Prove $\displaystyle \int_{C}fdr=\int_{S}dS\times\nabla f$. where $C=\partial S$ and the usual relationship between orientations hold.

Apply Stokes's theorem to $F=af$ where $a$ is an arbitrary constant vector.

From this identity and because $\nabla \times \mathbf{a = 0 }, $ thus $ \nabla\times F=0 + \nabla f \times a$.

Thus $(\nabla\times F)\cdot d\mathbf{ S }= (\nabla f \times a) \cdot d\mathbf{ S } = (d\mathbf{ S }\times \nabla f) \cdot a $, thanks to the answer below.

Then Stokes's theorem for arbitrary a implies $ \int_{C}f \mathbf{ a } \; d\mathbf{ r } = \iint_S (d\mathbf{ S }\times \nabla f) \cdot a $.

My concern: How do I proceed from here? Please explain steps in detail?


Apply the Kelvin-Stokes theorem to the vector field $fa$:

$$\oint fa \cdot dr = \int (\nabla \times [af]) \cdot dS$$

Use vector calculus identities to write

$$\nabla \times [af] = f \nabla \times a + \nabla f \times a$$

Since $\nabla \times a = 0$, we get

$$\oint f a \cdot dr = \int (\nabla f \times a) \cdot dS$$

The triple product can be cyclically permuted, yielding

$$\oint fa \cdot dr = \int (dS \times \nabla f) \cdot a$$

Since $a$ is constant, it can be moved out of both integrals, the way you would move a constant scalar out of an integral:

$$a \cdot \oint f \, dr = a \cdot \int dS \times \nabla f$$

$a$ was chosen arbitrarily; this is true for all $a$ and thus we can "cancel" $a$. If one must think of it more rigorously, look at this expression above as a linear function of $a$. Take a gradient of that function with respect to $a$. The resulting gradients are equal on both sides. The result is

$$\oint f \, dr = \int dS \times \nabla f$$


Remember that $$ (v\times w)\cdot z = \det(v,w,z) = \det (z,v,w) = (z\times v)\cdot w $$ and apply this to $v= \nabla f$, $w=a$, $z=ds$

  • $\begingroup$ Thanks, but how to proceed after? $\endgroup$ – Greek - Area 51 Proposal May 30 '14 at 14:40
  • $\begingroup$ @LePressentiment what do you mean? true for all a..remove a $\endgroup$ – username Jun 3 '14 at 22:46

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.