Documente Academic
Documente Profesional
Documente Cultură
Faculty of Engineering
Engineering Mathematics
Part 1
Part 1
2 Matrices 42
3 Calculus 74
Equations
5 Linear Differential Equations 151
Of Higher Order
6 Laplace Transforms 190
Contents
Part 2
Chapter Title Page
No. No.
9 Power Series Solution Of 285
Differential Equations
10 Partial Differential Equations 346
Differential Equations
12 Special Functions 459
Appendix 543
References 574
Chapter 1
Mathematical Numbers
Fig.1
Every point on the line corresponds to a real number, and every
real number can be paired with a point on this number line. If the
real number is an integer, its point on the number line coincides with
one of the notches for an integer; otherwise, its point lies between
two successive notches. All real numbers represented by points to
Chapter One 3
the right of the number 0 are positive, while all real numbers
represented by points to the left of the number 0 are negative.
Fig.2
Example 1:
|2|=2
| -4.5 | = 4.5
|0|=0
4 Mathematical Numbers
1.6 Complex Numbers
1.6.1 Introduction
− b ± b 2 − 4ac
obtained by the famous formula, x =
2a
For example, if x 2 + x − 2 = 0 , then we have:
−1± (1 + 8) −1± 9 −1± 3
x= = =
2 2 2
∴ x = 1 or −2
As we see there is no problems with solving the above equation. But
5x 2 − 6x + 5 = 0
6± (36 − 100) 6 ± − 64 6 ± j8
x= = =
10 10 10
∴ x = 0.6 + j 0.8 or x = 0.6 − j 0.8
1.6.2 Powers of j
j = −1
j 2 = −1
( )
j 3 = j 2 * j = −1 * j = − j
j 4 = ( j 2 ) = (− 1)2 = 1
2
(2 + j3)(5 + j 7 ) = 2 * 5 + j3 * 5 + j 2 * 7 + j 2 3 * 7
= 10 + j15 + j14 + j 2 21
= 10 + j 29 − 21
= −11 + j 29
If the expression contains more than two factors, we multiply the
factors together in stages:
(2 + j3)(5 + ( )
j 7 )(1 − j 2) = 10 + j15 + j14 + j 2 21 (1 − j 2)
= (10 + j 29 − 21)(1 − j 2)
= (− 11 + j 29 )(1 − j 2)
= −22 + j 22 + j 29 − j 2 58
= −22 + j 51 + 58 = 36 + j 51
(5 + j8)(5 − j8) = 25 + j 40 − j 40 − j 2 64
= 25 + 64 = 89
In spite of what we said above, here we have a result containing no
imaginary term. The result is therefore entirely real. This is rather an
exceptional case. If we look at the two complex numbers we can
8 Mathematical Numbers
find that they are identical except for the middle sign in the brackets
are different. These two complex numbers called conjugate complex
numbers and the product of two conjugate complex numbers is
always entirely real. In general we can say:
-4 4
-4 -3 -2 -1 0 1 2 3 4
Fig.3
A line which represents a magnitude (by its length) and direction
(by the arrow head) is called a vector. We shall be using this word
quite a lot. Any vector therefore must include both magnitude (or
size) and direction. If we multiply (+4) by the factor (-1), we get
(-4), i.e. the factor (-1) has the effect of turning the vector through
-4 -3 -2 -1 0 1 2 3 4
Fig.4
the effect and rotate the vector through only 90o . So, the factor j
3
j4
4 2
o
3 1 180
-4 4
j4
2 -4 -3 -2 -1 0 1 2 3 4
-1
o
1 180 -2
-4 4 -j4
-3
-4 -3 -2 -1 0 1 2 3 4 -4
Fig.5 Fig.6
Let us denote the two reference lines by XX, and YY, as usual.
You will see that:
(i) The scale on the X-axis represents real numbers, XX1 is
therefore called the real axis.
12Mathematical Numbers
(ii) The scale on the Y-axis represents imaginary numbers, YY1 is
therefore called the imaginary axis.
If we now wish to represent 2+ 3 as the sum of two vectors, we
must draw them as a chain, the second vector starting where the first
one finishes as shown in Fig.7.
5
2 3
0 1 2 3 4 5
Fig.7
The two vectors, 2 and 3 are together equivalent to a single vector
drawn from the origin to the end of the final vector (giving naturally
that 2+3=5).
If we wish to represent the complex number (3 + j2), then we add
together the vectors which represent 3 and j2. Notice that the 3 is
2 j3
1
2
0 1 2 3
Fig.8
Chapter One 13
Example 6 Draw an Argand diagram to represent the following
vectors: z1 = 3 + j 2 , z 2 = −3 + j1 , z3 = 2 − j 4 , and z 4 = −4 − j 4
Solution: The Argand diagram of the above vectors are shown in
the following Fig.9.
3
2 z1
j1
z2 j2
1
3
-4 -3 -2 3
-1 0 1 2 3 4
-1
j4
j4
-2
-j4
-3
z4 -4 z3
Fig.9.
