Physics I Course Material 15

3 Vector Fields (contd.)

Contents: The curl of a vector field; line integrals; conservative vector fields.

3.1 One vector field from another: the curl

So far, we have introduced the gradient operator. which involves the application
of the del operator to a scalar field φ, to produce a vector field ∇ ~ φ; and the
divergence, which involves the dot product of the del operator with a vector field
~ resulting in a scalar field ∇
A, ~ · A.
~ It is natural to ask whether we can take the
cross product of the del operator with a vector field A ~ to produce another vector
~ × A.
field, ∇ ~ This is indeed possible, and the result is called the curl of the
~ Thus
vector field A.
~ r) = ∇
curl A(~ ~ × A(~
~ r) . (3.1)
In terms of individual components, using the Levi-Civita symbol introduced in
~ ×A
Unit 1, we have (denoting ∇ ~ by B),
~
~ × A)
Bi = (∇ ~ i = ǫijk ∂j Ak . (3.2)
Remember that summations over the repeated indices j and k are implicit in the
above equation, in keeping with the summation convention. To get more practice
with this notation, let us write out the above formula in components, explicitly:
B1 = ∂2 A3 − ∂3 A2 , B2 = ∂3 A1 − ∂1 A3 , B3 = ∂1 A2 − ∂2 A1 .
In more familiar terms, this is nothing but
∂Az ∂Ay ∂Ax ∂Az ∂Ay ∂Ax
Bx = − , By = − , Bz = − . (3.3)
∂y ∂z ∂z ∂x ∂x ∂y
Just as vector fields whose divergence vanishes (in some specified region) are
called solenoidal, vector fields whose curl vanishes (in some specified region) are
said to be irrotational. Again, this is a very important sub-class of vector fields.
For example, the gravitational and electrostatic fields are irrotational. In fact,
it will turn out that all conservative vector fields are irrotational vector
fields! This will be seen to have important physical consequences.
Exercise Use the formula (3.3) to verify that the following vector fields are
~ r) = ~r ; (ii) A(~
irrotational: (i) A(~ ~ r ) = −GM~r/r 3 , the gravitational field of a
point mass M at the origin.
It is important to note the following. We have written div A ~=∇ ~ ·A
~ and curl
A~=∇ ~ ×A ~ as if these were the definitions of the divergence and the curl of a
vector field. This is not strictly true. The divergence and curl have definitions
based on two other concepts (the flux and circulation of a vector field) that
have direct physical interpretations. We shall come to these very shortly. It will
then be realised that ∇ ~ ·A~ and ∇ ~ ×A ~ are in fact calculational formulas for
the divergence and curl, respectively, rather than basic definitions.
16 V. Balakrishnan & S. Govindarajan

3.2 Formula for the work done by a force field

Consider a particle moving in a force field F~ (x, y, z) – for example, the gravita-
tional field of the earth. The work done by the force when the particle moves
from a point ~r to a neighbouring point ~r + δ~r is given by

δW = F~ (~r) · δ~r . (3.4)

Note the dot product. Work is a scalar, whereas both force and displacement are
vectors. The work done comes from the component of the force in the direction
of the displacement, and hence the dot product.
Now suppose the particle moves from point P to point Q along an arbitrary
curved path C connecting the two points. The total work done is obtained
by dividing the curve into infinitesimal directed line segments δ~r1 , δ~r2 , . . . , and
summing up δW for each of the segments.

~i
X
Work done W = F~ (~ri ) · dr
segments i

Taking the limit in which the length of each segment goes to zero,
Z
W = ~
F~ (~r) · dl . (3.5)
C

R
Here the symbol C stands for integration along the path C. Such an integral is
~ rather
called the line integral of the vector field F~ (~r) along C. We have used dl
~ as this is the usual convention for line integrals.
than dr,

3.3 Line integral of a vector field

It is quite clear that line integrals can be defined for arbitrary vector fields, and
not just for force fields. The line integral of a vector field A over a curve C is
given by
Z
A(~ ~
~ r) · dl (3.6)
C
Rb
Such line integrals are a generalisation of the usual integral of the form a f (x)dx
with which you are already familiar. Just as we can “add” integrals according to
Z b Z c Z c
f (x)dx + f (x)dx = f (x)dx ,
a b a
Physics I Course Material 17

we can also “add” line integrals. Con-

sider the curve C1 which begins at P R

and ends at Q and the curve C2 which

begins at Q and ends at R (see Fig. 3.1). C2

We can add the two curves to obtain a Q

curve C which begins at P and ends at

P C1
R. We write this as C = C1 + C2 . Due
to the additive property of integrals, one
has Figure 3.1: C = C1 + C2
Z Z Z
A(~ ~ +
~ r) · dl A(~ ~ =
~ r ) · dl ~
~ r ) · dl
A(~ (3.7)
C1 C2 C

