Page 1 of 9

Vector Calculus
Suffix Notation
We define the Knonecker Delta and the Levi-Civita permutation symbol as if i = j 1 dij = 0 otherwise 1 if (i, j, k ) is an even permuation of (1,2, 3) if (i, j, k ) is an odd permuation of (1,2, 3) eijk = -1 0 otherwise We then have that
a b = eijka jbk det A = eijk A1iA2 j A3k

The general identity

dil eijk elmn = djl dkl dim djm dkm din djn dkn

Can be established by the following argument: o It is equal to 1 when (i, j, k ) = (l, m, n ) = (1,2, 3) (the matrix, in that case, is simply identity matrix). o Changes sign when any of (i, j, k ) or (l, m, n ) are

interchanged (by the rules of determinants).

o This last property also implies that if any of

the (i, j, k ) or (l, m, n ) are equal, it is equal to 0. [Swapping those two equal indices gives x = x, which gives x = 0]. These properties ensure that the RHS and LHS are equal for any index.
If we contract the identity once by setting l = i, we

get
eijk eimn = djm dkn - djn dkm
Maths Revision Notes Daniel Guetta, 2007

Page 2 of 9 This is the most useful form to remember.

Vector Differential Operators

GRAD o If we consider a scalar field, F(x , y, z ) = F(r ) ,

Taylors Theorem states that

F(x + dx , y + dy, z + dz ) = F(x , y, z ) + F F F dx + dy + dz + O(dx 2, dx dy,...) x y z
2

Or

F(r + dr ) = F(r ) + (F) dr + O ( dr Where

= ex + ey + ez x y z = ei x i

And for an infinitesimal increment, we can write

dF= (F) dr

And the grad operator is F .

o The geometrical interpretation of the grad

operator is that F t Is the directional derivative the rate of . change of F with distance in the direction t
o Note that

The

derivative

is

maximal

in

the

direction t F . The derivative is zero in directions such that t ^ F . These directions therefore lie in the plane tangent to the surface

F = constant .
In other words, F is in the direction of increase of the grad field.
o Furthermore

Maths Revision Notes

Daniel Guetta, 2007

Page 3 of 9

The unit vector normal to the surface

F = constant is then n = F / F . The rate of change of F with arclength s along a curve is
t F

where

t = dr / ds is the unit tangent vector to

the curve.
OTHER DIFFERENTIAL OPERATORS o The divergence of a vector field is the scalar field
Fi F = F = ei x i x i

o The curl of a vector field is the vector field

Fk e F = F = ei eijk i x i x i

This can also be written as a determinant

ex
F = / x

ey
Fy

ez
Fz
2F x i x i

/ y / z

Fx

o The Laplacian of a scalar field is the scalar field

2F = (F) =

The Laplacian of a vector field is

2F = ei 2Fi x j x j

VECTOR DIFFERENTIAL IDENTITIES o There are a number of identities relating

different vector differential operators.

o Two operators, one field
(F) = 2F ( F ) = 0 (F) = 0 ( F ) = ( F ) - 2F

Maths Revision Notes

Daniel Guetta, 2007

Page 4 of 9
o One operator, two fields (YF) = YF + FY
(FF ) = (F) F + F F (FF ) = (F) F + F F (F G ) = G ( F ) - F (G ) (F G ) = (G )F -G( F ) - (F )G + F ( G ) (F G ) = (G )F + G ( F ) + (F )G + F (G )

o As a result of some of these identities, we have

the interesting fact that:

If

vector

field

is

irrotational

( F = 0 ), it can be written as the

gradient of a scalar potential F = F .

If

vector

field

is

solenoidal

( F = 0 ), it can be written as the curl of a vector potential F = G .

Integral Theorems
The gradient theorem states that

r1 r2

(F) dr = F(r2 ) - F(r1 )

The divergence theorem (Gauss Theorem) states that

( F )dV =

F dS
S

Where V is the volume bounded by the closed surface

S, and the vector surface element is dS = ndS, where

n is the outward unit normal vector. For multiply connected volumes (eg: spherical shells),

all the surfaces must be considered.

Related results are as follows:

( F )dV =
V

(F)dV =

F dS dS F

Maths Revision Notes

Daniel Guetta, 2007

Page 5 of 9 The rule is, effectively, to replace the in the volume integral by n in the surface one, and the dV by a dS.
The curl theorem (Stokes Theorem) states that

( F ) dS =

F dr

Where S is an open surface bounded by the closed

curve C. The direction of dS and dr are chosen so

that they form a right-handed system. Again, a multiply connected surface (such as an annulus) maybe have more than one bounding curve.
We can use these integral theorems to get geometrical interpretations of grad, div and curl. o Consider the gradient theorem to a tiny line ds . Since the variation of F segment dr = t
along the line are negligible, we have and t t (F)ds dF

F ds 0 s The rate of change with distance. t (F) = lim

o Applying

the

divergence

theorem

to

an

arbitrarily small volume dV bounded by dS : 1 F = lim F dS V 0 d V dS The efflux per unit volume.
o Finally, applying the url theorem to an

arbitrarily small open surface dS with a unit normal vector n and bounded by a curve dC , we find:
1 F dr dS 0 dS dC The circulation per unit area. n ( F ) = lim

Orthogonal Curvilinear Coordinates

Cartesian coordinates can be replaced with any

independent

set

of

coordinates

Maths Revision Notes

Daniel Guetta, 2007

Page 6 of 9

(q1[x 1, x 2, x 3 ], q2[x 1, x 2, x 3 ], q 3[x 1, x 2, x 3 ]) .

axes are curves.

