arXiv:1411.3780v1 [math.

AG] 14 Nov 2014

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

WITH APPLICATIONS TO LANDAU-GINZBURG MIRROR SYMMETRY
AMANDA FRANCIS

Abstract. The Landau-Ginzburg A-model, given by FJRW theory, defines a cohomological field theory, but in most examples is very difficult to compute, especially when the
symmetry group is not maximal. We give some methods for finding the A-model structure.
In many cases our methods completely determine the previously unknown A-model Frobenius manifold structure. In the case where these Frobenius manifolds are semisimple, this
can be shown to determine the structure of the higher genus potential as well. We compute
the Frobenius manifold structure for 27 of the previously unknown unimodal and bimodal
singularities and corresponding groups, including 13 cases using a non-maximal symmetry
group.

Contents
1. Introduction
2. Background
3. Computational Methods
4. Summary of Computations
References

1
4
7
15
30

1. Introduction
Mirror symmetry is a phenomenon from physics that has inspired a lot of interesting
mathematics. In the Landau-Ginzburg setting, we have two constructions, the A- and Bmodels, each of which depends on a choice of a polynomial with a group of symmetries.
Both models yield Frobenius manifolds.
The A-model arising from FJRW theory produces a full cohomological field theory.
From the cohomological field theory we can construct correlators and assemble these into
a potential function. This potential function completely determines the Frobenius manifold
and determines much of the structure of the cohomological field theory as well. Although
some of these correlators have been computed in special cases, in many cases their computation is quite difficult, especially in the case that the group of symmetries is not maximal
(mirror symmetry predicts that these should correspond to an orbifolded B-model). We
give some computational methods for computing correlators, including a formula for concave genus-zero, four-point correlators, and show how to extend these results to find other
correlator values. In many cases our methods give enough information to compute the Amodel Frobenius manifold. We give the FJRW Frobenius manifold structure for 27 pairs
of polynomials and groups, 13 of which are constructed using a non-maximal symmetry
group.
Date: November 17, 2014.
1

AMANDA FRANCIS

Conjecture 1.1 (The Landau Ginzburg Mirror Symmetry Conjecture). There exist A- and
B-model structures, each constructed from a polynomial W and an associated group, G
(see discussion following Definition 2.1), such that the A-model for W and G is isomorphic
to the B-model for W T and GT , where W T and GT are dual to the original polynomial and
group.
Although physics predicted its existence, a mathematical construction of the A-model
was not known until 2007 when Fan, Jarvis, and Ruan, following the ideas of Witten,
proposed a cohomological field theory AW,G to satisfy Conjecture 1.1 [1, 2, 3].
A basis for the vector space of the A-model consists of pairs of a monomial and a group
element. When the group element acts nontrivially on each variable of W, we call this
a narrow element, otherwise, it is called broad. The structure in the A-model is determined by certain structure constants called genus-g, k-point correlators, which come from
the cohomology of the moduli space M g,k of genus-g curves with k marked points. The
Frobenius algebra structure is given by the genus-zero, three-point correlators, the Frobenius manifold structure by the genus-zero, k-point correlators for k 3, and the higher
genus structure by the genus-g, k-point correlators for all nonnegative integers g and k
such that 2g 2 + k > 0. These correlators are defined as integrals of certain cohomology classes over M g,k . Finding the values of these correlators is a difficult PDE problem,
which has not been solved in general. They are difficult to compute, especially when they
contain broad elements, so in many cases we still do not know how to compute even the
A-model Frobenius algebra structure. In most cases we do not know how to compute the
Frobenius manifold or higher genus structures. In this paper we make progress along these
lines.
In 2010, Krawitz [4] proved Conjecture 1.1 at the level of Frobenius algebra for almost
every invertible polynomial W and G = Gmax
W , the maximal symmetry group. It is more
difficult to determine the structure in the A-model when G , Gmax
W , because of the introduction of broad elements. In 2011 Johnson, Jarvis, Francis and Suggs [5] proved the
conjecture at the Frobenius algebra level for any pair (W, G) of invertible polynomial and
admissible symmetry group with the following property:
Property (). Let W be an invertible, nondegenerate, quasihomogeneous polynomial and
let G be an admissible symmetry group for W. We say the pair (W, G) has Property () if
PM
(1) W can be decomposed as W = i=1
Wi where the Wi are themselves invertible
polynomials having no variables in common with any other W j .
(2) For any element g of G, where some monomial dm ; gc is an element of HW,G , and
for each i {1, . . . , M}, g fixes either all of the variables in Wi or none of them.
Even for W and G satisfying Property (), where we know the isomorphism class of the
the Frobenius algebra, we usually still cannot compute the entire Frobenius manifold (the
genus-zero correlators), nor the higher genus potential for the cohomological field theory.
In fact, computing the full structure of either model is difficult, and has only been done
in a few cases. The easiest examples of singularities are the so-called simple or ADE
singularities. Fan, Jarvis, and Ruan computed the full A-model structure for these in [1].
The next examples come from representatives of the Elliptic singularities P8 , X9 , J10 ,
and their transpose singularities. Shen and Krawitz [6] calculated the entire A-model for
T
certain polynomial representatives of P8 , X9T and J10
, with maximal symmetry group.
In 2013, Guere [7] provided an explicit formula for the cohomology classes when W
is an invertible chain type polynomial, and G is the maximal symmetry group for W.

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

In 2008, Krawitz, Priddis, Acosta, Bergin, and Rathnakumara [8] worked out the Frobenius algebra structure for quasihomogeneous polynomial representatives of various singularities in Arnolds list of unimodal and bimodal singularities [9]. We expand on their
results, giving the Frobenius manifold structure for the polynomials on Arnolds list, as
well as their transpose polynomials. In many cases there is more than one admissible
symmetry group for a polynomial.
In this paper we consider all possible symmetry groups for each polynomial. Previously, almost no computations have been done in the case where the symmetry group is
not maximal, because of the introduction of broad elements. However, this case is particularly interesting because the corresponding transpose symmetry group, used for B side
computations, will not be trivial. Thus mirror symmetry predicts these cases will correspond to an orbifolded B-model. Much is still unknown about these orbifolded B-models.
Knowing the corresponding A-model structures, though, give us a very concrete prediction
for what these orbifolded B-models should be.
Much of the work that has been done so far in this area has used a list of properties
(axioms) of FJRW theory originally proved in [1] to calculate the values of correlators
in certain cases. Primarily, these axioms have been used to compute genus-zero threepoint correlators, but many of them can be used to find information about higher genus
and higher point correlators. One such axiom is the concavity axiom. This axiom gives a
formula for some of the cohomology classes of the cohomological field theory in terms
of the top Chern class of a sum of derived pushforward sheaves.
We use a result of Chiodo [10] to compute the Chern characters of the individual sheaves
in terms of some cohomology classes in the moduli stack of W-curves Wg,k in genus zero.
Then we use various properties of Chern classes to compute the top Chern class of the sum
of the sheaves given the Chern character of each individual sheaf. Much is known about
the cohomology M g,k , and not a lot about Wg,k , so we push down the cohomology classes
in Wg,k to certain tautological classes i , a and I over M 0,k . In this way, we provide a
method for expressing as a polynomial in the tautological classes of M g,k . Correlators
can then be computed by integrating over M g,k , which is equivalent to calculating certain
intersection numbers.
Algorithms for computing these numbers are well established, for example in [11, 12].
Code in various platforms (for example [11] in Maple and [13] in Sage) has been written
which computes the intersection numbers we need. I wrote code in Sage which performs
each of the steps mentioned above to find the top Chern class of the sum of the derived
pushforward sheaves, and then uses Johnsons intersection code [13] to find intersection
numbers. This allows us to compute certain correlator values which were previously unknown. In particular, in Lemma 3.5, we restate an explicit formula found in [1] for computing any concave genus-zero four-point correlators, with a proof not previously given.
We also describe how to compute higher point correlators. To do this, we use a strengthened version of the Reconstruction Lemma of [1] to find values of non-concave correlators,
with an aim to describe the full Frobenius manifold structure of many pairs (W, G) of singularities and groups. In many cases, these new methods allow us to compute Frobenius
manifold structures for certain singularities and groups which were previously unknown.
In 2014, Li, Li, Saito, and Shen [14] computed the entire A-model structure for the
14 polynomial representatives of exceptional unimodal singularities listed in [9], and their
transpose polynomial representatives, with maximal symmetry group. In fact, for these
polynomials and groups, they proved Conjecture 1.1. But there are also three non-maximal
symmetry groups that appear in this family of polynomials: the minimal symmetry groups

AMANDA FRANCIS

T
for Z13
, Q12 , and U12 . The A-model for the first of these is computed in this paper (see
Section 4.2), for the second, the A-model splits into the tensor product of two known
A-models (see Section 4.1). We do not currently know a way to compute the A-model
structure for U12 with minimal symmetry group (see Section 4.3).
There are 51 pairs of invertible polynomials and admissible symmetry groups corresponding to those listed in [9], whose A-model structure is still unknown. Of these, 16
pairs have FJRW theories which split into the tensor products of theories previously computed (see Section 4.1). For 27 of the remaining pairs, we are able to use our computational
methods to find the full Frobenius manifold structure of the corresponding A-model (see
Section 4.2), including 13 examples with non-maximal symmetry groups. There are 8 pairs
whose theories we still cannot compute using any known methods (see Section 4.3).

1.1. Acknowledgements. I would like to thank my advisor Tyler Jarvis for many helpful
discussions that made this work possible. Thank also to those who have worked on the
SAGE code which I used to do these computations, including Drew Johnson, Scott Mancuso and Rachel Webb. I would also like to thank Rachel Webb for pointing out some
errors in earlier versions of this paper.
2. Background
We begin by reviewing key facts about the construction of the A-models. This will
require the choice of an admissible polynomial and an associated symmetry group.
2.1. Admissible Polynomials and Symmetry Groups. A polynomial W C[x1 , . . . , xn ],
P Q
a
where W = i ci nj=1 x j i, j is called quasihomogeneous if there exist positive rational numbers q j (called weights) for each variable x j such that each monomial of W has weighted
P
degree one. That is, for every i where ci , 0, j q j ai, j = 1.
A polynomial W C[x1 , . . . , xn ] is called nondegenerate if it has an isolated singularity
at the origin.
Any polynomial which is both quasihomogeneous and nondegenerate will be considered admissible for our purposes. We say that an admissible polynomial is invertible if it
has the same number of variables and monomials. In this paper, we focus on computations
involving only invertible polynomials.
P
The central charge c of an admissible polynomial W is given by c = j (1 2q j ).
Next we define the maximal diagonal symmetry group.
Definition 2.1. Let W be an admissible polynomial. The maximal diagonal symmetry
n
group Gmax
W is the group of elements of the form g = (g1 , . . . , gn ) (Q/Z) such that
W(e2ig1 x1 , e2ig2 x2 , . . . , e2ign xn ) = W(x1 , x2 , . . . , xn ).
Remark 2.2. If q1 , q2 , . . . , qn are the weights of W, then the element J = (q1 , . . . , qn ) is an
element of Gmax
W .
FJRW theory requires not only a quasihomogeneous, nondegenerate polynomial W, but
also the choice of a subgroup G of Gmax
W which contains the element J. Such a group is
min
called admissible. We denote hJi = Gmin
and Gmax is
W , and any subgroup between G
admissible.
Example 2.3. Consider the polynomial W = W1,0 = x4 + y6 . In this case that there are two
admissible symmetry groups:
hJi = h(1/4, 1/6)i, and Gmax
W = h(1/4, 0), (0, 1/6)i.
We shall use W = W1,0 and G = hJi for the rest of the examples in this section.

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

2.2. Vector Space Construction. We now briefly review the construction of the A-model
state space, as a graded vector space.
We use the notation Ig = {i | g xi = xi } to denote the set of indices of those variables fixed by an element g, and Fix(g) to denote the subspace of Cn which is fixed by g,
Fix(g) = {(a1 , . . . , an )| such that ai = 0 whenever g xi , xi }. The notation Wg will denote
the polynomial W restricted to Fix(g).
For the polynomial W with symmetry group G, the state space HW,G is defined in terms
of Lefschetz thimbles and is equipped with a natural pairing, h, i : HW,G HW,G C
which we will not define here.
Definition 2.4. Let Hg,G be the G-invariants of the middle-dimensional relative cohomology
G
Hg,G = H mid (Fix(g), (W)1
g ()) ,
where Wg1 () is a generic smooth fiber of the restriction of W to Fix(g). The state space
is given by

Hg,G .
HW,G =
gG

C[x1 . . . , xn ]
E . It has a natural
Recall the Milnor Ring QW of the polynomial W is D
W
W
x1 , . . . , xn
residue pairing.
Lemma 2.5 (Wall). Let = dx1 . . . dxn , then
n
 QW .
dW n1
are isomorphic as GW -spaces, and this isomorphism respects the pairing on both.
H0,G = H mid (Cn , (W)1 ()) 

The isomorphism in Lemma 2.5 certainly will hold for the restricted polynomials Wg as
well. This gives us the useful fact
M
M
G
Qg g ,
Hg,G 
(1)
HW,G =
gG

gG

where g = dxi1 . . . dxis for i j Ig .

Notation 1. An element of HW,G is a linear combination of basis elements. We denote these
basis elements by dm ; gc, where m is a monomial in C[xi1 , . . . , xir ] and {i1 , . . . , ir } = Ig . We
say that dm ; gc is narrow if Ig = , and broad otherwise.
l
k
The complex degree degC of a basis element = m ; (g1 , . . . , gn ) is given by
n

degC =

X
1
Ni +
(g j q j ),
2
j=1

where Ng is the number of variables in Fix(gi ).

