You are on page 1of 29

Chapter 1

Vector Calculus

J. W. Elliott
Department of Mathematics
University of Hull

1.1 Motivation
In the absence of any mean flow, viscous dissipation and heat transfer, the propagation of sound in a stationary medium of uniform mean density po and pressure
po is governed by the linearised Euler equations of motion

Here co is the uniform speed of sound, p = p(r, t ) is the acoustic pressure and
+
u = u ( r ,t ) is the fluid velocity at a point P with position vector r = O P at
a time t. In addition we have considered the presence of both an unsteady body
force f , which vanishes in the undisturbed state, and volume sources q.
We can eliminate the velocity u as follows

"

1 a2p
- - ( f o g - p o v . u ) = Po-v
C; at2
at
at

Thus the governing equation for the pressure p is

(:)

Po-

Lecture Notes on the Mathematics of Acoustics

This is the inhomogeneous wave equation describing the production of sound by


the volume source q and body force f .

1.1.1

Velocitypotentiul

In the absence of body forces, where f = 0 , the flow is irrotational since


dw
dt

d
dt

- -- - ( V x u ) = V x

(2)

= --&V

x ( V p ) = 0.

(4)

Since the fluid is disturbed from an initial state of rest, we can set
w=Vxu=O,

so

u=Vc$(r,t),

(5)

where q5 is the velocity potential. The linearised equations of motion are then

where the momentum equation integrates to yield the linearised Bernoulli equation. Furthermore, it follows that

Outside the source region, where q = 0, the fluctuations u and p all propagate as
sound waves governed by the homogeneous wave equation, namely

The velocity u associated with the sound is called the acoustic particle velocity.
1.1.2 Plane waves

A plane wave is one in which

Here x measures distance in the direction of the unit vector i, which is (say) to the
right. In the absence of body forces the linearised equations of motion are

au
-ap +Po-

'

C;

dt

dx

= 0,

au = --ap

Po-

dt

dx'

It is easily seen that both p and u satisfy the classical wave equation

Vector Calculus

These have a general solution

p ( x , t ) = F ( x - cot) G ( x cot),
1
U ( X , t ) = -[ F ( x - cot) poco
where F and G are arbitrary, known as d'hlembert's solution. This solution
represents the linear superposition of two arbitrary disturbances of invariant form
both moving with a speed c, one to the right and one to the left. In particular for a
wave simply travelling to the right

P = P o c o ~ = F ( x - c ~ t ) ~ and

P
Z=-=poco,

(13)

where Z is the acoustic impedance and /3 = l/Z is the acoustic admittance.


1.1.2.1 Planar harmonic wave
More generally a planar harmonic wave of angular frequency w is of the form

p(rl t ) = a c o s ( k l x k2y ~ Q -Z w t ) bsin(klx


= Rcos(k1x
k z y + k3z - wt + c).

+ k 2 y + k3z

wt)
(14)

Taking t a n c = bla, with r = (z, y, z ) the position vector and k = ( k l , k 2 ,k 3 )


the wavenumber vector, then we can write

where A = ~ e ' 4is the complex amplitude. The points of constant phase, where
$ ( r l t ) k . r - wt = constant define a wave front, and such points satisfy
Thus at any given time t the wave-front is given by dq5 = k . dr = 0, so the
wavenumber vector is perpendicular to any wave-front (crest or trough). If we
move with a given crest (or trough) then we must have

where v+ is the phase-velocity. Thus the wave propagates in the direction of the
wavenumber vector at a wave-speed

and the wavelength, the distance between two adjacent crests, is X = 2 ~ / l k l .

Lecture Notes on the Mathematics of Acoustics

1.1.3 Spherical waves


A spherical wave is one in which
U =

u(r,t)P,

p = p(r, t ) ,

(19)

where r measures distance in the radial direction P. In the absence of body forces
the linearised equations of motion are

It is easily seen that p and u satisfy the (spherical) wave equation

since here

Thus we have a general solution

where F and G correspond to waves radiating outward to infinity from the origin,
and inward to the origin from infinity respectively.

