A Unified Approach to Holomorphic Anomaly Equations and Quantum Spectral Curves

© All Rights Reserved

Просмотров: 1

A Unified Approach to Holomorphic Anomaly Equations and Quantum Spectral Curves

© All Rights Reserved

- Buses For Anonymous Message Delivery
- fem-ex21
- Peter Hamburger and Attila Sali- 11 -Doilies with Vertex Sets Of Sizes 275,286,...,462
- 0803.3435v1
- VCE Further Maths 1 & 2 Notes
- Metrics of Software Architecture Changes Based on Structural Distance
- Lecture 37
- Gerard
- tmpE147
- D1 Course Companipon and Revision Guide
- Fibonacci Identities and Graph Coloring
- documentclass
- hw4
- EM Algo
- Super Fibonacci Graceful Labeling
- Detecting Connection Traps in Complex Relational Databases
- Business Mathematics
- diff 510 lesson 1 draw and label a picture graph to represent data
- SAE Toward Efficient Cloud Data Analysis
- Science and Religion1

Вы находитесь на странице: 1из 47

arXiv:1808.05343v1 [math-ph] 16 Aug 2018

and some well-known quantum spectral curves. We develop a formalism of

abstract quantum ﬁeld theory based on the diagrammatics of the Deligne-

Mumford muduli spaces Mg,n and derive a quadratic recursion relation for

the abstract free energies in terms of the edge-cutting operators. This ab-

stract quantum ﬁeld theory can be realized by various choices of a sequence

of holomorphic functions and suitable propagators, and the realized quantum

ﬁeld theory can be represented by formal Gaussian integrals. Various applica-

tions are given.

Contents

1. Introduction 2

2. Diagrammatics of Moduli Space of Curves 4

2.1. Moduli space of stable curves and Feynman diagrams 4

2.2. Operators on vector space associated to the set of stable graphs 6

2.3. Abstract free energy 7

2.4. A recursion relation for the abstract free energy 8

3. A Generalization to Labelled Graphs 15

3.1. Diagrammatics for labelled graphs 15

3.2. A recursion relation for the abstract free energy 16

4. Realization of the Abstract Quantum Field Theory 17

4.1. Realization of the abstract QFT by Feynman rules 17

4.2. Representation by formal Gaussian integrals 18

4.3. Realization of the operators on stable graphs 20

4.4. Realization of the recursion relations 21

4.5. The general case 22

5. The holomorphic anomaly equation 23

6. Application to Topological 1D Gravity and Quantum Spectral Curves 25

6.1. Some results in topological 1D gravity 26

6.2. Topological 1D gravity as realization of the diagrammatics of stable

graphs 27

6.3. Example: The Airy function and quantum Airy curve 30

6.4. Example: The Kontsevich-Penner matrix model for N = 1 32

6.5. Example: Enumeration of stable graphs with genus zero vertices 34

6.6. Example: Enumeration of graphs 37

A. Some explicit expressions for[ Fg,n 43

References 46

Bibliography 46

1

2 ZHIYUAN WANG AND JIAN ZHOU

1. Introduction

There are various types of quadratic recursions which have played an important

role in Gromov-Witten theory. The Virasoro constraints are a sequence of recursion

relations [7, 10, 14] for the free energy deﬁned on the big phase space. It involves

diﬀerential operators of inﬁnitely many variables. There are also quadratic recursion

relations for the free energy restricted to the small phase space that involves only

ﬁnitely many variables. For example, the BCOV holomorphic anomaly equation

[3] and the Eynard-Orantin topological recursion relations [12]. The former was

developed by Bershadsky, Cecotti, Ooguri and Vafa to compute Gromov-Witten

invariants of the quintic Calabi-Yau threefold [2, 3]. The latter was ﬁrst discovered

by Eynard and Orantin in the setting of matrix models and later formulated in

a form which is conjectured and proved to hold in general for n-point functions.

For an approach to the holomorphic anaomaly equation from the point of view of

Eynard-Orantin topological relation, see Eynard-Marinõ-Orantin [13]. In the case

of Witten-Kontsevich tau-function, it has been shown by the second author that

the Virasoro constraints are equivalent to the EO topological recursion relations

[28]. It is an interesting problem to establish a relationship between the Virasoro

constraints and the EO topological relations in general. The basic ingredient for EO

topological recursion is an algebraic curve with some extra data, called the spectral

curve. A suitable quantization of the spectral curve leads to a Schrödinger equation

satisﬁed by a partition function constructed from the EO topological recursion,

called the quantum spectral curve (cf. Gukov-Sulkowski [16]). It is a natural

problem to understand the relationship between the holomorphic anomaly equation

and the quantum spectral curves. In this work we will present a formalism that

gives a uniﬁed construction of the holomorphic anomaly equation and the quantum

spectral curves in some well-known examples.

To formulate the BCOV holomorphic anomaly equation, it is crucial to ﬁrst ex-

tend the free energy Fg (t) to a non-holomorphic free energy Fbg (t, t̄) whose geometric

meaning is not mathematically clear at present. It is supposed to be modular-

invariant, and involve contributions from the boundary strata of the moduli spaces.

Originally the holomorphic anomaly is an equation for ∂t̄i Fbg which involves Fbh for

h < g of the following form:

g−1

X

1 (0)jk

∂¯t̄i Fbg = C̄ī Dj Dk Fbg−1 + Dj Fbr Dk Fbg−r .

2 r=1

It is shown in [3] that this system of equations can be recursively solved to get

(r<g)

(1) Fbg (t, t̄) = Γ(g) (∆ij , ∆i , ∆, Ci1 ,...,in ) + f (g) (t),

(r<g)

where Γ(g) is a polynomial equation in ∆ij , ∆i , ∆ (propagators) and Ci1 ,...,in (lower

genus vertices). Furthermore, the authors of that work present some explicit expres-

sions of Fbg for g = 2 and 3 and the corresponding Feynman graphs and Feynman

rules (cf. [3, (6.7), Figure 17] and [3, (6.8), Figure 18]). Inspired by Witten [23],

Aganagic, Bouchard and Klemm [1, (2.16)] obtain the following expression for Fbg :

−1 IJ

(2) b

Fg (t, t̄) = Fg (t) + Γg − (τ − τ̄ ) , ∂I1 · · · ∂In Fr<g (t) ,

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 3

2

∂ F0

where τ = (τij ) = ∂t i ∂tj

). This was interpreted for matrix models using Eynard-

Orantin topological recursion by Eynard, Mariño and Orantin [13], and further-

more, they also reformulate the partition function as a formal Gaussian integral [13,

(4.27)] and present the Feynman graphs and Feynman rules for the terms that con-

tribute to Γg . These results have been generalized to other models by Grimm,

Klemm, Mariño and Weiss [15]. These authors reformulate the holomorphic anom-

aly equation as a quadratic recursion relation for the derivative of Fbg with respective

to the propagators ∆IJ (cf. [15, (7.50)]:

g−1

∂ Fbg 1 1X b

(3) IJ

= DI ∂J Fbg−1 + ∂I Fr ∂J Fbg−r .

∂∆ 2 2 r=1

1

(4) ∆IJ = − √ ((τ − τ̄ )−1 )IJ + E IJ ,

2 −1

where E IJ is a holomorphic function. They also derive the formal Gaussian integral

representation of the partition function and the Feynman expansions for Fbg .

To summarize, one can see the common features of the above works on holo-

morphic anomaly equations. One starts with a sequence of functions Fg (t) (g ≥ 0)

for ﬁnitely many variables t1 , . . . , tn , and seek for another sequence Fbg (t, t̄) such

that limt̄→∞ Fbg (t, t̄) = Fg (t), and such that the sequence Fbg (t, t̄) satisﬁes qua-

dratic recursion equations of the form (3). Furthermore, the partition function

b t̄) = exp P

Z(t, 2g−2 b

g≥0 λ Fg (t, t̄) can be represented as a formal Gaussian integral,

and there are Feynman rules that express Fbg as a polynomial in the propagators

and the derives of Fr for r ≤ g. As pointed out in [3], the holomorphic anomaly has

contributions from the degenerate Riemann surfaces, from lower boundary strata

of the Deligne-Mumford moduli spaces. It turns out that the Feynman graphs that

appear in the Feynman expansion of Fbg in the above works are just the stable

graphs that index the stratiﬁcation of the Deligne-Mumford moduli spaces Mg .

In the present work, we will formulate a general construction of some kind of

abstract quantum field theory that have the above works on holomorphic anomaly

equations as special realizations. The key ingredient in our construction is the

diagrammatics for the stratiﬁcation of the Deligne-Mumford moduli spaces Mg,n ,

using the language of the dual graphs. There are some natural operators acting

on these graphs, including an edge-cutting operator K. We deﬁne an abstract free

energy

X 1

Fbg = Γ,

c

| Aut(Γ)|

Γ∈Gg,0

g−1

X

1

K Fbg = (D∂ Fbg−1 + ∂ Fbr ∂ Fbg−r )

2 r=1

for g ≥ 2 using the operators ∂ and D which correspond to adding external edges

on the graphs. For details, see Section 2. For generalizations that involves more

labelling on the graphs, see Section 3.

4 ZHIYUAN WANG AND JIAN ZHOU

Another ingredient in our formalism is what we call the realizations of the ab-

stract quantum field theory, we need a sequence of functions Fg (t), t = (t1 , . . . , tn ),

g ≥ 0, and a symmetric nondegenerate matrix κ that depends on t and t̄. They will

be used to give the Feynman rules that associate a weight ωΓ to each Γ. This kind

of realization of the above abstract quantum ﬁeld theory can also be represented

as a formal Gaussian integral. We will refer to this as the formal Gaussian inte-

gral representation of the abstract quantum field theory. These will be presented in

Section 4. P 2g−2

Using this formalism, given a holomorphic function F (t) = λ Fg (t), we

will obtain a family of free energies Fbg (κ, t). Here Fbg (κ, t) is a polynomial in κ of

degree 3g − 3, whose coeﬃcients are some diﬀerential polynomials in {Fk }0≤k≤g for

g ≥ 2. In particular, Fbg (0, t) = Fg (t) for g ≥ 2.

Let us now mention some applications of our formalism. First of all, for the choice

κ = (τ̄ − τ )−1 , we recover the holomorphic anomaly for matrix models [13] [11], or

more generally, when the propagator κIJ takes the form of ∆IJ in (4), we recover

the construction of holomorphic anomaly equation in Gromov-Witten theory as

discussed in [15]. For details, see Section 5. Secondly, the propagators κ can be also

chosen to be holomorphic. We will use this formalism for holomorphic propagators

to reinterpret some results of the second author [27] in the theory of topological

1D gravity. And as applications, we will reinterpret some well-known examples

of quantum spectral curves [16] in our formalism. We will deal with the case of

propagators of the form (4) in a subsequent work that generalizes the formalism of

this work [22].

The rest of this paper is arranged as follows. In Section 2 we formulate the

diagrammatics and derive a quadratic recursive relation for the free energies using

the edge-cutting operators. In Section 3 we generalize the diagrammatics to the

case of labelled graphs. We give a ﬁeld theory realization of the diagrammatics in

Section 4, where the free energies can be obtained from formal Gaussian integrals

and Feynman rules. The holomorphic anomaly equation is derived in Section 5

in our formalism using non-holomorphic propagators. In Section 6 we apply our

formalism to topological 1D gravity and derive quadratic recursion relations for

this theory. As applications, we relate the recursion relations to quantum spectral

curves in some important examples. Other applications include recursion relations

in enumeration problems of stable graphs and general graphs.

In this section we recall the diagrammatics related to the stratiﬁcation of the

Deligne-Mumford moduli spaces Mg,n . We deﬁne the edge-cutting operators and

some other operators acting on the graphs, and use them to derive a recursive

relation of the abstract free energy.

2.1. Moduli space of stable curves and Feynman diagrams. We ﬁrst recall

the stratiﬁcation of the Deligne-Mumford moduli space Mg,n of stable curves [6,17].

The Deligne-Mumford moduli space Mg,n is a smooth orbifold of complex di-

mension 3g − 3 + n. There are two types of basic maps over these spaces, namely

the forgetful map

πn+1 : Mg,n+1 → Mg,n ,

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 5

which simply forgets the (n + 1)-th marked point and contract the unstable com-

ponent; and the gluing maps

ξ1 : Mg1 ,n1 +1 × Mg2 ,n2 +1 → Mg1 +g2 ,n1 +n2 ,

ξ2 : Mg−1,n+2 → Mg,n ,

which glues two marked points together to produce a new nodal point. The union

of images of ξ1 and ξ2 is the boundary Mg,n − Mg,n ; moreover, it is of codimension

one in Mg,n . In this way, we get a stratiﬁcation of the compactiﬁed moduli space

Mg,n by decomposing it to the union of some products of the moduli spaces Mh,m

of smooth stable curves.

An eﬃcient way of describing this stratiﬁcation is to use the language of graphs.

Recall that for a stable curve (C; x1 , · · · , xn ) of genus g with n marked points, its

dual graph is deﬁned as follows. Let C̃ be its normalization, then we associate a

vertex to each of the connected components of C̃, and write down the genus of

this component at this vertex. For each nodal point of C, we draw an internal

edge connecting the corresponding vertices; and for each marked point, we draw

an external edge attaching to the corresponding vertex, and write the markings

i ∈ {1, 2, · · · , n} besides the external edges.

In what follows we will forget all the markings of the external edges, which

corresponds to the moduli space Mg,n /Sn parametrizing equivalent classes of stable

curves with n marked points while we do not distinguish these marked points. In

this way, we obtain a graph of genus g with n external edges, where the genus is

deﬁned to be the sum of the genus of the underlying one-dimensional manifold and

all the numbers associated to the vertices, which is exactly the genus of the stable

curve C.

Example 2.1. All possible dual graphs of curves in M1,1 , and M2,0 are listed in

Figure 1 and Figure 2 respectively.

1 0

2 1 1 1 0 1 0 0 0 0 0

A graph Γ is called stable if all its vertices are stable, that is, the valence of each

vertex of genus 0 is at least three, and the valence of each vertex of genus 1 is at

least one. It is clear that the stability of curves is equivalent to the stability of its

dual graph. The stable graphs will be the Feynman graphs used in this paper.

Let Gg,n be the set of all stable graphs of genus g with n external edges (not

c

necessarily connected), and Gg,n be the subset of connected stable graphs.