2 z1
j2
1
3
0 1 2 3 4 5
-1 z1 + z2
-2
-j4
-3
-4
z2
Fig.10
5 z1 − z 2
4 u
Fig.11
3
− z2
2 z1
j2
1
3
3 -2 3
-1 0 1 2 3 4
-1
-2
-j4
-3
-4
z2
1.6.8 Polar Form of a Complex Numbers
Complex numbers in the form a + jb is called rectangular form.
Sometimes, it is convenient to express it in a different form. On an
Argand diagram shown in Fig.12, let OA be a vector a + jb . Let r
= length of the vector and θ the angle made with OX. Since
z = a + jb , this can be written z = r cos θ + jr sin θ or
r
b
θ
o a x
Fig.12
Example 8 Express z = 4 + j 3 in polar form.
Solution: First draw a sketch diagram (that always helps). We can
3
see that: r = 4 2 + 32 = 16 + 9 = 5 and θ = tan −1 = 36.87 o
4
(
z = a + jb = r (cos θ + j sin θ ) ∴ z = 5 cos 36.87 o + j sin 36.87 o )
(i) r is called the modulus of the complex number z and is often
abbreviated to mod( z ) or indicated by z
( )
Thus if z = 3 + j 4 , ∴ z = 32 + 4 2 = 9 + 16 = 5
(ii) θ is called the argument of the complex number and can be
Second First
90 o < θ < 180o 0 < θ < 90o
o
o x
o o
o
180 < θ < 270 o
270 < θ < 360
Third Fourth
Fig.13
Example 9 Find arg( z ) when z = −3 − j 4
Solution: The vector has been drawn as shown in Fig. 14.
θ is measured from OX to OP. We first find E the equivalent
acute angle from the trinngle shown.
4
tan E = = 1.3333 then E = 53.13o
3
18Mathematical Numbers
But from Fig.14. This angle is 180 o < θ < 270 o then we have
∴θ = 180o + E = 233.13o
Fig.14
Example 10 Find arg( z ) when z = −5 + j 2
Solution: The vector has been drawn as shown in Fig.15.
Fig.15
2
∴ tan E = = 0.4 ∴ E = 21.8o
5
In this particular case, θ = 180 − E o ∴θ = 158.2 o
Complex numbers in polar form are always of the same shape and
differ only in the actual values of r and θ . We often use the
Chapter One 19
shorthand version r∠θ o to denote the polar form as shown in the
following examples:
Then, if z = −5 + j 2 r = (25 + 4) = 29 = 5.385 and from
above θ = 158.2 . Then, the full polar form is
( )
z = 5.385 cos158.2 o + j sin 158.2 o and this can be shortened to
z = 5.385∠158.2o .
Solution: r = (42 + 32 ) = 5
tan E = 0.75 , ∴ E = 36.87 o ∴θ = 360 − E = 323.13o
( )
∴ z = 5 cos 323.13o + j sin 323.13o = 5∠323.8o
Of course, given a complex number in polar form, you can convert
it into the basic rectangular form a + jb simply by evaluating the
cosine and the sine and multiplying by the value of r.
Solution:
( )
z = 5 cos 35o + j sin 35o = 5(0.8192 + j 0.5736)
= 4.096 + j 3.868
e jθ
=1+ jθ +
( jθ )2 ( jθ )3 ( jθ )4 ( jθ )5
+ + + (4)
2! 3! 4! 5!
jθ j 2 (θ )2 j 3 (θ )3 j 4 (θ )4 j 5 (θ )5
∴e = 1 + jθ + + + + ++
2! 3! 4! 5!
∴e jθ
= 1+ jθ −
(θ )2 j (θ )3 (θ )4 j (θ )5
− + + −−
2! 3! 4! 5!
jθ
(θ )2 (θ )4 j (θ )3 j (θ )5
∴e = 1 − + − .... +
j θ− + − .... (5)
2! 4! 3! 5!
It is clear from (2),(3) and (5) that the first bracket is in the form of
cosine and the second bracket in the form of sine.
e jθ = cos θ + j sin θ
if we replace θ by − θ in this result, we get the following:
e − jθ = cos(− θ ) + j sin (− θ )
= cos θ − j sin θ
r= (42 + 32 ) = 5
Fig.16
3
From this r = 5 , tan E = − = −0.7, ∴ E = −36.87 o
4
∴θ = 360 o − 36.87 o = 323.13o
(
Then, the polar form is z = 5 cos 323.13o + j sin 323.13o . )
j 323.13o
And the polar form is z = 5.e And the the shortened
Fig.17.
But the direction of the vector, measured from OX, could be given
( ( ) (
direction. We could write z = 5 cos − 36.87 o + j sin − 36.87 o . ))
But you already known as cos(− θ ) = cos(θ ) and
sin (− θ ) = − sin (θ ) .