1.1.3.1

Causality

Usually when sound is generated in open space we insist that the sound must not
anticipate its cause, the so-called causality condition. Thus G(r cot) = 0
since it must have existed for an infinitely long time in the past. This condition is
indistinguishable from Sommerfeld's radiation condition, namely

1.1.3.2 Harmonic spherical waves


A harmonic outward propagating spherical wave of frequency w is of the form

Vector Calculus

where K = W / Q is and A is a complex amplitude. Now if u = Re [ul(r,t ) ]then

which we can write as

1
-- poco
- I

p l ( r , t - ; )4

where q5 = tan-'(co/wr) = 7r/2 - tan-'(wr/co) is the phase angle. Thus, in


contrast to the planar case, the velocity and pressure are not in phase. Indeed as
r 4 0 , 4 -+ 7r/2 and p1 -iwpOru'. We also see the radiation impedance
N

':I I
-

Poco

dl + ( c o / w r )

I
poco.

As r 4 oo,4 -+ 0 the pressure and velocity are very nearly in phase, with the
spherical wave acting like a plane wave with p' pocoul.

1.2 Scalars and Vectors


Any entity that can be represented by a single real number, X E R, is called a
scalar. A vector, a E R3, is usually defined to be a quantity that has both a
magnitude and a direction.

1.2.1 Rectangular Cartesian coordinates


In 3-D Euclidean R3-space we draw, through a fixed point 0 , the origin, three
fixed, mutually perpendicular, lines Ox, Oy and Oz called the x-axis, y-axis and
z-axis respectively, and collectively known as rectangular Cartesian axes Oxyz,
as shown in Fig. 1.1. Place the thumb, index finger and middle finger of the
right hand, at right angles in the most natural way. If the index finger points
along Ox and the middle finger point along Oy then, for a right-handed set, the
thumb points along Oz. This statement exhibits cyclic symmetry in x, y, z since
it remains true when we replace x by y, y by z, and z by x.
The position of a general point P may be specified by drawing a line from 0
to P, of length r = OP. If a,P, y denote the angles O P makes with Ox, Oy Oz,

Lecture Notes on the Mathematics of Acoustics

Fig. 1.1 Rectangular Cartesian axes.

then we define the Cartesian coordinates of P to be the numbers

By considering the right-angled triangles OPL, O P M , O P N where L, M , N


are the feet of the perpendiculars from P to the x, y and z-axes, we see that
x

flength of OL,

y = flength of O M ,

= flength

of ON

We refer to P as the point P : (x,y, z ) . By Pythagoras

1.2.2 Geometric vectors

A geometrical representation of a vector is the directed line segment a = OA,


which has both a magnitude (the length OA) and a direction (indicated by an
arrow in Fig. 1.2).
+
--f
Let the vectors a = O A and b = O B represent adjacent sides of a parallelo+
gram. This representation is not unique, since we also have a = BC. The triangle
+
(or parallelogram) law states that c = a b is represented by the diagonal OC.
A unit vector is a vector of unit magnitude, often denoted by a circumflex.
We denote the modulus, or magnitude, of a vector a by the scalar a = la[,so ii =
a/lal is a unit vector lying in the direction of a. The zero vector 0, represented by
the point 0 ,is the only vector without an associated direction.

Vector Calculus

Fig. 1.2 A vector in the plane

1.2.3 Vector algebra


Let i, j and & denote unit vectors in the directions of the x-axis, y-axis and z-axis
respectively, then for any vector a E R3, we can uniquely write

where a l , az, a3 E R . For example i = (1,0,O),j = (0,1,O ) , 1; = (0,0,1).Then


the modulus or magnitude of a vector a = ( a l ,a2, a s ) is the scalar

Given vectors a = (a1,an, as), b = (bl ,bz , bg) and a scalar X E R , then

1.2.4

The scalar product

The scalar product of two vectors a = ( a l ,a:!,a s ) , b = ( b l ,bz, b3) is the scalar
a .b = albl

+ a2b2+ asb3 = JaJIbJcosQ

(35)

+
---+
where if a = O A and b = O B then 0 is the angle between O A and OB. Consequently

Two non-zero vectors a and b are orthogonal to each other if a . b = 0.

