Proof of the Umbral Moonshine Conjecture

The Umbral Moonshine Conjectures assert that there are infinite-dimensional graded modules, for prescribed finite groups, whose McKay-Thompson series are certain distinguished mock modular forms. Gannon has proved this for the special case involving the largest sporadic simple Mathieu group. Here we establish the existence of the umbral moonshine modules in the remaining 22 cases.


1 Introduction and Statement of Results
Monstrous moonshine relates distinguished modular functions to the representation theory of the Monster, M, the largest sporadic simple group. This theory was inspired by the famous observations of McKay and Thompson in the late 1970s [17,49]  Thompson conjectured that there is a graded infinite-dimensional M-module satisfying dim(V ♮ n ) = c(n). For g ∈ M, he also suggested [48] to consider the graded-trace functions T g (τ ) := ∞ n=−1 tr(g|V ♮ n )q n , now known as the McKay-Thompson series, that arise from the conjectured M-module V ♮ . Using the character table for M, it was observed [17,48] that the first few coefficients of each T g (τ ) coincide with those of a generator for the function field of a discrete group Γ g < SL 2 (R), leading Conway and Norton [17] to their famous Monstrous Moonshine Conjecture: This is the claim that for each g ∈ M there is a specific genus zero group Γ g such that T g (τ ) is the unique normalized hauptmodul for Γ g , i.e., the unique Γ g -invariant holomorphic function on H which satisfies T g (τ ) = q −1 + O(q) as ℑ(τ ) → ∞.
In a series of ground-breaking works, Borcherds introduced vertex algebras [2], and generalized Kac-Moody Lie algebras [3,4], and used these notions to prove [5] the Monstrous Moonshine Conjecture of Conway and Norton. He confirmed the conjecture for the module V ♮ constructed by Frenkel, Lepowsky, and Meurman [28][29][30] in the early 1980s. These results provide much more than the predictions of monstrous moonshine. The M-module V ♮ is a vertex operator algebra, one whose automorphism group is precisely M. The construction of Frenkel, Lepowsky and Meurman can be regarded as one of the first examples of an orbifold conformal field theory. (Cf. [21].) Here the orbifold in question is the quotient R 24 /Λ 24 /(Z/2Z), of the 24-dimensional torus Λ 24 ⊗ Z R/Λ 24 ≃

3
In 2010, Eguchi, Ooguri, and Tachikawa reignited moonshine with their observation [26] that dimensions of some representations of M 24 , the largest sporadic simple Mathieu group (cf. e.g. [19,20]), are multiplicities of superconformal algebra characters in the K3 elliptic genus. This observation suggested a manifestation of moonshine for M 24 : Namely, there should be an infinitedimensional graded M 24 -module whose McKay-Thompson series are holomorphic parts of harmonic Maass forms, the so-called mock modular forms. (See [43,52,53] for introductory accounts of the theory of mock modular forms.) Following the work of Cheng [10], Eguchi and Hikami [25], and Gaberdiel, Hohenegger, and Volpato [31,32], Gannon established the existence of this infinite-dimensional graded M 24 -module in [35].
It is natural to seek a general mathematical and physical setting for these results. Here we consider the mathematical setting, which develops from the close relationship between the monster group M and the Leech lattice Λ 24 . Recall (cf. e.g. [19]) that the Leech lattice is even, unimodular, and positive-definite of rank 24. It turns out that M 24 is closely related to another such lattice. Such observations led Cheng, Duncan and Harvey to further instances of moonshine within the setting of even unimodular positive-definite lattices of rank 24. In this way they arrived at the Umbral Moonshine Conjectures (cf. §5 of [14], §6 of [15], and §2 of [16]), predicting the existence of 22 further, graded infinite-dimensional modules, relating certain finite groups to distinguished mock modular forms.
To explain this prediction in more detail we recall Niemeier's result [41] that there are 24 (up to isomorphism) even unimodular positive-definite lattices of rank 24. The Leech lattice is the unique one with no root vectors (i.e. lattice vectors with norm-square 2), while the other 23 have root systems with full rank, 24. These Niemeier root systems are unions of simple simply-laced root systems with the same Coxeter numbers, and are given explicitly as A 24 1 , A 12 2 , A 8 3 , A 6 4 , A 4 6 , A 2 12 , A 4 5 D 4 , A 2 7 D 2 5 , A 3 8 , A 2 9 D 6 , A 11 D 7 E 6 , A 15 D 9 , A 17 E 7 , A 24 , D 6 4 , D 4 6 , D 3 8 , D 10 E 2 7 , D 2 12 , D 16 E 8 , D 24 , E 4 6 , E 3 8 , in terms of the standard ADE notation. (Cf. e.g. [19] or [37] for more on root systems.) For each Niemeier root system X let N X denote the corresponding unimodular lattice, let W X denote the (normal) subgroup of Aut(N X ) generated by reflections in roots, and define the umbral group of X by setting G X := Aut(N X )/W X .
(1.2) (See §A.1 for explicit descriptions of the groups G X .) Let m X denote the Coxeter number of any simple component of X. An association of distinguished 2m X -vector-valued mock modular forms H X g (τ ) = (H X g,r (τ ))-called umbral McKay-Thompson series-to elements g ∈ G X is described and analyzed in [14][15][16].
In (  2) The vector-valued mock modular forms H X = (H X g,r ) have "minimal" principal parts. This minimality is analogous to the fact that the original McKay-Thompson series T g (τ ) for the Monster are hauptmoduln, and plays an important role in our proof.
Example. Many of Ramanujan's mock theta functions [44] are components of the vector-valued umbral McKay-Thompson series H X g = (H X g,r ). For example, consider the root system X = A 12 2 , whose umbral group is a double cover 2.M 12 of the sporadic simple Mathieu group M 12 . In terms

