝔅,ω)(\mathfrak{B},\omega) satisfies the following:

  1. 1.

    𝔅\mathfrak{B} is basic and complete.

  2. 2.

    All strata of the base 𝔅\lfloor\mathfrak{B}\rfloor have the integral affine structure of either a standard simplex or 𝔱[0,)n\mathfrak{t}^{[0,\infty)^{n}}

If the above hold, then there exists some civilized almost complex structure JJ and a strict taming Ω\Omega of (𝔅,J)(\mathfrak{B},J) so that ωΩ\omega\in\Omega.

Proof:

Each strata 𝔅i\mathfrak{B}_{i} is in the form of some 𝔗𝔅in\mathfrak{T}^{n}_{\lfloor\mathfrak{B}_{i}\rfloor} bundle over the interior of a smooth compact symplectic manifold (Mi,ωMi)(M_{i},\omega_{M_{i}}) with a collection of orthogonally intersecting codimension 22 embedded symplectic submanifolds MjM_{j} like in example 2.53 on page 2.53. (There is one such MjM_{j} for each strata 𝔅j\lfloor\mathfrak{B}_{j}\rfloor which has 𝔅i\lfloor\mathfrak{B}_{i}\rfloor as a boundary face). By interior we mean the complement of these symplectic submanifolds.

As in example 2.53, there exists a neighborhood of MjM_{j}, Ui,jMiU_{i,j}\subset M_{i} which is identified with a complex line bundle over MiM_{i} with radial coordinate ri,jr_{i,j} and connection one form αi,j\alpha_{i,j} so that ωMi=ωMj+d(ri,j2αi,j)\omega_{M_{i}}=\omega_{M_{j}}+d(r_{i,j}^{2}\alpha_{i,j}). We can make these identifications compatible in the sense that if MkM_{k} is the intersection of Mj1M_{j_{1}} and Mj2M_{j_{2}}, then on Ui,j1Ui,j2U_{i,j_{1}}\cap U_{i,j_{2}},

ωMi=ωMk+d(ri,j12αi,j1)+d(ri,j22αi,j2)\omega_{M_{i}}=\omega_{M_{k}}+d(r^{2}_{i,j_{1}}\alpha_{i,j_{1}})+d(r^{2}_{i,j_{2}}\alpha_{i,j_{2}})

the 𝕋\mathbb{T} actions and projections from the line bundles commute, and ri,j1r_{i,j_{1}} and αi,j1\alpha_{i,j_{1}} are the pullback of rj2,kr_{j_{2},k} and αj2,k\alpha_{j_{2},k} under the projection to Mj2M_{j_{2}}. (To construct these identifications compatibly, start with the strata 𝔅i\lfloor\mathfrak{B}_{i}\rfloor of highest dimension.)

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\psfrag{ETF19a}{$M_{i}:=\lceil\mathfrak{B}_{i}\rceil$}\psfrag{ETF19b}{$U_{i,j_{1}}$}\psfrag{ETF19c}{$U_{i,j_{2}}$}\psfrag{ETF19d}{$M_{j_{1}}$}\psfrag{ETF19e}{$M_{j_{2}}$}\psfrag{ETF19f}{$M_{k}$}\psfrag{ETF19g}{$r_{i,j_{1}}$}\psfrag{ETF19h}{$r_{i,j_{2}}$}\psfrag{ETF19i}{$r_{j_{1},k}$}\psfrag{ETF19j}{$r_{j_{2},k}$}\includegraphics{ETF19}

We can arrange this so that in local coordinates in Ui,jU_{i,j} the fiber coordinate zz is actually the smooth part of some coordinate z~\tilde{z} from a coordinate chart on 𝔅\mathfrak{B}. As 𝔅\mathfrak{B} is complete, there exists some r0>0r_{0}>0 so that the tubular neighborhood around MjM_{j} of radius r0r_{0} is always contained inside Ui,jU_{i,j}.

We can choose an almost complex structure JJ tamed by ω\omega so that on Ui,jU_{i,j}, JJ is the usual complex structure restricted to the fibers of our line bundle, and is determined by this and the lift (using our connection αi,j\alpha_{i,j}) of some almost complex structure JjJ_{j} on MjM_{j}. (Note that such a JJ will be civilized in the sense of definition 2.36 on page 2.36.)

Denote by 𝔘j{\mathfrak{U}}_{j} a connected open subset of 𝔅\mathfrak{B} which is the preimage of the union of all Ui,jU_{i,j} in 𝔅\lceil\mathfrak{B}\rceil. (This is an open subset that contains the strata 𝔅j\mathfrak{B}_{j}.) There exist 𝔱+\mathbb{R}^{*}\mathfrak{t}^{\mathbb{R}^{+}} valued functions r~i,j\tilde{r}_{i,j} defined on 𝔘j{\mathfrak{U}}_{j} who’s smooth part is equal to ri,jr_{i,j}, so that locally there exist standard coordinates so that r~i,j=c~k|z~k|\tilde{r}_{i,j}=\tilde{c}_{k}\left\lvert\tilde{z}_{k}\right\rvert, and so that ir~i,j=𝔱c\prod_{i}\lfloor\tilde{r}_{i,j}\rfloor=\mathfrak{t}^{c} if Bj\lfloor B_{j}\rfloor is simplex of finite size. Note that if 𝔅j\lfloor\mathfrak{B}_{j}\rfloor has nonzero dimension, then any adjacent strata 𝔅k\mathfrak{B}_{k} with 𝔅k\lfloor\mathfrak{B}_{k}\rfloor of nonzero dimension has the same size, and must obey a similar equation r~i,k=𝔱c\prod\lfloor\tilde{r}_{i,k}\rfloor=\mathfrak{t}^{c} with exactly the same constant 𝔱c\mathfrak{t}^{c}.

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\psfrag{ETF19a}{$M_{i}:=\lceil\mathfrak{B}_{i}\rceil$}\psfrag{ETF19b}{$U_{i,j_{1}}$}\psfrag{ETF19c}{$U_{i,j_{2}}$}\psfrag{ETF19d}{$M_{j_{1}}$}\psfrag{ETF19e}{$M_{j_{2}}$}\psfrag{ETF19f}{$M_{k}$}\psfrag{ETF19g}{$r_{i,j_{1}}$}\psfrag{ETF19h}{$r_{i,j_{2}}$}\psfrag{ETF19i}{$r_{j_{1},k}$}\psfrag{ETF19j}{$r_{j_{2},k}$}\psfrag{ETF20a}{$\lfloor\tilde{r}_{1,j}\rfloor$}\psfrag{ETF20b}{$\lfloor\tilde{r}_{2,j}\rfloor$}\psfrag{ETF20c}{$\lfloor\tilde{r}_{3,j}\rfloor$}\psfrag{ETF20d}{$\lfloor\tilde{r}_{1,j}\rfloor\lfloor\tilde{r}_{2,j}\rfloor\lfloor\tilde{r}_{3,j}\rfloor=\mathfrak{t}^{c}$}\psfrag{ETF20e}{$\lfloor\mathfrak{B}_{j}\rfloor$}\includegraphics{ETF20}

We now provide a strict taming Ω\Omega which for each point p𝔅p\longrightarrow\mathfrak{B} includes some symplectic form ωp\omega_{p} taming JJ well in a neighborhood of pp as in part 2 of the above definition of a strict taming. We do this as follows: Suppose that p𝔘jp\in\mathfrak{U}_{j} so that r~i,j(p)<r02\tilde{r}_{i,j}(p)<\frac{r_{0}}{2} and r~j,ir02\tilde{r}_{j,i}\geq\frac{r_{0}}{2} for all ii so that this makes sense. Consider the set Sp𝔱S_{p}\subset\mathbb{R}^{*}\mathfrak{t}^{\mathbb{R}} given by the coordinates of pp, r~i,j(p)\tilde{r}_{i,j}(p). We shall define ωp\omega_{p} on a refinement of 𝔅\mathfrak{B} determined by a subdivision of the base 𝔅\lfloor\mathfrak{B}\rfloor as in example 2.49 on page 2.49. Subdivide the strata 𝔅j\lfloor\mathfrak{B}_{j}\rfloor by the set of all planes of the form {r~i,jSp}\{\lfloor\tilde{r}_{i,j}\rfloor\in\lfloor S_{p}\rfloor\}. (This is the set of planes through pp parallel to the boundary of the strata and their image under any symmetry of the strata.) Similarly subdivide all strata 𝔅k\mathfrak{B}_{k} connected to 𝔅j\mathfrak{B}_{j} through strata with nonzero dimensional bases by the planes {r~i,jSp}\{\lfloor\tilde{r}_{i,j}\rfloor\in\lfloor S_{p}\rfloor\}. This subdivision of 𝔅\lfloor\mathfrak{B}\rfloor defines a refinement of 𝔅\mathfrak{B} as in example 2.49.

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\psfrag{ETF19a}{$M_{i}:=\lceil\mathfrak{B}_{i}\rceil$}\psfrag{ETF19b}{$U_{i,j_{1}}$}\psfrag{ETF19c}{$U_{i,j_{2}}$}\psfrag{ETF19d}{$M_{j_{1}}$}\psfrag{ETF19e}{$M_{j_{2}}$}\psfrag{ETF19f}{$M_{k}$}\psfrag{ETF19g}{$r_{i,j_{1}}$}\psfrag{ETF19h}{$r_{i,j_{2}}$}\psfrag{ETF19i}{$r_{j_{1},k}$}\psfrag{ETF19j}{$r_{j_{2},k}$}\psfrag{ETF20a}{$\lfloor\tilde{r}_{1,j}\rfloor$}\psfrag{ETF20b}{$\lfloor\tilde{r}_{2,j}\rfloor$}\psfrag{ETF20c}{$\lfloor\tilde{r}_{3,j}\rfloor$}\psfrag{ETF20d}{$\lfloor\tilde{r}_{1,j}\rfloor\lfloor\tilde{r}_{2,j}\rfloor\lfloor\tilde{r}_{3,j}\rfloor=\mathfrak{t}^{c}$}\psfrag{ETF20e}{$\lfloor\mathfrak{B}_{j}\rfloor$}\psfrag{ETF21a}{$\lfloor p\rfloor$}\includegraphics{ETF21}

We can now define a symplectic form ωp\omega_{p} on the refinement of 𝔅\mathfrak{B} defined above as follows: First choose a function

ϕ:𝔱\phi:\mathbb{R}^{*}\mathfrak{t}^{\mathbb{R}}\longrightarrow\mathbb{R}

so that

ϕ(r~)=0r~>2𝔱0\phi(\tilde{r})=0\ \forall\tilde{r}>2\mathfrak{t}^{0}
ϕ(r~)+ϕ(r~1)=1\phi(\tilde{r})+\phi(\tilde{r}^{-1})=-1

and considered as a function ϕ:\phi:\mathbb{R}^{*}\longrightarrow\mathbb{R}, ϕ\phi is smooth, increasing and

dϕ(r)dr>1r for all 12<r<2\frac{d\phi(r)}{dr}>\frac{1}{r}\text{ for all }\frac{1}{\sqrt{2}}<r<\sqrt{2}

Also choose some smooth monotone function ρ:\rho:\mathbb{R}\longrightarrow\mathbb{R} so that ρ(x)=x\rho(x)=x for all xr02x\geq r^{2}_{0} and ρ(x)=r022\rho(x)=\frac{r_{0}^{2}}{2} for all x<r024x<\frac{r_{0}^{2}}{4}.

Now define h:𝔱+h:\mathbb{R}^{*}\mathfrak{t}^{\mathbb{R}^{+}}\longrightarrow\mathbb{R} by

h(r~):=ρ(r~2)+r022|Sp|x~Spϕ(r~x~1)h(\tilde{r}):=\rho(\lceil\tilde{r}^{2}\rceil)+\frac{r_{0}^{2}}{2\left\lvert S_{p}\right\rvert}\sum_{\tilde{x}\in S_{p}}\phi(\tilde{r}\tilde{x}^{-1})

We shall be replacing r2r^{2} with h(r~)h(\tilde{r}) in what follows: the relevant properties of hh is that it agrees with r2r^{2} for r~\tilde{r} large enough, is 0 for r~\tilde{r} small enough, and has sufficiently large derivative close to points where r~Sp\tilde{r}\in S_{p}, and is monotone increasing. On 𝔘j\mathfrak{U}_{j} we have that ω=ωMj+d(ri,j2αi,j)\omega=\omega_{M_{j}}+\sum d(r^{2}_{i,j}\alpha_{i,j}), we shall replace this with

ωp:=ωMj+d(h(r~i,j)αi,j))\omega_{p}:=\omega_{M_{j}}+\sum d(h(\tilde{r}_{i,j})\alpha_{i,j}))

Now we do the same for all strata 𝔅k\mathfrak{B}_{k} connected to 𝔅j\mathfrak{B}_{j} through strata with bases of non zero dimension setting on 𝔘k\mathfrak{U}_{k}

ωp:=ωMk+d(h(r~i,k)αi,k))\omega_{p}:=\omega_{M_{k}}+\sum d(h(\tilde{r}_{i,k})\alpha_{i,k}))

For all other 𝔘i\mathfrak{U}_{i}, we simply leave ω\omega unchanged. This defines ωp\omega_{p} as a two form on our refinement. It is now not difficult to check that the set of forms Ω\Omega given by ω\omega and ωp\omega_{p} for all pp is a strict taming of (𝔅,J)(\mathfrak{B},J).

\square

For example, the exploded fibration constructed in example 2.53 from a compact symplectic manifold with orthogonally intersecting codimension 22 symplectic submanifolds admits an almost complex structure with a strict taming. This is relevant for considering Gromov Witten invariants relative to these submanifolds.

The following picture of a symplectic form on a refinement in a toric case may be helpful for understanding the construction in the above proof. (We have drawn 𝔅\lceil\mathfrak{B}\rceil in moment map coordiantes.)

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\psfrag{ETF19a}{$M_{i}:=\lceil\mathfrak{B}_{i}\rceil$}\psfrag{ETF19b}{$U_{i,j_{1}}$}\psfrag{ETF19c}{$U_{i,j_{2}}$}\psfrag{ETF19d}{$M_{j_{1}}$}\psfrag{ETF19e}{$M_{j_{2}}$}\psfrag{ETF19f}{$M_{k}$}\psfrag{ETF19g}{$r_{i,j_{1}}$}\psfrag{ETF19h}{$r_{i,j_{2}}$}\psfrag{ETF19i}{$r_{j_{1},k}$}\psfrag{ETF19j}{$r_{j_{2},k}$}\psfrag{ETF20a}{$\lfloor\tilde{r}_{1,j}\rfloor$}\psfrag{ETF20b}{$\lfloor\tilde{r}_{2,j}\rfloor$}\psfrag{ETF20c}{$\lfloor\tilde{r}_{3,j}\rfloor$}\psfrag{ETF20d}{$\lfloor\tilde{r}_{1,j}\rfloor\lfloor\tilde{r}_{2,j}\rfloor\lfloor\tilde{r}_{3,j}\rfloor=\mathfrak{t}^{c}$}\psfrag{ETF20e}{$\lfloor\mathfrak{B}_{j}\rfloor$}\psfrag{ETF21a}{$\lfloor p\rfloor$}\psfrag{ETF8a}{$\lceil\mathfrak{B}\rceil$}\psfrag{ETF8b}{$\lfloor\mathfrak{B}\rfloor$}\psfrag{ETF8c}{$\lceil\mathfrak{B}^{\prime}\rceil$}\psfrag{ETF8d}{$\lfloor\mathfrak{B}^{\prime}\rfloor$}\includegraphics{ETF8}
Example 2.57.

Given any symplectic 𝔅\mathfrak{B} with a compact tropical part 𝔅\lfloor\mathfrak{B}\rfloor satisfying the conditions of Lemma 2.56, we can construct a family 𝔅^𝔗11\hat{\mathfrak{B}}\longrightarrow\mathfrak{T}^{1}_{1} of symplectic exploded fibrations so that the fiber over z~=1𝔱1\tilde{z}=1\mathfrak{t}^{1} is 𝔅\mathfrak{B}, and the fiber over z~=1\tilde{z}=1 is a smooth compact symplectic manifold.

2.8 Moduli stack of smooth exploded 𝕋\mathbb{T} curves

We shall use the concept of a stack without giving the general definition. (See the article [3] for a readable introduction to stacks). The reader unfamiliar with stacks may just think of our use of stacks as a way of keeping all information about families of holomorphic curves in case it is needed for future papers.

When we say that we shall consider an exploded 𝕋\mathbb{T} fibration 𝔅\mathfrak{B} as a stack, we mean that we replace 𝔅\mathfrak{B} with a category 𝔅¯\underline{\mathfrak{B}} over the category of exploded 𝕋\mathbb{T} fibrations (in other words a category 𝔅¯\underline{\mathfrak{B}} with a functor to the category of exploded 𝕋\mathbb{T} fibrations) as follows: objects are maps into 𝔅\mathfrak{B}:

𝔄𝔅\mathfrak{A}\longrightarrow\mathfrak{B}

and morphisms are commutative diagrams

𝔄𝔅id𝔅\begin{array}[]{lll}\mathfrak{A}&\longrightarrow&\mathfrak{B}\\ \downarrow&&\downarrow\operatorname{id}\\ \mathfrak{C}&\longrightarrow&\mathfrak{B}\end{array}

The functor from 𝔅¯\underline{\mathfrak{B}} to the category of exploded 𝕋\mathbb{T} fibrations is given by sending 𝔄𝔅\mathfrak{A}\longrightarrow\mathfrak{B} to 𝔄\mathfrak{A}, and the above morphism to 𝔄\mathfrak{A}\longrightarrow\mathfrak{C}.

Note that a maps 𝔅𝔇\mathfrak{B}\longrightarrow\mathfrak{D} are equivalent to functors 𝔅¯𝔇¯\underline{\mathfrak{B}}\longrightarrow\underline{\mathfrak{D}} which commute with the functor down to the category of exploded fibrations. Such functors are morphism of categories over the category of exploded fibrations, and this is the correct notion of maps of stacks.

For example, a point thought of as a stack is equal to the category of exploded fibrations itself, with the functor down to the category of exploded fibrations the identity. We shall refer to points thought of this way simply as points.

Definition 2.58.

The moduli stack sm(𝔅)\mathcal{M}^{sm}(\mathfrak{B}) of smooth exploded 𝕋\mathbb{T} curves in 𝔅\mathfrak{B} is a category over the category of exploded fibrations with objects being families of smooth exploded curves consisting of the following

  1. 1.

    A exploded 𝕋\mathbb{T} fibration \mathfrak{C}

  2. 2.

    A pair of smooth exploded 𝕋\mathbb{T} morphisms

    𝔅π𝔉\begin{split}&\mathfrak{C}\longrightarrow\mathfrak{B}\\ \pi&\downarrow\\ &\mathfrak{F}\end{split}
  3. 3.

    A section jj of ker(dπ)(T/π(T𝔉))\ker(d\pi)\otimes\left(T^{*}\mathfrak{C}/\pi^{*}(T^{*}\mathfrak{F})\right)

so that

  1. 1.

    π:𝔉\pi:\mathfrak{C}\longrightarrow\mathfrak{F} is a family (definition 2.45 on page 2.45).

  2. 2.

    The inverse image of any point p𝔉p\longrightarrow\mathfrak{F} is an exploded 𝕋\mathbb{T} curve with complex structure jj.

A morphism between families of curves is given by exploded morphisms ff and cc making the following diagram commute

𝔉11𝔅fcid𝔉22𝔅\begin{array}[]{lllll}\mathfrak{F}_{1}&\longleftarrow&\mathfrak{C}_{1}&\longrightarrow&\mathfrak{B}\\ \downarrow f&&\downarrow c&&\downarrow\operatorname{id}\\ \mathfrak{F}_{2}&\longleftarrow&\mathfrak{C}_{2}&\longrightarrow&\mathfrak{B}\end{array}

so that cc is a jj preserving isomorphism on fibers.

The functor down to the category of exploded fibrations is given by taking the base 𝔉\mathfrak{F} of a family.

Note that morphisms are not quite determined by the map f:𝔉1𝔉2f:\mathfrak{F}_{1}\longrightarrow\mathfrak{F}_{2}. 1\mathfrak{C}_{1} is non-canonically isomorphic to the fiber product of 2\mathfrak{C}_{2} and 𝔉1\mathfrak{F}_{1} over 𝔉2\mathfrak{F}_{2}.

This is a moduli stack in the sense that a morphism 𝔉¯sm(𝔅)\underline{\mathfrak{F}}\longrightarrow\mathcal{M}^{sm}(\mathfrak{B}) is equivalent to a family of smooth curves 𝔉𝔅\mathfrak{F}\longleftarrow\mathfrak{C}\longrightarrow\mathfrak{B} (this is the family which is the image of the identity map 𝔉𝔉\mathfrak{F}\longrightarrow\mathfrak{F}).

Definition 2.59.

A holomorphic curve 𝔅\mathfrak{C}\longrightarrow\mathfrak{B} is stable if it has a finite number of automorphisms, and is not a nontrivial refinement of another holomorphic curve. (If 𝔅\mathfrak{B} is basic, this is equivalent to all smooth components of \mathfrak{C} which are mapped to a point in 𝔅\lceil\mathfrak{B}\rceil being stable as punctured Riemann surfaces.)

Definition 2.60.

A family of stable holomorphic curves in 𝔅\mathfrak{B} is a family of smooth curves so that the map restricted to fibers is holomorphic and stable. The moduli stack of stable holomorphic curves in 𝔅\mathfrak{B}, is the substack (𝔅)sm(𝔅)\mathcal{M}(\mathfrak{B})\subset\mathcal{M}^{sm}(\mathfrak{B}) with objects consisting of all families of stable holomorphic curves, and morphisms the same as in sm\mathcal{M}^{sm}.

Definition 2.61.

Given a closed 22 form ω\omega on the exploded 𝕋\mathbb{T} fibration 𝔅\mathfrak{B}, the smooth exploded 𝕋\mathbb{T} curve f:𝔅f:\mathfrak{C}\longrightarrow\mathfrak{B} is ω\omega-stable if the integral of fωf^{*}\omega is nonnegative on any smooth component, and positive on any smooth component which is unstable as a punctured Riemann surface.

Use the notation sm,ω(𝔅)\mathcal{M}^{sm,\omega}(\mathfrak{B}) to indicate the substack of sm\mathcal{M}^{sm} consisting of families of ω\omega-stable smooth curves.

We have (𝔅)sm,ω(𝔅)sm(𝔅)\mathcal{M}(\mathfrak{B})\subset\mathcal{M}^{sm,\omega}(\mathfrak{B})\subset\mathcal{M}^{sm}(\mathfrak{B}). The moduli stack of ω\omega-stable curves is a little better behaved than the moduli stack of stable curves because ω\omega-stable curves have only a finite number of automorphisms.

We want to make compactness statements about holomorphic curves in families, so we generalize the above definitions for a family 𝔅^𝔊\hat{\mathfrak{B}}\longrightarrow\mathfrak{G} as follows:

Definition 2.62.

The moduli stack of smooth curves in a family 𝔅^𝔊\hat{\mathfrak{B}}\longrightarrow\mathfrak{G}, 𝔐sm(𝔅^𝔊)\mathfrak{M}^{sm}(\hat{\mathfrak{B}}\rightarrow\mathfrak{G}) is the substack of 𝔐sm(𝔅^)\mathfrak{M}^{sm}(\hat{\mathfrak{B}}) which is the full subcategory which has as objects families which admit commutative diagrams

(,j)𝔅^𝔉𝔊\begin{array}[]{cll}(\mathfrak{C},j)&\longrightarrow&\hat{\mathfrak{B}}\\ \downarrow&&\downarrow\\ \mathfrak{F}&\longrightarrow&\mathfrak{G}\end{array}

The moduli stack of ω\omega-stable smooth curves sm,ω(𝔅^𝔊)\mathcal{M}^{sm,\omega}(\hat{\mathfrak{B}}\rightarrow\mathfrak{G}) and the moduli stack of stable holomorphic curves (𝔅^𝔊)\mathcal{M}(\hat{\mathfrak{B}}\rightarrow\mathfrak{G}) are defined as the appropriate substacks of sm(𝔅^𝔊)\mathcal{M}^{sm}(\hat{\mathfrak{B}}\rightarrow\mathfrak{G}). Note that there is a morphism sm(𝔅𝔊)𝔊¯\mathcal{M}^{sm}(\mathfrak{B}\rightarrow\mathfrak{G})\longrightarrow\underline{\mathfrak{G}} which sends the object given by the diagram above to 𝔉𝔊\mathfrak{F}\longrightarrow\mathfrak{G}. The appropriate compactness theorem for families states that if we restrict to the part of the moduli space with bounded energy and genus, the map (𝔅^𝔊)𝔊¯\mathcal{M}(\hat{\mathfrak{B}}\rightarrow\mathfrak{G})\longrightarrow\underline{\mathfrak{G}} is topologically proper.

