I have to decide whether the following statement is true or false : every permutation of a basic sequence is equivalent to the entire sequence ! where a sequence $(x_n)$ in a Banach space $X$ is called basic if it's a basis of $[x_1,x_2,x_3,.........]$ (it's closed span).
I think it's false, and I think that if it's true then $\displaystyle\sum_{1}^{\infty}x_n$ converges unconditionaly . Any ideas. Thank you!
-
A basis is unconditional if and only if every permutation of the basis is a basic sequence. The property that you require is stronger. A basis for which every permutation is basic and equivalent to to the original sequence is called symmetric. So yes, symmetric basic sequences are unconditional. These facts can be found in Lindenstrauss and Tzafriri's Classical Banach Spaces I. – David Mitra Feb 13 '13 at 02:26
-
Thank you very much, I really appreciate that ! – Feb 13 '13 at 02:33
-
Do you have an idea how to prove this statement : A basis is unconditional if and only if every permutation of the basis is a basic sequence ! – Feb 13 '13 at 03:21
-
It follows from the definition of "basic" (existence of unique representations in terms of the basis elements ) and the definition of "unconditional" ( every expansion $\sum_{i=1}^\infty \alpha_i x_i$ in terms of the basis converges unconditionally). – David Mitra Feb 13 '13 at 03:28
1 Answers
A (Schauder) basis $(x_i)_{i=1}^\infty$ of a Banach space $X$ is unconditional if and only if for every $x\in X$, its expansion $\sum\limits_{i=1}^\infty \alpha_i x_i$ in terms of the basis converges unconditionally. This is equivalent to saying that any permutation of the basis is a basic sequence.
There are, of course, bases that are not unconditional. For instance, it can be shown that the sequence $\{e_1, e_1-e_2, e_1-e_3,\ldots\}$ is a basis of $\ell_1$ that is not unconditional.
The property that you require is stronger. A basis $(x_i)_{i=1}^\infty$ for which every permutation of $(x_i)_{i=1}^\infty$ is equivalent to $(x_i)_{i=1}^\infty$ is called symmetric (recall that two basic sequences $(x_i)$ and $(y_i)$ are equivalent if the convergence of $\sum\alpha_i x_i$ is equivalent to that of $\sum\alpha_i y_i$).
Of note:
- For $1<p<\infty$, the space $L_p(0,1)$ does not have a symmetric basis (it in fact does not possess a subsymmetric basis)$\,^1$. So the Haar system in these spaces gives an example of an unconditional basis that is not symmetric.
- The sequence $(x_i)$ constructed in my answer to this post is an unconditional basic sequence in $\ell_1$. It follows from the results there that this sequence is not equivalent to the standard unit vector basis of $\ell_1$. So from the last point below, it follows that $(x_n)$ is not symmetric.
- If $X$ is one of $c_0$, $\ell_1$, or $\ell_2$, then every unconditional basis of $X$ is symmetric. This follows from the fact that these spaces have unique unconditional bases$\,^2$.
- If $X$ is one of $c_0$, $\ell_p$ $1\le p<\infty$, then every symmetric basis of $X$ is equivalent to the unit vector basis of $X$. In fact, any symmetric basic sequence in $X$ is equivalent to the standard unit vector basis of $X$$\,^3$.
$^1$ c.f., Ivan Singer, Bases in Banach Spaces, Chapter 22.
$^2$ This is a well known fact. See, e.g., Singer or Joeseph Diestel's Sequences and Series in Banach Spaces.
$^3$ c.f. Lindenstrauss and Tzafriri Classical Banach Spaces I, proposition 3.b.5 and the remark following.
- 74,748