0
$\begingroup$

Is there a way to apply integration by parts or the divergence theorem to the integral:

$$ \int_{\Omega} \langle F'(x)n(x),n(x)\rangle dx $$

where

  • F: $\mathbb{R}^m\rightarrow\mathbb{R}^m$
  • n: $\mathbb{R}^m\rightarrow\mathbb{R}^m$
  • $F^\prime(x) = \begin{bmatrix}\nabla F_1(x)^T\\\vdots\\\nabla F_m(x)^T\end{bmatrix}$
  • $\langle x,y\rangle = x^Ty$

To write it another way, I'd like to rewrite the integral

$$ \int_{\Omega} n(x)^T F'(x)n(x) dx $$

as a surface integral on $\partial \Omega$ where we shed the derivative off of $F$. Please assume $\Omega$ to be as smooth as necessary and other properties on $F$ to get us as close as possible. Likely, $m$ will be 3, but I'd like a result for a general $m$ if possible. Thanks!

$\endgroup$
1
$\begingroup$

This can be done using index notation (and also vector notation). We have \begin{multline} \left<F'(x)n(x),n(x)\right> = \mathbf{n}\cdot(\mathbf{n}\cdot\boldsymbol\nabla\mathbf{F})= n_in_j\partial_iF_j = \partial_i(n_in_jF_j) - \partial_i(n_in_j)F_j \\= \boldsymbol\nabla \cdot [\mathbf{n}(\mathbf{n}\cdot \mathbf{F})] - [\boldsymbol \nabla\cdot(\mathbf{n}\mathbf{n})]\cdot\mathbf{F} \end{multline} Divergence theorem then gives $$ \int_{\Omega}\mathbf{n}\cdot(\mathbf{n}\cdot\boldsymbol\nabla\mathbf{F})d^mV = \int_{\partial\Omega} (\mathbf{n}\cdot\mathbf{F})(\mathbf{n}\cdot d^{m-1}\mathbf{A}) - \int_\Omega[\boldsymbol\nabla\cdot(\mathbf{n}\mathbf{n})]\cdot\mathbf{F}d^mV $$ where $d^m V$ is the $m$-dimensional volume element and $d^{m-1}\mathbf{A}$ is the directed $m-1$-dimensional area element of the $m-1$-dimensional surface $\partial\Omega$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.