Question: Let $\phi$ be a smooth scalar field defined on a region $R \subseteq \mathbb{R}^{3}$ with a smooth boundary $\partial R$. Then show that the following identity holds:

$$\iiint_R \mathbf{r} \times \nabla\phi \,dV = \iint_{\partial R} \mathbf{r} \times \phi \mathbf{n} \,dS$$

In my course up until this point the divergence theorem has been covered, and I can see the similarities between the integrals above and the divergence theorem identity but am unsure how to go from $\nabla . F$ in the divergence theorem to $\mathbf{r} \times \nabla\phi$ above.

How should one begin to attempt showing an identity like this given they have use of the divergence theorem?


In Cartesian coordinates we have $\mathbf{r} = x \mathbf{i} + y \mathbf{j} + z \mathbf{k},$ and, without showing every term,

$$\iiint_R\mathbf{r} \times \nabla \phi \, dV \\ = \left(\iiint_R y \frac{ \partial\phi}{\partial z} \, dV - \iiint_R z \frac{ \partial\phi}{\partial y} \, dV \right)\mathbf{i} + \cdot \cdot \cdot \\ = \left(\iiint_R \frac{ \partial }{\partial z}(y \phi) \, dV - \iiint_R \frac{ \partial}{\partial y}(z \phi) \, dV \right)\mathbf{i} + \cdot \cdot \cdot \\ = \left(\iiint_R \nabla \cdot (y \phi \mathbf{k}) \, dV - \iiint_R \nabla \cdot (z \phi \mathbf{j}) \, dV \right)\mathbf{i} + \cdot \cdot \cdot \\ = \left(\iint_{\partial R} (y \phi \mathbf{k}) \cdot \mathbf{n} \, dS - \iint_{\partial R} (z \phi \mathbf{j}) \cdot \mathbf{n} \, dS \right)\mathbf{i} + \cdot \cdot \cdot \\ = \left(\iint_{\partial R} y \phi n_z \, dS - \iint_{\partial R} z \phi n_y \, dS \right)\mathbf{i} + \cdot \cdot \cdot \\ = \iint_{\partial R} \mathbf{r} \times \phi \mathbf{n} \, dS. $$

The divergence theorem familiar to you was applied in going from the third to fourth step.


A proof with much more compact notation uses the component form of the divergence theorem

$$\int_R \partial_i f \, dV = \int_{\partial R} f \,n_i \, dS,$$

and the expression for curl and cross-product using the Levi-Cevita symbol

$$(\mathbf{r} \times \nabla \phi)_i = \varepsilon_{ijk} \,r_j \,\partial_k \phi, \\ (\mathbf{r} \times \phi \mathbf{n})_i = \varepsilon_{ijk}\,r_j\phi \,n_k.$$


$$\begin{align} \int_R (\mathbf{r} \times \nabla \phi)_i \, dV &= \int_{\partial R} \varepsilon_{ijk} \,r_j \,\partial_k \phi \, dS \\ &= \varepsilon_{ijk}\int_{\partial R} \ \,\partial_k (r_j\phi) \, dS \\ &= \varepsilon_{ijk}\int_{\partial R} \, r_j\,\phi \,n_k \, dS \\ &= \int_{\partial R} \varepsilon_{ijk} \, r_j\,\phi \,n_k \, dS \\ &= \int_{\partial R} (\mathbf{r} \times \phi \mathbf{n})_i \, dS \end{align}$$


While the solution above is elegant and correct, I believe the following approach better demonstrates how the identity was obtained in the first place.

Consider setting $\mathbf{F} = \phi\mathbf{c} \times \mathbf{r}$ in the Divergence Theorem as you quoted above, where $\mathbf{c}$ is an arbitrary yet constant vector field. By the divergence product rule for a scalar field and a vector field:

$ \nabla \cdot (\phi\mathbf{c} \times \mathbf{r}) = \nabla \phi \cdot (\mathbf{c} \times \mathbf{r}) + \phi \nabla \cdot (\mathbf{c} \times \mathbf{r}) = \mathbf{c} \cdot (\mathbf{r} \times \nabla \phi)$, by the cyclic properties of the triple vector product and the fact that $ \nabla \cdot (\mathbf{c} \times \mathbf{r}) = \mathbf{0}$ (check this directly).

Now note from the Divergence Theorem for the vector field $\mathbf{F}$:

$ \iiint_R \mathbf{c} \cdot (\mathbf{r} \times \nabla \phi) \,dV = \iint_{\partial R} (\phi\mathbf{c} \times \mathbf{r}) \cdot \mathbf{n} \,dS $

$ \implies \mathbf{c} \cdot \iiint_R (\mathbf{r} \times \nabla \phi) \,dV = \mathbf{c} \cdot \iint_{\partial R} \mathbf{r} \times \phi \mathbf{n}\,dS $

where the right hand side of the last expression follows again from the cyclic properties of the triple product. Recalling $\mathbf{c}$ is arbitrary, we're done.


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.