of Ramanujan's 3rd order mock theta functions
, we have that

6
in this work will ultimately admit an analogous algebraic interpretation. Such a result remains an important goal for future work.
In the statement of Theorem 1.2, replicable means that there are explicit recursion relations for the coefficients of the vector-valued mock modular form in question. For example, we recall the recurrence formula for Ramanujan's third order mock theta function f (q) = ∞ n=0 c f (n)q n that was obtained recently by Imamoglu, Raum and Richter [39]. If n ∈ Q, then let Then for positive integers n, we have that where N := 1 6 (−3a + b − 1) and N := 1 6 (3a + b − 1), and the sum is over integers a, b for which N, N ∈ Z. This is easily seen to be a recurrence relation for the coefficients c f (n). The replicability formulas for all of the H X g,r (τ ) are similar (although some of these relations are slightly more complicated and involve the coefficients of weight 2 cusp forms).
It is important to emphasize that, despite the progress which is represented by our main results, Theorems 1.1 and 1.2, the following important question remains open in general.
Question. Is there a "natural" construction ofǨ X ? IsǨ X equipped with a deeper algebra structure as in the case of the monster module V ♮ of Frenkel, Lepowsky and Meurman?
We remark that this question has been answered positively, recently, in one special case: A vertex operator algebra structure underlying the umbral moonshine moduleǨ X for X = E 3 8 has been described explicitly in [23]. See also [13,24], where the problem of constructing algebraic structures that illuminate the umbral moonshine observations is addressed from a different point of view.
The proof of Theorem 1.1 is not difficult. It is essentially a collection of tedious calculations. We use the theory of mock modular forms and the character table for each G X (cf. §A.2) to solve for the multiplicities of the irreducible G X -module constituents of each homogeneous subspace in the alleged G X -moduleǨ X . To prove Theorem 1.1 it suffices to prove that these multiplicities are non-negative integers. To prove Theorem 1.2 we apply recent work [40] of Mertens on the holomorphic projection of weight 1 2 mock modular forms, which generalizes earlier work [39] of Imamoglu, Raum and Richter.

7
In §2 we recall the facts about mock modular forms that we require, and we prove Theorem 1.2. We prove Theorem 1.1 in §3. The appendices furnish all the data that our method requires. In particular, the umbral groups G X are described in detail in §A, and the umbral McKay-Thompson series H X g (τ ) are given explicitly in §B.

