Arxiv:2008.06742V1 [Hep-Th] 15 Aug 2020 I.Mntosgaiyin Gravity Monstrous III

Home , Monster vertex algebra, Monstrous moonshine, Sporadic group

arXiv:2008.06742v1 [hep-th] 15 Aug 2020 I.MntosGaiyin Gravity Monstrous III. ‡ † ∗ V osru -hoyin M-theory Monstrous IV. I Conclusion VI. lcrncades [email protected] address: Electronic [email protected] address: Electronic lcrncades [email protected] address: Electronic I raiyof Triality II. .CenSmn arnintrsfrMntosM-theory Monstrous for terms Lagrangian Chern-Simons A. lattices of construction Cohomological V. .Introduction I. .Classification A. A. References Acknowledgements .Lagrangian(s) A. C. B. B. Weak Weak Weak 1. B = N asesfil n2 ,ehbt h aenme fbsncan bosonic of number whe same the M-theory, exhibits an Monstrous 1, + rather 26 of but in Susskind’ subsector field spaces, al and massless certain unrelated Griess Horowitz of a the includes sum Remarkably, of which a automorphism M-theory, merely the Monstrous not as is Monster algebra the of definition and iet ltoao o-uesmerc gravito-dilaton non-supersymmetric, of plethora a to rise 196 dimension pisudrteMntra 196 as Monster the under splits F uegaiyi 6+1? + 26 in Supergravity 1 = edfiea define We λ ult mn st f p-forms of) (sets among duality ψ n2 1 + 26 in -triality -triality 196 d 4 n t eeaiain in generalizations its and , 883 , eoe h mletnntiilrpeetto fteMon the of representation non-trivial smallest the denotes 8,i ce pnb h ose ru.Uo euto to reduction Upon group. Monster the by upon acted is 884, D n srnm,Uiest fPdv,adIF,Pdv,Ital Padova, INFN, and Padova, of University Astronomy, and etoRcrh nioFri oa n et fPhysics of Dept. and Roma, Fermi, Enrico Ricerche Centro 6+1Mntos ueybsncMter,woemsls s massless whose M-theory, bosonic purely Monstrous, 1 + 26 = 5+1 + 25 6+1 + 26 unu rvt eerh o nee,C,USA CA, Angeles, Los Research, Gravity Quantum ync CQ,LsAgls A USA CA, Angeles, Los ICMQG, Dyonica , 8 = 884 eiae oJh .Conway H. John to Dedicated osru M-theory Monstrous 1 lsi Marrani Alessio ⊕ d ai Chester David ihe Rios Michael 12 196 Contents , 883 hr h ige sietfidwt h dilaton, the with identified is singlet the where , † ‡ ∗ ctere,woesetu irreducibly spectrum whose theories, ic ooi -hoya subsector. a as M-theory bosonic s emoi ere ffreedom. of degrees fermionic d ope oaRarita-Schwinger a to coupled n er,b hwn htsc an such that showing by gebra, ler fmsls ttsfra for states massless of algebra tr hscaie the clarifies This ster. D 5+1 hsgives this 1, + 25 = y etu,of pectrum, 6 5 4 4 4 3 2 20 19 14 12 18 16 11 10 9 2

I. INTRODUCTION

The Monster group M was predicted to exist by Fischer and Griess [1]. It is the automorphism group of the Griess algebra, as well as the automorphism group of the Monster vertex algebra [3, 5]. Conway and Norton defined Monstrous moonshine as the observation that the Fourier coefficients of the j-function decompose into dimensions of representations of the Monster group [2] and this was proven by Borcherds using generalized Kac-Moody algebras [6]. In the language of conformal field theory, Monstrous moonshine is the statement that the states of an orbifold theory, 24 which is D = 25+1 bosonic string theory on (R /Λ24)/Z2 (where Λ24 is the Leech lattice [4, 31, 32]), are organized in representations of the Monster group, with partition function equivalent to the j-function [8, 12, 14]. Witten also found the Monster in three-dimensional pure gravity [22], for AdS3, where the dual CFT is expected to be that of Frenkel, Lepowsky and Meurman [5]. Eguchi, Ooguri and Tachikawa later noticed the elliptic genus of the K3 surface has a natural decomposition in terms of dimensions of irreducible representations of the largest Mathieu group M24 [9], and this was named Umbral moonshine [10, 11], which generalizes the moonshine correspondence for other sporadic groups [13]. With Witten’s proposal [21] that M-theory unifies all the ten-dimensional string theories with N = 1 supergravity in D = 10+1, Horowitz and Susskind argued [23] there exists a bosonic M-theory in D = 26 + 1 that reduces to the bosonic string upon compactification. As the Monster has a string theoretic interpretation in D = 25 + 1 [14–16], it is also natural to consider its action on fields from D = 26 + 1. Support for this is found from bosonic M-theory’s 23 2-brane near horizon geometry AdS4 × S , discussed by Horowitz and Susskind as an evidence for a 2+1 CFT dual with global SO24 symmetry [23]. Noting that the automorphism group of the Leech lattice, the Conway group Co0 [4, 32], is a maximal finite subgroup of SO24, and its Z2 quotient Co1 is a maximal subgroup of the Monster [4, 5], this provides a means to realize Monstrous symmetry as a finite subgroup of R-symmetry from D = 26 + 1 [25]. In the present paper, we introduce a purely bosonic gravity theory coupled to p-forms in 26+1 space-time dimensions, which contains the aforementioned bosonic string theory [23] as a subsector. We name such a theory Monstrous M- theory, or shortly M 2-theory, because its massless spectrum has the same dimension (196, 884) as the Griess algebra and it can thus be acted upon by the Monster group M. When reducing to 25 + 1, a plethora of gravito-dilatonic theories is generated, in which the decomposition 196, 884 = 196, 883⊕1, which first hinted at Monstrous moonshine [2], entails the fact that the dilaton in 25 + 1 is a singlet of M itself. As such, the irreducibility under M is crucially related to dilatonic gravity in 25 + 1. Thanks to existence of some maps among iso-dimensional representations of the massless little group SO24 in 25 + 1 (which generalize, in a “weak” sense, the triality of SO8), most of the gravito- dilatonic theories in 25 + 1, named Monstrous gravity theories, display a fermionic (massless) spectrum, as well, such that the whole spectrum is still acted upon by M. All this gives a more concise description of the Monster’s minimal representation 196, 883, coming from the field content of Monstrous gravity in D = 25 + 1; as such, this also elucidates the definition of the Monster as the automorphism group of the Griess algebra (the degree two piece of the monster vertex algebra), which has been considered artificial in that it was thought to involve an algebra of two or more unrelated spaces [4, 5, 7].

The plan of the paper is as follows. At the start of Sec. II, we brieﬂy recall the triality among the 8-dimensional representations of the Lie algebra d4, as well as the duality among all semispinor representations and their conjugates (and among all the tensor products thereof) in Lie algebras dn. Then, we introduce some “weak” generalizations within the Lie algebra d12, namely weak λ-triality in Sec. II A and weak ψ-triality in Sec. II B. Furthermore, we also introduce a “weak” generalization of the aforementioned duality in Sec. II C. In all these cases, the “weakness” relies on the reducibility of the bosonic representations involved. Next, in Sec. III we introduce and classify non-supersymmetric, gravito-dilatonic theories, named Monstrous gravities, in 25+1 space-time dimensions, whose massless spectrum (also including fermions in most cases) has dimension 196, 884, namely the same dimension as the Griess algebra [3, 5], and it can thus be acted upon by the Monster group M. A purely bosonic uplift to 26 + 1 space-time dimensions is discussed in Sec. IV, in which the so-called M 2-theory is deﬁned, along with some discussion of its Lagrangian in Sec. IVA; furthermore, Sec. IVB discusses a subsector of the M2-theory diplaying the same number of bosonic and fermionic massless degrees of freedom in 26 + 1; in this framework, after a discussion of some reductions to 25+1, in Sec. IVB1 a Lagrangian and local supersymmetry transformations are conjectured for the would-be N =1 supergravity in 26+1 space-time dimensions. Then, in Sec. V a cohomological construction of both the e8 root lattice and the Leech lattice Λ24 (respectively determining optimal sphere packing in 8 and 24 dimensions [4]) is discussed, and all this is again related to M-theory (i.e., N = 1 supergravity) in 10+1 and to the would-be N = 1 supergravity 26 + 1, respectively. Final comments are contained in the conclusive Sec. VI. 3

II. TRIALITY OF d4 AND ITS GENERALIZATIONS IN d12

By triality, denoted by T, in this paper we refer to a property of the Lie algebra d4 (see e.g. [19]), namely a map of its three 8-dimensional irreducible representations

1 ∧ ≡ 8v := (1, 0, 0, 0) (vector);  d4 :  λ ≡ 8s := (0, 0, 0, 1) (semispinor); (2.1)  ′ ′ λ ≡ 8 ≡ 8c := (0, 0, 1, 0) (conjugate semispinor)  s  among themselves :

∧1 T :  ↑↓ ցտ  . (2.2) λ′ ⇄ λ  

This property can be traced back to the three-fold structural symmetry of the Dynkin diagram of d4, and to the existence of an outer automorphism of d4 which interchanges 8v, 8s and 8c; in fact, the outer automorphism group of d4 (or, more precisely, of the corresponding spin group Spin8, the double cover of the Lie group SO8) is isomorphic to the symmetric group S3 that permutes such three representations. Thence, triality aﬀects all tensor products stemming from 8v, 8s and 8c. In particular, for later convenience, we stress that it maps also the three 56-dimensional irreducible representations of d4 :

