Modular moonshine. II (Q1918384)

It is well known that the monster simple group acts on the monster vertex algebra, and the moonshine conjectures state that the traces of elements of the monster on the vertex algebra are hauptmoduls. \textit{A. J. E. Ryba} [Contemp. Math. 193, 307-336 (1996; Zbl 1032.17052)] conjectured the existence of similar vertex algebras over fields of characteristic \(p\) acted on by the centralizers of certain elements of prime order \(p\) in the monster, and he conjectured that the Brauer traces of \(p\)-regular elements of the centralizers were certain hauptmoduls. Here the authors prove these conjectures when the centralizer involves a sporadic group \((p \leq 11\), corresponding to the sporadic groups \(B\), \(Fi_{24}'\), \(Th\), \(HN\), \(He\), and \(M_{12})\). For the proof of the modular moonshine conjectures for some elements \(g\) given in this paper, the authors observe (Proposition 2.3) that the definitions of Ryba's vertex algebras \(^gV\) in Part I (loc. cit.) is equivalent to defining it to be the Tate cohomology group \(\widehat H^0(g,V)\), or at least it would be if a good integral form \(V\) of the monster Lie algebra was known to exist. It is difficult to work out the dimension of \(\widehat H^0(g, V[1/2]_n)\) directly. However, it is easy to work out the difference of the dimensions of \(\widehat H^0\) and \(\widehat H^1\), which can be thought of as a sort of Euler characteristic (and is closely related to the Herbrand quotient in number theory). This suggests that one should really be looking at \(\widehat H^*(g,V [1/2])= \widehat H^0(g,V[1/2]) \oplus\widehat H^1(g,V[1/2])\), and it turns out for essentially formal reasons that this has the structure of a vertex superalgebra. The traces of elements \(h\) of the centralizer of \(g\) on this superalgebra are just given by the traces of the elements \(gh\) on the monster vertex algebra. The proof of the modular moonshine conjectures for the element \(g\) of the monster is now completed by showing that \(\widehat H^1(g,V [1/2])=0\). This is done for certain elements of odd prime order of the monster coming from \(M_{24}\); this is a long but straightforward calculation. There are plenty of elements \(g\) of the monster for which \(\widehat H^1(g,V[1/2])\) does not vanish; this happens whenever some coefficient of the hauptmodul of \(g\) is negative; for example \(g\) of type 3B. There are also some elements of large prime order for which the authors have been unable to prove the modular moonshine conjectures because they have not proved that \(\widehat H^1(g,V [1/2])=0\) (though this is probably true), but they do at least cover all the cases when \(g\) has prime order and its centralizer involves a sporadic simple group. The case when \(p=2\) has several extra complications due partly to the existence of several extensions of groups by 2-groups, and due partly to the fact that the Frenkel-Lepowsky-Meurman (FLM) construction can be carried out over \(\mathbb{Z}[1/2]\) but it is not clear how to do it over \(\mathbb{Z}\). They deal with these extra problems by using a \(\mathbb{Z} [1/3]\)-form of the monster vertex algebra. To construct this \(\mathbb{Z}[1/3]\)-form, they need to make a mild assumption (which has not been checked) about the construction of the monster vertex algebra from an element of order 3 announced by Dong and Mason and by Montague. (In particular, the proof of the modular moonshine conjectures for elements of type 2A in the monster uses this assumption.) Then they give some calculations illustrating the case when \(g\) is an element of type 7A in the monster (so the centralizer \(C_M(g)\) of \(g\) is \(\langle g\rangle \times He)\). They give the characters and the decomposition into irreducible modular representations of the first few graded pieces of \(\widehat H^0(g,V [1/2])\). Finally, they discuss some open problems, in particular, how one might construct a good integral form of the monster vertex algebra. For Part III, see Duke Math. J. 93, 129-154 (1998; Zbl 1047.17014).