Now let Γ be a stable graph of genus g with n external edges, and deﬁne MΓ to

be the subset of Mg,n /Sn consisting of all equivalent classes of curves whose dual

6 ZHIYUAN WANG AND JIAN ZHOU

graph is Γ, then

G

(5) Mg,n /Sn = MΓ

c

Γ∈Gg,n

2.2. Operators on vector space associated to the set of stable graphs.

Denote by V c the vector space over Q generated by all connected stable graphs, i.e.,

M

(6) Vc = QΓ;

c

Γ∈Gg,n

2g−2+n>0

similarly, denote by V the vector space over Q generated by all stable graphs, not

necessarily connected, i.e.,

M

(7) V= QΓ.

Γ∈Gg,n

2g−2+n>0

understood as the inverse procedures of the gluing maps and forgetful maps.

Deﬁne an operator K acting on V:

X

(8) K : V → V, Γ ∈ Gg,n 7→ K(Γ) = Γ′ ,

Γ′

′

where the summation is over all graphs Γ obtained by cutting an internal edge of

Γ. We will refer to this operator as the edge cutting operator. It is easy to see that

the operator K does not aﬀect the stability of the graph. It decreases the number

of internal edges by one, and increases the number of external edges by two. In

the picture of stable curves, this operator corresponds to breaking up a node and

regard it as two new marked points. So the operator K can be regarded as the

inverse operator of the gluing maps ξ1 and ξ2 .

Example 2.2.

K 2 =0,

K 0 0 =2 0 0 + 0 0 ,

K 0 0 =3 0 0 .

Let us deﬁne another operator ∂. This operator has two parts, one is to attach

an external edge to a vertex and sum over all vertices, and the other is to break up

an internal edge and insert a stable vertex of genus 0 with three edges.

Example 2.3.

∂ 1 = 1 ,

∂ 0 0 =2 0 0 +2 0 0 0 + 0 0 0 ,

0

∂ 0 1 = 0 1 + 0 1 +3

0 1.

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 7

In the dual picture, the operator ∂ is just adding one marked point, or breaking

up a node of the curve and gluing a 3-pointed sphere.

Now we deﬁne a third operator γ, which acts on Γ by simply attach a stable

vertex of genus 0 with three edges to an external edge of Γ, and sum over all external

edges.

Example 2.4.

γ 1 = 1 0 ,

γ 0 =2 0 0 .

If the graph Γ has no external edge, then we may simply have γ(Γ) = 0. This

operator is to glue a 3-pointed sphere to the curve along marked points to get a

new node.

Clearly the operators ∂ and γ also preserves the stability, and they both increase

the number of external edges by one. Intuitively, ∂ + γ is the inverse of the forgetful

map. Write D = ∂ + γ. Then D preserves the subspaces V c .

2.3. Abstract free energy. Let us now deﬁne an abstract quantum field theory

based on the diagrammatics of stable graphs discussed above.

Definition 2.1. For g ≥ 2, we define the abstract free energy of genus g to be

X 1

(9) Fbg = Γ.

c

| Aut(Γ)|

Γ∈Gg,0

X 1

(10) Fbg,n = Γ,

c

| Aut(Γ)|

Γ∈Gg,n

Example 2.5. Here we list some explicit expressions of Fbg,n for small (g, n).

Fb0,3 = 1

6 0

,

Fb0,4 = 1

24 0 + 18 0 0

,

Fb0,5 = 1

120 0 + 81 0 0 0 1

+ 12 0 0

,

Fb1,1 = 1 + 21 0

,

Fb1,2 = 1

2 1 + 14 0 + 12 1 0 + 41 0 0 + 14 0 0

,

8 ZHIYUAN WANG AND JIAN ZHOU

Fb1,3 = 1

6 1 1

+ 12 0 + 21 1 0 + 61 1 0 + 41 0 0

+ 41 0 0 1

+ 12 0 0 + 21 1 0 0 + 41 0 0 0

0

+ 41 0 0 0 + 61 0 0 ,

Fb2,0 = 2 + 12 1 + 21 1 1 + 18 0 + 12 1 0 + 81 0 0 1

+ 12 0 0

,

Fb2,1 = 2 + 12 1 + 1 1 + 81 0 + 12 1 0 + 21 1 0

+ 12 1 0 + 21 1 0 1 + 41 0 0 + 14 0 0

+ 61 0 0 + 21 0 0 1 + 21 0 0 1 + 41 0 0 0

0

+ 81 0 0 0 + 41 0 0.

2.4. A recursion relation for the abstract free energy.

Example 2.6. Using the expressions in the above example, one can directly check

the following identities:

9 b 2

K Fb0,4 = (F0,3 ) ,

2

K Fb0,5 = 12Fb0,3 Fb0,4 ,

K Fb1,1 = 3Fb0,3 ,

K Fb1,2 = 6Fb0,4 + 3Fb1,1 Fb0,3 ,

K Fb1,3 = 10Fb0,5 + 4Fb1,1 Fb0,4 + 6Fb1,2 Fb0,3 ,

1

K Fb2 = Fb1,2 + (Fb1,1 )2 ,

2

K Fb2,1 = 3Fb1,3 + 2Fb1,1 Fb1,2 .

In general, we have the following relation.

Theorem 2.1. For 2g − 2 + n > 0, we have

X

n+2 b 1

(11) K Fbg,n = Fg−1,n+2 + (n1 Fbg1 ,n1 )(n2 Fbg2 ,n2 ),

2 2 g +g =g,n +n

1 2 1 2 =n+2,

n1 ≥1,n2 ≥1

where the sum on the right hand side is taken over all stable cases.

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 9

c

Proof. First note that cutting oﬀ an internal edge of a stable graph in Gg,n gives

us two types of new stable graphs, one type is still connected, and must be of

genus g − 1 with n + 2 external edges; the other type of graphs have two connected

components, and have total genus g with n + 2 external edges. Thus we write

where Ig,n and Jg,n are sums over connected graphs and disconnected graphs re-

spectively.

c

We already know that graphs appearing in Ig,n all belong to Gg−1,n+2 ; conversely,

c b

every graph in Gg−1,n+2 may be transformed to a graph appearing in Fg,n once we

glue two external edges together. This tells us that the graphs appearing in Ig,n are

the exactly the same graphs in Fbg−1,n+2 . Now we check the coeﬃcients of every

c

graph in the two sums coincide. For a stable graph Γ ∈ Gg−1,n+2 , there are n+2

2

ways to choose two external edges to be glued together. Denote by Γ e one of the

equivalence classes of graphs obtained from Γ by gluing two external edges. Then

we need to show that

X

′ 1 e n+2 1

(13) K ·Γ = · Γ,

e

| Aut(Γ)| 2 | Aut(Γ)|

e

Γ∈G Γ

where GΓ is the set of equivalence classes of stable graphs Γe such that one can

e ′ e e e is the number

obtain Γ by cutting an edge of Γ, and K (Γ) = N (Γ) · Γ, where N (Γ)

e

of ways to cut an edge of Γ to get Γ. It suﬃces to show that

X

(14) e · | Aut(Γ)| = n + 2 .

N (Γ)

e

| Aut(Γ)| 2

e Γ

Γ∈G

Consider

the set of the choices of picking a pair of external edges of Γ. This set has

n+2

2 elements. We partition this set into disjoint union of HΓe consisting of those

choices for which one can get Γ e by gluing the pair of external edges, where Γe runs

over the set GΓ of all possibilities. Therefore,

X

n+2

(15) |HΓe | = .

2

e Γ

Γ∈G

(16) e · | Aut(Γ)| ,

|HΓe | = N (Γ)

e

| Aut(Γ)|

or equivalently,

|HΓe | e

N (Γ)

(17) = .

| Aut(Γ)| e

| Aut(Γ)|

Let Γ′′ be the graph obtained from Γ e by changing the internal edge that we

cut to get Γ into a dotted edge, and let Γ′ be the graph with two dotted external

edges obtained from Γ′′ by cutting the dotted edge. For examples, see Figure 3 and

Figure 4.

10 ZHIYUAN WANG AND JIAN ZHOU

1 1 1 1

v v v v

1 0 1 1 0 1 1 0 1 1 0 1

1 1 1 1

Γ Γ′ Γ′′ e

Γ

v1 v2 v1 v2 v1 v2 v1 v2

0 0 0 0 0 0 0 0

1 1 1 1 1 1 1 1

Γ Γ ′

Γ ′′ e

Γ

When we talk about automorphisms of Γ′ and Γ′′ , the dotted edges can not be

mapped to solid edges. Now we claim:

|HΓe | 1 1 e

N (Γ)

(18) = = = .

| Aut(Γ)| ′

| Aut(Γ )| ′′

| Aut(Γ )| e

| Aut(Γ)|

The second equality is trivial. We only need to consider the ﬁrst and third

equalities. Recall that |HΓe | is the number of ways to choose two external edges

of Γ to be glued together to get Γ,e thus |He | equals to the number of all possible

Γ

ways to change a pair of external edges of Γ to dotted edges to obtain Γ′ . Similarly,

e equals to the number of all possible ways to change an internal edge of Γ

N (Γ) e to

a dotted edge to obtain Γ′′ .

In the case where the dotted edge is a loop in Γ′′ , denote by v the vertex where

it is incident to. Then let Autv (Γ) be the group of automorphisms of Γ that ﬁx the

vertex v (e.g., for Figure 3, Autv (Γ) ∼

= (S3 )3 × (S2 )2 ). Let Ov be the orbit of v in

Γ under Aut(Γ), then

| Aut(Γ)|

(19) = |Ov |.

| Autv (Γ)|

It is clear that any automorphism of Γ′ ﬁxes the vertex v since there is no other

vertex with dotted half edges, and so Aut(Γ′ ) is a subgroup of Autv (Γ). To compute

the index of this subgroup, it suﬃces to consider the actions of these automorphisms

groups on the external half edges incident at v. Assume there are l external edges

attaching to v in Γ′ including two dotted edges, then we have

| Autv (Γ)| l! 1

(20) ′

= · ,

| Aut(Γ )| (l − 2)! 2

here the factor 21 comes from the symmetry between two dotted edges. Now (19)

and (20) give us

| Aut(Γ)| l

(21) = |Ov | · .

| Aut(Γ′ )| 2

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 11

l

Noticing that |Ov | equals to the number of ways to choose the vertex v, and 2 is

the number of ways to choose two external edges on v, thus

l

|Ov | · = |HΓe |

2

and this proves the ﬁrst equality in (18).

Using the same argument, we now consider Autv (Γ), e i.e., the group of automor-

phisms of Γ e ∼

e that ﬁx v (e.g., for Figure 3 Autv (Γ) ev be the

= (S3 )2 × (S2 )3 ). Let O

e e

orbit of v in Γ under Aut(Γ), then

e

| Aut(Γ)|

(22) ev |.

= |O

e

| Autv (Γ)|

Assume there are k loops attaching to v in Γ′′ including a dotted one, then

e

| Autv (Γ)| k! · 2k 1

(23) ′′

= k−1

· = k.

| Aut(Γ )| (k − 1)! · 2 2

Thus (22) and (23) give us

e

| Aut(Γ)| ev |,

= k · |O

| Aut(Γ′′ )|

e Therefore the third equality in (18) also

and the right hand side equals to N (Γ).

holds.

The case where the dotted edge in Γ′′ is not a loop can be proved similarly. Let

v = (v1 , v2 ) be the ordered pair of end points of the dotted edge. As above, we

consider Autv (Γ) (here the order of (v1 , v2 ) should be preserved, e.g., for Figure

4, Autv (Γ) ∼ = S2 ), and the orbit Ov of the ordered pair v in V (Γ) × V (Γ) under

Aut(Γ). We still have the equality:

| Aut(Γ)|

(24) = |Ov |.

| Autv (Γ)|

Assume there are li external edges attached to vi in Γ′ , then

| Autv (Γ)| l1 ! l2 !

(25) ′

= · · δ = l1 l2 · δ,

| Aut(Γ )| (l1 − 1)! (l2 − 1)!

1

here δ = 2 if (v2 , v1 ) ∈ Ov , and δ = 1 if (v2 , v1 ) 6∈ Ov . The equations (24) and (25)

give us

| Aut(Γ)|

(26) = l1 l2 |Ov | · δ,

| Aut(Γ′ )|

which equals |HΓe |.

e to be the group of automorphisms of Γ

Similarly deﬁne Autv (Γ) e that ﬁx v (e.g.,

e ∼ e e e under Aut(Γ),

for Figure 4, Aut(Γ) = S3 ). Let Ov be the orbit of v in V (Γ) × V (Γ) e

then

e

| Aut(Γ)|

(27) ev |.

= |O

e

| Autv (Γ)|

12 ZHIYUAN WANG AND JIAN ZHOU

Assume there are k internal edges connecting v1 and v2 in Γ′′ including a dotted

one, then

e

| Autv (Γ)| k!

(28) = · δ = k · δ.

| Aut(Γ′′ )| (k − 1)!

Then the equations (27) and (28) imply

e

| Aut(Γ)| ev | · δ,

= k|O

| Aut(Γ′′ )|

e Therefore the ﬁrst and third equalities in (18) also hold in

which equals to N (Γ).

this case.

In summary, we’ve proved the identity

n+2 b

(29) Ig,n = Fg−1,n+2 .

