You are on page 1of 575

Theoretical Physics I-Mathematical Methods

Daniel Erenso and Victor Montemayor

February 25, 2019


ii
Contents

Introduction xi

I Mathematical Methods in Physics I 1


1 Series and Convergence 3
1.1 Sequence and series . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Testing series for convergence . . . . . . . . . . . . . . . . . . . . 7
1.3 Series representations of real functions . . . . . . . . . . . . . . . 9
1.4 Homework Assignment 1 . . . . . . . . . . . . . . . . . . . . . . . 18
1.5 Homework Assignment 2 . . . . . . . . . . . . . . . . . . . . . . . 19

2 Complex Numbers, functions, and series 21


2.1 Complex numbers . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2 Complex In…nite Series and the Circle of Convergence . . . . . . 28
2.3 Powers and Roots of complex numbers . . . . . . . . . . . . . . . 32
2.4 Algebraic vs transcendental functions . . . . . . . . . . . . . . . . 34
2.5 Homework Assignment 3 . . . . . . . . . . . . . . . . . . . . . . . 40
2.6 Homework Assignment 4 . . . . . . . . . . . . . . . . . . . . . . . 42
2.7 Homework Assignment 5 . . . . . . . . . . . . . . . . . . . . . . . 43

3 Vectors, Lines, and Planes 45


3.1 An Overview of Vector Fundamentals . . . . . . . . . . . . . . . 45
3.2 Vector Multiplication . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.3 Equations of a straight line . . . . . . . . . . . . . . . . . . . . . 49
3.4 Equation of a plane . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.5 Homework Assignment 6 . . . . . . . . . . . . . . . . . . . . . . . 53

4 Matrix and determinants 55


4.1 Important Terminologies . . . . . . . . . . . . . . . . . . . . . . . 55
4.2 Matrix Arithmetic and Manipulation . . . . . . . . . . . . . . . 56
4.3 Matrix representation of a set of linear equations . . . . . . . . . 57
4.4 Matrix application: solving a set of linear equations . . . . . . . 58
4.5 Determinant of a square matrix . . . . . . . . . . . . . . . . . . . 63

iii
iv CONTENTS

4.6 Cramer’s Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66


4.7 The adjoint and inverse of a matrix . . . . . . . . . . . . . . . . . 67
4.8 Orthogonal matrices and the rotation matrix . . . . . . . . . . . 72
4.9 Linear independence . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.10 Gram-Schmidt orthogonalization . . . . . . . . . . . . . . . . . . 78
4.11 Homework Assignment 7 . . . . . . . . . . . . . . . . . . . . . . . 80
4.12 Homework Assignment 8 . . . . . . . . . . . . . . . . . . . . . . . 81
4.13 Homework Assignment 9 . . . . . . . . . . . . . . . . . . . . . . . 83

5 Introduction to di¤erential calculus-I 85


5.1 Partial di¤erentiation . . . . . . . . . . . . . . . . . . . . . . . . 85
5.2 Total di¤erential . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.3 The multivariable form of the Chain rule . . . . . . . . . . . . . . 90
5.4 Extremum (Max/Min) problems . . . . . . . . . . . . . . . . . . 94
5.5 The Method of Lagrangian Multipliers . . . . . . . . . . . . . . . 97
5.6 Change of Variables . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.7 Legendre Transformations . . . . . . . . . . . . . . . . . . . . . . 106
5.8 Homework Assignment 10 . . . . . . . . . . . . . . . . . . . . . . 107
5.9 Homework Assignment 11 . . . . . . . . . . . . . . . . . . . . . . 108

6 Introduction to Di¤erential Calculus II 109


6.1 Linear ordinary di¤erential equations . . . . . . . . . . . . . . . . 109
6.1.1 First-order LODE with constant coe¢ cients . . . . . . . . 110
6.2 The …rst-order DE and exact di¤erential . . . . . . . . . . . . . . 117
6.3 First-order DE and none exact total di¤erential . . . . . . . . . . 122
6.4 Higher-Order ODE’s . . . . . . . . . . . . . . . . . . . . . . . . . 126
6.4.1 Homogeneous ODE’s with constant coe¢ cient . . . . . . . 127
6.4.2 None homogeneous ODE’s with Constant Coe¢ cients . . 137
6.5 The method of superposition and the particular solution. . . . . 141
6.6 The method of successive integration . . . . . . . . . . . . . . . . 145
6.7 Partial Di¤erential Equations . . . . . . . . . . . . . . . . . . . . 154
6.8 Homework Assignment 12 . . . . . . . . . . . . . . . . . . . . . . 159
6.9 Homework Assignment 13 . . . . . . . . . . . . . . . . . . . . . . 159
6.10 Homework Assignment 14 . . . . . . . . . . . . . . . . . . . . . . 160
6.11 Homework Assignment 15 . . . . . . . . . . . . . . . . . . . . . . 161

7 Integral calculus-scalar functions 163


7.1 Integration in Cartesian coordinates . . . . . . . . . . . . . . . . 163
7.1.1 Single and multiple integrals . . . . . . . . . . . . . . . . 163
7.1.2 Physical Applications . . . . . . . . . . . . . . . . . . . . 171
7.2 Integration in curvilinear coordinates . . . . . . . . . . . . . . . . 183
7.2.1 One- and two–dimensional curvilinear Coordinates . . . . 184
7.2.2 3-D Curvilinear coordinates-cylindrical . . . . . . . . . . 186
7.2.3 3-D Curvilinear Coordinate Systems: spherical Coordinates 188
7.3 Homework Assignment 16 . . . . . . . . . . . . . . . . . . . . . . 191
7.4 Homework Assignment 17 . . . . . . . . . . . . . . . . . . . . . . 192
CONTENTS v

7.5 Homework Assignment 18 . . . . . . . . . . . . . . . . . . . . . . 193

8 Vector Calculus 195


8.1 Vector Products . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
8.1.1 Physical applications . . . . . . . . . . . . . . . . . . . . . 196
8.1.2 Vectors derivatives . . . . . . . . . . . . . . . . . . . . . . 198
8.2 Di¤erential Operators . . . . . . . . . . . . . . . . . . . . . . . . 202
8.2.1 The gradient operator and directional derivative . . . . . 202
8.2.2 The Divergence, the Curl, and the Laplacian . . . . . . . 206
8.3 Single vector function integrals . . . . . . . . . . . . . . . . . . . 217
8.3.1 Line vector integrals . . . . . . . . . . . . . . . . . . . . . 217
8.3.2 Conservative vector …elds . . . . . . . . . . . . . . . . . . 219
8.3.3 Conservative Fields and Exact Di¤erentials . . . . . . . . 221
8.4 Multiple vector integrals . . . . . . . . . . . . . . . . . . . . . . . 222
8.4.1 Green’s Theorem . . . . . . . . . . . . . . . . . . . . . . . 222
8.4.2 The Stokes’Theorem . . . . . . . . . . . . . . . . . . . . 226
8.4.3 The Divergence Theorem . . . . . . . . . . . . . . . . . . 231
8.4.4 More Examples on Divergence and Stokes theorems . . . 233
8.4.5 Applications of the Divergence and Stoke’s theorems . . . 238

II Mathematical Methods in Physics II 241


9 Introduction to the Calculus of Variations 243
9.1 Geodesic and stationary points . . . . . . . . . . . . . . . . . . . 243
9.2 The general problem . . . . . . . . . . . . . . . . . . . . . . . . 247
9.3 Appplications: The Brachystochrone Problem . . . . . . . . . . . 250
9.4 Applications: classical mechanics . . . . . . . . . . . . . . . . . . 253
9.5 Physical application of the Euler-Lagrange equation . . . . . . . 254

10 Introduction to the Eigenvalue Problem 261


10.1 Matrix Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
10.2 Orthogonal matrices and the rotational operator . . . . . . . . . 264
10.3 Eigenvalues and Eigenvectors . . . . . . . . . . . . . . . . . . . . 265
10.4 Physical applications . . . . . . . . . . . . . . . . . . . . . . . . . 270

11 Special functions 279


11.1 The factorial and gamma function . . . . . . . . . . . . . . . . . 279
11.2 The Beta Function . . . . . . . . . . . . . . . . . . . . . . . . . . 283
11.3 Stirling’s Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
11.4 The Error Function . . . . . . . . . . . . . . . . . . . . . . . . . . 288
11.5 Elliptic Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
11.6 The Dirac delta function . . . . . . . . . . . . . . . . . . . . . . . 294
vi CONTENTS

12 Power Series Solutions to Di¤erential Equations 303


12.1 Power series substitution . . . . . . . . . . . . . . . . . . . . . . . 303
12.2 Orthogonal vectors and Dirac Notation . . . . . . . . . . . . . . . 307
12.3 Complete Sets of Functions . . . . . . . . . . . . . . . . . . . . . 310
12.4 The Legendre Di¤erential Equation . . . . . . . . . . . . . . . . . 313
12.5 The Legendre Polynomials . . . . . . . . . . . . . . . . . . . . . . 316
12.6 The Generating Function for the Legendre Polynomials . . . . . 319
12.7 Legendre Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
12.8 The Associated Legendre Di¤erential Equation . . . . . . . . . . 329
12.9 Spherical Harmonics and the addition theorem . . . . . . . . . . 333
12.10The Method of Frobenius and the Bessel equation . . . . . . . . 341
12.11The orthogonality of the Bessel Functions . . . . . . . . . . . . . 348
12.12Fuch’s theorem for 2-nd DE . . . . . . . . . . . . . . . . . . . . . 354

13 A partial di¤erential equation 359


13.1 PDE in physics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
13.2 Laplace’s equation in Cartesian coordinates . . . . . . . . . . . . 361
13.3 Laplace’s equation in spherical coordinates . . . . . . . . . . . . . 367
13.4 Laplace’s equation in cylindrical coordinates . . . . . . . . . . . . 373
13.5 Poisson’s Equation . . . . . . . . . . . . . . . . . . . . . . . . . . 380

14 Functions of Complex Variables 387


14.1 Complex variables and functions . . . . . . . . . . . . . . . . . . 387
14.2 Analytic Functions . . . . . . . . . . . . . . . . . . . . . . . . . . 389
14.3 Important Terminologies . . . . . . . . . . . . . . . . . . . . . . . 391
14.4 Contour Integration and Cauchy’s Theorem . . . . . . . . . . . . 396
14.5 Cauchy’s Integral formula . . . . . . . . . . . . . . . . . . . . . . 398
14.6 Laurent’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . 404
14.7 Residues and the Residue Theorem . . . . . . . . . . . . . . . . . 405
14.8 Methods of …nding residues . . . . . . . . . . . . . . . . . . . . . 407
14.9 Applications of the Residue Theorem . . . . . . . . . . . . . . . . 412
14.10Standard Methods of Integration using Contour Integrals . . . . 418
14.11The Kramers-Kronig Relations . . . . . . . . . . . . . . . . . . . 419
14.12The revised Residue Theorem . . . . . . . . . . . . . . . . . . . . 420

15 Laplace transform 433


15.1 Inverse Laplace Transform . . . . . . . . . . . . . . . . . . . . . . 438
15.2 Applications of Laplace transforms . . . . . . . . . . . . . . . . . 440

16 Fourier series and transform 449


16.1 Introduction to Fourier Series . . . . . . . . . . . . . . . . . . . . 449
16.2 The Fourier Series Expansion of a Periodic Function . . . . . . . 452
16.3 Dirichlet Conditions . . . . . . . . . . . . . . . . . . . . . . . . . 455
16.4 Fourier series with spatial and temporal arguments . . . . . . . . 457
16.5 The Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . . 460
16.6 The Dirac Delta Function and the Fourier transform . . . . . . . 465
CONTENTS vii

16.7 Applications of the Fourier Transform . . . . . . . . . . . . . . . 467


16.8 Fourier Transforms and Convolution . . . . . . . . . . . . . . . . 473

III Mathematical Methods in physics III 477


17 Manifolds 479
17.1 What is a Manifold? . . . . . . . . . . . . . . . . . . . . . . . . . 479
17.2 Curves and surfaces in a Manifold . . . . . . . . . . . . . . . . . 481
17.3 Coordinate transformations and summation convention . . . . . . 484
17.4 The Riemannian geometry . . . . . . . . . . . . . . . . . . . . . . 488
17.5 Intrinsic and extrinsic geometry and the metric . . . . . . . . . . 491
17.6 Length, area, and volume . . . . . . . . . . . . . . . . . . . . . . 498
17.7 Local Cartesian coordinates and tangent space . . . . . . . . . . 506
17.8 The signature of a manifold . . . . . . . . . . . . . . . . . . . . . 511

18 Vector Calculus on manifolds 515


18.1 The tangent vector . . . . . . . . . . . . . . . . . . . . . . . . . . 515
18.2 The basis vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . 516
18.3 The metric function and coordinate transformations . . . . . . . 517
18.3.1 Raising and lowering vector indices . . . . . . . . . . . . . 523
18.4 The inner product and null vectors . . . . . . . . . . . . . . 524
18.5 The a¢ ne connections . . . . . . . . . . . . . . . . . . . . . . . . 525
18.6 Local geodesic and Cartesian coordinates . . . . . . . . . . . . . 533
18.7 The gradient, the divergence, the curl on a manifold . . . . . . . 536
18.8 Intrinsic derivative of a vector along a curve . . . . . . . . . . . . 538
18.9 Parallel transport . . . . . . . . . . . . . . . . . . . . . . . . . . . 539
18.10Null curves, non-null curves, and a¢ ne parameter . . . . . . . . . 540
18.11Refreshment from theoretical physics: the calculus of variation . 542
18.12The geodesic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 544
18.13Stationary property of the non-null geodesic . . . . . . . . . . . . 546

19 Tensor Calculus on manifolds 549


19.1 Tensors …elds and rank of a tensor . . . . . . . . . . . . . . . . . 549
19.2 Mapping tensors into tensors . . . . . . . . . . . . . . . . . . . . 554
19.3 Elementary tensor operations . . . . . . . . . . . . . . . . . . . . 555
19.4 Tensors and coordinate transformations . . . . . . . . . . . . . . 557
19.5 Tensor equations and the quotient theorem . . . . . . . . . . . . 558
19.6 Covariant derivatives of a tensor . . . . . . . . . . . . . . . . . . 559
19.7 Intrinsic derivative . . . . . . . . . . . . . . . . . . . . . . . . . . 561
viii CONTENTS
Preface

This material is built upon the lecture I have given for Topics in Theoretical
Physics I and II (PHYS 3150 and 3160) at Middle Tennessee State University
for more than a decade.

ix
x PREFACE
Introduction

xi
xii INTRODUCTION
Part I

Mathematical Methods in
Physics I

1
Chapter 1

Series and Convergence

This lecture is based on the assumption that you have some basic knowledge
of sequence, series, and tests for convergence of series. Therefore, I will only
summarize the basics with more emphasis on examples.

1.1 Sequence and series


A sequence: a sequence is an ordered list of objects (or events). Like a set, it
contains members (also called elements, or terms), and the number of ordered
element (possibly in…nite) is called the length of the sequence. Unlike a set,
order matters, and exactly the same elements can appear multiple times at
di¤erent positions in the sequence. A sequence is a discrete function.
A series: The sum of terms of a sequence is a series. More precisely, if
(x1 ; x2 ; x3 ; :::) is a sequence, one may consider the sequence of partial sums
(S1 ; S2 ; S3 ; ::Sn :), with
n
X
Sn = x1 + x2 + x3 :: + xn = xk : (1.1)
k=0

Geometric series: Let’s consider two simple idealistic biological and physical
processes that can be described by geometric progressions.
i. bacteria in a culture doubles every hour

2; 4; 8; 16; 32; :::; (1.2)


ii. A bouncing ball that raises each time 2/3 of the height of the previous
bounce:
2 4 8 16
1; ; ; ; ; :::; (1.3)
3 9 27 81
Suppose we want to determine the total number of bacteria or the total height
the ball raises in a given period of time we add the terms in the above expres-
sions. If we denote the sum of those terms by Sn , we may write the above

3
4 CHAPTER 1. SERIES AND CONVERGENCE

Figure 1.1: Bacteria growth in diluted human blood sample.

Figure 1.2: A bouncing ball.

expressions as
Sn = 2 1 + 2 + 22 + 23 + :::2n (1.4)
2 3 n
2 2 2 2
Sn = 1 + + + + ::: (1.5)
3 3 3 3
P
Using the summation sign, ; these can be put in the form
n
X n
X
Sn = 2 2k = 2 rk ;
k=0 k=0
n
X k Xn
2
Sn = = rk ; (1.6)
3
k=0 k=0

where r = 2 for the doubling bacteria and r = 2=3 for the bouncing ball. In
general as n ! 1; the series
1
X
S = lim Sn = a rk (1.7)
n!1
k=0

is known as geometric series. Using


1
X
rk = 1 + r + r2 + r3 + r4 + ::: (1.8)
k=0
1.1. SEQUENCE AND SERIES 5

and
1
X
rk+1 = r + r2 + r3 + r4 + ::: (1.9)
k=0

we can write
1
X 1
X 1
X
k k+1
r r = (1 r) rk
k=0 k=0 k=0
= 1 + r + r2 + r3 + r4 + ::: r + r2 + r3 + r4 + ::: = 1: (1.10)

Then for jrj < 1; we have


1
X 1
rk = (1.11)
1 r
k=0

so that the geometric series becomes


1
X a
S = lim Sn = a rk = : (1.12)
n!1 1 r
k=0

Example 1.1 Two trains move toward one another with a speed vo . A bird
(an “ideal physics bird”) ‡ies back and forth between the two trains with
a speed vbird = vo , where > 1. Find the total distance traveled by the
bird before the trains crash if they start a distance Do apart.
Solution: Consider the …rst four consecutive back and forth movement of the
bird as shown in the …gure below.The bird starts from train one ( T1 ) and
moves towards train two ( T2 ). Let the time it takes to reach to T2 is t1
and during this time it has traveled a distance x1 as measured from the
initial position of T1 . Then we can express the distance x1

x1 = D0 v0 t1 : (1.13)

Since the bird is moving with a constant speed vo ; the distance the bird
traveled, x1 ; over the time, t1 ; is related to its speed by
x1 x1
vo = ) t1 = : (1.14)
t1 vo
Substituting Eq. (1.14) into Eq. (1.13), we …nd
x1
x1 = D 0 ) x1 = D0 : (1.15)
1+
From Fig. 1.3, we note that

D1 = D0 2 (D0 x1 ) = 2x1 D0 ) D1 = 2 D0 D0
1+
1
) D1 = D0 : (1.16)
+1
6 CHAPTER 1. SERIES AND CONVERGENCE

Figure 1.3: The motion of the bird and the two trains.

Following a similar procedure and referring to Fig. 1.3, the distance the
bird traveled when it ‡ies back to the …rst train, x2 ; can be expressed as

x2 = D1 v0 t 2 (1.17)

so that using
x2 x2
vo = ) t2 = (1.18)
t2 vo
we …nd
x2
x2 = D1 v0 t2 = D1 ) x2 = D1 : (1.19)
1+
Substituting Eq. (1.16) into Eq. (1.19), one …nds

1
x2 = D0 : (1.20)
1+ +1

In a similar way for x3 ; we can easily show that

x3 = D2 : (1.21)
1+
Using
2
1 1
D2 = D1 = D0 (1.22)
+1 +1
1.2. TESTING SERIES FOR CONVERGENCE 7

we …nd
2
1
x3 = D0 : (1.23)
1+ +1
Then the total distance the bird ‡ies before the two trains crash can be
expressed as an in…nite series,
1
X = x1 + x2 + x3 + ::: = D0 + D0
1+ 1+ +1
2
1
+ D0 :::
1+ +1
" #
0 1 2
1 1 1
)X= D0 + + + ::: ; (1.24)
1+ +1 +1 +1

which can be put in the form


1
X
X=a rn ; (1.25)
n=0

where
1
a= D0 ; r = < 1: (1.26)
1+ +1
Eq. (1.25) is a geometric series and its sum can easily be determined using
the relation in Eq. (1.12)
1
!
X a 1 1
k
X=a r = = D0 1 = D0 : (1.27)
1 r 1+ 1 +1
2
k=0

The result in Eq. (1.27) is the total distance that the bird has traveled
before the two trains crashed.

1.2 Testing series for convergence


If an in…nite series, like the geometric series we studied in the previous section,
has a …nite sum it is said to be convergent otherwise it is divergent. There are
di¤erent ways of testing whether a given in…nite series,
1
X
a0 + a1 + a2 + :::: = an ; (1.28)
n=0

is a convergent or divergent series. Here we discuss some of the commonly used


to test the convergence of an in…nite series
(1) Preliminary test: If an does not tend to zero as n ! 1;
lim an 6= 0; (1.29)
n!1

the in…nite series is a divergent series.


8 CHAPTER 1. SERIES AND CONVERGENCE

(2) Absolute convergence test: Suppose the sum of the absolute values of all
the terms in the in…nite series in Eq. (1.28) converges, then the in…nite
series is called absolutely convergent series. An in…nite series that is ab-
solutely convergent is a convergent series. In other words, if
1
X
ja0 j + ja1 j + ja2 j + :::: = jan j (1.30)
n=0

is a convergent series, the in…nite series in Eq. (1.28) is a convergent series.


(3) The comparison test: Suppose there exists a convergent in…nite series with
all the terms positive,
1
X
bn = b0 + b1 + b2 + ::::; (1.31)
n=0

(i.e. bn > 0). Compare the nth term in Eq. (1.31) (bn ) with the cor-
responding nth term in Eq. (1.28) (an ). If jan j bn for all n; then the
series in Eq. (1.28) is a convergent series.
(4) The integral test: In the series in Eq. (1.28), if 0 < an+1 an for n > N ,
then one can use the integral test,
Z 1
I= an dn: (1.32)

The series in Eq. (1.28) is convergent if this integral is …nite otherwise it is


a divergent series. In Eq. (1.32), only the upper limit of the integration is
included. This means the integral need to be evaluated only at the upper
limit to test the convergence.
(5) The ratio test: In the ratio test, for the in…nite series in Eq. (1.28), we
…rst evaluate the ration de…ned by

an+1
= lim : (1.33)
n!1 an

The convergence of the series will then be determined by the value of the
ratio, ; according to
8
< < 1; the series is convergent,
= > 1; the series is divergent, (1.34)
:
= 1 use a di¤erent test.

Example 1.2 Consider the in…nite series given by


1 2 6 24
1+ + + + ; :::; (1.35)
6 120 5040 362; 880
1.3. SERIES REPRESENTATIONS OF REAL FUNCTIONS 9

(a) Rewrite this series using standard summation notation.

(b) Use the ratio test to check the convergence of this series.

Solution:

(a) We …rst rewrite the series using some general expression in terms of n
that generates all the terms in the series. In most cases we look for some
kind of relation that can be expressed using factorial and/or exponential
function. To this end, we note that

1 2 6 1 1 2 1 3 2 4 3 2
1+ + + + ::: = + + + + :::
6 120 5040 0! 3! 5! 7! 9!
0! 1! 2! 3! 4! n!
= + + + + :: :::; (1.36)
1! 3! 5! 7! 9! (2n + 1)!

so that the series can be expressed as

X1
1 2 6 24
1+ + + + ; ::: = an ; (1.37)
6 120 5040 362; 880 n=0

where
n!
an = : (1.38)
(2n + 1)!

(b) We use the ratio test to determine if the series is convergent or divergent.
Using the result determined in (a), Eqs. (?? and (??, the ratio, ;

(n+1)! (n+1)(n!)
an+1 (2(n+1)+1)! (2n+3)(2n+2)(2n+1)!
= lim = lim n!
= lim n!
n!1 an n!1
(2n+1)!
n!1
(2n+1)!

n+1 n 1
= lim ) ' lim = lim =0
n!1 (2n + 3) (2n + 2) n!1 4n2 n!1 4n

which shows, < 1; and the series is a convergent series according to Eq.
(1.34).

1.3 Series representations of real functions


This section introduces to how we express a given well de…ned real function in a
series form. We focus on three form of series representation of a function P (x):
Power series, Taylor, and Maclaurin’s series. The series representation of this
function could be convergent or divergent. The convergence or divergence is
determined by the domain of the variable, x, that the function depends on.
We will see how to determine the range of the domain for which the series is
convergent (i.e. the interval of convergence).
10 CHAPTER 1. SERIES AND CONVERGENCE

i. Power Series: A series in which the nth term is a constant, bn ; times the a
function, (x a)n ,

1
X n 2
P (x) = bn (x a) = b0 + b1 (x a) + b2 (x a) + :::; (1.39)
n=0

where a is real constant and x is real variable (i.e. x 2 R). Note that a
power series can be convergent or divergent. It depends on the variable x.

ii. Taylor series: Taylor series is a power series in which the coe¢ cient in the
nth term of the series, bn ; is given by the nth derivative of the function
P (x) evaluated at x = a; and divided by n factorial,

1
X n 1 dn P (x)
P (x) = bn (x a) , where bn = : (1.40)
n=0
n! dxn x=a

iii. Maclaurin’s Series: Maclaurin’s Series is a Taylor series in which the con-
stant, a = 0:

1
X 1 dn P (x)
P (x) = bn xn , where bn = (1.41)
n=0
n! dxn x=0

Any function, f (x); that is di¤ erentiable for all values of x in the speci…ed
n
domain, can be expressed in Taylor or Maclaurin’s series. That means if d dx f (x)
n

exists for all n 0 and x 2 R domain, we can write

1
X 1 dn f (x)
f (x) = bn (x a)n , where, bn = (1.42)
n=0
n! dxn x=a
1.3. SERIES REPRESENTATIONS OF REAL FUNCTIONS 11

Taylor Series for some common functions:


X1
( 1)n 2n+1
sin x = x
n=0
(2n + 1)!
x3 x5
=x + + ::: convergent for all x 2 R (1.43a)
3! 5!
X1
( 1)n 2n
cos x = x
n=0
(2n)!
x2 x4
=1 + :::convergent for all x 2 R (1.43b)
2! 4!
X1
1 n
ex = x
n=0
n!
x2
=1+x+ + :::convergent for all x 2 R (1.43c)
2!
X1
( 1)n+1 n
ln(1 + x) = x
n=1
n
x2 x3
=x + :::convergent for 1<x 1; (1.43d)
2 3
1
X p
(1 + x)p = xn
n
n=1
p(p 1) 2
= 1 + px + x + :::convergent for all jxj < 1 (1.43e)
2!
where
p p (p 1) (p 2) (p 3) ::: (p n + 1)
= (1.44)
n n!
and it is called binomial coe¢ cient.
Power series expansions of functions are unique. That is, if you use di¤ erent
methods for …nding a power-series expansion of a function, then you must get
the same result. A power series expansion for a function must be the power
series expansion for that function!

Example 1.3
(a) Show that the Maclaurin’s Series expansion for the function

f (x) = ln(1 + x) (1.45)

is expressible as
X1
( 1)n+1 n
ln(1 + x) = x : (1.46)
n=1
n

(b) Find the interval of convergence for this series.


12 CHAPTER 1. SERIES AND CONVERGENCE

Solution:
(a) Using Eq. (1.41)
1
X 1 dn P (x)
P (x) = bn xn , where bn = (1.47)
n=0
n! dxn x=0

we have
1
b0 =
ln(1 + x)jx=0 ) b0 = 0; (1.48)
0!
1 d 1
b1 = ln(1 + x) = ) b1 = 1; (1.49)
1! dx x=0 1 + x x=0

1 d2 1 d 1
b2 = ln(1 + x) =
2! dx2 x=0 2! dx 1+x x=0
1
1 1 ( 1)
= ) b2 = ; (1.50)
2! (1 + x)2 2
x=0

!
1 d3 1 d 1
b3 = ln(1 + x) = 2
3! dx3 x=0 3! dx (1 + x) x=0
2 2
1 ( 1) (1 2) ( 1)
= 3 ) b3 = ; (1.51)
3 2! (1 + x) 3
x=0

and
!
1 d4 1 d 1
b4 = ln(1 + x) = 2
4! dx4 x=0 4! dx (1 + x) x=0
3 3 2
1 ( 1) (1 2 3) 1 ( 1) 3! ( 1)
= 4 ) b4 = 4 = ;
4 3! (1 + x) 4 3! (1 + x) 4
x=0 x=0
(1.52)

so that one can easily see that


(
( 1)n+1
bn = n ; for n = 1; 2; 3::: (1.53)
0 n=0

There follows that


1
X X1
( 1)n+1 n
ln(1 + x) = bn xn = x : (1.54)
n=0 n=1
n

This is the Maclaurin’s Series.


1.3. SERIES REPRESENTATIONS OF REAL FUNCTIONS 13

(a) To …nd the interval of convergence we use the ratio test. The series con-
verges when
an+1 (x)
= lim < 1: (1.55)
n!1 an (x)
Using
( 1)n+2 n+1
an+1 (x) = x (1.56)
n+1
and
( 1)n+1 n
an (x) = x ; (1.57)
n
we …nd

( 1)n+2 n+1 n nx
= lim x = lim = jxj < 1
n!1 n+1 ( 1)n+1 xn n!1 n+1
) 1<x<1 (1.58)

We note that at x = 1, the series becomes


1
( P
X ( 1)n+1 n 1 ( 1)n+1
x = n=1P1 n 1 ; for x = 1 (1.59)
n=1
n n=1 n for x = 1

and the ratio test can not tell us whether the series is a convergent or
divergent series. So one must use a di¤ erent test. A simple integral test
Z1
dn 1
= ln (x)j = 1 (1.60)
n

shows that for x = 1, the series


X1 X1
( 1)n+1 n 1
x = (1.61)
n=1
n n=1
n

is a divergent series. However, for x = 1 the series becomes an alternating


series
X1 1
( 1)n+1 n X ( 1)n+1
x = (1.62)
n=1
n n=1
n
which is a convergent series.
Exercise: Prove that an alternating series of the form
1
X
( 1)n+1 an (1.63)
n=1

with an > 0 is a convergent series if an is monotonically decreas-


ing.
14 CHAPTER 1. SERIES AND CONVERGENCE

Therefore, the interval of convergence for the Maclaurin’s Series is 1<x 1:


Example 1.4 In Special Relativity the total energy of an object of (rest) mass
2 1=2
m0 moving with a speed v is given by E = m0 c2 , where = 1=(1 ) ,
8
= v=c, and c = 3:0 10 m=s is the speed of light in a vacuum. This
total energy is the sum of the kinetic energy T and the rest-mass energy
Eo = m0 c2 (this is the object’s energy when the object is not moving, so
v = 0 and = 1):

E = T + Eo = T + m0 c2 = m0 c2 : (1.64)

Show that the relativistic kinetic energy as deduced from these equations
reduces to the expected (classical) form in the non-relativistic limit— that
is, for v << c. That means we want to show that the kinetic energy of
the object is given by
1
T ' m0 v 2 (1.65)
2
for v << c which is what you know from Intro. Physics.
Solution: The relativistic kinetic energy, in terms of the kinetic energy and the
total energy can be expressed as
h 1=2
i
2
T = m0 c2 m0 c2 = m0 c2 1 1 (1.66)

Applying the series expansion


1
X p p(p 1) 2
(1 + x)p = xn = 1 + px + x :::; (1.67)
n 2!
n=1

2
for x = and p = 1=2; we may write
1 1
2 1=2 1 2 2( 2 1) 4
1 =1+ + + ::: . (1.68)
2 2!
In the classical limit, v << c ) = v=c << 1, we can drop all the terms
with n for n 4;

2 1=2 1 2 1 v2
1 '1+ =1+ . (1.69)
2 2 c2
Therefore, the relativistic kinetic energy in the classical limit becomes

1 v2 1
T ' m0 c2 1 + 1 )T ' m0 v 2 : (1.70)
2 c2 2

Example 1.5 Two identical small balls of charge Q and mass m are hung from
two silk threads of length L. The threads are attached at a common point
on the ceiling. The balls hang in equilibrium separated by a distance
1.3. SERIES REPRESENTATIONS OF REAL FUNCTIONS 15

x. Show that, for small charges Q and large lengths L, the equilibrium
position is given by
1=3
Q2 L
x= : (1.71)
2 0 mg

Solution: If the balls are at equilibrium the net force acting on each ball must
be zero. There are three forces acting on each of the balls. These are
the repulsive electrostatic force ( Fe ), the tension force ( T ), and earth’s
gravitational force ( W ) [Fig. 1.4].The magnitude of the electrical ( Fe )

Figure 1.4: Two charged balls suspended from a massless string.

and the gravitational ( W ) forces are given by

1 Q2
Fe = and W = mg;
4 0 x2
respectively. Since the balls are at equilibrium, the net force on each balls
must add up to zero. Considering the ball on the right side in Fig. 1.4,
for the net for in the y-direction, we have

F1y = Ty W = T sin mg = 0 ) T sin = mg (1.72)

and in the x direction


1 Q2 1 Q2
F1x = Fe Ty = T cos ) T cos = (1.73)
4 0 x2 4 0 x2

Dividing Eq. (1.72) by (1.73), we …nd


2
sin 4 0 mgx
tan = = 2
: (1.74)
cos Q
16 CHAPTER 1. SERIES AND CONVERGENCE

In terms of the length of the threads, L; and the separation distance, x,


we may write
q
2
L2 x4 2L x2
1=2
tan = = 1 : (1.75)
x=2 x 4L2
Applying the series expansion
1
X
p p p(p 1) 2
(1 + x) = xn = 1 + px + x ::: (1.76)
n 2!
n=1

we can express Eq. (1.75) as


1=2 2
#
1 1
2L x2 2L 1 x2 2(2 1) x2
1 = 1 + + ::: :
x 4L2 x 2 4L2 2! 4L2
(1.77)
For large L >> x we can drop all the terms in the series that involve L
and approximate Eq.(1.77) as
1=2
2L x2 2L
tan = 1 ' (1.78)
x 4L2 x
Substituting Eq. (1.78) into Eq. (1.74) and solving for x, we …nd
2 2 1=3
4 0 mgx 2L 4 0 mgx Q2 L
tan = ) ' )x' (1.79)
Q2 x Q2 2 0 mg
Another simple way: For the …gure shown belowwe may write

F1y = Ty W = 0 ) T cos = mg (1.80)

and
1 Q2
F1x = Fe Ty = 0 ) T sin = (1.81)
4 0 x2
which leads to
sin Q2
= : (1.82)
cos 4 0 mgx2
Apply the Maclaurin’s series expansion
3 5 7
sin = + ::: (1.83)
3! 5! 7!
2 4 6
cos = 1 +
::: (1.84)
2! 4! 6!
for small angle (which is the case for large x << L which what we should
expect for small charge Q), we have

sin ' and cos ' 1: (1.85)


1.3. SERIES REPRESENTATIONS OF REAL FUNCTIONS 17

so that
1=3
Q2 x Q2 Q2 L
sin ' 2
) ' 2
)x' : (1.86)
4 0 mgx 2L 4 0 mgx 2 0 mg
18 CHAPTER 1. SERIES AND CONVERGENCE

1.4 Homework Assignment 1


1. Land for sell: Imagine you are given a speci…c procedure to sell a big
chunk of land to investors. This land is triangularly shaped plateau. The
triangle can be approximated as an equilateral triangle. The size of this
land, approximately, about square kilometers. The procedure for selling
the land reads as follows:

(a) Connect the midpoints of the sides (of the equilateral triangle shaped)
chunk of land to form a total of 4 smaller triangularly shaped chunks of
land with equal sides (i.e. 4 equilateral triangles).
(b) Sell the middle triangularly shaped chunk of land.

(c) For each of the other 3 remaining unsold chunk of lands draw lines con-
necting the midpoints. You will form 4 small chuck of land from each.
Again you sell each middle triangular chuck of land for another investor.
Find the in…nite series for the total size (i.e. area) of land sold to the
investors if the process described above continued inde…nitely.

2. Find the limit of the sequence an given by

n2 + 5n3
an = p :
2n3 + 3 4 + n6

3. Use the ratio test to …nd whether the following series converge or diverge

(a)
X1 2
5n (n!)
:
n=0
(2n)!

(b) p
1
X (2n)!
:
n=0
n!

4. Find the interval of convergence for the series


1
X n
(x 1)
:
n=0
2n

5. Prove that
n
X n (n + 1)
i=
i=1
2

is true for any integer n > 0 (even or odd).


1.5. HOMEWORK ASSIGNMENT 2 19

1.5 Homework Assignment 2


For the functions listed in problem 1 and 2: (a) Find the …rst four none zero
terms of the Maclaurin’s series; (b) Find the general term and write the series in
summation form; (c) Check your result in (a) by computer; (d) Use a computer
to plot the functions and several approximating partial sum of the series.

1.
x
f (x) = p
1 x2
2.
1 x
g (x) =
1+x
Hint: Express the function as sum or product of two functions.
3. The little boy at the play ground: you wanted to introduce a little boy
to a swing set at the play ground. One sunny summer day you took him
to the play ground. You put him on the swing and secure his sit belt.
Then you pushed his sit in a horizontal direction with a force F so that
the rope of the swing is displaced by an angle and his sit is at a distance
x from the vertical (see the …gure below). At this instant the little boy
is at equilibrium position. Let the weight of the little boy be W and the
length of the rope is l:

(a) Find F=W as a series of power of .


(b) In the play ground it is not easy to estimate or measure the angle but
we can estimate x and l. Find F=W as a series of power of x=l.

4. Evaluate the following indeterminate functions using ’L’Hospital’s rule

(a)
ln (x a)
f (x) = lim p ;
x!1 x a
where a > 0.
(b)
g (x) = lim (x 1) ln [4 (x 1)] :
x!1

5. Use the series you know to show that


2 4 6
+ ::: = 1:
3! 5! 7!
20 CHAPTER 1. SERIES AND CONVERGENCE
Chapter 2

Complex Numbers,
functions, and series

2.1 Complex numbers


If you punch the real number, x = 1; into your calculator and try to …nd
its square roots, your calculator displays an error. That is because in reality
the square root for a negative number does not exist. For Mathematicians and
Physicists the square roots of a negative number do exist. It is just not in a set
of real numbers but rather in more universal set of numbers known as complex
numbers. A complex number, z, is de…ned by

z = a + ib; (2.1)

with a and b are real numbers and i is know as the imaginary, which is de…ned
as,
1=2
i2 = 1) i = ( 1) : (2.2)

For any complex number, z = a + ib; a is know as the real part and b is known
as the imaginary part,
Re z = a; Im z = b: (2.3)

Rectangular and polar representation: For any complex number, z = a +


ib, we can use a rectangular representation using the Cartesian coordinates
(x; y) or polar representation using the polar coordinates (r; ). In rectangular
representation the real part a is the x coordinate and the imaginary part b is
the y coordinate,
a = x; b = y: (2.4)

In polar coordinate
a = r cos ; b = r sin ; (2.5)

21
22 CHAPTER 2. COMPLEX NUMBERS, FUNCTIONS, AND SERIES

where r is called the modulus (or magnitude) and is called the phase of the
complex number z: Thus one can express the complex number z = a + ib

z = x + iy = r cos + ir sin : (2.6)

The Cartesian and Polar coordinates, as shown in Fig. 2.1, are related by the
equations
x = r cos ; y = r sin ; (2.7)

Figure 2.1: Rectangular and Polar representation of a complex number z =


x + iy:

Euler’s equation and exponential representation: The Euler’s Equation, that


relates an exponential function and trigonometric functions, is given by

exp (i ) = cos ( ) + i sin ( ) : (2.8)

We will derive Euler’s equation in section 2.2 when we introduce to series of


complex function. Using Euler’s equation one can also represent the complex
number in Eq.(2.6); in an exponential form as

z = x + iy = r (cos ( ) + i sin ( )) = r exp (i ) : (2.9)

The magnitude, r, (which is also represented as jzj) and the phase angle for
the complex number z are related to the Cartesian coordinates (x; y) by
p q
y
jzj = x2 + y 2 = r2 cos2 ( ) + r2 sin2 ( ) = r; = tan 1 : (2.10)
x
Note that the phase angle is always measured from the positive x-axis in a
counterclockwise direction.
2.1. COMPLEX NUMBERS 23

Complex conjugate of a complex Number : The complex conjugate of a com-


plex number, z = x + iy; is denoted by z and de…ned as

z =x iy (2.11)

The magnitude of a complex number, z; is related to its complex conjugate by


p p p
jzj = zz = (x + iy) (x iy) = x2 + y 2 : (2.12)

Example 2.1 Find the real and imaginary part of the complex number

3 + 4i
z= (2.13)
3 4i

Solution:

To …nd the real and imaginary parts …rst we need to write it in the form,
z = x + iy. This can be done by making the denominator real. In order to
make it real we multiply it (and also the numerator) by complex conjugate,

3 + 4i (3 + 4i) (3 + 4i) 9 + 12i + 12i 16 7 + 24i


z= = = =
3 4i (3 4i) (3 + 4i) 32 + 42 25
7 24
= + i ) Re z = 7=25; Im z = 24=25 (2.14)
25 25

Example 2.2 Find the real and imaginary part of the complex conjugate of
the complex quantity p
3 + 2i sin
z= ; (2.15)
4 + 2e4i
where is real.

Solution:

The complex conjugate, z ; is given by replacing i by i


p
3 2i sin
z = (2.16)
4 + 2e 4i
Applying Euler’s formula, we can write
4i
2e = 2 cos (4 ) 2i sin (4 ) (2.17)

so that p
3 2i sin
z = (2.18)
4 + 2 cos (4 ) 2i sin (4 )
To …nd the real and imaginary part of z we need to make the denominator
in the above expression real. We can do this by multiplying it with its
24 CHAPTER 2. COMPLEX NUMBERS, FUNCTIONS, AND SERIES

complex conjugate. The complex conjugate of 2 cos (4 ) 2i sin (4 ) is


2 cos (4 ) + 2i sin (4 ) . Thus
p
3 2i sin (2 cos (4 ) + 2i sin (4 ))
z =
(2 cos (4 ) 2i sin (4 )) (2 cos (4 ) + 2i sin (4 ))
p
3 2i sin (4 + 2 cos 4 + 2i sin 4 )
= 2 (2.19)
(4 + 2 cos 4 ) + 4 sin2 4
p p
3 (4 + 2 cos (4 )) + 4 sin (4 ) sin + 2i 3 sin (4 ) 2i sin (4 + 2 cos (4 ))
= 2
(4 + 2 cos (4 )) + 4 sin2 (4 )
(2.20)
p
3 (4 + 2 cos (4 )) + 4 sin (4 ) sin
) z = 2
(4 + 2 cos (4 )) + 4 sin2 (4 )
p
2 3 sin (4 ) 2 sin (4 + 2 cos (4 ))
+i 2 : (2.21)
(4 + 2 cos (4 )) + 4 sin2 (4 )
Therefore, the real and imaginary parts of z are
p
3 (4 + 2 cos 4 ) + 4 sin 4 sin
Re (z ) = 2 ; (2.22)
(4 + 2 cos (4 )) + 4 sin2 (4 )
p
2 3 sin 4 2 sin (4 + 2 cos 4 )
Im (z ) = 2 (2.23)
(4 + 2 cos (4 )) + 4 sin2 (4 )

Example 2.3 Consider the complex function


3
Z(x) = 2f (x) ig(x) (2.24)
2
where f (x) and g(x) are possibly complex functions of the real variable x.
(a) Find an expression for the complex conjugate of the function Z(x).
(b) Evaluate Z (x) for x = 1 if
p
f (x) = 3ix2 and g(x) = 2x + 4 2i (2.25)

Solution:
(a) since f (x) and g(x) are possibly complex functions, the complex conjugate
of Z(x) should be written as
3
Z (x) = 2f (x) + ig (x) (2.26)
2
where f (x) and g (x) are the complex conjugates of f (x) and g(x); re-
spectively.
2.1. COMPLEX NUMBERS 25

(b) We are given p


f (x) = 3ix2 , g(x) = 2x + 4 2i (2.27)
from which we …nd
p
f (x) = 3ix2 ; g (x) = 2x 4 2i (2.28)

Then
3
Z (x) = 2f (x) + ig (x) (2.29)
2
becomes
3 p p
Z (x) = 2 3ix2 + i 2x 4 2i ) Z (x) = 6ix2 + 3xi + 6 2
2
(2.30)
Now substituting x = 1; we …nd
p
Z (x) = 6 2 3i: (2.31)

Example 2.4 Find the values of the real quantities x and y if

(5 + i) x 5yi = 4 + 3i (2.32)

Solution: To determine the values of the real quantities x and y; we …rst need
to write the left hand side in the form z = a + ib

(5 + i) x 5yi = 5x + i (x 5y) = 4 + 3i (2.33)

Two complex number z1 and z2 are equal if and only if their corresponding
Re and Im parts are equal. Thus

5x = 4; x 5y = 3 ) x = 4=5; y = 11=25: (2.34)

Example 2.5 The Intensity of a Traveling Electromagnetic Waves: An elec-


tromagnetic wave is described by two mutually perpendicular oscillating
~ and magnetic …eld (B)
electric (E) ~ vectors. An electromagnetic wave
traveling in the positive z-direction the electric …eld can be expressed as
~ t) = E0 exp [i (kz
E(z; !t)] x
^;

where the amplitudes of the electric (E0 ) and the angular frequency (!)
are real. The wave vector (~k = k^
z ) given by

~k = 2 z^ = 2 n z^ = 2 n !
z^ = n z^; (2.35)
0 c c
in which 0 is the wave length, c is the speed of light in vacuum and
= c=n is the wavelength in the medium, with refractive index n; that
the wave is traveling in. The refractive index of the medium is possibly
26 CHAPTER 2. COMPLEX NUMBERS, FUNCTIONS, AND SERIES

complex for some media such as metals and should be expressed, generally,
as
n = nR + inI ; (2.36)
where both nR and nI are real. Note that Eq. (2.36) shows the wave ~k in
Eq. (2.35) is also possibly complex and should be written as

~k = n ! z^ = ! (nR + inI ) z^: (2.37)


c c
~ usually determined from the electric …eld
The magnetic …eld vector (B)
~ using the relation
vector (E)

~ t) = ~k
B(z; ~ (z; t):
E (2.38)

The “Poynting vector” is an important physical quantity in describing


electromagnetic waves. It is proportional to the cross product of the elec-
tric and magnetic …eld vectors,

~ t) = 1 E(z;
S(z; ~ t) ~ t):
B(z; (2.39)
0

The direction of Poynting vector shows the direction of propagation of


the traveling wave and the magnitude gives the intensity of the wave as a
function of position and time in the medium that the wave is propagat-
ing in. For a an electromagnetic wave traveling in a medium where the
refractive index is complex [Eq. (2.36)],

(a) …nd the complex conjugate of the electric …eld vector

(b) Find the magnetic …eld vector

(c) Shown that the magnitude of the Poynting vector is given by


q p
~ ~ ~ n2R + n2I 2 2!
S(z; t) = S(z; t) S (z; t) = E0 exp nI z : (2.40)
0 c c

Solution:

(a) Using
~ t) = E0 exp [i (kz
E(z; !t)] x
^; (2.41)
we may write the complex conjugate as
~ (z; t) = E0 exp [ i (k z
E !t)] x
^; (2.42)

Taking the complex conjugate of Eq. (2.37), we have


!
k = (nR inI ) : (2.43)
c
2.1. COMPLEX NUMBERS 27

so that
h ! i
~ (z; t)
E = E0 exp i (nR inI ) z !t x ^
c h i
! !
= E0 exp nI z exp i nR z !t x ^ (2.44)
c c
(b) In view of the result in part (a) and Eq. (2.37), the magnetic …eld becomes
~ t) = ~k E
B(z; ~ (z; t) = ! (nR + inI ) z^
! h c! i
E0 exp nI z exp i nR z !t x ^
c c h i
~ t) = ! (nR + inI ) E0 exp
) B(z;
!
nI z exp i
!
nR z !t z^ x ^
c c h c i
~ t) = ! (nR + inI ) E0 exp
) B(z;
!
nI z exp i
!
nR z !t y^
c c c
(2.45)

(c) Using the electric …eld


h ! i
~ t)
E(z; = E0 exp [i (kz !t)] x
^ = E0 exp i (nR + inI ) z !t x^
h c i
~ t) = E0 exp ! !
) E(z; nI z exp i nR z !t y^; (2.46)
c c
the result in part (b) for the magnetic …eld, the pointing vector
~= 1E
S ~ ~
B (2.47)
0
can be expressed as
~ = 1 n ! h ! i o
S E0 exp nI z exp i nR z !t x ^
0 c c
n! ! h ! i o
(nR + inI ) E0 exp nI z exp i nR z !t y^
c c c
) S ~ = 1 ! (nR + inI ) E02 exp !
2 nI z (^x y^) (2.48)
0 c c
) S ~ = 1 ! (nR + inI ) E02 exp !
2 nI z z^ (2.49)
0 c c
Then the magnitude of the pointing vector (the intensity)
q
~ t) = S(z;
S(z; ~ t) S ~ (z; t); (2.50)
becomes
~ t) = 1 ! ! 1 !
S(z; (nR + inI ) E02 exp 2 nI z z^ (nR inI ) E02
0 c c 0 c
! i1=2
exp 2 nI z z^
p c
! n2R + n2I 2 2!
) jS(z; t)j = E0 exp nI z : (2.51)
0c c
28 CHAPTER 2. COMPLEX NUMBERS, FUNCTIONS, AND SERIES

This result shows that the intensity decays exponentially as the wave pen-
etrates inside the medium.

2.2 Complex In…nite Series and the Circle of


Convergence
Complex In…nite Series: An in…nite series
1
X
zn = z0 + z1 + z2 + z3 :::zn + ::: (2.52)
n=0

is said to be a complex in…nite series if

zn = an + ibn ; (2.53)

where an and bn are real number. An in…nite complex series is convergent if


and only if
X1
an = a0 + a1 + a2 + a3 :::an + ::: = A (2.54)
n=0

and
1
X
bn = b0 + b1 + b2 + b3 :::bn + ::: = B; (2.55)
n=0

where A and B are real constants.


Test of convergence: For a complex in…nite series the convergence can be
tested using absolute convergence test. This means

(a) You can use any convergence test that involves the absolute value sign,
which in this case will mean the magnitude of a complex quantity. This
is applied to the full complex in…nite series.

(b) Use any convergence test to separately test the convergence of the real and
imaginary parts of the complex series. The full series converges only if
both the real and imaginary parts of the series converge separately.

Series of Complex function: In the previous chapter we have seen a real


power series
1
X
an xn = a0 + a1 + a2 + a3 :::an + ::: (2.56)
n=0

where an and x are real. A power series


1
X n 2 3 n
cn (z z0 ) = c0 +c1n (z z0 )+c2 (z z0 ) +c3 (z z0 ) :::cn (z z0 ) +:::
n=0
(2.57)
2.2. COMPLEX INFINITE SERIES AND THE CIRCLE OF CONVERGENCE29

is a series of complex function when z is a complex variable and z0 or cn are


complex.
Radius of convergence: Consider the complex power series in Eq. (2.57).
This series converges, applying the ratio test, for
n+1
cn+1 (z z0 ) cn+1
= lim n < 1 ) lim jz z0 j < 1 (2.58)
n!1 cn (z z0 ) n!1 cn
For example, for
cn+1 n
= (2.59)
cn n+1
we have
cn+1
lim =1 (2.60)
n!1 cn
which leads to
= jz z0 j < 1: (2.61)
For z = x + iy and z0 = a + ib; we note that
z z0 = x + iy (a + ib) = (x a) + i(y b)
p 2
) jz z0 j = (x a)2 + (y b)2 < 1 ) (x a) + (y b)2 < 12 : (2.62)
We recall the equation
(x a)2 + (y b)2 = R2 ; (2.63)
de…nes a circle with radius, R, centered at (a; b) : Thus the series converges for
all complex numbers, z; inside the circle (disk) of radius R = 1 on the complex
plane: This circle (disk) is known as circle (disk) of convergence for the in…nite
complex series in Eq. (2.57) and the radius of the circle (disk) is called the
radius of convergence.
Example 2.6 Determine the convergence of the following complex in…nite se-
ries
1
X i 1 i 1
zn = 1 + (2.64)
n=0
2 4! 6! 8!

Solution: This series can be expressed as


1
X X1 n
( i)
zn = (2.65)
n=0 n=0
(2n)!
Applying the ratio test we have
,
n+1 n n+1
zn+1 ( i) ( i) ( i) (2n)!
lim = lim = lim n
n!1 zn n!1 (2 (n + 1))! (2n)! n!1 (2 (n + 1))! ( i)

i 1 Zn+1
= lim = lim ) lim =0<1
n!1 (2n + 2) (2n + 1) n!1 (2n + 2) (2n + 1) n!1 Zn
(2.66)
which means the series is convergent.
30 CHAPTER 2. COMPLEX NUMBERS, FUNCTIONS, AND SERIES

Example 2.7 Find and sketch the circle of convergence in the complex plane
for the following complex in…nite series:
1
X 1
X n
n (z 2 + i)
cn z = (2.67)
n=0 n=0
2n

Solution: A complex series


1
X n
an (z z0 ) (2.68)
n=0
is convergent when
n+1
an+1 (z z0 )
= lim n < 1: (2.69)
n!1 an (z z0 )

Then using
n
n (z 2 + i)
an (z z0 ) =
2n
n+1
n+1 (z 2 + i)
an+1 (z z0 ) = (2.70)
2n+1
we can write
,
n+1 n+1
(z 2 + i) (z 2 + i)
= lim <1
n!1 2n+1 2n+1
n+1
(z 2 + i) 2n z 2+i
) = lim n <1) <1
n!1 2n+1 (z 2 + i) 2
) jz 2 + ij < 2 ) jz (2 i)j < 2: (2.71)
This means the series is convergent inside a disk of radius R = 2 centered
at z0 = 2 i:
Example 2.8 Evaluate f (2 i) and …nd the real and imaginary parts if
z 1
f (z) = (2.72)
z
Solution: We can write f (2 i) as
2 i 1 1 i
f (2 i) = = (2.73)
2 i 2 i
which can be rewritten as
(1 i) (2 + i) 3 i
f (2 i) = = (2.74)
(2 i) (2 + i) 5
and the real and imaginary part
Re (f (2 i)) = 3=5; Im (f (2 i)) = 1=5 (2.75)
2.2. COMPLEX INFINITE SERIES AND THE CIRCLE OF CONVERGENCE31

Example 2.9 Derive


(b) Euler’s Formula
ei = cos ( ) + i sin ( ) (2.76)
using series expansion of complex variables.
(b) the double-angle formulae for sin(2 ) and cos(2 )

sin(2 ) = 2 sin ( ) cos ( ) ;


cos(2 ) = cos2 ( ) sin2 ( ) : (2.77)

Solution:
(a) Using the series expansion for the complex function
1
X zn z2 z3
exp (z) = =1+z+ + + :: (2.78)
0
n! 2! 3!

we can write
1
X n 2 3
(i )
exp (i ) = =1+i i + :: (2.79)
0
n! 2! 3!

which can be put in the form


2 4 3 3
exp (i ) = 1 + + ::: + i i +i ::: (2.80)
2! 4! 3! 5!
2 4 3 3
exp (i ) = 1 + + ::: + i + ::: (2.81)
2! 4! 3! 5!
Applying the the series expansions for sine and cosine functions
X1
( 1)n 2n+1 x3 x5 x7
sin x = x =x + ::: convergent for all x
n=0
(2n 1)! 3! 5! 7!
(2.82)
X1
( 1)n 2n x2 x4 x6
cos x = x =1 + :::convergent for all x (2.83)
n=0
(2n)! 2! 4! 6!
we may write
exp (i ) = cos ( ) + i sin ( ) (2.84)

(b) using the result above we may write


2
[exp (i )] = exp (2i ) = cos (2 ) i sin (2 ) (2.85)

But we also know that


2 2 2
[exp (i )] = [cos ( ) i sin ( )] ) [exp (i )] = cos2 ( ) sin2 ( )
2i sin ( ) cos ( ) (2.86)
32 CHAPTER 2. COMPLEX NUMBERS, FUNCTIONS, AND SERIES

Therefore, we can see that

cos (2 ) i sin (2 ) = cos2 ( ) sin2 ( ) 2i sin ( ) cos ( ) (2.87)

which leads to

cos (2 ) = cos2 ( ) sin2 ( ) ; sin (2 ) = 2 sin ( ) cos ( ) : (2.88)

2.3 Powers and Roots of complex numbers


The powers and roots of a complex number is easily determined by applying the
Euler’s formula. Let’s consider a complex number

z = x + iy = r exp (i ) ; (2.89)

where p
1 y
r= x2 + y 2 ; = tan : (2.90)
x
This complex number to the power n (z n ) can then be written as
n n
zn = (x + iy) = [r exp (i )] = rn exp (in )
n
) z n = (x + iy) = rn (cos (n ) + i sin (n )) : (2.91)

Similarly, for the nth roots to this complex number (z 1=n ), one …nds
1=n
z 1=n = (x + iy) = r1=n exp (i k ) = r1=n (cos ( k ) + i sin ( k )) ; (2.92)

where
+2 k
k = ; for k = 0; 1; 2::n 1 (2.93)
n
Example 2.10 Solve the equation

z 5 = 32 (2.94)

and show the solutions on a complex plane.

