1
$\begingroup$

Let $T$ be a compact Hausdorff space and $X$ be an infinite-dimensional complemented subspace of $C(T)$.

Question 1. Assume that $X$ has a subspace $U$ that is isomorphic to $c_{0}$. Given any positive integer $n$ and a finite-dimensional subspace $M$ of $X$. Do there exist a constant $K$, depending only on the Banach-Mazur distance $\textrm{d}(U,c_{0})$, and a $n$-dimensional subspace $N$ of $U$ so that the Banach-Mazur distance $\textrm{d}(N,l_{\infty}^{n})\leq K$ and $$\max(\|x\|,\|y\|)\leq K\|x+y\|, \quad x\in M,y\in N ?$$

Question 2. J. Lindenstrauss and H. P. Rosenthal (1969) proved that $X$ is an $\mathcal{L}_{\infty,\lambda}$-space for some $\lambda$. But I want to know more about it. It is known that $C(T)$ is an $\mathcal{L}_{\infty,1+\epsilon}$ for every $\epsilon>0$. We further assume that $X$ is $C$-complemented in $C(T)$. Then what is the $\lambda$ ?

Thank you!

$\endgroup$
2
  • $\begingroup$ @TomaszKania I just want to estimate how many the $\lambda$ is, depending only on the constant $C$, not to say that $X$ is an $\mathcal{L}_{\infty,1+}$. That is, I want to prove a quantitative version of Theorem 3.2 in J. Lindenstrauss and H. P. Rosenthal's paper in 1969. $\endgroup$ – Dongyang Chen Apr 13 at 13:24
  • $\begingroup$ Question 2 follows from Question 1. $\endgroup$ – Dongyang Chen Apr 13 at 14:35
1
$\begingroup$

Question 1 has an affirmative answer. You don't need $X$ complemented in a $C(K)$ space and for $U$ you only need that it fails cotype. Use the Mazur technique for constructing basic sequences: norm to $1+\epsilon$ the finite dimensional $M$ by finitely many linear functionals of norm one. The intersection of the kernels of these functionals with $U$ (call the intersection $U_1$) is a finite codimensional subspace of $U$, and hence $U_1$ also fails cotype, which implies that $U_1$ contains for all $n$ a $1+\epsilon$ isomorphic copy of $\ell_\infty^n$. This gives you what you want with $K$ at most $2+\epsilon$.

I think that you know everything I said, so maybe I misunderstood your question?

$\endgroup$
4
  • $\begingroup$ Question 1 comes from the proof of Theorem 2.1 in J. Lindenstrauss and H. P. Rosenthal's paper in 1969. You do not misunderstand my question and you are right. But I have to check your answer. $\endgroup$ – Dongyang Chen Apr 21 at 15:23
  • $\begingroup$ As for Question 2, J. Lindenstrauss and H. P. Rosenthal pointed out that it follows from the proof of Theorem 2.1 and James's distortion theorem that if $X$ is $C$-complemented in $C(K)$ , then $X$ is an $\mathcal{L}_{\infty, 9C+\epsilon}$ space for every $\epsilon>0$. $\endgroup$ – Dongyang Chen Apr 21 at 15:28
  • $\begingroup$ In you answer, norm to $1+\epsilon$ the finite dimensional $M$ by finitely many linear functionals of norm one may means that there exists $x^{*}_{1},\cdots,x^{*}_{n}$ of norm one in $X^{*}$ so that $\max_{k}|\langle x^{*}_{k},m\rangle\|\geq \frac{1}{1+\epsilon}\|m\|$ for $m\in M$. Is that right? $\endgroup$ – Dongyang Chen Apr 22 at 14:01
  • $\begingroup$ Yes; that is right. $\endgroup$ – Bill Johnson Apr 22 at 16:19

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.