2

The relation

1 X

(30) Jg,n = (n1 Fbg1 ,n1 )(n2 Fbg2 ,n2 )

2g

1 +g2 =g,n1 +n2 =n+2,

n1 ≥1,n2 ≥1

can be established in a similar fashion. For two graphs Γ1 ∈ Ggc1 ,n1 , Γ2 ∈ Ggc2 ,n2 ,

denote by Γe one of the equivalence classes of graphs obtained from Γ1 and Γ2 by

gluing two external edges, one from each graph. Then we need to show that

X

1 e = n1 n2

(31) K ′′ ·Γ Γ1 · Γ2 ,

e

| Aut(Γ)| | Aut(Γ1 )| | Aut(Γ2 )|

e

Γ∈G ′′

Γ

where GΓ′′ is the set of equivalence classes of stable graphs Γ

obtain Γ1 and Γ2 by cutting an edge of Γ,e and

e = N ′′ (Γ)

K ′′ (Γ) e · Γ1 Γ2 ,

where N ′′ (Γ) e to get Γ1 and Γ2 . It suﬃces

to show that

X 1 n1 n2

(32) e ·

N ′′ (Γ) = · .

e

| Aut(Γ)| | Aut(Γ 1 )| | Aut(Γ 2 )|

e

Γ∈G

′′

Γ

Consider the set of the choices of picking one external edge from each of Γ1 and

Γ2 . This set has n1 n2 elements. We partition this set into disjoint union of HΓe′′

consisting of those choices for which one can get Γ e by gluing the pair of external

edges, where Γe runs over the set G ′′ of all possibilities. Therefore,

Γ

X

(33) |HΓe′′ | = n1 n2 .

e

Γ∈G ′′

Γ

e · | Aut(Γ1 )| · | Aut(Γ2 )|

(34) |HΓe′′ | = N ′′ (Γ) ,

| Aut(Γ)|e

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 13

or equivalently,

|HΓe′′ | e

N ′′ (Γ)

(35) = .

| Aut(Γ1 )| · | Aut(Γ2 )| e

| Aut(Γ)|

e by changing the internal edge that we cut

Let Γ′′ be the graph obtained from Γ

to get Γ into a dotted edge, and Γ be the disconnected graph obtained from Γ′′ by

′

|HΓe′′ | 1 1 e

N ′′ (Γ)

(36) = = =

| Aut(Γ1 )| · | Aut(Γ2 )| ′

| Aut(Γ )| ′′

| Aut(Γ )| e

| Aut(Γ)|

can be proved using exactly the same method as the proof of (18). Suppose that

the two external edges to be glued together are incident at v1 in Γ1 and v2 in Γ2

respectively. Denote by v = (v1 , v2 ) the ordered pair of these two vertices. Then the

e and the orbits Ovi , O

subgroups Autvi (Γi ), Autv (Γ) ev are similarly deﬁned. Then

the following equalities hold:

| Aut(Γi )| | Autv1 (Γ1 )| · | Autv1 (Γ1 )|

= |Ovi |, = δ ′′ ,

| Autvi (Γi )| | Aut(Γ′ )|

e

| Aut(Γ)| e

ev |, | Autv (Γ)| = δ ′′ ,

= |O

e

| Autv (Γ)| | Aut(Γ′′ )|

otherwise. Furthermore, we have

|Ov1 | · |Ov2 | · δ ′′ = |HΓe′′ |, ev | · δ ′′ = N ′′ (Γ).

|O e

The equalities (12), (29), (30) completes the proof of the theorem.

Furthermore, the right hand side of (11) can be rewritten in terms of operators

D. To explain this we need the following observation.

Proof. The proof is similar to the proof of Theorem 2.1. First the graphs appear-

ing in the two sums are the same, since we’ve already taken all possible unstable

contractions into consideration in the deﬁnition of the operator D. Thus we only

need to check the coeﬃcients.

c c

Let Γ ∈ Gg,n+1 be a stable graph, and GΓ be the subset of Gg,n consisting of

graphs obtained from Γ by removing an external edge and stabilizing. We only

need to show

X 1 1

D′ ( e = (n + 1) ·

Γ) Γ,

e

| Aut(Γ)| | Aut(Γ)|

e Γ

Γ∈G

e · Γ with N (Γ)

where D′ (Γ) = N (Γ) e the number of ways to obtain Γ from Γ̃ via one

of the operations in D = ∂ + γ. Then it suﬃces to show

X

(37) e · | Aut(Γ)| = n + 1.

N (Γ)

e

| Aut(Γ)|

e Γ

Γ∈G

14 ZHIYUAN WANG AND JIAN ZHOU

Consider the set of the choices of picking an external edge of Γ, and partition it

e after removing

to subsets HΓe consisting of those choices for which one can obtain Γ

this edge and stabilizing. Clearly

X

n+1

|HΓe | = = n + 1.

1

e

Γ∈G Γ

e · | Aut(Γ)|

N (Γ) = |HΓe |,

e

| Aut(Γ)|

or equivalently,

|HΓe | e

N (Γ)

= .

| Aut(Γ)| e

| Aut(Γ)|

Denote by Γ′ the graph obtained from Γ by changing the external edge that we

remove to a dotted edge. We will show that

|HΓe | 1 e

N (Γ)

(38) = = .

| Aut(Γ)| ′

| Aut(Γ )| e

| Aut(Γ)|

Let v be the vertex of Γ and Γ′ that the dotted edge attaches to. The ﬁrst equality

follows from

| Aut(Γ)| | Autv (Γ)| k!

= |Ov |, = ;

| Autv (Γ)| | Aut(Γ′ )| (k − 1)!

k

|HΓe | = · |Ov |,

1

where Autv (Γ) is the set of automorphisms of Γ that preserve v, Ov is the orbit of

v in Γ under Aut(Γ), and k is the number of external edges attaching to v in Γ.

e that is, ṽ = v if the graph is still

Let ṽ be the corresponding vertex of v in Γ,

stable after removing the dotted edge, and ṽ the vertex that v contracts to if the

graph is unstable. The second equality in (38) follows from

e

| Aut(Γ)| eṽ |,

= |O e = | Aut(Γ′ )|;

| Autṽ (Γ)|

e

| Autṽ (Γ)|

e = |O

N (Γ) eṽ |,

e is the set of automorphisms of Γ

where Autṽ (Γ) e that preserve ṽ, and O

eṽ is the orbit

e under Aut(Γ).

of ṽ in Γ e

Therefore the equality (38) holds, which proves the lemma.

Even though so far we have not deﬁned Fb1 , Fb0,2 , and Fb0,1 , it will be convenient

for later use to deﬁne DFb1 , DFb0,2 , DDFb0,1 by formally applying Lemma 2.1 as

follows:

DFb1 := Fb1,1 ,

(39) DFb0,2 := 3Fb0,3 ,

DDFb0,1 := 6Fb0,3 .

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 15

Now using Lemma 2.1 and conventions in (39), our Theorem 2.1 can be rewritten

in the following way.

1 X

(40) K Fbg,n = (DDFbg−1,n + DFbg1 ,n1 DFbg2 ,n2 ).

2 g +g =g,

1 2

n1 +n2 =n

energy:

g−1

X

b 1 b

(41) K Fg = (D∂ Fg−1 + ∂ Fbr ∂ Fbg−r ).

2 r=1

In this section, we generalize the construction of Section 2 to the case where all

the half-edges of graphs may have labels.

3.1. Diagrammatics for labelled graphs. Fix a positive integer N , for a stable

graph Γ ∈ Gg,n we label some indices on the edges as follows. For an external edge,

we label an index in {1, 2, · · · , N }; and for an internal edge, we label an index in

{1, 2, · · · , N } at each of its end points. In other words, we label each half edge

incident at a vertex.

As generalizations of the edge-cutting operator K deﬁned in Section 2.2, we

deﬁne the edge-cutting operators Kij as the operator that cuts oﬀ an internal

edges with two labels i and j, then sum over all such internal edges.

1 1 2 1 1 1 2 1

K11 0 0 =2 0 0

1 1 1 1 ,

1 1 2 1 1 1 2 1

K12 0 0 = 0 0

1 1 1 1 ,

1 1 2 1

K22 0 0 =0

1 1 .

Similarly, we also have operators ∂i and γi generalizing the operators ∂ and γin

Section 2.2 respectively. The operator ∂i has two parts, one is to add an external

edge labelled by i, and the other is to break up an internal edge and attach a 3-

pointed Riemann sphere such that the new external edge has label i. The operator

γi is to attach a 3-pointed Riemann sphere to an external edge e, and move the

label of e to be the label of one of the new external edge, while the other new

external edge has label i. The two labels of the new internal edge can be chosen

arbitrarily.

16 ZHIYUAN WANG AND JIAN ZHOU

2 2

∂1 1 = 1 1,

2 1 2

∂1 0 = 0 1 + 1 01 20 1 + 1 02 20 1 + 2 1 01 20 1

1 1 2 2 2 2 2

2 2 .

Example 3.3. For N = 2, here are some examples of the operator γi .

γ1 2 1 1 2 + 1 2 2 + 2 1 2 + 2 2 2

1 = 1 0 1 0 1 0 1 0

1 1 1 1,

2 1 2 11 2 12 2 21 2 22

γ1 0 = 0 0 1 + 0 0 1 + 0 0 1 + 0 0 1

1 1 1 1

2 2 2 2 2 .

Now let Gg,n (N ) be the set of all labelled stable graphs of genus g with n external

c

edges and N possible indices, and Gg,n (N ) be the set of connected ones. In the

labelled case we also deﬁne the abstract free energy in the same way.

Definition 3.1. For g ≥ 2, the abstract free energy Fbg is defined by:

X 1

(42) Fbg = Γ.

c

| Aut(Γ)|

Γ∈Gg,0 (N )

1 1 1 1 2 2 1 2

Fb2 = 2 + 2 1 1 + 2 1 1 + 1 1

1

1 1

2 1

+ 2 1 + 2 1 + 1 +···

1 2 2

The complete expressions for Fb2 can be obtained by adding all possible labels on

the graphs of Fb2 for the N = 1 case and modifying the number of automorphisms.

3.2. A recursion relation for the abstract free energy. In the labelled case,

we also have a recursion relation generalizing (41). Here we deﬁne Di = ∂i + γi .

Theorem 3.1. For g ≥ 2, we have

g−1

X

(43) Kij Fbg = Di ∂j Fbg−1 + ∂i Fbr ∂j Fbg−r

r=1

for i 6= j, and

g−1

X

1

(44) Kii Fbg = (Di ∂i Fbg−1 + ∂i Fbr ∂i Fbg−r ).

2 r=1

The proof is an easy modiﬁcation of the proof of Theorem 2.2 and is omitted.

For the case of g = 2 we use the following convention

P j P k j P k j

1

∂j Fb1 =: j( 1 + k 2 0 + k6=l 0 )

(45) k l

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 17

In this section we formulate a Feynman rule for the diagrammatics discussed in

the previous sections, in such way we are able to deﬁne a new free energy Fb from a

given holomorphic free energy F and a choice of propagator κ. We also represent

this new ﬁeld theory as formal Gaussian integrals. This construction realizes the

abstract free energy and its quadratic recursion relations discussed in preceding

sections.

4.1. Realization of the abstract QFT by Feynman rules. In this subsection

we will present a construction to realize the abstract quantum ﬁeld theory deﬁned

in Section 2.3 and Section 3.1 by Feynman rules that assign the contribution of a

vertex and the contribution of edge:

Y Y

(46) Γ 7→ ωΓ = ωv · ωe .

v∈V (Γ) e∈E(Γ)

for ωe , we need ﬁx a nondegenerate symmetric matrix κ = (κij )1≤i,j≤N of functions

in t1 , . . . , tN as propagators.

For a given positive integer N , let t1 ,t2 ,· · · ,tN be the coordinates on an N -

dimensional vector space, and κ = (κij ) be a nondegenerate symmetric matrix of

size N × N . Moreover, we ﬁx a sequence of holomorphic function Fg (t1 , · · · , tN )

for g ≥ 0. Let

∞

X

(47) F (t; λ) = λ2g−2 Fg (t).

g=0

c

Now for a labelled stable graph Γ ∈ Gg,0 (N ), we associate a polynomial ωΓ as

ext

follows. Let V (Γ), E (Γ), and E(Γ) be the sets of vertices, external half edges,

and internal edges of Γ respectively. For each e ∈ E ext (Γ), write l(e) as the label

on the external e; and for each e ∈ E(Γ), write l1 (e) and l2 (e) as the two labels

on e (the order is not important). For each v ∈ V (Γ), write g(v) as the genus

associated

P to v, and vali (v) as the number of half-edges labelled by i incident at v,

then i vali (v) is the valence of v. Then ωΓ is deﬁned by

Y (val (v),··· ,valN (v))

Y

(48) ωΓ = Fg(v)1 (t) κl1 (e)l2 (e) ,

v∈V (Γ) e∈E(Γ)

(val (v),··· ,valN (v))

where Fg(v)1 (t) := ( ∂t∂1 )val1 (v) · · · ( ∂t∂N )valN (v) Fg(v) (t). For N = 1, this

expression simpliﬁes to

Y (val(v))

(49) ωΓ = ( Fg(v) (t))κ|E(Γ)| .

v∈V (Γ)

X 1

(50) Fbg = ωΓ .

c

| Aut(Γ)|

Γ∈Gg,0 (N )

Example 4.1. For N = 1, we have

1 1 1 (4) 1 5

(51) Fb2 = F2 + κ[ F1′′ + (F1′ )2 ] + κ2 ( F0 + F1′ F0′′′ ) + κ3 (F0′′′ )2 .

2 2 8 2 24

18 ZHIYUAN WANG AND JIAN ZHOU

1 (2,0) 1 (1,0) 2 1 (0,2) 1 (0,1) 2

Fb2 = F2 + ( F1 + (F1 ) )κ11 + ( F1 + (F1 ) )κ22

2 2 2 2

1 (4,0) 1 (1,0) (3,0) 2 1 (0,4) 1 (0,1) (0,3) 2

+ ( F0 + F1 F0 )κ11 + ( F0 + F1 F0 )κ22

8 2 8 2

(1,1) (1,0) (0,1) 1 (2,2) (1,0) (1,2) (0,1) (2,1)

+ (F1 + F1 F1 )κ12 + ( F0 + F1 F0 + F1 F0 )κ212

2

1 (2,2) 1 (1,0) (1,2) 1 (0,1) (2,1)

+ ( F0 + F1 F0 + F1 F0 )κ11 κ22

4 2 2

1 (3,1) 3 (1,0) (2,1) 1 (0,1) (3,0)

+ ( F0 + F1 F0 + F1 F0 )κ11 κ12

2 2 2

1 (1,3) 3 (0,1) (1,2) 1 (1,0) (0,3)

+ ( F0 + F1 F0 + F1 F0 )κ22 κ12

(52) 2 2 2

5 (3,0) 2 3 5 (0,3) 2 3 1 (3,0) (0,3) 3 (2,1) (1,2) 3

+ (F0 ) κ11 + + (F0 ) κ22 + ( F0 F0 + F0 F0 )κ12

24 24 6 2

3 (2,1) 2 1 (3,0) (1,2) 2

+ ( (F0 ) + F0 F0 )κ11 κ22

8 4

3 (1,2) 2 1 (0,3) (2,1) 2

+ ( (F0 ) + F0 F0 )κ22 κ11

8 4

3 (2,1) 2 (3,0) (1,2) 3 (1,2) (0,3) (2,1)

+ ( (F0 ) + F0 F0 )κ11 κ212 + ( (F0 )2 + F0 F0 )κ22 κ212

2 2

5 (3,0) (2,1) 5 (0,3) (1,2)

+ F0 F0 κ211 κ12 + F0 F0 κ222 κ12

4 4

1 (3,0) (0,3) 9 (2,1) (1,2)

+ ( F0 F0 + F0 F0 )κ11 κ22 κ12 .

4 4

4.2. Representation by formal Gaussian integrals. In fact, the free energy

Fbg deﬁned in last subsection has a representation using formal Gaussian integrals.

