Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I want to find out for which $p, q$ exists $C > 0$ with $||f||_p \leq C ||f||_q$ for all $f \in C([0,1]), p \in \mathbb{R}_{\geq 1} \cup \{\infty\}$. I first let $p,q \geq 1$ and I looked at the case where $q > p$.

I defined $\tilde{f} := |f|^p$ and $\tilde{p} := \frac{q}{p}$, $f \in C([0,1])$ of course. Also, I let $g(x) = 1$ and applied Hölder's inequality:

$$||\tilde{f} \cdot g||_1 \leq ||\tilde{f}||_\tilde{p} \cdot ||g||_\tilde{q}$$

where of course $||g||_\tilde{q} = 1$.

$$\implies \int_0^1 |f|^p \mathrm dx \leq \left ( \int_0^1 |f|^q \mathrm dx \right )^\frac{p}{q} \iff ||f||_p \leq ||f||_q \cdot C \;\; \forall C \geq 1$$

So my solution is, that for $q > p \geq 1$, there always exists $C > 0$ such that the inequality holds good. Now, before I look at $p > q$, I wanted to know whether my solution is correct. It doesn't seem to be complete to me, as there might also exist $C \in (0,1)$ such that the inequality holds good. Am I wrong? If not, how can I find out for which $p$ and $q$ there exists a $C \in (0,1)$?

Thanks in advance for any answers...

share|cite|improve this question
up vote 3 down vote accepted

You have proved what you set out to: if $q > p \ge 1$ there exists $C > 0$ (namely, $C=1$) such that $||f||_p \le C ||f||_q$ for all $f \in C([0,1])$.

Note you are not trying to prove that $||f||_p \le C ||f||_q$ for all $C > 0$. This is easily seen to be false.

You might ask whether this result is sharp; whether there exists a number $0 < C < 1$ which could work. To find out, consider $f=1$.

share|cite|improve this answer
Thanks a lot for this answer. I indeed haven't considered the case $f=1$. – Huy Mar 13 '11 at 16:48

Your argument seems to be correct. It shows that indeed $\Vert f\Vert_{p} \leq \Vert f\Vert_{q}$ for $1 \leq p \leq q \lt \infty$.

Instead of using Hölder's inequality, you could also argue with Jensen's inequality (which is often used in the proofs of Hölder's inequality) as follows: For $p \leq q$ the function $\varphi(x) = x^{q/p}$ is convex, hence $\left(\int |f|^p \right)^{q/p} = \varphi\left(\int |f|^p\right) \leq \int \varphi(|f|^p) = \int |f|^{q}$ and extracting $q$th roots on both sides gives the desired $\Vert f \Vert_{p} \leq \Vert f \Vert_{q}$ for $p \leq q \lt \infty$. The case $q = \infty$ needs to be dealt with separately in this argument and I leave this to you.

To see that $C = 1$ is the best possible constant simply take $f$ to be the constant function $f = 1$, which has $\Vert f \Vert_{p} = 1$ for all $p$, so there can't be a constant $C \lt 1$.

share|cite|improve this answer
You say my argument shows the inequality for $1 \leq p \leq q < \infty$. But in my question, I also allow $p = \infty$. What exactly changes in that case? – Huy Mar 13 '11 at 17:24
@Huy: The case $q = \infty$ does not work as written because you'd integrate $|f|^{\infty}$. The easiest way to remedy this is to replace $f$ by $g = f/\Vert f\Vert_{\infty}$ (assuming $f \neq 0$ of course). Then observe that $\Vert g\Vert_{p} \leq 1$ and use that the $p$-norm is homogeneous, so $\Vert f\Vert_{p} \leq \Vert f\Vert_{\infty}$ by multiplying the last inequality by $\Vert f\Vert_{\infty}$. – t.b. Mar 13 '11 at 17:39

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.