10

Lecture Notes on the Mathematics ofAcoustics

1.2.5 The vector product


The vector product of two vectors a = (al,an, as) and b = (bl, b2, b3) then

+
d
where the unit vector fi is perpendicular to both a = OA and b = OB. Here 8,
the angle between
and
is such that when looking along
the sense
of 8 increasing is clockwise. Two non-zero vectors a and b are parallel to each
+
other if a x b = 0. Also if a = OA, b = O B represent two adjacent sides of a
parallelogram, then la x bl = la1 Ibl sin B represents the area of the parallelogram.

d.

z,

a,

1.2.6 The triple scalar product


For three vectors a = (al,a2, ag),b = (bl , b2, bg) and c = (cl , cz, cg) the triple
scalar product, sometimes denoted by [a,b, c],is defined to be the scalar

If a = OA, b = OB and c = oc represent adjacent edges of a parallelepiped,


then la. (b x c) I represents the volume of the parallelepiped. Consequently three
non-zero vectors a, b and c are co-planar if and only if a . (b x c) = 0.

1.2.7 The triple vector product


The triple vector product of three vectors a = (al,a2, ag), b = (bl , b2,b3) and
c = (cl,cz ,c3) is the vector defined by
(a x b) x c = ( a . c)b - (b . c)a.

From which it follows that


(a x b) . (c x d) = (a.c)(b.d) - (a.d)(b.c).

(39)

11

Vector CaIcuIus

1.2.8 The standard basis


The set of vectors {i,j,&}, also denoted by {GI,62,63) and {il ,i2,is), is known
as the standard basis. It is a right-handed, orthonormal basis set satisfying

1.2.9 The position vector


The position vector of a general point P with Cartesian coordinates (x,y,t)rela+
tive to an origin 0 is the directed line segment r = OP, which we write as

The modulus r = O P = Irl, in agreement with Eq. (31) is given by

+
An alternative notation to the above is to write x = O P as the position vector of
the point P: (xl,x2,x3),and correspondingly write

1.3 Vector and Scalar Functions

1.3.1 Vector-vahedfunctions of a real variable


A vector-valued function F : R
IR3 of a real variable t assigns to each scalar
t E I = [a,b] c R a unique vector F E IR3.In Cartesian coordinates we write

where u ,v ,w : IR H IR are scalar functions oft. As {i,j,&) are uniform, we have

F'(t)

dF
du, dv
dw
== -1 + -j + -k
dt

dt

dt

dt

du dv dw
( d t ' dt ' dt .

(48)

12

Lecture Notes on the Mathematics of Acoustics

The product rules of differentiation for scalar functions f and R , are

1.3.2

Vector and scalar fields

+ +

Let r = O P = ZE
yj zG. A scalar function of position, or scalar field,
R : R3 + R assigns to every point P E R3,a unique scalar R = R ( r ) =
R ( x , y, z). A vector-valued function of position, or vector field, F : R3 -+ R3
assigns to every point P E R3,a unique vector F = F ( r ) = F ( x , y, z ) . If we
write

then u , v , w : R3 H R are the scalar functions of position. Furthermore

Similarly for R , = d R / d y , R , = d R / d z , F ,

=dF/dy

and F ,

= dF/dz.

1.3.3 Digerential operators


The gradient of a scalar field R = R ( r ) is a vector field defined by

The divergence of a vector field F = ( u ,v , w) is a scaiar field defined by

The curl of a vector field F = ( u ,v,W ) is another vector field defined by

curl F

=V x F =

k
d/ax d/dy 8/82 .
U

(56)

Vector Calculus

The directional derivative of a scalar field R = R ( r ) is defined to be

where n measures distance in the direction of some unit vector ii.

1.3.4 The chain rules


Let R = R ( x ,y, z ) and x = x ( t ) ,y = y(t) and z = z ( t ) ,then R = R ( t ) and

Let R = R ( x ,Y , z ) and x = x(J, 7 ,C), Y =


R = R ( J ,q, C) and the chain rule gives

Y(J,V,C)

and

= z(J,7 ,C ) then

with similar expressions for dR/dq and dR/d<.

1.3.5 The del operator


We define the (vector) del operator, V, also called simply del or nabla, by

We treat V as a symbolic, but non-commutative, vector, with (F . V )R = F .