Consider the following partition function

Z

−2 ∂F0 (t) 1 ∂ 2 F0 (t)

Z(t, κ) = exp F (η) − λ (η − t) + (η − t)T (η − t)

∂t 2 ∂t2

(53)

1

+ (η − t)T κ−1 (η − t) dη,

2

where the integral is over η ∈ RN . Now do the Taylor expansion of F (η), we get

Z X YN

λ2g−2 (l(1) ,··· ,l(N ) ) (i)

Z(t, κ) =Z(t) exp QN (i) Fg (t) (ηi − ti )l

i=1 l !

P

2g−2+ l(i) >0 j=1

−2

λ

− (η − t)T κ−1 (η − t) dη

2

(54) P k

X X 1 λ2 k gk −2k Y (l

(1) (N )

,··· ,lj )

=Z(t) Q Q (i)

Fgj j ×

k!

k≥0 2gj −2+P l(i) >0 1≤j≤k i lj ! j=1

i j

Z Y

N Pk (i) λ−2

(ηi − ti ) j=1 lj · exp − (η − t)T κ−1 (η − t) dη,

i=1

2

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 19

where Z(t) = exp(F (t)) is the holomorphic partition function. Then evaluate the

Gaussian integral on the right-hand side of the last equality, we have

Z Y N Pk (i) λ−2

(ηi − ti ) j=1 lj · exp[− (η − t)T κ−1 (η − t)]dη

i=1

2

(55) l

(2π)n 1 1 X Y 2

=( ) · l

2 (λ κaσ(2j−1) aσ(2j) ),

det(λ−2 κ−1 ) 2 · l! j=1 σ∈S2l

P (i) P (i)

for i,j lj = 2l even, and equals zero if is odd, where the sequence

i,j lj

P (i)

(a1 , · · · , a2l ) consists of indices in {1, · · · , N } with the number of i equals j lj .

Let the normalized partition function be

(56) e κ) = Z(t, κ)/(2πλ2 ) n2 ,

Z(t,

then this relates to our free energy (50) in the following way.

Theorem 4.1. We have

X∞

(57) e κ) = exp(

Z(t, λ2g−2 Fbg (t)),

g=0

1

(58) Fb0 = F0 , Fb1 = F1 + log(det(κ)).

2

p

e κ)/(Z(t) det(κ)) is

Proof. From (54) and (55), the coeﬃcient of λn in Z(t,

(1) (N )

k (l ,··· ,lj ) l

X 1 1 Y Fgj j X Y

· l · κaσ(2j−1) aσ(2j) .

P k! 2 · l! j=1 l(1) ! · · · l(N ) !

2 gj −2k+2l=n j j σ∈S2l j=1

This corresponds to the summation of ωΓ over all possible ways to glue some sta-

(1) (N )

(l ,··· ,l )

ble vertices of types {Fgj j j

}1≤j≤l together to obtain a stable graph Γ (not

1

necessarily connected), with coeﬃcients | Aut(Γ)| . Then since

l = |E(Γ)|, k = |V (Γ)|,

we have

k

X

g(Γ) = 1 + l − k + gj .

j=1

Thus

X λ2g−2

e κ) =

Z(t, ωΓ

| Aut(Γ)|

Γ∈Gg,0

and

X X ∞

e = λ2g−2

log(Z) ωΓ = λ2g−2 Fbg .

c

| Aut(Γ)| g=0

Γ∈Gg,0

20 ZHIYUAN WANG AND JIAN ZHOU

X∞

Z̃(t, κ) = exp( λ2g−2 Fbg (t))

g=0

=Z(t) ×

k! l1 ! · · · lk !

k≥0 2gi −2+li >0 l

earlier work on holomorphic anomaly equations [11, 13, 15], where the propagator

κ was chosen to be the Zamolodchikov metric which is non-holomorphic. Later

in this paper we will see that the propagator κ admits diﬀerent choices, which

give us diﬀerent theories in mathematical physics. In particular, by taking suitable

holomorphic propagators we will see that partition functions for some well-known

quantum spectral curves can be constructed in our formalism.

4.3. Realization of the operators on stable graphs. Now we discuss the re-

alization of the operators K, ∂ and γ deﬁned in Section 2.2.

The operator K is realized by the diﬀerential operator ∂κ . In fact, by (49),

Y

(val(v))

∂κ ωΓ = ∂κ ( Fg(v) (t))κ|E(Γ)|

v∈V (Γ)

Y (val(v))

= |E(Γ)| · ( Fg(v) (t))κ|E(Γ)|−1

v∈V (Γ)

X

= ωΓ′ = ωKΓ ,

Γ′

where the summation on the right-hand side of the second equality is over all graphs

Γ′ obtained from Γ by cutting an edge.

The operator γ is realized by the operator of multiplication by |E ext (Γ)| · κF0′′′

on ωΓ . In fact, the eﬀect of γ is to change each external edge incident at v of a

stable graph Γ to an internal edge joining the vertex v to an additional trivalent

vertex of genus 0. Let Γ′ be one of stable graphs obtained from Γ in this way, then

by the Feynman rule,

Therefore,

X

(61) ωγΓ = ωΓ′ = |E ext (Γ)| · κF0′′′ · ωΓ .

Γ′

of contributions from the vertices and the contributions from the internal edges.

Hence ∂t ωΓ is a summation of the contributions of the actions on each vertices and

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 21

Y

(val(v))

∂t ωΓ = ∂t ( Fg(v) (t))κ|E(Γ)|

v∈V (Γ)

X (val(v)+1)

Y (val(v ′ ))

= Fg(v) (t) · Fg(v′ ) (t)) · κ|E(Γ)|

v∈V (Γ) v ′ ∈V (Γ)−{v}

Y (val(v))

+ Fg(v) (t)) · |E(Γ)|κ|E(Γ)|−1 · ∂t κ.

v∈V (Γ)

X X

(62) ∂Γ = Γv + Γe ,

v∈V (Γ) e∈E(Γ)

V (Γ), and Γe is obtained from Γ by changing an internal edge e ∈ E(Γ) to two

internal edges incident at an addition trivalent vertex w of genus zero. Therefore,

by the Feynman rule we get:

X X

ω∂Γ = ωΓv + ωΓe

v∈V (Γ) e∈E(Γ)

X (val(v)+1)

Y (val(v ′ ))

= Fg(v) (t) · Fg(v′ ) (t)) · κ|E(Γ)|

v∈V (Γ) v ′ ∈V (Γ)−{v}

X Y (val(v))

+ Fg(v) (t)) · κ|E(Γ)|+1 · F0′′′ (t).

e∈E(Γ) v∈V (Γ)

of t which satisﬁes

∂κ

(63) = κ2 · F0′′′ (t),

∂t

here κ needs not to be holomorphic. A general solution to this equation is

1

κ(t) = ,

C − F0′′ (t)

where C may be either a constant or an anti-holomorphic function of t.

the quadratic recursion relations in Theorem 2.2 and using the realizations of the

operators on graphs discussion in the preceding subsection, we obtain the recursion

relations for Fbg immediately.

Theorem 4.2. When the propagator κ satisfies the condition (63), the free energy

Fbg for g ≥ 2 satisfies the equations

g−1

X

1

(64) ∂κ Fbg = (Dt ∂t Fbg−1 + ∂t Fbr ∂t Fbg−r ),

2 r=1

22 ZHIYUAN WANG AND JIAN ZHOU

1 ′′ 1 ′ 2 1 (4) 5

∂κ Fb2 = F + (F1 ) + κ( F0 + F1′ F0′′′ ) + κ2 (F0′′′ )2 ,

2 1 2 4 8

1 1 (5) 1 (4) 1

∂t Fb2 = F2′ + κ[ F1′′′ + F1′ F1′′ ] + κ2 ( F0 + F1′ F0 + (F1′ )2 F0′′′ + F1′′ F0′′′ )

2 8 2 2

2 (4) 5

+ κ3 [F1′ (F0′′′ )2 + F0′′′ F0 ] + κ4 (F0′′′ )3 .

3 8

Here in the second equality we have used (63).

Example 4.4. Direct computation gives

1 (4)

Dt ∂t Fb1 = F1′′ + κ( F0 + F1′ F0′′′ ) + κ2 (F0′′′ )2 ,

2

one may check that

1

∂κ Fb2 = (Dt ∂t Fb1 + (∂t Fb1 )2 ).

2

4.5. The general case. Now we return to the general case for an arbitrary dimen-

sion N . Similar to N = 1 case, the operators Kij and ∂i deﬁned in Section 3 are

realized by ∂κij and ∂ti respectively. To realize the operator ∂i as the diﬀerential

operator ∂ti , we need to impose some conditions on the propagators κij analogous

to (63):

∂κjk X ∂ 3 F0

(65) = κjl κkm ,

∂ti ∂ti ∂tl ∂tm

l,m

or in matrix form:

∂κ ∂H(F0 )

(66) =κ κ

∂tk ∂tk

2

∂ F0

where H(F0 ) = ( ∂ti ∂tj

) is the Hessian of F0 . A general solution is given by

where C is a symmetric anti-holomorphic or constant matrix such that (C − H(F0 ))

is invertible.

The realization of the operator Di = ∂i + γi is more subtle. For our purpose,

it suﬃces to deﬁne the operator Dti ∂tj as a realization of the operator Di ∂j as

follows.

c

Definition 4.2. For Γ ∈ Gg,0 , we define

Dti ∂tj ωΓ =ωDi ∂j Γ

N

X

(67) ∂2 ∂ ∂ 3 F0

= ωΓ + ωΓ · κlm (t) ,

∂ti ∂tj ∂tl ∂tm ∂ti ∂tj

l,m=1

where

Y (val (v),··· ,valN (v))

Y

ωΓ = Fg(v)1 (t) · κl1 (e)l2 (e) (t, t̄).

v∈V (Γ) e∈E2 (Γ)

Then by applying the Feynman rules, Theorem 3.1 gives us

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 23

Theorem 4.3. When the condition (66) is satisfied, for g ≥ 2, we have the equa-

tions

g−1

X

(68) ∂κij Fbg = Dti ∂tj Fbg−1 + ∂ti Fbr ∂tj Fbg−r

r=1

for i 6= j, and

g−1

X

1

(69) ∂κii Fbg = (Dti ∂ti Fbg−1 + ∂ti Fbr ∂ti Fbg−r )

2 r=1

In the rest of this paper we will present some examples of our formalism developed

above. In this section we will focus on the case of holomorphic anomaly equations.

As we have mentioned in the Introduction, the holomorphic anomaly equation

was introduced by Bershadsky et al in physics literatures [2] [3] in order to compute

the Gromov-Witten invariants of the quintic Calabi-Yau threefold. They developed

a method to solve the non-holomorphic free energies Fg (t, t̄) recursively using the

holomorphic anomaly equations

g−1

X

1

(70) ∂¯ī Fg = C̄īj̄ k̄ e2K Gj j̄ Gkk̄ (Dj Dk Fg−1 + Dj Fr Dk Fg−r ), g ≥ 2,

2 r=1

Fg (t) = lim Fg (t, t̄)

t̄→∞

they got the generating series of Gromov-Witten invariants F (t) of the threefold

on the small phase space. The non-holomorphic free energies Fg (t, t̄) are supposed

to have modularity while the holomorphic free energies Fg (t) does not, and this

has been exploited to use the theory of modular forms. Since the ring of modular

forms are often polynomial algebras with ﬁnitely many generators, this suggests

some polynomiality properties of the non-holomorphic free energies [25]. This has

led to many recent progress on the computation of Gromov-Witten invariants of

quintic Calabi-Yau 3-folds.

Unfortunately the mathematical or geometric deﬁnition of the nonholomorphic

free energy is lacking in the mathematical literature. According to [3], the nonholo-

morphic free energy should be closely related to some geometric structures on the

relevant moduli spaces. In the genus zero case, it is known that the free energy is

related to the special Kähler geometry of the moduli space [4]. In the genus one

case, the free energy is related to the tt∗ -geometry of the moduli space [5] [8] and

the theory of analytic torsions [2]. In the higher genus case, the nonholomorphic

free energy is supposed to be a suitable non-holomorphic section of a holomorphic

line bundle on the moduli space [3]. Such an approach depends heavily on the

special properties of Calabi-Yau 3-folds.

It is interesting to further clarify the geometric meaning of non-holomorphic free

energy, and furthermore, to ﬁnd analogue of holomorphic anomaly equation for

Gromov-Witten theory of general symplectic manifolds or other related theories.

The formalism we develop in this paper shed some lights on this problem. In

24 ZHIYUAN WANG AND JIAN ZHOU

anomaly equations, and it does not need to start with a Calabi-Yau 3-fold.

Let us brieﬂy recall the direct integration approach. According to [3], the non-

holomorphic free energy should count the contributions from degenerate Riemann

surfaces, and so it should involve the boundary strata of the Deligne-Mumford

moduli spaces of algebraic curves. Witten [23] interpreted the holomorphic anomaly

equation from the point of view of geometric quantization of symplectic vector

spaces and background independence. Inspired by this work, Aganagic, Bouchard

and Klemm [1] obtained via change of polarizations the following general form for

Fbg :

(71) b IJ

Fg (t, t̄) = Fg (t) + Γg ∆ , ∂I1 · · · ∂In Fr<g (t) ,

in some particular cases, ∆IJ can be taken to be −(τ − τ̄ )IJ , where τ = (τij ) =

∂ 2 F0

∂ti ∂tj ). These particular cases were interpreted for matrix models using Eynard-

Orantin topological recursion by Eynard, Mariño and Orantin [13]. Furthermore,

they also represented the partition function as a formal Gaussian integral [13, (4.27)]

and presented the Feynman graphs and Feynman rules for the terms that contribute

to Γg . These results have been generalized to other models by Grimm, Klemm,

Mariño and Weiss [15]. These authors reformulated the holomorphic anomaly equa-

tion as a quadratic recursion relation for the derivative of Fbg with respective to the

propagators ∆IJ (cf. [15, (7.50)]:

X g−1

∂ Fbg 1 bg−1 + 1

(72) = D I ∂J F ∂I Fbr ∂J Fbg−r .

∂∆IJ 2 2 r=1

From the consideration of modularity of Fbg , the propagators ∆IJ has the form

1

(73) ∆IJ = √ ((τ − τ̄ )−1 )IJ + E IJ ,

−1

where E IJ is a holomorphic function. They also derived the formal Gaussian integral

representation of the partition function and the Feynman expansions for Fbg .

