Operators

in scalar and vector fields

• gradient of a scalar field
• divergence of a vector field
• curl of a vector field
 GRADIENT OF A SCALAR FIELD

We will denote by f(M) a real function of a point M in an area

A. If A is two dimensional, then

f M   f  x, y 

and, if A is a 3-D area, then

f M   f  x, y, z 

We will call f(M) a scalar field defined in A.

Let f(M) = f(x,y). Consider an equation f(M) = c. The curve defined
by this equation is called a level line (contour line, height line) of
the scalar field f(M). For different values c1, c2, c3, ... we may get a
set of level lines. y

c5
c3 c 4
c2
c1
Similarly, if A is a 3-D area, the equation f(M) = c defines a
surface called a level surface (contour surface). Again, for
different values of c, we will obtain a set of level surfaces.
z

c1

c2

c3

c4 y

x
 Directional derivative

Let F(M) be a 3-D scalar field and let us construct a value that
characterizes the rate of change of F(M) at a point M in a
direction given by the vector
e = (cos , cos , cos ) (the direction cosines).

z F M 1 
M1
F M 
M
  y


x
F F M 1   F M 
 lim
e M 1  M MM 1

Let M = [x,y,z] and M1=[x+x, y+y, z+z ], then

F M1   F M   F  x  x, y  y, z  z   F  x, y, z  

F F F
 dF  x, y, z     x  y  z 
x y z
F F F
  cos   cos    cos   
x y z

where   MM 1 and   0 as   0.
This gives us

F  F F F 
 lim cos   cos   cos    
e  0 x y z 

However, the first three terms of the limit do not depend

on  and   0 as   0 so that

F F F F
 cos  cos   cos 
e x y z
 F
Clearly, e assumes its greatest value for

F F F
e i j k
x y z

This vector is called the gradient of F denoted by

F F F
grad F  x, y, z   i j k
x y z

or, using the Hamiltonian operator  nabla  :

grad F  F
Thus grad F points in the direction of the steepest increase in F
or the steepest slope of F.
Geometrically, for a c, grad F at a point M is parallel the unit
normal n at M of the level surface F  x, y, z   c

z grad F F  x, y, z   c

n
M

x
Vector field, vector lines

Let a vector field f(M) be given in a 3-D area , that is,

each M  Ω is assigned the vector

f  M   f 1  x, y , z  i  f 2  x, y , z  j  f 3  x, y , z  k

A vector line l of the vector field f(M) is defined as a line

with the property that the tangent vector to l at any point L
of l is equal to f(L).
This means that if we denote by dS  dx i  dy j  dz k
the tangent vector of l, then, at each point M, the following
equations hold
dx dy dz
 
f1  M  f 2  M  f 3  M 
This yields a system of two differential equations that can be
used to define the vector lines for f(M). This system has a
unique solution if f1, f2, f3 together with their first order
partial derivatives are continuous and do not vanish at the
same point in Ω . Then, through each M  Ω , there passes
exactly one vector line.
Note: For a two-dimensional vector field we similarly get the
differential equation
dx dy

f1  M  f 2  M 
Example

Find the vector line for the vector field f M   x i  y j  2 z k

passing through the point M = [1, -1, 2].

Solution
We get the following system of differential equations
dx dy dz dx dy dy dz
  or  
x  y  2z x y y 2z

this clearly has a solution

xy  c1 , y 2  c2 z or, using parametric equations,

c1 t2
x  , y  t , z  , t  1
t c2
Since the resulting vector line should pass through M, we get
1
c1  1, c2 
2
1
x   , y  t , z  2t 2 , t  1
t
Example
Find the vector lines of a planar flow of fluid characterized by
the field of velocities f(M)=xy i - 2x(x-1) j.

Solution

The differential equation defining the vector lines is

dx dy
 integrating this differential equation
xy 2 x x  1 with separated variables, we obtain
1 2
2  x  1 dx    y dy  x  2 x   y  c1
2

2
2
y
which yields  x  12
  c c  0
2
y c=4

c=3

c=2

c=1

x
 VECTOR FIELD

We will denote by f(M) a real vector function of a point M in

an area A. If A is two dimensional, then

f M   f1 x, y i  f 2 x, y  j

and, if A is a 3-D region, then

f M   f 1 x, y i  f 2 x, y  j  f 3 x, y k

We will call f(M) a vector field defined in A.