Curvilinear

(as

opposed to rectilinear) means that the coordinates

In general, curvilinear coordinates, the line element is

given by dr = h1dq1 + h2 dq2 + h3 dq 3 Where:

hi = eihi =

r qi

(No sum)

This determines the displacement associated with an

increment in qi. We have o hi is the scale factor associated with the

coordinate

qi.

It

converts

the

coordinate

increment (which might be an angle, for example) into a length. This depends on position.
o ei is the corresponding unit vector. In general,

this will also depend on positive. If, at any point, hi = 0, then we have a coordinate
singularity however much we change the component

qi, we go nowhere.
To find the surfaces described by keeping a certain

coordinate constant, assume that it is constant and twiddle with the expressions obtained to get something recognisable. (Eliminate all non-Cartesian variables except for the one we want to keep constant). To prove that they are perpendicular, show that (for example) z / x u z / x v = -1 .
The Jacobian of (x , y, z ) with respect to (q1, q2, q 3 ) is

defined as:
x / q1 x / q 2 (x , y, z ) = y / q 1 y / q 2 J = (q1, q 2, q 3 ) z / q1 z / q 2 x / q 3 y / q 3 z / q 3

Maths Revision Notes

Daniel Guetta, 2007

Page 7 of 9 The columns of the Jacobian are the vectors hi as defined above. Therefore
J = h1 h2 h3
The volume element

in

general

curvilinear

coordinate system is therefore

dV = (x , y, z ) dq 1 dq 2 d q 3 (q1, q2, q 3 )

The Jacobian therefore appears whenever changing variables in a multiple integral. The modulus sign appears because when changing the limits of integration, one is likely to place them in the right direction (ie: upper limits greater than lower limits), even if q actually decreases as x increases.
If we consider three sets of n variables, ai , bi and gi ,

then, by the chain rule for partial differentiation: n ai a bk = i g j k =1 bk g j Taking the determinant of this matrix equation, we find that: (a1, , an ) (a1, , an ) (b1, , bn ) = (g1, , gn ) (b1, , bn ) (g1, , gn )
In other words, the Jacobian of a composite transformation is the product of the Jacobians of the transformations of which it is composed. In the special case where gi = ai for all i, we get a rule for the inversion of a Jacobian.
Things are made easier when the coordinates we

choose are orthogonal:

ei e j = dij

and right-handed

e1 e2 = e3
In this case
o The line element is given by dr = e1 h1dq1 o The surface element

is

given

by

dS = e 3 h1h2dq1dq 2 .

Maths Revision Notes

Daniel Guetta, 2007

Page 8 of 9
o The volume element

is

given

by

dV = h1h2h3dq1dq2dq 3 .
o The Jacobian is simply J = h1h2h3 . In general

F =
F = 1 h1h2h3

e1 F e2 F e3 F + + h1 q1 h2 q2 h3 q 3

(h2h3F1 ) + (h3h1F2 ) + (h1h2F3 ) q 1 q 2 q 3 h1e1 h2e2 h3e3 1 F = / q1 / q 2 / q 3 h1h2h3 h1F1 h2F2 h3F3

2F =

1 h1h2h3

h h F h1h2 F h3h1 F 2 3 + + q h q q h q q h q 1 1 2 2 2 3 3 3 1

Commonly used orthogonal coordinate systems are: o Cartesian coordinates o Cylindrical polar coordinates z

(x , y, z )

y
f r

x
Where:

0 < r < , 0 f < 2p , - < z < . r = (x , y, z ) = (r cos f, r sin f, z ) r hr = = (cos f, sin f, 0) r r hf = = (-r sin f, r cos f, 0) f r hz = = (0, 0,1) z hr = 1 e r = (cos f, sin f, 0)

Maths Revision Notes

Daniel Guetta, 2007

Page 9 of 9

hf = r ef = (- sin f, cos f, 0)

hz = 1 ez = (0, 0,1) d V = r dr df d z The system is singular on the axis

r = 0.

o Spherical polar coordinates z (x , y, z )

q f

r y

x
Where:

0 < r < , 0 < q < p , 0 f < 2p . r = (x , y, z ) = (r sin q cos f, r sin q sin f, r cos q) r hr = = (sin q cos f, sin q sin f, cos q) r r hq = = (r cos q cos f, r cos q sin f, -r sin q) q r hf = = (-r sin q sin f, r sin q cos f, 0) f hr = 1
hq = r hf = r sin q

er = (sin q cos f, sin q sin f, cos q)

eq = (cos q cos f, cos q sin f, - sin q) ez = (- sin f, cos f, 0)

dV = r 2 sin q dr dq df The system is singular on the axis

r = 0 , q = 0 and q = p .

Related to cylindrical polars by

r = r sin q

z = r cos q

Maths Revision Notes

Daniel Guetta, 2007