( ( ) (
∴ z = 5 cos 36.87 o − j sin 36.87 o ))
i.e. very much like the polar form but with a minus sign in the
middle. This comes about whenever we use negative angles.
In the same way we can say the following:
( ) ( ( )
z = 5 cos 250o + j sin 250o = 5 cos − 110o + j sin − 110o ( ))
It is sometimes convenient to use this form when the value of θ is
greater than 180 o , i.e. in the 3rd and 4th quadrants. In the same
way we can write the following:
24Mathematical Numbers
( ) ( )
z = 3 cos 230o + j sin 230o = 3 cos130o − j sin130o = 3∠ − 130o
z = 4(cos 290o + j sin 290o ) = 4(cos 70 o − j sin 70 o ) = 4∠ − 70 o .
The polar form at first sight seems to be a complicated way of
representing a complex number. However it is very useful as we
shall see.
Suppose we multiply together two complex numbers in this form:
Let z1 = r1 (cos θ1 + j sin θ1 ) and z 2 = r2 (cos θ 2 + j sin θ 2 )
( )
2∠30 * 3∠40 = (2 * 3)∠ 30o + 40 o = 6∠70o
Now let us see if we can discover a similar set of rules for
division. We already know that to simplify 5 + j 4 we first
obtain a denominator that is entirely real by multiplying top and
bottom by the conjugate of the denominator i.e (5 − j 4 ) Right.
10∠95o
Example 16 Simplify the following expressions: ,
2∠44 o
Solution:
10∠95o
2∠44 o
( )
= 5∠ 95o − 44 o = 5∠51o
26Mathematical Numbers
1.6.10 DeMoivre’s Theorem
There is very important rule is called DeMoivre’s Theorem. It says
that to raise a complex number in polare form to any power n, we
raise the r to the power n and multiply the angle by n.
This is where the polar form really comes into its own. For
DeMoivre's theorem also applies when we are raising the complex
number to a fractional power, i.e. when we are finding the roots of a
complex number as shown in the following example.
( ) (
cos 3θ + j sin 3θ = c 3 − 3cs 2 + j 3c 2 s − s 3 )
Now equating real parts and imaginary parts we get:
( )
cos 2 θ = 1 − sin 2 θ . So that we could write the results above as
= c 4 + j 4c 3 s − 6c 2 s 2 − j 4cs 3 + s 4
( ) (
= c 4 − 6c 2 s 2 + s 4 + j 4c 3 s − 4cs 3 )
Equating real parts:
(
∴ cos 4θ = c 4 − 6c 2 s 2 + s 4 )
( ) (
= c 4 − 6c 2 1 − c 2 + 1 − c 2 )2
= c 4 − 6c 2 + 6c 4 + 1 − 2c 2 + c 4
= 8c 4 − 8c 2 + 1
= 8 cos 4 θ − 8 cos 2 θ + 1
( ) (
Similarly, sin 4θ = 4c 3 s − 4cs 3 = 4cs c 2 − s 2 )
(
= 4cs c 2 − s 2 − c 2 − s 2 + 1 )
= 4cs (1 − s 2 ) = 4cs * c 2 = 4c 3 s
1 1
z+ = 2 cos θ z− = j 2 sin θ
z z
1 1
z n + n = 2 cos nθ z n − n = j 2 sin nθ
z z
1 1 1
∴ (2 cos θ )3 = z 3 + 3z 2 + 3z 2 + 3
z z z
1 1
= z 3 + 3z + 3 + 3
z z
Now here is the trick: we rewrite this, collecting the terms up in
pairs from the two extreme ends, thus:
(2 cos θ )3 = z 3 + 1
+ 3 z +
1
z3 z
But from the previous results
1 1
∴z + = 2 cos θ , and z 3 + = 2 cos 3θ
z 3
z
30Mathematical Numbers
∴ (2 cos θ )3 = 2 cos 3θ + 3 * 2 cos θ
equal if reduced to the proper range. So, for example 10o and 370 o
1.7 Polynomials
We have learned how to manipulate complex numbers, and
suggested that they will prove valuable in engineering calculations.
The original motivation for introducing them was to give the
p ( x ) = an x n + an −1 x n −1 + .... + a1 x + ao
Chapter One 35
where the a's are (real or complex) numbers and an ≠ 0 . For
example: p ( x ) = x 3 − 2 x + 4, q (t ) = 5t 8 − t 4 + 6t 3 − 1
The highest power in the polynomial is called the degree of the
polynomial. The above examples have degrees 3 and 8.
A number a (real or complex) is said to be a root of the polynomial
p( x ) if p (a ) = 0 . Thus x = 1 is a root of x 2 − 2 x + 1 = 0
The first important result about polynomials is that a number a (real
or complex) is a root of the polynomial p ( x ) if and only if ( x − a )
(
p(x ) = (x − 2) x 2 + 3x − 1 )
Note that necessarily the polynomial q has degree one less than the
degree of p. It may be the case that you can pull more than one
factor of x − a out of the polynomial. For example, 2 is a root of
of the root is the number of factors ( x − a ) that you can take out. In
the above example, 2 is a root of multiplicity 2, or a double root. A
36Mathematical Numbers
root is called a simple root if it produces only one factor. Multiple
roots are a considerable pain in the neck in many applications.