3 ∧ ≡ 56v := (0, 0, 1, 1) (3-form);  d4 :  ψ ≡ 56s := (1, 0, 0, 1) (vector-semispinor ); (2.3)  ′ ′ ψ ≡ 56 ≡ 56c := (1, 0, 1, 0) (conjugate vector-semispinor )  s  among themselves

∧3 T :  ↑↓ ցտ  . (2.4) ψ ⇄ ψ′   α We recall here that, in order to correspond to an irreducible representation, the spinor-vector ψµ must be gamma- traceless :

µ β Γαβψµ =0, (2.5)

µ µ where µ and α are the vector resp. spinor indices, and Γαβ ≡ (Γ )αβ denote the gamma matrices of d4. ψ is a 3 Rarita-Schwinger (RS) ﬁeld of spin/helicity 2 , and, in the context of supersymmetric theories, it is named gravitino (being the spartner of the graviton gµν ). As (2.2) denotes the action of triality T on (semi)spinors, (2.4) expresses the triality T acting on RS ﬁelds. Triality plays an important role in type II string theory, in which so8 (compact real form of d4) is the algebra of the massless little group (cfr. e.g. [20]). Within Lie algebras dn, one can always have a duality between the (semi)spinor representation

λ ≡ 2n−1 := 0n−1, 1 (2.6) and its conjugate representations

′ λ′ ≡ 2n−1 := 0n−2, 1, 0 (2.7) (as well as between tensor products thereof, such as the aforementioned gravitino ψ and conjugate gravitino ψ′ representations). In fact, λ and λ′, as well as ψ and ψ′, have the same dimension but diﬀerent Dynkin labels; in all these cases, the “duality” map is nothing but the spinor conjugation between the chiral components of the spinor bundle. On the other hand, there are no known examples of bosonic representations having the same dimension but diﬀerent Dynkin labels. 4

A. Weak λ-triality

However, there may be a “weaker” instance of the aforementioned triality, in which λ and λ′ have the same dimension of a reducible (bosonic) representation, namely of a sum of irreducible (bosonic) representations, of dn. In fact, for n = 12 (i.e. in d12) something remarkable takes place: in d12, the following three representations have the same dimension 2, 048: λ ≡ 211 = 2048 := (011, 1);  ′ 11 ′ ′ 10 d12 :  λ ≡ 2 = 2048 := (0 , 1, 0); (2.8)   1 3 11 2 9  ∧ ⊕ ∧ = 24 ⊕ 2024 = (1, 0 ) ⊕ (0 , 1, 0 ).   1 In other words, in d12 the reducible bosonic representation given by the sum of the vector (1-form) representation ∧ and of the 3-form representation ∧3 has the same dimension of each of the (semi)spinors λ and λ′. Analogously to the aforementioned case of d4, one can then deﬁne a “triality-like” map, named weak λ-triality and denoted by T˜λ, between the corresponding representation vector spaces, ∧1 ⊕ ∧3 T˜λ :  ↑↓ ցտ  (2.9) λ ⇄ λ′   At a glance, one realizes that a crucial diﬀerence with (2.2) relies in the reducibility of the bosonic sector of the map, which we will henceforth associate to the “weakness” of T˜λ. However, since no other Dynkin diagram (besides d4) has an automorphism group of order greater than 2, one can also conclude that (2.8)-(2.9) cannot be realized as an automorphism of d12, nor it can be traced back to some structural symmetry of the Dynkin diagram of d12 itself.

B. Weak ψ-triality

As triality T of d4 (2.2) affects all tensor products stemming from 8v, 8s and 8c, implying in particular (2.4), so 1 3 ′ the weak λ-triality T˜λ of d12 (2.9) affects all tensor products stemming from ∧ ⊕ ∧ , λ and λ ; in particular, in d12, the following three representations have the same dimension 47, 104: ψ ≡ 47, 104 := (1, 010, 1);  ′ ′ 9  ψ ≡ 47, 104 := (1, 0 , 1, 0); d12 :  (2.10)  2 · 2 · ∧4 ⊕ ∧3 ⊕ ∧2 =2 · (2 · 10, 626 ⊕ 2, 024 ⊕ 276)  =2· 2 · (03, 1, 08) ⊕(02, 1, 09) ⊕ (0, 1, 010) .   4 3 In other words, in d12 the reducible bosonic representation given by the sum of the 4-form ∧ , 3-form ∧ and 2- form ∧2 representations (with multiplicity 4, 2 and 2, respectively) has the same dimension of each of the RS field ′ representations ψ and ψ . Analogously to the aforementioned case of d4, one can then define a “triality-like” map, named weak ψ-triality and denoted by T˜ψ, between the corresponding representation vector spaces, 2 · 2 · ∧4 ⊕ ∧3 ⊕ ∧2 T˜ψ :  ↑↓ ցտ  . (2.11) ψ ⇄ ψ′   Again, (2.8)-(2.9) cannot be realized as an automorphism of d12, nor it can be traced back to some structural symmetry of the Dynkin diagram of d12 itself.

C. Weak duality among (sets of) p-forms

Representations with the same dimensions can also be only bosonic. Still, d12 provides the following example of such a phenomenon : the following two sets of representations have the same dimension 42, 504, ∧5 ≡ 42, 504 := (04, 1, 07); d12 :  (2.12)  4 · ∧4 =4 · 10, 626 =4 · (03, 1, 08).  5

5 In other words, in d12 the 5-form representation ∧ has the same dimension, namely 42, 504, of four copies of the 4-form representation ∧4. Again, one can then deﬁne a “duality-like” map, named (bosonic) weak duality and denoted by B, between the corresponding representation vector spaces1, B : ∧5 ↔ 4 · ∧4 . (2.13)

III. MONSTROUS GRAVITY IN 25 + 1

In the previous Section we have introduced some maps among fermionic and bosonic representations of d12, having the same dimension but diﬀerent Dynkin labels:

• the weak λ-triality T˜λ (2.8)-(2.9), generalizing the triality T (2.2) of d4 to d12;

• the weak ψ-triality T˜ψ (2.10)-(2.11), extending the weak triality of d12 to its RS sector;

• the bosonic weak duality B (2.12)-(2.13) among sets of bosonic (p-form) representations of d12.

As triality T (2.2) of d4 plays a role in the type II string theories having so8 as the algebra of the massless little group, one might ask the intriguing question whether (2.9), (2.11) and (2.13) have some relevance in relation to bosonic string theory [23], or in more general field theories defined in D = s + t = 25 + 1 space-time dimensions, in which so24 is the algebra of the massless little group. Below, we will show that this is actually the case for a quite large class of non-supersymmetric dilatonic gravity theories in 25 + 1, named Monstrous gravities, which we are now going to introduce. To this aim, we start and display various massless fields in D = 25 + 1 space-time dimensions. As mentioned, each massless field fits into the following irreducible representation R of the massless little group SO24:

field R Dynkin labels g : 299 2, 011 ψ : 47, 104 1, 010, 1 ψ′ : 47, 104′ (1 , 09, 1, 0) ∧1 24 (1, 011) λ : 2, 048 (011, 1) (3.1) λ′ : 2, 048′ (010, 1, 0) φ : 1 (012) ∧5 : 42, 504 (04, 1, 07) ∧4 : 10, 626 (03, 1, 08) ∧3 : 2, 024 (02, 1, 09) ∧2 : 276 (0, 1, 010) We are now going to classify field theories in 25 + 1 which share the following features : a They all contain gravity (in terms of one 26-bein, then yielding a metric tensor gµν ) and one dilaton scalar field φ; thus, the Lagrangian density of their gravito-dilatonic sector reads2 −2φ µ L = e (R − 4∂µφ∂ φ) . (3.2) b The relations among all such theories exploit the weak λ-triality T˜λ (2.8)-(2.9), the weak ψ-triality T˜ψ (2.10)-(2.11), as well as the bosonic weak duality B (2.12)-(2.13) of so24 (real compact form of d12), which is the Lie algebra of the massless little group. c The total number of degrees of freedom of the massless spectrum sums up to 299 · (#g)+47, 104 · (#ψ)+24 · #∧1 +2, 048 · (#λ)+1 · (#φ) +42, 504 · #∧5 + 10, 626 · #∧4 +2,024 · #∧3 + 276 · #∧2 = 196, 884. (3.3)

1 Of course, all instances of iso-dimensionality among representations given by (2.8)-(2.9), (2.10)-(2.11) and (2.12)-(2.13), hold up to Poincar´e/Hodge duality (in the bosonic sector); cfr. (3.7) further below. 2 Throughout our analysis, we rely on the conventions and treatment given in Secs. 22 and 23 of [36]. 6

Consequently, the whole set of massless degrees of freedom of such theories may be acted upon by the so- called Monster group M, the largest sporadic group, because 196, 883 is the dimension of its smallest non-trivial representation [1]. For this reason, the gravito-dilatonic theories under consideration will all be named Monstrous gravities. Concerning this, we would like to stress that the inclusion of the 300 massless degrees of freedom given by the graviton gµν (299) and the dilaton φ (1), rather than by the 2-form Aµν (276) and the vector Aµ (24), yields the splitting 196, 884 = 196, 883+1, (3.4) which is at the origin of the so-called Monstrous moonshine [2, 6]. The dilaton φ, which is a singlet of M, coincides with the vacuum state |Ωi of the chiral Monster conformal theory discussed in [8, 17, 18]. Thus, dilatonic gravity in 25 + 1 dimensions seems intimately related to the 196, 883-dimensional representation of M. In the context of Witten’s three-dimensional gravity [22], this suggests the 196, 883 primary operators that create black holes are carrying dilatonic gravity field content. As done in [22], it is enlightening to compare the number 196, 883 of primaries with the Bekenstein-Hawking entropy of the corresponding black hole : an exact quantum degeneracy of 196, 883 yields an entropy of Witten’s BTZ black hole given by ln (196, 883) ≃ 12.19, whereas the Bekenstein-Hawking entropy-area formula yields to 4π ≃ 12.57. Of course, one should not expect a perfect agreement between such two quantities, because the Bekenstein-Hawking entropy-area formula holds in the semi-classical regime and not in the exact quantum one. As given in (3.3), 196, 883 comes from gauge fields (potentials), graviton, etc., albeit without dilaton; in this sense, the quantum entropy ln (196, 883) ≃ 12.19 has a clear higher-dimensional interpretation, since the BTZ black hole degrees of freedom are expressed in terms of massless degrees of freedom of fields in 25 + 1 space-time dimensions.