We can also define the curve −C to be the curve C traversed in the opposite
direction, i.e., if C begins at P and ends Q along a path, then −C begins at Q
and ends at P along the same path.
Curves which begin and end at the same point are called closed curves. A
~ ~
H
line integral over a closed curve is written as C A(~r) · dl,
with a small circle over the integral sign
to indicate a closed path. It is called the
circulation of the vector field over the Q
closed path. Notice that a closed path
as shown in Fig. 3.2 can be regarded as C1
made up of two curves C1 and C2 which
have the same end-points i.e., they both -C 2
begin at P and end at Q. Notice that the P
path from Q to P along the right hand
branch is −C2 . Therefore C = C1 − C2 is Figure 3.2: C = C1 − C2
a closed curve, and
I Z Z
~ ~
A(~r) · dl = ~ ~
A(~r) · dl − ~
~ r ) · dl
A(~ (3.8)
C C1 C2

The important
Rb thing Rto notice is the following. In the case of usual integrals on a
a
real line, a f (x)dx+ b f (x)dx ≡ 0. However, it is clear that a line integral along
~ is in general not equal to zero: this will be
~ r) · dl,
H
a closed path, such as C A(~
~ r ) along C1 and C2 happen to
so only if the line integrals of the vector field A(~
be equal, which is the case we now turn to.

3.4 Conservative vector fields

~ ≡ 0 when a vector field V~ is a conservative vector field. Recall
V~ · dl
H
C
that this means that V~ is of the form V~ (~r) = ∇φ(~
~ r). Then, the line integral over
C1 is Z Z
~ ~
V (~r) · dl = ~ r ) · dl
∇φ(~ ~ = φ(Q) − φ(P ) , (3.9)
C1 C1
18 V. Balakrishnan & S. Govindarajan

~ = dφ ( Eqn. (2.6)). Exactly the same result holds

~ · dl
on using the property ∇φ
good for C2 , as the RHS of the above equation depends only on the end points
P and Q of the path. We therefore have an important result of great physical
significance which can be expressed in three entirely equivalent ways, of which
we shall make extensive use:
(i) The line integral of a conservative vector field V~ (~r) around any
closed path is equal to zero.
(ii) The value of the line integral of a conservative vector field V~ (~r)
from any point P to any point Q depends only on the end points
P and Q and not on the actual path taken from P to Q.
(iii) A conservative vector field V~ (~r) can be written as the gradient of
a scalar field, i.e., V~ (~r) = ∇φ(~
~ r).
(It has been assumed in the above that the vector field V~ (~r) is well defined at all
points in the region of interest.)
We will now verify that the inverse square (e.g., electrostatic or gravita-
tional) force field is conservative by establishing property (i) above. Figure
3.3 shows an arbitrary closed planar path C, not passing through the origin of
coordinates O. The integral over C can be broken up into a sum over a large
number of small closed paths (loops like P QRS) by a mesh formed of concentric
circles and radial lines as shown. [This is because every internal edge (like P Q,
QR, etc.) is shared by two loops traversed in opposite directions, and therefore
gives contributions to the overall integral which cancel out.] Now consider the
line integral over the loop P QRS. On the segments QR and SP , the line element
is perpendicular to the force field, which is radially directed. Hence these con-
tributions vanish. Further, since the magnitude of the force field only depends
on the radial distance and not on the angle variables, the contribution from P Q
exactly cancels that from RS. Hence the line integral over the loop P QRS is
zero. This is true for every infinitesimal loop in the mesh. Therefore the line
integral over C is zero, establishing property (i).

Note two additional points: (i) Since the

force does not depend on any angular vari-
ables, the argument can be extended eas-
ily to non-planar closed paths. (ii) The Q C
P
argument did not make use of the 1/r 2 de-
R
pendence of the force field. Therefore, it S
is true for any radially directed force O

field whose magnitude only depends

on the distance from the origin.
Figure 3.3: