Reflection groups of Lorentzian lattices (Q1580259)

The main idea of the paper is as follows. If an even lattice \(L\) in \((n+1)\)-dimensional Lorentzian space of signature \(1-n\) is ``interesting'', then there should be a nonzero so-called reflective modular form for \(L\). Such a form, of weight \((1-n)/2\), is allowed to have poles at cusps, and these are restricted by a technical condition involving the discriminant quadratic form of \(L\). The first eleven sections of the paper summarize the necessary background on modular forms, the main section 12 then gives many impressive examples supporting the principle above for levels \(N=1,\ldots,30\). For a fixed \(N\) the following program is carried through: Work out the relevant ring of (holomorphic) modular forms, try to find (meromorphic) modular forms with reflective singularities for each discriminant form, find corresponding lattices and see if they are connected with interesting reflection groups, automorphic forms on Grassmannians, moduli spaces or Lie algebras. The author points out that, however, he does ``not have any sort of proof that interesting Lorentzian reflection groups always correspond to reflective modular forms (or even a good definition what an interesting group is)''. So in the final Section 13 there are many open questions suggested for further research.