It is clear that our formalism is inspired by [1,13,15]. The results in these works

when the propagators are

(74) κ = −(τ − τ̄ )−1

are special cases of our formalism. Indeed, the condition (66) clearly holds for

such propagators, and so our formalism can be applied to the free energy Fg of

Gromov-Witten theory of a compact Calabi-Yau 3-fold to get a sequence Fbg that

satisﬁes the holomorphic anomaly equations. It is remarkable that the application

of our formalism does not need the Calabi-Yau condition, and so it applies to the

Gromov-Witten theory of any compact symplectic manifold, any Gromov-Witten

type theory as the FJRW theory.

t30

Example 5.1. For FJRW theory of type A1 , F0 (t0 ) = 3! , and so we have

∂ 2 F0

(75) τ= = t0 ,

∂t20

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 25

and so

1

(76) κ= .

t̄0 − t0

The free energy functions Fg (t0 ) are generating functions of intersection numbers

on Mg,n , restricted to the small phase space ti = 0 for i > 0. Therefore, for g ≥ 1,

(77) Fg (t0 ) = 0.

Then our construction yields a sequence Fbg (t, t̄):

Example 5.2. For the Gromov-Witten theory of P1 , the free energy function re-

stricted to the small phase space is given by

t20 t1

F0 (t0 , t1 ) = + qet1 ,

2

t1

F1 (t0 , t1 ) = − ,

24

Fg (t0 , t1 ) = 0, g ≥ 2,

where q is the degree tracking parameter. The period matrix τ can be directly com-

puted as follows:

t1 t0

(78) τ=

t0 qet1

and so the propagator is

−1

t1 − t̄1 t0 − t̄0

(79) κ=−

t0 − t̄0 qet1 − qet̄1

It is an interesting problem to extend our construction to the big phase space that

includes all the gravitational descendants of primary observables. It is well-known

that Gromov-Witten theory and its various generalizations in the mathematical

literature describes what physicists refer to as the topological string theory. The

mathematical meaning or deﬁnition of the partition function of what physicists refer

to as the string theory needs to be clariﬁed at present. For example, for Gromov-

Witten theory, the free energy Fg depends on inﬁnitely many coupling constants

{ti,n }1≤i≤N,n≥0 , we expect the string theory behind it to have free energy Fbg that

also depends on {t̄i,n }1≤i≤N,n≥0 . We hope our construction shed some lights on

this problem.

When the propagators take the form

(80) κ = −(τ − τ̄ )−1 + E,

where E is a symmetric matrix whose entries are holomorphic functions, the condi-

tion (66) do not hold in general. This means that we need to modify the diagram-

matics we discuss in this paper. We will do this in a subsequent work [22].

Curves

In this section we present the second class of applications of our formalism.

We will apply it to topological 1D gravity [27]. The propagator in this case is

holomorphic. By making particular choices of coupling constants, this gives us

examples concerning the quantum spectral curves [16].

26 ZHIYUAN WANG AND JIAN ZHOU

6.1. Some results in topological 1D gravity. In [27], the second author deﬁned

the partition function of the topological 1D gravity as a formal Gaussian integral:

Z

1 1

(81) Z 1D = 1 dx exp 2 S(x) ,

(2πλ2 ) 2 λ

1 X xn

(82) S(x) = − x2 + tn−1 .

2 n!

n≥1

X X Pk k

Y

(2n − 1)!! 2n−2 mj m j

(83) Z= Qk λ j=1 · tj−1 .

n≥0

Pk

mj j=2n j=1 (j!)mj m j! j=1

j=1

Using Wick’s theorem, it is also straightforward to write down the Feynman graphs

and the Feynman rules

X 1 Y

(84) Z= λval(v)−2 tval(v)−1 ,

| Aut(Γ)|

Γ∈G v∈V (Γ)

where the sum is taken over the set G all possible graphs, with the following Feyn-

man rules:

(86) w(e) = 1.

One of the results obtained in [27] is that after a natural change of variables, the

free energy functions Fg have nice simple expressions. This coordinate change can

be obtained by the idea of renormalization. Alternatively, one can formally apply

the stationary phase method and ﬁnd the critical point of S(x) and consider the

Taylor expansion of S(x) at its critical point. This gives us the following change of

coordinates:

∞

X X

1 tp1 tp

I0 = ··· k ,

k p1 ! pk !

k=1 p1 +···+pk =k−1

(87)

X In

Ik = tn+k 0 , k ≥ 1,

n!

n≥0

and

X∞

(−1)n I0n

(88) tk = In+k ,

n=0

n!

where I0 is the critical point of S(x), and In −δn,1 (n ≥ 1) are the Taylor coeﬃcients

at x = I0 . Using the new coordinate {Ik }k≥0 , the action (82) can be rewritten as

∞

X X∞

(−1)k (In−1 − δn,2 )

(89) S(x) = (Ik + δk,1 )I0k+1 + (x − I0 )n .

(k + 1)! n=2

n!

k=0

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 27

Then the partition function (81) can be rewritten in terms of another formal Gauss-

ian integral:

X ∞

1D 1 (−1)k k+1 1 1

Z = exp 2 (Ik + δk,1 )I0 + log

λ (k + 1)! 2 1 − I1

k=0

(90) Z

1 1 1 2 X In−1 xn

· 1 dx exp (− x + n · ) .

(πλ2 ) 2 λ2 2 (1 − I1 ) 2 n!

n≥3

gravity. Using this new equivalent theory, the free energy F 1D = g≥0 λ2g−2 Fg1D =

log(Z 1D ) of this theory can be written down as follows:

∞

X (−1)k

F01D = (Ik + δk,1 )I0k+1 ,

(k + 1)!

k=0

1 1

(91) F11D = log ,

2 1 − I1

X 2g−1

Y lj

l 1 Ij

Fg1D = hτ2l2 · · · τ2g−1

2g−1

ig , g ≥ 2.

P2g−1 j−1 j=2

lj ! (1 − I1 )(j+1)/2

j=2 2 lj =g−1

By rewriting the theory in the new coordinates {Ik }, the free energy F 1D has a

diﬀerent Feynman diagram expansion [27, (130)]:

X 1 Y Y 1

(92) Fg1D = Ival(v)−1 · ,

c,0

| Aut(Γ)| 1 − I1

Γ∈Gg,0 v∈V (Γ) e∈E(Γ)

c,0 c

for g ≥ 2, where Gg,0 is the subset of Gg,0 consists of stable graphs whose vertices

are all of genus 0.

graphs. Now we see that the Feynman rules (92) is indeed a special case of the

realization of diagrammatics of stable graphs described in Section 4. I.e., Fg1D = Fbg

(g ≥ 2) for suitable κ and a suitable sequence Fg (X) for g ≥ 0. By comparing

(92) with the Feynman rules (49), it is clear that for topological 1D gravity, the

propagator is taken to be

1

(93) κ= .

1 − I1

Since in (92), the summation is taken over stable graphs with only genus zero

vertices, so we have Fg (X) = 0 for g ≥ 1, and

(val(v))

(94) F0 (X) = Ival(v)−1 ,

where val(v) ≥ 3. In other words, we need a function F0 (X) such that

dn

(95) F0 (X) = In−1

dX n

for n ≥ 3. Since In−1 involves an inﬁnite set of indeterminates t0 , t1 , . . . , it is not

clear how to choose one of them or a combination of them to be X. We will take

28 ZHIYUAN WANG AND JIAN ZHOU

new independent variable. Now we deﬁne a function F0 (X) by

X∞ X∞

(−1)k (In−1 − δn,2 )

(96) F0 (X) = (Ik + δk,1 )I0k+1 + (X − I0 )n .

(k + 1)! n=2

n!

k=0

Write X̃ = X − I0 , then

∞

X X∞

(−1)k (In−1 − δn,2 ) n

(97) F0 (I0 + X̃) = (Ik + δk,1 )I0k+1 + X̃ .

(k + 1)! n=2

n!

k=0

This is equivalent to say

∞

X (−1)k

F0 (I0 ) = (Ik + δk,1 )I0k+1 ,

(k + 1)!

k=0

dF0 (X)

(98) = 0,

dX X=I0

dn F0 (X)

= In−1 − δn,2 , n ≥ 2.

dX n X=I0

Now using the formalism in Section 4, we can construct a sequence Fbg (X) from

F (X) = λ−2 · F0 (X), and the propagator

1

(99) κ = − ′′

F0 (X)

which clearly satisﬁes the condition (63). Hence (93), (94)and (95) hold when

X = I0 . For g ≥ 2 we have

X Y

1 (val(v))

(100) Fbg (X) = F0 (X) · κ|E(Γ)| ,

c,0

| Aut(Γ)|

Γ∈Gg,0 v∈V (Γ)

(101)

X Y

1 (val(v)) 1

Fbg (I0 ) = F0 (X) · (− )|E(Γ)|

c,0

| Aut(Γ)| X=I0 F0′′ (X)|X=I0

Γ∈Gg,0 v∈V (Γ)

X Y |E(Γ)|

1 1

= Ival(v)−1 · ,

c,0

| Aut(Γ)| 1 − I1

Γ∈Gg,0 v∈V (Γ)

this is exactly the Feynman rule (92), i.e., we have proved the following

Theorem 6.1. The free energy of the topological 1D gravity can be realized by the

abstract free energy:

(102) Fg1D (I0 , I1 , I2 , · · · ) = Fbg (X)|X=I0

for g ≥ 2. Here F = λ−2 F0 and κ are given by (96) and (99) respectively.

Example 6.1. F21D and F31D are explicitly given by:

1

F21D = 12 + 18 + 18

,

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 29

1 1 1 1

F31D = 48 + 48 + 16 + 16

1

+ 12 + 81 1

+ 16

1

+ 12 + 81 + 81 1

+ 48

1 1 1

+ 24 + 16 + 16 + 81

,

or,

5 2 3 1

F21D = I κ + I3 κ2 ,

(103) 24 2 8

1 1 7 25 5

F31D = I5 κ + ( I32 + I2 I4 )κ4 + I22 I3 κ5 + I24 κ6 ,

3

48 12 48 48 16

1

where κ = 1−I1 .

relation for F 1D as follows. First we have

g−1

X

1

(104) ∂κ Fbg = ((∂X + κF0′′′ (X))∂X Fbg−1 + ∂X Fbr ∂X Fbg−r ),

2 r=1

2 log κ, next we restrict to X = I0 to get:

g−1

X

1

(105) ∂κ Fg1D = 1D

((dX + κI2 )dX Fg−1 + dX Fr1D dX Fg−r

1D

),

2 r=1

(106) dX Ik = Ik+1 , k ≥ 1, dX κ = κ2 · I2 .

This also provides a way to compute Fg1D recursively without listing all possible

Feynman graphs.

∂F41D

2 = (dX + κI2 )∂I0 F31D + 2dX F11D · dX F31D + (dX F21D )2

∂κ

1 5 25 21 113 2 33

= I7 κ3 + ( I2 I6 + I3 I5 + I42 )κ4 + ( I I5 + I33 + 7I2 I3 I4 )κ5

48 16 48 64 48 2 24

3115 2 2 1127 3 985 4 1105 6 8

+( I I + I I4 )κ6 + I I3 κ7 + I κ ,

144 2 3 96 2 24 2 64 2

30 ZHIYUAN WANG AND JIAN ZHOU

1 1 5 21 2 5

F41D = I7 κ4 + ( I2 I6 + I3 I5 + I )κ

384 32 96 640 4

113 2 11 7 445 2 2 161 3

(107) +( I I5 + I33 + I2 I3 I4 )κ6 + ( I I + I I4 )κ7

576 2 96 12 288 2 3 192 2

985 4 1105 6 9

+ I I3 κ8 + I κ .

384 2 1152 2

6.3. Example: The Airy function and quantum Airy curve. Our ﬁrst ex-

ample is the Airy function

Z

t3

(108) Z A (t) = C dη exp − λ−2 (−ηt + ) ,

3

3

where C is a suitable normalization constant. Let f (η) = −ηt + t3 , then the critical

1

points of f (η) is η = ±t 2 . Denote by c one of the critical points and η̃ = η − c,

then

1 2 1

(109) f (η) = −(η̃ + c)c2 + (η̃ + c)3 = − c3 + cη̃ 2 + η̃ 3 .

3 3 3

Therefore in this case

I0 = c, I1 = 1 − 2c, I2 = −2;

(110)

Ik = 0, k ≥ 3.

1 1

(111) κ= = .

1 − I1 2c

Using the coordinate change (88), this particular point (110) in the space of

coupling constants is

t0 = c 2 , t1 = 1, t2 = −2,

(112)

tk = 0, k ≥ 3.

∂

In what follows we will use ∂c to replace the operator dT since we have ∂I

∂c = Ik+1

k

∂κ 2

for k ≥ 1 and ∂c = κ I2 . Now using our formalism we can compute the free energies

FgA associated with Z A . The Feynman graphs of this integral only involves cubic

graphs without external edges, whose vertices are all of genus zero.

Example 6.3. The free energies F0A and F1A are given by

2 1 1

(113) F0A = − c3 , F1A = log .

3 2 2c

And F2A and F3A are given by

1

F2A = 12 + 18

,

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 31

1 1 1

F3A = 8 + 16 + 16

1 1

+ 24 + 48

,

or,

5 5

(114) F2A = , F3A = .

48c3 64c6

Using (111), for g ≥ 1 we have

∂FgA

∂κ FgA = (−2c2 ) · ,

∂c

thus Theorem 4.2 give us the following recursion relation.

Theorem 6.2. For g ≥ 2 we have

g−1

X ∂F A ∂Fg−r

∂FgA 1 ∂ A

1 ∂Fg−1 r

A

(115) =− 2 ( − )+ .

∂c 4c ∂c c ∂c r=1

∂c ∂c

1

Note that in this case FgA must be a monomial in κ = 2c of degree 3g − 3, thus

A

Fg can be solved recursively using this relation. For example,

1105 565

(116) F4A = , F5A = , ···

9216c9 2048c12

The recursion relation (115) is equivalent to the equation

∂ 2

(117) (~ ) − t Z A (t) = 0

∂t

if we set ~ = λ2 . This is the Schrödinger equation of the quantum Airy curve.

Now let us recall some results about the quantum Airy curve [16,26] and compare

with the above construction. First let u(z) = 21 z 2 , v(z) = z be a parametrization

of the Airy curve

1 2

A(u, v) = v − u = 0,

2

then using the Eynard-Orantin topological recursion [12] one may get a family of

diﬀerentials

Wg,n (p1 , · · · , pn ) = Wg,n (z1 , · · · , zn )dz1 · · · dzn .

These invariants associated to the Airy curve encode the information of the Witten-

Kontsevich tau-function [18, 24, 28]. Following [16], deﬁne

∞

X

(118) ZAiry = exp( ~n−1 Sn ),

n=0

32 ZHIYUAN WANG AND JIAN ZHOU

where

Z z

S0 (z) = v(z)du(z),

1 du

(119) S1 (z) = − log( ),

2 dz

X Z z Z z

(−1)k

Sn (z) = ··· Wg,k (z1′ , · · · , zk′ )dz1′ · · · dzk′ , n ≥ 2.

k!