By fixing an order for the basis, we can create a matrix which contains all the pairing
information. This pairing matrix is given by
hD
Ei
= i , j
where {i } is a basis for HW,G .

AMANDA FRANCIS

2.3. The Moduli Space and basic properties. The FJRW cohomological field theory
arises from the construction of certain cohomology classes on a finite cover of the moduli
space of stable curves, the moduli space of W-orbicurves.
2.3.1. Moduli Spaces of Curves. The Moduli Space of stable curves of genus g over C
with k marked points M g,k can be thought of as the set of equivalence classes of (possibly
nodal) Riemann surfaces C with k marked points, p1 , . . . pk , where pi , p j if i , j. We
require an additional stability condition, that the automorphism group of any such curve
be finite. This means that 2g 2 + k > 0 for each irreducible component of C, where k
includes the nodal points.

M g,k .
We denote the universal curve over M g,k by C
The dual graph of a curve in M g,k is a graph with a node representing each irreducible
component, an edge for each nodal point, and a half edge for each mark. Componenents
with genus equal to zero are denoted with a filled-in dot while higher-genus components
are denoted by a vertex labeled with the genus.
Example 2.6. A nodal curve in M 1,3 and its dual graph are shown below.
1

For each pair of non-negative integers g and k, with 2g 2 + k > 0, the FJRW cohok
mological field theory produces for each k-tuple (1 , . . . , k ) HW,G
a cohomology class

W
g,k (1 , 2 , ..., k ) H (M g,k ). The definition of this class can be found in [2].
A genus-g, k-point correlator with insertions 1 , . . . , k HW,G is defined by the integral
Z

h1 , . . . , k ig,n =

M g,k

g,k (1 , . . . , k ) .

Finding the values of these correlators is a difficult PDE problem, which has not been
solved in general. In spite of this, we can calculate enough of them to determine the full
Frobenius manifold structure in many cases. Fan, Jarvis and Ruan [1] provide some axioms
the classes must satisfy, which, in some cases, allow us to determine their values. For
example, these axioms give the following selection rules for non-vanishing correlators.
Axiom 2.7 (FJR). Here we let i = dmi ; gi c, with gi = (g1i , . . . , gni ).


(1) h1 , . . . , k ig,k = (1) , . . . , (k) g,k . For any permutation S k .
P
(2) h1 , . . . , k i0,k = 0 unless ki=1 degC i = c + k 3.
P
(3) h1 , . . . , k i0,k = 0 unless q j (k 2) ki=1 gij Z for each j = 1, . . . n
Notice that the last selection rule above gives the following lemma.
Pk1
Lemma 2.8. The correlator h1 , . . . , k i0,k = 0 unless gk = (k 2) J i=1
gi , where
max
J GW is the element whose entries are the quasihomogeneous weights of W.
The following splitting axiom will also be useful for us.
Axiom 2.9 (FJR). if W1 C[x1 , . . . , xr ] and W2 C[xr+1 , . . . , xn ] are two admissible
polynomials with symmetry groups G1 , and G2 , respectively, then
HW1 +W2 ,G1 G2  HW1 ,G1 HW2 ,G2

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

where the classes are related by

W1 +W2 (1 1 , . . . , k k ) = W1 (1 , . . . , k ) W2 (1 , . . . , k ).
We omit the rest of the axioms (except the Concavity Axiom, which we will discuss
later), but refer the reader to [1] for the axioms, and [8] for a detailed explanation of how
to use them to find genus-zero, three-point correlator values.
2.3.2. B-model Structure. For the unorbifolded B-model the Frobenius manifold is given
by the Saito Frobenius manifold for a particular choice of primitive form (see [15]). For
the orbifolded B-model the Frobenius manifold structure is still unknown.
3. Computational Methods
Here we give results for computing concave genus-zero correlators and discuss how to
use the reconstruction lemma to find values of other correlators.
3.1. Using the Concavity Axiom. We give a formula for as a polynomial in the tautological classes i , a , and I in H (M g,k ). We then give a formula for computing concave
genus-zero four-point correlators.
Axiom 3.1
Suppose that all i are narrow insertions of the form d1 ; gc (See Notation
L(FJR).
n
1). If i=1 Li = 0, then the cohomology class W
. . , k ) can be given in terms of
g,k (1 , .L
n
1
the top Chern class of the derived pushforward sheaf R i=1 Li :

n
1
1 M

|G|g
W
D
(2)
g,k (1 , . . . , k ) =
PDst PD (1) cD R
Li
deg(st)
i=1

Recall that integration of top dimensional cohomology classes is the same as pushing
them forward to a point, or computing intersection numbers.
In this section we express (1)D cD R1 (L1 . . . LN ) as a polynomial f in terms of
pullbacks of , , and I classes, then, which gives


|G|g
PDst PD1 (st f (1 , . . . , D , 1 , . . . , k , {I }II ))
h1 , . . . , k i = p deg(st)
= |G|g p ( f (1 , . . . , D , 1 , . . . , k , {I }II )) ,
and this allows us to use intersection theory to solve for these numbers.
Recall the following well-known property of chern classes from K-theory.

X
1
1
=
=
(ct (E ))i
(3)
ct (E ) =
ct (E ) 1 (ct (E )) i=0
A vector bundle E is concave when R0 E = 0, or when
R E = R0 E R1 E = R1 E
Equation 3 gives
ct (R
1

M
i

j

X
M

Li ) =
Li )
ct (R
i

j=0

Also, since the total Chern classes are multiplicative, we have,

n

M
X Y

(4)
ck (R
Li ) =
ci j (R L j ).
P
i

i j =k

j=1

Together Equations 3 and 4 give a formula for finding the ith chern class of R1
terms of the chern classes of the R L j .

L
i

Li in

AMANDA FRANCIS

It is well-known that Chern characters can be expressed in terms of Chern classes, but
it is also possible to express Chern classes in terms of Chern characters (for example, in
[16]). We have
P

i1

i
ct (R Lk )
= exp
i=1 (i 1)!(1) chi (R Lk )t
j
(5)
P 1  P
i1

i
i=1 (i 1)!(1) chi (R Lk )t
j=0 j!
Thus, there exists a polynomial f such that
M
(6)
cD (R1
Li ) = f (ch1 (R Lk ), . . . , chD (R Lk ))
i

In Section 3.2 we will use a result of Chiodo [10] to express chi (R Lk ) in terms of tautological cohomology classes on M g,n and this will allow us to compute the corresponding
correlators, but first we need to discuss some cohomology classes on Wg,n and M g,n and
their relations.
3.1.1. Orbicurves. An orbicurve C with marked points p1 , . . . , pk is a stable curve C with
orbifold structure at each pi and each node. Near each marked point pi there is a local
group action given by Z/mi for some positive integer mi . Similarly, near each node p there
is again a local group Z/n j whose action on one branch is inverse to the action on the other
branch.
In a neighborhood of pi , C maps to C via the map,
% : C C,
where if z is the local coordinate on C near pi , and x is the local coordinate on C near pi ,
then %(z) = zr = x.
Let KC be the canonical bundle of C. The log-canonical bundle of C is the line bundle
KC,log = KC O(p1 ) . . . O(pk ),
where O(pi ) is the holomorphic line bundle of degree one whose sections may have a
simple pole at pi .
The log-canonical bundle of C is defined to be the pullback to C of the log-canonical
bundle of C:
KC ,log = % KC,log .
Given an admissible polynomial W, a W-structure on an orbicurve C is essentially
P
a choice of n line bundles L1 , ..., Ln so that for each monomial of W = j M j , with
a
M j = x1a j,1 . . . xn j,n , we have an isomorphism of line bundles
a j,1

j : L1

a j,n

. . . Ln

C ,log .
K

Recall that Fan, Jarvis and Ruan [1] defined a stack

WW,g,k = {C , p1 , . . . , pk , L1 , . . . , LN , 1 . . . s }
of stable orbicurves with the additional W-structure, and the canonical morphism,
st
- Mg,k
Wg,k
from the stack of W-curves to the stack of stable curves, Mg,k .
Notice that % will take global sections to global sections, and a straightforward compus
tation (see [1] 2.1 ) shows that if L on C such that L r  KC
, then
,log
s
(% L )r = KC,log
((r mi )p).

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

Recall that if L is the line bundle associated to the sheaf L and the action at an orbifold
point pi on L is given in local coordinates by (z, v) 7 (z, mi v), then the action on L will
s
take a generator s of L near pi and map it to (rmi ) s. Thus, if L r  KC
on a smooth
,log
mi
orbicurve with action of the local group on L defined by for mi > 0 at each marked
point pi , then

r
r
s
(% L) = |L| = C,log
O((mi )pi ) .
i

Example 3.2. Suppose W and G are as in Example 2.3 with notation as in Notation
1. If, for a curve in M 0,4 the four marked points correspond to the A-model elements
d1 ; (1/4, 1/2)c, d1 ; (1/4, 1/2)c, d1 ; (1/4, 1/2)c, and d1 ; (3/4, 5/6)c, then
|L x |4 = C,log O((1)p1 ) O((1)p2 ) O((1)p3 ) O((3)p4 )
|Ly |6 = C,log O((3)p1 ) O((3)p2 ) O((3)p3 ) O((5)p4 )
3.1.2. Some Special Cohomology Classes in M g,k and Wg,k . For i {1, . . . , k}, i
H q (M g,k ) is the first Chern class of the line bundle whose fiber at (C, p1 , . . . , pk ) is the
cotangent space to C at pi . In other words, if k+1 : M g,k+1 = C M g,k is the universal curve, and it is also the morphism obtained by forgetting the (k + 1)-st marked point,
k+1 = is the relative dualizing sheaf, and i is the section of k+1 which attaches a
genus-zero, three-pointed curve to C at the point pi , and then labels the two remaining
marked points on the genus-zero curve i and k + 1,
M g,k+1
i

k+1
?
M g,k

then, L = () is the cotangent line bundle and its first Chern class is i :
i = c1 ( ()),
Let Di,k+1 be the image of i in M g,k+1 , then we define

k
X

K = c1
Di,k+1 ,
i=1

and for a {1, . . . , 3g 3 + k},

a = (K a+1 ).

Each partition I t J = {1, . . . , k} and g1 + g2 = g of marks and genus such that 1 I,

2g1 2 + |I| + 1 > 0 and 2g2 2 + |J| + 1 > 0, gives an irreducible boundary divisor, which
we label g1 ,I . These boundary divisors are the nodal curves in M g,k . For example, the
boundary divisor 1,{1,2} in M 1,5 and its dual graph are given below.
2

3
4

1
2

4
5

We will use the following well-known lemma for M 0,k for expressing classes in terms
of boundary divisors.

10

AMANDA FRANCIS

Lemma 3.3.
i =

aI
b,c<I

Now we consider some cohomology classes on Wg,k . Consider the diagram:

Cg,k
i

Wg,k

st

M g,k

where Cg,k is the universal orbicurve over Wg,k .

The stack Wg,k has cohomology classes i , a and I defined in the same manner as i ,
a and I in M g,k . They satisfy the following properties (see 2.3 in [1]).
i = st (i ),

(7)

a = st (a ),

rI = st I

s
3.2. Chiodos Formula. Chiodos formula states that for the universal rth root L of log
on the universal family of pointed orbicurves : Cg,k Wg,k (1 , . . . , k ), with local group
hi i of order mi at the ith marked point, we have

k
X Bd+1 (s/r)
X Bd+1 (i )

X
1 X rBd+1 (+ )
j

d
i

ch(R (L )) =
d
i +
% cut
(+ ) ,
(d + 1)!

(d + 1)!
2
(d + 1)!
i=1
d0
i+ j=d1
cut
i, j0

where the second sum is taken over all decorated stable graphs cut with one pair of tails
labelled + and , respectively, so that once the + and edges have been glued, we get a
single-edged, n-pointed, connected, decorated graph of genus g and with additional decoration (+ and ) on the internal edge. Each such graph cut has the two cut edges, decorated
with group elements
+ and , respectively, and the map % cut is the corresponding gluing

r/s

r/s

map M cut , M g,k (1 , . . . , k ).

In the genus zero case, a choice of cut is the same as a a partition K t K 0 = {1, . . . , k}.
We will sum over all partitions containing the marked point 1, so we will not need to
multiply the last sum by 12 .
Also, recall from Equation 7 that I = 1r st I . So,


n
X
X Bd+1 (s/r)

Bd+1 (mi /r) d X Bd+1 (+K )

cht (R L ) = st
d
i +
( jK ) (d1 ) td
(d + 1)!
(d + 1)!
(d + 1)!
K
i=1
d0
We can use Lemma 3.3 to rewrite + and in terms of boundary divisors of M 0,n+
and M 0,n . This will enable us to easily push down these classes to M 0,n . This idea comes
from [17], and yields the following formulas:

(8)

( jK ) (+ ) = 0
P
P
( jK ) (+ ) = {1,a,b}IK K I + 1IK{a,b} K IK c
( jK ) (+ ) = 0
P
( jK ) (+ ) = ,IK c K IK
a,b,<I

if |K| 2
if |K| > 2
if n |K| 2
if n |K| > 2

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

11

Using the formulas in Equation 8 and the polynomial defined in Equation 6 we can now
express as a polynomial in , and classes,
W
g,k (1 , . . . , k )

P
P B (K )
(s/r)
(mi /r)
= (1)D f B2(2)!
1 ni=1 B2 (2)!
i + K 2(2)!+ ( jK ) ( + ) , . . .


