V15.

Relation to Physics
The three theorems we have studied: the divergence theorem and Stokes theorem in space, and Greens theorem in the plane (which is really just a special case of Stokes theorem) are widely used in physics and continuum mechanics, in the study of elds, potentials, heat ow, wave motion in liquids, gases, and solids, and thermodynamics, to name some of the uses. Often partial dierential equations which model some physical situation are derived using the vector integral calculus theorems. This section is devoted to a brief account of where you will rst meet the theorems: in electromagnetic theory.

1. Symbolic notation: the del operator

To have a compact notation, wide use is made of the symbolic operator del (some call it nabla): (1) = i+ j+ k x y z

M and the function M (x, y, z ) is understood to be . Then Recall that the product of x x we have (2) grad f = f = f f f i+ j+ k x y z

The divergence is sort of a symbolic scalar product: if F = M i + N j + P k , (3) div F = F = M N P + + x y z

while the curl, as we have noted, as a symbolic cross-product: i curl F = F =

x

j
y

k
z

Notice how this notation reminds you that F is a scalar function, while F is a vector function. We may also speak of the Laplace operator (also called the Laplacian), dened by (5) lap f = 2 f = ( ) f = 2f 2f 2f + + . x2 y 2 z 2

Thus, Laplaces equation may be written: 2 f = 0. (This is for example the equation satised by the potential function for an electrostatic eld, in any region of space where there are no charges; or for a gravitational eld, in a region of space where there are no masses.) 1

V. VECTOR INTEGRAL CALCLUS

In this notation, the divergence theorem and Stokes theorem are respectively (6)
S

F dS =
D

F dV
C

F dr =
S

F dS

Two important relations involving the symbolic operator are: (7) (7 )

curl (grad f ) = 0 f = 0

div curl F = 0 F = 0

The rst we have proved (it was part of the criterion for gradient elds); the second is an easy exercise. Note however how the symbolic notation suggests the answer, since we know that for any vector A, we have A A = 0, AAF = 0 ,

and (7 ) says this is true for the symbolic vector as well.

2. Application to Maxwells equations.

Each of Maxwells equations in electromagnetic theory can be written in two equivalent forms: a dierential form which involves only partial derivatives, and an integrated form involving line, surface, and other multiple integrals. In a sense we have already seen this with our criterion for conservative elds; we assume F is continuously dierentiable in all of 3-space. Then the integrated form of the criterion is on the left, and the dierential form is on the right: F dr = 0
C

for all closed C

curl F = 0

for all x, y, z .

And we know that it is Stokes theorem which provides the bridge between these two equivalent forms of the criterion. The situation with respect to Maxwells equations is similar. We consider here two of them, as typical. Gauss-Coulomb Law. Let E be an electrostatic eld, arising from a distribution in space of positive and negative electric charge. Then the Gauss-Coulomb Law may be written in either of the two forms (8) (8 )
S

E = 4, E dS = 4Q,

= charge density; Q = total net charge inside S .

(dierential form) (integrated form)

These are two equivalent statements of the same physical law. The integrated form is perhaps a little easier to understand, since the left hand side is the ux of E through S , which is a more intuitive idea than div E. On the other hand, quite a lot of technique is required actually to calculate the ux, whereas very little is needed to calculate the divergence. Neither (8) nor (8 ) is mathematics both are empirically established laws of physics. But their equivalence is a purely mathematical statement that can be proved by using the divergence theorem.

V15. RELATION TO PHYSICS

Proof that (8) (8 ). Let D be the interior of the closed surface S . Then E dS =
S D

E dV dV
D

by the divergence theorem; by (8) by denition of and Q.

= 4 = 4Q ,

Proof that (8 ) (8). We reason by contraposition: that is, we show that if (8) is false, then (8 ) must also be false. If (8) is false, this means that we can nd some point P0 : (x0 , y0 , z0 ) where E is dened and such that E = 4 at P0 ; we write this inequality as E 4 = 0, at P0 .

Say the quantity on the left is positive at P0 . Then by continuity, it is also positive in the interior of a small sphere S0 centered at P0 ; call this interior B0 . Then ( E 4) dV > 0 ,
B0

which we write E dV > 4

B0 B0

dV.

The integral on the right gives the total net charge Q0 inside S0 ; applying the divergence theorem to the integral on the left, we get E dS > 4Q0
S0

which shows that (8 ) is also false, since it fails for S0 .

