5

I am trying to evaluate a surface integral, but instead of using a surface in $\mathbb{R}^3$, using a surface in $\mathbb{R}^4$.

That is to say,

$\oint_S f(x,y,z,w)\,dS$, where S is given by some $r(u,v,t) = \left( x(u,v,t) , y(u,v,t) , z(u,v,t) , w(u,v,t)\right)$

So like a line integral has a $|r'(t)|$, a surface integral has a factor of $|r_u \times r_v|$, I read up on a generalization of this using the square root of a Gramian matrix, which I had never heard of before researching it now, but I don't know how to calculate it exactly for a parametric function from $\mathbb{R}^3 \to \mathbb{R}^4$, like we have here for $r(u,v,t)$.

Can someone help me with this evaluation? Does it involve integrating differential forms and manifolds? I know a little bit about differential geometry, but not much.

How do I evaluate these integrals, and what is the $\mathbb{R}^3 \to \mathbb{R}^4$ analog of $|r_u \times r_v|$ ?

  • 1
    If you decide to brave learning about differential forms and integrating on manifolds, you may find my lectures on YouTube of interest (linked in my profile). Lots of examples, including one day of integrating over 3-dimensional manifolds and applying Stokes's Theorem. – Ted Shifrin Aug 21 '20 at 01:29

1 Answers1

1

If you want to learn more about the general setting, take a look at this previous answer of mine on Integrating using surface and volume elements. The $|r'(t)|$ and $|r_u \times r_v|$ you mention for line and surface integrals (in $\Bbb{R}^3$) are simply the square root of the determinant Gramian matrix (I leave it to you to verify this).

In your particular case, since $S$ sits inside of some Euclidean space, we can give it the induced Riemannian metric (i.e we can take dot/inner products of vectors which are tangent to the surface $S$). So, here's what we do: first we're going to construct a $3\times 3$ matrix-valued function $G$ as follows: \begin{align} G &= \begin{pmatrix} \left\langle \frac{\partial r}{\partial u}, \frac{\partial r}{\partial u}\right\rangle & \left\langle \frac{\partial r}{\partial u}, \frac{\partial r}{\partial v} \right\rangle & \left\langle \frac{\partial r}{\partial u}, \frac{\partial r}{\partial t}\right\rangle \\ \left\langle \frac{\partial r}{\partial v}, \frac{\partial r}{\partial u} \right\rangle & \left\langle \frac{\partial r}{\partial v}, \frac{\partial r}{\partial v} \right\rangle & \left\langle \frac{\partial r}{\partial v}, \frac{\partial r}{\partial t} \right\rangle\\ \left\langle \frac{\partial r}{\partial t}, \frac{\partial r}{\partial u} \right\rangle & \left\langle \frac{\partial r}{\partial t}, \frac{\partial r}{\partial v} \right\rangle & \left\langle \frac{\partial r}{\partial t}, \frac{\partial r}{\partial t} \right\rangle \end{pmatrix} \end{align} Note that this is a matrix-valued function which means for each $(u,v,t)$, $G(u,v,t)$ is a $3\times 3$-symmetric matrix of numbers obtained by evaluating all the above partial derivatives at the point $(u,v,t)$.

Since the inner product is symmetric: $\langle v,w\rangle = \langle w,v\rangle$ (and in the case of this Euclidean inner product it is just $\sum_i v^iw^i$), it follows that $G$ is a symmetric matrix, so if you have to actually compute a specific example, you only have to compute the upper triangular portion. As a very explicit example, the $(1,3)$ entry of this matrix is \begin{align} \left\langle \frac{\partial r}{\partial u}, \frac{\partial r}{\partial t} \right\rangle &= \dfrac{\partial x}{\partial u}\dfrac{\partial x}{\partial t} + \dfrac{\partial y}{\partial u}\dfrac{\partial y}{\partial t} + \dfrac{\partial z}{\partial u}\dfrac{\partial z}{\partial t} + \dfrac{\partial w}{\partial u}\dfrac{\partial w}{\partial t}. \end{align} Now, suppose the parametrization is $r:A\subset \Bbb{R}^3\to r[A] = S\subset\Bbb{R}^4$. Then, \begin{align} \int_S f \, dS &= \int_A f\circ r \cdot \sqrt{\det G} \\ &\equiv \int_A f(r(u,v,t)) \cdot \sqrt{\det[G(u,v,t)]}\, du\,dv\,dt. \end{align} (where $\equiv$ means "same thing in different notation"). Now, this triple integral over $A\subset \Bbb{R}^3$ can be calculated for example using Fubini's theorem.

peek-a-boo
  • 55,725
  • 2
  • 45
  • 89
  • The specific example I'm using is the outside of the 4-dimensional hypersphere, so: r(u,v,t) = (sintsinvcosu,sintsinvsinu,sintcosv,cosu). I understand how to compute the Gramian matrix, but I'm not too thrilled to compute ALL these partial derivatives and then have them all cancel out with some trig identities. Do I need to calculate all these inner products for the gramian, and then take the determinant, or is there a shorter way to calculator its determinant?

    Thanks

    –  Aug 21 '20 at 01:36
  • @Cjw123 off the top of my head I can't think of a quicker way (though I just did the computation, and it took only about 4-5 minutes) – peek-a-boo Aug 21 '20 at 01:59
  • Hey! I got for the gramian that entries (1,2) and (1,3) are zero, so the determinant is only important for the (1,1) section when calculating the determinant (not sure the name for it.) I got 0 for the (2,3) and (3,2) entries, so the overall determinant being the (1,1) x (2,2) x (3,3) multiplied together. Are you able to verify this for this example, and check if anything comes out of that? I did it and am not getting anything simplified. Any help with this? –  Aug 21 '20 at 02:23
  • @Cjw123 yes, that's right. It turns out to be a diagonal matrix (just reiterating: you don't have to calculate every entry; only the upper triangular portion, because the matrix is always symmetric). And since it is diagonal, the determinant is just the product of diagonal entries. – peek-a-boo Aug 21 '20 at 02:45
  • Thank you. I didn't read that part. What did you get as the end result for the sqrt(detG)? –  Aug 21 '20 at 02:53
  • @Cjw123 I found $\det G = (\sin^2u \sin^2t)(\sin^2t) (\cos^2t)$, so $\sqrt{\det G} = |\sin u \sin^2t \cos t|$ – peek-a-boo Aug 21 '20 at 12:17