4.

Integration along curves

How do we integrate scalar and vector elds over curves rather
than straight lines? Before we do that, we better nd out how to
parametrise curves C(t) in 2 and 3 dimensions. Also, what is the
arclength dr along a curve in 3 dimensions? How do we use vector
calculus to dene the work done in moving along a path C. Why
does calculating the work done in certain force elds lead to inde-
pendence of path and in others does not? What is a conservative
vector eld?
4.1 Path integrals of scalar and vector elds
Motivation: Consider the work done by a force in moving a load
from x = a to x = b
Work = Force distance
But Force is a vector. So
W = F ((b a)i)
Now, what if F were not constant, F = F(x)? Then
W = lim
n
n

i
(F(x
i
) i)(x)
i
=
_
b
a
(F i)dx
=
_
t=(ba)
t=0
F dr,
where
r = (a, 0, 0) + (t, 0, 0)
is the position vector of the load, so that in this case dr = dt i
33
More generally, if F = F(r) is a vector eld which varies with
spatial position r = (x, y, z) and distance moved is in direction b,
then motion occurs along the line
r(t) = a + bt, for which dr(t) = bdt,
so that
W =
_
(ba)
0
F(r(t)) b dt
Worked example 4.1 Calculate the work done by moving along
the straight line from (2,1,1) to (3,2,2) in a force eld
F =
r
|r
2
|
=
xi + yj + zk
x
2
+ y
2
+ z
2
But what if the path we move on is a general curve C rather than a
straight line? We need to calculate
_
C
F dr
If we consider two end points with position vectors a and b does the
work done depend on the path we take?
Denition The work integral is an example of a path integral
of a vector eld
_
C
v(r) dr. (4.1)
We can interpret this in 2D as summing the components of the
vectors in the direction of the tangent to the curve C at each point
along C whose position vector is r.
34
F. r d
C
F
C
The answer is a scalar
_
C
v dr =
_
C
(v
1
dx + v
2
dy + v
3
dz).
In order to evaluate such path integrals we need to have a parametri-
sation of the curve C(t), just like we did for the straight line in worked
example 4.1. That is we write
r(t) = x(t)i + y(t)j + z(t)k
Then
dr =
_
_
_
dr(t)
dt
dt =
dx(t)
dt
i +
dy(t)
dt
j +
dz(t)
dt
k
_
_
_ dt
We shall see that for many physically important vector elds v (to
be made precise shortly) the path integral is independent of the path
taken. That is, if two curves C
1
and C
2
connect the same endpoints
with position vectors a and b, then the integral of v over both paths
is the same.
Before considering how to parametrise various curves lets mention
another type of path integral:
Denition: path integral of a scalar eld
_
C
f(r)ds =
_
C
f(r)|dr| (4.2)
where ds is the innitesimal arclength along C (see later).
35
We can interpret this in 2D as taking a path C across the contours
f =const. ssumming up the height.
x
y

C
=const.
x
y

C
C
1
2
Clearly in general, if two dierent curves C
1
and C
2
connect the
same end points a and b then the integrals along C
1
and C
2
will be
dierent.
One can also dene other path integrals of vector quantities, e.g.
_
C
v(r)ds =
_
C
v
1
(r)|dr|i +
_
C
v
2
(r)|dr|j +
_
C
v
3
(r)|dr|k
which is just 3 integrals of the form (4.2). So for the rest of this chapter
we shall just consider integrals of the form (4.1) and (4.2).
4.2 Parametrisation of Curves
The key to evaluating such integrals is to dene a single (intrinsic)
co-ordinate t that parametrises the curve C.
Consider rst curves in 2D. For some curves it is obvious how to do
this, e.g. use the x-co-ordinate as the parameter:
straight line y = bx + a x = t, y = a + bt or
r(t) = (t, a + bt) = (0, a) + (1, b)t, dr =
dr
dt
dt = (1, b) dt
parabola y = bx
2
+ a
r(t) = (t, bt
2
+ a) dr = (1, 2bt) dt
For other curves one can use an angular formulation
36
circle x
2
+ y
2
= a
2
x = a cos t, y = a sin t
dr = (a sin t, a cos t)
ellipse x
2
/a
2
+ y
2
/b
2
= 1 x = a cos t y = b sin t
dr = (a sin t, b cos t)
In 3D the concepts are similar; e.g.
straight line
r = a + bt (x, y, z) = (a
1
, a
2
, a
3
) + (b
1
, b
2
, b
3
)t

