Sie sind auf Seite 1von 54

1

VECTOR CALCULUS


2
Part 1 (Syllabus) Curves and Surfaces
-Euclidean Spaces
-Coordinate transformations in 2D Euclidean Spaces and the Jacobian
-Area integrals in the plane
-Vector valued functions of a scalar variable
-Parametric curves in 3D Euclidean Spae. Curvature and torsion.
-Parametric surfaces in 3D Euclidean Space and curvilinear coordinates
-Elements of length and area in curvilinear coordinates
-Scalar and vector Fields
-Line, surface and volume integrals involving scalar and vector elds

Part 2 (Syllabus) Vector Differential Operators
-Vector differential operators: grad, div and curl
-Identities involving vector differential operators
-Integral Theorems (Greens, Gausss, Stokess)
-Irrotational and Solenoidal Fields
-Scalar and Vector potentials
-Grad, div and curl (cylindrical and spherical polar coordinates)
3
Part 1
Curves and Surfaces
in 3 D Euclidean Space
444
3-D Euclidean Spaces
Spaces whose geometry satisfies the postulates of Euclid arose from
the first attempts to describe the geometry of the space we live in.
Such spaces are called Euclidean spaces and are examples
of "intrinsically flat" spaces. These spaces satisfy in prticular
Euclid's 5th postulate: "Given a straight line, and a point not on
the line, there exists exactly one parallel line".
In 3-D Euclidean spaces (E
3
) a rectangular cartesian coordinate system OXYZ
can be found such that the "metric element of length" ds between neighbouring
points (x, y, z), (x +dx, y +dy, z +dz) is given by
ds
2
= dx
2
+dy
2
+dz
2
(in essence, Pythogoras's Theorem)
The vector algebra developed in first year was based on
such spaces and coordinate systems.
5

NOTATION: RH: Right Handed RC: Rectangular Cartesian

OXYZ or OX
1
X
2
X
3
: RH, RC coordinate systems in 3 D Euclidean Space E
3


Position Vector r = x i + y j + z k x = x
1
e
1
+ x
2
e
2
+ x
3
e
3

= (x, y, z) = (x
1
, x
2
, x
3
)


Z
X
3

X
2

X
1

r =(x, y, z)
x=(x
1
, x
2
, x
3
)
X
Y
We use (x,y,z) or (x
1
, x
2
, x
3
) in different contexts depending on convenience of notation.
O
O
6
If we use x = (x
1
, x
2
.x
3
) to be the position vector of a point in a
RH, RC coordinate system OX
1
X
2
X
3
, the vector element of length
between neighbouring points (the displacement vector)
is
dx = dl = dx
1
e
1
+dx
2
e
2
+dx
3
e
3
The scalar element of length is
dl = dx
If we use r = (x, y, z) to be the position vector of a point in a
RH, RC coordinate system OXYZ, the vector element of length
between neighbouring points is dr = ds = dxi +dy j +dzk
Scalar element of length is ds = dr
dx
dr
O
O
r
x
77
Vectors should be viewed as mathematical objects that have an existence
that is independent of the coordinate system used for its representation.

Thus, the displacement vector joining two adjacent points remains the
same even though its components change under coordinate transformations.
So does the position vector of a point.
To illustrate this point we show below the displacement vector
and its components relative to two RH RC coordinate systems: OX
1
X
2
X
3

and OY
1
Y
2
Y
3
obtained by rotating the original coordinate system.

Y
1

X
1

X
2

X
3
Y
3

Y
2

dl = dr = (dx
1
, dx
2
, dx
3
)
= (dy
1
, dy
2
, dy
3
)
O
88

Differentiation
Consider a vector v which is function of the scalar u.
We define its derivative in the standard way with the
difference in v between adjacent values of u taken as a
vectorial difference (see figure).
dv
du
= lim
!u"0
!v
!u
= lim
!u"0
v(u+!u) #v(u)
!u
= lim
!u"0
v
1
(u+!u) #v
1
(u)
!u
i + lim
!u"0
v
2
(u+!u) #v
2
(u)
!u
j + lim
!u"0
v
3
(u+!u) #v
3
(u)
!u
k
=
dv
1
du
i +
dv
2
du
j +
dv
3
du
k
That is, simply differentiate the components, and sum vectorially.
v(u+!u)
v(u)
!v
Vector valued functions
of a scalar variable
9
Differential of a vector function v(u)
dv =
dv
du
du
and is in the limiting direction of !v as
!u "0.

Differential
v(u+!u)
v(u)
!v
10
X
Y
Z
P
Q
r=r(t)
r=r(t+dt)
dr A particle moves so that its position vector
is given by
r(t) = ti +t
2
j +t
3
k !!!!!a vector valued function of t.
The first derivative gives the velocity
v(t) =
dr
dt
= i +2t j +3t
2
k --another vector valued function of t.
v = v = 1+ 4t
2
+9t
4
------ speed
Note:
dr =
dr
dt
dt = vdt
so the velocity is in the limiting direction of
!r.
That is the velocity is tangent to the space curve
r = r(t).
Speed, Velocity and Acceleration (revision)
11
X
Y
Z
P
Q
r=r(t)
r=r(t+dt)
v(t)
v (t+dt)
X
Y
Z
v(t)
v (t+dt)
dv
P
It is often useful to visualise derivatives of vectors in a coordinate
free way as we have done here.
The derivative of the velocity v with respect to the scalar t is
the acceleration
a(t) =
dv
dt
= 2 j +6tk ---- another vector valued function
12
Differentiation of sums and products
a(u), b(u), c(u) are differentiable vector valued functions (i.e. the components
are differentiable), and f (u) is a differentiable scalar valued function of u.
The following can be proved .
d
du
(a(u) + b(u))=
d
du
a(u) +
d
du
b(u)
d
du
( f (u)a(u))=
df
du
a(u) + f (u)
d
du
a(u)
d
du
(a(u) . b(u))=
da(u)
du
.b(u) +a(u).
db(u)
du

d
du
(a(u) ! b(u))=
da(u)
du
!b(u) +a(u)!
db(u)
du

13
Example:
To prove
d
du
( f (u)a(u))=
df (u)
du
a(u) +f (u)
da(u)
du
, we take components
and prove the result for each component.
d
du
( f (u)a
k
(u)) (k =1, 2, 3)
lim
!u!0
f (u+!u)a
k
(u+!u)-f (u)a
k
(u)
!u
= lim
!u!0
( f (u)+ " f (u)!u+O(!u
2
))((a
k
(u) + " a
k
(u)!u+O(!u
2
))-f (u)a
k
(u)
!u
(assuming differentiability)
= " f (u)a
k
(u) + f (u) " a
k
(u)
Example:
p(u) = p
1
(u)e
1
+ p
2
(u)e
2
+ p
3
(u)e
3
d
du
p(u) =
dp
1
(u)
du
e
1
+ p
1
(u)
de
1
du
+....... ( only because basis vectors are constant vectors)
=
dp
1
(u)
du
e
1
+
dp
2
(u)
du
e
2
+
dp
3
(u)
du
e
3
..(i.e. can simply differentiate components)
14
!
d
du
(a .( b"c)) =
da
du
. (b"c) +a . (
db
du
"c) +a .( b"
dc
du
)
d
du
(a "( b"c)) =
da
du
" (b"c) +a " (
db
du
"c) +a "( b"
dc
du
)
Rule : differentiate term by term preserving order of vectors
in each expression.
Differentiating triple products
(see appendix for revision of scalar product, vector product,
scalar triple product and vector triple product)
15
Example
Suppose p = p(t) where t is the scalar variable (time),
! p =
d
dt
p(t) etc.
(1)
d
dt
( p. ! p!!! p) = ! p. ! p!!! p+ p. !! p!!! p+ p. ! p!!!! p
= p. ! p!!!! p
(2)
d
dt
( p! ! p) = p!!! p+ ! p! ! p
= p!!! p
16
If a vector is a function of more than one scalar variable,
we can define partial derivatives.
Suppose p(u, v)
p
u
=
! p
!u
= lim
"u!0
p(u+"u, v) " p(u, v)
"u
provided the limit exists.
The rules we had previously carry over also to partial derivatives:
Thus
!
!u
(a(u, v) . b(u, v))=
!a(u, v)
!u
. b(u, v) +a(u, v).
!b(u, v)
!u

etc.
17
Chain Rule for Vector Valued Functions
If
a(u
1
, u
2
,....u
n
)
and
u
j
= u
j
(v
1
, v
2
.....v
n
) ( j =1, n)
then
!a
!v
i
=
!a
!u
k k=1
n
"
!u
k
!v
i

[Follows from the chain rule applied to the p
th
component
a
p
(u
1
, u
2
,....u
n
) of a, which is a scalar function.
!a
p
!v
i
=
!a
p
!u
k k=1
n
"
!u
k
!v
i
(from first year) ]
18
Sum implicitly over repeated indices. The above can be written
more compactly as
!a
!v
i
=
!a
!u
k
!u
k
!v
i
k is called the dummy index and
k=1
n
"
is implied.
Einstein Summation Convention

19 19
Integration of a vector function of a scalar variable
Suppose b(u) is the vector function of a scalar variable u
possessing a derivative
a(u) =
db
du
Then the indefinite integral of a(u) is
a(u)
!
du = b(u) + c
where c is an arbitrary constant vector.
If a(u) = a
1
(u)e
1
+a
2
(u)e
2
+a
1
(u)e
3
b(u) = e
1
a
1
(u)
!
du+e
2
a
2
(u)
!
du+e
3
a
3
(u)
!
du
20
Example: Given a(u), evaluate
c
!
"
da
du
du
where c is a constant vector.
Now
d
du
(c"a) =
dc
du
"a +c"
da
du
= c"
da
du
(since c is a constant)
Hence
c"
da
du
!
du = c"a +d
where d is a constant vector.
Note that we could also have written
c
!
"
da
du
du = c"
da
du
!
du = c"a +d
because c is independent of u and can therefore be taken "out of
the integral". (Strictly, this has to be justified by taking
components - try to show this!)
21
Notation:
A function f defined in an open domain D is said to be
-- of class C
0
if it exists and is continuous in D
-- of class C
1
if its first partial derivatives exist and are continuous
in D
---of class C
2
if its second partial derivatives exist and are continuous
in D
22 22
Consider a RH, RC coordinate system (x
1
, x
2
) which specifies points in
R
2
. Introduce a new coordinate system (x
1
!
, x
2
!
) by rotating the coordinate axes
through an angle !. By construction, this is also a RH, RC coordinate system.
The components of the position vector OP
! " !!
are different in the two
coordinate systems and are related by
x
1
= ! x
1
cos! " ! x
2
sin! = f
1
(x
1
!
, x
2
!
)
x
2
= ! x
1
sin! + ! x
2
cos! = f
2
(x
1
!
, x
2
!
)
This is an example of an "orthogonal" coordinate
transformation (the matrix L =
cos! "sin!
sin! cos!
#
$
%
&
'
(
connecting x and x
!
is orthogonal" in the linear algebra sense
i.e. L
"1
= L
T
)
Orthogonal Transformations in 2D Euclidean Space
O
P
P
1

P
1
/

X
1

X
1
/

x
2

x
1
1

x
2
1

X
2

X
2
/

x
1


!
!
23
Consider a RH, RC coordinate system (x
1
, x
2
) which specifies points in
R
2
. Introduce a new coordinate system (u, v) through the transformation
x
1
= f
1
(u, v), x
2
= f
2
(u, v)
or x = f (u, v)
We assume that f !C
1
in some domain R
uv
. We can look at this as a mapping from
the region R
uv
in (u,v) coordinate space to a region R in (x
1
, x
2
) space.
The Jacobian of the transformation is
J =
! f
1
!u
! f
1
!v
! f
2
!u
! f
2
!v
(definition)
General Coordinate Transformations in 2D Euclidean Space
x1
R
u
v
Ruv
x2
f
24
From the transformation, and the differentiability
of f
1
(u,v) and f
2
(u,v) [because any C
1
function is differentiable] , we have
dx
1
=
! f
1
!u
du+
! f
1
!v
dv
dx
2
=
! f
2
!u
du+
! f
2
!v
dv
For these equations to be invertible, (solvable for (du, dv))
we require
J =
! f
1
!u
! f
1
!v
! f
2
!u
! f
2
!v
! 0
We choose transformations that are (1-1) on R
uv
- that is J ! 0.
25
Note that the Jacobian of the inverse transformation is

!(u, v)
!(x, y)
=
!u
!x
!u
!y
!v
!x
!v
!y
It can be shown that
!(x, y)
!(u, v)
=1
!(u, v)
!(x, y)
,
!(x, y)
!(x, y)
=1 ----- (show using chain rule)
We also have transitivity under successive coordinate transformations
(x, y) !(s, t) !(u, v)
!(x, y)
!(s, t)
=
!(x, y)
!(u, v)

!(u, v)
!(s, t)
[See Adams for examples on Jacobians]
26
C
2
: x=f(uP,v)
C
1
: x=f(u,v
p
)
v coordinate
curves
u coordinate
curves
X
2
X
1
P
By keeping one parameter fixed and varying the other,
we get coordinate curves
C
1
: x = f (u, v
P
) = f
1
(u, v
P
)e
1
+ f
2
(u, v
P
)e
2
C
2
: x = f (u
P
, v) = f
1
(u
P
, v)e
1
+ f
2
(u
p
, v)e
2
Along C
1
, u varies. This is the u-coordinate curve through P.
(u,v) are sometimes called curvilinear coordinates.
Curvilinear coordinate and coordinate curves
27
!
Example : Plane polar coordinates
x = f
1
(r,") = rcos"
y = f
2
(r,") = rsin"
r = c = const (" coordinate curves)
# x = ccos", y = csin"
# x
2
+ y
2
= c
2
$$$$circle
" = % = const (r coordinate curves)
# x = rcos%, y = rsin%
# y = x tan%$$$$$straight lines through origin
x 0
r-coordinate
curves
!"coordinate
curves
y
28
Example: Orthogonal transformation
x
1
= ! x
1
cos! " ! x
2
sin! = f
1
(x
1
!
, x
2
!
)
x
2
= ! x
1
sin! + ! x
2
cos! = f
2
(x
1
!
, x
2
!
)
! x
2
= c = const --> ! x
1
coordinate curve
x
1
= ! x
1
cos! "csin!
x
2
= ! x
1
sin! +ccos!
# x
1
+csin! = (x
2
"ccos!)cot!
# x
2
= (tan!)x
1
+c / sin!
O X
1

X
2


! x
1
coordinate curves
! x
2
coordinate curves
!
29
Elements of length and area
x = f (u, v) !dx = f
u
du+ f
v
dv (because f is differentiable)
where f
u
=
! f
!u
, f
v
=
! f
!v

Distance between two points P(u,v), Q (u+du, v+dv)
is given by
dl
2
= dx. dx = dx
1
2
+dx
2
2
= f
u
du+ f
v
dv
( )
. f
u
du+ f
v
dv
( )
= f
u
. f
u
du
2
+2 f
u
. f
v
dudv + f
v
. f
v
dv
2
Elements of length along u and v coordinate curves are
dl
u
2
= f
u
. f
u
du
2
= h
1
2
du
2
(with dv = 0)
dl
v
2
= f
v
. f
v
dv
2
= h
2
2
dv
2
(with du = 0)
h
1
= f
u
=
! f
!u
, h
2
= f
v
=
! f
!v
X2
X1
h2dv
h1du
P(u,v)
Q(u+du,v+dv)
R(u+du, v)
Arc Lengths
T(u,v+dv)
Element of length
P
x=f(u,v)
Q
x=f(u+du,v+dv)
=x+dx
dx
30
We can form unit tangents !
1
to coordinate curve C
1
and !
2
to coordinate
curve C
2
at P:
!
1
=
1
h
1
" f
"u
!
"
#
$
%
&
P
, !
2
=
1
h
2
" f
"v
!
"
#
$
%
&
P
h
1
=
" f
"u
P
, h
2
=
" f
"v
P

(The subscript P is usually omitted)
!
1
'!
2
=
1
h
1
h
2
" f
"u
'
" f
"v
(
)
*
+
,
-
=
1
h
1
h
2
(
" f
1
"u
" f
2
"v
.
" f
1
"v
" f
2
"u
)e
3
=
1
h
1
h
2
" f
1
"u
" f
1
"v
" f
2
"u
" f
2
"v
e
3
Note that for a non-degenerate transformation (i.e distinct families of
of coordinate curves required for a 1-1 transformation),we require
!
1
'!
2
/ 0 . Since the Jacobian is non-zero, this is guaranteed.
C
2
: x=f(uP,v)
C
1
: x=f(u,v
p
)
X
2
X
1
!
2
!
1
P
31
If !
1
and !
2
satisfy
!
1
!!
2
= e
3
or equivalently if
" f
1
"u
" f
1
"v
" f
2
"u
" f
2
"v
=
"( f
1
, f
2
)
"(u, v)
= h
1
h
2
we have an orthogonal curvilinear coordinate system.
Through any point, the two coordinate curves are
orthogonal to each other.

