On the Essential Spectrum of Adiabatic Stellar Oscillations Dedicated to Professor Shih-Hsien Yu to celebrate his sixtieth birthday

Tetu Makino · 2025

Read On the Essential Spectrum of Adiabatic Stellar Oscillations Dedicated to Professor Shih-Hsien Yu to celebrate his sixtieth birthday by Tetu Makino (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-2504-00284.txt.

Introduction

Tetu Makino, in a solo paper from twenty twenty-five, asked if a star's wobble operator hides a solid band of tones. The work deals with adiabatic oscillations of a gas star. The driver of the linearized flow is a second order operator with an integral part. It acts as a self adjoint operator on a space weighted by the star's density. Past work read its lone tones and shapes from a practical view, for mode sums.

What stayed open was whether a continuous part sits in its spectrum. Makino proves that a close cousin of the flow task does have such an essential spectrum. That result bears on the star task's own read. Readers in stellar math learn to check the base ground before they trust a mode sum.

Catalog record

CREATORTetu Makino
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE63 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-00284

Browse the full library · About the Tower of Babel

Full text

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

On the Essential Spectrum of Adiabatic Stellar Oscillations Dedicated to Professor Shih-Hsien Yu to celebrate his sixtieth birthday Tetu Makino Abstract The generator 𝑳 \bm{L} of the linearized evolution equation of adiabatic oscillations of a gaseous star, ELASO, is a second order integro-differential operator and is realized as a self-adjoint operator in the Hilbert space of square integrable unknown functions with weight, which is the density distribution of the compactly supported background. Eigenvalues and eigenfunctions of the operator 𝑳 \bm{L} have been investigated in practical point of view of eigenmode expansion of oscillations. But it should be examined whether continuous spectra are absent in the spectrum of 𝑳 \bm{L} or not. In order to discuss this question, the existence of essential spectra in a closely related evolution problem is established. Key Words and Phrases. Stellar oscillation, Euler-Poisson equation, Stellar rotation, Essential spectrum, Eigenfunction expansion 2020 Mathematical Subject Classification Numbers. 