2g−1+k=n

Then Sk can be used to construct the quantization of the Airy curve. The following

result was proposed by Gukov-Sulkowski in [16], and proved in [26].

Theorem 6.3. ( [26]) The function ZAiry satisfies

(120) ÂZAiry = 0,

where

1 2 1

(121) v̂ − û = (~∂u )2 − u·

Â =

2 2

is the quantization of the Airy curve.

One computes

1 3 1

S0 =

z , S1 = − log(z),

(122) 3 2

5 5 1105

S2 = , S3 = , S4 = , ···

24z 3 16z 6 1152z 9

We make a comparison between this and (114), (116), then we might simply expect

FgA = Sg for g ≥ 2 once we set z 3 = 2c3 . In fact, this can be proved by the following

recursion relation.

Lemma 6.1. ( [26]) For n > 2, we have

1 2 1 X

(123) ∂u Sn−1 + ∂u S0 · ∂u Sn + ∂u S1 · ∂u Sn−1 + ∂u Si · ∂u Sj = 0.

2 2 i+j=n

i,j≥2

This lemma can be obtained from Theorem 6.3 by expanding (120) directly as

a series in ~ and comparing the coeﬃcients. Using z 3 = 2c3 and u = 12 z 2 , this

recursion is equivalent to (115).

In conclusion, for the particular choices (110) of coupling constants, the pro-

cedure to produce FgA is equivalent to the solution of quantization of Airy curve

obtained by the Eynard-Orantin topological recursion.

example is the Kontsevich-Penner matrix model for N = 1 [18,21], which is related

to the quantization of the Catalan curve

1

x=z+ ,

z

see [16, 19]. The Schrödinger equation in this case is

∂2 ∂

(124) [~2 + ~t + (1 − ~)]Z(t) = 0,

∂t2 ∂t

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 33

Z

1

(125) Z (t) = dη exp − λ−2 ( η 2 + ηt + logη) ,

C

2

Now let us understand this integral using the model of topological 1D gravity. √

2

Write f (η) = 21 η 2 + ηt + log(η), we expand f (η) at its critical points c = −t± 2 t −4 .

Let η̃ = η − c, then

f (η) =f (η̃ + c)

(126) c2 1 η̃ 3 η̃ 4 η̃ 5

=(− − 1 + log(c)) + (1 − c−2 )η̃ 2 + 3 − 4 + 5 − · · ·

2 2 3c 4c 5c

By taking

1

(127) κ=

1 − c−2

and

(−1)k+1 · k!

(128) Ik = , k ≥ 1,

ck+1

we see Ik′ = Ik+1 and κ′ = κ2 · I2 where the prime means taking derivative with

respect to c. Thus we get a special case of the theory in Section 6.2, thus we obtain

a sequence of free energies FgC . Using the formula for FgC for g = 2, 3, 4 given by

(103) and (107), we get:

c2 1 1

F0C = − − 1 + log(c), F1C = log ,

2 2 1 − c−2

3 5

F2C = 2 2

+ ,

4(c − 1) 6(c − 1)3

2

5 10 25 5

F3C = + 2 + + 2 ,

2(c2 − 1)3 (c − 1)4 2(c2 − 1)5 (c − 1)6

105 507 6391 767

F4C = 2 4

+ 2 5

+ 2 6

+

8(c − 1) 5(c − 1) 24(c − 1) 2(c − 1)7

2

985 1105

+ + .

4(c2 − 1)8 18(c2 − 1)9

g−1

X ∂

1 ∂ ∂ C ∂ C

(129) ∂κ FgC = [( + κ · I2 ) Fg−1 + FrC · Fg−r ], g ≥ 2.

2 ∂c ∂c r=1

∂c ∂c

∂

To write this relation completely in terms of ∂c , we need the following lemma.

∂ C 2κ

(130) F = (κ2 · I2 − )∂κ FgC .

∂c g c

34 ZHIYUAN WANG AND JIAN ZHOU

Proof. Using the homogeneity we could see that every monomial in Fg1D is of the

form ag,i1 ···ik Ii1 · · · Iik · κl with i1 + · · · ik + k = 2l. Thus

Pk l l−1

∂

∂c (Ii1 · · · Iik κ )

l

j=1 Ii1 · · · Iij +1 · · · Iik κ + Ii1 · · · Iik lκ · κ2 I2

=

∂κ (Ii1 · · · Iik κl ) Ii1 · · · Iik lκl−1

k

1 X Iij +1

= · κ + κ2 · I2

l j=1 Ij

k

1 X (−1)(ij + 1)

= · κ + κ2 · I2

l j=1 c

2

= − · κ + κ2 · I2 ,

c

which proves the lemma.

2κ 2c3

κ2 · I2 − =− 2 ,

c (c − 1)2

we rewrite the recursion relation as follows.

Theorem 6.4. For the Kontsevich-Penner model, for g ≥ 2, we have

g−1

X ∂

∂ C c3 ∂ 2 ∂ C ∂ C

(131) Fg = − 2 − Fg−1 + FrC · Fg−r .

∂c (c − 1)2 ∂c c(c2 − 1) ∂c r=1

∂c ∂c

Or in terms of t = −(c + 1c ),

g−1

X ∂

1 ∂ ∂2 C ∂ C

(132) (c − ) · FgC = 2 Fg−1 + FrC · Fg−r .

c ∂t ∂t r=1

∂t ∂t

The second equation in the above theorem is exactly the same as the recursive

formula for the quantization of the Catalan curve [19, (4.13)]. Similar to the case

of quantum Airy curve, the free energies FgC here can also be constructed from the

Catalan curve using the Eynard-Orantin topological recursion.

this subsection we apply topological 1D gravity and our formalism to study the

problem of enumeration of stable graphs with only genus zero vertices. In this

model, all vertices of the Feynman graphs have valence ≥ 3. The partition function

is given by

Z X xn

st κ1/2 −2 x2

(133) Z (T, κ) = dx exp[λ (T − )].

(2πλ2 )1/2 n≥3

n! 2κ

X 2g−2

X

(134) F st (T, κ) = agd T d κg−1+d .

g≥2 d=1

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 35

By (83)

X X X

st λ2l−2k

(135) F (T, κ) = log · (2l − 1)!! · T k · κl · δl1 +···+lk ,2l .

k! · l1 ! · · · lk !

k≥0 li ≥3 l

The coeﬃcients agd counts the number of connected stable graphs of genus g with d

vertices of marked genus 0, weighted by the inverse of the orders of their automor-

phism groups. Such graphs has g − 1 + d edges. Using this interpretation one gets

the following explicit formulas for agd , d = 1, 2:

1 1

ag1 = = g ,

(2g)!! 2 g!

g+1

(136) 1X X 1

ag2 = .

2 k! · (2l)!! · (2g + 2 − 2k − 2l)!!

k=1 k+2l≥3

2g+2−k−2l≥3

22g − 2g−1 − (g + 1)2

(137) ag2 = .

2g · (g + 1)!

Since ag2g−2 counts trivalent graphs of genus g, it equals to bg−1 in [9, (65), (66)],

and so

X g X∞

g−1 (6m)! x m

(138) a2g−2 x = log .

m=0

(3m)!(2m)! 288

g≥2

It is interesting to compute agd for 3 ≤ d < 2g − 2. Formula (135) and the direct

counting of graphs are not very eﬀective for this purpose. We will apply the qua-

dratic recursion relations developed in Section 6.2 to derive some recursion relations

for agd .

In (105) and (106), we take Ik = T for all k ≥ 1, and then dX changes to an

operator D such that

DT = T, Dκ = κ2 T

then we get from (105) the following recursion relation:

g−1

X

1

(139) ∂κ Fgst st

= (D + κT )DFg−1 + DFrst · DFg−r

st

,

2 r=1

1

(140) F1st = log κ.

2

From this we recursively:

T κ2 5T 2 κ3

F2st = + ,

8 24

T κ3 11T 2 κ4 25T 3κ5 5T 4 κ6

F3st = + + + ,

48 48 48 16

T κ4 223T 2κ5 515T 3κ6 1373T 4κ7 985T 5κ8 1105T 6κ9

F4st = + + + + + .

384 1920 576 576 384 1152

36 ZHIYUAN WANG AND JIAN ZHOU

We have

X X

DFgst = agd dT d κg−1+d + agd (g − 1 + d)T d+1 κg+d ,

d d

X X

D 2

Fgst = agd d2 T d κg−1+d + agd (g − 1 + d)dT d+1 κg+d

d d

X X

+ agd (g − 1 + d)(d + 1)T d+1 g+d

κ + agd (g − 1 + d)(g + d)T d+2 κg+d+1 .

d d

X

2 agd (g − 1 + d)T d κg−2+d

d

g−1

X

st

= (D + κT )DFg−1 + DFrst DFg−r

st

r=1

X X

= ag−1

d d2 T d κg−2+d + ag−1

d (g − 2 + d)dT d+1 κg−1+d

d d

X

+ ag−1

d (g − 2 + d)(d + 1)T d+1 κg−1+d

d

X

+ ag−1

d (g − 2 + d)(g − 1 + d)T d+2 κg+d

d

X X

+2 ag−1

d dT d+1 κg−1+d + ag−1

d (g − 2 + d)T d+2 g+d

κ

d d

XX

g−2 X

+ ard dT d κr−1+d + ard (r − 1 + d)T d+1 r+d

κ ·

r=2 d d

X X

ag−r

d dT d κg−r−1+d + ag−r

d (g − r − 1 + d)T d+1 κg−r+d

d d

for g ≥ 3. Therefore the following quadratic recursion relations hold for agd :

X X g−1

1

agd = ag−1

d d2 + ad−2 (g − 4 + d)(g − 1 + d)

2(g − 1 + d)

d d

X g−1

+ ad−1 (g − 3 + d)(2d − 1) + 2(d − 1)

d

g−2

X X X

+ ard1 ag−r

d2 d1 d2 + ard1 ag−r

d2 (r − 1 + d1 )d2

r=2 d1 +d2 =d d1 +d2 =d−1

X

+ ard1 ag−r

d2 d1 (g − r − 1 + d2 )

d1 +d2 =d−1

X

+ ard1 ag−r

d2 (r − 1 + d1 )(g − r − 1 + d2 ) .

d1 +d2 =d−2

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 37

example: the model describing enumeration of graphs, not necessarily stable, in-

troduced and studied in [9]. The partition function for this case is

Z

1 η2

(141) Z(t) = √ dη exp λ−2 (t · eη − ) .

2πλ2 2

η2

The critical point η = T of the function f (η) = t · eη − 2 is given by

T

(142) T =t·e ,

whose solution may be explicitly given by the Lambert series [9]:

∞

X dd−1

(143) T = td .

d!

d=1

∞

T2 teT − 1 2 X teT n

f (η) = (teT − )+ x + x

2 2 n=3

n!

∞

T2 T −1 2 X T n

= (T − )+ x + x .

2 2 n=3

n!

Therefore in this case, the propagator is

1

(144) ,

κ=

1−T

and the point on the space of coupling constants is speciﬁed by

(145) I0 = T ; Ik = T, k ≥ 1,

or in the coordinates {tk },

(146) tk = t, k ≥ 0.

By (92), the free energy is

X 1 T |V (Γ)|

(147) Fggr = · ,

c,0

| Aut(Γ)| (1 − T )|E(Γ)|

Γ∈Gg,0

For example,

T2 1 1

F0gr = T− , F1gr = log ,

2 2 1−T

T 5T 2

F2gr = + ,

8(1 − T )2 24(1 − T )3

T 11T 2 25T 3 5T 4

F3gr = + + + ,

48(1 − T )3 48(1 − T )4 48(1 − T )5 16(1 − T )6

T 223T 2 515T 3 1373T 4

F4gr = 4

+ 5

+ 6

+

384(1 − T ) 1920(1 − T ) 576(1 − T ) 576(1 − T )7

985T 5 1105T 6

+ 8

+ .

384(1 − T ) 1152(1 − T )9

These match with Gk in [9, (63)].

38 ZHIYUAN WANG AND JIAN ZHOU

1

(148) Fggr = Fgst (T, ).

1−T

Therefore, for g ≥ 2 the free energy Fggr can be written in the form

2g−2

X agd · T d

Fggr = .

(1 − T )d+g−1

d=1

fg (T )

(149) Fggr = ,

(1 − T )3g−3

where fg (T ) is a polynomial of degree ≤ 2g −2 without constant term. For example,

T

F2gr = (3 + 2T ),

24(1 − T )3

T

F3gr = (1 + 8T + 6T 2 ),

48(1 − T )6

T

F4gr = (15 + 594T + 2624T 2 + 2144T 3 + 164T 4 − 16T 5),

5760(1 − T )9

T

F5gr = (3 + 465T + 6730T 2 + 21940T 3 + 18940T 4

11520(1 − T )12

+ 3012T 5 − 240T 6),

T

F6gr = (63 + 35568T + 1349298T 2 + 11582816T 3 + 31178616T 4

2903040(1 − T )15

+27897072T 5 + 6526912T 6 − 266448T 7 − 36576T 8 + 2304T 9).

Let

2g−2

X

fg (T ) = bgn · T n ,

n=1

and {bgk } are related by

k

X

2g − 2 − l g

(150) bgk = (−1)k+l al .

k−l

l=1

In particular,

1

bg1 =ag1 = ,

(2g)!!

(151) bg2 =ag2 − (2g − 3)ag1

22g − 2g−1 − (g + 1)2 2g − 3

= − .

2g · (g + 1)! (2g)!!

P3g−3

As pointed out in [9, (64), (65)], Fggr is of the form Fggr = n=g−1 λg,n κn , and

Bg

λg,g−1 = , λg,3g−3 = bg−1 ,

g(g − 1)

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 39

X∞ X∞

r (6m)! x m

br x = log .

n=1 m=0

(3m)!(2m)! 288

The numbers {bg } are exactly the coeﬃcients appearing in the expansion of the

Airy function (122). Here are some examples of Fggr written as polynomials in κ:

5 3 7 1

F2gr = κ − κ2 + κ,

24 24 12

5 6 35 5 13 4 1 3

F3gr = κ − κ + κ − κ ,

16 48 24 8

1105 9 1225 8 2273 7 313 6 227 5 17 4 1 3

F4gr = κ − κ + κ − κ + κ − κ − κ ,

1152 384 576 144 480 1440 360

565 12 14665 11 76367 10 11191 9 2557 8 7993 7

F5gr = κ − κ + κ − κ + κ − κ

128 768 2304 384 192 2880

37 6 1

+ κ + κ5 ,

320 48

82825 15 441245 14 493235 13 16116187 12 2827135 11

F6gr = κ − κ + κ − κ + κ

3072 3072 1536 41472 10368

1884983 10 567289 9 7489 8 10249 7 47 6 1 5

− κ + κ − κ − κ + κ + κ ,

17280 25920 6480 60480 10080 1260

19675 18 7969325 17 65405005 16 453853985 15

F7gr = κ − κ + κ − κ

96 6144 18432 82944

215237149 14 64035527 13 23126555 12 11204309 11

+ κ − κ + κ − κ

41472 20736 20736 51840

1352989 10 6481 9 1927 8 1

+ κ + κ − κ − κ7 .

