Class numbers of quadratic fields, Hasse invariants of elliptic curves, and the supersingular polynomial (Q1434337)

The main discovery of this paper (Theorem 1) is the detection of a close connection between the number of linear resp. quadratic factors modulo \(p\), where \(p > 3\) is a prime, of the Hasse invariant \(W_m(X)\) of the elliptic curve in Legendre form \[ Y^2 = X(X-1)(X-\lambda) \] over \({\mathbb Q}\) and the class number \(h(-p)\) of the complex quadratic field \({\mathbb Q}(\sqrt{-p})\). In \(W_m(X)\) one takes the integers \[ m = \frac{p-1}{2},\;\frac{p-1}{3}, \;\frac{p-2}{3},\;\frac{p-1}{4},\;\frac{p-3}{4}, \] considering \(p \bmod 2,3\) and \(4\), and the value \(x = \lambda\). \(W_m(X)\) is not to be confused with the Hasse invariant \(\gamma\) mod squares of the basic field \(K\) which together with the \(j\)-invariant characterizes isomorphism classes of elliptic curves over the field \(K\) [see \textit{P. Roquette}'s book ``Analytic theory of elliptic functions over local fields'' (Vandenhoeck \& Ruprecht, Göttingen) (1970; Zbl 0194.52002)]. Between \(W_m(X)\) and the Legendre polynomial \(P_m(X) \bmod p\) there holds the identity \[ W_m(X) = (1-X)^m P_m \biggl(\frac{1+X}{1-X}\biggr). \] The somewhat sophisticated proof is essentially classical. It refers back to old results of Hasse and Deuring, especially to the theory of meromorphisms of function fields. But of course, also newer results are applied, we mention only the papers of Kaneko-Zagier, Elkies and N. Yui and the textbooks of Husemöller and of Silverman on elliptic curves and of Cox on certain primes containing also important results on elliptic curves and on complex quadratic fields. Altogether, the reference section is rather comprehensive. In an appendix a well-known theorem of Pellet-Stickelberger-Voronoi attributed in Hasse's book to Stickelberger and in Narkiewicz's book to Pellet and Stickelberger is stated. A slightly generalized version of this theorem is applied in the publication. In the course of the paper the Hasse invariant of the Hessian elliptic curve \[ y^2+axy+y = x^3 \] plays an important role too. Reviewer's remark: It would be interesting to find out if Theorem 1 could be used to construct elliptic curves of high rank over \({\mathbb Q}\) by starting from complex quadratic fields \({\mathbb Q}(\sqrt{-p})\) of a large class number \(h(-p)\). This was successfully done in the dissertation of Jordi Quer for Mordell's elliptic cures which were not especially known as curves of high rank over \({\mathbb Q}\) [\textit{J. Quer}, Sobra el 3-rang del cosso quadratics i la corba elliptica \(Y^2 = X^3+M\). Barcelona (1987)].