A. Classiﬁcation

All such theories will be classiﬁed by using two sets of numbers :

3 1 • s1, a 5-characters string, providing the number of independent “helicity”-h massless fields, with h =2, 2 , 1, 2 , 0, respectively denoted by g (graviton), ψ (Rarita-Schwinger field), Λ1 (vector, i.e. 1-form, potential), λ (spinor field3), and φ (dilaton) : 1 1 s1 := #g, #ψ, #Λ , #λ, #φ = 1, #ψ, #Λ , #λ, 1 ; (3.5) • s2, a 4-character string, providing the number of independent p-form brane potentials, for p =5, 4, 3, 2, 5 4 3 2 s2 := #∧ , #∧ , #∧ , #∧ . (3.6) Before starting, we should point out that the classification below is unique up to Poincaré/Hodge duality in the p-form potentials’ sector, namely, for p= 1, ..5 : ∗ p-form pot. ←→ p’-form pot. ∧1 ∧23 ∧2 ∧22 (3.7) ∧3 ∧21 ∧4 ∧20 ∧5 ∧19 as well as up to chiral/non-chiral arrangements in the fermionic sector,

#ψ chiral/non-chiral arr.s 2 (2, 0) , (0, 2), (1, 1) (3.8) 4 (4, 0), (0, 4), (3, 1), (1, 3), (2, 2); #λ chiral/non-chiral arr.s 1 (1, 0), (0, 1) (3.9) 2 (2, 0), (0, 2), (1, 1) 3 (3, 0), (0, 3), (2, 1), (1, 2).

3 The spinor ﬁeld gets named gaugino (or dilatino) in presence of supersymmetry. 7

Clearly, (3.7) and (3.8)-(3.9) are particularly relevant if supersymmetry in 25 + 1 were considered; however, we will not be dealing with such an interesting issue here, and we will confine ourselves to make some comments further below (in 26 + 1). We will split the Monstrous gravity theories, sharing the features a - c listed above, in five groups, labelled with Latin numbers : 0, 1, 2, 3, 4, specifying the number #ψ of h =3/2 RS fields. The weak ψ-triality T˜ψ (2.10)-(2.11) of so24 maps such five groups among themselves. Then, each of these groups will be split into four subgroups, labelled with Greek letters : α, β, γ and δ, respectively characterized by the following numbers #∧1 and #λ of h = 1 vectors resp. h =1/2 spinors :

#∧1, #λ = (3, 0), (2, 1), (1, 2), (0, 3) . (3.10) α β γ δ

The weak λ-triality T˜λ (2.8)-(2.9) of so24 allows to move among such four subgroups (within the same group). The theories belonging to each of such four subgroups will share the same split of the massless degrees of freedom into bosonic and fermionic ones, respectively specified by the numbers B and F . Each of such four subgroups is a set of a varying number of theories, which will be labelled in lowercase Latin letters : i, ii, iii, etc. Such theories will be connected by the action of the bosonic weak duality B (2.12)-(2.13) of so24, and thus they will differ for the content of 5-form ∧5 and 4-form ∧4 (potential) fields. Modulo all possibilities arising from the combinations of (3.7)-(3.9), the classification of Monstrous gravity theories in 25 + 1 space-time dimensions is as follows.

0 Group 0 (ψ-less theories):

s1 s2 features 5  i (1, 0, 3, 0, 1) (0, 16, 12, 8) bosonic, ∧ -less ii ′′ (1, 12, 12, 8) bosonic α :  ′′  (3.11) (B|F )=(196,884|0),λ-less iii (2, 8, 12, 8) bosonic   ′′  iv (3, 4, 12, 8) bosonic   v ′′ (4, 0, 12, 8) bosonic, ∧4-less   s1 s2 features 5  i (1, 0, 2, 1, 1) (0, 16, 11, 8) ∧ -less  ii ′′ (1, 12, 11, 8) − β :  ′′  (3.12) (B|F )=(194,836|2,048) iii (2, 8, 11, 8) −   ′′  iv (3, 4, 11, 8) −   v ′′ (4, 0, 11, 8) ∧4-less    s1 s2 features 5  i (1, 0, 1, 2, 1) (0, 16, 10, 8) ∧ -less  ii ′′ (1, 12, 10, 8) − γ :  ′′  (3.13) (B|F )=(192,788|4,096)  iii (2, 8, 10, 8) −   ′′   iv (3, 4, 10, 8) −   v ′′ (4, 0, 10, 8) ∧4-less    s1 s2 features 5  i (1, 0, 0, 3, 1) (0, 16, 9, 8) ∧ -less  ii ′′ (1, 12, 9, 8) − δ : (3.14) 1  ′′  (B|F )=(190,740|6,144), ∧ -less  iii (2, 8, 9, 8) −   ′′   iv (3, 4, 9, 8) −   v ′′ (4, 0, 9, 8) ∧4-less    8

1 Group 1 (#ψ = 1 theories):

s1 s2 features 5  i (1, 1, 3, 0, 1) (0, 12, 10, 6) ∧ -less  α : ii ′′ (1, 8, 10, 6) − (3.15) (B|F )=(149,780|47,104), λ-less  ′′   iii (2, 4, 10, 6) −   iv ′′ (3, 0, 10, 6) ∧4-less    s1 s2 features 5  i (1, 1, 2, 1, 1) (0, 12, 9, 6) ∧ -less  β : ii ′′ (1, 8, 9, 6) − (3.16) (B|F )=(147,732|49,152)  ′′   iii (2, 4, 9, 6) −   iv ′′ (3, 0, 9, 6) ∧4-less    s1 s2 features 5  i (1, 1, 1, 2, 1) (0, 12, 8, 6) ∧ -less  γ : ii ′′ (1, 8, 8, 6) − (3.17) (B|F )=(145,684|51,200)  ′′   iii (2, 4, 8, 6) −   iv ′′ (3, 0, 8, 6) ∧4-less    s1 s2 features 5  i (1, 1, 0, 3, 1) (0, 12, 7, 6) ∧ -less  δ : ii ′′ (1, 8, 7, 6) − (3.18) (B|F )=(143,636|53,248), ∧1-less  ′′   iii (2, 4, 7, 6) −   iv ′′ (3, 0, 7, 6) ∧4-less    2 Group 2 (#ψ = 2 theories):

s1 s2 features i (1, 2, 3, 0, 1) (0, 8, 8, 4) ∧5-less α :  ′′  (3.19) (B|F )=(102,676|94,208), λ-less ii (1, 4, 8, 4) −  ′′ 4   iii (2, 0, 8, 4) ∧ -less  s1 s2 features i (1, 2, 2, 1, 1) (0, 8, 7, 4) ∧5-less β :  ′′  (3.20) (B|F )=(100,628|96,256) ii (1, 4, 7, 4) −  ′′ 4   iii (2, 0, 7, 4) ∧ -less  s1 s2 features i (1, 2, 1, 2, 1) (0, 8, 6, 4) ∧5-less γ :  ′′  (3.21) (B|F )=(98,580|98,304) ii (1, 4, 6, 4) −  ′′ 4   iii (2, 0, 6, 4) ∧ -less  s1 s2 features i (1, 2, 0, 3, 1) (0, 8, 5, 4) ∧5-less δ :  ′′  (3.22) (B|F )=(96,532|100,352), ∧1-less ii (1, 4, 5, 4) −  ′′ 4   iii (2, 0, 5, 4) ∧ -less  3 Group 3 (#ψ = 3 theories):

s1 s2 features α :  i (1, 3, 3, 0, 1) (0, 4, 6, 2) ∧5-less  (3.23) (B|F )=(55,572|141,312), λ-less ii ′′ (1, 0, 6, 2) ∧4-less   s1 s2 features β :  i (1, 3, 2, 1, 1) (0, 4, 5, 2) ∧5-less  (3.24) (B|F )=(53,524|143,360) ii ′′ (1, 0, 5, 2) ∧4-less   s1 s2 features γ :  i (1, 3, 1, 2, 1) (0, 4, 4, 2) ∧5-less  (3.25) (B|F )=(51,476|145,408) ii ′′ (1, 0, 4, 2) ∧4-less   s1 s2 features δ :  i (1, 3, 0, 3, 1) (0, 4, 3, 2) ∧5-less  (3.26) (B|F )=(49,428|147,456), ∧1-less ii ′′ (1, 0, 3, 2) ∧4-less   9

4 Group 4 (#ψ = 4 theories):

s1 s2 features α : 5 4 2 (3.27) (B|F )=(8,468|188,416), λ-less (1, 4, 3, 0, 1) (0, 0, 4, 0) ∧ , ∧ , ∧ -less

s1 s2 features β : 5 4 2 (3.28) (B|F )=(6,420|190,464) (1, 4, 2, 1, 1) (0, 0, 3, 0) ∧ , ∧ , ∧ -less

s1 s2 features γ : 5 4 2 (3.29) (B|F )=(4,372|192,512) (1, 4, 1, 2, 1) (0, 0, 2, 0) ∧ , ∧ , ∧ -less

s1 s2 features δ : 5 4 2 1 (3.30) (B|F )=(2,324|194,560) (1, 4, 0, 3, 1) (0, 0, 1, 0) ∧ , ∧ , ∧ , ∧ -less

The above classiﬁcation contains 60 Monstrous gravity theories, from the purely bosonic, ∧5-less, 0.α.i theory (3.11) to the theory with the highest F , i.e. the 4.δ theory (3.30). As it can be realized at a glance, Monstrous gravity theories are not supersymmetric, as it is evident from the mismatch between B and F . Also, it should be noted that all such theories (but the ones of the group 4 (3.27)-(3.30)) contain bosonic string theory, whose (massless, closed string) ﬁeld content is #g = #φ = #∧2 = 1 (see e.g. [23]) as a subsector.

