On a relation between modular functions and Dirichlet series: found in the estate of Adolf Hurwitz (Q2014262): Difference between revisions

Consider the functional equation \(\pi^{-s/2}\*\Gamma(s/2)\*f(s)=\pi^{-(1-s)/2}\*\Gamma((1-s)/2)\*f(1-s)\). \textit{B. Riemann} [``Über die Anzahl der Primzahlen unterhalb einer gegebenen Größe'', Königl.\ Preuss.\ Akad.\ Wiss.\ Berlin Monatsber., 671--680 (1859)] proved that the Riemann zeta-function is a solution of this equation, and \textit{H. Hamburger} [Math. Z. 10, 240--254 (1921; JFM 48.1210.03)] proved conversely that it is, up to a multiplicative constant, the only solution admitting a Dirichlet series expansion that converges in the half-plane \(\text{Re}(z)>1\) and of the form \(f=G/P\) with \(G\)~entire of finite order and \(P\)~a polynomial. The author provides a transcription of a short note in a notebook of Adolf Hurwitz, \textit{Über Modulfunktionen in Verbindung mit Dirichlet'schen Reihen} (On modular functions related to Dirichlet series) [Adolf-Hurwitz-Nachlass, ETH Zürich, Hs.~582:128; \url{http://www.e-manuscripta.ch/}] from the early 1880s which contains an argument showing that if \(f\) is a solution of this functional equation and \(\pi^{-s}\Gamma(s)f(2s)\) is written as \(\int_0^\infty\psi(x)x^{s-1}dx\), then the ``modular function'' \(\psi\) satisfies the ``transformation formula for half of the Jacobi theta-function'' by applying a Mellin transform, which yields \(\psi(y)=\sum_{n=1}^\infty e^{-n^2y\pi}\). She also provides a very detailed commentary of this note in the form of a free translation that fills the gaps in Hurwitz's arguments, but does not discuss the concept of modular function. Furthermore, she unearthes an intermediate link between these two converses, namely \textit{E. Cahen} [Ann. Sci. Éc. Norm. Supér. (3) 11, 75--164 (1894; JFM 25.0702.01)], and indicates their generalisation, \textit{E. Hecke}'s converse theorem [Math. Ann. 112, 664--699 (1936; Zbl 0014.01601)] (in its statement on page~357, it could be helpful to note that \(\varepsilon=\pm1\)), as well as further generalisations at the end of the article. On line~4 of page~346 of the article, it should read~``1882--'' before~``1919''; on line~14 of page~348, ``\(x^{\text{Re}s-2}\)''; on line~12 of page~354 and on line~21 of page~360, ``Jahrbuch'' instead of ``Jahresberichte''; on line~7 of page~358, ``\(\varepsilon\)'' instead of ``\(\gamma\)''; the misprint pointed out in Footnote~8 is not a misprint in Hurwitz's review, but a bug in the pdf rendering by the \textit{Zentralblatt} interface.