2
$\begingroup$

This result is very useful in continuum mechanics involving discontinuities, for example, multiphase flows. It is basically a variant of the Stokes theorem - converting area integral into a line integral.

Consider a closed two-dimensional surface $\mathcal{A}$ with boundary $\partial\mathcal{A}$. The normal to $\partial\mathcal{A}$ is $\widehat{n}$. Prove that

$-\int_\mathcal{A}\nabla'[\delta(x-x')\delta(y-y')]\,da'= -\int_{\partial\mathcal{A}}\delta(x-x')\delta(y-y')\widehat{n}\,ds$

where $\delta$ is the Dirac delta function. Proof for a general tensor would be great, but a simplified, rigorous proof (maybe for symmetric tensors, or even just the Dirac delta case) would work as well.

  • 0
    Maybe [a relevant question with some good answers](http://math.stackexchange.com/a/394818/301977)2017-02-14

0 Answers 0