Consider the multiplication operator $T_f:L^2(\mathbb{R}) \to L^2(\mathbb{R})$ given by

$$ \phi(x) \mapsto (T_f\phi)(x) = f(x)\phi(x), $$ where $$ f(x) = \begin{cases} 1, \quad x \ge 0, \\ 0, \quad x < 0. \end{cases} $$

So I checked and found that the spectrum of this operator is $\sigma(T_f) = \sigma_p(T_f) = \{0,1\}$. That is, $(T_f-\lambda I)$ is not injective when $\lambda = 0,1$, and hence the spectrum is simply the point spectrum.

  1. Now I have read that the point spectrum $\sigma_p$ has a subset $\sigma_d$ called the discrete spectrum. Are $0$ and $1$ in the discrete spectrum $\sigma_d$ or are they in $\sigma_p\setminus \sigma_d$? One of the conditions for $\lambda$ to be in $\sigma_d$ is that 'the root subspace $L_\lambda(A)$ corresponding to $\lambda$ is finite-dimensional. It seems to me that the eigenfunctions corresponding to, for example, $\lambda = 0$ form an infinite dimensional subspace. I.e. $\phi(x) \in \mathbb{R}$ for $x <0$ and $\phi(x) = 0$ for $x \ge 0$. So this means that $0$ and $1$ are not in the discrete spectrum?
  2. Is it possible to construct a simple explicit example of a multiplication operator that has a continuous spectrum? In this case $(T_f-\lambda I)$ should be injective but only have a dense image (i.e. not the whole space). Can a multiplication operator have this type of spectrum?

Your first multiplication operator is a projection because $T_f^2 = T_f$. This operator is selfadjoint, which makes $T_f$ an orthogonal projection.

The operator $T_f$ where $f=\tan^{-1}x$ has continuous spectrum $[-\pi/2,\pi/2]$, and it has no discrete spectrum because there is no set $E$ of positive measure on which $f$ is constant.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.