(VR)and(FxV)R=Fx(VR)butF.V#V.FandFxV#VxF.
In addition

where
1.3.6

v2= V . V is the Laplacian, or (scalar) del-squared operator.


Vector identities

Given scalar fields A, 0 and vector fields F, G we have the product rules

V(X0)= XVR

+ RVX,

(62)

14

1.3.7

Lecture Notes on the Mathematics of Acoustics

Taylor's theorem

For a sufficiently differentiable scalar field, R, Taylor's theorem states that

Alternatively Taylor's theorem can be written more compactly as

60

R(r

+ 6r) - R(r) = V R . Sr + 0 (ISrI2) ,

(70)

where to first order we have the linear approximation SR = V R . Sr.

1.4 Curves and Surfaces in 3-D Space


1.4.1 Curves
A general point P on a curve C C R3has a position vector r

--t

= O P given

by

We regard Eq. (71) as a mapping r : R H R3, from I = [a,b] onto C. Here


Eq. (71) is not a unique parametric representation since any transformation t =
t(q) yields a new representation r = r(q), where dtldq 2 0 preserves the sense
of direction.
The unit tangent vector T to the curve C is given by

I/

ds
dr
where - = dt
dt

dr

SO
1

ds

dt = %T, (72)

15

Vector Calculus

where s measures the arc length along the curve C. As dsldt > 0, s = s(t) has
a unique inverse t = t(s), yielding the intrinsic equation of C,

1.4.2

Surfaces

1.4.2.1 Cartesian representations


The implicit and explicit representations of a surface S c R3 are given by
R(x,y,z)=0,

and

z=f(x,y),

for(x,y)~D,

where D is a region of the xy-plane. One possible choice for R is R


The unit normal ii to a surface R = z - f (x, y) = 0 is given by

(74)

= z - f (x, y).

since there are two sides to each surface.


1.4.2.2 Parametric representation
A general point P on a surface S C R3has a position vector r

O P given by

Here Eq. (76) is a mapping r : R2 ++ R3 from D*, a region of the uv-plane, onto
S . Now r = r(u, vO),in which v = vo is fixed, but u varies, is the equation of
one of the u-coordinate curves. Similarly r = r(uo, v) is the equation of one of
the v-coordinate curves. From Eq. (72) the tangents to the coordinate curves are

Consequently the unit normal ii to S must be of the form

For a surface S given by z

= f (x, y),

we see that

r = r(x, y) = r [x,Y, f (x, Y)I = xi

+ y j + f (5, Y)&

is the equation of S pararneterised by x and y. In this case

(79)

16

Lecture Notes on the Mathematics of Acoustics

1.5 Curvilinear Coordinate Systems


1.5.1 Cartesian coordinates
For every point P we can associate a right-handed orthonormal basis set of uni----+
form vectors { i , j ,L), where r = O P = xi yj z&. Let Q be a neighbouring
+
point to P, with PQ = 6r = (62,by, bz), then the elementary (cuboidal) volume
element at P, of volume SV, with bxi, 6yj and bzk representing adjacent sides,
has

+ +

6s'

1brI2 = bx2

+ by2 + 6z2,

6V = bx by bz.

(81)

1.5.2 Curvilinear coordinates


A new coordinate system is defined by a continuously differentiable mapping

x = x ( u ,0 , w ) ,
--+
Thus r = O P = xi

= y(u, u , w ) ,

= z ( u , u, w ) .

+ yj + +&, is given by the mapping r : IR3

(82)

IR3, where

Assuming a one-to-one mapping then there must exist a unique inverse mapping,
such that, except at isolated points, the Jacobian J of the mapping

Suppose the point P is specified uniquely by u = uo, u = uo and w = W O ,then


the vector equations of the u-, u- and w-coordinate lines through P are

Given scale factors h, = Ir, 1, hv = Ir, 1 and hw = Irw 1 , the unit tangents at P
to coordinate lines are given by

e,,

1 dr
--

h, du'

1dr
e, = -h, du '

For a neighbouring point Q, where

1 dr
hw dw

ew=--

17

Vector Calculus

the arc length 6s = 16rl is given by

bs2 = ) & I 2 = h: bu2 h: bv2 h i 6w2


2 [huh, 6u 6v (G, . 6,) h,h, 6v 6w (G, .6,)

(88)
huhw bu 6w (6, . G,)] .

A (cuboidal) volume element in uvw-space, with 6u Q,, 6v 6,, 6w 6 , representing adjacent sides, corresponds to a curvilinear volume element in xyz-space, with
volume

Finally the equations for the coordinate surfaces through P are given by

r = r(uO,v, w ) ,

r = r ( u ,v , wO).

= r ( u ,v0, w ) ,

(90)

Thus the coordinate surface, w = const, has a unit normal ii with an elementary
surface element of area bA, given by

1.5.3 Orthogonal curvilinear coordinates


For orthogonal curvilinear coordinates we require

with u , v and w chosen so that {G,, G,, 6,) is a right-handed orthonormal basis
set satisfying

e.., . e.,.= e , .. e , = e w ..e w = l ,


A

6,xGw=Gwr

e , . e.
,.= e , . e , = e , ~ e , = .
A

..

e,xe,=G,,

e, x e,

= e,.

0, (93)

(94)

We can show that the arc length 6s = l6rl is given by

bs2 = /brl2= h: 6u2 + hz 6v2 + h;dw2.

(95)

Similarly the elementary volume element has a volume

6V = Jbubvbw

h,h,hWSu6v6w.

and for the coordinate surface, w = const, with unit normal G,, we have

(96)

18

1S.3.l

Lecture Notes on the Mathematics of Acoustics

Vector#elds

A vector field F = ( F l ,F2,F3) can also be written in the form

where F,, F, and Fw are the components of F along the coordinate lines. For
orthogonal curvilinear coordinates F, = F B,, F, = F . B, and Fw = F .GW so
that

For example, as regards the position vector r = (x, y , z ) , we have

In general the basis set of vectors {B,, B,, 6,) are not uniform. Indeed

with similar formulae for d F / a v and d F / d w obtained by cyclic permutation.


1S.3.2 The differential operators

Given a scalar field R ( r ) and a vector field F ( r ) we can show that

Vector Calculus

1.5.4

19

Cylindricalpolar coordinates: ( R ,4, z )

The general point P with position vector r = O P = xi

In terms of the right-handed orthonormal vectors { R,

+ yj + t k is given by

6,& ) we have

so ~ ~ = ( ~ R R + R ~ R ) + s ~ ~ = G R R + R G
(107)
yielding scale factors hR = 1, h$ = Rand h, = 1. Hence we have

r=RR+zk,

1S.4.1 Scalar and vector operators

+ F,k then

If R = R(r) and F = F(r) = F ~ R F ~ $

1.5.5

Spherical polar coordinates: ( r ,8 , 4 )


---4

The general point P with position vector r = OP = zi


r = r s i n B ( c o s ~ i + s i n ~ j ) + r c o s O k , so

In terms of the right-handed orthonormal vectors {i.,

so

+ yj + z k is given by
lrI2=r2.

(113)

e l & )we have

b r = ( 6 r i . + r 6 t ) = ~ r f + r d B ~ + r s i n B b q 5 ~(114)
,

yielding scale factors h, = 1, he = r , h+ = r sin 8. Hence we have

6s2 = (6r12 = 6r2 + r26d2 + r2sin2 8 6 4 ~bV~ =~r2sin 8 67- 68 64. (1 15)

20

Lecture Notes on the Mathematics of Acoustics

1.5.5.1 Scalar and vector operators

If R = R ( r ) and F

V .F

= F ( r ) = F,f

1 a
r 2 dr

= --(r2F,)

+ ~~6+ F@$then

1 a
1 dF,
+ -(sin9 Fe) + -r sin 6 d6
r sin 6 d 4 '
1
r2 sin 6

VxF=-

1.6 Integrals
1.6.1 Line integrals
Suppose the position vector of a general point P on a curve C : A

= r ( t )= x(t)i

+ y(t)j + z ( t ) k ,

for to

B is

< t < tl.

