Module-IV: Vector Calculus

Lesson 42

Gradient and Directional Derivative

42.1 Gradient of a Scalar Field

Let 𝑓𝑓(𝑥𝑥, 𝑦𝑦, 𝑧𝑧) be a real valued function defining a scalar field. To define the gradient of a scalar
field, we first introduce a vector operator called del operator denoted by ∇. We define the vector
differential operator in two and three dimensions as
𝜕𝜕 𝜕𝜕 𝜕𝜕 𝜕𝜕 𝜕𝜕
∇= 𝑖𝑖 𝜕𝜕𝜕𝜕 + 𝑗𝑗 𝜕𝜕𝜕𝜕 and ∇= 𝑖𝑖 𝜕𝜕𝜕𝜕 + 𝑗𝑗 𝜕𝜕𝜕𝜕 + 𝑘𝑘 𝜕𝜕𝜕𝜕

The gradient of a scalar field 𝑓𝑓(𝑥𝑥, 𝑦𝑦, 𝑧𝑧), denoted by ∇𝑓𝑓 or grad (𝑓𝑓) is defined as
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
∇f = 𝑖𝑖 𝜕𝜕𝜕𝜕 + 𝑗𝑗 𝜕𝜕𝜕𝜕 + 𝑘𝑘 𝜕𝜕𝜕𝜕

Note that the del operator ∇ operates on a scalar field and produces a vector field.

42.1. 1 Example

Find the gradient of the following scalar fields

(𝑖𝑖) 𝑓𝑓(𝑥𝑥, 𝑦𝑦) = 𝑦𝑦 2 − 4𝑥𝑥𝑥𝑥 at (1,2),

Solution
𝜕𝜕 𝜕𝜕
∇f(x, y) = �𝑖𝑖 𝜕𝜕𝜕𝜕 + 𝑗𝑗 𝜕𝜕𝜕𝜕 � (𝑦𝑦 2 − 4𝑥𝑥𝑥𝑥) = −4𝑦𝑦 𝑖𝑖 + (2𝑦𝑦 − 4𝑥𝑥)𝑗𝑗

42.1. 2 Example
1
𝒓𝒓 = 𝑥𝑥𝑥𝑥 + 𝑦𝑦𝑦𝑦 + 𝑧𝑧𝑧𝑧, |𝒓𝒓| = 𝑟𝑟 and 𝑟𝑟̂ = 𝒓𝒓/𝑟𝑟, then show that grad�𝑟𝑟 � = −𝑟𝑟̂ /𝑟𝑟 2 .

Solution
1 𝜕𝜕 𝜕𝜕 𝜕𝜕 1 1 𝜕𝜕𝜕𝜕 1 𝜕𝜕𝜕𝜕 1 𝜕𝜕𝜕𝜕 1 𝑥𝑥 𝑦𝑦
Grad�𝑟𝑟 � = �𝑖𝑖 𝜕𝜕𝜕𝜕 + 𝑗𝑗 𝜕𝜕𝜕𝜕 + 𝑘𝑘 𝜕𝜕𝜕𝜕 � �𝑟𝑟 � = 𝑖𝑖 �− 𝑟𝑟 2 𝜕𝜕𝜕𝜕 � + 𝑗𝑗 �− 𝑟𝑟 2 𝜕𝜕𝜕𝜕 � + 𝑘𝑘 �− 𝑟𝑟 2 𝜕𝜕𝜕𝜕 � = − 𝑟𝑟 2 �𝑟𝑟 𝑖𝑖 + 𝑟𝑟 𝑗𝑗 +
𝑧𝑧
𝑟𝑟
𝑘𝑘�

1 𝒓𝒓 𝑟𝑟̂
= − 𝑟𝑟 2 �𝑟𝑟 � = − 𝑟𝑟 2

where 𝑟𝑟̂ = (𝑥𝑥𝑥𝑥 + 𝑦𝑦𝑦𝑦 + 𝑧𝑧𝑧𝑧)/𝑟𝑟

 Gradient and Directional Derivative

42.1. 3 Geometrical Representation of the Gradient

Let 𝑓𝑓(𝑃𝑃) = 𝑓𝑓(𝑥𝑥, 𝑦𝑦, 𝑧𝑧) be a differentiable scalar field. Let 𝑓𝑓(𝑥𝑥, 𝑦𝑦, 𝑧𝑧) = 𝑘𝑘 be a level surface and
𝑃𝑃0 (𝑥𝑥0 , 𝑦𝑦0 , 𝑧𝑧0 ) be a point on it. There are infinite number of smooth curves on the surface passing
through the point 𝑃𝑃0 . Each of these curves has a tangent at 𝑃𝑃0 . The totality of these tangent lines
form a tangent plane to the surface at a point 𝑃𝑃0 . A vector normal to this plane at 𝑃𝑃0 is called the
normal vector to the surface at this point.

Consider now a smooth curve 𝐶𝐶 on the surface passing through a point 𝑃𝑃 on the surface. Let
𝑥𝑥 = 𝑥𝑥(𝑡𝑡), 𝑦𝑦 = 𝑦𝑦(𝑡𝑡), 𝑧𝑧 = 𝑧𝑧(𝑡𝑡) be the parametric representation of the curve 𝐶𝐶. Any point 𝑃𝑃 on 𝐶𝐶
has the position vector 𝑟𝑟(𝑡𝑡) = 𝑥𝑥(𝑡𝑡)𝑖𝑖 + 𝑦𝑦(𝑡𝑡)𝑗𝑗 + 𝑧𝑧(𝑡𝑡)𝑘𝑘. Since the curve lies on the surface, we
have

𝑓𝑓(𝑥𝑥(𝑡𝑡), 𝑦𝑦(𝑡𝑡), 𝑧𝑧(𝑡𝑡)) = 𝑘𝑘

𝑑𝑑
Then, 𝑑𝑑𝑑𝑑 𝑓𝑓(𝑥𝑥(𝑡𝑡), 𝑦𝑦(𝑡𝑡), 𝑧𝑧(𝑡𝑡)) = 0

𝜕𝜕𝜕𝜕 𝑑𝑑𝑑𝑑 𝜕𝜕𝜕𝜕 𝑑𝑑𝑑𝑑 𝜕𝜕𝜕𝜕 𝑑𝑑𝑑𝑑

By chain rule, we have 𝜕𝜕𝜕𝜕 𝑑𝑑𝑑𝑑
+ 𝜕𝜕𝜕𝜕 𝑑𝑑𝑑𝑑
+ 𝜕𝜕𝜕𝜕 𝑑𝑑𝑑𝑑
=0

𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑

or �𝑖𝑖 𝜕𝜕𝜕𝜕 + 𝑗𝑗 𝜕𝜕𝜕𝜕 + 𝑘𝑘 𝜕𝜕𝜕𝜕 � . �𝑖𝑖 𝑑𝑑𝑑𝑑 + 𝑗𝑗 𝑑𝑑𝑑𝑑 + 𝑘𝑘 𝑑𝑑𝑑𝑑 � = 0

or ∇𝑓𝑓. 𝑟𝑟 ′ (𝑡𝑡) = 0

Let ∇𝑓𝑓(𝑃𝑃) ≠ 0 and 𝑟𝑟 ′ (𝑡𝑡) ≠ 0. Now 𝑟𝑟 ′ (𝑡𝑡) is a tangent to 𝐶𝐶 at the point 𝑃𝑃 and lies in the tangent
plane to the surface at . Hence ∇𝑓𝑓(𝑃𝑃) is orthogonal to every tangent vector at 𝑃𝑃. Therefore,
∇𝑓𝑓(𝑃𝑃) is the vector normal to the surface 𝑓𝑓(𝑥𝑥, 𝑦𝑦, 𝑧𝑧) = 𝑘𝑘 at the point 𝑃𝑃.

42.1. 4 Example

We will find a unit normal vector to the surface 𝑥𝑥𝑦𝑦 2 + 2𝑦𝑦𝑦𝑦 = 8 at the point (3, −2,1).