The goal of this paper is to prove that the moduli stack of finite energy stable holomorphic curves with a fixed number of punctures and genus in a complete exploded 𝕋\mathbb{T} fibration with a strict taming is topologically compact. We should say what this means in this context. First, we need a notion of topological convergence. The following definition can be thought of as a notion of what it means for a sequence of points in sm\mathcal{M}^{sm} to converge topologically.

Definition 2.63.

A sequence of points p¯i(𝔅)\underline{p}^{i}\longrightarrow\mathcal{M}(\mathfrak{B}) corresponding to smooth curves fi:i𝔅f^{i}:\mathfrak{C}^{i}\longrightarrow\mathfrak{B} converges topologically to f:𝔅f:\mathfrak{C}\longrightarrow\mathfrak{B} corresponding to p¯(𝔅)\underline{p}\longrightarrow\mathcal{M}(\mathfrak{B}) in C,δC^{\infty,\delta} if there exist a sequence of families

𝔉(^,ji)f^i𝔅\mathfrak{F}\longleftarrow(\mathfrak{\hat{C}},j_{i})\xrightarrow{\hat{f}^{i}}\mathfrak{B}

so that this sequence of families converges in C,δC^{\infty,\delta} to

𝔉(^,j)f^𝔅\mathfrak{F}\longleftarrow(\mathfrak{\hat{C}},j)\xrightarrow{\hat{f}}\mathfrak{B}

and a sequence of maps pip^{i} of a point into 𝔉\mathfrak{F} so that pip^{i} converges to pp in 𝔉\lceil\mathfrak{F}\rceil, fif^{i} is the map given by the restriction of f^i\hat{f}^{i} to the fiber over pip^{i}, and ff is given by the restriction of ff to the fiber over pp.

(It may seem strange that we use C,δC^{\infty,\delta} instead of smooth convergence for the moduli stack of smooth curves. This is an artifact of our methods of proof. It is highly likely that in the case of integrable complex structures the analogous theorem can be proven with smooth convergence replacing C,δC^{\infty,\delta} convergence. It is also perhaps more natural for our case to consider the moduli stack of C,δC^{\infty,\delta} families. The required results in this setting follow trivially from results in the smooth setting. It is expected that \mathcal{M} carries a natural type of C,δC^{\infty,\delta} Kuranishi structure.)

We say that a sequence of points in sm(𝔅^𝔊)\mathcal{M}^{sm}(\hat{\mathfrak{B}}\rightarrow\mathfrak{G}) converge topologically if they converge topologically in sm(𝔅^)\mathcal{M}^{sm}(\hat{\mathfrak{B}}). Topological convergence in \mathcal{M} and sm,ω\mathcal{M}^{sm,\omega} is defined to simply be topological convergence in sm\mathcal{M}^{sm}. We say that a sequence of points in 𝔊¯\underline{\mathfrak{G}} converges topologically if and only if the corresponding sequence of points in 𝔊\mathfrak{G} converges topologically.

Definition 2.64.

To say that \mathcal{M} is topologically compact in C,δC^{\infty,\delta} means that given any sequence of points in \mathcal{M}, there exists a subsequence that converges topologically in C,δC^{\infty,\delta} to some point in \mathcal{M}.

To say that map of stacks f:𝔊¯f:\mathcal{M}\longrightarrow\underline{\mathfrak{G}} is topologically proper in C,δC^{\infty,\delta} is to say that if p¯i\underline{p}^{i}\longrightarrow\mathcal{M} is some sequence of points so that f(p¯i)f(\underline{p}^{i}) converges topologically in 𝔊¯\underline{\mathfrak{G}}, then there exists a subsequence of {p¯i}\{\underline{p}^{i}\} that converges topologically in C,δC^{\infty,\delta} to some point in \mathcal{M}.

In some sense, the moduli stack of stable holomorphic curves should be complete. (The sense in which this is true involves dealing with transversality issues.) We shall not say much about the other requirement for completeness in this paper, but note that if 𝔘𝔗\mathfrak{U}\subset\mathfrak{T} is a connected open subset of 𝔗\mathfrak{T}, and 𝔘¯sm,ω\underline{\mathfrak{U}}\longrightarrow\mathcal{M}^{sm,\omega} is a family of smooth curves so that at least one curve is holomorphic, then the family is holomorphic 𝔘¯\underline{\mathfrak{U}}\longrightarrow\mathcal{M}.

The main theorem of the paper, on page 4.1 can now be understood.

3 Estimates

In this section, we shall prove the analytic estimates required for our compactness theorem. These estimates do not differ much from the standard estimates used to study (pseudo)holomorphic curves in smooth symplectic manifolds. They are also proved using roughly the same methods as in the smooth case. We shall make the following assumptions:

  1. 1.

    We’ll assume that our target 𝔅\mathfrak{B} is complete. This is required to give us the kind of bounded geometry found in the smooth case when the target is compact or has a cylindrical end. We shall also assume that 𝔅\mathfrak{B} is basic. The assumption that 𝔅\mathfrak{B} is basic is mainly for convenience in the arguments that follow. It is most useful in establishing coordinates and notation for the construction of families of smooth curves in the proof of Theorem 4.1.

  2. 2.

    We’ll assume our almost complex structure JJ on 𝔅\mathfrak{B} is civilized. This artificial (but easy to achieve) assumption means that JJ induces a smooth almost complex structure on the smooth parts of 𝔅\mathfrak{B}. This allows us to use standard estimates from the study of holomorphic curves in smooth manifolds with minimal modification. This assumption is needed to have holomorphic curves be smooth morphisms instead of just C,δC^{\infty,\delta} morphisms.

  3. 3.

    We’ll assume that JJ has a strict taming Ω\Omega. This is required to get a local area bound for holomorphic curves. We shall also work with a particular symplectic form ωΩ\omega\in\Omega. This together with JJ defines a pseudo metric on 𝔅\mathfrak{B} which collapses all tropical directions (so it is like a metric on the smooth part 𝔅\lceil\mathfrak{B}\rceil.) The strategy of proof for most of our estimates is roughly as follows: first prove a weak estimate in ω\omega’s pseudo metric, and then improve this to get an estimate in an actual metric on 𝔅\mathfrak{B}. We shall use this strategy to prove estimates for the the derivative at the center of holomorphic disks with bounded Ω\Omega energy and small ω\omega energy, and to get strong estimates on the behavior of holomorphic cylinders with bounded Ω\Omega energy and small ω\omega energy.

This section ends with Proposition 3.17, which is a careful statement of the fact that any holomorphic curve with bounded Ω\Omega energy, genus, and number of punctures can be decomposed into a bounded number of components, which either have ‘bounded conformal structure and bounded derivative’, or are annuli with small ω\omega energy, (and so we have strong estimates on their behavior.)

Throughout this section, we shall use the notation (𝔅,J,Ω,ω,g)(\mathfrak{B},J,\Omega,\omega,g) to indicate a smooth, basic, complete exploded fibration 𝔅\mathfrak{B}, a civilized almost complex structure JJ, a strict taming Ω\Omega, a choice of symplectic taming form ωΩ\omega\in\Omega, and a metric gg. If we say that a constant depends continuously on (𝔅,J,Ω,ω,g)(\mathfrak{B},J,\Omega,\omega,g), we mean that if we have a family 𝔅^𝔊\hat{\mathfrak{B}}\longrightarrow\mathfrak{G} with such a structure on each fiber, the constant can be chosen a continuous function on 𝔊\mathfrak{G}. In what follows, we will generally only prove lemmas for a single exploded fibration 𝔅\mathfrak{B}, and leave the proof of the corresponding statement for families to the diligent reader.

We shall use the following lemma which bounds the geometry of a complete exploded 𝕋\mathbb{T} fibration. Note that given a metric gg on some complete 𝔅\mathfrak{B}, the injectivity radius is a continuous function on 𝔅\mathfrak{B}, and is therefore bounded below.

Lemma 3.1.

Given a complete exploded 𝕋\mathbb{T} fibration 𝔅\mathfrak{B} with a metric gg and any finite number of smooth tensor fields θi\theta_{i} (such as an almost complex structure or a symplectic structure), for any R>0R>0 which is smaller than the injectivity radius of 𝔅\mathfrak{B}, if a sequence of points pn𝔅p_{n}\longrightarrow\mathfrak{B} converges topologically to p𝔅p\longrightarrow\mathfrak{B}, then (BR(pn),g,θi)(B_{R}(p_{n}),g,\theta_{i}) converges to (BR(p),g,θi)(B_{R}(p),g,\theta_{i}) in the following sense:

Considering BR(pi)B_{R}(p_{i}) as a smooth manifold, there exist a sequence of diffeomorphisms fn:BR(pn)BR(p)f_{n}:B_{R}(p_{n})\longrightarrow B_{R}(p) so that fn(g)f_{n}^{*}(g) and fn(θi)f_{n}^{*}(\theta_{i}) converges in CC^{\infty} on compact subsets of BR(pn)B_{R}(p_{n}) to gg and θi\theta_{i}.

Proof:

This follows from the fact that all smooth tensor fields (and their derivatives) can be locally written as some sum of smooth functions on 𝔅\mathfrak{B} times some basis tensor fields.

\square

This tells us that the geometry of a complete exploded 𝕋\mathbb{T} fibration is bounded in the same way that the geometry of a compact manifold is bounded. The analogous result for families is also true.

The following is a consequence of our bounded geometry and the standard monotonicity lemma for holomorphic curves first proved in [5]. (The proof involves a bound on the covariant derivatives of JJ, the curvature of gg and the injectivity radius, all of which vary continuously in families.)

Lemma 3.2.

Given (𝔅,J,Ω,g)(\mathfrak{B},J,\Omega,g), for any ϵ>0\epsilon>0 there exists some E>0E>0 depending continuously on (𝔅,J,Ω,g)(\mathfrak{B},J,\Omega,g) so that any non constant JJ holomorphic curve which passes through a point pp and which is a complete map when restricted to the ϵ\epsilon ball around pp has Ω\Omega energy greater than EE.

Lemma 3.3.

Given (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega), define the pseudo-metric

v,wω:=12(ω(v,Jw)+ω(w,Jv))\langle v,w\rangle_{\omega}:=\frac{1}{2}(\omega(v,Jw)+\omega(w,Jv))

For any such (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega) and a choice of ϵ>0\epsilon>0, there exists some EE depending continuously on (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega) so that any non constant JJ holomorphic curve which passes through a point pp and which is a complete map when restricted to the ϵ\epsilon ball around pp in the above pseudo metric has ω\omega energy greater than EE.

Proof:

The fact that 𝔅\mathfrak{B} is basic and complete means in particular that it can be covered by a finite number of coordinate charts, U2n×𝔗AmU\subset\mathbb{R}^{2n}\times\mathfrak{T}^{m}_{A}. 2n×𝔗Am\mathbb{R}^{2n}\times\mathfrak{T}^{m}_{A} with JJ and ω\omega can be regarded as subsets of more standard charts 2n×𝔗lk\mathbb{R}^{2n}\times\mathfrak{T}^{k}_{l} cut out by setting some monomials equal to 11. This has the advantage that the smooth part of this is just equal to 2n×l\mathbb{R}^{2n}\times\mathbb{C}^{l}, and we can assume our symplectic and almost complex structures just come from smooth ones here (this uses that JJ is civilized), so we can use standard holomorphic curve results. Note also that the above pseudo metric is actually a metric on this smooth part. The size of covariant derivatives of JJ in this metric give continuous functions on 𝔅\mathfrak{B}, and are therefore bounded because 𝔅\mathfrak{B} is complete. To have the bounded geometry required to apply the standard monotonicity lemma for holomorphic curves, we also need some kind of injectivity radius estimate. For a point qUq\in\lceil U\rceil in the smooth part of a coordinate chart, we can define iU(q)i_{U}(q) to be the injectivity radius at qq of U\lceil U\rceil with our metric. We can then define an injectivity radius for a point p[𝔅]p\in[\mathfrak{B}] to be given by

i(p):=infq(d(p,q)+maxUiU(q))i(p):=\inf_{q}\left(d(p,q)+\max_{U}i_{U}(q)\right)

where d(p,q)d(p,q) indicates distance in our pseudo metric. This defines a continuous function on 𝔅\mathfrak{B}, and at each point i(p)>0i(p)>0, so the fact that 𝔅\mathfrak{B} is complete tells us that there is some lower bound 0<ci(p)0<c\leq i(p).

To prove our lemma we can now take 0<ϵ<c0<\epsilon<c. If our holomorphic curve is complete restricted to the ϵ\epsilon ball around pp, it is also complete (which implies that the maps of smooth components are proper maps) when restricted to the ϵ\epsilon ball around any qUq\in\lceil U\rceil so that d(p,q)=0d(p,q)=0. We can apply the standard monotonicity lemma for holomorphic curves to these balls, so we can choose some E>0E>0 so that any holomorphic curve passing through qq which is complete restricted to the ball around qq is either constant in U\lceil U\rceil or has energy greater than EE. So either our holomorphic curve must have energy greater than EE or it must have some connected component which is constant in the smooth part of every coordinate chart. This connected component must therefore not touch the boundary of our ϵ\epsilon ball, and must be a complete map to 𝔅\mathfrak{B}. The fact that Ω\Omega is a strict taming then allows us to use Lemma 3.2 to complete our proof.

\square

The above proof also implies the following:

Lemma 3.4.

Given (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega), there exists some ϵ>0\epsilon>0 which can be chosen to depend continuously on (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega) so that given any complete holomorphic curve in 𝔅\mathfrak{B}, the ω\omega energy of any smooth component is either greater than ϵ\epsilon, or the image of that smooth component has zero size in the ω\omega pseudo-metric.

Lemma 3.5.

Given (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega), for any ϵ>0\epsilon>0, there exists some E>0E>0 depending continuously on (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega) so that any JJ holomorphic map ff of e(R+1)<|z|<e(R+1)e^{-(R+1)}<\left\lvert z\right\rvert<e^{(R+1)} with ω\omega-energy less than EE is contained inside a ball of radius ϵ\epsilon (in ω\omega’s pseudo metric) on the smaller annulus eR|z|eRe^{-R}\leq\left\lvert z\right\rvert\leq e^{R}.

Proof:

Suppose to the contrary that this lemma is false.

Choose EE small enough so that Lemma 3.3 holds for ϵ8\frac{\epsilon}{8} balls. There must be a path in e(R+1)<|z|<eR+1e^{-(R+1)}<\left\lvert z\right\rvert<e^{R+1} joining z=1z=1 with an end of this annulus contained entirely inside the ball Bϵ8(f(1))B_{\frac{\epsilon}{8}}(f(1)). (Otherwise, ff restricted to some subset would be a proper map to Bϵ8(f(1))B_{\frac{\epsilon}{8}}(f(1)), and we could use Lemma 3.3 to get a contradiction.) Suppose without loss of generality that it connects 11 with the circle |z|=e(R+1)\left\lvert z\right\rvert=e^{-(R+1)}.

Suppose for a second that there exists some point zz so that e(R+34)<|z|<e(R+14)e^{-(R+\frac{3}{4})}<\left\lvert z\right\rvert<e^{-(R+\frac{1}{4})} and f(z)B3ϵ8(f(1))f(z)\notin B_{\frac{3\epsilon}{8}}(f(1)). As above, there must exist some path connecting zz with an end which is contained in the ball Bϵ8(f(1))B_{\frac{\epsilon}{8}}(f(1)). There must therefore be some region conformal to [0,π]×[0,14][0,\pi]\times[0,\frac{1}{4}] so that the dist(f(0,t),f(π,t))>ϵ8\operatorname{dist}\left(f(0,t),f(\pi,t)\right)>\frac{\epsilon}{8}. Then the Cauchy-Schwartz inequality tells us that

0π014|df|ω2>14(ϵ8)2π\int_{0}^{\pi}\int_{0}^{\frac{1}{4}}\left\lvert df\right\rvert^{2}_{\omega}>\frac{1}{4}\frac{\left(\frac{\epsilon}{8}\right)^{2}}{\pi}

This in turn gives a uniform lower bound for the energy of our curve. This case can therefore be discarded, and we can assume that all zz satisfying e(R+34)<|z|<e(R+14)e^{-(R+\frac{3}{4})}<\left\lvert z\right\rvert<e^{-(R+\frac{1}{4})} are contained in B3ϵ8(f(1))B_{\frac{3\epsilon}{8}}(f(1)).

As we are assuming this lemma is false, there must be some point z2z_{2} so that eR|z2|eRe^{-R}\leq\left\lvert z_{2}\right\rvert\leq e^{R} and f(z2)Bϵ(f(1))f(z_{2})\notin B_{\epsilon}(f(1)). Repeating the above argument (with z2z_{2} in place of the point z=1z=1) gives us that zz must be contained in B3ϵ8(f(z2))B_{\frac{3\epsilon}{8}}(f(z_{2})) when eR+14|z|eR+34e^{R+\frac{1}{4}}\leq\left\lvert z\right\rvert\leq e^{R+\frac{3}{4}}.

We then have ff contained in balls which are at least ϵ4\frac{\epsilon}{4} apart on the boundary of an annulus, thus we can apply Lemma 3.3 to some point in the image under ff of the interior at least ϵ8\frac{\epsilon}{8} from both balls to obtain a lower bound for the energy and a contradiction to the assumption that the lemma was false.

\square

The following gives useful coordinates for the analysis of holomorphic curves:

Lemma 3.6.

Given (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega), there exist constants cic_{i} and ϵ>0\epsilon>0 depending continuously on (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega) so that for all points p𝔅p\longrightarrow\mathfrak{B}, there exists some coordinate neighborhood UU of pp so that the smooth part, U\lceil U\rceil can be identified with a relatively closed subset of the open ϵ\epsilon ball in n\mathbb{C}^{n} with almost complex structure J^\hat{J} and some flat J^\hat{J} preserving connection \nabla so that

  1. 1.

    pp is sent to 0, and the restriction of J^\hat{J} to U\lceil U\rceil is JJ.

  2. 2.

    The metric ,ω\langle\cdot,\cdot\rangle_{\omega} on U\lceil U\rceil is close to the standard flat metric on n\mathbb{C}^{n} so that

    12v,vω|v|22v,vω\frac{1}{2}\langle v,v\rangle_{\omega}\leq\left\lvert v\right\rvert^{2}\leq 2\langle v,v\rangle_{\omega}
  3. 3.

    J^\hat{J} at the point znz\in\mathbb{C}^{n} converges to the standard complex structure J0J_{0} on n\mathbb{C}^{n} as z0z\rightarrow 0 in the sense that

    J^J0c0|z|\left\lVert\hat{J}-J_{0}\right\rVert\leq c_{0}\left\lvert z\right\rvert
  4. 4.

    The torsion tensor

    𝐓(v,w):=vwwv[v,w]\mathbf{T}_{\nabla}(v,w):=\nabla_{v}w-\nabla_{w}v-[v,w]

    is bounded by c0c_{0}, and has its kkth derivatives bounded by ckc_{k}.

The only point in this lemma which does not follow from the definition of a civilized almost complex structure and calculation in local coordinates is the fact that the constants involved do not have to depend on the point pp. This follows from the fact that 𝔅\mathfrak{B} is complete.

Lemma 3.7.

Given a holomorphic map ff of the unit disk to a space with a flat JJ preserving connection \nabla, fxf_{x} defines a map to n\mathbb{C}^{n} defined by parallel transporting fxf_{x} back to the tangent space at f(0)f(0) which we then identify with n\mathbb{C}^{n}. Such an fxf_{x} satisfies the following equation involving the torsion tensor of \nabla, 𝐓\mathbf{T}_{\nabla}.

¯fx=12J𝐓(fy,fx)\bar{\partial}f_{x}=\frac{1}{2}J\mathbf{T}_{\nabla}(f_{y},f_{x})

Proof:

¯fx=12(fxfx+Jfyfx)=12fx(fx+Jfy)+12J(fyfxfxfy)=12J𝐓(fy,fx)\begin{split}\bar{\partial}f_{x}&=\frac{1}{2}(\nabla_{f_{x}}f_{x}+J\nabla_{f_{y}}f_{x})\\ &=\frac{1}{2}\nabla_{f_{x}}(f_{x}+Jf_{y})+\frac{1}{2}J(\nabla_{f_{y}}f_{x}-\nabla_{f_{x}}f_{y})\\ &=\frac{1}{2}J\mathbf{T}_{\nabla}(f_{y},f_{x})\end{split}

\square

As the ¯\bar{\partial} above is the standard linear ¯\bar{\partial} operator, this expression is good for applying the following standard elliptic regularity lemma for the linear ¯\bar{\partial} equation. This allows us to get bounds on higher derivatives from bounds on the first derivative of holomorphic functions.

Lemma 3.8.

For for a given number kk and 1<p<1<p<\infty, there exists a constant cc so that given any map ff from the unit disk D(1)D(1),

fLk+1p(D(12))c(¯fLkp(D(1))+fLkp(D(1)))\left\lVert f\right\rVert_{L_{k+1}^{p}(D(\frac{1}{2}))}\leq c\left(\left\lVert\bar{\partial}f\right\rVert_{L_{k}^{p}(D(1))}+\left\lVert f\right\rVert_{L_{k}^{p}(D(1))}\right)
Lemma 3.9.

Given (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega), there exists some energy EE, some distance r>0r>0, and a constant cc, each depending continuously on (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega) so that any holomorphic map ff of a disk {|z|1}\{\left\lvert z\right\rvert\leq 1\} into 𝔅\mathfrak{B} with ω\omega-energy less than EE or contained inside a ball of ω\omega radius rr, satisfies

|df|ω<c at z=0\left\lvert df\right\rvert_{\omega}<c\text{ at }z=0

We shall omit the proof of the above Lemma which is a standard bubbling argument, similar, but easier than the proof of the following:

Lemma 3.10.

Given (𝔅,J,Ω,ω,g)(\mathfrak{B},J,\Omega,\omega,g), for any E>0E>0, there exists some ϵ>0\epsilon>0, distance r>0r>0 and constant cc, each depending continuously on (𝔅,J,Ω,ω,g)(\mathfrak{B},J,\Omega,\omega,g) so that any holomorphic map ff of a disk |z|1\left\lvert z\right\rvert\leq 1 into 𝔅\mathfrak{B} with Ω\Omega-energy less than EE, and either contained in a ω\omega-ball of radius rr or having ω\omega-energy less than ϵ\epsilon satisfies

|df|g<c at z=0\left\lvert df\right\rvert_{g}<c\text{ at }z=0

Proof:

Suppose that this lemma was false. Then there would exist some sequence of maps fif_{i} satisfying the above conditions with |dfi(0)|\left\lvert df_{i}(0)\right\rvert\rightarrow\infty. First, we obtain a sequence of rescaled JJ holomorphic maps f~i:D(Ri)𝔅\tilde{f}_{i}:D(R_{i})\longrightarrow\mathfrak{B} from the standard complex disk of radius RiR_{i} so that

|df~i|2\left\lvert d\tilde{f}_{i}\right\rvert\leq 2
|df~i(0)|=1\left\lvert d\tilde{f}_{i}(0)\right\rvert=1
limiRi=\lim_{i\rightarrow\infty}R_{i}=\infty

We achieve this in the same way as in any standard bubbling off argument such as the proof of lemma 5.11 in [1]. We can then use Lemma 3.6, Lemma 3.7, and Lemma 3.8 to get a bound on the higher derivatives of f~i\tilde{f}_{i} on D(Ri1)D(R_{i}-1).

