12
$\begingroup$

This last summer, I read up on Tartar's so called Method of Compensated Compactness (or at least how it applied to scalar conservation laws). I used this theory to prove the existence of $L^{\infty}$ solutions to the scalar conservation law $u_{t}+(f(u))_{x}=0$ with initial data in $L^{\infty}$. Here we do not need to have a strictly hyperbolic conservation law. Perhaps just assume $f\in C^{2}$. The sources I used for my presentation were from Yunguang Lu's book Hyperbolic Conservation Laws and the Compensated Compactness Method and Dafermos's book Hypberbolic Conservation Laws in Continuum Physics.

I am aware that Compensated Compactness can be used to prove existence of solutions to $2\times 2$ systems of conservation laws. I haven't looked over this proof, but I believe Lu covers this.

My question is as follows: Where else is the method of Compensated Compactness used? If you can say, if possible, could you give a rough, brief sketch of how the method was used? Any input/thoughts would be appreciated.

$\endgroup$
2
  • $\begingroup$ It seems me that the application of the Compensated Compactness approach - which is essentially a qualitative approach do not give any interesting practical new informations regarding the problem of the <modelisation/simulation and computation> in Fluid Dynamics! In fact, the application of a such rigorous pure mathematical qualitative approach seems me - ONLY - motivated by the desired demonstration of the practical importance of a such abstract approach! It is very unfortunate that young researchers in mathematics do not have interested by the problems of fluid dynamics! Obviously, (cont.) $\endgroup$
    – user90173
    Apr 10, 2016 at 16:08
  • $\begingroup$ (cont.) the fluid dynamics is characterized by its FRAGMENTATION, by the fact that various problems in it are often solved by an ad hoc methods, and that no aspect of it is understood down to the ground - all this makes it difficult for a working student of mathematics to enter the field! At the beginning, such a student is faced with what seems a NEVER-EDING LIST OF EQUATIONS DESCRIBING SOME PORTION OF FLUID DYNAMIC? WHERE IS HE TO BEGING? R. Kh. Zeytounian 10/04/2016 $\endgroup$
    – user90173
    Apr 10, 2016 at 16:08

2 Answers 2

13
$\begingroup$

Compensated compactness helps when one needs to find the limit of $u_n \cdot v_n$, where the sequences of vector fields $u_n$ and $v_n$ converge weakly in $L^2$: $u_n\rightharpoonup u$, $v_n\rightharpoonup v$. If none of the sequences converges strongly in $L^2$, the vector fields can still possess some additional properties which would compensate for the lack of strong convergence. For instance, the following result and its versions are often used in the homogenization theory.

Lemma. Let $u_n$, $v_n\in L^2(\Omega)$, where $\Omega$ is a bounded domain in $\mathbb R^d$ and let $u_n\rightharpoonup u$, $v_n\rightharpoonup v$ in $L^2(\Omega)$. Assume that $$\mbox{curl }v_n=0\qquad \mbox{and}\qquad \mbox{div } u_n\to w\quad \mbox{in }\ H^{-1}(\Omega).$$ Then $u_nv_n\to uv$ in the weak-* topology of $L^1(\Omega)$.

The lemma can be used to justify the convergence of solutions of the Dirichlet problem with strongly oscillating coefficients $$\mbox{div }\left(a\left(\frac{x}{\varepsilon}\right)\nabla u_\epsilon\right)=f,\qquad \left.u\right|_{\partial\Omega}=0,$$ to the solutions of some averaged problem (see, e.g., "Homogenization of differential operators and integral functionals" by V. V. Jikov, S. M. Kozlov, O. A. Oleinik).

$\endgroup$
8
$\begingroup$

In addition to proving the existence of solutions to a large class of nonlinear partial differential equations, among them conservation laws, compensated compactness is also used to show the convergence of numerical methods to approximate those solutions. Given an equation $P(u)=0$, one constructs by whatever method (finite differences, finite elements, spectral,...) a sequence of approximations $u_n$. Then one tries to prove:

  1. the sequence $\{u_n\}$ converges in a some sense to some function $u$;
  2. this function $u$ is a solution of the original problem;
  3. an error estimate for $u-u_n$.

Then, if sufficiently good a priori estimates are known for $u_n$, the method of compensated compactness can give 1. and 2.This plan is carried out for instance here and here,

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.