3
$\begingroup$

I'm studying a paper, which I'll include a small piece here. And I'm struggling to calculate

$$C_n\|u_{m,n}\|^{\left(\frac{2*}{2}\right)^k\frac{2*-q}{(r_k)^k}}_{L^{2*}(\Omega)}$$ Where $\Omega$ is an open, bounded subset of $\mathbb{R}^N$. We also are considering values of $m$ and $n$ such that $m>\overline{t}$, and $n>\max\{\overline{t}, \overline{n}\}$, given that $\overline{n}>0$, $\overline{t}=t_0+1$ with $t_0>0$. We also are considering $\gamma>1$, $2^{*}=\frac{2N}{N-2}> \gamma +2$ and $p$, $q$ are such that \begin{align*} \gamma+2&<p<\frac{2^*}{2}(\gamma +2)\\ \gamma+2&< q< \min\{p,2^*\} \end{align*} Remark: $\mathcal{S}^2$ is a constant of Sobolev, I mean appears after I used one of Sobolev embeddings theorem. Here $\alpha$ is a positive constant.

I tried to use Moser iteration. However, I do not understand where I'm making mistakes. I found the expression in \eqref{5} which is clearly different. Would you check if the expression in \eqref{6} is right?

I bet there is a very big problem, because since the basis $\left(\frac{(r_{k-1}-q+2)^2}{4\alpha S^2 (r_{k-1}-q+1)}\right)$ goes to infinity (by \eqref{4}), how can the whole expression obtained in \eqref{5} be smaller or equal to $\tilde{C}_n$? partial of page 165

I've included the original paper in the link of google drive below. This part is on page 165 of the following paper:

David Arcoya, Lucio Boccardo, Luigi Orsina, "Critical points for functionals with quasilinear singular Euler-Lagrange equations.", Calculus of Variations and Partial Differential Equations 47, No. 1-2, 159-180 (2013), MR3044135, Zbl 1266.35102.

MY ATTEMPT

Let's see the iteration

\begin{equation} |u_{m,n}|_{L^{\frac{r-q+2}{2}2^*}(\Omega)}\leq \left(\frac{(r-q+2)^2}{4\alpha S^2 (r-q+1)}\right)^{\frac{1}{r-q+2}}n^{\frac{p-q}{r-q+2}}|u_{m,n}|_{L^r(\Omega)}^{\frac{r}{r-q+2}}\label{1}\tag{1} \end{equation}

Defining $r_0=2^*$ and $r_k=\left(\frac{2^*}{2}r_{k-1}+\frac{2^*}{(2-q)}\right)$. From \eqref{1} one has that