Solution: Noting that


z 5 = 32 = r exp [ ] (2.95)
where
r = 25 ; = 0 (2.96)
the solutions to
z 5 = 32 (2.97)
given by
zk = r1=5 exp (i k ) = r1=5 (cos ( k ) + i sin ( k )) ; (2.98)
2.3. POWERS AND ROOTS OF COMPLEX NUMBERS 33

where

r1=5 = 2;
+2 k 2 k 2
k = = ; for k = 0; 1; 2; 3; 4 ) 0 = 0; 1 = = 72 ;
n n 5
4 6 8
2 = = 144 ; 3 = = 216 ; 4 = = 288 (2.99)
5 5 5
Then the solution to the equation are found to be

Figure 2.2: The root on the complex plane. The circle has a radius r = 2

z0 = 2 [cos (0) + i sin (0)] = 2 + i0


2 2
z1 = 2 cos + i sin = 0:62 + i1:90
5 5
4 4
z2 = 2 cos + i sin = 1:62 + i1:18
5 5
6 6
z3 = 2 cos + i sin = 1:62 i1:18
5 5
8 8
z4 = 2 cos + i sin = 1:62 i1:90 (2.100)
5 5

These roots are shown on the complex plane in Fig. 2.2


34 CHAPTER 2. COMPLEX NUMBERS, FUNCTIONS, AND SERIES

2.4 Algebraic vs transcendental functions


The logarithm and the exponential function are examples of transcendental func-
tions. Transcendental function is a term often used to describe the trigonometric
functions, i.e., sine, cosine, tangent, cotangent, secant, cosecant, and so forth. A
function that is not transcendental is said to be algebraic. Examples of algebraic
functions are rational functions and the square root function. A composition of
transcendental functions can give an algebraic function. For example
p
f (x) = cos (arcsin (x)) = 1 x2 (2.101)
Euler’s Equation, trigonometric, and hyperbolic Functions: We recall the
Euler’s formula in terms of a real variable x is given by
exp (ix) = cos (x) + i sin (x) (2.102)
and taking the complex conjugate we have
exp ( ix) = cos (x) i sin (x) (2.103)
so that adding the two equations we …nd
eix + e ix
cos (x) = (2.104)
2
and subtracting
eix e ix
sin (x) = : (2.105)
2i
If x is a complex variable z, using Euler’s Equation, we can express the sine and
cosine functions as
eiz + e iz
cos (z) = (2.106)
2
and
eiz e iz
sin (z) = (2.107)
2i
If z is a pure imaginary complex number z = iy, the sine and cosine functions
become
e y + ey
cos (iy) = ) cos (iy) = cosh (y) (2.108)
2
and
e y ey ey e y
sin (iy) = =i ) sin (iy) = i sinh (y) (2.109)
2i 2
where
ey + e y ey e y
cosh (y) = ; sinh (y) = (2.110)
2 2
are called the cosine hyperbolic and sine hyperbolic functions. It can be gener-
alized for a complex number z as
z
e + ez ez + e z
sinh (z) = ; cosh (z) = : (2.111)
2 2
2.4. ALGEBRAIC VS TRANSCENDENTAL FUNCTIONS 35

Other hyperbolic functions: named and de…ned in a way similar to the


trigonometric functions:
sinh (z) 1
tanh (z) = ; coth (z) = (2.112)
cosh (z) tanh (z)
1 1
csc h (z) = ; sec h (z) =
sinh (z) cosh (z)

Logarithms: Consider a complex number z = rei , then the logarithm of z


to the base e which is denoted by ln z can be expressed as

ln z = ln rei = ln r + ln ei = ln r + i ln e ) ln z = ln r + i (2.113)

The principal value: the principal value of a complex number z expressed


in polar coordinates as
z = rei (2.114)
is the value of the complex number when 0 2 : The principal value of
ln z is the value of ln z = ln rei for 0 2 :

Example 2.11 Evaluate the following integral applying Euler’s formula,


Z
I= sin ( x) cos ( x) dx (2.115)

where and are positive, nonzero integers.


Solution: Applying Euler’s formula we can write
ei x
e i x
ei x
+e i x
sin ( x) = ; cos ( x) = (2.116)
2i 2
so that
ei x
e i x
ei x
+e i x
sin ( x) cos ( x) =
2i 2
ei( + )x
e i( + )x
+ ei( )x
e i( )x
= (2.117)
:
4i
Then substituting the above expression for the integrand, we have
Z h i
1
I= ei( + )x e i( + )x + ei( )x
e i( )x
dx (2.118)
4i
so that applying Euler’s formula, once again, one can write

ei( + )x
e i( + )x
= 2i sin [( + ) x] ; ei( )x
e i( )x
= 2i sin [( ) x] :
(2.119)
Using Eq. (2.119), we can write the integral as
Z
1
I= [sin [( + ) x] + sin [( ) x]] dx: (2.120)
2
36 CHAPTER 2. COMPLEX NUMBERS, FUNCTIONS, AND SERIES

which leads to
1 cos [( + ) x] cos [( ) x]
I= + (2.121)
2 +

Noting that cosine is an even function,

cos (x) = cos ( x)

we can easily see that Eq. (2.121) reduces to


cos [( + ) ] cos [( + ) ] cos [( ) ] cos [( ) ]
I= + =0
+
(2.122)
Therefore Z
I= sin ( x) cos ( x) dx = 0: (2.123)

Orthonormal Sets of Functions: The set of functions {f1 (x); f2 (x); :::fj (x); :::}
is said to be orthonormal if
Z x2
1 i=j
fi (x)fj (x)dx = ij = (2.124)
x1 0 i 6= j

Example 2.12 Evaluate the following integral, where and are positive,
nonzero integers: Z
I= sin ( x) sin ( x) dx (2.125)

and show that the sine function form an orthonormal set of functions.
Solution: Applying Euler’s formula, we may write

ei x
e i x
ei x
e i x
sin ( x) = ; sin ( x) = (2.126)
2i 2i
so that
ei x
e i x
ei x
e i x
sin ( x) sin ( x) =
2i 2i
ei( + )x
ei( )x
e i( )x
+e i( + )x
= (2.127)
:
4
This can be rewritten as
1 ei( )x
+e i( )x
ei( + )x
+e i( + )x
sin ( x) sin ( x) = :
2 2 2
(2.128)
or
1
sin ( x) sin ( x) = fcos [( ) x] cos [( + ) x]g (2.129)
2
2.4. ALGEBRAIC VS TRANSCENDENTAL FUNCTIONS 37

Then substituting the above expression for the integrand, we have


Z
1
I= fcos [( ) x] cos [( + ) x]g dx (2.130)
2

1 sin [( ) x] sin [( + ) x]
)I= : (2.131)
2 +
We recall sin is an odd function, which means

sin ( x) = sin (x) (2.132)

we have

sin [ ( ) ]= sin [( ) ] ; sin [ ( + ) ] = sin [( + ) ]


(2.133)
so that
sin [( ) ] sin [( + ) ]
I= (2.134)
+
Since and are positive, none zero integers, we always have

sin [( + ) ] = 0 (2.135)

which leads to
sin [( ) ]
I= : (2.136)

This leads to
Z
=
I= sin ( x) sin ( x) dx = ; = (2.137)
0 6 =

which means the sine function

fj (x) = sin (jx) (2.138)

form an orthogonal set of functions in the domain ( ; ):


Example 2.13 Evaluate the complex expression

cos [i ln 2] (2.139)

Solution: Evaluating this expression is much easier if we use the Euler’s for-
mula:
eiz + e iz
cos (z) = (2.140)
2
we can express cos [i ln 2] as

ei(i ln 2) + e i(i ln 2)
e ln 2
+ eln 2
cos [i ln 2] = = (2.141)
2 2
38 CHAPTER 2. COMPLEX NUMBERS, FUNCTIONS, AND SERIES

and using the relation


ab = eb ln a ; (2.142)
for a = 2; b = 1 (or 1), we have
ln 2 1
e =2 = 0:5; eln 2 = 21 = 2 (2.143)
so that
3=2 3
cos [i ln 2] = = = 0:75: (2.144)
2 4
Example 2.14 Find the principal value of the complex logarithm
z = ln (1 i) (2.145)

Solution: To …nd the principal value …rst we need to express z = 1 i using


polar coordinates. To this end, we note that
q p
2 7
r = j1 ij = 12 + ( 1) = 2; = +2 n (2.146)
4
where n = 0; 1; 2:::But to …nd the principal value for ln z we use = 47 :
Thus p 7
z = 1 i = 2ei 4 (2.147)
which leads to
7 i 1 7 i
ln z = ln 21=2 + ) ln z = ln 2 + (2.148)
4 2 4
Example 2.15 Find the principal value of the complex quantity
i
z = (1 i) (2.149)

Solution: To …nd the principal value we …rst put the complex number in the
form rei : Recalling that
p 7
1 i= 2 exp i +2 n (2.150)
4
we have
p i i
i 7
(1 i) = 2 exp i
+2 n (2.151)
4
i 7
i
(1 i) = 21=2 exp i i +2 n
4
i i 7
(1 i) = 2 2 exp +2 n
4
The principal value is when n = 0:That means
i i 7
(1 i) = 2 2 exp (2.152)
4
2.4. ALGEBRAIC VS TRANSCENDENTAL FUNCTIONS 39

Example 2.16 Find the value for the following complex expression
2 !12 3
p
3+i
z = sin 1 4 p 5 (2.153)
3 i

Solution: Let the value of this expression be so that


p !12
3+i
sin ( ) = p : (2.154)
3 i

Noting that

1 1 p p
tan p = ; 3 + i = 2 ) 3 + i = 2ei 6 (2.155)
3 6

1 1 11 p 11
tan p = ) 3 i = 2ei 6 (2.156)
3 6
we may write

2ei 6
12 h i12
= e i( 6 )
11
sin ( ) = 11
6 (2.157)
2ei 6
ei12( 6 )=e
11
20i 10i(2 )
sin ( ) = 6 =e =1
) = 2k ; k = 0; 1; 2::: (2.158)
2
40 CHAPTER 2. COMPLEX NUMBERS, FUNCTIONS, AND SERIES

2.5 Homework Assignment 3


1. For the complex number p i
z= 2e 4

(a) Find x; y; r; and write it in the forms

z = (x; y); z = x + iy; z = (r; ); and z = r(cos + i sin ):

(b) Find z and also write it in the forms

z = (x; y); z = x + iy; z = (r; ); and z = r(cos + i sin ):

(c) Plot both z and z and label it.


2. Write the complex expression
1
z= 2
(2 3i)
in the form z = a + bi:
3. Prove that
z1 z1
=
z2 z2
(i.e. the conjugate of the quotient of two complex numbers is the quotient
of the conjugates.) Also prove that

(z1 z2 ) = z1 z2
(z1 z2 ) = z1 z2

Hint: it is easier to prove the statements about the quotient and the product
using the polar coordinates rei form; for the sum or di¤ erence it is easier
to use the rectangular form x + iy:
4.
(a) Find the absolute value of a complex number
3
z = (1 + 2i)

(b) Solve for all possible values of the real number x and y in the equation

2ix + 3 = y i

5. Physical application: A particle moves in the (x; y) plane so that its position
(x; y) as a function of time is given by

z = cos (3t) + i sin(3t)


2.5. HOMEWORK ASSIGNMENT 3 41

(a) Find the velocity (v) and acceleration (a) of the particle

dz d2 z
v= ;a = 2
dt dt

(b) Find the magnitude (absolute value) of the velocity and acceleration
(c) Describe the motion of the particle.
42 CHAPTER 2. COMPLEX NUMBERS, FUNCTIONS, AND SERIES

2.6 Homework Assignment 4


1. Test each of the following series for convergence.
(a)
1
X 1 1 1 1
n =1+ + 2 + :::: n + :::
n=0
(1 + i) (1 + i) (1 + i) (1 + i)

(b)

1
X n 2 n
1 i 1 i 1 i 1 i
=1+ + + ::: + :::
n=0
1+i 1+i 1+i 1+i

2. Find the disk of convergence for each of the following power series of complex
variable
(a)
X1 3 3 3
(n!) z n z (2!) z 2 (n!) z n
=1+ + ::: :::
n=0
(3n)! 3! 6! (3n)!

(b)
1
X n 2 3 n
(z i) z i (z i) (z i) (z i)
= + + + ::: :::
n=1
n 1 2 3 n

3.
(a) Show that for any real y, eiy = 1: Hence show that jez j = ex for every
complex z
jez j = ex :
(b) For the complex numbers z1 and z2 ; prove that
jz1 z2 j = jz1 j jz2 j
z1 jz1 j
=
z2 jz2 j
Hint: write the numbers in rei form
(c) Using the relations in part (a) and (b) …nd the value for
1 i
W =
1+i

4. Find the roots for


z = i1=5 :
and show your results on the complex plane.
5. Find the formula for sin (3 ) and cos (3 ) :
Hint: refer to the example for double angle formula that we did in class.
2.7. HOMEWORK ASSIGNMENT 5 43

2.7 Homework Assignment 5


Find the values in rectangular form x+iy for the complex expressions in problem
1-4

1.
z = sin ( i ln 3)

2.
z = tan i

3.
z = cos (2i ln i)

4. p
1
z = tanh i 3

5. Physical Application: Consider the RLC-ac circuit shown in the …gure below.
The circuit has a resistor and an inductor connected in series and then a
capacitor in parallel with them (R and L in series„and then C in parallel
with them). From introductory physics, the impedance for a resistor (ZR ),

capacitor (ZC ), and inductor (ZL ), are given by

ZR = R; ZC = 1= (i!c) ; and ZL = i!L;

where R is the resistance of the resistor, C is the capacitance of the ca-


pacitor, L is the inductance of the inductor, and is the angular frequency
of the ac-source voltage, V (t) = Vmax sin (!t) in the circuit. In ac cir-
cuit for series connection of two impedances, Z1 and Z2 the equivalence
impedance Z12 is given by

Z12 = Z1 + Z2
44 CHAPTER 2. COMPLEX NUMBERS, FUNCTIONS, AND SERIES

where as for parallel connection


1 1 1
= + :
Z12 Z1 Z2

(a) Find the equivalent impedance of the circuit


(b) Find the magnitude of the equivalent impedance of the circuit.
(c) A circuit is said to be at resonance when the total impedance of the circuit
is real. Find ! in terms of R, L, and C at resonance.
Chapter 3

Vectors, Lines, and Planes

3.1 An Overview of Vector Fundamentals


A vector : A vector is a quantity that can be described by a magnitude and
direction. We shall represent a vector by a boldface letter (for example A) or
~ Here we will use
by a letter with an arrow on top of the letter (for example A).
the second way of representation. The component of a vector is represented by
~
a subscript (for example Ax represent the x component of the vector A).

Figure 3.1: A two-dimensional vector and its components on the x-y plane.

~ is
Magnitude of a vector : the length of the arrow representing a vector A
~ ~
called the length or the magnitude of A (written A or A) and is given by
q
A = A2x + A2y (2-Dimensions)
q
A = A2x + A2y + A2z (3-Dimensions) (3.1)

45
46 CHAPTER 3. VECTORS, LINES, AND PLANES

A unit vector : a unit vector is a vector whose magnitude is one. For any
~ we can …nd a unit vector denoted by A^ (a vector along the direction
vector A
~ with unit magnitude). This unit vector is given by
of A

~
A
A^ = (3.2)
~
A

3.2 Vector Multiplication


Two vectors can be multiplied in two di¤erent ways. One way of multiply-
ing two vectors leads to a scalar. Such multiplication of vectors is referred as
Scalar product (dot product). The other way of multiplying two vectors leads
to another vector and is known as cross product (vector product)

A. The Scalar Product (The Dot Product): The scalar product of two vectors
~ = Ax^{ + Ay |^+ Az k^ and B
A ~ = Bx^{ + By |^+ Bz k^ (written as A
~ B)
~ is given
by
A~ B
~ = Ax Bx + Ay By + Az Bz : (3.3)
If the magnitude of the two vectors and the angle between the two, ; is
known, the scalar product of these two vectors can also be determined
using
~ B
A ~ = A ~ B~ cos : (3.4)

A dot product can be used to determine the angle between two vectors
0 1 0 1
A~ B~ A B + A B + A B
= cos 1 @ A = cos 1 @ x x y y z zA
(3.5)
~
A B ~ ~
A B ~

Application: Suppose you pull an object on a frictionless surface by ap-


plying a force and moved the object horizontally for some distance. If the
force applied is F~ and the displacement of the object is S
~ , the work done
W is given by
W = F~ S ~ = F~ S~ cos ( ) (3.6)

B. The Vector Product (The Cross Product): The vector product of two vectors
A~ and B ~ (written as A
~ B) ~ gives a third vector C
~ perpendicular to the
plane formed by the two vectors. If the angle between the two vectors is
; the magnitude of C~ is given by

~
C= A ~ = A
B ~ B
~ sin (3.7)

and the direction can be determined using the right-hand rule.


3.2. VECTOR MULTIPLICATION 47

The Right-Hand Rule: To …nd the direction of C~ =A ~ B ~ put your right hand
~
on the …rst vector (A) curl your …ngers towards the second vector (B) ~
~
then your thumb points in the direction of vector C.

Right-Handed Coordinate Systems: in right-hand coordinate system the right


hand rule is valid. For example in Cartesian coordinate system we have

^{ ^ |^
|^ = k; k^ = ^{; k^ ^{ = |^: (3.8)

If we know the components of the two vectors, A ~ = Ax^{ + Ay |^ + Az k^ and


~ ^
B = Bx^{ + By |^ + Bz k; then you can …nd the components of the vector
product C~ =A~ B ^ in Cartesian coordinates, using
~ = Cx^{ + Cy |^ + Cz k;

Cx = Ay B z Az B y (3.9)
Cy = Az Bx Ax Bz
Cz = Ax B y Ay B z

Application of cross product: torque due to a force F~ on a ridged object


placed at a position ~r with respect to an axis (through the origin) is given
by ~ = ~r F~ .

Example 1 A force of magnitude 4:0N pointing in the positive-x direction is


applied to a large solid object at the position (0; 2; 0)m. Find the resulting
torque exerted on the object about the origin using the

(a) the magnitude and direction form of the cross product.

(b) component form of the cross product.

Solution:
48 CHAPTER 3. VECTORS, LINES, AND PLANES

(a) The torque is given by


~ = ~r F~ (3.10)
where the force, pointing along the positive-x direction is given by

F~ = (4; 0; 0) N = 4N x
^ (3.11)

and the position vector

~r = (0; 2; 0)m = 2m^


y (3.12)

We use the relation for the magnitude of the cross product of two vectors
given by
j~ j = j~rj F~ sin ( ) (3.13)

where is the angle between the position and the force vector. The mag-
nitude of the position and force vectors are obtained using
q
j~rj = rx2 + ry2 + rz2 = 2m (3.14)

and q
F~ = Fx2 + Fy2 + Fz2 = 4N: (3.15)

Noting that the angle between ~r and F~ is

= (3.16)
2
the magnitude of the torque would be

j~ j = F~ j~rj sin ( ) = 8N m: (3.17)

To …nd the direction use the right hand rule and you will …nd the direction
to be along the negative-z direction. Then the torque may be expressed
as
~ = 8N m^ z (3.18)

(b) The torque is given by


~ = ~r F~ (3.19)
where the force, pointing along the positive-x direction is given by

F~ = (4; 0; 0) N = 4N x
^ (3.20)

and the position vector

~r = (0; 2; 0)m = 2m^


y (3.21)
3.3. EQUATIONS OF A STRAIGHT LINE 49

We apply the relation


x^ y^ z^
~
A ~ =
B Ax Ay Az (3.22)
Bx By Bz
which gives
x
^ y^ z^
2 0 0 0 0 2
~ = ~r F~ = 0 2 0 =x
^ y^ + z^ (3.23)
0 0 4 0 4 0
4 0 0
~ = ~r F~ = 8N m^
z

3.3 Equations of a straight line


Symmetric equations (in a two dimensional space):
x x0 y y0 z z0
= = (3.24)
a b c
where a; b;and c 6= 0: When c = 0 we should write the symmetric equation of
the straight line as
x x0 y y0
= ; z = z0 (3.25)
a b
Parametric equations:
8
< x = x0 + at
~ or
~r = ~r0 + At y = y0 + bt (3.26)
:
z = z0 + ct
The symmetric equations can be obtained from the parametric equations:
x x0 y y0 z z0
t= = = (3.27)
a b c

3.4 Equation of a plane


~ = ai + bj + ck is normal (perpendicular) to a plane, then the dot product
If N
~ and the vector ~r ~r0
of the vector N
~r ~r0 = (x x0 ) x
^ + (y y0 ) y^ + (z z0 ) z^ (3.28)
is zero,
a (x x0 ) + b (y y0 ) + c (z z0 ) = 0 (3.29)
This de…nes the equation of the plane. It can be rewritten as
ax + by + cz = d (3.30)
where
d = ax0 + by0 + cz0 (3.31)
50 CHAPTER 3. VECTORS, LINES, AND PLANES

Example 2 The position of a particle at time t = 0 is given by (1; 2; 3)m. The


particle moves with a constant velocity (0; 2; 1)m=s. What is the equation
of the particle’s trajectory?
Solution: The initial position of the particle can be expressed, using Cartesian
unit vectors, as
~r0 = x
^ + 2^
y + 3^z (3.32)
at a later time t, let the position of the particle is described by the vector

~r = x (t) x
^ + y (t) y^ + z (t) z^ (3.33)

If the particle is traveling by a constant velocity, the vector

~r ~r0 = [x (t) 1] x
^ + [y (t) 2] y^ + [z (t) 3] z^ (3.34)

must be proportional to the velocity vector

~v = 0^
x + (2m=s) y^ + (1m=s) z^: (3.35)

This means

~r ~r0 = t~v ) [x (t) 1] x


^ + [y (t) 2] y^ + [z (t) 3] z^ (3.36)
= (2m=s) t^
y + (1m=s) t^
z

) x (t) x
^ + y (t) y^ + z (t) z^ = 1^
x + 2^
y + 3^
z + (2m=s) t^
y + (1m=s) t^
z (3.37)
or
~r (t) = ~r0 + ~v t (3.38)
where
~r0 = x
^ + 2^
y + 3^
z ; ~v = (2m=s) y^ + (1m=s) z^ (3.39)

Example 3 Consider the particle in the previous example. What is the dis-
tance of closest approach (the “impact parameter”) of the particle to the
origin?
Solution: Let the distance of the closest approach be described by a point
(x0; y0; z0)
~r0 = x0 x
^ + y 0 y^ + z 0 z^ (3.40)
The closest distance is determined by the magnitude of the vector normal
to the trajectory of the particle. That means we must have ~r0 ? ~r0 ~r0 for
the impact parameter. We recall that when two vectors are perpendicular
their dot product is zero

~r0 (~r0 ~r0 ) = 0: (3.41)

Since ~r0 describes a point on the trajectory of the particle given by

~r (t) = ~r0 + ~v t (3.42)


3.4. EQUATION OF A PLANE 51

at which the particle becomes closest to the origin, we may write

~r0 = ~r0 + ~v t0 ) ~r0 ~r0 = ~v t0 : (3.43)

Then

(~r0 + ~v t0 ) ~v t0 = 0
0 02
(~r0 ~v ) t + (~v ~v ) t = 0 (3.44)

Noting that
2 2
~v ~v = (2m=s) + (1m=s) = 5m=s (3.45)
~r0 ~v = r0x vx + r0y vy + r0z vz = 2 2+3 1=7

Therefore
(~r0 ~v ) t0 + (~v ~v ) t02 = 0 (3.46)
becomes
7
7t0 + 5t02 = 0 ) t0 = 0; t0 = (3.47)
5
The acceptable value is t0 = 0. Thus the shortest distance from the origin
is given by the magnitude of the vector
p p
~r0 = ~r0 + ~v t0 = ~r0 ) r0 = 12 + 22 + 32 = 14 (3.48)

Example 4 Find the shortest distance from the origin to the plane

3x 2y 6z = 7 (3.49)

Solution: Comparing
3x 2y 6z = 7 (3.50)
to the equation of a plane

ax + by + cz = d (3.51)

we have
~ = 3^
N x 2^
y 6^
z (3.52)
which is a vector normal to the plane. The unit vector along normal vector
is given by

~
^ = N = q 3^
N
x 2^ y 6^z ^ = 3x
)N ^
2
y^
6
z^ (3.53)
~
N 2 2 7 7 7
32 + ( 2) + ( 6)

We now pick any point on the plane. This point must satisfy the equation

3x 2y 6z = 7: (3.54)
52 CHAPTER 3. VECTORS, LINES, AND PLANES

We may pick (3; 1; 0) which we write as

~r = 3^
x + 1^
y (3.55)

Then the distance from the origin to the plane is the component of this
~ ; because the distance
vector along the vector perpendicular to the plane, N
is measured by the length of the normal to the plane. We can write this
distance as
d = ~r N^ = 3 3 2 1 ) d = 1units (3.56)
7 7

Example 5 Find the symmetric and parametric equations of the line through
(0; 2; 4) and (3; 2; 1).

Solution: The symmetric equations of a line are given by


x x0 y y0 z z0
= = (3.57)
a b c
where
~ = a^
A x + b^
y + c^
z (3.58)
is a vector parallel to the line passing through (0; 2; 4) and (3; 2; 1) .
Noting that a vector parallel to the line passing through can be expressed
as

~ = (3
A 0) x
^+( 2 2) y^ + (1 ~ = 3^
4) z^ ) A x + 0^
y 3^
z (3.59)

we have
a = 3; b = 0; c = 3 (3.60)
so that the symmetric equations becomes

x 0 z 4
= ;y = 2 (3.61)
3 3
The parametric equations are given by
8
< x = x0 + at
~ or
~r = ~r0 + At y = y0 + bt (3.62)
:
z = z0 + ct

which give 8
< x = 3t
~r = y= 2 (3.63)
:
z = 4 3t

Example 6 Find the equation of the plane containing the three points (0; 1; 1),
(2; 1; 3), and (4; 2; 1).
3.5. HOMEWORK ASSIGNMENT 6 53

Solutions: We …rst construct two vectors A ~ and B~ that lie on the plane formed
by these three points. Consider the vector pointing from (0; 1; 1) to (2; 1; 3)
and the vector from (0; 1; 1) to (4; 2; 1) and let denote these vectors by A ~
~
and B, respectively. The we can write

~ = (2
A 0) x
^ + (1 1) y^ + (3 1) z^ = 2^
x + 2^
z (3.64)
~ = (4
B 0) x
^ + (4 1) y^ + (1 1) z^ = 4^
x + 3^
y

The vector, N ~ , that is normal to the plane formed by these two vectors
(i.e. the three points) can be determined using the cross product of the
two vectors

x
^ y^ z^
~ =A
~ ~ = ~ =x 2 2 0 2 0 2
N B 0 2 2 )N ^ y^ + z^
3 0 4 0 4 3
4 3 0
(3.65)
~ =
N 6^
x + 8^
y 8^
z (3.66)
We recall that equation of a plane is given by

a (x x0 ) + b (y y0 ) + c (z z0 ) = 0 (3.67)

where
~ = a^
N x + b^
y + c^
z (3.68)
and
~r0 = x0 x
^ + y0 y^ z0 z^ (3.69)
is a point on the plane as measured from the origin. Then we can write
the equation of the plane formed by the three points as

6 (x x0 ) + 8 (y y0 ) 8 (z z0 ) = 0 (3.70)

where
~r0 = x0 x
^ + y0 y^ z0 z^ (3.71)
is given by ~r0 = (0; 1; 1), (2; 1; 3), or (4; 2; 1):

3.5 Homework Assignment 6


~=
1. Find the angles between A 2i + j ~ = 2i
2k and B 2j
~ = 2i j + 4k and B
2. Show that the vectors A ~ = 5i + 2j 2k are orthogonal
(perpendicular) and …nd a third vector perpendicular to both.

3. Show that the lines that join (0; 0; 0) to (1; 2; 1) and (1; 1; 1) to (2; 3; 4)
intersect and …nd the acute angle between them.
54 CHAPTER 3. VECTORS, LINES, AND PLANES

4. A particle is traveling along the line (x 3)=2 = (y + 1)=( 2) = z 1:


~ Find the distance
Write the equation of this path in the form ~r = ~r0 + At.
of closest approach of the particle to the origin (that is, the distance from
the origin to the line). If t represents time, show that the time of the
closest approach is
A~ ~r0
t= :
A2
Use this value to check your answer for the distance of closest approach.
If P is the point of closest approach, what is A ~ ~r?

5. The vectors A~ = ai + bj and B


~ = ci + dj form two sides of a parallelogram.
Show that the area of the parallelogram is given by the absolute value of
the following determinant
a b
c d
Chapter 4

Matrix and determinants

4.1 Important Terminologies


A matrix : A rectangular array of quantities, usually inclosed in large parenthe-
ses with n rows and m columns.
2 3
a11 a12 a1m 1 a1m
6 a21 a22 a2m 1 a2m 7
6 7
6 7
6 7
A=6 6 7 (4.1)
7
6 7
6 7
4 an 11 an 12 an 1m 1 an 1m 5
an1 an2 anm 1 anm

when n = m the matrix is called a square matrix.


Transpose of a Matrix : The transpose of a Matrix A in Eq. (4.1) which we
denote by AT is obtained by simply write the rows of matrix A as columns.
2 3
a11 a21 an 11 an1
6 a12 a22 an 12 an2 7
6 7
6 7
6 7
T
A =6 6 7 (4.2)
7
6 7
6 7
4 a1m 1 a2m 1 an 1m 1 anm 1 5
a1m a2m an 1m anm

0 1
2 2
2 3 1
A = ; AT = @ 3 1 A (4.3)
2 1 0
1 0
0 1
2 4
2 1 3
B = @ 1 1 A ; BT =
4 1 1
3 1

55
56 CHAPTER 4. MATRIX AND DETERMINANTS

The Identity Matrix; I (The Unit Matrix, U )

IA = AI = A (4.4)

4.2 Matrix Arithmetic and Manipulation


Consider the three matrices:
0 1 0 1
2 4 2 1 3
2 3 1
A = ;B = @ 1 1 A;C = @ 4 1 2 A ; (4.5)
2 1 0
3 1 1 0 1
0 1
2 0 1
D = @ 1 1 2 A
3 1 0

Multiplication by a Scalar : Any matrix can be multiplied by a scalar:

2 2 3 2 1 2 4 2 4 6 2 8
2A = = (4.6)
2 2 1 2 0 2 5 2 4 2 0 10

Addition and subtraction: Two matrices can be added or subtracted if and


only if they have the same dimensions. From matrices A; B; C, and D we can
add/ subtract only matrices C and D
0 1 0 1
2 1 3 2 0 1
C +D = @ 4 1 2 A+@ 1 1 2 A (4.7)
1 0 1 3 1 0
0 1 0 1
2 2 1+0 3+1 0 1 4
= @ 4+1 1 1 2+2 A=@ 5 2 0 A
1+3 0+1 1+0 2 1 1

Matrix Multiplication: two matrices can be multiplied if and only if the num-
ber of columns of the …rst matrix is equal to the number of rows of the second
matrix. If matrices have the same dimension, then they can be multiplied. From
the above matrices we can make the multiplications:
0 1
2 4
2 3 1 @ (ab)11 (ab)12
AB = 1 1 A= (4.8)
2 1 0 (ab)21 (ab)22
3 1
0 10 1 0 1
2 1 3 2 0 1 (cd)11 (cd)12 (cd)13
CD = @ 4 1 2 A@ 1 1 2 A = @ (cd)21 (cd)22 (cd)23 A
1 0 1 3 1 0 (cd)31 (cd)32 (cd)33
(4.9)
but we can not make the matrix multiplications BC or BD
4.3. MATRIX REPRESENTATION OF A SET OF LINEAR EQUATIONS57

The element in row i and column j of the product matrix AB is equal to


row i of A times column j of B. In index notation
n
X
(ab)ij = aik bkj ; (4.10)
k=1

where (ab)ij is the element of the product matrix AB:


Commutativities: For any two multiplyable matrices C and D;

CD 6= DC (4.11)

Commutator : For square matrices C and D the Commutator [C; D] is de-


…ned as
[C; D] = CD DC (4.12)
For any three matrices, F; G;and H that can be multiplied we can write
The Associative Law :
F (GH) = (F G)H (4.13)
The Distributive Law :

F (G + H) = F G + F H (4.14)

4.3 Matrix representation of a set of linear equa-


tions
A set of linear equations

a1 x + b1 y + c1 z + d1 w = c1 ; (4.15)
a2 x + c2 z + d2 w = c2 ;
b3 y + c3 z = c4 ;
a4 x = c4 :

can be expressed using matrices as


2 32 3 23
a1 b1 c1 d1 x c1
6 a2 0 c2 d2 7 6 7 6 c2 7
6 76 y 7=6 7
4 0 b3 c3 0 54 z 5 4 c3 5 : (4.16)
a4 0 0 0 w c4
or
M r = k; (4.17)
where 2 3
a1 b1 c1 d1
6 a2 0 c2 d2 7
M =6
4 0
7 (4.18)
b3 c3 0 5
a4 0 0 0
58 CHAPTER 4. MATRIX AND DETERMINANTS

is called the coe¢ cient matrix,


2 3
x
6 y 7
r=6 7
4 z 5; (4.19)
w

and 2 3
c1
6 c2 7
k=6 7
4 c3 5 : (4.20)
c4
One can augment the matrices M and k and write a matrix
2 3
a1 b1 c1 d1 c1
6 a2 0 c2 d2 c2 7
A=6 4 0 b3 c3 0
7 (4.21)
c3 5
a4 0 0 0 c4

which is referred as the Augmented Matrix.

4.4 Matrix application: solving a set of linear


equations
With application of matrices we can solve a set of linear equations. There are
two di¤erent methods. The …rst involves the Gaussian Elimination method and
Row Echelon Form where as the second method involves Cramer’s Rule. We
…rst discuss the …rst method. The second method will be discussed after we get
introduced to square matrices and determinants.
Gaussian Elimination method (named after German mathematician and sci-
entist Carl Friedrich Gauss): in this method we follow the following basic steps
to solve a set of linear equations:

1st Write the set of linear equations using matrices.

2nd Obtain the augmented matrix.

3rd Apply any of the following elementary row operations

(a) Multiply any row of the matrix by a non-zero constant,

(b) Interchange any two rows,

(c) Replace any row by a linear combination of the rows of the matrix (as long
as the linear combination includes the row being replaced). In other words
you can add to or subtract from a multiple of one row to another,
4.4. MATRIX APPLICATION: SOLVING A SET OF LINEAR EQUATIONS59

4th Repeat step 2 until the augmented matrix is in Row Echelon Form and
entirely row reduced.

Row Echelon Form: a matrix is in Row Echelon Form if all of the following
conditions are satis…ed

All nonzero rows (rows with at least one nonzero element) are above any
rows of all zeroes.
Note: All zero rows, if any, belong at the bottom of the matrix.

The leading coe¢ cient (the …rst nonzero number from the left, also called
the pivot) of a nonzero row is always strictly to the right of the leading
coe¢ cient of the row above it

All entries in a column below a leading entry are zeroes (implied by the
…rst two criteria).

Reduced row echelon form (row canonical form): A matrix that is in row
echelon form and also its every leading coe¢ cient is 1.

Example 1 The matrix 2 3


0 1 4 0
6 0 0 1 0 7
B=6
4 0
7 (4.22)
0 0 1 5
0 0 0 0
in Row Echelon Form and in a reduced row form where as the matrix
2 3
1 1 1 1
C=4 0 9 0 2 5 (4.23)
0 0 0 3

is in Row Echelon Form but not in reduced row form. The following
matrix is not in Row Echelon Form
2 3
1 2 3 4
C=4 0 3 7 2 5 (4.24)
0 2 0 0

as the leading coe¢ cient of row 3 is not strictly to the right of the leading
coe¢ cient of row 2.

Rank of a Matrix : the number of nonzero rows remaining when a matrix


has been row reduced is called the rank of the matrix.

(a) If (rank M )<(rank A), the equations are inconsistent and there is no solu-
tion.

(b) If (rank M )=(rank A)= n (number of unknowns), there is one solution.


60 CHAPTER 4. MATRIX AND DETERMINANTS

(c) If (rank M )=(rank A)= R < n, then R unknowns can be found in terms of
the remaining n R unknowns.
Example 2 Consider the circuit shown. The circuit element values are R =
1:0W and V = 2:0V .
(a) Applying Kirchho¤’s voltage and current rules …nd set of linear equations.
(b) Find the coe¢ cient and augmented matrices for the set of linear equations.
(c) Find the values of the three branch currents I1 ; I2 ; and I3 using Gaussian
elimination method.
(d) Find the rank of the coe¢ cient and augmented matrices. Is the ranks
consistent with what we discussed earlier?
Kirchho¤ ’s Voltage Law : Start at one point in a circuit and go around any
closed loop in the circuit. The voltage at the starting point of the loop must be
the same as the voltage at the ending point, since they are the same point in
the circuit. Another way of stating this is simply that the net change in voltage
as you go around any closed loop in the circuit must be zero (so the starting
potential must be the same as the ending potential). This is just a restatement
of the conservation of energy.
Kirchho¤ ’s Current Law :The net current ‡owing into any node of a circuit
must equal the net current ‡owing out of that same node. This law says that
there is no place from which charge magically appears in the circuit, and no
place that the charge disappears. There is a net conservation of charge in an
electric circuit.
Solution:
(a) Applying Kirchho¤’s voltage law for the closed loop abef a we can write
V
V +I1 R+2I2 R 2 V +I1 R = 0 ) 2I1 R+2I2 R = V ) I1 +I2 =
2R
(4.25)
and for closed loop bcdeb
I3 R 2 V + I3 R + 2 V 2I2 R = 0 ) 2I3 R 2I2 R = 0 I2 + I3 = 0:
(4.26)
4.4. MATRIX APPLICATION: SOLVING A SET OF LINEAR EQUATIONS61

Using Kirchho¤ ’s Current Law at node b; we may write

I1 = I2 + I3 ) I1 I2 I3 = 0 (4.27)

(b) Substituting R = 1and V = 2V in the results we obtained above, we have

I1 + I2 = 1 (4.28)
I2 + I3 = 0
I1 I2 I3 = 0

or

I1 + I2 + 0I3 = 1 (4.29)
0I1 I2 + I3 = 0
I1 I2 I3 = 0

so that we can put these equations using matrices as

M I = K; (4.30)

where coe¢ cient matrix M is given by


2 3
1 1 0
M =4 0 1 1 5 (4.31)
1 1 1

and 2 3 2 3
I1 1
I = 4 I2 5 ; K = 4 0 5 : (4.32)
I3 0
The augmented matrix is a combination of the coe¢ cient and the constant
matrix given by 2 3
1 1 0 1
A=4 0 1 1 0 5 (4.33)
1 1 1 0

(c) We apply elementary row operations to row reduce the augmented matrix.
2 3
1 1 0 1
A=4 0 1 1 0 5 (4.34)
1 1 1 0

Subtract row 1 from row 3 :


2 3
1 1 0 1
A=4 0 1 1 0 5 (4.35)
0 2 1 1
62 CHAPTER 4. MATRIX AND DETERMINANTS

Multiply row 2 by 2 and subtract from row three:


2 3
1 1 0 1
A=4 0 1 1 0 5 (4.36)
0 0 3 1
Divide row 3 by 3:
2 3
1 1 0 1
A=4 0 1 1 0 5 (4.37)
1
0 0 1 3
Subtract row 3 from row 2:
2 3
1 1 0 1
A=4 0 1 0 1
3
5 (4.38)
1
0 0 1 3
Divide row 2 by 1:
2 3
1 1 0 1
A=4 0 1 0 1
3
5 (4.39)
1
0 0 1 3
Subtract row 2 from row 1:
2 2
3
1 0 0 3
A=4 0 1 0 1
3
5 (4.40)
1
0 0 1 3
Therefore the values of the three currents would be
2 3 2 2 3
I1 3
4 I2 5 = 4 1 5 (4.41)
3
1
I3 3
which means
2 1 1
I1 = A; I2 = A; I3 = A (4.42)
3 3 3
(d) The number of nonzero rows for the row reduced augmented matrix
2 3
1 0 0 23
A=4 0 1 0 13 5 (4.43)
0 0 1 13
is 3 and the rank is 3. For the reduced coe¢ cient matrix (the …rst three
columns in the reduced augmented matrix)
2 3
1 0 0
M =4 0 1 0 5 (4.44)
0 0 1
we have also 3 nonzero rows and the rank is 3. There are 3 unknowns,
this means that (rank M )=(rank A)= 3 (number of unknowns), therefore
we …nd one solution for each unknowns.
4.5. DETERMINANT OF A SQUARE MATRIX 63

4.5 Determinant of a square matrix


The determinant of a 2 2 matrix : Suppose we are given a 2 2 matrix
a11 a12
A= ; (4.45)
a21 a22
the determinant is given by
2
X
jAj = ( 1)i+j aij Mij ; (4.46)
j=1

where Mij is called the minor of aij : The minor of aij , Mij ; is the determinant
of the matrix formed by removing the row and the column containing aij : Since
we are given a 2-dimensional square matrix i = 1 or 2. If we chose i = 1, we
have
2
X
jAj = ( 1)1+j a1j M1j = ( 1)1+1 a11 M11 + ( 1)1+2 a12 M12 (4.47)
j=1

so that noting that


M11 = a22 ; M12 = a21 (4.48)
one …nds
jAj = a11 a22 a12 a21 : (4.49)
The determinant of a 3 3 matrix: Consider a 3-dimensional matrix
2 3
^{ |^ k^
C = 4 Ax Ay Az 5 (4.50)
Bx By Bz
constructed from two 3-dimensional vectors in Cartesian coordinates and the
^ Using Eq. (4.59) we may write the determinant of this
unit vectors, ^{, |^, and k.
matrix as
3
X
jAj = ( 1)1+j aij Mij (4.51)
j=1

= ( 1)1+1 a11 M11 + ( 1)1+2 a12 M12 + ( 1)1+3 a13 M13 ;


where we chose i = 1. Note that for a 3-dimensiona square matrix i = 1; 2; or
3: Noting that
a11 = ^{; a12 = |^; and a13 = k^ (4.52)
the corresponding Minors M11 ; M12 , and M13 are given by
Ay Az
M11 = = Ay Bz Az B y ; (4.53)
By Bz
Ax Az
M12 = = Ax B z Az B x ; (4.54)
Bx Bz
Ax Ay
M13 = = Ax B y Ay B x ; (4.55)
Bx By
64 CHAPTER 4. MATRIX AND DETERMINANTS

we …nd for Eq. (4.51)


jCj = ( 1)1+1^{ (Ay Bz Az By ) + ( 1)1+2 |^(Ax Bz Az Bx )
+( 1)1+3 k^ (Ax By Ay Bx )
) jCj = ^{ (Ay Bz Az By ) + |^(Az Bx Ax Bz ) + k^ (Ax By Ay B(4.56)
x)

Note: The result in Eq. ( 4.56) shows that the determinant of the matrix C is
~ and B,
the cross product of vector A ~

x^ y^ z^
~
A ~ =
B Ax Ay Az : (4.57)
Bx By Bz
We now consider a square, n n; matrix A (a matrix with n rows and n
columns) 2 3
a11 a12 a1j a1n 1 a1n
6 a21 a22 a2j a2n 1 a2n 7
6 7
6 7
6 7
6 7
6 7
6 7
6 7
A=6 6 a i1 ai2 aij ain 1 ain
7
7 (4.58)
6 7
6 7
6 7
6 7
6 7
6 7
4 an 11 an 12 an 1j an 1n 1 an 1n 5
an1 an2 anj ann 1 ann
Then the determinant of this matrix (represented as jAj or det A ) is given by
n
X n
X
jAj = ( 1)i+j aij Mij = ( 1)i+j aij Mij (4.59)
j=1 i=1

where Mij is called the minor of aij : The minor of aij , Mij ; is the determinant
of the matrix formed by removing the row and the column containing aij :
Useful properties of determinants: It is useful to know the following proper-
ties of determinants of a square matrix which can be proved using Eq. (4.59).
1. Multiplying by a constant: Suppose a row or a column of a square matrix A
is multiplied by a constant k, then the determinant of the new matrix is
k times the determinant of the matrix A.
Example 3 For the 2-dimensiona square matrix
a11 a12
A= ; (4.60)
a21 a22
we may write a new matrix, B given by
ka11 a12 ka11 ka12
B= or B = ; (4.61)
ka21 a22 a21 a22
4.5. DETERMINANT OF A SQUARE MATRIX 65

then
det B = k det A; (4.62)
2. Zero determinant value: The value of the determinant of square matrix is
zero if
(a) all elements of a row (or a column) are zero
Example 4
0 a12 a11 a12
A= or A = ) det A = 0; (4.63)
0 a22 0 0

(b) two rows (or columns) are identical


Example 5
a11 a12 = a11 a11 a21
A= or A = ) det A = 0
a21 a22 = a21 a21 = a11 a22 = a21
(4.64)
(c) two rows (or columns) are proportional
Example 6
a11 a12 = ka11 a11 a12
B= or B = ) det A = 0;
a21 a22 = ka21 a21 = ka11 a22 = ka12
(4.65)
where k is the proportionality constant.
3. Interchanging a row or a column: If two rows (or two columns) of a square
matrix are interchanged, the determinant of the matrix remain the same.
Example 7
a11 a12 a12 a11
A = ,B =
a21 a22 a22 a21
) det A = det B (4.66)

Example 8 Evaluate the determinant of the matrix A:


0 1
2 2 0
A=@ 0 2 2 A (4.67)
1 1 1

Solution: Applying the relation above we can express the determinant as


1+1 2 2 1+2 0 2
jAj = 2 ( 1) + 2 ( 1) (4.68)
1 1 1 1
0 1
2 2 0
1+3 @ 0
+0 ( 1) 2 2 A (4.69)
1 1 1
66 CHAPTER 4. MATRIX AND DETERMINANTS

2 2 0 2
jAj = 2 2 (4.70)
1 1 1 1

jAj = 2 (( 2 1) (2 1)) 2 ((0 1) (2 1)) = 8 + 4


) jAj = 12 (4.71)

4.6 Cramer’s Rule


Let a system of simultaneous equations be represented by the augmented matrix
A, and let M be the corresponding coe¢ cient matrix. Also, let Mj be the
coe¢ cient matrix with its j th column replaced by the constant column from
Aaug . We de…ne the following determinants:

D = jM j ; Dj = jMj j (4.72)

The value of the j th unknown, xj ; is then equal to

Dj
xj = (4.73)
D
provided D 6= 0

Example 9 Solve the system of equations from Example 2 using Cramer’s rule.
( D = 12; D1 = 8; D2 = 4; and D3 = 4.)

Solution: We recall that the coe¢ cient matrix is given by


2 3
1 1 0
M =4 0 1 1 5 (4.74)
1 1 1

and the corresponding Augmented matrix by


2 3
1 1 0 1
A=4 0 1 1 0 5: (4.75)
1 1 1 0

Replacing the …rst column of M by the last column of A; we …nd


2 3
1 1 0
M1 = 4 0 1 1 5; (4.76)
0 1 1

the second column of M by the last column of A


2 3
1 1 0
M2 = 4 0 0 1 5 (4.77)
1 0 1
4.7. THE ADJOINT AND INVERSE OF A MATRIX 67

and the third column of M by the last column of A


2 3
1 1 1
M3 = 4 0 1 0 5: (4.78)
1 1 0
For the determinants of the matrices M; M1 ; M2 ; and M3 , one can easily
…nd
1 1 0 1
jM j = ) jM j = 2 + 1 = 3; (4.79)
1 1 1 1

1 1 0 1
jM1 j = ) jM1 j = 2; (4.80)
1 1 0 1
0 1 0 1
jM2 j = ) jM2 j = 1; (4.81)
0 1 1 1
and
1 0 0 0 0 1
jM3 j = + ) jM3 j = 1: (4.82)
1 0 1 0 1 1
Therefore, the three currents are given by
jM1 j 2 jM2 j 1 jM2 j 1
I1 = = ; I2 = = ; I2 = = : (4.83)
jM j 3 jM j 3 jM j 3

4.7 The adjoint and inverse of a matrix


Adjoint of a Matrix : The adjoint of a square matrix, A, is given by

adj(A) = [cof (A)]T (4.84)

where cof (A) is the cofactor of the matrix A. We recall that the minor of matrix
A (Mij ) is the determinant of the matrix formed from matrix A by removing
the ith row and j th column. For the cofactor matrix the elements are expressed
as
i+j
[cof (A)]ij = ( 1) Mij : (4.85)
1
Inverse of a (Square) Matrix : A
1 1
A A = AA =I (4.86)

We can determine the inverse of an invertible matrix (det jAj 6= 0) using row
reduction or the adjoint matrix.

a. Row reduction in this approach for the matrix, for example,


2 3
a11 a12 a13
A = 4 a21 a22 a23 5 (4.87)
a31 a32 a33
68 CHAPTER 4. MATRIX AND DETERMINANTS

we start from 2 3
a11 a12 a13 1 0 0
4 a21 a22 a23 0 1 0 5 (4.88)
a31 a32 a33 0 0 1
and do elementary row operation until we end up with
2 3
1 0 0 b11 b12 b13
4 0 1 0 b21 b22 b23 5 (4.89)
0 0 1 b31 b32 b33

so that the inverse of the Matrix A is given by


0 1
b11 b12 b13
A 1 = @ b21 b22 b23 A : (4.90)
b31 b32 b33

b. Using the adjoint matrix: Using the adjoint matrix the inverse can be ex-
pressed as
T
[cof (A)]
A 1=
det jAj
Example 10 Find the inverse of the matrix
0 1
1 2 3
A=@ 2 0 4 A (4.91)
1 1 1

using

a. Row reduction approach


b. The adjoint matrix approach

Solution: a. In the row reduction approach we start from


2 3
1 2 3 1 0 0
4 2 0 4 0 1 0 5 (4.92)
1 1 1 0 0 1

and try to get


2 3
1 0 0 a11 a12 a13
4 0 1 0 a21 a22 a23 5 (4.93)
0 0 1 a31 a32 a33

so that we can get the inverse matrix


0 1
a11 a12 a13
A 1 = @ a21 a22 a23 A : (4.94)
a31 a32 a33
4.7. THE ADJOINT AND INVERSE OF A MATRIX 69

In order to get this matrix let’s make the elementary row operation.
Add row 1 to row 3:
2 3
1 2 3 1 0 0
4 2 0 4 0 1 0 5 (4.95)
0 1 4 1 0 1

Add row 3 to row 2:


2 3
1 2 3 1 0 0
4 2 1 0 1 1 1 5 (4.96)
0 1 4 1 0 1

Multiply row 2 by 2 and subtract the result from row 1


2 3
5 0 3 1 2 2
4 2 1 0 1 1 1 5 (4.97)
0 1 4 1 0 1

Divide row 1 by -5 and row 2 by 2:


2 3
1 0 3=5 1=5 2=5 2=5
4 1 1=2 0 1=2 1=2 1=2 5 (4.98)
0 1 4 1 0 1

Subtract row 1 from row 2


2 3
1 0 3=5 1=5 2=5 2=5
4 0 1=2 3=5 3=10 1=10 1=10 5 (4.99)
0 1 4 1 0 1

Divide row 2 by 1/2


2 3
1 0 3=5 1=5 2=5 2=5
4 0 1 6=5 3=5 1=5 1=5 5 (4.100)
0 1 4 1 0 1

subtract row 2 from row 3


2 3
1 0 3=5 1=5 2=5 2=5
4 0 1 6=5 3=5 1=5 1=5 5 (4.101)
0 0 14=5 2=5 1=5 4=5

Divide row 1 by 3=5, row 2 by 6=5, and row 3 by 14=5


2 3
5=3 0 1 1=3 2=3 2=3
4 0 5=6 1 1=2 1=6 1=6 5 (4.102)
0 0 1 1=7 1=14 2=7
70 CHAPTER 4. MATRIX AND DETERMINANTS

subtract row 3 from row 2


2 3
5=3 0 1 1=3 2=3 2=3
4 0 5=6 0 5=14 5=21 5=42 5 (4.103)
0 0 1 1=7 1=14 2=7
subtract row 3 from row 1
2 3
5=3 0 0 10=21 25=42 20=21
4 0 5=6 0 5=14 5=21 5=42 5 (4.104)
0 0 1 1=7 1=14 2=7

Divide row 1 by 5=3 and row 2 by 5=6


2 3
1 0 0 1=7 5=14 4=7
4 0 1 0 3=7 2=7 1=7 5 (4.105)
0 0 1 1=7 1=14 2=7

Then the inverse matrix is given by


0 1
2=7 5=14 4=7
A 1 = @ 3=7 2=7 1=7 A : (4.106)
1=7 1=14 2=7
Using Matematica

N.B. Use Mathematica only to check your result!!!


b. Here we …rst need to …nd the adjoint matrix of A. In order to …nd the
adjoint, …rst we need to …nd the cofactor matrix. Let this cofactor
matrix be 0 1
c11 c12 c13
cof (A) = @ c21 c22 c23 A (4.107)
c31 c32 c33
where the elements
i+j
cij = ( 1) Mij (4.108)
in which Mij is the minor for the matrix A. Now using
0 1
1 2 3
A=@ 2 0 4 A (4.109)
1 1 1
4.7. THE ADJOINT AND INVERSE OF A MATRIX 71

we can write
0 4 2
M11 = det = 4 ) c11 = ( 1) M11 = 4 (4.110)
1 1

2 4 3
M12 = det = 6 ) c12 = ( 1) M12 = 6 (4.111)
1 1
2 0 4
M13 = det = 2 ) c13 = ( 1) M11 = 2 (4.112)
1 1
2 3 3
M21 = det = 5 ) c21 = ( 1) M21 = 5 (4.113)
1 1
1 3 4
M22 = det = 4 ) c22 = ( 1) M22 = 4 (4.114)
1 1
1 2 5
M23 = det = 1 ) c23 = ( 1) M23 = 1 (4.115)
1 1
2 3 4
M31 = det = 8 ) c31 = ( 1) M31 = 8; (4.116)
0 4
1 3 5
M32 = det = 2 ) c32 = ( 1) M32 = 2; (4.117)
2 4
1 2 6
M33 = det = 4 ) c33 = ( 1) M33 = 4: (4.118)
2 0
There follows that
2 3
4 6 2
cof (A) = 4 5 4 1 5 (4.119)
8 2 4
and 2 3
4 5 8
T
[cof (A)] = 4 6 4 2 5: (4.120)
2 1 4
The determinant of A is given by
1 2 3
det jAj = 2 0 4
1 1 1
0 4 2 4 2 0
= 2 +3 (4.121)
1 1 1 1 1 1
) det jAj = 14

Then the inverse matrix


T
1 [cof (A)]
A = (4.122)
det jAj
72 CHAPTER 4. MATRIX AND DETERMINANTS

becomes
2 3 0 1
4 5 8 2=7 5=14 4=7
1 4
A 1
= 6 4 2 5 = @ 3=7 2=7 1=7 A
14
2 1 4 1=7 1=14 2=7
(4.123)

4.8 Orthogonal matrices and the rotation ma-


trix
Orthogonal Matrices: matrices that make an orthogonal transformation of vec-
tors. In an orthogonal transformation of vectors the magnitude of the vectors
remains the same. For an orthogonal matrix
1
M = MT (4.124)

The Rotation Operator : Consider a rectangular coordinate system (XY Z)


rotated by an angle about the z axis in the counterclockwise direction to give
another coordinate system (X0Y 0Z0) as shown in the …gure below.

Consider a vector ~r which has components (x; y; z) in the XY Z coordinate


system. We want to …nd the components of this vector in the X 0 Y 0 Z 0 coordinate
system. From the …gure above we can write that the new components (x0 ; y 0 ; z 0 )
can be expressed as

x0 = x cos ( ) + y cos (90 ) = x cos ( ) + y sin ( ) (4.125)


0
y = x sin ( ) + y cos ( )
z0 = z
4.8. ORTHOGONAL MATRICES AND THE ROTATION MATRIX 73

Using matrix representation we can write that


0 0 1 0 10 1
x cos ( ) sin ( ) 0 x
@ y 0 A = @ sin ( ) cos ( ) 0 A @ y A ) r0 = Rr (4.126)
z0 0 0 1 z

where the matrix R( ) = R is the rotation matrix given by


0 1
cos ( ) sin ( ) 0
R = @ sin ( ) cos ( ) 0 A (4.127)
0 0 1

~ (I used Mathematica)
The inverse of the matrix R

which is the same as the transpose of the matrix, RT given by


0 1
cos ( ) sin ( ) 0
RT = @ sin ( ) cos ( ) 0 A : (4.128)
0 0 1

Therefore, since
1
R = RT (4.129)

the rotation matrix is orthogonal.


Linear Operators: The operator/function F (x) is a linear operator/function
if
F (ax) = aF (x) (4.130)

and
F (x + y) = F (x) + F (y) (4.131)

If any one or these conditions is not met, then the operator/function is not
linear !

