-1
$\begingroup$

Setting

  • For $i \in \mathbb{N}$, consider two sequences $f_i,g_i \in L^1(\mathbb{R})$ such that $$ f_i \rightarrow_{L^1} f \in L^1(\mathbb{R}) $$ and also $$ g_i \rightarrow_{L^1} g \in L^1(\mathbb{R}).$$

  • Assume that all the functions involved are (1) striclty positive and that for any $i \in \mathbb{N}$, we have (2) $$ 0 < \int_\mathbb{R} f_i^2 g_i^{-1} < \infty$$ and also (3) $$ 0 < \int_\mathbb{R} f^2 g^{-1} < \infty $$

Question

Do we have the following convergence $$\int_\mathbb{R} f_i^2 g_i^{-1} \rightarrow \int_\mathbb{R} f^2 g^{-1} $$

$\endgroup$
1
  • $\begingroup$ If you look at $$ \int f_i^{\frac{1}{2}}g^{\frac{1}{2}}_i $$ instead then you do have convergence... But you need to use Holder whereas in this setting $(2,-1)$ you cannot (the natural scaling that pops up is Reverse holder...) $\endgroup$
    – Anthony
    Commented Apr 17 at 8:34

1 Answer 1

5
$\begingroup$

No, this is not true. The following argument is about spaces of finite measure, but it can easily be adapted to your question about $\mathbb R$.

Let $(X, m)$ be a space with $m(X) < \infty$. Take $g_i = g = 1$, and $f=0$. You are then asking the following:

if $f_i \in L^2 (X) \subseteq L^1 (X)$, and if $f_i \to 0$ in $L^1 (X)$, does it follow that $f_i \to 0$ in $L^2 (X)$?

No, it does not: take $X = [0,1] \subset \mathbb R$, and $f_i = i \; 1_{[0, \frac 1 {i^2}]}$ (with $1_A$ the indicator function of the subset $A$). Then $f_i \to 0$ in $L^1$ but $\| f_i \| _{L^2} = 1 \not\to 0$ in $L^2$.

$\endgroup$
4
  • $\begingroup$ Thank you ! But I think your counter example does not work because $f$ is not strictly positive as assume in the statement... $\endgroup$
    – Anthony
    Commented Apr 17 at 8:56
  • 2
    $\begingroup$ @Anthony shift $f_i$ to $f_i + c$ for a positive constant $c$ $\endgroup$
    – Hannes
    Commented Apr 17 at 9:17
  • 1
    $\begingroup$ @Anthony: As Hannes says, take for instance $f_i = 1 + i \; 1_{[0, \frac 1 {i^2}]}$ and $f=1$. $\endgroup$
    – Alex M.
    Commented Apr 17 at 9:25
  • $\begingroup$ Yeah you right sry.. $\endgroup$
    – Anthony
    Commented Apr 17 at 9:53

You must log in to answer this question.

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