As 𝔅\mathfrak{B} is complete, we can choose a subsequence so that fi(0)f_{i}(0) converges topologically to some p𝔅p\longrightarrow\mathfrak{B}. Lemma 3.1 tells us that the geometry of (g,J,ω)(g,J,\omega) around fi(0)f_{i}(0) converges to that around pp. Considering all our maps as maps sending 0 to pp, we can choose a subsequence that converges on compact subsets to a non constant holomorphic map f:𝔅f:\mathbb{C}\longrightarrow\mathfrak{B}. Note that ff either has ω\omega-energy less than ϵ\epsilon or is contained in a ball of radius rr in the ω\omega pseudo-metric. Because ff must have finite ω\omega energy, we can use Lemma 3.5 to tell us that ff must converge in the ω\omega pseudo-metric as |z|\left\lvert z\right\rvert\rightarrow\infty. Then by choosing ϵ\epsilon or rr small enough, we can prove using Lemma 3.3 and the standard removable singularity theorem for holomorphic curves, that ff must actually have zero size in the ω\omega pseudo-metric because ff either has ω\omega energy less than ϵ\epsilon or is contained in an ω\omega ball of radius rr.

Now that we have that the image of ff has zero size in the ω\omega pseudo metric, we know that ff must be contained entirely within some ()n(\mathbb{C}^{*})^{n} worth of points over a single topological point in [𝔅][\mathfrak{B}]. This ()n(\mathbb{C}^{*})^{n} has the standard complex structure, so ff gives us nn entire holomorphic maps from \mathbb{C}\longrightarrow\mathbb{C}^{*}. As ff is non constant, at least one of these maps must be non constant. This map must have infinite degree, as it must have dense image in the universal cover of \mathbb{C}^{*} which is just the usual complex plane. It therefore has infinite Ω\Omega-energy, contradicting the fact that it must have Ω\Omega-energy less than EE.

To prove the family case, it is important to note that (as stated at the start of this section), when we say our constants depend continuously on (𝔅,J,Ω,ω,g)(\mathfrak{B},J,\Omega,\omega,g), we mean that given any finite dimensional family, (𝔅^𝔊,J,Ω,ω,g)(\hat{\mathfrak{B}}\longrightarrow\mathfrak{G},J,\Omega,\omega,g), the constants can be chosen to depend continuously on 𝔊\mathfrak{G}. With this understood, the proof in the family case is analogous to the above proof.

\square

Lemma 3.11.

Given (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega), for any 0δ<10\leq\delta<1, there exists an energy bound E>0E>0, a distance r>0r>0, and a constant cc, each depending continuously on (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega) so that any holomorphic map ff from the annulus e(R+1)<|z|<e(R+1)e^{-(R+1)}<\left\lvert z\right\rvert<e^{(R+1)} to 𝔅\mathfrak{B} with ω\omega energy less than EE, or contained inside a ball of ω\omega radius rr satisfies the following inequality

dist(f(z),f(1))ωceδR(|z|δ+|z|δ) for eR|z|eR\operatorname{dist}(f(z),f(1))_{\omega}\leq ce^{-\delta R}(\left\lvert z\right\rvert^{\delta}+\left\lvert z\right\rvert^{-\delta})\text{ for }e^{-R}\leq\left\lvert z\right\rvert\leq e^{R}

Actually, with a more careful argument, this is true with exponent δ=1\delta=1, but we will only prove the easier case.

Proof:

For this proof we shall use coordinates z=et+iθz=e^{t+i\theta}, and the cylindrical metric in which {t,θ}\{\frac{\partial}{\partial t},\frac{\partial}{\partial\theta}\} are an orthonormal frame.

We choose our energy bound E>0E>0 small enough that we can use the derivative bound from Lemma 3.9, and Lemma 3.5 implies that the smaller annulus is contained a small enough ball that we can use the coordinates of Lemma 3.6 (we also choose rr small enough that this is true). We then have the following estimate for the (standard) ¯\bar{\partial} of ff in these coordinates.

|¯f|c1|df||f|\left\lvert\bar{\partial}f\right\rvert\leq c_{1}\left\lvert df\right\rvert\left\lvert f\right\rvert

We can run this through the inequality from Lemma 3.8 to obtain the following estimate on ¯f\bar{\partial}f restricted to the interior of a disk (which holds on the interior of the cylinder where we have the derivative bound from Lemma 3.9.)

¯fLp(D(12))c1fL1pD(12)fL(D(12))c2fL(D(1))fL(D(12))\left\lVert\bar{\partial}f\right\rVert_{L^{p}(D(\frac{1}{2}))}\leq c_{1}\left\lVert f\right\rVert_{L^{p}_{1}D(\frac{1}{2})}\left\lVert f\right\rVert_{L^{\infty}(D(\frac{1}{2}))}\leq c_{2}\left\lVert f\right\rVert_{L^{\infty}(D(1))}\left\lVert f\right\rVert_{L^{\infty}(D(\frac{1}{2}))} (1)

Here c2c_{2} is a constant depending only on pp and (𝔅,J,ω)(\mathfrak{B},J,\omega). We can fix pp to be something bigger than 22. By choosing EE or rr small, we can force |f|\left\lvert f\right\rvert to be as small as we like on the smaller cylinder using Lemma 3.5.

Now we can use Cauchy’s integral formula

2πif(z0)=|z|=1f(z)zz0𝑑z+|z|=e2lf(z)zz0𝑑z+1|z|e2l¯f(z)zz0dz2\pi if(z_{0})=-\int_{\left\lvert z\right\rvert=1}\frac{f(z)}{z-z_{0}}dz+\int_{\left\lvert z\right\rvert=e^{2l}}\frac{f(z)}{z-z_{0}}dz+\int_{1\leq\left\lvert z\right\rvert\leq e^{2l}}\frac{\bar{\partial}f(z)}{z-z_{0}}\wedge dz

or in our coordinates,

f(t0,θ0)=12π02πf(0,θ)1et0ei(θθ0)𝑑θ+12π02πf(2l,θ)1et02lei(θθ0)𝑑θ+12π02π02l¯f(θ,t)1et0tei(θθ0)𝑑t𝑑θ\begin{split}f(t_{0},\theta_{0})&=-\frac{1}{2\pi}\int_{0}^{2\pi}\frac{f(0,\theta)}{1-e^{t_{0}}e^{i(\theta-\theta_{0})}}d\theta+\frac{1}{2\pi}\int_{0}^{2\pi}\frac{f(2l,\theta)}{1-e^{t_{0}-2l}e^{i(\theta-\theta_{0})}}d\theta\\ &+\frac{1}{2\pi}\int_{0}^{2\pi}\int_{0}^{2l}\frac{\bar{\partial}f(\theta,t)}{1-e^{t_{0}-t}e^{i(\theta-\theta_{0})}}dtd\theta\end{split}

Let us now consider each term of this expression for f(l,θ0)f(l,\theta_{0}) in the middle of a cylinder under the assumption that the average of f(2l,θ)f(2l,\theta) is 0.

The first term:

|12π02πf(0,θ)1elei(θθ0)𝑑θ|1el1max|f(0,θ)|\left\lvert\frac{1}{2\pi}\int_{0}^{2\pi}\frac{f(0,\theta)}{1-e^{l}e^{i(\theta-\theta_{0})}}d\theta\right\rvert\leq\frac{1}{e^{l}-1}\max\left\lvert f(0,\theta)\right\rvert

The second term using that the average of f(2l,θ)f(2l,\theta) is 0,

|12π02πf(2l,θ)1el2lei(θθ0)𝑑θ|=|12π02πf(2l,θ)(11elei(θθ0)1)𝑑θ|=|12π02πf(2l,θ)(elei(θθ0)1elei(θθ0))𝑑θ|1el1max|f(2l,θ)|\begin{split}\left\lvert\frac{1}{2\pi}\int_{0}^{2\pi}\frac{f(2l,\theta)}{1-e^{l-2l}e^{i(\theta-\theta_{0})}}d\theta\right\rvert&=\left\lvert\frac{1}{2\pi}\int_{0}^{2\pi}f(2l,\theta)\left(\frac{1}{1-e^{-l}e^{i(\theta-\theta_{0})}}-1\right)d\theta\right\rvert\\ &=\left\lvert\frac{1}{2\pi}\int_{0}^{2\pi}f(2l,\theta)\left(\frac{e^{-l}e^{i(\theta-\theta_{0})}}{1-e^{-l}e^{i(\theta-\theta_{0})}}\right)d\theta\right\rvert\\ &\leq\frac{1}{e^{l}-1}\max\left\lvert f(2l,\theta)\right\rvert\end{split}

The third term:

|12π02π02l¯f(θ,t)1et0tei(θθ0)𝑑t𝑑θ|¯fL312π11eltei(θθ0)L32c3(l+1)(maxt[1,2l+1]|f(t,θ)|)(maxt[0,2l]|f(t,θ)|)\begin{split}\left\lvert\frac{1}{2\pi}\int_{0}^{2\pi}\int_{0}^{2l}\frac{\bar{\partial}f(\theta,t)}{1-e^{t_{0}-t}e^{i(\theta-\theta_{0})}}dtd\theta\right\rvert&\leq\left\lVert\bar{\partial}f\right\rVert_{L^{3}}\frac{1}{2\pi}\left\lVert\frac{1}{1-e^{l-t}e^{i(\theta-\theta_{0})}}\right\rVert_{L^{\frac{3}{2}}}\\ &\leq c_{3}(l+1)\left(\max_{t\in[-1,2l+1]}\left\lvert f(t,\theta)\right\rvert\right)\left(\max_{t\in[0,2l]}\left\lvert f(t,\theta)\right\rvert\right)\end{split}

The constant c3c_{3} depends only on (𝔅,J,ω)(\mathfrak{B},J,\omega). It is zero if JJ is integrable.

To summarize the above, we have the following expression which holds if the average of f(θ,2l)f(\theta,2l) is 0. (It also holds if the average of f(θ,0)f(\theta,0) is 0 due to the symmetry of the cylinder.)

|f(l,θ)|1el1(max|f(0,θ)|+max|f(2l,θ)|)+c3(l+1)(maxt[1,2l+1]|f(t,θ)|)(maxt[0,2l]|f(t,θ)|)\begin{split}\left\lvert f(l,\theta)\right\rvert\leq&\frac{1}{e^{l}-1}\left(\max\left\lvert f(0,\theta)\right\rvert+\max\left\lvert f(2l,\theta)\right\rvert\right)\\ &+c_{3}(l+1)\left(\max_{t\in[-1,2l+1]}\left\lvert f(t,\theta)\right\rvert\right)\left(\max_{t\in[0,2l]}\left\lvert f(t,\theta)\right\rvert\right)\end{split} (2)

The amount that the average of f(t0,θ)f(t_{0},\theta) changes with t0t_{0} inside [0,2l][0,2l] is determined by the integral of ¯f\bar{\partial}f, which is dominated as above by a term of the form

change in average c4l(maxt[1,2l+1]|f(t,θ)|)(maxt[0,2l]|f(t,θ)|)\text{change in average }\leq c_{4}l\left(\max_{t\in[-1,2l+1]}\left\lvert f(t,\theta)\right\rvert\right)\left(\max_{t\in[0,2l]}\left\lvert f(t,\theta)\right\rvert\right) (3)

Now let’s put these estimates into slightly more invariant terms: Define the variation of f(t,θ)f(t,\theta) on [a,b]×𝕋1[a,b]\times\mathbb{T}^{1} as follows:

Vf([a,b]):=maxt1,t2[a,b]dist(f(t1,θ1),f(t2,θ2))ωVf([a,b]):=\max_{t_{1},t_{2}\in[a,b]}\operatorname{dist}(f(t_{1},\theta_{1}),f(t_{2},\theta_{2}))_{\omega}

Now the above two estimates give the following:

Vf([a,a+l])(8el1+c5(l+1)Vf([al1,a+2l+1]))Vf([al,a+2l])Vf([a,a+l])\leq\left(\frac{8}{e^{l}-1}+c_{5}(l+1)Vf([a-l-1,a+2l+1])\right)Vf([a-l,a+2l])

Now if we choose ll large enough, and then make VfVf small enough by making our energy bound EE small (or directly making rr small), we can get the estimate

Vf([a,a+l])eδl3Vf([al,a+2l]) for [al1,a+2l+1][R,R]Vf([a,a+l])\leq\frac{e^{-\delta l}}{3}Vf([a-l,a+2l])\text{ for }[a-l-1,a+2l+1]\subset[-R,R]

Applying this 33 times, we get that on the appropriate part of the cylinder,

Vf([al,a+2l])eδlVf([a2l,a+3l])Vf([a-l,a+2l])\leq e^{-\delta l}Vf([a-2l,a+3l])
 so Vf([a,a+l])eδ2l3Vf([a2l,a+3l])\text{ so }Vf([a,a+l])\leq\frac{e^{-\delta 2l}}{3}Vf([a-2l,a+3l])

Inductively continuing this argument gives that if [anl,a+(n+1)l][R+1,R1][a-nl,a+(n+1)l]\subset[-R+1,R-1],

Vf([a,a+l])eδnl3Vf([anl,a+(n+1)l])Vf([a,a+l])\leq\frac{e^{-\delta nl}}{3}Vf([a-nl,a+(n+1)l])

The required estimate follows from this.

\square

Lemma 3.8, Lemma 3.7, and the coordinates from Lemma 3.6 from page 3.6 can be used with inequality 1 on page 1 in a standard fashion to get a similar estimate on the derivative of ff in the ω\omega pseudo metric. This gives the following corollary:

Corollary 3.12.

Given (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega), there exists some energy E>0E>0 depending continuously on (𝔅,J,Ω,ω)(\mathfrak{B},J,\Omega,\omega) so that given any c>0c>0, there exists some distance RR depending continuously on (𝔅,J,Ω,ω,c)(\mathfrak{B},J,\Omega,\omega,c) so that given any holomorphic map ff of a cylinder elR<|z|<el+Re^{-l-R}<\left\lvert z\right\rvert<e^{l+R} to 𝔅\mathfrak{B} with ω\omega energy less than EE, the ω\omega energy of ff restricted to el<|z|<ele^{-l}<\left\lvert z\right\rvert<e^{l} is less than cEcE.

Lemma 3.13.

Given (𝔅,J,Ω,g)(\mathfrak{B},J,\Omega,g), and any energy bound E<E<\infty and exponent 0δ<10\leq\delta<1, there exists a covering of 𝔅\mathfrak{B} by a finite number of coordinate charts and a constant cc so that the following is true:

Given any holomorphic map ff of Ω\Omega energy less than EE from a cylinder e(R+1)|z|e(R+1)e^{-(R+1)}\leq\left\lvert z\right\rvert\leq e^{(R+1)} to 𝔅\mathfrak{B} contained inside a coordinate chart, there exists a map FF given in coordinates as

F(z):=(c1zα1𝔱a1,,ckzαk𝔱ak,ck+1,,cn)F(z):=(c_{1}z^{\alpha_{1}}\mathfrak{t}^{a_{1}},\dotsc,c_{k}z^{\alpha_{k}}\mathfrak{t}^{a_{k}},c_{k+1},\dotsc,c_{n})

so that

dist(f(z)F(z))ceδR(|z|δ+|z|δ) for eR|z|eR\operatorname{dist}\left(f(z)-F(z)\right)\leq ce^{-\delta R}\left(\left\lvert z\right\rvert^{\delta}+\left\lvert z\right\rvert^{-\delta}\right)\text{ for }e^{-R}\leq\left\lvert z\right\rvert\leq e^{R}

where for i[1,k]i\in[1,k], cic_{i}\in\mathbb{C}^{*}, aia_{i}\in\mathbb{R}, αi\alpha_{i}\in\mathbb{Z}, and for i[k+1,n]i\in[k+1,n], cic_{i}\in\mathbb{R}. dist\operatorname{dist} indicates distance in the metric gg.

In the case that we a have a family (𝔅^𝔊,J,Ω,g)(\hat{\mathfrak{B}}\longrightarrow\mathfrak{G},J,\Omega,g) with the above structure, we can choose cc continuous on 𝔊\mathfrak{G} and a set of coordinate charts 𝔘^\hat{\mathfrak{U}} on 𝔅^\hat{\mathfrak{B}} so that the above coordinate charts on any fiber are the non empty intersections of these with the fiber.

Proof:

First, choose any symplectic form ωΩ\omega\in\Omega. Because 𝔅\mathfrak{B} is basic and complete, we can then choose a finite number of coordinate charts covering 𝔅\mathfrak{B} which are small enough in the ω\omega pseudo metric so that Lemma 3.11 tells us that for any holomorphic map from a cylinder as above contained inside one of these coordinate charts satisfies

distω(f(z),f(1))ceδR(|z|δ+|z|δ)\operatorname{dist}_{\omega}(f(z),f(1))\leq ce^{-\delta^{\prime}R}(\left\lvert z\right\rvert^{\delta^{\prime}}+\left\lvert z\right\rvert^{-\delta^{\prime}}) (4)

where δ=δ+12\delta^{\prime}=\frac{\delta+1}{2}.

In our coordinates,

f(1)=(c1𝔱a1,,ck𝔱ak,ck+1,,cn)f(1)=(c_{1}\mathfrak{t}^{a_{1}},\dotsc,c_{k}\mathfrak{t}^{a_{k}},c_{k+1},\dotsc,c_{n})

We can choose αi\alpha_{i} so that the winding numbers of the first kk coordinates of f(eiθ)f(e^{i\theta}) are the same as our model map FF:

F:=(c1zα1𝔱a1,,ckzαk𝔱ak,ck+1,,cn)F:=(c_{1}z^{\alpha_{1}}\mathfrak{t}^{a_{1}},\dotsc,c_{k}z^{\alpha_{k}}\mathfrak{t}^{a_{k}},c_{k+1},\dotsc,c_{n})

The metric gg can be compared to the pseudo metric from ω\omega on the last coordinates ck+1,,cnc_{k+1},\dotsc,c_{n}, so we only need to prove convergence in the first kk coordinates. Let us do this for ff restricted to the first coordinate f1f_{1}.

For this, use coordinates z=et+iθz=e^{t+i\theta} on the domain with the usual cylindrical metric and the similar cylindrical metric on our target. (Choose our coordinate charts so that this metric on the target is comparable to gg). Use the notation

f1(z)c1zα1𝔱a1=eh(z)\frac{f_{1}(z)}{c_{1}z^{\alpha_{1}}\mathfrak{t}^{a_{1}}}=e^{h(z)}

where hh is a \mathbb{C} valued function so that h(0)=0h(0)=0. The metric we are using on the target is the standard Euclidean metric on \mathbb{C}. Our goal is to prove the appropriate estimate for hh.

First note that Lemma 3.8 and Lemma 3.7 can be used with the inequality 4 in a standard fashion to get a similar estimate on the derivative of ff in the ω\omega pseudo metric.

|df(z)|ωceδR(|z|δ+|z|δ) for eR|z|eR\left\lvert df(z)\right\rvert_{\omega}\leq ce^{-\delta^{\prime}R}(\left\lvert z\right\rvert^{\delta^{\prime}}+\left\lvert z\right\rvert^{-\delta^{\prime}})\text{ for }e^{-R}\leq\left\lvert z\right\rvert\leq e^{R} (5)

(The cc in the above inequality is some new constant which is independent of f). We can choose our coordinate charts so that if J0J_{0} indicates the standard complex structure, there exists some constant MM so that

|JvJ0v|gM|v|ω\left\lvert Jv-J_{0}v\right\rvert_{g}\leq M{\left\lvert v\right\rvert_{\omega}}

This follows from writing down JJ in coordinates, and the fact that our ω\omega pseudo metric is a metric on the smooth part of our coordinate chart. This then tells us that |¯f1|g\left\lvert\bar{\partial}f_{1}\right\rvert_{g} in our coordinates is controlled by |df|ω\left\lvert df\right\rvert_{\omega}. This then gives that for some new constant cc independent of ff we get the following inequality:

|¯h(z)|ceδR(|z|δ+|z|δ) for eR|z|eR\left\lvert\bar{\partial}h(z)\right\rvert\leq ce^{-\delta^{\prime}R}(\left\lvert z\right\rvert^{\delta^{\prime}}+\left\lvert z\right\rvert^{-\delta^{\prime}})\text{ for }e^{-R}\leq\left\lvert z\right\rvert\leq e^{R} (6)

If we choose our coordinate charts small enough in the ω\omega pseudo-metric, Lemma 3.10 gives us a bound for |df|g\left\lvert df\right\rvert_{g} on the interior of the cylinder, so we have that there exists some cc independent of ff so that

|dh(z)|<c for eR|z|eR\left\lvert dh(z)\right\rvert<c\text{ for }e^{-R}\leq\left\lvert z\right\rvert\leq e^{R} (7)

We can now proceed roughly as we did in the proof of Lemma 3.11. In particular, Cauchy’s integral theorem tells us that

h(t0,θ0)=12π02πh(t0l,θ)1elei(θθ0)𝑑θ+12π02πh(t0+l,θ)1elei(θθ0)𝑑θ+12π02πt0lt0+l¯h(t,θ)1et0tei(θθ0)𝑑t𝑑θ\begin{split}h(t_{0},\theta_{0})&=-\frac{1}{2\pi}\int_{0}^{2\pi}\frac{h(t_{0}-l,\theta)}{1-e^{l}e^{i(\theta-\theta_{0})}}d\theta+\frac{1}{2\pi}\int_{0}^{2\pi}\frac{h(t_{0}+l,\theta)}{1-e^{-l}e^{i(\theta-\theta_{0})}}d\theta\\ &+\frac{1}{2\pi}\int_{0}^{2\pi}\int_{t_{0}-l}^{t_{0}+l}\frac{\bar{\partial}h(t,\theta)}{1-e^{t_{0}-t}e^{i(\theta-\theta_{0})}}dtd\theta\end{split}

We can now bound each term in the above expression as in the proof of Lemma 3.11, except we use the estimate 6 to bound |¯h|\left\lvert\bar{\partial}h\right\rvert.

Then we have the following estimate:

|h(t0,θ0)12π02πh(t0+l,θ)𝑑θ|1el1max|h(t0l,θ)12π02πh(t0+l,θ)𝑑θ|+1ellmax|h(t+l,θ)12π02πh(t0+l,θ)𝑑θ|+c(l+1)eδReδl(eδt0+eδt0)\begin{split}\left\lvert h(t_{0},\theta_{0})-\frac{1}{2\pi}\int_{0}^{2\pi}h(t_{0}+l,\theta)d\theta\right\rvert\leq&\frac{1}{e^{l}-1}\max\left\lvert h(t_{0}-l,\theta)-\frac{1}{2\pi}\int_{0}^{2\pi}h(t_{0}+l,\theta)d\theta\right\rvert\\ &+\frac{1}{e^{l}-l}\max\left\lvert h(t+l,\theta)-\frac{1}{2\pi}\int_{0}^{2\pi}h(t_{0}+l,\theta)d\theta\right\rvert\\ &+c(l+1)e^{-\delta^{\prime}R}e^{\delta^{\prime}l}\left(e^{\delta^{\prime}t_{0}}+e^{-\delta^{\prime}t_{0}}\right)\end{split}

(Of course, this is a new constant cc, which is independent of ll or hh.)

Note that the change in the average of hh is determined by the integral of ¯h\bar{\partial}h, which we can bound using estimate 6. The change in this average can then be absorbed into the last term of the above inequality. Define the variation of hh for a particular tt as follows:

Vh(t):=maxθ|h(t,θ)12π02πh(t,θ)𝑑θ|Vh(t):=\max_{\theta}\left\lvert h(t,\theta)-\frac{1}{2\pi}\int_{0}^{2\pi}h(t,\theta)d\theta\right\rvert

We then have the following estimate (with a new constant cc):

Vh(t)1el1(Vh(t+l)+Vh(tl))+c(l+1)eδ(lR)(eδt+eδt)Vh(t)\leq\frac{1}{e^{l}-1}(Vh(t+l)+Vh(t-l))+c(l+1)e^{\delta^{\prime}(l-R)}\left(e^{\delta^{\prime}t}+e^{-\delta^{\prime}t}\right)

Recalling that δ<δ\delta<\delta^{\prime} and VhVh is bounded by equation 7, we can choose ll large enough so that for RR sufficiently large, using the above estimate recursively tells us that there exists some cc independent of ff or RR so that

Vh(t)ceδR(eδt+eδt) for RtRVh(t)\leq ce^{-\delta R}(e^{\delta t}+e^{-\delta t})\text{ for }-R\leq t\leq R

(This estimate follows automatically from the bound on VhVh for RR bounded.) The required estimate for hh then follows from the fact that the change in the average of hh is bounded by the estimate 6, which is stronger than what we need (which is the same equation with δ\delta in place of δ\delta^{\prime}). We therefore have

h(z)ceδR(|z|δ+|z|δ) for eR|z|eRh(z)\leq ce^{-\delta R}(\left\lvert z\right\rvert^{\delta}+\left\lvert z\right\rvert^{-\delta})\text{ for }e^{-R}\leq\left\lvert z\right\rvert\leq e^{R}

which is the required estimate.