Example 11 Consider the operator that takes its argument (scalar or vector),
squares it, and adds 3:
O(A) = A2 + 3 (4.132)

Is this a linear operator?


74 CHAPTER 4. MATRIX AND DETERMINANTS

Solution: We need to check the two conditions stated above. First condition:
2
O(aA) = (aA) + 3; (4.133)
2 2
aO(A) = a A + 3 = aA + 3a ) O(aA) 6= aO(A)

You can concluded the function is not linear because if one of the condi-
tions is violated then the function is not linear. But let’s check also the
second condition,
2
O(A+B) = (A + B) +3 ) O(A+B) = A2 +AB +BA+B 2 +3 (4.134)

O(A) + O(B) = A2 + 3 + B 2 + 3 ) O(A) + O(B) = A2 + B 2 + 6 (4.135)

which shows
O(A + B) 6= O(A) + O(B) (4.136)

4.9 Linear independence


A set of linear equations or vectors could be linearly dependent or independent.
In order to obtain a complete solution to a set of linear equations using ma-
trices the equations must be linearly independent. Sometimes in some physical
problems it may be necessary to express a given vector as a set of linearly in-
dependent vectors. In this section we will see how linear independency in set of
linear equations, vectors, and functions can be veri…ed and how we can obtain
a set of linearly independent orthogonal vectors.
Linear independency-set of linear equations: The existence of the solutions
to a set of linear equations with n unknowns depends on the linear independency
of the equations which is determined by the Rank of the Augmented (A) and
Coe¢ cient matrices (M ). The Rank of a Matrix is the number of nonzero rows
remaining when a matrix has been row reduced.

a. If (rank M )<(rank A), the equations are inconsistent and there is no solution.

b. If (rank M )=(rank A)= n (number of unknowns), there is one solution.

c. If (rank M )=(rank A)= m < n, then m unknowns can be found in terms of


the remaining n m unknowns.

Example 12 [Linear independence-set of linear equations: in this example the


current rule is excluded in order to illustrate linear independency in a set
of linear equations.] Consider the circuit shown below. The circuit element
values are R = 1:0 and V = 2:0V . Find the values of the three branch
currents I1 ; I2 , and I3 by solving the system of three simultaneous equa-
tions obtained by applying only Kirchho¤’s voltage law to three di¤erent
loops.
4.9. LINEAR INDEPENDENCE 75

Solution: Applying Kirchho¤’s voltage law to three di¤erent loops we can write

V + I1 R + 2I2 R 2 V + I1 R = 0 ) 2I1 R + 2I2 R = V; (4.137)

2 V 2I2 R + I3 R 2 V + I3 R = 0 ) 2I2 R + 2I3 R = 0; (4.138)


and

V + I1 R + I3 R 2 V + I3 R + I1 R = 0 ) 2I1 R + 2I3 R = V: (4.139)

Substituting the values for R and V , we …nd

I1 + I2 = 1; I2 + I3 = 0; I1 + I3 = 1; (4.140)

which may be put in a matrix form


0 10 1 0 1
1 1 0 I1 1
@ 0 1 1 A @ I2 A = @ 0 A : (4.141)
1 0 1 I3 1

There follows that for the Augmented matrix, we have


0 1
1 1 0 1
A=@ 0 1 1 0 A (4.142)
1 0 1 1

and for the coe¢ cient matrix


0 1
1 1 0
M =@ 0 1 1 A: (4.143)
1 0 1

To …nd the Rank of the Augmented and the Coe¢ cient matrix we carry
out elementary row operation:
Add row two to row one: 0 1
1 0 1 1
@ 0 1 1 0 A (4.144)
1 0 1 1
76 CHAPTER 4. MATRIX AND DETERMINANTS

and subtract row three from row one


0 1
0 0 0 0
@ 0 1 1 1 A: (4.145)
1 0 1 1

We note that the Augmented matrix can not be row reduced any further.
The rank of the coe¢ cient matrix is 2 and also the rank of the augmented
matrix is 2. That means (rank M )=(rank A), but the number of unknown
variables is 3. This means only two unknowns can be found in terms of
the third unknown.

Linearly dependent and independent vectors: If we de…ne three vectors from


the three equations we obtained in the example above
~ = (1; 1; 0) ; B
A ~ = (0; 1; 0) ; C
~ = (1; 0; 1) (4.146)

we can see that


~+B
A ~ ~ =0
C (4.147)
These three vectors are linearly dependent vectors. In general for a set of vectors
~1; A
A ~2; A
~ 3 ; :::A
~ n ; if

~ 1 + k2 A
k1 A ~ 2 + k3 A
~ 3 + ::: + kn A
~ n = 0; (4.148)

where ki is none zero real number, the vectors are said to be linearly dependent
otherwise the vectors are linearly independent. Let’s consider three dimensional,
three vectors
~1
A = a1 x
^ + a2 y^ + a3 z^; (4.149)
~1
B = b1 x
^ + b2 y^ + b3 z^;
~3
C = c1 x
^ + c2 y^ + c3 z^:

Using matrix representation, we may write


~ 1 + k2 A
k1 A ~ 2 + k3 A
~ 3 + ::: + kn A~n = 0
0 10 1
k1 a1 k1 a2 k1 a2 x^
) @ k2 b1 k2 b2 k2 b2 A @ y^ A = 0: (4.150)
k3 c1 k3 c2 k3 c2 z^

This equality holds if and only if

k1 a1 k1 a 2 k1 a2
det k2 b1 k2 b2 k2 b2 =0 (4.151)
k3 c1 k3 c2 k3 c2

N.B. This condition is applicable if and only if the dimension of the vector
is the same as the number of the vectors. In other words when the coe¢ cient
matrix forms a square matrix.
4.9. LINEAR INDEPENDENCE 77

Example 13 Consider the set of three mutually orthogonal Cartesian unit vec-
tors. Prove that these three unit vectors form a linearly independent set
of vectors.
Solution: For the three unit vectors we may write that
~1
A = x
^; ) a1 = 1; a2 = a3 = 0 (4.152)
~1
B = y^; b2 = 1; b1 = b3 = 0
~3
C = z^ ) c1 = c2 = 0; c3 = 1

so that
k1 a1 k1 a2 k1 a2 k1 0 0
k2 0
det k2 b1 k 2 b2 k 2 b2 = det 0 k2 0 = k1 = k1 k2 k3 :
0 k3
k3 c1 k3 c2 k 3 c2 0 0 k3
(4.153)
Therefore
k1 a1 k1 a2 k1 a2
det k 2 b1 k 2 b2 k 2 b2 = k1 k2 k3 ; (4.154)
k 3 c1 k 3 c2 k3 c2
can be zero if and only if k1 = 0; k2 = 0; or k3 = 0: This means that the
Cartesian unit vectors are linearly independent.

In general, any set of linearly independent vectors either is already a


mutually orthogonal set of vectors or else the vectors in the set can be combined
to form a set of mutually orthogonal vectors (in a process called Gram-Schmidt
orthogonalization).
A Basis Set of Vectors: these are set of linearly independent vectors which
span a vector space.
A Spanning Set of Vectors: a set of vectors spans a space if all the vectors
in the space can be written as a linear combination of the spanning basis set of
vectors.

Example 14 Consider the four set of vectors given by


~
A = ~ = (5; 2; 1) ;
(1; 4; 5) ; B (4.155)
~
C = ~ = (3; 6; 11)
(2; 1; 3) ; D

and the two dimensional vector space (a plane) formed by the two vectors

~r1 = (9; 0; 7) ; ~r2 = (0; 9; 13) (4.156)

and the origin. Show that


~ B;
(a) The four vectors A; ~ C;and
~ ~ are a spanning set of vectors in the two
D
dimensional space formed by the two vectors ~r1 and ~r2 .
(b) Show that the vectors ~r1 and ~r2 are a basis set of vectors for the two
dimensional vector space.
78 CHAPTER 4. MATRIX AND DETERMINANTS

Solution: If we perform row reduction for the matrix formed by the four vec-
tors, 2 3
1 4 5
6 5 2 1 7
M =6 4 2
7 (4.157)
1 3 5
3 6 11
we …nd 2 3
9 0 7
6 0 9 13 7
M =6
0
4 0
7: (4.158)
0 0 5
0 0 0
All the row operations are reversible and we can write the vectors
~ = (1; 4; 5) ; B
A ~ = (5; 2; 1) ; (4.159)
~ = (2; 1; 3) ; D
C ~ = (3; 6; 11)

in terms of the basis set of vectors


~r1 = (9; 0; 7) ; ~r2 = (0; 9; 13) (4.160)

4.10 Gram-Schmidt orthogonalization


We call a set of vectors orthonormal if they are mutually orthogonal (perpen-
dicular), and each vector is normalized (that is its norm is one). For example
the vectors x^; y^; z^ form an orthonormal set. Gram-Schmidt orthogonalization
is a systematic process of obtaining orthonormal basis vectors from a set of
basis vectors. Suppose we have basis vectors A, ~ B,
~ and C;~ we can obtain an
orthonormal set of basis vectors applying the following procedure:
~ you get the …rst orthonormal basis vector
1. Normalize A-
~
A
A^ = : (4.161)
~
A

2.
~ along the direction of A
(a) Find the component of the second vector, B; ~

~A = B
B ~ A^ A:
^ (4.162)

~ and normalize the resulting vector,


(b) Subtract this component from A
~
B ~ A^ A^
B
^=
B : (4.163)
~
B ~ A^ A^
B

This is the second orthonormal vector in the set. It is orthogonal to A^


and it has a unit magnitude.
4.10. GRAM-SCHMIDT ORTHOGONALIZATION 79

3.
~ along the directions of A
(a) Find the components of the third vector, C; ~ and
~
B
C~A = C ~ A^ A;^ C~B = C ~ B^ B:^ (4.164)

~ and normalize the resulting vector,


(b) Subtract these components from C

~
C ~ A^ A^
C ~ B
C ^ B
^
C^ = : (4.165)
~
C ~ A^ A^
C ~ B
C ^ B
^

This is the third orthonormal vector in the set. It is orthogonal to both


A^ and B^ it also has a unit magnitude.

4. For a space of higher dimensions, we continue to implement this procedure.


Example 15 For the given set of basis vectors, use Gram-Schmidt method to
…nd an orthonormal set
~ = (0; 2; 0; 0) ; B
A ~ = (3; 4; 0; 0) ; C
~ = (1; 2; 3; 4) : (4.166)

~
Solution: Following the …rst step we normalize vector A

~
A
A^ = = (0; 1; 0; 0) : (4.167)
~
A

and using step 2(a), we have

B~A = B ~ A^ A^ = [(3; 4; 0; 0) (0; 1; 0; 0)] (0; 1; 0; 0) = (0; 4; 0; 0)


(4.168)
so that following step 2(b) the second orthonormal vector in the set can
be expressed

~
B ~ A^ A^
B
^= (3; 4; 0; 0) (0; 4; 0; 0) (3; 0; 0; 0)
B = = = (1; 0; 0; 0)
~
B ~ A^ A^
B j(3; 4; 0; 0) (0; 4; 0; 0)j j(3; 0; 0; 0)j
(4.169)
Using
~ B
C ^ = (1; 2; 3; 4) (1; 0; 0; 0) = 1
~ A^ = (1; 2; 3; 4) (0; 1; 0; 0) = 2
C (4.170)

one …nds
~
C ~ A^ A^
C ~ B
C ^ B
^ = (1; 2; 3; 4) 2 (0; 1; 0; 0) (1; 0; 0; 0) = (0; 0; 3; 4) :
80 CHAPTER 4. MATRIX AND DETERMINANTS

The third orthonormal basis vector, following step 3(a) and (b), becomes

~
C ~ A^ A^
C ~ B
C ^ B
^
(0; 0; 3; 4) 3 4
C^ = = = 0; 0; ;
~
C ~ A^ A^
C ~ B
C ^ B
^ j(0; 0; 3; 4)j 5 5
(4.171)

Function Spaces and Linear Independence of Sets of Functions: If f1 (x) ; f2 (x) ; f3 (x) ;
:::fn (x) have derivatives of order n 1, and if the determinant

f1 (x) f2 (x) ::: fn (x)


f10 (x) f20 (x) ::: fn0 (x)
: : :
W = det 6= 0 (4.172)
: : :
: : :
f1n 1 (x) f2n 1 (x) :: fnn 1 (x)

then the functions are linearly independent. The determinant W is called the
Wronskian of the functions.

Example 17 Is the set of functions {sin (!t), sin2 (!t)} a linearly independent
set? (! = constant)
Solution: We can write the Wronskian for the two functions as
f1 (x) f2 (x) sin (!t) sin2 (!t)
W = det = det
f10 (x) f20 (x) d
dt [sin (!t)] d
dt sin2 (!t)

sin (!t) sin2 (!t)


= det
! cos (!t) 2! sin (!t) cos (!t)
= 2! sin2 (!t) cos (!t) ! sin2 (!t) cos (!t)

W = ! sin2 (!t) cos (!t) 6= 0


and the functions are independent.

4.11 Homework Assignment 7


1. Solve the following set of equations for the four unknowns by reducing the
matrix using elementary row operations..

x 2y 4 = 0; 2z + w = 5; w + y = 2; 3x 1 = 5z:

2. Solve the following dc circuit problems from the algebra-based physics


class using row reduction of matrices
(a) The circuit diagram for this problem is shown in the …gure below
4.12. HOMEWORK ASSIGNMENT 8 81

(b)
3.
(a) What is meant by the speci…c gravity (s:g) of a substance?
(b) What is Archimedes’Principle? Provide a clear statement of this principle.
(c) An object is composed of x grams of lead (sg = 11) and y grams of tin
(sg = 7). The object has a mass of 82grams in air, and an apparent mass
of 77grams when suspended in oil having a speci…c gravity of 0:5. Find
the values of x and y.
4. The half-life of a radioactive substance is the time it takes for half of it
to decay into some other substances called decay products. Suppose that
a radioactive sample consists of components A and B having half-lives
2hr and 3hr, respectively. Assume that the decay products are gases that
escape at once. At the end of 12hr the sample is found to have a mass of
56g (grams), and is found to have a mass of 12g at the end of 18hr. Find
the masses of A and B that were originally present in the sample.
5. Do all the problems in the sample test. No need of turning your solutions
in!

4.12 Homework Assignment 8


1.
82 CHAPTER 4. MATRIX AND DETERMINANTS

(a) Given the following sets of equations


1 p 1 p
x0 = x + y 3 ; y0 = x 3 + y ; and z 0 = z
2 2
and
00 1 p 00 1 0p 00
x = x0 + y 0 3 ; y = x 3 + y 0 ; and z = z 0
2 2
write each set as a matrix equation and then solve for x00 and y 00 in terms
of x and y by multiplying matrices. Theses equations represent rotations
of axes about the z-axis. By comparing the equations and matrices with
the rotation matrix derived in class, …nd the rotation angles and check
your result.
(b) Prove that the matrices corresponding to the two sets of equations given
in (a) above can correspond to rotation matrices by demonstrating that
they are orthogonal matrices.
2. Show that the function
~ ~r + 3
f (~r) = A
is linear or none linear function.
3. Show that the vector function
~
f (~r) = A ~r

is a linear or none linear vector function.


4. De…ne the integral with respect to x from 0 to 1; the object being operated
on are function of x. Show that this integral operator is linear or none
linear operator. (This means we want to show that the operator

Z1
() dx
0

is a linear or none linear operator.)


5.
(a) For the matrix 2 3
1 0 5i
A=4 2i 2 0 5
1 1+i 0
…nd the transpose, the inverse , the complex conjugate, and the transpose
conjugate.
(b) Show that the product AAT is a symmetric matrix.
4.13. HOMEWORK ASSIGNMENT 9 83

4.13 Homework Assignment 9


1. For each of the following problems write and tow reduce the augmented
matrix to …nd out whether the given set of equations have exactly one
solution, no solution, or an in…nite set of solutions.
(a)
2x + y z = 2
4x + y 2z = 3

(b)
2x + 5y + z = 2
x + y + 2z = 1
x + 5z = 3

2. Find the rank of the matrix


0 1
1 1 4 3
B 3 1 10 7 C
M =B
@ 4
C:
2 14 10 A
2 0 6 4

3. Write the vectors


~ = (1; 4; 5); B
A ~ = (5; 2; 1); C
~ = (2; 1; 3); D
~ = (3; 6; 11)

as a linear combination of the vectors


~a = (9; 0; 7) ; ~b = (0; 9; 13) :

4. In problem (a) and (b) show that the given functions are linearly inde-
pendent functions.

(a)
f1 (x) = sin (x) ; f2 (x) = cos (x)
(b)
f1 (x) = x; f2 (x) = ex ; f3 (x) = xex

5. Show that the matrix


p p !
1+i 3 p3 (1 +
4 2 2p
i)
M= p
p3 (1 + i) 3+i
2 2 4

is a unitary matrix. That means we need to show that


1
M = MT :
84 CHAPTER 4. MATRIX AND DETERMINANTS
Chapter 5

Introduction to di¤erential
calculus-I

In advanced undergraduate or graduate physics courses we often deal with dif-


ferential calculus. In this chapter we will get introduced to the basics in dif-
ferential calculus. These includes partial di¤erentiation and total di¤erential of
multivariable functions and its applications to real physical problems.

5.1 Partial di¤erentiation


Consider a well-de…ned function z(x; y) that de…nes the z coordinate of a point
in space. This function depends on two coordinates (x and y). This function is
di¤erentiable with respect to both x and y for all x and y. In the case where
z depends only one coordinates, for example z (x) ; the di¤erentiation of this
function with respect to x is given by

dz (x) z (x) z (x + x) z (x)


= lim = lim : (5.1)
dx x!0 x x!0 x

Thus for z(x; y) the partial di¤erentiation with respect to x or y; when x and
y are independent variables, is determined by keeping y or x constant and are
given by

@z (x; y) f (x; y) f (x + x; y) f (x; y)


= lim = lim ...y is kept constant
@x x!0 x x!0 x
(5.2)
and

@z (x; y) f (x; y) f (x; y + y) f (x; y)


= lim = lim ...x is kept constant.
@y y!0 y y!0 y
(5.3)

85
86 CHAPTER 5. INTRODUCTION TO DIFFERENTIAL CALCULUS-I

Generally for a function that depends on the independent set of variables fx1 ; x2 ; :::xn g ;
f (x1 ; x2 ; :::xn ) ; the partial di¤erentiation with respect to one of these variables,
xi is given by

@ f (x1 ; x2 ; :::xi + xi ::xn ) f (x1 ; x2 ; :::xi :::xn )


f (x1 ; x2 ; :::xn ) = lim
@xi xi !0 xi
...all x 6= xi are kept constant (5.4)

Example 1 Consider the function z(x; y) = x2 + y 2 , where x and y are inde-


pendent.

(a) Sketch this function to show that it de…nes a surface in a 3-D space described
by the Cartesian coordinates (x; y; z), where the z coordinate depends on
the x and y coordinates.

(b) Evaluate the following partial di¤erentiations at the point (x; y) = (1; 0):

@z @z @ 2 z @ @z @ 2 z @ @z @2z @ @z
; ; = ; 2 = ; = (5.5)
@x @y @x@y @x @y @x @x @x @y@x @y @x

Solution:

(a) The 3-D plot of this function obtained using Mathematica is shown below
5.1. PARTIAL DIFFERENTIATION 87

Using the function z(x; y) = x2 + y 2 ; we can evaluate the partial di¤erenti-


ations at point (x; y) = (1; 0) as:

@z @ @z
= x2 + y 2 = 2x ) = 2; (5.6a)
@x @x @x x=1;y=0
@z @ @z
= x2 + y 2 = 2y ) = 0; (5.6b)
@y @y @y x=1;y=0
@2z @ @ @ @2z
= x2 + y 2 = (2y) = 0 ) = 0;
(5.6c)
@x@y @x @y @x @x@y x=1;y=0
@2z @ @ @ @2z
= x2 + y 2 = (2x) = 2 ) = 2;(5.6d)
@x2 @x @x @x @x2 x=1;y=0
@2z @ @ @ @2z
= x2 + y 2 = (2x) = 0 ) = 0:
(5.6e)
@y@x @y @x @y @y@x x=1;y=0

Important notation: In some branches of physics, such as in thermodynamics


and statistical physics, a function may depend on set of variables which are not
necessarily independent as it may be related by one or more variables. However,
it may physically be important to …nd the partial di¤erentiation with respect
to one variable by keeping one or more of the other variables constant. In such
cases, the partial di¤erentiation is expressed using subscripts for the variables
that need to be kept constant. For example

@z
(5.7)
@r x

mean "the partial of z with respect to r, with x held constant" with z expressed
as a function of r and x.

Example 2a For z (x; y) = x2 + 2y 2 ; x = r cos ( ) ; and y = r sin ( ) ;…nd the


following partial derivative
@z
: (5.8)
@r x

Solution: First the function z (x; y) needs to be expressed as z (x; r) : To this


end, we note that
p
x2 + y 2 = r2 ) y = r2 x2 (5.9)

and
z (x; r) = x2 + 2y 2 = x2 + 2 r2 x2 = 2r2 x2 :
Then one can easily show that

@z @
= 2r2 x2 = 4r: (5.10)
@r x @r x
88 CHAPTER 5. INTRODUCTION TO DIFFERENTIAL CALCULUS-I

Example 2b An ideal gas is de…ned to be a gas that obeys the ideal gas equa-
tion of state
P V = nRT (5.11)
and
3
U= nRT (5.12)
2
where p is the pressure, V is the volume, n mass of the ideal gas mea-
sured in moles, T is the temperature, U is the internal energy, and R
the universal gas constant. Note that P; V; T; and U are the thermody-
namic variables. Obviously, as we can see from these two equations are
not independent set of variables. Find expressions for the following partial
di¤erentiations:

@P @P @P @2P @2P
; ; ; ; and : (5.13)
@V T @T V @V U @V 2 T @T 2 V

Solution: Using the ideal gas equations

3
P V = nRT; U = nRT (5.14)
2
we may write

nRT 3 2U nR 2U 2U
P = ; U = nRT; T = )P = = (5.15)
V 2 3nR V 3nR 3V
so that
@P @P (V; T ) @ nRT nRT
= = = ; (5.16a)
@V T @V T @V V V2
@P @P (V; T ) @ nRT nR
= = = ; (5.16b)
@T V @V V @T V V
@P @P (V; U ) @ 2U 2U
= = = ; (5.16c)
@V U @V U @V 3V 3V 2
@P @P (U; V ) @ 2U 2
= = = ; (5.16d)
@U V @U V @U 3V 3V

@2P @ 2 P (V; T ) @ @P (V; T )


= =
@V 2 T @V 2 T @V @V
@ nRT 2nRT
= 2
= ; (5.17)
@V V V3
@2P @ 2 P (V; T ) @ @P @ nR
= = = = 0:(5.18)
@T 2 V @T 2
V @T @T @T V
5.2. TOTAL DIFFERENTIAL 89

5.2 Total di¤erential


Consider a straight wire with length L that lies along the positive x-axis with one
end at the origin. Imagine you glued a charge that is not uniformly distributed
along the wire. That means the charge Q on the wire depends on the coordinate
x, (Q (x)). Thus the in…nitesimal charge Q (x) on the wire over the length in
between points x and x + x depends on the length ( x) and how the charge
changes over x ( Q(x) x ),

Q (x)
Q (x) = x: (5.19)
x
In the limit as the interval length, x; goes to zero, we note
dQ (x) Q (x)
= lim ; (5.20)
dx x!0 x
so that the the corresponding in…nitesimal charge dQ (x) becomes
dQ (x)
dQ = dx; (5.21)
dx
which is the di¤erential of the function Q (x) : Physically, this means the in…n-
itesimal charge on the wire in the interval dx depends on how the change Q
changes with respect to x; dQ(x)
dx :
Consider a mountain with hills and valleys. The elevation of the mountain
(z) depends on where you are located on the mountain (x; y) because of the hills
and valleys. If you hike on this mountain, what you will travel on the surface
of the mountain depends on what you traveled along the directions in x and y
and also the slope of the mountain along these directions. Thus an in…nitely
displacement on the mountain dz is given by the total di¤erential
@z (x; y) @z (x; y)
dz = dx + dy: (5.22)
@x @y
Example 3 Application of the total di¤ erential : The height and radius of a
right-circular cylinder are measured to be

H H = 7:6 0:2cm; R R = 3:7 0:2cm (5.23)

Find an approximate value for the uncertainty in the cylinder’s volume,


dV:
Solution: We recall that volume of a cylinder is given by

V (R; H) = R2 H (5.24)

so that the total di¤erential can be expressed as


@V @V
dV = dR + dH (5.25)
@R @H
90 CHAPTER 5. INTRODUCTION TO DIFFERENTIAL CALCULUS-I

replacing dV; dR; and dH by V; R; and H; respectively, we may ex-


press the uncertainty in volume as
@V @V
V = R+ H: (5.26)
@R @H
Evaluating
@V @
= R2 H = 2 RH (5.27a)
@R @R
@V @
= R2 H = R2 (5.27b)
@H @H
at H = 7:6 and R = 3:7; we have
@V
= 176:6 (5.28a)
@R H=7:6;R=3:7
@V
= 42:9 (5.28b)
@H H=7:6;R=3:7

so that using these results along with R = 0:2 and H = 0:2; the
uncertainty in the volume becomes

V = 176:6 0:2 + 42:9 0:2 = 44:76 (5.29)

Since uncertainty is expressed using one signi…cant …gure, we must write

V = 50 (5.30)

The volume is
V (R; H) = R2 H = 326:7 (5.31)
then we may write
V V = 330 50cm3 : (5.32)

5.3 The multivariable form of the Chain rule


Function of one variable: In a function of a single variable, y (x), the variable
x could be a function of another variable such as time, t; (x (t)). In such cases
to …nd the time dependence of the function y (x) ; we need to use the chain rule
for a function of one variable. We recall from the previous section, the total
di¤erential for y (x) can be written as
dy
dy = dx (5.33)
dx
so that diving this by dt, we …nd the Chain rule for a function of single variable
dy dy dx
= : (5.34)
dt dx dt
5.3. THE MULTIVARIABLE FORM OF THE CHAIN RULE 91

We often need to apply the chain rule in one, two, or three dimensions in physics.
Function of two variables: For a function of two variable, z (x; y) ; we recall
that the total di¤erential is given by
@z @z
dz = dx + dy: (5.35)
@x @y
If both x and y depend on another variable, like time, t, dividing this equation
by dt, we …nd the Chain rule for a function of two variables
dz @z dx @z dy
= + : (5.36)
dt @x dt @y dt
This can be generalized for function of multivariable.

Example 4 Consider the function

z = exp x2 y2 (5.37)

where
x(t) = et and y(t) = e t : (5.38)
Find
dz
: (5.39)
dt
Solution: Since x and y are a function of the same variable, t; we can write
dz @z dx @z dy
= + (5.40)
dt @x dt @y dt
Noting that
@z
= ( 2x) exp x2 y2
@x
@z
= ( 2y) exp x2 y2
@y
dx dy
= et = x; = et = y (5.41)
dt dt
we …nd
dz
= ( 2x) exp x2 y 2 x + ( 2y) exp x2 y 2 ( y)
dt
dz
) = 2 y 2 x2 exp x2 y 2 (5.42)
dt

Example 5 Show that the trajectory of the projectile motion of a ball with
mass m thrown from the ground (x0 = 0; y0 = 0) at an angle (measured
from the horizontal (+x-axis)) with a speed v0 is de…ned by equation of
hyperbola given by
y (x) = bx ax2 ; (5.43)
92 CHAPTER 5. INTRODUCTION TO DIFFERENTIAL CALCULUS-I

where
g
a= ; b = v0 tan ( ) : (5.44)
2v0 cos2 ( )
Note that x is the distance of the ball measured on the ground and y(x)
is the height of the ball from the ground.
Solution: From Newtons second law

Fx = max and Fy = may (5.45)

Since for a projectile motion there is no force on the x direction, Fx = 0;


we …nd
dvx dx
ax = = 0 ) vx (t) = v0x ) = v0x ) x (t) = x0 + v0x t
dt dt
) x (t) = x0 + v0 cos ( ) t: (5.46)

This shows that the x coordinate depends on time t. In the y direction


there is gravitational force, Fy = mg that leads to
dvy dy
ay = = g ) vy (t) = v0y gt ) = v0 sin ( ) gt: (5.47)
dt dt
Using the chain rule for a function of one variable, one can write
dy dy dx
= : (5.48)
dt dx dt
From the result in Eq. (5.46), we have
dx x (t) x0
= v0 cos ( ) and t = (5.49)
dt v0 cos ( )
so that substituting these equations along with the result in Eq. (5.47),
one may write Eq. (5.48) as
x (t) x0 dy
v0 sin ( ) g = v0 cos ( ) (5.50)
v0 cos ( ) dx
There follows that
dy g
= v0 tan ( ) (x x0 ) : (5.51)
dx v0 cos2 ( )
Upon integrating this equation with respect to x,
Z Z
g
dy = v0 tan ( ) (x x0 ) dx (5.52)
v0 cos2 ( )
we …nd
g x2
y (x) = y0 + v0 tan ( ) x x0 x : (5.53)
v0 cos2 ( ) 2
5.3. THE MULTIVARIABLE FORM OF THE CHAIN RULE 93

Assuming the initial position of the projectile is at the origin (x0 = 0; y0 = 0) ;


this can be put in the form

y (x) = bx ax2 ; (5.54)

where
g
a= ; b = v0 tan ( ) : (5.55)
2v0 cos2 ( )

Example 6 Implicit Di¤ erentiation and application of the total di¤ erential :
Consider the equation
y 3 x2 y = 8 (5.56)
where y is a function of x. Evaluate dy=dx and d2 y=dx2 at the point
(x; y) = (3; 1), where y = y (x; y)

Solution: The easy way to …nd both

dy d2 y
and (5.57)
dx dx2
is to apply implicit di¤erentiation. To this end, let

z (x; y) = y 3 x2 y = 8 (5.58)

so that
dz @z @z dy dz dy
= + =0) = 2xy + 3y 2 x2 =0
dx @x @y dx dx dx
dy 2xy
) = 2 (5.59)
dx 3y x2

Then at (x; y) = (3; 1); we …nd

dy
= 1: (5.60)
dx
To …nd the second derivative of the function we use the result for the …rst
derivative. Noting that

dy 2xy
= 2 = f (x; y) (5.61)
dx 3y x2
we can write
d2 y @ @ dy
2
= f (x; y) + f (x; y) : (5.62)
dx @x @y dx
Upon carrying out the partial di¤erentiations, we have

@ 2y 4x2 y
[f (x; y)] = 2 + 2 (5.63)
@x 3y x2 (3y 2 x2 )
94 CHAPTER 5. INTRODUCTION TO DIFFERENTIAL CALCULUS-I

and
@ 2x 12xy 2
[f (x; y)] = 2 2; (5.64)
@y 3y x2 (3y 2 x2 )
so that Eq. (5.62) becomes
!
d2 y 2y 4x2 y 2x 12xy 2 dy
= + 2 + 2 : (5.65)
dx2 3y 2 x2 (3y 2 x2 ) 3y 2 x2 (3y 2 x2 ) dx

This can be rewritten as


2
d2 y 1 2yx 1 2xy
= +
dx2 x 3y 2 x2 y 3y 2 x2
" #
2
1 2xy 3 2xy dy
+ (5.66)
y 3y 2 x2 x 3y 2 x2 dx

and using the result in Eq. (5.59), we …nd


" #
2 2
d2 y 1 dy 1 dy 1 dy 3 dy 2xy
= + + (5.67)
dx2 x dx y dx y dx x dx 3y 2 x2

Recalling the result we obtain for Eq. (5.59) at (x; y) = (3; 1)


dy
=1 (5.68)
dx
we …nd from Eq. (5.122)

d2 y 1 8
= 1 1 1= : (5.69)
dx2 3 3

5.4 Extremum (Max/Min) problems


For a function of single variable y (x), the …rst derivative
dy
f (x) = (5.70)
dx
tells us the slope of the tangent line at the point on the curve with the coordinate
(x; y (x)). The tangent line to an extremum point on a curve is a horizontal
line, the slope of which is zero. Therefore, if the x coordinates of the extremum
point on the curve de…ned by y (x) is xi ; since the slope of the tangent line at
(xi ; y (xi )) is zero,
dy
f (xi ) = = 0: (5.71)
dx x=xi

For a function of two variable, z (x; y) ; the extremum points with coordinates
(xi ; yi ; z (xi ; yi )) represent the coordinates for maxima on the hills or the minima
5.4. EXTREMUM (MAX/MIN) PROBLEMS 95

in the valleys on the surface de…ned by the function z (x; y) : The tangent plane
to these extremum points are horizontal planes (parallel to the x-y plane). For
this plane the slopes along the x and the y must be zero. Mathematically, these
slopes are de…ned by the partial derivative of the function z (x; y) evaluated at
the maxima or minima points with coordinates (xi ; yi ). Thus at the extremum
(maxima or minima) points

@f (x; y) @f (x; y)
= 0 and = 0: (5.72)
@x x=xi @y y=yi

These two equations de…nes the horizontal tangent plane at the extremum points
on the surface de…ned by z (x; y)

Example 7 An aquarium with rectangular sides and bottom has a …xed volume
V0 . Find its proportions so that it will require the least amount of glass
for construction.

Figure 5.1: The aquarium has length l, width w and height h:

Solution: As shown in Fig. 5.1, the length, width, and height of the aquarium
are l; w; and h, respectively. The the volume, which is …xed, is related to
l; w; and h by
V (l; w; h) = lwh = V0 : (5.73)
The aquarium has rectangular sides made of glass. The surface area,
A (l; w; h) of the …ve sides (excluding the top side which is open) is given
by
A (l; w; h) = lw + 2 (lh + wh) : (5.74)
We are interested in constructing an aquarium with least amount of glass.
It means the area of the glass sides that the aquarium is constructed from
must be minimized while the volume remains constant. As we can see
96 CHAPTER 5. INTRODUCTION TO DIFFERENTIAL CALCULUS-I

from Eq. (5.1) l; w; and h are not independent variables. We can reduce
the area from a function of three variable to a three variable by expressing
h in terms of l and w using Eq. (5.1) as
V0
h= : (5.75)
lw
Substituting Eq. (5.75) into Eq. (5.74) we can then express the area in
terms of two independent variables
1 1
A (l; w) = lw + 2V0 + : (5.76)
l w
For the minimum area we must have
@A (l; w)
=0 (5.77)
@l
and
@A (l; w)
=0 (5.78)
@w
so that using Eq. (??), one …nds
2V0
w =0 (5.79)
l2
and
2V0
l =0 (5.80)
w2
respectively. Now combining these two equations, we …nd
w4 w3 1=3
w 2V0 =0)w 1 = 0 ) w = 0; w = (2V0 ) (5.81)
4V02 2V0
But the acceptable solution is
1=3
w = (2V0 ) : (5.82)
Substituting combining this result and Eqs. (??) and (5.1) the correspond-
ing length and height of the aquarium are
2V0 2V0 1=3
l = = = (2V0 ) (5.83)
w2 (2V0 )
2=3

1=3
V0 V0
h = 1=3 1=3
= (5.84)
(2V0 ) (2V0 ) 4
Thus the proportions can be expressed as
l:w:h 1
1=3
=1:1:
(2V0 ) 8
These are the dimension of the Aquarium for least amount of glass for
construction and a volume V0 .
5.5. THE METHOD OF LAGRANGIAN MULTIPLIERS 97

5.5 The Method of Lagrangian Multipliers


In the Example 7 the volume of the rectangular aquarium, which is a function
of three variables, is speci…ed,

V (l; w; h) = lwh = V0 : (5.85)

That means the volume of the rectangular aquarium given can not be less or
greater than V0 : We were asked to determine the minimum total surface area,
A (l; w; h)
A (l; w; h) = lw + 2 (lh + wh) : (5.86)
This area appear to be a function three variables. But because of the condition
set for the volume, we have seen that it is actually a function of two variables.
Such kind of conditions are referred as constraints. The function that set the
volume to V0 in Eq. (5.85) is the constraint for the area de…ned by the function
in Eq. (??).
Suppose the function f (x; y; z; :::) describes some physical observable for
some system. In real physical problems the variables x; y; z; :::are subject to
constrains so that they are no longer independent. It is possible, at least in
principle, to use each constraint to eliminate one variable and to proceed with a
new and smaller set of independent variables. The use of Lagrangian multipliers
is an alternate technique that may be applied to determine extremum points for
the function when this elimination of variables is inconvenient or undesirable.
Suppose we are interested in …nding the extremum points for the function
f (x; y; z) under the constraint de…ned by the function (x; y; z) = k; where k is
a constant. We note that for the function to be extremum (including the saddle
point)

@f (x; y; z) @f (x; y; z) @f (x; y; z)


df = 0 ) dx + dy + dz = 0 (5.87)
@x @y @z
and the necessary and su¢ cient condition is

@f (x; y; z) @f (x; y; z) @f (x; y; z)


= = = 0: (5.88)
@x @y @z
Noting that for the constraint
@ @ @
d (x; y; z) = dk = 0 ) dx + dy + dz = 0 (5.89)
@x @y @z

and for a function F (x; y; z; :::) = f (x; y; z) + (x; y; z); for a constant, at
the extremum points we may write

F (x; y; z) = f (x; y; z) + (x; y; z) = f (x; y; z) + k ) dF = df = 0


@f @ @f @ @f @
) + dx + + dy + + dz = 0
@x @x @y @y @z @z
98 CHAPTER 5. INTRODUCTION TO DIFFERENTIAL CALCULUS-I

There follows that


@f @ @f @ @f @
+ = 0; + = 0; and + = 0: (5.90)
@x @x @y @y @z @z
Here we want x; y; and z so need not be determined. This method would fail
if the coe¢ cient of vanish at the extremum,
@ @ @
= = = 0:
@x @y @z
The following steps can be used to …nd extremum points for a function f (x; y; z; :::)
with a constraint (x; y; z; :::) = k
(1) Construct the function F (x; y; z; :::) = f (x; y; z; :::) + (x; y; z; :::), where
is the (unknown) Lagrangian multiplier.
(2) Compute the partial derivatives of F with respect to each of the variables
x; y; z; ::: and set them equal to zero.
(3) Solve the resulting system of simultaneous equations generated in Step 2
along with the constraint equation
(x; y; z; :::) = constant

Example 8 Consider a wire bent to …t a curve de…ned by y = 1 x2 as shown


in the …gure below.

Suppose a string is stretched from the origin to a point (x; y) on the curve.
Find the minimum length of the string.
Solution: We need to …nd the minimum value for the distance
p
d (x; y) = x2 + y 2 : (5.91)
with the constraint
y (x) = 1 x2 ) (x; y) = y + x2 = 1: (5.92)
Following the procedure for the Method of Lagrangian multipliers, we have
5.5. THE METHOD OF LAGRANGIAN MULTIPLIERS 99
p
(1) Construct the function F (x; y) = f (x; y)+ (x; y) = x2 + y 2 + y + x2
(2) Compute the partial derivatives of F with respect to each of the variables
x; y and set them equal to zero,
@ x 1
F (x; y) = 0) p +2 x=0) p = 2 (5.93)
@x x2 + y2 x2 + y2
@ y y
F (x; y) = 0) p + =0) p = (5.94)
@y x2 + y 2 x2 + y 2

(3) Solve the resulting system of simultaneous equations generated in Step 2


along with the constraint equation

(x; y) = constant

to determine the coordinates (x; y) at which the function f (x; y) takes a


maxima or minima values:
1 y
p = 2 ;p = ; y + x2 = 1 (5.95)
x2 + y2 x2 + y2
These three equations can be combined to …nd the values for x and y
1 2y
p = p ;y = 1 x2 ) 1 = 2 1 x2 (5.96)
x2 + y2 x + y2
2

1 1
) x = p ;y = (5.97)
2 2
Then the minimum distance is found to be
p
dmin = 3=2: (5.98)

Example 8 (a) The aquarium problem : An aquarium with rectangular sides


and bottom has a …xed volume V0 . Find its proportions so that it will
require the least amount of glass for construction.
Solution: In order to determine the minimum area of the glasses used using
the method of Lagrangian multiplies we need to write the constraint. The
constraint is the volume

V0 = lwh ) (l; w; h) = lwh = V0 = constant. (5.99)

The surface area of the aquarium is

A (l; w; h) = lw + 2 (wh + lh) ) f (l; w; h) = lw + 2 (wh + lh) : (5.100)

Constructing the function F (l; w; h);

F (l; w; h) = f (l; w; h)+ (l; w; h) ) F (l; w; h) = lw +2 (wh + lh)+ lwh:


(5.101)
100 CHAPTER 5. INTRODUCTION TO DIFFERENTIAL CALCULUS-I

We then minimize this function by partially di¤erentiation it with respect


to l; w; and h

@F (l; w; h)
= 0 ) w + 2h + wh = 0; (5.102)
@l
@F (l; w; h)
= 0 ) l + 2h + lh = 0; (5.103)
@w
and
@F (l; w; h)
= 0 ) 2w + 2l + lw = 0: (5.104)
@h
We solve the above three equations along with the constraint

lwh = V0 : (5.105)

to determine l; w; and h: Since we do not need the solution for ; from


Eq. (5.102) we can express as

w + 2h
= (5.106)
wh
and substitute it into Eqs. (5.103) and (5.104) we …nd

w + 2h w + 2h
l + 2h l=0)l 1 = 2h ) w = l (5.107)
w w

and
w + 2h w + 2h
2w + 2l lw = 0 ) l 2 = 2w ) l = 2h (5.108)
wh h

Now substituting these results in Eq. (5.105), we …nd


1=3
V0 1=3
4h3 = V0 ) h = ) w = l = 2h = (2V0 ) (5.109)
4

Example 8 (b) Particle in a box-Quantum mechanics:Consider the quantum


mechanical problem of a particle with mass m in a box. The box is a
rectangular box with length l; width w; and height h: The ground state
energy energy is given by

2 ~ 1 1 1
E (l; w; h) = + 2+ 2 :
8m l2 w h

Find the proportions to sides of the box that minimize the ground state
energy subject to the constraint that the volume is a constant

V (l; w; h) = lwh = V0
5.6. CHANGE OF VARIABLES 101

Solution: Constructing the function F (l; w; h);


1 ~ 1 1
F (l; w; h) = f (l; w; h)+ (l; w; h) ) F (l; w; h) = 2
+ 2 + 2 + lwh:
l 4m w h
(5.110)
We then minimize this function by partially di¤erentiation it with respect
to l; w; and h
@F (l; w; h) ~
=0) + wh = 0; (5.111)
@l 2ml3
@F (l; w; h) ~
=0) + lh = 0; (5.112)
@w 2mw3
and
@F (l; w; h) ~
=0) + lw = 0: (5.113)
@h 2mh3
We solve the above three equations along with the constraint
lwh = V0 : (5.114)
to determine l; w; and h: Since we do not need the solution for ; using
the …rst equation, we have
~
= (5.115)
2ml3 wh
so that using this expression for ; one can write
~ ~ ~
3
+ lh = 0 ) 3
= )w=l (5.116)
2mw 2mw 2ml2 w
and
~ ~ ~
3
+ lw = 0 ) 3
= )h=l (5.117)
2mh 2mh 2ml2 h
Now substituting these results into the constraint for the volume, we …nd
1=3 1=3
l3 = V0 ) l = (V0 ) ) w = l = h = (V0 ) (5.118)
The box must be a cube, w : l : h = 1 in order to minimize the energy of
the particle.

5.6 Change of Variables


Physical problems in most branches of physics requires solving di¤erential equa-
tions. These di¤erential equations could be a function of single or multivariable.
In order to …nd the solutions to these di¤erential equations change of variables
sometimes is necessary for di¤erent reasons. In order to change the di¤erential
equation from one variable to another when the di¤erential equation is a func-
tion of multivariable, we need to apply partial di¤erentiation of the function. In
the next examples we illustrates application of partial di¤erentiation in change
of variables in a second order di¤erential equation for single and multivariable
function.
102 CHAPTER 5. INTRODUCTION TO DIFFERENTIAL CALCULUS-I

Example 9 The Bessel di¤ erential equation is given by:


d2 y dy
x2 2
+x (1 x) y = 0 (5.119)
dx dx
In this equation y = y(x). Find the corresponding equation with the
following change of variable:
p
u = 2 x; y = y (u) (5.120)

Solution: Using partial di¤erentiation, we can express


dy @y du
= (5.121)
dx @u dx
since y = y (u) that depends only in one variable u, we may rewrite
dy dy du dy 1
= = p : (5.122)
dx du dx du x
Using the result above the second derivative in the bessel equation can be
expressed as
d2 y d dy 1 1 d dy dy d 1
= p =p + p : (5.123)
dx2 dx du x x dx du du dx x
Noting that
d dy d dy du d dy d2 y 1
= ) = 2p (5.124)
dx du du du dx dx du du x
and
d 1 1
p = p (5.125)
dx x 2x x
we …nd
d2 y 1 d dy dy d 1 1 d2 y 1 dy
2
=p + p = p : (5.126)
dx x dx du du dx x x du2 2x x du
Then substituting Eqs. (5.122) and (5.126) into Eq. (5.119) the Bessel
equation takes the form
1 d2 y 1 dy dy 1
x2 p +x p (1 x) y = 0 (5.127)
x du2 2x x du du x
which can simpli…ed into
p
d2 y x dy
x 2
+ (1 x) y = 0 (5.128)
du 2 du
From Eq. (5.120) we have x = u2 =4 and the Bessel equation becomes
u2 d2 y u dy u2
+ 1 y=0
4 du2 4 du 4
d2 y dy
) u2 2 + u + u2 4 y = 0: (5.129)
du du
5.6. CHANGE OF VARIABLES 103

In classical or quantum mechanical studies of the electrical or magnetic prop-


erties of particles or systems in physics often involve solving a second order
partial di¤erential equation for some form of multivariable a scalar or vector
function. One of these that commonly used is the Laplace equation. The
Laplace equation in Cartesian coordinates for some scalar function V (x; y; z)
is given by
@ 2 V (x; y; z) @ 2 V (x; y; z) @ 2 V (x; y; z)
+ + = 0:
@x2 @y 2 @z 2
In order to determine the solution to such kind of second order partial di¤er-
ential equations it may be necessary to change the variables from Cartesian
to cylindrical (r; '; z) or to spherical (r; ; ') coordinates if the system has a
cylindrical or spherical symmetry, respectively. Such change of variables for the
given partial di¤erential equation are determined by the equations that relates
a point in space de…ned by the Cartesian coordinates to Cylindrical coordinates

x = r cos ; y = r sin ; z = z (5.130)

and to spherical coordinates

x = r cos ' sin ; y = r sin ' sin ; z = r cos . (5.131)

In the next example we will see how we make change of variables for the two
dimensional Laplace equation in Cartesian to Cylindrical coordinates.

Example 10 In classical electrodynamics in a two-dimensional space where


there is no free charges the electrical potential, V (x; y) ; satis…es the two-
dimensional Laplace Equation given by
@ 2 V (x; y) @ 2 V (x; y)
+ =0 (5.132)
@x2 @y 2
Suppose the two dimensional space is a dielectric disk of radius R and you
are given a boundary condition for the potential at the edge of the disk is
zero V (x; y) = 0; for x2 + y 2 = R2 . For such kind of problem it is much
easier if we solve the Laplace equation in polar coordinates. Transform
the Laplace equation into polar coordinates using

x = r cos ( ) :y = r sin ( ) ; V = V (r; ) : (5.133)

Solution: There are two di¤erent way of transforming this equation. I will use
the short way. We note that
@V (r; ) @V (x; y) @x @V (x; y) @y
= + (5.134)
@r @x @r @y @r
so that using expressions for x and y in Eq. (5.133), one …nds
@V (r; ) @V (x; y) @V (x; y)
= cos ( ) + sin ( ) : (5.135)
@r @x @y
104 CHAPTER 5. INTRODUCTION TO DIFFERENTIAL CALCULUS-I

Similarly, for partial di¤erentiation with respect to ; we have

@V (r; ) @V (x; y) @x @V (x; y) @y


= +
@ @x @ @y @
@V (x; y) @V (x; y)
= r sin ( ) + r cos ( ) (5.136)
@x @y

dividing this equation by r; one …nds

1 @V (r; ) @V (x; y) @V (x; y)


= sin ( ) + cos ( ) : (5.137)
r @ @x @y

Now multiplying Eq. (5.135) by cos ( ) and Eq. (5.137) by sin ( ) ; we


…nd
@V (r; ) @V (x; y) @V (x; y)
cos ( ) = cos2 ( ) + sin ( ) cos ( ) (5.138)
@r @x @y

and
sin ( ) @V (r; ) @V (x; y) @V (x; y)
= sin2 ( ) + sin ( ) cos ( ) : (5.139)
r @ @x @y

Subtracting Eq. (5.139) from Eq. (5.138) one …nds

@V (x; y) @V (r; ) sin ( ) @V (r; )


= cos ( ) : (5.140)
@x @r r @
Similarly,multiplying Eq. (5.135) by sin ( ) and Eq. (5.137) by cos ( ) ;
we have
@V (r; ) @V (x; y) @V (x; y)
sin ( ) = sin ( ) cos ( ) + sin2 ( ) (5.141)
@r @x @y

and
cos ( ) @V (r; ) @V (x; y) @V (x; y)
= sin ( ) cos ( ) + cos2 ( ) : (5.142)
r @ @x @y

and adding these two equations, we …nd

@V (x; y) @V (r; ) cos ( ) @V (r; )


= sin ( ) + : (5.143)
@y @r r @

From Eqs. (5.140) and Eq. (5.143) follows that

@ @ cos ( ) @
= sin ( ) + ; (5.144)
@y @r r @
@ @ sin ( ) @
= cos ( ) (5.145)
@x @r r @
5.6. CHANGE OF VARIABLES 105

so that one can write

@2 @ cos ( ) @ @ cos ( ) @
= sin ( ) + sin ( ) + (5.146)
@y 2 @r r @ @r r @

and

@2 @ sin ( ) @ @ sin ( ) @
= cos ( ) cos ( ) : (5.147)
@x2 @r r @ @r r @

On carrying out the partial di¤erentiation in the …rst square bracket on


the second square bracket in Eq. (5.146) we have

@2 @2 @ 1 @ cos ( ) @ @
= sin2 ( ) + sin ( ) cos ( ) + sin ( )
@y 2 @r2 @r r @ r @ @r
cos ( ) @ @
+ cos ( ) (5.148)
r2 @ @

so that upon performing the di¤erentiations in the square brackets, one


…nds

@2 @2 sin ( ) cos ( ) @ 2 sin ( ) cos ( ) @


2
= sin2 ( ) 2 +
@y @r r @r@ r2 @
2 2
cos ( ) @ sin ( ) cos ( ) @ sin ( ) cos ( ) @ cos ( ) @ 2
2
+ + + :
r @r r @ @r r 2 @ r2 @ 2
(5.149)

that simpli…es into

@2 @2 2 sin ( ) cos ( ) @ 2 2 sin ( ) cos ( ) @


= sin2 ( ) +
@y 2 @r 2 r @r@ r2 @
2 2 2
cos ( ) @ cos ( ) @
+ + : (5.150)
r @r r2 @ 2

Following a similar procedure in Eq. (5.147), we have

@2 @2 @ 1 @
= cos2 ( ) sin ( ) cos ( )
@x2 @r2 @r r @
sin ( ) @ @ sin ( ) @ @
cos ( ) + sin ( ) (5.151)
r @ @r r2 @ @

that leads to

@2 @2 2 sin ( ) cos ( ) @ 2 sin ( ) cos ( ) @ 2


= cos2 ( ) +
@x2 @r 2 r2 @ r @r@
2 2
sin ( ) @ sin ( ) @ 2
+ + : (5.152)
r @r r2 @ 2
106 CHAPTER 5. INTRODUCTION TO DIFFERENTIAL CALCULUS-I

Now adding Eqs. (5.150) and (5.152), we …nd

@2 @2 @2 1 @ 1 @2
+ = + + (5.153)
@x2 @y 2 @r2 r @r r2 @ 2
which can be rewritten as
@2 @2 1 @ @ 1 @2
+ 2 = r + (5.154)
@x2 @y r @r @r r2 @ 2
Therefore the Laplace equation in polar coordinates can be expressed as

1 @ @V (r; ) 1 @ 2 V (r; )
r + =0 (5.155)
r @r @r r2 @ 2

5.7 Legendre Transformations


Legendre Transformation of a function f (x) denoted by g (p) is given by

g (p) = xp f (x) (5.156)

where
df
p=
: (5.157)
dx
The Legendre transformation in Eq. (5.156) can also be expressed using Eq.
(5.157) as
df df
g =x f (x) (5.158)
dx dx
It is commonly used in thermodynamics and in the Hamiltonian formulation
of classical mechanics. Here we will consider an example in thermodynamics.
As mentioned previously, in thermodynamics we often have many variables to
choose from in order to specify the state of a given system, but only a small
number of these variables are actually independent variables. So-called ther-
modynamic potentials are functions that result from a transformation of the
internal energy function U from one set of variables to another.
Terminology and Notation:

P = pressure; V = volume; T = temperature(K); S = entropy


W = work done on the system; Q = heat energy into the system;

The First Law of Thermodynamics:

dU = dQ + dW = T dS P dV (5.159)

Internal Energy and the Thermodynamic Potentials:

U = Internal Energy; U = U (S; V ); H = Enthalpy, H = H(S; P );


F = Helmholtz F ree Energy; F = F (T; V )
5.8. HOMEWORK ASSIGNMENT 10 107

Example 11 Demonstrate a Legendre Transformation on the First Law of


Thermodynamics by deriving the form of the thermodynamic potential
Helmholtz Free Energy such that F = F (T; V ).

Solution: Let the internal energy, U , is a function of entropy, S; and volume,


V: Then the total di¤erential for the internal energy can be written as

@U @U
dU (S; V ) = dS + dV: (5.160)
@S V @V S

Comparing this equation with the …rst low of thermodynamics

dU = T dS P dV (5.161)

we …nd
@U
= T; (5.162)
@S V
@U
= P: (5.163)
@V S

The Helmholtz free energy which is a function of T and V , can then be


written as
@U @U
F = F (T; V ) = F ;V =F ; (5.164)
@S V @S V

which is the Legendre Transformation of the internal energy U

@U @U
F =S U: (5.165)
@S V @S V

If we substitute
@U
=T (5.166)
@S V

we …nd
F = F (T; V ) = ST U (5.167)

5.8 Homework Assignment 10


1. For the function p
z (u; v; w) = ln u2 + v 2 + w2
…nd the following partial derivatives:

@z @z @z
; ; and :
@u @v @w
108 CHAPTER 5. INTRODUCTION TO DIFFERENTIAL CALCULUS-I

2. For the function


u (x; y) = ex cos (y)
verify that
(a)
@2u @2u
=
@x@y @y@x
(b)
@2u @2u
+ 2 = 0:
@x2 @y
If
z (x; y) = x2 + 2y 2 ; x = r cos ( ) ; y = r sin ( )
…nd the following partial derivatives
3.
@z
@ x
4.
@z
@y@
5. Use di¤erential to show that for large n and small a
p p a
n+a n' p :
2 n

5.9 Homework Assignment 11


1. The acceleration of gravity can be found from the length l and period T
of a pendulum; the formula is g = 4 2 l=T 2 : Find the relative error in g in
the worst case if the relative error in l is 5%, and the relative error in T
is 2%:
2. For an idea gas of N molecules, the number of molecules with speeds v
is given by the formula
Z
4a3 N v 2 a2 x2
n (v) = p x e dx
0

where a is a constant and N is the total number of molecules. If N = 1026 ,


estimate the number of molecules with speeds between v = 1=a and 1:01=a
y dz
3. Given z = xe ; x = cosh (t) ; y = cos (t) ; …nd dt :

4. If xy 3 yx3 = 6 is the equation of the curve, …nd the slope and the
equation of the tangent line at the point (1; 2) : Computer plot the curve
and the tangent line on the same axis.
5. In problem 4 …nd d2 y=dx2 at (1; 2).
Chapter 6

Introduction to Di¤erential
Calculus II

In the previous chapter we got introduced to the basics in di¤erential calculus


and its applications. Generally, we introduced to how to determine the di¤er-
entiation of functions of single and multivariable variables and its application
maximization or minimization in physical problems. In this chapter, we will
introduce to a sort of the reverse process in di¤erential calculus and its applica-
tion in real physical problems. We will learn the di¤erent methods in di¤erential
calculus used to determine a single or multivariable variable functions satisfying
a di¤ erential equation. A di¤erential equation (DE) is an equation involving
derivatives of functions with respect to one or more independent variables. In
real physical problems we could …nd linear or none linear di¤erential equations
for scalar (e.g. an electrical potential) or vector (e.g. an electric …eld) func-
tions. In real physical problems we could …nd linear or none linear di¤erential
equations for scalar (e.g. an electrical potential) or vector (e.g. an electric
…eld) functions. In this chapter we will consider primarily focus only in linear
di¤erential equations.