There is a simple test for multiplicity. Suppose a is a root of
p( x ) , so that p (a ) = 0 . If, in addition, p′(a ) = 0 (derivative) then a
is a multiple root. To take the above example:
Q p ( x ) = x 3 − x 2 − 8 x + 12 ∴ p′( x ) = 3 x 2 − 2 x − 8 and p (2 ) = 0
and we have p′(2 ) = 0 , so we know that 2 is a multiple root.
( )(
p ( z ) = z 4 − 4 z 3 + 9 z 2 − 16 z + 20 = z 2 − 4 z + 5 z 2 + 4 )
The roots of z 2 + 4 are 2j and its complex conjugate, − 2 j . Thus
the given polynomial, of degree four, has two pairs of complex
conjugate roots.
( ) ( )( )(
z 5 − 1 = ( z − 1) z 4 + z 3 + z 2 + z + 1 = ( z − 1)( z − α ) z − α 2 z − α 3 z − α 4 )
For convenience, write β = α 2 , and note that β = α 3 while
(
α = α 4 . Our problem is to factorize z 4 + z 3 + z 2 + z + 1 as a )
product of real quadratic factors. We know the roots are
α ,α , β , and β . Now construct the quadratic with roots α and α .
We have: ( z − α )( z − α ) = z 2 − (α + α ) + αα = z 2 − 2ℜ(α ) + 1
Chapter One 39
where ℜ(α ) is the real part of α . Since ( z − β )(z − β ) behaves in
the same way, we have:
( )( )
z 5 − 1 = ( z − 1) z 2 − 2ℜ(α ) + 1 z 2 − 2ℜ(β ) + 1
2π 4π
∴ z 5 − 1 = ( z − 1) z 2 − 2 cos + 1 z 2 − 2 cos + 1
5 5
and this is a product of real linear and quadratic factors.
Problems
١) Express the complex number z = 1 − 3 j exactly in
modulus - argument form. Hence find the modulus
٧) Let p ( z ) = z 5 − 5 z 4 + 8 z 3 − 2 z 2 − 8 z + 8
polynomial p ( z ) = z 3 + 2 z 2 − 6 z + 8
Hence write p as a product of linear factors.
١٢) (
If z1 = 12 cos 125o + j sin 125o and )
Chapter One 41
( )
z 2 = 5 cos 72 o + j sin 72 o Then, find (i) z1 * z 2 and
z1
z2
giving the
١٣) ( )
If z1 = 12 cos 125o + j sin 125o , find z 3 and
1
z3
١٤) If z = x + jy , find the equations of the two loci
π
defined by: (i) z − 4 = 3 and (ii) arg( z + 2 ) =
6
١٥) If z = x + jy , find the value of x and y when :
3z 3z 4
+ =
1− j j 3− j
١٦) Express 2 + j 3 and 1 − j 2 in polar form and
2.1 Introduction
A matrix is, by definition, a rectangular array of numeric or
algebraic quantities, which are subject to mathematical operations.
So a real matrix is an arrangement of real numbers into rows and
columns. Matrices can be defined in terms of their dimensions
(number of rows and columns). Let us take a look at a matrix with 4
rows and 3 columns (we denote it as a 4x3 matrix and call it A):
7 6 1
5 8 1
A =
2 12 0
9 5 0
The dimensions of this matrix are 4 by 3. The dimensions of a
matrix tell you the size of the matrix because they tell you the
number of rows and columns in the matrix. By convention, we list
the number of rows before the number of columns.
Definition 1 The dimensions of a matrix are the number of rows
and columns (listed in that order) of the matrix.
Each element of the matrix is named according to its position.
Typically, capital letters represent matrices and small letters with
subscripts represent elements in the matrix. Since vectors can be
Chapter Two 43
considered to be matrices with only one row or one column, they
could be labeled with capital letters also. However, small letters
usually represents vectors. The element 6 is in the position a12 (read
a one two) because it is in row 1 and column 2. Also by convention,
we list the row number of the element before the column number.
An element in row i and column j would be denoted by aij . This
AT .
Definition 2 By the transpose of the m by n matrix A, denoted by
element.
44 Matrices
Definition 3 We say that two m by n matrices, A and B are equal
if their corresponding elements are equal.
