Complex oscillations of non-definite Sturm-Liouville problems, II

Angelo B. Mingarelli · 2025

Read Complex oscillations of non-definite Sturm-Liouville problems, II by Angelo B. Mingarelli (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-16101.txt.

Introduction

Angelo Mingarelli revisits a classic result on wave-like math problems in a 2025 solo paper. The topic is non-definite Sturm-Liouville problems, a class of eigenvalue problems from physics and maths. An old result by Richardson dealt with how zeros of real and complex parts split apart. Mingarelli corrects and updates that result.

He then extends the work to a new case. In that case the weight function may be zero on a middle stretch of the interval. He studies how the zeros act when the end points are brought into the picture. Worked examples show the new claims are sharp and cannot be pushed further. The paper closes with open questions for later work.

Catalog record

CREATORAngelo B. Mingarelli
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE36 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-16101

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Complex oscillations of non-definite Sturm-Liouville problems, II Angelo B. Mingarelli Abstract Abstract. We correct and update a result of R.G.D. Richardson [ 13 ] dealing with the separation of zeros of the real and imaginary parts of non-real eigenfunctions of non-definite Sturm-Liouville eigenvalue problems. We then extend it to the case where the weight function is allowed to be identically zero on a subinterval that excludes the end-points and study the behavior of the zeros of the real and imaginary parts when the end-points are included. Examples are given illustrating the sharpness of the results along with open questions. Keywords: Sturm-Liouville, eigenvalues, non-definite, complex ghosts, oscillations, zeros, separation theorem 2020 Mathematics Subject Classification: 34B24, 34B09, 34C10 1. Introduction We consider the Sturm-Liouville eigenvalue problem with real piecewise continuous coefficients over an interval [a,b] , (1.1) \displaystyle-y^{\prime\prime}+q(x)y=\lambda r(x)y, (1.2) \displaystyle y(a)=y(b)=0. in the non-definite case, [ 11 ] , i.e., this is essentially the case where the quadratic forms induced by the two operators on the left and on the right are sign-indefinite on their domains. The more general equation -(p(x)y^{\prime})^{\prime}+q(x)y=\lambda r(x)y where p(x) is of one sign may be reduced to ( 1.1 ) by means of a Liouville transformation, and so leads to no new results. Thus, we shall restrict ourselves to the formulation ( 1.1 ) in this paper. Tacit assumptions and Notation In the sequel we will always assume that the eigenvalue problem ( 1.1 )-( 1.2 ) is non-definite and has at least one pair of non-real eigenvalues, usually denoted by \lambda,\bar{\lambda}\in\mathbf{C} . The problem of the actual existence of non-real eigenvalues for problems of this type is difficult and very few results exist in this direction, see, e.g., [ 1 ] , [ 4 ] . Given a non-real eigenvalue of ( 1.1 )-( 1.2 ), an associated non-real eigenfunction will be denoted by y=\varphi+i\psi , where \varphi,\psi are its, normally unspecified, real and imaginary parts. A non-real eigenfunction will be called a complex ghost state or a complex ghost , for short [ 11 ] . Hereinafter, all solutions will be assumed non-trivial unless otherwise specified. In addition, since {\rm Im}\ \lambda\neq 0 we will always assume that, unless otherwise specified, {\rm Im}\ \lambda>0 . This is always possible as we can replace \lambda by \bar{\lambda} and y by \bar{y} in ( 1.1 ). The following original result is essentially stated in Richardson [ 13 ] , see also [ 9 , 11 ] . Theorem 1 . [ [ 13 ] , Theorem X] Let r be continuous in [a,b] . If r(x) changes its sign precisely once in (a,b) then the roots of the real and imaginary parts of any complex ghost separate one another. A discussion of Theorem 1 is necessary. First, it was understood at the time of Richardson’s paper (as it is now in classical Sturmian cases, such as in Sturm’s separation theorem) that