Harmonic Maass forms and Mock modular forms
Here we recall some very basic facts about harmonic Maass forms as developed by Bruinier and Funke [9] (see also [43]). We begin by briefly recalling the definition of a harmonic Maass form of weight k ∈ 1 2 Z and multiplier ν (a generalization of the notion of a Nebentypus). If τ = x + iy with x and y real, we define the weight k hyperbolic Laplacian by Suppose Γ is a subgroup of finite index in SL 2 (Z) and k ∈ 1 2 Z. Then a real-analytic function F (τ ) is a harmonic Maass form of weight k on Γ with multiplier ν if: (a) The function F (τ ) satisfies the weight k modular transformation, and if k ∈ Z + 1 2 , the square root is taken to be the principal branch.
(b) We have that ∆ k F (τ ) = 0, (c) There is a polynomial P F (q −1 ) and a constant c > 0 such that F (τ ) − P F (e −2πiτ ) = O(e −cy ) as τ → i∞. Analogous conditions are required at each cusp of Γ.
We denote the C-vector space of harmonic Maass forms of a given weight k, group Γ and multiplier ν by H k (Γ, ν). If no multiplier is specified, we take ν 0 (γ) := θ(γτ )

Main properties
The Fourier expansion of harmonic Maass forms F (see Proposition 3.2 of [9]) splits into two components. As before, we let q := e 2πiτ .
where F + is the holomorphic part of F , given by

HARMONIC MAASS FORMS AND MOCK MODULAR FORMS
8 and F − is the nonholomorphic part, given by The holomorphic part of a harmonic Maass form is called a mock modular form. The differential operator ξ w := 2iy w ∂ ∂τ (see [9]) defines a surjective map The shadow of a Maass form f (τ ) ∈ H 2−k (Γ, ν), is the cusp form g(τ ) ∈ S k (Γ, ν) (defined, for now, only up to scale) such that ξ 2−k f (τ ) = g ||g|| , where || • || denotes the usual Petersson norm.