\begin{align*} &|u_{m,n}|_{L^{r_k}(\Omega)}=|u_{m,n}|_{L^{\frac{(r_{k-1}-q+2)}{2}2^*}(\Omega)}\label{2}\tag{2}\\ &\leq\underbrace{\left(\frac{(r_{k-1}-q+2)^2}{4\alpha S^2 (r_{k-1}-q+1)}\right)^{\frac{1}{r_{k-1}-q+2}}n^{\frac{p-q}{r_{k-1}-q+2}}}_{C_1}|u_{m,n}|_{L^{r_{k-1}}(\Omega)}^{\left(\frac{2^*}{2}\right)\frac{r_{k-1}}{r_k}}\\ &=C_1|u_{m,n}|_{L^{r_{k-1}}(\Omega)}^{\left(\frac{2^*}{2}\right)\frac{r_{k-1}}{r_k}}\\ &\leq C_1\left\{\left(\frac{(r_{k-2}-q+2)^2}{4\alpha S^2 (r_{k-2}-q+1)}\right)^{\frac{1}{r_{k-2}-q+2}}n^{\frac{p-q}{r_{k-2}-q+2}}|u_{m,n}|_{L^{r_{k-2}}(\Omega)}^{\left(\frac{2^*}{2}\right)\frac{r_{k-2}}{r_{k-1}}}\right\}^{\left(\frac{2^*}{2}\right)\frac{r_{k-1}}{r_k}}\\ &\leq C_1\underbrace{\left(\frac{(r_{k-2}-q+2)^2}{4\alpha S^2 (r_{k-2}-q+1)}\right)^{\frac{1}{r_{k-2}-q+2}\left(\frac{2^*}{2}\right)\frac{r_{k-1}}{r_k}}n^{\frac{p-q}{r_{k-2}-q+2}\left(\frac{2^*}{2}\right)\frac{r_{k-1}}{r_k}}}_{C_2}|u_{m,n}|_{L^{r_{k-2}}(\Omega)}^{\left(\frac{2^*}{2}\right)^2\frac{r_{k-2}}{r_{k-1}}\frac{r_{k-1}}{r_k}}\\ &=C_1C_2|u_{m,n}|_{L^{r_{k-2}}(\Omega)}^{\left(\frac{2^*}{2}\right)^2\frac{r_{k-2}}{r_k}}\\ &\leq C_1C_2\left\{\left(\frac{(r_{k-3}-q+2)^2}{4\alpha S^2(r_{k-3}-q+2)}\right)^{\frac{1}{r_{k-3}-q+2}}n^{\frac{p-q}{r_{k-3}-q+2}}|u_{m,n}|_{L^{r_{k-3}}(\Omega)}^{\frac{2^*}{2}\frac{r_{k-3}}{r_{k-2}}}\right\}^{\left(\frac{2^*}{2}\right)^2\frac{r_{k-2}}{r_k}}\\ &= C_1C_2\underbrace{\left(\frac{(r_{k-3}-q+2)^2}{4\alpha S^2(r_{k-3}-q+2)}\right)^{\frac{1}{r_{k-3}-q+2}\left(\frac{2^*}{2}\right)^2\frac{r_{k-2}}{r_k}}n^{\frac{p-q}{r_{k-3}-q+2}\left(\frac{2^*}{2}\right)^2\frac{r_{k-2}}{r_k}}}_{C_3}|u_{m,n}|_{L^{r_{k-3}}(\Omega)}^{\left(\frac{2^*}{2}\right)^3\frac{r_{k-3}}{r_{k-2}}\frac{r_{k-2}}{r_k}}\\ &= C_1C_2C_3|u_{m,n}|_{L^{r_{k-3}}(\Omega)}^{\left(\frac{2^*}{2}\right)^3\frac{r_{k-3}}{r_k}}\\ &\leq \cdots \leq\\ &\leq C_1C_2C_3\cdots C_{k-1}\underbrace{\left(\frac{(r_{0}-q+2)^2}{4\alpha S^2(r_{0}-q+2)}\right)^{\frac{1}{r_{0}-q+2}\left(\frac{2^*}{2}\right)^{k-1}\frac{r_1}{r_k}}n^{\frac{p-q}{r_{0}-q+2}\left(\frac{2^*}{2}\right)^{k-1}\frac{r_1}{r_k}}}_{C_k}|u_{m,n}|_{{L^{2^*}}(\Omega)}^{\left(\frac{2^*}{2}\right)^{k}\frac{r_0}{r_k}} \end{align*}

Now rewriting the first powers in $C_1, C_2,\ldots, C_k$, it follows that in

  1. item $C_1)$ \begin{align*} \frac{1}{r_{k-1}-q+2}&=\frac{\left(\frac{2^*}{2}\right)}{\left(\frac{2^*}{2}\right)(r_{k-1}-q+2)}\\ &=\left(\frac{2^*}{2}\right)\frac{1}{r_k} \end{align*}
  1. item $C_2)$ \begin{align*} \frac{1}{r_{k-2}-q+2}\left(\frac{2^*}{2}\right)\frac{r_{k-1}}{r_k}&=\frac{\left(\frac{2^*}{2}\right)}{\left(\frac{2^*}{2}\right)(r_{k-2}-q+2)}\left(\frac{2^*}{2}\right)\frac{r_{k-1}}{r_k}\\ &=\left(\frac{2^*}{2}\right)^2\frac{1}{r_{k-1}}\frac{r_{k-1}}{r_k}\\ &=\left(\frac{2^*}{2}\right)^2\frac{1}{r_k}\\ \end{align*}
  1. item $C_3)$ \begin{align*} \frac{1}{r_{k-3}-q+2}\left(\frac{2^*}{2}\right)^2\frac{r_{k-2}}{r_k}&=\frac{\left(\frac{2^*}{2}\right)}{\left(\frac{2^*}{2}\right)(r_{k-3}-q+2)}\left(\frac{2^*}{2}\right)^2\frac{r_{k-2}}{r_k}\\ &=\left(\frac{2^*}{2}\right)^3\frac{1}{r_{k-2}}\frac{r_{k-2}}{r_k}\\ &=\left(\frac{2^*}{2}\right)^3\frac{1}{r_k} \end{align*} $\vdots$
  1. item $C_k)$ \begin{align*} \frac{1}{{r_0}-q+2}\left(\frac{2^*}{2}\right)^{k-1}\frac{r_1}{r_k}&=\frac{\left(\frac{2^*}{2}\right)}{\left(\frac{2^*}{2}\right)(r_0-q+2)}\left(\frac{2^*}{2}\right)^{k-1}\frac{r_1}{r_k}\\ &=\left(\frac{2^*}{2}\right)^k\frac{1}{r_1}\frac{r_1}{r_k}\\ &=\left(\frac{2^*}{2}\right)^k\frac{1}{r_k} \end{align*}