_
_
x a
1
b
1
,
y a
2
b
2
,
z a
3
b
3
_
_
= (t, t, t)
Therefore
x a
1
b
1
=
y a
2
b
2
=
z a
3
b
3
which is equation for straight line in Cartesian coordinates
helix (shape of classical light spring)
r = (x, y, z) = (a cos t, a sin t, bt)
Note that x, y describe a circle, so that the helix lies inside the
cylinder, with arbitrary value of z .
Here a is the radius of the helix and b/a = tan where is the
helix angle.
37
2D curves in a general plane spanned by orthogonal unit vectors
a and

b e.g.
The length, or total arclength S of a curve r = (x(t), y(t), z(t)
from t = a to t = b is the sum of little pieces of curve (innitesimal
arcs)
ds = |dr| =

dr(t)
dt
dt

dr
dt

dt
=

dx
dt
i +
dy
dt
j +
dz
dt
k

dt =

_
dx
dt
2
+
dy
dt
2
+
dz
dt
2
dt;
so that
S =

i=0
ds
i
=
_
t=b
t=a

_
dx
dt
2
+
dy
dt
2
+
dz
dt
2
dt;
Worked example 4.2 Calculate the total arclength of the fol-
lowing curves
1. The catenary (shape of a perfectly exible heavy chain sus-
pended between two points x = X and x = X) whose height
is given by the dimensionless formula y = coshx.
2. One complete revolution of a helix of radius a and angle =
b/a.
Application of arclength
Now, arclength gives a special choice of the parametrisation t such
that the curve is parametrised by its length. We use the parameter s
to represent the arclength parametrisation of a curve r(s), dened by
s(t) =
_

dr
dt

dt,
38
so that

dr(s)
ds

dr(t)
dt
/
_
_
ds
dt
_
_

dr(t)
dt

dr
dt

= 1.
This enables us to dene simple geometric properties of the curve
(youre not expected to memorise these definitions)
1. unit tangent vector u(s) = r

(s)
2. curvature (s) = |u

(s)| = |r

(s)|
3. unit (principle) normal vector p(s) =
u

|u

|
=
r

4. torsion (tortuosity) = p(s) b

(s),
where b(s) = (u(s) p(s)) is so-called unit bi-normal.
These denitions only work if we use the arclength parametrisation.
4.3 Evaluation of scalar path integrals
To evaluate the scalar path integrals of the form (4.2), i.e.,
_
C
f(r)ds
from points A to B along the curve C:
1. parametrise the curve C as r(t) = (x(t), y(t), z(t)) (note: any
parametrisation t will do provided we strictly increase from A to
B along the curve)
2. work out the limits a and b on t
3. calculate
ds = |dr| =

dr(t)
dt

dt =

_
_
_
_
dx(t)
dt
_
_
_
2
+
_
_
_
dy(t)
dt
_
_
_
2
+
_
_
_
dz(t)
dt
_
_
_
2
dt
4. evaluate the scalar eld f along (x, y, z) = (x(t), y(t), z(t)) and
integrate with respect to t.
_
C
f(r)ds =
_
t=b
t=a
f(x(t), y(t), z(t))