In other words, A = B if A and B have the same dimensions and
Notice that aij = a ji for all i and j; as is true for all symmetric
Example 2
2 2 1 0
1 3
If S = [1 4 3] And R = 2 0 0 Find S * R
4 4
2 1
1 1 2
3
Chapter Two 49
Solution:
2 2 1 0
1 3
S * R = [1 4 3] 2 0 0
4 4
2 1
1 1 2
3
1
Column 1 of S , R = 1 * 2 + 4 * 2 + 3 * 2 = 17
4
Column 2 of S , R = 1 * 2 + 4 * 0 + 3 *1 = 5
3 1
Column 3 of S , R = 1 *1 + 4 * + 3 *1 = 8
4 3
Column 4 of S , R = 1* 0 + 4 * 0 + 3 * 2 = 6
∴ S * R = [17 5 8 6]
2.4 Equations
Solving equations is an important part of mathematics. If we are
working with more than one unknown at a time, we need to solve
systems of equations. You may already know how to solve a system
of linear equations, but matrices provide a more compact way to
arrive at the solution. Matrices are also easier to manipulate on a
computer or calculator. Both of these facts will become more
important when you work with larger systems.
Example 5
Solve the following system of equations:
5 x1 + 3 x2 = 93
− 4 x1 − 2 x2 = −66
Solution: Let's look at a system of linear equations:
5 x1 + 3 x2 = 93
− 4 x1 − 2 x2 = −66
Can be written in matrix form as AX = B where
Chapter Two 51
5 3 x1 93
A= ; X = , and B =
− 4 −2 x2 − 66
When you learned to solve systems of linear equations, you
learned that
(a) You arrive at the same solution no matter which equation you
write first,
(b) The solution doesn't change if you multiply an equation by a
scalar other than zero, and,
(c) You can replace an equation with the sum of that equation and
another equation without changing the solution.
These may not be exactly the words you used when you were
solving a system of linear equations, but you did all these things.
Experiment with the system above to convince yourself that these
statements are true. We can also solve this system entirely in matrix
form. We use the same rules, and we call them Elementary Row
Operations (EROs). The EROs tell us that we can
(a) Interchange any two rows;
(b) Multiply any row by a non-zero scalar; and
(c) Replace any row by the sum of that row and any other row.
Proper use of EROs will leave us with a system that has the same
solution as our original system, but is much easier to solve. If you
were presented the system
x1 = a, x2 = b
52 Matrices
You would be able to solve it instantly because you only have to
read the solution. If this system were written using matrix notation,
it would look like this:
1 0 x1 a 1 0
= The matrix is the 2 by 2 identity
0
1 x2 b 0
1
matrix. Because you can just read of the solution when a system is
in this form, our first goal is to transform our system into this form.
Let's solve the system above using matrices. We can represent
this entire system with a 2 by 3 matrix, which looks like this:
5 3 93
. This is called an augmented matrix because we
− 4 −2 − 66
combined 2 matrices (a matrix and a vector for this system). In this
case, we combined the 2 by 2 coefficient matrix which is made of
the coefficients for our unknowns and the 2 by 1 matrix from the
right-hand side of the equations into one 2 by 3 matrix. In other
words, we put A to the left of the bar and put b to the right of the
bar. The application of an ERO to the augmented matrix does not
change the solution set of the linear system that the augmented
matrix represents because whatever you do to the left side of an
equation, you also do to the right side. Therefore, we will arrive at
the same solution whether we use augmented matrices or not, and
augmented matrices are more compact to write. Using matrix
notation, our goal is to transform our system into one that looks like
the following form:
Chapter Two 53
1 0 a
0 1 b
In other words, we want the identity matrix to the left of the bar and
the solution to the right of the bar.
Remark 1 The bar is not a formal part of the matrix, so it is not
necessary. It is placed there so that we can refer to the different
parts of the augmented matrix and easily move back and forth
between the augmented matrix and the linear system that it
represents. In this book, r1 represents row 1 and so on.
5 393
Original augment matrix
− 4 −2 − 66
1 0.6 18.6
r1 ÷ 5
− 4 −2 − 66
1 0.6 18.6
4 r1 + r2
0 0. 4 8. 4
1 0.6 18.6
r2 ÷ 0.4
0 1 21
1 0 6
−0.6 * r2 + r1
0 1 21
When we convert this from augmented matrix notation back to the
algebraic notation for a system of equations, it looks like this:
54 Matrices
1x1 + 0 x2 = 6
0 x1 + 1x2 = 21
This tells us that x1 = 6 and x2 = 21 . Substitute this solution into
the system to assure yourself that we are correct. If we
systematically use elementary row operations (ERO) to obtain the
identity matrix to the left of the bar, we call this the Gauss Jordan
Elimination Method.
Example 6
Now, let's solve the system using Gauss Jordan elimination.
5 x1 + 3 x2 = 70
− 4 x1 − 2 x2 = −56
5 70
3
Original augmented matrix.
− 4 −2 − 56
1 0.6 14
r1 ÷ 5
− 4 −2 − 56
1 0.6 14
4 * r1 + r2
0 0. 4 0
1 0.6 14
r2 ÷ 4
0 1 0
1 0 14
− 0.6r2 + r1
0 1 0
Let's look at the scalar version of this equation, ax = b ; to help us
1
Example 2 Find the limits of f ( x) = as x approaches 5.
2 x − 10
Solution: It is clear as x approach 5, 2 x − 10 approaches zero the
1 1
approaches which is undefined. Thus;
2 x − 10 0
1
lim = ∞ (undefiend )
x → 5 2 x − 10
Chapter Three 75
This limit lim f ( x) = f ( x) represents a horizontal line, which
x→a
lim f ( x) = lim f ( x) = b
x→a − x→a +
lim f ( x) = b
x→a
3.2 Derivatives
Suppose y = f (x) is shown in Fig.1, the slope of the curve is the
slope of the secant line between point A and another point P on the
graph is shown in the following equation:
m AP =
( f (x + h ) − f (x )) = ( f (x + h ) − f (x ))
(x + h ) − x h
Notice that h can change and with it the location of point P,
therefore h is the limiting factor of the slope of the curve. As h gets
close to point A, the slope of the curve becomes the tangent of the
graph at point A.
So, f ′( x) or
dy
= lim
( f (x + h ) − f (x ))
dx h →0 h
∴
dx
(
d tan ( x )
e )
= e tan ( x ) * sec 2 x
d 3 2
Example 4 Find − x
dx
Solution:
2 1
d 3 2 2 3 −1 2 − 3
− x = − x =− x
dx 3 3
d − 3
Example 5 Find
dx x 2 − 1
Solution:
d − 3 d − 3
1 3
=
dx x − 1 dx
2
2 2
( −
)
− 3 * x − 1 2 = * x − 1 2 * 2x
2
( )
d − 3 3x 3x * x 2 − 1
∴ = =
(
dx x 2 − 1 x 2 − 1 x 2 − 1 )
x2 − 1
2
( )
d
Example 6 Find (5 x + 7 )4
dx
d
Solution: (5 x + 7 )4 = 4(5 x + 7)3 * 5 = 20(5 x + 7 )3
dx
Chapter Three 79
d
Example 7 Find (sin (5 x + 6))
dx
d
Solution: (sin (5 x + 6)) = 5 cos(5 x + 6)
dx
Example 8 Find
d
dx
( ( ))
cos x 2
Solution:
d
dx
( ( )) ( ) ( )
cos x 2 = − sin x 2 * 2 x = −2 x * sin x 2
d
Example 9 Find (ln(3 − 4 cos x ))
dx
d 1 4 sin x
Solution: (ln(3 − 4 cos x )) = * (4 sin x ) =
dx 3 − 4 cos x 3 − 4 cos x
d
Example 10 Find (log10 (2 x − 1))
dx
d 1 2
Solution: (log10 (2 x − 1)) = *2 =
dx (2 x − 1) ln(10) (2 x − 1) ln(10)
Example 11 Find
d 5x
dx
(
e * ln (2 x − 1) )
Solution: Assume y = e5 x * ln (2 x − 1) , u = e5 x and v = ln (2 x − 1)
dy dv du
∴ y = uv , =u +v
dx dx dx
dy 1
∴ = e5 x * 2 + ln (2 x − 1) * 5 e5 x
dx (2 x − 1)
dy 2e 5 x
∴ = + 5 e5 x ln (2 x − 1)
dx (2 x − 1)
80 Calculus
Example 12 Find
d
dx
(
3 x 5 ln(sin x ) )
Solution: Assume y = 3 x 5 ln (sin x ) , u = 3 x 5 and v = ln (sin x )
dy dv du
∴ y = uv , =u +v
dx dx dx
∴
d
dx
( )
3x 5 ln (sin x ) = 3x 5
1
sin x
* cos x + ln (sin x ) * 3 * 5 x 4
∴
d
dx
( )
3x 5 ln (sin x ) = 3 x 5 cot x + 15 x 4 * ln (sin x )
d e 2 x
Example 13 Find
dx ln (3x )
e2x
Solution: Assume y = , u = e 2 x and v = ln(3 x )
ln (3x )
du dv
v −u
dy
= dx 2 dx
dx v
3
2 x ln (3 x ) * 2e
2x
− e2x *
d e 3x
∴ =
dx ln (3 x )
(ln(3x )) 2
1
e 2 x (2 ln ( x ) + 2 ln (3) − )
x e 2 x (2 x ln ( x ) + 2 x ln (3) − 1)
= =
(ln(3x )) 2
x(ln (3 x ))2
d x 2 sinh( 2 x)
Example 14 Find
dx cosh (3 x )
Chapter Three 81
x 2 sinh( 2 x)
Solution: Assume y = , u = x 2 , v = sinh (2 x ) ,
cosh (3 x )
and w = cosh (3x )
d uv uv 1 du 1 dv 1 dw
Where = + − .
