No label defined (Q4350123)

Let \(V\in L^1_{\text{loc}}([0,\infty),{\mathbb{R}})\) be such that the Sturm-Liouville equation (i) \(-y''+Vy=\lambda y\) is in the limit point case at infinity. For \(\lambda\in{\mathbb{R}}\) let \(u(x,\lambda)\), \(v(x,\lambda)\) be the fundamental system of (i) defined by the boundary conditions \(u=1, u'=0, v=0, v'=1\) at \(x=0\), and \({\mathcal S}(\lambda)\) the set of all real linear combinations of \(u\) and \(v\). Finally, let \(m(\lambda)\) denote the boundary value (if it exists) of the corresponding Weyl-Titchmarsh ``\(m\)-function'' (which originally is only defined for the spectral parameter lying in the upper half-plane). The significance of \({\Im}(m(\lambda))\) as the spectral density of the associated Schrödinger operator is well-known and it is the main goal of the author's recent project (sketched in the paper under review) to supplement the classical theory of Weyl-Titchmarsh by some equally general new methods allowing to develop representation formulas for \(m(\lambda)\) based exclusively on the use of the real fundamental system \(u, v\) defined above. The shortest way to summarize the first method seems to be as follows: Let \(y, y_1, y_2\in{\mathcal S}(\lambda)\) and for abbreviation set \[ \langle y_1,y_2\rangle :=\lim_{N\to\infty}{{{\int_0^N}y_1y_2dt}\over{{\int_0^N}v^2dt}},\quad {|}y|:=\langle y,y\rangle^{1\over 2}\tag{1} \] assuming the limits to exist. Note that \(|y|=0\) not necessarily implies \(y=0\) but at least that \(y\) is subordinate. Now define the so-called ``\(M\)-function'' by \(M(\lambda):=\infty\) or \(M(\lambda):=-\langle u,v\rangle+i\sqrt{|u|^2-\langle u,v\rangle^2}\) according as \(v\) is subordinate or not. (Reviewer's remark: In the fragmentary Hilbert space structure suggested by (1) the real or imaginary part of \(M(\lambda)\) can be considered as to give the length of the component of \(u\) parallel or orthogonal to \(v\), respectively.) Evidently, \(M(\lambda)\) is a solution of the quadratic equation \[ |u+Mv|=0,\tag{2} \] (for non-real \(M\) at least formally). For real \(M\), (2) asserts the subordinacy of \(u+Mv\). For non-real \(M\), where subordinate solutions of the usual type (i.e. real ones) do not exist, (2) attributes to \(u+Mv\) a very interesting new property which could be considered as an extension of the notion of subordinacy to the complex realm. In this case, (2) is also equivalent to the somewhat less transparent ``condition (A)'' (as it is called) preferred in the present paper, probably because it offers a starting point better adapted to the rather subtle proof of the central theorem, viz. \(m(\lambda)=M(\lambda)\), of the present theory [cf. \textit{I. Al-Naggar} and \textit{D. B. Pearson}, Helv. Phys. Acta 67, No. 2, 144-166 (1994; Zbl 0849.34064)]. A second (not yet published) method of the desired kind is embodied in the representation formula \[ m(\lambda)=\lim_{N\to\infty} {1\over N} \Biggl\{-P{\int_0^N}{u\over v}dt+ i\sum_{t_j\leq N}{1\over (v'(t_j))^2}\Biggr\} \tag{3} \] where \(P\) denotes the principal value and \(t_j\) the zeros of \(v\). Note that the imaginary part of the right hand side of (3) does not depend on \(u\). For certain hydrogen-like potentials this part thus can easily be transformed into the classical expression for the spectral density (cf. e.g. Problem 4 on p. 255 in \textit{E. A. Coddington} and \textit{N. Levinson}, Theory of ordinary differential equations. New York, Toronto, London, McGraw-Hill (1955; Zbl 0064.33002)]). Finally, a third method is discussed for which we refer the reader to the papers of \textit{D. B. Pearson} [J. Phys. A, Math. Gen. 26, N. 16, 4067-4080 (1993; Zbl 0795.34071) and Proc. Lond. Math. Soc., III. Ser. 68, No. 1, 127-144 (1994; Zbl 0809.34037)]. For the entire collection see [Zbl 0866.00046].