Orthogonal curvilinear coordinate systems
x 0
r-coordinate
curves
!"coordinate
curves
y
Example
32
dA = (h
1
du !
1
)!(h
2
dv !
2
) = h
1
h
2
dudv !
1
!!
2
= h
1
h
2
1
h
1
h
2
f
u
! f
v
dudv = f
u
! f
v
dudv
= Abs value of
" f
1
"u
" f
1
"v
" f
2
"u
" f
2
"v
dudv
dA =
"( f
1
, f
2
)
"(u, v)
dudv
or equivalently
dx
1
dx
2
=
"( f
1
, f
2
)
"(u, v)
dudv
(Jacobian measures local magnification of area dudv)
For an orthogonal system,
dA = h
1
h
2
dudv
X2
X1
h2dv
h1du
P(u,v)
Q(u+du,v+dv)
R(u+du, v)
Arc Lengths
T(u,v+dv)
Element of area
u
v
(u,v) (u+du,v)
(u+du,v+dv) (u,v+dv)
33
Example: x = r cos!, y = rsin! (plane polar coordinates)
"(x, y)
"(r,!)
=
"x
"r
"x
"!
"y
"r
"y
"!
=
cos! !rsin!
sin! r cos!
= r
"(r,!)
"(x, y)
=
1
r
(check directly)
dA =
"(x, y)
"(r,!)
drd! = rdrd!
Note:
h
1
=
" f
"r
= (
" f
1
"r
)
2
+(
" f
2
"r
)
2
= cos
2
! +sin
2
! =1
h
2
=
" f
"!
= (
" f
1
"!
)
2
+(
" f
2
"!
)
2
= r
2
sin
2
! +r
2
cos
2
! = r
dA = (h
1
dr)(h
2
d!) !!!!!because coordinate curves orthogonal
x 0
r-coordinate
curves
!"coordinate
curves
y
r
!
dr
d!
dr
rd!
34
!
Single integrals : x = x(u)
f (x)dx = f (x(u))
dx
du u
1
u
2
"
x
1
x
2
" du
Double Integrals; x = x(u,v)
f (x
1
, x
2
)dA
R
"" = f ((x
1
(u,v), x
2
(u,v))
R
uv
""
#(x
1
, x
2
)
#(u,v)
dudv
or using (x, y) as variables
f (x, y)dxdy
R
"" = f ((x(u,v), y(u,v))
R
uv
""
#(x, y)
#(u,v)
dudv
Area integrals by coordinate transformation
35
X2
X1
u
v
R
uv
R
X1
X2
R
uu
u
R
uv
No Obvious
advantage
Clear
Advantage!
Choose a transformation which simplies the boundaries
or the integrand or both
36
EX: Calculate area bounded by the curves
y=0, y = x
2
, y =1! x
2
, y = 4! x
2
.
The form of the boundaries suggests that we
cook up a transformation that makes the two
families of curves coordinate curves: So we
take
y = ux
2
(1" u "1), y = v ! x
2
(1" v " 4)
The four boundaries are u = 0, u =1, v =1, v = 4
u = y x
2
, v = x
2
+ y
#
!(u, v)
!(x, y)
=
!
2y
x
3
1
x
2
2x 1
= !
2
x
(
y
x
2
+1) = !2
(u+1)
3
2
v
1
2
!(x, y)
!(u, v)
= !
v
1
2
2(u+1)
3
2
y=x
2
y=4-x
2

y=1-x
2

y=0
v=1
v=4 u=1
u=0
Y
X
37
!
dA = dxdy =
"(x, y)
"(u,v)
## ## ## dudv
$ dA =
1
2
(
v=1
v=4
# ##
v
1
2
(u +1)
3
2
du )dv
u=0
u=1
#
=
1
2
( v
1
2
v=1
v=4
# dv)(
1
(u +1)
3
2
du
u=0
u=1
# )

=
1
2
2
3
v
3
2
%
&
'
(
)
*
1
4
+2(u +1)
+
1
2
%
&
'
(
)
*
0
1
=
7
3
( 2 +1)
u
v
0 1
4
0
38
Scalar and Vector Fields
!
Consider a doman D " R
3
Scalar Field #(x)
For every point P $ D assign a scalar # $ R
#(x) = #(x
1
, x
2
, x
3
)
Ex : #(x) = x
1
2
+ x
2
2
+ x
3
2
= r (distance from origin)
%(x) = density at position x in a fluid
Vector Field v(x)
For every point P $ D assign a vector v $ R
v(x) = v
1
(x
1
, x
2
, x
3
)e
1
+v
2
(x
1
, x
2
, x
3
)e
2
+v
3
(x
1
, x
2
, x
3
)e
3
Ex : v(x) = x (Big bang model of universe)
v = v = x = r
39
!
Example : v = (x,0,0)
Velocity has only a "x" component.
The magnitude is
v = x
2
+0+0 = x
Example : v = (y,0,0)
Velocity has only a "x" component.
The magnitude of the velocity is
v(x, y,z) = y
2
+0+0 = y
Example : v = (
x
x
2
+ y
2
,
y
x
2
+ y
2
,0)
The magnitude of the velocity is
v =
1
x
2
+ y
2
x
2
+ y
2
=
1
x
2
+ y
2
=
1
r
X
Y
X
Y
Y
X 40
Curves and Surfaces in R
3

41 !
A curve C is a set (" R
3
) of points having
at least one parametric representation
x = f (t) = f
1
(t)e
1
+ f
2
(t)e
2
+ f
3
(t)e
3
where f is continuous in [a,b] (that is f #C
0
)
We say that f maps C
t
:[a,b] onto C.
C is an arc if f (a) $ f (b)
C is a closed curve if f (a) = f (b)
a b
t
X1
X2
X3 t=a
t=b
Ct
C
Curves
42
!
C is simple if there exists at least one C
0
mapping f (t) of [a,b] on to C
which is one to one on [a,b]. That is, a simple curve cannot intersect
itself
C is smooth if there exists at least one C
1
mapping
(
df
j
dt
= " f
j
continuous) with
" f (t) = f
3
"
2
+ f
2
"
2
+ f
3
"
2
# 0 on [a,b]
(the latter is the requirement for the curve to have a tangent)
A simple curve C is piecewise smooth if the images C
1
,C
2
,C
3
....C
N
(N finite) of [a, t
1
],[t
1
, t
2
],........[t
N$1
,b] are smooth.
X1
X2
X3
t=a
t=b
C
A simple
piecewise-
smooth curve
43
A plane curve (e.g. x
3
= f
3
= 0)
x(t) = f
1
(t)e
1
+ f
2
(t)e
2
is said to be positively directed if t increasing
corresponds to counter clockwise
passage in the x
1
! x
2
plane.
Each simple curve has two orientations. A simple curve with a sense of
direction is said to be an oriented simple curve or a directed simple
curve. The orientation is induced by the parametrisation.
Ex: P to Q , or Q to P in the simple curve shown below.




P
Q
Q
P
u increasing
v increasing
x = f (u)
x = g(v)
44
Suppose C is a simple smooth curve given by x= f (t) a ! t ! b.
Let P and Q be two adjacent points on C
P: x = x
P
= f (t)
Q:x = x
Q
= f (t +"t)
Then PQ= "x = f (t +"t) # f (t)
Thus "x "t is a vector in direction PQ. In limit "t $0,
PQ becomes tangent to curve at P. Thus
T =
dx
dt
= lim
"t$0
"x
"t
=
d f
dt
= % f (t)
is a vector which is tangent to the curve at P.
Unit Tangent
! =
dx dt
dx dt
=
% f (t)
% f (t)
=
1
h
% f (t)
h = % f (t) is related to arc length along curve (see later)
O
P(t)
Q(t +!t)
xP
xQ
"
45
dl
2
= dx
2
= dx
1
2
+dx
2
2
+dx
3
2
=
dx
1
dt
!
"
#
$
%
&
2
+
dx
2
dt
!
"
#
$
%
&
2
+
dx
3
dt
!
"
#
$
%
&
2
'
(
)
)
*
+
,
,
dt
2
= - f
1
2
+ - f
2
2
+ - f
3
2
'
(
*
+
dt
2
dl = - f
1
2
+ - f
2
2
+ - f
3
2
= - f dt = hdt
h is called the length scale factor that converts parametric coordinate difference
to metric (physical) length. Sometimes we use ds instead of dl. The arc length is
s(t) = l(t) = dl =
.
h( - t )d - t
t
0
t
.
Note that dx ( = dr) can be looked upon as a vectorial element of length
along the curve. We sometimes write
dl = ds = dx = (hdt)!
O
P(t)
x
x +dx
Q(t+dt)
dl=IdxI = hdt
Element of Length
46 46
Curvature

!
Suppose that C is a simple smooth curve parametrised as
x = f (s) a " s " b
where s is the arc length along the curve (for an arbitrary parametrisation t,
use
ds
dt
to change variables from t #s)
With this parametrisation the tangent
T =
dx
ds
is now also the unit tangent because the
scale factor
h =
dx
ds
=1 (!dx = ds)
$% =
dx
ds
X1
X2
X3
P(x(s)
s
P0
Unit tangent
!(s)
47 47
!
Since "." =1
".
d"
ds
+
d"
ds
." = 0 #2".
d"
ds
= 0
which implies that " is perpendicular to
d"
ds
.
We define the curvature of x(s) to be
$ =
d"
ds
and the principal normal n by
d"
ds
= $n
The plane containing " and n at any point P of the curve
is called the osculating plane.
Note that
n =
1
$
d"
ds
=
1
$
d
2
x(s)
ds
2
48 48
x1
X3
!"
Unit tangent #(s)
X2
Unit tangent #(s+!s)
P[x(s)]
Q[x(s+!s)]
P0
s
!
d"
ds
= lim
#s$0
#"
#s
= lim
#s$0
#%
#s
= &
So curvature measures the
rate of rotation of the
unit tangent as we move along the curve.
!(s)
!(s+"s)
"#
"!
"! = "#
Radius of Curvature:
!
" =
1
#
49 49
!
Ex : Consider the circular helix defined by
x(t) = coste
1
+sinte
2
+te
3
T =
dx
dt
= "sinte1 +coste2 +e3 """"tangent
ds = dx =
dx
dt
dt = 2dt #scale factor h = 2
and
ds
dt
= 2
$ =
dx
dt
/
dx
dt
=
1
h
dx
dt
=
1
2
("sinte1 +coste2 +e3) - - - unit tangent
d$
dt
=
1
2
("coste
1
"sinte
2
)
%
d$
ds
=
dt
ds
d$
dt
=
1
2
1
2
("coste
1
"sinte
2
)
%& =
d$
ds
=
1
2
and
Principal normal is n = "(coste1 +sinte2 )
X3
X1
X2
1
50 50
Torsion
Define the unit bi-normal vector by
b(s) =! (s)!n(s) (i.e. complete the orthonormal triad).
b(s).b(s) =1"2b(s).
db(s)
ds
= 0
"
db
ds
is perpendicular to b
Also
!.b = 0 "
d!
ds
. b+!.
db
ds
= 0
"n. b+!.
db
ds
= 0
!.
db
ds
= 0
"
db
ds
is perpendicular to !
#
db
ds
is in the direction of n
b(s)
!(s)
n(s)
51 51
We define torsion by
db
ds
= !!n
That is,
! =
db
ds
Since b is a unit vector, we see as before that !
is the rate of rotation of b, or equivalently of
the osculating plane with distance along the curve.
! is positive if b rotates positively (in a right handed
sense) around " as s increases.
The torsion of a curve can also be seen to be the extent
to which a curve fails to be planar.
b(s)
b(s+!")
!"
!b
!b =!"
b(s)
!(s)
n(s)
52 52
Example : Calculate the torsion of the circular helix x(t) = coste
1
+sinte
2
+te
3
We had
ds
dt
= 2
! =
1
2
(!sinte
1
+coste
2
+e
3
) ---unit tangent
n = !(coste
1
+sinte
2
) ---------unit normal
"b =! #n =
1
2
e
1
e
2
e
3
!sint cost 1
!cost !sint 0
=
1
2
(sinte
1
!coste
2
+e
3
)
----unit binormal
db
dt
=
1
2
(coste
1
+sinte
2
)
$
db
ds
=
db
dt
dt
ds
=
1
2
(coste
1
+sinte
2
)
Using
db
ds
= !"n, torsion " =
1
2
X3
X1
X2
1
53 53
Example: Show that
dn
ds
= !!" +#b
(Hint: differentiate n = b"" )
Collecting all the results, we have a set of three
equations which are known as the Frenet-Serret Relations.

d"
ds
=!n
dn
ds
= !!" +#b

db
ds
= !#n
54
Line Integrals
Line integrals along a path C given by
x = x
1
(t)e
1
+ x
2
(t)e
2
+x
3
(t)e
3
= f
1
(t)e
1
+ f
2
(t)e
2
+f
3
(t)e
3

may be of one of several forms;
A(x). dx
C
!
, A(x)"dx
C
!
, !(x)dl
C
!
, A
1
(x)dx
2
C
!
etc.
where A(x) is a vector field and !(x) is a scalar field.
Now
A(x).dx
C
!
= (e
1
A
1
+e
2
A
2
+e
3
A
3
). (e
1
dx
1
+e
2
dx
2
+e
3
dx
3
)
C
!
= A
1
(x)dx
1
+
C
!
A
2
(x)dx
2
+
C
!
A
3
(x)dx
3
C
!

X1
X2
X3
C
a
b Cx1
O
55
I = A!dx
C
"
= (e
1
A
1
+e
2
A
2
+e
3
A
3
)!(e
1
dx
1
+e
2
dx
2
+e
3
dx
3
)
C
"
= e
1
(A
2
dx
3
# A
3
C
"
dx
2
) +e
2
(A
3
dx
1
# A
1
C
"
dx
3
) +e
3
(A
1
dx
2
# A
2
C
"
dx
1
)
and
[ A(x)!dx
C
"
]
1
= (A
2
dx
3
C
"
# A
3
dx
2
) etc.
56
Noting that dx
k
= ! f
k
(t)dt, dl = dx =
dx
dt
dt = ! f (t) dt
all such line integrals can be built up from the following
two basic forms:
!
C
"
(x)dl = !( f (t)) ! f (t) dt
a
b
"
and
!
C
"
(x)dx
k
= !
a
b
"
( f (t)) ! f
k
(t)dt (k=1,2,3)
where !(x) is some scalar field.
So we now focus on how to evaluate such integrals.
57
Note:
It is important to visualise any integral as the limit of a
sum (Riemann Sum) formed as follows:
Partition the variable of integration into N discrete intervals.
Weight each interval by the value of the integrand evaluated
at a point in the interval
.
!
C
!
(x)dx
1
= Lim
N"#
"x
1
p
"0
!(x
1
p
)"x
1
p
p=1
N
$
!
C
!
(x)dl = Lim
N"#
"l"0
!(x
k
p
)"l
p
p=1
N
$
X1
X2
X3
C
a
b Cx1
O
58
The following theorem can easily be proved (left as an excercise!)
Theorem:
If C is an oriented simple curve or an oriented simple closed curve
we can unambigously evaluate
!(x)
C
!
dx
k
or !(x)dl
C
!
using any orientation preserving parametrisation x = f (t) of C. The sign
of the integrals will be reversed if we change the orientation.
59
If we use one of the coordinates in E
3
as an parameter - say x
1
(an "explicit representation")
C: x
2
= g(x
1
), x
3
= h(x
1
) a ! x
1
! b
where h and g are continuous functions in [a, b],
the parametric equation can be written as
x = e
1
x
1
+e
2
x
2
+e
3
x
3
= e
1
x
1
+e
2
g(x
1
) +e
3
h(x
1
) (here parameter t = x
1
)
dx = e
1
dx
1
+e
2
" g (x
1
)dx
1
+e
3
" h (x
1
)dx
1
dl = ds = dx =
dx
dx
1
dx
1
= 1+ " g
2
+ " h
2
dx
1
Then these basic elements become
!
C
#
(x)dl = !(
a
b
#
x
1
, g(x
1
), h(x
1
)) 1+ " g
2
+ " h
2
dx
1
!
C
#
(x)dx
1
= !(
a
b
#
x
1
, g(x
1
), h(x
1
))dx
1
!
C
#
(x)dx
2
= !(
a
b
#
x
1
, g(x
1
), h(x
1
)) " g (x
1
)dx
1
...etc.
Explicit representation
X1
X2
X3
C
a
b Cx1
O
60
Example: Evaluate xy
2
ds
C
!
where C is the quarter circle
x=cost, y=sint (0 " t "
!
2
)
r = (x, y) = (cost, sint)
ds =
dr
dt
dt = sin
2
t +cos
2
tdt = dt
xy
2
ds
C
!
= cost sin
2
0
!/2
!
tdt = sin
2
0
!/2
!
td(sint) =
sin
3
t
3
#
$
%
&
'
(
0
!/2
=
1
3
Example: Evaluate xy
2
dx
C
!
, xy
2
dy
C
!
along the parabola y = x
2
from (0,0) to (2,4).
r = (x, y) = (x, x
2
) (using explicit representation with x as parameter)
dr = (dx, dy) = (dx, 2xdx)
xy
2
dx
C
!
= x. x
4
0
2
!
dx =
x
6
6
#
$
%
&
'
(
0
2
=
32
3
xy
2
dy
C
!
= x.x
4
.2x
0
2
!
dx = 2
x
7
7
#
$
%
&
'
(
0
2
=
256
7
61
Given a force field F(r) acting on a particle and a curve C,
the work done in moving the particle along C is
W= F. dr
C
!

Example: Calculate work done by the force of gravity F=(0,0,-mg) in
moving a particle of mass m from a = (x
0
,y
0
, z
0
) to b = (x,y, z).
Work done: W = F. dr
C
1
!

= (
C1
!
F
x
, F
y
, F
z
). (dx, dy, dz)
= (F
x
dx +
C1
!
F
y
dy + F
z
dz)
=- mgdz
z0
z
!
= "mg(z " z
0
)
(Note : The path C
1
never came into the calculation!)
B(x,y,z)
A(x0,y0,z0)
C
F
X
Z
Y
C
1
Work Integral
62
To calculate the work integral along a general path, first
write the equation of the path as a parametrised curve.
In the above example, the equation of path C (a straight line) is
r = a +t(b!a) = (x
0
, y
0
, z
0
) +(x - x
0
, y ! y
0
, z - z
0
)t
(0 " t "1)
W = F. dr
C
#
= F(r(t)).
dr
dt
0
1
#
dt
= (
0
1
#
(0, 0, -mg)).(x - x
0
, y ! y
0
, z - z
0
)dt = !mg(z ! z
0
) dt
0
1
#
= !mg(z ! z
0
)
as before
In fact, W = F.dr
C
1
#
= F.dr
C
#
independently of path because
the force field is "conservative" (see later)
63
If C is a closed curve,
I = A . dx
!
!
or A . dl or A . dr
!
!
or A . ds
!
!

!
!
is called the CIRCULATION of A about C and written
Example : Calculate the circulation of v = (x
1
, 0, 0) over unit circle.
Parametrise circle:x = (cost, sint, 0) (0 " t " 2!)
v. dx
unit circle
!
!
= v .
dx
dt
dt
unit circle
!
!
= (cost, 0, 0).(#sint, cost, 0)dt
0
2!
!
=- sint cost dt
0
2!
!
= 0
Circulation
Integral
x1
x2
64
Another type of integral is
I = A!dx
C
"

This is interpreted as
I = (e
1
A
1
+e
2
A
2
+e
3
A
3
)!(e
1
dx
1
+e
2
dx
2
+e
3
dx
3
)
C
"
= e
1
(A
2
dx
3
# A
3
C
"
dx
2
) +e
2
(A
3
dx
1
# A
1
C
"
dx
3
) +e
3
(A
1
dx
2
# A
2
C
"
dx
1
)
Components are
I
1
= A
2
dx
3
# A
3
dx
2
C
"
etc. of the standard form.
C
"
Another type of circulation integral would be
I = A!dx
C
!
"

where C is a closed curve.

65
!
Example : In the previous example with v = (x
1
,0,0) calculate
I = v
unit circle
" # dx
Now x = (cost,sint,0) and dx = ($sint,cost,0)
I
1
= (v
2
dx
3
$v
3
C
" dx
2
) = 0
I
2
= (v
3
dx
1
$v
1
C
" dx
3
) = 0
I
3
= v
1
unit circle
" dx
2
$ v
2
unit circle
" dx
1
v
1
unit circle
" dx
2
= x
1
unit circle
" dx
2
= cost d(sint)
0
2%
" = cos
2
t dt
0
2%
"
=
1
2
[1+cos2t] dt
0
2%
" = %
and v
2
unit circle
" dx
1
= 0
&I = % e
3

x
1
x
2
66
Not all surfaces can be represented as a graph of a function
of 2 variables using rectangular cartesian coordinates

z = f(x,y)

Thus, it is not possible to represent the surface of a doughnut
by such an equation.

We are therefore led to consider more general parametric
representations of surfaces.

Just as curves can be seen as mappings of a portion of a straight line
surfaces can be seen as being mappings of a region of the plane.
Surfaces
67
!
C
uv
is positively directed (simple piecewise smooth) curve in the (u,v) plane.
S
uv
is the union of C
uv
and its interior.
A surface S is a set (" R
3
) of points having at least one parametric representation
x = g(u,v) = e
1
g
1
(u,v) +e
2
g
2
(u,v) +e
3
g
3
(u,v) u,v #S
uv
g(u,v) is continuous on S
uv
. As u,v vary, the tip of x describes a
surface in x
1
x
2
x
3
space. g is a mapping of S
uv
onto S.
C
S
x =g(u,v)
x1
x2
x3
P
S
uv
u
v
Cuv
Mapping
Parametric Surfaces
g (u,v)
68
S is simple if there exists at least one mapping x = g(u,v)
(of S
uv
onto S) that is one to one.
If S is simple, C
uv
maps onto the edge of C.
If g(u, v) !C
1
, we also say that the surface is differentiable,
or a C
1
surface.
S is smooth if the functions
!g
!u
and
!g
!v
are continuous
on S
uv
(that is, g(u, v) !C
1
) with
!g
!u
"
!g
!v
# 0 ( u, v ! S)
(This is the condition for there to be a tangent plane at
every point -int uitively, no corners)
69
Example: Consider the parametrised surface
x = g(r, !) = r cos!e
1
+rsin!e
2
+re
3
(r ! 0, 0 "! " 2")
Here
g
1
(r, !) = r cos!, g
2
(r, !) = rsin!, g
3
(r, !) = r
This is the surface of a cone of semi vertex angle 45
0
because
x
3
= x
1
2
+ x
2
2

with (0,0) #(0,0,0) (....the vertex)
Clearly, all partial derivatives of g are continuous everywhere
in (r,!) plane. But
T
1
=
#g
#r
= cos!e
1
+sin!e
2
+e
3
T
2
=
#g
#!
= $rsin!e
1
+r cos!e
2
and
T
1
%T
2
= r($cos!e
1
$sin!e
2
+e
3
) = 0 at (0,0)
This is a differentiable surface, but it is not smooth (by definition).
r
!
x3
x1
x2
45
70
Let S be a simple smooth surface. Consider a point P ! (u
P
,v
p
) on S.
The curves
C
1
: x = g(u, v
P
)
C
2
: x = g(u
P
, v)
which clearly lie on S and pass through P, are
called the coordinate curves through P.
We can form tangents T
1
to C
1
and T
2
to C
2
at P.
T
1
=
!g
!u
"
#
$
%
&
'
u
P
,v
P
, T
2
=
!g
!v
"
#
$
%
&
'
u
P
,v
P
!g
!u
(
!g
!v
P
) 0 (smooth * coordinate curves through P are distinct).
Note: The directions that we get for the tangent vectors (i.e. which way
they point) depends on the parametrisation.
Coordinate Curves
n
+
P
C1: x=g(u,vP)
C2: x=g(uP,v)
n
-
71 71
We can form everywhere on S two unit normals
n
+
=
g
u
!g
v
g
u
!g
v
, n
"
=-n
+

Note that what we define as the + and - directions is arbitrary
depending on our parametrisation.
For a given parametrisation, the positive sign is used for the
normal direction that forms a RH triad with respect to !
1
and !
2
.
This is turn depends on the sense of travel long the bounding curve
of the surface (assuming an open surface) as u and v increase.
!1
!2
P
n
+
n
-
A surface has two normals at every point
72
Unit tangents to coodinate curves:
!
1
=
"g
"u
"g
"u
=
1
h
1
"g
"u
, !
2
=
"g
"v
"g
"v
=
1
h
2
"g
"v
h
1
= g
u
, h
2
= g
v
can be identified as the length scales
along the coordinate curves (see below)
The plane containing !
1
, !
2
is called the tangent plane at P.
Every tangent to the surface at P lies in the tangent plane
through P.
Now for any two adjacent points P,Q on surface,
dl = dx =
"g
"u
du+
"g
"v
dv = g
u
du+ g
v
dv
!1
!2
P
n
+
n
-
x1
x2
x3
P
Q
S
g(u,v)
g(u+!u,v+!v)
dl
73
The metric distance between two points P(u,v), Q (u+du, v+dv) is given by
dl
2
= dx
1
2
+dx
2
2
+dx
3
2
= dx . dx = g
u
du+ g
v
dv
( )
. g
u
du+ g
v
dv
( )
= g
u
.g
u
du
2
+2g
u
. g
v
dudv + g
v
. g
v
dv
2
dl is called the "metric element" on the surface.
Length element between two points P(u,v), S (u+du, v)
along a u coordinate curve is
dl
u
= g
u
. g
u
du =
!g
!u
du = h
1
du (dv = 0)
Similarly, along a v coordinate curve
dl
v
= g
v
. g
v
dv =
!g
!v
dv = h
2
dv (du = 0)
R (u+du,v)
h1du
h2dv
P(u,v)
T(u,v+dv)
74
!
If every u curve is perpendicular to every v - curve, we have an orthogonal
curvilinear coordinate system. A necessary condition is that
g
u
.g
v
= 0 ("
1
."
2
= 0 - tangent curves orthogonal)
In this case
dl
2
= h
1
2
du
2
+h
2
2
dv
2

75
!
Example : Parametric representation on the surface of a sphere
using co - latitude " and longitude #.
x = g(",#) = Rsin" cos#e
1
+ Rsin"sin#e
2
+ Rcos"e
3
T
1
=
$g
$"
= R(cos" cos#e
1
+cos"sin#e
2
%sin"e
3
)
T
2
=
$g
$#
= R(%sin"sin#e
1
+sin" cos#e
2
)
T
1
.T
2
= 0 & orthogonal curvilinear coordinates
h
1
=
$g
$"
= R
h
2
=
$g
$#
= Rsin"
&dl
2
= R
2
d"
2
+ R
2
sin
2
"d#
2
X1
X2
X3
R sin
R sin cos
!
76
!
Example : Parametric representation on the surface of a cylinder :
(",z) : " is plane polar angle, and z in vertical coordinate.
x = g(",z) = Rcos"e
1
+ Rsin"e
2
+ ze
3
T
1
=
#g
#"
= R($sin"e
1
+cos"e
2
)
T
2
=
#g
#z
= Re
3
T
1
.T
2
= 0 % orthogonal curvilinear coordinates
h
1
=
#g
#"
= R
h
2
=
#g
#z
=1
dl
2
= R
2
d"
2
+dz
2
P
z
!
R
X
Y
Z
! coordinate
curve
R coordinate
curve
z coordinate
curve
77 77
Example: On surface of sphere
x = g(!, ") = Rsin! cos"e
1
+ Rsin! sin"e
2
+ Rcos!e
3
g
!
= R(cos! cos"e
1
+cos! sin"e
2
!sin!e
3
)
g
"
= R(!sin! sin"e
1
+sin! cos"e
2
)
dS = (g
!
"g
"
)d!d" = R
2
sin!d!d"(sin! cos"e
1
+sin! sin"e
2
+cos!e
3
)
e
r
= sin! cos"e
1
+sin! sin"e
2
+cos!e
3

is unit vector in outward radial direction (by projection)
# dS = R
2
sin!d!d" e
r
h
1
=
#g
#!
= R
h
2
=
#g
#"
= Rsin!
Note: Since this is an orthogonal curvilinear coordinate system, we also
could have written directly
dS= h
1
h
2
d!d" = R
2
sin!d!d"
X1
X2
X3
R sin
R sin cos !
78
Element of area dS
Now, between adjacent points on surface,
dl = dx =
!g
!u
du+
!g
!v
dv = g
u
du+ g
v
dv
and along coordinate curves
dl
u
= g
u
du, dl
v
= g
v
dv
dS is the area of the parallelogram formed by dl
u
and dl
v
This can be seen to be
dS = dl
u
"dl
v
= (dl
u
!
1
)"(dl
v
!
2
) = g
u
"g
v
dudv
dS = dl
u
!dl
v
= (g
u
!g
v
)dudv = g
u
!g
v
dudv n
+
X
1

X
3

P(u,v) Q(u+du,v+dv)
X
2

dl
g(u,v)
g(u+dv,v+dv)
Vector element of area dS
S
n
+

u
u+du
v
v+dv
X
1

X
2

X
3

d
79
C
n
+
S
Orientable
An orientable surface is a two sided surface with one side specified as the outside
or positive side and the other side the inside or the negative side.
Consider a surface with a bounding curve C.
For a two sided surface we can assign a unit normal n at
every point of the surface such that the resulting vector field n(x) of
normals is continuous everywhere. For such a surface, when the foot of n describes
any closed curve not crossing C, the initial and final directions of n are the same.
Such a surface has two orientations (two sides) determined
by the normals n(x), -n(x).
80
If the direction of the bounding curve C ("the positive
orientation") induced by the parametrisation
and n(x) form a RH system, C is said to be positively directed with
respect to n(x). This choice of direction n(x) (written as n
+
(x)) determines
the positive side (that is the positive orientation) of the surface. It is
arbitrary to the extent that it depends on the prametrisation.
C is positively directed with respect to n
+
(x) in accompanying figure.
C
n
+
S
One orientation of a
two sided surface
81
The other orientation (negative side) of the same surface.
The eld of normals is now n
-

n-
C2
Note: We have assigned a direction to C
2
to make it positively
directed with respect to the normal n
- .
The direction of C
which is the natural direction that corresponds to the parametisation
is now negatively directed with respect to n
- .


82 82
If a surface is one sided, any two points on the surface can be connected
by a curve without crossing the boundary and when one returns to the starting
point the normals are in the opposite direction!
Twist
n1
n2
Not all surfaces are two sided, even though we can assign two normals to every
point on the surface.
Non orientable
-Mobius strip-
83
A piecewise smooth surface need not be orientable even if it can be
divided into components, each of which is separately orientable.
The Mobius strip is such an example (see next slide).

A sum of surfaces is orientable

(i) If each component is orientable
(ii) An orientation can be induced on each piece such that positive
direction on common boundaries are opposite.
n+
C1
The theorems that we derive will be applicable to orientable smooth surfaces
and more generally to orientable sums of smooth surfaces. 84
Two Orientations of S
n
+
P
n
-
C1
C2
n
+
n
+
Orientable Sums
n
-
n
+
Twist!
Mobius Strip-
-non orientable sum
85
Surface integrals can be of one of many types:
a(x). dS
S
!
, a(x)" dS
S
!
, !(x)dS
S
!
where a(x), is a vector field and !(x) is a scalar field.
Here
a(x). dS
S
!
= a(x).(ndS) =
S
!!
a
1
(x)n
1
dS
S
!
+ a
2
(x)n
2
dS
S
!
+ a
3
(x)n
3
dS
S
!
and
a(x)" dS
S
!
#
$
%
&
'
(
1
= (a
2
S
!
dS
3
)a
3
dS
2
) = (a
2
S
!
n
3
)a
3
n
2
)dS
where dS = ndS
All integrals can be built up from the following two basic types:
"(x)dS
S
!
and "(x)n
+
3
dS
S
!

Surface Integrals
86
Now, for a parametric surface
x = g(u, v)

the unit normal is
n =
g
u
!g
v
g
u
!g
v
=
1
g
u
!g
v
e
1
e
2
e
3
g
1,u
g
2,u
g
3,u
g
1,v
g
2,v
g
3,v
"
#
$
$
$
$
%
&
'
'
'
'
This is also taken to be n
+
(the direction induced by the parametrisation).
The vector element of area is
dS = ndS = (g
u
!g
v
)dudv
dS = dS = g
u
!g
v
dudv and
n
3
+
dS =
(g
u
!g
v
)
3
g
u
!g
v
g
u
!g
v
dudv =(g
1,u
g
2,v
(g
2,u
g
1,v
)dudv
87
Using the above
!
S
!!
(x)dS = ![g(u, v)]
"g
"u
"
"g
"v
S
uv
!!
dudv
and
!
S
!!
(x)n
3
+
dS = ![g(u, v)]
"g
1
"u
"g
2
"v
#
"g
2
"u
"g
1
"v
$
%
&
'
(
)
S
uv
!!
dudv
+ two similar integrals (with n
1
+
, n
2
+
)
If we have an explicit representation of a surface,
x
3
= f (x
1
, x
2
) x
1
,x
2
* S
x
1
x
2

the parametric representation is
x = g(x
1
, x
2
) = e
1
x
1
+e
2
x
2
+e
3
f (x
1
, x
2
) (x
1
= u, x
2
= v)
where S
x
1
x
2
is simply the projection of S onto the OX
1
X
2
plane.
S: x =g(u,v)
x1
x2
x3
P
n
+
C
Sx1x2
88
!
"g
"x
1
#
"g
"x
2
= e
1
$"f
"x
1
%
&
'
(
)
*
+e
2
"f
"x
2
%
&
'
(
)
*
+e
3
"g
"x
1
#
"g
"x
2
=
"f
"x
1
%
&
'
(
)
*
2
+
"f
"x
2
%
&
'
(
)
*
2
+1
dS =
"g
"x
1
#
"g
"x
2
%
&
'
(
)
*
dx
1
dx
2
dS
x1
x2
x3 n
+
dS cos !3
=dS n3
+
=
dx1dx2
!3
dS
!(x
S
!!
)dS = !
S
x1x2
!!
(x
1
, x
2
, f (x
1
, x
2
))
" f
"x
1
"
#
$
%
&
'
2
+
" f
"x
2
"
#
$
%
&
'
2
+1 dx
1
dx
2
!(x
S
!!
)n
3
+
dS = !
S
x1x2
!!
(x
1
, x
2
, f (x
1
, x
2
))dx
1
dx
2
89 89
As for parametric curves, the following theorem can easily
be proved for surfaces (left as an excercise!)
Theorem:
If S is an simple smooth orientable surface, we can unambigously evaluate
!(x
S
!!
)dS, !(x
S
!!
)n
3
+
dS
using any orientation preserving parametrisation of S. The sign
of the integrals will be reversed if we change the orientation.
90
!
EX: Find the surface area of the paraboloid of revolution z = 4 " x
2
" y
2
about the z axis for z > 0.
We use the explicit parametrisation (x, y) since the equation of the
surface is given in the form z = f (x, y). Any point on this surface
can be written in vector notation as
r(x, y) = xi + y j +(4 " x
2
" y
2
)k
#r
#x
= i " 2xk
#r
#y
= j " 2yk
Element of area is
dS = (
#r
#x
$
#r
#y
)dxdy = (2xi +2y j + k)dxdy """(outward)
dS = 4x
2
+ 4y
2
+1 dxdy """ Note : chosen to be positive
S = 4x
2
+ 4y
2
+1 dxdy
S
xy
%%
Z
X
Y
r(x,y)
Sxy
(0,0,4)
91
!
We do this integral by changing to
plane polar coordinates;
S
xy
" S
r#
S = 1+ 4r
2
0
2
$
0
2%
$ rdrd#
= 2%
1
12
(1+ 4r
2
)
3/ 2
&
'
(
)
*
+
0
2
=
%
6
(17
3/ 2
,1)
X
Y
r
!
r
!
0 2
2
"
Sr!
Sxy
2
92
n
a
P
Surface S
a.n
volume V
!
S : Closed surface about P with outward
unt outward normal n
V : Interior of S
a(x) : Vector field
(a. n)dS ="flux" of a across dS
a. n = flux of a per unit area of surface
Flux of a out of S is
a. dS
S
"" = a
S
"" . n dS (definition)
If a is velocity (m s
-1
), flux is rate of outflow of volume of fluid.

Dimension of a.dS
S
"" :
m
s
m
2
=
m
3
s
=
volume
time

Outward ux of a vector eld across a closed surface
93
If !(r) is the density of a fluid and v(r) is the velocity of a fluid,
the "mass flux" across element of area dS is ! v. n dS.
! v. n dS
S
!!

represents the total rate of outflow of mass across S and has
dimension [
Mass
time
].
One can also define upward and downward fluxes across open surfaces.
Ex: calculate the upward flux of the vector field
F=(i + j +(x
2
+ y
2
)k) across the unit disc x
2
+ y
2
=1.
Of the two normal directions of the disc we select n=k (upward).
F
unit disc
!!
.n dS = (i + j +(x
2
+ y
2
)k)
unit disc
!!
.k dS = (x
2
+ y
2
)
unit disc
!!
dS = r
2
0
2"
!
0
1
!
(rdrd#) =
1
4
2"
94
Coordinate Transformations in 3D Euclidean Space
A coordinate transformation is a mapping
x = f (u) = e
1
f
1
(u
1
, u
2
, u
3
) +... +e
3
f
3
(u
1
, u
2
, u
3
) ( f continuous)
with u ! D
u
" x = f (u) ! D
where f and D
u
are so chosen such that for every point u ! D
u
(1) The functions
! f
j
!u
k
are continuous
(2) The mapping is one to one ( f (u) = f (u
0
) #u = u
0
)
with the Jacobian J
J
f
1
, f
2
, f
3
u
1
, u
2
, u
3
$
%
&
'
(
) =
! f
1
!u
1
! f
1
!u
2
! f
1
!u
3
! f
3
!u
1
! f
3
!u
3
* 0
X1
X2
X3
u1
u2
u3
f(u)
D Du
95
Curvilinear coordinates, coordinate surfaces
and coordinate curves
!
Set u
1
= const = k
1
, (and let u
2
,u
3
vary).
x = f (k
1
,u
2
,u
3
) describes a two parameter (u
2
,u
3
) " coordinate surface.
Set u
1
= k
1
,u
2
= k
2
(and let u
3
vary)
x = f (k
1
,k
2
,u
3
) describes a one parameter u
3
" coordinate curve.
Suppose P # D and has parametric coordinates (u
1P
,u
2P
,u
3P
)
S
12P
: (u
1
" u
2
surface) x = f (u
1
,u
2
,u
3P
)
S
31P
: (u
1
" u
3
surface) x = f (u
1
,u
2P
,u
3
)
S
23P
: (u
2
" u
3
surface) x = f (u
1p
,u
2
,u
3
)
P
u1
u2
u3
S12P
C1P
C2P
C3P
96
!
C
1P
: (u
1
curve) x = f (u
1
,u
2P
,u
3P
) "1 =
# f
#u
1
# f
#u
1
C
2P
: (u
2
curve) x = f (u
1P
,u
2
,u
3P
) "2 =
# f
#u
2
# f
#u
2
C
3P
: (u
3
curve) x = f (u
1P
,u
2P
,u
3
) "3 =
# f
#u
3
# f
#u
3
P
u1
u2
u3
S12P
C1P
C2P
C3P
97
Element of length is given by
dl
2
= dx. dx =
! f
!u
1
du
1
+
! f
!u
2
du
2
+
! f
!u
3
du
3
!
"
#
$
%
&.
! f
!u
1
du
1
+
! f
!u
2
du
2
+
! f
!u
3
du
3
!
"
#
$
%
&
=
! f
!u
1
2
du
1
2
+
! f
!u
2
2
du
2
2
+
! f
!u
3
2
du
3
2
+(....)du
1
du
2
+(.........)du
2
du
3
+(.........)du
3
du
1
Note that in the (u
1
, u
2
, u
3
) coordinate system, the form of the metric element
of length may be such that it is not immediately obvious that the underlying
space is Euclidean ( flat).
That is, it may not be obvious that there is a transformation
(u
1
, u
2
, u
3
) '(w, s, t) such that
dl
2
= ds
2
+dt
2
+dw
2
The Element of Length
98
Element of length along u
k
curve is
d
k
l =
! f
!u
k
du
k
= h
k
du
k
(k =1,2,3)
h
k
=
! f
!u
k
= (
! f
1
!u
k
)
2
+.. +(
! f
3
!u
k
)
2
! 0
"
k
=
1
h
k
! f
!u
k
Here h
k
=
! f
!u
k
is the length scale factor.
"
1
.("
2
""
3
) =
1
h
1
h
2
h
3
{
! f
!u
1
#
! f
!u
2
"
! f
!u
3
} =
1
h
1
h
2
h
3
J
Note: J ! 0 $ "
1
, "
2
, "
3
are not coplanar, and none are null.
P
u1
u2
u3
S12P
C1P
C2P
C3P
!1
!2
!3
99
Any vector a can be resolved into components in directions
!
1
, !
2
, !
3
:
a ="
1
!
1
+"
2
!
2
+"
3
!
3
The components are the lengths of the sides of the
parellogram as in diagram.
(Note: For a general curvilinear coordinate system "
1
! a.!
1
)
a
!
1
!
2
!
3
Components of Vector
100 100
Volume of the elementary parallelopiped
with sides h
1
du
1
!
1
, h
2
du
2
!
2
, h
3
du
3
!
3
dV = (h
1
du
1
!
1
).(h
2
du
2
!
2
)!(h
3
du
3
!
3
)
= h
1
h
1
h
1
!
1
.(!
2
!!
3
) du
1
du
2
du
3
= J du
1
du
2
du
3
"dV =
"( f
1
, f
2
, f
3
)
"(u
1
, u
2
, u
3
)
du
1
du
2
du
3
The Element of Volume
101
Volume Integrals
(change of variable)
!
"(x)dV
D
###
= "(x)dx
1
D
### dx
2
dx
3
= "(x(u
1
,u
2
,u
3
)
D
u
### )
$(x
1
, x
2
, x
3
)
$(u
1
,u
2
,u
3
)
du
1
du
2
du
3
= "(x(u
1
,u
2
,u
3
)
D
u
### ) J du
1
du
2
du
3
102 102
If !
1
.(!
2
!!
3
) = 1 curvilinear coordinate system
orthogonal (RH:+1, LH: -1).

Arc length:
dl
2
= h
1
2
du
1
2
+h
2
2
du
2
2
+h
3
2
du
3
2
Jacobian: J = h
1
h
2
h
3
Element of volume:
dV = h
1
h
2
h
3
du
1
du
2
du
3
Volume integrals:
"
V
"""
(x)dV = "(x(u))
V
u
"""
h
1
h
2
h
3
du
1
du
2
du
3
Orthogonal Curvilinear Coordinates
103
!
Example : Consider the ellipsoid
x
2
a
2
+
y
2
b
2
+
z
2
c
2
=1
We can find the volume by changing variables
(x, y,z) "(u,v,w) via
x = au, y = bv,z = cw
In (u,v,w) space the ellipsoid is the sphere
u
2
+v
2
+w
2
=1
J =
#(x, y,z)
#(u,v,w)
=
#x
#u
#x
#v
#x
#w
. . .
. . .
=
a 0 0
0 b 0
0 0 c
= abc
V = dxdydz
ellipsoid
$$$ = abcdudvdw
unit sphere
$$$ =
4%
3
abc
104
Spherical Polar Coordinates
(orthogonal curvilinear)
!
Spherical Polars : u = (r,",#)
x = rsin" cos#e1 +rsin"sin#e2 +r cos"e3
$x
$r
= (sin" cos#,sin"sin#,cos")
$x
$"
= r(cos" cos#,cos"sin#,%sin")
$x
$#
= (%sin"sin#,sin" cos#,0)
h
1
=
$x
$r
=1
h
2
=
$x
$"
= r
h
3
=
$x
$#
= rsin"
Z
X
Y
!
P " coordinate
curve
! coordinate
curve
r coordinate
curve
" r
!
!
dV = r
2
sin" drd"d#
105
Cylindrical Polar Coordinates
(orthogonal curvilinear)
!
Cylindrical Polars : u = (R,",z)
x = Rcos"e1 + Rsin"e2 + ze3
#x
#R
= (cos",sin",0)
#x
#"
= R($sin",cos",0),
#x
#z
= (0,0,1)
h
1
=
#x
#R
=1
h
2
=
#x
#"
= R
h
3
=
#x
#z
=1
P
z
!
R
X
Y
Z
! coordinate
curve
R coordinate
curve
z coordinate
curve
!
dV = R dRd"dz
106
Exercise: Show that the volume of the region bounded above
by the sphere x
2
+ y
2
+ z
2
= a
2
and below by the cone
x
2
+ y
2
= tan
2
! z
2
is
2"a
3
3
(1!cos!)
Z
O
107
VECTOR CALCULUS

Part 2: Vector differential operators
108
The vector differential operator "Del" or "Nabla" is defined by it effect on a
scalar or vector functions;
!= e
1
!
!x
1
+e
2
!
!x
2
+e
3
!
!x
3
! operates on a scalar field "(x) as follows:
Gradient:
grad " = !"(x) =
(e
1
!
!x
1
+e
2
!
!x
2
+e
3
!
!x
3
)"(x) =e
1
!"
!x
1
+e
2
!"
!x
2
+e
3
!"
!x
3
So when "del" operates on a scalar field "(x), the result is the
vector field grad ".
Grad, div and curl in Cartesians
109
"del" operates on a vector field a(x) in two different ways, producing either
another scalar field, or another vector field:

Divergence: div a =!. a = (e
1
!
!x
1
+e
2
!
!x
2
+e
3
!
!x
3
). (a
1
e
1
+a
2
e
2
+a
3
e
3
)
=
!a
1
!x
1
+
!a
2
!x
2
+
!a
3
!x
3
Curl: curl a =!" a = (e
1
!
!x
1
+e
2
!
!x
2
+e
3
!
!x
3
) " (a
1
e
1
+a
2
e
2
+a
3
e
3
)
=
e
1
e
2
e
3
!
!x
1
!
!x
2
!
!x
3
a
1
a
2
a
3
= (
!a
3
!x
2
#
!a
2
!x
3
)e
1
+..........
The above definitions of grad, div and curl use cartesian components. However,
there are coordinate free definitions which allow us to interpret these in a
geometrical way;
110
!
Example :
F = r = xi + y j + zk
F = x
2
+ y
2
+ z
2
= r
divF = ".F =
#x
#x
+
#y
#y
+
#z
#z
= 3
div measures divergence
Example :
F = $yi + x j (2D field)
F = x
2
+ y
2
= r (in 2D)
curl F =
i j k
#
#x
#
#y
#
#z
$y x 0
= 0i +0 j +2k
= 2k
Curl measures rotation
X
Y
X
Y
111
!
" = "(x) - Scalar Field
Consider level (equipotential) surfaces
through neighbouring points P, # P
"(x) = "
P
, "(x) = "
P
+$"
P
n is unit normal in direction of " increasing :
Rate of change of " in direction n (unit normal)
%"
%n
= lim
NP&0
"
N
'"
P
NP
Rate of change of " in arbitrary direction u (unit vector)
%"
%u
= lim
# P P&0
"
# P
'"
P
# P P
= lim
NP&0
{
"
N
'"
P
NP
(u.n)}
=
%"
%n
(u.n)
(maximum rate of change of " is
%"
%n
(when u.n =1)
Grad: a coordinate free
denition
P
P
/
N
u
n
! = !P + "!P
! = !p
n, u
unit vectors
112
!
Define vector field by
"# =
$#
$n
n (n direction of # increasing)
Then
$#
$u
= u."#
"# is a vector perpendicular to # = const
at every point on surface. It is in the direction of
# increasing
n."# gives maximum rate of change of #

P
P
/
N
u
n
! = !P + "!P
! = !p
n, u
unit vectors
!
"#
"x
1
= e
1
.$#,
"#
"x
2
= e
2
.$#,
"#
"x
3
= e
3
.$#
%$# =
"#
"x
1
e
1
+
"#
"x
2
e
2
+
"#
"x
3
e
3
as before
! = const
!
"#
113
!
Ex : Calculate a unit normal to the surface
S: x
2
+ y
2
+ z
2
=1
Consider the scalar field "(x, y,z) = x
2
+ y
2
+ z
2
#1
The surface S corresponds to the equipotential "(x, y,z) = 0.
$The required direction is given by %" .
%" = (2x,2y,2z)
n =
%"
%"
=
1
4x
2
+ 4y
2
+ 4z
2
(2x,2y,2z) =
1
r
(x, y,z) =
r
r
n is in the radial direction, which is not surprising because S is a sphere.
Note : had we chosen "(x, y,z) =1- x
2
# y
2
# z
2
, we would have obtained
n = #
r
r
n
r
114
O
! = const
n
P
P0
Tangent Plane
r0
r
Tangent plane to a surface
!
Ex : Calculate equation of tangent plane to the surface x
2
+ 2y
2
+ xz = 6
at r
0
= (1,1, 3)
"(x, y, z) = x
2
+ 2y
2
+ xz #6
$" = (2x + z, 4y, x)
$"(r
0
) = (5, 4,1)
Equation of tangent plane is
(r # r
0
).$"(r
0
) = 0
5(x #1) + 4(y #1) +1(z # 3) = 0
5x + 4 y + z =12
115
n
a
P
Surface S
a.n
volume V
S : Closed surface about P with outward
unt normal n such that x-x
P
<!, x ! S
V : Interior of S
a(x) : Vector field (visualise as fluid velocity)
(a. n)dS ="flux" of a across dS
a. n = "flux" per unit area
[div a]
P
= lim
!"0

a. n dS
##
vol V
= lim
!"0

total outward flux of a through S
vol V
$
%
&
'
(
)
If a is velocity, outward flux is volume rate of outflow of fluid.