Replacing properly the expressions above in every power of $C_1, C_2,\ldots, C_k$ it follows that \begin{align*} C_1&=\left(\frac{(r_{k-1}-q+2)^2}{4\alpha S^2(r_{k-1}-q+1)}\right)^{\left(\frac{2^*}{2}\right)\frac{1}{r_k}}n^{(p-q)\left(\frac{2^*}{2}\right)\frac{1}{r_k}}\\ C_2&=\left(\frac{(r_{k-2}-q+2)^2}{4\alpha S^2(r_{k-2}-q+1)}\right)^{\left(\frac{2^*}{2}\right)^2\frac{1}{r_k}}n^{(p-q)\left(\frac{2^*}{2}\right)^2\frac{1}{r_k}}\\ &\vdots\\ C_k&=\left(\frac{(r_0-q+2)^2}{4\alpha S^2(r_0-q+1)}\right)^{\left(\frac{2^*}{2}\right)^k\frac{1}{r_k}}n^{(p-q)\left(\frac{2^*}{2}\right)^k\frac{1}{r_k}}\\ \end{align*} Rewriting the whole expression in \eqref{2} replacing the exponents as above, one has that \begin{align*} |u_{m,n}|_{L^{r_k}}&\leq \left[\left(\frac{(r_{k-1}-q+2)^2}{4\alpha S^2(r_{k-1}-q+1)}\right)n^{(p-q)}\right]^{\left(\frac{2^*}{2}\right)\frac{1}{r_k}}\\ &\cdot \left[\left(\frac{(r_{k-2}-q+2)^2}{4\alpha S^2(r_{k-2}-q+1)}\right)n^{(p-q)}\right]^{\left(\frac{2^*}{2}\right)^2\frac{1}{r_k}}\\ &\vdots\\ &\cdot\left[\left(\frac{(r_0-q+2)^2}{4\alpha S^2(r_0-q+1)}\right)n^{(p-q)}\right]^{\left(\frac{2^*}{2}\right)^k\frac{1}{r_k}}\cdot |u_{m,n}|_{L^{2^*}(\Omega)}^{\left(\frac{2^*}{2}\right)^k\frac{r_0}{r_k}}\label{3}\tag{3}\\ \end{align*} Considering $r_0=2^*$. Since $\frac{2^*}{2}>1$, let's prove that $r_k$ is an increasing sequence which diverges to infinity.

Indeed, note that if

  1. item $(k=1)$ \begin{align*} r_1&=\frac{2^*}{2}(r_0-q+2)\\ &=\frac{2}{2^*}=r_0+(2-q)\\ \end{align*} Subtracting $\frac{2}{2^*}r_0$ in both sides, it follows that \begin{align*} \frac{2}{2^*}(r_1-r_0)&=r_0-\frac{2}{2^*}r_0+(2-q)\\ &=2^*-\frac{2}{2^*}2^*+2-q \\ &=2^*-q>0 \end{align*} Thus $r_1>r_0$.
  2. item $(k=2)$ , similarly \begin{align*} \frac{2}{2^*}(r_2-r_1)&=r_1-\frac{2}{2^*}r_1+(2-q)\\ &=\frac{2^*}{2}(r_0-q+2)-r_0+q-2-q+2\\ &=\frac{(2^*)^2-2^*q+22^*-22^*}{2}\\ &=\frac{2^*}{2}(2^*-q)>0 \end{align*} Therefore, $r_2>r_1$.

