Are you sure?
This action might not be possible to undo. Are you sure you want to continue?
theorems
Flux and divergence:
proof of the divergence
theorem, à lá Purcell.
Circulation and curl: proof
of Stokes’ theorem, also
following Purcell.
See Purcell, chapter 2, for more
information.
1
dl
1
S
1
C
2
dl
2
S
2
C
B
1 2
1 2
C
C C
d
d d
I =
= +
l
l l
v l
v l v l
v
v v
The divergence theorem
Consider a vector function v
that exists everywhere in space,
and a specific volume V,
bounded by surface S. The flux
of v through S is given by
where the infinitesimal area
element vectors on S, da, point
outward and are perpendicular
to S.
Now consider dividing V in
two:
V
S
da
S
d u =
l
v a
v
The divergence theorem (continued)
The two new volumes have more
total surface area on their
boundaries than they used to, by
twice the area of the dividing
surface D. But because the area
elements point outward from each
volume, the flux through D from
V
1
is equal and opposite to that
from V
2
, so the total flux through
all the surfaces is the same as it was
before:
D
1
da
2
da
1
V
2
V
1
S
2
S
1 2
1 2
S S S
d d d u = = +
l l l
v a v a v a
v v v
The divergence theorem (continued)
In fact, if we were to subdivide V into
N (>>1) cells, the flux through all the
dividing surfaces would cancel out in
the sum, though the total surface area
would be much larger than originally:
We now claim that
i
V
1
i
N
i
i
S S
d d
=
u = =
¯
l l
v a v a
v v
bounds .
i i
S V
( )
0
lim .
i
i
i
S
N
V
i
d
V
÷·
÷
 

=


\ .
l
v a
v
v
³
Note:
We’re about to examine the variation of a function v in a very
small volume near the point x,y,z. Recall that if the volume is
small enough that v changes very little within it, then the
value of v at x+δx,y+δy,z+δz would be
One can see this from the definition of the derivative, or by
expanding v in a Taylor series about the point x,y,z and
neglecting all the terms of higher order than those with first
derivatives – OK, because they have successively higher
powers of the intervals δx,δy,δz and are thus much smaller
than the zeroth and firstderivative terms.
( )
( )
, ,
, ,
x x y y z z
x y z x y z
x y z
o o o
o o o
+ + + ~
c c c
+ + +
c c c
v
v v v
v
The divergence theorem (continued)
To justify the claim, consider a small
box within V, lying with one corner
at r
i
= (x,y,z) with volume ΔxΔyΔz, as
one of the V
i
. Its contribution to the
flux is
Look at the top and bottom sides first. Their area vectors are
equal and opposite and lie along z, and the top lies Δz above
the bottom. If the box is indeed small, the value of the z
component of v in the center of the top face is approximately
sides
i
i i
S
d = A
¯
l
v a v a
v
( )
, top
, ,
2 2
z z z
z z
y v v v x
v v x y z z
x y z
A c c c A
~ + + + A
c c c
x
y
z
z A
top
ˆ
x y A = A A a z
bot.
ˆ
x y A = ÷ A A a z
i
r
The divergence theorem (continued)
The value of v
z
in the center of the bottom
face is, on the other hand,
Take the value at the center of each face to
serve as the average of v
z
on the face. Then
the flux contributed by the top and bottom is
x
y
z
z A
top
ˆ
x y A = A A a z
bot.
ˆ
x y A = ÷ A A a z
( )
, bot.
, ,
2 2
z z
z z
y v v x
v v x y z
x y
A c c A
~ + +
c c
( ) ( )
( ) ( )
, top top , bot. bot.
, ,
2 2
, ,
2 2
z z z
z z z
z z
z
z
y v v v x
v a v a v x y z z x y
x y z
y v v x
v x y z x y
x y
v
x y z
z
 
A c c c A
A + A ~ + + + A A A

c c c
\ .
 
A c c A
+ + + ÷A A

c c
\ .
c
= A A A
c
i
r
The divergence theorem (continued)
Similarly, the front and back, and right and left, pairs contribute
( ) ( )
( ) ( )
, fr. fr. , ba. ba.
, ,
2 2
, ,
2 2
x x x
x x x
x x
x
x
y v v v z
v a v a v x y z x y z
z y x
y v v z
v x y z y z
z y
v
x y z
x
 
A c c c A
A + A ~ + + + A A A

c c c
\ .
 