6.1 Linear ordinary di¤erential equations

As we just stated a di¤erential equation is an equation involving derivatives of


functions with respect to one or more independent variables. If there is only
one independent variable, then the equation is said to be an ordinary di¤ erential
equation (ODE). The order of the ODE or any DE is the order of the highest-
order derivative in the DE. An nth order ODE for the function y (x) (with x as

109
110 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

independent and y as dependent variable) has the generic form

dy d2 y d3 y d4 y
a0 (x) y + a1 (x) + a2 (x) 2 + a3 (x) 3 + a4 (x) 4
dx dx dx dx
dn y
+:::an (x) n = f (x) (6.1)
dx
or, more concisely,
n
X di y
ai (x) = f (x) : (6.2)
i=0
dxi
Note that in ODE ai (x) are generally function of the independent variable x.
When ai (x) =Constant for all i, the ODE is known as linear ordinary di¤erential
equations (LODE).
In this section we learn how can determine the solutions to LODEs. By a
solution to a DE we mean a function which, when substituted back into the
DE, yields an identity. The general solution to an nth -order LDE contains n
independent, arbitrary constants of integration. A particular solution to a DE
is the general solution evaluated for particular values of its arbitrary constants.
These values for the constants are determined by applying boundary conditions
(BC’s) that pertain to the DE, ‘usually obtained from an analysis of the physical
system being modeled by the DE.

6.1.1 First-order LODE with constant coe¢ cients


First-order LODE with constant coe¢ cients, generally, has the form
dy (x)
a0 y (x) + a1 = f (x) (6.3)
dx
The solutions to …rst order LODE can generally be determined by a method of
integration. For a1 6= 0; one can put Eq. (6.3) in the form

a0 dy (x) 1
y (x) + = f (x) : (6.4)
a1 dx a1
Noting that
a0
x
a0 a0
x a1 d h aa0 x i a0 a0 d h aa0 x i
e a1
e a1 x = 1 ) e a1
e 1 =1) =e a1 x
e 1 (6.5)
a0 dx a1 dx
Eq. (6.3) can be rewritten as
a0
x d h aa0 x i dy (x) 1
y (x) e a1
e 1 + = f (x) : (6.6)
dx dx a1
a0
Multiplying both sides of this equation by e a1 t ; we …nd
d h aa0 x i a0 dy (x) 1 a0
y (x) e 1 + e a1 x = f (x) e a1 t (6.7)
dx dx a1
6.1. LINEAR ORDINARY DIFFERENTIAL EQUATIONS 111

which can be put in the form


d h a0 i 1 a0
y (x) e a1 x = f (x) e a1 x : (6.8)
dx a1
Integrating Eq. (6.8), we may write,
Z x Z x
d h a0 0 i
a1 x
1 a0 0

0
y (x 0
) e dx 0
= f (x0 ) e a1 x dx0 : (6.9)
x0 dx a1 x0

which leads to
Z x
a0 a0 1 a0
y (x) e a1 x y (x0 ) e a1 x0 = f (x) e a1 x dx (6.10)
a1 x0
Z x
a0 a0 1 a0 0
) y (x) e a1 x = y (x0 ) e a1 x0 + f (x0 ) e a1 x dx0
a1 x0

In this equation a0 are a1 are constants in the DE and x0 and y (x0 ) are also
constants de…ned by the boundary conditions of the function y (x) that pertain
to the DE. It usually obtained from an analysis of the physical system being
modeled by the DE. We also note that the de…nite integral can be expressed as
Z x Z
1 a0 0 1 a0
f (x0 ) e a1 x dx0 = f (x) e a1 x dx C (x0 ) ;
a1 x0 a1

where we have replaced the result of the integral when it is evaluated at x0 by


a constant C (x0 ) : Thus Eq. (6.10) can be expressed as
Z
a0
a1 x
a0
a1 x0
1 a0
y (x) e = y (x0 ) e C (x0 ) + f (x) e a1 x dx: (6.11)
a1
Thus by introducing another constant C for
a0
C = y (x0 ) e a1 x0 C (x0 ) ; (6.12)

the general solution to the …rst-order LODE in Eq. (6.3) is given by


a0
x Z
a0
x e a1 a0
y (x) = Ce a1
+ f (x) e a1 x dx: (6.13)
a1

In the next example we will see how we apply Eq. (6.13) to solve real physical
problems described by a …rst-order LODE.

Example 1 An object is dropped from rest and has a vertically downward


acceleration whose magnitude is given by a = g exp ( kt), where k is a
(real) positive constant and g is the acceleration due to gravity at the
earth’s surface.
(a) Find the general solution for the distance fallen as a function of time.
112 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

(b) Evaluate the general solution at the appropriate BCs in order to …nd the
particular solution for the distance fallen by the object described above
as a function of time.

(c) Comment on limiting behavior of your answer, and physically describe the
system being modeled.

Solution:

(a) we recall that the acceleration, a, and the velocity, v are related by the
equation
dv
a= (6.14)
dt
and using the expression for the given acceleration, we …nd

dv
= g exp ( kt) : (6.15)
dt
In view of Eqs. (6.3) and (6.13), for a LODE of the form

dv (t)
a0 v (t) + a1 = f (t) (6.16)
dt
the solution can be expressed as
a0
t Z t
a0
t e a1 a0 0
v (t) = C1 e a1
+ f (t0 ) e a1 t dt0 ; (6.17)
a1 t0

From Eqs. (6.15) and (??), we note that

a0 = 0; a1 = 1; f (t) = g exp ( kt)

and the solution becomes


Z t
v (t) = C1 + g exp ( kt0 ) dt0 ; (6.18)
0

where we set the initial time t0 = 0: Upon carrying out the integration,
we …nd
g
v (t) = C1 + (1 exp ( kt)) (6.19)
k
To …nd the position we note that the velocity is related to the displacement
by
dy dy g
v (t) = ) = C1 + (1 exp ( kt)) (6.20)
dt dt k
Noting that this equation can be put

dy (t)
b0 y (t) + b1 = f (t) (6.21)
dt
6.1. LINEAR ORDINARY DIFFERENTIAL EQUATIONS 113

where
g g
b0 = 0; b1 = 1; f (t) = C1 + exp ( kt)) (6.22)
k k
Applying Eq. (6.17) we can write the solution to Eq. (6.17) as
b0
t Z t
b0
t e b1 b0 0
y (t) = C2 e b1
+ f (t0 ) e b1 t dt0 ; (6.23)
b1 t0

so that substituting the values in Eq. (6.22), we have


Z th i
g g
y (t) = C2 + C1 + exp ( kt0 )) dt0 ; (6.24)
t0 k k

and upon carrying out the integration, one …nds


g g
y (t) = C2 + C1 + t + 2 [exp ( kt) 1] (6.25)
k k

(b) We are given the initial conditions (boundary conditions) (v (t = 0) = 0)


and assume that the object is dropped from an initial position (y (t = 0) = y0 ) :
Then evaluating Eqs. (6.19) and (6.25) at t = 0, we …nd

C1 = 0; C2 = y0

and the distance fallen as function of time becomes


g g
d (t) = y0 y (t) = t + 2 [1 exp ( kt)] (6.26)
k k
and the speed
g
v (t) = (1 exp ( kt)) (6.27)
k
(c) For small t (using Taylor series expansion), we have
1 1 2
exp ( kt) ' 1 kt + kt2 ) 1 exp ( kt) ' kt kt (6.28)
2 2
so that
g g 1 2 1
d (t) ' t + 2 kt kt ) d (t) ' gt2 : (6.29)
k k 2 2
and the speed
g 1 2
v (t) '
(kt kt ) ) v (t) ' gt (6.30)
k 2
where we have dropped the higher order in t. For large time; the velocity
g
v (t) = (1 exp ( kt)) (6.31)
k
becomes
g
v (t) ' = v0 : (6.32)
k
114 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

For the distance


g g
d (t) = y0 y (t) = t + 2 [1 exp ( kt)] ; (6.33)
k k
if we let the large time be t = t0 ; we …nd
g
d (t) = d (t) = y0 y (t) = v0 t0 + : (6.34)
k2

Example 2 A conducting bar illustrated in the …gure below moves on two


frictionless parallel semi-in…nite rails connected by a conducting wire of
length l in the presence of uniform magnetic …eld directed into the page.
The bar has mass m and its length is also l. The bar is given an initial
velocity, v0 ; at t = 0 directed to the right.

(a) Using the di¤erential form of Faraday’s Law of induction …nd the current
induced in the circuit assuming that the resistance of the circuit is R.

(b) Find the magnetic force on the conducting bar.

(c) Using Newton’s second law …nd the speed of the bar as function of time.

(d) Show that the energy dissipated in the circuit (Joule Heat energy) is equal
to the initial kinetic energy of the bar.

Solution:

(a) According to Faraday’s law a changing magnetic ‡ux, ; induces voltage,


Vin given by
d
Vin = : (6.35)
dt
Over a time interval t = t and t = t+dt, the position of the conducting rod
changes from x0 to x0 +dx, as the as it slides along the positive x-direction
with a speed v (t). This distance can expressed as

dx = vv (t) dt: (6.36)


6.1. LINEAR ORDINARY DIFFERENTIAL EQUATIONS 115

In the process the area bounded by part of the rails, the wire, and the
rod increases from an area A = xl at t = t; to an area, A = xl + ldx: The
increase in area, dA; will then be

dA = ldx = lvdt: (6.37)

The magnetic ‡ux, ; for a uniform magnetic …led B directed perpendic-


ular to the area, A; bounded a closed conducting wire can be expressed
as
= BA: (6.38)
Since the change in the magnetic ‡ux with time is due to the change in
area, we can write
d = BdA: (6.39)
so that using Eq. (6.37) one …nds for the change ‡ux over the time interval
dt
d = Blvdt: (6.40)
The induced voltage due to the change in magnetic ‡ux with time becomes
d
Vin = = Blv: (6.41)
dt
Applying Ohm’s law the resulting induced current, Iin can be expressed
as
Vin Blv
Iin = = : (6.42)
R R
According to Lenz’s law the induced current must oppose the cause for
the induction of the current. The cause is increase in magnetic ‡ux. To
oppose this increase, the induced current must ‡ow in a counterclockwise
direction.

(b) The bar experiences a magnetic force as a result of the induced current and
the uniform magnetic …eld. This magnitude of this force is given by

l2 vB 2
Fm = Iin lB = (6.43)
R
The direction is opposite to the direction of the velocity as one easily
determine using the right-hand-rule.
(c) Using Newtons second law

dv l2 vB 2 l2 B 2 dv
m = Fm = ) v+ = 0: (6.44)
dt R mR dt
that can be put in the form

dv (t)
a0 v (t) + a1 = f (t) (6.45)
dt
116 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

where
l2 B 2
a0 = ; a1 = 1; f (t) = 0
mR
We recall the solution to Eq. (6.45) is given by
a0
t Z t
a0
t e a1 a0 0
v (t) = C1 e a1
+ f (t0 ) e a1 t dt0 ; (6.46)
a1 t0

which gives
l2 B 2 t
a0
t
v (t) = C1 e a1
= C1 e mR (6.47)
Noting that initial velocity,v (t = 0) = v0 ; the velocity of the rod becomes
l2 B 2 t
v(t) = v0 e mR : (6.48)

(d) The energy delivered to the resistor per unit time (power) is given by
dW 2
P = = Iin (t) R (6.49)
dt
Using the results we found for the induced current in Eq. (6.42) along
with the velocity in Eq. (6.48), Eq. (6.42) can be put in the form
2l2 B 2 t
dW l2 v 2 (t) B 2 B 2 l2 v02
= R = e mR (6.50)
dt R2 R
or
dW
= f (t) : (6.51)
dt
Then the total energy delivered to the resistor is

Z1 2l2 B 2 t
B 2 l2 v02 B 2 l2 v02 mR
W = e mR dt ) W =
R R 2l2 B 2
0
1
) W = mvo2 (6.52)
2
Example 3 Discuss the variation in the intensity of radiation as it passes
through a stellar atmosphere assuming a plane-parallel model for the at-
mosphere.
Starting Point: Consider a di¤erential slab of stellar atmosphere of thickness
dz at the position z above the reference position z = 0. Let (z) be the
density of the atmosphere at that position, and let I(z) be the intensity
of radiation incident on the slab. Let Io be the intensity of radiation at
the reference position z = 0 within the atmosphere. Find an expression
for I(z) for z > 0.
6.2. THE FIRST-ORDER DE AND EXACT DIFFERENTIAL 117

Solution: The change in intensity due to absorption is related to the density


of the atmosphere by
dI (z) = K (z)dz (6.53)

where K is a constant of proportionality.


Z I(z) Z z Z z
dI (z) = K (z)dz ) I (z) = I0 K (z)dz (6.54)
I0 0 0

6.2 The …rst-order DE and exact di¤erential


In the previous section, we have seen how the solution to a …rst-order LODE

dy (x)
a0 (x) y (x) + a1 (x) = f (x) (6.55)
dx
can be determined when the coe¢ cients are constant. Generally, a …rst-order
DE could be ODE where the coe¢ cients could be function of one variable (only
x) as described by Eq. (6.55). It could also be a DE in which the coe¢ cients
are a function of two variables (x and y) that has the form

dy (x)
Q (x; y) = P (x; y) : (6.56)
dx
In this section we will develop the method to solve these two type of …rst-order
DEs. But we …rst consider the later case. To this end, we note that Eq. (6.56)
can be rewritten as

dF (x; y) = P (x; y) dx + Q (x; y) dy = 0; (6.57)

which is the total di¤erential of some constant function, F (x; y) =constant,


that depends on both x and y if

@F (x; y) @F (x; y)
P (x; y) = ; Q (x; y) = : (6.58)
@x @y

This indicates if one can determine this constant function from the total di¤er-
ential obtained from the …rst-order DE in Eq. (6.56), one can solve the DE by
…nding y from the function. This depends on certain conditions that the total
deferential needs to satisfy that we will discuss. But …rst we will see how to
determine the total di¤erential, generally, for any function F (x; y) ; that is not
necessarily a constant function.

Example 4 Find dF for the function

F (x; y) = 3x2 + 2x sin (2y) : (6.59)


118 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

Solution: The total di¤erential, dF; for the function of two variables, F (x; y) ;
is given by
dF = P (x; y) dx + Q (x; y) dy: (6.60)
For the given function in Eq. (6.59), we have
@F @
P (x; y) = = 3x2 + 2x sin (2y) = 6x + 2 sin (2y) (6.61)
@x @x
and
@F @
Q (x; y) = = 3x2 + 2x sin (2y) = 4x cos (2y) (6.62)
@y @y
so that total di¤erential becomes

dF = [6x + 2 sin (2y)] dx + [4x cos (2y)] dy (6.63)

In Example 4 we found the total di¤erential

dF = P (x; y) dx + Q (x; y) dy; (6.64)

where
@F @F
P (x; y) = = 6x + 2 sin (2y) and Q (x; y) = = 4x cos (2y) : (6.65)
@x @y
Taking the partial derivative of the function P (x; y) with respect to y and the
function Q (x; y) with respect to x, we …nd
@P @F @
= = [6x + 2 sin (2y)] = 4 cos (2y) (6.66)
@y @y@x @y
@Q @F @
= = [4x cos (2y)] = 4 cos (2y) : (6.67)
@x @x@y @x
This shows that for the total di¤erential

dF = P (x; y) dx + Q (x; y) dy; (6.68)

we found
@P (x; y) @Q (x; y)
= : (6.69)
@y @x
Such kind of di¤erential is called an exact di¤ erential. If the di¤erential is an
exact di¤ erential, then there is a function F (x; y) such that
@F @F
P (x; y) = ; Q (x; y) =
@x @y
Example 5 For the …rst-order DE
dy
x3 + 2y = 3x (2 xy) ; (6.70)
dx
6.2. THE FIRST-ORDER DE AND EXACT DIFFERENTIAL 119

(a) Find the total di¤erential for a function of two variable F (x; y) =constant
in form
dF (x; y) = P (x; y) dx + Q (x; y) dy = 0 (6.71)

(b) Show that the total di¤erential is an exact di¤erential and …nd the function
F (x; y) =constant.
(c) Using the function F (x; y) =constant, …nd the solution to the given di¤er-
ential equation.
Solution:
(a) Multiplying the given …rst-order di¤erential equation by dx, we …nd

x3 + 2y dy = 3x (2 xy) dx: (6.72)

This can be put in the form

dF (x; y) = P (x; y) dx + Q (x; y) dy = 0; (6.73)

where
P (x; y) = 3x (2 xy) ; Q (x; y) = x3 + 2y : (6.74)

(b) Now taking the partial derivative of P with respect to y and Q with respect
to x, we …nd
@P @
= [3x (2 xy)] = 3x2 (6.75)
@y @y
and
@Q @
= x3 + 2y = 3x2 : (6.76)
@x @x
There follows that
@P @Q
= (6.77)
@y @x
Thus, the total di¤erential is an exact di¤ erential and there exists a func-
tion F (x; y) =constant, such that
@F @F
P = and Q = : (6.78)
@x @y
Now using
@F
P = = 3x (2 xy) (6.79)
@x
and integrating with respect to x
Z x Z x
@F (x0 ; y) 0
dx = 3x0 (2 x0 y) dx0 (6.80)
x0 @x0 x0

we …nd
F (x; y) F (x0 ; y) = 3x2 x3 y 3x20 + x30 y: (6.81)
120 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

This equation can be rewritten as


F (x; y) = 3x2 x3 y + C (y) ; (6.82)
where we introduced the function
C (y) = F (x0 ; y) 3x20 + x30 y; (6.83)
is a constant with respect to x but not with respect to y. Now substituting
Eq. (6.82) into
@F
Q (x; y) = = x3 + 2y ; (6.84)
@y
we …nd
@
3x2 x3 y + C (y) = x3 + 2y
@y
dC (y) dC (y)
) x3 + = x3 2y ) = 2y (6.85)
dy dy
Now integrating with respect to y,
Z y Z y
dC (y 0 ) 0
dy = 2 y 0 dy 0 ; (6.86)
y0 dy 0 y0

we …nd
C (y) = y 2 + C1 ; (6.87)
where
C1 = y02 + C (y0 ) ;
is constant independent of x and y. Therefore, the solution to the DE is
found to be
F (x; y) = 3x2 x3 y + C (y) = 3x2 x3 y y 2 = C. (6.88)

(c) The equation


3x2 x3 y y2 = C (6.89)
can be rewritten as
y 2 + x3 y 3x2 + C = 0; (6.90)
or
ay 2 + by + c = 0; (6.91)
which is a quadratic equation with
a = 1; b = x3 ; c = 3x2 + C: (6.92)
Using the well known solutions for a quadratic equation,
p
b b2 4ac
y= (6.93)
2a
we …nd
1h 3 p 6 i
y (x) = x x + 4 (3x2 C) (6.94)
2
6.2. THE FIRST-ORDER DE AND EXACT DIFFERENTIAL 121

Example 6 Solve the DE


dy
y2 3x2 e3y = 2xe3y + ex : (6.95)
dx
in the same ways as Example 6.
Solution:
(a) Multiplying the given …rst-order di¤erential equation by dx, we …nd

y2 3x2 e3y dy = 2xe3y + ex dx: (6.96)

or
dF (x; y) = P (x; y) dx + Q (x; y) dy = 0; (6.97)
where
P (x; y) = 2xe3y + ex ; Q (x; y) = y2 3x2 e3y (6.98)

(b) For an exact di¤erential, we must be able to show that


@P (x; y) @Q (x; y)
= : (6.99)
@y @x
Using Eq. (6.98), we …nd
@P (x; y) @
= 2xe3y + ex = 6xe3y ;
@y @y
@Q (x; y) @
= y 2 3x2 e3y = 6xe3y ; (6.100)
@x @x
that shows
@P (x; y) @Q (x; y)
= : (6.101)
@y @x
Therefore, the di¤erential is an exact di¤ erential and there exists a func-
tion F (x; y) (=Constant, in this case) such that
@F @F
P (x; y) = ; Q (x; y) = : (6.102)
@x @y
This leads to
Z x Z x
@F 0 0
dx = 2x0 e3y + x0 ex dx0
x0 @x0 x0
) F (x; y) F (x0 ; y) = x2 e3y + xex x20 e3y + x0 ex0 + C1 (y) (6.103)

or
F (x; y) = x2 e3y + xex + C1 (y) :
where we introduced the function de…ned by

F (x; y) = x2 e3y + xex + C1 (y) : (6.104)


122 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

Substituting this result into


@F
= Q (x; y) = y2 3x2 e3y = 3x2 e3y y2 ; (6.105)
@y
we …nd
@ dC1 (y)
x2 e3y + xex + C1 (y) = 3x2 e3y y2 ) = y2 : (6.106)
@y dy
Integrating this equation with respect to y yields
Z
y3
C1 (y) = y 2 dy + K ) C1 (y) = + C2 ;
3
where C2 is a constant of integration with respect to y and it is independent
of both x and y. Now substituting this result into Eq. (6.104), the solution
to the exact di¤erential in Eq. (6.104) is found to be
1 3
F (x; y) = x2 e3y + xex y = C: (6.107)
3
Note that we have de…ned C representing all the constants independent
of the both variables x and y.

6.3 First-order DE and none exact total di¤er-


ential
We recall that the di¤erential equation

dF = P (x; y) dx + Q (x; y) dy (6.108)

is said to be an exact di¤ erential equation, when


@P (x; y) @Q (x; y)
= : (6.109)
@y @x
For an exact di¤ erential there exists a function, F (x; y); such that

@F (x; y) @F (x; y)
P (x; y) = ; Q (x; y) =
@x @y
2 2
@ F (x; y) @ F (x; y)
) = : (6.110)
@y@x @x@y
Sometimes, we may …nd a …rst order LDE with none constant coe¢ cients that
may lead to a total di¤erential that is not exact. Let’s consider the …rst-order
ODE with none constant function
dy (x)
a0 (x) y (x) + a1 (x) = f (x) : (6.111)
dx
6.3. FIRST-ORDER DE AND NONE EXACT TOTAL DIFFERENTIAL 123

This ODE can be rewritten as


dy (x) a0 (x) f (x)
= y (x) (6.112)
dx a1 (x) a1 (x)
that can be put in the form
dy (x)
Q (x; y) = P (x; y) : (6.113)
dx
There follows the total deferential

dF (x; y) = P (x; y) dx + Q (x; y) dy = 0; (6.114)

where
a0 (x) f (x)
P (x; y) = y (x) ; Q (x; y) = 1: (6.115)
a1 (x) a1 (x)
Here we note that
@P (x; y) @ a0 (x) f (x) @Q (x; y)
= y (x) = y; = 0; (6.116)
@y @y a1 (x) a1 (x) @x
and
@P (x; y) @Q (x; y)
6= (6.117)
@y @x
that mean the resulting DE from the …rst-order ODE is not an exact di¤erential
and the method we developed in the previous section does not work. However,
this method can be used if we can …nd a function that can make the ODE in Eq.
(6.111) to lead to an exact di¤erential. Next we will determine this function
which is referred as "The integration factor".
The Integration Factor: The …rst-order ODE with none constant coe¢ cient
in Eq. (6.111) can put in the form
dy
+ P (x)y = Q(x); (6.118)
dx
where we introduced the functions de…ned by
a0 (x) f (x)
P (x) = ; Q(x) = : (6.119)
a1 (x) a1 (x)
This is the generic …rst-order ODE. Let’s say this ODE does not lead to an exact
di¤erential. Let the function V (x) be the function that when it multiplied to
the ODE in Eq. (6.118) leads to a total di¤erential that is an exact di¤erential.
Thus multiplying Eq. (6.118) by V (x) ; we have
dy
V (x) + V (x) P (x)y = V (x) Q(x)
dx
dy
) V (x) + V (x) P (x)y V (x) Q(x) = 0 (6.120)
dx
124 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

which can put in the form

dF (x; y) = [V (x) P (x)y V (x) Q(x)] dx + V (x) dy = 0: (6.121)

We recall that for di¤erential to be an exact di¤erential the partial derivative of


the coe¢ cient of dx with respect to y and the partial derivative of the coe¢ cient
of dy with respect to x must be equal. This means for the di¤erential in Eq.
(6.121), we must have

@ @
[V (x) P (x)y V (x) Q(x)] = V (x) (6.122)
@y @x
that leads
dV (x) dV (x)
V (x) P (x) = ) P (x)dx = : (6.123)
dx V (x)
Upon integrating this equation one …nds
R
P (x)dx
V (x) = e (6.124)

The function V (x) given by the integral expression Eq. (6.124) is the multiplying
factor to the ODE in Eq. (6.118) that leads to an exact di¤erential. It is referred
as the integration factor. Thus multiplying Eq. (6.118) by the integration factor,
we …nd R R R
dy
e P (x)dx + e P (x)dx P (x)y = e P (x)dx Q(x): (6.125)
dx
Applying the product rule for di¤erentiation, we put this equation in the form
d h R P (x)dx i R
ye = e P (x)dx Q(x); (6.126)
dx
so that integrating it with respect to x; one can write
R
Z h R i
ye P (x)dx = e P (x)dx Q(x) dx + C; (6.127)

where C is the constant of integration. Dividing this equation by the integration


factor; we …nd
R
Z h R i R
y(x) = e P (x)dx e P (x)dx Q(x) dx + Ce P (x)dx : (6.128)

This is the general solution to the …rst-order ODE in Eq. (6.118) and the
constant C is determined from the given boundary condition set to the speci…c
problem that the DE is representing. In the next examples, we will see its
application. But let’s examine Eq. (6.128) if it gives the general solution we
determined for a …rst-order ODE with constant coe¢ cients. We recall, for a
ODE with constant coe¢ cient
dy (x)
a0 y (x) + a1 = f (x) (6.129)
dt
6.3. FIRST-ORDER DE AND NONE EXACT TOTAL DIFFERENTIAL 125

the solution were shown to be


a0
x Z
a0
x e a1 a0
y (x) = Ce a1
+ f (x) e a1 x dx: (6.130)
a1

Noting that Eq. (6.129) can rewritten as Eq. (6.118)

dy
+ P (x)y = Q(x); (6.131)
dx
where
a0 f (x)
P (x) = :Q (x) = : (6.132)
a1 a1
Substituting this functions into Eq. (6.128), we …nd
R a0 Z R a0 R
a1 dx
f (x) a0
y(x) = e e a1 dx dx + Ce a1 dx
: (6.133)
a1

that leads to a0
x Z
a0
x e a1 a0
y(x) = Ce a1
+ e a1 x f (x) dx: (6.134)
a1
which the same as Eq. (6.130).

Example 7 For the …rst-order DE


dy
2x + 4x2 = y (6.135)
dx

(a) Express the DE in the form

dy
+ P (x)y = Q(x); (6.136)
dx

(b) Using the general solution for such kind of …rst-order DE (Eq. (6.118)),
…nd the solution for y (x) subject to the boundary condition,

26
y (x0 = 1) = : (6.137)
3

Solution:

(a) Noting that the DE can be rewritten as

dy 1 dy
+ 2x = y) + P (x) y = Q (x) ; (6.138)
dx 2x dx
we have
1
P (x) = ; Q (x) = 2x: (6.139)
2x
126 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

(b) Using the function P (x) ; we …nd


Z Z
1 1
P (x)dx = dx = ln x: (6.140)
2x 2
Substituting this result and Q (x) = 2x into Eq. (6.118), we have
Z h i
1 1 1
ln x
y(x) = e 2 e 2 ln x ( 2x) dx + Ce 2 ln x
p
Z p p
= 2eln x eln(1= x) xdx + Celn x (6.141)

so that
Z
p 1 p 4 2 p
y(x) = 2 x p xdx + C x ) y(x) = x + C x; (6.142)
x 3
where we used the relations
p p p 1
eln x
= x and eln(1= x)
=p : (6.143)
x
Using the given boundary condition
26
y (x = 1) = (6.144)
3
the constant of integration, C, is found to be
4 26
+C = ) C = 10: (6.145)
3 3
Therefore, the general solution in Eq. (6.142) becomes
4 3=2 p
y(x) = 10 x x:
3
This is the solution to the ODE in Eq. (6.135) under the given boundary
condition.

6.4 Higher-Order ODE’s


In the previous section we have introduced to various methods for solving a …rst-
order ODE. Next we will get introduced to how to solve higher-order (n > 2)
ODEs. We recall that an nth order ODE is given by
dy d2 y d3 y dn 1 y
a0 (x) y + a1 (x) + a2 (x) 2 + a3 (x) 3 ::::an 1 (x) n 1 (6.146)
dx dx dx dx
dn y
+an (x) n = f (x) :
dx
We will develop the methods for solving this equation in two parts. In the
…rst part we will consider various techniques for Homogenous ODE. Generally,
a Homogeneous DE is a DE in which the function, f (x) = 0. In the second
part, we will develop the techniques for none-Homogeneous ordinary di¤erential
equations (HODE) that will be built upon the techniques for HODE.
6.4. HIGHER-ORDER ODE’S 127

6.4.1 Homogeneous ODE’s with constant coe¢ cient


From Eq. (6.146) an nth order HODE when all the coe¢ cients, a0 (x) =
a0 ; a1 (x) = a1 ; a2 (x) = a2 ::::an (x) = an ; can be written as

dy d2 y d3 y dn 1 y
a0 y + a1 + a2 2 + a3 3 ::::an 1 (6.147)
dx dx dx dxn 1
dn y
+an n = 0:
dx
We will introduce to how the general solution to such kind of HODE be deter-
mined by solving what is know as the indicial equation that can be derived from
the HODE.
Indicial Equation: The indicial equation is founded on an educational guess.
We guess the solution to Eq.(6.147) be

y(x) = Aekx : (6.148)

Substituting this into Eq. (6.147), one can easily …nd

a0 + a1 k + a2 k 2 + a3 k 3 ::::an 1k
n 1
+ an k n Aekx = 0: (6.149)

There follows that

a0 + a1 k + a2 k 2 + a3 k 3 ::::an 1k
n 1
+ an k n = 0: (6.150)

This equation is called the indicial equation. The solution to the indicial equa-
tion in Eq. (6.150) gives n roots that could be real or complex. If these roots
are none degenerate (i.e. no repeated identical roots) and are given by

k = k1 ; k2 :::kn ; (6.151)

the general solution to the HODE with constant coe¢ cients in Eq. (6.147) is
given by
y (x) = C1 ek1 x + C2 ek2 x + :::Cn ekn x (6.152)
As we have seen in the in the solutions to …rst-order ODE, the constants
C1 ; C2 ; :::Cn ; are determined from the boundary conditions pertinent to the
particular problem.

Example 8 An inductor and capacitor are connected in a series circuit, as


shown. At t = 0, the charge on the capacitor plates is Qo and the current
‡owing in the circuit is Io . Find an expression for the charge on the
capacitor, Q(t).

Solution: Using Kircho¤’s rule we may write

Q (t) dI (t)
+L =0 (6.153)
C dt
128 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

and recalling that I(t) = dQ=dt, we may write


d2 Q 1
+ Q = 0: (6.154)
dt2 LC
Eq. (6.154) is a HODE with constant coe¢ cients. Thus, we can assume a
solution of the form
Q (t) = Aekt : (6.155)
Substituting this into Eq. (6.154), we …nd
d2
Aekt + ! 2 Aekt = 0 ) k 2 + ! 2 = 0 (6.156)
dt2
the solution of which is given by
k1 = i!; k2 = i! (6.157)
where we introduced the constant de…ned by
1
!=p : (6.158)
LC
The general solution can then be expressed as
Q(t) = A exp (i!t) + B exp ( i!t) : (6.159)
At t = 0 the capacitor is fully charged and the charge is Q0 ; this means
Q(0) = A exp (i!t) + B exp ( i!t)jt=0 = Q0 ) A + B = Q0 : (6.160)
We are also given that the current at t = 0 is I0 : Thus using the de…nition
of an electrical current, we may write
dQ
I(t = 0) = = I0
dt t=0
d I0
) [A exp (i!t) + B exp ( i!t)] = I0 ) A B= (6.161)
dt t=0 i!
6.4. HIGHER-ORDER ODE’S 129

Combining Eqs. (6.160) and (6.161), one can easily show that
1 I0 1 I0
A= Q0 + ;B = Q0 : (6.162)
2 i! 2 i!
Therefore, the charge at a given time t can be expressed as
1 I0 1 I0
Q(t) = Q0 + exp (i!t) + Q0 exp ( i!t) ; (6.163)
2 i! 2 i!
or
exp (i!t) + exp ( i!t) I0 exp (i!t) exp ( i!t)
Q(t) = Q0 + :
2 ! 2i
(6.164)
Applying Euler’s formula Eq. (6.164) can be put in the form
Q(t) = A cos (!t) + B sin (!t) ; (6.165)
where
I0
C = Q0 ; D = : (6.166)
!
Introducing the constants de…ned by
I0
Q0 = E sin '; = E cos '; (6.167)
!
Eq. (6.165) can also be expressed as
Q (t) = E [sin ' cos (!t) + cos ' sin (!t)] = E sin (!t + ') ; (6.168)
with s
2
Q0 I0
tan (') = ;E = Q20 + : (6.169)
I0 =! !
If we de…ne another constant de…ned in a slightly di¤erent way
I0
Q0 = G cos '; = G sin '; (6.170)
!
one can rewrite Eq. (6.165) as
Q(t) = G cos ' cos (!t) + G sin ' sin (!t) = G cos (!t ') ; (6.171)
where s
2
I0 =! I0
tan (') = ;G = Q20 + : (6.172)
Q0 !
The current
dQ
I (t) = (6.173)
dt
can be determined using any one of the equations for the charge in Eqs.
(6.164), (6.165), (6.168), or (6.171). If we chose Eq. (6.168), we …nd
I (t) = I0 cos (!t) Q0 ! sin (!t) (6.174)
for the current in the circuit.
130 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

In Example 8, we saw that the solution to the second-order DE in Eq. (6.154)


can be expressed in four di¤erent forms given by Eqs. (6.164), (6.165), (6.168),
and (6.171). In view of this results, we can make two generalizations to the
solution of a second-order HODE that can be put in form

d2 y
! 2 y = 0; (6.175)
dx2
when ! a real constant. For the plus case,

d2 y
+ ! 2 y = 0; (6.176)
dx2
as we saw in the example above, the roots to the indicial equation

k 2 + ! 2 = 0; (6.177)

are complex, k1 = i! and k2 = i! and the general solution can be expressed


in four di¤erent forms given by:
8
>
> A exp (i!x) + B exp ( i!x) ;
<
C cos (!x) + D sin (!x) ;
y (x) = (6.178)
>
> E sin (!x + ') ;
:
G cos (!x + ') :

For the minus case


d2 y
! 2 y = 0; (6.179)
dx2
the roots to the corresponding indicial equation

k2 ! 2 = 0; (6.180)

are real, k1 = ! and k2 = !: It can easily be shown that the solution can be
expressed as 8
>
> A exp (!x) + B exp ( !x) ;
<
C cosh (!x) + D sinh (!x) ;
y (x) = (6.181)
>
> E sinh (!x + ') ;
:
G cosh (!x + ') :
Problem 1: show the solutions in Eq. (6.181)
Problem 2:Find the current and the charge in the previous example if a re-
sistor R is included in series with C and L:

Example 10 The Schröedinger Equation: The one-dimensional time-independent


Schröedinger equation for a particle traveling along the x-direction with a
potential energy function U (x) is given by

~2 d2 (x)
+ U (x) (x) = E (x); (6.182)
2m dx2
6.4. HIGHER-ORDER ODE’S 131

where ~ = h=2 (h is a constant called Planck’s constant), m is the mass of


the particle, and E is the total energy of the particle. Consider a particle
of mass m and energy E traveling in the potential shown in the …gure.
The x-axis is divided into two regions by the potential energy function.
Region I has U (x) = 0, which means that there are no forces acting on
the particle. In this region the particle is a free particle. Region II has
U (x) = Uo , where Uo is a constant such that Uo > E. This is a forbidden
region for the particle according to classical physics. The signi…cance of
the wave function (x) that is the solution to the Schröedinger equation
2
above for a given potential energy function is that j (x)j dx gives the
probability of …nding the particle within an interval dx at each value of x.
Solve the Schröedinger equation for the wave function of the particle in
each of the two regions I and II shown in the …gure. Comment on your
answers.

Solution: In region I, the potential energy is zero, U (x) = 0; and the Schröedinger
Equation becomes

~2 d2 (x) d2 (x) 2mE


= E (x) ) = (x) (6.183)
2m dx2 dx2 ~2
which we may write as

d2 (x) d2 (x)
= k 2 (x) = + k 2 (x) = 0 (6.184)
dx2 dx2
where
2mE
k2 = : (6.185)
~2
Substituting
x
(x) = Ae (6.186)
132 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

we …nd the indicial equation


2
+ k2 = 0 ) = ik (6.187)

so that the solution can be written as

(x) = A1 eikx + A2 e ikx


: (6.188)

In region II, the potential energy is zero, U (x) = U0 ; and the Schröedinger
Equation becomes

~2 d2 (x)
= (U0 E) (x)
2m dx2
d2 (x) 2m (U0 E)
) = (x) (6.189)
dx2 ~2
which we may write as

d2 (x) d2 (x)
2
= q 2 (x) = q 2 (x) = 0 (6.190)
dx dx2
where
2m (U0 E)
q2 = : (6.191)
~2
Substituting
x
(x) = Ae (6.192)
we …nd the indicial equation
2
q2 = 0 ) = q (6.193)

so that the solution can be written as

(x) = A1 eqx + A2 e qx
: (6.194)

But physical condition requires A1 = 0, which means


qx
(x) = A2 e : (6.195)

Indicial equation with degenerate roots: As we stated earlier the gen-


eral solution in Eq. (??) is valid when there are no repeated roots (none-
degenerate roots). However, it is possible that a HODE with constant coe¢ -
cients could lead to an indicial equation with degenerate roots. For example,
for the second-order HODE
d2 y dy
2a + a2 y = 0; (6.196)
dx2 dx
6.4. HIGHER-ORDER ODE’S 133

the roots of the quadratic indicial equation

k2 2ak + a2 = 0; (6.197)

are degenerate
k1 = k2 = a: (6.198)
When the roots are degenerate the general solution in Eq. (??) can be used
with some changes that depend on the degeneracy in the roots of the indicial
equation. For example, for the HODE in Eq. (6.196), the values for the roots
are identical (k1 = k2 = a), where the degeneracy, m = 2, the general solution
should be written as
y(x) = (A + Bx) eax : (6.199)
Note that for m = 2, the coe¢ cient to eax is a polynomial function, P1 (x) =
A + Bx; of degree l = m 1 = 1. Similarly, for a HODE that leads to an indicial
equation, with m = 3 degenerate roots, (k1 = k2 = k3 = a), the general solution
is given by
y(x) = A + Bx + Cx2 eax = P2 (x) eax : (6.200)
In this case the coe¢ cient polynomial function, P2 (x) ; has a degree, l = m 1 =
2:
Thus we can make the following changes to the general solution in Eq. (??)
in relation to the degenerate roots to the indicial equation:

(a) n-th order di¤erential equation leads to an indicial equation with degen-
erate roots, m = n; the general solution is given by

y(x) = A + Bx + Cx2 ::: + Dxn 1


eax (6.201)

(b) n-th order di¤erential equation that leads to an indicial equation with
three di¤erent set of roots. Suppose the …rst set of roots has m = m1
degenerate roots (i.e. k1 = k2 = k3 = :::km1 = a), the second set has
m = m2 degenerate roots (i.e. k10 = k20 = k30 = :::km2 = a0 ), and the
third set consist of m3 none-degenerate roots (k100 = a001 ; k200 = a002 ; k300 =
a003 :::km
00
3
= a00m3 ) (Note that m1 + m2 + m3 = n). For this HODE equation,
the general solution is given by
a0 x a0 x 00 00
y(x) = Pm 1 1 (x)e + Pm 2 1 (x)e + C100 ea1 x + C200 ea2 x(6.202)
00
00
+:::Cm 3
eam3 x ;

where

Pm 1 1 (x) = C0 + C1 x + C2 x2 ::: + Cm1 1x


m1 1
; (6.203)
Pm 2 1 (x) = C00 + C10 x + C20 x2 ::: + 0
Cm2 1 xm2 1 ;

are polynomials of degree m1 1 and m2 1; respectively.


134 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

Example 11 In each of the following parts, …nd the indicial equation, list the
roots of the indicial equation {ki }, and write out the general solution to
the DEs:

(a)
d3 y dy
3
+ = 0: (6.204)
dx dx
(b)
d4 s d3 s
2 3 = 0: (6.205)
dt4 dt3
Solution:

(a) Upon substituting

dy d3 y
y (x) = Aekx ) = ky (x) ) 3 = k 3 y (x) (6.206)
dx dx
into the given DE, the indicial equation becomes

k 3 + k = 0: (6.207)

The is a cubic equation with three distinct roots

k 3 + k = 0 ) k k 2 + 1 = 0 ) k1 = 0; k2 = i; k3 = i: (6.208)

The roots do not show any degeneracy in this case and the solution to the
DE can be written as

y (x) = A + Beix + Ce ix
: (6.209)

or
y (x) = A + D cos (x) + E sin (x) (6.210)

(b) Substituting

d3 s d4 s
s (t) = Aekt ; ) = k 3
s (t) ) = k 4 s (t) (6.211)
dt3 dt4
in the given DE, the roots to the indicial equation are found to be
3
2k 4 3k 3 = k 3 (2k 3) = 0 ) k1 = ; k2 = k3 = k4 = 0: (6.212)
2
Since the the indicial equation has three identical roots, the solution can
be written as
3
2k 4 3k 3 = 0 ) k 3 (2k 3) = 0 ) k1 = ; k2 = k3 = k4 = 0; (6.213)
2
6.4. HIGHER-ORDER ODE’S 135

This shows that the three roots (k2 ; k3 ; k4 ) are degenerate with degeneracy,
m = 3; and one none degenerate root (k1 ). Thus applying the relation in
Eq. (6.202), the general solution can written as

y (x) = A + Bx + Cx2 eax + Aek1 x ; (6.214)

where a is the value for the degenerate roots and k1 is the value for the
none degenerate root. Using these values the general solution to the DE
becomes
3
y (x) = A + Bx + Cx2 + Ae 2 x : (6.215)

N.B. You may use Mathematica to check you results. Here is the result for
the previous two examples

Here C[1], C[2], C[3], and C[4] are constants of integration. note that -1 in
8
the …rst solution can be absorbed in C[2] and 27 in the second solution can also
be absorbed in C[1].

Example 12 Damped Simple Harmonic Motion: An ideal spring of spring con-


stant k hangs vertically from a support rod and has a mass m attached to
its lower end. The mass hangs suspended in a viscous ‡uid. The mass is
pulled down some distance from its equilibrium position at y = 0 and re-
leased. Discuss the resulting motion of the mass if the viscous ‡uid exerts
a damping force on it given by fy = bvy , where b is a damping constant.
136 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

Solution: Since the mass initially was at equilibrium and also consider the dis-
placement from this equilibrium position, we do not consider the buoyant
and gravitational force on the mass. Therefore, the equation of motion for
the mass is given by Newton’s second law
d2 y
Fnet = may = m : (6.216)
dt2
The net force acting on the mass is given by
dy
Fnet = ky bvy = ky b (6.217)
dt
which leads to
d2 y dy d2 y dy
m = ky b ) 2 +2 + !2 y = 0 (6.218)
dt2 dt dt dt
where
k b
; !2 = = : (6.219)
m 2m
The indicial equation would then be
2
+2 + ! 2 = 0; (6.220)

and the solutions


1 = + D; 2 = D; (6.221)
where p
D= 2 !2 (6.222)
Case 1 ( = !) we have
p
D= 2 !2 = 0 (6.223)

and the roots to the indicial equation becomes

1 = 2 = (6.224)

and the general solution is given by


t
y(t) = (At + B) e (6.225)

This describes a Critically damped motion


Case 2 ( > !) we note that D is real
p
D= 2 !2 < (6.226)

which means the solution is given by


( D)t ( +D)t
y(t) = Ae + Be : (6.227)

This shows over-damped motion


6.4. HIGHER-ORDER ODE’S 137

Case 3 ( < !) we may write


p p
D= 2 !2 = (! 2 2) =i (6.228)

where p
= !2 2 (6.229)
which means the roots for the indicial equation becomes

1 = +i ; 2 = i (6.230)

and the solution is given by


( i )t ( +i )t t
y(t) = Ae + Be ) y(t) = e Aei t
+ Be i t
(6.231)

or
t
y(t) = e [A0 cos ( t) + B 0 sin ( t)] (6.232)
or 00
t
y(t) = e A cos ( t ) : (6.233)

It shows under-damped oscillatory motion.

6.4.2 None homogeneous ODE’s with Constant Coe¢ cients


th
An n -order none homogeneous ODE with constant coe¢ cient has the form

dy d2 y d3 y dn 1 y dn y
a0 y + a1 + a2 2 + a3 3 ::::an 1 n 1
+ an n = f (x) : (6.234)
dx dx dx dx dx
In this section, we will see the procedure that is built upon on the methods we
developed in the previous sections for HODE and an educated guess to part of
the general solution related to the none homogeneity of the DE. This procedure
consist of the following successive steps.

1-st Find the general solution to the HODE (i.e. with f (x) = 0). This solu-
tion is called the complementary function, or the homogeneous solution,
denoted by yh (x). Note that this solution contains n constants of integra-
tion for the nth -order ODE.
2-nd Find in any way you can a particular solution, yp (x); that satis…es the
full none homogeneous ODE. Note that this function contains no arbitrary
constants of integration.
3-rd The general solution to the none homogeneous ODE is then given by

y(x) = yh (x) + yp (x): (6.235)

4-th Any boundary conditions given to the problem must then be applied to
the general solution y(x).
138 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

Example 13 Find the general solution to the following di¤erential equation:

d2 s ds
3 + 2s = 16: (6.236)
dt2 dt

Solution:
Step 1 : Find the homogeneous solution. That means determine the solution
for
d2 s ds
2
3 + 2s = 0: (6.237)
dt dt
Noting that the indicial equation is

k2 3k + 2 = 0 ) k1 = 2; k2 = 1 (6.238)

the homogeneous solution can be written as

sh = Ae2t + Bet : (6.239)

Step 2 : By inspecting the none homogeneous ODE, I guess a particular


solution of the form
dsp d2 sp
sp = C ) = = 0; (6.240)
dt dt2
so that when we substitute it back into the none homogeneous ODE in
Eq. (6.236), we …nd
2C = 16 ) C = 8: (6.241)
Thus the particular solution is found to be

sp = 8: (6.242)

Step 3 : Write the general solution

s (t) = sh (t) + sp (t) ) s (t) = Ae2t + Bet + 8: (6.243)

Step 4 : Apply the boundary conditions if there are any. Here we are not
given any.
Example 14 Find the general solution to the following di¤erential equation:

d2 R
+ 4R = sin (3x) (6.244)
dx2

Solution:
Step 1 : Find the homogeneous solution for

d2 R
+ 4R = 0: (6.245)
dx2
6.4. HIGHER-ORDER ODE’S 139

Noting that the roots to the indicial equation are

k 2 + 4 = 0 ) k1 = 2i; k2 = 2i (6.246)

the homogeneous solution can be written as

Rh (x) = A cos (2x) + B sin (2x) : (6.247)

Step 2 : By inspecting the given none homogeneous ODE, I made a guess


(an educated guess) to the particular solution that has the form

Rp = C sin (3x) : (6.248)

Note that this guess will be valid because 3i is di¤ erent from the roots of
the indicial equation. See the next example.
Since in the particular solution we are not allowed to have arbitrary constant
of integration, we must determine C: Upon substituting Eq. (6.248) into
Eq. (6.244), this constant can be determined to be

9C sin (3x) + 4C sin (3x) = sin (3x) ) 5C sin (3x) = sin (3x)
1
) C= : (6.249)
5
Thus the particular solution becomes
1
Hp = sin (3x) : (6.250)
5
Step 3 : Write the general solution
1
R (x) = Rh (x) + Rp (x) ) R (x) = A cos (2x) + B sin (2x) sin (3x) :
5
(6.251)
Step 4 : Apply the boundary conditions if there are any. Here there are
no any.
Example 16 Find the general solution to the following di¤erential equation:

d2 y
+ 4y = sin (2x) (6.252)
dx2

Solution:
Step 1 : Find the solution for the homogeneous LDE

d2 y
+ 4y = 0: (6.253)
dx2
The roots to the indicial equation

k2 + 4 = 0 (6.254)
140 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

are
k1 = 2i; k2 = 2i: (6.255)
Then the homogeneous part of the solution can be expressed as

yh (x) = A cos (2x) + B sin (2x) ; (6.256)

or preferably
yh (x) = A0 e2ix + B 0 e 2ix
: (6.257)
Step 2 : When the non-homogenous part f (x) is a trigonometric or hyper-
bolic function (i.e. sin (ax) ; cos (ax) ; sinh (ax) ; or cosh (ax)), before we
try to …nd the particular solution it is recommended to express the func-
tions as exponential function applying Euler’s formula. Thus for function
in Eq. (6.252), one should write
1 2ix 2ix
f (x) = sin (2x) = e e : (6.258)
2i
Now comparing Eqs. (6.257) and (6.258), we note that values to the ex-
ponents to the Homogeneous solution (the roots of the indicial equation)
are the same as the exponents to the function, f (x) : Under these circum-
stances, one can make an educated guess to the particular solution based
on the property of an exponential function and the order of the di¤eren-
tial equation. For example, for the DE in Eq. (6.252), if one guesses a
particular solution
yp (x) = C1 e2ix + C2 e 2ix
which has the same form as the homogeneous solution in Eq. (6.257), we
…nd
d2 yp
+ 4yp = 4yp + 4yp = 0;
dx2
as one would have guessed. Therefore, we must guess a function that must
have a di¤erent form the homogenous solution. In this case, our educated
guess be
yp (x) = x C1 e2ix + C2 e 2ix ; (6.259)
that leads to
dyp
= C1 e2ix + C2 e 2ix + 2ix C1 e2ix C2 e 2ix
dx
d 2 yp
) = 2i C1 e2ix C2 e 2ix + 2i C1 e2ix C2 e 2ix
dx2
4x C1 e2ix + C2 e 2ix (6.260)
2
d yp
) + 4yp = 4i C1 e2ix C2 e 2ix
: (6.261)
dx2
Substituting this result into Eq. (??), we …nd
1 2ix
4i C1 e2ix C2 e 2ix
= e e 2ix
:
2i
6.5. THE METHOD OF SUPERPOSITION AND THE PARTICULAR SOLUTION.141

There follows that


1
C1 = C2 =
: (6.262)
8
Therefore, the particular solution becomes
1
yp (x) = xe2ix + xe 2ix
; (6.263)
8
that can be put in the form

x e2ix + e 2ix
x
yp (x) = = cos (2x) : (6.264)
4 2 4

Step 3 : Write the general solution, y (x) = yh (x) + yp (x)


x
y (x) = A cos (2x) + B sin (2x) cos (2x) (6.265)
4

Step 3 : In this example also we are not given boundary conditions.

6.5 The method of superposition and the par-


ticular solution.
In the previous example the particular solution in Eq. (6.259) that we came up
with an educated guess that can be put in the form

yp (x) = yp1 (x) + yp2 (x) ; (6.266)

where
yp1 (x) = Pn (x) e2ix ; yp2 (x) = Qn (x) e 2ix
: (6.267)
with Pn (x) = C1 x and Qn (x) = C2 x are polynomial functions with degree,
n = 1: To determine the particular solution, we also wrote the function, f (x) =
sin (2x) ; as a sum of two functions such that

d2 yp (x)
+ 4yp (x) = f (x) = f1 (x) + f2 (x) ; (6.268)
dx2
where
f1 (x) = Pm (x) e2ix ; f2 (x) = Qm (x) e 2ix
; (6.269)
1 1
with Pm (x) = 2iand Qm (x) = are polynomial of degree, m = 0: In a
2i
similar way, the particular solutions to

d2 yp0 1 (x) 0
d2 yp0 2 (x)
+ 4yp1 (x) = f 1 (x) and + 4yp0 2 (x) = f2 (x) ; (6.270)
dx2 dx2
can be guess to be

yp0 1 (x) = Pn0 (x) e2ix ; yp0 2 (x) = Q0n (x) e 2ix
: (6.271)
142 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

with Pn0 (x) = C10 x and Q0n (x) = C20 x are polynomial functions with degree, n =
1; respectively. Upon adding these two equations, we …nd a none homogeneous
ODE in Eq. (6.272) has the same form as Eq. (6.268)

d2 yp0
+ 4yp0 = f1 (x) + f2 (x) = f (x) ; (6.272)
dx2
where

yp0 (x) = 2yp0 1 (x) + 2yp0 2 (x) = Pn00 (x) e2ix + Q00n (x) e 2ix
; (6.273)

with Pn00 (x) = 2C10 x = C100 x and Q00n (x) = 2C20 x = C200 x are still the same
polynomials with the same degree, n = 1 obtained the previous example by
directly considering the function f (x). This can be generalized for an n-th
order none homogeneous ODE
X dn y
an = f (x) (6.274)
n
dxn

that can be written as


X dn y
an = f1 (x) + f2 (x) : (6.275)
n
dxn

The particular solution can be determined the principle of superposition that


states
y (x) = yh (x) + yp1 (x) + yp2 (x) (6.276)
where yp1 (x) and yp2 (x) satisfy the ODE with the non-homogeneous terms
f1 (x) and f2 (x),
X dn y
an n = f1 (x) (6.277)
n
dx
and
X dn y
an = f2 (x) ; (6.278)
n
dxn
respectively.
From what we discussed in relation to the previous example, we can also
state that for a second-order Non-homogenous second-order ODE that can be
expressed as
d d
a b y (x) = f1 (x) = ecx Pn (x) ; (6.279)
dx dx
where Pn (x) is a polynomial of degree n; the particular solution of such kind of
non-homogenous LDE is given by
8 cx
< e Qn (x) ; ( if c 6= a & c 6= b)
yp (x) = xecx Qn (x) ; (if c = a or b & a 6= b) (6.280)
: 2 cx
x e Qn (x) ; (if c = a = b)
6.5. THE METHOD OF SUPERPOSITION AND THE PARTICULAR SOLUTION.143

where
Qn (x) = a0 + a1 x + a2 x2 + :::an xn : (6.281)
is a polynomial of the same degree as Pn (x) with undetermined coe¢ cients
determined by substituting the particular solution to the given none homogenous
DE.

Example 18 Using the principle of superposition solve the following di¤eren-


tial equation:
d2 y
y = 2 sin (x) + 4x cos (x) (6.282)
dx2
Solution: For the homogenous DE

d2 y
y=0 (6.283)
dx2
the indicial equation
k2 1=0 (6.284)
gives
k1 = 1; k2 = 1: (6.285)
The the solution for the homogenous DE can be written as

yh (x) = C1 ex + C2 e x
: (6.286)

For the particular solutions …rst we rewrite the DE

d2 y
y = 2 sin (x) + 4x cos (x) (6.287)
dx2
as
d2 y eix e ix
eix + e ix
y=2 + 4x (6.288)
dx2 2i 2
d d
+1 1 = (2x i) eix + (2x + i) e ix
(6.289)
dx dx
or
d d
+1 1 = f1 (x) + f2 (x) ; (6.290)
dx dx
where
f1 (x) = (2x i) eix ; f2 (x) = (2x + i) e ix
; (6.291)
and
a = 1; b = 1 (6.292)
Then for the non-homogenous DE with the function f1 (x)

d d
+1 1 = f1 (x) = (2x i) eix (6.293)
dx dx
144 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

noting that
c = i 6= a; b (6.294)
we may write the particular solution as

yp1 (x) = (A + Bx) eix ; (6.295)

that gives

dyp1
= Beix + i (A + Bx) eix = Beix + iyp1 (x) ; (6.296)
dx
d 2 y p1 dBe ix
dyp1
= +i = iBeix + i Beix + iyp1 (x)
dx2 dx dx
= 2iBeix yp1 (x) (6.297)

Substituting these equations into

d2 yp1 (x)
yp1 (x) = f1 (x) = (2x i) eix : (6.298)
dx2
we …nd

2iBeix 2yp1 (x) = (2x i) eix ) 2iBeix 2 (A + Bx) eix


= (2x i) eix (6.299)

) [ 2Bx + 2Bi 2A] eix = (2x i) eix : (6.300)


There follows that

2B = 2; 2Bi 2A = i)B= 1) 2i 2A = i
i
) B= 1; A = (6.301)
2
and the …rst particular solution can be written as

i
yp1 (x) = + x eix : (6.302)
2

Similarly, for the second function we may write the second particular so-
lution
yp2 (x) = (A0 + B 0 x) e ix (6.303)
which gives

dyp2 d2 yp2
) = B0e ix
iyp2 (x) ) = 2iB 0 e ix
yp2 (x) : (6.304)
dx dx2
Then
d2 yp2 (x)
yp2 (x) = f2 (x) : (6.305)
dx2
6.6. THE METHOD OF SUCCESSIVE INTEGRATION 145

becomes
2iB 0 e ix
2 (A0 + B 0 x) e ix
= [ 2B 0 x 2B 0 i 2A] eix
ix
= (2x + i) e (6.306)
There follows that
2B 0 = 2; 2Bi 2A = i ) B 0 = 1 ) 2i 2A = i
i
) B0 = 1; A0 = (6.307)
2
and the second particular solution is
i ix
yp2 (x) = x e (6.308)
2
Therefore, the general solution is given by
i i
y (x) = C1 ex + C2 e x
+ x eix + x e ix
(6.309)
2 2
or
y (x) = C1 ex + C2 e x
+ sin (x) 2x cos (x) : (6.310)
Using Mathematica!