Divergence: coordinate free denition
div a measures divergence of uid ow per unit volume

116
Demonstration for a rectangular volume element
Q,R belong to S, x
1R
= x
1Q
+!x
1
"a
1
"x
1
!
"
#
$
%
&
Q
=
"a
1
"x
1
!
"
#
$
%
&
P
+o(1),
"a
1
"x
1
!
"
#
$
%
&
R
=
"a
1
"x
1
!
"
#
$
%
&
P
+o(1)
'a
1R
(a
1Q
=
"a
1
"x
1
!
"
#
$
%
&
P
!x
1
+o(!x
1
)
lim
rate of outflow
volume
)
*
+
,
-
.
= lim
(
"a
1
"x
1
)
P
!x
1
+o(!x
1
)
/
0
1
2
3
4
!x
2
!x
3
+two similar expressions
!x
1
!x
2
!x
3
)
*
+
+
+
+
,
-
.
.
.
.
=
"a
1
"x
1
+... +
"a
3
"x
3
= div a
limit independent of shape of boundary (later)
Q R
a1R a1Q
!x1
!x2
!x3
!x2
!x3
P
117
Curl: coordinate free denition
a
!
a.! C
m
P
!
m: unit vector in some direction
C : boundary of a (small) plane surface
having area A normal to m and such that
x - x
p
< ", x # C
$ : unit tangent to C with direction related to m
by RH rule
Then
m. (curl a) = lim
"%0
1
A
a.$
C
& dl (dl element of length)
= lim
"%0
{
circulation of a about closed loop C
area A
}
Exercise: Show this for a rectangular area
Curl a measures circulation per unit area
Curl
118
!
Example 1: v = (x,0,0)
div v = ".v =1, curl v = "#v = 0
Excess of flux leaving region ABCD
No rotation imparted to region.
Note : v = "(
1
2
x
2
) - -can be derived from scalar potential
Example 2 : v = (y,0,0)
div v = ".v = 0, curl v = "#v = (0,0,$1)
Flux leaving region balances flux
entering region.
Rotation about - Z direction
No scalar potential exists for flow.
Example 3: v = (
x
x
2
+ y
2
,
y
x
2
+ y
2
,0)
div v = ".v = 0 (except at origin)
curl v = "#v = 0
Zero net flux and rotation
v = "[
1
2
ln(x
2
+ y
2
)]
X
Y
X
Y
Y
X
119
!= e
1
!
!x
1
+e
2
!
!x
2
+e
3
!
!x
3
(cartesians)
! behaves essentially like b in b f , b. a, b"a but since ! is a differential
operator, the sequence in which the symbols are written matter. Thus
(a. b) c = c (a. b),
but (a.!) c # (!. a) c
(a.!) c = (a
1
!
!x
1
+a
2
!
!x
2
+a
3
!
!x
3
) (e
1
c
1
+e
2
c
2
+e
3
c
3
)
= (a
1
!c
1
!x
1
+a
2
!c
1
!x
2
+a
3
!c
1
!x
3
)e
1
+two similar terms
(!. a) c = (
!a
1
!x
1
+
!a
2
!x
2
+
!a
3
!x
3
) (e
1
c
1
+e
2
c
2
+e
3
c
3
)
Some properties of del
120
!
If " and# are scalar fields, a and b are vector fields and
$ and % are scalars
& ($" +%#) = $&" +%&#
&. (a +b) = &. a +&. b
&'(a +b) = &' a +&' b
These, and other identities can be proved by
taking components
Differentiation of Sums
121
!
div ("a) = " diva +grad". a #. ("a) = "#.a +#". a
curl ( "a) = " curla +grad" $a #$("a) = "#$a +#" $a
div (a$b) = b.curla - a.curl b #. (a$b) = (#$a).b - (#$b).a
curl (a$b) = a divb - b diva +(b .grad)a - (a .grad)b
grad (a.b) = a $curlb +b $ curla +(b .grad)a +(a .grad)b
#$(a$b) = (b.#)a % b(#.a) %(a.#)b+a(#.b)
#(a.b) = b$(#$a) +a$(#$b) +(b.#)a +(a.#)b
Differentiation of Products
122
!
!
Example of proof by components :
[curl "a]
1
=
#("a
3
)
#x
2
$
#("a
2
)
#x
3
=
#"
#x
2
a
3
+"
#a
3
#x
2
$
#"
#x
3
a
2
$"
#a
2
#x
3
= "(
#a
3
#x
2
$
#a
2
#x
3
) +(
#"
#x
2
a
3
$
#"
#x
3
a
2
)
= "[curl a]
1
+[grad" %a]
1
&curl "a = " curl a +grad" %a
123
Example: !
1
r
=
d
dr
(
1
r
) r = "
1
r
2
r
Example:
curl (
1
r
2
r) =
1
r
2
curlr +grad(
1
r
2
)#r
=
1
r
2
0 "
2
r
3
r
r
#r = 0
Exercise: If f (r) = f (r) show that
!f =
df
dr
r
r
=
df
dr
r
where r is a unit vector in the radial r direction.
124
Second Order Operators
Starting with a vector field a(x) and a scalar field !(x)
we have constructed
First order operators: div a, curl a and grad!
We can differentiate again and construct
Second order operators: grad (div a)
div (curl a)
curl (curl a)
curl (grad !)
div (grad !)
Some operations like curl( div a) do not make sense
in vector calculus (may in Tensor calculus!)
125
LAPLACIAN : !
2
!
2
! = div grad ! = !.(!!) --- (coordinate free definition)
=
"
2
!
"x
1
2
+
"
2
!
"x
2
2
+
"
2
!
"x
3
2
--- (in RC system)
!
2
= (
"
2
"x
1
2
+
"
2
"x
2
2
+
"
2
"x
3
2
) when operating on a scalar field and is the "Laplacian".
When !
2
operates on a vector field it is defined by its action as
!
2
a = grad (div a)-curl(curl a) ----- (coordinate free definition)
Caution ...
!
2
a = (!
2
a
1
, !
2
a
2
, !
2
a
3
) only in RC system. In any other system,
!
2
a = grad (div a)-curl(curl a), where in evaluating grad and div
one must take into consideration the fact that the unit vectors are
functions of positions in space. The expressions for grad, div and
curl in a general orthogonal curvilinear coordinate system are given
later .
The Laplacian

126
!
Proof (by taking components in RC system)
"# = (
$#
$x
1
,
$#
$x
2
,
$#
$x
3
)
"%a =
e
1
e
2
e
3
$
$x
1
$
$x
2
$
$x
3
a
1
a
2
a
3
&["%"#]
1
=
$
$x
2
$#
$x
3
'
(
)
*
+
,
-
$
$x
3
$#
$x
2
'
(
)
*
+
,
= 0
similarly for other two components
&curl grad # = "%"# = 0
!
curl (grad ") = #$#" = 0
div (curl a) = #.#$a = 0
Two important identities
127
Laplaces and Poissons equation occurs in many branches of applied
mathematics.
Electrostatics
Gravitation theory
Incompressible Fluid Flows
where is a scalar potential function that describes the particular
vector eld
(e.g. gravitational potential, electrostatic potential etc.)
Discussed in partial differential equations course.
Two partial differential equations of
mathematical physics involving the Laplacian
!
"
2
#(x, y,z) = 0 Laplace' s Equation
"
2
#(x, y,z) = $(x,y,z) Poisson' s Equation
where $(x,y,z) is a source term.
!(x, y, z)
128
Vector Integral Theorems
In evaluating definite integrals, we effectively "take out" one dimension

df
dx
a
b
!
dx = f (b) " f (a)
"a 1-D integral along a line is reduced to evaluating a function at the
end points of the line

S
!!
#
C
!
""""Stokes theorem
-a 2-D surface integral across an open surface S is reduced to evaluating a 1-D line integral
over the bounding curve C

V
!!!
#
S
!!
----- Divergence theorem or Gauss's theorem
-a 3-D volume integral across a volume V is reduced to evaluating a 2-D surface integral
over the bounding surface S
We prove these theorems by first proving Green's theorem in the plane.
129
Greens Theorem in the Plane
C: Closed positively directed plane curve
S: Union of C and its interior
Under suitable restrictions
!"
!x
!
!#
!y
"
#
$
%
&
'
S
((
dxdy = # dx +"dy
C
!
(
We prove the result for
(a) standard domain (convex) region with simple piecewise smooth
boundary curve C with #, " )C
1
on S (functions and first partial derivatives
continuous)
The result follows also
(b) for sum of standard domains S
n
where the boundary C
n
is
simple piecewise smooth with #, " )C
1
on S
n
, #, " )C
0
on S.
130
!
Proof of (a) : Standard domain (convex : x simple and y simple)
g
1
(x) " y " g
2
(x), f
1
(y) " x " f
2
(y) describe C
#dx = #dx + #dx
C2
$
C1
$
C
$
= #(x, g
1
(x))dx +
a
b
$
#(x, g
2
(x))dx
b
a
$
= % #(x, g
2
(x) % #(x, g
1
(x) [ ]dx
a
b
$
= %
&#
&y
dy
g1 (x)
g2 (x)
$
'
(
)
)
*
+
,
,
dx
a
b
$
= %
&#
&y
dxdy
S
$$
Similarly
-dy =
C
$
+
&-
&x
dxdy
S
$$
Hence
#dx +-dy =
C
$
&-
&x
%
&#
&y
'
(
)
*
+
,
dxdy
S
$$
C2
x=f1(y)
x=f2(y)
X
C1
c
d
C2
y=g1(x)
y=g2(x)
X
Y
a b
C1
131
Example : Evaluate
I = (y
2
+sin x
2
)dx +(cos y
2
!
!
" x)dy
over the square C: 0 # x #1: 0 # y #1
I =
$
$x
(cos y
2
" x) "
$
$y
(y
2
+sin x
2
)
%
&
'
(
)
*
S
!!
dxdy (using Green's Theorem)
= ("1"2y)dx dy = "
0
1
!
0
1
!
(1+2y)
0
1
!
dy = "2
Note: In this case area integral is easier to evaluate than
line integral!
132
!
(Proof of b) :
For more complicates regions, divide into sub- regions each
of which is standard (convex).
S = S
1
+ S
2
+ ...S
n
We show result for two regions S
1
and S
2
.
"#
"x
$
"%
"y
&
'
(
)
*
+
dxdy
S1 +S2
,,
=
"#
"x
$
"%
"y
&
'
(
)
*
+
dxdy
S1
,,
+
"#
"x
$
"%
"y
&
'
(
)
*
+
dxdy
S2
,,
= %dx +#dy +
C1
,
%dx +#dy +
C3
,
%dx +#dy +
C4
,
%dx +#dy
C2
,
But contributions to line integrals from common boundaries
cancel (because %,# - C
0
). Thus
%dx +#dy = $ %dx +#dy
C4
,
C3
,
.
"#
"x
$
"%
"y
&
'
(
)
*
+
dxdy
S1 +S2
,,
= %dx +#dy +
C1
,
%dx +#dy
C2
,
Similarly for more than two regions.
S1
S2
C1
C2
C3
C4
133
Example: Verify Green's theorem for y
2
dx +2x dy
C
!
!
x
2
+ y
2
= 9.
y
2
dx +2x dy
C
!
!
=
!
!x
(2x) "
!
!y
(y
2
)
#
$
%
&
'
(
S
!!
= 2 "2y [ ]
S
!!
dxdy
To evaluate the area integral, use plane polar coordinates;
x = r cos", y = rsin", dA = rdrd"
RHS: 2 "2y [ ]
S
!!
dxdy = (2 "2.3rsin")r dr d"
0
3
!
=
0
2#
!
(9 "18sin
0
2#
!
")d" =18#
To evaluate the line integral, parametrise circle by
x = 3cost, y = 3sint (0 ) t ) 2#)
LHS: y
2
dx +2x dy
C
!
!
= 9sin
2
t("3sint)dt +
0
2#
!
2.3cost.3cost dt = 0 +18#
0
2#
!
X
Y
3
3
S
C
134
Example: Show that for a closed plane curve C
-ydx + xdy
x
2
+ y
2
C
!
!
= 2! -if C encircles origin
= 0 -otherwise
" =
-y
x
2
+ y
2
, # =
x
x
2
+ y
2
""
"y
=
y
2
# x
2
(x
2
+ y
2
)
2
,
"#
"x
=
y
2
# x
2
(x
2
+ y
2
)
2

First order partial derivatives are continuous except at (0,0).
Expect to be able to apply Green's theorem to any region
not containing origin.