IV. MONSTROUS M-THEORY IN 26 + 1

At this point, the natural question arises whether Monstrous gravities classified above can be uplifted to 26 + 1 space-time dimensions, in which the massless little group is SO25. At least in one case, namely for the purely bosonic Monstrous gravity labelled by 0.α.iii , the answer is positive. The field content of such a theory is specified by the following s1 and s2, as from (3.11) :

s1 = (1, 0, 3, 0, 1); 0.α.iii :  (4.1) (B|F )=(196,884|0)  s2 = (2, 8, 12, 8) ,  or equivalently :

ﬁeld R of so24 # g : 299 1 ψ : 47, 104 0 ψ′ : 47, 104′ 0 ∧1 24 3 λ : 2, 048 0 (4.2) λ′ : 2, 048′ 0 φ : 1 1 ∧5 : 42, 504 2 ∧4 : 10, 626 8 ∧3 : 2, 024 12 ∧2 : 276 8

It is then easy to realize that all such bosonic massless (SO24-covariant) fields in 25 + 1 can be uplifted to a smaller set of bosonic massless fields (SO25-covariant) fields in 26 + 1; namely, since

so25 so24 fields g −→ g , ∧1, φ 324 299 24 1 ∧5 −→ ∧5 , ∧4 53,130 42,504 10,626 ∧4 −→ ∧4 , ∧3 (4.3) 12,650 10,626 2,024 ∧3 −→ ∧3 , ∧2 2,300 2,024 276 ∧2 −→ ∧2 , ∧1 300 276 24 10 the purely bosonic field content (4.2) of the theory 0.α.iii in 25+1 can be obtained by the S1 reduction of the following purely bosonic field content in 26 + 1 :

ﬁeld R of so25 # g : 324 1 ∧1 25 0 ϕ : 1 0 (4.4) ∧5 : 53, 130 2 ∧4 : 12, 650 6 ∧3 : 2, 300 6 ∧2 : 300 2

So, we have obtained a purely bosonic gravity theory (coupled to p-forms, with p= 2, 3, 4, 5) in 26+1 whose massless spectrum, given by 196, 884 degrees of freedom, is acted upon by the Monster group M, because it corresponds to the sum of its two smallest representations, namely the trivial (singlet) 1 and the non-trivial one 196, 883. Such a theory will be henceforth named Monstrous M-theory, or simply M 2-theory. Note that the disentangling of the 196, 884 degrees of freedom into 196, 883 ⊕ 1 occurs only when reducing the theory to 25 + 1, in which case the dilaton φ is identiﬁed with the singlet of M. We would like here to stress that the possibility of an uplift to 26 + 1 is far from being trivial, and when possible, this uniquely ﬁxes the content of the higher dimensional massless spectrum.

A. Lagrangian(s)

A priori, the 196, 884-dimensional degrees of freedom of the massless spectrum of M2-theory can be realized in various ways at the Lagrangian level. Here, we will attempt at writing down a general Lagrangian for M2-theory. We start and label the massless ﬁelds of M2-theory, given by (4.4), as follows :

ﬁeld label # g : gµν 1 5 (5)A ∧ : Cλµνρσ 2 4 (4)i (4.5) ∧ : Cλµνρ 6 3 (3)i ∧ : Cλµν 6 2 (2)A ∧ : Cλµ 2

The uppercase Latin indices take values 1, 2, whereas the lowercase Latin indices run 1, 2, .., 6. A general deﬁnition of the ﬁeld strengths reads

(3)A (2)A A (3)j G := dC + Aj C ; (4)i (3)i i (2)A (2)B (4)j G := dC + B(AB)C ∧ C + Cij C ; (5)i (4)i i (2)A (3)j i (5)A (4.6) G := dC + DAj C ∧ C + EAC ; (6)A (5)A A (2)B (2)C (2)D A (3)i (3)j A (4)i (2)B G := dC + F(BCD)C ∧ C ∧ C + Gij C ∧ C + HBiC ∧ C , where the uppercase bold Latin tensors are constant4, and they are possibly given by suitable representation theoretic projectors5. Then, a general Lagrangian density can be written as 1 1 L = R − A G(3)A · G(3)B − B G(4)i · G(4)j 2 · 3! AB 2 · 4! ij 1 1 − C G(5)i · G(5)j − D G(6)A · G(6)B + L , (4.7) 2 · 5! ij 2 · 6! AB CS-like

4 All (uppercase and calligraphic) Latin tensors introduced in (4.7)-(4.9) are constant, because there is no scalar field in the (massless) spectrum of the theory. 5 Here we will not analyze possible characterizations of such tensor as (invariant) projectors. We confine ourselves to remark that, in a very simple choice of covariance (namely, A = 1, 2 and i = 1, 2,..., 6 running over the spin-1/2 and spin-5/2 representations 2 and 6 of sl2), most of them vanish. 11 where the calligraphic Latin constant tensors are (symmetric and) positive definite in order for all kinetic terms of p-forms to be consistent. A minimal, Maxwell-like choice is AAB = DAB = δAB and Bij = Cij = δij , such that (4.7) simplifies down to

2 6 1 1 L = R − G(3)AG(3)A|µνρ − G(4)i G(4)i|λµνρ 2 · 3! µνρ 2 · 4! λµνρ AX=1 Xi=1 6 2 1 1 − G(5)i G(5)i|λµνρσ − G(6)A G(6)A|λµνρστ + L . (4.8) 2 · 5! λµνρσ 2 · 6! λµνρστ CS-like Xi=1 AX=1 The “topological”, “Chern-Simons-like” Lagrangian occurring in (4.7) and (4.8) is composed by a number of a priori non-vanishing terms, such as for instance :

ABCDi (6) (6) (6) (6) (3) |g|LCS = ǫE3 GA GB GC GD Ci p ijklmA (5) (5) (5) (5) (5) (2) +ǫI2 Gi Gj Gk Gl Gm CA + ... ijklmnp (4) (4) (4) (4) (4) (4) (3) +ǫS3 Gi Gj Gk Gl Gm Gn Cp + ... ABCDEFGHi (3) (3) (3) (3)j (3) (3) (3) (3) (3) +ǫW3 GA GB GC GD GE GF GG GH Ci + ... (4.9) where the full Lagrangian is shown in Appendix (A). We leave the study of the constant tensors A, ..., H, A, ..., D, and E, ..., W. in (4.6), (4.7), and (4.9), respectively, for further future work. It is immediate to realize that M2-theory includes Horowitz and Susskind’s bosonic M-theory [23] as a truncation; indeed, by setting

C(2)A = 0; C(3)i = δi1C; (4.10) C(4)i = 0; C(5)A =0, one obtains (F = dC) 1 L = R − F 2, (4.11) 2 · 4! matching the Lagrangian discussed in [23]. Finally, we observe that a Scherk-Schwarz reduction of the Lagrangian (4.7) to 25 + 1 would provide a quite general Lagrangian for 0.α.iii gravito-dilatonic theory; we leave this task for further future work.