_
_
_
_
dx(t)
dt
_
_
_
2
+
_
_
_
dy(t)
dt
_
_
_
2
+
_
_
_
dz(t)
dt
_
_
_
2
dt
39
We have already seen examples of this: total arclength S is just
dened to be the scalar path integral of the function f = 1.
Worked example 4.3 Integrate f(r) = y/x from A = (0, 0) to
B = (1, 1) along the two curves y = x and y = x
2
. Are the two
answers the same?
4.4 Evaluation of work integrals
The evaluation of work integrals of the form (4.1) is even easier. You
dont need to calculate |
dr
dt
|. Simply
1. parametrise the curve C as r(t) = x(t), y(t), z(t) (again any
reasonable parametrisation will do)
2. work out the limits a and b on t
3. evaluate the vector eld v along (x, y, z) = (x(t), y(t), z(t)), form
the dot product and integrate w.r.t. t:
_
C
v(r)dr =
_
t=b
t=a
v(x(t), y(t), z(t))
_
_
_
dx(t)
dt
i +
dy(t)
dt
j +
dz
dt
(t)k
_
_
_ dt
Worked example 4.4
Find the work done in moving a particle from (0, 0, 0) to (1, 1, 1)
in the force eld
F = (2x + y
2
)i 3xyj + k
along the following paths
1. the straight path C
1
, the straight line joining (0, 0, 0) to (1, 1, 1)
2. the path C
2
given parametrically by x = t, y = t
2
, z = t
3
Note that the answers are not the same!
Properties of work integrals
Linear:
_
C
(v
1
+ v
2
) dr =
_
C
v
1
dr +
_
C
v
2
dr
40
decomposition into pieces: if C = C
1
+ C
2
_
C
r dr =
_
C
1
r dr +
_
C
2
r dr
Worked example 4.5
Find the work done in moving from (0, 0, 0) to (1, 1, 1) in the
force eld F given by
F = (2xy + z
3
)i + x
2
j + 3xz
2
k
along the paths
1. C
1
the straight line joining (0, 0, 0) to (1, 1, 1)
2. C
2
composed of the three straight lines joining (0, 0, 0) to (1, 0, 0)
to (1, 1, 0) to (1, 1, 1).
Note that both answers are the same as they would be for any curve
C joining (0, 0, 0) and (1, 1, 1). This is because the above force eld F
is conservative.
4.5 Conservative vector elds and independence
of path
Recall from Chapter 2, a vector eld v is conservative curl v =
0 v = , for some scalar function .
In addition, we have the following result:
If the vector eld F is conservative then the work integral F dr
between A and B (with position vectors a and b) is independent
of the path C chosen between these two points (provided F remains nite
within a simple domain containing A and B). Moreover, writing F = we have
_
C
F dr =
_
B
A
F dr = (a) (b) (4.3)
41
Proof. We showed in Chapter 2 that if F is conservative then there
exists a scalar eld such that F = . So
_
B
A
F dr =
_
B
A
dr
=
_
_

x
i +

y
j +

z
k
_
_
(dxi + dyj + dzk)
=
_
B
A
_
_

x
i +

y
j +

z
k
_
_
(dxi + dyj + dzk)
=
_
B
A
d = (b) (a).
Hence the integral only depends on the value of at A and B and is
independent of the the path taken between them. 2
Worked example 4.6 Caclulate curl F for the force eld F
taken in worked example 4.5, hence show the eld is conservative.
Find the scalar potential eld and hence calculate the work done
in moving from (1, 2, 1) to (3, 1, 4)
Remarks
The converse is also true; if the work integral is independent of
path taken between any two points then there must exist a scalar
potential function such that F = .
If C is a closed curve, then we can show that for a conservative
vector eld
_
C
F dr = 0
(for any closed curve C inside a simple domain in which F is nite)
note the special notation
_
for integrals around closed paths.
42
So we have (provided F is nite everywhere of interest)
Conservative curl F = 0 F =
_
C
F dr = 0.
Physical interpretation: is the scalar potential function associ-
ated with the force F = . Consider the earths gravitational
eld. The gravitational force F associated with a body of mass m
at position vector r from the center of the earth is
F = r
k
|r|
3
=
k
|r|
2
e
r
(4.4)
where (k/r
2
0
) = g = 9.98, r
0
being the radius of the earth and e
r
is a unit vector in the radial direction = r/|r|.
F is conservative ( = k/|r|, see Worked example 2.3). Inde-
pendence of path means that the energy required to overcome the
gravitational potential in order to escape the atmosphere is inde-
pendent of the path taken by the space rocket.
Many physical force elds can be expressed as a force in the direc-
tion of the origin which is a function only of distance r := |r| from
the origin:
F = f

(r)e
r
, for some scalar function f.
Then F is conservative, because it can be shown that
F = , where = f(r). (4.5)
Taking the special case f = k/r we get the 1/r
2
force law (4.4)
for gravity (applies also to electrostatics).
Worked example 4.7 Verify the result (4.5) in the particular
case f = r
3
.
43
Force elds that are not conservative are called non-conservative
or dissipative. An example is the force on a mass in a rotating
fram due to an angular acceleration a
F = ma r
Then curl F = 2ma (see Worked example 3.2) and clearly from
the physics the work done in moving in such a force eld is not
independent of path:
Summary
Two forms of integrals:
scalar path integral
_
C
f(r)ds, work integral
_
C
F(r) dr
Evaluate by parametrising the curve; C := r(t). Note special
parametrisation by arclength r(s).
Work integrals of conservative elds are independent of path
_
C
F dr = (B) (A)
_
F dr = 0.
44