Let 𝑓𝑓(𝑥𝑥, 𝑦𝑦, 𝑧𝑧) = 𝑥𝑥𝑦𝑦 2 + 2𝑦𝑦𝑦𝑦 = 8 then

𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
𝜕𝜕𝜕𝜕
= 𝑦𝑦 2 , 𝜕𝜕𝜕𝜕 = 2𝑥𝑥𝑥𝑥 + 2𝑧𝑧 and 𝜕𝜕𝜕𝜕
= 2𝑦𝑦

Therefore
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
∇f = 𝑖𝑖 𝜕𝜕𝜕𝜕 + 𝑗𝑗 𝜕𝜕𝜕𝜕 + 𝑘𝑘 𝜕𝜕𝜕𝜕 = 𝑦𝑦 2 𝑖𝑖 + (2𝑥𝑥𝑥𝑥 + 2𝑧𝑧)𝑗𝑗 + 2𝑦𝑦𝑦𝑦

At (3, −2,1), we obtain the normal vector as ∇f(3, −2,1) = 4i − 10j − 4k. The unit normal
vector at (3, −2,1) is given by
4𝑖𝑖−10𝑗𝑗 −4𝑘𝑘 2𝑖𝑖−5𝑗𝑗 −2𝑘𝑘
= .
√16+100+16 √33

2
 Gradient and Directional Derivative

42.1. 5 Example

Here we will find the angle between the two surfaces 𝑥𝑥𝑥𝑥𝑥𝑥𝑥𝑥 𝑧𝑧 = 𝑦𝑦 2 − 1 and 𝑥𝑥 2 𝑦𝑦 = 2 − 𝑧𝑧 at the
given point (1,1,1).

First note that the angle between two surfaces at a common point is the angle between their
normals at that point. Now we have

𝑓𝑓1 (𝑥𝑥, 𝑦𝑦, 𝑧𝑧) = 𝑥𝑥𝑥𝑥𝑥𝑥𝑥𝑥 𝑧𝑧 − 𝑦𝑦 2 + 1 = 0, ∆𝑓𝑓1 (𝑥𝑥, 𝑦𝑦, 𝑧𝑧) = (log 𝑧𝑧)𝑖𝑖 − 2𝑦𝑦𝑦𝑦 + (𝑥𝑥/𝑧𝑧)𝑘𝑘

∆𝑓𝑓1 (1,1,1) = −2𝑗𝑗 + 𝑘𝑘 = 𝑛𝑛1

𝑓𝑓2 (𝑥𝑥, 𝑦𝑦, 𝑧𝑧) = 𝑥𝑥 2 𝑦𝑦 − 2 + 𝑧𝑧 = 0, ∆𝑓𝑓2 (𝑥𝑥, 𝑦𝑦, 𝑧𝑧) = 2𝑥𝑥𝑥𝑥𝑥𝑥 + 𝑥𝑥 2 𝑗𝑗 + 𝑘𝑘

∆𝑓𝑓2 (1,1,1) = 2𝑖𝑖 + 𝑗𝑗 + 𝑘𝑘 = 𝑛𝑛2

𝑛𝑛 .𝑛𝑛 1 1
Therefore cos 𝜃𝜃 = �|𝑛𝑛 1||𝑛𝑛2 |� = or 𝜃𝜃 = 𝑐𝑐𝑐𝑐𝑐𝑐 −1 � �.
1 2 √30 √30

42.1.6 Properties of Gradient

Let 𝑓𝑓 and 𝑔𝑔 be any two differentiable scalar fields. The gradient satidfies the following algebraic
properties,