(120)

Then, on C, the scalar field R = R ( t ) ,and the line integral of R along C from A
to B reduces to a standard (Riemann) integral of the form

ds

C : A-B

We also write & R ( r )ds to denote a line integral around a closed curve C.

(121)

1.6.2 Surface integrals


Suppose the position vector of a general point P on a surface S is

r = r ( u ,v ) = X ( U , v ) i + Y ( ~v ), j

Z(U,

v)k,

for ( u ,v ) E D.

(122)

Then on S, the scalar field R = R(u, v ) , and the surface integral of R over S
reduces to a double integral over the region D of thejut uv-plane, of the form

Vector Calculus

where we regard SSs as a single symbol.

1.6.3 Multiple integrals


We evaluate a double or triple integral by expressing it as a repeated integral. For
example the triple, or volume, integral of a scalar field R = R(r) over a region
C c R3 is given by

where dV = dx dy dz. If we integrate first with respect to z, keeping both x and


y fixed, then the result is a double integral of the form

where D is a domain of the xy-plane. This double integral is, in turn, evaluated by
first integrating with respect to x (or y), keeping y (or x) fixed, and then integrating
with respect to y (or x). Note that for a separable integrand, over a rectangular
domain, a multiple integral reduces to a product of standard (Riemann) integrals.
Moreover if the position vector of a general point P in C is given by

+ y(u, u, w)j + Z(U,u, w)&,

for (u, u, W) E C*,


(126)
where C* is the corresponding region in uuw-space, then R = R(u, v,w)and we
can represent Eq. (124) as the alternative triple integral

r = r ( u , u, w) = X ( U , v, w)i

where J is the Jacobian of the mapping.

Example 1.1

Using cylindrical polar coordinates, namely

< <

find the curved surface area A of the cylinder S : x2 + y2 = a2, 0 z H and


the volume V of the region C : x2 Y2 5 a2, 0 5 z 5 H within the cylinder.

Solution
of S is

----+

The position vector r = O P of a general point on the curved surface

r = r($,

Z) =

( a cos $, a sin 4, z ) .

(129)

Lecture Notes on the Mathematics of Acoustics

22

Here 8 and z parameterise S with D: 0 5 8 5 27~,0 5 z 5 H mapping to S.


Here

Ir4 x r,l = a (- sin 4i

+ cos 4j) x k-I = a lcos 4i + sin 4jJ= a ,

(130)

or SA = h$h, 64 6z = a 64 Sz. Thus the curved surface area of the cylinder is


given by

Here SV = J6R 64 6z where the Jacobian J = hah4h, = R. Thus if C* is the


region 0 R a, 0 I 4 27r, 0 < z < H which maps to C , then

< <

Example 1.2

<

Using spherical polar coordinates, namely

+ +

find the surface area A of the sphere S: x2 y2 z2 = a2 and the volume V of


the region C : x2 y2 + .z2 5 a2 within the sphere.
---+
Solution The position vector r = O P of a general point on the surface S is

Here 8 and
Also

4 parameterise S
re

=a

with D : 0

< 8 < T , 0 < 4 < 2 n mapping to S.

[cos 8 (cos 4; + sin 4j) - sin


+

r+ = a sin 6 (- sin +i cos 44) ,

(135)
(136)

or 6A = he h+68 64 = a2 sin 0 68 64. Thus the surface area of the sphere is given
by

Vector Calculus

23

Here 6V = J 6r 68 64 where the Jacobian J = h,hsh+ = r2sin 8. Thus if C* is


the region 0 L r 5 a, 0 5 8 n,0 4 5 2n which maps to C, then

<

V=

//L //L,
dV =

<

r 2 sin 8 d r do dq5

1.6.3.1 Alternative notation


+
When we take x = (x, y , z ) = O P to denote the position vector of the point P,
the triple integral Eq. (124) may be written as

Moreover, when the integrand is a function of several variables, then

Both notations indicate that here y is the dummy vector of integration, its value
ranging over all the points of the region C.

1.7 Integral Theorems


1.7.1 Green's lemma

Lemma 1.1 Consider a curve C in the xy-plane enclosing a convex simplyconnected region D in a positive anticlockwise sense, then

where fC fl(r)ds, denotes a line integral around a closed curve C.


