Rational period functions and indefinite binary quadratic forms. II (Q2641330): Difference between revisions

[For Part I, cf. Math. Ann. 286, No. 4, 697--707 (1990; Zbl 0673.10018)]. A rational period function of weight \(2k\) for \(\text{SL}_ 2{\mathbb Z}\) is a rational function \(q(\tau)\) satisfying the two relations \[ q(\tau)+\tau^{-2k}q((-1)/\tau)=0,\quad q(\tau)+\tau^{-2k}q((\tau - 1)/\tau)+(\tau -1)^{-2k}q((-1)/(\tau -1)). \] It has been known that the finite poles of rational period functions occur only at zero or at real quadratic irrationals. By looking at possible minimal or irreducible systems of poles it is possible to classify rational period functions of positive weight. If the minimal pole set contained \(\beta\), the conjugate of \(\alpha\), then the rational period function of weight \(2k\), \(k\) odd, with poles only at elements of the pole set was of the form \[ \sum_{{a>0>c,\atop (a,b,c)\in A}}\frac{1}{(a\tau^ 2-b\tau +c)^ k}, \] where the sum is over primitive indefinite binary quadratic forms \((a,b,c)\) in a narrow equivalence class \(A\). The situation when the irreducible pole set fails to be symmetric is studied in this paper. Given a pole set which is minimal and not symmetric, it is always possible to construct a rational period function of weight \(2k\) with poles only at elements of that pole set whenever \(k=1,2,3,4,5\), or \(7\). For all other weights certain obstructions occur in the construction which may act as barriers to the existence of rational period functions. In fact, the space of such obstructions is isomorphic to \(S_{2k}\oplus S_{2k}\) where \(S_{2k}\) is the space of cusp forms of weight \(2k\). By the way of example an infinite family of irreducible pole sets is given for which it is impossible to construct a rational period function of weight \(12\) with poles in exactly one of the pole sets.