The arithmetic of Borcherds' exponents (Q1416721): Difference between revisions

From the text: ``Recently, \textit{R. E. Borcherds} [Invent. Math. 120, No. 1, 161--213 (1995; Zbl 0932.11028)] provided a striking description for the exponents in the naive infinite product expansion of many modular forms. For example, if \(E_k(z)\) denotes the usual normalized weight \(k\) Eisenstein series, let \(c(n)\) denote the integer exponents one obtains by expressing \(E_4(z)\) as an infinite product: \[ E_4(z) = 1 + 240\sum_{n=1}^\infty\sum_{d\mid n}d^3q^n= (1 - q)^{-240}(1-q^2)^{26760}\cdots =\prod_{n=1}^\infty(1 - q^n)c(n) \] (\(q := e^{2\pi iz}\) throughout). Although one might not suspect that there is a precise description or formula for the exponents \(c(n)\), Borcherds provided one. He proved that there is a weight \(1/2\) meromorphic modular form \[ G(z) = \sum_{n\geq -3}b(n)q^n = q^{-3} + 4 - 240q + 26760q^4 +\dots -4096240q^9 + \dots \] with the property that \(c(n) = b(n^2)\) for every positive integer \(n\). It is natural to examine other methods for studying such exponents. Here we point out a \(p\)-adic method which is based on the fact that the logarithmic derivative of a meromorphic modular form is often a weight two p-adic modular form. Let \(K\) be a number field and let \(O_v\) be the completion of its ring of integers at a finite place \(v\) with residue characteristic \(p\). Moreover, let \(\lambda\) be a uniformizer for \(O_v\). Following Serre, we say that a formal power series \(f = \sum_{n=0}^\infty a(n)q^n \in O_v[[q]]\) is a \(p\)-adic modular form of weight \(k\) if there is a sequence \(f_i \in O_v[[q]]\) of holomorphic modular forms on \(\text{SL}_2(\mathbb Z)\), with weights \(k_i\), for which \(\text{ord}_\lambda(f_i - f ) \to +\infty\) and \(\text{ord}_\lambda(k - k_i)\to +\infty\). Theorem 1. Let \(F(z) = q^h \left(1 +\sum_{n=1}^\infty a(n)q^n\right)\in O_K[[q]]\) be a meromorphic modular form on \(\text{SL}_2(\mathbb Z)\), where \(O_K\) is the ring of integers in a number field \(K\). Moreover, let \(c(n)\) denote the numbers defined by the formal infinite product \[ F(z) = q^h \prod_{n=1}^\infty (1-q^n)c(n). \] If \(p\) is prime and \(F(z)\) is good at \(p\) (see the paper for the definition), then the formal power series \[ B = h -\sum_{n=1}^\infty \sum_{d\mid n} c(d)dq^n \] is a weight two \(p\)-adic modular form.'' Although the authors emphasize those forms \(F(z)\) which have Heegner divisors, it is stressed that Theorem 1 holds for many forms which do not have a Borcherds product. For example, \(E_{p-1}(z)\) is good at \(p\) for every prime \(p\geq 5\). Therefore their Theorem 2 describes some forms with Heegner divisors which are good at a prime \(p\). They also note that Theorem 2 admits a generalization to those forms with algebraic integer coefficients and Heegner divisors and that it has interesting consequences regarding class numbers of imaginary quadratic fields (Corollary 3).