5
$\begingroup$

I recently read a result (in Jarchow's book) that any ultrabornological space can be expressed as a colimit (in the category LCS) of Banach spaces. My question is the following.

Let $\mu$ be a finite measure on $\mathbb{R}$. Is there an uncountable set of $p_i \in [0,\infty)$ and finite Borel measures $\mu_i$ on $\mathbb{R}$ for which one can construct an inductive system $\{L^{p_i}_{\mu_i}(\mathbb{R})\}$ of locally convex spaces for which $$ \injlim L^{p_i}_{\mu_i}(\mathbb{R}) \cong L^{\infty}_{\mu}(\mathbb{R})? $$

Edit: Alternatively, I look for a system such that $ \injlim L^{p_i}_{\mu_i}(\mathbb{R}) $ is dense in $L^{\infty}_{\mu}(\mathbb{R})$ when its topology is relaxed from the final topology (which need not be metrizable) to the (relative) Banach topology of $L^{\infty}_{\mu}(\mathbb{R})$.

Improve this question
$\endgroup$

3 Answers 3

Reset to default
10
$\begingroup$

The answer to this question is YES -- but it is useless!

In fact, a theorem of Valdivia (which you can find, e.g., as Theorem 6.5.8 in the book Barrelled Locally Convex Spaces of Bonet and Perez-Carreras) states that given any infinite-dimensional separable Banach space $F$, one can represent any ultrabornological locally convex space $E$ (this means, that $E$ has some representation as an inductive limit of Banach spaces) which does not carry the finest locally convex topology as an inductive limit of Banach spaces isomorphic to $F$ such that, in addition,the linking maps between the steps are nuclear operators.

This theorem is rather useless because it is too good: Every ultrabornological space has such a representation and thus you cannot deduce any nice properties from it.

Improve this answer
$\endgroup$
5
  • $\begingroup$ Oh, so these representaitons are never "explicit"; I'm assuming... $\endgroup$
    – ABIM
    Commented Jan 31, 2020 at 13:44
  • 16
    $\begingroup$ "The theorem is so good that it is useless." Got to remember that one. $\endgroup$
    – Bill Johnson
    Commented Jan 31, 2020 at 15:36
  • 1
    $\begingroup$ @AIM_BLB Just a rewording of the penultimate line of Jochen's answer. $\endgroup$
    – Bill Johnson
    Commented Mar 23, 2020 at 15:53
  • $\begingroup$ @JochenWengenroth I have been thinking about this answer for some time (as you may notice) and is there a characterization/(can more be said about) of LF spaces which are the limit of finite-dimensional Fréchet spaces? (Possibly after the topology is relaxed in the limit). $\endgroup$
    – ABIM
    Commented Jun 9, 2020 at 19:39
  • $\begingroup$ Then every linear map into any locally convex space is continuous because this is so for finite dimensional spaces. I think that there is only one such LF-space. $\endgroup$
    – Jochen Wengenroth
    Commented Jun 9, 2020 at 20:31
3
$\begingroup$

When I read your question, I assumed that you were asking whether there is a way to do this in the context of the underlying structure rather in the spirit of the answer above which addresses the situation where the spectrum consists of spaces which are merely isomorphic to the appropriate ones. Here are some suggestions (rather than an answer).

For clarity, I will work in the context of a space $\Omega$ with a $\sigma$–algebra $\cal A$ (rather than the reals) and the $L^1$-case. The family of positive finite measures forms a lattice and so the corresponding family $\{L^1(\mu)\}$ can be regarded both as an inductive and a projective spectrum in a natural way. So we can take inductive and projective limits in various senses.

Your query refers to the former but I am afraid that it seems to me that the results are not of interest (a purely subjective opinion, of course). One can take projective limits in three senses—that of Banach spaces (with linear contractions as morphisms), strict topologies, and locally convex spaces. Only the first two are of interest in the context of your query.

The first case yields the Banach space of bounded, measurable functions, the second one the same space with a weaker, but still complete, lc topology. If you are interested in duality theory, I would suggest that the latter space is the interesting one—its dual is the space of $\sigma$-additive measures on $\cal A$.

Improve this answer
$\endgroup$
2
  • $\begingroup$ do you happen to have a reference to this type of literature? $\endgroup$
    – ABIM
    Commented Mar 22, 2020 at 11:43
  • 2
    $\begingroup$ I am afraid not. I’m afraid that this is a deeply unfashionable area, so one has to work these things out for oneself with no hope of publication. $\endgroup$
    – user131781
    Commented Mar 23, 2020 at 10:46
1
$\begingroup$

It's a very short answer but this may be of interest to you:

Davis, Henry W.; Murray, F. J.; Weber, J. K. Jr., Families of $L_p$-spaces with inductive and projective topologies, Pac. J. Math. 34, 619-638 (1970). ZBL0187.05802.

https://projecteuclid.org/download/pdf_1/euclid.pjm/1102971942

Improve this answer
$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .