I'm looking for a book with a proof that for an infinite dimensional Hilbert space, $B(H)$ is not reflexive. Thank you.

- 53,687

- 723
- 3
- 10
-
I'm pretty sure this is in Conway's Functional Analysis. – Jul 10 '14 at 05:47
-
Thanks for you answer. I tried finding it here http://books.google.de/books?id=ix4P1e6AkeIC&printsec=frontcover&hl=de&source=gbs_ge_summary_r&cad=0#v=onepage&q=reflexive&f=false but without success. Is this what you meant? – Zolf69 Jul 10 '14 at 06:10
2 Answers
Clearly this is not answer you requested, but may be this will be helpful for someone else.
We may regard $c_0$ as a closed subspace of $\mathcal{B}(H)$. Indeed, take any orthnormal basis of vectors $(e_i)_{i\in I}\subset H$ and for each $s\in c_0(I)$ consider bounded linear operator $T_s:H\to H$ well defined by $T(e_i)=s_ie_i$. Consider map $T:c_0(I)\to\mathcal{B}(H):s\mapsto T_s$. It is an isometric embedding, because $\Vert T_s\Vert=\sup_{i\in I}|s_i|$. Therefore $c_0(I)$ is a closed subspace of $\mathcal{B}(H)$. Since $H$ is infinite dimensional, then $I$ is infinite. Take any countable $J\subset I$, then $c_0=c_0(J)$ is closed subspace of $c_0(I)$ and a fortiori of $\mathcal{B}(H)$. Closed subspace of reflexive space is reflexive, so if $\mathcal{B}(H)$ is reflexive, then so is $c_0$. But $c_0$ is not reflexive, so $\mathcal{B}(H)$ is not reflexive either.

- 56,803
-
Surely one of the simplest arguments and nicely put. If I recall correctly, Pedersen proves in Analysis Now that $\mathcal{B}(H)$ is the bidual of $\mathcal{K}(H)$ using the trace pairing, from which one can deduce the result as well. – user160629 Jul 10 '14 at 15:16
There is a more general fact: A C*-algebra $A$ is reflexive if and only if it is finite-dimensional. In the infinite-dimensional case, there is a normal element $x\in A$ with infinite spectrum (actually, one can find a positive element with infinite spectrum). Therefore, by the spectral theorem, $C^*(x)$ is an infinite-dimensional commutative C*-algebra, hence of the form $C_0(X)$ for some infinite locally compact space $X$. Thus, there is a closed subspace of $C_0(X)$ (and in particular of $A$), isomorphic to $c_0$, hence $A$ is non-reflexive as this property passes to closed subspaces.
You will find other proofs of this fact in this thread.
As for the reference, if $X$ is an infinite-dimensional Banach space with the approximation property, then $\mathscr{B}(X)$ is not reflexive. This is Theorem 4 on p. 247 in
J. Diestel and J. J. Uhl, Jr., Vector Measures, Providence, R.I., AMS, Mathematical Surveys, (15), 1977.

- 16,361