Lecture 12: Fields (RHB 8.6, 8.7.

1; D chapter 4; BK chapter 4)
In Physics we often have to consider properties that vary in some region of space e.g. tem-
perature of a body. To do this we require the concept of fields.
If to each point r in some region of in ordinary 3-d space there corresponds a scalar
φ(x1 , x2 , x3 ), then φ(r) is a scalar field.

Scalar Fields: Temperature distribution in a body T (r) or pressure in the atmosphere P (r);
Electric charge density or mass density ρ(r);
Electrostatic potential φ(r).

Similarly a vector field assigns a vector V (x1 , x2 , x3 ) to each point r of some region.

Vector Fields: Velocity in a fluid v(r);

Electric current density J(r);
Electric field E(r); Magnetic field B(r)

A vector field in 2-d can be represented graphically, at a carefully selected set of points r,
by an arrow whose length and direction is proportional to V (r) e.g. wind velocitymap on
the weather forecast.

12. 1. Level Surfaces of a Scalar Field

If φ(r) is a non-constant scalar field, then the equation

φ(r) = c

where c is a constant, defines a level surface (or equipotential) of the field. Level surfaces
do not intersect (else φ would be multi-valued at the point of intersection).
Familiar examples in two dimensions (level curves) are the contours of constant height on a
geographical map, h(x1 , x2 ) = c . Also isobars on a weather map are level curves of pressure
P (x1 , x2 ) = c.

Examples in three dimensions:

(i) Let φ(r) be φ(r) = x2 + y 2 + z 2
√
The level surface φ(r) = c is a sphere of radius c centred on the origin. As c is varied, we
obtain a family of level surfaces.
(ii) Electrostatic potential due to a point charge q situated at the point a is

q 1
φ(r) =
4π0 |r − a|

The level surfaces are concentric spheres centred on the point a.

(iii) Let φ(r) = k · r . The level surfaces are planes k · r = constant with normal k.

45
(iv) Let φ(r) = exp(ik · r) . Note that this a complex scalar field. Since k · r = constant is
the equation for a plane, the level surfaces are planes.
The function φ goes through phase 2π when we go from r to r + ∆r where ∆r · k = 2π i.e.
we move distance 2π/|k| along the unit normal k̂.
Therefore exp(ik · r) is referred to as a ‘plane wave’ of wavelength 2π/|k| in direction k̂.

12. 2. Gradient of a Scalar Field

How does a scalar field change as we change position?

As an example think of a 2-d contour map of the height h=h(x,y) of a hill say. The height is
a scalar field. If we are on the hill and move in the x-y plane then the change in height will
depend on the direction in which we move (unless the hill is completely flat!). For example
there will be a direction in which the height increases most steeply (‘straight up the hill’)
We now introduce a formalism to describe how a scalar field φ(r) changes as a function of r.
Mathematical Note: A scalar field φ(r) = φ(x1 , x2 , x3 ) is said to be continuously differentiable
in a region R if its first order partial derivatives

∂φ(r) ∂φ(r) ∂φ(r)

, and
∂x1 ∂x2 ∂x3
exist and are continuous at every point r ∈ R. We will generally assume scalar fields are
continuously differentiable.

Let φ(r) be a scalar field. Consider 2 nearby points: P (position vector r) and Q (position
vector r + δr). Assume P and Q lie on different level surfaces as shown:

Now use Taylor’s theorem for a function of 3 variables to evaluate the change in φ as we
move from P to Q
δφ ≡ φ(r + δr) − φ(r)
= φ(x1 + δx1 , x2 + δx2 , x3 + δx3 ) − φ(x1 , x2 , x3 )
∂φ(r) ∂φ(r) ∂φ(r)
= δx1 + δx2 + δx3 + O( δx2i )
∂x1 ∂x2 ∂x3

46
where we have assumed that the higher order partial derivatives exist, and that terms of
order ( δx2i ) can always be neglected. Thus we can write

δφ = ∇ φ(r) · δr

where the 3 quantities

∂φ(r)
∇ φ(r) i =
∂xi
form the Cartesian components of a vector field. We write

∂φ(r) ∂φ(r) ∂φ(r) ∂φ(r)

∇ φ(r) ≡ ei = e1 + e2 + e3
∂xi ∂x1 ∂x2 ∂x3

or in the old ‘x, y, z’ notation (where e1 = i, e2 = j and e3 = k)

∂φ(r) ∂φ(r) ∂φ(r)

∇ φ(r) = e1 + e2 + e3
∂x ∂y ∂z

The vector field ∇ φ(r), pronounced “grad phi”, is called the gradient of φ(r).

Example: calculate the gradient of φ = r2 = x2 + y 2 + z 2

 
∂ ∂ ∂
∇φ(r) = e1 + e2 + e3 (x2 + y 2 + z 2 )
∂x ∂y ∂z
= 2x e1 + 2y e2 + 2z e3 = 2r

12. 3. Interpretation of the gradient

In deriving the expression for δφ above, we assumed that the points P and Q lie on different
level surfaces. Now consider the situation where P and Q are nearby points on the same
level surface. In that case δφ = 0 and so

δφ = ∇φ(r) · δr = 0

The infinitesimal vector δr vector lies in the level surface at r̂, and the above equation holds

for all such δr, hence

47
 ∇φ(r) is normal to the level surface at r.

To construct a unit normal n(r) to the level surface at r, we divide ∇φ(r) by its length
∇φ(r)

n(r) = valid for |∇φ(r)| =
6 0
|∇φ(r)|

12. 4. Directional Derivative

Now consider the change in δφ poduced by moving distance δs in some direction say ŝ.
Then δr = ŝδs and
δφ = (∇φ(r) · ŝ) δs
As δs → 0, the rate of change of φ as we move in the direction of ŝ is
dφ(r)
= ŝ · ∇φ(r) = |∇φ(r)| cos θ (1)
ds
where θ is the angle between ŝ and the normal to the level surface at r.

ŝ · ∇φ(r) is the directional derivative of the scalar field φ in the direction of ŝ.

Note that the directional derivative has its maximum value when s is parallel to ∇φ(r), and
is zero when s lies in the level surface. Therefore

∇φ points in the direction of maximum increase in φ

Also recall that this direction is normal to the level surface. For a familiar example think of
the contour lines on a map. The steepest direction is perpendicular to the contour lines.
Example:
√ Find the directional derivative of φ = xy(x + z) at point (1, 2, −1) in the (e1 +
e2 )/ 2 direction.
∇φ = (2xy + yz)e1 + x(x + z)e2 + xye3 = 2e1 + 2e3
at (1,2,-1). Thus at this point
1 √
√ (e1 + e2 ) · ∇φ = 2
2
Physical example: Let T (r) be the temperature of the atmosphere at the point r. An
object flies through the atmosphere with velocity v. Obtain an expression for the rate of
change of temperature experienced by the object.
As the object moves from r to r + δr in time δt, it sees a change in temperature
 
δr
δT (r) = ∇T (r) · δr = ∇T (r) · δt
δt
Taking the limit δt → 0 we obtain
dT (r)
= v · ∇T (r)
dt

48