IS THAT NOT MUCH EASIER?

6.6 The method of successive integration


In the previous section the methods introduced to determine the general solu-
tions whether for homogeneous or none homogeneous ODE with constant coe¢ -
cients is based on an educated guess. In this section, we will derive a general an
integral expression for the general solution to a second-order none homogeneous
ODE by successively integrating the ODE. The general solutions determined by
an educated guess in the previous sections will be shown that it can be derived
from this general integral expression. To this end, we put the second-order ODE
with constant coe¢ cients
d2 y dy
a2 2
+ a1 + a0 y = F (x) (6.311)
dx dx
146 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

in the form
d2 d d d
a2 + a1 + a0 y = a b y = F (x) ; (6.312)
dx2 dx dx dx

where a and b are the roots for a quadratic equation given by


p p
a1 a21 4a2 a0 a1 + a21 4a2 a0
a= ;b = : (6.313)
a2 a2
Introducing a new variable u (x) de…ned by

d
u (x) = b y (x) ; (6.314)
dx

Eq. (6.312) can be rewrite as

d du
a u (x) = au = F (x) : (6.315)
dx dx
ax
Multiplying this equation by e ; we have

ax du ax ax
e ae u=e F (x) ; (6.316)
dx
so that applying the product rule, one can write
d ax ax
ue =e F (x) : (6.317)
dx
Integrating this equation with respect to x, we …nd
Z x
0
ue ax = e ax F (x0 ) dx0 + C1
Z x
0
) u = eax e ax F (x0 ) dx0 + C1 eax ; (6.318)

that leads to Z x
ax0
u (x) = eax e F (x0 ) dx0 + C1 eax ; (6.319)

where C1 is a constant of integration. Now substituting Eq. (6.319) into Eq.


(6.314), one can write
Z x
dy ax 0
by = e e ax F (x0 ) dx0 + C1 eax ; (6.320)
dx
bx
and multiplying by e , we …nd
Z x
bx dy bx (a b)x ax ax0
e be y=e e e F (x0 ) dx0 + C1 e(a b)x
: (6.321)
dx
6.6. THE METHOD OF SUCCESSIVE INTEGRATION 147

Once again applying the product rule this can be put in the form
Z x
d bx (a b)x 0
e y =e e ax F (x0 ) dx0 + C1 e(a b)x ; (6.322)
dx
so that integrating with respect to x, we can write
Z x Z x
00 0
e bx y = e(a b)x e ax F (x0 ) dx0 dx00
Z x
00
+C1 e(a b)x dx00 + C2 ; (6.323)

where again C2 is a constant of integration. Thus the solution to the DE in Eq.


(6.312) can be expressed as
Z " x Z 0 #
x
b)x00 ax0
y (x) = ebx e(a e F (x0 ) dx0 dx00 (6.324)
Z x
b)x00
+C1 ebx e(a dx00 + C2 ebx ;

where a and b are given by Eq. (6.313). Noting that the …rst term involving the
function F (x) is the a result of the none homogeneity of the DE and the second
and third terms are independent of this function, we can rewrite Eq. (6.324) as

y (x) = yH (x) + yp (x) ; (6.325)

where Z x
b)x00
yH (x) = C1 ebx e(a dx00 + C2 ebx (6.326)

and " #
Z x Z x00
00 0
yp (x) = ebx e(a b)x
e ax
F (x0 ) dx0 dx00 (6.327)

are the homogeneous and particular solutions, respectively. Next we will use
Eqs. (??) and (??) to derive the solutions we determined in the previous sections
using an educated guess.
Problem 1: Derive the integral expression for 3-rd and 4-th order none ho-
mogeneous ODE. From the results obtained try to write an expression for an
n-th order none homogeneous ODE with constant coe¢ cients.
Problem 2: Applying the general solution derived for a …rst-order ODE (Eq.
(6.128) in section 6.3, derive Eqs. (6.319) and (6.324).

A. HODE with none degenerate roots: We recall that for a second-order


homogeneous ODE (F (x) = 0)

d2 y dy
a2 2
+ a1 + a0 y = 0; (6.328)
dx dx
148 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

that leads to an indicial equation

a2 k 2 + a1 k + a0 = 0 (6.329)

the roots are given by Eq. (6.313). When these roots are none degenerate
(a 6= b), we saw that the general solution to DE is given by

y (x) = C1 eax + C2 ebx : (6.330)

The result in Eq. (6.324), for F (x0 ) = 0; becomes


Z x
00
y (x) = C1 e(a b)x dx00 + C2 ebx ; (6.331)

and upon carrying out the integration, one …nds

C1 ebx e(a b)x C1 eax


y (x) = + C2 ebx = + C2 ebx
a b a b
= C1 eax + C2 ebx : (6.332)

where we have absorbed the constant 1/ (a b) in C1 : This is the same


as Eq. (6.330).
B. HODE with degenerate roots: For a homogeneous ODE that leads to an
indicial equation with degenerate roots (i.e. a = b), Eq. (6.331) becomes
Z x
y (x) = C1 ebx dx00 + C2 ebx = (C2 + C1 x) ebx ; (6.333)

which is in agreement with Eq. (6.199).


C. None HODE: We have seen that the particular solutions, yp (x) ; for the
second-order none HODE
d2 y dy d d
a2 2
+ a1 + a0 y = a b = ecx Pn (x) ; (6.334)
dx dx dx dx

with a polynomial Pn (x) of degree n; were determined based on an edu-


cated guess. For such none HODE, depending on the degeneracy of the
values for a; b; and c; we saw that the particular solutions are given by
8 cx
< e Qn (x) ; ( when c 6= a & c 6= b) ;
yp (x) = xecx Qn (x) ; (when c = a or b & a 6= b) ; (6.335)
: 2 cx
x e Qn (x) ; (when c = a = b) :

Next we will derive these particular solutions from Eq. (6.327). To this
end, using the expression for n-th degree polynomial function, Pn (x) ;
n
X
Pn (x) = Ai x i ; (6.336)
n=0
6.6. THE METHOD OF SUCCESSIVE INTEGRATION 149

one can write the function, F (x) ; that makes the ODE none homogeneous,
as
X n
F (x) = ecx Pn (x) = ecx Ai x i : (6.337)
n=0

Then the particular solution in Eq. (6.327) can be expressed as,

Z " Z #
x n
X x00
00
yp (x) = ebx e(a b)x
Ai e(c a)x0 0i
x dx0 dx00 (6.338)
n=0
Z " Z #
n
X x x00
bx x00 x0 0i 0
= Ai e e e x dx dx00
n=0

where
=c a and =a b: (6.339)

Next we will evaluate Eq. (6.338) to derive the particular solutions given
in Eq. (6.335).

(i.) When c 6= a & c 6= b: Under this condition, =c a 6= 0: Under this


condition one can use the relation

di x0 x 0i
[e ]=e x ; (6.340)
d i

and show that


Z x00 Z x00
x0 0i 0 di x0 di x00
e x dx = e dx0 = e : (6.341)
d i d i

Substituting this equation into Eq. (6.338), we can write the particular
solution as
n
X Z x
x00 di x00
yp (x) = ebx Ai e e dx00
i=0
d i
n
X Z x
bx di + )x00
= e Ai i e( dx00 : (6.342)
i=0
d

Upon carrying out the integration, we …nd

n
X
bx di e( + )x
yp (x) = e Ai : (6.343)
i=0
d i +
150 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

Noting that
d0 e( + )x
e( + )x
= = C 0 e( + )x
;
d 0 + +
" #
d e( + )x
d 1 x
= 2 + e( + )x
d + d ( + ) +
= (C0 + C1 x) e( + )x ;
" #
d2 e( + )x
1 x
= 2 + e( + )x
d 2 + ( + ) +
" #
2 x x x2
= 3 2 2 + e( + )x
( + ) ( + ) ( + ) +
= C0 + C1 x + C2 x2 e( + )x
; (6.344)
one can easily show that
Xn Z
di x ( + )x00 00
Ai i e dx = C0 + C1 x + C2 x2 + :::Cn xn e( (6.345)
+ )x

i=0
d
= Qn (x) e( + )x
;
where Qn (x) is a polynomial of degree n. Note that the constants Ai are
absorbed in the constants Ci :Thus, substituting, = c a and = a b
in Eq. (6.345), the particular solution in Eq. (6.343) becomes
yp (x) = ebx Qn (x) e(c a+a b)x
= Qn (x) ecx : (6.346)
(ii.) When c = a & c 6= b: for this case we have = c a = 0 and Eq.
(6.338) becomes
n
X Z x" Z x00 #
bx x00
yp (x) = Ai e e x dx dx00 :
0i 0
(6.347)
n=0

Substituting this into Eq. (6.327), the particular solution can be expressed
as
Z
Ai bx x h x00 00 i+1 i 00
Xn
yp (x) = i+1
e e (x ) dx (6.348)
i=0
n+1
X Z x
Aj 1 x00 j
= ebx j
e (x00 ) dx00 ;
j=1

where we have introduced the summation index de…ned by j = i + 1.


Applying the relation in Eq. (6.340), one can easily show that
X Aj 1 dj Z x 00
n+1 n+1
X Aj 1 dj e x
bx x 00 bx
yp (x) = e e dx = e :
j=1
j
d j j=1
j
d j
(6.349)
6.6. THE METHOD OF SUCCESSIVE INTEGRATION 151

Following a similar procedure we used in (ii) earlier, one can show that
n+1
X Aj 1 dj e x
= C1 x + C2 x2 + :::Cn+1 xn+1 e x

j=1
j
d j
= x C0 + C1 x + C2 x2 :::Cn xn e x :(6.350)
Thus the particular solution in Eq. (6.349) becomes
yp (x) = ebx xe x Qn (x) = xecx Qn (x) ; (6.351)
where Qn (x) is a polynomial of degree n. This result is the same as the
solution in Eq. (6.335) when c = a & c 6= b or c = b & c 6= a.
(iii.) When c = a =b : for this case we have =c a = 0 and =a b and
Eq. (6.348) becomes
n+1
X Z x
Aj 1 j
yp (x) = ebx j
(x00 ) dx00 : (6.352)
j=1

Upon integrating this, we …nd


n+1
X Aj 1 j+1
yp (x) = ebx (x00 ) ; (6.353)
j=1
j (j + 1)

that can be put in the form


yp (x) = C2 x2 + C2 x3 + :::Cn xn+2 ebx = x2 ecx Qn (x) ; (6.354)
where Qn (x) is a polynomial of degree n. This result is also the same as
the solution in Eq. (6.335) when c = a =b.
Example 17 A forced Harmonic oscillator : Consider an ideal spring with
spring constant k = 8N=m with a mass m = 2kg attached to its one end
with the other end …xed. The spring and mass are sitting on a frictionless
surface. The mass is constantly acted by a periodic external force given by,
F~e = 2 sin (3t) x
^:What would be the amplitude of the oscillation. Initially

the mass is at rest with the spring at its equilibrium position. (i.e. the
spring is neither stretched nor compressed.)
152 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

Solution: In order to …nd the amplitude of oscillation one needs to solve the
equation of motion of the mass. Applying Newtons’s second low one can
write the equation of motion for the mass, m, as
d2 x
m = Fe kx = 2 sin (3t) kx: (6.355)
dt2
Using the values for the spring constant, k = 8N=m; and mass, m = 2kg;
that can be put in the form
d2 x d2 d d
+ 4x = +4 x= + 2i 2i x = sin (3t) : (6.356)
dt2 dt2 dt dt
In view of the general solutions in Eqs. (6.326) and (6.327) that we de-
rived by successively integrating the second-order none homogeneous ODE
given in Eq. (6.312), the homogeneous and the particular solutions to Eq.
(6.356) can be expresses as
Z t
00
bt
xH (t) = C1 e e(a b)t dt00 + C2 ebt (6.357)

and Z t" Z #
t00
bt (a b)t00 at0 0 0
xp (t) = e e e F (t ) dt dt00 : (6.358)

From Eq. (6.356), we have


e3it e 3it
a = 2i; b = 2i; and F (t) = sin (3t) = ; (6.359)
2i
so that Eqs. (6.357) becomes
Z t
2it 00 C1 2it
xH (t) = C1 e e4it dt00 + C2 e 2it
= e + C2 e 2it
: (6.360)
4i
Using Euler’s formula the homogeneous solutions can be expressed as
C1
xH (t) = [cos (2t) + 2i sin (2t)] + C2 [cos (2t) 2i sin (2t)]
4i
C1 C1
= + 2iC2 cos (2t) + 2iC2 sin (2t) ; (6.361)
4i 4i
we can rewrite as this equation as
xH (t) = A cos (2t) + B sin (2t) : (6.362)
Substituting the values for a, b, and F (t) from Eq. (6.359) into Eq.
(6.358), the particular solution, becomes
Z t" Z t00 0
! #
2it 4it00 2it0 e3it e 3it0
xp (t) = e e e dt0 dt00
2i
Z " Z 00 #
e 2it t 4it00 t 0
= e eit e 5it0 dt0 dt00 : (6.363)
2i
6.6. THE METHOD OF SUCCESSIVE INTEGRATION 153

Upon carrying out the inner integral, we have


" !#
2it Z t it00 5it00
e e 00 e
xp (t) = e4it + dt00
2i i 5i
Z " 5it00 00
!#
e 2it t e e it
= + dt00 ; (6.364)
2i i 5i

so that one can also carry out similar integration to this equation and …nd

2it
e e5it e it 1 1 e3it e3it
xp (t) = = sin (3t) : =
2i 5i2 5i2 5 5 2i
(6.365)
Using the results in Eqs. (6.362) and (6.365), the general solution, x (t) =
xH (t) + xp (t) ; becomes

1
x (t) = A cos (2t) + B sin (2t) sin (3t) : (6.366)
5

Initially (t = 0), the mass is at rest with the spring at its equilibrium
position. (i.e. the spring is neither stretched nor compressed.). This
means for the position of the mass given by Eq. (6.366), we …nd

x (t = 0) = 0 ) A = 0; (6.367)

and for the velocity, that can easily be determined from x (t) ;

dx (t) 3
v (t) = = 2 ( A sin (2t) + B cos (2t)) cos (3t) ; (6.368)
dt 5

we also obtain

3 3
v (t = 0) = 0 ) 2B =0)B= : (6.369)
5 10

Therefore, with the given initial conditions, the position is given by

3 1
x (t) = sin (2t) sin (3t) : (6.370)
10 5

and the velocity


1
v (t) = [2 cos (2t) 3 cos (3t)] : (6.371)
5

Using Mathematica:
154 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

6.7 Partial Di¤erential Equations


In the previous sections we saw that physical properties of some real systems
can be described by homogeneous or none homogeneous ordinary di¤erential
equations (ODE) of unknown functions of one variable. The techniques for
solving such kind of di¤erential equation were also discussed in these sections.
Generally, most real physical systems or process are described by unknown
functions of two or more variables that must be determined from a homogeneous
or none homogeneous Partial Di¤ erential Equation (PDE ).
A Partial Di¤ erential Equation (PDE ). PDE is a di¤erential equation in-
volving partial derivatives of an unknown function with respect to two or more
independent variables. There are various physical processes that requires use of
partial di¤erential equations. Some are listed below:

1. The gravitation potential of a given mass distribution described by given


mass density or electrical potential of a given charge distribution with a
given charge density satisfy the partial di¤erential equation (for example
in Cartesian coordinates) given by

@ 2 V (x; y; z) @ 2 V (x; y; z) @ 2 V (x; y; z)


+ + = (x; y; z); (6.372)
@x2 @y 2 @z 2

where V (x; y; z) is the gravitational potential (electrical potential) of a


given mass (charge) distribution and (x; y; z) is the mass (charge) density.
If you know the density, then by solving this partial di¤erential equation
you can determine the potential at some point in space. This PDE is
known as the Poisson’s equation. If (x; y; z) = 0; it is called Laplace’s
equation
6.7. PARTIAL DIFFERENTIAL EQUATIONS 155

2. The Schrödinger equation:

~2 @ 2 (x; y; z; t) @ 2 (x; y; z; t) @ 2 (x; y; z; t)


+ +
2m @x2 @y 2 @z 2
@
+U (x; y; z) (x; y; z; t) = i~ (x; y; z; t); (6.373)
@t
where (x; y; z; t) is the wave function U (x; y; z) is the potential energy.

3. The wave equation:

@ 2 u(x; y; z; t) @ 2 u(x; y; z; t) @ 2 u(x; y; z; t) 1 @2


+ + = u(x; y; z; t);
@x2 @y 2 @z 2 v 2 @t2
(6.374)

In this last section of this chapter, we will be introduced to the technique for
solving PDE and …nd the unknown function as it applied to real physical systems
or processes. This technique is known as the separation of variables technique
described by the following successive steps that build upon the methods for solve
ODE that we were introduced to in the previous sections.

1st . Assume a product form for the solution.

2nd . Substitute the product function into the original PDE and simplify.

3rd . Divide both sides of the resulting DE by the product function.

4th . Try to separate the variables, getting only one variable on one side of the
equation.

5th . The two sides of the equation must then be equal to the same constant,
called the separation constant. (Think about what to name this constant—
you can save yourself a lot of needless work by choosing a good name!)

6th . Solve the resulting ordinary DE’s using techniques that you were intro-
duced to in the previous sections.

7th . The general solution is then a linear combination of all possible product-
form solutions to the PDE, including all possible values of the separation
constant!

8th . Apply any BC’s to the general solution.

Example 18 Traveling Waves on a Stretched String:

(a) Show that the string supports a wave motion by deriving the wave equation.

@ 2 u (z; t) 1 @ 2 u (z; t)
= ; (6.375)
@z 2 v2 @t2
156 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

where s
T
v= ; (6.376)

T is the tension on the string, and is the linear mass density.

(b) Solve the 1-D traveling wave equation for a string of length L having both
ends …xed. Which means, apply the BC’s

u(y; t = 0) = u(y = 0; t) = u(y = L; t) = 0: (6.377)

Solution:

(a) Because of the weight of the mass hanging the string experiences a tension
force T . When is displaced from the equilibrium by the vibrator as shown
in the …gure below:If the string is displaced from the equilibrium, the net
6.7. PARTIAL DIFFERENTIAL EQUATIONS 157

transverse force on the segment between y and y + y can be expressed


as
F = T sin 0 sin ( ) (6.378)
If we assume the displacement is small, we can replace sin by tan,
0
F = T tan tan ( ) : (6.379)

Noting that

0 @u @u
tan = ; tan ( ) = ; (6.380)
@y y=y+ y @y y=y

we may write !
@u @u
F =T : (6.381)
@y y=y+ y @y y=y

We recall that the de…nition of the …rst derivative of a function u(y) at y

@u u(y + y) u(y)
= lim (6.382)
@y y=y
y!0 y

and the second derivative


2 3
@u @u
@2u 6 @y
y+ y
@y
y7
= lim 4 5 (6.383)
@y 2 y=y
y!0 y
" #
@2u @u @u
) y = lim (6.384)
@y 2 y=y
y!0 @y y+ y @y y

so that for small y; we can write

@2u
F 'T y (6.385)
@y 2
Noting that for the transverse motion, Newton’s second law can be written
as
@2u
F = ma = m 2 (6.386)
@t
If the string has a linear mass density , we can write m = y so that

@2u
F = y (6.387)
@t2
Combining the two expressions for the force F; we …nd

@2u @2u
T y= y (6.388)
@y 2 @t2
158 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

Which leads to
@2u @2u @2u 1 @2u
= or = (6.389)
@y 2 T @t2 @y 2 v 2 @t2
where s
T
v= : (6.390)

Equation (6.389) is the one-dimensional wave equation. It describes a


wave traveling in the y-direction with a speed, v; given by Eq. (6.390)
that depends on the magnitude of the tension on the string, T; and the
linear mass density of the string, :
(b) Let’s assume the solution of the PDE can be expressed as
u(y; t) = Y (y) T (t) (6.391)
so that we can write
d2 Y (y) Y (y) d2 T (t)
T (t) = (6.392)
dy 2 v2 dt2
dividing both sides by Y (y) T (t)
1 d2 Y (y) 1 d2 T (t) 2 d2 Y (y)
= = k ) + k 2 Y (y) = 0;
Y (y) dy 2 v 2 T (t) dt2 dy 2
d2 T (t)
+ k 2 v 2 T (t) = 0 (6.393)
dt2
The solutions to these HDEs is given by
Y (y) = A cos (ky) + B sin (ky) ; T (t) = C cos (kvt) + D sin (kvt) (6.394)
Therefore the general solution can be expressed as
X
u(y; t) = [Ak cos (ky) + Bk sin (ky)] [Ck cos (kvt) + Dk sin (kvt)]
k
(6.395)
Now we apply the boundary conditions. For u(y = 0; t) = 0; we have
X
u(0; t) = Ak (Ck cos (kvt) + Dk sin (kvt)) = 0 (6.396)
k

we have
X
u(y; t) = Ak (Ck cos (kvt) + Dk sin (kvt)) = 0 (6.397)
k

This can be zero for all time t; if an only if when Ak = 0: This leads to
simpli…ed expression for the general solution
X
u(y; t) = Bk sin (ky) [Ck cos (kvt) + Dk sin (kvt)] : (6.398)
k
6.8. HOMEWORK ASSIGNMENT 12 159

Then applying the second boundary conditions for u(y; t = 0) = 0; we …nd


X
u(y; t = 0) = (Bk sin (ky)) Ck = 0 ) Ck = 0: (6.399)
k

Thus the further simpli…ed solution can be rewritten as


X X
u(y; t) = Bk Dk sin (ky) sin (kvt) = Bk0 sin (ky) sin (kvt) (6.400)
k k

Then, applying the second boundary condition u(y = L; t) = 0; we have


X
u(L; t) = Bk0 sin (kL) sin (kvt) = 0 ) sin kL = 0
k
n
) kL = n ) k = ; where n = 0; 1; 2:::: (6.401)
L
Therefore, the general solution becomes
X
u(y; t) = Bk0 sin (kn y) sin (kn vt) : (6.402)
n=0

where
n
kn = ; n = 0; 1; 2:3::: (6.403)
L
This concludes the chapter to the introduction to di¤erential equations.

6.8 Homework Assignment 12


1. Boas 8.1 #6
2. Boas 8.1 #7
3. Boas 8.2 #6
5. Boas 8.2 #32

6.9 Homework Assignment 13


1. Consider the …rst-order di¤erential
dy
2x + y = 2x5=2
dx
(a) Show that the di¤erential equation is an exact di¤erential equation or not.
(b) Write the di¤erential equation in the form
dy
+ P (x) y = Q (x)
dx
and determine the solution using the relation derived for such di¤erential
equation.
160 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II

(c) Verify your solution using computer programing (Mathematica is recom-


mended).
2. Do part (a)-(c) in problem 1 for the di¤erential equation
dy
+ y tanh x = 2ex
dx

3. Consider 109 Gallons volume salty lake. Suppose a salty water with a salt
density
sw = 5 103 lb=gal
begins to ‡ow at a rate of
dVin
=4 105 gal=h:
dt
into the salty lake. At the same time the salty water in the lake begin to
‡ow out at a rate of
dVout
= 105 gal=h:
dt
(a) Determine the di¤erential equation describing the volume of the salty
water, V (t) ; in the lake. Solve the di¤erential equation to …nd the volume
of the salty water as a function of time.
(b) Find the di¤erential equation that describes the mass, m (t) of the salt in
the salty lake. Solve the di¤erential equation to …nd the mass of the salt
in the lake as function of time if the mass of the salt in the lake at the
initial time, before the in and out ‡ows begin, is m(0) = 107 lb:

6.10 Homework Assignment 14


1. Find the general solution for the None-homogeneous DE
d2 y dy
2
+ 2y = e2x :
dx dx
2. Find the general solution for the None-homogeneous DE
d2 y dy
5 +6 + 2y = x2 + 6x:
dx2 dx
3. Find the general solution for the None-homogeneous DE
d2
+ 1 y = 8x sin x:
dx2

4. Using the method of superposition …nd the general solution to the DE


d2 y dy
5 + 6y = 2ex + 6x 5:
dx2 dx
6.11. HOMEWORK ASSIGNMENT 15 161

6.11 Homework Assignment 15


1. In classical physics, the motion of a particle can be described for all time
once the “initial conditions” (position, velocity) of the particle’s motion
are known. The so-called “equation of motion” in this case is Newton’s
second law. In order to apply Newton’s second law, one must be able to
specify all of the forces acting on the particle in question. In quantum me-
chanics, the equation of motion is called the Schröudinger equation. This
equation is a partial di¤erential equation involving a function called the
wave function for the particle, denoted (x; t) (for 1-D problems), which
contains all observable information about the particle’s motion (position,
velocity, momentum, energy,...). Instead of precisely specifying all future
2
motion of the particle, in quantum mechanics the quantity j (x; t) is
used to predict the probability of a particular future motion. The 1-D
Schröudinger equation is given by

~2 @ 2 (x; t) @
+ U (x) (x; t) = i~ (x; t)
2m @x2 @t
where m is the mass of the particle, ~ is Planck’s constant (1:05 10
34J:s), i is the pure imaginary number, and U (x) is the net potential
energy function of the particle at position x. (Note that, since Fx =
dU (x)=dx, specifying the potential energy function in quantum mechan-
ics is equivalent to specifying all of the forces acting on the particle in
classical mechanics.) The form of the Schröudinger equation given above
is called the time-dependent (1-D) Schröudinger equation (TDSE) since
it involves the full time-dependent wave function, (x; t).
(a) Using the separation of variables technique, and letting the separation
constant be denoted E (which turns out to be the total energy of the
particle), show that the resulting di¤erential equation that is independent
of time— the so-called time-independent Schröudinger equation (TISE)—
has the (1-D) form

~2 d2 (x; t)
+ U (x) (x; t) = E (x; t)
2m dx2
where we have set (x; t) = (x) T (t). In applications, the potential
energy function U (x) must be speci…ed, and the resulting TISE solved for
the spatial wave function (x)
(b) Also, show that the total wave function in this case has the form
iEt=~
(x; t) = (x) e :
162 CHAPTER 6. INTRODUCTION TO DIFFERENTIAL CALCULUS II
Chapter 7

Integral calculus-scalar
functions

In this chapter we will be introduced to single and multiple integrals involving


scalar functions. We focus on the methods for carrying out such integrals. In
almost all branches of physics, integrations of scalar functions are often used
to determine various quantities describing the static or dynamic properties of
an object. The object could be in a microscopic particle like an electron in an
atom or massive objects like a planet in the solar system. The simplicity of
carrying out integrations of scalar functions depends on the geometry of the
object and system that object is existing in. The methods of integrations in
Cartesian coordinates for rectangular geometry and the methods of integrations
in curvilinear coordinates for curved geometry can greatly simplify the integra-
tions. We will be introduced the methods of integrations of single and multiple
integrals in Cartesian and curvilinear coordinates and its applications in real
physical problems.

7.1 Integration in Cartesian coordinates


We are interested in the methods of integrations of functions of single, two, and
three variables in Cartesian coordinates.

7.1.1 Single and multiple integrals


A. Single integral and arc length
Consider the two points P1 and P2 in the x-y plane shown in Fig. 7.1.1
described by the Cartesian coordinates (x1 ; y1 (x1 )) and (x2 ; y2 (x2 ))), respec-
tively. You took two paths to get from the …rst to the second point. The …rst
is a straight line path shown in red and the second is a curved path shown by
the black curve. The slope

163
164 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

y (x) dy (x)
m (x) = lim = : (7.1)
x!0 x dx
is a constant for all x from the …rst to the second point along the …rst path
but not along the second path. For any of these paths the net distance traveled
along the y-direction, that can be determined from the line integral
Z y2 (x2 ) Z x2
y = y2 (x2 ) y1 (x1 ) = dy = m (x) dx: (7.2)
y1 (x1 ) x1

are the same. Similarly, along the x-direction, the net distance given by
Z y2
x = x2 x1 = m0 (y) dy; (7.3)
y1

where
x (y) dx (y)
m0 (y) = lim = ; (7.4)
y!0 y dy
are the same for the two paths.
However, for the total distances (arc length), l, for these two paths are di¤er-
ent, as one can imagine. For the …rst path this length can easily be determined
using Pythagorean theorem and is given by
q q
2 2 2 2
l = (y2 y1 ) + (x2 x1 ) = ( y) + ( x) : (7.5)

That can also be put in the form


s s
2 2
y x
l= x +1= y + 1: (7.6)
x y
7.1. INTEGRATION IN CARTESIAN COORDINATES 165

For the second path, since the slope is not constant, one need to divide the
path into an in…nitessimal intervals, li ; …nd the lengths using Pythagorean
theorem, and sum it up all to …nd all. This means
s s
X X yi
2 X xi
2
l= lim li = lim xi +1= lim yi + 1:
i
li !0
i
xi !0 xi i
yi !0 yi
(7.7)
Noting that, generally, integration of a function f (x) is de…ned as

X Z2
lim fi ! df; : (7.8)
fi !0
i 1

and di¤erentiation as
df (x) f (x)
= lim (7.9)
dx x!0 x
one can write
v" # s
Z2 Zx2u
u dy 2 Zy2 2
dx
l= dl = t + 1 dx = 1+ dy: (7.10)
dx dy
1 x1 y1

This is the arc-length (or the total distance) for any path one can follow to go
from P1 to P2 :
B. Double integral-Area
Let’s add two points P3 and P4 with coordinates (x1 ; y2 ) and (x2 ; y1 ), re-
spectively, to form the rectangular region shown in Fig. 7.1.1. The area of this

rectangle depends on the length l (the distance between P1 and P4 ) and the
width w (the distance between P1 and P3 )

A = lw: (7.11)
166 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

The points along the length of the rectangle, l, are de…ned by the equation

x (y) = y1 = constant or x (y) = y2 = constant (7.12)

and those along the width, w, by

y (x) = x1 = constant or y (x) = x2 = constant. (7.13)

There follows that


dy dx
= 0; =0 (7.14)
dx dy
Thus the length and the width of the rectangle can be expressed using the
arc-length relation that can be expressed as
v" #
Zx2u
u dy 2 Zx2
l= t + 1 dx = dx (7.15)
dx
x1 x1

and s
Zy2 2 Zy2
dx
w= 1+ dy = dy; (7.16)
dy
y1 y1

respectively. The area of the rectangle can then be expressed using the double
(or surface) integral
Zx2 Zy2 Zx2Zy2
A = dx dy = dxdy; (7.17)
x1 y1 x1 y1

or
Zy2 Zx2 Zy2Zx2
A = dy dx = dydx: (7.18)
y1 x1 y1 x1

Note that the order of integration does not bring any di¤erence in the result.
This result can easily be generalized for any arbitrary shape in which the bound-
aries can be de…ned by a well-de…ne real function. Consider the region bounded
by the two curves shown in Fig. 7.1.1. Let the points on the lower boundary are
de…ned by the function y1 (x) and the upper boundary by the function y1 (x) :
In this case the double integral for the area in Eq. () should be expressed as
2 3
Zx2 yZ2 (x)
6 7
A= 4 dy 5 dx: (7.19)
x1 y1 (x)

On the other hand, if the points in the lower boundary and the upper boundary
are de…ned by the functions x1 (y) and x2 (y), respectively, instead (see Fig. ),
The order of integration needs to be switched and the area should be expressed
7.1. INTEGRATION IN CARTESIAN COORDINATES 167

as 2 3
Zx2 xZ2 (y)
6 7
A= 4 dy 5 dx; (7.20)
x1 x1 (y)

Note that even though the order of integrations are switched, the result should
be the same.
C. Triple integral-Volume
ZZZ
V = dxdydx: (7.21)

Example 1 Line and surface integral : Consider a wire bent into the arc of a
circle of radius R, as shown in the …gure below.
(a) You climbed along this wire. Determine the height of the elevation that
you climbed if you have moved a distance of half of the radius along the
x-direction.
(b) Find the length of the wire
(c) Find the area bounded by the curve, the x and y axes
Solution:
168 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

(a) The height is the change in y coordinate ( y),


Z x2
y= m (x) dx; (7.22)
x1

where m (x) is the slope of the wire given by


dy (x)
m (x) = :
dx
The function de…ning the curve shown in the …gure is
p
y (x) = R2 x2
and slope becomes
dy (x) x
m (x) = = p : (7.23)
dx R2 x2
The distance traveled along the x-direction can then be written as
Z 0 p !
xdx 0 3
2 2 1=2
y= p = R x = 1 R: (7.24)
R=2 R 2 x 2 R=2 4
7.1. INTEGRATION IN CARTESIAN COORDINATES 169

(b) Find the length of the wire is given by using the are-length in Eq. (7.10)
v" #
Zx2u
u dy 2
l= t + 1 dx: (7.25)
dx
x1

Using the result in Eq. (7.23), we may write


v" #
ZRu
u 2 ZR
t x Rdx
l= p + 1 dx = p dx: (7.26)
R 2 x2 R 2 x2
0 0

Introducing the transformation de…ned by

x = R sin ( ) ) dx = R cos ( ) (7.27)

the length of the wire is found to be

Z=2 Z=2
R2 cos ( ) R
l= q d =R d = : (7.28)
2
R2 1 sin ( ) 2
0 0

Note that we have expressed the limits of integration in terms of the


variable : That means for x = 0

x = R sin ( ) ) =0

and x = R
1
x = R sin ( ) ) = sin (1) = =2

Homework problem: Using the expression


s
Zy2 2
dx
l= 1+ dy; (7.29)
dy
y1

show that you will …nd the same result.


(c) The area is given by the surface integral
ZZ
A= dxdy: (7.30)

Noting that the x and y axes are de…ned by the equations y = 0 and
x = 0; respectively, for the area bounded and integration with respect to
x …rst, we have
0p 1
R2 y 2
ZR Z ZRp
B C
A= B @ dxC
A dy = R2 y 2 dy: (7.31)
0 0 0
170 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

Introducing the transformation of variable de…ned by


y = R sin ) dy = R cos d (7.32)
and noting that
y=0) = 0; y = R ) = =2 (7.33)
we may rewrite the area as
Z=2p Z=2
2 2 R2
A= R2 R2 sin R cos d = R cos2 d = : (7.34)
4
0 0

Example 2 Volume integral : A triangle in the x-y plane has vertices at the
points (x; y; z) = (0; 0; 0); (1; 0; 0);and (0; 2; 0). Find the volume of space
that lies above this triangle, but below the plane z = 2+x+y. All distance
units are in centimeters.
Solution: The volume is given by
ZZZ
V = dxdydx: (7.35)

The equation of the line joining the point (1; 0; 0) and (0; 2; 0) is given by
y 2 2 0
= ) y = 2x + 2 (7.36)
x 0 0 1
Then the volume is given by
8 2 3 9
Z < y=Z 2x+2 z=2+x+y
x=1 Z =
V = 4 dz 5 dy dx: (7.37)
: ;
x=0 y=0 z=0

Integrating with respect to z,


8 9
Z < y=Z 2x+2
x=1
=
V = (2 + x + y) dy dx; (7.38)
: ;
x=0 y=0

with respect to y
Z
x=1 2x+2 Z
x=1
y2
V = 2y + xy + dx = [2 ( 2x + 2) + x ( 2x (7.39)
+ 2)
2 0
x=0 x=0
# Z
x=1
2
( 2x + 2)
+ dx = [6 6x] dx;
2
x=0

and with respect to x; results in


Z
x=1
1
V = [6 6x] dx = 6x 3x2 0
= 3cm2 : (7.40)
x=0
7.1. INTEGRATION IN CARTESIAN COORDINATES 171

7.1.2 Physical Applications


There are several physical quantities that require carrying out single and mul-
tiple integrals. Here are some examples in electrostatics and mechanics.
I. Electrostatics

A. Electrostatic Potential
The electrostatic properties of a given charge distribution can all be derived
from the electrostatic potential, V (r). In an introductory physics courses
you have introduced to the electrostatic potential of a point charge. This
electrostatic potential at a point in space described by the position vector;
~r; measured from a point charge with charge, q; that is placed at the origin,
is given by
q
V (r) = ; (7.41)
4 0 j~rj
if the point charge is placed in a free space (with an electrical permittivity,
0 ) in the absence of any other charge. When the charge is not a point
charge and instead the charge is a line, surface, or volume charge with
none uniform distribution, the electrostatic potential, V (r); is determined
by carrying out single or multiple integrations that depends on the given
charge distributions. For the given charge distribution let’s consider an
in…nitely small charge dq 0 at a point described by the position vector, ~r0 ;
from the origin of the coordinate system. The corresponding in…nitessimal
electrostatic potential, dV (r); at a point in space described by the position
vector, ~r; that is also measured from the origin, depends on the distance of
this point from the position of the in…nitesimal charge, dq 0 : This distance
R (r0 ) is determined from the magnitude of the position vector directed
from the position of dq 0 to the position of the point in space that we want
to determine the corresponding electrostatic potential given by
0
R (r0 ) = ~r ~r : (7.42)

In view of Eq. (7.41), one can then express dV (r) as

dq 0
dV (r) = : (7.43)
4 r ~r0 j
0 j~

Line charge density : Suppose you are given a charge that is distributed on
a …nite length of wire with length, L. Also the charge is not uniformly
distributed on the wire and the charge distribution is described by a line
a line charge density,
dq 0
(~r0 ) = ) dq 0 = (~r0 )dr0 : (7.44)
dr0
Using this equation, the total electrostatic potential due to the entire
charge over the entire length can be determined from Eq. (7.43) using the
172 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

line integral
Z Z
dq 0 1 (~r0 )
V (r) = = dr0 (7.45)
4 r ~r0 j
0 j~ 4 0 L j~r ~r0 j

Surface charge density: Any external charge that one may put on a metal-
lic ball is distributed only on the surface of the sphere no matter the
magnitude of the charge. This charge distribute itself uniformly or none
uniformly depending on the absence or presence of charges in surround-
ing regions of the sphere. let’s consider the general case none uniform
distribution that can be described by a surface charge density,
dq 0
(~r0 ) = ) dq 0 = (~r0 )da0 ; (7.46)
da0
where da0 is the in…nitessimal area over which the in…nitessimal charge
dq 0 is distributed on: In this case the electrostatic potential at a point ~r
due to the entire charge on the surface of the sphere is given
Z Z
1 (~r0 )
V (r) = da0 : (7.47)
4 0 j~r ~r0 j
Area

This is a surface integral that requires multiple integration. Note that


the in…nitessimal area, da0 ; on the surface of the sphere with radius r0 , in
spherical coordinates can be expressed as
0
da0 = r02 sin d 0 d'0 (7.48)

Volume charge density: Similarly, for a charge described by a volume


charge density,
dq 0
(~r0 ) = 0 ) dq 0 = (~r0 )dv 0 ; (7.49)
dv
the electrostatic potential at a point ~r for the entire volume of charge can
be expressed as
ZZ Z
1 (~r0 )
V (r) = dv 0 : (7.50)
4 0 j~r ~r0 j
V olum

B. The total charge: Like the total electrostatic potential depends on the
charge densities and determined by single or multiple integrations, so does
the total charge. The total charge, Q, distributed on a line is determined
by single integral Z
Q= (~r0 )dr0 ; (7.51)
Line
on a surface and volume by multiple integrations given by
Z Z
Q= (~r0 )da0 ; (7.52)
Area
7.1. INTEGRATION IN CARTESIAN COORDINATES 173

and
ZZ Z
Q= (~r0 )dv 0 ; (7.53)
V olum

respectively.

II. Classical Mechanics

A. Moment of inertial, I: In classical mechanics, moment of inertia is also


called mass moment of inertia, rotational inertia, polar moment of iner-
tia of mass, or the angular mass, (SI units kg m2 ). It is a property of a
distribution of mass in space that measures its resistance to rotational ac-
celeration about an axis. Newton’s …rst law, which describes the inertia of
a body in linear motion, can be extended to the inertia of a body rotating
about an axis using the moment of inertia. That is, an object that is ro-
tating at constant angular velocity will remain rotating unless acted upon
by an external torque. In this way, the moment of inertia plays the same
role in rotational dynamics as mass does in linear dynamics, describing
the relationship between angular momentum and angular velocity, torque
and angular acceleration. The symbols I and sometimes J are usually
used to refer to the moment of inertia or polar moment of inertia. For a
discrete mass that consist of N entities each with mass mi , the moment
of inertial is given by
XN
I= mi ri2 ; (7.54)
i=1

for continuum mass distribution in volume, V , described by the mass


density, (r) ; the moment of inertia can be determined using the volume
integral
Z ZZZ ZZZ
I= dmr2 = (r) r2 dv = (x; y; z) r2 dxdydz (7.55)
M V V

where ri (r) is the perpendicular distance from the axis of rotation and

dm dm
(r) = = (7.56)
dv dxdydz

is the volume mass density in Cartesian coordinates.

B. The total mass: The total mass, M , in a volume with mass density, (r)
is given by
ZZ Z
M= (r)dxdydz: (7.57)
V ol m
174 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

C. Center of Mass: The center of mass (gravity) for a given mass with uni-
formly or nonuniform distribution is a point at which the mass can be
at rotational equilibrium about an axis passing through this point. The
center mass, in Cartesian coordinates, (x; y; z) can be determined using
Z Z Z Z Z Z
x dm = xdm; y dm = ydm; and z dm = zdm: (7.58)

For a mass distributed on a volume with volume mass density


dm dm
(x; y; z) = = ) dm = (x; y; z) dxdydx (7.59)
dv dxdydz
the center of mass can expressed as
ZZZ ZZZ
x (x; y; z) dxdydx y (x; y; z) dxdydx
x = ZZZ ; y = ZZZ ;(7.60)
(x; y; z) dxdydx (x; y; z) dxdydx
ZZZ
z (x; y; z) dxdydx
z = ZZZ : (7.61)
(x; y; z) dxdydx

If mass is distributed along a line on the x-axis with linear mass density,
we have
dm
(x) = ) dm = (x) dx (7.62)
dx
and the center of mass becomes
R
(x) xdx
x= R ; y = 0; and z = 0: (7.63)
dm

Similarly for surface mass distribution on the x-y plane, with surface mass
density
dm dm
(x; y) = = ) dm = (x; y) dxdy (7.64)
da dxdy
the center of mass is given by
RR RR
x (x; y) dxdy y (x; y) dxdy
x= RR ;y = R R ; and z = 0: (7.65)
(x; y) dxdy (x; y) dxdy

The next examples illustrates the application of single and multiple integra-
tions to determine the physical quantities described.

Example 3 A thin rod of length L lies along the positive x-axis with one end
at the origin. It has a linear charge density given by (x) = Ax, where A
is a constant.
7.1. INTEGRATION IN CARTESIAN COORDINATES 175

(a) What are the MKS units of the constant A?


(b) Find an expression for A if the total charge on the rod is Q.
(c) Point P is located a distance b from the origin along the negative x-axis.
Find an expression for the electrostatic potential at the point P , V (r).
Solution:
(a) the unit must be the unit of charge=length2 = C=m2
(b) The total charge for a line charge distribution is given by
Z Z L
0 0 AL2 2Q
Q= (~r )dr = Axdx = )A= 2 (7.66)
line 0 2 L

(c) For a line charge distribution, the potential is given by


Z L
1 (~r0 )
V (r) = dr0 (7.67)
4 0 0 j~r ~r0 j

Using the charge density (~r0 ) = Ax0 and


0
~r ~r = jb ( x
^) x0 (^
x)j = b + x0

one can write Z L


A x0
V (r) = dx0 : (7.68)
4 0 0 b + x0
Noting that
x0 b
=1 ; (7.69)
b + x0 b + x0
one can put the integral in the form
Z "Z Z #
L L L
A b A dx0
V (r) = 1 0
dx0 = 0
dx b
4 0 0 b+x 4 0 0 0 b + x0
(7.70)
which leads to
x=L
A A b+L
V (r) = [L b ln (b + x)] = L b ln : (7.71)
4 0 x=0 4 0 b
176 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

Using the result in part b for A; we may express the electrostatic potential
as
Q b b+L
V (r) = 1 ln : (7.72)
2 0L L b

Example 4 A semi-circular plate of radius R is arranged in the upper-half x-y


plane with its center-of-curvature at the origin. The plate has a surface-
charge density given by (x; y) = Ax2 y, where A is a constant.

(a) What are the MKS units of the constant A?

(b) What is the total charge on the plate in terms of A?

(c) Re-do part (b), switching the order of integration.

Solution:
charge
(a) The MKS unit of A must be the unit of length 5
since we must get a unit
charge
for the surface charge density (x; y) after we multiply A by x2 y
length 2
length3 :

(b) The total charge is given by surface integral


2 3
y 0 (x0 )
Z Zx2 Z
6 7
Q= (~r0 )da0 = 4 (x0 ; y 0 )dy 0 5 dx0 (7.73)
area
x1 y1 (x0 )

Using equation of a circle

x02 + y 02 = R2 (7.74)

the equation of curved part of the semi-circle in the …gure can be written
as p
y 0 (x0 ) = R2 x02 : (7.75)
7.1. INTEGRATION IN CARTESIAN COORDINATES 177

For the area shown in the …gure the limits of integration


p would then be
x1 = R to x2 = R and y10 (x0 ) = 0 to y20 (x0 ) = R2 x02 and the
integral becomes
2 0p 13
Z R RZ2 x02
6 02 B C7
Q=A 4x @ y 0 dy 0 A5 dx0 ; (7.76)
R
0

where we used the surface charge density


(x0 ; y 0 ) = Ax02 y 0 : (7.77)
Upon carrying out the integration with respect to y …rst, we have
Z R " p
R 2 x2
# Z
2 y 2
A R
Q=A x dx = R2 x2 x4 dx; (7.78)
R 2 0 2 R

so that integration with respect to x0 leads to


" R
#
A R 2 x3 x5 R5 R5 2AR5
Q= =A = (7.79)
2 3 5 R 3 5 15

(c) If we switch the order of integration, we have


0
2 (y )
Z Zy2 xZ
Q= (~r0 )da0 = [ (x0 ; y 0 )dy 0 ] dx0 (7.80)
area
y1 x1 (y 0 )

Referring to the …gure,p the limits of integration p would then be y1 = 0 to


y2 = R and x1 (y 0 ) = R2 y 02 to x2 (y 0 ) = R2 y 02 and the integral
becomes
2 p 3
R2 y 02 2 p 2 02 3
Z R Z Z R R y
6 7 x 03
Q = A y0 64 x02
dx07
5 dy 0
= A y 04 5 0
p 2 02 dy
0 p 2 02 0 3 R y
R y
2 3
Z R
3
2 R2 y 02 2

= A y 4
0 5 dy 0 (7.81)
0 3

The above integration can be carried out using transformation of variables.


If we introduce
du
u = R2 y 2 ) = ydy; (7.82)
2
one can write
2 3
Z 3
Z 3
2 R2 y 2 2 u2 2 5
y 4 5 dy = du = u2 (7.83)
3 3 15
178 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

Therefore, the charge becomes


2 3
Z R 3
R
2 R2 y 2 2 2A 5 2AR5
Q=A y4 5 dy = R2 y2 2
= (7.84)
0 3 15 0 15

Example 5 A right-circular cylinder has a height H, a radius R, and a mass M


distributed uniformly throughout its volume. Find the moment of inertia
of the cylinder about its axis of symmetry.
Solution: Since the mass is distributed uniformly in the cylinder, the volume
density, (r) ; is a constant given by
dm M M
(r) = = = : (7.85)
dv V R2 H
The moment of inertia about the axis of symmetry is an axis that passes
through the center of the cylinder. For a cylinder center about the z-axis,
we can take the z-axis as the center of symmetry. Thus using Eq. (7.55),
the moment of inertia can be expressed using a triple integral in Cartesian
coordinates as
Z ZZZ ZZZ
2 2 M
I = dmr = (r) r dxdydz r2 dxdydz: (7.86)
R2 H
M V V

Noting that r is perpendicular distance of the mass, dm; (in the volume,
dv = dxdydz, that is centered at a point with coordinates (x; y; z)) from
the axis of symmetry (z-axis), we can write
p
r = x2 + y 2 ; (7.87)
where
0 r R: (7.88)
Furthermore, since the cylinder, with height H; is centered about the z-
axis,
H H
z : (7.89)
2 2
Thus the moment of inertia can be expressed as
Z z2 =H=2 "Z y2 =R "Z x2 =pR2 y2 # #
M 2 2
I= x + y dx dy dz:
R2 H z1 = H=2 y1 = R x1 = pR2 y2
(7.90)
Since the integrand is independent of z we can carry out integration with
respect to z followed by integration with respect to z. This leads to
2 p 2 2 3
Z R 3 R y
M 4 x 5
I = + y2 x p 2 2 dy H
R2 H R 3 R y
Z R
M R2 + 2y 2 p 2
= 2
2 R y 2 dy: (7.91)
R R 3
7.1. INTEGRATION IN CARTESIAN COORDINATES 179

Introducing the transformation of variable de…ned by

y = R sin ( ) ) dy = R cos ( ) d (7.92)

we may write
Z =2
M R2 + 2R2 sin2 ( )
I = 2 cos2 ( ) d
=2 3
Z =2
2M R2
= cos2 ( ) + 2 sin2 ( ) cos2 ( ) d : (7.93)
3 =2

Note that we have re-expressed the limits of integration in terms of


(i.e. R x R) =2 =2) : Using the result obtained by Math-
ematica (which you can also easily show),

the integral becomes


Z =2
3
cos2 ( ) + 2 sin2 ( ) cos2 ( ) = : (7.94)
=2 4

The moment of inertia is then found to be


M R2
I= : (7.95)
2

Example 6 Consider a wire of constant linear mass density bent into the arc
of a circle of radius R, as shown in the …gure below.
(a) What is the total mass of the wire?
(b) Find the wire’s center-of-mass.
(c) Find the area under the arc of the wire.
(d) The wire is now rotated about the x-axis. Find the volume inside the
surface thus generated.
(e) Exercise: Find the surface area of this curved surface of rotation.
180 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

Solution:

(a) The total mass of the wire is given by


Z
M = dm (7.96)

For uniformly distributed linear mass over curved length the linear mass
density, (l) ; given by

dm
(l) = = ) dm = dl
dl
the total mass becomes Z
M= dl = l (7.97)

Using the result in Example 1(b) for the arc-length of the curve

R
l= ;
2
the total mass becomes
R
M= : (7.98)
2
(b) The mass is distributed uniformly on a curved line on the x-y plane. Thus
coordinates of the center of mass are given by
Z Z Z Z
x dm = xdm; y dm = ydm; and z = 0 (7.99)

Noting that from the result in (a)


Z
R
M = dm = (7.100)
2
7.1. INTEGRATION IN CARTESIAN COORDINATES 181

The x and y coordinates of the center of mass can be expressed as


R Z Z
xdm 2 2
x= = xdm = xdl
M R R
and R Z Z
ydm 2 2
y= R = ydm = ydl; (7.101)
dm R R
where we replaced dm = dl and take into account that the density is
constant. From Eq. (7.10), we note that
v" # s
u
u dy 2 dx
2
dl = t + 1 dx = 1 + dy (7.102)
dx dy

we may write v"


Z u #
2 R u dy 2
x= xt + 1 dx; (7.103)
R 0 dx

and s
Z R 2
2 dx
y= y 1+ dy: (7.104)
R 0 dy
The curve is de…ned by the equation of a circle
p p
x2 + y 2 = R2 ) y = R2 y 2 or x = R2 y2 ; (7.105)

we chose the positive roots for both x and y since the in the …rst quadrant
both are positives. Using the expressions for x and y we have
dx y dy x
= p ; = p (7.106)
dy R 2 y2 dx R2 x2
so the one can write
v" #
Z R uu 2 Z R
2 x 2 xdx
x= xt p + 1 dx = p xdx;
R 0 R 2 x2 0 R 2 x2
(7.107)
and
v2 3
Z u !2 Z
R u R
2 u y 2 ydy
y= y t4 p + 15dx = p : (7.108)
R 0 R 2 y2 0 R2 y 2

Upon carrying out the integrations,

2 p R
2 p R
x= R2 x2 ;y = R2 x2 ; (7.109)
0 0
182 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

and one …nds


2R
x=y= : (7.110)

The center of mass is ( 2R ; 2R ; 0).


(c) The area can be determined using Eqs. (7.19)
2 3
Zx2 yZ2 (x)
6 7
A= 4 dy 5 dx: (7.111)
x1 y1 (x)

or (7.20) 2 3
Zx2 xZ2 (y)
6 7
A= 4 dy 5 dx: (7.112)
x1 x1 (y)

Here we chose to use Eq. (7.19). For the boundaries shown in the …gure,
we have p
x1 (y) = 0; x2 (y) = R2 x2 (7.113)
and
x1 = 0; x2 = R: (7.114)
so that the area becomes
Z R " Z pR 2 x2
# Z R p
A= dy dx = R2 x2 dx: (7.115)
0 0 0

Introducing the transformation of variable de…ned by


x = R sin ) dx = R cos d ; (7.116)
we may write
Z Rp Z R
A= R2 R2 sin2 R cos d = R2 cos2 d : (7.117)
0 0

Employing the relation


cos (2 ) + 1
cos (2 ) = cos2 ( ) sin2 ( ) ) cos2 ( ) = (7.118)
2
the area is found to be
Z =2 =2
2 cos 2 + 1 sin 2 R2
A=R d = + = : (7.119)
0 2 4 2 0 4
1
Note that the limits of integration in terms of = sin (x=R) ;

x=0) = 0; and x = R ) = (7.120)


2
are used in the integration.
7.2. INTEGRATION IN CURVILINEAR COORDINATES 183

(d) The rotation of the wire about the x-axis creates a hemisphere of radius, R;
shown in the …gure. A point on the surface of the sphere with coordinates
(x; y; z) is de…ned by the equation of a sphere
p
x2 + y 2 + z 2 = R2 ) x = R2 (y 2 + z 2 ) (7.121)

and you can determine the volume by evaluating the integral


Z Z p 2 2 Z p 2 2 2
R R x R (y +z )
V = p dzdydx: (7.122)
R R 2 x2 0

But here we will use a short cut. If we consider the in…nitesimal volume
to be a cylinder of radius y and thickness (i.e. height) dx; we can write

dv = y 2 dx (7.123)

where y can be related to the radius of the sphere by


p
y = R 2 x2 : (7.124)

The volume will then be


Z R Z R
R3 2 R3
V = y 2 dx = R2 x2 dx = R3 = : (7.125)
0 0 3 3

7.2 Integration in curvilinear coordinates


We now consider the methods of integrations in curvilinear coordinates. We …rst
consider one and two dimensional curvilinear coordinates. For three-dimensional
184 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

coordinates, integrations in cylindrical and spherical coordinates will be consid-


ered. For problems with curved geometries, the simplicity of carrying out the
integrations in curvilinear instead of Cartesian coordinates will also be demon-
strated in this section. The demonstrations will re-consider some of the problems
we saw as examples in the previous section.

