In this problem we explore a little further the analogy between Sturm-
Liouville boundary value problems and Hermitian matrices. Let \(\mathbf{A}\) be
an \(n \times n\) Hermitian matrix with eigenvalues \(\lambda_{1}, \ldots,
\lambda_{n}\) and corresponding orthogonal eigenvectors \(\xi^{(1)}, \ldots,
\xi^{(n)} .\) Consider the nonhomogeneous system of equations
$$
\mathbf{A x}-\mu \mathbf{x}=\mathbf{b}
$$
where \(\mu\) is a given real number and \(\mathbf{b}\) is a given vector. We will
point out a way of solving Eq. ( \(i\) ) that is analogous to the method
presented in the text for solving Eqs. (1) and (2).
(a) Show that \(\mathbf{b}=\sum_{i=1}^{n} b_{i} \xi^{(i)},\) where
\(b_{i}=\left(\mathbf{b}, \boldsymbol{\xi}^{(i)}\right)\)
(b) Assume that \(\mathbf{x}=\sum_{i=1}^{n} a_{i} \xi^{(i)}\) and show that for
Eq. (i) to be satisficd, it is necessary that \(a_{i}=b_{i}
/\left(\lambda_{i}-\mu\right) .\) Thus
$$
\mathbf{x}=\sum_{i=1}^{n} \frac{\left(\mathbf{b},
\boldsymbol{\xi}^{(i)}\right)}{\lambda_{i}-\mu} \boldsymbol{\xi}^{(i)}
$$
provided that \(\mu\) is not one of the eigenvalues of \(\mathbf{A}, \mu \neq
\lambda_{i}\) for \(i=1, \ldots, n .\) Compare this result with \(\mathrm{Eq.}(13)
.\)