Pn BD+1 (mi /r) D P BD+1 (+K )
P
BD+1 (s/r)
i j
(D+1)! D i=1 (D+1)! i + K (D+1)! ( jK ) ( i+ j=D1 (+ ) )
The following lemma allows us to always choose a +K in a way that makes sense in
Chiodos formula.
Lemma 3.4. Let B be a degree one boundary graph, with decorations 11 , . . . , 1k1 for the
first node and 21 , . . . , 2k2 for the second, and genera g1 and g2 , respectively. If a smooth
curve with decorations 11 , . . . , 1k1 , 21 , . . . , 2k2 and genus g1 + g2 has integer line bundle
degree, then it is possible to assign decorations to the edge of B such that each node will
have integral line bundle degree.
Proof. If the line bundle degree of the smooth curve is integral then:
X
X
1i
2j Zn
S 0 = J(2(g1 + g2 ) 2 + k1 + k2 )
i

Similarly let S 1 and S 2 be the equivalent sums in Q , corresponding to the nodes with
decorations 11 , . . . , 1k1 and 21 , . . . , 2k2 , respectively. To find 0 we take
X
0 J(2g1 2 + (k1 + 1))
1i
i

Then, by Lemma 2.8 this will force the S 1 to be an integer vector.

Also,
X
S 0 = S 1 + 0 + J(g2 2 + (k2 + 1))
2j = S 1 + S 2
j

Which implies that S 2 Z.

We have now described the virtual class in the concave case explicitly in terms of
tautological classes on the moduli of stable curves. Now we will talk about how these can
actually be computed. As a special case we have the following explicit formula for concave
genus-zero four-point correlators.
Lemma 3.5. If hd1 ; g1 c , d1 ; g2 c , d1 ; g3 c , d1 ; g4 ci is a genus-zero, four-point correlator
which satisfies parts (2) and (3) of Axiom 2.7, and if it is also satisfies the hypotheses of
the concavity axiom, (all insertions are narrow, all line bundle degrees are negative, and
all line bundle degrees of the nodes of the boundary graphs 1,2 , 1,3 , and 1,4 are all
negative), then

N
4
3
X
X

1 X
j
B2 (qi )
B2 ((g j )i ) +
B2 (+ )
hd1 ; g1 c , d1 ; g2 c , d1 ; g3 c , d1 ; g4 ci =
2 i=1
j=1
k=1

N
4
3
X
X

1 X
qi (qi 1)
=
ij (ij 1) +
+j (+j 1)
2 i=1
j=1
k=1
where for each j g j = (1j , . . . , nj ), +1 J g1 g2 , +2 J g1 g3 , and +3 J g1 g4 ,
and B2 is the 2nd Bernoulli polynomial.

12

AMANDA FRANCIS

Proof.
cht (R1 Li ) =

X  Bd+1 (qi )
d0

(d + 1)!

i
4 B
X
d+1 ( j )

dj
(d
+
1)!
j=1
d1

X Bd+1 (( K )i )
X
+
+r
(+ )k ( )d1k td
(pK )
(d + 1)!
K
k=0

Which means that

ch1 (R1 Li ) =

4 B (i )
X
X B2 ((+ )K )
2 j
B2 (qi )
1
j + r
(pK ) (1)
(2)!
(2)!
(2)!
K
j=1

Notice that (pK ) (1M 0,3 ) = K , and that for M 0,4 , our choices for K cut are just
{1, 2}, {1, 3}, and {1, 4}. Numbering these gives:
(+ )1 = J g1 g2 ,
(+ )2 = J g1 g3 ,
(+ )3 = J g1 g4 ,

for K = {1, 2},

for K = {1, 3}, and
for K = {1, 4}.

Since B2 (x) = x2 x + 61 ,
ch1 (R1 Li )

1
1 
qi (qi 1) + 1
2
6
4 
X
1
ij (ij 1) + j

6
j=1
!
X
1
K K
+ (+ 1) + K .
+r
6
K

The psi and kappa classes in W0,4 are all pullbacks of the equivalent psi and kappa
classes in M , 0,4 , and the I classes are scalar multiples of the equivalent classes in M , 0,4 ,
as in Equation 7
So,
4
X
X






1 
ch1 (R1 Li ) = st qi (qi 1) 1
ij (ij 1) j +
+K (+K 1) K .
2
K
j=1
We use Equation 5 to convert to chern classes,

1
i1
1
i

ct (R Li ) = exp (1) (i 1)!chi (R Li )t

i=1

1 X
i1
1
=
(1) (i 1)!cht (R Li )ti .
j! i=1
j=0
Which means that c0 (R1 Li ) = 1 and c1 (R1 Li ) = ch1 (R1 Li ), then, by Equation
4 since D = 1,
X Y
c1 (R1 i Li ) =
c ji (R1 Li )

0 j1 ,... jN
j1 +...+ jN =1
N
X

N
X

i=1

i=1

c1 (R1 Li ) =

ch1 (R1 Li ).

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

Finally we recall that ct (R1 Li ) =

j (ct (R

13

Li )) j , so,

c1 (R1 Li ) = c1 (R1 Li ) =

N
X

ch1 (R1 Li ).

i=1

And, from Equation 2

N
X
1
ch1 (R1 Li ))
PDst PD1 (1)1 (
deg(st)
i=1
N
X
1
ch1 (R1 Li ))
=
PDst PD1
deg(st)
i=1
!
4 
X
X
1 
1
1
1
i i
=
qi (qi 1) + 1
j ( j 1) + j + r
+K (+K 1) + K .
2
6
6
6
K
j=1

0,4 (1 , . . . , 4 ) =

Next, we notice that if p : M 0,4 () is the map sending all of M 0,4 to a point, then
the push forward of any of the cohomology classes mentioned above is equal to 1. That is,
p 1 = p i = p K = 1.
So,
hd1 ; g1 c , d1 ; g2 c , d1 ; g3 c , d1 ; g4 ci
N
X
1
= p
PDst PD1
ch1 (R1 Li ))
deg(st)
i=1

N
4 
X
X


1 X
K
K
i
i
(qi (qi 1)) 1
+ (+ 1) K
j ( j 1) j +
= p
2 i=1
K
j=1

N
4
X

1 X
i
i
1
1
2
2
3
3
qi (qi 1) + + (+ 1) + + (+ 1) + + (+ 1)
=
j ( j 1) .
2 i=1
j=1

Example 3.6. We will compute the genus-zero four-point correlator
hd1 ; (1/4, 1/2)c , d1 ; (1/4, 1/2)c , d1 ; (1/4, 1/2)c , d1 ; (3/4, 5/6)ci0,4 for the A-model HW1,0 ,hJi .
It is straightforward to verify that it satisfies Axiom 2.7, and that each of the degree one
nodal degenerations of the dual graph are concave. Thus,
1
2

hd1 ; (1/4, 1/2)c , d1 ; (1/4, 1/2)c , d1 ; (1/4,


 1/2)c , d1 ; (3/4, 5/6)ci0,4 =
3
3
3
3
3
3
3
3
5
5
5
5
16 16 16 16 + 16 + 16 + 16 + 16 + 36
36
36
36
+ 14 + 14 + 14 +


1
1
1
2 0+ 3 = 6

5
36



There are ten concave four-point correlators in HW1,0 ,hJi and 2 which are not concave.
We will see in the next section that we can use the values of the concave correlators to find
other correlator values.
3.3. Using the Reconstruction Lemma. In this section we show how to use known correlator values to find unknown correlator values. In some cases our new methods for computing concave correlators will allow us to compute all genus-zero correlators in the A-model.
The WDVV equations are a powerful tool which can be derived from the Composition
axiom. Applying these equations to correlators, we get the following lemma,

14

AMANDA FRANCIS

Lemma 3.7. [17] Reconstruction Lemma.

Any genus-zero, k-point correlator of the form
h1 , ..., k3 , , ,  ? i0,k
where 0 < degC (), degC () < c , can be rewritten as
X X
D
E
cI,J hkI , , , l i 0l , , , jJ
h1 , ..., k3 , , ,  ? i =

ItJ=[k3]

ItJ=[k3]
J,

D
E
cI,J hkI , , , l i 0l , , , jJ .

where the l are the elements of Q

some basis B and 0l are the corresponding elements of
(k )!
the dual basis B 0 , and cI,J = Q nI (in)!KQ
n JQ
( j )! . Here nX ( x ) refers to the number of elements
equal to x in the tuple X. The product nX ( x )! is taken over all distinct elements x in
X.
We say that an element HW,G is non-primitive if it can be written  ? = for some
 and in HW,G with 0 < degC , degC < degC . Otherwise, we say that is primitive.
Corollary 3.8. A genus-zero, four-point correlator containing a non-primitive insertion
can be rewritten:
X
D
E X
D
E
h, , , l i 0l , , +
h, , l i 0l , , ,
h, , ,  ? i0.4 =
l

D
E
h, , l i 0l , , , .

In fact, using the reconstruction lemma, it is possible to write any genus-zero k-point
correlator in terms of the pairing, genus-zero three-point correlators and correlators of the
form h1 , . . . , k0 i for k0 k where i is primitive for i k0 2 (see [2]).
We say that a correlator is basic if at most two of the insertions are non-primitive.
P
Recall that the central charge of the polynomial W is c = j (1 2q j ).
Lemma 3.9 ([2]). If degC () < c for all classes , P is the maximum complex degree of
any primitive class, and P < 1, then all the genus-zero correlators are uniquely determined
by the pairing, the genus-zero three-point correlators, and the basic genus-zero k-point
correlators for each k that satisfies
(9)

k 2+

1 + c
.
1P

Example 3.10. In the case of HW1,0 ,hJi , the maximal primitive degree is 7/12 and c = 7/6,
so the Frobenius manifold structure is determined by the genus-zero three-point correlators
and the basic genus-zero k-point correlators for k 7.
All of the three-point correlators can be found using the methods described in [8]. There
are 165 distinct genus-zero three point correlators. The selection rules show that 158 of
them vanish. All of the seven remaining correlator values can be found using a pairing
axiom found in [1]. There are ten concave and two nonconcave four-point correlators
which satisfy Axiom 2.7, which might therefore be nonzero. We can use the values of the
concave correlators to find values of the others using Lemma 3.7. For example, if we let
l
k
X = d1 ; (1/4, 1/2)c , Y = d1 ; (1/2, 1/3)c , Z = d1 ; (3/4, 1/6)c , and W = xy2 ; (0, 0) ,

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

15

then the nonconcave four point correlators are hX, W, W, X 2 i and hZ, Z, W, Wi. Lemma 3.7
gives
X
hX, W, W, X 2 i =
2hX, W, X, ih0 , W, Xi hW, W, ih0 , X, X, Xi.

The selection rules, methods in [8], and computations from Example 3.6 reduce this to
hX, W, W, X 2 i = hW, W, 1ihXY 2 , X, X, Xi=

1
1 1
=
.
24 6
144

To find the value of hZ, Z, W, Wi we have to look for ways to reconstruct it using fivepoint correlators.
hY, W, W, XY, XZi = hX, Y, W, XZihX, Z, W, XYi + hX, W, W, X 2 ihY, Z, Z, XYi
hW, XY, XZihX, X, Y, Z, Wi hY, Y, W, XYihX, Z, W, XYi.
The correlators hX, Z, W, XYi and hW, XY, XZi are both equal to zero, which can be shown
using the methods in [8]. The correlators hY, Z, Z, XYi, and hX, W, W, X 2 i are both concave.
Using Lemma 3.5, it is a straightforward calculation to find their values, which are 16 and
1
4 , respectively. So we have
(10)

hY, W, W, XY, XZi = hX, W, W, X 2 ihY, Z, Z, XYi =

1
1 1
=
.
144 4
576

We can reconstruct this same correlator in a different way:

(11)

hW, Y, XY, W, XZi = hX, Y, W, XZihY, Z, W, X 2 i + hX, Y, X 2 , XYihZ, Z, W, Wi

hW, XY, XZihX, X, Y, Z, Wi hY, Y, W, XYihX, Z, W, XYi
= hX, Y, X 2 , XYihZ, Z, W, Wi = 16 hZ, Z, W, Wi.

1
.
Together, Equations 10 and 11 tell us that hZ, Z, W, Wi = 96
To compute the full Frobenius manifold structure of HW1,0 ,hJi , we still need to find the
values of all basic five, six, and seven-point correlators. There are fifteen basic five-point
correlators, seven of which are not concave. There are five basic six-point correlators,
three of which are not concave, and there is one basic seven-point correlator, which is not
concave. All of these correlators can be computed using the methods established here.
Their values are given in Section 4.2.

Remark 3.11. In certain examples, if we use Lemma 3.7 together with the computed
values of concave genus-zero higher point correlators, we can find values of previously
unknown three-point correlators. This allows us to find even Frobenius algebra structures
that were previously unknown.
4. Summary of Computations
We compute the FJRW theories with all possible admissible symmetry groups coming from the quasihomogeneous polynomial representatives corresponding to an isolated
singularity in Arnolds [9] list of singularities. Recall that there are potentially several
polynomial representatives corresponding to a singularity. We note here that the polynomial representatives of X9 and J10 which appear in this list of singularities [9] were not
considered in [6] for any choice of symmetry group. The non-maximal symmetry groups
for P8 , also, have not yet been treated.

