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

Here is my problem. Let $C$ be the space of continuous and nondexcreasing functions defined on $[0,1]$ and endowed with the sup norm. Take any $z\in C$ and consider the following: $$ U(x;z)=\int_{0}^{x}\left[1-\int_{s}^{1}F(z(\xi))f(\xi)d\xi\right]^{n-1}ds $$ where $F:[0,1]\rightarrow [0,1]$ is continuously differentiable and increasing; $F'=f$, $f(s)>0, s\in[0,1]$ and $f(s)=0,\,s\notin [0,1]$, $x\in[0,1]$, and $n>2$. Fix the value of $x$. I would like to show that $U$ is Lipschitz continuous with respect to $z$, i.e., I'd like to show that there is a constant $K>0$ such that:

$$ ||U(x;z)-U(x;y)||\leq K||z-y|| \qquad z,y\in C $$

This is what I've done so far. For a fixed value of $x$ the above expression can be treated as a functional. I know that the Frechet derivative of this expression looks like the following:

$$ \frac{dU(z)}{dz}=-(n-1)\int_{0}^{x}\left[1-\int_{s}^{1}F(z(\xi))f(\xi)d\xi\right]^{n-2}\left(\int_{s}^{1}f(z(\xi))f(\xi)h(\xi)d\xi\right)ds $$ where $h$ is the 'perturbation' function. Since $f$ is continuous and $f(s)>0$ if $s\in[0,1]$, it must be the case that:

$$ \left|\frac{dU(z)}{dz}\right|\leq (n-1)M_0\int_{0}^{x}\left|\left[1-\int_{s}^{1}F(z(\xi))f(\xi)d\xi\right]^{n-2}\left(\int_{s}^{1}h(\xi)d\xi\right)\right|ds\\ \leq (n-1)M_0\int_{0}^{x}\left(\int_{s}^{1}|h(\xi)|d\xi\right)ds $$ where $M_0=\sup_{s\in[0,1]} |f(s)f(s)|$, and the second line follows because $f(z(s))f(s)\leq \sup_{s\in[0,1]}|f(z(s))f(s)|$ for any $z\in C$, and $\left[1-\int_{s}^{1}F(z(\xi))f(\xi)\right] \leq 1$ for any $z\in C$ and $s\in[0,1]$.

I know that the perturbation $h$ is bounded because it belongs to $C$ but any such bound depends on the choice of the perturbation and hence, the bound of the whole expression will depend on $h$. Does this means that I am not free to choose the overall bound that would allow me to show the existence of a Lipschitz constant $K$?

Any help/suggestion/insight/reference is greatly appreciated it!

share|cite|improve this question
The derivative is a linear functional of $h$, so of course it depends on $h$. What you want is a bound on the norm of this linear functional. Recall that to give a bound for linear functional means to prove an inequality of the form $|Th|\le M\|h\|$, with $M$ independent of $h$. As long as $M$ is independent of $z$, you have uniformly bounded derivative, hence the Lipschitz condition. – user31373 Jul 10 '12 at 22:20
Thanks! Please correct me if I am wrong, but isn't enough that the Frecht derivative above exists to claim that the linear operator $Th$ is bounded? From the little I know about Frecht derivatives, I can recall that a functional is Frecht diff if there exists a bounded linear operator $T$ $\lim_{h\rightarrow 0}\frac{||U(z+h)-U(z)-Th||}{||h||}=0$. So, if I found the Frecht derivative, wouldn't this be sufficient to claim that $|Th|\leq M||h||$ holds true? Sorry if what I's saying is nonsense but I am truly confused at this stage. Thanks for any help you can provide me with this. – Cristian Jul 10 '12 at 22:44
The Frechet derivative is a bounded functional by definition. In your post I did not see a proof that what you found is the Frechet derivative. But let's assume that it is. You still need a bound with $M$ independent of $z$; this does not come from any definitions, and you actually need some estimates. – user31373 Jul 10 '12 at 22:51
Thanks again Leonid. So you're basically saying that my work so far is no god to show what I want to show. Do you have any suggestion/hint how I should proceed? I've been stuck here for quite some time and I just don't see any way out, although intuitively I would expect that my claim (about lipschitz) is true... – Cristian Jul 10 '12 at 23:04
I did not say that. You did estimate $|Th|$ in terms of $h$ (I use $T$ as a shorter notation for the derivative). All that's left to do is bring it to the form $|Th|\le M\|h\|$. Recall that $\|h\|$ is the supremum of $h$, and it's easy to estimate integrals from above in terms of supremum. – user31373 Jul 10 '12 at 23:15
up vote 0 down vote accepted

You could continue your estimates as follows: $$\left|\frac{dU(z)}{dz}\right|\leq (n-1)M_0\int_{0}^{x}\left(\int_{s}^{1}|h(\xi)|d\xi\right)ds \le (n-1)M_0\int_{0}^{x} (1-s)\|h\| ds \le (n-1)M_0 \|h\|$$ Hence, $U$ is Lipschitz with the Lipschitz constant at most $(n-1)M_0$.

share|cite|improve this answer
thanks! I did exactly that (before seeing your post). I am confident now of what I did! Thanks a lot! – Cristian Jul 11 '12 at 0:17

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.