dx w w u dx v dx w dx
d x 2 sinh( 2 x) x 2 sinh( 2 x) 1 1
∴ = * 2 * 2x + * 2 cosh (2 x )
dx cosh (3 x ) cosh (3 x ) x sinh (2 x )
1
− * 3 sinh (3 x )
cosh (3 x )
d x 2 sinh(2 x) x 2 sinh(2 x) 2 2 cosh(2 x ) 3 sinh (3x )
∴ = * + −
dx cosh(3 x ) cosh(3 x ) x sinh (2 x ) cosh(3 x )
Example 15 Find
d 5
dx
(
x sin 2 x cos 4 x )
Solution:
d
Example 16 Find (1 + tan 2 x )3
dx
d
Solution: (1 + tan 2 x )3 = 3(1 + tan 2 x )2 * d (1 + tan 2 x )
dx dx
d d
∴ (1 + tan 2 x )3 = 3(1 + tan 2 x )2 * 0 + sec 2 (2 x ) (2 x )
dx dx
d
(
∴ (1 + tan 2 x )3 = 3(1 + tan 2 x )2 * 2 * sec 2 (2 x )
dx
)
d
∴ (1 + tan 2 x )3 = 6 sec 2 (2 x ) * (1 + tan 2 x )2
dx
d 2 x
Example 17 Find 1 − cot
dx 3
Solution:
1
−1
d 2 x 1 2 x 2 d x
1 − cot = 1 − cot * 1 − cot 2
dx 3 2 3 dx 3
Chapter Three 83
x d x
0 − 2 cot * cot
d 2 x 3 dx 3
∴ 1 − cot =
dx 3 x
2 1 − cot 2
3
x x 1
− 2 cot * − csc 2
d x 3 3 3
∴ 1 − cot 2 =
dx 3 x
2 1 − cot 2
3
x x
cot * csc 2
d x 3 3
∴ 1 − cot 2 =
dx 3 x
3 * 1 − cot 2
3
2 y + xy 3 = 6 xy + y 2
This is where implicit differentiation is applied. Implicit
differentiation is taking the derivative of both sides of the equation
with respect to one of the variables. Most commonly, used is the
derivative of y with respect to x. or dy / dx . Since we have not
solved for y as a function of x, the derivative of y must be left as
dy / dx .
Solution:
d
dx
(
3x + y 3 = )
d 2
dx
(
y +4 )
dy dy dy dy
∴ 3 + 3y2 = 2y ∴ 2y − 3y2 =3
dx dx dx dx
dy
∴ y (2 − 3 y ) =3
dx
dy 3
∴ =
dx y (2 − 3 y )
Then the slope of the curve at point (1,3) is
dy 3 3 1
∴ = = =−
dx (1,3) y (2 − 3 y ) (1,3) 3(2 − 3 * 3) 7
Chapter Three 85
3.3 Integration
3.3.1 Introduction
You are now familiar with differentiation principles and have had a
lot of examples about the differentiation in the previous sections.
Now we are going to do the same work with integration.
Integration is the reverse of differentiation. In differentiation we
start with a function and proceed to find its differential coefficient.
In integration, we start with the differential coefficient and have to
work back to find the function from which it has been derived. The
following example clearify the meaning of Integration.
d 5
dx
( )
x + 4 = 5x 4
So, ∫ 5 x 4 dx = x 5 + c
Where c is constant and always called the constant of integration.
This constant is very important to be included in the result of
integration. If you don t put the constant of integration the results is
not genuine.
dy
So, if = f ( x ) , then y is the function whose derivative is f ( x )
dx
and is called the anti-derivative of f ( x ) or the indefinite integral of
f ( x ) , denoted by ∫ f (x )dx . Similarly, if y = ∫ f (u )du , then
86 Calculus
dy
= f (u ) . Since the derivative of a constant is zero, all indefinite
du
integrals differ by an arbitrary constant.
There is some important integration rules are shown in the
Appendix of this book.
b c b
∫a f ( x ) dx = ∫a f ( x ) dx + ∫c f ( x ) dx where a < c < b
Chapter Three 87
3.3.3 Methods of Integrations:
1- Substitution
Sometimes it s not easy to solve some integrals without using
intermediate function and we can integrate with respect to this
function and then we can substitute to get the results in terms of x.