X
Y
C
135
Case 1: C excludes origin;
!
-ydx + xdy
x
2
+ y
2
C
!
"
=
= ! dx
C
!
"
+!dy = (
#!
#x
S
""
$
#"
#y
)dxdy
= [
y
2
$ x
2
(x
2
+ y
2
)
2
S
""
$
y
2
$ x
2
(x
2
+ y
2
)
2
]dxdy = 0
136
X
Y
C
C0
L
1

L
2

!
Case 2 : C encloses origin. In this case construct a closed contour which excludes
the origin (see diagram). Here we take C
0
to be a small circle of radius a.
-ydx + xdy
x
2
+ y
2
C+L1+L2+C0
" = 0 becase C+ L
1
+ L
2
+C
0
excludes O.
-ydx + xdy
x
2
+ y
2
C
" = #
-ydx + xdy
x
2
+ y
2
C0
"
On circle set x = acost, y = asint (Note : t decreases from 2$ to 0)

-ydx + xdy
x
2
+ y
2
C
" = #
#asint(#asint) +acost (acost)
a
2
2$
0
" dt
= 2$
Note : C+C
0
is traversed so that the S is to the "left " of C
(positively directed - with respect to enclosed area which has
a normal in the k direction).
137
Divergence Theorem (3D)
!
S : Closed surface outward normal n
V : Volume of S and its interior
"#
"x
k
V
$$$ dV = n
k
#
S
$$ dS (k =1,2, 3)
Under suitable restrictions :
[ Rule :
"
"x
k
on V becomes n
k
on S]
Notation : As before
f (x) !C
m
on domain D if f has continuous partial
derivatives (formed by values of f on D only) of orders 0,1,.....m
(e
1
, e
1
, e
1
) "(i, j, k)
(x
1
, x
2
, x
3
) "(x, y, z)
138
!
V : f (x, y) " z " g(x, y), f ,g continuous on S
xy
+ two similar expressions describe V
#$
#z V
%%% dxdydz = dxdy
S
xy
%%
#$
#z
dz
f
g
%
= $(x, y,g)dxdy
S
xy
%% & $(x, y, f )dxdy
S
xy
%%
n
z
dS = dxdy on z = g(x, y)
n
z
dS = &dxdy on z = f (x, y)
'
#$
#z V
%%% dxdydz = n
z
$
S
%% dS
Similarly
#
#x V
%%% ,
#
#y V
%%%
X
Y
Z
n = n
+ (
nz > 0)
n = n
- (
nz< 0 )
z = g(x,y)
z = f(x,y)
S
xy
Standard Domain
139
!
Alternative form:
Set " = a
k
(k =1,2,3)
a = (a
1
,a
2
,a
3
) = (a
x
,a
y
,a
z
)
#a
1
#x V
$$$ dxdydz = n
1
a
1
S
$$ dS
#a
2
#y V
$$$ dxdydz = n
2
a
2
S
$$ dS
#a
3
#z V
$$$ dxdydz = n
3
a
3
S
$$ dS
and sum;
(
#a
1
#x
+
#a
2
#y
+
#a
3
#z V
$$$ )dxdydz = (a.n)
S
$$ dS
!
diva
V
""" dV = (n.a)dS
S
"" = a.dS
S
""
140

(a) Proved for standard domain V
(b) True for sum of standard domains V
n
:
Apply result to each V
n
. Integrals along common boundaries cancel.
(c) Sufficient Restrictions:
Boundary S
n
of each V
n
is orientable sum of simple smooth surfaces
! !C
1
on each V
n
! !C
0
on V
Finite number of standard domains
V1 V2
141
!
Example : A surface S encloses a volume V.
Show that the volume V is given by
V =
1
3
(r.n)dS
S
""
where r is the position vector.
We use the divergence theorem:
1
3
(r.n)dS
S
"" =
1
3
r.dS
S
""
=
1
3
div r dV
V
"""
=
1
3
3 dV
V
"""
= dV """ = V
142
X Y
!
z=1-x-y
y=1-x
Example: Verify the divergence theorem for the tetrahedron bounded
by the coordinate planes and the plane x+y+z=1 and the vector field
A=3x
2
i + xy j + zk
div A = 6x + x +1= 7x +1
(7x +1)
V
!!!
dV = (7x +1)dz dydx =
11`
24
0
1"x"y
!
0
1"x
!
0
1
!
A.ndS
S
xy
!!
= (3x
2
S
xy
!!
i + xy j +ok).("k)dS = 0
A.ndS
S
yz
!!
= (0
S
yz
!!
i + xy j + zk).("i)dS = 0
A.ndS
S
zx
!!
= (3x
2
S
zx
!!
i +0 j + zk).("j)dS = 0
143
!
On inclined plane, we parametrise explicitly in terms of (x, y) :
Equation of plane is
r = xi + y j +(1- x - y)k
"r
"x
= i # k,
"r
"y
= j # k
"r
"x
$
"r
"y
= i + j + k
dS = (i + j + k)dxdy
A.
S
%% dS = (3x
2
S
xy
%% i + xy j + zk).(i + j + k)dxdy
= (
0
1#x
%
0
1
% 3x
2
+ xy +(1# x # y))dydx =
11
24
144
2-D divergence theorem from Greens Theorem
Consider a vector field a = i! ! j" and a closed 2D curve C enclosing a surface S
div
S
""
a dxdy = a
C
!
"
.!dl (Divergence Theorem-2D)
Now unit tangent " =
dr
dl
= i
dx
dl
+ j
dy
dl
. Hence unit normal is
! = i
dy
dl
! j
dx
dl
(" #k =!)
and div a =
#$
#x
!
#%
#y
[
#$
#x
!
#%
#y
]
S
""
dxdy = [i$ ! j%]
C
!
"
.[i
dy
dl
! j
dx
dl
]dl (2D divergence T)
[
!"
!x
!
!#
!y
]
S
""
dxdy = # dx +
C
!
"
!dy (Green's Theorem in the Plane!)
X
y
C
!
"
145
2D Stokes Theorem from Greens theorem
X
y
C
!
"
Let a be a 2-D vector field given by
a = i!(x, y) + j"(x, y) and let k be unit vector in z direction.
curla = (0, 0, !
#!
#y
+
#"
#y
)
k. curla dx dy
S
""
= (!
#!
#y
+
#"
#y
)dx dy
S
""
= ! dx
C
!
"
+! dy (using 2D Green's Theorem)
= (a." )
C
!
"
dl
= a.
C
!
"
dl
For the enclosing area in the xy plane, dS = k dxdy
#k. curla dx dy
S
""
= curla . dS
S
""
= a.
C
!
"
dl
This is 2D Stoke's Theorem which generalises to 3D as we now show.
146 146
S: orientable smooth surface with a field of normals n
C: Edge of S positively directed with respect to n
Under suitable restrictions:
(i) n!"!
S
##
dS = ! dx
C
!
#
= ! " dl
C
!
#
(ii) ("!a
S
##
).dS = a.! dl
C
!
#
$$$$$(standard form)
Stokess Theorem (3D)
(The Curl Theorem)
X
Y
Z
n
Surface S
Boundary C
!
a
147
div (curl a) = 0 identically
! curl
S
""
a.dS = div curl
V
"""
a dV = 0 - any closed S
Let S=S
1
#S
2
: C common boundary
curl
S
1
""
a. ndS + curl a. ndS
S
2
""
= 0
curl
S
1
""
a. n
1
dS + curl a. (-n
2
)dS
S
2
""
= 0
curl
S
1
""
a. n
1
dS = curl
S
1
""
a. n
2
dS
With n
2
taken as the inward normal, C is positively
directed with respect to both surfaces.
$Surface integral over orientable surfaces with same boundary curve
must have same value!. Therefore the value can depend only on integral
over the boundary curve.
X
Y
Z
n1
Surface S1
C
Surface S2
n2 Volume V
A plausibility argument
148
X
Y
Z
n1
Surface S1
C
Surface S2
n2 Volume V
Proof of 3D Stokes theorem
Consider a simple smooth surface S
Let x = g(u, v) be a mapping which shows it to be so.
Then the unit "outward" normal is
n
+
=
x
u
!x
v
x
u
!x
v

and the element of area is
dS = x
u
!x
v
dudv


150
X
Y
Z
n
Surface S
C
!
Suv
Cuv
u
v
Then
n
+
!"!dS =
x
u
!x
v
x
u
!x
v
!"! x
u
!x
v
dudv
=[(x
u
."!)x
v
#(x
v
."!)x
u
]dudv
[n
+
!"! ]
i
dS =[(x
u
."!)
$x
i
$v
#(x
v
."!)
$x
i
$u
]dudv
=
"!
"x
k
"x
k
"u
"x
i
"v
#
"!
"x
k
"x
k
"v
"x
i
"u
%
&
'
(
)
*
dudv (implicit summation over k)
=
"!
"u
"x
i
"v
#
"!
"v
"x
i
"u
=
"
"u
(!
"x
i
"v
) #
"
"v
(!
"x
i
"u
)
151 151
n
+
!"! dS
S
##
$
%
&
'
(
)
i
=
"
"u
(!
"x
i
"v
) *
"
"v
(!
"x
i
"u
)
$
%
&
'
(
)
S
u
##
dudv
= (
C
u
!
#
!
"x
i
"u
du+!
"x
i
"v
dv) (by Green's Theorem for plane)
= ! dx
i
C
!
#
+ n
+
!"! dS
S
##
= !dx
C
!
#
(ii) For an orientable sum of simple smooth surfaces, apply the result to each,
integrals along common boundaries cancel.
Sufficient Restrictions: ! ,C
1
on each S
n
, ! ,C on S
152
(iii) Set ! = a
1
, take the 1
st
component of above.
Now (n!"!) =
e
1
e
2
e
3
n
1
n
2
n
3
"a
1
"x
1
"a
1
"x
2
"a
1
"x
3
#
$
%
%
%
%
%
%
&
'
(
(
(
(
(
(
) [n
2
"a
1
"x
3
*n
3
"a
1
"x
2
]dS
S
++
= a
1
dx
1
C
!
+
! = a
2
, take the 2
nd
component
[n
3
"a
2
"x
1
*n
1
"a
2
"x
3
]dS
S
++
= a
2
dx
2
C
!
+
and similarly
[n
1
"a
3
"x
2
*n
2
"a
3
"x
1
]dS
S
++
= a
3
dx
3
C
!
+

Adding
[n
1
S
++
(
"a
3
"x
2
*
"a
2
"x
3
) +n
2
(
"a
1
"x
3
*
"a
3
"x
1
) +...]dS = (a
1
dx
1
+a
2
dx
2
C
!
+
+..)
n. curla
S
++
dS = a. dx
C
!
+
= a
C
!
+
.#dl
153
The divergence theorem can be used to justify the
coordinate free denition of divergence (recall we previously
used the special case of a rectangular volume element to demonstrate
its plausibility)
Likewise, Stokess theorem can be used to justify the
coordinate free denition of curl.
154
!
Consider a surface S given by the part of the sphere
x
2
+ y
2
+ z
2
=1
bounded by the planes z = 0, z =
1
2
; a = zi + x j + yk
Unit outward normal (radial) :
n = sin" cos#i +sin"sin# j +cos"k (
$
3
%" %
$
2
,0 % # % 2$)
dS =
&r
&"
'
&r
&#
d"d# = sin"d"d#
(('a
S
)) ).ndS = (i + j + k).(sin" cos#i +sin"sin# j +cos"k)
S
"#
)) sin"d"d#
= (
0
2$
)
$ 3
$ 2
) sin
2
" cos# +sin
2
"sin# +sin" cos")d"d#
= 2$ sin" cos"
$ 3
$ 2
) d" =
$
4
Z
X
Y
C1
C2 n
n
!/3
Example: 3D Stokes theorem
155
!
Now consider the line integrals along C
1
and C
2
.
On C
1
: r = costi + sin t j + 0k (t : 0 "2#)
a
C
1
$
.dr = a
0
2#
$
.
dr
dt
dt = (0i + cost j + sintk).(
0
2#
$
%sin ti + cost j + 0k)dt
= cos
2
0
2#
$ tdt = #
On C
2
: x
2
+ y
2
= (
3
2
)
2
r =
3
2
costi +
3
2
sint j +
1
2
k (t : 2# "0)
a
C
1
$ .dr = a
2#
0
$ .
dr
dt
dt = %
3#
4
(check!)
Hence sum is
#
4
Z
X
Y
C1
C2 n
n
156
!
Now consider the line integrals along C
1
and C
2
.
On C
1
: r = costi + sin t j + 0k (t : 0 "2#)
a
C
1
$
.dr = a
0
2#
$
.
dr
dt
dt = (0i + cost j + sintk).(
0
2#
$
%sin ti + cost j + 0k)dt
= cos
2
0
2#
$ tdt = #
On C
2
: x
2
+ y
2
= (
3
2
)
2
r =
3
2
costi +
3
2
sint j +
1
2
k (t : 2# "0)
a
C
1
$ .dr = a
2#
0
$ .
dr
dt
dt = %
3#
4
(check!)
Hence sum is
#
4
Z
X
Y
C1
C2 n
n
157
Greens Theorems (I and II)
If ! and " are scalar fields, the following theorems follow;
(!!
2
"
V
"""
+!!.!")dV = !
#"
#n
S
""
dS
where ! #C
1
piecewise on V, " #C
2
piecewise on V
(!!
2
"
V
"""
$"!
2
!)dV = (!
#"
#n
S
""
$"
#!
#n
)dS
where !," #C
2
piecewise on V
158
!
("#
2
$
V
%%% +#".#$)dV = "
&$
&n S
%% dS
Using
div(" a) = " diva +#" .a
" div(grad$) = div(" grad$) 'grad".grad$
" div(grad$
V
%%% )dV = [div(" grad$) 'grad".grad$]dV
V
%%%
= " grad$.dS
S
%% ' grad".grad$dV
V
%%%
( (" #
2
$
V
%%% +#".#$)dV = "
&$
&n S
%% dS
Proof of rst theorem
159
An application of integral Identities
!
Equation of continuity in fluid dynamics :
Rate of increase of mass of material within
a fixed surface S = rate of flow of mass (mass flux) intoS
i.e.
"
"t
#
V
$$$ dV =
"#
"t V
$$$ dV = % (#v
S
$$ .n)dS
(
V
$$$
"#
"t
+div #v)dV = 0 applying divergence theorem.
But above is true for every volume V.
"#
"t
+div #v = 0%%%%% at all points of material.
n
S
V
160
Irrotational (or conservative) vector elds

Denition: F(x) is said to be irrotational (or curl free) in a domain D


if curl F=0 in D
!
If F(x)("C
1
) is irrotational on a simply connected volume D, then
there exists a scalar potential # such that F = $#. Conversely,
if there exists # such that F = $#, then F is irrotational.
(a) suppose $%F = 0 (that is, F is irrotational)
Take a fixed reference point B" D, and define
#(xP ) = F.dx
B
P
&
# is independent of the path joining B and P because
F.dx
BMP
& ' F.dx
BNP
& = F.dx
BMPNB
& = n
S
&& .curlF dS (Using Stoke' s Theorem)
= 0
B
P
S
N
M
Theorem
161
B
P
Q
X1
X2
X3
!x
1
!
curl F = 0 " F = #$
!
Now consider neighbouring points P and Q where PQ is
in the x
1
direction :
"#
"x
1
$
%
&
'
(
)
P
= lim
#
Q
*#
P
PQ
= lim
+x
1
,0
1
+x
1
F
1
(-,
x
1P
x
1P
++x
1
. x
2P
, x
3P
)d- = F
1
(x
P
)
(using MVT)
Similarly for the other two components of /#
0F = /#
(b)Now suppose F = /#
Then we have, using the standard identity,
/1F = /1/# = 0
162
Uniqueness
!
Suppose " # is another potential for the vector field F.
Then F = $ " # and F = $# . Writing U = # % " # we see that
$U = 0
or
&U
&x
=
&U
&y
=
&U
&z
= 0
'U = # % " # = a constant.
The scalar potential is unique apart for an arbitrary constant.
163
An equivalent theorem
F is irrotational on a simply connected domain D
if and only if the work integral
F.dx
C
AB
!
is independent of path connecting A and B,
for all pairs of points A,B in D.
Note: In a more general form of the above theorems,
F is allowed to be undefined at "exceptional points"
(a finite number). A scalar potential still exists, but it is
also not defined at this point (cf: force field of gravity)
164
Irrotational elds (example)
X
Y
Example: Given a = (x, 0, 0) show that a is
irrotational, and find a potential that describes the field.
curl a =
i j k
!
!x
!
!y
!
!z
x 0 0
= 0
Consider a field point P (x,y,z) and take as
reference point P
0
(0,0,0).
Take C: r = (tx, ty, tz) (0 ! t !1) " (straight line joining P
0
and P)
"(x,y,z) = a
C
#
. dr = a(r(t))
t=0
t=1
#
.
dr
dt
dt
= (tx, 0, 0).(x, y, z)
t=0
t=1
#
dt = x
2
t dt
0
1
#
=
1
2
x
2
165
(b) Integrate PDE s directly
Set a = !! "
"!
"x
= x,
"!
"y
= 0,
"!
"z
= 0
"!
"x
= x
! =
1
2
x
2
+ g(y, z)
"!
"y
= 0 "
"g(y, z)
"y
= 0 "g = c +h(z)
#! =
1
2
x
2
+c +h(z)
"!
"z
= 0 "
dh
dz
= 0 "h = const
! =
1
2
x
2
+d
166
Solenoidal Fields and the Vector Potential
Suppose F ( !C
1
) is a solenoidal fileld (div F = 0) on a
domain D. If D has the property that every closed surface
in D bounds a domain contained in D (i.e not doughnut type!),
then " a vector potential A such that F = curl A.
Conversely, if F = curl A, then F is solenoidal.
Note: In this case, the vector field is not allowed
to have exceptional points. (- thus there is no vector
potential for the gravitational field F(r) =
r
r
3
).
Pr oof (part b):
Clearly if F = curl A, then div F = div(curl A) = 0
Hence F is solenoidal.
Denition: A vector eld F is solenoidal (or divergence free) in a
Domain D iff div F=0 in D
Theorem
167
!
Proof (Part a) : Suppose div F = 0.
Suppose
"(y, z) = F
1
(0, y, z)
is the values of the x - component of F, at the
point(0, y, z) of the plane x = 0, and F
2
and F
3
are
the y and z components of F.
We now define a vector field A by the following
equation (Note A
x
= 0)
A(x, y, z) =[ F
3
( # x , y, z)d # x
0
x
$
% "(y, # z )d # z ] j
0
z
$
%[ F
2
( # x , y, z)d # x
0
x
$
]k
where F
2
dx
0
x
$
and F
3
dx
0
x
$
are the integrals of F
2
(x, y, z) and F
3
(x, y, z)
with (y, z) fixed from Q(0, y, z) to the point P(x, y, z), and "dz
0
z
$
is the integral
from R(0, y,0) to the point Q(0, y, z).
O
x
y
z
R(0,y,0)
Q(0,y,z)
P(x,y,z)
168
!
Now
A(x, y, z) = ( F
3
( " x , y, z)d " x
0
x
#
$ %(y, " z )d " z ) j
0
z
#
$( F
2
( " x , y, z)d " x
0
x
#
)k
&' A = (
(A
z
(y
$
(A
y
(z
)i + (
(A
x
(z
$
(A
z
(x
) j + (
(A
y
(x
$
(A
x
(y
)k
(A
z
(y
$
(A
y
(z
=
(
(y
(- F
2
( " x , y, z)d " x
0
x
#
) $
(
(z
( F
3
( " x , y, z)d " x
0
x
#
$ %(y, " z )d " z )
0
z
#
= -
(F
2
( " x , y, z)
(y
d " x
0
x
#
$
(F
3
( " x , y, z)
(z
d " x
0
x
#
+%(y, z)
(A
x
(z
$
(A
z
(x
= F
2
(A
y
(x
$
(A
x
(y
= F
3
169
Curl A = (!
!
!y
F
2
0
x
"
dx !
!
!z
F
3
0
x
"
dx +")i + F
2
j + F
3
k
= ! (
!F
2
!y
+
!F
3
!z
)dx !"
0
x
"
#
$
%
&
'
(
i + F
2
j + F
3
k (!F continuously differentiable)
But divF =
!F
1
!x
+
!F
2
!y
+
!F
3
!z
= 0
)Curl A =
!F
3
!x
dx +"
0
x
"
#
$
%
&
'
(
i + F
2
j + F
3
k
= ( F
1 [ ]
0
x
+!)i + F
2
j + F
3
k (using !(y,z) = F
1
(0,y,z))
= F
1
i + F
2
j + F
3
k
Uniqueness:
If F = *+A and F = *+ , A then
*+( , A ! A) = 0
Hence there exists a scalar field - such that
, A ! A = *-
) , A = A+*-
170
As for the case of scalar potentials, vector potentials can also be obtained by solving
PDEs. Suppose v is solenoidal (i.e. div v=0). We have to solve the PDEs
!A
3
!y
!
!A
2
!z
= v
1
!A
1
!z
!
!A
3
!x
= v
2
!A
2
!x
!
!A
1
!y
= v
3
We have 3 equations relating the 9 independent partial derivatives
of the 3 components of A. If we use them all we have 6 degrees of
freedom at our disposal in finding a solution, which can be used in many ways.

