Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

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

Let $$\varphi(s) = \begin{cases} s &: s < 0\\ 0 &: s \in [0,1]\\ s-1 &: s > 1 \end{cases}$$ Note that $\varphi$ is Lipschitz. Consider where $f \in L^2(0,T;L^2(\Omega)$, $$u_t - \Delta (\varphi(u)) = f \text{ in $\Omega \times (0,T)$}$$ $$u|_{\partial\Omega} =0$$ $$u(0) = u_0$$ Apparently if I take $\varphi(u)$ as a test function, I can obtain an estimate on $u$ in $L^\infty(0,T;L^2(\Omega))$, but I can't see this. Can someone help me? Thanks.

share|cite|improve this question
up vote 2 down vote accepted

Since $\varphi$ is just Lipschitz, second order weak derivatives of $\varphi(u)$ may not exist no matter how smooth be a solution $u$. That is why $u$ is to be treated only as a weak solution. So, let $u\in H^1\bigl(\Omega \times (0,T))\bigr)$ be a weak solution of the problem, i.e., let $u$ possessing traces $u|_{t=0}=u_0$ and $u|_{\partial\Omega}=0$ satisfy the integral identity $$ \int\limits_{\Omega}u_t v\,dx+\int\limits_{\Omega}\nabla\varphi(u)\cdot \nabla v\,dx=\int\limits_{\Omega}fv\,dx \quad \forall\, v\in H^1\bigl(\Omega\times (0,T))\bigr)\colon\,v|_{\partial\Omega}=0 $$ a.e. on $(0,T)$. The required bound follows by a routine trick of introducing a function $\Phi\in C^1(\mathbb{R})$ such that $\Phi'=\varphi$ where $$ \Phi(s)= \begin{cases} \frac{s^2}{2}, &s<0,\\ 0, &s\in [0,1],\\ \frac{(s-1)^2}{2}, &s>1. \end{cases}\tag{$\ast$} $$ Notice that $\varphi^2(s)=2\Phi(s)\leqslant s^2\;\forall\,s\in\mathbb{R}$. Choosing a test function $v=\varphi(u)$ yields $$ \frac{d\,}{dt}\int\limits_{\Omega}\Phi(u)\,dx+\int\limits_{\Omega}|\nabla\varphi(u)|^2dx =\int\limits_{\Omega}f\varphi(u)\,dx $$ a.e. on $(0,T)$, whence by the Cauchy–Bunyakovski inequality follows $$ \frac{d\,}{dt}\int\limits_{\Omega}\Phi(u)\,dx\leqslant \|f\|_{L^2(\Omega)} \|\varphi(u)\|_{L^2(\Omega)}=\|f\|_{L^2(\Omega)}\biggl(2\int\limits_{\Omega} \Phi(u)\,dx\biggr)^{1/2} $$ a.e. on $(0,T)$, implying the bound $$ \biggl(2\int\limits_{\Omega}\Phi\bigl(u(x,t)\bigr)\,dx\biggr)^{1/2}\leqslant \int\limits_0^T\|f\|_{L^2(\Omega)}dt+\|u_0\|_{L^2(\Omega)} $$ a.e. on $(0,T)$. The rest is easy due to the definition $(\ast)$ of function $\Phi$.

Remark. At points $x,t$ where $0\leqslant u(x,t)\leqslant 1$ solution's smoothness w.r.t. $x$ is not generally to be expected higher than that of a solution $u(x,t)\overset{\rm def}{=}\psi(x)$ with some $L^2(\Omega)$ function $\psi\,\colon \Omega\to [0,1]$ vanishing near the boundary $\partial\Omega$ and possessing no weak derivatives, which solves the problem corresponding to $f=0$ and $u_0=\psi$.

share|cite|improve this answer

Well, let me forget $f$. Then, notice that $$ \int u_t\phi(u)=-\int (\nabla \phi(u))^2, $$ where we used the boundary conditions and the definition of $\phi$ to remove the boundary term $$ \int_{\partial\Omega}\phi(u)\nabla\phi(u)nds=0. $$ Now $$ \int u_t\phi(u)=\int u_t u\textbf{1}_{u<0}+\int u_t (u-1)\textbf{1}_{u>1}=\int u_t u\textbf{1}_{u<0}+\int u_t u\textbf{1}_{u>1}-\int u_t \textbf{1}_{u>1}. $$ Now $$ 0.5\frac{d}{dt}|u|_{L^2}=\int uu_tdx=\left(\int uu_t\textbf{1}_{u>1}+\int uu_t\textbf{1}_{0\leq u\leq 1}+\int uu_t\textbf{1}_{u>1}\right). $$ So $$ 0.5\frac{d}{dt}|u|_{L^2}=\int u_t\phi(u)+\int u_t\textbf{1}_{u>1}+\int uu_t\textbf{1}_{0\leq u\leq 1}, $$ $$ 0.5\frac{d}{dt}|u|_{L^2}=-\int (\nabla \phi(u))^2+\int u_t\textbf{1}_{u>1}+\int uu_t\textbf{1}_{0\leq u\leq 1}. $$ Now, if $u>1$, $u_t=\Delta (u-1)=\Delta u$, so $$ \int u_t\textbf{1}_{u>1}=0. $$ On the other hand, if $|u|<1,$ $\partial_t u=0$. Putting all together we conclude, right? Is that correct?

share|cite|improve this answer

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.