∆(𝑓𝑓 + 𝑔𝑔) = ∆𝑓𝑓 + ∆𝑔𝑔

∆(𝑐𝑐1 𝑓𝑓 + 𝑐𝑐2 𝑔𝑔) = 𝑐𝑐1 ∆𝑓𝑓 + 𝑐𝑐2 ∆𝑔𝑔, where 𝑐𝑐1 , 𝑐𝑐2 are arbitrary constants

∆(𝑓𝑓𝑓𝑓) = 𝑓𝑓∆𝑔𝑔 + 𝑔𝑔∆𝑓𝑓

𝑓𝑓 𝑔𝑔∆𝑓𝑓−𝑓𝑓∆𝑔𝑔
∆ �𝑔𝑔 � = 𝑔𝑔 2

42.2 Directional Derivative

Let 𝑓𝑓(𝑃𝑃) = 𝑓𝑓(𝑥𝑥, 𝑦𝑦, 𝑧𝑧) be a differentiable scalar field.
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
Then 𝜕𝜕𝜕𝜕 , 𝜕𝜕𝜕𝜕 , 𝜕𝜕𝜕𝜕 denotes the rates of change of 𝑓𝑓 in the direction of 𝑥𝑥, 𝑦𝑦 and 𝑧𝑧 axis, respectively.

𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

If 𝑓𝑓(𝑥𝑥, 𝑦𝑦, 𝑧𝑧) = 𝑘𝑘 is the level surface and 𝑃𝑃0 is any point, then 𝜕𝜕𝜕𝜕
, 𝜕𝜕𝜕𝜕
, 𝜕𝜕𝜕𝜕 at 𝑃𝑃0 (𝑥𝑥0 , 𝑦𝑦0 , 𝑧𝑧0 ) denote
the slopes of the tangent lines in the directions of 𝑖𝑖, 𝑗𝑗, 𝑘𝑘 respectively. It is natural to give the
definition of derivative in any direction which we call as the directional derivative.

Let 𝑏𝑏� = 𝑏𝑏1 𝑖𝑖 + 𝑏𝑏2 𝑗𝑗 + 𝑏𝑏3 𝑘𝑘 be any unit vector. Let 𝑃𝑃0 be any point 𝑃𝑃0 : 𝑎𝑎 = 𝑎𝑎1 𝑖𝑖 + 𝑎𝑎2 𝑗𝑗 + 𝑎𝑎3 𝑘𝑘.

Then, the position vector of any point 𝑄𝑄 on the line passing through 𝑃𝑃0 and in the direction of 𝑏𝑏�
is given by
3
 Gradient and Directional Derivative

𝑟𝑟 = 𝑎𝑎 + 𝑡𝑡𝑏𝑏� = (𝑎𝑎1 + 𝑡𝑡𝑏𝑏1 )𝑖𝑖 + (𝑎𝑎2 + 𝑡𝑡𝑏𝑏2 )𝑗𝑗 + (𝑎𝑎3 + 𝑡𝑡𝑡𝑡3 )𝑘𝑘 = 𝑥𝑥(𝑡𝑡)𝑖𝑖 + 𝑦𝑦(𝑡𝑡)𝑗𝑗 + 𝑧𝑧(𝑡𝑡)𝑘𝑘

This is, the point 𝑄𝑄(𝑎𝑎1 + 𝑡𝑡𝑏𝑏1 , 𝑎𝑎2 + 𝑡𝑡𝑏𝑏2 , 𝑎𝑎3 + 𝑡𝑡𝑡𝑡3 ) is on this line. Now, the vector formthe point
𝑃𝑃0 to 𝑄𝑄 is given by 𝑡𝑡𝑏𝑏�. Since | 𝑏𝑏�|=1, the distance from 𝑃𝑃0 to 𝑄𝑄 is 𝑡𝑡. Then
𝜕𝜕𝜕𝜕 𝑓𝑓(𝑄𝑄)−𝑓𝑓(𝑃𝑃)
𝜕𝜕𝜕𝜕
= lim𝑡𝑡→0 𝑡𝑡

if it exists, is called the directional derivative of 𝑓𝑓 at the point 𝑃𝑃0 in the direction to 𝑏𝑏� .
𝜕𝜕
Therefore 𝜕𝜕𝜕𝜕 𝑓𝑓(𝑥𝑥(𝑡𝑡), 𝑦𝑦(𝑡𝑡), 𝑧𝑧(𝑡𝑡)) is rate of change of 𝑓𝑓 with respect to the distance 𝑡𝑡.