\square

To prove compactness results, we shall divide our domain up into annuli with small energy, and other compact pieces with derivative bounds. For this we shall need some facts about annuli. Recall the following standard definition for the conformal modulus of a Riemann surface which is an annulus:

Definition 3.14.

The conformal modulus of an annulus AA is defined as follows. Let S(A)S(A) denote the set of all continuous functions with L2L^{2} integrable derivatives on AA which approach 11 at one boundary of AA and 0 at the other. Then the conformal modulus of AA is defined as

R(A):=supfS(A)2πA(dfj)df=supfS(A)2πA|df|2R(A):=\sup_{f\in S(A)}\frac{2\pi}{\int_{A}(df\circ j)\wedge df}=\sup_{f\in S(A)}\frac{2\pi}{\int_{A}\left\lvert df\right\rvert^{2}}

We can extend the definition of conformal modulus to include ‘long’ annuli inside exploded curves as follows:

Definition 3.15.

Call a (non complete) exploded curve 𝔄\mathfrak{A} an exploded annulus of conformal modulus logx𝔱l\log x\mathfrak{t}^{-l} if it is connected, and there exists an injective holomorphic map f:𝔄𝔗f:\mathfrak{A}\longrightarrow\mathfrak{T} with image {1<|z~|<x𝔱l}\{1<\left\lvert\tilde{z}\right\rvert<x\mathfrak{t}^{-l}\}. Call it an exploded annulus with semi infinite conformal modulus if it is equal to (a refinement of) {|z~|<1}𝔗11\{\left\lvert\tilde{z}\right\rvert<1\}\subset\mathfrak{T}^{1}_{1}.

(We use x𝔱lx\mathfrak{t}^{-l} in the above definition because the tropical part of the resulting annulus will have length ll.) Two exploded annuli with the same conformal modulus may not be isomorphic, but they will have a common refinement.

We shall need the following lemma containing some useful properties of the (usual) conformal modulus.

Lemma 3.16.
  1. 1.

    An open annulus AA is conformally equivalent to {1<|z|<eR}\{1<\left\lvert z\right\rvert<e^{R}\} if and only if the conformal modulus of AA is RR. If the conformal modulus of AA is infinite, then AA is conformally equivalent to either a punctured disk, or a twice punctured sphere.

  2. 2.

    If {Ai}\{A_{i}\} is a set of disjoint annuli AiAA_{i}\subset A none of which bound a disk in AA, then

    R(A)R(Ai)R(A)\geq\sum R(A_{i})
  3. 3.

    If A1A_{1} and A2A_{2} are annuli contained inside the same Riemann surface which share a boundary so that R(A2)<R(A_{2})<\infty, and the other boundary of A1A_{1} intersects the other boundary of A2A_{2} and the circle at the center of A2A_{2}, then

    R(A2)+16π2R(A2)R(A1)R(A_{2})+\frac{16\pi^{2}}{R(A_{2})}\geq R(A_{1})

    (This is not sharp.)

    \psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\psfrag{ETF19a}{$M_{i}:=\lceil\mathfrak{B}_{i}\rceil$}\psfrag{ETF19b}{$U_{i,j_{1}}$}\psfrag{ETF19c}{$U_{i,j_{2}}$}\psfrag{ETF19d}{$M_{j_{1}}$}\psfrag{ETF19e}{$M_{j_{2}}$}\psfrag{ETF19f}{$M_{k}$}\psfrag{ETF19g}{$r_{i,j_{1}}$}\psfrag{ETF19h}{$r_{i,j_{2}}$}\psfrag{ETF19i}{$r_{j_{1},k}$}\psfrag{ETF19j}{$r_{j_{2},k}$}\psfrag{ETF20a}{$\lfloor\tilde{r}_{1,j}\rfloor$}\psfrag{ETF20b}{$\lfloor\tilde{r}_{2,j}\rfloor$}\psfrag{ETF20c}{$\lfloor\tilde{r}_{3,j}\rfloor$}\psfrag{ETF20d}{$\lfloor\tilde{r}_{1,j}\rfloor\lfloor\tilde{r}_{2,j}\rfloor\lfloor\tilde{r}_{3,j}\rfloor=\mathfrak{t}^{c}$}\psfrag{ETF20e}{$\lfloor\mathfrak{B}_{j}\rfloor$}\psfrag{ETF21a}{$\lfloor p\rfloor$}\psfrag{ETF8a}{$\lceil\mathfrak{B}\rceil$}\psfrag{ETF8b}{$\lfloor\mathfrak{B}\rfloor$}\psfrag{ETF8c}{$\lceil\mathfrak{B}^{\prime}\rceil$}\psfrag{ETF8d}{$\lfloor\mathfrak{B}^{\prime}\rfloor$}\psfrag{ETF9int}{$A_{1}\cap A_{2}$}\psfrag{ETF9cylinder}{$A_{2}$}\includegraphics{ETF9}
  4. 4.

    If A1A_{1} and A2A_{2} are annuli contained inside a Riemann surface so that every circle homotopic to the boundary inside A2A_{2} contains a segment inside A1A_{1} that intersects both boundaries of A1A_{1}, then

    R(A1)4π2R(A2)R(A_{1})\leq\frac{4\pi^{2}}{R(A_{2})}
    \psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\psfrag{ETF19a}{$M_{i}:=\lceil\mathfrak{B}_{i}\rceil$}\psfrag{ETF19b}{$U_{i,j_{1}}$}\psfrag{ETF19c}{$U_{i,j_{2}}$}\psfrag{ETF19d}{$M_{j_{1}}$}\psfrag{ETF19e}{$M_{j_{2}}$}\psfrag{ETF19f}{$M_{k}$}\psfrag{ETF19g}{$r_{i,j_{1}}$}\psfrag{ETF19h}{$r_{i,j_{2}}$}\psfrag{ETF19i}{$r_{j_{1},k}$}\psfrag{ETF19j}{$r_{j_{2},k}$}\psfrag{ETF20a}{$\lfloor\tilde{r}_{1,j}\rfloor$}\psfrag{ETF20b}{$\lfloor\tilde{r}_{2,j}\rfloor$}\psfrag{ETF20c}{$\lfloor\tilde{r}_{3,j}\rfloor$}\psfrag{ETF20d}{$\lfloor\tilde{r}_{1,j}\rfloor\lfloor\tilde{r}_{2,j}\rfloor\lfloor\tilde{r}_{3,j}\rfloor=\mathfrak{t}^{c}$}\psfrag{ETF20e}{$\lfloor\mathfrak{B}_{j}\rfloor$}\psfrag{ETF21a}{$\lfloor p\rfloor$}\psfrag{ETF8a}{$\lceil\mathfrak{B}\rceil$}\psfrag{ETF8b}{$\lfloor\mathfrak{B}\rfloor$}\psfrag{ETF8c}{$\lceil\mathfrak{B}^{\prime}\rceil$}\psfrag{ETF8d}{$\lfloor\mathfrak{B}^{\prime}\rfloor$}\psfrag{ETF9int}{$A_{1}\cap A_{2}$}\psfrag{ETF9cylinder}{$A_{2}$}\psfrag{ETF10a}{$0$}\psfrag{ETF10b}{$1$}\includegraphics{ETF10}

Proof:

The first two items are well known. To prove item 3, first set R=R(A2)R=R(A_{2}) and put coordinates (0,R)×/2π(0,R)\times\mathbb{R}/2\pi\mathbb{Z} on A2A_{2}. Consider any function fS(A1)f\in S(A_{1}). Without losing generality, we can assume that the shared boundary is (0,θ)(0,\theta), and that some segment of the other boundary of A1A_{1} is a curve in A2A_{2} between (R2,0)(\frac{R}{2},0) and (R,θ0)(R,\theta_{0}) where θ00\theta_{0}\geq 0. Then consider integrating |df|2\left\lvert df\right\rvert^{2} along diagonal lines (Rt,4πt+c)(Rt,-{4\pi}t+c). Each of these lines contains a segment inside A1A_{1} on which the integral of dfdf is 11. The length of each segment is bounded by ((4π)2+R2)12((4\pi)^{2}+R^{2})^{\frac{1}{2}}. The integral of |df|2\left\lvert df\right\rvert^{2} along this segment is therefore at least (16π2+R2)12(16\pi^{2}+R^{2})^{-\frac{1}{2}}. This tells us that the integral of |df|2\left\lvert df\right\rvert^{2} over A1A_{1} is at least 2πR((4π)2+R2)\frac{2\pi R}{((4\pi)^{2}+R^{2})}, and therefore,

R(A1)(16π2+R2)R=R(A2)+16π2R(A2)R(A_{1})\leq\frac{(16\pi^{2}+R^{2})}{R}=R(A_{2})+\frac{16\pi^{2}}{R(A_{2})}

To prove item 4, consider a function fS(A1)f\in S(A_{1}). We shall integrate |df|2\left\lvert df\right\rvert^{2} along segments of the form (c,t)(c,t) inside A2A1A_{2}\cap A_{1} traveling from one boundary of A1A_{1} to the other. The integral of dfdf along such a segment is 11, and the length of the segment is at most 2π2\pi, so the integral of |df|2\left\lvert df\right\rvert^{2} along the segment is at least 12π\frac{1}{2\pi}, and the integral of |df|2\left\lvert df\right\rvert^{2} over A1A_{1} is at least R(A2)2π\frac{R(A_{2})}{2\pi}. Therefore, we have

R(A1)4π2R(A2)R(A_{1})\leq\frac{4\pi^{2}}{R(A_{2})}

\square

Proposition 3.17.

Given (𝔅,J,Ω,ω,g)(\mathfrak{B},J,\Omega,\omega,g), an energy bound EE and a number NN, and small enough ϵ>0\epsilon>0, there exists a number bound MM depending lower semicontinuously on (𝔅,J,Ω,ω,g,E,N,ϵ)(\mathfrak{B},J,\Omega,\omega,g,E,N,\epsilon) and for any large enough distance RR, a derivative bound cc and conformal bound R^\hat{R} depending continuously on (𝔅,J,Ω,ω,g,E,N,ϵ,R)(\mathfrak{B},J,\Omega,\omega,g,E,N,\epsilon,R) so that the following is true:

Given any complete, stable holomorphic curve f:𝔅f:\mathfrak{C}\longrightarrow\mathfrak{B} with energy at most EE, and with genus and number of punctures at most NN, there exists some collection of at most MM exploded annuli 𝔄i\mathfrak{A}_{i}\subset\mathfrak{C}, so that:

  1. 1.

    Each 𝔄i\mathfrak{A}_{i} has conformal modulus larger than 2R2R

  2. 2.

    Put the standard cylindrical metric on 𝔄i\mathfrak{A}_{i}, and use the notation 𝔄l,i\mathfrak{A}_{l,i} to denote the annulus consisting of all points in 𝔄i\mathfrak{A}_{i} with distance to the boundary at least ll.

    𝔄R2,i𝔄R2,j= if ij\mathfrak{A}_{\frac{R}{2},i}\cap\mathfrak{A}_{\frac{R}{2},j}=\emptyset\text{ if }i\neq j
  3. 3.

    ff restricted to 𝔄i\mathfrak{A}_{i} has energy less than ϵ\epsilon.

  4. 4.

    Each component of 𝔄R,i\mathfrak{C}-\bigcup\mathfrak{A}_{R,i} is a smooth Riemann surface with bounded conformal geometry in the sense that any annulus inside one of these smooth components with conformal modulus greater than R^\hat{R} must bound a smooth disk inside that component.

  5. 5.

    The following metrics can be put on 𝔄i\mathfrak{A}_{i} and each smooth component of 𝔄R,i\mathfrak{C}-\bigcup\mathfrak{A}_{R,i}:

    1. (a)

      On any component which is a smooth torus, use the unique flat metric in the conformal class of the complex structure so that the area of the torus is 11.

    2. (b)

      If the component is equal to a disk, an identification with the standard unit disk can be chosen so that if 𝔄R,i\mathfrak{A}_{R,i} is the bounding annulus, 0 is in the complement of 𝔄i\mathfrak{A}_{i}. Give components such as this the standard Euclidean metric.

    3. (c)

      If the component is equal to some annulus, give it the standard cylindrical metric on /×(0,l)\mathbb{R}/\mathbb{Z}\times(0,l). Give each 𝔄i\mathfrak{A}_{i} the analogous standard metric.

    4. (d)

      Any component not equal to a torus, annulus, or disk will admit a unique metric in the correct conformal class with curvature 1-1 so that boundary components are geodesic (there will be no components which are smooth spheres). Give these components this metric.

    On any component of 𝔄9R10,i\mathfrak{C}-\bigcup\mathfrak{A}_{\frac{9R}{10},i}, the derivative in this metric is bounded by cc

    |df|g<c\left\lvert df\right\rvert_{g}<c

    Moreover, on 𝔄6R10,i𝔄9R10,i\mathfrak{A}_{\frac{6R}{10},i}-\mathfrak{A}_{\frac{9R}{10},i}, the ratio between the two metrics defined above is less than cc.

  6. 6.

    There exists some lower energy bound ϵ0>0\epsilon_{0}>0 depending only on ϵ\epsilon, RR, and EE so that ff restricted to any component of 𝔄R,i\mathfrak{C}-\bigcup\mathfrak{A}_{{R},i} which is a disk, annulus or torus has ω\omega energy greater than ϵ0\epsilon_{0}

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\psfrag{ETF19a}{$M_{i}:=\lceil\mathfrak{B}_{i}\rceil$}\psfrag{ETF19b}{$U_{i,j_{1}}$}\psfrag{ETF19c}{$U_{i,j_{2}}$}\psfrag{ETF19d}{$M_{j_{1}}$}\psfrag{ETF19e}{$M_{j_{2}}$}\psfrag{ETF19f}{$M_{k}$}\psfrag{ETF19g}{$r_{i,j_{1}}$}\psfrag{ETF19h}{$r_{i,j_{2}}$}\psfrag{ETF19i}{$r_{j_{1},k}$}\psfrag{ETF19j}{$r_{j_{2},k}$}\psfrag{ETF20a}{$\lfloor\tilde{r}_{1,j}\rfloor$}\psfrag{ETF20b}{$\lfloor\tilde{r}_{2,j}\rfloor$}\psfrag{ETF20c}{$\lfloor\tilde{r}_{3,j}\rfloor$}\psfrag{ETF20d}{$\lfloor\tilde{r}_{1,j}\rfloor\lfloor\tilde{r}_{2,j}\rfloor\lfloor\tilde{r}_{3,j}\rfloor=\mathfrak{t}^{c}$}\psfrag{ETF20e}{$\lfloor\mathfrak{B}_{j}\rfloor$}\psfrag{ETF21a}{$\lfloor p\rfloor$}\psfrag{ETF8a}{$\lceil\mathfrak{B}\rceil$}\psfrag{ETF8b}{$\lfloor\mathfrak{B}\rfloor$}\psfrag{ETF8c}{$\lceil\mathfrak{B}^{\prime}\rceil$}\psfrag{ETF8d}{$\lfloor\mathfrak{B}^{\prime}\rfloor$}\psfrag{ETF11d}{{\footnotesize derivative bounded here}}\psfrag{ETF11c}{{\footnotesize components with $\omega$ energy bounded below}}\psfrag{ETF11b}{$\mathfrak{A}$}\psfrag{ETF11a}{$\mathfrak{A}_{R}${\footnotesize:well behaved annulus}}\includegraphics{ETF11}

Proof:

Lemma 3.10 tells us that for ϵ\epsilon small enough, any holomorphic map of the unit disk with ω\omega-energy less than ϵ\epsilon and Ω\Omega-energy less than EE must have derivative at 0 bounded by c0c_{0}. We shall prove our theorem for ϵ\epsilon small enough so that this is true, and small enough that Corollary 3.12 also holds with an energy bound of ϵ\epsilon. We shall also choose our distance RR greater than 4π4\pi (for use with Lemma 3.16), and large enough that Corollary 3.12 tells us that if ff restricted to some smooth annulus 𝔄i\mathfrak{A}_{i} has ω\omega-energy less than ϵ\epsilon, then the ω\omega-energy of 𝔄R2,i\mathfrak{A}_{\frac{R}{2},i} is less than ϵ5\frac{\epsilon}{5}.

Let us now begin to construct our annuli. First, note that any edge of \mathfrak{C} has zero ω\omega-energy, so we can choose an exploded annulus 𝔄i\mathfrak{A}_{i} containing each edge with ω\omega-energy less than ϵ5\frac{\epsilon}{5}. (Note for use with item 6 that in the case that this bounds a disk, the ω\omega-energy of the resulting disk will be at least 4ϵ5\frac{4\epsilon}{5} ). We can do this so that these 𝔄i\mathfrak{A}_{i} are mutually disjoint. Note that the complement of these 𝔄i\mathfrak{A}_{i} is a smooth Riemann surface with boundary. (Note also that no component of \mathfrak{C} can be isomorphic to 𝔗\mathfrak{T} as the Ω\Omega-energy of any complete component is equal to the ω\omega-energy which must be zero on 𝔗\mathfrak{T}. Similarly, there is no complete component of \mathfrak{C} which is locally modeled everywhere on 𝔗\mathfrak{T}.)

For each connected component of \mathfrak{C} which is a smooth sphere, we shall now remove an annulus. First, note that we can put some round metric in the correct conformal class of the sphere so that there exist 33 mutually perpendicular geodesics which divide our sphere into 88 regions each of which has equal ω\omega-energy. As our sphere must have energy at least ϵ\epsilon in order to be stable, we can choose two antipodal regions that each have energy ϵ8\frac{\epsilon}{8}. By choosing ϵ0>0\epsilon_{0}>0 small enough, we can then get that there exist two disks with ω\omega-energy at least ϵ0\epsilon_{0} which intersect each of these regions, and which have radius as small as we like. Choose ϵ0>0\epsilon_{0}>0 small enough that the complement of these disks is an annulus of conformal radius at least k(2R+1){k(2R+1)} for some integer k>5Eϵk>\frac{5E}{\epsilon}. How small ϵ0\epsilon_{0} is required to achieve this depends only on RR, ϵ\epsilon and EE. Then we can divide this annulus into kk annuli with conformal modulus at least (2R+1)(2R+1), at least one of which has ω\omega-energy at most ϵ5\frac{\epsilon}{5}. Add this annulus to our collection. Note that it bounds disks which have energy at least ϵ0\epsilon_{0}.

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\psfrag{ETF19a}{$M_{i}:=\lceil\mathfrak{B}_{i}\rceil$}\psfrag{ETF19b}{$U_{i,j_{1}}$}\psfrag{ETF19c}{$U_{i,j_{2}}$}\psfrag{ETF19d}{$M_{j_{1}}$}\psfrag{ETF19e}{$M_{j_{2}}$}\psfrag{ETF19f}{$M_{k}$}\psfrag{ETF19g}{$r_{i,j_{1}}$}\psfrag{ETF19h}{$r_{i,j_{2}}$}\psfrag{ETF19i}{$r_{j_{1},k}$}\psfrag{ETF19j}{$r_{j_{2},k}$}\psfrag{ETF20a}{$\lfloor\tilde{r}_{1,j}\rfloor$}\psfrag{ETF20b}{$\lfloor\tilde{r}_{2,j}\rfloor$}\psfrag{ETF20c}{$\lfloor\tilde{r}_{3,j}\rfloor$}\psfrag{ETF20d}{$\lfloor\tilde{r}_{1,j}\rfloor\lfloor\tilde{r}_{2,j}\rfloor\lfloor\tilde{r}_{3,j}\rfloor=\mathfrak{t}^{c}$}\psfrag{ETF20e}{$\lfloor\mathfrak{B}_{j}\rfloor$}\psfrag{ETF21a}{$\lfloor p\rfloor$}\psfrag{ETF8a}{$\lceil\mathfrak{B}\rceil$}\psfrag{ETF8b}{$\lfloor\mathfrak{B}\rfloor$}\psfrag{ETF8c}{$\lceil\mathfrak{B}^{\prime}\rceil$}\psfrag{ETF8d}{$\lfloor\mathfrak{B}^{\prime}\rfloor$}\psfrag{ETF11d}{{\footnotesize derivative bounded here}}\psfrag{ETF11c}{{\footnotesize components with $\omega$ energy bounded below}}\psfrag{ETF11b}{$\mathfrak{A}$}\psfrag{ETF11a}{$\mathfrak{A}_{R}${\footnotesize:well behaved annulus}}\psfrag{ETF12c}{$\omega$-energy $<\frac{\epsilon}{5}$}\psfrag{ETF12b}{$\mathfrak{A}$}\psfrag{ETF12a}{$\omega$-energy $>\epsilon_{0}$}\includegraphics{ETF12}

Consider a holomorphic injection of the unit disk i:Di:D\longrightarrow\mathfrak{C} into the complement of our annuli 𝔄R,i\mathfrak{A}_{R,i} constructed up to this point. Suppose that i(0)i(0) is in the complement of 𝔄R4,i\mathfrak{A}_{\frac{R}{4},i}. Then Lemma 3.16 tells us that the restriction of ii to |z|<e16π2R\left\lvert z\right\rvert<e^{-\frac{16\pi^{2}}{R}} is in the complement of 𝔄R2,i\mathfrak{A}_{\frac{R}{2},i}. Now suppose that |d(fi)|>c0e16π2R+k(2R+1)\left\lvert d(f\circ i)\right\rvert>c_{0}e^{\frac{16\pi^{2}}{R}+k(2R+1)} where kk is some integer greater than 5Eϵ\frac{5E}{\epsilon}. Then the restriction of ff to the disk |z|<e16π2Rk(2R+1)\left\lvert z\right\rvert<e^{-\frac{16\pi^{2}}{R}-k(2R+1)} must have energy greater than ϵ\epsilon due to Lemma 3.10. This disk is surrounded by kk disjoint annuli of conformal modulus (2R+1)(2R+1) which are contained in the complement of 𝔄R2,i\mathfrak{A}_{\frac{R}{2},i}. At least one of these must have energy less than ϵ5\frac{\epsilon}{5}. Add this annulus to our collection. Continue adding annuli in this manner. After a finite (but not universally bounded) number of times, no more annuli can be added in this way. We shall argue this below after adding some more annuli.