16

AMANDA FRANCIS

4.1. FJRW Theories which split into known tensor products. Recall that if a polynomial W can be written W = W1 + W2 , where W1 C[x1 , . . . , xl ] and W2 C[xl+1 , . . . , xn ],
and if G can be written G = G1 G2 , where for each i, Gi is an admissible symmetry group
for Wi , then then the A-model associated to the pair (W, G) splits into the tensor product of
the A-models for the pairs (W1 , G1 ) and (W2 , G2 ).
The ADE singularities (except for D4 ) with all possible symmetry groups were computed in [1]. The D4 case was done in [17]. For the following pairs of polynomial and
group, the A-model structure splits into a product of known FJRW theories. We use the
notation AW,G to denote the full FJRW theory associated to the pair (W, G). Symmetry
groups which are maximal are denoted with the subscript max.
W
X9 = x4 + y4
J10 = x3 + y6
Q12 = x3 + y5 + yz2
J3,0 = x3 + y9
W1,0 = x4 + y6
E18 = x3 + y10
E20 = x3 + y11
U16 = x3 + xz2 + y5
U16 = x3 + xz2 + y5
T
U16
= x3 z + z2 + y5
W18 = x4 + y7
Q16 = x3 + yz2 + y7
Q16 = x3 + yz2 + y7
QT16 = x3 + z2 + y7 z
Q18 = x3 + yz2 + y8

G
h(1/4, 0), (0, 1/4)imax
h(1/3, 0), (0, 1/6)imax
h(1/3, 0, 0), (0, 1/5, 2/5)i
h(1/3, 0), (0, 1/9)imax
h(1/4, 0), (0, 1/6)imax
h(1/3, 0), (0, 1/10)imax
h(1/3, 0), (0, 1/11)imax
h(1/3, 0, 5/6), (0, 1/5, 0)imax
h(1/3, 0, 1/3), (0, 1/5, 0)i
h(1/6, 0, 1/2), (0, 1/5, 0)imax
h(1/4, 0), (0, 1/7)imax
h(1/3, 0, 0), (0, 1/7, 13/14)imax
h(1/3, 0, 0), (0, 1/7, 3/7)i
h(1/3, 0, 0), (0, 1/14, 1/2)imax
h(1/3, 0, 0), (0, 1/8, 7/16)imax

dimension
9
10
12
16
15
18
20
20
16
16
18
26
16
16
18

A-models
AA3 ,J AA3 ,J
AA2 ,J AA5 ,J
AA2 ,J AD6 ,J
AA2 ,J AA8 ,J
AA3 ,J AA5 ,J
AA2 ,J AA9 ,J
AA2 ,J AA10 ,J
AD4 ,Gmax AA4 ,J
AD4 ,J AA4 ,J
ADT ,J AA4 ,J
4
AA3 ,J AA6 ,J
AA2 ,J AD8 ,Gmax
AA2 ,J AD8 ,J
AA2 ,J ADT ,J
8
AA2 ,J AD9 ,J

4.2. Previously Unknown FJRW Theories we can compute. Recall that in Lemma 3.9
we saw that the Frobenius manifold structure can be found from the genus-zero threepoint correlators, together with the genus-zero k-point correlators for k as in Equation 9.
The value of k depends on the central charge c of the polynomial and on the maximum
degree P of any primitive element in the state space. The value of k is given for each of the
A-models below, together with all necessary correlators. Correlators marked with a must
be found using the Reconstruction Lemma. Unmarked k-point correlators for k > 3 can be
found using the concavity axiom.
P8 = x3 + y3 + z3 , G = h = (1/3, 0, 2/3), 2 = (0, 1/3, 2/3)i
k 6, X = e0 , Y = xyze0 . P = 1/2, c = 1
Relations: X 2 = Y 2 = 0; dim = 4
4 pt correlators
3 pt correlators
None.
6 pt correlators
h1, 1, XYi = 1/27
hX, X, Y, Y, XY, XYi = 0
h1, X, Yi = 1/27
5pt correlators
None.

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

P8 = x3 + y3 + z3 , G = hJ = (1/3, 1/3, 1/3)i

k 6, X = e0 , Y = xyze0 . P = 1/2, c = 1
Relations: X 2 = Y 2 = 0; dim = 4
4 pt correlators
3 pt correlators
None.
6 pt correlators
h1, 1, XYi = 1/27
None.
h1, X, Yi = 1/27
5pt correlators
None.
X9 = x4 + y4 , G = hJ = (1/4, 1/4), = (0, 1/2)i
k 6, X = xye0 , Y = e J+ , Z = e2J , W = e3J+ . P = 1/2, c = 1
Relations: 16X 2 = YW = Z 2 , X 3 = Z 3 = Y 2 = W 2 = 0. dim = 6
3 pt correlators
5 pt correlators
6 pt correlators
h1, 1, Z 2 i = 1
hY, Y, Y, Y, Z 2 i = 81
hY, Y, Y, W, W 2 , W 2 i = 0
1
h1, Y, Wi = 1
hY, Y, Z, Z, W 2 , W 2 i = 32
hY, Y, W, W, Z 2 i = 0
1
h1, Z, Zi = 1
hY, Z, Z, W, W 2 i = 16
hY, W, W, W, W 2 , W 2 i = 0
1
1
h1, X, Xi = 161
hZ, Z, Z, Z, W 2 i = 16
hZ, Z, W, W, W 2 , W 2 i = 32
1
1
2
2
2
hW, W, W, W, W i = 8
hX, X, Y, Y, W , W i = 512
1
1
4 pt correlators
hX, X, X, X, W 2 i = 4096
hX, X, Z, Z, W 2 , W 2 i = 512
1
hX, X, Y, W, W 2 i = 256
hY, Y, Y, Wi = 0
1
hX, X, Z, Z, W 2 i = 256
hY, Y, Z, Zi = 1/4
hY, W, W, Wi = 0
hZ, Z, W, Wi = 1/4
hX, X, Y, Yi = 1/64
X9 = x4 + y4 , G = hJ = (1/4, 1/4)i
k 6, X = y2 e0 , Y = xye0 , Z = x2 e0 , W = e2J . P = 1/2, c = 1
Relations: 16XZ = 16Y 2 = W 2 , X 2 = Z 2 = Y 3 = W 3 = 0. dim = 6
3 pt correlators
5 pt correlators
6 pt correlators
1
h1, 1, W 2 i = 1
hW, W, W, W, W 2 i = 16
hX, X, Y, W, W 2 , W 2 i = 8a
1
h1, W, Wi = 1
hX, X, X, X, W 2 i = 32a2
hY, Z, Z, W, W 2 , W 2 i = 65536a
1
2
h1, X, Zi = 16
hX, X, Z, Z, W i = 0
1
1
h1, Y, Yi = 16
hX, Y, Y, Z, W 2 i = 4096
1
2
hX, Z, W, W, W i = 256
1
2
4 pt correlators
hY, Y, Y, Y, W i = 4096
1
hY, Y, W, W, W 2 i = 256
hX, X, Y, Wi = a , 0
1
1
hZ, Z, Z, Z, W 2 i = 134217728a
hY, Z, Z, Wi = 8192a
2

17

18

AMANDA FRANCIS

J10 = x3 + y6 , G = hJ = (1/3, 1/6)i

k 6, X = y3 e0 , Y = xye0 , Z = e4J , W = e2J . P = 1/2, c = 1
Relations: 18XY = ZW, X 2 = Y 2 = Z 2 = W 2 = 1, dim = 6
5 pt correlators
3 pt correlators
hY, Y, Y, Y, ZWi = 19
h1, 1, ZWi = 1
hY, Z, Z, Z, ZWi = 181
h1, Z, Wi = 1
hX, X, X, Z, ZWi = a3
h1, X, Yi = 181
1
hX, Y, W, W, ZWi = 324
1
hY, Y, Y, Z, ZWi = 314,928a
4 pt correlators
hZ, Z, Z, Wi = 1/6
6 pt correlators
hW, W, W, Wi = 1/3
hX, X, X, Wi = a , 0
hZ, Z, W, W, ZW, ZWi = 541
1
1
hX, Y, Z, Zi = 108
hX, X, Y, Y, ZW, ZWi = 17,496
1
1
hY, Y, Y, Wi = 104,976a
hX, Y, Z, W, ZW, ZWi = 1944
T
Z13
= x3 + xy4 , G = hJ = (1/3, 1/9)i
k 6, X = e4J , Y = e2J , Z = y5 e0 , W = xy2 e0 . P = 5/9, c = 10/9.
Relations: X 2 Y = 6Z 2 = 18W 2 , X 3 = Y 2 = Z 3 = W 3 = 0, dim = 8.
3 pt correlators
4 pt correlators
h1, 1, X 2 Yi = 1
h1, X, XYi = 1
hX, X, X, X 2 Yi = 1/9
h1, Y, X 2 i = 1
hX, X, Yi = 1
hX, X, X 2 , XYi = 2/9
h1, Z, Zi = 1/6 h1, W, Wi = 1/18
hX, Y, X 2 , X 2 i = 1/9
hY, Y, Y, XYi = 1/3
5 pt correlators
hX, Y, Z, XYi = 0
hX, Y, Y, XY, X 2 Yi = 1/27
hX, Z, Z, X 2 i = 1/54
hX, X, Z, XY, X 2 Yi = 1/27
hX, W, W, X 2 i = 1/162
hX, Y, Z, X 2 , X 2 Yi = 0
2
hY, Y, Z, X 2 i = 0
hX, Z, Z, Z, X Yi = 0
2
hX, Z, Z, Zi = a = 1/9
hX, Z, W, W, X Yi = 2a/9
hX, Z, W, Wi = a/36
hY, Y, Y, X 2 , X 2 Yi = a/162
6 pt correlators
hY, Y, Z, Z, X 2 Yi = 1/162
hY, Y, W, W, X 2 Yi = 1/486
hX, Y, Z, Z, X 2 Y, X 2 Yi = 1/1458
hY, Z, Z, XY, XYi = 1/162
hX, Y, W, W, X 2 Y, X 2 Yi = 1/4374
hY, W, W, XY, XYi = 1/486
hY, Y, Y, Z, X 2 Y, X 2 Yi = 0
hZ, Z, Z, X 2 , XYi = 2a/9
hZ, Z, Z, Z, XY, X 2 Yi = 1/2916
hZ, W, W, X 2 , XYi = a/162
hZ, Z, W, W, XY, X 2 Yi = 1/26244
hW, W, W, W, XY, X 2 Yi = 1/26244

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

J3,0 = x3 + y9 , G = hJ = (1/3, 1/9)i

k 6, X = e4J , Y = e2J , Z = y5 e0 , W = xy2 e0 . P = 5/9, c = 10/9.
Relations: 27ZW = X 2 Y, ab = 1/531441, X 3 = Y 2 = Z 2 = W 2 = 0, dim = 8.
3 pt correlators
5 pt correlators
h1, 1, X 2 Yi = 1
h1, X, XYi = 1
hX,
Y,
Y, XY, X 2 Yi = 1/27
h1, Y, X 2 i = 1 h1, Z, Wi = 1/27
hX, Z, Z, Z, X 2 Yi = 2a/9
hX, X, Yi = 1
hX, W, W, W, X 2 Yi = 2b/9
hY, Y, Y, X 2 , X 2 Yi = 1/27
4 pt correlators
hY,
Y, Z, W, X 2 Yi = 1/729
hX, X, X, X 2 Yi = 1/9
hY, Z, W, XY, XYi = 1/729
hX, X, X 2 , XYi = 2/9
hZ, Z, Z, X 2 , XYi = 2a/9
2
2
hX, Y, X , X i = 1/9
hW,
W, W, X 2 , XYi = 2b/9
hX, Z, W, X 2 i = 1/243
6 pt correlators
hY, Y, Y, XYi = 1/3
hX, Y, Z, W, X 2 Y, X 2 Yi = 1/6561
hY, Z, Z, Zi = a
hZ, Z, W, W, XY, X 2 Yi = 4/177147
hY, W, W, Wi = b
Z1,0 = x3 y + y7 , Gmax = h = (1/21, 6/7)i
k 5, X = e5 , Y = e13 . P = 1/3, c = 8/7.
Relations: X 7 = 3Y 2 , X 13 = Y 3 = 0, dim = 19.
3 pt correlators
4 pt correlators
h1, 1, X 5 Y 2 i = 1
h1, X, X 4 Y 2 i = 1
2
3 2
hX, X, X 5 Y, X 5 Y 2 i = 1/7
h1, X , X Y i = 1
h1, X 3 , X 2 Y 2 i = 1
hX, Y, X 6 , X 5 Y 2 i = 1/7
h1, Y, X 5 Yi = 1
h1, X 4 , XY 2 i = 1
4
5
2
hX, Y, Y 2 , X 4 Y 2 i = 1/21
h1, XY, X Yi = 1
h1, X , Y i = 1
2
3
6
6
hX, Y, XY 2 , X 3 Y 2 i = 1/21
h1, X Y, X Yi = 1
h1, X , X i = 3
hX, Y, X 2 Y 2 , X 2 Y 2 i = 1/21
hX, X, X 3 Y 2 i = 1
hX, X 2 , X 2 Y 2 i = 1
hY, Y, Y, X 5 Y 2 i = 2/7
hX, X 3 , XY 2 i = 1
hX, Y, X 4 Yi = 1
4
2
3
hY, Y, XY, X 4 Y 2 i = 5/21
hX, X , Y i = 1
hX, XY, X Yi = 1
2
5
6
hY, Y, X 2 Y, X 3 Y 2 i = 4/21
hX, XY, X Yi = 1
hX, X , X i = 3
hY, Y, X 3 Y, X 2 Y 2 i = 1/7
hX, X 2 Y, X 2 Yi = 1 hX 2 , X 2 , XY 2 i = 1
hY, Y, X 4 Y, XY 2 i = 2/21
hX 2 , X 3 , Y 2 i = 1
hX 2 , Y, X 3 Yi = 1
hY, Y, Y 2 , X 5 Yi = 1/21
hX 2 , X 4 , X 6 i = 3 hX 2 , XY, X 2 Yi = 1
2
5
5
3
3
6
hX , X , X i = 3 hX , X , X i = 3
5 pt correlators
hX 3 , Y, X 2 Yi = 1
hX 3 , X 4 , X 5 i = 3
hX 3 , XY, XYi = 1
hY, Y, X 5 i = 1
None
hY, X 4 , XYi = 1
hX 4 , X 4 , X 4 i = 3