This result is also known as Stokes's theorem in the plane.

1.7.2 Gauss's divergence theorem


Theorem 1.1 Let C be a finite region bounded by a simple closed surface S,
which has an outward-drawn unit normal ii. If F(r) = (PI,F2F3) is a vector

24

Lecture Notes on the Mathematics of Acoustics

field deJined and continually differentiable throughout C and on S, then

Corollary 1.1
on S, then

I f both R(r) and F(r) are continuously diflerentiable in C and

grad R dV =

JJ,

Rfi dA,

1.7.3 Stokes's theorem


Theorem 1.2 Let S be an open surface with a unit normal ii and C be a simple
closed boundary curve to S with a unit normal N and unit tangent vector T taken
in the positive anticlockwise sense ( N x T = ii). I f F(r ) is a vectorfield then

where ds is an element of arc length along the curve C.


By the triple scalar product identity Eq. (146) can be re-stated as

Corollary 1.2
on C, then

I f both R(r) and F(r) are continuously differentiable in S and

Vector Calculus

1.7.4 Green's theorems


Following from the divergence theorem (143) we have

Theorem 1.3

Green'sjrst theorem (or identity)

JJs$2d~ JJJ, div($V$) d v JJJ, [mvZ$+ ~ 9~ $. dv.


1
=

(150)

Theorem 1.4

Green's second theorem (or identity)

1.7.5 Fundamental solutions


The fundamental solution to the inhomogeneous Helmholtz equation

(v2+
where C

K')

9(r) = p(r),

for all r E C

c R3,

(152)

c R3 is a finite region of 3-D space, is any solution G = G(r;y ) to

Here 6(r) is the 3-D Dirac delta function, which has the property
f(a)
0

i f a E C,
i f a e C.

Thus G(r; y) is the solution at the point r due to a point source placed at the point
y. By the principle of superposition we have

where y is now the dummy variable of integration. For then

In the absence of boundaries, where C

= R3, we have the free-space solution

26

Lecture Notes on the Mathematics of Acoustics

Here i =
is the imaginary unit, and the sign is fixed by the causality
Sommerfeld radiation condition. This, in turn, yields

with the case K = 0 corresponding to Laplace's equation, V2+ = 0. Indeed near


r = y, Helmholtz's equation effectively reduces to Laplace's equation, having the
same singular behaviour. Moreover for any C c R it can be shown that

where H satisfies the homogeneous Helmholtz equation


( ~ ~ + ~ ~ ) H ( r ; y ) = f0o, r a l l r C ~ ~ ~ .

1.7.6

( 160)

The Green 'sfunction

Let C C R3 be a finite region of 3-D space bounded by a surface S , with an


outward-drawn unit normal fi. Suppose we wish to find 4 = 4(r) such that

(v2+ 6')

+(r) = p(r),

for all r E C,

(161)

subject to the boundary condition


4=f(r),

forallr~S,

where f or g are given scalar fields on S. From Green's 2nd identity Eq. (151) we
have

Now if

4 satisfies Eq. (161) and G satisfies Eq. (153), then

Moreover if the Green's function G also satisfies the boundary condition

G = 0,

for all r E S,

(165)

Vector Calculus

1.8 Suffix Notation


In suffix notation the equation c = a + b is simply written as

where a = ( a l ,a2, a s ) and b = (bl ,b2, b3). It is understood that this equation
holds for all values of the free suffix i. The summation convention states that
when a suffix is repeated in a single term, we sum over all values of that suffix.
Thus

To avoid ambiguity, never use any suffix more than twice in any one term. Thus

where we use two dummy variables to indicate that it is b dotted with a.

1.8.1 The Kronecker delta


The Kronecker delta dij consists of 9 quantities defined by

Jij

= [ I ].. =

1 for i = j,

( )
100

w e e1=

(1")

We see that the trace of the matrix I

Tr(I) = bii = d I 1 + d 2 p

+ ds3 = 3.

(172)

For any vector u = ( u ,up, u3)we have

62.3.u3. .
- 6zlul

di2~2 di3~3,

SO

d i j ~=
j Uir

(1 73)

Lecture Notes on the Mathematics of Acoustics

28

since as j runs over all values dij

# 0 only for j

= i.