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\psfrag{ETF19a}{$M_{i}:=\lceil\mathfrak{B}_{i}\rceil$}\psfrag{ETF19b}{$U_{i,j_{1}}$}\psfrag{ETF19c}{$U_{i,j_{2}}$}\psfrag{ETF19d}{$M_{j_{1}}$}\psfrag{ETF19e}{$M_{j_{2}}$}\psfrag{ETF19f}{$M_{k}$}\psfrag{ETF19g}{$r_{i,j_{1}}$}\psfrag{ETF19h}{$r_{i,j_{2}}$}\psfrag{ETF19i}{$r_{j_{1},k}$}\psfrag{ETF19j}{$r_{j_{2},k}$}\psfrag{ETF20a}{$\lfloor\tilde{r}_{1,j}\rfloor$}\psfrag{ETF20b}{$\lfloor\tilde{r}_{2,j}\rfloor$}\psfrag{ETF20c}{$\lfloor\tilde{r}_{3,j}\rfloor$}\psfrag{ETF20d}{$\lfloor\tilde{r}_{1,j}\rfloor\lfloor\tilde{r}_{2,j}\rfloor\lfloor\tilde{r}_{3,j}\rfloor=\mathfrak{t}^{c}$}\psfrag{ETF20e}{$\lfloor\mathfrak{B}_{j}\rfloor$}\psfrag{ETF21a}{$\lfloor p\rfloor$}\psfrag{ETF8a}{$\lceil\mathfrak{B}\rceil$}\psfrag{ETF8b}{$\lfloor\mathfrak{B}\rfloor$}\psfrag{ETF8c}{$\lceil\mathfrak{B}^{\prime}\rceil$}\psfrag{ETF8d}{$\lfloor\mathfrak{B}^{\prime}\rfloor$}\psfrag{ETF11d}{{\footnotesize derivative bounded here}}\psfrag{ETF11c}{{\footnotesize components with $\omega$ energy bounded below}}\psfrag{ETF11b}{$\mathfrak{A}$}\psfrag{ETF11a}{$\mathfrak{A}_{R}${\footnotesize:well behaved annulus}}\psfrag{ETF12c}{$\omega$-energy $<\frac{\epsilon}{5}$}\psfrag{ETF12b}{$\mathfrak{A}$}\psfrag{ETF12a}{$\omega$-energy $>\epsilon_{0}$}\psfrag{ETF13a}{$\mathfrak{A}_{R}$}\psfrag{ETF13b}{$\mathfrak{A}_{\frac{R}{2}}$}\psfrag{ETF13c}{New low energy annulus}\psfrag{ETF13d}{Disk with $\omega$-energy $>\epsilon_{0}$}\includegraphics{ETF13}

We now want to add annuli so that condition 4 is satisfied. Suppose some smooth component of 𝔄R,i\mathfrak{C}-\bigcup\mathfrak{A}_{R,i} contains some annulus of conformal modulus greater than 2R(3+5Eϵ)2R(3+\frac{5E}{\epsilon}) that does not bound a disk. Note that as we have chosen R>4πR>4\pi, the size of such an annulus is greater than 2R(1+5Eϵ)+2R+32π2R2R(1+\frac{5E}{\epsilon})+2R+\frac{32\pi^{2}}{R}. The annulus of size RR at its end can’t be contained entirely within 𝔄i\mathfrak{A}_{i}, and then applying Lemma 3.16 part 4 with the next annulus of size 16π2R\frac{16\pi^{2}}{R} in the place of A1A_{1}, we see that our annulus minus annuli at each end of conformal modulus R+16π2RR+\frac{16\pi^{2}}{R} must be contained entirely in the complement of 𝔄R2,i\mathfrak{A}_{\frac{R}{2},i}. We then obtain more than 5Eϵ\frac{5E}{\epsilon} disjoint annuli in the compliment of 𝔄R2,i\mathfrak{A}_{\frac{R}{2},i} with conformal modulus greater than 2R2R. The restriction of f to at least one of these annuli must have ω\omega-energy less than ϵ\epsilon. Add this annulus to our collection.

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\psfrag{ETF19a}{$M_{i}:=\lceil\mathfrak{B}_{i}\rceil$}\psfrag{ETF19b}{$U_{i,j_{1}}$}\psfrag{ETF19c}{$U_{i,j_{2}}$}\psfrag{ETF19d}{$M_{j_{1}}$}\psfrag{ETF19e}{$M_{j_{2}}$}\psfrag{ETF19f}{$M_{k}$}\psfrag{ETF19g}{$r_{i,j_{1}}$}\psfrag{ETF19h}{$r_{i,j_{2}}$}\psfrag{ETF19i}{$r_{j_{1},k}$}\psfrag{ETF19j}{$r_{j_{2},k}$}\psfrag{ETF20a}{$\lfloor\tilde{r}_{1,j}\rfloor$}\psfrag{ETF20b}{$\lfloor\tilde{r}_{2,j}\rfloor$}\psfrag{ETF20c}{$\lfloor\tilde{r}_{3,j}\rfloor$}\psfrag{ETF20d}{$\lfloor\tilde{r}_{1,j}\rfloor\lfloor\tilde{r}_{2,j}\rfloor\lfloor\tilde{r}_{3,j}\rfloor=\mathfrak{t}^{c}$}\psfrag{ETF20e}{$\lfloor\mathfrak{B}_{j}\rfloor$}\psfrag{ETF21a}{$\lfloor p\rfloor$}\psfrag{ETF8a}{$\lceil\mathfrak{B}\rceil$}\psfrag{ETF8b}{$\lfloor\mathfrak{B}\rfloor$}\psfrag{ETF8c}{$\lceil\mathfrak{B}^{\prime}\rceil$}\psfrag{ETF8d}{$\lfloor\mathfrak{B}^{\prime}\rfloor$}\psfrag{ETF11d}{{\footnotesize derivative bounded here}}\psfrag{ETF11c}{{\footnotesize components with $\omega$ energy bounded below}}\psfrag{ETF11b}{$\mathfrak{A}$}\psfrag{ETF11a}{$\mathfrak{A}_{R}${\footnotesize:well behaved annulus}}\psfrag{ETF12c}{$\omega$-energy $<\frac{\epsilon}{5}$}\psfrag{ETF12b}{$\mathfrak{A}$}\psfrag{ETF12a}{$\omega$-energy $>\epsilon_{0}$}\psfrag{ETF13a}{$\mathfrak{A}_{R}$}\psfrag{ETF13b}{$\mathfrak{A}_{\frac{R}{2}}$}\psfrag{ETF13c}{New low energy annulus}\psfrag{ETF13d}{Disk with $\omega$-energy $>\epsilon_{0}$}\psfrag{ETF14a}{New low energy annulus}\psfrag{ETF14b}{$\mathfrak{A}_{\frac{R}{2}}$}\psfrag{ETF14c}{$\mathfrak{A}_{R}$}\includegraphics{ETF14}

We shall now argue that we can only add a finite number of annuli in this manner. The number of annuli that bound disks is bounded by Eϵ\frac{E}{\epsilon} plus the twice the number of connected components of \mathfrak{C} that are spheres (which is bounded by Eϵ\frac{E}{\epsilon}). There are also a finite number of exploded annuli with infinite conformal modulus. The number of these is bounded by our topological bounds, and the fact that there are at most Eϵ\frac{E}{\epsilon} spherical components with fewer than 33 punctures. Call the complement of all the above annuli 0\mathfrak{C}_{0}. There is then a bound on the number of homotopy classes in 0\mathfrak{C}_{0} of our remaining annuli, given by our bounds on the topology of \mathfrak{C}, the fact that we have removed a bounded number of annuli, and the observation that for any Riemann surface with boundary, there are only a finite number of homotopy classes which contain annuli of conformal modulus greater than 4π4\pi (Lemma 3.16 part 4 can help to prove this). If there were an infinite number of annuli 𝔄R2,i\mathfrak{A}_{\frac{R}{2},i} in our collection in the same homotopy class, then there would exist an annulus in 0\mathfrak{C}_{0} in that homotopy class which contains all of them. As the 𝔄R2,i\mathfrak{A}_{\frac{R}{2},i} are disjoint and have conformal modulus at least RR, this annulus would have to have infinite conformal modulus. Note that there are no nontrivial annuli of infinite conformal modulus in 0\mathfrak{C}_{0} (because we have removed the annuli containing edges and at least one annulus from each spherical component), and if an annulus of infinite conformal modulus surrounds a disk with energy greater than ϵ0\epsilon_{0}, then the derivative of ff must have been unbounded there. This is not possible, so we must only have a finite number of annuli in our collection. Note that everything apart from our bound on the number of annuli in a fixed homotopy class is bounded independent of ff.

Now we shall merge some annuli so that there exist no annular component of 𝔄R,i\mathfrak{C}-\bigcup\mathfrak{A}_{R,i} with ω\omega-energy less than ϵ5\frac{\epsilon}{5}. Then we will have a bound on the number of annuli which is independent of ff. Suppose that we have some collection {𝔄1,,𝔄n}\{\mathfrak{A}_{1},\dotsc,\mathfrak{A}_{n}\} of our annuli so that 𝔄R,i\mathfrak{A}_{R,i} and 𝔄R,i+1\mathfrak{A}_{R,i+1} bound an annulus which has ω\omega-energy less than ϵ5\frac{\epsilon}{5}. Use the notation 𝔄[m,n]\mathfrak{A}_{[m,n]} to denote the annulus that consists of 𝔄m\mathfrak{A}_{m}, 𝔄n\mathfrak{A}_{n} and everything in between. Then, as we’ve chosen each of our 𝔄i\mathfrak{A}_{i} to have ω\omega-energy less than ϵ5\frac{\epsilon}{5}, 𝔄[i,i+2]\mathfrak{A}_{[i,i+2]} has ω\omega-energy less than ϵ\epsilon, so we can apply Corollary 3.12 to show that far enough into the interior of this cylinder, there is very little ω\omega-energy. (Note that as the parts of exploded annuli which are locally modeled on 𝔗\mathfrak{T} always have no ω\omega-energy, so there is no difficulty in applying this lemma in the seemingly more general setting of exploded annuli.) Applying Lemma 3.16 part 3, and noting that as we have chosen R>4πR>4\pi, we have 2R>R+16π2R2R>R+\frac{16\pi^{2}}{R}, we see that 𝔄[i,i+2]𝔄i𝔄i+2𝔄R2,[i,i+2]\mathfrak{A}_{[i,i+2]}-\mathfrak{A}_{i}-\mathfrak{A}_{i+2}\subset\mathfrak{A}_{\frac{R}{2},[i,i+2]}. We have chosen RR large enough that Corollary 3.12 tells us that the energy of ff restricted to 𝔄R2,[i,i+2]\mathfrak{A}_{\frac{R}{2},[i,i+2]} is less than ϵ5\frac{\epsilon}{5}. We can now repeat this argument inductively to show that the energy of ff restricted to 𝔄[1,n]\mathfrak{A}_{[1,n]} is less than ϵ\epsilon and 𝔄R2,[1,n]\mathfrak{A}_{\frac{R}{2},[1,n]} is less than ϵ5\frac{\epsilon}{5}.

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\psfrag{ETF19a}{$M_{i}:=\lceil\mathfrak{B}_{i}\rceil$}\psfrag{ETF19b}{$U_{i,j_{1}}$}\psfrag{ETF19c}{$U_{i,j_{2}}$}\psfrag{ETF19d}{$M_{j_{1}}$}\psfrag{ETF19e}{$M_{j_{2}}$}\psfrag{ETF19f}{$M_{k}$}\psfrag{ETF19g}{$r_{i,j_{1}}$}\psfrag{ETF19h}{$r_{i,j_{2}}$}\psfrag{ETF19i}{$r_{j_{1},k}$}\psfrag{ETF19j}{$r_{j_{2},k}$}\psfrag{ETF20a}{$\lfloor\tilde{r}_{1,j}\rfloor$}\psfrag{ETF20b}{$\lfloor\tilde{r}_{2,j}\rfloor$}\psfrag{ETF20c}{$\lfloor\tilde{r}_{3,j}\rfloor$}\psfrag{ETF20d}{$\lfloor\tilde{r}_{1,j}\rfloor\lfloor\tilde{r}_{2,j}\rfloor\lfloor\tilde{r}_{3,j}\rfloor=\mathfrak{t}^{c}$}\psfrag{ETF20e}{$\lfloor\mathfrak{B}_{j}\rfloor$}\psfrag{ETF21a}{$\lfloor p\rfloor$}\psfrag{ETF8a}{$\lceil\mathfrak{B}\rceil$}\psfrag{ETF8b}{$\lfloor\mathfrak{B}\rfloor$}\psfrag{ETF8c}{$\lceil\mathfrak{B}^{\prime}\rceil$}\psfrag{ETF8d}{$\lfloor\mathfrak{B}^{\prime}\rfloor$}\psfrag{ETF11d}{{\footnotesize derivative bounded here}}\psfrag{ETF11c}{{\footnotesize components with $\omega$ energy bounded below}}\psfrag{ETF11b}{$\mathfrak{A}$}\psfrag{ETF11a}{$\mathfrak{A}_{R}${\footnotesize:well behaved annulus}}\psfrag{ETF12c}{$\omega$-energy $<\frac{\epsilon}{5}$}\psfrag{ETF12b}{$\mathfrak{A}$}\psfrag{ETF12a}{$\omega$-energy $>\epsilon_{0}$}\psfrag{ETF13a}{$\mathfrak{A}_{R}$}\psfrag{ETF13b}{$\mathfrak{A}_{\frac{R}{2}}$}\psfrag{ETF13c}{New low energy annulus}\psfrag{ETF13d}{Disk with $\omega$-energy $>\epsilon_{0}$}\psfrag{ETF14a}{New low energy annulus}\psfrag{ETF14b}{$\mathfrak{A}_{\frac{R}{2}}$}\psfrag{ETF14c}{$\mathfrak{A}_{R}$}\psfrag{ETF15a}{$\mathfrak{A}_{1}$}\psfrag{ETF15b}{$\mathfrak{A}_{2}$}\psfrag{ETF15c}{$\mathfrak{A}_{3}$}\psfrag{ETF15d}{$\mathfrak{A}_{4}$}\psfrag{ETF15e}{$\mathfrak{A}_{\frac{R}{2},[1,3]}$ has low $\omega$-energy}\includegraphics{ETF15}

Now, replace all of our sets of annuli {𝔄1,,𝔄n}\{\mathfrak{A}_{1},\dotsc,\mathfrak{A}_{n}\} of maximal size obeying the above conditions with the annulus given by 𝔄R2,1\mathfrak{A}_{\frac{R}{2},1}, 𝔄R2,n\mathfrak{A}_{\frac{R}{2},n} and everything in between. (Note that if we had some collection of annuli which bounded small energy annuli in a cyclic fashion, we would obtain a connected component of \mathfrak{C} which had energy less than ϵ\epsilon, and was not stable.)

We shall now check that all the conditions we require are satisfied by this new set of annuli. First, the number is bounded independent of ff. The size of each annulus is greater than 2R2R. Our resulting set of annuli also obey the non-intersection condition 2 because the original set did. We showed above that the energy of each new annulus is less than ϵ\epsilon. Because the complement of our original set of annuli obeyed condition 4 with a bound of 2R(3+5Eϵ)2R(3+\frac{5E}{\epsilon}), by increasing our conformal bound appropriately and using Lemma 3.16, we can achieve condition 4. The lower bound on the ω\omega-energy of unstable components is also satisfied by construction.

All that remains is condition 5. Note first that the complement of our annuli 𝔄R,i\mathfrak{A}_{R,i} admits metrics as described in 5. The fact that there exists some cc so that the metrics we choose on different components differ by a factor of less than cc on 𝔄6R10,i𝔄9R10,i\mathfrak{A}_{\frac{6R}{10},i}-\mathfrak{A}_{\frac{9R}{10},i} follows from conditions 4 and 2, and the bound on topology as follows: The complement of 𝔄R,i\mathfrak{A}_{R,i} is a Riemann surface with bounded genus, a bounded number of boundary components and bounded conformal geometry from condition 4. An easy compactness argument tells us that there exists a constant cc^{\prime} so that any injective holomorphic map of the unit disk into either 𝔄i\mathfrak{A}_{i} or the complement of 𝔄R,i\mathfrak{A}_{R,i} has derivative at 0 bounded by cc^{\prime}. We can use this (remembering that R>4π>10R>4\pi>10 so we can get a disk of unit size centered on any point inside 𝔄6R10𝔄9R10\mathfrak{A}_{\frac{6R}{10}}-\mathfrak{A}_{\frac{9R}{10}}) to get the ratio of the metric on the complement of 𝔄R,i\mathfrak{A}_{R,i} divided by the metric on 𝔄i\mathfrak{A}_{i} is bounded by cc^{\prime} here.

Then, we can use Lemma 3.16 and our conformal bound from condition 4 to get some lower bound r0>0r_{0}>0 so that the boundary of 𝔄9R10\mathfrak{A}_{\frac{9R}{10}} and 𝔄6R10\mathfrak{A}_{\frac{6R}{10}} is further than r0r_{0} from the boundary of the complement of 𝔄R,i\mathfrak{A}_{R,i}. A second easy compactness argument then tells us that if we fix any distance r0>0r_{0}>0, there exists some constant cr0c_{r_{0}} (only depending on the bounds in the above paragraph) so that for every point further than r0r_{0} from the boundary there exists an injective holomorphic map of the unit disk sending 0 to that point, with derivative at 0 greater than cr0c_{r_{0}}. We then get that the above ratio of metrics above is bounded below by cr0c\frac{c_{r_{0}}}{c^{\prime}}.

We must now prove our derivative bound on the complement of 𝔄9R10\mathfrak{A}_{\frac{9R}{10}}. We shall do this by proving that given any holomorphic injection of the unit disk i:D𝔄R,ii:D\longrightarrow\mathfrak{C}-\bigcup\mathfrak{A}_{R,i} so that i(0)𝔄9R10,ii(0)\notin\mathfrak{A}_{\frac{9R}{10},i}, then |d(fi)|\left\lvert d(f\circ i)\right\rvert is bounded. Then, our conformal bound from condition 4 will tell us that we have a bound on |df|\left\lvert df\right\rvert in the metric that we have chosen.

We do this in two cases. First, suppose that i(0)i(0) is in the complement of 𝔄R4,i\mathfrak{A}_{\frac{R}{4},i} for all our old annuli. Then, Lemma 3.16 tells us that the image of ii restricted to the disk of radius e16π2Re^{-16\frac{\pi^{2}}{R}} is contained in the complement 𝔄R2,i\mathfrak{A}_{\frac{R}{2},i} for all our old annuli, as argued above. If the derivative of fif\circ i on this disk was large enough, we would be able to add another annulus to our old collection in the manner described above. As this process terminated, |d(fi)|\left\lvert d(f\circ i)\right\rvert is bounded. Second, suppose that i(0)i(0) is contained inside 𝔄R4,i\mathfrak{A}_{\frac{R}{4},i} for one of our old annuli. Then Lemma 3.16 tells us that the restriction of ii to the disk of radius e16π2Re^{-16\frac{\pi^{2}}{R}} must be contained inside 𝔄i\mathfrak{A}_{i}, and therefore have energy less than ϵ\epsilon. This means that |d(fi)|c0e16π2R\left\lvert d(f\circ i)\right\rvert\leq c_{0}e^{16\frac{\pi^{2}}{R}}, and our derivative is bounded as required.

\square

4 Compactness

Theorem 4.1.

Given a basic, complete exploded 𝕋\mathbb{T} fibration 𝔅\mathfrak{B} with a civilized almost complex structure JJ and strict taming Ω\Omega, the moduli stack g,n,E(𝔅)\mathcal{M}_{g,n,E}(\mathfrak{B}) of stable holomorphic curves with a fixed genus gg and number of punctures nn, and with Ω\Omega energy less than EE is topologically compact in C,δC^{\infty,\delta} for any 0<δ<10<\delta<1.

More generally, if (𝔅^𝔊,J,Ω)(\hat{\mathfrak{B}}\longrightarrow\mathfrak{G},J,\Omega) is a family of such (𝔅,J,Ω)(\mathfrak{B},J,\Omega), the map g,n,E(𝔅^𝔊)𝔊¯\mathcal{M}_{g,n,E}(\hat{\mathfrak{B}}\rightarrow\mathfrak{G})\longrightarrow\underline{\mathfrak{G}} is C,δC^{\infty,\delta} topologically proper.

In particular, this means that given any sequence of the above holomorphic curves in the fibers over a topologically convergent sequence in 𝔊\mathfrak{G}, there exists a subsequence fif^{i} which converges to a holomorphic curve ff as follows: There exists a sequence of families of smooth curves,

(^,ji)f^i𝔅^𝔉𝔊\begin{array}[]{ccc}(\mathfrak{\hat{C}},j_{i})&\xrightarrow{\hat{f}^{i}}&\hat{\mathfrak{B}}\\ \downarrow&&\downarrow\\ \mathfrak{F}&\longrightarrow&\mathfrak{G}\end{array}

so that this sequence of families converges in C,δC^{\infty,\delta} to the smooth family

(^,j)f^𝔅^𝔉𝔊\begin{array}[]{ccc}(\mathfrak{\hat{C}},j)&\xrightarrow{\hat{f}}&\hat{\mathfrak{B}}\\ \downarrow&&\downarrow\\ \mathfrak{F}&\longrightarrow&\mathfrak{G}\end{array}

and a sequence of points pi𝔉p^{i}\rightarrow\mathfrak{F} so that pip\lceil p^{i}\rceil\rightarrow\lceil p\rceil, fif^{i} is the map given by the restriction of f^i\hat{f}^{i} to the fiber over pip^{i}, and ff is given by the restriction of ff to the fiber over pp. Of course, the case where we just have a single (𝔅,J,ω)(\mathfrak{B},J,\omega) is the same as the family case when 𝔊\mathfrak{G} is a point.

The definition of C,δC^{\infty,\delta} convergence can be found in section 2.4 starting on page 2.4 and the definition of the moduli stack is contained in section 2.8, which starts on page 2.8. The notion of topological convergence is introduced on page 2.14.

The proof of this theorem uses Lemma 3.13 on page 3.13 and Proposition 3.17 on page 3.17. Together, these allow us to decompose holomorphic curves into pieces with bounded behavior. A standard Arzela-Ascoli type argument gets a type of convergence of these pieces. If we were working in the category of smooth manifolds, this would be sufficient to prove the compactness theorem because the type of convergence involved would have a unique limit, and therefore the limiting pieces would glue together. The extra problem that must be dealt with in this case is that we are not dealing with a type of convergence that has a unique limit, so we must work much harder to show that our resulting ‘limiting family’ for each piece glues together to the limiting family 𝔉\mathfrak{F}.

Example 4.2.

The following are examples of the types of non uniqueness we have to deal with:

  1. 1.

    In 𝔗11\mathfrak{T}^{1}_{1}, let z~(pi)=1i𝔱0\tilde{z}(p^{i})=\frac{1}{i}\mathfrak{t}^{0}. Then pip\lceil p^{i}\rceil\rightarrow\lceil p\rceil where p𝔗11p\rightarrow\mathfrak{T}^{1}_{1} is any point so that z~(p)=c𝔱a\tilde{z}(p)=c\mathfrak{t}^{a} where a>0a>0. Note that there is no ‘point’ p𝔗11p\rightarrow\mathfrak{T}^{1}_{1} so that z~(p)=0𝔱0\tilde{z}(p)=0\mathfrak{t}^{0}. (The author did try modifying the definitions of exploded fibrations so that such limits exist, however if this is tried, the theory of holomorphic curves becomes much more complicated, and the whole setup is a lot less natural.) This is one reason that non unique limits need to be considered.

  2. 2.

    Expanding example 1, consider the maps fi:P1P1×𝔗11f^{i}:\mathbb{C}P^{1}\longrightarrow\mathbb{C}P^{1}\times\mathfrak{T}^{1}_{1} given by fi(u)=(u,pi)f^{i}(u)=(u,p_{i}). This sequence of maps does not have a unique limit. We instead consider the family

    P1×𝔗11f^:=idP1×𝔗11𝔗11\begin{split}\mathbb{C}P^{1}&\times\mathfrak{T}^{1}_{1}\xrightarrow{\hat{f}:=\operatorname{id}}\mathbb{C}P^{1}\times\mathfrak{T}^{1}_{1}\\ &\downarrow\\ &\mathfrak{T}^{1}_{1}\end{split}

    Then the sequence of maps fif^{i} are given by the restriction of the inverse image of f^\hat{f} to the inverse image of pip^{i}. These converge topologically to f^\hat{f} restricted to the inverse image of pp, where p𝔗11p\rightarrow\mathfrak{T}^{1}_{1} is any point so that pip\lceil p^{i}\rceil\rightarrow\lceil p\rceil.

  3. 3.

    Expanding example 1 to see different behavior, consider the map π:𝔗22𝔗11\pi:\mathfrak{T}^{2}_{2}\longrightarrow\mathfrak{T}^{1}_{1} so that πz~=w~1w~2\pi^{*}\tilde{z}=\tilde{w}_{1}\tilde{w}_{2}. This was studied in example 2.47 on page 2.47. This can be considered to be a family of annuli by restricting to the subset where |w~1|<1\left\lvert\tilde{w}_{1}\right\rvert<1 and |w~2|<1\left\lvert\tilde{w}_{2}\right\rvert<1. Then π1(pi)\pi^{-1}(p^{i}) is an annulus 1>|w~1|>1i1>\left\lvert\tilde{w}_{1}\right\rvert>\frac{1}{i}, or 1i<|w~2|<1\frac{1}{i}<\left\lvert\tilde{w}_{2}\right\rvert<1. These domains converge topologically to π1(p)\pi^{-1}(p) where p=c𝔱ap=c\mathfrak{t}^{a} for any cc\in\mathbb{C}^{*} and a>0a>0. π1(c𝔱a)\pi^{-1}(c\mathfrak{t}^{a}) has two coordinate charts which are subsets of 𝔗11\mathfrak{T}^{1}_{1}, with coordinate w~i\tilde{w}_{i} so that |w~i|<1\left\lvert\tilde{w}_{i}\right\rvert<1, and w~i>𝔱a\lfloor\tilde{w}_{i}\rfloor>\mathfrak{t}^{a}. The transition map between these coordinate charts is given by w1=c𝔱aw21w_{1}=c\mathfrak{t}^{a}w_{2}^{-1}.