We have
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝑑𝑑𝑑𝑑 𝜕𝜕𝜕𝜕 𝑑𝑑𝑑𝑑 𝜕𝜕𝜕𝜕 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕
= 𝜕𝜕𝜕𝜕 𝑑𝑑𝑑𝑑
+ 𝜕𝜕𝜕𝜕 𝑑𝑑𝑑𝑑
+ 𝜕𝜕𝜕𝜕 𝑑𝑑𝑑𝑑

where
𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
,
𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
, are evaluated at 𝑡𝑡 = 0 .

We write
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕
= �𝑖𝑖 𝜕𝜕𝜕𝜕 + 𝑗𝑗 𝜕𝜕𝜕𝜕 + 𝑘𝑘 𝜕𝜕𝜕𝜕 � . �𝑖𝑖 𝑑𝑑𝑑𝑑 + 𝑗𝑗 𝑑𝑑𝑑𝑑 + 𝑘𝑘 𝑑𝑑𝑑𝑑 � = ∇𝑓𝑓. 𝑑𝑑𝑑𝑑

𝑑𝑑𝑑𝑑
But 𝑑𝑑𝑑𝑑
= 𝑏𝑏�(a unit vector). Therefore, the directional derivative of 𝑓𝑓 in the direction of 𝑏𝑏� in given
by

Directional derivative =∇𝑓𝑓. 𝑏𝑏� = grad(𝑓𝑓). 𝑏𝑏�,

which is denoted by 𝐷𝐷𝑏𝑏 (𝑓𝑓). Note that 𝑏𝑏� is a unit vector. If the direction is specified by a vector
𝑢𝑢, then 𝑏𝑏� = 𝑢𝑢/|𝑢𝑢|.

42.2.1 Example

We will determine the directional derivative of 𝑓𝑓(𝑥𝑥, 𝑦𝑦, 𝑧𝑧) = 𝑥𝑥𝑦𝑦 2 + 4𝑥𝑥𝑥𝑥𝑥𝑥 + 𝑧𝑧 2 at the point
(1,2,3) in the direction of 3𝑖𝑖 + 4𝑗𝑗 − 5𝑘𝑘.

Consider

∇𝑓𝑓 = (𝑦𝑦 2 + 4𝑦𝑦𝑦𝑦)𝑖𝑖 + (2𝑥𝑥𝑥𝑥 + 4𝑥𝑥𝑥𝑥)𝑗𝑗 + (4𝑥𝑥𝑥𝑥 + 2𝑧𝑧)𝑘𝑘.

At the point (1,2,3), we have ∇𝑓𝑓 = 28𝑖𝑖 + 16𝑗𝑗 + 14𝑘𝑘. The unit vector in the given direction is
𝑏𝑏� = (3𝑖𝑖 + 4𝑗𝑗 − 5𝑘𝑘)/5√2.

Therefore

4
 Gradient and Directional Derivative

1 78
𝐷𝐷𝑏𝑏 (1,2,3) = 5√2 (28𝑖𝑖 + 16𝑗𝑗 + 14𝑘𝑘). (3𝑖𝑖 + 4𝑗𝑗 − 5𝑘𝑘) = 5√2

Suggested Readings

Courant, R. and John, F. (1989) Introduction to Calculus and Analysis, Vol. II, Springer-Verlag,
New York.

Jain, R.K. and Iyengar, S.R.K. (2002) Advanced Engineering Mathematics, Narosa Publishing
House, New Delhi.

Jordan, D.W. and Smith, P. (2002) Mathematical Techniques, Oxford University Press, Oxford.

Kreyszig, E. (1999) Advanced Engineering Mathematics, John Wiley, New York.

Piskunov, N. (1974) Differentail and Integral Calculus, Vol. II, MIR Publishers, Moscow.

Wylie, C. R. and Barrett, L.C. (2003) Advanced Engineering Mathematics, Tata McGraw-Hill,
New Delhi.