9
$\begingroup$

Considering $L^p$ $( 1 \leq p < \infty)$ as a normed vector space, each element of $L^p$ is actually an Equivalent class. Take $[f] \in L^p $ as an Equivalent class, What is the Nicest possible function $g$ such that $g \in [f]$ (i.e. $g=f$ almost everywhere).

By the word Nice you are free to consider any good topological or algebraic property like continuity, differentiability, boundedness, etc... (as many as you can)

Second question: Let $N$ denotes the set of such Nice Properties, imposing $N$ as set of conditions on $g$, can one identify $g$ uniquely ? In other words can we rewrite $L^p$ in following way

$$L^p = \{ g : R\rightarrow R \cup \{\pm \infty\} \quad | \quad \|g\|_p < \infty ,~g\text{ satisfies }N \}$$

This makes we can think about $L^p$ as a set of nice functions, free of any confusing equivalent classes.

Clarification: My intention is to find a set of finite conditions called $N$ , when we imposing them, $g$ is determined uniquely, in order to replace this $g$ by $[f]$ to rewrite $L^p$ as above. Lusin's theorem says one can pick $g \in [f]$ such that $g$ is continuous except on very small set (as small as you want but might have positive measure).This might be helpful but still can't give us the set of condition $N$ !

$\endgroup$
2
  • 4
    $\begingroup$ Lifting theory is well developed. Start with $$ $$ en.wikipedia.org/wiki/Lifting_theory $\endgroup$ Commented May 26, 2017 at 6:22
  • $\begingroup$ Thanks, I am not sure using that, I can identify $N$. $\endgroup$
    – Red shoes
    Commented May 26, 2017 at 6:28

1 Answer 1

2
$\begingroup$

In harmonic analysis one frequently defines a modification of $f$ by taking the new value $f(x)$ as the limiti of averages of $f$ over balls centered at $x$, with radius tending to 0. You do this at every point for which the limit exists. If $f$ is equivalent to a continuous function, this method produces the continuous representative. By Lebesgue differentiation this limit exists at almost every $x$ and coincides with the original value of $f(x)$ for any locally $L^1$ function; of course it remains the question of how to redefine $f$ at points where the limit do not exist; maybe one can set $f=0$ there. I guess you can not do better than this, but I might be wrong since I do not know lifting theory.

$\endgroup$
9
  • $\begingroup$ what is significant about this limit when $f$ is not continuous ? Actually when $f$ is equivalent to continuous function? if we now $f$ is equivalent to a continuous then we are done with $[f]$. no need think about that procedure . $\endgroup$
    – Red shoes
    Commented May 26, 2017 at 6:50
  • $\begingroup$ Is it better than Lusin's Theorem? Thanks for your answer. $\endgroup$
    – Red shoes
    Commented May 26, 2017 at 7:04
  • 1
    $\begingroup$ I would take lower or upper limit for any point (depending on what we need). $\endgroup$ Commented May 26, 2017 at 9:22
  • $\begingroup$ @FedorPetrov Takin lower/upper limit of what ? what limit? can clarify it? $\endgroup$
    – Red shoes
    Commented May 26, 2017 at 10:52
  • $\begingroup$ Of averages over small balls. $\endgroup$ Commented May 26, 2017 at 10:59

You must log in to answer this question.

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