103680 4032 20160 72

1

Plug T = 1 − κ into (135), we get the generating series for {λg,k }:

∞

X 3g−3

X

λ2g−2 λg,d κd

g=2 d=g−1

(152) X X X

λ2l−2k k l−k

= log · (2l − 1)!! · (κ − 1) κ · δl1 +···+lk −2l,0 .

k! · l1 ! · · · lk !

k≥0 li ≥3 l

The relations among {agk }, {bgk } and {λg,k } are given by (150) and

X l g

2g−2

k

λg,3g−3−k = (−1) b ,

k l

l=k

X 3g − 3 − j

3g−3

g n

bn = (−1) λg,j ,

j=g−1

n

k

X

k+l 2g − 2 − l

λg,3g−3−k = (−1) a2g−2−l .

k−l

l=0

in Section 6.2 to this problem. For this purpose, we need to ﬁnd an operator that

∂ ∂

satisﬁes the conditions in (106). We can take dX to be D = ∂(log T ) = T ∂T , because

40 ZHIYUAN WANG AND JIAN ZHOU

we have

1

Ik = T, κ= ,

1−T

the conditions in (106) are satisﬁed by D. So we get the following equation from

(105):

g−1

X

1 gr gr

(153) ∂κ F gr = ((D + κT )DFg−1 + DFrgr DFg−r ),

2 r=1

(154) (κ + κ2 T )∂κ Fggr − DFggr = (g − 1)Fggr .

Proof. Every monomial in Fggr for g ≥ 2 is of the form ag,k T k κk+g−1 . Then the

lemma follows from

κ∂κ (T k κk+g−1 ) = (k + g − 1)T k κk+g−1

and

D(T k κk+g−1 ) = kT k κk+g−1 + (k + g − 1)T k κk+g−2 · κ2 T.

1

Since κ + κ2 T = (1−T )2 , (153) and Lemma 6.3 give us:

g−1

X

1 T gr gr

(155) (D + g − 1)Fggr = [(D + )DFg−1 + (DFrgr )(DFg−r )].

2(1 − T )2 1−T r=1

This theorem provides a way to compute Fggr recursively. In fact, the left-hand

side of (155) can be rewritten as

(D + g − 1)Fg1D =T Fg′ (T ) + (g − 1)Fg (T )

(156) d

=T 2−g (T g−1 · Fg (T )),

dT

thus Fg1D is determined by the formula

(157)

Z T g−1

X

1D 1−g T g−2 T 1D 1D 1D

Fg (T ) = T · 2

(D + )DFg−1 + (DFr )(DFg−r ) dT.

0 2(1 − T ) 1−T r=1

κ = (1 − T )−1 using T = 1 − κ1 , the recursion relation (155) can be written as

follows.

Theorem 6.6. For g ≥ 2, we have

gr gr

2 d gr κ2 d2 Fg−1 dFg−1

(κ − κ) + g − 1 Fg = (κ2 − κ)2 + (3κ − 2)(κ 2

− κ)

dκ 2 dκ2 dκ

(158) g−1 gr

X dF gr dFg−r

r

+ (κ2 − κ)2 · .

r=1

dκ dκ

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 41

In particular, for g ≥ 3,

gr gr

2 d κ2 d2 Fg−1 dFg−1

(κ − κ) + g − 1 Fggr = (κ2 − κ)2 + (4κ − 3)(κ 2

− κ)

dκ 2 dκ2 dκ

(159) g−2 gr

X dF gr dFg−r

r

+ (κ2 − κ)2 · .

r=2

dκ dκ

d g−1 g

F gr

(1 − κ)

dκ κg−1

1 1 g−2 gr 2

dFggr

=− 2 −1 (g − 1)Fg + (κ − κ)

κ κ dκ

(160) g−2

2 gr gr

1 (1 − κ) 2 2

d F g−1 2

dFg−1

=− (κ − κ) + (4κ − 3)(κ − κ)

2 κg−2 dκ2 dκ

g−2

X dF gr dFg−r gr

r

+ (κ2 − κ)2 · .

r=2

dκ dκ

Now if we want to integrate this equality with respect to κ, we need to determine

Fggr

the constant of integration. Note that (1 − κ)g−1 κg−1 is a polynomial in κ whose

Bg

constant term is λg,g−1 = g(g−1) , therefore

Z κ gr

gr κg−1 Bg 1 (1 − κ)g−2 2 2

d2 Fg−1

Fg = + − (κ − κ) +

(1 − κ)g−1 g(g − 1) 0 2 κg−2 dκ2

(161) gr g−2 gr

dFg−1 X dFrgr dFg−r

2 2 2

(4κ − 3)(κ − κ) + (κ − κ) · dκ .

dκ r=2

dκ dκ

Now let us derive some recursion relations for the coeﬃcients {λg,k } using The-

orem 6.6. Expand the two sides of (159), we get

X

l(κ − 1) + (g − 1) λg,l κl

l

κ2 X X

= (κ − 1)2 l(l − 1)λg−1,l κl + (4κ − 3)(κ − 1)lλg−1,l κl

2

l l

g−2

XX

2 l1 +l2

+ (κ − 1) l1 l2 λr,l1 λg−r,l2 κ ,

r=2 l1 ,l2

which gives us

(l − 1)λg,l−1 + (g − l − 1)λg,l

1 1 1

= (l − 4)(l − 1)λg−1,l−4 − (l − 3)(2l − 1)λg−1,l−3 + (l − 2)lλg−1,l−2

2 2 2

g−2

X X X

(162) 1

+ l1 l2 λr,l1 λg−r,l2 − 2 l1 l2 λr,l1 λg−r,l2

2 r=2

l1 +l2 =l−4 l1 +l2 =l−3

X

+ l1 l2 λr,l1 λg−r,l2 .

l1 +l2 =l−2

42 ZHIYUAN WANG AND JIAN ZHOU

determines {λg,n } uniquely. In fact, ﬁrst we set l = 3g − 2, then (162) gives

1 (3g − 5)(6g − 5)

λg,3g−3 = (3g − 6)λg−1,3g−6 − λg−1,3g−5

2 2(3g − 3)

(3g − 4)(3g − 2) 1 X X

g−2

+ λg−1,3g−4 + l1 l2 λr,l1 λg−r,l2

2(3g − 3) 2(3g − 3) r=2

l1 +l2 =3g−6

X X

− 2 l1 l2 λr,l1 λg−r,l2 + l1 l2 λr,l1 λg−r,l2 .

l1 +l2 =3g−5 l1 +l2 =3g−4

(3g − 4)λg,3g−4 + (2 − 2g)λg,3g−3

1 1

= (3g − 7)(3g − 4)λg−1,3g−7 − (3g − 6)(6g − 7)λg−1,3g−6

2 2

g−2

1 1X X

+ (3g − 5)lλg−1,3g−5 + l1 l2 λr,l1 λg−r,l2

2 2 r=2

l1 +l2 =3g−7

X X

− 2 l1 l2 λr,l1 λg−r,l2 + l1 l2 λr,l1 λg−r,l2 ,

l1 +l2 =3g−6 l1 +l2 =3g−5

from which we can solve λg,3g−4 using λg,3g−3 and lower genus λr,n . Similarly λg,n

can be solved recursively from n = 3g − 3 to n = g − 1 using (162).

11661131005.

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 43

In this appendix, we give more examples of the abstract free energies.

For the case N = 1, here are the explicit expressions for Fb2,2 and Fb3 in terms of

Feynman graphs.

Fb2,2 = 1

2 2 + 14 1 + 12 1 1 + 12 2 0 + 12 1 1 1

+ 16 1

+ 41 1 0 + 21 0 1 + 21 0 1 + 41 0 1 + 41 1 0

+ 14 0 1 + 12 1 1 0 + 1 0 1 + 14 1 0 1

1 + 12 0 0

+ 18 0 0 + 18 0 0 + 18 0 0 + 18 0 0 + 18 0 0

0

+ 14 0 0 + 14 0 1 0 + 12 1 + 12 1 0 0

0

+ 12 0 0 1 + 14 1 0 0 + 12 0 0 1 + 14 1 0 0

1

+ 14 0 0 1 + 14 1 0 0 + 12 1 0 0 + 14 0 0

1

+ 12 1 0 0 1 + 18 0 0 0 + 18 0 0 0 + 14 0 0 0

+ 18 0 0 0 1

+ 16 0 0 0 + 14 0 0 0 1

+ 12 0 0 0

0 0

+ 14 0 + 14 0 0 0 + 18 0 + 18 0 0 0

0 0

0 0

+ 12 0 + 12 0 0 0 1 + 14 1 0 0 0 + 12 1 0

0 0

1 0 0

+ 21 1 0 0 0 + 41 0 0 1

+ 16 0 0 + 41 0 0

0 0 0

0 0 0 0

+ 18 0 0 0 0 + 14 + 18 0 0 + 14 0 0

0 0 0 0

+ 14 0 0 0 0 + 81 0 0 0 0 + 18 0 0 0 0

.

44 ZHIYUAN WANG AND JIAN ZHOU

1

Fb3 = 3 + 2 2 + 2 1 + 81 1 + 41 1 1 + 21 2 0 + 21 1 1

+ 12 1 1 1

1 + 48 0 + 14 1 0 + 16 1 0 + 14 1 0

1

+ 18 1 0 + 12 1 1 0 + 12 1 0 1 + 14 1 0 1 + 16 1 0

1

1 1 1 1

+ 48 0 0 + 16 0 0 + 16 0 0 + 12 0 0 + 81 0 1 0

0

+ 14 1 + 41 1 0 0 + 14 1 0 0 + 14 1 0 0 + 16 1 0 0

0

1

+ 14 0 0 + 14 1 0 0 1 + 18 0 0 0 1

+ 16 0 0 0

1

1

0 0

+ 12 0 0 0 + 18 0 + 18 0 + 14 1 0 0 0

0 0

0 0 0 0

+ 14 1 0 + 18 1 0 1

+ 48 0 0 1

+ 24 0 0

0 0 0 0

1

0 0 1

0

+ 16 + 16 0 0 0 0 + 18 0 0

0 0 0.

And in terms of the derivatives of the holomorphic free energy F (t) together

with the propagator κ, the explicit expressions for Fb3 and Fb4 are as follows.

1

Fb3 = F3 + ( F2′′ + F1′ F2′ )κ

2

1 (4) 1 1 1 1

+[ F1 + (F1′′ )2 + F0′′′ F2′ + F1′ F1′′′ + (F1′ )2 F1′′ ]κ2

8 4 2 2 2

1 (6) 1 (4) 5 1 (5)

+[ F0 + F0 F1′′ + F0′′′ F1′′′ + F0 F1′ + F0′′′ F1′ F1′′

48 4 12 8

1 (4) ′ 2 1 ′′′ ′ 3 3

+ F0 (F1 ) + F0 (F1 ) ]κ

4 6

1 (4) 2 7 ′′′ (5) 5 ′′′ 2 ′′ 2 ′′′ (4) ′ 1 ′′′ 2 ′ 2 4

+[ (F0 ) + F0 F0 + (F0 ) F1 + F0 F0 F1 + (F0 ) (F1 ) ]κ

12 48 8 3 2

25 ′′′ 2 (4) 5 ′′′ 3 ′ 5 5 ′′′ 4 6

+[ (F0 ) F0 + (F0 ) F1 ]κ + (F0 ) κ .

48 8 16

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 45

1 1

Fb4 = F4 + [ (F2′ )2 + F1′ F3′ + F3′′ ]κ

2 2

1 1 1 1 1

+[F1′ F1′′ F2′ + (F1′ )2 F2′′ + F1′′ F2′′ + F0′′′ F3′ + F1′′′ F2′ + F1′ F2′′′

2 2 2 2 2

1 (4)

+ F2 ]κ2

8

1 1 1

+[ (F1′ )2 (F1′′ )2 + (F1′′ )3 + F0′′′ (F1′ )2 F2′ + F0′′′ F1′′ F2′ + F0′′′ F1′ F2′′

2 6 2

1 ′ 3 ′′′ 5 5 1 (4)

+ (F1 ) F1 + F1 F1 F1 + (F1′′′ )2 + F0′′′ F2′′′ + F0 F1′ F2′

′ ′′ ′′′

6 24 12 2

1 (4) ′′ 1 ′ 2 (4) 1 ′′ (4) 1 (5) ′ 1 (5) ′ 1 (6)

+ F0 F2 + (F1 ) F1 + F1 F1 + F0 F2 + F1 F1 + F1 ]κ3

4 4 4 8 8 48

1 3 5

+[ F0′′′ (F1′ )3 F1′′ + F0′′′ F1′ (F1′′ )2 + (F0′′′ )2 F1′ F2′ + (F0′′′ )2 F2′′ + F0′′′ (F1′ )2 F1′′′

2 2 8

5 1 (4) 3 (4) 3 (4)

+ F0′′′ F1′′ F1′′′ + F0 (F1′ )4 + F0 (F1′ )2 F1′′ + F0 (F1′′ )2

4 24 4 8

2 ′′′ (4) ′ 2 (4) ′ ′′′ 2 ′′′ ′ (4) 1 (4) (4) 1 (5)

+ F0 F0 F2 + F0 F1 F1 + F0 F1 F1 + F0 F1 + F0 (F1′ )3

3 3 3 6 12

3 (5) 7 (5) 7 (5) 1 (6) 1 (6)

+ F0 F1′ F1′′ + F0 F1′′′ + F0′′′ F1 + F0 (F1′ )2 + F0 F1′′

8 48 48 16 16

1 (7) 1 (8) 4

+ F0 F1′ + F ]κ

48 384 0

1 5 5

+[ (F0′′′ )2 (F1 )4 + 2(F0′′′ )2 (F1′ )2 F1′′ + (F0′′′ )2 (F1′′ )2 + (F0′′′ )3 F2′

8 4 8

15 ′′′ 2 ′ ′′′ 7 ′′′ (4) ′ 3 8 ′′′ (4) ′ ′′ 25 ′′′ (4) ′′′

+ (F0 ) F1 F1 + F0 F0 (F1 ) + F0 F0 F1 F1 + F0 F0 F1

8 12 3 24

1 (4) 2 ′ 2 1 (4) 2 ′′ 25 ′′′ 2 (4) 25 ′′′ (5) ′ 2

+ (F0 ) (F1 ) + (F0 ) F1 + (F0 ) F1 + F0 F0 (F1 )

3 3 48 48

7 ′′′ (5) ′′ 5 (4) (5) ′ 21 (5) 2 5 ′′′ (6) ′

+ F0 F0 F1 + F0 F0 F1 + (F ) + F0 F0 F1

12 16 640 0 24

5 (4) (6) 1 ′′′ (7) 5

+ F0 F0 + F0 F0 ]κ

96 32

2 ′′′ 3 ′ 3 25 ′′′ 3 ′ ′′ 5 ′′′ 3 ′′′ 109 ′′′ 2 (4) ′ 2

+[ (F0 ) (F1 ) + (F0 ) F1 F1 + (F0 ) F1 + (F ) F0 (F1 )

3 8 4 48 0

125 ′′′ 2 (4) ′′ 11 ′′′ (4) 2 ′ 11 (4) 3 53 ′′′ 2 (5) ′

+ (F ) F0 F1 + F0 (F0 ) F1 + (F0 ) + (F0 ) F0 F1

48 0 8 96 48

7 ′′′ (4) (5) 113 ′′′ 2 (6) 6

+ F0 F0 F0 + (F ) F0 ]κ

12 576 0

25 15 185 ′′′ 3 (4) ′ 445 ′′′ 2 (4) 2

+[ (F0′′′ )4 (F1′ )2 + (F0′′′ )4 F1′′ + (F ) F0 F1 + (F ) (F0 )

16 8 48 0 288 0

161 ′′′ 3 (5) 7

+ (F ) F0 ]κ

192 0

15 985 ′′′ 4 (4) 8 1105 ′′′ 6 9

+[ (F0′′′ )5 F1′ + (F ) F0 ]κ + (F ) κ .