A c c A
+ + + ÷A A

c c
\ .
c
= A A A
c
( ) ( )
( ) ( )
, R R , L L
, ,
2 2
, ,
2 2
y y y
y y y
y y
y
y
v v v
x z
v a v a v x y z y x z
x z y
v v
x z
v x y z x z
x z
v
x y z
y
c c c
 
A A
A + A ~ + + + A A A


c c c
\ .
c c
 
A A
+ + + ÷A A


c c
\ .
c
= A A A
c
The divergence theorem (continued)
Thus
and
Back to the flux:
( )
sides
,
i
y
x z
i i
S
i i
v
v v
d x y z
x y z
V
c
 
c c
~ A = + + A A A


c c c
\ .
=
¯
l
v a v a
v r
v
³
0
lim , just as we claimed.
i
i
i
S
V
i
d
V
÷
 

=


\ .
l
v a
v
v
³
( )
1 1 1
, or
i
i
N N N
i
S
i i i
i
i i i
S S V
i
S V
d
d d V V d
V
d d
t
t
= = =
 

u = = ~ = ÷


\ .
=
l
¯ ¯ ¯
l l l
l l
v a
v a v a v
v
v a v
v
v v
v
³
³
³ Divergence theorem
The divergence theorem (continued)
The meaning of the divergence theorem is illustrated by
applying it to fluid flow, with v as the fluid velocity:
The divergence theorem will eventually lead us to Gauss’
law.
Total flow through Difference of numbers of faucets
bounding surface and drains within volume
S V
d dt =
¹ ¦
=
` ´
) ¹
l l
v a v
v
³
Stokes’ theorem
Consider a vector function v that
exists everywhere in space, and a
specific surface S (not necessarily
planar), bounded by curve C. The
circulation of v around C is given by
where the infinitesimal line element
vectors on C, dl, point parallel to C in
the righthandrule sense (as in
) relative to the area
element vectors on S.
C
d I =
l
v l
v
S
C
da
dl
(points out
of page)
a b
d d d = × a l l
Stokes’ theorem (continued)
Now divide the area in two. The
two new areas have more total
circumference than they used to,
by twice the length of the dividing
line D. But because the line
elements follow each curve
counterclockwise, the line integral
along B around S
1
is equal and
opposite to that from S
2
, so the
sum of the two line integrals is the
same as the original:
1
dl
1
S
1
C
2
dl
2
S
2
C
B
1 2
1 2
C C C
d d d I = = +
l l l
v l v l v l
v v v
Stokes’ theorem (continued)
In fact, if we were to subdivide S into N
(>>1) cells, the line integrals along all
the dividing lines would cancel out in
the sum, though the sum of
circumferences would be much larger
than originally:
I bet you can guess what we’ll claim
now:
1
i
N
i
i
C C
d d
=
I = =
¯
l l
v l v l
v v
( )
( )
0
ˆ
lim .
i
i
N
i
C
a
i
d
a
÷·
÷
 

= ×


\ .
l
v l
n v
v
³
i
S
bounds ; area of
ˆ
is
i i i
i i i
C S S
a = a n
Stokes’ theorem (continued)
To justify the claim, consider a small
box within S, lying with one corner
at r
i
= (x,y,z) with area , as
one of the S
i
. Its contribution to the
total circulation is
Look at the top and bottom sides first. Their line element
vectors are equal and opposite and lie along x, and the top
lies Δy above the bottom. If the box is indeed small, the value
of the x component of v in the center of the top side is
approximately
x
y
y A
T
ˆ
x A = ÷ A l x
B
ˆ
x A = A l x
sides
i
i i
C
d = A
¯
l
v l v l
v
( )
, T
, ,
2
x x
x x
v v x
v v x y z y
x y
c c A
~ + + A
c c
ˆ
i
x y = A A a z
Stokes’ theorem (continued)
The value of v
x
in the center of the bottom
side is, on the other hand,
Take the value at the center of each side to
serve as the average of v
x
on the side. Then
the contribution to the circulation by the top
and bottom is
( )
, B
, ,
2
x
x x
v x
v v x y z
x
c A
~ +
c
( ) ( )
( ) ( )
, T T , B B
, ,
2
, ,
2
x x
x x x
x x
x
v v x
v l v l v x y z y x
x y
v v x
v x y z x x y
x y
 
c c A
A + A ~ + + A ÷A

c c
\ .
c c A
 
+ + A = ÷ A A

c c
\ .
x
y
y A
T
ˆ
x A = ÷ A l x
B
ˆ
x A = A l x
Stokes’ theorem (continued)
Similarly, the left and right sides give
So, noting that , we have for all four sides together
Rest assured that we would get the same result with the area vector in any
direction; thus our claim is justified.
x
y
x A
L
ˆ
y
A =
÷ A
l
y
R
ˆ
y A = A l y
( ) ( )
( ) ( )
, R R , L L
, ,
2
, ,
2
y y
y y
y
y y
y
v l v l
v v
y
v x y z x y
y x
v v
y
v x y z y x y
y x
A + A ~
c c
 
A
+ + A A


c c
\ .
c c
 
A
+ + ÷A = A A


c c
\ .
( )
sides
i
y
x
i i i
C
v
v
d x y
x y
c
 
c
= A ~ ÷ A A = ×


c c
\ .
¯
l
v l v l v a
v
³
ˆ
i
x y = A A a z
Stokes’ theorem (continued)
Thus
or
( ) ( )
1 1
,
i
N N
i i
N
i i
C C S
d d d
÷·
= =
I = = = × ÷÷÷÷÷ ×
¯ ¯
l l l
v l v l v a v a
v v
³ ³
( )
.
C S
d d = ×
l l
v l v a
v
³
Stokes’ theorem
This action might not be possible to undo. Are you sure you want to continue?
Use one of your book credits to continue reading from where you left off, or restart the preview.