19

20

AMANDA FRANCIS

Z1,0 = x3 y + y7 , G = hJ = (2/7, 1/7)i

k 7, X = e4J , Y = e2J , Z = y4 e0 , W = xy2 e0 . P = 4/7, c = 8/7.
Relations: X 3 = 3Y 2 , 21ZW = XY 2 , a , 0, X 5 = Y 3 = Z 2 = W 2 = 0, dim = 9.
5 pt correlators
hX, X, X, XY 2 , XY 2 i = 2/49
hX, Y, Y, Y 2 , XY 2 i = 0
hX, Z, Z, Z, XY 2 i = 2a/7
3 pt correlators
hX, Z, W, X 2 , XY 2 i = 1/3087
h1, 1, XY 2 i = 1
h1, X, Y 2 i = 1
hX, Z, W, Y 2 , Y 2 i = 1/1029
h1, Y, XYi = 1
h1, Z, Wi = 1/21
2
2
hX,
W, W, W, XY 2 i = 2l/7
h1, X , X i = 3
hX, Y, Yi = 1
hY, Y, Y, XY, XY 2 i = 2/49
hX, X, X 2 i = 3
hY, Y, Z, W, XY 2 i = 2/1029
hY, Z, W, XY, Y 2 i = 1/1029
hZ, Z, X, X 2 , Y 2 i = 2a/7
4 pt correlators
hZ, Z, Z, XY, XYi = 2a/7
hX, X, Y, XY 2 i = 1/7
hZ, Z, W, W, Y 2 i = 2/64827
hX, X, XY, Y 2 i = 2/7
hW, W, W, X 2 , Y 2 i = 2/(1361367k)
hX, Y, XY, XYi = 1/7
2
2
hW, W, W, XY, XYi = 2l/7
hX, Y, X , Y i = 3/7
hX, Z, W, XYi = 1/147
6 pt correlators
hY, Y, Y, Y 2 i = 1/7
hX, Y, Z, W, XY 2 , XY 2 i = 5/7203
hY, Y, X 2 , XYi = 0
hY, Z, Z, Z, Y 2 , XY 2 i = 6a/49
hY, Z, Z, Zi = a
hY, W, W, W, Y 2 , XY 2 i = 6a/49
hY, Z, W, X 2 i = 1/147
hZ, Z, W, W, XY, XY 2 i = 4/151263
hY, W, W, Wi = 1/(194, 481a)
7 pt correlators
None

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

W1,0 = x4 + y6 , G = hJ = (1/4, 1/6)i

k 7, X = e9J , Y = e2J , Z = e7J , W = xy2 e0 . P = 7/12, c = 7/6.
Relations: XZ = Y 2 , X 2 Z = 24W 2 , X 3 = Y 3 = Z 2 = W 3 = 0, dim = 9.
3 pt correlators
h1, 1, XY 2 i = 1
h1, X, Y 2 i = 1
4 pt correlators
h1, Y, XYi = 1
h1, Z, X 2 i = 1
hX, X, X, XY 2 i = 1/6
h1, W, Wi = 1/24 hX, X, Zi = 1
hX, X, X 2 , Y 2 i = 1/6
hX, Y, Yi = 1
hX, X, XY, XYi = 1/3
5 pt correlators
hX, Y, X 2 , XYi = 1/6
2
hX, Z, Z, XZi = 0
hX, X, Z, XZ, X Zi = 0
hX, Z, X 2 , X 2 i = 0
hX, Y, Y, Y 2 , XY 2 i = 1/24
hX, W, W, X 2 i = 1/144
hX, Y, Z, XY, XY 2 i = 1/24
2
2
hY, Y, Z, Y 2 i = 1/4
hX, Z, Z, X , X Zi = 0
2
hY, Y, X 2 , X 2 i = 1/6
hX, Z, W, W, X Zi = 1/576
hY, Z, Z, XYi = 1/4
hX, W, W, XZ, XZi = 0
hZ, Z, Z, X 2 i = 0
hY, Y, Y, XY, XY 2 i = 1/24
2
2
hZ, Z, W, Wi = 1/96
hY, Y, Z, X , X Zi = 1/24
hY, Y, W, W, XY 2 i = 1/576
hY, W, W, XY, Y 2 i = 1/576
6 pt correlators
hZ, Z, Z, Z, X 2 Zi = 1/8
hX, Z, Z, Z, X 2 Z, X 2 Zi = 0
hZ, Z, Z, XZ, XZi = 1/8
hX, X, W, W, X 2 Z, X 2 Zi = 1/1728
hZ, W, W, X 2 , XZi = 1/576
hY, Y, Z, Z, XY 2 , XY 2 i = 1/24
hZ, W, W, XY, XYi = 0
hZ, Z, W, W, XZ, X 2 Zi = 1/1152
hW, W, W, W, XZi = 1/13824
hW, W, W, W, X 2 , X 2 Zi = 1/27648
7 pt correlators
hZ, W, W, W, W, X 2 Z, X 2 Zi = 1/55296
Q2,0 = x3 + xy4 + yz2 , Gmax = h = (1/3, 11/12, 1/24)i
k 5, X = e7 , Y = e11 . P = 7/24, c = 7/6.
Relations: 2X 7 = Y 3 , X 8 = Y 4 = 0 , dim = 17.
4 pt correlators
3 pt correlators
hX, X, X 6 , X 7 Yi = 2/3
h1, 1, X 7 Yi = 2
h1, X, X 7 Yi = 2
hX, X, X 7 , X 6 Yi = 1/3
h1, X 2 , X 6 Yi = 2
h1, Y, X 7 i = 2
3
5
6
hX, Y, X 7 , X 7 i = 1/3
h1, X , X Yi = 2
h1, XY, X i = 2
4
3
2
5
hX, Y, Y 2 , Y 3 i = 4/3
h1, X , X Yi = 2
h1, X Y, X i = 2
hY, Y, XY, X 7 Yi = 2/3
h1, Y 3 , Y 3 i = 4
hX, X, X 5 Yi = 2
hY, Y, X 2 Y, X 6 Yi = 2/3
hX, X 2 , X 4 Yi = 2
hX, Y, X 6 i = 2
3
3
5
hY, Y, X 3 Y, X 5 Yi = 2/3
hX, X , X Yi = 2
hX, XY, X i = 2
4
2
2
5
hY, Y, X 4 Y, X 4 Yi = 2/3
hX, X , X Yi = 2
hX , Y, X i = 2
hX 2 , X 2 , X 3 Yi = 2 hX 2 , XY, X 4 i = 2
5 pt correlators
hX 2 , X 3 , X 2 Yi = 2
hY, X 3 , X 4 i = 2
2
hY, Y, Y i = 4
hX 3 , X 3 , XYi = 2
None

21

22

AMANDA FRANCIS

Q2,0 = x3 + xy4 + yz2 , G = hJ = (1/3, 1/6, 5/12)i

k 7, X = e10J , Y = e8J , Z = y3 e6J , W = xye6J . P = 7/12, c = 7/6.
Relations: X 3 Y = 8Z 2 = 24W 2 , X 4 = Y 2 = Z 3 = W 3 = 0 a = 1, dim = 10.
3 pt correlators
4 pt correlators
h1, 1, X 3 Yi = 2 h1, X, X 2 Yi = 2
hX, X, X 2 , X 3 Yi = 2/3
h1, Y, X 3 i = 2
h1, X 2 , XYi = 1
hX, X, X 3 , X 2 Yi = 4/3
h1, Z, Zi = 1/4 h1, W, Wi = 1/12
hX, Y, Z, X 2 Yi = 0
2
hX, X, XYi = 2
hX, Y, X i = 1
hX, Y, X 3 , X 3 i = 2/3
5 pt correlators
hX, Z, Z, X 3 i = 1/12
2
3
hX, Z, XY, XYi = 0
hX, X, Z, X Y, X Yi = 0
hX, W, W, X 3 i = 1/36
hX, Y, Y, X 2 Y, X 3 Yi = 2/9
hY, Y, Y, X 2 Yi = 2/3
hX, Y, Z, X 3 , X 3 Yi = 0
3
hY, Y, Z, X 3 i = 0
hX, Z, Z, Z, X Yi = a/72
3
hY,
Y, XY, XYi = 2/3
hX, Z, W, W, X Yi = a/216
hY, X 2 , Z, XYi = 0
hY, Y, Y, X 3 , X 3 Yi = 2/9
hY, Z, Z, Zi = a/48
hY, Y, Z, Z, X 3 Yi = 1/36
hY, Z, W, Wi = a/144
hY, Y, W, W, X 3 Yi = 1/108
hX 2 , X 2 , Z, Zi = 1/12
hY, Z, Z, XY, X 2 Yi = 1/36
hX 2 , X 2 , W, Wi = 1/36
hY, W, W, XY, X 2 Yi = 1/36
6 pt correlators
hX 2 , Z, Z, Z, X 2 Yi = a/72
hX 2 , Z, W, W, X 2 Yi = a/216
hX, Y, Z, Z, X 3 Y, X 3 Yi = 1/108
hZ, Z, Z, XY, X 3 i = a/72
hX, Y, W, W, X 3 Y, X 3 Yi = 1/324
hZ, W, W, XY, X 3 i = a/216
hY, Y, Y, Z, X 3 Y, X 3 Yi = 0
7 pt correlators
hZ, Z, Z, Z, XY, X 3 Yi = 1/288
2
3
3
hZ, Z, Z, Z, X 2 Y, X 2 Yi = 1/288
hX , Z, Z, Z, Z, X Y, X Yi = 1/216
2
3
3
hZ, Z, W, W, XY, X 3 Yi = 1/2592
hX , Z, Z, W, W, X Y, X Yi = 1/1944
hZ, Z, W, W, X 2 Y, X 2 Yi = 1/2592
hX 2 , W, W, W, W, X 3 Y, X 3 Yi = 1/1944
hW, W, W, W, XY, X 3 Yi = 1/2592
hW, W, W, W, X 2 Y, X 2 Yi = 1/2592

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

QT2,0 = x3 y + y4 z + z2 ,

Gmax = hJ = (7/24, 1/8, 1/2)i

k 5, X = e11J , Y = e9J , Z = y3 e8J . P = 5/12, c = 7/6.

Relations: X 4 = 3Y 2 , aX 3 Y = Z 2 , X 8 = Y 3 = Z 3 = 0, a , 0, dim = 14.
4 pt correlators
hX, X, X 2 Y, X 3 Y 2 i = 1/8
hX, X, X 3 Y, X 2 Y 2 i = 1/8
hX, Y, Y 2 , X 2 Y 2 i = 1/12
3 pt correlators
hX, Y, X 3 Y, X 3 Yi = 0
3 2
2 2
hX, Y, XY 2 , XY 2 i = 1/12
h1, 1, X Y i = 1
h1, X, X Y i = 1
2
2
3
hX, Y, X 3 , X 3 Y 2 i = 1/8
h1, X , XY i = 1
h1, Y, X Yi = 1
hX, Z, Z, X 3 Y 2 i = a/8
h1, X 3 , Y 2 i = 1
h1, XY, X 2 Yi = a
hX, Z, YZ, X 3 Yi = a/8
h1, Z, YZi = 1
hX, X, XY 2 i = 1
2
2
2
hX 2 , Z, Z, X 2 Y 2 i = a/8
hX, X , Y i = 1
hX, Y, X Yi = 1
3
3
hY, Y, Y, X 3 Y 2 i = 7/24
hX, XY, XYi = 1 hX, X , X i = 3
hY, Y, XY, X 2 Y 2 i = 5/24
hX 2 , Y, XYi = 1 hX 2 , X 2 , X 3 i = 3
hY, Y, X 2 Y, XY 2 i = 1/8
hY, Z, Zi = a
hY, Y, Y 2 , X 3 Yi = 1/24
5 pt correlators
3 2
3 2
hY, Y, YZ, YZi = a/4
hX, Y, Y, X Y , X Y i = 1/96
hY, Z, Y 2 , YZi = a/24
hY, Z, Z, XY 2 , X 3 Y 2 i = a/96
hZ, Z, X 3 , XY 2 i = a/8
hY, Z, Z, X 2 Y 2 , X 2 Y 2 i = a/96
hZ, Z, XY, X 3 Yi = a/8
hZ, Z, X 2 Y, X 2 Yi = a/8
hZ, Z, Y 2 , Y 2 i = a/24
S 1,0 = x2 y + y2 z + z5 , Gmax = h = (1/20, 9/10, 1/5)i
k 4, X = e7 , Y = e . P = 1/4, c = 6/5.
Relations: X 5 = Y 3 , X 9 = Y 4 = 0, dim = 17.
3 pt correlators
h1, 1, X 3 Y 3 i = 2
h1, X, X 2 Y 3 i = 2
4 pt correlators
3 2
h1, Y, X Y i = 2
h1, X 2 , XY 3 i = 2
hX,
X,
X 3 Y, X 3 Y 3 i = 2/5
h1, XY, X 2 Y 2 i = 2
h1, X 3 , Y 3 i = 2
hX, X, X 3 Y 2 , X 3 Y 2 i = 2/5
h1, Y 2 , X 3 Yi = 2 h1, X 2 Y, XY 2 i = 2
hX, Y, X 4 , X 3 Y 3 i = 2/5
h1, X 4 , X 4 i = 2
hX, X, XY 3 i = 2
hX,
Y, Y 3 , X 2 Y 3 i = 2/5
hX, Y, X 2 Y 2 i = 2
hX, X 2 , Y 3 i = 2
hX, Y, XY 3 , XY 3 i = 1/5
hX, XY, XY 2 i = 2
hX, Y 2 X 2 Yi = 2
hY, Y, Y 2 , X 3 Y 3 i = 8/5
hX, X 3 , X 4 i = 2
hY, Y, X 3 Yi = 2
hY, Y, XY 2 , X 2 Y 3 i = 6/5
hY, X 2 , XY 2 i = 2 hY, XY, X 2 Yi = 2
hY, Y, Y 3 , X 3 Y 2 i = 2/5
hY, X 3 , Y 2 i = 2
hX 2 , X 2 , X 4 i = 2
hY, Y, X 2 Y 2 , XY 3 i = 4/5
hX 2 , XY, Y 2 i = 2 hX 2 , X 3 , X 3 i = 2
hXY, XY, XYi = 2