Proceeding in this way indefinitelyt, one can find that $r_k$ is an increasing sequence, that is $r_0<r_1<r_2<...<r_k<...$ and its limit is $$\lim_{k\to \infty}r_k=+\infty.$$ One claims that $$\left(\frac{(r_0-q+2)^2}{4\alpha S^2(r_0-q+1)}\right)<\left(\frac{(r_1-q+2)^2}{4\alpha S^2(r_1-q+1)}\right)<\dots<\left(\frac{(r_k-q+2)^2}{4\alpha S^2(r_k-q+1)}\right)<\dots\label{4}\tag{4}$$ Indeed, taking $t=r_0, r_1, r_2, ..., r_k,...$ and defining $$f(t)=\frac{(t-q+2)^2}{4\alpha S^2 (t-q+1)}$$ one has that \begin{align*} f'(t)&=\frac{2(t-q+2)4\alpha S^2(t-q+1)-(t-q+2)^24\alpha S^2}{4\alpha S^2(t-q+1)^2}\\ &=\frac{(t-q+2)[2t-2q+2-t+q-2]}{(t-q+1)^2} \end{align*} In order to show that $f$ is an increasing function one only needs to get that $(t-q+2)(t-q)>0$ and this is true, first note that if $t=r_0$ then $(2^*-q+2)(2^*-q)>0$ once $2^*-q>0$. Now, by \eqref{4}, replacing every single value for $t$ one has that $f'(t)>0$ which implies that $f$ is an increasing function. Hence one has proved the claim. Now turning back to \eqref{3} replacing the result in \eqref{4}, one can rewrite \eqref{3} as $$|u_{m,n}|_{L^{r_k}(\Omega)}\leq \left[\left(\frac{(r_{k-1}-q+2)^2}{4\alpha S^2(r_{k-1}-q+1)}\right)n^{(p-q)}\right]^{\frac{1}{r_k}\displaystyle\sum_{i=1}^{k}\left(\frac{2^*}{2}\right)^i}|u_{m,n}|_{L^{2^*}(\Omega)}^{\left(\frac{2^*}{2}\right)^k\frac{r_0}{r_k}}.\label{5}\tag{5}$$ On the other hand, notice that, taking $A=\left(\frac{2^*}{2}\right)(2-q)$ one can rewrite $r_k$ as \begin{align*} r_k&=r_{k-1}\left(\frac{2^*}{2}\right)+\left(\frac{2^*}{2}\right)(2-q)\\ &=r_{k-1}\left(\frac{2^*}{2}\right)+A\\ &=\left(r_{k-2}\left(\frac{2^*}{2}+A\right)\right)\left(\frac{2^*}{2}\right)+A\\ &=\left(\frac{2^*}{2}\right)^2r_{k-2}+\left(1+\frac{2^*}{2}\right)A\\ &=\left(\frac{2^*}{2}\right)^2\left[r_{k-3}\left(\frac{2^*}{2}+A\right)\right]+\left(1+\frac{2^*}{2}\right)A\\ &=\left(\frac{2^*}{2}\right)^3r_{k-3}+\left(1+\left(\frac{2^*}{2}\right)+\left(\frac{2^*}{2}\right)^2\right)A\\ &\vdots\\ &=\left(\frac{2^*}{2}\right)^kr_0+A+A\left(\frac{2^*}{2}\right)+A\left(\frac{2^*}{2}\right)^2+...+A\left(\frac{2^*}{2}\right)^{k-1} \end{align*} what is the same as $r_k=\left(\frac{2^*}{2}\right)^kr_0+A\displaystyle\sum_{i=0}^{k-1}\left(\frac{2^*}{2}\right)^i$, this is $$r_k=\left(\frac{2^*}{2}\right)^kr_0+(2-q)\left(\frac{2^*}{2}\right)\left[\frac{1-\left(\frac{2^*}{2}\right)^{k-1}}{1-\left(\frac{2^*}{2}\right)}\right].\label{6}\tag{6}$$

I appreciate if someone can help me! Thanks in advance.

$\endgroup$
8
  • 5
    $\begingroup$ I suggest you rewrite your calculations and this question completely. What you've written here appears to be using the same notation as in the original paper and contains many constants whose significance is unclear, such as $m$ and $q$. A link to the paper does not suffice. The question should be self-contained. I suggest rewriting this using fewer constants, only the essential ones. Also, I suggest replacing $2^*$ by its actual value. That makes the algebra much clearer. I also suggest writing the initial analytic inequalities in a simpler form. $\endgroup$
    – Deane Yang
    Commented Mar 13, 2022 at 14:52
  • 1
    $\begingroup$ All else aside, there's so much MathJax here that my browser completely chokes on it, and produces a jumbled mess. $\endgroup$
    – LSpice
    Commented Mar 13, 2022 at 20:11
  • $\begingroup$ @DanieleTampieri I would like to thank you for all corrections you've made. $\endgroup$
    – Silvinha
    Commented Mar 14, 2022 at 12:52
  • 1
    $\begingroup$ @Silvinha, here's my perspective: The odds that there's someone reading MathOverflow who has or is reading the same paper are pretty low. On the other hand, the odds that there is someone who knows this stuff inside out are pretty good. Lots of such experts here. The easier you make it for them to read and understand the chain of inequalities, the more likely you'll get an answer. Here's the other thing: When I'm trying to understand something like this, I always rewrite the argument myself in the simplest way possible. Just doing that often clarifies what's going on a lot. $\endgroup$
    – Deane Yang
    Commented Mar 16, 2022 at 16:27
  • 1
    $\begingroup$ @DeaneYang I agree with you. I will make changes as soon as possible again. I intend to create a new question with fewer variables. I intend to keep this question because I sent the link to authors. I appreciate all advice, thanks. $\endgroup$
    – Silvinha
    Commented Mar 17, 2022 at 20:18

0

You must log in to answer this question.