Inverse scattering, the coupling constant spectrum, and the Riemann hypothesis (Q1610855): Difference between revisions

The article deals with an approach to the Riemann hypothesis using inverse scattering theory. The idea, originally from \textit{K. Chadan} and \textit{M. Musette} [Sur un potentiel singulier intéressant. (On an interesting singular potential.), C. R. Acad. Sci., Paris, Sér. II 316, 1-6 (1993; Zbl 0773.34062)], is to consider the equation \(-\psi''+ \lambda V\psi= k^2\psi\) on \([0,\infty)\) which is the s-wage equation related to the three-dimensional Schrödinger equation with rotationally invariant potential. Under certain conditions on \(V\) it is known that the Jost functions \(M^{(\pm)}(\lambda,k)\) are entire functions of the coupling constant \(\lambda\). Furthermore, one has infinitely may \(\lambda_n(k)\) with \(|\lambda_n(k)|\to \infty\) for \(n\to\infty\) where \(k\in\mathbb{C}\) and \(\text{Im}(k)\geq 0\) such that \(M(\lambda_n(k), k)= 0\). For \(V\geq 0\) the criterion \((\text{Im}(\lambda_n(0))) \int^\infty_0|f(\lambda_n(0), 0,x)|^2 V(x) dx= 0\) implies that the zeroes are real and negative. Here \(f\) is the Jost solution for zero energy of the reduced Schrödinger equation. The author states that the criterion can be extended to a one-parameter family of potentials \(V(x,g)\) where \(g={1\over\lambda}\) and \(V\) has to be substituted by the first-order term \(V_1\) in the asymptotic expansion of \(V(x,g)\) in \(g\). Restricting the class of potentials allowed to locally integrable functions satisfying \(\displaystyle\int^\infty_0 x|V(x)|dx<\infty\), \textit{A. Martin} [On the analytic properties of partial wave scattering amplitudes obtained from the Schrödinger equation, Nuovo Cimento, X. Ser. 14, 403-425 (1959; Zbl 0088.23103) and \(S\)-matrix, left-hand cut discontinuity and potential, Nuovo Cimentio (10) 19, 1257-1265 (1961)] proved a representation for Jost functions. Setting \(\nu+{1\over 2}= s\) and \(\lambda= s(s-1)\) the author considers the ansatz \(M^{(\pm)}(\nu, 0)= 2\xi(\nu+{1\over 2})\) where \(\xi(s)={1\over 2} s(s-1) \pi^{-s/2}\Gamma({s\over 2}) \zeta(s)\) and \(\zeta(s)\) is the Riemannian \(\zeta\)-function and makes then a specific choice for \(M^{(\pm)}(\nu, k)\). This is the starting point for the construction of a one-parameter family \(U(\nu,x)\) of potentials satisfying the reduced Schrödinger equation given the Jost function as above within the class of potentials allowing the representation of the Jost function due to Martin. This is done first in the limiting case \(|\nu|\to \infty\) using the Marchenko equation. Then the construction is extended to the case of finite \(\nu\). To this end, one has in this case that \(\int^\infty_0|f(0,0, x)|^2 V_1(x) dx= 0\) and the criterion above gives no information about the position of the zeroes of the Jost function at zero energy. In the last section the reasons for the vanishing of the integral are given and a program to overcome the difficulties is alluded to.