23

24

AMANDA FRANCIS

S 1,0 = x2 y + y2 z + z5 , G = hJ = (3/10, 2/5, 1/5)i

k 7, X = e2J , Y = e7J , Z = e6J , W = z2 e5J . P = 3/5, c = 6/5.
Relations: 2XZ = aY 2 , X 3 = 2aYZ, XYZ = 10W 2 , X 5 = Y 4 = Z 2 = 0, a , 0, dim = 10.
3 pt correlators
4 pt correlators
h1, 1, XYZi = 1
h1, X, YZi = 1
hX, X, Y, XY 2 i = 1/5
h1, Y, XZi = 1
h1, Z, XYi = 1
hX, X, XY, YZi = 1/5
h1, W, Wi = 1/10 h1, X 2 , X 2 i = 2a
hX, X, XZ, XZi = a/5
hX, Y, Zi = 1
hX, X, X 2 i = 2a
hX, Y, X 2 , YZi = 2/5
hY, Y, Yi = 2/a
hX, Y, XY, XZi = 1/5
5 pt correlators
hX, Z, Z, YZi = 1/10
hX, Z, X 2 , XZi = a/5
hX, X, Z, YZ, XYZi = 1/25
hX, Z, XY, XYi = 0
hX, Y, Y, YZ, XYZi = 0
hX, W, W, XYi = 1/50
hX, Y, Z, XZ, XYZi = 1/50
hY, Y, Z, YZi = 1/(5a)
hX, Z, Z, XY, XYZi = 0
hY, Y, X 2 , XZi = 0
hX, Z, W, W, XYZi = 3/500
hY, Y, XY, XYi = 2/(5a)
hX, W, W, XZ, YZi = 1/500
hY, Z, Z, XZi = 1/5
hY, Y, Y, XZ, XYZi = 2/(25a)
hY, Z, X 2 , XYi = 0
hY, Y, Z, XY, XYZi = 2/(25a)
hY, W, W, X 2 i = 1/50
hY, Y, W, W, XYZi = 1/(125a)
hZ, Z, Z, XYi = 1/10
hY, Z, Z, X 2 , XYZi = 0
hZ, Z, W, Wi = 1/50
hY, Z, Z, YZ, YZi = 1/(25a)
hZ, Z, X 2 , X 2 i = 2/5
hY, W, W, XY, YZi = 1/(250a)
6 pt correlators
hY, W, W, XZ, XZi = 1/250
hZ, Z, Z, Z, XYZi = 3/25
hX, Z, Z, Z, XYZ, XYZi = 0
hZ, Z, Z, XZ, YZi = 2/25
hX, X, Z, Z, XYZ, XYZi = 1/625
hZ, W, W, X 2 , YZi = 1/125
hY, Y, Z, Z, XYZ, XYZi = 4/(125a)
hZ, W, W, XY, XZi = 1/500
hY, Z, W, W, YZ, XYZi = 1/(2500a)
hW, W, W, W, XZi = 1/5000
hZ, Z, W, W, XZ, XYZi = 3/2500
7 pt correlators
hW, W, W, W, X 2 , XYZi = 9/25000
hW, W, W, W, YZ, YZi = 3/(12500a)
hY, W, W, W, W, XYZ, XYZi = 3/(62500a)
E19 = x3 + xy7 , Gmax = hJ = (1/3, 2/21)i
k 5, X = e13J , Y = e11J . P = 2/7, c = 8/7.
Relations: 7X 6 = Y 3 , X 7 = Y 5 = 0, dim = 15.
3 pt correlators
4 pt correlators
h1, 1, X 6 Yi = 1
h1, X, X 5 Yi = 1
hX, X, X 5 , X 6 Yi = 2/21
6
2
4
h1, Y, X i = 1
h1, X , X Yi = 1
hX, X, X 6 , X 5 Yi = 1/21
h1, XY, X 5 i = 1
h1, X 3 , X 3 Yi = 1
hX, Y, X 6 , X 6 i = 1/21
h1, X 2 Y, X 4 i = 1
h1, Y 2 , Y 2 i = 7
hX, Y, Y 2 , X 6 Yi = 1/3
hX, X, X 4 Yi = 1
hX, Y, X 5 i = 1
hY, Y, XY, X 6 Yi = 1/3
hX, X 2 , X 3 Yi = 1
hX, XY, X 4 i = 1
hY,
Y, X 2 Y, X 5 Yi = 1/3
hX, X 3 , X 2 Yi = 1
hX 2 , X 2 , Y 2 i = 1
hY,
Y, X 3 Y, X 4 Yi = 1/3
hX 2 , Y, X 2 Yi = 1 hX 2 , XY, XYi = 1
hX 3 , Y, XYi = 1
hY, X 2 , X 4 i = 1
5 pt correlators
hY, X 3 , X 3 i = 1
hY, Y, Y 2 i = 7
None
2
2
2
2
3
hX , X , X Yi = 1 hX , XY, X i = 1

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

T
Z17
= x3 + xy8 , Gmax = h = (2/3, 1/24)i
k 5, X = e5 , Y = e . P = 7/24, c = 7/6.
Relations: 8X 7 = Y 3 , X 8 = XY 2 = Y 4 = 0, dim = 17.
3 pt correlators
4 pt correlators
h1, 1, X 7 Yi = 1
h1, X, X 6 Yi = 1
hX,
X,
X 6 , X 7 Yi = 1/12
h1, X 2 , X 5 Yi = 1
h1, Y, X 7 i = 1
hX, X, X 7 , X 6 Yi = 1/24
h1, X 3 , X 4 Yi = 1
h1, XY, X 6 i = 1
hX, Y, X 7 , X 7 i = 1/24
h1, X 4 , X 3 Yi = 1
h1, X 2 Y, X 5 i = 1
hX, Y, Y 2 , X 7 Yi = 1/3
h1, Y 2 , Y 2 i = 8
hX, X, X 5 Yi = 1
hY, Y, XY, X 7 Yi = 1/3
hX, X 2 , X 4 Yi = 1
hX, Y, X 6 i = 1
hY, Y, X 2 Y, X 6 Yi = 1/3
hX, X 3 , X 3 Yi = 1
hX, XY, X 5 i = 1
hY, Y, X 3 Y, X 5 Yi = 1/3
hX, X 4 , X 2 Yi = 1 hX 2 , X 2 , X 3 Yi = 1
hY,
Y, X 4 Y, X 4 Yi = 1/3
hX 2 , Y, X 5 i = 1
hX 2 , X 3 , X 2 Yi = 1
hX 2 , XY, X 4 i = 1
hX 3 , Y, XYi = 1
5 pt correlators
hY, Y, Y 2 i = 1
hY, X 3 , X 4 i = 1
None
3
3
hX , X , XYi = 1
T
Z17
= x3 + xy8 , G = hJ = (1/3, 1/12)i
k 7, X = e4J , Y = e2J , Z = y7 e0 , W = xy3 e0 . P = 7/12, c = 7/6.
Relations: X 3 Y = 8Z 2 = 24W 2 , X 4 = Y 2 Z 3 = W 3 , a, b , 0 , dim = 10.
5 pt correlators
hX, Y, Y, X 2 Y, X 3 Yi = 1/36
3 pt correlators
hX, X, Z, X 2 Y, X 3 Yi = 0
h1, 1, X 3 Y i = 1
h1, X, X 2 Yi = 1
hX, Y, Z, X 3 , X 3 Yi = 0
h1, Y, X 3 i = 1
h1, X 2 , XYi = 1
hX, Z, Z, Z, X 3 Yi = a/6
h1, Z, Zi = 1/8 h1, W, Wi = 1/24
hX, Z, W, W, X 3 Yi = b/6
hX, X, XYi = 1
hX, Y, X 2 i = 1
hY, Y, Y, X 3 , X 3 Yi = 1/36
4 pt correlators
hY,
Y, Z, Z, X 3 Yi = 1/288
hX, X, X 2 , X 3 Yi = 1/12
hY, Y, W, W, X 3 Yi = 1/864
hX, X, X 3 , X 2 Yi = 1/6
hY, Z, Z, XY, X 2 Yi = 1/288
hX, Y, X 3 , X 3 i = 1/12
hY, W, W, XY, X 2 Yi = 1/864
hX, Y, Z, X 2 Yi = 0
hX 2 , Z, Z, Z, X 2 Yi = a/6
hX, Z, Z, X 3 i = 1/96
hX 2 , Z, W, W, X 2 Yi = b/6
hX, Z, XY, XYi = 0
hZ, Z, Z, XY, X 3 i = a/6
hX, W, W, X 3 i = 1/288
hZ,
W, W, XY, X 3 i = b/6
hY, Y, Y, X 2 Yi = 1/3
hY, Y, XY, XYi = 1/3
6 pt correlators
hY, Y, Z, X 3 i = 0
hX, Y, Z, Z, X 3 Y, X 3 Yi = 1/3456
2
hY, X , Z, XYi = 0
hX, Y, W, W, X 3 Y, X 3 Yi = 1/10368
hY, Z, Z, Zi = a = 1/192
hY, Y, Y, Z, X 3 Y, X 3 Yi = 0
hY, Z, W, Wi = b = 1/576
hZ,
Z, Z, Z, XY, X 3 Yi = 1/9216
hX 2 , X 2 , Z, Zi = 1/96
hZ, Z, Z, Z, X 2 Y, X 2 Yi = 1/9216
hX 2 , X 2 , W, Wi = 1/288
hZ, Z, W, W, XY, X 3 Yi = 1/82944
hZ, Z, W, W, X 2 Y, X 2 Yi = 1/82944
7 pt correlators
hW, W, W, W, XY, X 3 Yi = 1/82944
None
hW, W, W, W, X 2 Y, X 2 Yi = 1/82944

25

26

AMANDA FRANCIS

Z19 = x3 y + y9 , Gmax = hJ = (8/27, 1/9)i

k 5, X = e11J , Y = e19J . P = 1/3, c = 32/27.
Relations: X 9 = 3Y 2 , X 16 = Y 3 = 0, dim = 25.
4 pt correlators
3 pt correlators
hX, X, X 7 Y, X 7 Y 2 i = 1/9
7 2
6 2
h1, 1, X Y i = 1
h1, X, X Y i = 1
hX, X 2 , X 6 Y, X 7 Y 2 i = 1/9
h1, X 2 , X 5 Y 2 i = 1
h1, X 3 , X 4 Y 2 i = 1
hX, X 2 , X 7 Y, X 6 Y 2 i = 1/9
h1, X 4 , X 3 Y 2 i = 1
h1, Y, X 7 Y 2 i = 1
hX, X 3 , X 5 Y, X 7 Y 2 i = 1/9
h1, X 5 , X 2 Y 2 i = 1
h1, XY, X 6 Yi = 1
hX, X 3 , X 6 Y, X 6 Y 2 i = 1/9
h1, X 6 , XY 2 i = 1
h1, X 2 Y, X 5 Yi = 1
hX, X 3 , X 7 Y, X 5 Y 2 i = 1/9
h1, X 7 , Y 2 i = 1
h1, X 3 Y, X 4 Yi = 1
hX, X 4 , X 4 Y, X 7 Y 2 i = 1/9
h1, X 8 , X 8 i = 3
hX, X, X 5 Y 2 i = 1
hX, X 4 , X 5 Y, X 6 Y 2 i = 1/9
hX, X 2 , X 4 Y 2 i = 1
hX, X 3 , X 3 Y 2 i = 1
hX, X 4 , X 6 Y, X 5 Y 2 i = 1/9
hX, X 4 , X 2 Y 2 i = 1
hX, Y, X 6 Yi = 1
hX, X 4 , X 7 Y, X 4 Y 2 i = 1/9
hX, X 5 , XY 2 i = 1
hX, XY, X 5 Yi = 1
2
, X 2 , X 5 Y, X 7 Y 2 i = 1/9
hX
hX, X 6 , Y 2 i = 1
hX, X 2 Y, X 4 Yi = 1
2
hX , X 2 , X 6 Y, X 6 Y 2 i = 2/9
hX, X 7 , X 8 i = 3
hX, X 3 Y, X 4 Yi = 1
hX 2 , X 2 , X 7 Y, X 5 Y 2 i = 1/9
hX 2 , X 2 , X 3 Y 2 i = 1 hX 2 , X 3 , X 2 Y 2 i = 1
hX 2 , X 3 , X 4 Y, X 7 Y 2 i = 1/9
hX 2 , X 4 , XY 2 i = 1
hX 2 , Y, X 5 Yi = 1
hX 2 , X 3 , X 5 Y, X 6 Y 2 i = 2/9
hX 2 , X 5 , Y 2 i = 1
hX 2 , XY, X 4 Yi = 1
hX 2 , X 3 , X 6 Y, X 5 Y 2 i = 2/9
hX 2 , X 6 , X 8 i = 3 hX 2 , X 2 Y, X 3 Yi = 1
hX 2 , X 3 , X 7 Y, X 4 Y 2 i = 1/9
hX 2 , X 7 , X 7 i = 3 hX 3 , X 3 , XY 2 i = 1
hX 2 , X 4 , X 3 Y, X 7 Y 2 i = 1/9
hX 3 , X 4 , Y 2 i = 1
hX 3 , Y, X 4 Yi = 1
hX 2 , X 4 , X 4 Y, X 6 Y 2 i = 2/9
hX 3 , X 5 , X 8 i = 3 hX 3 , XY, X 3 Yi = 1
hX 2 , X 4 , X 5 Y, X 5 Y 2 i = 2/9
hX 3 , X 6 , X 7 i = 3 hX 3 , X 2 Y, X 2 Yi = 1
hX 2 , X 4 , X 6 Y, X 4 Y 2 i = 2/9
hX 4 , X 4 , X 8 i = 3
hX 4 , Y, X 3 Yi = 1
hX 2 , X 4 , X 7 Y, X 3 Y 2 i = 1/9
hX 4 , X 5 , X 7 i = 3 hX 4 , XY, X 2 Yi = 1
hX 3 , X 3 , X 3 Y, X 7 Y 2 i = 1/9
hX 4 , X 6 , X 6 i = 3
hY, Y, X 7 i = 1
hX 3 , X 3 , X 4 Y, X 6 Y 2 i = 2/9
hY, X 5 , X 2 Yi = 1
hY, XY, X 6 i = 1
hX 3 , X 3 , X 5 Y, X 5 Y 2 i = 1/3
hX 5 , X 5 , X 6 i = 3
hX 5 , XY, XYi = 1
hX 3 , X 3 , X 6 Y, X 4 Y 2 i = 2/9
hX 3 , X 3 , X 7 Y, X 3 Y 2 i = 1/9
5 pt correlators
hX 3 , X 4 , X 2 Y, X 7 Y 2 i = 1/9
hX 4 , X 4 , X 4 , X 6 Y 2 , X 7 Y 2 i = 4/81
hX 3 , X 4 , X 3 Y, X 6 Y 2 i = 2/9
hX 3 , X 4 , X 4 , X 7 Y 2 , X 7 Y 2 i = 2/81
hX 3 , X 4 , X 4 Y, X 5 Y 2 i = 1/3
hX 3 , X 4 , X 5 Y, X 4 Y 2 i = 1/3
hX 3 , X 4 , X 6 Y, X 3 Y 2 i = 2/9
hX 3 , X 4 , X 7 Y, X 2 Y 2 i = 1/9
hX 4 , X 4 , XY, X 7 Y 2 i = 1/9
hX 4 , X 4 , X 2 Y, X 6 Y 2 i = 2/9
hX 4 , X 4 , X 3 Y, X 5 Y 2 i = 1/3
hX 4 , X 4 , X 4 Y, X 4 Y 2 i = 4/9
hX 4 , X 4 , X 5 Y, X 3 Y 2 i = 1/3
hX 4 , X 4 , X 6 Y, X 2 Y 2 i = 2/9
hX 4 , X 4 , X 7 Y, XY 2 i = 1/9