 Flux through a surface

Let  be a simple (closed) surface and f(M) a 3-D vector field.

The surface integral

 f M  dS   f x, y, z  dydz  f x, y, z  dxdz  f x, y, z  dxdy

1 2 3

is referred to as the flux of the vector field f(M) through

the surface .
 Divergence of a vector field

closed 3-D area V

vector field f x, y, z 
with S as border
S = V
M
an internal point M
V

 f x, y, z dS |V| is the volume of V

D V
V

D is the flux of f(x, y, z) through V per unit volume

Let us shrink V to M, that is, the area V becomes a point and see
what D does. If f (x, y, z) has continuous partial derivatives, the
below limit exists, and we can write symbolically:

 f x, y, z dS
DM   lim D  lim V
V M V M V

If we view f (x, y, z) as the velocity of a fluid flow, D(M)

represents the rate of fluid flow from M.
for D(M) > 0, M is a source of fluid;
for D(M) < 0, M is a sink.
if D(M) = 0, then no fluid issues from M.
If we perform the above process for every M in the region in
which the vector field f (x ,y, z) is defined, we assign to the
vector field f (x, y, z) a scalar field D(x,y,z) = D(M).

This scalar field is called the divergence of f (x, y, z).

We use the following notation:

DM   div f x, y, z   div  f1 x, y, z i  f 2 x, y, z  j  f 3 x, y, z k 

It can be proved that, in Cartesian coordinates, we have

  
div f x,y,z  f1 x, y, z   f 2 x, y, z   f 3  x, y , z 
x y z

Or, using the Hamiltonian or nabla operator

  
 i j k
x y z

we can write

div f x, y, z     f x, y, z 

 Solenoidal vector field

If div f(M) = 0 at every M in a 3D region A, we say that the

field f (x, y, z) is solenoidal in A.

If a vector field f (x, y, z) is solenoidal in a region A, then it has

neither sources nor sinks in A.
 Gauss's - Ostrogradski's theorem

Let us take a closed surface  that contains a 3D region V where

a vector field f (x, y, z) is defined and "add-up" the divergence of
f (x, y, z) within V, that is, calculate the tripple integral

 div f x, y, z  dx dy dz

V

From what was said about the divergence being the rate of
flow through points of 3D space, we could conclude that this
integral should represent what flows through  as the
boundary of V. However this is actually the flux of f (x, y, z)
through  or, in symbols

 f x, y, z dS

This is what Gauss's - Ostrogradsky's theorem says:
If a vector function f(x, y, z) has continuous partial derivatives in
a 3D region V bounded by a finite boundary , we can write

 f x, y, z  dS   div f x, y, z  dx dy dz

V

where the surface on the left-hand side of the equation is

oriented so that the normals point outwards.

The most general form of this formula was first proved by

Mikhail Vasilevich Ostrogradsky in 1828.
Example
Calculate the volume of an area V bounded by a closed
surface  given by the parametric equations
x   u, v, y   u, v, z   u, v, u, v M
Solution
Let us define a vector field f x, y, z   xi  yj  zk
By Gauss's-Ostrogradsky's theorem, we have

 f ( x, y, z)dS   div f x, y, z  dx dy dz  3 dx dy dz

V V

and thus
1
V   A ' u , v   B ' u, v   C ' u , v  du dv
3M
where A, B, C are the coordinates of a normal to , that is,

 'u  'u  'u  'u  'u  'u

A B C
 'v  'v  'v  'v  'v  'v
For example, if B is a ball of radius r, we have

x  r cos u sin v, y  r sin u sin v, z  r cos v, u, v 0,2  0, 

And so

1
B     v du dv 
3 2 3 3 2 3 3 2
r cos u sin v r sin u sin v r sin v cos
3M


1 2r 3 2r 3 4r 3
   r sin v du dv 
3
 sin v dv   cos v0 


3M 3 0 3 3
 The curl of a vector field

n

A
M l

vector field f (x, y, z)

