DIVERGENCE, GRADIENT, CURL AND LAPLACIAN

Content

DIVERGENCE
GRADIENT
CURL
DIVERGENCE THEOREM
LAPLACIAN
HELMHOLTZ’S THEOREM

DIVERGENCE
Divergence of a vector field is a scalar operation that in once view tells us whether flow
lines in the field are parallel or not, hence “diverge”.

For example, in a flow of gas through a pipe without loss of volume the flow lines
remain parallel, but if the pipe narrows and the gas experiences compression then the flow
lines in the gas will converge (i.e. divergence is not zero)

Another term for the divergence operator is the „del vector‟, „div‟ or „gradient operator‟
(for scalar fields). The divergence operator acts on a vector field and produces a scalar. In
contrast, the gradient acts on a scalar field to produce a vector field.

When the divergence operator acts on a vector field it produces a scalar. In contrast,
the gradient operator acts on a scalar field to produce a vector field.

The divergence vector operator is  (also known as „del‟ operator ) and is defined as

  
 
 x1 
     
  , or,   ˆx1  ˆx2  ˆx3
 x2  x1 x2 x3
  
 
 x3 

in either indicial notation, or Einstein notation as

 
 ˆxi   ,i ˆxi
xi

Note, carefully that the Del operator does not cummute. That is,

a  a
 a  a 

Also note that:

 
  a.b    a a.b   b a.b 

For example,

  
 
 x1 
  
 V =  a1 a2 a3   
 x2 
  
 
 x3 

a1 a2 a3

 V =   , or,
x1 x2 x3

in indicial notation, as

Vi
V   Vi ,i
xi
Vi ,i  ai ,i

We will see more on indicial notation later but note that the summation is implied by
repeated indices (ii) and that a “,i” denotes derivative with respect to the variable i (=1,2,3)

(Note that I have omitted the dot; the dot is not essential, but like a „dot product‟ of two
vectors the outcome is a scalar value.)

(Note that I have omitted the dot; the dot is not essential—it represents abuse of
mathematical notation, although it is still correct)

Schey p. 36-40 “Div, Grad and All That”

   
 
 x1 
  
 V =  a1 a2 a3   
 x2 
  
 
 x3 

 a a2 a3 
 V =  1  , or,
 x1 x2 x3 

If we consider the physical meaning of a divergence. Start with a cube and estimate the
surface integral of some function such as the force applied at all points on its surface:
 n is the unit vector normal to the faces of a cube

dS is the area of each face of the cube

( x1 ,x2 ,x3 ) is a point at the centre of the cube. We are interested in the ratio of the
integral of this function to the volume enclosed by the cube. This limit of this ratio is called
the divergence:

1
  F  " divF"  lim  F  ndS
V  0 V (Gauss‟ Theorem)

  F  is a scalar.
If, for example we examine the divergence of the electrostatic field, then the sum of the
field over the faces can give us an idea of the charge included in the volume. If we sum the
flow of the field over the faces and ad the total flow on all 6 faces and then make the cube
diminish to a very small size, then we should be able to determine the flux at the point.

If we measure the sum of all the displacements/strains around the face of the cube and
in the limit they tend to zero then we can say that there is no net change in the volume of
the object. That would mean that as much strain is moving the faces in one direction as
there are net strains compensating for deformation on the other side of the cube.

For cases of pure and simple shear there is no net change in area of volume, i.e. volume
is conserved. In this situation,

  u =0

  u can also be viewed as the volumetric strain:

V
 u 
V

ui u1 u2 u3

u    
xi x1 x2 x3

You will see later that the strain tensor is defined in general as:

1 u j u j
eij  (  )
2 x j xi

The strain tensor is equal to the divergence of the displacement field, when i=j .
 Show that the divergence is the volumetric strain by the geometry of a deforming cube.
Assume that the cube is growing only in the direction of the basis vectors.

GRADIENT
Gradient is the derivative of a scalar field and is also known as the “grad”.

The gradient of a scalar field produces a vector. The gradient is written as:

  
 ˆx1  ˆx2  ˆx3
x1 x2 x3

The greatest spatial rate of change occurs in the direction of the gradient.

The gradient is invariant to the coordinate transformation:

For a simple 2D example:

h  x, y  h  x, y 
h( x, y )  ˆx  ˆy
x y

In indicial notation, the above example becomes:

h( xi )
h( x, y )  xˆ i
xi

The magnitude of this vector in the two-dimensional case is:

2
 h( x, y )   h( x, y ) 
2

h     
 x   y 

CURL
The “curl” or “rot” of a vector field is defined as

 ˆx1 ˆx2 ˆx3 

 
   
u  
 x1 x2 x3 
 
 u1 u2 u3 

In indicial notation, this can be written as:

 f i     u i
 ijk  j uk

fi  ijk uk , j , or in mixed notation:

fi   ijk
uk
x j

For clarification, we can write out the individual components long-hand:

f3 f 2
f1   , because if i=1, j=2 or 3 alone for non-zero results
x2 x3

Similarly,

f1 f3
f2  
x3 x1

f 2 f1
f3  
x1 x2

DIVERGENCE THEOREM AND THE LAPLACIAN

Gauss‟s Theorem (or divergence theorem) states that the flux of a property over the
surface of a volume equals the divergence of the property added up over the whole volume
enclosed by the same surface. The integral of the divergence over the volume tells use
whether that property is changing in size. That is,

   FdV   F  ndS
V S

Where n is the vector normal to the surface at any point and F is the vector field
property in question.

If we make:

F  u , then

  udV   u  ndS ,
V S

  udV   u  ndS or
2

V S
  udV   u  ndS
V S

“The sum of the Laplacian (for a scalar field) over the volume is also the sum of the
gradient over the entire surface”

Note that we have just introduced the Laplacian operator.

LAPLACIAN
(From Lay and Wallace, 1995)

When the Laplacian operator acts on a scalar field (u) it is equivalent to taking first the
gradient followed by the divergence of the result, i.e.:

Laplacian = “   ”

 2u  u,ii
  u  x1 ,x2 ,x3   u  x1 ,x2 ,x3   u  x1 ,x2 ,x3  
  , , 
 x1 x2 x3
 
 2u  x1 ,x2 ,x3   2u  x1 ,x2 ,x3   2u  x1 ,x2 ,x3 
 , ,
x12 x22 x32

The Laplacian can also operate on a vector field ( F ), in which case it is equivalent to

another vector field whose components are the Laplacian of the original vector components
(if Cartesian coordinates are used)

 F 
2
i
 Fi ,i ˆxi
 2 F1  2 F2  2 F3
 ˆ
x  ˆ
x  ˆx3
x12 x22 x32
1 2

In the general case for any coordinate system we can use the following vector identity, seen
elsewhere (->)

HELMHOLTZ’S THEOREM
(From Lay and Wallace, 1995)

This theorem states that any vector field F , can be represented in terms of a vector
potential (  ) and a scalar potential (  ) by
 F     

if      0 and  (   )  0

which are useful vector identities are:

 (   )  0 (div of curl of vector field)

     0 (curl of grad of scalar field)