Normally I read the Divergence Theorem written as (\oiint doesn't exist here):

\begin{align} \oint_{\partial \Omega} \vec{F} \cdot \hat{n} \; dS = \iiint_{\Omega} \nabla \cdot \vec{F} \; dV \end{align}

Where $\partial \Omega$ is the boundary of region $\Omega$ and $dS$ are infinitesimal surface of $\partial \Omega$ and $dV$ infinitesimal volume of $\Omega$.

But reading a paper CASTILLO and GRONE, A matrix analysis approach to higher order approximations for divergence and gradients satisfying global conservation law, it states the Divergence Theorem as:

\begin{align} \int_{\partial \Omega} f \, \vec{v} \cdot \hat{n} \; dS = \int_{\Omega} \nabla \cdot \vec{v} \, f \; dV + \int_{\Omega} \vec{v} \, \nabla f \; dV \end{align}

Are those equivalent? if so how can I correlate the RHS of both equations?

I really think that he is abusing notation using a single integral instead a triple one and not directly specifying the loop integral in left hand side, but even with that I cannot correlate, maybe I'm missing some definition here.


The only error here is the second integral on the RHS should have the dot product as the integrand:

$$ \int_{\Omega} \vec{v} \cdot \, \nabla f \; dV, $$

since $\nabla \cdot (f \,\vec{v}) = f \, \nabla \cdot \vec{v} + \nabla f \cdot \vec{v}$.

Aside from appearance, given an integrable function $f:\Omega \subset \mathbb{R}^3 \to \mathbb{R}$ there is no difference between

$$\int_\Omega f, \quad \int_\Omega f \, dV, \quad\text{and } \quad \int\int\int_\Omega f \, dV$$

There is difference between a multiple Riemann integral and an iterated Riemann integral, for example,

$$\int_{[0,1]^3} f \quad \text{and} \quad\int_0^1 \left(\int_0^1 \left(\int_0^1 f(x,y,z) \, dz \right)\, dy \right) \, dx$$

where the object on the right may exist but the object on the left may not. However, if $|f|$ is integrable on $[0,1]^3$, then the two must be equal, as one consequence of Fubini's theorem.

  • $\begingroup$ I double checked the paper, the $\cdot$ really doesn't exist. Maybe a mistype. $\endgroup$ – Lin Jun 29 '18 at 3:04
  • $\begingroup$ Definitely a mistype because $\vec{v}\, \nabla f$ is a vector. $\endgroup$ – RRL Jun 29 '18 at 3:06
  • $\begingroup$ Just look up the vector identity for $\nabla \cdot (f \, \vec{v})$ in Wikipedia if you want confirmation. $\endgroup$ – RRL Jun 29 '18 at 3:08
  • $\begingroup$ Is $f \vec{v}$ equivalent to $\vec{F}$? since $f$ is a scalar valued function and $\vec{v}$ a vector. (of course I could write $\vec{f}$ but I'm also using the capital here to show up the vector function). $\endgroup$ – Lin Jun 29 '18 at 3:09
  • $\begingroup$ Yes, they are equivalent. You can regard $\vec{v}$ as carrying all the information about the direction of the vector field at every point, while $f$ just rescales $\vec{v}$ so that it matches the magnitude of $\vec{F}$. This is acomplished by setting $$f := \sqrt{\vec{F}\cdot\vec{F}}$$ and $$\vec{v} := \frac{1}{f}\vec{F}$$. Then $\vec{F} = f\vec{v}$ (quite obviously). $\endgroup$ – Jackozee Hakkiuz Jun 29 '18 at 3:42

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.