All we need to show is that we can find a potential satisfying above
equations. Here is one method - (the potentials are not unique!.).
We demonstrate the result by example- a formal proof follows a similar
argument but is not given here
171
Example: Show that B = zi + x j + yk is solenoidal and find a vector potential.
divB =
!z
!x
+
!x
!y
+
!y
!z
= 0
!B is solenoidal
Look for a vector field A with A
y
= 0 (we use 3 degrees of freedom here
since 3 of the partial derivatives are zero).
!A
z
!y
= z
!A
x
!z
"
!A
z
!x
= x
"
!A
x
!y
= y
First equation gives
!A
z
!y
= z #A
z
= yz + f (x, z) = yz (take f (x,z)=0)
(we have used up 2 degrees of freedom here - those related
to the other two partial derivatives of A
z
)
172
!
Second equation gives
"A
x
"z
#
"A
z
"x
= x $
"A
x
"z
= x $ A
x
= xz + g(x, y)
We cannot set g(x, y) = 0 (that will mean using 2 more degrees of freedom),
so we try to evaluate it using 3rd eqn:
#
"A
x
"y
= y $#
"g
"y
= y $ g = #
1
2
y
2
+h(x)
Take h(x) = 0 (last degree of freedom used here)
A = (xz #
1
2
y
2
)i +(yz) k
Vector potentials are not unique. Another possible vector potential is (check)
A =
1
3
{(zx # y
2
)i +(xy # z
2
) j +(yz # x
2
)k }
173
Summary: B(x) is said to be solenoidal if div B=0.
F(x) is said to be irrotational if curl F=0.


!
div B = 0 " B = curl A
curl F = 0 "F = grad #
Vector potential
Helmholtzs Theorem (more difcult to prove!)
Any vector eld F for which curl F and div F tend to zero as x
tends to innity can be written as the sum of two parts, one
solenoidal and one irrotational
!
F = "# +"$B
Scalar potential
Vector potential
Scalar potential
174
Gradient (orthogonal curvilinear coordinates)
!
grad " : #
1
.$" = lim
"
Q
%"
P
PQ
= lim
"
& Q
+ o('u
1
) %"
P
h
1
'u
1
+ o('u
1
)
= lim
("
(u
1
'u
1
+ o('u
1
)
h
1
'u
1
+ o('u
1
)
=
1
h
1
("
(u
1
(all evaluated at P)
$" = #
1
1
h
1
("
(u
1
+ #
2
1
h
2
("
(u
2
+ #
3
1
h
3
("
(u
3
!
"f = (
#f
#r
,
1
r
#f
#$
,
1
rsin$
#f
#%
) &&&Spherical Polars
"f = (
#f
#R
,
1
R
#f
#%
,
#f
#z
) - - - - - Cylindrical Polars
P(u1,u2P,u3P)
u1
C1P
Q
!1
h1"u1
Q
1
175
Gradient (orthogonal curvilinear coordinates)
!
grad " : #
1
.$" = lim
"
Q
%"
P
PQ
= lim
"
& Q
+o('u
1
) %"
P
h
1
'u
1
+o('u
1
)
= lim
("
(u
1
'u
1
+o('u
1
)
h
1
'u
1
+o('u
1
)
=
1
h
1
("
(u
1
(all evaluated at P)
$" = #1
1
h
1
("
(u
1
+#2
1
h
2
("
(u
2
+#3
1
h
3
("
(u
3
!
"f = (
#f
#r
,
1
r
#f
#$
,
1
rsin$
#f
#%
) &&&Spherical Polars
"f = (
#f
#R
,
1
R
#f
#%
,
#f
#z
) - - - - - Cylindrical Polars
P(u1,u2P,u3P)
u1
C1P
Q
!1
h1"u1





!
h
1
= h
r
=1
h
2
= h
"
= r
h
3
= h
#
= rsin"
176
Divergence (orthogonal curvilinear coordinates)
Let a =!
1
"
1
+!
2
"
2
+!
3
"
3
where "
1
, "
2
, "
3
are unit
tangents to coordinate curves and form an orthogonal triad.
Construct an elementary curvilinear parallelopiped about
a point P(u
1
, u
2
, u
3
) with coordinate increments #u
1
, #u
2
, #u
3
.
Flux out of volume through [face centered at Q + face centered at R]
= !!
1Q
(h
2Q
"u
2
h
3Q
"u
3
) +!
1R
(h
2R
"u
2
h
3R
"u
3
)
=[!
1R
h
2R
h
3R
!!
1Q
h
2Q
h
3Q
]"u
2
"u
3
Now !
1
(u
1
, u
2
, u
3
), h
2
(u
1
, u
2
, u
3
), h
3
(u
1
, u
2
, u
3
) and between Q and R only u
1
changes.
Hence expression is
=[
#
#u
1
(!
1R
h
2R
h
3R
)"u
1
+o("u
1
)]"u
2
"u
3
(Using Taylor expansion)
=
#
#u
1
(!
1P
h
2P
h
3P
)"u
1
"u
2
"u
3
where we have used
#
#u
1
(!
1R
h
2R
h
3R
) =
#
#u
1
(!
1P
h
2P
h
3P
) +0(1)

177
div a = lim [
a.n dS
!!
V
]
= lim
1
h
1
!u
1
h
2
!u
2
h
3
!u
3
[
"
"u
1
(#
1
h
2
!u
2
h
3
!u
3
)]!u
1
+ two similar terms
"
#
$
%
&
'
=
1
h
1
h
2
h
3
"
"u
1
(#
1
h
2
h
3
) +
"
"u
2
(#
2
h
3
h
1
) +
"
"u
3
(#
3
h
1
h
2
)
"
#
$
%
&
'
!
h
1
= h
r
=1
h
2
= h
"
= r
h
3
= h
#
= rsin"
!
".F =
1
r
2
#(r
2
F
r
)
#r
+
1
rsin$
#(sin$F
$
)
#$
+
1
rsin$
#F
%
#%
Spherical polars
178
Construct a (curvilinear) rectangular area bounded
by coordinate curves perpendicular to !
3
. Then
!
3
.Curl a = Lim
a.!
!
!
dl
area A
h1!u1
h2!u2
u1 u2
"1
"2
Curl (orthogonal curvilinear coordinates)
Circulation of vector field about rectangle has a contribution
from sides parallel to u
1
curves of
(!!
1
h
1
"u
1
)
u2+"u2
+(!
1
h
1
"u
1
)
u2
=
#
#u
2
(!!
1
h
1
"u
1
)]"u
2
..(using Taylor's theorem incrementing in u
2
)
and from sides parallel to u
2
curves of
(!
2
h
2
"u
2
)
u1+"u1
!(!
2
h
2
"u
2
)
u1
=
#
#u
1
(!
2
h
2
"u
2
)]"u
1
..(using Taylor's theorem incrementing in u
1
)
179
Hence
!
3
.Curl a = Lim
a.!
!
!
dl
area A
=
1
h
1
h
2
!u
1
!u
2
[
"
"u
1
(#
2
h
2
!u
2
)]!u
1
+[
"
"u
2
("#
1
h
1
!u
1
)]!u
2
#
$
%
&
'
(
=
1
h
1
h
22
[
"
"u
1
(#
2
h
2
)]
1
+[
"
"u
2
("#
1
h
1
)]
#
$
%
&
'
(
h1!u1
h2!u2
u1 u2
"1
"2
180
!
Exercise : Show by taking a (curvilinear) rectangular area bounded
by coordinate curves perpendicular to "
3
that
"
3
.Curl a = Lim
a." # dl
area A
=
1
h
1
h
2
$
$u
1
(%
2
h
2
) &
$
$u
2
(%
1
h
1
)
'
(
)
*
+
,
and hence
Curl a =
1
h
1
h
2
h
3
h
1
"
1
h
2
"
2
h
3
"
3
$
$u
1
$
$u
2
$
$u
3
h
1
%
1
h
2
%
2
h
3
%
3
h1!u1
h2!u2
u1 u2
"1
"2
Curl (orthogonal curvilinear coordinates)
!
" #F =
1
rsin$
%(sin$F
&
)
%$
'
%F
$
%&
(
)
*
+
,
-
,
1
r
1
sin$
%Fr
%&
'
%(rF
&
)
%r
(
)
*
+
,
-
,

1
r
%(rF
$
)
%r
'
%F
r
%$
(
)
*
+
,
-
Spherical polars
181
!
Now
grad V=
1
h
1
"V
"u
1
#
1
+
1
h
2
"V
"u
2
#
2
+
1
h
3
"V
"u
3
#
3
and
div a =
1
h
1
h
2
h
3
"
"u
1
($
1
h
2
h
3
) +
"
"u
2
($
2
h
3
h
1
) +
"
"u
3
($
3
h
1
h
2
)
%
&
'
(
)
*
+,
2
V = div(grad V) =
1
h
1
h
2
h
3
[
"
"u
1
h
2
h
3
h
1
"V
"u
1
-
.
/
0
1
2
+
"
"u
2
h
1
h
3
h
2
"V
"u
2
-
.
/
0
1
2
+
"
"u
3
h
1
h
2
h
3
"V
"u
3
-
.
/
0
1
2
]
Laplacian (general orthogonal)
182
Laplacian (spherical and cylindrical polars)
!
"
2
V =
1
h
1
h
2
h
3
[
#
#u
1
h
2
h
3
h
1
#V
#u
1
$
%
&
'
(
)
+
#
#u
2
h
1
h
3
h
2
#V
#u
2
$
%
&
'
(
)
+
#
#u
3
h
1
h
2
h
3
#V
#u
3
$
%
&
'
(
)
]
"
2
V =
1
r
2
#
#r
r
2
#V
#r
$
%
&
'
(
) +
1
r
2
sin*
#
#*
sin*
#V
#*
$
%
&
'
(
) +
1
r
2
sin
2
*
#
2
V
#+
2
,,Spherical
"
2
V =
1
R
#
#R
R
#V
#R
$
%
&
'
(
) +
1
R
2
#
2
V
#+
2
$
%
&
'
(
)
+
#
2
V
#z
2
,,,,,,,,,,,,,,Cylindrical
Spherical Symmetry; (V =V(r)) - e.g. gravitational potential of a star
"
2
V =
1
r
2
#
#r
r
2 #V
#r
$
%
&
'
(
)
Cylindrical Symmetry; (V =V(R)) - e.g. gravitational potential of a galaxy
"
2
V =
1
R
#
#R
R
#V
#R
$
%
&
'
(
)
183
Laplacian (spherical and cylindrical polars)
!
"
2
V =
1
h
1
h
2
h
3
[
#
#u
1
h
2
h
3
h
1
#V
#u
1
$
%
&
'
(
)
+
#
#u
2
h
1
h
3
h
2
#V
#u
2
$
%
&
'
(
)
+
#
#u
3
h
1
h
2
h
3
#V
#u
3
$
%
&
'
(
)
]
"
2
V =
1
r
2
#
#r
r
2
#V
#r
$
%
&
'
(
) +
1
r
2
sin*
#
#*
sin*
#V
#*
$
%
&
'
(
) +
1
r
2
sin
2
*
#
2
V
#+
2
,,Spherical
"
2
V =
1
R
#
#R
R
#V
#R
$
%
&
'
(
) +
1
R
2
#
2
V
#+
2
$
%
&
'
(
)
+
#
2
V
#z
2
,,,,,,,,,,,,,,Cylindrical
Spherical Symmetry; (V =V(r)) - e.g. gravitational potential of a star
"
2
V =
1
r
2
#
#r
r
2 #V
#r
$
%
&
'
(
)
Cylindrical Symmetry; (V =V(R)) - e.g. gravitational potential of a galaxy
"
2
V =
1
R
#
#R
R
#V
#R
$
%
&
'
(
)






!
h
1
= h
r
=1
h
2
= h
"
= r
h
3
= h
#
= rsin"
184
Some Applications
185
M
m
+e
-e
Electron
of charge
plectron
of charge
Apple
of mass
Earth of
of mass
r
r
!
Force of Gravity (attractive) :
F(r) = "
GMm
r
2
r = "
GMm
r
2
r
r
= "GMm
r
r
3
Electrostatic Force (between charges e
1
,e
2
)
F(r) =
e
1
e
2
r
2
r =
e
1
e
2
r
2
r
r
= e
1
e
2
r
r
3

In both cases, force law is an inverse square law
F = k
r
r
3
- - - - Inverse square law
Inverse square force laws
186
!
A force field representing an inverse square law
is conservative because
"#
kr
r
3
= "#[
k
r
3
r] =
k
r
3
"#r +"(
k
r
3
)#r
= 0$ 3
k
r
4
r
r
#r = 0
From our previous theorem, a scalar potential exists, and can be found by
carrying out a line intregral from a reference point P(r
B
) to a field point P(r)
!(r) = F. dr
P(r
B
)
P(r)
!
= (
kr
r
3
). dr
P(r
B
)
P(r)
!
= (
k
r
3
)(r. dr
P(r
B
)
P(r)
!
)
=
1
2
(
k
r
3
)d(r. r
P(r
B
)
P(r)
!
) =
1
2
(
k
r
3
)d(r
2
P(r
B
)
P(r)
!
)
=
1
2
(
k
r
3
)2r dr
P(r
B
)
P(r)
!
=
k
r
2
dr
P(r
B
)
P(r)
!
= "
k
r
+
k
r
B
= "
k
r
+const
187
Exercise: Show that F(r) =
kr
r
3
is also solenoidal. In this case
it is not possible to find a vector potential since F has an
exceptional point!
188
We consider a fixed (source) point Q(!) = Q(!
1
, !
2
, !
3
)
and a general (field) point P(x) = P(x
1
, x
2
, x
3
).

