1

It is well known that on the Banach space $l^{\infty}\left(\mathbb{N}\right)$ of bounded functions $f:\mathbb{N}\to\mathbb{C}$ with the sup-norm, there exists a (non-unique) Banach limit. This is a linear functional $L$ that is positive, bounded and shift-invariant. It can be seen that such a functional satisfies, for every real-valued $f\in l^{\infty}\left(\mathbb{N}\right)$, $$\liminf_{n\to\infty} f\left(n\right) \leq L\left(f\right) \leq \limsup_{n\to\infty}f\left(n\right).$$

I was wondering what if we replace $\mathbb{N}$ by a directed set. More specifically, let $\mathbf{N}$ be a directed set and consider the Banach space $l^{\infty}\left(\mathbf{N}\right)$ of bounded functions (nets) $f:\mathbf{N}\to\mathbb{C}$ with the sup-norm. While it seems that there is no a natural analog to the shift-invariance property, there are notions of liminf and limsup for nets (see for instance here). Then my question is as follows.

Let $\mathbf{N}$ be a directed set. Does there exist a bounded linear functional $L$ on $l^{\infty}\left(\mathbf{N}\right)$ such that $$\liminf_{n\in\mathbf{N}} f\left(n\right) \leq L\left(f\right) \leq \limsup_{n\in\mathbf{N}}f\left(n\right)$$ for every real-valued net $f\in l^{\infty}\left(\mathbf{N}\right)$?

When reading the standard proofs that a Banach limit exists for $l^{\infty}\left(\mathbb{N}\right)$, there is an essential use of partial sums. Namely, we are looking at the subspace of those elements in $l^{\infty}\left(\mathbb{N}\right)$ whose finite partial sums are uniformly bounded, and using Hahn-Banach theorem to extend a seminorm from it to the whole space. For general nets (say uncountable) it seems that this should not work anymore.

N A-R
  • 191
  • There is no much point in replacing sequences by nets in normed spaces or metric spaces for that matter as continuity can be completely specified by convergence of sequences. Nets will be more appropriate in non-metrizable topologies. – Mittens Apr 12 '22 at 21:36
  • You could see "universal subnet" ... https://math.stackexchange.com/q/1555013/442 [The reason partial sums are used for Banach limits is to get translation invariance.] – GEdgar Apr 12 '22 at 21:41

0 Answers0