Holomorphic projection of weight 1 2 mock modular forms
As noted above, the modular transformations of a weight 1 2 harmonic Maass form may be simplified by multiplying by its shadow to obtain a weight 2 nonholomorphic modular form. One can use the theory of holomorphic projections to obtain explicit identities relating these nonholomorphic modular forms to classical quasimodular forms. In this way, we may essentially reduce many questions about the coefficients of weight 1 2 mock modular forms to questions about weight 2 holomorphic modular forms. The following theorem is a special case of a more general theorem due to Mertens (cf. Theorem 6.3 of [40]). Theorem 2.2 (Mertens). Suppose g(τ ) and h(τ ) are both theta functions of weight 3 2 contained in S 3 2 (Γ, ν g ) and S 3 2 (Γ, ν h ) respectively, with Fourier expansions where each χ i and ψ i is a Dirichlet character. Moreover, suppose h(τ ) is the shadow of a weight 1 2 harmonic Maass form f (τ ) ∈ H 1 2 (Γ, ν h ). Define the function If f (τ )g(τ ) has no singularity at any cusp, then f + (τ )g(τ ) + D f,g (τ ) is a weight 2 quasimodular form. In other words, it lies in the space CE 2 (τ ) ⊕ M 2 (Γ, ν g ν h ).
Two Remarks. 1) These identities give recurrence relations for the weight 1 2 mock modular form f + in terms of the weight 2 quasimodular form which equals f + (τ )g(τ ) + D f,g (τ ). The example after Theorem 1.2 for Ramanujan's third order mock theta function f is an explicit example of such a relation. 2) Theorem 2.2 extends to vector-valued mock modular forms in a natural way.
Proof of Theorem 1.2. Fix a Niemeier lattice and its root system X, and let M = m X denote its Coxeter number. Each H X g,r (τ ) is the holomorphic part of a weight 1 2 harmonic Maass form H X g,r (τ ). To simplify the exposition in the following section, we will emphasize the case that the root system X is of pure A-type. If the root system X is of pure A-type, the shadow function S X g,r (τ ) is given byχ depending on the parity of r is the twisted Euler character given in the appropriate table in §A.3, a character of G X . (If X is not of pure A-type, then the shadow function S X g,r (τ ) is a linear combination of similar functions as described in §B.2.) Given X and g, the symbol n g |h g given in the corresponding table in §A.3 defines the modularity for the vector-valued function ( H X g,r (τ )). In particular, if the shadow (S X g,r (τ )) is nonzero, and if for γ ∈ Γ 0 (n g ) we have that (S X g,r (τ ))| 3/2 γ = σ g,γ (S X g,r (τ )), then ( H X g,r (τ ))| 1/2 γ = σ g,γ ( H X g,r (τ )). Here, for γ ∈ Γ 0 (n g ), we have σ g,γ = ν g (γ)σ e,γ where ν g (γ) is a multiplier which is trivial on Γ 0 (n g h g ). This identity holds even in the case that the shadow S X g,r vanishes. The vector-valued function (H X g,r (τ )) has poles only at the infinite cusp of Γ 0 (n g ), and only at the component H X g,r (τ ) where r = 1 if X has pure A-type, or at components where r 2 ≡ 1 (mod 2M ) otherwise. These poles may only have order 1 4M . This implies that the function ( H X g,r (τ )S X g,r (τ )) has no pole at any cusp, and is therefore a candidate for an application of Theorem 2. S X M,r (τ ), viewed as a scalar-valued modular function, is modular on Γ(4M ), and so ( H X g,r (τ )S X g,r (τ )) is a weight 2 nonholomorphic scalar-valued modular form for the group Γ(4M ) ∩ Γ 0 (n g ) with trivial multiplier.
Applying Theorem 2.2, we obtain a function F X g,r (τ )-call it the holomorphic projection of H X g,r (τ )S X e,r (τ )-which is a weight 2 quasimodular form on Γ(4M ) ∩ Γ 0 (n g ). In the case that S X g,r (τ ) is zero, we substitute S X e,r (τ ) in its place to obtain a function F X g,r (τ ) = H X g,r (τ )S X e,r (τ ) which is a weight 2 holomorphic scalar-valued modular form for the group Γ(4M ) ∩ Γ 0 (n g ) with multiplier ν g (or alternatively, modular for the group Γ(4M ) ∩ Γ 0 (n g h g ) with trivial multiplier).
The function F X g,r (τ ) may be determined explicitly as the sum of Eisenstein series and cusp forms on Γ(4M ) ∩ Γ 0 (n g h g ) using the standard arguments from the theory of holomorphic modular forms (i.e. the "first few" coefficients determine such a form). Therefore, we have the identity 10 where the function D X g,r (τ ) is the correction term arising in Theorem 2.2. If X has pure A-type, then B X g,r (n)q n . Then by Theorem 2.2, we find that (2. 3) The function F X g,r (τ ) may be found by considering its first few coefficients as determined using the explicit prescriptions given in §B. 4. It may also be found exactly as a sum of Eisenstein series and cusp forms in the following manner. The Eisenstein component is determined by the constant terms at cusps. Since D X g,r (τ ) (and the corresponding correction terms at other cusps) has no constant term, these are the same as the constant terms of H X g,r (τ )S X g,r (τ ), which are determined by the poles of H X g,r . The cuspidal component can be found by considering the order of vanishing of H X g,r (τ )S X g,r (τ ) at cusps. Once the B X g,r (N ) are known, equation (2.3) provides a recursion relation which may be used to calculate the coefficients of H X g,r (τ ). If the shadows S X g,r (τ ) are zero, then we may apply a similar procedure in order to determine F X g,r (τ ). For example, suppose F X g,r (τ ) = ∞ N =0 B X g,r (n)q n , and X has pure A-type. Then we find that the coefficients B X g,r (N ) satisfy Proceeding in this way we obtain the claimed results.
3 Proof of Theorem 1.1 Here we prove Theorem 1.1. The idea is as follows. For each Niemeier root system X we begin with the vector-valued mock modular forms (H X g (τ )) for g ∈ G X . We use their q-expansions to 3 PROOF OF THEOREM 1.1