B. B = F in 26 + 1

It is remarkable to notice that a certain subsector of M2-theory, when coupled to an h =3/2 ﬁeld, exhibits an equal number of bosonic and fermionic degrees for freedom (namely, B = F ), which is a necessary condition for (linearly realized) supersymmetry to hold. Such a subsector of M2-theory is given by

ﬁeld R of so25 # g : 324 1 ∧1 25 0 ϕ : 1 0 (4.12) ∧5 : 53, 130 1 ∧4 : 12, 650 3 ∧3 : 2, 300 3 ∧2 : 300 0

Thus, when coupled to a an h = 3/2 RS ﬁeld ψ ﬁtting the 98, 304 irreducible representation (with Dynkin labels 1, 010, 1 ), the resulting theory has B = F = 98, 304. (4.13) 12

Trivially, bosonic M-theory [23] is a subsector of (the purely bosonic sector of) such a theory in 26 + 1. By recalling (4.3) and observing that the RS representation branches from 26+1 to 25+1 as

98, 304 = 47, 104 ⊕ 47, 104′ ⊕ 2, 048 ⊕ 2, 048′, (4.14) ψ ψ ψ′ λ λ′ the theory with B = F = 98, 304 gives rise to the following massless spectrum, when reduced to 25 + 1 :

ﬁeld R of so24 # g : 299 1 ψ : 47, 104 2 ≡ (ψ ⊕ ψ′) ∧1 : 24 1 λ : 2, 048 2 ≡ (λ ⊕ λ′) (4.15) ϕ : 1 1 ∧5 : 42, 504 1 ∧4 : 10, 626 4 ∧3 : 2, 024 6 ∧2 : 276 3

By recalling the treatment of previous Section, one may recognize (4.15) as a subsector (in which (4.13) holds) of the Monstrous gravity 2.γ.ii in (3.21), simply obtained by reducing the number of 2-forms from 4 to 3. Finally, we would like to point out that other examples of subsectors of Monstrous gravity in 25 + 1 exist such that B = F ; below, we list some of them :

s1 s2 B = F 0.γ.i − iv (0, 0, 1, 1, 0) (0, 0, 1, 0) 2, 048 α.i − ii β.i − ii 2.  (0, 2, 0, 0, 0) (1, 4, 4, 4) 94, 208 (4.16)  γ.i − ii δ.i − ii  2.γ.ii (0, 2, 1, 1, 0) (1, 4, 5, 4)  96, 256. 2.δ.ii (1, 2, 0, 1, 1) (1, 4, 5, 3)

In the ﬁrst column we report the labels pertaining to the possible parent theories among the 60 Monstrous gravity theories in 25 + 1 classiﬁed above. Note that, among the B = F subsectors in 25+ 1 reported above, only (4.15) and the second in the last line of (4.16) (i.e., the subsector of the 2.δ.ii Monstrous gravity) contain gravity.

1. = 1 Supergravity in 26 + 1? N As pointed out, B = F is just a necessary but not suﬃcient condition for supersymmetry to hold. Nevertheless, it is tantalizing to conjecture that the theory in 26 + 1 whose massless spectrum is given by (4.12) and one Rarita- Schwinger ﬁeld ψ is actually an N = 1 supergravity theory. Inspired by M-theory6 (i.e., N = 1 supergravity) in 10 + 1, and exploiting a truncation of the purely bosonic Lagrangians discussed in Sec. IVA (the capped lowercase

6 Throughout our treatment, we refer to the conventions used in Sec. 22 of [36]. 13

Latin indices run ˆı =1, 2, 3 throughout), one can also write down a tentative Lagrangian for such a theory :

3 3 1 1 1 L = R − G(4)ˆı · G(4)ˆı − G(5)ˆı · G(5)ˆı − G(6) · G(6) + L 2 · 4! 2 · 5! 2 · 6! CS-like Xˆı=1 Xˆı=1

i µνρ ω +˜ω −a ψ Γ ∇ν ψρ 2 µ 2 3 [µ (4) ν] (4)ˆı ˜(4)ˆı + bˆıψµΓ Γ Γ ψν · G + G Xˆı=1 3 [µ (5) ν] (5)ˆı ˜(5)ˆı + cˆıψµΓ Γ Γ ψν G + G Xˆı=1 [µ (6) ν] (6) ˜(6) +dψµΓ Γ Γ ψν · G + G , (4.17) where

(4) (4)ˆı (4)ˆı|αβγδ Γ · G =ΓαβγδG ,etc. (4.18) and, upon truncation of (4.6) resp. (4.9),

(4)ˆı (3)ˆı (4)ˆ G := dC + CˆıˆC ; G(5)ˆı := dC(4)ˆı + EˆıC(5); (4.19) (6) (5) (3)ˆı (3)ˆ G := dC + GˆıˆC ∧ C ;

(6) (6) (6) (6) (3)ˆı |g|LCS-like = ǫEˆıG G G G C p (4)ˆı (4)ˆ (4)kˆ (4)ˆl (4)m ˆ (4)ˆn (3)ˆp +ǫGˆıˆkˆˆlmˆ nˆpˆG G G G G G C (6) (6) (4)ˆı (4)ˆ (4)kˆ (3)ˆl +ǫHˆıˆkˆlˆG G G G G C (6) (5)ˆı (5)ˆ (4)kˆ (4)ˆl (3)m ˆ +ǫIˆıˆkˆlˆmˆ G G G G G C (6) (6) (6) (5)ˆı (4)ˆ +ǫJˆıG G G G C (6) (6) (6) (4)ˆı (5) +ǫKˆıG G G G C (6) (5)ˆı (4)ˆ (4)kˆ (4)ˆl (4)m ˆ +ǫLˆıˆkˆˆlmˆ G G G G G C (6) (6) (5)ˆı (5)ˆ (5) +ǫMˆıˆG G G G C (6) (4)ˆı (4)ˆ (4)kˆ (4)ˆl (5) +ǫNˆıˆkˆlˆG G G G G C (5)ˆı (5)ˆ (5)kˆ (5)ˆl (4)m ˆ (3)ˆn +ǫOˆıˆkˆlˆmˆ nˆ G G G G G C (5)ˆı (5)ˆ (5)kˆ (4)ˆl (4)m ˆ (4)ˆn +ǫPˆıˆkˆˆlmˆ nˆ G G G G G C . (4.20) Moreover,

G˜(4)ˆı := G(4)ˆı + eˆıψΓ(2)ψ; G˜(5)ˆı := G(5)ˆı + fˆıψΓ(3)ψ; (4.21) G˜(6) := G(6) + gψΓ(4)ψ are the supercovariant ﬁeld strengths, and

ab ∇µ (ω) ψν := ∂µψν + hωµ Γabψν (4.22) is the covariant derivative with

ab ab abαβ ω˜µ : = ωµ + ilψαΓµ ψβ; (4.23) ab ab ab ωµ : = ωµ (e)+ Kµ ; (4.24) ab abαβ b a a b b a Kµ : = i mψαΓµ ψβ + n ψµΓ ψ − ψµΓ ψ + ψ Γµψ . (4.25) h i 14

The Lagrangian (4.17) should be invariant under the following local supersymmetry transformations with parameter ε (a Majorana spinor) : i δ ea = − ε¯Γaψ ; (4.26) ε µ 2 µ 3 αβγδ βγδ α ˜(4)ˆı δεψµ = p∇µ (˜ω) ε + qˆı Γ µ + rˆıΓ δµ εGαβγδ Xˆı=1 3 αβγδρ βγδρ α ˜(5)ˆı αβγδρσ βγδρσ α ˜(6) + sˆı Γ µ + tˆıΓ δµ εGαβγδρ + u Γ µ + vΓ δµ εGαβγδρσ; (4.27) Xˆı=1 (3)ˆı ˆı δεCµνρ = w ε¯Γ[µν ψρ]; (4.28) (4)ˆı ˆı δεCµνρσ = x ε¯Γ[µνρψσ]; (4.29) (5) δεCµνρστ = yε¯Γ[µνρρψτ]. (4.30) To prove (or disprove) the invariance of the Lagrangian (4.17) (with deﬁnitions (4.18)-(4.25)) under the local supersymmetry transformations (4.26)-(4.30), and thus ﬁxing the real parameters a,...,y as well as the tensors C, E, G and E, G, seems a formidable task, which deserves to be pursued in a separate paper. Under dimensional reduction to 25+ 1, one would then get a would-be type IIA N = (1, 1) supergravity theory; as observed above, this would correspond to a suitable truncation of the Monstrous gravity 2.γ.ii in (3.21); again, we leave this interesting task for further future work.

V. COHOMOLOGICAL CONSTRUCTION OF LATTICES

Let us consider the following (commutative) diagram, starting from the Lie algebra e8,

3 3 a8 ⊕ ∧ ⊕ ∧ ր ց 3 3 e8 b4 ⊕ g ⊕ ∧ ⊕ ∗∧ (5.1) ց ր d8 ⊕ λ

2 where g := S0 denotes the D = 10+1 graviton representation (which has been related to “super-Ehlers” embeddings p D−p in [37]), and ∗ stands for the Hodge dual (∗∧ := ∧ ). Thus, the number #e8 of roots of the e8 root lattice reads

3 3 #e8 = dim e8 − 8 = (dim b4 − 4) + (dim g − 4) + dim(∧ ⊕ ∗∧ ). (5.2) 240 248 32 40 84·2

+ Therefore, the number #e8 of positive roots of e8 is

+ 1 3 #e8 = (dim b4 − 4) + (dim g − 4) + dim ∧ = 120. (5.3) 2 32 40 84 Note that it also holds that 1 (dim b4 − 4) + (dim g − 4) = dim b4. (5.4) 2 32 40

10 It should be also remarked that 120 = 3 , i.e. it matches the number of degrees of freedom of a massless 3-form 3 3 2 3 potential in 11 + 1 space-dimensions; indeed, ∧ in 11 + 1 gives rise to ∧ ⊕ ∧ = ∧ ⊕ b4 in 10 + 1.

The case of e8 is peculiar, because the closure (as well as the commutativity) of the diagram (5.1) relies on the existence of the “anomalous” embedding

d ⊃ b ; 8 4 (5.5) 16 = 16 ≡ λ, where λ is the spinor representation. 15

By replacing b4 and ∧ respectively as follows,

b4 → b12; (5.6) ∧3 → ∧5 ⊕ 3 · ∧4 ⊕ 3 · ∧3, (5.7) one can deﬁne the “Leech algebra” L24 in analogy with e8 (albeit with D = 26+1 graviton g), through the following diagram :

5 4 3 a24 ⊕ ∧ ⊕ 3 · ∧ ⊕ 3 · ∧ ⊕ ∧5⊕ 3 · ∧4 ⊕ 3 · ∧3 ր ց b ⊕ g ⊕ ∧5 ⊕ 3 · ∧4 ⊕ 3 · ∧3 (5.8) L 12 24 ⊕ ∗ ∧5 ⊕3 · ∧4 ⊕ 3 · ∧3 ց ր ?