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

T
Z19
= x3 + xy9 , Gmax = hJ = (1/3, 2/27)i
k 5, X = e16J , Y = e14J . P = 8/27, c = 32/27.
Relations: 9X 8 = Y 3 , X 9 = Y 5 = 0, dim = 19.
3 pt correlators
h1, 1, X 8 Yi = 1
h1, X, X 7 Yi = 1
2
6
4 pt correlators
h1, X , X Yi = 1
h1, Y, X 8 i = 1
hX, X, X 7 , X 8 Yi = 2/27
h1, X 3 , X 5 Yi = 1
h1, XY, X 7 i = 1
hX, X, X 8 , X 7 Yi = 1/27
h1, X 4 , X 4 Yi = 1
h1, X 2 Y, X 6 i = 1
5
3
2
2
hX, Y, X 8 , X 8 i = 1/27
h1, X , X Yi = 1
h1, Y , Y i = 9
hX, Y, Y 2 , X 8 Yi = 1/3
hX, X, X 6 Yi = 1
hX, X 2 , X 5 Yi = 1
hY, Y, XY, X 8 Yi = 1/3
hX, Y, X 7 i = 1
hX, X 2 , X 5 Yi = 1
6
3
4
hY, Y, X 2 Y, X 7 Yi = 1/3
hX, XY, X Yi = 1 hX, X , X Yi = 1
2
5
4
3
hY, Y, X 3 Y, X 6 Yi = 1/3
hX, X Y, X i = 1 hX, X , X Yi = 1
hY, Y, X 4 Y, X 5 Yi = 1/3
hX 2 , X 2 , X 4 Yi = 1
hX 2 , Y, X 6 i = 1
5 pt correlators
hX 2 , X 3 , X 3 Yi = 1 hX 2 , XY, X 5 i = 1
hX 2 , X 4 , X 2 Yi = 1
hY, Y, Y 2 i = 9
None
hY, X 3 , X 5 i = 1
hY, X 4 , X 4 i = 1
hX 3 , X 3 , X 2 Yi = 1 hX 3 , XY, X 4 i = 1

W17 = x4 + Xy5 , Gmax = hJ = (1/4, 3/20)i

k 4, X = e14J , Y = e9J . P = 1/5, c = 6/5.
Relations: X 4 = 5Y 4 , X 7 = Y 5 = 0, dim = 16.
3 pt correlators
h1, 1, X 2 Y 4 i = 1
h1, X, XY 4 i = 1
4 pt correlators
h1, Y, X 2 Y 3 i = 1
h1, X 2 , Y 4 i = 1
3
h1, XY, XY i = 1 h1, Y 2 , X 2 Y 2 i = 1
hX, X, X 2 Y, X 2 Y 4 i = 1/4
h1, X 2 Y, Y 3 i = 1 h1, XY 2 , XY 2 i = 1
hX, X, X 2 Y 2 , X 2 Y 3 i = 1/4
h1, X 3 , X 3 i = 5
hX, X, Y 4 i = 1
hX, Y, X 3 , X 2 Y 4 i = 1/4
hX, Y, XY 3 i = 1
hX, X 2 , X 3 i = 5
hX, Y, Y 4 , XY 4 i = 1/20
hX, XY, Y 3 i = 1
hX, Y 2 , XY 2 i = 1
hY, Y, Y 3 , X 2 Y 4 i = 3/20
2 2
2
3
hY, Y, X Y i = 1
hY, X , Y i = 1
hY, Y, XY 3 , XY 4 i = 1/10
hY, XY, XY 2 i = 1
hY, Y 2 , X 2 Yi = 1
hY, Y, Y 4 , X 2 Y 3 i = 1/20
2
2
2
2
2
hX , Y , Yi = 1 hX , X , X i = 5
hXY, XY, Y 2 i = 1
T
W17
= x4 y + y5 , Gmax = h = (1/20, 4/5)i
k 4, X = e3 , Y = e9 . P = 1/4, c = 6/5.
Relations: X 5 = 4Y 3 , X 8 = Y 4 = 0, dim = 17.
3 pt correlators
h1, 1, X 3 Y 3 i = 1
h1, X, X 2 Y 3 i = 1
4 pt correlators
3 2
h1, Y, X Y i = 1
h1, X 2 , XY 3 i = 1
hX, X, X 3 Y, X 3 Y 3 i = 1/5
2 2
3
3
h1, XY, X Y i = 1
h1, X , Y i = 1
hX, X, X 3 Y 2 , X 3 Y 2 i = 1/5
h1, Y 2 , X 3 Yi = 1
h1, X 2 Y, XY 2 i = 1
hX, Y, X 4 , X 3 Y 3 i = 1/5
h1, X 4 , X 4 i = 4
hX, X, XY 3 i = 1
hX,
Y, Y 3 , X 2 Y 3 i = 1/20
hX, Y, X 2 Y 2 i = 1
hX, X 2 , Y 3 i = 1
hX, Y, XY 3 , XY 3 i = 1/20
hX, XY, XY 2 i = 1
hX, X 3 , X 4 i = 4
hY, Y, Y 2 , X 3 Y 3 i = 1/5
hX, Y 2 , X 2 Yi = 1
hY, Y, X 3 Yi = 1
hY,
Y, XY 2 , X 2 Y 3 i = 3/20
hY, X 2 , XY 2 i = 1
hY, XY, X 2 Yi = 1
hY,
Y, Y 3 , X 3 Y 2 i = 1/20
hY, X 3 , Y 2 i = 1
hX 2 , XY, Y 2 i = 1
hY, Y, X 2 Y 2 , XY 3 i = 1/20
hX 2 , X 2 , X 4 i = 4 hX 2 , X 3 , X 3 i = 4
hXY, XY, XYi = 1

27

28

AMANDA FRANCIS
T
W17
= x4 y + y5 , G = h = (1/10, 3/5)i
k 7, X = e , Y = e4 , Z = e7 , W = xy2 e0 . P = 3/5, c = 6/5.
Relations: XZ = Y 2 , X 3 = 4YZ, XYZ = 20W 2 , X 5 = Y 5 = Z 2 = W 3 = 0, dim = 10.
3 pt correlators
4 pt correlators
h1, 1, XYZi = 1
h1, X, YZi = 1
hX, X, Y, XYZi = 1/5
h1, Y, XZi = 1
h1, Z, XYi = 1
hX, X, XY, YZi = 1/5
h1, W, Wi = 1/20 h1, X 2 , X 2 i = 4
hX, X, XZ, XZi = 2/5
2
hX, X, X i = 4
hX, Y, Zi = 1
hX, Y, XY, XZi = 1/5
hY, Y, Yi = 1
hX, Y, X 2 , YZi = 2/5
5 pt correlators
hX, Z, Z, YZi = 1/5
hX, X, Z, YZ, XYZi = 1/25
hX, Z, XY, XYi = 0
hX, Y, Y, YZ, XYZi = 0
hX, Z, X 2 , XZi = 2/5
hX, Y, Z, XZ, XYZi = 1/50
hX, W, W, XYi = 1/100
hX, Z, Z, XY, XYZi = 0
hY, Y, Z, YZi = 1/10
hX, Z, W, W, XYZi = 3/1000
hY, Y, XY, XYi = 1/5
hX, W, W, XZ, YZi = 1/1000
hY, Y, X 2 , XZi = 0
hY, Y, Y, XZ, XYZi = 1/25
hY, Z, Z, XZi = 1/5
hY, Y, Z, XY, XYZi = 1/25
hY, Z, X 2 , XYi = 0
hY, Y, W, W, XYZi = 1/500
hY, W, W, X 2 i = 1/100
hY, Z, Z, YZ, YZi = 1/50
hZ, Z, Z, XYi = 1/10
hY, Z, Z, X 2 , XYZi = 0
hZ, Z, W, Wi = 1/100
hY, W, W, XY, YZi = 1/1000
hZ, Z, X 2 , X 2 i = 4/5
hY, W, W, XZ, XZi = 1/500
6 pt correlators
hZ, Z, Z, Z, XYZi = 3/25
hX, X, W, W, XYZ, XYZi = 1/1250
hZ, Z, Z, XZ, YZi = 2/25
hX, Z, Z, Z, XYZ, XYZi = 0
hZ, W, W, X 2 , XZi = 1/250
hY, Y, Z, Z, XYZ, XYZi = 2/125
hZ, W, W, XY, XZi = 1/1000
hY, Z, W, W, YZ, XYZi = 1/10000
hW, W, W, W, XYZi = 1/20000
hZ, Z, W, W, XZ, XYZi = 3/5000
7 pt correlators
hW, W, W, W, X 2 , XYZi = 9/100000
hY, W, W, W, W, XYZ, XYZi = 3/500000
hW, W, W, W, YZ, YZi = 3/100000
hZ, Z, W, W, W, XYZ, XYZi = 0

Q17 = x3 + xy5 + yz2 , Gmax = hJ = (1/3, 2/15, 13/30)i

k 5, X = e10J , Y = e8J . P = 3/10, c = 6/5.
Relations: 5X 9 = 2Y 3 , X 10 = Y 4 = 0, dim = 21.
3 pt correlators
h1, 1, X 9 Yi = 2
h1, X, X 8 Yi = 2
h1, X 2 , X 7 Yi = 2
h1, Y, X 9 i = 2
4 pt correlators
3
6
h1, X , X Yi = 2
h1, XY, X 8 i = 2
hX,
X,
X 8 , X 9 Yi = 8/15
h1, X 4 , X 5 Yi = 2
h1, X 5 , X 4 Yi = 2
hX, X, X 9 , X 8 Yi = 4/15
h1, X 2 Y, X 7 i = 2
h1, X 3 Y, X 6 i = 2
hX, Y, Y 2 , X 9 Yi = 2/3
h1, Y 2 , Y 2 i = 5
hX, X, X 7 Yi = 2
hX,
Y, X 9 , X 9 i = 4/15
hX, X 2 , X 6 Yi = 2
hX, Y, X 8 i = 2
hY,
Y,
XY, X 9 Yi = 2/3
hX, X 3 , X 5 Yi = 2
hX, XY, X 7 i = 2
hY, Y, X 2 Y, X 8 Yi = 2/3
hX, X 4 , X 4 Yi = 2
hX, X 5 , X 3 Yi = 2
hY, Y, X 3 Y, X 7 Yi = 2/3
hX, X 2 Y, X 6 i = 2
hX 2 , X 2 , X 5 Yi = 2
hY, Y, X 4 Y, X 6 Yi = 2/3
hX 2 , Y, X 7 i = 2
hX 2 , X 3 , X 4 Yi = 2
hY, Y, X 5 Y, X 5 Yi = 2/3
hX 2 , XY, X 6 i = 2 hX 2 , X 4 , X 3 Yi = 2
5 pt correlators
hX 2 , X 2 Y, X 5 i = 2
hY, Y, Y 2 i = 5
None
3
6
4
5
hY, X , X i = 2
hY, X , X i = 2
3
3 3
3
5
hX , X X Yi = 2
hX , XY X i = 2
hX 3 , X 4 X 2 Yi = 2
hXY, X 4 X 4 i = 2