11
solve for the q-series whose coefficients are the alleged multiplicities of the irreducible components of the alleged infinite-dimensional G X -modulě These q-series turn out to be mock modular forms. The proof requires that we establish that these mock modular forms have non-negative integer coefficients.
Proof of Theorem 1.1. As in the previous section, we fix a root system X and set M := m X , and we emphasize the case when X is of pure A-type.
The umbral moonshine conjecture asserts that where the second sum is over the irreducible characters of G X , and the m X χ,r (n) are non-negative integers which are the multiplicities of the irreducible G X -modules in the graded components of the alleged G X -moduleǨ X . Moreover, the umbral moonshine conjecture is true if and only if the coefficients of certain weight 1 2 mock modular forms are non-negative integers. Indeed, it turns out that the multiplicities m X χ,r (n) are the Fourier coefficients of if and only if the conjecture is true. To see this, we recall the orthogonality of characters. We have that for irreducible characters χ i and χ j , We also have the relation for g and h ∈ G X , Here |C G X (g)| is the order of the centralizer of g in G X . Since the order of the centralizer times the order of the conjugacy class of an element is the order of the group, (3.1) and (3.3) together imply the inverse relation H X g,r (τ ) = χ χ(g)H X χ,r (τ ).
These lead to the key identity

12
Therefore, in order to prove the theorem it suffices to prove that the coefficients of the mock modular forms H X χ,r (τ ) are all non-negative integers. For holomorphic modular forms, we may answer questions of this type by making use of Sturm's theorem [47] (see also Theorem 2.58 of [42]). This theorem provides a bound B associated to a space of modular forms such that a modular form f (τ ) is uniquely identified by it's first B coefficients. This bound reduces many questions about the Fourier coefficients of modular forms to finite calculations. In particular, because of the existence of integral bases, this bound B may be used to show that if the first B coefficients of the form f (τ ) are integral, then all coefficients of f (τ ) must be integral.
Sturm's Theorem relies on the finite dimensionality of certain spaces of modular forms, and so it can not be applied directly to spaces of mock modular forms. However, by making use of holomorphic projection we can adapt Sturm's theorem to this setting.
Let H X χ,r (τ ) be defined as above. Recall that the transformation matrix for the vector-valued function H X g,r (τ )) is σ g,γ , the conjugate of the transformation matrix for (S X e,r (τ )) when γ ∈ Γ 0 (n g h g ), and σ g,γ is the identity for γ ∈ Γ(4M ). Therefore if . Formulas for the shadow functions (cf. §B.2) show that the leading coefficient of S X e,1 (τ ) is 1 and has integral coefficients. This implies that the function A χ,r (τ ) := H X χ,r (τ )S X e,1 (τ ) also has integral coefficients up to the Sturm bound for Γ(4M ) ∩ Γ 0 (N X χ ) and that every coefficient of A χ,r (τ ) is integral if and only if every coefficient of H X χ,r is integral. The shadow of H X χ,r (τ ) is given by If X is pure A-type, then S X g,r (τ ) = χ X A g,r S M,r (τ ) = (χ ′ (g) + χ ′′ (g))S M,r (τ ) for some irreducible characters χ ′ and χ ′′ , according to §A.3 and §B.2. Therefore, When X is not of pure A-type the shadow is some sum of such functions, but in every case has integer coefficients, and so, applying Theorem 2.2 to A χ,r (τ ), we find that the holomorphic projection of this function has only integer coefficients if and only if A χ,r (τ ) has only integer coefficients. But the holomorphic projection is modular on Γ(4M )∩ Γ 0 (N X χ ) and has integer coefficients up to the Sturm bound for Γ(4M ) ∩ Γ 0 (N X χ ). Therefore, in order to check that H X χ,r (τ ) has only integer coefficients, it suffices to check up to the Sturm bound for Γ(4M ) ∩ Γ 0 (N χ ). These calculations were carried out using the sage mathematical software [45].
To complete the proof, it suffices to check that the multiplicities m X χ,r (n) are non-negative. The proof of this claim follows easily by modifying step-by-step the argument in Gannon's proof of non negativity in the M 24 case [35] (i.e. X = A 24 1 ). Here we describe how this is done.

