Most densely-defined unbounded linear operators on Hilbert spaces have a very large domain. In fact, for a lot of natural operators the intersection of their domains are still dense.

Let us consider some counterexamples:

  • Let $T : \mathbb Q^n \subset \mathbb R^n \rightarrow \mathbb R^m$ be a densely-defined unbounded operator.

    Let $Q$ be orthogonal.

    Then $T_Q : Q \mathbb Q^n \subset \mathbb R^n \rightarrow \mathbb R^m$, which maps $Qx$ to $Tx$ for $x \in \mathbb Q^n$, is a densely-defined unbounded operator, too.

    Then $\mathbb Q^n \cap Q \mathbb Q^n = \{ 0 \}$ for most choices $Q$.

  • Let $\Omega \subseteq \mathbb R^n$ be a smoothly bounded domain.

    Consider the differential operator $D : C_0^\infty(\Omega) \subset L^2(\Omega) \rightarrow L^2(\Omega)$, which maps compactly-supported smooth functions over $\Omega$ to their total differential.

    Pick an arbitrary weakly differentiable $\psi \in H^1(\Omega)$ that satisfies $\psi > 0$ essentially.

    Then $\psi \cdot C_0^\infty(\Omega)$ is a linear subspace of weakly differentiable functions (partial differentiation), and the operator $D : \psi \cdot C_0^\infty(\Omega) \subset L^2(\Omega) \rightarrow L^2(\Omega)$ is densely defined.

    However, for most $\psi$, we have $C_0^\infty(\Omega) \cap \psi \cdot C_0^\infty(\Omega) = \{ 0 \}$.

These examples are artificial. In both cases, the unbounded are not closed. Taking the closure provides a common intersection for the domains of the operators. In the first case, the closure is even bounded. Most Hilbert spaces of functions have a common subspace of their domains that is dense (e.g., compactly supported smooth functions).

I do not expect that "all closed densely-defined unbounded linear operators have a dense common domain". But is there a theory that provides an abstract point of view on situations where all relevant operators' domains have a large (i.e. dense) intersection?

  • 1
    $\begingroup$ I don't understand your first example. $T$ cannot be unbounded as it is an operator on a finite dimensional space. Or am I getting this completely wrong? $\endgroup$ – user51527 Jul 18 '14 at 9:25
  • $\begingroup$ All bounded operators are unbounded operators. en.wikipedia.org/wiki/Unbounded_operator $\endgroup$ – shuhalo Jul 18 '14 at 9:31

Perhaps not what you're looking for, but you may be interested in the following result. Let $\cal H$ be a separable infinite-dimensional Hilbert space, and $H$ any self-adjoint unbounded linear operator on $\cal H$ with purely discrete spectrum. Consider $T = U H$ where $U$ is a unitary operator. Then the set of $U$ for which ${\cal D}(T) \cap {\cal D}(T^*) = \{0\}$ is a dense $G_\delta$.

See R.B. Israel, ``Some Generic Results in Mathematical Physics'', Markov Processes and Related Fields 10 (2004), 517-521.

| cite | improve this answer | |
  • $\begingroup$ What topology is being put on the set of all unitary operators? $\endgroup$ – Nate Eldredge Jul 18 '14 at 19:09
  • $\begingroup$ The paper uses the norm topology, but it seems to me you can use the strong operator topology or the weak operator topology as well. $\endgroup$ – Robert Israel Jul 18 '14 at 19:35

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.