The question mark in (5.8) occurs because there is no analogue of the “anomalous embedding” (5.5) for L24. Thus, it holds that

#L24 = dim L24 − 24 196,560 196,584 5 4 3 5 4 3 = (dim b12 − 12) + (dim g − 12) + dim(∧ ⊕ 3 · ∧ ⊕ 3 · ∧ + ∗ ∧ ⊕ 3 · ∧ ⊕ 3 · ∧ ) 288 312 2·(53,130+3·12,650+3 ·2,300) = 196, 560, (5.9) where #L24 denotes the number of minimal, non-trivial vectors (thus, of norm 4) of the Leech lattice Λ24. Therefore, the Z2-modded number of minimal, non-trivial vectors of Λ24 is

+ 1 5 4 3 #L24 = (dim b12 − 12) + (dim g − 12) + dim( ∧ ⊕ 3 · ∧ ⊕ 3 · ∧ )=98, 280, (5.10) 2 288 312 53,130+3·12,650+3·2,300 which is the number entering the construction of the minimal non-trivial representation of the Monster group M (cfr. [39]). Note that it also holds that

1 (dim b12 − 12) + (dim g − 12) = dim b12. (5.11) 2 288 312

28 It should be also remarked that 98, 280 = 5 , i.e. it matches the number of degrees of freedom of a massless 5-form potential in 29+1 space-dimensions; indeed, it can be checked that ∧5 in 29+1 gives rise to ∧5 ⊕ 3 · ∧4 ⊕ 3 · ∧3 ⊕∧2 = 5 4 3 ∧ ⊕ 3 · ∧ ⊕ 3 · ∧ ⊕ b12 in 26 + 1. (5.3) and (5.10) deﬁne a cohomological construction of the 8-dimensional e8 root lattice and the 24-dimensional Leech lattice Λ24, respectively based on the analogy between :

• M theory in 10 + 1 space-time dimensions, with SO9 massless little group and massless spectrum given by 128 (gravitino) = 84 (3-form potential)+44 (graviton); this corresponds to D0-branes (supergravitons) in BFSS M(atrix) model, carrying 256 = 128(B)+ 128(F ) Kaluza-Klein states [24];

• the would-be N = 1 supergravity in 26 + 1 space-time dimensions, with SO25 massless little group and massless spectrum given by 98, 304 (would-be gravitino) = 3 · 2, 300 + 3 · 12, 650 + 53, 130 (set of massless p-forms which is the “(26 + 1)-dimensional analogue” of the 3-form in 10 + 1)+324 (graviton); this would correspond to D0-branes (would-be “supergravitons”) in the would-be BFSS-like M(atrix) model, carrying 196, 608 = 98, 304(B)+98, 304(F ) KK states.

There are many analogies, but the big diﬀerence is supersymmetry in 26+1 (and possibly in 25+1), whose nature is at present still conjectural. The “Leech algebra” L24 encodes dim su25 = 624 = 324+300, and 2 · 97, 980= 2 · (3 · 2, 300+3 · 12, 650+53, 130) = 195, 960 to get 624+195, 960 = 196, 584. Removing the 12+12 = 24 Cartans gives 196, 560, the number of minimal Leech vectors. It is thus tempting to conjecture “Monstrous supergravitons” as D0 branes, as L24 “sees” 98, 304 of the bosonic KK states. On the other hand, the Monster M acts on almost all of these, albeit seeing only 299 + 1 16 of the 324 graviton degrees of freedom from 324 + 300, giving 299 + 1+(300+97, 980) = 299+1+98, 280 of the Griess algebra [1, 4]. More concisely, the relation between the “Leech algebra” L24 and the Griess algebra is precisely given by the relation between M2-theory and its subsector coupled to a RS ﬁeld (the would-be gravitino) in 26 + 1, discussed in Sec. IV B. The dimensional reduction 26+1 −→ 10+1 can proceed along a decomposition proved by Wilson [32], characterizing the aforementioned number of minimal Leech vectors as

196, 560= 3 · 240 · (1+16+256). (5.12)

We identify 1 + 16 + 256 = 273 with the (Hermitian part of) Vinberg’s T-algebra7 [26, 33]

1α 16 128 8,2 ′ T3 =  ∗ 1β 128  (5.13) ∗ ∗ 1γ  so16 covariant  with spin factor lightcone coordinates 1α and 1β removed, giving 128+128+16+1= 273 dimensions. The spin factor 8,2 8 8,2 1α ⊕ 1β ⊕ 16 of T3 (5.13) enjoys an enhancement from so16 to so17,1 Lorentz symmetry , and der T3 = so17 [28]. Breaking the so25 Lie algebra of massless little group in 26 + 1 with respect to so17, as well as its 4096 spinor 3 (both encoded in the so-called “Magic Star algebra” f4 [27]), one obtains the decomposition

3 f4 := so25 ⊕ 4096 = so17 ⊕ so8 ⊕ (17, 8v) ⊕ (256, 8s) ⊕ (256, 8c). (5.14)

7 As so8 acts on S , one can take the 240 roots as forming a discrete 7-sphere, and the 273 is constructed as 17+256 = 273 by picking one of the 256 spinors. This gives a discrete form of the maximal Hopf fibration S7 ֒→ S15 → S8, and the three maps yield three charts of the form 196, 560 = 3 · 240 · 273 (cfr. (5.12)) in a discrete Cayley plane [25, 26, 31]. ′ Through the super Ehlers embedding [37], e8(8) = sl9 (R)⊕84⊕84 = so9 ⊕44⊕84⊕84, we can identify each discrete 7 S fiber of 240 E8 roots with the M2 and M5 brane gauge fields of D = 10 + 1 M-theory, as well as with little group (so9) and graviton (44) degrees of freedom, albeit with all 4 + 4 Cartans removed. This is understood with so9 ⊂ so25 acting isometrically on the S8 base. From this perspective, the reduction from D =26+1 to D = 10+1 occurs first along three charts, and gauge and gravity data are encoded in discrete S7 chart fibers therein. This picture is further supported by noting the Conway group Co0 is a maximal finite subgroup of SO24, and that Co0 can be generated by unitary 3 × 3 octonion matrices [32] of F4 type. In general, the stabilizer subgroup of 3 × 3 unitary matrices over the octonions lies in SO9 ⊂ F4 through Peirce decomposition; since there are three independent primitive idempotents O in the exceptional Jordan algebra J3 , there are three such embedded copies of SO9, providing three charts for the reduction to D = 10 + 1. As it is well known, a remarkable class of M-theory compactifications is provided by G2 compactifications [34] to D = 3 + 1, where the internal manifold with G2 holonomy is characterized by its invariant 3-form (which comes from an octonionic structure). In the 26+1 framework under consideration, a compactification down to 3 + 1 dimensions can involve a 23-sphere S23, which in turn can be fibrated with an OP2 base and S7 fibers. 7 Since S is the quintessential G2 manifold [35], this provides a natural 26+1 −→ 10+1 −→ 3+1 pattern of reduction along a G2 manifold from the Monstrous M-theory introduced above.

VI. CONCLUSION

We have shown that in 26 + 1 space-time dimensions there exists a purely bosonic theory (named Monstrous M- theory, or simply M2-theory), whose massless spectrum, of dimension 196, 884, may be acted upon by the Monster group M. A subsector of such a theory yields Horowitz and Susskind’s bosonic M-theory [23]. When reducing to 25 + 1 space-time dimensions, a certain purely bosonic gravito-dilatonic theory (namely, the 0.α.iii theory, whose massless spectrum is given by (4.1)-(4.2)), arises. Remarkably, this theory contains a subsector given by the massless excitations of the closed and open bosonic string in 25 + 1, namely a graviton, an antisymmetric rank-2 ﬁeld, a dilaton, and a vector potential, respectively. By generalizing the triality of SO8 (massless little group of string theory in 9+1) to SO24 (massless little group of bosonic string theory in 25 + 1), such a theory can be shown to be actually part of a web of (at least) some 60 gravito-dilatonic theories, named Monstrous gravity theories. The

7 In (5.13) the Greek subscripts discriminate among so16-singlets. 8 so17,1 would be the Lorentz symmetry of the 18-dimensional string suggested by Lorentzian Kac-Moody algebras [38]. 17

