The Divergence Theorem Here’s a really cool idea relating the flux of a vector field through a surface to the integral of that field throughout the space enclosed by the surface. Start by considering an arbitrary surface S. Divide up the interior into N little cuboids. The flux of F through the surface S equals the sum of the fluxes through the surfaces of each of the N little cuboids. In other words, N S F n dS F n dS i 1 Si If you think about it, this makes sense because anywhere one of the little cuboids touches another little cuboid, the flux through the face of the first contributes some amount to the total flux, but the flux through the face of the second that’s touching it is just the negative of this amount because the unit vector of the first face is just the negative of the unit vector of the face on the second one. Think of two little boxes touching on one face: the normal vectors to the common face are negatives of one another, so whatever one of them contributes to the flux the other one cancels out. This will be true for all faces of all the little cuboids except for ones that touch the surface of S. These have no “corresponding face” on a neighboring cuboid to cancel out the flux through the exterior face, so that little bit of flux contributes to the total. Since we want to see if there’s a relationship between the flux through the surface and the vector field in the interior, we want in some way to do something with our surface integral so that it represents a volume. The simplest way to do that is to introduce a volume element, Vi that corresponds to each one of our little cuboids. But what the Lord giveth, the Lord taketh away, so if we multiply part of the integral by the volume element, we have to divide another part by the same thing. So, we do: N 1 F n dS F n dS Vi S Si i 1 Vi However, by the discussion we just went through above, you should be able to recognize that if we let N approach infinity, the quantity in square brackets simply becomes div F through the little piece of volume Vi . Furthermore, in the limit of N’s approaching infinity, the sum of all the little volume elements Vi is simply the volume of the region V, which can be found as simply the triple integral V dV Of course, we don’t have just the volume; we have a scalar field permeating that volume and hence being integrated over that volume, so we have on the right instead of just the simple volume integral above, the triple integral V F dV Putting all this together, we’ve now found a relationship between the flux of a vector function through a surface and the integral of the divergence of that field over the volume enclosed by the surface: S F n dS V F dV This result, important enough to have its own name—The Divergence Theorem—is either reasonable or amazing, depending on how you look at it. If you think just about the math, it’s amazing that a vector field integrated over a surface can be equal to a kind of derivative of the field integrated over a volume. On the other hand, if you think about it in physics terms, it makes more sense: the flux through a surface (i.e., the amount of “stuff” streaming out through the surface of an object) is probably related to the divergence of the field producing the flux in the interior (i.e., the amount of “stuff” being generated in the volume inside the surface). One of the reasons we’ve been defining quantities like divergence and flux in physics terms and then deriving the math is so that the results would seem more intuitive. If they haven’t, you can be like all the math majors and memorize these formulas and think, “Wow! What a coincidence that these formulas apply to the real world!” Another reason for doing things this way is so that if you follow through the same procedure, you can derive the results in other coordinate systems should you ever need to do so. A simple illustration For a simple illustration of the divergence theorem, take the field F = {x, y, z} over the region of space represented by the surface that’s composed of the hemisphere of radius 1 and z > 0 centered at the origin and the unit circle in the xy-plane. Together, those two surfaces enclose a volume of space. On the hemisphere, the normal unit vector simply points radially outward and can be represented in Cartesian coordinates by {x, y, z}. Hence F n = {x,y,z}{x,y,z}= x2 + y2 + z2 , which in turn equals 1 since it’s a unit hemisphere. Thus, F n = 1. Hence, the flux through the hemisphere is simply F n dS dS 2 hemisphere because the surface area of the hemisphere is 2. On the bottom of the hemisphere (a.k.a. the unit circle), n = -k and hence F n = - z. Hence, circle F n dS z dydx 0 because z = 0 everywhere in the xy-plane. Thus, there is no contribution to the flux from the bottom of the hemisphere. Hence, the total flux is 2. Now, the divergence of F is F Fx x F y y Fz z x y z 111 3 x y z Hence, hemisphere F dV 3 hemisphere dV 3 2 2 3 since the volume of the hemisphere is half that of the unit sphere, or 2/3. Hence, we’ve just illustrated that for the given vector field F over the given hemisphere, the divergence theorem holds. You’ll get to use it with harder examples. As you should have noticed by now, it’s much easier to teach math than to take it—the teachers take all the easy examples for themselves and give the hard ones as exercises for students. No doubt, that’s why Kenny enjoyed teaching the ninth-graders so much. There are really only two things that make teaching math work (or three, if you count the huge amount of work making up these labs): 1) sometimes students insist on asking questions and depending on your level of knowledge it’s either very hard to answer them (if you know nothing) or very hard to answer them in such a way that the students are led to the answer rather than just having it delivered to them; 2) it’s often very timeconsuming to make problems that are either realistic or unify several disparate topics you’ve studied (or both) without being so hard that they can’t be done by students just starting their study of the material (or can’t be done in a reasonable time frame). I find high-school students usually complain if you give them take-home tests that require more than three or four hours to finish….