3610,35Q65, 35P05, 35R35,35B35, 47N20, 76N15, 76U05, 85A30 1 Introduction We consider the equation of linearized adiabatic stellar oscillations, ELASO : \frac{\partial^{2}\bm{u}}{\partial t^{2}}+\mathcal{B}\frac{\partial\bm{u}}{\partial t}+\mathcal{L}\bm{u}=0,\quad t\geq 0,\bm{x}\in\mathfrak{R}_{b}, (1.1) where the unknown is \bm{u}=\bm{u}(t,\bm{x})\in\mathbb{R}^{3} and \displaystyle\mathcal{B}\bm{v} \displaystyle=2\Omega\begin{bmatrix}-v^{2}\\ \\ v^{1}\\ \\ 0\end{bmatrix}, (1.2a) \displaystyle\mathcal{L}\bm{u} \displaystyle=\mathcal{L}_{0}\bm{u}+4\pi\mathsf{G}\mathcal{L}_{1}\bm{u}, (1.2b) \displaystyle\mathcal{L}_{0}\bm{u} \displaystyle=\frac{1}{\rho_{b}}\nabla\delta P-\frac{\nabla P_{b}}{\rho_{b}^{2}}\delta\rho, (1.2c) \displaystyle\delta\rho=-\mathrm{div}(\rho_{b}\bm{u}),\quad\delta P=\frac{\gamma P_{b}}{\rho_{b}}\delta\rho+\gamma P_{b}(\bm{u}|\mathfrak{a}_{b}), (1.2d) \displaystyle\mathfrak{a}_{b}=-\frac{1}{\gamma\mathsf{C}_{V}}\nabla S_{b}=\frac{\nabla\rho_{b}}{\rho_{b}}-\frac{\nabla P_{b}}{\gamma P_{b}}, (1.2e) \displaystyle\mathcal{L}_{1}\bm{u} \displaystyle=\nabla\mathcal{K}[\delta\rho], (1.2f) \displaystyle\mathcal{K}[g](\bm{x})=\frac{1}{4\pi}\int_{\mathfrak{R}_{b}}\frac{g(\bm{x}^{\prime})}{\|\bm{x}-\bm{x}^{\prime}\|}d\bm{x}^{\prime}. (1.2g) Here \mathsf{G},\gamma,\mathsf{C}_{V} are positive constants, 1<\gamma<2 , and (\rho,S,\bm{v})=(\rho_{b}(\bm{x}),S_{b}(\bm{x}),\bm{0}) is a stationary solution of the Euler-Poisson equations in the rotating co-ordinates with a constant angular velocity \Omega : \displaystyle\frac{D\rho}{Dt}+\rho\mathrm{div}\bm{v}=0, (1.3a) \displaystyle\rho\Big[\frac{D\bm{v}}{Dt}+\bm{\Omega}\times\bm{v}+\bm{\Omega}\times(\bm{\Omega}\times\bm{x})\Big]+\mathrm{grad}P+\rho\mathrm{grad}\Phi=0, (1.3b) \displaystyle\rho\frac{DS}{Dt}=0, (1.3c) \displaystyle\Phi(t,\dot{)}=4\pi\mathsf{G}\mathcal{K}[\rho(t,\cdot)], (1.3d) where \displaystyle\bm{\Omega}=\Omega\frac{\partial}{\partial x^{3}}, \displaystyle\frac{D}{Dt}=\frac{\partial}{\partial t}+\sum v^{k}\frac{\partial}{\partial x^{k}} , under the equation of state P=\rho^{\gamma}e^{\frac{S}{\mathsf{C}_{V}}}. (1.4) The equations ( 1.3a ) - ( 1.3d ), ( 1.4 ) govern the adiabatic inviscid interior motion of a gaseous star, where \rho\geq 0 is the density, P the pressure, S the specific entropy, \bm{v} the velocity field, \Phi is the gravitational potential. We assume that \mathfrak{R}_{b}=\{\bm{x}|\rho_{b}(\bm{x})>0\} is a bounded domain of class C^{3,\alpha} , and \rho_{b}^{\gamma-1},S_{b}\in C^{\infty}(\mathfrak{R}_{b})\cap C^{3,\alpha}(\mathfrak{R}_{b}\cup\partial\mathfrak{R}_{b}) , \alpha being a positive number such that 0<\alpha<\Big(\frac{1}{\gamma-1}-1\Big)\wedge 1 , \inf_{0<r<r_{0}}\Big(-\frac{1}{r}\frac{\partial\rho_{b}}{\partial r}\Big)>0\quad\mbox{for}\quad 0<r_{0}\ll 1, (1.5) and -\infty<\frac{\partial{c_{b}}}{\partial\bm{n}}<0\quad\mbox{on}\quad\partial\mathfrak{R}_{b}, (1.6) where \bm{n} is the outer normal vector at the boundary point, and {c_{b}}=\sqrt{\Big(\frac{\partial P}{\partial\rho}\Big)_{S}}\Big|_{\rho=\rho_{b},S=S_{b}}=\sqrt{\frac{\gamma P_{b}}{\rho_{b}}}, (1.7) the