Let $\mathcal{B}(H)$ denote the space of bounded linear operators on a complex Hilbert space $H$. The von Neumann bicommutant theorem says:

Theorem. Suppose that $\mathcal{A}$ is a $C^*$-subalgebra of $\mathcal{B}(H)$ and that $\mathcal{A}$ contains the identity operator. If $\mathcal{A}$ is closed with respect to the strong operator topology, then $\mathcal{A}$ coincides with its bicommutant $\mathcal{A}''$.

[Note: The commutant $\mathcal{S}'$ of a subset $\mathcal{S} \subseteq \mathcal{B}(H)$ is defined as the set of all operators in $\mathcal{B}(H)$ that commute with all operators in $\mathcal{S}$.]

Question. Does this theorem remain true if we replace the assumption "$\mathcal{A}$ is a $C^*$-subalgebra of $\mathcal{B}(H)$" with the assumption "$\mathcal{A}$ is a commutative subalgebra of $\mathcal{B}(H)$"?

The question seems to be a bit bold at first glance, but it is motivated by the following simple finite-fimensional observation:

Motivation. The bicommutant theorem fails if we merely assume $\mathcal{A}$ to be a subalgebra of $\mathcal{B}(H)$ that is strongly closed and contains the identity operator. As a simple counterexample we can choose $\mathcal{A}$ to be the set of all upper triangular matrices in $\mathbb{C}^{2\times 2} = \mathcal{B}(\mathbb{C}^2)$.

On the other hand, the algebra $\mathcal{A}$ is not commutative in this example, and if we try to "adjust" the example by choosing $\mathcal{A}$ as the commutative algebra \begin{align*} \mathcal{A} := \{ \begin{pmatrix} a & b \newline 0 & a \end{pmatrix}: \; a, b \in \mathbb{C} \}, \end{align*}, then we indeed have $\mathcal{A} = \mathcal{A}''$.


The answer is negative. Consider the Hardy space $H^{\infty}(\mathbb{T})$ consisting of holomorphic functions admitting bounded extension to the unit circle and view it is a subalgebra of $B(L^{2}(\mathbb{T})$. Let us compute the commutant. Denote the operator of multiplication by $z$ by $M_z$. If $T$ belongs to the commutant of $H^{\infty}(\mathbb{T})$, it commutes with $M_z$. But it also has to commute with its inverse $M_{\overline{z}}$. But these two generate $L^{\infty}(\mathbb{T})$, so the commutant is equal to $L^{\infty}(\mathbb{T})$, hence also the bicommutant, as this a maximal abelian subalgebra of $B(L^{2}(\mathbb{T}))$. It remains to see that $H^{\infty}(\mathbb{T})$ is strongly closed, but this is simple: a function $f\in L^{\infty}(\mathbb{T})$ belongs to $H^{\infty}(\mathbb{T})$ iff all of its negative Fourier coefficients vanish and this property is clearly preserved by strong convergence (even by weak convergence).

  • $\begingroup$ That's great, thanks! $\endgroup$ – Jochen Glueck Feb 19 at 21:16

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.