Enclosing solutions of an inverse Sturm-Liouville problem with finite data (Q1340886)

Consider the Sturm-Liouville problem with Dirichlet boundary conditions \[ - u''+ q(x) u= \lambda u,\quad u(0)= u(\pi)= 0.\tag{1} \] In this paper, \(q(x)\) is assumed to be a continuous function in \([0,\pi]\), symmetric about \(\pi/2\), that is \(q(x)= q(\pi- x)\) for all \(x\in [0,\pi]\). A real number \(\lambda\) is called an eigenvalue of (1) if there is a nontrivial solution \(u(x)\) of the boundary value problem (1). In this case, \(u(x)\) is called an eigenfunction of (1). The set of all eigenvalues is the spectrum of (1). As is well-known, the spectrum of (1) is an infinite sequence of real numbers which is bounded from below and tends to infinity. We refer to (1) as Dirichlet problem and regard eigenvalues of (1) as functionals of \(q(x)\), denoted by \(\lambda_ i(q)\), \(i\in \mathbb{N}\), where the \(\lambda_ i(q)\) are ordered increasingly: \(\lambda_ 1(q)< \lambda_ 2(q)\dots\) The opposite of the computation of eigenvalues and eigenfunctions of (1) for a given potential \(q(x)\), the inverse Dirichlet problem is concerned with the reconstruction of \(q(x)\) in (1) from spectral data. The author of the paper is concerned with the reconstruction of an unknown potential \(q(x)\) in the Sturm-Liouville problem with Dirichlet boundary conditions, when only a finite number of eigenvalues are known. Developing a new reconstruction scheme, one of the author's goals is a precise error estimation in the infinity norm, bounding both the errors committed by the termination of an infinite iteration and roundoff errors in the calculations and in the representation of numbers and functions on a computer. In finitely many iterations, the author constructs bounds of a solution of the inverse Dirichlet problem (in a given function set), without an a priori-restriction on the infinity norm of \(q\). The inverse problem is transformed into a system of nonlinear equations. Applying interval Newton's method to this system, the existence and local uniqueness of a solution of the system is validated. This solution is enclosed in an interval vector. From the interval vector, an interval function \([q](x)\) is constructed that encloses a potential function \(q(x)\) corresponding to the prescribed eigenvalues.