13
Conjectural expressions for the alleged McKay-Thompson series H X g,r (τ ) in terms of Rademacher sums and unary theta functions are given in §B.3. These expressions are known to hold in many cases, but in any case, the difference between H X g,r (τ ) and the corresponding Rademacher sum (cf. (B.26), (B.27)) is a unary theta function of bounded level, according to the Serre-Stark theorem [46] on modular forms of weight 1 2 . The unary theta functions have bounded coefficients, and so the non-negativity depends on the asymptotic growth of the coefficients of the Rademacher sums.
Exact formulas are known for all the coefficients of Rademacher sums because they are defined by averaging the special function r 1/2 (γ, τ ) (see (B.22)) over cosets of a specific modular group modulo Γ ∞ , the subgroup of translations. Therefore, Rademacher sums are standard Maass-Poincaré series, and as a result we have formulas for each of their coefficients as convergent infinite sums of Kloosterman-type sums weighted by values of the I 1/2 modified Bessel function. (For example, see [8] for the general theory, and [12] for the specific case that X = A 24 1 .) More importantly, this means also that the generating function for the multiplicities m X χ,r (n) is a weight 1 2 harmonic Maass form, which in turn means that exact formulas (modulo the unary theta functions) are also available in similar terms. For positive integers n, this then means that (cf. Theorem 1.1 of [8]) where the sums are over the cusps ρ of the group Γ 0 (N X g ), and finitely many explicit negative rational numbers m. The constants a X ρ (m) are essentially the coefficients which describe the generating function in terms of Maass-Poincaré series. Here I is a suitable normalization and change of variable for the standard I 1/2 modified Bessel-function.
The Kloosterman-type sums K X ρ (m, n, c) are well known to be related to Salié-type sums (for example see Proposition 5 of [38]). These Salié-type sums are of the form where ǫ X ρ (m, n) is a root of unity, −D(m, n) is a discriminant of a positive definite binary quadratic form, and β X is a nonzero positive rational number.
These Salié sums may then be estimated using the equidistribution of CM points with discriminant −D(m, n). This process was first introduced by Hooley [36], and it was first applied to the coefficients of weight 1 2 mock modular forms by Bringmann and Ono [7]. Gannon explains how to make effective the estimates for sums of this shape in §4 of [35], thereby reducing the proof of the M 24 case of umbral moonshine to a finite calculation. In particular, in equations (4.6-4.10) of [35] Gannon shows how to bound coefficients of the form (3.4) in terms of the Selberg-Kloosterman zeta function, which is bounded in turn in his proof of Theorem 3 of [35]. We follow Gannon's proof mutatis mutandis. We find, for each root system, that the coefficients of each multiplicity generating function are positive beyond the 390th coefficient. Moreover, the coefficients exhibit subexponential growth. A finite computer calculation in sage has verified the non-negativity of the finitely many remaining coefficients.
Remark. It turns out that the estimates required for proving nonnegativity are the worst for the M 24 case considered by Gannon.

A The Umbral Groups
In this section we present the facts about the umbral groups that we have used in establishing the main results of this paper. We recall (from [15]) their construction in terms of Niemeier root systems in §A.1, and we reproduce their character tables (appearing also in [15]) in §A.2. Note that we use the abbreviations a n := √ −n and b n := (−1 + √ −n)/2 in the tables of §A.2.
The root system description of the umbral groups (cf. §A.1) gives rise to certain characters called twisted Euler characters which we recall (from [15]) in §A. 3. The data appearing in §A.3 plays an important role in §B.2, where we use it to describe the shadows S X g of the umbral McKay-Thompson series H X g explicitly.