9 relation between SO8 and SO24 can be interpreted in terms of the Conway group Co0, which is a maximal finite subgroup of SO24 itself; as shown by Wilson [32], Co0 is generated by unitary 3 × 3 octonion matrices, namely by F4 matrices. Interestingly, SO9 can be maximally embedded into F4 in three possible ways, each one providing the manifestly triality-symmetric breaking f4 → so9 = so8 ⊕ 8v ⊕ 8s ⊕ 8c; in this sense, no triality is needed for so24, but rather just the threefold nature of the (symmetric) embedding of SO9 in F4. In turn, the “anomalous” embedding F4(−20) ⊂ SO25,1 [29] allows one to reduce from 27D to lower dimensions in a non-trivial way, namely along the chain 27 → 26 → 11 → 4 dimensions. In turn, this, as remarked in [25], confirms and strengthens Ramond and Sati’s argument that D = 10 + 1 M-theory has hidden Cayley plane fibers [30]. It is worth recalling that the Moonshine decomposition 196, 884 = 196, 883⊕1 always holds in Monstrous gravities, due to the very existence of the dilatonic scalar field in their spectra. In particular, the dilaton φ is a singlet of M, and as such it coincides with the vacuum state |Ωi of the chiral Monster conformal theory discussed in [8, 17, 18]. On the other hand, Witten in [22] identifies the 1 as the stress energy tensor while the other 196, 883 are primary fields transforming in the smallest non-trivial rep of the Monster, according to FLM theory [5]. Remarkably, a certain subsector of M2-theory, when coupled to one massless Rarita-Schwinger field in 26 + 1, gives rise to a theory which has the same number of bosonic and fermionic massless degrees of freedom, namely B = F = 98, 304, for a total of 196, 608 degrees of freedom. It is then very tempting to ask whether this gravity theory can actually be supersymmetric in 26 + 1, giving rise to the would-be N = 1 supergravity in 26 + 1 space-time dimensions. In this perspective, we have conjectured a quite straightfowardly “M-theory-inspired” Lagrangian density as well as the corresponding local supersymmetry transformations in 26 + 1. The invariance of such a Lagrangian under those supersymmetry transformations is still conjectural, and to prove (or disprove) it seems quite a formidable, though absolutely worthy task, and we leave it for further future work. At any rate, the reduction of the bosonic sector of such a would-be N = 1 supergravity from 26+1 to 25+1 yields a suitable subsector of the Monstrous gravity 2.γ.ii, simply obtained by reducing the number of 2-forms from 4 to 3. It thus becomes evident that there is a certain mismatch, essentially amounting to the 276 degrees of freedom of a massless 2-form in 25+ 1, between the total degrees of freedom of the would-be N = 1 supergravity in 26 + 1 (98, 304+98, 304 = 196, 608) and the 196, 884 degrees of freedom of (purely bosonic) M2-theory : 196, 884−196, 608 = 276. Thus, “monstrousity” and “supersymmetry” in 26 + 1 (as well as, predictably, in 25 + 1) seem to be slightly mismatching, yet tightly related, concepts. All this suggests that the Monster group M has its origin in a gravity theory in 26 + 1 dimensions, as its definition as the automorphism of the Griess algebra [1, 5, 7] is clarified by showing that such an algebra is not merely a sum of unrelated spaces, but rather the whole massless spectrum of Monstrous gravities in 25 + 1, which in at least one case (namely, the 0.α.iii theory, whose massless spectrum is given by (4.1)-(4.2)) uplifts to Monstrous M-theory (i.e., M2-theory) in 26+1. The spectrum of M2-theory dimensionally reduced to 25 + 1 contains a subsector given by the massless excitations of the closed and open bosonic string in 25 + 1, namely a graviton, an antisymmetric rank-2 field, a dilaton, and a vector potential, respectively; then, since the Monster VOA originates from D = 25+1 bosonic string through FLM theory [5], this shows how the Monster group M acts on fields originating from D = 26 + 1. Unlike the FLM construction, the Leech lattice need not be assumed, but can be derived from D = 26+1 field content. The Z2 orbifold is then the elimination of the dual magnetic brane gauge fields, and half the degrees of freedom of the SO25 little group and graviton, giving the desired 98, 280. On the other hand, the discussion of the analogies between the e8 root lattice and the Leech lattice Λ24 seems to suggest that M-theory in 10 + 1 and the would-be N = 1 supergravity in 26 + 1 are tightly related to the lattices e8 resp. Λ24, which determine the optimal lattice packings in D = 8 resp. D = 24. All in all, it would be interesting to explore the implications of the characterization of the Monster group as acting on the whole massless spectrum of M2-theory in 26 + 1. Developments for future work may concern further study and formalization of the maps discussed in Sec. II; as pointed out above, no other Dynkin diagram (besides d4) has an automorphism group of order greater than 2, thus such maps cannot be realized as an automorphism of d12, nor they can be traced back to some structural symmetry of the Dynkin diagram of d12 itself. Moreover, the study of the Lagrangian structure of M2-theory, as well as of its Scherk-Schwarz reduction to 25 + 1, should be the object of further investigations. Last but not least, we leave the interesting issue of the study of the massive spectrum of (massive variants of) Monstrous gravities and M2-theory for further future work.

We would like to conclude with a sentence by John H. Conway, to whom this paper is dedicated, on the Mon-

9 The Conway group Co0 is the full automorphism of the Leech lattice Λ24; however, it is not a simple group, nor is it contained in the Monster. In fact, its quotient by its center Z2, namely the Conway simple group Co1 ∼ Co0/Z2 is contained in M. This means the Monster’s maximal ﬁnite subgroup Co1 has the Z2 action built in, which acts on only half the minimal Leech vectors 196, 560/2 = 98, 280. 18 ster group [40]: “There’s never been any kind of explanation of why it’s there, and it’s obviously not there just by coincidence. It’s got too many intriguing properties for it all to be just an accident.”

Acknowledgements

We would like to acknowledge an informal yet inspiring discussion with Eric Weinstein during the Advances in Quantum Gravity conference, hosted by Laura Deming at Topos House, San Francisco, CA, on July 2016. Since that occasion, we started thinking about the idea of a weak triality for d12. 19

Appendix A: Chern-Simons Lagrangian terms for Monstrous M-theory

The full Lagrangian from Eq. (4.9) is given by

ABCiDE (6) (6) (6) (4) (3) (2) ABijCD (6) (6) (5) (5) (3) (2) |g|LCS = εE2 GA GB GC Gi GD CE + εF2 GA GB Gi Gj GC CD p ABijkC (6) (6) (5) (4) (4) (2) AijklB (6) (5) (5) (5) (4) (2) +εG2 GA GB Gi Gj Gk CC + εH2 GA Gi Gj Gk Gl CB ijklmA (5) (5) (5) (5) (5) (2) ABiCDEF (6) (6) (4) (3) (3) (3) (2) +εI2 Gi Gj Gk Gl Gm CA + εJ2 GA GB Gi GC GD GE CF AijBCDE (6) (5) (5) (3) (3) (3) (2) AijkBCD (6) (5) (4) (4) (3) (3) (2) +εK2 GA Gi Gj GB GC GD CE + εL2 GA Gi Gj Gk GB GC CD ijklABC (5) (5) (5) (4) (3) (3) (2) AijklBC (6) (4) (4) (4) (4) (3) (2) +εM2 Gi Gj Gk Gl GA GB CC + εN2 GA Gi Gj Gk Gl GB CC ijklmAB (5) (5) (4) (4) (4) (3) (2) ijklmnA (5) (4) (4) (4) (4) (4) (2) +εP2 Gi Gj Gk Gl Gm GA CB + εW2 Gi Gj Gk Gl Gm Gn CA AiBCDEFG (6) (4) (3) (3) (3) (3) (3) (2) ijABCDEF (5) (5) (3) (3) (3) (3) (3) (2) +εQ2 GA Gi GB GC GD GE GF CG + εR2 Gi Gj GA GB GC GD GE CF ijkABCDE (5) (4) (4) (3) (3) (3) (3) (2) ijklABCD (4) (4) (4) (4) (3) (3) (3) (2) +εS2 Gi Gj Gk GA GB GC GD CE + εT2 Gi Gj Gk Gl GA GB GC CD iABCDEFGH (4) (3) (3) (3) (3) (3) (3) (3) (2) ABCDi (6) (6) (6) (6) (3) +εU2 Gi GA GB GC GD GE GF GG CH + εE3 GA GB GC GD Ci ABCDEi (6) (6) (6) (3) (3) (3) ABijCk (6) (6) (5) (4) (3) (3) +εF3 GA GB GC GD GE Ci + εG3 GA GB Gi Gj GC Ck AijkBl (6) (5) (5) (5) (3) (3) ABijkl (6) (6) (4) (4) (4) (3) +εH3 GA Gi Gj Gk GB Cl + εI3 GA GB Gi Gj Gk Cl Aijklm (6) (5) (5) (4) (4) (3) ijklmn (5) (5) (5) (5) (4) (3) +εJ3 GA Gi Gj Gk Gl Cm + εK3 Gi Gj Gk Gl Gm Cn ABCDEFi (6) (6) (3) (3) (3) (3) (3) AijBCDk (6) (5) (4) (3) (3) (3) (3) +εL3 GA GB GC GD GE GF Ci + εM3 GA Gi Gj GB GC GD Ck ijkABCl (5) (5) (5) (3) (3) (3) (3) AijkBCl (6) (4) (4) (4) (3) (3) (3) +εN3 Gi Gj Gk GA GB GC Cl + εO3 GA Gi Gj Gk GB GC Cl ijklABm (5) (5) (4) (4) (3) (3) (3) ijklmAn (5) (4) (4) (4) (4) (3) (3) +εP3 Gi Gj Gk Gl GA GB Cm + εR3 Gi Gj Gk Gl Gm GA Cn (A1) ijklmno (4) (4) (4) (4) (4) (4) (3) ABCDEFGi (6) (3) (3) (3) (3) (3) (3) (3) +εS3 Gi Gj Gk Gl Gm Gn Co + εT3 GA GB GC GD GE GF GG Ci ijABCDEk (5) (4) (3) (3) (3) (3) (3) (3) ijkABCDl (4) (4) (4) (3) (3) (3) (3) (3) +εU3 Gi Gj GA GB GC GD GE Ck + εV3 Gi Gj Gk GA GB GC GD Cl ABCDEFGHi (3) (3) (3) (3) (3) (3) (3) (3) (3) ABCij (6) (6) (6) (5) (4) +εW3 GA GB GC GD GE GF GG GH Ci + εE4 GA GB GC Gi Cj ABiCDj (6) (6) (5) (3) (3) (4) ABijCk (6) (6) (4) (4) (3) (4) +εF4 GA GB Gi GC GD Cj + εG4 GA GB Gi Gj GC Ck AijkBl (6) (5) (5) (4) (3) (4) ijklAm (5) (5) (5) (5) (3) (4) +εH4 GA Gi Gj Gk GB Cl + εI4 Gi Gj Gk Gl GA Cm Aijklm (6) (5) (4) (4) (4) (4) ijklmn (5) (5) (5) (4) (4) (4) +εJ4 GA Gi Gj Gk Gl Cm + εK4 Gi Gj Gk Gl Gm Cn AiBCDEj (6) (5) (3) (3) (3) (3) (4) AijBCDk (6) (4) (4) (3) (3) (3) (4) +εL4 GA Gi GB GC GD GE Cj + εM4 GA Gi Gj GB GC GD Ck ijkABCl (5) (5) (4) (3) (3) (3) (4) ijklABm (5) (4) (4) (4) (3) (3) (4) +εN4 Gi Gj Gk GA GB GC Cl + εO4 Gi Gj Gk Gl GA GB Cm ijklmAn (4) (4) (4) (4) (4) (3) (4) iABCDEFj (5) (3) (3) (3) (3) (3) (3) (4) +εP4 Gi Gj Gk Gl Gm GA Cn + εQ4 Gi GA GB GC GD GE GF Cj ijABCDEk (4) (4) (3) (3) (3) (3) (3) (4) ABCiD (6) (6) (6) (4) (5) +εR4 Gi Gj GA GB GC GD GE Ck + εE5 GA GB GC Gi CD ABijC (6) (6) (5) (5) (5) ABiCDE (6) (6) (4) (3) (3) (5) +εF5 GA GB Gi Gj CC + εG5 GA GB Gi GC GD CE AijBCD (6) (5) (5) (3) (3) (5) AijkBC (6) (5) (4) (4) (3) (5) +εH5 GA Gi Gj GB GC CD + εI5 GA Gi Gj Gk GB CC ijklAB (5) (5) (5) (4) (3) (5) AijklB (6) (4) (4) (4) (4) (5) +εJ5 Gi Gj Gk Gl GA CB + εK5 GA Gi Gj Gk Gl CB ijklmA (5) (5) (4) (4) (4) (5) AiBCDEF (6) (4) (3) (3) (3) (3) (5) +εL5 Gi Gj Gk Gl Gm CA + εM5 GA Gi GB GC GD GE CF ijABCDE (5) (5) (3) (3) (3) (3) (5) ijkABCD (5) (4) (4) (3) (3) (3) (5) +εN5 Gi Gj GA GB GC GD CE + εO5 Gi Gj Gk GA GB GC CD ijklABC (4) (4) (4) (4) (3) (3) (5) iABCDEFG (4) (3) (3) (3) (3) (3) (3) (5) +εP5 Gi Gj Gk Gl GA GB CC + εQ5 Gi GA GB GC GD GE GF CG . 20

