I am interested in knowing whether the space of trace class operators is (crudely) finitely representable in an $L^1$-space. I suspect that the answer is negative but I am unable to find any argument confirming my intuition.

As for motivation, I am working on matrix-valued versions of some inequalities coming from harmonic analysis, and I would like to know if the generalisation I seek is non-trivial, if true.

Definition: A Banach space $X$ is said to be crudely finitely representable in a Banach space $Y$ if there exists a constant $C>0$ such that every finite-dimensional subspace $V$ of $X$ is $C$-isomorphic to a subspace of $Y$, i.e. there exists an isomorphism $T: V \to T(V) \subset Y$ satisfying $\|T\|\cdot \|T^{-1}\| \leqslant C$.

  • $\begingroup$ crudely finitely represented in an $L^1$-space - is there a definition? $\endgroup$
    – Marc Palm
    Dec 13, 2013 at 16:03
  • $\begingroup$ books.google.de/… $\endgroup$
    – Marc Palm
    Dec 13, 2013 at 16:06
  • $\begingroup$ Thanks for the comment, Marc; I've added the definition to the body of the question. $\endgroup$ Dec 13, 2013 at 16:13
  • $\begingroup$ The answer is no, but right now I don't recall a proof. $\endgroup$ Dec 13, 2013 at 16:14
  • 2
    $\begingroup$ BTW: A separable Banach space $X$ is crudely finitely representable in an $L_1$ space iff $X$ is isomorphic to a subspace of $L_1(0,1)$. This is proved in the Lindenstrauss-Pelczynski "Absolutely summing operators" paper. (Use ultra products to embed into an abstract $L_1$ space and quote Kakutani's representation theorem.) $\endgroup$ Dec 13, 2013 at 16:18

2 Answers 2


You can deduce that $S_1$ is not finitely crudely representable in an $L_1$ space from the paper

Pisier, Gilles Some results on Banach spaces without local unconditional structure. Compositio Math. 37 (1978), no. 1, 3–19.

However, I think that the result you want might have been known earlier. Maybe it follows from Kwapien and Pelczynski's "Main triangle projection" paper, which I do not now have at hand.

  • $\begingroup$ Thanks a lot for this answer; it crossed my mind that local unconditional structure might be relevant. To be more specific, the result follows from theorem 2.1, because $L^1$ has local unconditional structure (as na $\mathcal{L}^1$-space) and finite cotype. $\endgroup$ Dec 14, 2013 at 10:24
  • $\begingroup$ On a tangential note, do we know what are the reflexive subspaces of $S_1$? At least do we know if they are super-reflexive? $\endgroup$ Mar 15, 2014 at 20:22

The question was answered long ago. But my favorite answer (to myself) is: The trace class $S_1$ has not Analytic UMD property while commutative $L^1$ has this property. Since Analytic UMD property is a local property, it follows immediately that $S_1$ is not finitely representable in $L^1$.


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.