7.2.1 One- and two–dimensional curvilinear Coordinates


In polar coordinates a point in the x-y plane with coordinates (x; y) can be
described by (r; ). These two system of coordinates are related by

x = r cos ; y = r sin : (7.126)

The in…nitessimal arc-length in Cartesian coordinates that we saw in the previ-


ous section p
dl = dx2 + dy 2
in polar coordinates can then be written as
q q
2 2 2 2
dl = [d (r cos )] + [d (r sin )] = [cos dr r sin d ] + [sin dr + r cos d ] :

This can be simpli…ed into


q p
2
dl = cos2 + sin2 dr2 + r2 sin2 + cos2 d = dr2 + r2 d 2 :

By factoring dr or rd ; one can establish the relations


s
2
d
ds = 1 + r dr (7.127)
dr
or s
2
1 dr
ds = + 1rd : (7.128)
rd
respectively. Similarly for the in…nitessimal area in Cartesian coordinates,

da = dxdy; (7.129)

Using
@x @x
dx = dr + d = cos ( ) dr r sin ( ) d (7.130)
@r @
and
@y @y
dy = dr + d = sin ( ) dr + r cos ( ) d (7.131)
@r @
one can write

da = dxdy = (cos ( ) dr r sin ( ) d ) (sin ( ) dr + r cos ( ) d ) : (7.132)


7.2. INTEGRATION IN CURVILINEAR COORDINATES 185

There follows that


2 2
da = cos2 ( ) rdrd sin2 ( ) rdrd + sin ( ) cos ( ) (dr) sin ( ) cos ( ) (rd ) :
h i (7.133)
2 2
da = rdrd 2 sin2 ( ) rdrd + sin ( ) cos ( ) (dr) (rd ) : (7.134)
which is given by

da = rdrd : (7.135)

Example 7 Using curvilinear coordinates, repeat Example 6a-d


(a) The total mass of the wire is given by
Z Z
M = dm = L ds (7.136)

In polar coordinates this can be expressed as


s
Z =2 2
1 dr
M= L + 1rd (7.137)
0 rd
Referring to the …gure, we have
dr
r=R) =0 (7.138)
d
so that Z =2
LR
M= L Rd = (7.139)
0 2
(b) The area is given by Z Z Z
A= da = dxdy (7.140)

Noting that the in…nitesimal area da can be written as

da = rdrd (7.141)

we may express the area as


Z =2 Z R
R2
A= rdrd = (7.142)
0 0 4

Example 8 (This is Example 4b) A semi-circular plate of radius R is arranged


in the upper-half x-y plane with its center-of-curvature at the origin. The
plate has a surface-charge density given by (x; y) = Ax2 y, where A is a
constant.
Determine the total charge on the plate using polar coordinates.
Solution:
186 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

7.2.2 3-D Curvilinear coordinates-cylindrical


A point p in space described by the Cartesian coordinates (x; y; z) can be de-
scribed in cylindrical coordinates by (r; '; z). These coordinates are related
by
x = r cos (') ; y = r sin (') ; z = z (7.143)

In cylindrical coordinates an in…nitesimal volume dv can be expressed as

dv = dr (rd') dz = rdrd'dz (7.144)

in…nitesimal area da on the curved surface

da = dz (rd') = rdzd' (7.145)

on the planar surface


da = (dr) (rd') = rdrd' (7.146)
7.2. INTEGRATION IN CURVILINEAR COORDINATES 187

Moment of Inertia:The moment of inertia of an in…nitessimal mass dm in


the in…nitessimal volume dv about a given axis of rotation is given by
Z Z
I = r? dm = r? dv (7.147)

where r? is the perpendicular distance of the mass dm from the given axis
of rotation and is the mass density. In cylindrical coordinates this can be
expressed as Z Z Z Z
I = r? dm = r? rdrd'dz (7.148)

Example 9 (This is the same problem considered in Example 17.4 ) A right-


circular cylinder has a height H, a radius R, and a mass M distributed
uniformly throughout its volume. Find the moment of inertia of the cylin-
der about its axis of symmetry using cylindrical coordinates.

Solution: Moment of inertia of a solid object is given by


Z Z
I= r2 dm = r2 (x; y; z) dxdydz (7.149)
vol vol

The mass is distributed uniformly and therefore the density is constant


Z Z Z
M M
I= r2 dxdydz = r2 dxdydz = r2 dxdydz (7.150)
V R2 H
vol vol vol

The axis of symmetry is the z-axis if we assume the cylinder is vertical.


Then we may express the inertia about this axis as
Z Z
M
I= r2 dm = r2 rdrd'dz (7.151)
R2 H
vol vol

ZH ZR Z2
M 1
I= r3 drd'dz = M R2 (7.152)
R2 H 2
z=0x=0 0

Example 10 Using Cylindrical coordinates prove that the volume of a right-


circular cone of base radius R and height H to be

R2 H
V = : (7.153)
3

Solution: In cylindrical coordinates an in…nitessimal volume dv is given by


Z Z Z
dv = rdrd'dz ) V = rdrd'dz (7.154)
188 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

Noting that for a right circular cone, we have

(H z) R
0 < ' < 2 ;0 < r < ; 0 < z < H; (7.155)
H
and the volume can be written as
Z 2 "Z H Z (H
H
z)R ! #
V = rdr dz d': (7.156)
0 0 0

Upon integrating with respect to ' and then r, one …nds


Z H
(H z)R Z H
r2 H
R2
V =2 dz = H2 2Hz + z 2 dz: (7.157)
0 2 0 H2 0

There follows that


" H
#
R2 2 z3
2 R2 H
V = H z Hz + = : (7.158)
H2 3 0 3

7.2.3 3-D Curvilinear Coordinate Systems: spherical Co-


ordinates
A point p in space described by the Cartesian coordinates (x; y; z) can be de-
scribed in spherical coordinates by (r; ; '). These coordinates are related by

x = r cos (') sin ( ) ; y = r sin (') sin ( ) ; z = r cos ( )

In spherical coordinates an in…nitessimal volume dv can be expressed as

dv = dr (rd ) (r sin ( ) d') = r2 dr sin ( ) d d' (7.159)

in…nitessimal area da on the surface

da = (rd ) (r sin ( ) d') = r2 sin ( ) d d' (7.160)


7.2. INTEGRATION IN CURVILINEAR COORDINATES 189

Example 11 A thin spherical dielectric shell of radius R has a surface-charge


density given by
'
( ; ') = 0 sin ( ) cos (7.161)
4
where 0 is a constant. Find the total charge.

Solution: An in…nitesimal area on the surface of a sphere of radius R is given


by
da = R2 sin ( ) d d' (7.162)
Thus the total surface charge given by
Z Z
Q= ( ; ') da (7.163)

can be expressed as
Z Z
'
Q= 0 sin ( ) cos R2 sin ( ) d d' (7.164)
4

For a full sphere the limits of integration for is (0 ) and for ' is
(0 ' 2 ): Thus the total charge can be determined by evaluating the
integral
Z 2 Z
'
Q= 0R
2
sin2 ( ) cos d d'
0 0 4
Z Z 2
'
= 0R
2
sin2 ( ) d cos d' (7.165)
0 0 4
190 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

Recalling that
Z
sin 2
sin2 ( ) d = = (7.166)
0 2 4 0 2
and Z 2 2
' '
cos d' = 4 sin =4 (7.167)
0 4 4 0

The total charge is found to be


2
Q=2 0R : (7.168)

Example 12 A zone of a sphere is de…ned to be that section of the sphere


that lies between two parallel planes intersecting the sphere. Consider
two planes separated by a distance h that intersect a sphere of radius
R. Prove that the surface area of a zone cut out by these two planes is
A = 2 Rh. Prove this somewhat surprising result (surprising in that it
does not depend on where the slice of the sphere is taken— toward the
equator of the sphere or closer to the top or bottom!)

Solution: An in…nitesimal area da on the surface of the sphere with radius R


is given by
da = R2 sin d d': (7.169)
Then the surface area of the zone is given by
Z Z 2 Z 2

R2 sin d d' = 2 R2 cos ( )


2
A= da =
1
0 1

= 2 R2 [cos ( 1 ) cos ( 2 )] : (7.170)


7.3. HOMEWORK ASSIGNMENT 16 191

where the limits of integration, 1 and 2 represent the angular positions


of the two planes forming the zone. Suppose in Cartesian coordinates, the
upper plane is de…ned by z = z1 and the bottom by z = z1 ; we can easily
see that
z1 z2
cos ( 1 ) = ; cos ( 2 ) = : (7.171)
R R
There follow that

R [cos ( 1 ) cos ( 2 )] = z1 z2 = h; (7.172)

where h is the height of the zone. Thus the area of the surface area of the
zone of a sphere becomes
A = 2 Rh; (7.173)
which is indeed independent of the position of the two planes forming the
zone.

7.3 Homework Assignment 16


1. Consider an area on the x-y plane bounded by the y-axis, the line de…ned
p
by the equation, x = 2; and a curve de…ned by the equation, y = x;.
Using integration in Cartesian coordinates, …nd
(a) the length of the curved side of the area,
(b) the area.
2. The area in the previous problem is rotated about the y-axis to form a
volume with curved surface. Using integration in Cartesian coordinates,
…nd the surface area for
(a) the curved surface,
(b) the plane surface.
3. For an area (A) bounded by the x-axis (y = 0), y = sin (x) ; and x = =2
evaluate the integral
Z Z
I= 6y 2 cos xdxdy:
A

4. For the area (A) bounded by the parabola y = x2 and the straight line
2x y + 8 = 0; evaluate the integral
Z Z
I= xdxdy = 36
A

5. Using double integrals determine the volume bounded above the square
plane formed by the vertices (0; 0),(2; 0),(0; 2), and (2; 2) and below the
plane z = 8 x + y:
192 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS

7.4 Homework Assignment 17


1. A rope of length, L = 10cm, is stretched along the positive y-axis with one
end at the origin. The linear mass density, (y) ; over this rope increases
uniformly from 4 gm=cm at its one end at the origin to 24 gm=cm at the
other end: Find
(a) the linear mass density,
(b) the total mass of the rope,
(c) the center of mass,
(d) the moment of inertia about an axis passing through the center of mass,
(e) the moment of inertia about an axis passing through the heavy end.
2. A solid disk with radius, r = a; and negligible thickness is sitting on the
x-y plane centered at the origin. It has a uniform surface mass density,
0 : Using integration in Cartesian coordinates, …nd

(a) the circumference of the circle,


(b) the area of the disk,
(c) the centroid (or center of mass) for the portion of the disk in the …rst
quadrant,
(d) the moment of inertia of the disk about an axis along the diameter of the
disk,
(e) the centroid for the portion of the side of the disk in the …rst quadrant,
3. The surface charge density ( ) on the area bounded by the x-axis (y = 0)
and curve (y = 16 x2 ) is proportional to y: That means
(y) = ky:
Find the total charge (Q) on the area.
4. A light with an intensity, I0 = 42J=m2 ; is incident on a square silvered
mirror sitting on the x-y plane. The square has side length 2 m and
is centered about the origin. The fraction of the incident light that is
2
re‡ected at a point on the mirror (x; y) is (x y) =4. In other words, the
intensity of the re‡ected light at the point is
2
IR (x; y) = I0 (x y) =4
Find the fraction of the re‡ected light from the entire mirror.
5. Using double integrals determine the volume formed below the surface
de…ned by z = py(x + 2) and above the surface bounded by x + y = 0,
y = 1; and y = x:
7.5. HOMEWORK ASSIGNMENT 18 193

7.5 Homework Assignment 18


1. Repeat Problem 2 in Homework Assignment 16 using Polar coordinates.
1. Consider the volume of the space bounded by the curved surface of the
cylinder de…ned x2 + y 2 = 4, z = 2x2 + y 2 ; and the x-y plane. Find the
volume of the cylinder using integrations in cylindrical coordinates.
3. Consider a solid sphere of radius, r = a and mass, M . The mass is
distributed uniformly over the volume of the sphere. Using integrations
in spherical coordinates, …nd

(a) The centroid of a solid half ball,


(b) the moment of inertia of the of the sphere about an axis passing through
the diameter (along the z-axis)
(c) the volume of the sphere

4. In problem 3 we considered a solid sphere of radius, r = a; and mass,


M that is distributed uniformly over the volume of the sphere. Now let’s
imagine the sphere is hollow and the same mass is distributed over the
surface of the sphere. Using integrations in spherical coordinates, …nd
(a) the surface area of the sphere,

(b) the centroid of the curved area of the upper hemisphere,


(c) the moment of inertia of the of the sphere about an axis passing through
the diameter (along the z-axis)
5. Consider a sphere with constant mass density, 0 and radius r = 2a:
Determine the gravitational attraction force on a unit mass (m = 1) at
the center of the sphere due to the portion of the mass of the sphere
occupying the volume above the plane de…ned by, z = a:
194 CHAPTER 7. INTEGRAL CALCULUS-SCALAR FUNCTIONS
Chapter 8

Vector Calculus

We were introduced to the basic to vector algebra in chapter 3. Here we will


introduced to calculus of vector functions (di¤erentiation and integration). But
…rst we begin by reviewing vector products and their applications in physics

8.1 Vector Products


Two vectors A~ = Ax x ~ = Bx x
^+Ay y^ +Az z^ and B ^+By y^+Bz z^ can be multiplied in
two di¤erent ways one resulting a scalar and the other resulting a third di¤erent
vector.
Dot product (Scalar product): The dot product of A ~ and B ~ is given by

~ B
A ~ = Ax B x + Ay B y + Az B z (8.1)

If the angle between the two vectors is the dot product can be determined
using
~ B
A ~ = A ~ cos
~ B (8.2)
~ and B
Cross product (vector product): the cross product of A ~ results in a
~ given by
third vector C
x^ y^ z^
~ =A
C ~ ~ = det
B Ax Ay Az (8.3)
Bx By Bz

~ = (Ay Bz
C Az B y ) x
^ + (Az Bx Ax Bz ) y^ + (Ax By Ay Bx ) z^: (8.4)
~ is given by
If the angle between the two vectors is ; the magnitude of vector C

~ = A
C ~ ~ = A
B ~ B
~ sin (8.5)

~ is always
and the direction is determined using the right-hand rule. Vector C
perpendicular to the plane formed by the two vectors.

195
196 CHAPTER 8. VECTOR CALCULUS

~ B;
Triple scalar and vector products: Consider the three vectors A; ~ and C
~
in Cartesian coordinates
~ = Ax x
A ~ = Bx x
^ + Ay y^ + Az z^; B ~ = Cx x
^ + By y^ + Bz z^; C ^ + Cy y^ + Cz z^: (8.6)

These three vectors can be multiplied to give a scalar or a vector. The product
of these vectors that leads to a scalar is called a triple scalar product and is
given by
Cx Cy Cz
~ A
C: ~ B ~ = det Ax Ay Az : (8.7)
Bx By Bz
Recalling that interchanging two rows changes only make the value of the de-
terminant negative, one can write

Cx Cy Cz Bx By Bz Ax Ay Az
Ax Ay Az = Ax Ay Az = Bx By Bz : (8.8)
Bx By Bz Cx Cy Cz Cx Cy Cz

Referring to the expression for the triple scalar product, one can easily see that

~ A
C: ~ ~ =
B ~ A
B: ~ ~ = A:
C ~ B~ ~ :
C (8.9)

Taking into account the relation


~
A ~ =
C ~
C ~
A;

one can establish the relation for triple scalar product of three vectors

~ A
C: ~ ~ = B:
B ~ C~ ~ = A:
A ~ B~ ~ :
C (8.10)

These same three vectors can also be multiplied to yield another vector instead
of a scalar. Such multiplication is known as triple vector product that can be
expressed as
A~ B~ C ~ =B ~ A~ C
~ C~ A ~ B~ .

The expression for triple vector product can be memorized using the phrase
"BAC-CAB RULE ".

8.1.1 Physical applications


Work: Suppose a constant force, F~ ; is applied on an object to move it for from
a position ~r1 to a position ~r2 for a displacement, d~ = ~r = ~r2 ~r1 : The angle
between the force and the displacement vectors is : We have seen that the work
done by this force, W; is given by the scalar product of the vectors F~ and d;~

W = F~ d~ cos = F~ d:
~ (8.11)
8.1. VECTOR PRODUCTS 197

For none constant force that depends on the position of the object, F~ = F~ (~r) ;
the work done is determined from an integral form of the scalar product for the
force and displacement vectors given by
Z r2
W = F~ d~r: (8.12)
r1

Angular and linear velocities: Consider an object under a rotational motion.


Suppose the object is rotating with a constant angular velocity, ! ~ ; about some
axis of rotation. Over a time interval, t = t2 t1 ; the position of the object
relative to the axis of rotation has changed from ~r1 to ~r2 and the linear displace-
ment becomes, d~ = ~r = ~r2 ~r1 : The linear velocity, ~v ; of the object, which
is a constant for a constant angular velocity, !~ ; can be determined using the
vector product of the two vectors expressed as

~v = !
~ ~
d: (8.13)

However, when the angular velocity is not a constant and depends on the relative
position of the object, !
~ =!
~ (~r), the average linear velocity is determined using
the line integral,
Z ~r2
~v = !
~ d~r: (8.14)
~
r1

Angular momentum: Suppose the object for which we described the constant
linear velocity, ~v ; and constant angular velocity, !
~ has a mass, m. This mass
has a linear momentum, p~ = m~v : The angular momentum, J; ~ of this object, is
given by the vector product,
J~ = ~r p~: (8.15)
When the linear momentum depends on the position of the object the average
angular momentum becomes
Z
J~ = d~r p~: (8.16)

Torque: The mechanical torque, ~ ; is vector physical quantity that we often


use to describe the rotational motion of an object acted by a force, F~ . Suppose
the force acted on the objected has rotated the object about an axis such that the
position relative to the axis has changed from ~r1 to ~r2 for a total displacement,
d~ = ~r = ~r2 ~r1 : For a constant force the torque about the axis of rotation is
given by the vector product,
~ = d~ F~ : (8.17)
However, when the force is not constant and depends on the position of the
object, the torque given by the integral,
Z ~
r2
~= d~r F~ : (8.18)
~
r1
198 CHAPTER 8. VECTOR CALCULUS

Replace the linear velocity, ~v = !


~ ~r; in the expression for the linear
momentum p~ = m~v ; the angular momentum can be expressed as a triple vector
product. Z
J~ = m d~r (~
! ~r) (8.19)

Magnetic …eld and force: Suppose an object with charge, q; is moving with a
~ The magnetic
constant velocity, ~v ; in a region with a uniform magnetic …eld, B:
force experienced by the object is given by

F~B = q~v ~
B: (8.20)

For a none uniform magnetic …eld, B ~ (r) ; such as the magnetic …eld due to a
wire of length l that carries a current, I: where the magnetic …eld is given by

~l
~ (r) = 0I ~r
B ; (8.21)
4 r3
the magnetic force can be expressed using a triple vector product

~v ~l ~r
q 0I
F~B (~r) = : (8.22)
4 r3

Solid state Physics: Reciprocal lattice vectors {~bi }

~bi = ~aj ~ak


, where (i; j; k) = (1; 2; 3) (8.23)
~a1 (~a2 ~a3 )

{~ai } is called a lattice vector. It is used to describe crystal structure in solids.


The reciprocal lattice vectors are useful to study waves in solids (lattice waves)
known as phonons. These waves are described by the wave vector, ~k; expressed
in terms of the reciprocal vectors.

8.1.2 Vectors derivatives


Next we shall see how we carry out the derivatives of a vector functions. To
this end, let’s consider time dependent scalar function, c (t) ; and two vector
functions that can be expressed in Cartesian coordinates as,

~ (t) = Ax (t) x
A ~ (t) = Bx (t) x
^ + Ay (t) y^ + Az (t) z^; B ^ + By (t) y^ + Bz (t) z^: (8.24)

Using these vectors we can establish the following derivatives with respect to
time:
Derivative of a vector:

~
dA dAx dAy dAz
= x
^+ y^ + z^: (8.25)
dt dt dt dt
8.1. VECTOR PRODUCTS 199

Derivative of a vector multiplied by a scalar:

d ~
cA ~ dc + dA c
~ =A (8.26)
dt dt dt
Derivative of scalar vector product:

d ~ ~ ~ ~
~ dB + dA B
A B =A ~ (8.27)
dt dt dt
Derivative of vector product:

d ~ ~
dB ~
dA
A ~ =A
B ~ + ~
B (8.28)
dt dt dt
Example 8.1 The position of an object at time, t; is described by the two
dimensional vector, ~r (t) : This vector, using Polar coordinates, (r; ) ; can
be expressed as
~r (t) = r cos ( ) x
^ + r sin ( ) y^: (8.29)
Find the general expressions for the velocity, ~v (t) ; and acceleration, ~a (t) ;
and express your answer in Polar Coordinates.
Solution:
The velocity: ~v (t) ; is given by
d~r d d
~v (t) = = (r cos ( )) x
^+ (r sin ( )) y^ (8.30)
dt dt dt
Assuming the general case where both r and are time dependent, upon
di¤ erentiating, we …nd
dr d cos ( ) dr d sin ( )
~v (t) = cos ( ) x
^+r x
^+ sin ( ) y^ + r y^
dt dt dt dt
dr d dr d
) ~v (t) = cos ( ) x
^ r sin ( ) x ^+ sin ( ) y^ + r cos ( (8.31)
) y^
dt dt dt dt
that can put in the form
dr d
~v (t) = (cos ( ) x
^ + sin ( ) y^) + r ( sin ( ) x
^ + cos ( ) y^) : (8.32)
dt dt
Noting that in two-dimensional Polar coordinates, the unit vectors along
the radial direction (i.e. along ~r) is

d~r d~r
r^ = = cos ( ) x
^ + sin ( ) y^; (8.33)
dr dr
and along the angular direction is

^ = d~r d~r
= sin ( ) x
^ + cos ( ) y^; (8.34)
d d
200 CHAPTER 8. VECTOR CALCULUS

the instantaneous velocity can be put in takes the form


dr d
~v (t) = r^ + r^: (8.35)
dt dt
In terms of the angular speed, !;
d
!= (8.36)
dt
and the radial speed
dr
vr = ; (8.37)
dt
the instantaneous velocity can be expressed as

~v (t) = vr r^ + !r^: (8.38)

Further more, using the equation relating the linear tangential and the
angular speeds,
vt = !r; (8.39)
the instantaneous can also be written as

~v (t) = vr r^ + vt ^: (8.40)

When the magnitude of the radial displacement is constant (i.e. the motion
is in a circular orbit), we have
dr
vr = =0 (8.41)
dt
and the instantaneous velocity becomes

~v (t) = vt ^ = !r^: (8.42)

This is in agreement with what we learned in into physics I. Acceleration:


~a(t) is given by the time derivative of the instantaneous velocity,

d~v d dvr d^
r dvt ^ d^
~a (t) = = vr r^ + vt ^ = r^ + vr + + vt (8.43)
dt dt dt dt dt dt
Noting that as it can be noted in the expressions, both unit vectors are
time dependent and we have

d^
r d d^ d ^
= (cos ( ) x
^ + sin ( ) y^) = ( sin ( ) x
^ + cos ( ) y^) = = !^
dt dt dt dt
(8.44)
and
d^ d d^
= ( sin ( ) x
^ + cos ( ) y^) = ( cos ( ) x
^ sin ( ) y^)
dt dt dt
d^ d
) = r^ = !^ r: (8.45)
dt dt
8.1. VECTOR PRODUCTS 201

so that the acceleration can be put in the form


dvr dvt ^ dvr
~a (t) = r^ + vr ! ^ + vt !^
r= vt ! r^
dt dt dt
dvt ^:
+ vr ! + (8.46)
dt
In terms of the magnitude of the acceleration in the radial and angular
directions
dvr d2 r
ar = = 2 (8.47)
dt dt
and
dvt d (r!) dr d!
at = = =! + r; (8.48)
dt dt dt dt
the instantaneous acceleration becomes

~a (t) = (ar vt !) r^ + (vr ! + at ) ^: (8.49)

This can also be put in the form

d2 r dr ^;
~a (t) = r! 2 r^ + 2 !+r (8.50)
dt2 dt
where we used
dr d! d2 r
at = ! + r; ar = 2 ; vt = !r (8.51)
dt dt dt
and the angular acceleration de…ned by

d2
= (8.52)
dt2
For a circular orbit where r is constant of motion (i.e. ~r = R^
r), we have

dr d2 r
= 0; 2 = 0 (8.53)
dt dt
and
~v (t) = vt ^ = !R^ (8.54)
so that
vt2
~a (t) = R! 2 r^ + R ^ = r^ + R ^ = ~ac + ~ajj (8.55)
R
where
vt2
~ac = r^ (8.56)
R
and
d!
~ajj = R ^ = R ^ (8.57)
dt
are the centripetal and tangential accelerations, respectively.
202 CHAPTER 8. VECTOR CALCULUS

8.2 Di¤erential Operators


8.2.1 The gradient operator and directional derivative
In the previous section we considered the derivative of scalar and vector func-
tions of one variable, time t. In most physical problems we often encounter
products of scalars and vectors that are functions of two or more variables.
Suppose we are given the scalar function in Cartesian coordinates, f (x; y; z).
We recall that the in…nitesimal change in this function, df (the total di¤eren-
tial), is given by
@f @f @f
df = dx + dy + dz: (8.58)
@x @y @z
~ and B
In view of the expression for scalar vector product of two vectors A ~ in
Cartesian coordinates

~ B
A ~ = (Ax x^ + Ay y^ + Az z^) (Bx x
^ + By y^ + Bz z^) = Ax Bx + Ay By + Az Bz ;
(8.59)
one can re-write df as

@f @f @f
df = (dx^
x + dy y^ + dz z^) x
^+ y^ + z^ (8.60)
@x @y @z

Noting that the di¤erential for the position vector in Cartesian coordinates,

~r = x^
x + y y^ + z z^;

is given by
d~r = dx^
x + dy y^ + dz z^; (8.61)
one can put df in the form

@f @f @f @ @ @
df = d~r x
^+ y^ + z^ = d~r x
^+ y^ + z^ f; (8.62)
@x @y @z @x @y @z

where we factored out the function f to the right side of the partial di¤eren-
tial operators. The operators in the bracket are represented by the gradient
operator,O
~;
@ @ @
O
~ =x ^ + y^ + z^ : (8.63)
@x @y @z
The gradient operator operates on a scalar function, f (x; y; z) to generate a
vector function
@f @f @f
O
~f = x
^+ y^ + z^: (8.64)
@x @y @z
Thus the in…nitesimal change, df , in the function f (x; y; z) becomes

df = O
~ f dr: (8.65)
8.2. DIFFERENTIAL OPERATORS 203

The directional derivative: The function, f (x; y; z); de…nes a surface in three
dimensional space. Suppose this function is a constant, f (x; y; z) = C. Then
the total di¤erential for the function, df; becomes
~
df = rf d~r = 0; (8.66)

indicating that the scalar product of the vectors rf ~ and d~r is zero and the
vectors are perpendicular to one another at the point (x; y; z). Since d~r =
dx^
x + dy y^ + dz z^; is an in…nitesimal displacement on the surface de…ned by
the function, f (x; y; z); the vector d~r is tangent to the surface at the point.
~ must be a vector normal to the surface at the same point.
Therefore, rf;
Now let’s consider an in…nitesimal displacement, d~s; on the surface de…ned
by the function, f (x; y; z). This displacement can be expressed as

d~s = ds^
n = ds (nx ; ny ; nz ) ; (8.67)

where
n
^ = nx x
^ + ny y^ + nz z^ (8.68)
is a unit vector along the direction of d~s: Thus one can write

d~s = ds^
n = d~r; ) ds (nx ; ny ; nz ) = (dx; dy; dz)

there follows that


dx dy dz
dsnx = dx; dsny = dy; dsnz = dz ) nx = ; ny = ; nz = : (8.69)
ds ds ds
For a function f (x; y; z); the directional derivative is the derivative along the
direction n
^ . It is given by

df @f dx @f dy @f dz
= + + : (8.70)
ds @x ds @y ds @z ds

Using the results above, we may write

df @f @f @f
= nx + ny + nz
ds @x @y @z
@f @f @f
= x
^+ y^ + z^ (nx x
^ + ny y^ + nz z^) (8.71)
@x @y @z

and the directional derivative can then be expressed as

df
= rf n
^;
ds

where n^ is the unit vector along the direction of the vector S ~ that is tangent
to the surface de…ned by the function, f (x; y; z) at a point on the surface with
coordinates (x; y; z).
204 CHAPTER 8. VECTOR CALCULUS

The Gradient Operator in curvilinear Coordinates: In the previous chapter


we have introduced to two curvilinear coordinate systems, namely Cylindrical
and spherical coordinates system. Using the relation between the Cartesian
coordinates and these curvilinear coordinates, it can be derived that the gradient
in Cylindrical Coordinates is given by

@f '
^ @f @f
rf = r^ + + z^ (8.72)
@r r @' @z

and in Spherical Coordinates by

@f ^ @f '
^ @f
rf = r^ + + ; (8.73)
@r r@ r sin @'

where (^ ^ ; z^) and r^; ^; '


r; ' ^ are the unit vectors in Cylindrical and spherical
coordinates, respectively. Note that with the exception of z^; these unit vectors
are not constant unlike the Cartesian unit vectors.

Example 8.2 A point charge Q is located at the origin. The electrostatic po-
tential at a point P a distance r from the point charge is given by

kQ
(r) = ; (8.74)
r
where k is the Coulomb’s law constant
1
k= ; (8.75)
4 0

where 0 is the electrical permittivity of a free space. In classical electro-


dynamics, the electric …eld is related to the electrostatic potential, (r) ;
by the equation
E~ (~r) = r (r) : (8.76)
Find an expression for the electric …eld at point P using

(a)
@ 1 @ '
^ @
r = r^ +^ + (8.77)
@r r@ r sin @'

(b)
@ @ @
O= x
^+ y^ + z^ (8.78)
@x @y @z

(c) Find the directional derivative of the electrostatic potential at ( 1; 2; 1) in


~ = 2^
the direction S x 2^ y + z^

Solution:
8.2. DIFFERENTIAL OPERATORS 205

(a) Since the electric potential depends only on r, using spherical coordinates
is easier than Cartesian. Using the gradient in spherical coordinates, one
can then write the electric …eld as

@ kQ 1 @ kQ 1 @ kQ
~ (~r) =
E r (r) = r^ +^ +'^
@r r r@ r r sin @' r
~ (~r) = @ kQ kQ
)E r^ = 2 r^ (8.79)
@r r r

(b) In Cartesian coordinates, the electric potential is given by

kQ
(r) = ; (8.80)
r (x; y; z)

where p
~r = x^
x + y y^ + z z^ ) r = x2 + y 2 + z 3 : (8.81)
Then we may write

~ (~r) = @ (r) @ (r) @ (r)


E r (r) = x
^+ y^ + z^ : (8.82)
@x @y @z

Noting that the partial derivative, for example with respect to x; can be
determined as
@ (r) d (r) dr d kQ d hp 2 i
= = x + y2 + z3
@x dr dx dr r dx
@ (r) kQ x kQ x kQ
) = p = = x: (8.83)
@x r 2 x2 + y 2 + z 3 r2 r r3

Similarly,the partial derivatives with respect to y and z, we have

@ (~r) kQ y kQ @ (~r) kQ z kQ
= = y; = = z: (8.84)
@y r2 r r3 @z r2 r r3

so that the gradient of (~r) becomes

kQ kQ kQ kQ kQ
r (~r) = x^
x y y^ z z^ = (x^
x + y y^ + z z^) = ~r:
r3 r3 r3 r3 r3
(8.85)
Noting that the unit vector along ~r is

~r
r^ = (8.86)
r
the electric …eld is found to be

~ (~r) = kQ
E r (~r) = r^: (8.87)
r2
206 CHAPTER 8. VECTOR CALCULUS

(c) We are given


~ = 2^
S x 2^
y + z^ (8.88)
and we want to …nd
d (r)
= r (r) n
^ : (8.89)
ds (1;2; 1)

Here n^ is the unit vector along the vector S~ tangent to the surface de…ned
by the electrostatic potential function (r) at a point ( x; y; z). This unit
vector is given by
S~ 2 2 1
n
^= = x ^ y^ + z^: (8.90)
S~ 3 3 3

We are interested in the change in the potential, (r) ; along this direction
at a point, (1; 2; 1) ; on the surface. This is given by the directional
derivative
d (r)
= r (r) n ^ : (8.91)
ds (1;2; 1)

Using the result we obtained in part (a) or (b) and the unit vector, n
^ ;one
can then write
d (~r) kQ 2 2 1
= (x^
x + y y^ + z z^) x
^ y^ + z^ (8.92)
ds r3 3 3 3 (1;2; 1)

d (~r) kQ 2 2 1
= x y+ z (8.93)
ds r3 3 3 3 (1;2; 1)

Using the expression for r in Cartesian coordinates, we have

d (~r) kQ 2 2 1
= 3=2
x y+ z (8.94)
ds (x2 + y2 + z2) 3 3 3
(1;2; 1)

so that the directional derivative for the electrostatic potential becomes


d (~r) kQ 2 4 1 kQ
= p = p : (8.95)
ds 6 3 3 3 6

Exercise: Do you thing we can …nd the angle between the electric …eld vector
E~ (~r) at point ( 1; 2; 1) ( if we assume KQ = 1)and S?
~

8.2.2 The Divergence, the Curl, and the Laplacian


We have seen the gradient operator on a scalar function. It can provide infor-
mation regarding how the scalar function changes along a given direction. Now
we will consider how the gradient operator acts on a vector functions, such as
the electric …eld due to a point charge we determined in the previous example,
8.2. DIFFERENTIAL OPERATORS 207

gravitational …eld due to a given mass, or magnetic …eld due to a current carry-
ing wire. For vector functions the gradient operator acts in two di¤erent ways
to describe the divergence and the curl (rotation) of the vector function. For a
vector function
~ (x; y; z) = Vx (x; y; z) x
V ^ + Vy (x; y; z) y^ + Vz (x; y; z) z^ (8.96)
~ (x; y; z) ; denoted by r V
the divergence of V ~ is given by

~ = @Vx + @Vy + @Vz .


r V (8.97)
@x @y @z
Note that the divergence of a vector function results in a scalar function. For the
~ (x; y; z) ;, is denoted by r V
curl of the vector, V ~ . In Cartesian Coordinates,
~
r V is given by the determinant expression

x
^ y^ z^
r ~ = det
V @ @ @
(8.98)
@x @y @z
Vx Vy Vz

that leads to

~ = @Vz @Vy @Vx @Vz @Vy @Vx


r V x
^+ y^ + z^: (8.99)
@y @z @z @x @x @y
~ (x; y; z) is expressible as a gradient of some scalar
Suppose the vector …eld V
function, (x; y; z) ;
~ (x; y; z) = r (x; y; z) ;
V (8.100)
the divergence of this vector becomes
~ = r [r (x; y; z)] = (r r) (x; y; z) = r2 (x; y; z) ;
r V (8.101)

where
@2 @2 @2
r2 = + + (8.102)
@x2 @y 2 @z 2
is called the Laplacian operator. As an example, let’s consider the electric …eld
vector, which is given by
E~ (~r) = r (~r) : (8.103)
~ (~r) leads to
The divergence of E

~ (~r) = @2 @2 @2
r E r r (x; y; z) = + + = r2 (x; y; z) :
@x2 @y 2 @z 2
(8.104)
In classical electrodynamics the divergence of the electric …eld is related to the
volume charge density, (x; y; z) ; by the di¤erential form of Gauss’law,

~ (~r) = (x; y; z)
r E ; (8.105)
0
208 CHAPTER 8. VECTOR CALCULUS

that can be expressed using the Laplacian as

~ (~r) = @2 @2 @2 (x; y; z)
r E r r (x; y; z) = 2
+ 2 + 2 =
@x @y @z 0
2
)r (x; y; z) = F (x; y; z) : (8.106)
where
(x; y; z)
F (x; y; z) = : (8.107)
0
Equation (8.106) is known as Poison’s equation. In a free space where there is
no charge, (x; y; z) = 0; we …nd
r2 (x; y; z) = 0; (8.108)
which is known as Laplace’s equation. Often you encounter Poison’s and Laplace’s
equations in classical electromagnetism and mechanics theories.
You also encounter, more frequently, in most branches of theoretical physics
that could involve the gradient, the divergence, and the curl of a scalar and/or
a vector functions. Thus it worth to remember the following relations:
~ ; multiplied
(a) The divergence of a vector and a scalar : For a vector function, V
by a scalar functions f; the divergence can be expressed as
~
r: f V ~ rf + f r:V
=V ~ (8.109)

(b) The curl of the curl : In view of the relation for triple vector product
~
A ~
B ~ =B
C ~ A
~ C
~ ~ A
C ~ B
~ (8.110)

~ ; one can establish the


for the curl of the curl of the vector function, V
relation
r r V ~ =r r V ~ ~:
r2 V (8.111)
that involves the divergence and the Laplacian of the a vector function.
Thus for the Laplacian of a vector function one can write
~ =r r V
r2 V ~ r r ~
V

~ ; multiplied by
(c) The curl of a vector and a scalar : For a vector function, V
a scalar functions f; the curl can be expressed as

r ~
fV = (rf ) ~ +f r
V ~
V =f r ~
V ~
V rf (8.112)

Example 8.3 Consider two functions— a scalar function and a vector function—
de…ned as follows:
~ (x; y; z) = (xy; yz; zx) :
f (~r) = xyz; V (8.113)
At the point (1; 1; 1), evaluate the following quantities :
8.2. DIFFERENTIAL OPERATORS 209

(a) rf
~
(b) r V

(c) r ~
V
(d) r2 f
~
(e) r2 V

~
(f ) r2 f V

Solution:
(a) The gradient of the scalare function, f , given by

@f @f @f
rf = x
^+ y^ + z^: (8.114)
@x @y @z

For f (x; y; z) = xyz;

@ (xyz) @ (xyz) @ (xyz)


rf = x
^+ y^ + z^ = yz x
^ + xz y^ + xy^
z: (8.115)
@x @y @z

Upon evaluating this at (x; y; z) = (1; 1; 1); we …nd

rf = x
^ + y^ z^: (8.116)

~ = (Vx ; Vy ; Vz ), is given by
(b) The divergence of the vector function, V

~ = @Vx + @Vy + @Vz :


r V (8.117)
@x @y @z
Noting that
Vx = xy; Vy = yz; Vz = xz
one …nds

~ = @ (xy) + @ (yz) + @ (xz) = y + z + x:


r V (8.118)
@x @y @z

At the point (1; 1; 1); the divergence is found to be

~ = 1:
r V (8.119)

(c) For r ~ ; we have


V

x
^ y^ z^
r ~ = det
V @ @ @
(8.120)
@x @y @z
xy yz xz
210 CHAPTER 8. VECTOR CALCULUS

that leads to

~ = @ (xz) @ (yz) @ (xy) @ (xz)


r V x
^+ y^
@y @z @z @x
@ (yz) @ (xy)
+ z^
@x @y
)r ~ = (0
V y) x
^ + (0 z) y^ + (0 x) z^ = y^
x z y^ x^
z : (8.121)

Evaluating this at the point (1; 1; 1);

r ~ =x
V ^ y^ z^: (8.122)

(d) For the Laplacian of the scalar function, f ,


@2f @2f @2f
r2 f = + + (8.123)
@x2 @y 2 @z 2
we …nd
@ 2 (xyz) @ 2 (xyz) @ 2 (xyz)
r2 f = + + = 0: (8.124)
@x2 @y 2 @z 2
Noting that
r2 f = r rf; (8.125)
one can also …nd the same result
@ (yz) @ (xz) @ (xy)
r2 f = r (yz x
^ + xz y^ + xy^
z) = + + = 0; (8.126)
@x @y @z
using the expression for the gradient of the function determined in (a).
(e) Here we want to determine the Laplacian of a vector function instead of a
scalar function. In such cases, one should evaluate
~
@2V ~
@2V ~
@2V
~ =
r2 V + + (8.127)
@x2 @y 2 @z 2
where
~ = Vx x
V ^ + Vy y^ + Vz z^:
Thus for
~ = xy^
V x + yz y^ + zx^
z
one …nds
~
@2V @ 2 (xy^
x + yz y^ + zx^
z)
= =0 (8.128)
@x2 @x 2

~
@2V @ 2 (xy^
x + yz y^ + zx^
z)
= =0 (8.129)
@y 2 @y 2
~
@2V @ 2 (xy^
x + yz y^ + zx^
z)
= =0 (8.130)
@z 2 @z 2
8.2. DIFFERENTIAL OPERATORS 211

The Laplacian becomes


~ = 0:
r2 V (8.131)
We can also apply the relation

~ =r r V
r2 V ~ r r ~
V

along with the results we obtained earlier


~ =y+z+x
r V (8.132)

and
~ =r
U ~ =
V y^
x z y^ x^
z: (8.133)
Noting that

~ @ @
r r V = r (y + z + x) = x
^ (y + z + x) + y^ (y + z + x)
@x @y
(8.134)
@f ~
+^
z (y + z + x) ) r r V =x
^ + y^ + z^
@z
and
x
^ y^ z^
r r ~
V =r ~ = det
U @ @ @
(8.135)
@x @y @z
y z x
that can be simpli…ed into

~ @ (y z) @ ( x) @ ( y) @ ( x)
r r V = x
^+ y^
@y @z @z @x
@ ( z) @ ( y) ~ =x
+ z^ ) r r V ^ + y^ + z^: (8.136)
@x @y

Using these results, one also …nds

~ =r r V
r2 V ~ r r ~
V =0 (8.137)

(f ) Noting that

~
r2 f V ^r2 (f Vx ) + y^r2 (f Vz ) + z^r2 (f Vz )
=x (8.138)

and
2 2 2
@ 2 (xy) z @ 2 (xy) z @ 2 (xy) z
2
r (f Vx ) = + + = 2y 2 z + 2x2 z
@x2 @y 2 @z 2
(8.139)
212 CHAPTER 8. VECTOR CALCULUS

2 2 2
@ 2 (yz) x @ 2 (yz) x @ 2 (yz) x
2
r (f Vy ) = + + = 2z 2 x + 2y 2 x
@x2 @y 2 @z 2
(8.140)
2 2 2 2 2 2
@ (zx) y @ (zx) y @ (zx) y
r2 (f Vz ) = + + = 2z 2 y + 2x2 y
@x2 @y 2 @z 2
(8.141)
we …nd
~
r2 f V ^ 2y 2 z + 2x2 z + y^ 2z 2 x + 2y 2 x + z^ 2z 2 y + 2x2 y (8.142)
=x

Using Mathematica

The gradient, the divergence, the Laplacian, and the curl that we have seen
so far is for scalar and vector functions expressed in terms of Cartesian coordi-
nates. Depending on the problem we want to solve, sometimes it is convenient
to evaluate these quantities in curvilinear coordinates by expressing the func-
tions using the corresponding transformations for the Cartesian to the conve-
nient curvilinear coordinates. Thus it is important to know the expression for
these quantities in the two curvilinear coordinates we studied (cylindrical and
spherical coordinates). These expressions can be derived from its corresponding
expressions in Cartesian. For now we list these expressions in cylindrical and
spherical coordinates.
(a) Cylindrical coordinates (r; '; z): For a scalar function, f (r; '; z); the gradi-
ent and Laplacian are given by
@f 1 @f @f
rf = r^ +'
^ + z^ (8.143)
@r r @' @z
8.2. DIFFERENTIAL OPERATORS 213

and
1 @ @f 1 @2f @2f
r2 f = r + 2 2
+ 2; (8.144)
r @r @r r @' @z
respectively. For a vector function,

~ (r; '; z) = Vr (r; '; z) r^ + V' (r; '; z) '


V ^ + Vz (r; '; z) z^;

the divergence and the curl can determined using

~ = 1 @ (rVr ) + 1 @ (V' ) + @Vz


r V (8.145)
r @r r @' @z

and

r ~ = r^ 1 @Vz
V
@V'
+'
^
@Vr @Vz
+ z^
1 @
(rV' )
@Vr
:
r @' @z @z @r r @r @'
(8.146)
respectively.

(b) Spherical coordinates (r; ; '): For a scalar function, f (r; ; '); the gradient
can be expressed as

@f ^ 1 @f 1 @f
rf = r^ + +'
^ ; (8.147)
@r r@ r sin @'

and the Laplacian as

1 @ @f 1 @ @f 1 @2f
r2 f = r2 + sin ( ) + 2 :
r2 @r @r r2 sin ( ) @ @ r2 sin ( ) @'2
(8.148)
On the other hand for a vector function,

r + V (r; '; )^ + V' (r; '; )^


V (r; '; ) = Vr (r; '; )^ ';

the divergence and the curl are given by

~ = 1 @ r2 Vr +
r V
1 @
(sin ( ) V ) +
1 @V'
(8.149)
r2 @r r sin ( ) @ r sin ( ) @'

and

1 @ @V 1 @Vr 1 @
r ~ = r^
V (sin ( ) V' ) +^ (rV' )
r sin ( ) @ @' r sin ( ) @' r @r
1 @ @Vr
+^' (rV ) ; (8.150)
r @r @

respectively.
214 CHAPTER 8. VECTOR CALCULUS

Example 8.4 Consider an in…nitely long cylindrical wire of radius a carrying


a current I along the positive z-direction and this current is uniformly
distributed.

The magnetic …eld is given by

~ =
B 0 Ir=2 a2 '
^ for r < a (8.151)

at a distance r from the wire in cylindrical coordinates ( r; '; z); 0 is


magnetic permeability of a free space.

(a) Using parametric plot show that the magnetic …eld lines form a concentric
circle.

~ 6=
(b) Show that magnetic …eld vector is a rotational vector …eld (i.e. the r B
0) inside the cylinder

~ = 0:
(c) Show that r B

Solution:

(a)
8.2. DIFFERENTIAL OPERATORS 215

(b) We note that the components of the magnetic …eld in cylindrical coordinates
are
Ir
Br = 0; B' = 0 2 ; Bz = 0:
2 a
Then using the expression for the curl of a vector in cylindrical coordinates

~ = r^ 1 @Bz
r B
@B'
+'
^
@Br @Bz
+ z^
1 @
(rB' )
@Br
r @' @z @z @r r @r @'
(8.152)
we …nd

~ @B' 1 @
r B = r^ + z^ (rB' )
@z r @r
@ 0 Ir 1 @ Ir
= r^ 2
+ z^ r 0 2 (8.153)
@z 2 a r @r 2 a

2
)r ~ = z^ 1 @
B 0 Ir
= 0I
z^ (8.154)
r @r 2 a 2 a2
216 CHAPTER 8. VECTOR CALCULUS

Noting the current density, which is de…ned as


Current
J= ; (8.155)
Area
is
I
J= (8.156)
a2
we may write
)r ~ =
B 0J z
^ = ~
0J (8.157)
which is known as Ampere’s law.
(c) The divergence in cylindrical coordinates is given by

~ = 1 @ (rVr ) + 1 @ (V' ) + @Vz


r B
r @r r @' @z
and using
0 Ir
Br = 0; B' = ; Bz = 0:
2 a2
we …nd
~ =0
r B
[Electricity and Magnetism I...PHYS 4310]
Example 8.5 In example 8.2 we have shown that the electric …eld of a point
charge, Q, position at the origin, is given

~ (~r) = kQ
E r (~r) = r^: (8.158)
r2
Show that this electric …eld vector is irrotational vector …eld.
Solution:
The electric …eld vector in spherical coordinates has components
kQ
Er = ; E = E' = 0
r2
then the curl of the electric …eld vector in spherical coordinates

~ = r^ 1 @ @E
r E (sin ( ) E' ) (8.159)
r sin ( ) @ @'
1 @Er 1 @ 1 @ @Er
+^ (rE' ) + '
^ (rE ) (8.160)
r sin ( ) @' r @r r @r @
becomes
r ~ = 0:
E (8.161)
This means that the electric …eld vector is irrotational vector …eld.
Exercise: Find the divergence of the electric …eld and try to make a physical
interpretation for the result.
8.3. SINGLE VECTOR FUNCTION INTEGRALS 217

8.3 Single vector function integrals


8.3.1 Line vector integrals
Example 8.6 Consider the vector …eld
~ r) = x2 y^
A(~ x + xy 2 y^ + a3 e y
x cos ( ) z^; (8.162)

where a; ; and are known constants. Evaluate the line integral


I
I= ~ r) d~r
A(~ (8.163)
C

around the curve C shown in the diagram below.

Solution:
The closed line integral can easily be carried out if we divide the integral into
three line integrals corresponding to the three sides ( s1; s2;and s3),
I Z Z Z
~ r) d~r =
A(~ ~ r) d~r +
A(~ ~ r) d~r +
A(~ ~ r) d~r
A(~ (8.164)
C s1 s2 s3

Let’s represent a point on any of these line segments by a vector

~r = x^
x + y y^ (8.165)

so that
d~r = dx^
x + dy y^ (8.166)
218 CHAPTER 8. VECTOR CALCULUS

For a point on side s1

x = 0 ) ~r = y y^ ) d~r = dy y^ (8.167)

for s2 p p
y= 2 ) ~r = x^
x+ 2^
y ) d~r = dx^
x (8.168)
and for s3
p p dx
y= x ) ~r = x^
x+ x^ x + p y^:
y ) d~r = dx^ (8.169)
2 x
The function
~ r) = x2 y^
A(~ x + xy 2 y^ + a3 e y
x cos ( ) z^; (8.170)

also takes di¤ erent expressions on these three sides of the curve. On s1
~ r) = 0
x = 0 ) A(~ (8.171)

on s2
p p p
y= ~ r) = x2 2^
2; z = 0 ) A(~ y + a3 e
x + 2x^ 2
x cos ( ) z^ (8.172)

and on s3
p p
~ r) = x5=2 x
y = x; z = 0 ) A(~ ^ + x2 y^ + a3 e x
x cos ( ) z^ (8.173)

Therefore, using the three line integrals on s1


Z
~ r) d~r = 0
A(~ (8.174)
s1

on s2
Z Z p p
p Z
2
2 p 2
~ r) d~r = 2 2 2 3 8 2
A(~ x 2^
x + 2x^
y dx^
x= 2 x dx = x =
s2 0 0 3 3
0
(8.175)
and on s3
Z Z 0 Z 0
~ r) d~r = dx x2
A(~ x5=2 x
^ + x2 y^ x + p y^ =
dx^ x5=2 + p dx
s3 2 2 x 2 2 x
Z 0 0
1 2 1
x5=2 + x3=2 dx = x7=2 + x5=2
2 2 7 5 2
Z p p
) ~ r) d~r = 16 2 4 2 :
A(~ (8.176)
s3 7 5
we …nd
I p p p p
~ r) d~r = 8 2 16 2 4 2 44 2
A(~ = (8.177)
C 3 7 5 105
8.3. SINGLE VECTOR FUNCTION INTEGRALS 219

8.3.2 Conservative vector …elds


Conservative Vector Fields: a vector …eld
~ (x; y; z) = Vx (x; y; z) x
V ^ + Vy (x; y; z) y^ + Vz (x; y; z) z^ (8.178)

is said to be conservative if any one of the following conditions is satis…ed:

If the line integral over a closed curve is zero independent of the path,
I
~ (x; y; z) d~r = 0:
V (8.179)
C

If the curl of the vector …eld is zero,

r ~ (x; y; z) = 0:
V (8.180)

If the vector …eld can be expressed as a gradient of some scalar function


U (x; y; z);
V~ (x; y; z) = rU (x; y; z): (8.181)
Gravitational and electric …elds are conservative …elds.

Example 8.7 Consider the moon and the earth: Imagine the origin of a spher-
ical coordinate system coincides with the center of the earth. Then the
position of the moon can be described by ( r; ; ') in spherical coordinates.
The gravitational potential energy of the earth and the moon is given by
GM m
U= (8.182)
r
where r is the distance between the earth and the moon, M is mass of the
earth, and m is mass of the moon. (See Fig. 8.1)

Figure 8.1: Moon’s orbit around the earth.


220 CHAPTER 8. VECTOR CALCULUS

(a) Find the gravitational force

F~ = r U (8.183)
on the moon using the gravitational potential energy.
(b) Show that gravitational force is a conservative force using the work done
by the gravitational force when the moon makes one complete revolution
around the earth. I
W = F~ d~r
C

Solution:
(a) The gravitational force is given by

F~ = r U (8.184)
using the gradient in spherical coordinates
@U ^ 1 @U 1 @U
rU = r^ + +'
^ (8.185)
@r r @ r sin @'
we …nd
@U @ GM m GM m
F~ = r^ = r^ = r^ (8.186)
@r @r r r2
(b) The work done is given by
I I
GM m
W = F~ (x; y; z) d~r = r^ d~r (8.187)
C C a2
for a circular orbit r is a constant. That means in spherical coordinates
the position vector
~r = a^
r (8.188)
becomes
d~r = ad^
r: (8.189)
Recalling that in spherical coordinates
~r = r sin ( ) cos (') x
^ + r sin ( ) sin (') y^ + r cos ( ) z^
the unit vector along the radial direction
@~r @~r
r^ =
@r @r
= sin ( ) cos (') x
^ + sin ( ) sin (') y^ + cos ( ) z^ (8.190)
we may write
d^
r = [cos ( ) cos (') d sin ( ) sin (') d'] x
^
+ [cos ( ) sin (') d + sin ( ) cos (') d'] y^ sin ( ) d z^: (8.191)
8.3. SINGLE VECTOR FUNCTION INTEGRALS 221

There follows that

r^ d~r = sin ( ) cos ( ) cos2 (') d sin2 ( ) sin (') cos (') d'
+ sin ( ) cos ( ) sin2 (') d + sin2 ( ) sin (') cos (') d'
sin ( ) cos ( ) d :
) r^ d~r = 0 (8.192)

Then the work done


I
GM m
W = r^ d~r = 0 (8.193)
C a2

8.3.3 Conservative Fields and Exact Di¤erentials


Exact vs Inexact Di¤ erentials: Consider the scalar function U (x; y; z). we recall
that the di¤erential for this function is given by
@U @U @U
dU =dx + dy + dz
@x @y @z
@U @U @U
dU = x
^+ y^ + z^ (dx^x + dy y^ + dz z^) = rU d~r: (8.194)
@x @y @z
Let’s assume that the scalar potential U (x; y; z) be a potential function for some
conservative vector …eld, V ~ (x; y; z) = Vx (x; y; z) x
^ + Vy (x; y; z) y^ + Vz (x; y; z) z^;
~
(i.e. V (x; y; z) = rU (x; y; z)) then we can write

~ d~r ) Vx = @U ; Vy = @U ; Vz = @U :
dU = rU d~r = V (8.195)
@x @y @z
For a conservative vector …eld, we know that

x
^ y^ z^
~ (x; y; z) = det @ @ @ @Vz @Vy @Vx @Vz
r V @x @y @z = x
^+ y^
@y @z @z @x
Vx Vy Vz
@Vy @Vx
+ z^ = 0 (8.196)
@x @y
which leads to
@Vz @Vy @Vx @Vz @Vy @Vx
= 0; = 0; =0
@y @z @z @x @x @y
@Vz @Vy @Vx @Vz @Vy @Vx
) = ; = ; = : (8.197)
@y @z @z @x @x @y
Recalling that we assumed the scalar function U is a function related to a
conservative vector …eld V by
@U @U @U
Vx = ; Vy = ; Vz = (8.198)
@x @y @z
222 CHAPTER 8. VECTOR CALCULUS

we may write

@Vz @Vy @2U @ 2 U @Vx @Vz @2U @2U


= or = ; = or = ;
@y @z @y@z @z@y @z @x @z@x @x@z
@Vy @Vx @2U @2U
= or = : (8.199)
@x @y @x@y @y@x

These relations are known as the Reciprocity Relations. The di¤erential

@U @U @U
dU = dx + dy + dz; (8.200)
@x @y @z

is said to be an exact di¤erential if the Reciprocity Relations are fully satis…ed.


If not, then the di¤erential is inexact. We derived the Reciprocity Relations
assuming the scalar function U is a function that its gradient gives a conservative
~ (x; y; z) = rU (x; y; z)). Therefore, we can conclude that
vector …eld V (i.e. V
a di¤erential
@U @U @U
dU = dx+ dy + dz = f (x; y; z)dx+g(x; y; z)dy +h(x; y; z)dz (8.201)
@x @y @z

is an exact di¤erential if the vector …eld constructed as

~ (x; y; z) = f (x; y; z)^


V x + g(x; y; z)^
y + h(x; y; z)^
z (8.202)

is a conservative vector …eld. Which means we can also say if


I I I I
~ (x; y; z) d~r =
V f (x; y; z)dx + g(x; y; z)dy + h(x; y; z)dz = 0;
C C C C
(8.203)
the di¤erential is exact otherwise it is inexact.

8.4 Multiple vector integrals


8.4.1 Green’s Theorem
Let’s consider two functions P (x; y) and Q(x; y). If these two functions and its
…rst derivatives are continuous for all x and y in the region bounded by the
curve C of area A, then Green’s theorem states
ZZ I
@Q(x; y) @P (x; y)
dxdy = [P (x; y)dx + Q(x; y)dy] ; (8.204)
A @x @y C

where the line integral is counterclockwise around the boundary of the area A
(i.e. on the curve C).