For a plane  determined by its unit normal n, containing a closed

curve l with a fixed point M inside the area A bounded by l, define
the quantity
l f x, y, z  ds
C n , l , M   |A| is the surface
A area of A
If f (x, y, z) has continuous partial derivatives, we can calculate

C n , M   lim C n , l , M 

l M

and C (n, M) does not depend on the choice of l and the

way it shrinks to M. It is only determined by the point M
and the direction of n

It can further be proved that there exists a universal vector c(M)

such that
cM  n  C n , M 
for every normal n determining the plane 
Using the above method we can assign a vector c(M) to every
point M in the 3D-region in which the vector field satisfies the
assumptions (continuous partial derivatives). In other words, we
have defined to the original vector field f (x, y, z) a new vector
field c (x, y, z).

This vector field is called the curl of f (x, y, z) and denoted by

curl f (x, y, z) or rot f (x ,y, z)

If f (x, y, z) = f1(x, y, z) i + f2(x, y, z) j + f3(x, y, z) k, it can be

shown that

 f 2 f 3   f 3 f1   f1 f 2 
curl f x, y, z     i     j    k
 z y   x z   y x 
The last formula can be written using the following formal
determinant

i j k
  
curl f x, y, z  
x y z
f1 x, y, z  f 2  x, y , z  f 3  x, y , z 

or

curl f x, y, z     f x, y, z 

Suppose the vector f (x, y, z) field represents a flow of fluid.
Let us think of a very small turbine on a shaft positioned at a
point M.
Through the fluid flow, the turbine will turn at a speed s.
Let us move the shaft changing its direction while leaving the
turbine at the point M.
In a certain direction c(M), the speed s of the turbine will reach
its maximum. Then c(M) is clearly the curl of f (x, y, z) at M.

s s
M

s
 Irrotational vector field

If curl f x, y, z   0 in a 3D-region A, we say that the field

f x, y, z  is irrotational.

Recall that f (x, y, z) has to be irrotational if line integrals of

the vector field are to be independent of the line along which
the integral is calculated being only functions of the initial
and final points.
Every continuously differentiable vector function f(M) defined
in a region A (subject to mild restrictions) can be expressed as
a sum of two vector functions g(M) and h(M) such that

g(M) is solenoidal and h(M) is irrotational.

The possibility of such decomposition greatly simplifies

the study of many velocity and force fields occurring in physics.
 Stokes formula

Let a vector field f (x, y, z) be defined in a 3D region A with

continuous first partial derivatives. Let a simple 3D surface  be
given in A bounded by a closed regular curve l. Then the flux of
the curl of f (x, y, z) through  equals the line integral of f (x, y, z)
along l. Formally

 curl f x, y, z dS   f x, y, z ds

l
 Example

Use the Stokes formula to calculate the line integral of the

second type of the vector field
f (x, y, z) = y i + z j + x k
along the circle C given by the equation x2 + y2 = r2.

Solution
We have
curl f (x, y, z) = (-1) i + (-1) j + (-1) k
Let us consider the semi-sphere S:

x  r cos u sin v, y  r sin u sin v, z  r cos v, u, v 0,2  0, / 2

Since, clearly, C is the boundary of S, we can use the Stokes
formula:

I   y dx  z dy  x dz    dy dz  dx dz  dx dy 
C S

2  /2
   du
0
  A  B  C  dv
0

where A  r 2 cos u sin 2 v, B  r 2 sin u sin 2 v, C  r 2 sin v cos v

2  /2
I  r 2  du  cos u sin 
v  sin u sin 2 v  sin v cos v dv
2

0 0

2  /2 2  /2  /2
 
I  r   cos u du  sin v dv   sin u du  sin v dv  2  sin v cos v dv 
2 2 2

0 0 0 0 0 
 2 2
Since, clearly,  sin u du   cos u du  0 , we have
0 0

 /2  /2
1
I  2r  sin v cos v dv  r  sin 2v dv  r  cos 2v0  r 2
2 2 2  /2

0 0
2