Gauss Divergence Theorem.

The Gauss Divergence theorem relates flux integrals and triple integrals.
Recall that when we evaluate a double integral over a surface we compute the
integral over the *skin* of the object enclosed by the surface (the boundary). In
triple integrals we integrate inside of the object:
Theorem 1 Gauss divergence theorem: Let T be a closed bounded region
in the space whose boundary T is a piecewise smooth orientable surface. Let
F~ be a continuous vector field whose partial derivatives are continuous in some
domain containing T . Then:
ZZZ

div F~ dV =

ZZ

F~ ~n dA

Example 1 Verify the divergence theorem when F~ (x, y, z) = (7x, 0, z) and T

is the region enclosed by the sphere of radius 2.
1

In this case:
T = {(x, y, z) : x2 + y 2 + z 2 4}
and
T = {(x, y, z) : x2 + y 2 + z 2 = 4}
We first compute the triple integral over T using a change of variable to spherical
coordinates:
(x, y, z) = (r cos cos , r sin cos , r sin ), r [0, 2], [0, 2], [
and the Jacobian of change

 x

 r

J =  x

 x


, ]
2 2

of variables is:

y z 
r r 
y z  = (x, y, z) = r2 cos .
 (r, , )
y z 


Then:
RRR
T

div F~ dV

=
=
=
=
=

RRR (7x) 0 (z)

+
+
dx dy dz
T
x
y
z
RRR
R 2 R 2 R 2 2
r cos dr d d
6
dxdydz = 6 0
0
T
2
h
i
3 r=2
R 2 R 2
6 0
cos r3
dd
2
r=0
R 2 R 2
16 0 cos d d
2

R 2
2
16 0 [sin ]
d = 64 (4)
2

On the other hand, a parametric representation for T , the sphere of radius 2,

could be:
~r(u, v) = (2 cos u cos v, 2 sin u cos v, 2 sin v), u [0, 2], v [
Then:
~ru = (2 sin u cos v, 2 cos u cos v, 0)
~rv = (2 cos u sin v, 2 sin u sin v, 2 cos v)
2

, ]
2 2

therefore
~ru ~rv = (4 cos u cos2 v, 4 sin u cos2 v, 4 cos v sin v)
and
F~ (~r(u, v)) = F~ (2 cos u cos v, 2 sin u cos v, 2 sin v) = (14 cos u cos v, 0, 2 sin v)

RR

~ du dv =
F~ (~r) N
R

R 2 R 2

R 2 R 2

4 cos u cos2 v

2
(14 cos u cos v, 0, 2 sin v) 4 sin u cos v
2

4 cos v sin v

dv du

(56 cos3 v cos2 u 8 cos v sin2 v)dvdu

Notice that:
R 2 R 2
0

=
=
=
=

R 2 R 2

cos2 u cos v(1 sin2 v)dvdu

R 2 R
R 2 R 2
56 0
cos2 u cos vdvdu 56 0
cos2 u cos v sin2 vdvdu
2

3
R 2
R 2
2
2
56 0 cos2 u[sin v]
du 56 0 cos2 u[ sin3 v ]
du
2
2
R 2
56(2 23 ) 0 cos2 udu
R 2
2u du
56(2 23 ) 0 1 + cos
2

56 cos3 v cos2 udvdu = 56

= 56(2 32 )
because

R 2
0

()

cos 2u = 12 [sin 2u]2

0 = 0.

And:

R 2 sin3 v 2
[ 3 ] du
0
0
2
32
= 3 (F)
Adding () and (F) we obtain the same result that we got doing the triple
8

R 2 R 2

cos v sin2 vdvdu = 8

integral of the divergence of F~ over T (4).

Example 2 Use the Divergence Theorem to calculate

RR
T

F~ ~n dA, when

T = {(x, y, z) : x2 + y 2 = z 2 , 0 z 2}
and F~ (x, y, z) = (4x, 3z, 5y)
3

By the Divergence Theorem:

RR
T

F~ ~ndA =

RRR
T

divF~ dV,

where T = {(x, y, z) : x2 + y 2 z 2 , 0 z 2}
RRR
T

divF~ dV

= 4
= 4

RRR

1dxdydz
T
R 2 R z R z2 x2
0

z 2 x2

1dydxdz,

Using polar coordinates, x = r cos and y = r sin and r [0, z] and [0, 2]:
4

R 2 R z R z2 x2
0

z 2 x2

1dydxdz = 4

R 2 R z R 2

= 8

R2Rz
0

rddrdz

rdrdz

z 2 dz = 32
3
RR
Example 3 For F~ (x, y, z) = (x3 , x2 y, x2 z), evaluate T F~ ~n dA when:
= 4

R2
0

T = {x2 + y 2 = a2 , 0 z b} {x2 + y 2 a2 , z = 0} {x2 + y 2 a2 , z = b}

4

in two different ways.