Faradays Law A changing magnetic eld B(x, y, z ; t) produces an electric eld E. The relation between the two elds is given by Faradays law, which can be stated (in suitable units) in two equivalent forms (c is the velocity of light): (9) (9 ) 1 c 1 E dr = c C E = B t d dt dierential form B dS
S

integrated form

V. VECTOR INTEGRAL CALCLUS

As before, it is the integrated form which is more intuitive, though harder to calculate. The line integral on the left is called the electromotive force around the closed loop C ; Faradays law (9 ) relates it to the magnetic ux through any surface S spanning the loop C. A few comments on the two forms. The derivative in (9) is taken by just dierentiating each component of B with respect to the time t. It is a partial derivative, since the components of B are also functions of x, y, z . In (9 ) on the other hand we have an ordinary derivative, since after the integration, the ux is a function of t alone. It is understood in physics that on S the positive direction for ux and the positive direction on C must be compatibly chosen. The magnetic ux through S is the same for all surfaces S spanning the loop C . (This is a consequence of the physical law B = 0.) As a result, one speaks simply of the ux through the loop C , meaning the ux through any surface spanning C , i.e. having C as its boundary. Once again, though (9) and (9 ) both express the same physical law, the equivalence between them is a mathematical statement; to prove it we use Stokes theorem. Proof that (9) (9 ), E dr =
C S

E dS , B dS , t B dS ,
S

by Stokes theorem, by (9)

1 c 1 d = c dt

if B has a continuous derivative and S is smooth, and nite in extent and in area. (This last equality is fairly subtle, and is taken up in theoretical advanced calculus courses.) Proof that (9 ) (9). We show that if (9) is false, then (9 ) is false: If (9) is false, this means that at some point P0 , E = (10) E + 1 B = 0. c t 1 B ; we write this c t

C0 S0 P0 i

This means that at least one of the components of this vector is not 0 at P0 ; say it is the i -component, and its positive. Then by continuity it will remain positive in a small ball around P0 . Inside this little ball, draw a little disc S0 as shown with center at P0 , having normal vector i ; orient its circular boundary C0 compatibly. Since the vector on the left in (10) has a positive i -component on S0 , E +
S0

1 B c t

dS > 0,

which we may write E dS >

S0

1 c

S0

B dS ; t

V15. RELATION TO PHYSICS

applying Stokes theorem to the left-hand side, and interchanging the order of dierentiation and integration on the right (this is valid under the reasonable hypotheses we stated before), we get

E dr >
C

1 d c dt

B dS ,
S

which shows the integrated form (9 ) is false for this little circle and disc, and therefore not true in general.

V. VECTOR INTEGRAL CALCLUS

3. Harmonic functions in space.

A harmonic function in space is by denition a function f (x, y, z ) which satises Laplaces equation 2 f = 0, or written out (see (5)): 2f 2f 2f + + = 0. x2 y 2 z 2 For example, the potential function for an electrostatic eld E is harmonic in any region of space which is free of electrostatic charge. Similarly, the potential function for a gravitational eld F is harmonic in any region where there is no mass. These statements are mathematical consequences of physical laws, and therefore are also physical laws i.e., experimental facts, not mathematical facts. To see why the potential function for E is harmonic, suppose we are in a simply-connected region of space where there is no charge. We then have Gauss-Coulomb law Faradays law E = 0 E = 0; since = 0 in the region

the second equation is valid since the eld arises from a distribution of static electric charges there is no changing magnetic eld. Faradays law shows that E is conservative, so that it has a (mathematical) potential function f (x, y, z ); the physical potential function would be f (x, y, z ). By the Gauss-Coulomb law, noting that E = f , we get E = f = 0, showing that f (x, y, z ) is a harmonic function. Because harmonic functions can represent potential functions, there is great interest in nding harmonic functions in a region D of space. Typically, one prescribes the values that f (x, y, z ) should have on the boundary of D, and then searches analytically (or numerically by computer) for the values of f (x, y, z ) inside D. In this work, the divergence theorem gives an important theoretical tool; some of the Exercises use it to explore the situation a little further. In general, this aspect of the subject properly belongs to the realm of partial dierential equations, i.e., to Dierential Equations and Advanced Calculus courses: see you there, maybe. or 2 f = 0,

Exercises: Section 6H

18.02 Notes and Exercises by A. Mattuck, with the assistance of T.Shifrin and S. LeDuc
c M.I.T. 2010