8 384 0 1152 0

(k+1)

In particular, by setting Fg = 0 for all g ≥ 1 and F0 = Ik , we recover the

expressions (103) and (107) for F31D and F41D .

46 ZHIYUAN WANG AND JIAN ZHOU

References

[1] Aganagic M, Bouchard V, Klemm A. Topological Strings and (Almost) Modular Forms. Com-

munications in Mathematical Physics, 2008, 277(3):771-819.

[2] Bershadsky M, Cecotti S, Ooguri H, Vafa C. Holomorphic anomalies in topological ﬁeld theo-

ries. arXiv preprint hep-th/9302103, 1993.

[3] Bershadsky M, Cecotti S, Ooguri H, Vafa C. Kodaira-Spencer theory of gravity and exact

results for quantum string amplitudes. Communications in Mathematical Physics, 1994, 165(2):

311-427.

[4] Candelas P, de la Ossa X C, Green P S, Parkes L. A pair of Calabi-Yau manifolds as an exactly

soluble superconformal theory. Nuclear Physics B, 1991, 359(1): 21-74.

[5] Cecotti S, Vafa C. Topologicalanti-topological fusion. Nuclear Physics B, 1991, 367(2):359-461.

[6] Deligne P, Mumford D. The irreducibility of the space of curves of given genus. Publications

Mathmatiques de l’Institut des Hautes tudes Scientiﬁques, 1969, 36(1): 75-109.

[7] Dijkgraaf R, Verlinde H, Verlinde E. Topological strings in d¡ 1. Nuclear Physics B, 1991,

352(1): 59-86.

[8] Dubrovin B. Geometry and integrability of topological-antitopological fusion. Communications

in Mathematical Physics, 1993, 152(3):539-564.

[9] Dubrovin B, Yang D, Zagier D. Classical Hurwitz numbers and related combinatorics. Moscow

Mathematical Journal, 2017, 17(4): 601-633.

[10] Eguchi T, Hori K, Xiong C S. Quantum cohomology and Virasoro algebra. arXiv preprint

hep-th/9703086, 1997.

[11] Eynard B. Large N expansion of convergent matrix integrals, holomorphic anomalies, and

background independence. Journal of High Energy Physics, 2009, 2009(03): 003.

[12] Eynard B, Orantin N. Invariants of algebraic curves and topological expansion. Communica-

tions in Number Theory & Physics, 2007, 1(2):347-452.

[13] Eynard B, Mariño M, Orantin N. Holomorphic anomaly and matrix models. Journal of High

Energy Physics, 2007, 2007(06): 058.

[14] Getzler E. The Virasoro conjecture for Gromov-Witten invariants. Contemporary Mathemat-

ics, 1999, 241: 147-176.

[15] Grimm T W, Klemm A, Mariño M, Weiss M. Direct integration of the topological string.

Journal of High Energy Physics, 2007, 2007(08): 058.

[16] Gukov S, Sulkowski P. A-polynomial, B-model, and quantization. Journal of High Energy

Physics, 2012, 2012(2):70.

[17] Knudsen F F. The projectivity of the moduli space of stable curves, II: The stacks Mg,n .

Mathematica Scandinavica, 1983, 52(2): 161-199.

[18] Kontsevich M. Intersection theory on the moduli space of curves and the matrix Airy function.

Communications in Mathematical Physics, 1992, 147(1): 1-23.

[19] Mulase M, Sulkowski P. Spectral curves and the Schroedinger equations for the Eynard-

Orantin recursion. arXiv preprint arXiv:1210.3006, 2012.

[20] Nishigaki S, Yoneya T. A nonperturbative theory of randomly branching chains. Nuclear

Physics B, 1991, 348(3): 787-807.

[21] Penner R C. Perturbative series and the moduli space of Riemann surfaces. Journal of Dif-

ferential Geometry, 1988, 27(1): 35-53.

[22] Wang Z., Zhou J., in preparation.

[23] Witten E. Quantum Background Independence In String Theory. arXiv preprint

arXiv:hep-th/9306122.

[24] Witten E. Two-dimensional gravity and intersection theory on moduli space. Surveys in

diﬀerential geometry, 1990, 1(1): 243-310.

[25] Yamaguchi S, Yau S T. Topological string partition functions as polynomials. Journal of High

Energy Physics, 2004, 2004(07): 047.

[26] Zhou J. Intersection numbers on Deligne-Mumford moduli spaces and quantum Airy curve.

arXiv preprint arXiv:1206.5896, 2012.

[27] Zhou J. On topological 1D gravity. I. arXiv preprint arXiv:1412.1604, 2014.

[28] Zhou J. Topological Recursions of Eynard-Orantin Type for Intersection Numbers on Moduli

Spaces of Curves. Letters in Mathematical Physics, 2013, 103(11):1191-1206.

HOLOMORPHIC ANOMALY EQUATIONS AND QUANTUM SPECTRAL CURVES 47

E-mail address: zhiyuan-14@mails.tsinghua.edu.cn

E-mail address: jzhou@math.tsinghua.edu.cn

- Buses For Anonymous Message DeliveryЗагружено:Griffin Boyce
- fem-ex21Загружено:Vikram Singh
- Peter Hamburger and Attila Sali- 11 -Doilies with Vertex Sets Of Sizes 275,286,...,462Загружено:Sprite090
- 0803.3435v1Загружено:iammegatron
- Metrics of Software Architecture Changes Based on Structural DistanceЗагружено:shivnair
- VCE Further Maths 1 & 2 NotesЗагружено:kwmaster
- Lecture 37Загружено:arunvinodh
- GerardЗагружено:Edwin Leena
- tmpE147Загружено:Frontiers
- D1 Course Companipon and Revision GuideЗагружено:Nicole Young
- Fibonacci Identities and Graph ColoringЗагружено:humejias
- documentclassЗагружено:api-240547275
- hw4Загружено:ashish3586
- EM AlgoЗагружено:Asad Lodhi
- Super Fibonacci Graceful LabelingЗагружено:RyanElias
- Detecting Connection Traps in Complex Relational DatabasesЗагружено:niranga81
- Business MathematicsЗагружено:amityscribd
- diff 510 lesson 1 draw and label a picture graph to represent dataЗагружено:api-269266578
- SAE Toward Efficient Cloud Data AnalysisЗагружено:Reddy Sumanth
- Science and Religion1Загружено:Ibrahim Isaac
- Ucinet GuideЗагружено:2anandsasi
- graphs_2Загружено:bijender
- Group {1, −1, i, −i} Cordial Labeling of Product Related GraphsЗагружено:International Journal of Advanced Scientific Research and Development
- BST & GraphsЗагружено:Varun Agarwal
- Generalized Interval Valued Neutrosophic Graphs of First TypeЗагружено:Anonymous 0U9j6BLllB
- RMM2018 Day2 EnglishЗагружено:G
- 44_ucsЗагружено:HALWA
- Are Field Quanta Real Objects Some Remarks on the Ontology of Quantum Fiel TheoryЗагружено:fisfil4681
- 1745806749maths Class Ix Periodic Test II Exam Sample Paper 02Загружено:Adt
- MSci Handbook 11Загружено:Roy Vesey

- 1809.10144.pdfЗагружено:Fábio Duarte
- On subfields of the second generalization of the GK maximal function fieldЗагружено:Fábio Duarte
- Duality and normalization, variations on a theme of Serre and ReidЗагружено:Fábio Duarte
- Sign-changing blowing-up solutions for the critical nonlinear heat equationЗагружено:Fábio Duarte
- X-UDS: The Chandra Legacy Survey of the UKIDSS Ultra Deep Survey FieldЗагружено:Fábio Duarte
- Riemann problems and dispersive shocks in self-focusing mediaЗагружено:Fábio Duarte
- On the gaps of the spectrum of volumes of tradesЗагружено:Fábio Duarte
- Using convolutional neural networks to predict galaxy metallicity from three-color imagesЗагружено:Fábio Duarte
- Dark Matter Searches at CollidersЗагружено:Fábio Duarte
- Bayesian Multi-Messenger Search Method for Common Sources of Gravitational Waves and High-Energy NeutrinosЗагружено:Fábio Duarte
- Origins of molecular clouds in early-type galaxiesЗагружено:Fábio Duarte
- Photoevaporation of Molecular Clouds in Regions of Massive Star Formation as Revealed Through H2 and Brγ EmissionЗагружено:Fábio Duarte
- Numerical Models for the Diffuse Ionized Gas in Galaxies. II. Three-dimensional radiative transfer in inhomogeneous interstellar structures as a tool for analyzing the diffuse ionized gasЗагружено:Fábio Duarte
- Deep multi-survey classification of variable starsЗагружено:Fábio Duarte
- Periodic Co/Nb pseudo spin-valve for cryogenic memoryЗагружено:Fábio Duarte
- 1809.10160.pdfЗагружено:Fábio Duarte
- Wide-Field Optical Spectroscopy of Abell 133: A Search for Filaments Reported in X-ray ObservationsЗагружено:Fábio Duarte
- The optical morphologies of galaxies in the IllustrisTNG simulation: a comparison to Pan-STARRS observationsЗагружено:Fábio Duarte
- The Large Early Galaxy Astrophysics Census (LEGA-C) Data Release II: dynamical and stellar population properties of z ~< 1 galaxies in the COSMOS fieldЗагружено:Fábio Duarte
- The LISA interferometer: impact of stray light on the phase of the heterodyne signalЗагружено:Fábio Duarte
- Observability of molecular species in a nitrogen dominated atmosphere for 55 Cancri eЗагружено:Fábio Duarte
- Enhanced Black-Hole Mergers in Binary-Binary InteractionsЗагружено:Fábio Duarte
- The MUSCLES Treasury Survey. V. FUV Flares on Active and Inactive M DwarfsЗагружено:Fábio Duarte
- Weak radius of the protonЗагружено:Fábio Duarte
- Multi-Conjugate Adaptive Optics Simulator for the Thirty Meter Telescope: Design, Implementation, and ResultsЗагружено:Fábio Duarte
- Opacity, variability and kinematics of AGN jetsЗагружено:Fábio Duarte
- Unistructurality of cluster algebrasЗагружено:Fábio Duarte
- Absence of unidirectionally propagating surface plasmon-polaritons in nonreciprocal plasmonicsЗагружено:Fábio Duarte
- Neutron Star Mergers Are the Dominant Source of the r-process in the Early Evolution of Dwarf GalaxiesЗагружено:Fábio Duarte

- Geometric Constructions Foldable IЗагружено:Wendy Millheiser Menard
- CramersRuleЗагружено:Lanre Jonathan Ige
- Seismic Analysis of Multi Storied RC Buildings- Part 4Загружено:Jaga Nath
- sols4Загружено:jiamin_tay5573
- 025Загружено:Fahad Almafakir
- Random Walk Dif 1Загружено:Yashanshu Gautam
- Homework 1Загружено:professorbell
- Leaf SpringЗагружено:Tiago Camargo Alves
- A DOCTOR.docxЗагружено:Anonymous sMAUpreg0
- 2 Approximations and Rounding ErrorsЗагружено:İsmail Eren
- CBSE 2014 – 2015 Class 09 SA1 Question Paper – MathsЗагружено:Raynaah
- SVD ChapterЗагружено:jdmena227
- Given a WeibullЗагружено:RagamadhuriNallam
- Obstacle Avoidance Methods for a Laser Based Intelligent WheelchairЗагружено:pvdai
- Vectors Summary Sheet122211Загружено:pfredaac
- IGCSE Mathematics 4400 May 2004 Question Paper and Mark Scheme Paper 3H N20710Загружено:Varun Panicker
- disclosure secondary math 1Загружено:api-278394494
- GRE - Graduate Assessment TestЗагружено:Hasnain1991
- Mechanical Translational Systems QuestionsЗагружено:djun033
- [8]+ easier..Загружено:Kelvin Fook
- Nicoleta Bostan Math IA - Origami Geometric ProofsЗагружено:Nicole Bostan
- 76-84Загружено:Carlos Moran Cepeda
- Ok Network 5Загружено:ttungl
- Advanced Gd & tЗагружено:pavan_03
- TIMSS 1999 Math HighlightsЗагружено:Marco Antonio Alarcón
- Dynamics Response Spectrum Analysis - Shear Plane FrameЗагружено:amrsaleh999
- Class Schedule Summer 2016Загружено:visayasstateu
- Multivariate Statistical Control ThesisЗагружено:Troshin Sergey
- CricketsЗагружено:bignelie
- ch8Загружено:Sajid Rasool

## Гораздо больше, чем просто документы.

Откройте для себя все, что может предложить Scribd, включая книги и аудиокниги от крупных издательств.

Отменить можно в любой момент.