TOPIC: WHAT THE DIVERGENCE AND CURL MEAN

Introduction

Consider a vector field F(x,y) = (P(x,y),Q(x,y)). Denoting partial derivatives with subscripts as usual, the divergence of F(x,y), div(F(x,y)), is given by

div(F(x,y)) = Px(x,y) + Qy(x,y)

and the curl of F(x,y), curl(F(x,y)), is given by

div(F(x,y)) = Qx(x,y) - Py(x,y)

The questions we will answer in this setion of the notes are:

  • What do divergence and curl have to do with work and flux -- two things that we are often calculuting for a given vector field and path?
  • What is the geometric meaning of the divergence and curl -- or, if I'm looking at a graph of a vector field, how do I recognize positive divergence or negative curl, for example?
  • If two vector fields have the same divergecne and curl, what else do I need to know about them to know that they are the same; i.e., have the same P(x,y) and Q(x,y) as well?
  • Suppose I'm given two functions f1(x,y) and f2(x,y). Is there a vector field F(x,y) with div(F(x,y)) = f1(x,y) and curl(F(x,y)) = f2(x,y)? If so, Is there a formula for finding P(x,y) and Q(x,y) given the divergence and curl?
  • All of these questions are closely related. We will adress them in four subsections below. The last two are rather deep; they adress the uniqueness and existence of the following question:

    To what extent do the divergence and curl of a vector field determine that vector field?

    This question isn't easy, but amazingly, it can be answered just using the vector calculus that we know.

    To see why it is interesting, recall that Maxwell's equations for electrostatics specify the electric field E(x,y) and the magnetic field B(x,y) exactly by specifying their divergence and curl. For example, in a static situation, the curl of the electric field is zero, and the divergence of the electric field is a multiple of the electric charge density c(x,y):

    div(E(x,y)) = c(x,y) and curl(E(x,y)) = 0

    Likewise, the electrostatic equations for hte magentic field are

    div(B(x,y)) = 0 and curl(B(x,y)) = j(x,y)

    where j(x,y) is a current density.

    Of course, one usually sees these in three dimensions. So here consider a situation in which neiter the charge density nor the current density depends on z. Then the third dimension frops out, as we have it here: planar electrstatics. Moreover, once we have understood the planar case thoroughly, it is very easy to move on to the full three dimensional case. So we will stay in this section of the notes.

    Looking at the elctrostatic eqautions, it looks as if they are only going to specify the electric and magnetic fields the divergence and curl uniquely specify a vector field.

    But it can't be that the divergence and curl are enough to specify a vector field F(x,y) = (P(x,y),Q(x,y)), because if we add any constants to P or Q, these constants will drop out when we take the partial derivatives in computing the divergence and curl. Thus

    F(x,y) = (xy, x+y) and G(x,y) = (xy+1,x+y-3)

    have the same divergence and the same curl.

    Why isn't this a problem for Mr. Maxwell's equations? Because of "boundry conditions"!

    For instance, if the total amount of charge is finite, then the electric field E(x,y) must tend to zero as (x,y) tends to infinity. This "boundary condition" picks out the physical solution of Maxwell's equations from among all the others.

    If you take a solution of Maxwell's eqautions that do satisfy this boundary condition, and you add a constant vector field to the solution, the equations will still be satisfied, but not the boundary condition -- the new limiting value at infinity is clearly the constant vector field that you just added on.

    We shall see below that among all solutions to Maxwell's equations, as above, there is at most one that satisfies the boundary condition. In fact, we will prove the more general fact: any vector field whose divergecne and curl both vanish tthroughout the plane is either constant or unbounded.

    Here are some vector fields that have both zero curl and zero divergence, and yet are not constant:

    F(x,y) = (x,-y)

    G(x,y) = (excos(y),-exsin(y))

    As indicated above, they are unbounded -- their magnitudes take on arbitrarily large values.

    But now we are getting a bit ahead of ourselves. That's O.K.; I just wanted to make clear that the considerations we are undertaking here are motivated by applications. It was partly because Mr. Maxwell understood the meaning of the divergence and curl, and the extent that they determine the vector field they came from, better than his predecessors that he was able to formulate his equations in the first place. Now, back to the beginning.

    Links to the Subsections

  • 1. The Divergence and Curl as Densities
  • 2. Visualizing Divergecne and Curl
  • 3. Equal Divergecne and Curl
  • 4. Finding F from its Divergecne and Curl