[1] R. L. Griess, The structure of the monster simple group, Proceedings of the Conference on Finite Groups (Univ. Utah, Park City, Utah 1975), Boston, MA: Academic Press, 113-118 (1976). [2] J. H. Conway, S. Norton, Monstrous moonshine, Bull. London. Math. Soc. 11: 308-339 (1979). [3] J. H. Conway, A simple construction for the Fischer-Griess monster group, Inventiones Mathematicae 79 3: 513-540 (1985). [4] J.H. Conway, N.J.A. Sloane, Sphere Packings, Lattices and Groups, Grundlehren der mathematischen Wissenschaften v. 290, Springer (2013). [5] I. B. Frenkel, J. Lepowsky, A. Meurman, Vertex operator algebras and the monster, Academic press (1988). [6] R. E. Borcherds, Monstrous moonshine and monstrous lie superalgebras, Inventiones Mathematicae 109 1: 405-444 (1992). [7] R. E. Borcherds, What is the monster?, Notices of the A.M.S 49: 9 1076-1077 (2002), arXiv:math/0209328 [math.GR]. [8] L. Dixon, P. Ginsparg, J. Harvey, Beauty and the beast: superconformal symmetry in a Monster module, Comm. Math. Phys. 119 2: 221-241 (1988). [9] T. Eguchi, H. Ooguri, Y. Tachikawa, Notes on the K3 Surface and the Mathieu group M24, Exper. Math 20: 91-96 (2011), arXiv:1004.0956 [hep-th]. [10] M. C. N. Cheng, J. F. R. Duncan, J. A. Harvey, Umbral Moonshine, Commun. Num. Th. Phys. 8 2: 101-242 (2014), arXiv:1204.2779 [math.RT]. [11] J. F. R. Duncan, M. J. Griffin, K. Ono, Proof of the Umbral Moonshine Conjecture, Res. Math. Sci. 2: 26 (2015), arXiv:1503.01472 [math.RT]. [12] N. M. Paquette, D. Persson, R. Volpato, Monstrous BPS-Algebras and the Superstring Origin of Moonshine, arXiv:1601.05412 [hep-th]. [13] J. B. Bae, J. A. Harvey, K. Lee, S. Lee, B. C. Rayhaun, Conformal Field Theories with Sporadic Group Symmetry, arXiv:2002.02970 [hep-th]. [14] B. Craps, M. R. Gaberdiel, J. A. Harvey, Monstrous branes, arXiv:hep-th/0202074. [15] M. P. Tuite, On the Relationship between the Uniqueness of the Moonshine Module and Monstrous Moonshine, Commun. Math. Phys. 166: 495 (1995), arXiv:hep-th/9305057. [16] C. J. Cummins, T. Gannon , Modular equations and the genus zero property of moonshine functions, Inventiones Mathe- maticae 129: 413-443 (1997). [17] L. Dolan, P. Goddard, P. Montague, Conformal field theory, triality and the Monster group, Phys. Lett. B236, 165 (1990). [18] L. Dolan, P. Goddard, P. Montague, Conformal field theory of twisted vertex operators, Nucl. Phys. B338, 529 (1990). [19] I. Porteous, “Clifford algebras and the classical groups”, Cambridge Studies in Advanced Mathematics 50, Cambridge University Press (1995). [20] J. F. Adams, Spin(8 ), Triality, F4 and all that, in : “Superspace and supergravity”, S. Hawking and M. Roˇcek, Cambridge University Press (1981). [21] E. Witten, String Theory Dynamics In Various Dimensions, Nucl. Phys. B443: 85-126 (1995), arXiv:hep-th/9503124. [22] E. Witten, Three-Dimensional Gravity Revisited, arXiv:0706.3359 [hep-th]. [23] G.T. Horowitz, L. Susskind, Bosonic M-theory, J. Math. Phys. 42: 3152 (2001), arXiv:hep-th/0012037. [24] T. Banks, W. Fischler, S. H. Shenker, L. Susskind, M Theory As A Matrix Model: A Conjecture, Phys. Rev. D 55: 5112-5128 (1997), arXiv:hep-th/9610043. [25] M. Rios, A. Marrani, D. Chester, Exceptional Super Yang-Mills in D = 27 + 3 and Worldvolume M-Theory, arXiv:1906.10709 [hep-th]. [26] M. Rios, A. Marrani, D. Chester, The Geometry of Exceptional Super Yang-Mills Theories, Phys. Rev. D 99, 046004 (2019), arXiv:1811.06101 [hep-th]. [27] Piero Truini, Michael Rios, Alessio Marrani, The Magic Star of Exceptional Periodicity, Cont. Mathematics, vol. 721, AMS, arXiv:1711.07881 [hep-th]. [28] P. Truini, W. De Graaf, A. Marrani, Exceptional Periodicity and Magic Star Algebras. III : The Algebra f(n)4 and the Derivations of HT-Algebras; in preparation. [29] P. Ramond, Exceptional groups and physics, talk at the 24th International Colloquium on Group Theoretical Methods in Physics (GROUP 24), 15-20 Jul 2002, Paris, France; Inst. Phys. Conf. Ser. 173 (2003), arXiv:hep-th/0301050. [30] H. Sati, OP2 bundles in M-theory, Commun. Num. Theor. Phys. 3 (2009) 495-530, arXiv:0807.4899 [hep-th]. [31] M. Rios, U-Duality and the Leech Lattice, arXiv:1307.1554 [hep-th]. [32] R. A. Wilson, Octonions and the Leech lattice, J. Algebra 322: 2186-2190 (2009). [33] E.B. Vinberg, The theory of Convex Homogeneous Cones, in : Transactions of the Moscow Mathematical Society for the year 1963, 340-403, American Mathematical Society, Providence RI 1965. [34] S. Gukov, M-theory on manifolds with exceptional holonomy, Fortschr. Phys. 51 (2003), 719–731. [35] D. Joyce, “Compact Manifolds with Special Holonomy”, Oxford University Press (2000). [36] T. Ort´ın, Gravity and Strings, 2nd edition, Cambridge Univ. Press (2015). [37] S. Ferrara, A. Marrani, M. Trigiante, Super-Ehlers in Any Dimension, JHEP 11: 068 (2012), arXiv:1206.1255 [hep-th]. [38] M. R. Gaberdiel, D. I. Olive, P. C. West, A class of Lorentzian Kac-Moody algebras, Nucl. Phys. B 645: 403-437 (2002), arXiv:hep-th/0205068 [39] J. H. Conway, The √Monster construction, in: “Groups, Difference Sets, and the Monster” (Arasu, Dillon et al. Eds.), De Gruyter (1996), pp. 345-348. [40] B. Haran, “Life, Death and the Monster (John Conway)” (2014).