  4. 4.

    Let us expand on example 3 by adding in the information of maps fif^{i} from our domains π1(pi)\pi^{-1}(p^{i}) to 𝔗2\mathfrak{T}^{2}. Suppose that

    fi(w~1)=(c~1iw~1,c~2iw~1)f^{i}(\tilde{w}_{1})=(\tilde{c}_{1}^{i}\tilde{w}_{1},\tilde{c}_{2}^{i}\tilde{w}_{1})

    so of course,

    fi(w~2)=(c~1i1iw~21,c~2i1iw~21)f^{i}(\tilde{w}_{2})=(\tilde{c}_{1}^{i}\frac{1}{i}\tilde{w}_{2}^{-1},\tilde{c}_{2}^{i}\frac{1}{i}\tilde{w}_{2}^{-1})

    If (c1i,c2i)𝔗2(c^{i}_{1},c_{2}^{i})\in\mathfrak{T}^{2} are generically chosen, we will not have a unique limit even if we restrict to the bounded domain 1>|w~1|>121>\left\lvert\tilde{w}_{1}\right\rvert>\frac{1}{2}. To put all our fif^{i} into an individual family, consider the family

    𝔗2×𝔗11f^(c~1,c~2,w~1,w~2):=(c~1w~1,c~2w~1)𝔗2id×π𝔗2×𝔗11\begin{split}\mathfrak{T}^{2}&\times\mathfrak{T}^{1}_{1}\xrightarrow{\hat{f}(\tilde{c}_{1},\tilde{c}_{2},\tilde{w}_{1},\tilde{w}_{2}):=(\tilde{c}_{1}\tilde{w}_{1},\tilde{c}_{2}\tilde{w}_{1})}\mathfrak{T}^{2}\\ &\downarrow\operatorname{id}\times\pi\\ \mathfrak{T}^{2}&\times\mathfrak{T}^{1}_{1}\end{split}

    Our individual maps fif^{i} are the restriction of f^\hat{f} to the fiber over (c~1i,c~2i,pi)(\tilde{c}^{i}_{1},\tilde{c}^{i}_{2},p^{i}). These converge topologically to the restriction of f^\hat{f} to the fiber over any point (c~1,c~2,c𝔱a)(\tilde{c}_{1},\tilde{c}_{2},c\mathfrak{t}^{a}) where a>0a>0. Note that if we restricted to the domains where 1>|w~1|>|w~2|1>\left\lvert\tilde{w}_{1}\right\rvert>\left\lvert\tilde{w}_{2}\right\rvert or 1>|w~2|>|w~1|1>\left\lvert\tilde{w}_{2}\right\rvert>\left\lvert\tilde{w}_{1}\right\rvert, then we get a 𝔗2\mathfrak{T}^{2} worth of valid topological limits for each domain. There is a non trivial requirement that needs to be satisfied for these topological limits to be glued together.

Apart from an easy use of an Arzela Ascoli type argument and elliptic bootstrapping, the main hurdle to proving Theorem 4.1 is dealing with the above types of non uniqueness for limits in showing that the different limiting pieces of our holomorphic curves glue together. This is not extremely difficult, but it requires a lot of notation to keep track of everything.

Note the assumption that 𝔅\mathfrak{B} is basic is mainly for convenience in the following arguments. In the family case, by restricting to a subset of 𝔊\mathfrak{G} which contains the image of a subsequence, we can also assume that 𝔅^\hat{\mathfrak{B}} is basic.

The following lemma will give us good coordinate charts on 𝔅\mathfrak{B} or 𝔅^\hat{\mathfrak{B}}. Recall that if 𝔅\mathfrak{B} is basic, we use the notation 𝔅i\lfloor\mathfrak{B}_{i}\rfloor for a strata of the tropical part 𝔅\lfloor\mathfrak{B}\rfloor, and 𝔅i¯\overline{\lfloor\mathfrak{B}_{i}\rfloor} for the polygon which after identifying some strata is equal to the closure of 𝔅i𝔅\lfloor\mathfrak{B}_{i}\rfloor\subset\lfloor\mathfrak{B}\rfloor. A neighborhood of a strata 𝔅i\mathfrak{B}_{i} is then equal to an open subset of 𝔅i𝔗𝔅i¯n\lceil\mathfrak{B}_{i}\rceil\rtimes\mathfrak{T}^{n}_{\overline{\lfloor\mathfrak{B}_{i}\rfloor}} using the construction of example 2.17 on page 2.17. This has a (sometimes defined) action of 𝔗n\mathfrak{T}^{n} action corresponding to the (sometimes defined) action of 𝔗n\mathfrak{T}^{n} on 𝔗𝔅i¯n\mathfrak{T}^{n}_{\overline{\lfloor\mathfrak{B}_{i}\rfloor}} given by coordinate wise multiplication. We will need this 𝔗n\mathfrak{T}^{n} action to move parts of our holomorphic curves around.

Lemma 4.3.

Given any basic exploded 𝕋\mathbb{T} fibration 𝔅^\hat{\mathfrak{B}}, and a family 𝔅^𝔊\hat{\mathfrak{B}}\longrightarrow\mathfrak{G}, for each strata 𝔅^i𝔅^\hat{\mathfrak{B}}_{i}\subset\hat{\mathfrak{B}}, there exists some 𝔘i𝔅^\mathfrak{U}_{i}\subset\hat{\mathfrak{B}} containing 𝔅^i\hat{\mathfrak{B}}_{i} so that

  1. 1.

    The image in the smooth part 𝔘i𝔅\lceil\mathfrak{U}_{i}\rceil\subset\lceil\mathfrak{B}\rceil is an open neighborhood of 𝔅i𝔅\lceil\mathfrak{B}_{i}\rceil\subset\lceil\mathfrak{B}\rceil.

  2. 2.

    If 𝔅^i\lfloor\hat{\mathfrak{B}}_{i}\rfloor is nn dimensional, then there is an identification of 𝔘i\mathfrak{U}_{i} with an open subset of 𝔅i𝔗𝔅i¯n\lceil\mathfrak{B}_{i}\rceil\rtimes\mathfrak{T}^{n}_{\overline{\lfloor\mathfrak{B}_{i}\rfloor}} using the construction on page 2.17. We can make this identification so that the (sometimes defined) free action of 𝔗n\mathfrak{T}^{n} given by considering 𝔅i𝔗𝔅i¯n𝔅i𝔗n\lceil\mathfrak{B}_{i}\rceil\rtimes\mathfrak{T}^{n}_{\overline{\lfloor\mathfrak{B}_{i}\rfloor}}\subset\lceil\mathfrak{B}_{i}\rceil\rtimes\mathfrak{T}^{n} satisfies the following:

    1. (a)

      The 𝔗n\mathfrak{T}^{n} action preserves fibers of the family 𝔅^𝔊\hat{\mathfrak{B}}\longrightarrow\mathfrak{G}, in the sense that if p1p_{1} and p2p_{2} have the same image in 𝔊\mathfrak{G}, then z~p1\tilde{z}*p_{1} and z~p2\tilde{z}*p_{2} also have the same image in 𝔊\mathfrak{G}.

    2. (b)

      For any point p𝔘ip\longrightarrow\mathfrak{U}_{i}, the set of values in 𝔗n\mathfrak{T}^{n} for which this action is defined is nonempty and convex in the sense that it is given by a set of inequalities of the form

      |z~α|>c𝔱x𝔱\left\lvert\tilde{z}^{\alpha}\right\rvert>c\mathfrak{t}^{x}\in\mathbb{R}\mathfrak{t}^{\mathbb{R}}

      where c𝔱xc\mathfrak{t}^{x} depends smoothly on p𝔅^p\longrightarrow\hat{\mathfrak{B}}.

    3. (c)

      If 𝔅^i\lfloor\hat{\mathfrak{B}}_{i}\rfloor is in the closure of 𝔅^j\lfloor\hat{\mathfrak{B}}_{j}\rfloor, then the action of 𝔗n\mathfrak{T}^{n} on 𝔘i𝔘j\mathfrak{U}_{i}\cap\mathfrak{U}_{j} is equal to the action of a subgroup of the 𝔗m\mathfrak{T}^{m} acting on 𝔘j\mathfrak{U}_{j}.

Proof:

The only point in the above lemma that requires proof is the compatibility of the actions of 𝔗n\mathfrak{T}^{n} with restriction and the family. We can construct this first for the strata 𝔅^i\lfloor\hat{\mathfrak{B}}_{i}\rfloor with the highest dimension and then extend it to lower dimensional strata.

Being able to do this amounts to constructing coordinate charts identified with open subsets of k×𝔗Bn\mathbb{R}^{k}\times\mathfrak{T}^{n}_{B}, where coordinates on the projection of this to 𝔊\mathfrak{G} consist of some sub collection of these coordinates (the normal form for coordinates on a family), and so that transition functions are of the form

(x,w~)(ϕ(x),f(x)𝔱aw~)(x,\tilde{w})\mapsto(\phi(x),f(x)\mathfrak{t}^{a}\tilde{w})

where ϕ\phi is a diffeomorphism, and ff is a smooth \mathbb{C}^{*} valued function. These transition functions preserve a 𝔗n\mathfrak{T}^{n} action of the form

z~(x,w~)=(x,z~w~)\tilde{z}*(x,\tilde{w})=(x,\tilde{z}\tilde{w})

Because of the assumption that the closure of strata in 𝔅^\lfloor\hat{\mathfrak{B}}\rfloor are simply connected, we can first reduce to the case that coordinate charts are in the normal form for a family, and transition functions are of the form

(x,w~)(ϕ(x),f(x,w)𝔱aw~)(x,\tilde{w})\mapsto(\phi(x),f(x,w)\mathfrak{t}^{a}\tilde{w})

where ff is smooth and ()n(\mathbb{C}^{*})^{n} valued. There is then no obstruction to modifying our coordinate charts one by one so that the transition functions no longer have any dependance on ww. (This amounts to replacing the coordinates w~\tilde{w} with f(x,w)f(x,0)w~\frac{f(x,w)}{f(x,0)}\tilde{w}. This change is well defined on the intersection with previously corrected coordinate charts, does not affect our charts being in the normal form for families, and there is no obstruction to extending it to the rest of a chart.)

\square

We shall now start working towards proving Theorem 4.1. The proof shall rely on Lemma 3.13 on page 3.13 and Proposition 3.17 on page 3.17. We shall be using the notation from Proposition 3.17. First, choose the following:

  1. 1.

    Choose a metric, a taming form ωΩ\omega\in\Omega, and finite collection of coordinate charts on 𝔅{\mathfrak{B}} each contained in some 𝔘i\mathfrak{U}_{i} from Lemma 4.3 and small enough to apply Lemma 3.13. For the case of a family, restrict 𝔅^𝔊\hat{\mathfrak{B}}\longrightarrow\mathfrak{G} to some small coordinate chart on 𝔊\mathfrak{G} in which the image of some subsequence converges, and choose our finite collection of coordinate charts on (the smaller, renamed,) 𝔅^\hat{\mathfrak{B}} satisfying the above.

  2. 2.

    Choose exploded annuli 𝔄nii\mathfrak{A}_{n}^{i}\subset\mathfrak{C}^{i} satisfying the conditions in Proposition 3.17 with ω\omega-energy bound ϵ\epsilon small enough and RR large enough that each 𝔄6R10,ni\mathfrak{A}^{i}_{\frac{6R}{10},n} is contained well inside some 𝔘i\mathfrak{U}_{i} in the sense that it is of distance greater than 22 to the boundary of 𝔘i\mathfrak{U}_{i}. (The fact that we can achieve this follows from Lemma 3.11 on page 3.11.) Also choose ϵ\epsilon small enough and RR large enough so that each connected smooth component of 𝔄6R10,ni\mathfrak{A}^{i}_{\frac{6R}{10},n} is contained in one of the above coordinate charts and can have Lemma 3.13 applied to it.

Our taming form ω\omega is a smooth two form on some refinement of 𝔅^\hat{\mathfrak{B}}. As convergence in a refinement is stronger, we will simply call this refinement 𝔅^\hat{\mathfrak{B}}.

Use the notation CmiC^{i}_{m} to indicate connected components of ik𝔄8R10,ki\mathfrak{C}^{i}-\bigcup_{k}\mathfrak{A}^{i}_{\frac{8R}{10},k}. We can choose a subsequence so that the number of such components is the same for each i\mathfrak{C}^{i}, CmiC^{i}_{m} has topology that is independent of ii, and choosing diffeomorphisms identifying them, CmiC^{i}_{m} converges as ii\rightarrow\infty to some CmC_{m} in the sense that the metric from Proposition 3.17 and the complex structure converges to one on CmC_{m}.

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\psfrag{ETF19a}{$M_{i}:=\lceil\mathfrak{B}_{i}\rceil$}\psfrag{ETF19b}{$U_{i,j_{1}}$}\psfrag{ETF19c}{$U_{i,j_{2}}$}\psfrag{ETF19d}{$M_{j_{1}}$}\psfrag{ETF19e}{$M_{j_{2}}$}\psfrag{ETF19f}{$M_{k}$}\psfrag{ETF19g}{$r_{i,j_{1}}$}\psfrag{ETF19h}{$r_{i,j_{2}}$}\psfrag{ETF19i}{$r_{j_{1},k}$}\psfrag{ETF19j}{$r_{j_{2},k}$}\psfrag{ETF20a}{$\lfloor\tilde{r}_{1,j}\rfloor$}\psfrag{ETF20b}{$\lfloor\tilde{r}_{2,j}\rfloor$}\psfrag{ETF20c}{$\lfloor\tilde{r}_{3,j}\rfloor$}\psfrag{ETF20d}{$\lfloor\tilde{r}_{1,j}\rfloor\lfloor\tilde{r}_{2,j}\rfloor\lfloor\tilde{r}_{3,j}\rfloor=\mathfrak{t}^{c}$}\psfrag{ETF20e}{$\lfloor\mathfrak{B}_{j}\rfloor$}\psfrag{ETF21a}{$\lfloor p\rfloor$}\psfrag{ETF8a}{$\lceil\mathfrak{B}\rceil$}\psfrag{ETF8b}{$\lfloor\mathfrak{B}\rfloor$}\psfrag{ETF8c}{$\lceil\mathfrak{B}^{\prime}\rceil$}\psfrag{ETF8d}{$\lfloor\mathfrak{B}^{\prime}\rfloor$}\psfrag{ETF11d}{{\footnotesize derivative bounded here}}\psfrag{ETF11c}{{\footnotesize components with $\omega$ energy bounded below}}\psfrag{ETF11b}{$\mathfrak{A}$}\psfrag{ETF11a}{$\mathfrak{A}_{R}${\footnotesize:well behaved annulus}}\psfrag{ETF12c}{$\omega$-energy $<\frac{\epsilon}{5}$}\psfrag{ETF12b}{$\mathfrak{A}$}\psfrag{ETF12a}{$\omega$-energy $>\epsilon_{0}$}\psfrag{ETF13a}{$\mathfrak{A}_{R}$}\psfrag{ETF13b}{$\mathfrak{A}_{\frac{R}{2}}$}\psfrag{ETF13c}{New low energy annulus}\psfrag{ETF13d}{Disk with $\omega$-energy $>\epsilon_{0}$}\psfrag{ETF14a}{New low energy annulus}\psfrag{ETF14b}{$\mathfrak{A}_{\frac{R}{2}}$}\psfrag{ETF14c}{$\mathfrak{A}_{R}$}\psfrag{ETF15a}{$\mathfrak{A}_{1}$}\psfrag{ETF15b}{$\mathfrak{A}_{2}$}\psfrag{ETF15c}{$\mathfrak{A}_{3}$}\psfrag{ETF15d}{$\mathfrak{A}_{4}$}\psfrag{ETF15e}{$\mathfrak{A}_{\frac{R}{2},[1,3]}$ has low $\omega$-energy}\psfrag{ETF16a}{$C^{i}_{1}=(C_{1},j_{i})$}\psfrag{ETF16b}{$(C_{2},j_{i})$}\psfrag{ETF16c}{$(C_{3},j_{i})$}\psfrag{ETF16d}{$\mathfrak{A}^{i}_{4}$}\psfrag{ETF16e}{$\mathfrak{A}^{i}_{5}$}\psfrag{ETF16f}{$\mathfrak{A}^{i}_{6}$}\psfrag{ETF16g}{Transition functions bounded}\includegraphics{ETF16}

We shall choose two coordinate charts on 𝔄ni\mathfrak{A}_{n}^{i} with coordinates z~n±\tilde{z}_{n}^{\pm} in a subset of 𝔗11\mathfrak{T}^{1}_{1} so that each boundary of 𝔄ni\mathfrak{A}_{n}^{i} is identified with |z~n±|=1\left\lvert\tilde{z}_{n}^{\pm}\right\rvert=1, and the coordinates are related by z~n+z~n=Qni𝔱+\tilde{z}_{n}^{+}\tilde{z}_{n}^{-}=Q^{i}_{n}\in\mathbb{R}^{*}\mathfrak{t}^{\mathbb{R}^{+}}. (This means that the conformal modulus of 𝔄ni\mathfrak{A}_{n}^{i} is logQni-\log Q^{i}_{n}. These QnQ_{n} will later become coordinates on our family 𝔉\mathfrak{F}.) Choose a subsequence so that the number of annuli is independent of ii, and {Qni}\{Q^{i}_{n}\} converges topologically. (In other words, calling any conformal modulus not a real number infinity, either the conformal modulus of 𝔄ni\mathfrak{A}_{n}^{i} converges to a finite number, or converges to infinity. )

We can define

𝔄n±:={|z~n±|e7R10}𝔗11\mathfrak{A}^{\pm}_{n}:=\{\left\lvert\tilde{z}^{\pm}_{n}\right\rvert\leq e^{-\frac{7R}{10}}\}\subset\mathfrak{T}^{1}_{1}
𝔄7R10,ni:={(z~n+,z~n) so that z~n+z~n=Qni}𝔄n+×𝔄n\mathfrak{A}^{i}_{\frac{7R}{10},n}:=\{(\tilde{z}^{+}_{n},\tilde{z}^{-}_{n})\text{ so that }\tilde{z}^{+}_{n}\tilde{z}^{-}_{n}=Q^{i}_{n}\}\subset\mathfrak{A}^{+}_{n}\times\mathfrak{A}^{-}_{n}

We shall need to keep track of what 𝔄n\mathfrak{A}_{n} is attached to, so use the notation n±n^{\pm} to define Cn±iC^{i}_{n^{\pm}} as the component attached to the end of 𝔄ni\mathfrak{A}_{n}^{i} with the boundary |z~n±|=1\left\lvert\tilde{z}^{\pm}_{n}\right\rvert=1. (Assume, by passing to a subsequence, that n±n^{\pm} is well defined independent of ii.)

We can consider transition maps between Cn±iC^{i}_{n^{\pm}} and 𝔄7R10,ni\mathfrak{A}^{i}_{\frac{7R}{10},n} to give transition maps between Cn±C_{n^{\pm}} and 𝔄n±\mathfrak{A}^{\pm}_{n}. The bound on the derivative of transition maps from Proposition 3.17 on the region 𝔄9R10,ni𝔄6R10,ni\mathfrak{A}^{i}_{\frac{9R}{10},n}-\mathfrak{A}^{i}_{\frac{6R}{10},n} (and standard elliptic bootstrapping to get bounds on higher derivatives on the smaller region 𝔄8R10,ni𝔄7R10,ni\mathfrak{A}^{i}_{\frac{8R}{10},n}-\mathfrak{A}^{i}_{\frac{7R}{10},n}) tells us that we can choose a subsequence so that the transition maps between Cn±iC^{i}_{n^{\pm}} and 𝔄n±\mathfrak{A}^{\pm}_{n} converge to some smooth transition map between Cn±C_{n^{\pm}} and 𝔄n±\mathfrak{A}^{\pm}_{n}. We can modify our diffeomorphisms identifying CmiC_{m}^{i} with CmC_{m} so that these transition maps all give exactly the same map. Denote the complex structure on CmC_{m} induced by this identification by jij_{i}. (We shall do something similar for the annuli 𝔄ni\mathfrak{A}^{i}_{n} later on.) We can do this so that jij_{i} and the metrics given by this identification still converge in CC^{\infty} to those on CmC_{m}.

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\psfrag{ETF19a}{$M_{i}:=\lceil\mathfrak{B}_{i}\rceil$}\psfrag{ETF19b}{$U_{i,j_{1}}$}\psfrag{ETF19c}{$U_{i,j_{2}}$}\psfrag{ETF19d}{$M_{j_{1}}$}\psfrag{ETF19e}{$M_{j_{2}}$}\psfrag{ETF19f}{$M_{k}$}\psfrag{ETF19g}{$r_{i,j_{1}}$}\psfrag{ETF19h}{$r_{i,j_{2}}$}\psfrag{ETF19i}{$r_{j_{1},k}$}\psfrag{ETF19j}{$r_{j_{2},k}$}\psfrag{ETF20a}{$\lfloor\tilde{r}_{1,j}\rfloor$}\psfrag{ETF20b}{$\lfloor\tilde{r}_{2,j}\rfloor$}\psfrag{ETF20c}{$\lfloor\tilde{r}_{3,j}\rfloor$}\psfrag{ETF20d}{$\lfloor\tilde{r}_{1,j}\rfloor\lfloor\tilde{r}_{2,j}\rfloor\lfloor\tilde{r}_{3,j}\rfloor=\mathfrak{t}^{c}$}\psfrag{ETF20e}{$\lfloor\mathfrak{B}_{j}\rfloor$}\psfrag{ETF21a}{$\lfloor p\rfloor$}\psfrag{ETF8a}{$\lceil\mathfrak{B}\rceil$}\psfrag{ETF8b}{$\lfloor\mathfrak{B}\rfloor$}\psfrag{ETF8c}{$\lceil\mathfrak{B}^{\prime}\rceil$}\psfrag{ETF8d}{$\lfloor\mathfrak{B}^{\prime}\rfloor$}\psfrag{ETF11d}{{\footnotesize derivative bounded here}}\psfrag{ETF11c}{{\footnotesize components with $\omega$ energy bounded below}}\psfrag{ETF11b}{$\mathfrak{A}$}\psfrag{ETF11a}{$\mathfrak{A}_{R}${\footnotesize:well behaved annulus}}\psfrag{ETF12c}{$\omega$-energy $<\frac{\epsilon}{5}$}\psfrag{ETF12b}{$\mathfrak{A}$}\psfrag{ETF12a}{$\omega$-energy $>\epsilon_{0}$}\psfrag{ETF13a}{$\mathfrak{A}_{R}$}\psfrag{ETF13b}{$\mathfrak{A}_{\frac{R}{2}}$}\psfrag{ETF13c}{New low energy annulus}\psfrag{ETF13d}{Disk with $\omega$-energy $>\epsilon_{0}$}\psfrag{ETF14a}{New low energy annulus}\psfrag{ETF14b}{$\mathfrak{A}_{\frac{R}{2}}$}\psfrag{ETF14c}{$\mathfrak{A}_{R}$}\psfrag{ETF15a}{$\mathfrak{A}_{1}$}\psfrag{ETF15b}{$\mathfrak{A}_{2}$}\psfrag{ETF15c}{$\mathfrak{A}_{3}$}\psfrag{ETF15d}{$\mathfrak{A}_{4}$}\psfrag{ETF15e}{$\mathfrak{A}_{\frac{R}{2},[1,3]}$ has low $\omega$-energy}\psfrag{ETF16a}{$C^{i}_{1}=(C_{1},j_{i})$}\psfrag{ETF16b}{$(C_{2},j_{i})$}\psfrag{ETF16c}{$(C_{3},j_{i})$}\psfrag{ETF16d}{$\mathfrak{A}^{i}_{4}$}\psfrag{ETF16e}{$\mathfrak{A}^{i}_{5}$}\psfrag{ETF16f}{$\mathfrak{A}^{i}_{6}$}\psfrag{ETF16g}{Transition functions bounded}\psfrag{ETF17a}{$(C_{1},j)$}\psfrag{ETF17a'}{$q_{1}$}\psfrag{ETF17b}{$(C_{2},j)$}\psfrag{ETF17b'}{$q_{2}$}\psfrag{ETF17c}{$(C_{3},j)$}\psfrag{ETF17c'}{$q_{3}$}\psfrag{ETF17d+}{$\mathfrak{A}^{+}_{4}$}\psfrag{ETF17d+a}{$4^{+}:=1$}\psfrag{ETF17d-}{$\mathfrak{A}^{-}_{4}$}\psfrag{ETF17d-b}{$4^{-}:=2$}\psfrag{ETF17e}{$\mathfrak{A}^{+}_{5}$}\psfrag{ETF17eb}{$5^{+}:=2$}\psfrag{ETF17f+}{$\mathfrak{A}_{6}^{+}$}\psfrag{ETF17f-}{$\mathfrak{A}_{6}^{-}$}\psfrag{ETF17f+c}{$6^{+}:=3$}\psfrag{ETF17f-b}{$6^{-}:=2$}\includegraphics{ETF17}

We shall now start to construct our family ^𝔉\mathfrak{\hat{C}}\longrightarrow\mathfrak{F}. The first step shall be to get some kind of convergence for each piece of our holomorphic curve.

