Integral equality in a matrix setting

30 Views Asked by At

While reading a proof I stumbled over the following equation:

\begin{align} \int_{B_1} \langle A,D\psi(x)\rangle dx=0 \quad \forall \psi\in W^{1,\infty}_0(B_1,\mathbb{R}^2)\forall A\in \mathbb{R}^{2\times 2} \end{align} I denote by $\langle\ \ ,\ \ \rangle$ the Frobenius inner product. This looks like something derived by partial integration, which I have never seen in that Frobenius setting. Also in the proof it is implied that $\int_{B_1} \langle D\psi(x),D\psi(x)\rangle dx=0$ does not hold. I would appreciate any hint why the equation holds and if one can adapt partial integration in that setting.

1

There are 1 best solutions below

0
On BEST ANSWER

You can literally just write out the inner product and use the well-know integration by parts formula: $$ \langle A, D\phi \rangle=A_{11}\partial_1\phi_1+A_{12}\partial_1\phi_2+A_{21}\partial_2\phi_1+A_{22}\partial_2\phi_2 $$ Since your matrix is constant you can just swap the derivates using integration by parts; ill do it for 1 term as an example $$ \int_{B_1} A_{11}\partial_1\phi_1 dx=-\int_{B_1} \phi_1\partial_1 A_{11}=0 $$ Summing over 4 zero terms is still zero. Boundary terms are getting dropped since your function vanishes on the boundary by assumption.
The second eqzuality only holds when $D\phi=0$ by properties of the scalar product.