speed of sound. Note that \mathcal{A}_{b}=(\mathfrak{a}_{b}|\bm{n}_{b}),\quad\mathcal{N}_{b}^{2}=(\mathfrak{a}_{b}|\bm{n}_{b})\Big(\frac{\nabla P_{b}}{\rho_{b}}\Big|\bm{n}_{b}\Big), where \displaystyle\bm{n}_{b}=-\frac{\nabla\rho_{b}}{\|\nabla\rho_{b}\|} , are the Schwarzschild discriminant, the square of the Brunt-Väisälä frequency (local buoyancy frequency). We note that the operator \mathcal{L}_{0} can be written as \displaystyle\mathcal{L}_{0}\bm{u} \displaystyle=\mathrm{grad}\Big[-\sigma_{b}\mathrm{div}(\rho_{b}\bm{u})+\sigma_{b}\rho_{b}(\bm{u}|\mathfrak{a}_{b})\Big]+ \displaystyle+\sigma_{b}\Big[-\mathrm{div}(\rho_{b}\bm{u})\mathfrak{a}_{b}+(\bm{u}|\mathfrak{a}_{b})\nabla\rho_{b}\Big], (1.8) where \sigma_{b}=\frac{\gamma P_{b}}{\rho_{b}^{2}}. (1.9) Let us keep in mind that \displaystyle\frac{1}{C}\mathsf{d}^{\frac{1}{\gamma-1}}\leq\rho_{b}\leq{C}\mathsf{d}^{\frac{1}{\gamma-1}},\quad\frac{1}{C}\mathsf{d}^{1+\frac{1}{\gamma-1}}\leq P_{b}\leq{C}\mathsf{d}^{1+\frac{1}{\gamma-1}}, \displaystyle\frac{1}{C}\leq{c_{b}}\leq C,\quad\frac{1}{C}\mathsf{d}^{-\frac{2-\gamma}{\gamma-1}}\leq\sigma_{b}\leq{C}\mathsf{d}^{-\frac{2-\gamma}{\gamma-1}} on \mathfrak{R}_{b} , where \mathsf{d}(\bm{x})=\mathrm{dist}(\bm{x},\partial\mathfrak{R}_{b}) When (\rho,S,\bm{v})=(\rho(t,\bm{x}),S(t,\bm{x}),\bm{v}(t,\bm{x})) is a solution of the rotating Euler-Poisson equations ( 1.3a )-( 1.3d ) ( 1.4 ) near the stationary solution (\rho_{b},S_{b},\bm{0}) , then the unknown variable \bm{u} of the ELASO means \bm{u}(t,\bm{x})=\bm{\varphi}(t,\bm{x})-\bm{x}+\bm{u}^{0}(\bm{x}), (1.10) where \bm{\varphi} is the flow of the velocity field \bm{v} defined by \frac{\partial}{\partial t}\bm{\varphi}(t,\bm{x})=\bm{v}(t,\bm{\varphi}(t,\bm{x})),\quad\bm{\varphi}(0,\bm{x})=\bm{x}, and \bm{u}^{0} is supposed to satisfy \displaystyle\rho(0,\bm{x})-\rho_{b}(\bm{x})=-\mathrm{div}(\rho_{b}(\bm{x})\bm{u}^{0}(\bm{x})), \displaystyle S(0,\bm{x})-S_{b}(\bm{x})=-(\bm{u}^{0}(\bm{x})|\nabla S_{b}(\bm{x})). The formal integro-differential operator \mathcal{L} considered on C_{0}^{\infty}(\mathfrak{R}_{b};\mathbb{C}^{3}) can be extended to a self-adjoint operator \bm{L} in the Hilbert space \mathfrak{H}=L^{2}(\mathfrak{R}_{b},\rho_{b}d\bm{x};\mathbb{C}^{3}) (1.11) endowed with the norm \|\bm{u}\|_{\mathfrak{H}}=\Big[\int_{\mathfrak{R}_{b}}\|\bm{u}(\bm{x})\|\rho_{b}(\bm{x})d\bm{x}\Big]^{\frac{1}{2}}. (1.12) Actually \bm{L} is given by \displaystyle\mathsf{D}(\bm{L}) \displaystyle=\Big\{\bm{u}\in\mathfrak{G}_{0}\quad\Big|\quad\mathcal{L}\bm{u}\in\mathfrak{H}\Big\}, (1.13) \displaystyle\bm{L}\bm{u} \displaystyle=\mathcal{L}\bm{u}. (1.14) Here \mathfrak{G}=\Big\{\bm{u}\in\mathfrak{H}\quad\Big|\quad\mathrm{div}(\rho_{b}\bm{u})\in