For each CmC_{m} choose some point qmCmq_{m}\in C_{m}, and consider Qmi:=fi(qm)𝔅^Q^{i}_{m}:=f^{i}(q_{m})\in\hat{\mathfrak{B}}. This sequence has a subsequence that converges in 𝔅^\lceil\hat{\mathfrak{B}}\rceil to some point Qm𝔅^Q_{m}\longrightarrow\hat{\mathfrak{B}}. Label the strata of 𝔅^\hat{\mathfrak{B}} that contains QmQ_{m} by 𝔅^m\hat{\mathfrak{B}}_{m}, and consider the chart 𝔘m\mathfrak{U}_{m} from Lemma 4.3 containing 𝔅^m\hat{\mathfrak{B}}_{m}. As the derivative of fif^{i} restricted to CmC_{m} is bounded, we can choose a subsequence so that fi(Cm)f^{i}(C_{m}) is contained well inside 𝔘m\mathfrak{U}_{m}. (We can choose a subsequence so that the image of fi(Cm)f^{i}(C_{m}) is contained in the subset of 𝔘m\mathfrak{U}_{m} which consists of all points some arbitrary distance from the boundary of 𝔘m\mathfrak{U}_{m}.)

As the derivative of fif^{i} is uniformly bounded on i𝔄9R10,ki\mathfrak{C}^{i}-\bigcup\mathfrak{A}^{i}_{\frac{9R}{10},k}, we can use lemmas 3.6, 3.7, and 3.8 to get bounds on the higher derivatives of fif^{i} restricted to CmiC^{i}_{m}. Lemma 3.1 then tells us that if fi(qm)f^{i}(q_{m}) converges topologically to Qm𝔅^Q_{m}\longrightarrow\hat{\mathfrak{B}}, the geometry around fi(qm)f^{i}(q_{m}) converges to that around QmQ_{m}, so we can choose a subsequence so that fif^{i} restricted to CmiC^{i}_{m} converges in some sense to a map

fm,Qm:Cm𝔅^ so that fm,Qm(qm)=Qm𝔘mf_{m,Q_{m}}:C_{m}\longrightarrow\hat{\mathfrak{B}}\text{ so that }f_{m,Q_{m}}(q_{m})=Q_{m}\subset\mathfrak{U}_{m}

More specifically, remembering that everything is contained inside 𝔘m\mathfrak{U}_{m}, we can use our 𝔗n\mathfrak{T}^{n} action on 𝔘m\mathfrak{U}_{m} to say this more precisely. There exists some sequence c~mi𝔗n\tilde{c}_{m}^{i}\in\mathfrak{T}^{n} so that

c~mifi:Cm𝔘m\tilde{c}_{m}^{i}*f^{i}:C_{m}\longrightarrow\mathfrak{U}_{m}

converges in CC^{\infty} to fm,Qm:Cm𝔘mf_{m,Q_{m}}:C_{m}\longrightarrow\mathfrak{U}_{m}.

Of course, there was a choice involved here. We could also have chosen a different topological limit QmQ^{\prime}_{m} of fi(qm)f^{i}(q_{m}), which would give a translate of fm,Qmf_{m,Q_{m}} by our 𝔗n\mathfrak{T}^{n} action. These choices will turn up as parameters on our family. They will need to be ‘compatible’ in the sense that these pieces will need to fit together.

We now consider the analogous convergence on annular regions. Because the ω\omega-energy of fif^{i} restricted to our annular regions 𝔄ni\mathfrak{A}^{i}_{n} is small, we can apply Lemma 3.13 on page 3.13 to tell us that if the limit of the conformal modulus of 𝔄ni\mathfrak{A}^{i}_{n} is infinite, then fif^{i} restricted to the smooth parts of 𝔄ni\mathfrak{A}^{i}_{n} converges in some sense (considered in more detail later) to some unique pair of holomorphic maps

fn,Q±:{e7R10>|z~n±|>0}𝔅^f^{\pm}_{n,Q}:\{e^{-\frac{7R}{10}}>\left\lvert\tilde{z}_{n}^{\pm}\right\rvert>0\}\longrightarrow\hat{\mathfrak{B}}

compatible with fn±,Qn±f_{n^{\pm},Q_{n^{\pm}}} and the transition maps. (Recall that Cn±C_{n^{\pm}} is the component attached to 𝔄n±\mathfrak{A}^{\pm}_{n}.) As we have made the assumption that JJ is civilized, we can use the usual removable singularity theorem for finite energy holomorphic curves to see that these limit maps extend uniquely to smooth maps on {e7R10>|z~n±|>𝔱x}𝔄n±\{e^{-\frac{7R}{10}}>\left\lvert\tilde{z}_{n}^{\pm}\right\rvert>\mathfrak{t}^{x}\}\subset\mathfrak{A}^{\pm}_{n} for some x>0x>0.

\psfrag{ETF1math}{$\lfloor\mathfrak{T}^{1}_{1}\rfloor:=\mathfrak{t}^{\mathbb{R}^{+}}$}\psfrag{ETF1mathC}{$\lceil\mathfrak{T}^{1}_{1}\rceil:=\mathbb{C}$}\psfrag{ETF1mathtot}{A picture of $[\mathfrak{T}^{1}_{1}]$}\psfrag{ETF1Cs}{$\mathbb{C}^{*}$}\psfrag{ETF3tot}{$[\mathfrak{C}]$}\psfrag{ETF3totl}{$\lceil\mathfrak{C}\rceil$}\psfrag{ETF3totb}{$\lfloor\mathfrak{C}\rfloor$}\psfrag{ETF4manifold}{$\lceil\operatorname{Expl}M\rceil=M$}\psfrag{ETF4totb}{$\lfloor\operatorname{Expl}M\rfloor$}\psfrag{ETF7v1}{$v^{1}$}\psfrag{ETF7v2}{$v^{2}$}\psfrag{ETF7v3}{$v^{3}$}\psfrag{ETF7v4}{$v^{4}$}\psfrag{ETF7tot}{$[\mathfrak{C}]$}\psfrag{ETF7totb}{$\lfloor f(\mathfrak{C})\rfloor\subset\lfloor\mathfrak{T}^{n}\rfloor$}\psfrag{ETF7eqn}{$\sum df(v^{i})=0$}\psfrag{ETF6M}{$\lceil\mathfrak{M}\rceil$}\psfrag{ETF6tot}{$[\mathfrak{M}]$}\psfrag{ETF6totb}{$\lfloor\mathfrak{M}\rfloor$}\psfrag{ETF19a}{$M_{i}:=\lceil\mathfrak{B}_{i}\rceil$}\psfrag{ETF19b}{$U_{i,j_{1}}$}\psfrag{ETF19c}{$U_{i,j_{2}}$}\psfrag{ETF19d}{$M_{j_{1}}$}\psfrag{ETF19e}{$M_{j_{2}}$}\psfrag{ETF19f}{$M_{k}$}\psfrag{ETF19g}{$r_{i,j_{1}}$}\psfrag{ETF19h}{$r_{i,j_{2}}$}\psfrag{ETF19i}{$r_{j_{1},k}$}\psfrag{ETF19j}{$r_{j_{2},k}$}\psfrag{ETF20a}{$\lfloor\tilde{r}_{1,j}\rfloor$}\psfrag{ETF20b}{$\lfloor\tilde{r}_{2,j}\rfloor$}\psfrag{ETF20c}{$\lfloor\tilde{r}_{3,j}\rfloor$}\psfrag{ETF20d}{$\lfloor\tilde{r}_{1,j}\rfloor\lfloor\tilde{r}_{2,j}\rfloor\lfloor\tilde{r}_{3,j}\rfloor=\mathfrak{t}^{c}$}\psfrag{ETF20e}{$\lfloor\mathfrak{B}_{j}\rfloor$}\psfrag{ETF21a}{$\lfloor p\rfloor$}\psfrag{ETF8a}{$\lceil\mathfrak{B}\rceil$}\psfrag{ETF8b}{$\lfloor\mathfrak{B}\rfloor$}\psfrag{ETF8c}{$\lceil\mathfrak{B}^{\prime}\rceil$}\psfrag{ETF8d}{$\lfloor\mathfrak{B}^{\prime}\rfloor$}\psfrag{ETF11d}{{\footnotesize derivative bounded here}}\psfrag{ETF11c}{{\footnotesize components with $\omega$ energy bounded below}}\psfrag{ETF11b}{$\mathfrak{A}$}\psfrag{ETF11a}{$\mathfrak{A}_{R}${\footnotesize:well behaved annulus}}\psfrag{ETF12c}{$\omega$-energy $<\frac{\epsilon}{5}$}\psfrag{ETF12b}{$\mathfrak{A}$}\psfrag{ETF12a}{$\omega$-energy $>\epsilon_{0}$}\psfrag{ETF13a}{$\mathfrak{A}_{R}$}\psfrag{ETF13b}{$\mathfrak{A}_{\frac{R}{2}}$}\psfrag{ETF13c}{New low energy annulus}\psfrag{ETF13d}{Disk with $\omega$-energy $>\epsilon_{0}$}\psfrag{ETF14a}{New low energy annulus}\psfrag{ETF14b}{$\mathfrak{A}_{\frac{R}{2}}$}\psfrag{ETF14c}{$\mathfrak{A}_{R}$}\psfrag{ETF15a}{$\mathfrak{A}_{1}$}\psfrag{ETF15b}{$\mathfrak{A}_{2}$}\psfrag{ETF15c}{$\mathfrak{A}_{3}$}\psfrag{ETF15d}{$\mathfrak{A}_{4}$}\psfrag{ETF15e}{$\mathfrak{A}_{\frac{R}{2},[1,3]}$ has low $\omega$-energy}\psfrag{ETF16a}{$C^{i}_{1}=(C_{1},j_{i})$}\psfrag{ETF16b}{$(C_{2},j_{i})$}\psfrag{ETF16c}{$(C_{3},j_{i})$}\psfrag{ETF16d}{$\mathfrak{A}^{i}_{4}$}\psfrag{ETF16e}{$\mathfrak{A}^{i}_{5}$}\psfrag{ETF16f}{$\mathfrak{A}^{i}_{6}$}\psfrag{ETF16g}{Transition functions bounded}\psfrag{ETF17a}{$(C_{1},j)$}\psfrag{ETF17a'}{$q_{1}$}\psfrag{ETF17b}{$(C_{2},j)$}\psfrag{ETF17b'}{$q_{2}$}\psfrag{ETF17c}{$(C_{3},j)$}\psfrag{ETF17c'}{$q_{3}$}\psfrag{ETF17d+}{$\mathfrak{A}^{+}_{4}$}\psfrag{ETF17d+a}{$4^{+}:=1$}\psfrag{ETF17d-}{$\mathfrak{A}^{-}_{4}$}\psfrag{ETF17d-b}{$4^{-}:=2$}\psfrag{ETF17e}{$\mathfrak{A}^{+}_{5}$}\psfrag{ETF17eb}{$5^{+}:=2$}\psfrag{ETF17f+}{$\mathfrak{A}_{6}^{+}$}\psfrag{ETF17f-}{$\mathfrak{A}_{6}^{-}$}\psfrag{ETF17f+c}{$6^{+}:=3$}\psfrag{ETF17f-b}{$6^{-}:=2$}\psfrag{ETF18a}{$C_{4^{+}}:=C_{1}$}\psfrag{ETF18a'}{$q_{1}$}\psfrag{ETF18aa}{${f_{1,Q_{1}}}:C_{1}\longrightarrow\mathfrak{U}_{1}$}\psfrag{ETF18ab}{$q_{1}\mapsto Q_{1}\in\mathfrak{U}_{1}$}\psfrag{ETF18d+}{$\xrightarrow{f^{+}_{4,Q}}\mathfrak{U}_{4}$}\psfrag{ETF18d-}{$\mathfrak{U}_{4}\xleftarrow{f^{-}_{4,Q}}$}\psfrag{ETF18dd}{$\tilde{z}_{4}^{+}\tilde{z}_{4}^{-}=Q_{4}$}\psfrag{ETF18ba}{$C_{2}\xrightarrow{f_{1,Q_{2}}}\mathfrak{U}_{2}$}\psfrag{ETF18bb}{$q_{2}\mapsto Q_{2}\in\mathfrak{U}_{2}$}\psfrag{ETF18b}{$C_{4^{-}}:=C_{2}$}\psfrag{ETF18b'}{$q_{2}$}\includegraphics{ETF18}

We may assume after passing to a subsequence that fi(𝔄6R10,ni)f^{i}(\mathfrak{A}^{i}_{\frac{6R}{10},n}) are all contained inside a single 𝔘n\mathfrak{U}_{n}. Recalling our convention that Cn±C_{n^{\pm}} intersects 𝔄ni\mathfrak{A}^{i}_{n}, note that this means that 𝔘n±\mathfrak{U}_{n^{\pm}} intersects 𝔘n\mathfrak{U}_{n} (and 𝔘n±𝔘n\lfloor\mathfrak{U}_{n^{\pm}}\rfloor\subset\lfloor\mathfrak{U}_{n}\rfloor), so our sequence c~n±i\tilde{c}^{i}_{n^{\pm}} defined earlier also has an action on 𝔘n\mathfrak{U}_{n}, and

c~n±ifi:𝔄7R10,ni𝔘n\tilde{c}^{i}_{n^{\pm}}*f^{i}:\mathfrak{A}_{\frac{7R}{10},n}^{i}\longrightarrow\mathfrak{U}_{n}

converges in CC^{\infty} on compact subsets adjacent to Cn±C_{n^{\pm}} to fn,Q±f^{\pm}_{n,Q} .

We now have a type of convergence of the individual pieces we have cut our holomorphic curve into. We shall now define a model for the exploded structure on our family that is too large, as it ignores the requirement that these pieces must fit together.

Define 𝔙m𝔘m\mathfrak{V}_{m}\subset\mathfrak{U}_{m} to be the exploded fibration consisting of all points Qm𝔘mQ_{m}^{\prime}\longrightarrow\mathfrak{U}_{m} so that

  1. 1.

    There exists some z~\tilde{z} so that Qm=z~QmQ_{m}^{\prime}=\tilde{z}*Q_{m}

  2. 2.

    Defining

    fm,Qm:=z~fm,Qmf_{m,Q_{m}^{\prime}}:=\tilde{z}*f_{m,Q_{m}}

    The image of CmC_{m}, fm,Qm(Cm)f_{m,Q^{\prime}_{m}}(C_{m}) is contained well inside 𝔘m\mathfrak{U}_{m} in the sense that the distance to the boundary of 𝔘m\mathfrak{U}_{m} is greater than 11.

  3. 3.

    If CmC_{m} is attached to 𝔄n±\mathfrak{A}^{\pm}_{n}, then defining

    fn,Q±:=z~fn,Q±f^{\pm}_{n,Q^{\prime}}:=\tilde{z}*f^{\pm}_{n,Q}

    The image of the smooth part of 𝔄n±\mathfrak{A}^{\pm}_{n}, {fn,Q±(z~), 0<|z~|<e7R10}\{f^{\pm}_{n,Q^{\prime}}(\tilde{z}),\ 0<\left\lvert\tilde{z}\right\rvert<e^{-\frac{7R}{10}}\} is also contained well inside 𝔘n\mathfrak{U}_{n}.

Note that 𝔙m\mathfrak{V}_{m} is a smooth exploded fibration which includes all QmQ^{\prime}_{m} which are topological limits for QmiQ^{i}_{m}. For such QmQ^{\prime}_{m}, the map fm,Qmf_{m,Q^{\prime}_{m}} will be holomorphic, but for other QmQ^{\prime}_{m}, this may not be the case.

We shall consider the family 𝔉\mathfrak{F} as a sub exploded fibration of

m𝔙mn𝔔nm𝔘mn𝔔n\prod_{m}\mathfrak{V}_{m}\prod_{n}\mathfrak{Q}_{n}\subset\prod_{m}\mathfrak{U}_{m}\prod_{n}\mathfrak{Q}_{n}

The 𝔔n\mathfrak{Q}_{n} above stands for the ‘gluing parameter’ for identifying coordinates z~n±\tilde{z}_{n}^{\pm} on 𝔄n±\mathfrak{A}^{\pm}_{n} via

z~n+z~n=Qn𝔔n:={|z~|<e2R}𝔗11\tilde{z}_{n}^{+}\tilde{z}_{n}^{-}=Q_{n}\in\mathfrak{Q}_{n}:=\{\left\lvert\tilde{z}\right\rvert<e^{-2R}\}\subset\mathfrak{T}^{1}_{1}

We shall consider the following sequence of points

Qi:=(fi(q1),fi(q2),,Qn1i,)m𝔘mn𝔔nQ^{i}:=(f^{i}(q_{1}),f^{i}(q_{2}),\dotsc,Q^{i}_{n_{1}},\dotsc)\in\prod_{m}\mathfrak{U}_{m}\prod_{n}\mathfrak{Q}_{n}

where the conformal modulus of 𝔄ni\mathfrak{A}^{i}_{n} is equal to logQni-\log Q^{i}_{n}.

Note that there is a transitive (sometimes defined) action of 𝔗k\mathfrak{T}^{k} on 𝔙m𝔔n\prod\mathfrak{V}_{m}\prod\mathfrak{Q}_{n} which is the action from Lemma 4.3 on 𝔙m𝔘m\mathfrak{V}_{m}\subset\mathfrak{U}_{m}, and multiplication by some coordinate of 𝔗k\mathfrak{T}^{k} on 𝔔n\mathfrak{Q}_{n}. Our family will be given by a complete inclusion

𝔉m𝔙mn𝔔nm𝔘mn𝔔n\mathfrak{F}\longrightarrow\prod_{m}\mathfrak{V}_{m}\prod_{n}\mathfrak{Q}_{n}\subset\prod_{m}\mathfrak{U}_{m}\prod_{n}\mathfrak{Q}_{n}

satisfying the following conditions:

  1. 1.

    The image of 𝔉\mathfrak{F} contains some point Qm𝔙mn𝔔nQ\longrightarrow\prod_{m}\mathfrak{V}_{m}\prod_{n}\mathfrak{Q}_{n} which is a topological limit of QiQ^{i}, and the image of 𝔉\mathfrak{F} is given by the orbit of QQ under the action of some subgroup of 𝔗k\mathfrak{T}^{k}.

  2. 2.

    The distance in any smooth metric from some subsequence of QiQ^{i} to the image of 𝔉\mathfrak{F} converges to 0.

  3. 3.

    There is no other inclusion satisfying the above conditions which has smaller dimension than 𝔉\mathfrak{F}.

It is clear that such an 𝔉\mathfrak{F} must exist. It can be seen from Lemma 4.3, item 2 that 𝔉\mathfrak{F} is smooth.

Now let us construct ^𝔉\hat{\mathfrak{C}}\longrightarrow\mathfrak{F}. We shall have charts on ^\hat{\mathfrak{C}} given by C^m:=Cm×𝔉\hat{C}_{m}:=C_{m}\times\mathfrak{F} and 𝔄^n\hat{\mathfrak{A}}_{n}. 𝔄^n\hat{\mathfrak{A}}_{n} has coordinates (z~n+,z~n,Q)(\tilde{z}_{n}^{+},\tilde{z}_{n}^{-},Q) where z~n±𝔄n±\tilde{z}^{\pm}_{n}\in\mathfrak{A}^{\pm}_{n} and Qn=z~n+z~nQ_{n}=\tilde{z}_{n}^{+}\tilde{z}_{n}^{-}. Transition maps between C^m\hat{C}_{m} and 𝔄^n\hat{\mathfrak{A}}_{n} are simply given by the transition maps between CmC_{m} and 𝔄n±\mathfrak{A}^{\pm}_{n} times the identity on the 𝔉\mathfrak{F} component. As an explicit example, if ϕ\phi is a transition map between CmC_{m} and 𝔄n+\mathfrak{A}^{+}_{n}, the corresponding transition map between C^m\hat{C}_{m} and 𝔄^n\hat{\mathfrak{A}}_{n} is given by

(z,Q)(ϕ(z),Qnϕ(z),Q)(z,Q)\mapsto\left(\phi(z),\frac{Q_{n}}{\phi(z)},Q\right)

This describes the exploded fibration ^\hat{\mathfrak{C}}. The map down to 𝔉\mathfrak{F} is simply given by the obvious projection to the second component of each of the above charts. This gives a smooth family of exploded curves. Note that condition 6 from Proposition 3.17 ensures that this is a family of stable curves.

Lemma 4.4.

If Q𝔉Q\longrightarrow\mathfrak{F}, and QniQn\lceil Q^{i}_{n}\rceil\rightarrow\lceil Q_{n}\rceil, then fn,Q±f^{\pm}_{n,Q} can be glued by the identification z~n+z~n=Qn\tilde{z}_{n}^{+}\tilde{z}_{n}^{-}=Q_{n}. This is automatic if the limit of the conformal modulus of 𝔄ni\mathfrak{A}^{i}_{n} is finite.

Proof:

In the case that the limit of the conformal modulus of 𝔄ni\mathfrak{A}^{i}_{n} is finite, limQni=Qn\lim Q^{i}_{n}=Q_{n} for any Q𝔉Q\longrightarrow\mathfrak{F}. If this was not true, we could simply restrict 𝔉\mathfrak{F} to the points that satisfy this. This would include any point in 𝔉\mathfrak{F} that is the topological limit of QiQ^{i}, it would also satisfy the other conditions required of 𝔉\mathfrak{F}, but have smaller dimension contradicting the minimality of 𝔉\mathfrak{F}. When we have this, as we have fif^{i} converging on Cn±C_{n^{\pm}} and 𝔄ni\mathfrak{A}^{i}_{n}, and transition maps between these also converging, we can glue the limit.

In the case that the limit of the conformal modulus of 𝔄ni\mathfrak{A}^{i}_{n} is infinite, we want to glue fn,Q+(z~n+)f^{+}_{n,Q}(\tilde{z}_{n}^{+}) to fn,Q(z~n)f^{-}_{n,Q}(\tilde{z}_{n}^{-}) over the region where the smooth coordinates z~n+=z~n=0\lceil\tilde{z}^{+}_{n}\rceil=\lceil\tilde{z}^{-}_{n}\rceil=0 via the identification z~n+z~n=Qn\tilde{z}_{n}^{+}\tilde{z}_{n}^{-}=Q_{n}. Define the following continuous function

ϕn:𝔉𝔗l (the group acting on 𝔘n)\phi_{n}:\mathfrak{F}\longrightarrow\mathfrak{T}^{l}\text{ (the group acting on $\mathfrak{U}_{n}$)}

which detects the failure for fn±f^{\pm}_{n} to glue for any Q𝔉Q\longrightarrow\mathfrak{F}: If Qn=0\lceil Q_{n}\rceil=0, and z~n+z~n=Qn\tilde{z}_{n}^{+}\tilde{z}_{n}^{-}=Q_{n} so that z~+=z~=0\lceil\tilde{z}^{+}\rceil=\lceil\tilde{z}^{-}\rceil=0, let ϕ(Q)\phi(Q) be the element of 𝔗l\mathfrak{T}^{l} so that ϕn(Q)fn,Q+(z~n+)=fn,Q(z~n)\phi_{n}(Q)*f_{n,Q}^{+}(\tilde{z}_{n}^{+})=f_{n,Q}^{-}(\tilde{z}_{n}^{-}). This may only fail to be defined because the left or right hand sides might not be inside 𝔘\mathfrak{U}. Note first that if defined, this doesn’t depend on our choice of z~n±\tilde{z}_{n}^{\pm}. Also note that this will be defined on some open subset of 𝔉\mathfrak{F}, and there exists some homomorphism from 𝔗k\mathfrak{T}^{k} to 𝔗l\mathfrak{T}^{l} so that this map is equivariant with respect to the 𝔗k\mathfrak{T}^{k} action on 𝔉\mathfrak{F} and the 𝔗l\mathfrak{T}^{l} action on 𝔘m\mathfrak{U}_{m}. We can uniquely extend ϕn\phi_{n} to the rest of 𝔉\mathfrak{F} in an equivariant way.