Proof:
8.4. MULTIPLE VECTOR INTEGRALS 223

We note that
ZZ ZZ ZZ
@Q(x; y) @P (x; y) @Q(x; y) @P (x; y)
dxdy = dxdy dydx:
A @x @y A @x A @y
(8.205)
For the area shown in the …gure belowwe may write
ZZ Z b Z yu
@P (x; y) @P (x; y)
dydx = dy dx (8.206)
A @y a yl @y

so that using the relation


Z b
df (t)
dt = f (b) f (a) (8.207)
a dt

we …nd
ZZ Z b
@P (x; y)
dydx = [P (x; yu ) P (x; yl )] dx
A @y a
Z b Z b Z a Z b
= P (x; yu )dx P (x; yl )dx = P (x; yu )dx P (x; yl )dx
a
ZaZ b
I a
@P (x; y)
) dydx = P (x; y)dx: (8.208)
A @y C

Note that the line integral over the closed curve C is the boundary of the
area A and it must be integrated in a counterclockwise direction. Because
the limits of integration in the two line integrals above indicate that we
perform the integration …rst on the upper part of the curve from b ! a and
then from a ! b: Similarly if we perform the integration in the reverse
224 CHAPTER 8. VECTOR CALCULUS

order (i.e. …rst with respect to x and then with respect to y) for the
integral
ZZ Z d Z xr
@Q(x; y) @Q(x; y)
dxdy = dx dy (8.209)
A @x c xl @x
as shown in the …gure below,for the same area then we can write

ZZ Z d Z d Z d
@Q(x; y)
dxdy = [Q(xr ; y) Q(xl ; y)] dy = Q(xr ; y)dy Q(xl ; y)dy
A @x c c c
Z d Z c ZZ I
@Q(x; y)
= Q(xr ; y)dy + Q(xl ; y)dy ) dxdy = Q(x; y)dy
c d A @x C
(8.210)

Here also the line integral over the same curve C is performed in the coun-
terclockwise direction like the previous integration. Therefore, combining
the two relations
ZZ I ZZ I
@P (x; y) @Q(x; y)
dydx = P (x; y)dx; dxdy = Q(x; y)dy;
A @y C A @x C
(8.211)
we …nd
ZZ I
@Q(x; y) @P (x; y)
dxdy = [P (x; y)dx + Q(x; y)dy] : (8.212)
A @x @y C

Example 8.8 Let the area A be the region bounded by the curves y = x2 and
x = y 2 . Verify Green’s theorem for the integral
I
I= 2xy x2 dx + x + y 2 dy (8.213)
C
8.4. MULTIPLE VECTOR INTEGRALS 225

Solution:
Comparing the given integral with
ZZ I
@Q(x; y) @P (x; y)
dxdy = [P (x; y)dx + Q(x; y)dy] : (8.214)
A @x @y C
we want to verify that the result of
I I
I1 = [P (x; y)dx + Q(x; y)dy] = 2xy x2 dx + x + y 2 dy
C C
(8.215)
is equal to
ZZ
@Q(x; y) @P (x; y)
I2 = dxdy
A @x @y
ZZ " #
@ x + y2 @ 2xy x2
= dxdy (8.216)
A @x @y
for the area shown in the …gure belowAnother short cut for the line integral

is the following:
Z
I1 = 2xy x2 dx + x + y 2 dy
Cl
Z
+ 2xy x2 dx + x + y 2 dy
Cu
226 CHAPTER 8. VECTOR CALCULUS

For the lower curve y = x2 ) dy = 2xdx and for the upper curve x =
y 2 ) dx = 2ydy and noting that the two curves intersect at two points
(0; 0) and (1; 1), we can write
Z 1 Z 0
I1 = 2x3 x2 dx + x + x4 2xdx + 2y 3 y 4 2ydy + 2y 2 dy
0 1
(8.217)
Z 1 Z 0
I1 = 2x3 x2 + 2x2 + 2x5 dx + 4y 4 2y 5 + 2y 2 dy (8.218)
0 1
Z 1 Z 0
5 3 2
I1 = 2x + 2x + x dx + 4y 4 2y 5 + 2y 2 dy (8.219)
0 1

1 0
x6 x4 x3 4y 5 y6 2y 3 1
I1 = + + + + = (8.220)
3 2 3 0 5 3 3 1 30
For the area integral
ZZ " # Z 1 Z py
@ x + y2 @ 2xy x2
I2 = dxdy = (1 2x) dxdy
A @x @y 0 y2
Z 1 p Z 1
2 y p
= x x y2 dy = y y y 2 + y 4 dy
0 0
2 3 5 1
2 3=2 y y y 2 1 1 1 1
= y + = + = (8.221)
3 2 3 5 0 3 2 3 5 30

8.4.2 The Stokes’Theorem


Stokes’ Theorem:
I Z
~ d~r =
D r ~ n
D ^ dA: (8.222)
Curve Surf ace A
bounding A

Consider the surface bounded by the curve, C, on the x-y plane as shown in the
…gure below.Using an in…nitesimal area dA in this region one can establish the
relation
n
^ dA = z^dxdy; (8.223)
where n^ = z^ is the unit vector normal to this area. On the curve, an in…nites-
simal displacement can be written as

d~r = dx^
x + dy y^: (8.224)

~ that depends on the x and y coordinates


For a vector D

~ (x; y) = Dx (x; y) x
D ^ + Dy (x; y) y^ + Dz (x; y) z^ (8.225)
8.4. MULTIPLE VECTOR INTEGRALS 227

the curl in Cartesian coordinates is given by


x
^ y^ z^
~ = det @ @ @ @Dz @Dy @Dx @Dz
r D @x @y @z = x
^+ y^
@y @z @z @x
Dx Dy Dz
@Dy @Dx
+ z^ (8.226)
@x @y
One can then write
~ d~r = Dx (x; y) dx + Dy (x; y) dy;
D

and
~ n @Dy @Dx
r D ^ dA = dxdy:
@x @y
Using these two expressions the Stoke’s theorem
I Z
~
D d~r = r ~ n
D ^ dA: (8.227)
Curve Surf ace A
bounding A

can be expressed as
I Z
@Dy @Dx
(Dx dx + Dy dy) = dxdy: (8.228)
Curve Surf ace A @x @y
bounding A

Some Notes about Stokes’ Theorem:


228 CHAPTER 8. VECTOR CALCULUS

1. The curve C bounds the open area A. The integral must therefore be the
same for any area A bounded by the curve C. This means that you can
pick an easy area A if you wish to make life simpler!

2. The integral Z
r ~ n
D ^ dA (8.229)
A

~ through the open area A.


is just the ‡ux of the curl of the vector D

3. The integral I
~ d~r
D (8.230)
C

~ around the closed curve C. If D


is the line integral of the vector D ~ is a
force, then this integral is just the work done by that force around the
curve C. If the force is a conservative force, then the integral is zero.

Recall Green’s Theorem in the x-y Plane: Let’s consider two functions P (x; y)
and Q(x; y). If these two functions and their …rst derivatives are continuous for
all x and y in the region bounded by the curve C of area A, then Green’s theorem
states
ZZ I
@Q(x; y) @P (x; y)
dxdy = [P (x; y)dx + Q(x; y)dy] ; (8.231)
A @x @y C

where the line integral is counterclockwise around the boundary of the area A
(i.e. on the curve C). Note that for

~ = Dx (x; y) x
D ^ + Dy (x; y) y^ = P (x; y)^
x + Q(x; y)^
y (8.232)

we …nd ZZ I
@Dy @Dx
dxdy = [Dx dx + Dy dy] : (8.233)
A @x @y C

This shows that Green’s theorem is the two-dimensional form of Stoke’s theorem.

Example 8.8 Consider the vector …eld


~
D(x; y; z) = 3y; xz; yz 2 (8.234)

and the paraboloid


2z = x2 + y 2 (8.235)
bounded by the plane z = 2.

(a) Evaluate the ‡ux integral directly


Z
r ~ n
D ^ dA (8.236)
A
8.4. MULTIPLE VECTOR INTEGRALS 229

(b) Evaluate the ‡ux integral indirectly by exploiting Stokes’ theorem.


Solution:
~
(a) The curl of the vector D(x; y; z) = 3y; xz; yz 2 is given by

x
^ y^ z^
r ~ = det
D @ @ @
(8.237)
@x @y @z
Dx Dy Dz
becomes
! !
~ = @ yz 2 @ ( xz) @ (3y) @ yz 2
r D x
^+ y^
@y @z @z @x
@ ( xz) @ (3y)
+ z^ = z 2 + x x
^ (3 + z) z^ (8.238)
@x @y
for z = 2, we …nd
r ~ = (4 + x) x
D ^ 5^
z (8.239)
Then the ‡ux becomes

Z Z
r ~ n
D ^ dA = [(4 + x) x
^ 5^
z] n
^ dA
A A
Z 2 Z p
4 x2
= p [(4 + x) x
^ 5^
z ] z^dxdy (8.240)
2 4 x2

which gives
Z Z Z p ! Z
2 4 x2 2 p
r D ~ n
^ dA = 5 dy dx = 10 4 x2 dx
p
A 2 4 x2 2
(8.241)
230 CHAPTER 8. VECTOR CALCULUS

Using the integral

we …nd Z
r ~ n
D ^ dA = 20 : (8.242)
A

(b) Applying Stokes’ theorem


Z I
r ~ n
D ^ dA = ~ d~r:
D (8.243)
Surf ace A Curve
bounding A

we have

~
D(x; y; z) d~r = 3y; xz; yz 2 (dx; dy) = 3ydx xzdy (8.244)

Using the circular curve of radius R = 2 on the plane z = 2 bounding the


circular surface we considered in part a, we may write
I Z 2 h p p i
~ d~r
D = 3 4 x2 dx 2xd 4 x2
C 2
Z 2 h p p i
+ 3 4 x2 dx 2xd 4 x2 (8.245)
2
8.4. MULTIPLE VECTOR INTEGRALS 231

I Z 2 p
~ d~r 2x2
D = 3 4 x2 + p dx
C 2 4 x2
Z 2 p 2x2
+ 3 4 x2 + p dx (8.246)
2 4 x2

I Z 2 p
~ d~r 2x2
D = 3 4 x2 + p dx
C 2 4 x2
Z 2 p 2x2
3 4 x2 + p dx (8.247)
2 4 x2
I Z 2 p
~ d~r = 2x2
D 2 3 4 x2 + p dx (8.248)
C 2 4 x2
By evaluating the integral we …nd
I
~ d~r =
D 20 (8.249)
C

8.4.3 The Divergence Theorem


~ (x; y; z)
For a vector …eld V
ZZZ ZZ
~
r V (x; y; z) dxdydz = ~ (x; y; z) n
V ^ da (8.250)
volum e Surface
enclosing

Example 8.9 Consider the vector …eld


~ (x; y; z) = (xy; yz; zx)
V (8.251)

(a) Compute the ‡ux v through the closed rectangular surface of sides a; b;
and c shown in the …gure by direct application of the de…nition of ‡ux.

(b) Compute v through the same surface indirectly by applying the Divergence
theorem.

Solution:
~ (x; y; z) over a given surface area A is
(a) The ‡ux v for the vector …eld V
de…ned as ZZ
v = ~ (x; y; z) n
V ^ da:

The …gure above have three pairs of surfaces normal to the three axis. The
‡ux through the surfaces normal to the x axis
ZZ ZZ
vx = ~
V (a; y; z) x
^ dydz + ~ (0; y; z) ( x
V ^) dydz (8.252)
232 CHAPTER 8. VECTOR CALCULUS

Z c Z b Z c Z b
) vx = ~ (a; y; z) x
V ^dydz ~ (0; y; z) x
V ^dydz (8.253)
0 0 0 0
Z c Z b Z c Z b
) vx = Vx (a; y; z)dydz Vx (0; y; z)dydz (8.254)
0 0 0 0

Noting that Vx (0; y; z) = 0 and Vx (a; y; z) = ay; we …nd


Z c Z b
ab2 c
vx = aydydz = : (8.255)
0 0 2

Similarly for the surfaces normal to the y axis, we may write


ZZ ZZ
vy = ~
V (x; b; z) y
^ dxdz + ~ (x; 0; z) ( y^) dxdz
V (8.256)

Z c Z a Z c Z a
) vy = Vy (x; b; z)dxdz Vy (x; 0; z)dxdz: (8.257)
0 0 0 0

Noting that Vy (x; 0; z) = 0 and Vy (x; b; z) = bz; we …nd


Z c Z a
abc2
vy = bzdxdz = (8.258)
0 0 2

and Z Z
c a
a2 bc
vz = cxdxdy = : (8.259)
0 0 2
Therefore, the total ‡ux will be

a2 bc ab2 c abc2
v = + + : (8.260)
2 2 2
8.4. MULTIPLE VECTOR INTEGRALS 233

(b) According to Divergence theorem


ZZZ ZZ
r V ~ (x; y; z) dxdydz = ~ (x; y; z) n
V ^ da
volum e Surface
enclosing
(8.261)
Thus the ‡ux can be expressed as
ZZZ
v = r V ~ (x; y; z) dxdydz
volum e
Z c Z bZ a
@Vx @Vy @Vz
= + + dxdydz
0 0 0 @x @y @z
Z c Z b Z a
@ (xy) @ (yz) @ (zx)
= + + dxdydz
0 0 0 @x @y @z
Z c Z b Z a
= (y + z + x) dxdydz (8.262)
0 0 0

which leads to
a2 bc ab2 c abc2
v = + + : (8.263)
2 2 2

8.4.4 More Examples on Divergence and Stokes theorems


The divergence in cylindrical coordinates:

~ = 1 @ (rV ) + 1 @ (V' ) + @ (Vz )


r V (8.264)
r @r r @' @z

Example 8.10 Verify the divergence theorem for the vector …eld

~ = ar^
V r + b^
' + cz z^ (8.265)

(where a; b; and c are constants) and the cylinder shown below.

Solution:

Noting that
Vr = ar; V' = b; V z = cz (8.266)
the divergence

~ = 1 @ (rV ) + 1 @ (V' ) + @ (Vz )


r V (8.267)
r @r r @' @z

becomes

~ = 1 @ r2 a + 1 @ (b) + @ (cz) = 2a + c
r V (8.268)
r @r r @' @z
234 CHAPTER 8. VECTOR CALCULUS

Now applying the Divergence theorem


ZZZ ZZ
r V~ (x; y; z) dxdydz = ~ (x; y; z) n
V ^ da
volum e Surface
enclosing
(8.269)
For the volume integral, we …nd
ZZZ
(2a + c) dxdydz = (2a + c) R2 H: (8.270)
volum e

For the surface integral over the closed cylindrical surface we can write
ZZ ZZ
~
V (r; '; z) n
^ da = ~b (r; '; z) n
V ^ b da
Surface bottom
enclosing
ZZ ZZ
+ ~t (r; '; z) n
V ^ t da + ~s (r; '; z) n
V ^ s da: (8.271)
top side

We note that
n
^b = z^; n
^ t = z^; n
^ s = r^ (8.272)
and
~b (r; '; z) = V
V ~ (r; '; z = 0) = ar^
r + b^
' (8.273)
~t (r; '; z) = V
V ~ (r; '; z = H) = ar^
r + b^
' + cH z^ (8.274)
8.4. MULTIPLE VECTOR INTEGRALS 235

~s (r; '; z) = V
V ~ (r = R; '; z) = aR^
r + b^
' + cz z^ (8.275)
so that
ZZ ZZ
~b (r; '; z) n
V ^ b da = (ar^
r + b^
') ( z^) da = 0 (8.276)
bottom bottom
ZZ ZZ
~t (r; '; z) n
V ^ t da = (ar^
r + b^
' + cH z^) z^da
top top
ZZ
= cHda = c R2 H (8.277)
top

and
ZZ ZZ
~s (r; '; z) n
V ^ s da: = (aR^
r + b^
' + cz z^) r^da
side side
ZZ ZZ
= aRda = aR (2 RH) ) ~s (r; '; z) n
V ^ s da = 2a R2 H:
side side
(8.278)
Therefore the integral over the closed cylindrical surface
ZZ ZZ
~
V (r; '; z) n
^ da = ~b (r; '; z) n
V ^ b da
Surface bottom
enclosing
ZZ ZZ
+ ~t (r; '; z) n
V ^ t da + ~s (r; '; z) n
V ^ s da: (8.279)
top side

becomes
ZZ
~ (r; '; z) n
V ^ da = c R2 H + 2a R2 H: = (2a + c) R2 H
Surface
enclosing
ZZZ
= ~ (x; y; z) dxdydz
r V (8.280)
volum e

The Divergence in spherical Polar Coordinates:

~ = 1 @ r 2 Vr +
r V
1 @
(sin ( ) V ) +
1 @
(V' ) (8.281)
r2 @r r sin ( ) @ r sin ( ) @'
Example 8.11 Verify the divergence theorem for the vector …eld
~ = ~r
V (8.282)
and the sphere shown below.
Solution: For the ‡ux, we …nd
ZZ Z2 Z
~ (r; ; ') n
V ^ da = r r^R2 sin ( ) d d'
R^
surf ace
0 0
ZZ
~ (r; ; ') n
V ^ da = 4 R3
surf ace
236 CHAPTER 8. VECTOR CALCULUS

~ = r^
The divergence of the vector V r

~ = 1 @ r 2 Vr +
r V
1 @
(sin ( ) V ) +
1 @
(V' ) (8.283)
2
r @r r sin ( ) @ r sin ( ) @'

becomes
~ = 1 @ r2 Vr = 1 @ r3 = 3
r V (8.284)
r2 @r r2 @r
Then the integral becomes

ZZZ Z2 Z
~ (r; ; ') d =
r V 3r2 dr sin ( ) d d' = 4 R3 :
volum e
0 0
(8.285)

Example 8.12 Verify Stokes’ theorem for the magnetic …eld vector

r + 3^
~ = 4^
B 2^
' (8.286)

and the curve C shown in the diagram below.

Solution:

We are given the magnetic …eld in spherical coordinates with components

Br = 4; B = 3; B' = 2 (8.287)
8.4. MULTIPLE VECTOR INTEGRALS 237

so that the curl of the magnetic …eld

~ = 1 @ @
r B (sin ( ) B' ) (B ) r^
r sin ( ) @ @'
1 @ 1 @ 1 @ @
+ (Br ) (rB' ) ^ + (rB ) (Br ) ' ^
r sin ( ) @' r @r r @r @
(8.288)
becomes
1 @ 1 @ 1 @
r ~ =
B ( 2 sin ( )) r^ ( 2r) ^ + (3r) '
^
r sin ( ) @ r @r r @r
2 cos ( ) 2 3 ~ = 2 cot ( ) r^ + 2 ^ + 3 '
= r^ + ^ + ' ^)r B ^ (8.289)
r sin ( ) r r r r r
Noting that the in…nitesimal area on the x-y plane in spherical coordinates
can be expressed as
h i
^ da = z^r sin ( ) drd' = cos ( ) r^ sin ( ) ^ r sin ( ) drd'
n (8.290)

then
2 2 3 h i
~ n
r B ^ da = cot ( ) r^ + ^ + '^ cos ( ) r^ sin ( ) ^ r sin ( ) drd'
r r r
(8.291)
~ n 2 2
)r B ^ da = cot ( ) cos ( ) sin ( ) r sin ( ) drd': (8.292)
r r
Noting that on the x-y plane = =2; we have

r ~ n
B ^ da = 2drd': (8.293)
so that the integral Z
r ~ n
B ^ dA (8.294)
A
238 CHAPTER 8. VECTOR CALCULUS

for the surface bounded by the curve shown in the …gure can be written as
Z Z R Z =2
r ~ n
B ^ dA = 2drd' = R : (8.295)
A 0 0

We recall the position vector in spherical coordinates is given by

~r = r^
r ) d~r = dr^
r + rd^
r (8.296)

where

r^ = sin ( ) cos (') x


^ + sin ( ) sin (') y^ + cos ( ) z^: (8.297)

Since the closed curve is on the x-y plane ( = =2), we have

r^ = cos (') x
^ + sin (') y^ ) d^
r = [ sin (') x
^ + cos (') y^] d' = '
^ d'
(8.298)
so that
~r = r^r ) d~r = dr^
r + r'
^ d': (8.299)
Then
~ d~r = Br dr + B' rd' = 4dr
B 2rd' (8.300)
which leads to
I Z Z Z
~
B d~r = (4dr 2rd')+ (4dr 2rd')+ (4dr 2rd') (8.301)
C l1 l2 l3

where l1 and l3 are the straight lines (on this lines d' = 0) and l2 is the
curved line (on this line r = R and dr = 0). Therefore
I Z R Z =2 Z 0
~ d~r =
B 4dr 2Rd' + 4dr = R : (8.302)
C 0 0 R

The two results for the magnetic …eld vector

r + 3^
~ = 4^
B 2^
' (8.303)

verify that Stokes’ theorem


I Z
B~ d~r = r ~ n
B ^ dA: (8.304)
C A

is valid.

8.4.5 Applications of the Divergence and Stoke’s theo-


rems
Gauss’ Law
~ =
r E (8.305)
0
8.4. MULTIPLE VECTOR INTEGRALS 239

Gravity
r ~g = 4 G (8.306)
Ampere’s Law : Ampere’s law states that
I
B~ d~r = 0 Ien (8.307)
C

where B~ is the magnetic …eld due to the current Ien passing through the area
enclosed by the curve C and 0 is the magnetic permeability of free space. The
di¤erential form of Ampere’s law can be obtained if we apply Stokes’theorem.
Using Stokes’theorem we can write
I Z
~ d~r =
B r B ~ n^ dA = 0 Ien (8.308)
C A

If the current passing through the area enclosed by the curve C is described by
~ the total current Ien can be expressed as
the current density J,
Z
Ien = J~ n^ dA (8.309)
A

so that
I Z Z
~ d~r =
B r ~ n
B ^ dA = 0 Ien = 0 J~ n
^ dA
C A A
Z Z
) r ~
B n
^ dA = 0 J~ n
^ dA ) r ~ =
B ~
0 J: (8.310)
A A

This is the di¤erential form of Ampere’s law.


Example 8.13 Who is wrong (Amperes or Stokes)?[From PHYS 4330]: Con-
sider a parallel plate capacitor connected to a battery as shown in the …gure
below
If we consider the ‡at circular area for the Amperian loop shown in the …gure
the current enclosed is just Ienc
Z
Ienc = J~ d~a = I: (8.311)
f lat surf ace

which leads to
I Z Z
~
B d~r = r ~ n
B ^ dA ) J~ n
^ dA = 0 Ien = 0I (8.312)
C A A

On the other hand, if we consider the balloon shaped surface shown in the …gure
there is no current passing through this surface and therefore the current enclosed
is zero Z
Ienc = J~ d~a = 0: (8.313)
balloon surf ace
240 CHAPTER 8. VECTOR CALCULUS

Figure 8.2: A capacitor connected to a dc current source.

which leads to a di¤ erent result


I Z
~
B d~r = r B ~ n^ dA (8.314)
C A
Z
J~ n
^ dA = 0 Ien = 0
A

Who is wrong here? Neither Ampere nor Stoke! However, in the


case considered above the current is not steady because of the build
up of charge on the capacitor. This suggested that Ampere’s Law
need to be …xed so that it can also be used when the current is not
steady and it was …xed by Maxwell. You want to learn more about
this? TAKE THEORETICAL II (PHYS 3160) in spring 2013 and
then E & M I & II (PHYS 4310 & 4330)!!!
Part II

Mathematical Methods in
Physics II

241
Chapter 9

Introduction to the
Calculus of Variations

9.1 Geodesic and stationary points


Geodesic: The curve along a surface which marks the shortest distance between
two neighboring points. Finding geodesics is one of the problems which can be
solved using the calculus of variation.
Stationary point: A point with coordinates, (x0 ; f (x0 ));on a curve de…ned
by the function f (x) is said to be a stationary point when
df (x)
= 0: (9.1)
dx x=x0

Figure 9.1: Stationary points.

Example 9.1 A ball of mass m is kicked from the ground level with an initial
speed, vo ; at an angle o above the horizontal. Find the time, t; at which
the height of the ball, y(t), becomes stationary.

243
244CHAPTER 9. INTRODUCTION TO THE CALCULUS OF VARIATIONS

Solution: We recall that from kinematics of a projectile the motion of the ball
along the y direction is determined by Newton’s second law
dvy dvy
m = mg ) = g ) vy (t) = v0y gt (9.2)
dt dt
Noting that
dy (t)
vy (t) = (9.3)
dt
the value of the time after the ball is kicked that makes the height function
of the ball,y(t), stationary is given by

dy (t) v0y v0 sin ( o )


vy (t) = =0)t= = (9.4)
dt g g

Example 9.2 Geodesic: Consider two points in a x-y plane P1 and P2 . Prove
that the shortest distance between the two points is the distance measured
along a straight line (i.e. show that the geodesic is given by an equation
of a straight line, y (x) = mx + b.)

Figure 9.2: Geodesic and none geodesic paths.

Solution: Let’s consider two points on the x-y plane. Let P1 be (x1 ; y1 ) and P2
be (x2 ; y2 ) : Then the distance between these points is given by the integral
Z (2)
L= ds; (9.5)
(1)

where
s s
p dy
2
dx
2
ds = dx2 + dy 2 = 1+ dx = 1+ dy: (9.6)
dx dy
9.1. GEODESIC AND STATIONARY POINTS 245

We may rewrite this distance as


s
Z (2) 2
dy
L= 1+ dx: (9.7)
(1) dx

Out of the in…nitely many functions that can be used to connect the two
points, we want to determine the one that would give the minimum dis-
tance. Let these function be denoted by Y (x) : From these in…nite number
of functions there is only one function that gives the minimum distance
between the two points. If this function is y (x) ; then we may write Y (x)
in terms of y (x) as

Y (x; ) = y (x) + (x) ; (9.8)

where (x) is an arbitrary function which must satisfy the condition

(x1 ) = (x2 ) = 0 (9.9)

so that at the two end points ( x = x1 = x2 ), we …nd

Y (x; ) = y (x) : (9.10)

Here is the constant of variation. It is this constant that determines by


how much Y (x) di¤ ers from y (x). Now in terms of Y (x), we may write
Z (2) p
L( ) = 1 + Y 0 2 dx; (9.11)
(1)

where
0 dY (x; )
Y = : (9.12)
dx
We are interested in the path that gives the the minimum distance be-
tween the two points (i.e. the geodesic). The necessary condition for the
distance, L ( ) ; to be minimum is that the length function, L ( ) ; must
have a stationary point at ( = 0; L ( = 0)). This requires

dL ( )
= 0; (9.13)
d =0

which leads to
"Z #
(2)
dL ( ) 1 1 dY 0 (x; )
= p (2Y 0 ) dx = 0:
d =0 (1) 2 1 + Y 02 d
=0
(9.14)
Using
Y (x; ) = y (x) + (x) (9.15)
246CHAPTER 9. INTRODUCTION TO THE CALCULUS OF VARIATIONS

we may write

dY dy d
= + ) Y 0 (x; ) = y 0 (x) + 0
(x) (9.16)
dx dx dx
so that
dY 0 ( ) d 0
= [y (x) + 0 (x)] = 0
(x) : (9.17)
d d
There follows that
Y 0 (x; )j =0 = y 0 (x) (9.18)
and
dY 0 ( ) 0
= (x) : (9.19)
d =0

In view of these results, one …nds for the stationary point


Z (2)
dL ( ) 1 1 dY 0 ( )
= p (2Y 0 ) dx
d =0 (1) 2 1 + Y 02 d
=0
Z (2)
y 0 (x) 0 (x)
= p dx = 0: (9.20)
(1) 1 + y 02 (x)

Using integration by parts


Z Z
udv = uv vdu (9.21)

for
0
(x) = dv ) v = (x) ; (9.22)
and !
y0 d y0
u= p ) du = p dx; (9.23)
1 + y0 2 dx 1 + y0 2
we may write the integral as
Z x2 Z !
(2) x2
y 0 0 (x) y0 d y0
p dx = p (x) (x) p dx = 0:
(1) 1 + y0 2 1 + y0 2 x1 dx 1 + y0 2
x1
(9.24)
Due to the conditions
(x1 ) = (x2 ) = 0 (9.25)
the …rst term in the above expression becomes zero. Thus one can write
Z (2) Z x1 !
dL ( ) y 0 0 (x) d y0
= p dx = (x) p dx = 0:
d =0 (1) 1 + y0 2 x1 dx 1 + y0 2
(9.26)
9.2. THE GENERAL PROBLEM 247

Since (x) is an arbitrary function, for the integral to be zero, we must


have !
d y0 y0
p = 0 ) p = c; (9.27)
dx 1 + y0 2 1 + y0 2
where c is a constant. Upon solving for y 0
1=2
c2
y0 = = m: (9.28)
1 c2

Note that we have introduced another constant in terms of the constant c.


There follows that
dy
= m ) y (x) = mx + b; (9.29)
dx
which is equation of a straight line.

9.2 The general problem


In the previous section we saw the application of the calculus of variation to
show that the shortest path connecting two points on a plane (the geodesic) is
a straight line
dy
= m ) y (x) = mx + b: (9.30)
dx
Next we shall consider the application of the calculus of variation to the general
problem. In an Euclidean space a surface is de…ned by the function, F (x; y; z) ;
where it depends on the Cartesian coordinates x; y; and z: Instead of the Euclid-
dy
ean space let’s consider a surface de…ned by the function, F x; y (x) ; y 0 (x) = dx :
This surface could, for example, be a surface in phase space (in Classical me-
chanics) if we replace

dy py
x ! t; y (x) ! y (t) ; y 0 (x) ! y 0 (t) = = vy (t) =
dt m
describing the dynamics of a particle mass, m, moving along the y-direction in
py y^
terms of the parameters (time = t; position = y (t) ; velocity = vy (t) y^ = m ),
where py y^ is the momentum: In classical mechanics, the dynamics of a particle
is determined by an equation derived from Newton’s second law. As we shall
see, this equation can be derived from a more general equation know as the
Euler-Lagrange Equation

@ @F @F
= 0; (9.31)
@x @y 0 @y

where F = F (x; y (x) ; y 0 (x)) is the function that de…nes the surface constructed
by the set of points with coordinates, (x; y (x) ; y 0 (x)):The Euler-Lagrange Equa-
tion is derived by applying the calculus of variation. In general, in the problem
248CHAPTER 9. INTRODUCTION TO THE CALCULUS OF VARIATIONS

that we want to solve applying the calculus of variation, we know the coordi-
nates of two di¤ erent points (x1 ; y (x1 ) ; y 0 (x1 )) and (x2 ; y (x2 ) ; y 0 (x2 )) on the
surface de…ned by F = F (x; y (x) ; y 0 (x)): From the in…nitely many trajectories
that can connect these two points, there is only one trajectory on this surface
that is the shortest (the Geodesic). Finding the Geodesic is the general problem
that can be solved applying the calculus of variation.
The surface is de…ned by the function F (x; y (x) ; y 0 (x)): The distance be-
tween these two points determined by evaluating the integral
Z x2
I= F (x; y (x) ; y 0 (x)) dx: (9.32)
x1

To determine the equation that the function F is governed by so that we …nd the
shortest length joining the two points, let the function for any path connecting
the two points be Y (x) : From these in…nite number of functions there is only
one function that gives the minimum distance between the two points. If this
function is y (x) ; then we may write Y (x) in terms of y (x) as

Y (x; ) = y (x) + (x) ; (9.33)

where (x) is an arbitrary function which must satisfy the condition

(x1 ) = (x2 ) = 0 (9.34)

so that at the two points (x = x1 = x2 ), we …nd

Y (x; ) = y (x) : (9.35)

We also have
dY (x; ) dY ( )
= (x) ) = (x) ; (9.36)
d d =0
9.2. THE GENERAL PROBLEM 249

and
dY dy d
= + or Y 0 (x; ) = y 0 (x) + 0
(x) ) Y 0 (x; )j =0 = y 0 (x) ; (9.37)
dx dx dx
which gives

dY 0 ( ) d 0 0 0 dY 0 ( ) 0
= [y (x) + (x)] = (x) ) = (x) : (9.38)
d d d =0

For the Geodesic the integral


Z x2
I( )= F (x; Y (x; ) ; Y 0 (x; )) dx; (9.39)
x1

must be stationary, that means


Z x2
dI ( ) d
= [F (x; Y (x; ) ; Y 0 (x; ))] dx = 0: (9.40)
d =0 x1 d =0

Noting that

d @F dY ( ) @F dY 0 ( )
[F (x; Y (x; ) ; Y 0 (x; ))] = +
d =0 @Y d @Y 0 d =0
0
@F dY ( ) @F dY ( )
= + 0
(9.41)
@Y =0 d =0 @Y =0 d =0

and substituting

dY ( )
Y (x; )j =0 = y (x) ; = (x) ; Y 0 (x; )j =0 = y 0 (x) ;
d =0
dY 0 ( ) 0
= (x) ; (9.42)
d =0

we …nd
d @
[F (x; Y (x; ) ; Y 0 (x; ))] = F (x; y (x) ; y 0 (x)) (x)
d =0 @y
@
+ F (x; y (x) ; y 0 (x)) 0 (x) ; (9.43)
@y 0

Then the integral for the Geodesic line becomes


Z x2
dI ( ) @
= F (x; y (x) ; y 0 (x)) (x) dx + (9.44)
d =0 x @y
Z1 x2
@
+ F (x; y (x) ; y 0 (x)) 0 (x) dx: (9.45)
x1 @y 0
250CHAPTER 9. INTRODUCTION TO THE CALCULUS OF VARIATIONS

Using integration by parts the second integral can be rewritten as


Z x2 x2
@ 0 0 @ @F
F (x; y (x) ; y (x)) (x) dx = (x)
x1 @y 0 @x @y 0 x1
Z x2
@ @F
(x) (x) dx; (9.46)
x1 @x @y 0

so that using
(x1 ) = (x2 ) = 0; (9.47)
we …nd
Z x2 Z x2
@ @ @F
F (x; y (x) ; y 0 (x)) 0
(x) dx = (x) (x) dx: (9.48)
x1 @y 0 x1 @x @y 0

Thus the stationary integral can be put in the form


Z x2
dI ( ) @
= F (x; y (x) ; y 0 (x)) (x) dx+
d =0 x1 @y
Z x2
@ @F
(x) (x) dx = 0: (9.49)
x1 @x @y 0
or Z x2
dI ( ) @ @F @F
= (x) dx = 0: (9.50)
d =0 x1 @x @y 0 @y
There follow that
@ @F @F
=0 (9.51)
@x @y 0 @y
where F = F (x; y (x) ; y 0 (x)):

Example 9.3 Do Example 9.2 using Euler-Lagrange Equation.

9.3 Appplications: The Brachystochrone Prob-


lem
The Brachystochrone Problem: If two points A and B are given, at di¤erent
heights but not lying one above the other (as shown in the …gure below), it is
required to …nd among all possible curves connecting them, that one along which
a material point slides from A to B under the in‡uence of gravity (neglecting
friction) in the shortest possible time. This curve is called a Brachistochrone
curve (Gr. o&, brachistos - the shortest, o o&, chronos - time), or
curve of fastest descent.[From Wikipedia, the free encyclopedia]
This problem occupied at the time of the leading mathematicians in the
whole of Europe: Newton, Leibniz, Bernoulli, L’Hospital, and others. From then
on, the calculus of variations developed as a special mathematical discipline.
9.3. APPPLICATIONS: THE BRACHYSTOCHRONE PROBLEM 251

Example 9.4 Using the Euler-Lagrange equation solve the brachystochrone prob-
lem, assuming the “material point” starts from rest.

Solution: We are given the two points (x1 ; y1 ) and (x2 ; y2 ); we chose axes
through the point 1 with the y axis positive downward as shown in Figure
below. We want to …nd the curve joining the two points, down which
a bead will slide (from rest) in the least time. That means we want to
minimize time t:In other words we want to …nd the stationary value for

the integral
Z 2 Z 2
ds
I= dt = :
1 1 v
The total energy of the bead is zero since it starts from rest assuming the
zero energy level is the origin. If there is no friction, then we can write
the energy at any point below the origin described by the coordinates (x; y)
as
1 p
mv 2 mgy = 0 ) v = 2gy: (9.52)
2
Then Z Z
2 2
ds ds
I= = p : (9.53)
1 v 1 2gy
Noting that
p
ds = dx2 + dy 2
252CHAPTER 9. INTRODUCTION TO THE CALCULUS OF VARIATIONS

we have
r
2
Z 2
p Z 2 1 + dx Z y2
p
dx2 + dy 2 dy 1 + x02 1
I= p = p p dy = p
dy;
1 2gy 1 2gy
y1 y 2g
(9.54)
so that the stationary value of this integral is determined from the Euler-
Lagrange equation
@ @F @F
0
= 0; (9.55)
@y @x @x
where p
1 1 + x02
F (y; x (y) ; x0 (y)) = p p ; (9.56)
2g y
and
dx (y)
x0 (y) = : (9.57)
dy
Noting that
@F
=0 (9.58)
@x
and
@F 1 x0
= p p p (9.59)
@x0 2g 1 + x02 y
so that
!
@ 1 x0 x0 p
p p p =0) p p = c; (9.60)
@y 2g 1 + x02 y 1+x 02 y
where c is a constant. Solving for x0 ; we …nd
x0 p
p p = c ) x02 = c 1 + x02 y
1 + x02 y
r Z yr
dx cy cy
) x02 (1 cy) = cy ) = )x= dy: (9.61)
dy 1 cy 0 1 cy
Introducing the transformation de…ned by
1 1
cy = sin2 = (1 cos ( )) ) dy = sin cos d (9.62)
2 2 c 2 2
we have
Z r Z s
y
cy 1 sin2 2
x = dy = sin cos d
0 1 cy c 0 1 sin2 2
2 2
Z Z
1 sin 1
= 2
sin cos d = sin2 d
c 0 cos 2
2 2 c 0 2
Z
1 1 1
) x= (1 cos ( )) d ) x = ( sin ( )) : (9.63)
c 0 2 2c
9.4. APPLICATIONS: CLASSICAL MECHANICS 253

Therefore, the trajectory of the bead that takes the smallest possible time
is given by
1 1
x= ( sin ( )) ; y = (1 cos ( )) : (9.64)
2c 2c
Cycloid: Consider a circle of radius r rolling along the positive x-axis with a
constant angular velocity staring from the origin. If you mark the point on the
circle coinciding with the origin at the initial time (as shown in the Fig. 9.3)

Figure 9.3: Cycloid motion.

and follow the trajectory of this point its x and y coordinates of this point are
given by
x = r( sin ( )) ; y = r (1 cos ( )) ; (9.65)
where is the angle that the circle (the point) rotated. For example the …gure
below shows this trajectory for a point on a circle of unit radius (r = 1).
For a given , the circle’s centre lies at

x = r ; y = r: (9.66)

On the other hand if the circles center is

x = r ;y = r: (9.67)

then the trajectory looks like the …gure shown below

9.4 Applications: classical mechanics


Multivariable Functions: Suppose we are given a function, F , that depends on
y; z; dy=dx; dz=dx, and x, and we want to …nd y = y(x) and z = z(x) which
make Z x 2

I= F (x; y; z; y 0 ; z 0 ) dx (9.68)
x1
254CHAPTER 9. INTRODUCTION TO THE CALCULUS OF VARIATIONS

stationary, then we must solve the Euler-Lagrange equations


@ @F @F
= 0; (9.69)
@x @y 0 @y
@ @F @F
= 0
@x @z 0 @z

9.5 Physical application of the Euler-Lagrange


equation
~ The Hamiltonian ( H): The sum of the kinetic energy (T ) and potential
energy (V )
H = T + V: (9.70)
~ The Lagrangian ( L): The kinetic energy minus the potential energy
L=T V: (9.71)

~ The classical action ( S): the integral of the Lagrangian with respect to time
over a given period of time
Z t2
S= Ldt: (9.72)
t1

~ Hamilton’s Principle: The motion of a given system from time t1 to time t2


is such that the classical action
Z t2
S= Ldt (9.73)
t1
9.5. PHYSICAL APPLICATION OF THE EULER-LAGRANGE EQUATION255

has a stationary value for the correct path of the motion. The path actu-
ally followed by a system, as speci…ed in terms of the generalized coordi-
nates qi , is that path that makes the action integral stationary:
Z t2
I= L(qi ; q_i ; t) dt = 0 (9.74)
t1

for i = 1; 2; 3:::n:This means that the Lagrangian must satisfy the set of
equations
@ @L @L
=0 (9.75)
@t @ q_i @qi
for i = 1; 2; 3:::n:

Example 9.5 Use Lagrange’s equations to …nd the equation of motion for a
particle traveling along the x-y plane under the in‡uence of a potential
energy function U (x).

Solution: The kinetic energy of a particle moving in the x-y plane can be ex-
pressed as
1 1
T = m vx2 + vy2 = m x_ 2 + y_ 2 : (9.76)
2 2
Then the Lagrangian
1
L=T U= m x_ 2 + y_ 2 U (x): (9.77)
2
256CHAPTER 9. INTRODUCTION TO THE CALCULUS OF VARIATIONS

Since L = L (x; y; x;
_ y;
_ t) which is a function of two variables and we must
have two Euler-Lagrange equations

@ @L @L @ @L @L
= 0; = 0: (9.78)
@t @ x_ @x @t @ y_ @y

Therefore, using the Lagrangian we …nd

@ @L @L @ @U (x)
=0) (mx)
_ + =0
@t @ x_ @x @t @x
@U (x)
) m•
x=
@x
@ @L @L
= 0 ) m•
y = 0 (Zero acceleration) (9.79)
@t @ y_ @y

Example 9.6 The Atwood’s Machine: A string of length, l , passes over a fric-
tionless pulley connecting two masses, m1 and m2 . Find an expression
for the acceleration of the masses in the system.

Figure 9.4: Atwood’s Machine.


9.5. PHYSICAL APPLICATION OF THE EULER-LAGRANGE EQUATION257

Solution: Let’s de…ne the origin of the y axis on the surface of the ground also
assume m1 > m2 ; and the length of the string is l: At a given time t let
the position of m1 and m2 be y1 and y2 ; as shown in Fig. 9.4. Then the
kinetic energy of the system can be expressed as
1 1
T = m1 y_ 12 + m2 y_ 22 (9.80)
2 2
and the gravitational potential energy

V = m1 gy1 + m2 gy2 (9.81)

Then the Lagrangian


L=T V (9.82)
of the system can be written as
1 1
L= m1 y_ 12 + m2 y_ 22 m1 gy1 m2 gy2 : (9.83)
2 2
This equation appears to be a function of two variables. However, because
of the constraint
y1 + y2 + l = C; (9.84)
where C is a constant, we end up with a Lagrangian that depends only on
one variable. If we replace

y2 = C y1 l ) y_ 2 = y_ 1 (9.85)

we have
1
L= (m1 + m2 ) y_ 12 m1 gy1 m2 g (C y1 l) : (9.86)
2
which we may put in the form
1
L= (m1 + m2 ) y_ 12 (m1 m2 ) gy1 C1 ; (9.87)
2
where we replaced, C1 = m2 g (C l) : Now using

@ @L @L
=0 (9.88)
@t @ q_i @qi

for qi = y1
@L @L
= (m1 m2 ) g; = (m1 + m2 ) y_ 1 (9.89)
@y1 @ y_ 1
we …nd
@ m1 m2
[(m1 + m2 ) y_ 1 ] = (m1 m2 ) g ) a1 = y•1 = g (9.90)
@t m1 + m2
258CHAPTER 9. INTRODUCTION TO THE CALCULUS OF VARIATIONS

Recalling that
y_ 2 = y_ 1 (9.91)
the acceleration of the second mass becomes
m1 m2
a1 = g: (9.92)
m1 + m2
The minus sign indicates the …rst mass is accelerating in the negative y-
direction.
Example 9.7 Central Forces: Describe the properties of the motion of a mass
m moving under the in‡uence of a central force (that is, a force acting
only along the radial direction) given by

F~ = f (r)^
r (9.93)

for some function f (r). Assume that the motion is con…ned to a plane.
Solution: The kinetic energy
1
T =mv 2 : (9.94)
2
Using polar coordinates the magnitude of the velocity can be expressed as
2
v = r_ 2 + r2 _ (9.95)

and the kinetic energy becomes


1 2
T = m r_ 2 + r2 _ : (9.96)
2
The potential energy is related to the central force by

F~ = r U (r) (9.97)

where U (r) is the potential energy. Since the force is a central force it is
directed along the radial direction and it depends on r only. Therefore the
potential energy can be expressed as
Z
U (r) = f (r) dr: (9.98)

Then the Lagrangian can be expressed as


Z
1 2
_ ;_ =T
L t; r; r; U= m r_ 2 + r2 _ + f (r) dr (9.99)
2
Then using the Euler-Lagrange’s equation

@ @L @L
=0 (9.100)
@t @ q_i @qi
9.5. PHYSICAL APPLICATION OF THE EULER-LAGRANGE EQUATION259

we have
@ @L @L @ @L @L
= 0; =0 (9.101)
@t @_ @ @t @ r_ @r
so that using
@L @L @L 2 @L
= 0; = mr2 _ ; = mr _ + f (r); = mr_ (9.102)
@ @_ @r @ r_
we …nd

@ @L @L
=0) = const ) mr2 _ = cont ) I! = cons:
@t @_ @_
(Conservation of Ang. Mom.)
@ @L @L 2
r = mr _ + f (r)
= 0 ) m• (9.103)
@t @ r_ @r
260CHAPTER 9. INTRODUCTION TO THE CALCULUS OF VARIATIONS
Chapter 10

Introduction to the
Eigenvalue Problem

10.1 Matrix Review


Matrix Arithmetic and Manipulation: Consider the following matrices:
0 1
2 4
2 3 1
A = ;B = @ 1 1 A
2 1 0
3 1
0 1 0 1
2 1 3 2 0 1
C = @ 4 1 2 A;D = @ 1 1 2 A (10.1)
1 0 1 3 1 0

Multiplication by a Scalar : Any matrix can be multiplied by a scalar:

2 2 3 2 1 2 4 2
2A =
2 2 1 2 0 2 5 2
4 6 2 8
2A = (10.2)
4 2 0 10

~ Addition and subtraction: Two matrices can be added or subtracted if and


only if they have the same dimensions. From matrices A; B; C, and D we
can add/subtract only matrices C and D
0 1 0 1
2 1 3 2 0 1
C +D =@ 4 1 2 A+@ 1 1 2 A
1 0 1 3 1 0
0 1 0 1
2 2 1+0 3+1 0 1 4
=@ 4+1 1 1 2+2 A=@ 5 2 0 A
1+3 0+1 1+0 2 1 1

261
262 CHAPTER 10. INTRODUCTION TO THE EIGENVALUE PROBLEM

Matrix Multiplication: two matrices can be multiplied if and only if the num-
ber of columns of the …rst matrix is equal to the number of rows of the second
matrix. If matrices have the same dimension, then they can be multiplied. From
the above matrices we can make the multiplications:
0 1
2 4
2 3 1 @ (ab)11 (ab)12
AB = 1 1 A= (10.3)
2 1 0 (ab)21 (ab)22
3 1
0 10 1 0 1
2 1 3 2 0 1 (cd)11 (cd)12 (cd)13
CD = @ 4 1 2 A@ 1 1 2 A = @ (cd)21 (cd)22 (cd)23 A
1 0 1 3 1 0 (cd)31 (cd)32 (cd)33
(10.4)
but we can not make the matrix multiplications BC or BD
The element in row i and column j of the product matrix AB is equal to
row i of A times column j of B. In index notation
n
X
(ab)ij = aik bkj (10.5)
k=1

where (ab)ij is the element of the product matrix AB:


~ Commutativities: For any two multiplyable matrices C and D;
CD 6= DC (10.6)

~ Commutator : For square matrices C and D the Commutator [C; D] is de…ned


as
[C; D] = CD DC (10.7)
~ For any three matrices, F; G;and H that can be multiplied, we can write
The Associative Law :
F (GH) = (F G)H (10.8)
The Distributive Law :
F (G + H) = F G + F H (10.9)

~ The Identity Matrix; I (Boas: The Unit Matrix, U )


IA = AI = A (10.10)

~ Transpose of a Matrix : The transpose of the matrix A is denoted by AT :


0 1
2 2
2 3 1
A = ; AT = @ 3 1 A
2 1 0
1 0
0 1
2 4
2 1 3
B = @ 1 1 A ; BT = (10.11)
4 1 1
3 1
10.1. MATRIX REVIEW 263

~ Adjoint of a Matrix : The adjoint of a square matrix, A, is given by


adj(A) = [cof (A)]T (10.12)
where cof (A) is the cofactor of the matrix A. We recall that the minor
of matrix A; (Mij ); is the determinant of the matrix formed from matrix
A by removing the ith row and j th column. For the cofactor matrix the
elements are expressed as
i+j
[cof (A)]ij = ( 1) Mij : (10.13)

~ Inverse of a (Square) Matrix : A 1

1 1
A A = AA =I (10.14)
We can determine the inverse of an invertible matrix (det jAj =
6 0) using
row reduction or the adjoint matrix.
a. Row reduction in this approach for the matrix, for example,
2 3
a11 a12 a13
A = 4 a21 a22 a23 5 (10.15)
a31 a32 a33
we start from 2 3
a11 a12 a13 1 0 0
4 a21 a22 a23 0 1 0 5 (10.16)
a31 a32 a33 0 0 1
and do elementary row operation until we end up with
2 3
1 0 0 b11 b12 b13
4 0 1 0 b21 b22 b23 5 (10.17)
0 0 1 b31 b32 b33
so that the inverse of the Matrix A is given by
0 1
b11 b12 b13
A 1 = @ b21 b22 b23 A : (10.18)
b31 b32 b33

b. Using the adjoint matrix: Using the adjoint matrix the inverse can be ex-
pressed as
T
[cof (A)]
A 1= (10.19)
det jAj
Example 10.1 Find the inverse of the matrix
0 1
1 2 3
A=@ 2 0 4 A (10.20)
1 1 1
using
264 CHAPTER 10. INTRODUCTION TO THE EIGENVALUE PROBLEM

a. Row reduction approach


b. The adjoint matrix approach

Solution: a. In the row reduction approach we start from


2 3
1 2 3 1 0 0
4 2 0 4 0 1 0 5 (10.21)
1 1 1 0 0 1
and try to get 2 3
1 0 0 a11 a12 a13
4 0 1 0 a21 a22 a23 5 (10.22)
0 0 1 a31 a32 a33
so that we can get the inverse matrix
0 1
a11 a12 a13
A 1 = @ a21 a22 a23 A : (10.23)
a31 a32 a33

b. Find the cofactor Cij of the element Aij in row i and column j which
is equal to (-1) i+j times the value of the determinant remaining when
we cross o¤ row i and column j. After you obtained all elements of
the cofactor matrix write the cofactor matrix C, transpose and divide
it with the determinant of the matrix A. The resulting matrix is A 1 :
1 1 T
A = C : (10.24)
jAj
Ans: 0 1
4 5 8
1@
A 1
= 2 2 2 A: (10.25)
2
2 3 4

10.2 Orthogonal matrices and the rotational op-


erator
Matrices that make an orthogonal transformation of vectors. In an orthogonal
transformation of vectors the magnitude of the vectors remains the same. For
an orthogonal matrix
M 1 = MT (10.26)
The Rotation Operator : For a counter-clockwise rotation about the z-axis
by an angle , we denote the rotation matrix by R: Rz ( ) = R. Then,
r0 = Rr (10.27)
0 0
1 0 10 1
x cos sin 0 x
) @ y 0 A = @ sin cos 0 A@ y A
z0 0 0 1 z
10.3. EIGENVALUES AND EIGENVECTORS 265

10.3 Eigenvalues and Eigenvectors


In general, a linear transformation of a vector, ~r; to a vector, ~r0 ; using matrix
can be expressed as
~r0 = M ~r: (10.28)

In Cartesian coordinates
0 1 0 10 1
x0 M11 M12 M13 x
@ y 0 A = @ M21 M22 M23 A @ y A : (10.29)
z0 M31 M32 M33 z

Under this transformation if the vector ~r0 is expressible as

~r0 = M ~r = ~r; (10.30)

where is a constant, then the vector ~r is called the eigenvector (characteristic


vector) and is the eigenvalue (characteristic value) of the Matrix M;
0 10 1 0 1
M11 M12 M13 x x
@ M21 M22 M23 A @ y A = @ y A: (10.31)
M31 M32 M33 z z

Using the identity matrix, I;


0 1
1 0 0
I=@ 0 1 0 A; (10.32)
0 0 1
266 CHAPTER 10. INTRODUCTION TO THE EIGENVALUE PROBLEM

one can put Eq. (10.31)


0 10 1
M11 M12 M13 x
@ M21 M22 M23 A @ y A
M31 M32 M33 z
0 10 1 0 10 1
1 0 0 x 0 0 x
= @ 0 1 0 A@ y A = @ 0 0 A@ y A (10.33)
0 0 1 z 0 0 z

which can be rewritten as


20 1 0 13 0 1
M11 M12 M13 0 0 x
4@ M21 M22 M23 A @ 0 0 A5 @ y A
M31 M32 M33 0 0 z
=0 (10.34)
0 10 1
M11 M12 M13 x
)@ M21 M22 M23 A @ y A = 0:
M31 M32 M33 z

The eigenvalues are obtained from the condition

M11 M12 M13


M21 M22 M23 = 0; (10.35)
M31 M32 M33

which is known as the eigenvalue equation (characteristic equation). To …nd the


eigenvectors we substitute the eigenvalues and solve the resulting equations.
From The VNR Concise Encyclopedia of Mathematics (Van Nostrand Rein-
hold Co., publishers, 1977):
Eigenvalues: Eigenvalue problems are important in many branches of physics.
They make it possible to …nd coordinate systems in which the transformations
in question take on their simplest forms. In mechanics for instance, the prin-
cipal moments of a rigid body are found with the help of the eigenvalues of the
symmetric matrix representing the inertia tensor.... Eigenvalues are of central
importance in quantum mechanics, in which the measured values of physical
“observables” appear as the eigenvalues of certain operators. The term “trans-
formation” is used predominantly in pure mathematical (geometrical) context,
whereas “operator” is more customary in applications (physics, technology).

Example 10.2 Find the eigenvalues and the corresponding eigenvectors of the
matrix
0 1
0 1 0
M =@ 1 0 0 A (10.36)
0 0 0
10.3. EIGENVALUES AND EIGENVECTORS 267

Solution: The eigenvalue equation


0 1 0
0 1 0
1 0 0 =0) =0
0 0
0 0 0
3
) + =0) 1 = 0; 2 = 1; 3 = 1: (10.37)
The corresponding eigenvectors are determined from
0 10 1
M11 i M12 M13 xi
@ M21 M22 i M23 A @ yi A = 0: (10.38)
M31 M32 M33 i zi
For 1 = 0, we …nd
0 10 1
0 1 0 x1
@ 1 0 0 A @ y1 A = 0 ) x1 = 0; y1 = 0: (10.39)
0 0 0 z1
We use the bra-ket notation to represent eigenvectors. The eigenvector for
an eigenvalue, ; is represented using a ket-vector, j i ; which is expressed
as a column matrix 0 1
x
j i = @ y A: (10.40)
z
The eigenvector for 1 can then be written as a column matrix
0 1
0
j 1i = @ 0 A : (10.41)
z1
The corresponding bra-vector, h j ; generally, is the transpose conjugate
of the ket-vector. But here since we will consider only real values, the
bra-vector is written as
h 1j = x y z : (10.42)
Thus we can write
h 1j = 0 0 z1 : (10.43)
Both bra and ket vectors must be normalized (be unit vectors). Thus for
any eigenvectors, we must have
h j i = 1:
Normalizing this vector
0 1 0 1
0 0
h 1 j 1i =1) 0 0 z1 @ 0 A = 1 ) z1 = 1 ) j 1i =
@ 0 A:
z1 1
(10.44)
268 CHAPTER 10. INTRODUCTION TO THE EIGENVALUE PROBLEM

Similarly for 2 =1

0 10 1
1 1 0 x2
@ 1 1 0 A @ y2 A = 0 ) x2 + y2 = 0; x2 y2 = 0; z2 = 0
0 0 1 z2
) x2 = y2 ; z2 = 0 (10.45)

and the eigenvector becomes

0 1
1
j 2i = x2 @ 1 A ; (10.46)
0
0 1
1
h 2 j 2i = 1 ) x22 1 1 0 @ 1 A = 1 ) x2 = p1 (10.47)
0 2
1 0
1
1
) j 2i = p @ 1 A (10.48)
2 0

and for 3 = 1

0 10 1
1 1 0 x3
@ 1 1 0 A @ y3 A = 0 ) x3 + y3 = 0; x3 + y3 = 0; z3 = 0
0 0 1 z3
) x3 = y3 ; z3 = 0 (10.49)

1 0 0 1
1 1
1 @
) j 3 i = y3 @ 1 A ) j 3i = p 1 A: (10.50)
0 2 0

Using Mathematica:
10.3. EIGENVALUES AND EIGENVECTORS 269

Eigenvalue equation for Hermitian Matrices: A Hermitian Matrix is a matrix


that is equal to its transposed and conjugated matrix
T
(M ) = M: (10.51)
The eigenvalues for a Hermitian matrix are all real and the corresponding eigen-
vectors are orthogonal. This means for a Hermitian matrix with the eigenvalue
equation
M j ii = i j ii ; (10.52)
we always …nd
i = i (10.53)
and
0 i 6= j
h j j ii = = ij (10.54)
1 i=j
The Similarity Transformation: The similarity transformation of the matrix
M is given by
D C 1 M C: (10.55)
where C is a matrix whose columns are the eigenvectors of the Eigenvalue equa-
tion for matrix M . The matrix D is a diagonal matrix where the diagonal
elements are the eigenvalues.
270 CHAPTER 10. INTRODUCTION TO THE EIGENVALUE PROBLEM

10.4 Physical applications


Normal modes of Vibration for couple harmonic oscillators

Example 10.4 Consider a system consisting of two equal masses m connected


by three identical springs of spring constant k.

Figure 10.1: Coupled harmonic oscillators.

The masses can slide on a horizontal, frictionless surface. The springs are
at their unstretched/uncompressed lengths when the masses are at their equi-
librium positions. At t = 0, the masses are displaced from their equilibrium
positions by the amounts x10 and x20 and released from rest, as shown in the
…gure above. Completely describe the resulting motion.