the expression separate one another meant that between any two consecutive zeros of one solution there is always exactly one zero (of any other independent solution) that is distinct from either of the two original zeros. It is unlikely that Richardson had in mind the possibility that the weight function, r(x) , could be identically zero in a subinterval, however the latter possibility does change the result(s). We will show below that Theorem 1 as stated is, in general, not true (see e.g., Example 1 and Figure 1 below). This was alluded to in [ 9 ] but without discussion. It is true if one excludes the endpoints as one of the two zeros in question, i.e., we are dealing only with interior zeros. However, counterexamples can be found in cases where the endpoints are included (see the examples at the end of this article). Indeed, we show in Remark 2 that if r(x) is identically zero in a right-neighborhood of x=a that includes the endpoint, then Theorem 1 is false. In fact, we go further and show in Lemma 1 below, that the statement in Theorem 1 cannot be true if the endpoints are included (if even the weight function never vanishes identically in any subinterval). As a result, Richardson’s Theorem X in [ 13 ] must be replaced by something like Theorem 2 . Indeed, in Theorem 2 we show that the single turning point or zero required in Theorem 1 may be replaced by an interval of zeros without affecting the conclusion. In addition, we will find conditions under which given two consecutive zeros (one of which is an endpoint) of either the real (or imaginary) part of a complex ghost there is or there isn’t a zero of the imaginary (or real) part. In this sense, this paper continues a study of the zeros of the real and imaginary parts of such ghosts, research that began with [ 13 ] , [ 10 ] , [ 9 ] , [ 12 ] and continued with [ 7 ] , [ 8 ] and the results therein. Remark 1 . We note that Theorem 1 appears to be exceptional in the sense that it has no apparent equivalent in the two turning point case (with a weight function that doesn’t vanish identically in a subinterval). For example, in [ 7 ] there are examples showing that the interior zeros of the real and imaginary parts of non-real eigenfunctions may not separate one another. 2. Main results First, we show that Theorem 1 cannot be true if one or more end-points are included in its conclusion. Lemma 1 . There is no non-trivial complex ghost whose real and imaginary parts have zeros that separate one another on an interval that includes the endpoints of [a,b] . Proof. Without loss of generality we may assume that the endpoint in question is x=a . Assuming the contrary, let y=\varphi+\psi i be a non-trivial complex ghost where \varphi satisfies \varphi(a)=0 and \varphi(x_{0})=0 where x_{0} is a zero of \varphi larger than x=a . Then, by hypothesis, \psi(x) has a unique zero t_{0} in (a,x_{0}) . Now \psi(a)=\psi(t_{0})=0 , thus, by assumption \varphi(x_{1})=0 for some unique x_{1}\in(a,t_{1}) . Since \varphi(a)=\varphi(x_{1})=0 there must be a unique zero t_{1}\in(a,x_{1}) of \psi . Continuing in this way we see that the now infinite sequence of zeros x_{i},t_{i} must accumulate at x=a . On the other hand the theorem of the mean then implies that both \varphi^{\prime}(a)=0 and \psi^{\prime}(a)=0 . Thus, y^{\prime}(a)=0 and so y is the trivial solution, a contradiction. ∎ Lemma 2 . Let \lambda be a non-real eigenvalue of ( 1.1 )-( 1.2 ) and y=\varphi+i\psi be a corresponding eigenfunction. Then \varphi,\psi are linearly independent over [a,b] . Proof. Assume the contrary. Then \psi=\mu\varphi where \mu\in\mathbf{C}\textbackslash{0} . Since y=(1+i\mu)\varphi satisfies ( 1.1 ) we get that -\varphi^{\prime\prime}+q\varphi=\lambda r\varphi . Since the left side of the previous equation is real and {\rm Im}\,\lambda\neq 0 we find that r(x)\varphi(x)\equiv 0 on [a,b] . Since, by hypothesis, ( 1.1 )-( 1.2 ) is non-definite, r(x) cannot be identically zero and so there is an interval J\subset[a,b] on which r(x)>0 , say. Thus, \varphi(x)\equiv 0 on J . Hence \psi(x)\equiv 0 on J so that y\equiv 0 on J . This, however,

Read the complete text