L^{2}(\mathfrak{R}_{b},\sigma_{b}d\bm{x};\mathbb{C})\quad\Big\} (1.15) is a Hilbert space endowed with the norm \|\bm{u}\|_{\mathfrak{G}}=\Big[\|\bm{u}\|_{\mathfrak{H}}^{2}+\int_{\mathfrak{R}_{b}}|\mathrm{div}(\rho_{b}\bm{u})|^{2}\sigma_{b}(\bm{x})d\bm{x}\Big]^{\frac{1}{2}} (1.16) and \mathfrak{G}_{0} is the closure of C_{0}^{\infty}(\mathfrak{R}_{b};\mathbb{C}^{3}) in \mathfrak{G} . Details of mathematically rigorous discussion of the above described situation can be found in [ 11 ] . We use the following notations: Notation 1 Let \mathsf{X},\mathsf{Y} be Hilbert spaces. For an operator T from a subspace of \mathsf{X} into \mathsf{Y} , \mathsf{D}(T) denotes the domain of T , \displaystyle\mathsf{R}(T)=\mbox{the range of }T=\{Tx|\quad x\in\mathsf{D}(T)\}, \displaystyle\mathsf{N}(T)=\mbox{the kernel of }T=\{x\in\mathsf{D}(T)|\quad Tx=0_{\mathsf{Y}}\}. \mathscr{B}(\mathsf{X};\mathsf{Y}) denotes the Banach space of all bounded linear operators from \mathsf{X} into \mathsf{Y} : |\|T\||_{\mathscr{B}(\mathsf{X};\mathsf{Y})}:=\sup\Big\{\|Tx\|_{\mathsf{Y}}\Big|\quad\|x\|_{\mathsf{X}}=1\Big\}<\infty. \mathscr{B}(\mathsf{X})=\mathscr{B}(\mathsf{X};\mathsf{X}) Let T be an operator in \mathsf{X} such that \mathsf{D}(T) is dense in \mathsf{X} . \displaystyle\mbox{}(T)=\mbox{the resolvent set of }T=\Big\{\lambda\in\mathbb{C}\Big|\quad\mathsf{N}(\lambda-T)=\{0_{\mathsf{X}}\},(\lambda-T)^{-1}\in\mathscr{B}(\mathsf{X})\Big\}, \displaystyle\mbox{}(T)=\mbox{the spectrum of }T=\mathbb{C}\setminus\mbox{}(T), \displaystyle\mbox{}_{p}(T)=\mbox{the set of all eigenvalues of }T=\Big\{\lambda\in\mathbb{C}\Big|\quad\mathsf{N}(\lambda-T)\not=\{0_{\mathsf{X}}\}\Big\}. We are interested the structure of the spectrum \mbox{}(\bm{L}) of the self-adjoint operator \bm{L} . If \mbox{}(\bm{L})=\mbox{}_{p}(\bm{L}) , where \mbox{}_{p}(\bm{L}) denotes the set of all eigenvalues of \bm{L} , then there is an orthonormal system of eigenfunctions, (\bm{\phi}_{n})_{n} , \bm{L}\bm{\phi}_{n}=\lambda_{n}\bm{\phi}_{n} , which is complete in \mathfrak{H} . See, e.g., [ 1 , Theorem X.3.4] . In this situation we have eigenfunction expansions \bm{u}=\sum c_{n}\bm{\phi}_{n} for \forall\bm{u}\in\mathfrak{H} , for which \bm{L}\bm{u}=\sum\lambda_{n}c_{n}\bm{\phi}_{n} , and, if \Omega=0 , the general solution of ELASO \frac{\partial^{2}\bm{u}}{\partial t^{2}}+\bm{L}\bm{u}=\bm{0} is given by \bm{u}(t,\bm{x})=\sum_{n}(c_{n}^{+}\bm{u}_{n}^{+}(t,\bm{x})+c_{n}^{-}\bm{u}_{n}^{-}(t,\bm{x})), where \bm{u}_{n}^{\pm}(t,\bm{x})=\begin{cases}e^{\pm\sqrt{\lambda_{n}}\mathrm{i}t}\bm{\phi}_{n}(\bm{x})\quad(\lambda_{n}\geq 0)\\ e^{\pm\sqrt{|\lambda_{n}|}t}\bm{\phi}_{n}(\bm{x})\quad(\lambda_{n}<0)\end{cases} and c_{n}^{\pm}=\begin{cases}\frac{1}{2}\Big((\bm{u}^{0}|\bm{\phi}_{n})_{\mathfrak{H}}\pm\frac{1}{\sqrt{\lambda_{n}}\mathrm{i}}(\bm{v}^{0}|\bm{\phi}_{n})_{\mathfrak{H}}\Big)\quad(\lambda_{n}>0)\\ \frac{1}{2}\Big((\bm{u}^{0}|\bm{\phi}_{n})_{\mathfrak{H}}\pm\frac{1}{\sqrt{|\lambda_{n}|}}(\bm{v}^{0}|\bm{\phi}_{n})_{\mathfrak{H}}\Big)\quad(\lambda_{n}<0)\\ \frac{1}{2}(\bm{u}^{0}|\bm{\phi}_{n})_{\mathfrak{H}}\quad(\lambda_{n}=0),\end{cases} with \displaystyle\bm{u}^{0}(\bm{x})=\bm{u}(0,\bm{x}),\bm{v}^{0}(\bm{x})=\frac{\partial\bm{u}}{\partial t}(0,\bm{x}) . However, if there are continuous spectra, that is, if \mbox{}(\bm{L})\setminus\mbox{}_{p}(\bm{L})\not=\emptyset , then the eigenfunction expansion does not work. In this sense, the question whether \mbox{}(\bm{L})=\mbox{}_{p}(\bm{L}) or not is important. The aim of this study is concerned with this question, namely Question 1 Is it the case that \mbox{}(\bm{L})=\mbox{}_{p}(\bm{L}) ? Note that \bm{L} is not of the Strum-Liouville type, say, with discrete spectrum, since the multiplicity of the eigenvalue 0 is infinite, or, \mathrm{dim}\mathsf{N}(\bm{L})=\infty , and, therefore, the resolvent is not compact. In fact, 1) Suppose \mathfrak{a}_{b}\not=0 ; Then \bm{u}(\bm{x})=\frac{1}{\rho_{b}(\bm{x})}\mathfrak{a}_{b}(\bm{x})\times\nabla f(\bm{x}), where f\in C_{0}^{\infty}(\mathfrak{R}_{b};\mathbb{R}) is arbitrary, enjoys \mathrm{div}(\rho_{b}\bm{u})=0,\quad(\bm{u}|\mathfrak{a}_{b})=0, since \mathfrak{a}_{b}=-\frac{1}{\gamma\mathsf{C}_{V}}\nabla S_{b} satisfies \mathrm{rot}\mathfrak{a}_{b}=0 ; Then \bm{u}\in\mathsf{N}(\bm{L}) , since \delta\rho=0,\delta P=0 ; 2) Suppose \mathfrak{a}_{b}=0 ; Then \bm{u}(\bm{x})=\frac{1}{\rho_{b}(\bm{x})}\mathrm{rot}\bm{f}(\bm{x}) with arbitrary \bm{f}\in C_{0}^{\infty}(\mathfrak{R}_{b};\mathbb{R}^{3}) belongs to the kernel \mathsf{N}(\bm{L}) . Here we note the following fact: Suppose that the background is isentropic, that is, \mathfrak{a}_{b}=0 everywhere. Let \bm{L}^{G} be the Friedrichs extension of the operator \mathcal{L}\restriction C_{0}^{\infty}(\mathfrak{R}_{b},;\mathbb{C}^{3}) in the functional space \mathfrak{G}=\{\bm{u}\in\mathfrak{H}|\mathrm{div}(\rho_{b}\bm{u})\in L^{2}(\mathfrak{R}_{b},\sigma_{b}d\bm{x};\mathbb{C})\} . Then \mbox{}(\bm{L}^{G})=\mbox{}_{\mathrm{p}}(\bm{L}^{G}) , and \mbox{}(\bm{L}^{G}) consists of \{0\} and a sequence of eigenvalues \lambda_{n},n\in\mathbb{N}, of finite multiplicities such that \lambda_{n}\not=0,\lambda_{n}<\lambda_{n+1}\rightarrow+\infty as n\rightarrow\infty . For proof see [ 7 , Sections III, IV ] for the case of spherically symmetric background for \Omega=0 , and [ 11 , Theorem 6] . Anyway we have \mbox{}_{p}(\bm{L})=\mbox{}_{p}(\bm{L}^{G})=\mbox{}(\bm{L}^{G}) , and the Question is: Is (\lambda-\bm{L})^{-1}\Big(\supset(\lambda-\bm{L}^{G})^{-1}\Big)\in\mathscr{B}(\mathfrak{H}) when \lambda\in\mbox{}(\bm{L}^{G}) , that is, (\lambda-\bm{L}^{G})^{-1}\in\mathscr{B}(\mathfrak{G}) ? On the other hand, Suppose \mathfrak{a}_{b}\not=0 , \Omega=0 . There can appear the so-called ‘ g -mode’ \{\lambda_{-n};n\in\mathbb{N}\}\subset\mbox{}_{p}(\bm{L}) such that \lambda_{-n}>0 , \lambda_{-n}\rightarrow 0 as n\rightarrow\infty . It is the case when \displaystyle\inf_{\mathfrak{R}_{b}}\frac{1}{r}\frac{dS_{b}}{dr}>0 , or, \displaystyle\inf_{\mathfrak{R}_{b}}\frac{\mathcal{N}_{b}^{2}}{r^{2}}>0 . For proof see [ 10 ] . 2 ELASO‡ The second order equation ELASO ( 1.1 ) is equivalent to the evolution equation, which we call ELASO‡ . : \frac{\partial U}{\partial t}+\mathcal{A}U=0 (2.1) with \mathcal{A}=\begin{bmatrix}\mathcal{B}&\mathcal{L}\\ \\ -I&O\end{bmatrix} (2.2) for the unknown U=\begin{bmatrix}U^{1}\\ U^{2}\\ U^{3}\\ U^{4}\\ U^{5}\\ U^{6}\end{bmatrix}=\begin{bmatrix}\bm{v}\\ \\ \bm{u}\end{bmatrix}=\begin{bmatrix}v^{1}\\ v^{2}\\ v^{3}\\ u^{1}\\ u^{2}\\ u^{3}\end{bmatrix},\quad\bm{v}=\begin{bmatrix}v^{1}\\ v^{2}\\ v^{3}\end{bmatrix},\quad\bm{u}=\begin{bmatrix}u^{1}\\ u^{2}\\ u^{3}\end{bmatrix}. (2.3) We consider the ELASO‡ in the Hilbert space \mathfrak{E}=\mathfrak{H}\times\mathfrak{G}_{0} with the densely defined closed operator \bm{A} in \mathfrak{E} , \mathsf{D}(\bm{A})=\mathfrak{G}_{0}\times\mathsf{D}(\bm{L}) , \bm{A}U=\mathcal{A}U , namely, \bm{A}=\begin{bmatrix}\bm{B}&\bm{L}\\ \\ -\bm{I}&O\end{bmatrix}, where \bm{B}:\bm{v}\mapsto\mathcal{B}\bm{v} is a bounded operator from \mathfrak{H} onto \mathfrak{H} . Note that \mathsf{N}(\bm{A})=\{\bm{0}\}\times\mathsf{N}(\bm{L}) . Then, given U_{0}\in\mathsf{D}(\bm{A}) , the initial value problem \frac{dU}{dt}+\bm{A}U=0,\quad U|_{t=0}=U_{0} (2.4) admits a unique solution U=U(t,\bm{x}) in C^{1}([0,+\infty[;\mathfrak{E})\cap C([0,+\infty[;\mathsf{D}(\bm{A})) . And, for this U(t,\bm{x})=(\bm{v}(t,\bm{x}),\bm{u}(t,\bm{x}))^{\top} , the component \bm{u}(t,\bm{x}) is a solution of ELASO ( 1.1 ) in C^{2}([0,+\infty[;\mathfrak{H})\cap C^{1}([0,+\infty[;\mathfrak{G}_{0})\cap C([0,+\infty[;\mathsf{D}(\bm{L})) , and \bm{v}(t,\bm{x})=\partial\bm{u}(t,\bm{x})/\partial t . Note that 1) \lambda\in\mbox{}(\bm{A}) if and only if \lambda^{2}-\lambda\bm{B}+\bm{L} does not have a bounded inverse, and 2) \lambda\in\mbox{}_{p}(\bm{A}) if and only if there is \bm{\phi}\in\mathsf{D}(\bm{L}) such that \bm{\phi}\not=0,(\lambda^{2}-\lambda\bm{B}+\bm{L})\bm{\phi}=\bm{0} . Hence, when \Omega=0,\bm{B}=O , then it holds that \mbox{}(\bm{L})=\mbox{}_{p}(\bm{L})\Leftrightarrow\mbox{}(\bm{A})=\mbox{}_{p}(\bm{A}) , since \lambda\in\mbox{}(\bm{A})\ [\!(\in\mbox{}_{p}(\bm{A}))\!]\quad\Leftrightarrow\quad-\lambda^{2}\in\mbox{}(\bm{L})\ [\!(\in\mbox{}_{p}(\bm{L}))\!] for \bm{B}=O . We consider

Read the complete text