Today’s problem is a commonly appearing exercise in the theory of  spaces.
 spaces.
Problem: Let  be a finite measure space, and let
 be a finite measure space, and let  for all
 for all  Show that
 Show that  
 
Solution: Let’s first understand intuitively why this might be true. As  the parts where
 the parts where  start contributing more to the
 start contributing more to the  -norm of
-norm of  (since
 (since  gets very large for large
 gets very large for large  and
 and  ) while the parts where
) while the parts where  contribute less and less to the norm. Thus, we would expect that in the limit as
 contribute less and less to the norm. Thus, we would expect that in the limit as  the part that will contribute to the norm is the largest value of
 the part that will contribute to the norm is the largest value of  or in other words, the essential supremum
 or in other words, the essential supremum  It also suggests that for this problem, instead of thinking horizontally, that it is instead helpful to think vertically, i.e. in terms of the sets
 It also suggests that for this problem, instead of thinking horizontally, that it is instead helpful to think vertically, i.e. in terms of the sets 
Now for a formal proof. We prove the existence and value of the limit by showing both  and
 and  By Hölder’s inequality with
 By Hölder’s inequality with  and
 and  we have
 we have 
      ![Rendered by QuickLaTeX.com \[\|f\|_p \leq \mu(X)^{\frac{1}{p}-\frac{1}{q}} \|f\|_q\]](https://www.stepanmalkov.com/wp-content/ql-cache/quicklatex.com-ae6d0286c98a0ae3b875e8678f7322f5_l3.png)
 In particular, for
 In particular, for  we get that
 we get that  Taking the limsup on both sides yields
 Taking the limsup on both sides yields  
For the other direction, we recall the decomposition
      ![Rendered by QuickLaTeX.com \[\int |f|^p dx = \int_{|f|>\lambda} |f|^p dx + \int_{|f| \leq \lambda} |f|^p dx \geq \lambda^p \mu(\{x:|f|>\lambda\}).\]](https://www.stepanmalkov.com/wp-content/ql-cache/quicklatex.com-39ebf8e76081f5447b1e3b909de3442e_l3.png)
 Thus, for any
 Thus, for any  let
 let  Taking
 Taking  -th roots on both sides yields
-th roots on both sides yields       ![Rendered by QuickLaTeX.com \[\|f\|_p \geq (\|f\|_\infty-\epsilon)\mu(\{x:|f|>\|f\|_\infty-\epsilon\})^{\frac{1}{p}}.\]](https://www.stepanmalkov.com/wp-content/ql-cache/quicklatex.com-baeeee39fb257cc0aca92f0d91a57621_l3.png)
 as
 as  on both sides yields
 on both sides yields  and since
 and since  is arbitrary, we obtain
 is arbitrary, we obtain       ![Rendered by QuickLaTeX.com \[\limsup_{p \to \infty} \|f\|_p \leq \|f\|_\infty \leq \liminf_{p \to \infty} \|f\|_p,\]](https://www.stepanmalkov.com/wp-content/ql-cache/quicklatex.com-490de0e5431ba32f5c604ccf1aefc7f4_l3.png)
 
Remark 1: By interpolation theorems, it suffices to take  for any
 for any  for the statement of the problem to hold, since then
 for the statement of the problem to hold, since then  for any
 for any  
 
Remark 2: This result still holds true if  The only issue in our argument is now the application of Hölder’s, which only works on finite measure spaces. In this case, we can instead use the following trick, taking
 The only issue in our argument is now the application of Hölder’s, which only works on finite measure spaces. In this case, we can instead use the following trick, taking  to be large enough so that
 to be large enough so that  :
: 
      ![Rendered by QuickLaTeX.com \[\|f\|_p^p = \|f^r f^{p-r}\|_1 \leq \|f^{p-r}\|_\infty \|f^{r}\|_1 = \|f\|_\infty^{p-r} \|f^r\|_1,\]](https://www.stepanmalkov.com/wp-content/ql-cache/quicklatex.com-f1a10a303ce7ce4bc21c379eb3e21874_l3.png)
 -th roots on both sides and taking the
-th roots on both sides and taking the  as
 as  yields
 yields       ![Rendered by QuickLaTeX.com \[\limsup_{p \to \infty} \|f\|_p \leq \|f\|_\infty.\]](https://www.stepanmalkov.com/wp-content/ql-cache/quicklatex.com-570b54cc11175a0389424190b1b00de6_l3.png)