A.1 Construction
As mentioned in §1, there are exactly 24 self-dual even positive-definite lattices of rank 24 up to isomorphism, according to the classification of Niemeier [41] (cf. also [18,50]). Such a lattice L is determined up to isomorphism by its root system L 2 := {α ∈ L | α, α = 2}. The unique example without roots is the Leech lattice. We refer to the remaining 23 as the Niemeier lattices, and we call a root system X a Niemeier root system if it occurs as the root system of a Niemeier lattice. The simple components of Niemeier root systems are root systems of ADE type, and it turns out that the simple components of a Niemeier root system X all have the same Coxeter number. Define m X to be the Coxeter number of any simple component of X, and call this the Coxeter number of X.
For X a Niemeier root system write N X for the corresponding Niemeier lattice. The umbral group attached to X is defined by setting where W X is the normal subgroup of Aut(N X ) generated by reflections in root vectors.
Observe that G X acts as permutations on the simple components of X. In general this action is not faithful, so define G X to be the quotient of G X by its kernel. It turns out that the level of the mock modular form H X g attached to g ∈ G X is given by the order, denoted n g , of the image of g inḠ X . (Cf. §A.3 for the values n g .) The Niemeier root systems and their corresponding umbral groups are described in Table 1. The root systems are given in terms of their simple components of ADE type. Here D 10 E 2 7 , for example, means the direct sum of one copy of the D 10 root system and two copies of the E 7 root system. The symbol ℓ is called the lambency of X, and the Coxeter number m X appears as the first summand of ℓ.
In the descriptions of the umbral groups G X , and their permutation group quotientsḠ X , we write M 24 and M 12 for the sporadic simple groups of Mathieu which act quintuply transitively on 24 and 12 points, respectively. (Cf. e.g. [20].) We write GL n (q) for the general linear group of a vector space of dimension n over a field with q elements, and SL n (q) is the subgroup of linear transformations with determinant 1, &c. The symbols AGL 3 (2) denote the affine general linear group, obtained by adjoining translations to GL 3 (2). We write Dih n for the dihedral group of order 2n, and Sym n denotes the symmetric group on n symbols. We use n as a shorthand for a cyclic group of order n.
We also use the notational convention of writing A.B to denote the middle term in a short exact sequence 1 → A → A.B → B → 1. This introduces some ambiguity which is nonetheless easily navigated in practice. For example, 2.M 12 is the unique (up to isomorphism) double cover of M 12 which is not 2 × M 12 . The group AGL 3 (2) naturally embeds in GL 4 (2), which in turn admits a unique (up to isomorphism) double cover 2.GL 4 (2) which is not a direct product. The group we denote 2.AGL 3 (2) is the preimage of AGL 3 (2) < GL 4 (2) in 2.GL 4 (2) under the natural projection. Table 2: Character table of

A.3 Twisted Euler Characters
In this section we reproduce certain characters-the twisted Euler characters-which are attached to each group G X , via its action on the root system X. (Their construction is described in detail in §2.4 of [15].) To interpret the tables, write X A for the (possibly empty) union of type A components of X, and interpret X D and X E similarly, so that if m = m X Then X = A d m−1 for some d, and X = X A ∪ X D ∪ X E , for example. Then g →χ X A g denotes the character of the permutation representation attached to the action ofḠ X on the simple components of X A . The characters g →χ X D g and g →χ X E g are defined similarly. The characters χ X A g , χ X D g , χ X E g andχ X D g incorporate outer automorphisms of simple root systems induced by the action G X on X. We refer to §2.4 of [15] for full details of the construction. For the purposes of this work, it suffices to have the explicit descriptions in the tables in this section. The twisted Euler characters presented here will be used to specify the umbral shadow functions in §B.2.
The twisted Euler character tables also attach integers n g and h g to each g ∈ G X . By definition, n g is the order of the image of g ∈ G X inḠ X (cf. §A.1). The integer h g may be defined by setting h g := N g /n g where N g is the product of the shortest and longest cycle lengths appearing in the cycle shape attached to g by the action of G X on a (suitable) set of simple roots for X.             [g] 1A 2A 4AB  [g] 1A 2A n g |h g 1|1 1|2 1 0 [g] 1A 2A n g |h g 1|1 1|2 n g |h g 1|1 1|2 [g] 1A 2A n g |h g 1|1 2|1 [g] 1A 2A n g |h g 1|1 2|2 [g] 1A 2A n g |h g 1|1 1|4

B The Umbral McKay-Thompson Series
In this section we describe the umbral McKay-Thompson series in complete detail. In particular, we present explicit formulas for all the McKay-Thompson series attached to elements of the umbral groups by umbral moonshine in §B.4. Most of these expressions appeared first in [14,15], but some appear for the first time in this work. In order to facilitate explicit formulations we recall certain standard functions in §B.1. We then, using the twisted Euler characters of §A.3, explicitly describe the shadow functions of umbral moonshine in §B.2. The Rademacher sum construction of the umbral McKay-Thompson series is described in §B.3.

B.1 Special Functions
The Dedekind eta function is η(τ ) := q 1/24 which is a modular form of weight two for Γ 0 (N ) if M |N . Define the Jacobi theta function θ 1 (τ, z) by setting According to the Jacobi triple product identity we have The other Jacobi theta functions are Define Ψ 1,1 and Ψ 1,−1/2 by setting These are meromorphic Jacobi forms of weight one, with indexes 1 and −1/2, respectively. Here, the term meromorphic refers to the presence of simple poles in the functions z → Ψ 1, * (τ, z), for fixed τ ∈ H, at lattice points z ∈ Zτ + Z.

29
The standard index m theta functions, for m a positive integer, are defined by where r ∈ Z. Evidently, θ m,r only depends on r mod 2m. Set S m,r (τ ) := 1 2πi ∂ z θ m,r (τ, z)| z=0 , so that For a m a positive integer define We recover Ψ 1,1 upon specializing to m = 1. Observe that and define also S m,r (τ ) := 1 2πi ∂ z θ m,r (τ, z)| z=0 , so that As in the integral index case, θ m,r depends only on r mod 2m. We recover −θ 1 upon specializing θ m,r to m = r = 1/2. For m ∈ Z + 1/2, m > 0, define Given α ∈ Q write [α] for the operator on q-series (in rational, possibility negative powers of q) that eliminates exponents not contained in Z + α, so that if f = β∈Q c(β)q β then [α]f := n∈Z c(n + α)q n+α (B.13)

B.2 Shadows
Let X be a Niemeier root system and let m = m X be the Coxeter number of X. For g ∈ G X we define the associated shadow function S X g = (S X g,r ) by setting where the S X A g , &c., are defined in the following way, in terms of the twisted Euler characters χ X A g , &c. given in §A.3, and the unary theta series S m,r (cf. (B.6)).
Note that if m = m X then S X g,r = S X g,r+2m = −S X g,−r for all g ∈ G X , so we need specify the S X A g,r , &c., only for 0 < r < m.
If X A = ∅ then S X A g := 0. Otherwise, we define S X A g,r for 0 < r < m by setting  If X E = ∅ then S X E g := 0. Otherwise, m is 12 or 18 or 30. In case m = 12 define S X E g,r for 0 < r < 12 by setting In case m = 18 define S X E g,r for 0 < r < 18 by setting
Conjecture B.1. Let X be a Niemeier root system and let g ∈ G X . If X = A 3 8 and g ∈ G X , or if X = A 3 8 and g ∈ G X does not satisfy o(g) = 0 mod 3, then we havě If X = A 3 8 and g ∈ G X satisfies o(g) = 0 mod 3 theň H X g,r (τ ) = −2R X Γ 0 (ng),ν X g (τ ) +ť (9) g (τ ). (B.27) Conjecture B.1 is a theorem in the case that X = A 24 1 . This is the main result of [12]. A number of other cases of Conjecture B.1 are proved in [16].
The functions f 23,a and f 23,b in Table 36 are cusp forms of weight two for Γ 0 (23), defined by Note that the definition of F (2) g appearing here for g ∈ 23A ∪ 23B corrects errors in [11,12].
For use later on, note that ψ  g,r ) associated to g ∈ G (5) is a 10-vector-valued function, with components indexed by r ∈ Z/10Z.

(B.87)
In the expression for g ∈ 8AB, we write F  We have m X = 13, so the umbral McKay-Thompson series H (13) g = (H (13) g,r ) associated to g ∈ G (13) = G X ≃ Z/4Z is a 26-vector-valued function, with components indexed by r ∈ Z/26Z.
Define H (13) g = (H (13) g,r ) for g ∈ G (13) by requiring that are meromorphic Jacobi forms of weight 1 and index 13 given explicitly in Table 42.