Lecture 6: Vector Calculus II Gradient, Divergence and Divergence Theorem
Gradient of Scalars Consider scalar field V and its change dV between two infinitesimal neighbouring points: V V V dV = dx + dy + dz x y z
V V V = ax + ay + a z ( dx a x + dy a y + dz a z ) . y z x
First bracketed factor is easily identified to be the gradient of V :
V =
V V V ax + ay + az x y z
Geometrically,
dV = V cos . dl Maximum when = 0 implies V gives the direction and dV = V d l = V cos dl
magnitude of the maximum rate of change of V per unit distance. If instead, the neighbouring points lie on the surface V = constant, then dV = 0 which occurs when = / 2 . Hence V is perpendicular to V = constant and V V = normal derivative of V . n Generalising,
V a = directional derivative of V along a .
1
Using earlier definitions of del operator and unit vectors for different coordinate system, we have
V 1 V V a + a + az z V 1 V 1 V V = ar + a + a r r r sin V =
Other properties are linearity
( cylindrical polar ) ( spherical polar )
(V + U ) = V + U
and (variants of) Leibniz property
(VU ) = V U + U V U V V U V = U2 U V n = n V n1 V
; ; .
If a vector field A is in the form of a gradient i.e. A = V , then V is called the scalar potential of A . Example (Sadiku, partial Example 3.3): Find the gradient of scalar field W = 10 r sin 2 cos . Solution:
W =
W 1 W 1 W ar + a + a r r r sin
= 10 sin 2 cos a r + 10 sin 2 cos a 10 sin sin a .
Example (Sadiku, Example 3.5): Find the angle of intersection between the line x = y = 2 z and the ellipsoid x 2 + y 2 + 2 z 2 = 10 . Solution: Note that the line will be parametrised by one parameter, say and thus the line given by
r ( ) = 2 a x + 2 a y + a z
Substituting this relation into equation of ellipse gives the intersecting point(s):
( 2 )
+ ( 2 ) + 2 2 = 10
2
= 1 .
Taking = 1 gives the point
r = 2 ax + 2 ay + az
( x, y, z)
= ( 2, 2,1) .
The other point is the inversion image about the origin. The equation of ellipsoidal surface can be written as
f ( x, y, z ) = x 2 + y 2 + 2 z 2 10
(equals zero). Its gradient is
f = 2x a x + 2 y a y + 4z a z
which should be normal to the ellipsoid.
At point ( 2, 2,1) , a unit normal to the ellipsoid will be
an =
f f
=
( 2,2,1)
2(2) a x + 2(2) a y + 4(1) a z 42 + 42 + 42 ax + ay + az 3 .
The angle it makes with line r ( 1) can be obtained using the dot product formula:
cos =
an r an r
2(1) + 2(1) + 1(1) 3 2 + 2 +1
2 2 2
5 3 3
which gives 15.79 . The angle required for the question is however the angle the line makes with tangent to the surface, which is
90 15.79 = 74.21 .
Note that there are other answers due to options of and the normal assignment.
Divergence of Vector Field and Divergence Theorem Earlier:
A dS as net outward flux of A from closed surface.
Propose to define divergence of A at a point P is the outward flux per unit volume in zero volume limit around the point:
div A = A =
lim v 0
A dS v .
Consider
A dS for the durface of infinitesimal cuboid below.
z dx P dy dz
A three-dimensional Taylor series of each component of A about P looks like
A x ( x, y, z ) = A x ( x0 , y0 , z0 ) + ( x x0 ) + ( z z0 ) A x z
P
A x x
P
+ ( y y0 )
A x y
P
+ higher order terms .
For the front side have x = x0 + dx / 2 and dS = dy dz a x , and thus
dx A x A dS = dy dz A x ( x0 , y0 , z0 ) + + higher order terms . 2 x P front For the back side have x = x0 dx / 2 and dS = dy dz ( a x ) , thus dx A x A dS = dy dz A x ( x0 , y0 , z0 ) 2 x back Together, + higher order terms . P
front +back
A dS = dx dy dz
A x x
P
+ higher order terms .
Similarly,
left +right
A dS = dx dy dz
A y y
P
+ higher order terms ;
top +bottom
A dS = dx dy dz
A z z
P
+ higher order terms .
Summing over all surfaces and dividing by v = dx dy dz , we have
lim v 0
Thus,
A dS v
A y A z A = x + + . x y z P
A =
A x x
A y y
A z z
For cylindrical polar coordinates, the divergence can be given by
A =
A 1 1 A ( A ) + + zz
The divergence for spherical polar coordinates is similarly given by
A =
1 2 1 1 A ( r Ar ) + r sin ( A sin ) + r sin r 2 r
Note that the divergence operates only on vector fields (dot product operation) and always give a scalar. The operation is distributive:
( A + B) = A + B ,
and also obeys a generalized Leibnitz rule:
(VA ) = V A + A V
Finally, by definition and summing over infinitesimal volumes, we have Divergence Theorem: The total outward flux of a vector field A through a closed surface S is the same as the volume integral of the divergence of A .
A dS =
A
v
dv
Example (Sadiku Example 3.7): If G ( r ) = 10 e 2 z ( a + a z ) , find flux of G out of cylinder = 1 , 0 z 1. Confirm result through divergence theorem.
Solution: For top end of cylinder, z = 1 and dS = d d a z . Flux through top end is
For bottom end of cylinder, z = 0 and dS = d d ( a z ) . Flux through bottom end is
1 2 2 G dS = 0 010 e d d = 10 e 2 2 0 = 10 e . = = top
1
1 2 G dS = 0 0 10 e d d = 10 ( 2 ) 2 0 = 10 . = = bottom
1 0
For curved surface of cylinder, = 1 and dS = dz d a . Flux through curved surface is
1 2
curved
G dS =
z =0 =0
10 e
2 z
e 2 z (1) d dz = 10 ( 2 ) = 10 (1 e2 ) . 2 0
2
Hence, total flux is
cylinder
G dS = 10 e 2 + ( 10 ) + 10 (1 e 2 ) = 0 .
Since surface is closed, apply divergence theorem:
G dS
S
( G ) dv
v
but
G =
1 1 ( G ) + G + z Gz 1 = ( 210 e2 z ) 20 e2 z = 0 .
Thus,
( G ) dv
v
= 0 =
G dS
S
as required.