Step 1: The Equations of Motion:


Step 2: Similarity Transformation (…nd the eigenvalues and eigenvectors)
Step 3: Solving the De-coupled Transformed Equations of Motion
Step 4: The Propagator Matrix, U
Step 5: The Normal Modes of Vibration
Solution:
Step 1 The Equations of Motion: To …nd the equations of motion I will use
Euler-Lagrange equations. You can use Newton’s second law if you want
to. To use the Euler-Lagrange equation we need to …nd the kinetic energy
and the potential energy. Suppose at a given instant of time the posi-
tion of the …rst mass is x1 and the second mass is x2 as measured from
their corresponding equilibrium positions. Then the corresponding kinetic
energy can be expressed as
1 1
K1 = mx_ 21 ; K2 = mx_ 22 (10.56)
2 2
10.4. PHYSICAL APPLICATIONS 271

so that the total kinetic energy be


1
K= m x_ 21 + x_ 22 : (10.57)
2
The potential energy (elastic) is due to the displacement of the springs
from their equilibrium position. If the …rst mass is displaced by x1 from
the equilibrium and the second mass by x2 in the positive x direction,
then the …rst spring will be stretched by x1 ; the second spring will be
compressed by x1 and at the same time it will be stretched by x2 ; as
a result the net displacement will be x2 x1 : The third spring will be
compressed by x2 : Therefore, the total potential energy will be
1 2 1 2 1 2
K= kx + kx + k (x2 x1 ) : (10.58)
2 1 2 2 2
Then the Lagrangian

L (t; x1 ; x2 ; x_ 1 ; x_ 2 ) = T U
1 1 2 1 2 1 2
) L (t; x1 ; x2 ; x_ 1 ; x_ 2 ) = m x_ 21 + x_ 22 kx kx k (x2 x1 )
2 2 1 2 2 2
(10.59)

which leads to
@L @L
= kx1 + k (x2 x1 ) ; = kx2 k (x2 x1 ) ;
@x1 @x2
@ @L @ @L
= m•
x1 ; = m•
x2 : (10.60)
@t @ x_ 1 @t @ x_ 2

Using the Euler-Lagrange’s equation

@ @L @L
= 0 for i = 1; 2 (10.61)
@t @ x_ i @xi

we …nd

m•
x1 [ kx1 + k (x2 x1 )] = 0 ) m•
x1 = 2kx1 + kx2 (10.62)

and

m•
x2 [ kx2 k (x2 x1 )] = 0 ) m•
x2 = 2kx2 + kx1 : (10.63)

If we introduce a constant r
k
; != (10.64)
m
then the above two equations can be put in the form

x
•1 = 2! 2 x1 + ! 2 x2 (10.65)
272 CHAPTER 10. INTRODUCTION TO THE EIGENVALUE PROBLEM

and
• 2 = ! 2 x1
x 2! 2 x2 : (10.66)
These two equations describe the equations of motion for the two masses.
As you can see the two equations are coupled second order di¤erential
equations.

Step 2: Similarity transformation: In a matrix form the two equations


can be put in the form

x
•1 2! 2 !2 x1
= (10.67)
x
•2 !2 2! 2 x2
or ::
~r = M~r (10.68)
where
~r = x1 e^1 + x2 e^2 ; (10.69)
and
2! 2 !2
M= : (10.70)
!2 2! 2
Note that e^1 and e^2 are column matrices given by

1 0
e^1 = ; e^2 = : (10.71)
0 1

Suppose if one solves the eigen value equation for the matrix M

Mj i= j i; (10.72)

we know that the matrix, C; constructed from the normalized eigen


vectors j i ; leads to a similarity transformation for the matrix M.
That means
1
C MC = D
1
a11 a12 M11 M12 a11 a12
)
a21 a22 M21 M22 a21 a22
1 0
= ; (10.73)
0 2

where
a11 a11
j 1i = ;j 2i = (10.74)
a21 a22
are the normalized eigen vectors. Noting that
1 1
C C = CC =I
10.4. PHYSICAL APPLICATIONS 273

one can write the matrix for the equation of motion as


:: ::
1
~r = M~r ) ~r = M CC ~r (10.75)
1
and multiplying from the left by the matrix C ; one can put the
equation of motion in the form
::
1 1 1
C ~r = C MC C ~r (10.76)

and introducing a new vector de…ned by

x1 y1 :: x
•1 y•1
1 1 1 1
C ~r = C = )C ~r = C = ;
x2 y2 x
•2 y•2
(10.77)
and noting that
1
C MC = D
is a similarity transformation, we …nd

y•1 1 0 y1
= :
y•2 0 2 y2

It is important to note that C is a matrix whose columns are the


eigenvectors of the Eigenvalue equation for matrix M . The matrix
D is a diagonal matrix in which the diagonal elements are the eigen-
~ such that
values. Suppose if we can …nd the eigenvectors, R;

~ = R;
MR ~ (10.78)

or using the bra-ket notation (also known as Dirac notation)

Mj i= j i; (10.79)

then for the eigenvalue, one can written

2! 2 !2 2 2
det = 0 ) 2! 2 + !2 =0
!2 2! 2
) 2! 2 + !2 2! 2 + + !2 = 0 ) 1 = !2 ; 2 = 3! 2 :
(10.80)

The corresponding eigenvectors are obtained from

2! 2 1 !2 X1
=0 (10.81)
!2 2! 2 1 X2

and
2! 2 2 !2 X1
= 0: (10.82)
!2 2! 2 2 X2
274 CHAPTER 10. INTRODUCTION TO THE EIGENVALUE PROBLEM

Solving these equations for, 1 = ! 2 ; we …nd


!2 !2 X1
=0) X1 + X2 = 0 ) X2 = X1
!2 !2 X2
(10.83)
for, 2 = 3! 2 ;
!2 !2 X1
= 0 ) X1 + X2 = 0 ) X2 = X1 : (10.84)
!2 !2 X2
Then the two eigenvectors becomes

~ 1 = X1 1 ~ 2 = X1 1
R ;R : (10.85)
1 1
Upon normalizing these eigen vectors, we …nd

^ 1 = p1
R
1 ^ 2 = p1
;R
1
; (10.86)
2 1 2 1
or using bra-ket notation
1 1 1 1
j 1i =p ;j 2i =p : (10.87)
2 1 2 1
You can use Mathematica to check you results. Here is an example
how to …nd eigenvalues and unnormalized eigenvectors determined
using Mathematica:
10.4. PHYSICAL APPLICATIONS 275

Step 3: Solving the De-coupled Transformed Equations of Motion: The matrix


C and its inverse C 1 can be expressed as
! !
p1 p1 p1 p1
2 2 1 2 2
C= p1 p1
;C = p1 p1
: (10.88)
2 2 2 2

N.B. You must use the method we studied to …nd the inverse of the matrix
C. But, here I am going to use mathematica to …nd the inverse

We note that
! !
p1 p1 p1 p1 1 0
1 2 2 2 2 1
CC = p1 p1 p1 p1
) CC =
2 2 2 2
0 1
1
) C C=I (10.89)
Recalling that ::
~r = M~r (10.90)
can be written as
y•1 1 0 y1
= :
y•2 0 2 y2
where
x1 y1 :: x
•1 y•1
1 1 1 1
C ~r = C = )C ~r = C = ;
x2 y2 x
•2 y•2
(10.91)
we can express the equations of motion as
y•1 !2 0 y1
= ) y•1 = ! 2 y1 ; y•2 = 3! 2 y2
y•2 0 3! 2 y2
(10.92)
276 CHAPTER 10. INTRODUCTION TO THE EIGENVALUE PROBLEM

so that the solutions can be expressed as


p p
y1 = A cos (!t) + B sin (!t) ; y2 = C cos 3!t + D sin 3!t : (10.93)
Now substituting these back into
1 x1 y1 1 x1 y1
C = ) CC =C
x2 y2 x2 y2
x1 y1
) =C (10.94)
x2 y2
we …nd
!
p1 p1 A cos
x1 (t)
= 2 2 p (!t) + B sin (!t)
p
x2 (t) p1 p1 C cos 3!t + D sin 3!t
2 2
(10.95)
Initially (t = 0) the …rst mass is displaced, x1 (0) = x10 and the second
mass displaced x2 (0) = x20 ; and both masses released from rest, x_ 1 (0) =
x_ 2 (0) = 0: Using these initial conditions we …nd
!
p1 p1 x10 A
2 2 =
p1 p1 x 20 C
2 2
1 1
) A = p (x10 + x20 ) ; C = p (x10 x20 ) (10.96)
2 2
and
!
p1 p1 !A sin
2 2 x_ 1
= p p (!t) + p
!B cos (!t) p
p1 p1 x_ 2 3!C sin 3!t + 3!D cos 3!t
2 2

)
0
= p!B ) B = 0; D = 0 (10.97)
0 3!D
Then using the results above, one can write
! !
x1 (t) p1 p1 p1 (x10 + x20 ) cos (!t)
= 2 2 2 p (10.98)
x2 (t) p1 p1 p1 (x10 x20 ) cos 3!t
2 2 2
or
1 1 p
x1 (t) = (x10 + x20 ) cos (!t) + (x10 x20 ) cos 3!t
2 2
1 1 p
x2 (t) = (x10 + x20 ) cos (!t) (x10 x20 ) cos 3!t (10.99)
:
2 2
Step 4: The Propagator Matrix, U : If upon simplifying the above expressions,
we …nd
1h p i 1h p i
x1 (t) = cos (!t) + cos 3!t x10 + cos (!t) cos 3!t x20
2 2
1h p i 1h p i
x2 (t) = cos (!t) cos 3!t x10 + cos (!t) + cos 3!t x20
2 2
(10.100)
10.4. PHYSICAL APPLICATIONS 277

which can be put using matrices as


p p
x1 (t) 1 cos (!t) + cos p3!t cos (!t) cos p3!t
=
x2 (t) 2 cos (!t) cos 3!t cos (!t) + cos 3!t
x10
; (10.101)
x20
or
x1 (t) x1 (0)
= U (t) ) ~r = U (t)~r (0) ; (10.102)
x2 (t) x2 (0)
where
x1 (0) x10
~r (0) = = ; (10.103)
x2 (0) x20
and
p p
1 cos (!t) + cos p3!t cos (!t) cos p3!t
U (t) =
2 cos (!t) cos 3!t cos (!t) + cos 3!t

is called the propagator matrix.

Step 5: The Normal Modes of Vibration: Suppose the initial state of the two
masses is described by the …rst eigenvector. That means

x10 1 1
~r (0) = =p (10.104)
x20 2 1

then
~r = U (t)~r (0) (10.105)
gives
p p
x1 (t) 1 cos (!t) + cos p3!t cos (!t) cos p3!t
= p
x2 (t) 2 2 cos (!t) cos 3!t cos (!t) + cos 3!t
1
(10.106)
1

which leads to

x1 (t) 1 cos (!t)


=p ) x1 (t) = x2 (t) : (10.107)
x2 (t) 2 cos (!t)

The two masses oscillate with a frequency, !; in the same direction. On


the other hand, if the initially state of the two masses is given by the
second eigenvector

x10 1 1
~r (0) = =p : (10.108)
x20 2 1
278 CHAPTER 10. INTRODUCTION TO THE EIGENVALUE PROBLEM

then we have
p p
x1 (t) 1 cos (!t) + cos p3!t cos (!t) cos p3!t
= p
x2 (t) 2 2 cos (!t) cos 3!t cos (!t) + cos 3!t
1
(10.109)
1

which gives
p
x1 (t) 1 cos p3!t
=p ) x1 (t) = x2 (t) : (10.110)
x2 (t) 2 cos 3!t
p
The two masses oscillate with a frequency 3! out of phase by .
The two modes of vibrations we saw above are called Normal Modes of vibration.
Chapter 11

Special functions

11.1 The factorial and gamma function


The Factorial, n!: We recall the factorial function, n!; for a positive integer or
zero is de…ned as

n! = n (n 1) (n 2) (n 3)(n 4):::3 2 1 0!; (11.1)

where
0! = 1: (11.2)
The integral form of the Factorial function: Consider the integral function
given by Z 1
F (p) = e x dx: (11.3)
0
For any real number, > 0, the value of this integral is
Z 1 1
e x 1
F (p) = e x dx = = : (11.4)
0 0

Now let’s di¤erentiate this integral with respect to as many as we can, say n
n
time (i.e. @@ n ). For the …rst derivative, n = 1
Z 1 Z 1
1 0!
e x dx = ) x0 e x dx = 1
0
Z 1 Z 01
@ 1 @ @ 1
) e x dx = ) e x dx =
@ 0 0 @ @
Z 1
1
xe x dx = 2
(11.5)
0

This can be put in the form


Z 1
1!
x1 e x
dx = 2
: (11.6)
0

279
280 CHAPTER 11. SPECIAL FUNCTIONS

For the second derivative (n = 2)


Z 1 Z 1
@2 x @2 1 @ x @ 1!
e dx = ) xe dx =
@ 2 0 @ 2 @ 0 @ 2
Z 1
2 1! 2!
) x2 e x
dx = 3
= 3
: (11.7)
0

For the third derivative (n = 3)


Z 1 Z 1
@3 x @3 1 @ @ 2!
e dx = ) x2 e x
dx =
@ 3 0 @ 3 @ 0 @ 3
Z 1 Z 1
3 x 3 2 1! 3!
) x e dx = 3
) x3 e x
dx = 3
: (11.8)
0 0

Therefore it is not di¢ cult to see for the nth derivative, we …nd
Z 1
n!
xn e x dx = n : (11.9)
0

Now we set = 1, we …nd


Z 1
n! = xn e x
dx; (11.10)
0

which is the integral form of the Factorial function which is valid for an integer,
n 0.
The Gamma function: The Gamma function is de…ned by the integral Func-
tion given by Z 1
(p) = xp 1
e x
dx: (11.11)
0
where p > 0 is any positive real number. For p = n + 1; with n 0 (positive
integer or zero), we …nd
Z 1
(n + 1) = xn e x dx; (11.12)
0

which is the Factorial function. Therefore, the factorial function in terms of the
Gamma function can be expressed as
Z 1
n! = (n + 1) = xn e x dx: (11.13)
0

The Recursion Relation: If we replace p by p + 1 in the expression for the


Gamma function Z 1
(p) = xp 1 e x dx: (11.14)
0
we …nd Z 1
(p + 1) = xp e x
dx: (11.15)
0
11.1. THE FACTORIAL AND GAMMA FUNCTION 281

If we denote

u = xp ; dv = e x
dx ) du = pxp 1
;v = e x
; (11.16)

then using integration by parts


Z Z
udv = uv vdu (11.17)

we …nd Z 1
x p 1
(p + 1) = e x 0 pxp 1
e x
dx: (11.18)
0
In the above expression the …rst term is zero
Z 1
(p + 1) = p xp 1
e x
dx: (11.19)
0

We know that Z 1
(p) = xp 1
e x
dx (11.20)
0
hence
(p + 1) = p (p): (11.21)

Example 11.1 Show that


1 p
= (11.22)
2
and Z 1 1
u2 2
e du = : (11.23)
0 2
Solution: Using the de…nition of the Gamma function, we can write
Z 1
1 1
= p e x dx. (11.24)
2 0 x

Introducing a new variable de…ned by

u2 = x ) 2udu = dx (11.25)

we can write
Z 1 Z 1
1 1 u2 u2
= e 2udu = 2 e du . (11.26)
2 0 u 0

Squaring both sides, we have


Z 1 Z 1 Z 1 Z 1
1 2
v2 2 2
2
=4 e u du e dv =4 e (u +v ) dudv
2 0 0 0 0
(11.27)
282 CHAPTER 11. SPECIAL FUNCTIONS

so that introducing the polar coordinates de…ned by

u = r cos ; v = r sin ) dudv = rdrd ; u2 + v 2 = r2 (11.28)

as shown in the …gure below


the above integral can be put in the form
Z 1 Z =2
2 1 r2
=4 e rdr d : (11.29)
2 0 0

Here the set the upper limit of integration for be =2 since both u and v
are positive and we must integrate only in the …rst quadrant (the region shown
in green). Therefore, integrating with respect to r and leads to
1
r2 p
2 1 e 1
=4 = ) = : (11.30)
2 2 2 2
0

Now substituting this result into


Z 1 Z 1
1 1 x u2
= p e dx = 2 e du; (11.31)
2 0 x 0

we can see that Z r


1
u2
e du = : (11.32)
0 2
11.2. THE BETA FUNCTION 283

11.2 The Beta Function


For p > 0 and q > 0, the beta function B(p; q) is de…ned by a de…nite integral
Z 1
q 1
B(p; q) = xp 1 (1 x) dx: (11.33)
0

There are di¤erent forms of representations of the beta function. These


includes the following

~ Replace x = y=a ) dx = dy=a


Z Z
1 a y p 1 y q 1 1 a
q 1
B(p; q) = 1 dy = yp 1
(a y) dy:
a 0 a a ap+q 1 0
(11.34)
~ Replace
x = sin2 ( ) ) dx = 2 sin ( ) cos ( ) d
which gives
Z =2
q 1
B(p; q) = sin2p 2
( ) 1 sin2 ( ) 2 sin ( ) cos ( ) d
0
Z =2
) B(p; q) = 2 sin2p 1
( ) cos2q 1
( )d (11.35)
0

~ Replacing x = y= (1 + y) ; we have
dy ydy dy
dx = 2 = 2
1+y (1 + y) (1 + y)
and
y y
x = 0 ) y = 0; x = 1 ) = 1 , lim =1 (11.36)
1+y y!1 1+y
so that
Z 1 p 1 q 1
y y dy
B(p; q) = 1 2
0 1+y 1+y (1 + y)
Z 1 p 1 q 1
y 1 dy
= 2
0 1+y 1+y (1 + y)
Z 1
yp 1
dy
) B(p; q) = p+q (11.37)
0 (1 + y)

Example 11.2 Prove that the Gamma and the Beta Functions are related by
(p) (q)
B(p; q) = : (11.38)
(p + q)
284 CHAPTER 11. SPECIAL FUNCTIONS

Solution: For the Gamma functions


Z 1 Z 1
(q) = xq 1 e x dx, (p) = yp 1
e y
dy (11.39)
0 0

introducing the transformation of variables de…ned by

u2 = x ) 2udu = dx; v 2 = y ) 2vdv = dy (11.40)

we …nd
Z 1 Z 1
2q 1 u2 v2
(q) = 2 u e du , (p) = 2 v 2p 1
e dv . (11.41)
0 0

Multiplying the two functions, we have


Z 1Z 1
2 2
(p) (q) = 4 u2q 1 v 2p 1
e (u +v ) du dv, (11.42)
0 0

so that using the polar coordinates

u = r cos ; v = r sin ) dudv = rdrd ; u2 + v 2 = r2 (11.43)

we …nd
Z 1 Z =2
2q 1 2p 1 r2
(p) (q) = 4 (r cos ) (r sin ) e rdrd . (11.44)
0 0

This can be rewritten as


Z 1 Z =2
r2 2p 1 2q 1
(p) (q) = 4 r2(p+q 1)
e rdr (sin ) (cos ) d .
0 0
(11.45)
The …rst integral
Z 1 Z 1
r2
4 r2(p+q 1) e rdr = 2 R(p+q 1)
e R
dR = 2 (p + q) (11.46)
0 0

and applying
Z =2
B(p; q) = 2 sin2p 1
( ) cos2q 1
( )d (11.47)
0

we note that the second integral can be expressed in terms of the Beta
function as Z =2
2q 1 2p 1 B(q; p)
(sin ) (cos ) d = : (11.48)
0 2
Therefore
(p) (q)
B(p; q) = . (11.49)
(p + q)
This equation relates the beta and gamma functions.
11.3. STIRLING’S FORMULA 285

11.3 Stirling’s Formula


We recall the Gamma Function
Z 1
(p) = xp 1
e x
dx; (11.50)
0

when p is zero or positive integer, gives the factorial function


Z 1
p! = (p + 1) = xp e x dx: (11.51)
0

Next we want to …nd an approximate formula when p is very large. This ap-
proximate formula is known as Stirling’s formula and is given by
p
p! = (p + 1) = pp e p 2 p; (11.52)

or if we take the natural logarithm of both sides


p 1
ln(p!) = ln pp e p
2 p = ln (pp ) + ln e p
+ ln (2 p) 2
1 1
= p ln (p) p ln (e) + ln (2 p) ) ln(p!) = p ln p p+ ln (2 p)
2 2
If p is very large, the last term is very small as compared to the …rst two terms
and the Stirling’s formula is given by

ln(p!) = p ln p p: (11.53)

Proof: Introducing a new variable de…ned by


p p p
x = p + y p ) dx = pdy; x = 0 ) y = p; x ! 1 ) y ! 1;
(11.54)
the function Z 1
p! = (p + 1) = xp e x
dx (11.55)
0
can be put in the form
Z 1
p p p p
p! = p
(p + y p) e (p+y p) pdy: (11.56)
p

Noting that
p p p p p
(p + y p) = eln[(p+y p) ] = ep ln(p+y p) ; (11.57)

one can write


Z 1 p p
Z 1
p p p p
p! = p ep ln(p+y p) e (p+y p) pdy = p p
ep ln(p+y p) p y p
dy:
p p
(11.58)
286 CHAPTER 11. SPECIAL FUNCTIONS

We recall that the Taylor series expansion for f (y) about y = 0 is given
by

1 df (y) 1 d2 f (y)
f (y) = f (0) + y+ y 2 + ::: (11.59)
1! dy y=0 2! dy 2 y=0

p
so that for f (y) = ln p + y p ; using
p p
df (y) p p
f (0) = ln (p) ; = p = ;
dy y=0 p + y p y=0 p
2 p p p
d f (y) d p p p 1
= p = p 2 = ; (11.60)
dy 2 y=0 dy p + y p y=0 p + y p y=0 p

we …nd an approximate expression


p
p py y2
ln (p + y p) = ln (p) + : (11.61)
p 2p

Then the approximate expression for the factorial becomes


Z 1
p p p
p! = p p ep ln(p+y p) p y p dy
p
Z 1 n h p
y2
i p o
Z 1 y2
p p ln(p)+ p
py
p y p p p ln(p) p
= p p
e 2p
dy = p p
e 2
dy
p p
Z 1 Z 1
p y2 p y2
= pep ln(p) p
p
e 2 dy ) p! = pep ln(p) p
p
e 2 dy: (11.62)
p p

Noting that Z 1 p
y2
e 2 dy = 2 (11.63)
1

we may write

p Z 1 Z p
p Z 1
y2 y2 y2
2 = e 2 dy = e 2 dy + p
e 2 dy
1 1 p
Z 1 Z 1 Z p
p p Z p
p
y2 y2 y2 y2
) p
e 2 dy = e 2 dy e 2 dy = 2 e 2 dy;
p 1 1 1
(11.64)

we can write
Z 1 p Z p
p
p y2 p y2
p! ' pep ln(p) p
p
e 2 dy = 2p ep ln(p) p
pep ln(p) p
e 2 dy:
p 1
(11.65)
11.3. STIRLING’S FORMULA 287

The second integral


Z p
p
y2
lim e 2 dy ! 0: (11.66)
p!1 1

Hence, the Stirling’s approximation for the factorial function will be


p p p p
p! ' 2p ep ln(p) p = 2p eln(p ) e p ' 2p pp e p : (11.67)

or with further approximation


p 1
ln(p!) ' ln 2p pp e p
= (ln + ln p + p ln p p)
2
) ln(p!) ' p ln p p: (11.68)

where we dropped the …rst two terms as compared to the last two terms.
Example 11.4 Consider a classroom full of gas molecules. There are approxi-
mately N = 5000NA = 3 1027 molecules in the room. From the Binomial
Theorem, it can be shown that the probability for n of the molecules to
be in the front half and n0 = N n molecules to be in the back half of
the room is given by

N 0 N!
P (n) = pn q n = pn q N n
; (11.69)
n n! (N n)!
where p is the probability that a molecule will be found in the front half
of the room, and q is the probability that it will be found in the back half.
From the symmetry of the problem, we must have
1 1
p= ;q = 1 p= =p (11.70)
2 2
On the average, we would expect to …nd half of the molecules in the front
half of the room and the other half of the molecules in the back half of the
room. Find the probability that 0:1% of the molecules in the room have
shifted from the front to the back half of the room. That is, …nd the value
of P (n) = P (0:499N ).
Solution: Since q = p, we can write
N! N!
P (n) = pn pN n
= pN : (11.71)
n! (N n)! n! (N n)!
On the average there are nave = 0:5N of molecules in the front half of the
room and n0ave = N nave = 0:5N in the back half of the room. Here we
want to …nd the probability that 0:1% of the molecules shifted to the back
half of the room. In other words we want to determine the probability
that the number of molecules in the front half (nave ) is reduced by 0:1%.
Which means we want to …nd P (n) for

n = 0:5N (N 0:1%) ) n = 0:499N (11.72)


288 CHAPTER 11. SPECIAL FUNCTIONS

which is given by
N!
P (n) = pN : (11.73)
n! (N n)!
Obviously, both N and n are very large number and we can make Stirling’s
approximation
ln n! = n ln n n (11.74)
for the factorial. Thus
N!
ln [P (n)] = ln pN = ln N ! ln n! ln(N n)!+ln pN (11.75)
n! (N n)!

can be approximated as

ln [P (n)] = N ln N N (n ln n n) ((N n) ln (N n) (N n))


+N ln p = N ln N N n ln n + n N ln (N n) + n ln (N n) + N n
+N ln p = N [ln N ln (N n) + ln p] + n [ln (N n) ln n]
N N n
= N ln p + n ln
N n n
N n N n
) ln [P (n)] ' n ln N ln : (11.76)
n Np

Substituting the values

N =3 1027 ; n = 0:499N ) N n = 0:501N; p = 0:5; (11.77)

we …nd

and
P (n) = exp[ 6 1021 ]: (11.78)

11.4 The Error Function


The error function is de…ned as the area under
Z x
2 2
erf(x) = p e t dt (11.79)
0

There are also other related integrals which sometimes referred as error func-
tions. These includes
11.4. THE ERROR FUNCTION 289

(a) The standard normal or Gaussian cumulative distribution function, (x);


Z x p
1 2 1 1
(x) = p e t =2 dt = + erf x= 2 : (11.80)
2 1 2 2
Noting that
Z x Z 0 Z x
1 2 1 2 1 2
p e t =2 dt = p e t =2 dt + p e t =2 dt
2 1 2 1 2 0
Z 1 Z x r Z x
1 2 1 2 1 1 2
=p e t =2 dt + p e t =2 dt = p +p e t =2 dt
2 0 2 0 2 2 2 0
Z x Z x
1 2 1 1 2
)p e t =2 dt = + p e t =2 dt (11.81)
2 1 2 2 0
The error function can also be expressed as
p Z x
1 1 1 t2 =2
erf x= 2 = (x) =p e dt: (11.82)
2 2 2 0

(b) The complementary error function:


Z 1 p
2 t2 =2
erf c(x) = p e dt = 1 erf x= 2 ,
x
r Z 1
x 2 t2 =2
) erf c( p ) = e dt (11.83)
2 x

Example 11.3 Consider a criterion that either is or is not satis…ed. We look


at a system that has many elements, each of which satis…es or does not
satisfy the criterion.
For example: Consider a test with many multiple-choice questions. Criterion:
the answer to a test question is correct. Each answer on the test is either
correct (satis…es criterion) or is incorrect (does not satisfy the criterion).
We then look at a large number of these systems (for example, a large number
of tests consisting of multiple-choice questions). We let x represent the
number of elements within a given system satisfying the criterion. We
then de…ne the following:
x The average number of elements satisfying the criterion
The standard deviation about the mean of the number of elements satis-
fying the criterion.
The probability that any one system will have x to x + dx elements satisfying
the criterion is then given by the Gaussian distribution:
1 (x x)2
(x)dx = p e 2 2 dx
2
290 CHAPTER 11. SPECIAL FUNCTIONS

Figure 11.1: Gaussian distribution.

Find an expression in terms of the error function for the probability that
the number of elements of a given system satisfying the criterion, x, will
be in the range
x n x x+n
for some real value of n (usually integral).

Solution: The probability that one system will have x to x + dx elements


satisfying the criterion is

1 (x x)2
(x)dx = p e 2 2 dx (11.84)
2
then the probability that the number of elements in the range x n
x x + n satisfying the criterion will be
Z x+n Z x+n
1 (x x)2
Pn (x) = (x)dx = p e 2 2 dx: (11.85)
x n x n 2

Introducing a new variable de…ned by


x x p
p = y ) dx = 2 dy (11.86)
2
and noting that for x1 = x n and x2 = x + n
x1 x x n x n
y1 = p = p = p
2 2 2
x2 x x+n x n
y2 = p = p =p (11.87)
2 2 2
11.4. THE ERROR FUNCTION 291

we can write
Z n
p
p Z n
p
2 1 y2 1 2
y2
Pn (x) = p e 2 dy = p e dy: (11.88)
pn
2
2 pn
2

If we split the integral in to two regions pn2 ; 0 and 0; pn2 ; we have


"Z Z pn #
0
1 2 2 2
Pn (x) = p e y dy + e y dy (11.89)
pn 0
2

and noting that


Z 0 Z 0 Z n
p
2
y2 y2 y2
e dy = e d ( y) = e dy (11.90)
pn n
p 0
2 2

we …nd Z n
p
2 2
y2
Pn (x) = p e dy: (11.91)
0
Recalling the de…nition for the error function
Z x
2 2
erf(x) = p e t dt (11.92)
0
we …nd that
n
p
Pn (x) = erf
: (11.93)
2
You can get the values of the error function for di¤erent values of n using,
for example, Mathematica. You will …nd the following results

n Pn (x)
1 68:26%
2 95:44%
3 99:74%
292 CHAPTER 11. SPECIAL FUNCTIONS

11.5 Elliptic Integrals


The Complete Elliptic Integral of the First Kind
Z =2
d'
K( ) p : (11.94)
2
0 1 sin2 '

Example 11.5 Consider a simple pendulum with a mass m suspended from


the end of a light rigid rod of length l. We pull the pendulum to the side
by an angle and release it from rest. Find an expression for the period
of the pendulum, T , assuming that = 0 and d =dt > 0 at t = 0, where
is the angle of the pendulum from the vertical. Then …nd an approximate
expression for the period of the pendulum for not-so-small amplitudes of
motion.

Figure 11.2: A simple pendulum. At the initial time, t = 0; the mass m was
displaced by an angle, ; from the vertical.

Solution: Using conservation of Mechanical energy,

M EI = M E (11.95)

where M EI is the initial mechanical energy (when the pendulum is pulled


to the side by an angle ) which is just only the gravitational potential
energy given by

M EI = mghmax = mgl(1 cos ) (11.96)


11.5. ELLIPTIC INTEGRALS 293

and M E is the mechanical energy at some instant of time (i.e. at an


angle ) which is the sum of kinetic and potential energy given by
2
1 1 d
M E = mgh + mv 2 = mgl(1 cos ) + ml2 : (11.97)
2 2 dt
Then
2
1 d
M EI = M E ) mgl(1 cos ) = mgl(1 cos ) + ml2
2 dt
2
r
1 d d 2g
) g cos = g cos ) + l ) = (cos cos ):
2 dt dt l
(11.98)
Noting that the period is the time for one complete oscillation which we
can express as Z t1=2
T =2 dt (11.99)
0
where t = 0 is the time at which the pendulum at the maximum displace-
ment from the vertical ( = ) and t = t1=2 is the time at which the
pendulum reached to the position ( = ). Therefore, using
r
d 2g d
= (cos cos ) ) dt = q (11.100)
dt l 2g
(cos cos ) l

we can write
Z t1=2 Z
d
T =2 dt = 2 q (11.101)
0 2g
l (cos cos )
Using the

cos = 1 2 sin2 ; cos =1 2 sin2 (11.102)


2 2
we have
r s
2g 2g
(cos cos ) = 1 2 sin2 1 + 2 sin2
l l 2 2
s
2g
= 2 sin2 2 sin2 (11.103)
l 2 2
so that
Z
d
T = 2 q
2g 2
l 2 sin 2 2 sin2 2
s Z s Z d
l d l sin( 2 )
= q = r (11.104)
:
g sin2 sin2 g sin2 ( 2 )
2 2 1 sin2 ( 2 )
294 CHAPTER 11. SPECIAL FUNCTIONS

Introducing the transformation


s
sin 2 sin2 2
sin ' = ) 1 = cos '
sin 2 sin2 2
1 cos 2 d 2 cos ' 2 cos '
) cos 'd' = d ) = d' = q
2 sin 2 sin 2 cos 2 1 sin2 2
d 2 cos '
) =q ; (11.105)
sin 2 1 sin2 sin2 '
2

= ) sin ' = 1)'= (11.106)


2
The expression for the period can be put in the form
s Z s Z
=2 =2
l d' l d'
T =2 q =4 p ;
g =2 1 sin2 2 sin2 ' g 0 1 2
sin2 '

where we introduced the constant


= sin (11.107)
2
and take into consideration the fact that sin2 ' is an even function such
that
Z =2 Z =2
d' d'
q =2 q :
2 2
=2 1 sin 2 sin ' 0 1 sin2 2 sin2 '

Therefore, the period is given by


s
l
T =4 K( ) (11.108)
g

where Z =2
d'
K( )= p (11.109)
2
0 1 sin2 '
is the elliptic integral of the …rst kind.

11.6 The Dirac delta function


Let’s consider a point in space that can be described by the Cartesian coordi-
nates (x; y; z) or the spherical coordinates (r; ; ') as shown in Fig. The position
vector
~r = x^
x + y y^ + z z^
= r sin ( ) cos (') x
^ + r sin ( ) sin (') y^ + r cos ( ) z^ = r^
r (11.110)
11.6. THE DIRAC DELTA FUNCTION 295

Figure 11.3: A point in space in spherical and Cartesian coordinates.

where
@~
r
@r
r^ = sin ( ) cos (') x
^ + sin ( ) sin (') y^ + cos ( ) z^ = @~
r
(11.111)
@r

is the unit vector along the radial direction. In spherical coordinates we recall
the gradient and the Laplacian for a scalar function, f (r; ; '), are given by
@f ^ 1 @f 1 @f
rf = r^ + +'
^ ; (11.112)
@r r@ r sin @'
and
1 @ @f 1 @ @f 1 @2f
r2 f = r2 + sin ( ) + 2 ; (11.113)
r2 @r @r r2 sin ( ) @ @ r2 sin ( ) @'2
respectively. On the other hand for a vector …eld,
V r + V (r; '; )^ + V' (r; '; )^
~ (r; '; ) = Vr (r; '; )^ '; (11.114)
instead of the gradient, most often, we are interested in the divergence of the
vector …eld. In spherical coordinates the divergence of a vector …eld is given by

~ = 1 @ r2 Vr +
r V
1 @
(sin ( ) V ) +
1 @V'
: (11.115)
r2 @r r sin ( ) @ r sin ( ) @'
Suppose we have some physical quantity described by some scalar function that
depend on the radial distance between two points. A good example of such phys-
ical quantity is electrical and gravitational potentials. Let’s say this function in
296 CHAPTER 11. SPECIAL FUNCTIONS

spherical coordinates can be expressed as


1
f (r; ; ') = ; for r > 0; (11.116)
r
and one can construct a vector …eld

~ =r 1 @ 1 r^
V = r^ = ; for r > 0: (11.117)
r @r r r2
The divergence of this vector …eld becomes

~ =r r^ 1 @ 1 0; for r > 0
r V = r2 = : (11.118)
r2 r2 @r r2 1 for r = 0;
There follows that
1 1 r^ 0; for r > 0
r2 =r r =r = (11.119)
r r r2 1 for r = 0;
which leads to
ZZZ ZZZ
~ d = r^
r V r d = 0; for r > 0 (11.120)
volum e volum e r2
Z1Z Z2
r^
) r d = 0; for r > 0 (11.121)
r2
0 0 0

On the other hand if one determines using the divergence theorem


ZZZ ZZ
r V ~ d = ~ (x; y; z) d~a
V (11.122)
volum e Surface
enclosing

we …nd
ZZ
~ (x; y; z) d~a
V
Surface
enclosing
Z Z2 Z Z2
r^ 2
= r^r sin ( ) cos (') d d' = sin ( ) cos (') d d' = 4
r2
0 0 0 0
(11.123)
This none zero value must be a result when r = 0 is included. Therefore, one
can write
Z1Z Z2
r^ 0 for r > 0
r d = (11.124)
r2 4 for r = 0
0 0 0

The function, (~r) ; de…ned as


r^ 1 1 0 for r 6= 0
4 (~r) = r = r r = r2 = (11.125)
r2 r r 1 for r = 0
11.6. THE DIRAC DELTA FUNCTION 297

is known as the Dirac delta function. Then for the Dirac delta function
Z1Z Z2
(~r) d = 1;
0 0 0
Z1Z Z2 Z1Z Z2
) f (0) (~r) d = f (0) ) f (~r) (~r) d = f (0) ; (11.126)
0 0 0 0 0 0

For one dimensional case,


0 for x 6= 0
(x) = (11.127)
1 for x = 0
and
Z1 Z1
(x) dx = 1; f (x) (x) dx = f (0) (11.128)
1 1

In the usual sense of functions the Dirac delta function does not exist. But
there are various forms of sequence functions in the limiting case display the
properties of the Dirac delta function. These functions include:

1. The Gaussian function:


n
n (x) = p exp( n2 x2 ); (11.129)

See Fig. 1.

2. Lorentz function
n 1
n (x) = ;
1 + n2 x2
See Fig. 2.
298 CHAPTER 11. SPECIAL FUNCTIONS

3. The sinc function:


Z n
sin (nx) 1
n (x) = = eixt dt
x 2 n

See Fig. ??.The general form of the Dirac delta function for, x = a

0 for x 6= a
(x a) =
1 for x = a

and

Z1 Z1
(x a) dx = 1; f (x) (x a) dx = f (a) : (11.130)
1 1
11.6. THE DIRAC DELTA FUNCTION 299

For 3D case
Z1Z Z2
(~r) r2 sin ( ) drd d' = 1
0 0 0
Z1Z Z2
f (~r) (~r) r2 sin ( ) drd d' = f (0)
0 0 0
Z1Z Z2
f (~r ~r0 ) (~r ~r0 ) r2 sin ( ) drd d' = f (~r0 ) (11.131)
0 0 0

Example 11.6 Form introductory physics, the electric potential, V (~r) ; due to
a point charge located at the origin (0; 0; 0) (i.e. r = 0) at a point in space
described by the position vector, ~r; is given by
1 q
V (~r) = :
4 0 r

Show that the volume charge density, (~r) ; for this point charge can be
expressed in terms of the Dirac delta function
dq
(~r) = = q (~r) = q (x) (y) (z) ; (11.132)
d
where dq an in…nitessimal charge in an in…nitesimal volume d :
Solution: The electric potential, dV (~r) of an in…nitessimal charge dq 0 in a
volume d 0 as shown in Fig. can be expressed as
ZZZ
1 dq 0 1 (~r0 ) d 0 1 (~r0 ) d 0
dV (~r) = 0j
= 0j
) V (~r) = ;
4 0 j~
r ~
r 4 0 j~
r ~
r 4 0 j~r ~r0 j
V
(11.133)
Using spherical coordinates, we can write
Z1Z Z2
1 (~r0 ) r02 sin ( ) dr0 d 0 d'0
V (~r) = (11.134)
4 0 j~r ~r0 j
0 0 0

This potential for a point charge becomes


Z1Z Z2
1 (~r0 ) r02 sin ( ) dr0 d 0 d'0 1 q
=
4 0 j~r ~r0 j 4 0r
0 0 0
Z1Z Z2
1 (~r0 ) 02 1
) r sin ( ) dr0 d 0 d'0 = : (11.135)
j~r ~r0 j q r
0 0 0
300 CHAPTER 11. SPECIAL FUNCTIONS

From the property of the Dirac Delta function

Z1Z Z2
0
f (~r0 ) (~r0 ~r0 ) r02 sin dr0 d 0 d'0 = f (~r0 ) (11.136)
0 0 0

one can easily …nd

1 (~r0 )
f (~r0 ) = ; (~r0 ~r0 ) =
j~r ~r0 j q
1 1
) f (~r0 ) = = ) ~r0 = 0 (11.137)
j~r ~r0 j r

which leads to
(~r0 )
= (~r0 ) ) (~r0 ) = q (~r0 ) : (11.138)
q

Example 11.7 The volume charge density, (~r) ; of a point charge, q, placed
at a point on the z-axis, ~r0 = a^
z ; can be expressed as

(~r) = q (~r ~r0 ) ; (11.139)

where (~r) is the Dirac Delta function. Show that the electric potential,
V (~r) ; due to this point charge is given by

1 q q 1
V (~r) = = q : (11.140)
4 0 j~r ~r0 j 4 0 2
x2 + y 2 + (z a)

The electric potential for a volume charge distribution is given by


ZZZ
1 (~r0 ) d 0
V (~r) = ; (11.141)
4 0 j~r ~r0 j
V

where ~r0 is the position of the in…nitesimal charge dq 0 = (~r0 ) d 0 ; in an


in…nitesimal volume d 0 , (~r0 ) is the charge density in the volume V .

Solution: Using the given charge density and the expression for the potential,
one can write
ZZZ ZZZ
q (~r0 ~r0 ) d 0 q (~r0 ~r0 ) d 0
V (~r) = 0
= ; (11.142)
4 0 j~r ~r j 4 0 j~r ~r0 j
V V

In Cartesian coordinates, we have

~r0 = x0 x
^ + y 0 y^ + z 0 z^; ~r = x^x + y y^ + z z^; ~r0 = a^
z
q
0 2 2 2
) j~r ~r j = (x x0 ) + (y y 0 ) + (z z 0 ) (11.143)
11.6. THE DIRAC DELTA FUNCTION 301

and

(~r0 ~r0 ) = (x0 x0 ) (y 0 y0 ) (z 0 z0 )


0 0 0
= (x ) (y ) (z a) (11.144)

so that
Z1 Z1 Z1
q (x0 ) (y 0 ) (z 0 a) dx0 dy 0 dz 0
V (~r) = q
4 0 2 2 2
1 1 1 (x x0 ) + (y y 0 ) + (z z0)
Z1 Z1 Z1
q 0 0 0 0
= (z a) dz (y ) dy f (x0 ; y 0 ; z 0 ) (x(11.145)
0
) dx0 ;
4 0
1 1 1

where
1
f (x0 ; y 0 ; z 0 ) = q : (11.146)
2 2 2
(x x0 ) + (y y 0 ) + (z z0)

Now applying the property of the Dirac Delta function


Z1
f (x) (x a) dx = f (a) (11.147)
1

one can easily see that


Z1 Z1
0 0 0 0 0
f (x ; y ; z ) (x ) dx = f (x0 ; y 0 ; z 0 ) (x0 0) dx0
1 1

0 0 1
= f (0; y ; z ) = q : (11.148)
2 2
x2 + (y y0 ) + (z z0)

The electric potential becomes


Z1 Z1
q 0 0
V (~r) = (z a) dz f (0; y 0 ; z 0 ) (y 0 ) dy 0 : (11.149)
4 0
1 1

Once again using the property of the Dirac delta function, we have
Z1 Z1
0 0 0 0
f (0; y ; z ) (y ) dy = f (0; y 0 ; z 0 ) (y 0 0) dy 0
1 1
1
= f (0; 0; z 0 ) = q : (11.150)
2
x2 + y 2 + (z z0)
302 CHAPTER 11. SPECIAL FUNCTIONS

and the expression for potential reduces to


Z1
q
V (~r) = f (0; 0; z 0 ) (z 0 a) dz 0 : (11.151)
4 0
1

One last time using the Dirac delta function property, we …nd for the
potential
Z1
q q
V (~r) = f (0; 0; z 0 ) (z 0 a) dz 0 = f (0; 0; a)
4 0 4 0
1
q 1
) V (~r) = q : (11.152)
4 0 2
x2 + y 2 + (z a)
Chapter 12

Power Series Solutions to


Di¤erential Equations

12.1 Power series substitution


The di¤erential equations we seek to …nd the solution to are linear di¤erential
equation like those we studied in PHYS 3150. However, unlike those equa-
tions here the coe¢ cients in the di¤erential equations we shall consider are not
constants and depend on the variable, x; like the following di¤erential equation

d2 y (x) dy (x)
+ f (x) + g(x)y(x) = 0: (12.1)
dx2 dx
Such kind of di¤erential equations or even those with constant coe¢ cients can
be solved using series substitution method. The method involves a simple pro-
cedure but a little too much algebra. We assume the solution to the di¤erential
equation can be expressed as a power series
1
X
y (x) = an xn = a0 + a1 x + a2 x2 + a3 x3 + :::an xn + ::: (12.2)
n=0

We then substitute this series into the di¤erential equation and determine the
expansion coe¢ cients. We will demonstrate the application of this method using
the following example which we already determined the solution using a di¤erent
method last semester.

Example 12.1 A mass m is attached to a horizontal spring of spring constant,


k; whose other end is attached to a rigid vertical wall. The mass slides
on a horizontal, frictionless surface. At time, t = 0; the mass is stretched
from its equilibrium position by a distance x = xmax and released from
rest. Find the equation for the position of the mass as a function of time,
x(t).

303
304CHAPTER 12. POWER SERIES SOLUTIONS TO DIFFERENTIAL EQUATIONS

Solution: Using Newton’s second law the equation of motion for the mass, m;
can be written as
d2 x (t)
Fnet = ma ) m = kx; (12.3)
dt2
which we may rewrite as
d2 x (t)
+ ! 2 x = 0; (12.4)
dt2
where r
k
;!= (12.5)
m
is the angular frequency. Let’s assume that the solution to this di¤erential
equation is given by the power series
1
X
x (t) = an tn = a0 + a1 t + a2 t2 + a3 t3 + :::an tn + ::: (12.6)
n=0

so that
1
X 1
dx (t) d (tn ) X
= an = nan tn 1
= a1 + 2a2 t + 3a3 t2 +
dt n=0
dt n=0
X1
:::nan tn 1
+ ::: = (n + 1) an+1 tn (12.7)
n=0

and
1 1
d2 x (t) X d2 (tn ) X
= an = n (n 1) an tn 2
dt2 n=0
dt2 n=0
= 2a2 + 3 2a3 t + 4 3a4 t2 + :::n (n 1) an tn 2
+ :::
X1
= (n + 2) (n + 1) an+2 tn (12.8)
n=0
12.1. POWER SERIES SUBSTITUTION 305

The di¤erential equation can then be rewritten as

1
X 1
X
(n + 2) (n + 1) an+2 tn + ! 2 an tn = 0
n=0 n=0
1
X
) (n + 2) (n + 1) an+2 + ! 2 an tn = 0: (12.9)
n=0

From the above equation we …nd the following recursion relation

an
(n + 2) (n + 1) an+2 + ! 2 an = 0 ) an+2 = !2 :
(n + 2) (n + 1)

Now let’s examine the …rst few terms for this recursion relation. First we
consider when n is even

a0 2
n = 0 ) a2 = !
2 1
1
a0 ( 1) 2 1
) n = 0 ) a2 1 = !
2!
2 2
a2 2 a0 ( 1) ! 2 2
a0 ( 1) ! 2 2
n = 2 ) a4 = ! = =
4 3 4 3 2 1 4!
a0 2
) n = 2 ) a2 2 = ( 1) ! 2 2
(2 2)!
3 3
a4 ! 2 a0 ( 1) ! 2 3 a0 ( 1) ! 2 3
n = 4 ) a6 = = =
6 5 6 5 4 3 2 1 6!
3 2 3
a0 ( 1) !
) n = 4 ) a2 3 = : (12.10)
(2 3)!

From the results we see above it is not hard to come up with the relation

m
a0 ( 1) ! 2m
a2m = (12.11)
(2m)!

that generates the values for the even term coe¢ cients in the series for
m = 0; 1; 2; ::.

Next we shall consider the odd terms for the recursion relation

an
an+2 = !2 : (12.12)
(n + 2) (n + 1)
306CHAPTER 12. POWER SERIES SOLUTIONS TO DIFFERENTIAL EQUATIONS

For the …rst three odd terms, we have


1 1
a1 ( 1) 2 a1 ( 1) 2
n = 1 ) a3 = ! = !
3 2 3!
1 2 1
a1 ( 1) !
) n = 1 ) a(2 1+1) =
[(2 1) + 1]!
2 2
a3 2 a1 ( 1) ! 4 a1 ( 1) ! 4
n = 3 ) a5 = ! = =
5 4 5 4 3 2 1 5!
2 2 2
a1 ( 1) !
) n = 3 ) a(2 3 1) =
(2 2 + 1)!
3 3
a5 ! 2 a1 ( 1) ! 6 a1 ( 1) ! 2 3
n = 5 ) a7 = = =
7 6 7 6 5 4 3 2 1 7!
3
a1 ( 1) ! 2 3
) n = 5 ) a(2 4 1) = (12.13)
(2 3 + 1)!

Then for the odd terms we can write


m
a1 ( 1) ! 2m
a(2m+1) = ; (12.14)
(2m + 1)!

where m = 0; 1; 2; 3:::Thus the general solution to the di¤erential equation


1
X 1
X 1
X
x (t) = an tn ) x (t) = a2m t2m + a2m+1 t2m+1 ; (12.15)
n=0 m=0 m=0

using the results obtained above, can be expressed as


1
X X1 1
a0 ( 1) ! 2m 2m X a1 ( 1) ! 2m 2m+1
m m
x (t) = an tn = t + t
n=0 m=0
(2m)! m=0
(2m + 1)!
X1 1
a1 X ( 1)
m m
( 1) 2m 2m+1
) x (t) = a0 (!t) + (!t) (12.16)
m=0
(2m)! ! m=0
(2m + 1)!

We recall that the Taylor series expansions for sin (x) and cos (x) are
1
X k X1 n
( 1) x2k+1 ( 1) x2n
sin (x) = ; cos (x) = (12.17)
(2k + 1)! n=0
(2n)!
k=0

so that the solution to the di¤erential equation can be put in the form
a1
x (t) = a0 cos (!t) + sin (!t) (12.18)
!
or
x (t) = A cos (!t) + B sin (!t) ; (12.19)
12.2. ORTHOGONAL VECTORS AND DIRAC NOTATION 307

where A = a0 and B = a1 =! are constants determined by the initial


conditions. Since initially (t = 0) the spring is stretched by xmax and the
mass is released from rest, we have

x (t = 0) = a0 cos (0) a1 ! sin (0) = xmax ) a0 = xmax (12.20)

and
dx (t)
= a0 sin (!t) a1 ! 2 cos (!t) ;
dt
dx (0)
= 0 ) a1 ! 2 = 0 ) a1 = 0: (12.21)
dt
Therefore, the equation for the position of the mass as a function of time,
x(t) is given by
x (t) = xmax cos (!t) ; (12.22)
where r
k
!= ; (12.23)
m
is the angular frequency.

12.2 Orthogonal vectors and Dirac Notation


~ and B;
We recall that if the scalar product of two real vectors, A ~ is zero, given
by
X3
~ B
A ~ = Ai Bi = 0; (12.24)
i=1

the two vectors are said to be orthogonal. If these vectors are complex and
orthogonal, we must write
3
X
~
A ~ =
B Ai Bi = 0; (12.25)
i=1

~ is the complex conjugate of A:


where A ~ Using Dirac notation any vector A ~ is
~
denoted by a ket vector jAi and its complex conjugate A by a bra vector hAj :
Then the dot product of two vectors, using Dirac notation, can be expressed as
3
X
hA jBi = Ai B i (12.26)
i=1

and if the vectors are orthogonal,


3
X
hA jBi = Ai Bi = 0: (12.27)
i=1
308CHAPTER 12. POWER SERIES SOLUTIONS TO DIFFERENTIAL EQUATIONS

Orthonormal Sets of Functions: two di¤erent functions A(x) and B(x) are
said to be orthogonal for all x (a; b) when
Z b
A (x)B(x)dx = 0; (12.28)
a

where A (x) is the complex conjugate of the function A(x): A set of functions

fA1 (x) ; A2 (x) ; A3 (x) :::An (x) :::g (12.29)

that meet the requirement


Z b
0 m 6= n;
An (x)Am (x)dx = (12.30)
a C m = n;

where C is a constant, are said to form an orthogonal set of functions. Using


the Kronicker delta
0 n=6 m
nm = ; (12.31)
1 n=m
and Dirac notation, we may write

hAn (x) jAm (x)i = Cn nm (12.32)

For m = n,
hAn (x) jAn (x)i = Cn (12.33)
so that the for the set of functions de…ned by

An (x)
Fn (x) = p (12.34)
Cn
one can write
Z b
hFn (x) jFm (x)i = Fn (x) Fm (x) dx = nm : (12.35)
a

These set of functions are said to form an orthonormal set of functions.

Example 12.2 Show that the set of functions de…ned by


1
Fn (x) = p sin (nx)

for n = 1; 2; 3::::form an orthonormal set of functions for all x ( ; ).


Solution: Using Euler’s formula, we can express

1 eimx e imx
e inx
einx
Fm (x) = p sin (mx) = p ) Fn (x) = p
2i 2i
(12.36)
12.2. ORTHOGONAL VECTORS AND DIRAC NOTATION 309

so that Z b
hFn (x) jFm (x)i = Fn (x) Fm (x) dx (12.37)
a

becomes
Z inx
e einx eimx e imx
hFn (x) jFm (x)i = p p dx
2i 2i
Z
1
= ei(m n)x e i(m+n)x
ei(m+n)x + e i(m n)x
dx
4
1 ei(m n)x e i(m+n)x ei(m+n)x e i(m n)x
= +
4 i m n m+n m+n m n
1 4 sin ((m n) ) 4 sin ((m + n) )
) hFn (x) jFm (x)i = ;
4 m n m+n
(12.38)

whether m = n or m 6= n the second term is always zero and we can write

sin [(m n) ]
hFn (x) jFm (x)i = = nm ;
(m n)

where we have applied applied L’Hospital’s rule. But it would be shorter


if one applies the double angle relations

cos ( + ) = cos ( ) cos ( ) sin ( ) sin ( ) ;


cos ( ) = cos ( ) cos ( ) + sin ( ) sin ( ) ;

that give
1
sin ( ) sin ( ) =[cos ( ) cos ( + )] :
2
In view of these relation, one can write the integral as
Z
hFn (x) jFm (x)i = sin (nx) sin (mx) dx
Z
1
= [cos ((n m) x) cos ((n + m) x)] dx (12.39)
2
1 sin ((n m) x) sin ((n + m) x)
= (12.40)
2 n m n+m
1 sin ((n m) ) sin ((n + m) )
= (12.41)
n m n+m

whether m is equal to or di¤erent from n; sin ((n + m) ) = 0; thus

1 sin ((n m) )
hFn (x) jFm (x)i = = nm
n m
310CHAPTER 12. POWER SERIES SOLUTIONS TO DIFFERENTIAL EQUATIONS

12.3 Complete Sets of Functions


We recall a set of functions, Fn (x) ; form an orthonormal set for all x (a; b) ;
when
Z b
hFn (x) jFm (x)i = Fn (x) Fm (x) dx = nm : (12.42)
a

Next we shall study when such kind of set of functions form a complete set for
all x (a; b) in the function space. But before we do that, we want to review
complete set of vectors in vector space.
Complete sets of vectors in vector space: Consider a three dimensional real
space R3 in Cartesian coordinate system. We recall the unit vectors x ^; y^; and z^
or x^1 ; x
^2 ; and x
^3 form an orthonormal set of vectors in 3-D vector space since

x
^n x
^m = nm (12.43)

or using Dirac notation


hxn j xm i = nm : (12.44)
Any vector ~r R3 can be expressed in terms of these three unit vectors
3
X
~r = ri x
^i (12.45)
i=1

or using Dirac notation


3
X
jri = ri jxi i : (12.46)
i=1

Multiplying both sides from the left by hxj j ;

3
X
hxj jri = ri hxj jxi i : (12.47)
i=1

Since the vectors jxi i form an orthonormal set of vectors

3
X
hxj jri = ri ji ) rj = hxj jri : (12.48)
i=1

If one of the vectors, x^i ; is missing, then we can not express any vector ~r R3
in terms of the remaining two vectors. But if we have all these three vectors
any vector ~r R3 can be expressed in terms of these vectors. Then we say that
the set of vectors f^
xi g = f^ x1 ; x ^3 g is a complete orthonormal set in R3 :
^2 ; x
We now consider the function space where we have the in…nite set of or-
thonormal functions fFn (x)g = fF0 (x) ; F1 (x) ; F2 (x) ; F3 (x) ; :::g for all x
(a; b) : These set of functions form a complete set if a function, g (x) ; de…ned
12.3. COMPLETE SETS OF FUNCTIONS 311

for all x (a; b) can be expressed as a linear combination of these orthonormal


functions. That means
1
X
jg (x)i = cn jFn (x)i ; (12.49)
n=0

where the expansion coe¢ cie