Similarly

1.8.2 The alternating tensor


The alternating tensor c i j k is a set of 27 quantities defined by
+1
-1
0

if (i,j,lc) = (1,2,3) o r ( 2 , 3 , 1 ) o r ( 3 , 1 , 2 ) ,
if (i,j , k ) = ( l , 3 ,2) Or (2,1,3) Or ( 3 , 2 ,I ) ,
if any i ,j, k equal.

There are only 6 non-zero elements. A cyclic permutation of suffixes yields


= Ejki = Ekij,

Eijk

since the sign of

fijk

but

fijk = -Ejik,

(176)

is changed if any two suffixes are interchanged.

1.8.2.1 The vector product


The vector product of a = ( a l la2, a s ) and b = (bl , b2,b3) has an ith component
[a x b],= eijkajbk

(177)

for i = 1,2 and 3. Thus the triple scalar product is given by


a .( b X
1.8.2.2

C) = ai

[bX cIi = aiEijkbjck = cijkaibjck.

(178)

The 4relation

A very useful result is that


f i j k fimn = bjmbkn

bjndkrn.

from which it follows that


623. f z3
. . k = E i .~k = ' 1
fijkeijk = djjdkk
EijkEijn

djkbkj = (6kk)2- 6kk = 9 - 3 = 6 ,

= d j j d k n - djnbkj = 3dkn - dkn = 2dkn.

(Igo)
(181)
(l82)

Vector Calculus

29

1.8.3 Cartesian coordinates


When using suffix notation we use Cartesian axes 0x1x2x3, and we can state that
the standard basis {GI, 62, G3) is both orthonormal and right-handed by writing

-+

If r = O P is the position vector of the general point P : (xl ,x2, xs), then
r = xi$

is equivalent to

r = xlel

+ x2G2+ 5363.

(184)

Note that r2 = lrI2 = xixi, but that r2 # x:.


1.8.4 Differential operators

For a scalar field R(r) : R3 H R the ith component of the vector field V R ( r ) is
dR
d ~1 i

[VR],
=-

d
where V = B.-.
dxj

(18 5 )

Thus for a vector field F = (Fl , F 2 , F3) we have

Note that, in suffix notation the chain rule for a mapping r = r(u, v, w), is simply

1.8.4.1

Second order operators

Here div(grad R), also known as the Laplacian of a scalar field R(r), is given by

Also the ith component of the Laplacian of a vector field F ( r ) is

Similarly we have grad(div F), of which the ith component is given by

Lecture Notes on the Mathematics of Acoustics

Problems
--t

Exercise 1.1 Let r = O P = (x, y, 2) be the position vector of the general point
P and G = G(r) be the scalar field

where y = (x,, y,, 2,) is some given uniform vector. Obtain V G and show that
G is a solution to

(v2

X2) G = 0,

for all R

# 0.

Show that for the case X = * i ~ ,where K > 0, then the condition of an outward
propagating wave Ge-'"Qt, is equivalent to the Sommerfeld condition

Exercise 1.2

(9

Show that one corollary of the divergence theorem is that

where C is a finite region C of 3-D space enclosed by the surface S , and


in which n measures distances in the direction of the outward-drawn unit
normal, fi, to S .
(ii) By considering the scalar field G of Ex. 1.1, and a sphere S : Ir - y 1 = ro,
show that

Hence confirm the above corollary for this G by direct numerical evaluation
for the case where C is the region enclosed by spheres Ir - yl = b and
Ir - yI = E where 0 < E < b. To evaluate the volume integral you may wish
to use spherical polars

Confirm that this corollary does not hold for the case where C is the region
inside a single sphere S: Ir - y 1 = ro and explain this result.

31

Vector Calculus

Exercise 1.3
identities

Use suffix notation and the summation convention to prove the

(i) V . ( F x G ) = G . ( V x F ) - F . ( V x G ) ,
(ii) V x (F x G) = ( V . G ) F - ( V . F ) G + ( G . V ) F - ( F . V ) G ,
(iii) V ( F . G ) = ( F . V ) G + ( G . V ) F + Gx ( V x F ) + F x

(Vx G).