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

QT17 = x3 y + y5 z + z2 , Gmax = h = (1/30, 9/10, 1/2)i

k 5, X = e7 , Y = e19 , Z = y4 e10 . P = 1/3, c = 6/5.
Relations: X 5 = 3Y 2 , aX 4 Y = Z 2 , X 10 = Y 3 = Z 3 = 0, a , 0, dim = 17.
4 pt correlators
3 pt correlators
hX, X, X 3 Y, X 4 Y 2 i = 1/10
h1, 1, X 4 Y 2 i = 1
h1, X, X 3 Y 2 i = 1
hX, X, X 4 Y, X 3 Y 2 i = 1/10
h1, X 2 , X 2 Y 2 i = 1
h1, Y, X 4 Yi = 1
hX, Y, Y 2 , X 3 Y 2 i = 1/15
3
2
3
h1, X , XY i = 1
h1, XY, X Yi = 1
hX, Y, XY 2 , X 2 Y 2 i = 1/15
4
2
2
2
h1, X , Y i = 1
h1, X Y, X Yi = 1
hX, Y, X 4 Y, X 4 Yi = 0
h1, Z, YZi = a
hX, X, X 2 Y 2 i = 1
hX, Y, X 4 , X 4 Y 2 i = 1/10
hX, X 2 , XY 2 i = 1
hX, Y, X 3 Yi = 1
hX, Z, Z, X 4 Y 2 i = a/10
3
2
2
hX, X , Y i = 1
hX, XY, X Yi = 1
hX, Z, YZ, X 4 Yi = a/10
4
4
2
2
2
hX, X , X i = 3
hX , X , Y i = 1
hX 2 , Z, Z, X 3 Y 2 i = a/10
hX 2 , Y, X 2 Yi = 1
hX 2 , X 3 , X 4 i = 3
hY, Y, Y, X 4 Y 2 i = 3/10
hY, Y, X 4 i = 1
hX 2 , XY, XYi = 1
hY, Y, XY, X 3 Y 2 i = 7/30
hY, X 3 , XYi = 1
hY, Z, Zi = a
hY, Y, X 2 Y, X 2 Y 2 i = 1/6
3
3
3
hX , X , X i = 3
hY, Y, Y 2 , X 4 Yi = 1/30
hY, Y, X 3 Y, XY 2 i = 1/10
hY, Y, YZ, YZi = a/3
hY, Z, Y 2 , YZi = a/30
hX 3 , Y, Y 2 , X 4 Yi = a/10
5 pt correlators
hZ, Z, XY, X 4 Yi = a/10
hX, Y, Y, X 4 Y 2 , X 4 Y 2 i = 1/150
hZ, Z, X 4 , XY 2 i = a/10
hY, Z, Z, XY 2 , X 4 Y 2 i = a/150
2 2
3 2
hZ, Z, X 2 Y, X 3 Yi = a/10
hY, Z, Z, X Y , X Y i = a/150
hZ, Z, Y 2 , Y 2 i = a/30
S 16 = x2 y + xz4 + y2 z, Gmax = hJ = (5/17, 7/17, 3/17)i
k 5, X = e8J , Y = e7J , Z = e6J . P = 7/17, c = 21/17.
Relations: 2XZ = Y 2 , X 4 = 2YZ, 4X 3 Y = Z 2 , X 4 = Y 3 = Z 3 = 0, dim = 16.
4 pt correlators
3 pt correlators
hX, X, X 2 Y, X 3 YZi = 3/17
hY, Y, YZ, X 3 Zi = 1/17
h1, 1, X 3 YZi = 1
hX, X, X 3 Y, X 2 YZi = 1/17
hY, Y, X 2 Z, XYZi = 3/17
h1, X, X 2 YZi = 1
hX, X, X 3 Z, X 3 Zi = 2/17
hY, Y, X 3 Y, X 3 Yi = 2/17
h1, Y, X 3 Zi = 1
2
2
hX, Y, YZ, X YZi = 2/17
hY, Z, XZ, X 4 Yi = 4/17
h1, X , XYZi = 1
3
3
3
hX,
Y,
X
Y,
X
Zi
=
1/17
hY,
Z, YZ, X 4 Yi = 1/17
h1, Z, X Yi = 1
hX, Y, XYZ, XYZi = 2/17
hY, Z, X 2 Z, X 4 Yi = 4/17
h1, XY, X 2 Zi = 1
hX, Y, X 3 , X 3 YZi = 3/17
hY, Z, XY, X 4 Yi = 5/17
h1, X 3 , YZi = 1
2
2
hX, Z, XZ, X YZi = 1/17
hY, Z, X 2 Y, X 4 Yi = 3/17
h1, XZ, X Yi = 1
3
hX,
Z,
YZ,
X
Zi
=
1/17
hX 2 , Z, Z, X 2 YZi = 6/17
hX, X, XYZi = 1
hX, Z, X 2 Z, XYZi = 1/17
hZ, Z, XY, X 3 Zi = 1/17
hX, Y, X 2 Zi = 1
hX, Z, X 3 Y, X 3 Yi = 2/17
hZ, Z, X 3 , XYZi = 7/17
hX, X 2 , YZi = 1
3
2
hX, Z, Z, X YZi = 5/17
hZ, Z, XZ, X 3 Yi = 3/17
hX, ZX Yi = 1
hY, Y, Z, X 3 YZi = 7/17
hZ, Z, X 2 Y, X 2 Zi = 2/17
hX, XY, XZi = 1
hY, Y, XZ, X 2 YZi = 5/17
hZ, Z, YZ, YZi = 4/17
hX, X 3 , Y 3 i = 2
2
5 pt correlators
hY, X , XZi = 1
hY, Z, X 3 i = 1
hX, X, Z, X 3 YZ, X 3 YZi = 4/289
hY, Y, X 2 Yi = 2
hX, Y, Y, X 3 YZ, X 3 YZi = 8/289
hY, XY, XYi = 2
hY, Z, Z, XYZ, X 3 YZi = 4/289
hX 2 , Z, XYi = 1
hY, Z, Z, X 2 YZ, X 2 YZi = 8/289
2
2
3
hX , X , X i = 2
hZ, Z, Z, X 2 Z, X 3 YZi = 20/289
hZ, Z, Zi = 4
hZ, Z, Z, X 3 Z, X 2 YZi = 12/289

29

30

AMANDA FRANCIS

S 17 = x2 y + y2 z + z6 , Gmax = hJ = (7/24, 5/12, 1/6)i

k 5, X = e8J , Y = e7J . P = 1/4, c = 5/4.
Relations: X 6 = Y 3 , X 11 = Y 4 = 0, dim = 21.
3 pt correlators
4 pt correlators
h1, 1, X 10 i = 2
h1, X, X 9 i = 2
hX,
X,
X 4 Y, X 10 i = 1/3
h1, Y, X 4 Y 2 i = 2
h1, X 2 , X 8 i = 2
3 2
hX, X, X Y , X 3 Y 2 i = 1/(3k)
h1, XY, X 3 Y 2 i = 2
h1, X 3 , X 7 i = 2
hX, Y, X 5 , X 10 i = 1/3
h1, X 2 Y, X 2 Y 2 i = 2
h1, X 4 , X 6 i = 2
hX, Y, X 6 , X 9 i = 1/3
h1, Y 2 , X 4 Yi = 2
h1, X 3 Y, XY 2 i = 2
hX, Y, X 7 , X 8 i = 1/3
h1, X 5 , X 5 i = 2
hX, X, X 8 i = 2
hY, Y, Y 2 , X 10 i = 5/3
hX, Y, X 3 Yi = 2
hX, X 2 , X 7 i = 2
hY,
Y, XY 2 , X 9 i = 1/3
hX, XY, X 2 Y 2 i = 2
hX, X 3 , X 6 i = 2
hY,
Y, X 6 , X 3 Y 2 i = 1/3
hX, X 2 Y, XY 2 i = 2
hX, Y 2 , X 3 Yi = 2
hY, Y, X 2 Y 2 , X 8 i = 1
hX, X 4 , X 5 i = 2
hY, Y, X 4 Yi = 2
hY, Y, Y 2 , X 4 Yi = 1/18
hY, XY, X 3 Yi = 2
hY, X 2 , X 2 Y 2 i = 2
hY,
Y, X 7 , X 3 Y 2 i = 1/3
hY, X 3 , XY 2 i = 2
hY, X 2 Y, X 2 Yi = 2
hY, X 4 , Y 2 i = 2
hX 2 , X 2 , X 6 i = 2
hX 2 , XY, XY 2 i = 2
hX 2 , X 3 , X 5 i = 2
5 pt correlators
hX 2 , X 2 Y, Y 2 i = 2
hX 2 , X 4 , X 4 i = 2
None
3
2
2
hXY, X , Y i = 2
hXY, XY, X Yi = 2
3
3
4
hX , X , X i = 2

4.3. FJRW Theories which we still cannot compute. Unfortunately, there are some Amodels which we are still unable to compute. In these examples, there are not enough
relations between concave correlators (whose values we can compute) and other correlators
(whose values we can only find using the Reconstruction Lemma).
W
P8 = x3 + y3 + z3
U12 = x3 + y3 + z4
T
S 1,0
= x2 + xz2 + y5 z
Z17 = x3 y + y8
Z18 = x3 y + xy6
T
W17
= x4 y + y5
T
Q17 = x3 y + y5 z + z2
T
S 17
= x2 + y6 z + z2 x
T
S 17
= x2 + y6 z + z2 x

G
h(1/3, 0, 0), (0, 1/3, 1/3)i
hJi
hJimax
hJimax
hJimax
hJi
hJi
Gmax
hJi

dimension
8
12
14
22
18
8
7
17
7

Remark 4.1. The code which we used to make our computations is available by email
request from the author. We made the computations in SAGE [18]. The SAGE computations depend on code written by Drew Johnson [13] for computing intersection numbers
of classes on M g,n .
References
[1] H. Fan, T. J. Jarvis, and Y. Ruan. The Witten equation, mirror symmetry and quantum singularity theory.
ArXiv e-prints, December 2007.
[2] H. Fan, T. J. Jarvis, and Y. Ruan. Geometry and analysis of spin equations. Communications on Pure and
Applied Mathematics, 61(6):745788, 2008.
[3] H. Fan, T. J. Jarvis, and Y. Ruan. The Witten equation and its virtual fundamental cycle. ArXiv e-prints,
December 2007.
[4] M. Krawitz. FJRW rings and Landau-Ginzburg Mirror Symmetry. ArXiv e-prints, June 2009.
[5] A. Francis, T. Jarvis, D. Johnson, and R. Suggs. Landau-Ginzburg Mirror Symmetry for Orbifolded Frobenius Algebras. ArXiv e-prints, November 2011.

COMPUTATIONAL TECHNIQUES IN FJRW THEORY

31

[6] M. Krawitz and Y. Shen. Landau-Ginzburg/Calabi-Yau Correspondence of all Genera for Elliptic Orbifold
$\mathbb{p}1$. ArXiv e-prints, June 2011.
[7] J. Guere. A LandauGinzburg mirror theorem without concavity. ArXiv e-prints, July 2013.
[8] M. Krawitz, N. Priddis, P. Acosta, N. Bergin, and H. Rathnakumara. FJRW-Rings and Mirror Symmetry.
Communications in Mathematical Physics, 296:145174, May 2010.
[9] V. I. Arnold, V. V. Goryunov, O. V. Lyashko, and V. A. Vasilliev. Singularity theory. I. In Dynamical Systems.
VI, volume 6 of Encycl. Math. Sci., pages iv+245. Springer-Verlag, New York, 1993.
[10] A. Chiodo. Towards an enumerative geometry of the moduli space of twisted curves and r-th roots. ArXiv
Mathematics e-prints, July 2006.
[11] C. Faber. Algorithms for computing intersection numbers on moduli spaces of curves, with an application
to the class of the locus of Jacobians. In eprint arXiv:alg-geom/9706006, page 6006, June 1997.
[12] E. Arbarello and M. Cornalba. Combinatorial and algebro-geometric cohomology classes on the moduli
spaces of curves. In eprint arXiv:alg-geom/9406008, page 6008, June 1994.
[13] Drew Johnson. Top Intersections on Mbarg,n, August 2011. https://bitbucket.org/drew j/top-intersectionson-mbar g-n.
[14] C. Li, S. Li, K. Saito, and Y. Shen. Mirror symmetry for exceptional unimodular singularities. ArXiv e-prints,
May 2014.
[15] K Saito. Primitive forms for a universal unfolding of a function with an isolated critical point. J. Fac. Sci.
Univ. Tokyo Sect. IA 28, 1982.
[16] D. Edidin, T. J. Jarvis, and T. Kimura. Chern Classes and Compatible Power Operations in Inertial K-theory.
ArXiv e-prints, September 2012.
[17] H. Fan, A. Francis, T. J. Jarvis, E. Merrell, and Y. Ruan. Wittens D 4 Integrable Hierarchies Conjecture.
ArXiv e-prints, August 2010.
[18] W. A. Stein et al. Sage Mathematics Software (Version 6.1). The Sage Development Team, 2014.
http://www.sagemath.org.
Department of Mathematics, Brigham Young University, Provo, UT 84602, USA
E-mail address: amanda@mathematics.byu.edu