We want to show that ϕn\phi_{n} is identically 11 on 𝔉\mathfrak{F}. For each QiQ^{i}, there exists a close point Qˇi\check{Q}^{i} inside 𝔉\mathfrak{F} so that the distance in any smooth metric between Qˇi\check{Q}^{i} and QiQ^{i} converges to 0 as ii\rightarrow\infty. Then ϕn(Qˇi)\phi_{n}(\check{Q}^{i}) converges to 11, because these points QiQ^{i} come from holomorphic curves that are converging on either end of 𝔄7R10,ni\mathfrak{A}^{i}_{\frac{7R}{10},n} to fn±f^{\pm}_{n} in a way given by Lemma 3.13. This tells us that the sub exploded fibration of 𝔉\mathfrak{F} given by ϕn1(1)\phi_{n}^{-1}(1) satisfies the conditions 1 and 2 above, so by the minimality of the dimension of 𝔉\mathfrak{F}, it must be all of 𝔉\mathfrak{F}. This tells us that if Qn1Q_{n}^{-1} is infinite, then we can glue together fn±f^{\pm}_{n} without any modifications.

\square

Note that choosing any point Q𝔉Q\longrightarrow\mathfrak{F} so that QQ is the topological limit of QiQ^{i} (such a point must exist by the definition of 𝔉\mathfrak{F}), the above lemma allows us to glue together fm,Qmf_{m,Q_{m}} and fn,Q±f^{\pm}_{n,Q} to obtain our limiting holomorphic curve ff for Theorem 4.1. If the above lemma held for every point in 𝔉\mathfrak{F}, we would have constructed our family. As it is, we need to make some gluing choices.

If QnQ_{n}\in\mathbb{C}^{*} and the limit of the conformal modulus of 𝔄ni\mathfrak{A}_{n}^{i} is infinite, we will need to make ‘gluing’ choices. We do not use the standard gluing and cutting maps as this will not give us strong enough regularity for the resulting family.

  1. 1.

    Chose some sequence Qˇi𝔉\check{Q}^{i}\longrightarrow\mathfrak{F} so that the distance between Qˇi\check{Q}^{i} and QiQ^{i} converges to 0.

    We need to take care of the different complex structures obtained by gluing 𝔄n±\mathfrak{A}^{\pm}_{n} by Qˇni\check{Q}^{i}_{n} and QniQ^{i}_{n}. Choose an almost complex structure jij_{i} on 𝔄n+\mathfrak{A}^{+}_{n} as follows: Choose smooth isomorphisms

    Φni:𝔄n+𝔄n+\Phi^{i}_{n}:\mathfrak{A}^{+}_{n}\longrightarrow\mathfrak{A}^{+}_{n}

    so that

    1. (a)
      Φni(z~n+)=z~n+ for |z~n+|e8R10\Phi^{i}_{n}\left(\tilde{z}_{n}^{+}\right)=\tilde{z}_{n}^{+}\text{ for }\left\lvert\tilde{z}_{n}^{+}\right\rvert\geq e^{-\frac{8R}{10}}
    2. (b)
      Φni(z~n+)=QˇniQniz~n+ for |z~n+|eR\Phi^{i}_{n}\left(\tilde{z}_{n}^{+}\right)=\frac{\check{Q}^{i}_{n}}{Q^{i}_{n}}\tilde{z}_{n}^{+}\text{ for }\left\lvert\tilde{z}_{n}^{+}\right\rvert\leq e^{-R}
    3. (c)

      On the region z~n+0\lceil\tilde{z}_{n}^{+}\rceil\neq 0, the sequence of maps {Φni}\{\Phi^{i}_{n}\} converges in CC^{\infty} to the identity.

    Now define jij_{i} on 𝔄n+\mathfrak{A}^{+}_{n} to be the pullback under Φni\Phi^{i}_{n} of the standard complex structure. Using the standard complex structure on 𝔄n\mathfrak{A}^{-}_{n}, we then get our jij_{i} defined on 𝔄^n\hat{\mathfrak{A}}_{n}. This is compatible with the jij_{i} already defined on C^m\hat{C}_{m}, so we get jij_{i} defined on ^\hat{\mathfrak{C}}. Note that jij_{i} restricted to the fiber over Qˇi\check{Q}^{i} is the complex structure on i\mathfrak{C}^{i}. From now on, we shall use these new coordinates on 𝔄ni\mathfrak{A}^{i}_{n}.

  2. 2.

    We now define the linear gluing map as follows:

    1. (a)

      Chose some smooth cutoff function

      ρ:[0,1]\rho:\mathbb{R}^{*}\longrightarrow[0,1]

      so that

      ρ(x)=0 for all xe8R10\rho(x)=0\text{ for all }x\geq e^{-\frac{8R}{10}}
      ρ(x)=1 for all xe9R10\rho(x)=1\text{ for all }x\leq e^{-\frac{9R}{10}}

      Extend this to

      ρ:𝔱[0,1]\rho:\mathbb{R}^{*}\mathfrak{t}^{\mathbb{R}}\longrightarrow[0,1]

      satisfying all the above conditions. (We defined this first on \mathbb{R}^{*} so that it was clear what ‘smooth’ meant.)

    2. (b)

      Given maps ϕ+,ϕ:𝔗11k\phi^{+},\phi^{-}:\mathfrak{T}^{1}_{1}\longrightarrow\mathbb{C}^{k} which vanish at z=0z=0 define the gluing map

      G(ϕ+,ϕ)(z~+,z~):=ρ(|z~|)ϕ+(z~+)+ρ(|z~+|)ϕ(z~)G_{(\phi^{+},\phi^{-})}(\tilde{z}^{+},\tilde{z}^{-}):=\rho\left(\left\lvert\tilde{z}^{-}\right\rvert\right)\phi^{+}(\tilde{z}^{+})+\rho\left(\left\lvert\tilde{z}^{+}\right\rvert\right)\phi^{-}(\tilde{z}^{-})

      Note that if ϕ+\phi^{+} and ϕ\phi^{-} are smooth, G(ϕ+,ϕ):𝔗22kG_{(\phi^{+},\phi^{-})}:\mathfrak{T}^{2}_{2}\longrightarrow\mathbb{C}^{k} is smooth. Note also that if ϕ+\phi^{+} and ϕ\phi^{-} are small in C,δC^{\infty,\delta}, then G(ϕ+,ϕ)G_{(\phi^{+},\phi^{-})} is small too.

  3. 3.

    We now define a linear ‘cutting’ map: as follows:

    1. (a)

      Choose a smooth cutoff function β:[0,1]\beta:\mathbb{R}^{*}\longrightarrow[0,1] satisfying the following:

      β(x)+β(x1)=1\beta(x)+\beta\left(x^{-1}\right)=1
      β(x)=1 for all x>eR10\beta(x)=1\text{ for all }x>e^{\frac{R}{10}}
    2. (b)

      Given a map ϕ:𝔄nik\phi:\mathfrak{A}^{i}_{n}\longrightarrow\mathbb{C}^{k} and Qˇni\check{Q}^{i}_{n}\in\mathbb{C}^{*}, where

      𝔄ni:={z+z=Qˇni}\mathfrak{A}^{i}_{n}:=\{z^{+}z^{-}=\check{Q}^{i}_{n}\}

      we can define the cutting of ϕ\phi as follows:

      ϕ+(z+):=β(|z+|2|Qˇni|)ϕ(z+)\phi^{+}(z^{+}):=\beta\left(\frac{\left\lvert z^{+}\right\rvert^{2}}{\left\lvert\check{Q}^{i}_{n}\right\rvert}\right)\phi(z^{+})
      ϕ(z):=β(|z|2|Qˇni|)ϕ(z):=β(|z|2|Qˇni|)ϕ(z+=Qˇniz)\phi^{-}(z^{-}):=\beta\left(\frac{\left\lvert z^{-}\right\rvert^{2}}{\left\lvert\check{Q}^{i}_{n}\right\rvert}\right)\phi(z^{-}):=\beta\left(\frac{\left\lvert z^{-}\right\rvert^{2}}{\left\lvert\check{Q}^{i}_{n}\right\rvert}\right)\phi\left(z^{+}=\frac{\check{Q}^{i}_{n}}{z^{-}}\right)

      Note that G(ϕ+,ϕ)G_{(\phi^{+},\phi^{-})} restricted to z~+z~=Qˇn\tilde{z}^{+}\tilde{z}^{-}=\check{Q}_{n} is equal to ϕ\phi.

  4. 4.

    Lemma 3.5 tells us that the image of our annuli of finite conformal moduli, fi(𝔄7R10,ni)f^{i}(\mathfrak{A}^{i}_{\frac{7R}{10},n}) is contained in some coordinate chart appropriate for Lemma 3.13. (We can choose this coordinate chart to be contained inside 𝔘n\mathfrak{U}_{n} so that the 𝔗k\mathfrak{T}^{k} action on 𝔘n\mathfrak{U}_{n} just consists of multiplying coordinate functions by a constant). In fact, we can choose a subsequence so that either 𝔄ni\mathfrak{A}^{i}_{n} has infinite conformal modulus for all ii or fi(𝔄7R10,ni)f^{i}(\mathfrak{A}^{i}_{\frac{7R}{10},n}) is contained in one of these coordinate charts for all ii. In that case, Lemma 3.13 together with the conditions on our coordinate change for 𝔄ni\mathfrak{A}^{i}_{n} above tells us that

    fni(z)=(eϕn,1i(z)cn,1izα1,,eϕn,ki(z)cn,kizαk,,cn,dk+ϕn,dki):=eϕnicnizα\begin{split}f^{i}_{n}(z)&=(e^{\phi^{i}_{n,1}(z)}c_{n,1}^{i}z^{\alpha_{1}},\dotsc,e^{\phi^{i}_{n,k}(z)}c_{n,k}^{i}z^{\alpha_{k}},\dotsc,c_{n,d-k}+\phi_{n,d-k}^{i})\\ &:=e^{\phi^{i}_{n}}c^{i}_{n}z^{\alpha}\end{split}

    where ϕni\phi^{i}_{n} is exponentially small on the interior of 𝔄R,ni\mathfrak{A}^{i}_{R,n} as required by Lemma 3.13. We can regard our local coordinate chart as giving us a local trivialization of T𝔅T\mathfrak{B} compatible with our 𝔗k\mathfrak{T}^{k} action. We can extend this trivialization using our 𝔗k\mathfrak{T}^{k} action to an open subset 𝔒n𝔘n\mathfrak{O}_{n}\subset\mathfrak{U}_{n} which will contain the image of any relevant translation of fnif^{i}_{n}. Note that 𝔒n\mathfrak{O}_{n} is an open subset of some refinement of 𝔗k×d2k\mathfrak{T}^{k}\times\mathbb{R}^{d-2k} (with the above coordinates). We shall define our maps as maps to 𝔗k×d2k\mathfrak{T}^{k}\times\mathbb{R}^{d-2k}. Note that in particular, if Q𝔉Q^{\infty}\longrightarrow\mathfrak{F} is some topological limit of QiQ^{i}, then fn,Q±f^{\pm}_{n,Q^{\infty}} in these coordinates are given by

    fn,Q+(z~+):=eϕn+(z~+)cn(z~+)αf^{+}_{n,Q^{\infty}}(\tilde{z}^{+}):=e^{\phi^{+}_{n}(\tilde{z}^{+})}c_{n}(\tilde{z}^{+})^{\alpha}
    :=(eϕn,1+(z~+)cn,1(z~+)α1,,eϕn,k+(z~+)cn,k(z~+)αk,,cn,dk+ϕn,dk+(z~+)):=(e^{\phi^{+}_{n,1}(\tilde{z}^{+})}c_{n,1}(\tilde{z}^{+})^{\alpha_{1}},\dotsc,e^{\phi^{+}_{n,k}(\tilde{z}^{+})}c_{n,k}(\tilde{z}^{+})^{\alpha_{k}},\dotsc,c_{n,d-k}+\phi^{+}_{n,d-k}(\tilde{z}^{+}))
    fn,Q(z~):=eϕn(z~)cn(Qnz~)αf^{-}_{n,Q^{\infty}}(\tilde{z}^{-}):=e^{\phi^{-}_{n}(\tilde{z}^{-})}c_{n}\left(\frac{Q^{\infty}_{n}}{\tilde{z}^{-}}\right)^{\alpha}

    where ϕ±\phi^{\pm} are smooth and vanish at z±=0z^{\pm}=0.

    Similarly, if 𝔄ni\mathfrak{A}^{i}_{n} is infinite, then we can consider fnif^{i}_{n} to map to 𝔒n\mathfrak{O}_{n} so that

    fni(z~+):=eϕni+cni(z~+)αf^{i}_{n}(\tilde{z}^{+}):=e^{\phi^{i+}_{n}}c^{i}_{n}(\tilde{z}^{+})^{\alpha}
    fni(z~):=eϕni(z~)cni(Qˇniz~)αf^{i}_{n}(\tilde{z}^{-}):=e^{\phi^{i-}_{n}(\tilde{z}^{-})}c^{i}_{n}\left(\frac{\check{Q}^{i}_{n}}{\tilde{z}^{-}}\right)^{\alpha}
  5. 5.

    Use the notation p1p2\frac{p_{1}}{p_{2}} for two points pi𝔙np_{i}\longrightarrow\mathfrak{V}_{n} to mean the element of 𝔗k\mathfrak{T}^{k} so that p1=p1p2p2p_{1}=\frac{p_{1}}{p_{2}}*p_{2}. Define Fni:𝔄^n𝔅^F^{i}_{n}:\hat{\mathfrak{A}}_{n}\longrightarrow\hat{\mathfrak{B}} by

    Fni(z~+,z~,Q):=Qn+Qˇn+icni(z~+)αF^{i}_{n}(\tilde{z}^{+},\tilde{z}^{-},Q):=\frac{Q_{n^{+}}}{\check{Q}^{i}_{n^{+}}}*c^{i}_{n}(\tilde{z}^{+})^{\alpha}

    and

    Fn(z~+,z~,Q):=Qn+Qn+cn(z~+)αF_{n}(\tilde{z}^{+},\tilde{z}^{-},Q):=\frac{Q_{n^{+}}}{Q^{\infty}_{n^{+}}}*c_{n}(\tilde{z}^{+})^{\alpha}

    (Recall that Qn+Q_{n^{+}} recorded the position of Cn+C_{n^{+}} which is attached to 𝔄n+\mathfrak{A}^{+}_{n}). Note that even though we defined this as a map to 𝔗k×d2k\mathfrak{T}^{k}\times\mathbb{R}^{d-2k}, and our chart is some refinement of this, these FniF^{i}_{n} are still smooth maps to 𝔅^\hat{\mathfrak{B}}. (This can be seen if we restrict this map to local coordinate charts.)

    Then we can define f^i:𝔄^n𝔅^\hat{f}^{i}:\hat{\mathfrak{A}}_{n}\longrightarrow\hat{\mathfrak{B}} using the cutting and gluing maps above by

    f^i(z~+,z~,Q):=eG(ϕni+,ϕni)(z~+,z~)Fni(z~+,z~,Q)\hat{f}^{i}(\tilde{z}^{+},\tilde{z}^{-},Q):=e^{G_{(\phi^{i+}_{n},\phi^{i-}_{n})}(\tilde{z}^{+},\tilde{z}^{-})}F^{i}_{n}(\tilde{z}^{+},\tilde{z}^{-},Q)

    We can similarly define f^:𝔄^n𝔅^\hat{f}:\hat{\mathfrak{A}}_{n}\longrightarrow\hat{\mathfrak{B}}.

    Also define

    f^:C^m𝔅^\hat{f}:\hat{C}_{m}\longrightarrow\hat{\mathfrak{B}}

    by

    f^(z,Q)=fm,Qm(z)\hat{f}(z,Q)=f_{m,Q_{m}}(z)

    and define f^i\hat{f}^{i} on C^m\hat{C}_{m} similarly by translating fif^{i} appropriately depending on the difference between Qˇmi\check{Q}^{i}_{m} and QmQ_{m}.

    This gives well defined smooth maps

    f^i:^𝔅^\hat{f}^{i}:\hat{\mathfrak{C}}\longrightarrow\hat{\mathfrak{B}}
    f^:^𝔅^\hat{f}:\hat{\mathfrak{C}}\longrightarrow\hat{\mathfrak{B}}

    so that fif^{i} is given by the restriction of f^i\hat{f}^{i} to the fiber of ^𝔉\hat{\mathfrak{C}}\longrightarrow\mathfrak{F} over Qˇi𝔉\check{Q}^{i}\longrightarrow\mathfrak{F}. Note that as the vectorfields ϕ±\phi^{\pm} are vertical with respect to 𝔅^𝔊\hat{\mathfrak{B}}\longrightarrow\mathfrak{G}, Gϕ+,ϕG_{\phi^{+},\phi^{-}} is too, and we have finally constructed our required families.

    (^,ji)f^i𝔅^(^,j)f^𝔅^𝔉𝔊𝔉𝔊\begin{array}[]{ccccccc}(\mathfrak{\hat{C}},j_{i})&\xrightarrow{\hat{f}^{i}}&\hat{\mathfrak{B}}&&(\mathfrak{\hat{C}},j)&\xrightarrow{\hat{f}}&\hat{\mathfrak{B}}\\ \downarrow&&\downarrow&&\downarrow&&\downarrow\\ \mathfrak{F}&\longrightarrow&\mathfrak{G}&&\mathfrak{F}&\longrightarrow&\mathfrak{G}\end{array}
Lemma 4.5.
f^i:(^,ji)𝔅^\hat{f}^{i}:(\hat{\mathfrak{C}},j_{i})\longrightarrow\hat{\mathfrak{B}}

converges in C,δC^{\infty,\delta} to

f^:(^,j)𝔅^\hat{f}:(\hat{\mathfrak{C}},j)\longrightarrow\hat{\mathfrak{B}}

Proof:

We work in coordinates. First note that f^i\hat{f}^{i} converges in CC^{\infty} restricted to f^\hat{f} on C^m\hat{C}_{m}. This also holds for 𝔄^n\hat{\mathfrak{A}}_{n} if the size of 𝔄ni\mathfrak{A}^{i}_{n} stays bounded. Note also that jij_{i} converges in CC^{\infty} to jj.

It remains only to show that f^i\hat{f}^{i} converges in C,δC^{\infty,\delta} to f^\hat{f} on 𝔄^n\hat{\mathfrak{A}}_{n}. Recall that 𝔄^n\hat{\mathfrak{A}}_{n} has coordinates (z~+,z~,Q)(\tilde{z}^{+},\tilde{z}^{-},Q) where Q𝔉Q\longrightarrow\mathfrak{F} and Qn=z~+z~Q_{n}=\tilde{z}^{+}\tilde{z}^{-}. Note that the maps Fni:𝔄^n𝔅^F^{i}_{n}:\hat{\mathfrak{A}}_{n}\longrightarrow\hat{\mathfrak{B}} defined above converge in CC^{\infty} to Fn:𝔄^n𝔅^F_{n}:\hat{\mathfrak{A}}_{n}\longrightarrow\hat{\mathfrak{B}}. We have that

f^i(z~+,z~,Q):=eG(ϕni+,ϕni)(z~+,z~)Fni(z~+,z~,Q)\hat{f}^{i}(\tilde{z}^{+},\tilde{z}^{-},Q):=e^{G_{(\phi^{i+}_{n},\phi^{i-}_{n})}(\tilde{z}^{+},\tilde{z}^{-})}F^{i}_{n}(\tilde{z}^{+},\tilde{z}^{-},Q)

Similarly, we write

f^(z~+,z~,Q):=eG(ϕn+,ϕn)(z~+,z~)Fn(z~+,z~,Q)\hat{f}(\tilde{z}^{+},\tilde{z}^{-},Q):=e^{G_{(\phi^{+}_{n},\phi^{-}_{n})}(\tilde{z}^{+},\tilde{z}^{-})}F_{n}(\tilde{z}^{+},\tilde{z}^{-},Q)

We can choose a subsequence so that ϕni±(z±)\phi^{i\pm}_{n}(z^{\pm}) converges in CC^{\infty} on compact subsets of {0<|z±|e7R10}\{0<\left\lvert z^{\pm}\right\rvert\leq e^{-\frac{7R}{10}}\} to ϕn±(z±)\phi^{\pm}_{n}(z^{\pm}). Lemma 3.13 and our cutting construction above tells us that for any δ<1\delta<1, there exists some constant cc independent of nn so that

|ϕni±(z~±)|c|z~±|δ\left\lvert\phi_{n}^{i\pm}(\tilde{z}^{\pm})\right\rvert\leq c\left\lvert\tilde{z}^{\pm}\right\rvert^{\delta}

and the same inequality holds for ϕn±\phi^{\pm}_{n} and all derivatives (with a different constant cc). This implies that ϕni±\phi^{i\pm}_{n} converges to ϕn±\phi^{\pm}_{n} in C,δC^{\infty,\delta} for any δ<1\delta<1. A quick calculation then shows that G(ϕni+,ϕni)(z~+,z~)G_{(\phi^{i+}_{n},\phi^{i-}_{n})}(\tilde{z}^{+},\tilde{z}^{-}) converges in C,δC^{\infty,\delta} to G(ϕn+,ϕn)(z~+,z~)G_{(\phi^{+}_{n},\phi^{-}_{n})}(\tilde{z}^{+},\tilde{z}^{-}). Adding the extra coordinates on which G(ϕni+,ϕni)G_{(\phi^{i+}_{n},\phi^{i-}_{n})} does not depend does not affect C,δC^{\infty,\delta} convergence, so we get that f^i\hat{f}^{i} converges to f^\hat{f} in C,δC^{\infty,\delta} as required.

\square

This completes the proof of Theorem 4.1.

References

  • [1] F. Bourgeois, Y. Eliashberg, H. Hofer, K. Wysocki, and E. Zehnder. Compactness results in symplectic field theory. Geom. Topol., 7:799–888 (electronic), 2003.
  • [2] Y. Eliashberg, A. Givental, and H. Hofer. Introduction to symplectic field theory. Geom. Funct. Anal., (Special Volume, Part II):560–673, 2000. GAFA 2000 (Tel Aviv, 1999).
  • [3] Barbara Fantechi. Stacks for everybody. In European Congress of Mathematics, Vol. I (Barcelona, 2000), volume 201 of Progr. Math., pages 349–359. Birkhäuser, Basel, 2001.
  • [4] Kenji Fukaya. Asymptodic Analyisis, Multivalued Morse theory and Mirror symmetry. Proceedings of Symposia in Pure Mathematics, 73:205–280.
  • [5] M. Gromov. Pseudoholomorphic curves in symplectic manifolds. Invent. Math., 82(2):307–347, 1985.
  • [6] Mark Gross and Bernd Siebert. Mirror symmetry via logarithmic degeneration data. I. J. Differential Geom., 72(2):169–338, 2006.
  • [7] Eleny-Nicoleta Ionel and Thomas H. Parker. The symplectic sum formula for Gromov-Witten invariants. Ann. of Math. (2), 159(3):935–1025, 2004.
  • [8] Maxim Kontsevich and Yan Soibelman. Affine structures and non-Archimedean analytic spaces. In The unity of mathematics, volume 244 of Progr. Math., pages 321–385. Birkhäuser Boston, Boston, MA, 2006.
  • [9] An-Min Li and Yongbin Ruan. Symplectic surgery and Gromov-Witten invariants of Calabi-Yau 3-folds. Invent. Math., 145(1):151–218, 2001.
  • [10] Jun Li. A degeneration formula of GW-invariants. J. Differential Geom., 60(2):199–293, 2002.
  • [11] Grigory Mikhalkin. Enumerative tropical algebraic geometry in 2\mathbb{R}^{2}. J. Amer. Math. Soc., 18(2):313–377 (electronic), 2005.
  • [12] Brett Parker. Exploded fibrations. In Proceedings of Gökova Geometry-Topology Conference 2006, 2007.