R = x - ! ------position vector of P relative to Q
R = (x
1
!!
1
)
2
+(x
2
!!
2
)
2
+(x
3
!!
3
)
2
The scalar function
"(x
1
, x
2
, x
3
) =
1
R
=
1
(x
1
!!
1
)
2
+(x
2
!!
2
)
2
+(x
3
!!
3
)
2
defines a potential, and has interesting properties.
Q
P
O
#
x
R
!
Some properties of the 1/R potential
189
!(x
1
, x
2
, x
3
) =
1
R
=
1
(x
1
!"
1
)
2
+(x
2
!"
2
)
2
+(x
3
!"
3
)
2
#R
#x
i
=
1
2
2(x
i
!"
i
)
R
=
(x
i
!"
i
)
R
(i =1, 2, 3)
"
#!
#x
i
=
#
#x
i
1
R
= !
1
R
2
#R
#x
i
= !
1
R
3
(x
i
!"
i
) (i =1, 2, 3)
or $! = $
1
R
= !
x !"
R
3
= !
R
R
3
This is an inverse square force law for a source located at
Q("). So generally, an inverse square force derives from a
1
R
type potential.
190
!
"
2
1
R
=
#
#x
1
(
#
#x
1
1
R
) +
#
#x
2
(
#
#x
2
1
R
) +
#
#x
3
(
#
#x
3
1
R
)
=
#
#x
1
($
1
R
3
(x
1
$%
1
)) +
#
#x
2
($
1
R
3
(x
2
$%
2
)) +
#
#x
3
($
1
R
3
(x
3
$%
3
))
=
3
R
4
[
(x
1
$%
1
)
R
(x
1
$%
1
) +
(x
2
$%
2
)
R
(x
2
$%
2
) +
(x
3
$%
3
)
R
(x
3
$%
3
)] $
3
R
3
=
3
R
4
[
(x
1
$%
1
)
2
+ (x
2
$%
2
)
2
+ (x
3
$%
3
)
2
R
] $
3
R
3
=
3
R
3
$
3
R
3
= 0
A more elegant proof :
"
2
1
R
= div grad
1
R
= div (-
1
R
3
R) = $[
1
R
3
divR + grad(
1
R
3
).R]
= $[
3
R
3
$
3
R
4
R
R
.R] = 0
1/R is a Harmonic Function
191
Any function that satisfies Laplaces equation !
2
! = 0 is called a
harmonic function. Thus
1
R
is a harmonic function, and forms
the starting point for developing more general solutions of
Laplaces equation.
192
Appendix 1

Vector Algebra (Revision)
193
Products of Vectors (revision)
!
If a and b are any two vectors in R
3
, then the scalar, or "dot" product of
the two vectors is
a.b = a b cos" = abcos" (0 #" # $)
From definition
(i) a.b = b.a
(ii) a.b = 0 %either a = 0, b = 0 or " = $ 2
(iii) Unit basis vectors for RC system satisfy
e
1
.e
2
= 0, e
2
.e
3
= 0 , e
1
.e
1
= e
2
.e
2
= e
3
.e
3
=1
(iii) a.b = (a
1
e
1
+ a
2
e
2
+ a
3
e
3
).(b
1
e
1
+ b
2
e
2
+ b
3
e
3
)
= a
1
b
1
+ a
2
b
2
+ a
3
b
3
Scalar or dot product
!
a
b
!
a
b
194
Vector or cross product
!
Let n be a unit vector orthogonl to both a and b such that a, b, n form a
right - handed system. The vector product of a and b is defined to be the vector
a"b = a b sin# n = absin#n (0 $# $ %)
(i) b"a = &basin#(&n ) (because b,a and - n form a RH system)
'a"b = &b"a ....anti - commutative
(ii) a"b = 0 (either a = 0, b = 0 or sin# = 0 (# = 0, %)
(iii) e
1
"e
1
= 0, e
2
"e
2
= 0, e
3
"e
3
= 0
e
1
"e
2
= &e
2
"e
1
= e
3
e
2
"e
3
= &e
3
"e
3
= e
1
e
3
"e
1
= &e
1
"e
3
= e
2
a
b
!
axb=ab sin ! n
195
!
(iv) a"b = (a
1
e
1
+a
2
e
2
+a
3
e
3
)"(b
1
e
1
+b
2
e
2
+b
3
e
3
)
= (a
1
e
1
)"(b
1
e
1
) +(a
1
e
1
)"(b
2
e
2
) +(a
1
e
1
)"(b
3
e
3
) +(a
2
e
2
)"(b
1
e
1
) +..
= 0+a
1
b
2
e
3
# a
1
b
3
e
2
# a
2
b
1
e
3
+....
= (a
2
b
3
# a
3
b
2
)e
1
+(a
3
b
1
# a
1
b
3
)e
2
+(a
1
b
2
# a
2
b
1
)e
3
=
e
1
e
2
e
3
a
1
a
2
a
3
b
1
b
2
b
3
(v) The area of a triangle whose sides are b and c
A =
1
2
bcsin$ =
1
2
b"c
We can also define a directed quantity known as the vector area by
A =
1
2
b"c
b
c
A=1/2(bxc)
!
c sin !
196
Products of Three Vectors
!
Given three vectors a,b and c , we can form two types of products which make
mathematical sense : the triple scalar product and the triple vector product.
The triple scalar product is defined to be
a.(b"c).
From diagram
b"c = bcsin#n
a.(b"c) = a b"c cos$ = (bcsin#)acos$
= volume of parellepiped formed by a,b,c
(area of base % height)
&a.(b"c) = b.(c"a) = c.(a"b)
= 'a.(c"b) = ...........................
The Triple Scalar Product
a.(bxc)
a
b
c
(bxc)
Volume =
!
"
197
!
Expanding using components
a = a
1
e
1
+a
2
e
2
+a
3
e
3
b"c =
e
1
e
2
e
3
b
1
b
2
b
3
c
1
c
2
c
3
= (b
2
c
3
# b
3
c
2
)e
1
+(b
3
c
1
# b
1
c
3
)e
2
+......
a.(b"c) = a
1
(b
2
c
3
# b
3
c
2
) +..... =
a
1
a
2
a
3
b
1
b
2
b
3
c
1
c
2
c
3
but
a
1
a
2
a
3
b
1
b
2
b
3
c
1
c
2
c
3
=
c
1
c
2
c
3
a
1
a
2
a
3
b
1
b
2
b
3
$a.(b"c) = c.(a"b) = (a"b).c
Rule : The dot and the cross can be interchanged in a triple scalar
product : a.(b"c) = (a"b).c 198
!
Example : Find the volume of the parallelepiped spanned by
the vectors i + j - k, 2i - 3j + 4k, - i +2 j - 3k
Solution :
V =
1 1 "1
2 "3 4
"1 2 "3
=1(9" 8) "1("6+ 4) "1(4 " 3) = 2
199
The Triple Vector Product
!
b"c is perpendicular to both b and c, and is therefore
perpendicular to the plane containing b and c
a"(b"c) is perpendicular to a and b"c and is
therefore a vector in the plane containing b and c.
a"(b"c) = #b+c ( #, constants)
Similarly
(a"b)"c = $a +%b ( $,% constants)
a"(b"c) and (a"b)"c are usually different vectors.
An Important Identity
b
c
(bxc)
a
b
c
ax(bxc)
(bxc)
!
a"(b"c)
!
a"(b"c) = (a.c)b#(a.b)c
200
!
Now, a = a
1
e
1
+a
2
e
2
+a
3
e
3
b"c =
e
1
e
2
e
3
b
1
b
2
b
3
c
1
c
2
c
3
= (b
2
c
3
# b
3
c
2
)e
1
+(b
3
c
1
# b
1
c
3
)e
2
+......
a"(b"c) =
e
1
e
2
e
3
a
1
a
2
a
3
b
2
c
3
# b
3
c
2
b
3
c
1
# b
1
c
3
b
1
c
2
# b
2
c
1
= [a
2
(b
1
c
2
# b
2
c
1
) # a
3
(b
3
c
1
# b
1
c
3
)]e
1
+..........
= [(a
2
c
2
+a
3
c
3
)b
1
#(a
3
b
3
+a
2
b
2
)c
1
]e
1
+...........
= [(a
1
c
1
+a
2
c
2
+a
3
c
3
)b
1
#(a
1
b
1
+a
2
b
2
+a
3
b
3
)c
1
]e
1
+.....
$[a"(b"c)]
1
= [(a.c)b#(a.b)c]
1
%a"(b"c) = [(a.c)b#(a.b)c]
201
Vector Equation of a Straight Line
!
Parametric equation of a straight line passing
through a point A(a) containing the direction vector l is
r = a + "l (" a parameters)
(r # a) $l = 0
Equation of straight line passing through two points A(a), B(b) is
r = a + "(b # a) (" a parameter)
or
r = %a + &b
% + & =1
a r
A
P
O
l
a
b
A
O
B
P
r
202
!
Parametric equation of a plane passing through
A(a) containing direction vectors l and m is
r = a +"l +m (" , parameters)
(r # a).l $m = 0
Equation of plane passing through three points A(a),B(b) and C(c) is
r = a +"(b# a) + (c # a) (", parameters)
or
r = %a +&b+' c
%+& +' =1
Vector Equation of a Plane

a
m
l
lxm
A
P
O
r
O
a
A
P
O
r
B
C
203
Appendix 2

Area and Volume Integrals (Revision)
204
Area Integrals
!
Consider a 2 - D Euclidean space. In a RC system points have coordinates
(x, y)
Suppose f (x, y) " C
0
in some closed region R of x, y plane.
Partition R into n sub - regions R
k
(i) area of R
k
is #A
k
(ii)(x
k
, y
k
) " R
k
The area integral (a Riemann integral) is defined as
the limit of a sum of the function value weigthed
by the areas;
f (x, y)dA = lim
n$%
#A
k
$0
R
&& f (x
k
, y
k
)
k=1
n
' #A
k
The limit must exist independently
of the nature of the partitioning.
R
Y
Y
!Ak
(xk,yk)
205
Geometrical interpretation of an area integral
The surface z=f(x,y)
R
X
Y
Z
Section defined by R
!Ak
!Vk=f(xk,yk)!Ak
!
Consider a 3D RH RC coordinate system
OXYZ.
We assume f (x, y) > 0 on R.
Each element of the region R
defines an element of volume
defined by the surface
z = f (x, y)
So the area integral is the volume under
this surface :
V = f (x, y)dA
R
""
206
!
Consider a rectangular partitioning;
I = f (x, y)dA = lim
R
""
j
# f (x
i
, y
j
)
i
# $x
i
$y
j
This double summation can first be sumed
in the x direction
I = lim {
j
# f (x
i
, y
j
)
i
# $x
i
}$y
j
or we can reverse the order of summation, and
first sum in the y direction :
I = lim {
i
# f (x
i
, y
j
)
j
# $y
j
}$x
i
In the limit, this leads to the following rule
for evaluating double integrals.
X
Y
R
xi
yj
Nj
M
(xi,yj)
Sum in x first
X
Y
R
xi
yj
N
Mi
(xi,yj)
Sum in y first
Changing the order of summation
207
!
Consider a simple region which can be
specified as;
f
1
(y) " x " f
2
(y) for y
1
" y " y
2
g
1
(x) " y " g
2
(x) for x
1
" x " x
2
Then
f (x, y)dA = { f (x, y)dx }dy
f
1
( y)
f
2
( y)
#
y
1
y
2
#
R
##
Double Integral
as a repeated integral
X
Y
R
(x,y)
x=f2(y) x=f1(y)
y1
y2
Integrate in x first
area dxdy
X
Y
R
(x,y)
y=g2(x)
y=g1(x)
x1
x2
Integrate in y first
area dxdy
!
f (x, y)dA = { f (x, y)dy }dx
g
1
( x)
g
2
( x)
"
x
1
x
2
"
R
""
208
!
Example : Evaluate
x
2
R
"" e
y
dA
where R is the area bounded by x =1, y = x, y = 2x.
We choose to integrate in the y direction first because then there
is a single equation for each of the boundaries
x
2
R
"" e
y
dxdy = (
0
1
" x
2
x
2 x
" e
y
dy )dx
= x
2
e
y
[ ]
0
1
"
x
2x
dx = x
2
(e
2 x
0
1
" # e
x
)dx
=
1
4
e
2
# e +
7
4
(integrating by parts)
y
x
y=2x
y=x
x=1
0
1
209
!
(i) Choose order of integration (x first and then y or vice - versa) carefully
to make calculations easy (in the previous example, y first, then x).
Had we done it the other way, we have to evaluate two integrals
over two simple regions;
I = { x
2
e
y
dx
y
2
y
"
0
1
" }dy + { x
2
e
y
dx
y
2
1
"
1
2
" }dy
#more complicated!
(ii) For more complex regions, divide into sum
of simple regions as required (as in second method above)
...dA
R
1
+R
2
"" = .....dA
R
1
"" + .....dA
R
2
""
y
x
y=2x
y=x
x=1
0
1
1
2
R1
R2
X1
X2
210
!
Example : Determine the area bounded by the curves
2y =16" x
2
, x +2y = 4.
Sketch curves and find points of intersection :
(-3,7/2),(4,0)
A = { dy
2"
x
2
8"
x
2
2
#
"3
4
# }dx = (8"
x
2
2
" 2+
x
2 "3
4
# )dx
= (6"
x
2
2
+
x
2 "3
4
# )dx
= 6x +
x
2
2
"
x
3
6
$
%
&
'
(
)
"3
4
=
343
12
(-3,7/2)
(4,0)
y
x
y=8-x
2
/2
y=2-x/2
211
Volume Integrals
!
f (x, y,z) continuous in some closed region " of x, y,z plane.
Partition " into n sub - regions "
i
(i) volume of "
i
is #V
i
(ii) (x
i
, y
i
,z
i
) $"
i
f (x, y,z)dV =
"
% lim
n&'
#V
i
&0
f (x
i
, y
i
,z
i
)
i=1
n
( #V
i
independently of partitioning.
Using rectangular cartesian coordinates,
f (x, y,z)dV =
"
% f (x, y,z)dxdydz %%%
As in the case of area integrals, a volume integral over a
simple region can be evaluated by repeated integrals.
212
!
Example :
Evaluate zdV
V
""" over the region between planes z = 0, y = 0, y = x
and cylinder x
2
+ z
2
= a
2
.
zdV = dx dy zdz
0
a
2
#x
2
"
0
x
"
0
a
" """ = [ [ zdz
0
a
2
#x
2
"
0
x
"
0
a
" ]dy]dx
= [
a
2
# x
2
2
$
%
&
'
(
)
dy]dx =
0
x
"
0
a
"
a
2
# x
2
2
$
%
&
'
(
)
xdx
0
a
"
= #
(a
2
# x
2
)
2
8
$
%
&
'
(
)
0
a
=
a
4
8
Y
Z
x
a
y=x
y=0
213 213
Appendix 3
213
Revision (differentiability)
If f (x) is differentiable at x then in a neighbourhood of x
f (x +h) = f (x) + ! f (x)h +O(h
2
)
If ! f (x) is continuous at x it is differentiable at x.
If f (x, y) is differentiable at (x, y) then for a neighbourhood of (x, y)
f (x +h, y +k) = f (x, y) + f
x
(x, y)h + f
y
(x, y)k +O(h
2
+k
2
)
If f
x
(x, y) and f
y
(x, y) are continuous at (x, y) then f is differentiable
at (x, y).
214 214
Appendix 4
Suppose two functions are denoted by f and ! and
! is positive. We are considering the behaviour of these functions
as some limit is approached.
(1) If there exists a constant k such that f < k! as the limit
is approached then
f = O(!)
f = O(1) means f is bounded
(2) If
f
!
!0 as the limit is approaced
f = o(!)
f = o(1) means f !0
(3) If
f
!
!l " 0 as the limit is approached, then
f # l!

Das könnte Ihnen auch gefallen