By Gauss divergence theorem:
ZZ

F~ ~n dA =

ZZZ

divF~ dV

where T = {x2 + y 2 a2 , 0 z b} and divF~ = 5x2 .

Let us compute the triple integral first noticing that x and y are in the disc
of radius a, x2 + y 2 a2 and then we can write:
ZZZ
T

divF~ dV =

Z b Z Z

5x dxdy dz
x2 +y 2 a2

We can use polar coordinates in the double integral as follows:

x = r cos , y = r sin , r [0, a], [0, 2]

Recall that the Jacobian of change of variable in this case is r.

R b RR
0



R b R 2 R a 3
2
2
5x
dx
dy
dz
=
5
r
cos
dr
d
dz
x2 +y 2 a2
0
0 0

h
i
r=a
R b R 2
r4
2
= 5 0
cos

0
4 r=0 d dz


4 R b R 2
5a
1
+
cos
2
= 4 0 0
d dz
2
4
4 Rb
5a
5ba4
=2
[sin
2]
dz
=
= 5ba
+
=0
4
8 0
4

()

On the other hand:

RR
T

F~ ~n dA =

RR
{x2 +y 2 =a2 ,0zb}

RR
{x2 +y 2 a2 ,z=b}

F~ ~n dA +

RR

F~ ~n dA

We compute these three integrals independently.

To compute:
ZZ

F~ ~n dA

{x2 +y 2 =a2 ,0zb}

{x2 +y 2 a2 ,z=0}

F~ ~n dA

we need a parametric representation of the circular cylinder of radius a, for example:

~r(u, v) = (a cos u, a sin u, v), u [0, 2], v [0, b]
Then
~ru = (a sin u, a cos u, 0) and ~rv = (0, 0, 1)
obtaining:
~ = ~ru ~rv = (a cos u, a sin u, 0)
N
Furthermore F~ (x, y, z) = (x3 , x2 y, x2 z) and therefore:
F~ (~r(u, v)) = (a3 cos3 u, a3 cos2 u sin u, a2 v cos2 u)
Then:

a cos u

~ = (a3 cos3 u, a3 cos2 u sin u, a2 v cos2 u)

F~ (~r(u, v))N
a sin u

= a4 cos4 u+a4 cos2 u sin2 u

But
a4 cos4 u + a4 cos2 u sin2 u = a4 cos2 u(cos2 u + sin2 u) = a4 cos2 u
Hence:
RR
{x2 +y 2 =a2 ,0zb}

because 2

R 2
0

F~ ~n dA =

R b R 2

a2 cos2 u du dv
R b R 2
= a2 0 0 (1 + cos 2u) du dv = a2 b,
0
2

cos 2udu = [sin 2u]2

0 = 0.

To compute:
ZZ

F~ ~n dA

{x2 +y 2 a2 ,z=0}

we need a parametric representation of the {x2 + y 2 a2 , z = 0}, for example:

~r(u, v) = (u cos v, u sin v, 0), u [0, a], v [0, 2].
7

Then:
~ru = (cos v, sin v, 0) and ~rv = (u sin v, u cos v, 0)
obtaining:
~ = ~ru ~rv = (0, 0, u)
N
and
F~ (~r(u, v)) = (u3 cos3 v, u3 cos2 v sin v, 0)
Therefore:

3
3
3
2
~
~
F (~r(u.v)) N = (u cos v, u cos v sin v, 0) 0 = 0

u
Hence:
ZZ

F~ ~n dA = 0

{x2 +y 2 a2 ,z=0}

Finally, to compute:
ZZ

F~ ~n dA

{x2 +y 2 a2 ,z=b}

we need a parametric representation of the {x2 + y 2 a2 , z = 0}, for example:

~r(u, v) = (u cos v, u sin v, b), u [0, a], v [0, 2].
Then:
~ru = (cos v, sin v, 0) and ~rv = (u sin v, u cos v, 0)
obtaining:
~ = ~ru ~rv = (0, 0, u)
N
F~ (~r(u, v)) = (u3 cos3 v, u3 cos2 v sin v, bu2 cos2 v)

and

~ = (u3 cos3 v, u3 cos2 v sin v, bu2 cos2 v)

F~ (~r(u.v)) N
0 = bu3 cos2 v

u
Therefore:
RR
{x2 +y 2 a2 ,z=b}

R 2 R a

F~ ~n dA =

4
bu3 cos2 vdu dv = ba4

(Check this last integral, we have done similar ones before!) Hence:
RR
T

F~ ~n dA =

RR
{x2 +y 2 =a2 ,0zb}

RR
{x2 +y 2 a2 ,z=b}
4

F~ ~n dA +
F~ ~n dA

4
= ba4 + 0 + ba4 = 5ba4 ,

as we expected from ()

RR
{x2 +y 2 a2 ,z=0}

F~ ~n dA