2 u3
above integral we get: ∫ u du =
3
+c
Once the solution has been found in terms of u, substitute back into
terms of x, the final solution is:
∫ 2 x(x + 3) dx =
2 2 (x 2
+ 3 )3
+c
3
Example21 Find the results of the following integral: ∫ x x − 1 dx
∴ ∫ (
x x − 1 dx = ∫ u 2 + 1 . u . 2udu )
( )
= ∫ x x − 1. dx = ∫ u 2 + 1 2u 2 du = ∫ 2u 4 + 2u 2 du ( )
=
2u 5 2u 3
5
+
3
2
+ c = u 3 3u 2 + 5 + c
15
( )
Substitute in the above equation for u = x −1,
88 Calculus
∴ ∫ x x − 1 dx =
2
15
( 3
) (
)
x − 1 3 x − 1 + 5 + c
2
2
∴ ∫ x x − 1 dx = (x − 1)3 / 2 (3(x − 1) + 5) + c
15
2
∴ ∫ x x − 1 dx = (x − 1)3 / 2 (3x + 2) + c
15
2 2 u3
above integral, we get: ∫ sin x cos x dx = ∫ u du =
3
+c
2 sin 3 x
∫ sin x cos x dx =
3
+c
2-
a2 + u2 u = a sinh θ
3-
u2 − a2 u = a cosh θ
Chapter Three 89
Example 23 Find the results of the following integral:
∫ 4 − x 2 dx
x
It is clear that sin θ = then from the following figure, it is clear
2
4 − x2
that cos θ = ,
2
Substitute in the above integral, we get:
2
∫ 4 − x 2 dx = ∫ 4 − 4 sin 2 θ * 2 cosθ dθ x
= ∫ 4 1 − sin 2 θ * cos θ dθ θ
4 − x2
1
But cos 2 θ = (1 + cos 2θ )
2
sin 2θ
∴∫ 4 − x 2 dx = 2θ + +c
2
But sin 2θ = 2 sin θ cos θ
−1 x x 4 − x2
∴ θ = sin , sin θ = , and cos θ =
2 2 2
90 Calculus
4 − x2
2 −1 x x +c
∴ ∫ 4 − x dx = 2 sin + *
2 2 2
2
2 −1 x x 4 − x +c
∴ ∫ 4 − x dx = 2 sin +
2 4
2 x 4 − x2
−1 x
∴ ∫ 4 − x dx = 2 sin +
2 2
+c
∫ x 2 + 4 dx
∫ x 2 + 4 dx = ∫ 4 sinh 2 θ + 4 2 cosh θ dθ
1
But cosh 2 θ = (1 + cosh 2θ )
2
∫ x 2 + 4 dx = 4 ∫
1
(1 + cosh 2θ ) dθ = 2θ + sinh 2θ + c
2 2
But sinh 2θ = 2 sinh θ * cosh θ . Substitute in the above equation,
∫ x 2 + 4 dx = 2 θ + sinh θ 1 + sinh 2 θ + c
2
2 −1 x x x
∴ ∫ x + 4 dx = 2 sinh + 1 + + c
2 2 2
∫ x 2 − 4 dx
∫ x 2 − 4 dx = ∫ 4 cosh 2 θ − 4 2 sinh θ dθ
∴∫ x 2 − 4 dx = 4 ∫ cosh 2 θ − 1 2 sinh θ dθ
∴∫ x 2 − 4 dx = 4 ∫ sinh 2 θ dθ
1
But sinh 2 θ = (cosh 2θ − 1) ,
2
∫ x 2 − 4 dx = 2 ∫ (cosh 2θ − 1) dθ
∴ ∫ x 2 − 4 dx = 2[sinh θ cosh θ − θ ] + c
∴∫ x 2 − 4 dx = 2 cosh 2 θ − 1 * cosh θ − θ + c
x
But cosh θ = ,
2
x2 x
2 −1 x
∴∫ x − 4 dx = 2 − 1 * − cosh +c
4 2 2
x x2 − 4
2 −1 x
∴∫ x − 4 dx = − 2 cosh +c
2 2
3- Integration By Parts
If u and v are functions of x. Then we know that:
d
(uv ) = u dv + v du
dx dx dx
Now integrate both sides with respect to x. On the left, we get back
to the function from which we started
d
∴∫ (uv )dx = ∫ u dv dx + ∫ v du dx
dx dx dx
dv du
∴ uv = ∫ u dx + ∫ v dx
dx dx
And rearranging the terms, we have
dv du
∫ u
dx
dx = uv − ∫ v dx
dx
Chapter Three 93
On the left-hand side, we have a product of two factors to
integrate. One factor is chosen as the function u; the other is thought
of as being the differential coefficient of some function v. To find v,
of course, we must integrate this particular factor separately. Then,
knowing u and v we can substitute in the right-hand side and so
complete the routine.
You will notice that we finish up with another product to integrate
on the end of the line, but, unless we are very unfortunate, this
product will be easier to tackle than the original one.
This is the key to the routine:
dv du
∴ ∫ u
dx
dx = uv − ∫ v dx or
dx ∫ u dv = u v − ∫ v du
Solution :
Assume u = x dv = cos x
du = dx and v = sin x
Solution:
Assume u=x dv = ln x
Then, du = dx and v = x ln x − x
∫ x ln x dx = x 2 ln x − x 2 − ∫ ( x ln x − x )dx
x2
∴ 2 * ∫ x ln x dx = x 2 ln x − x 2 + +c
2
2 x2
∴ 2 * ∫ x ln x dx = x ln x − +c
2
x 2 ln x x 2
∴ ∫ x ln x dx =
2
−
4
+c
Solution:
Assume u = x2 dv = sin x dx
Then, du = 2 xdx and v = − cos x
Solution:
Assume u = ex dv = sin x dx
∴ du = e x dx and v = − cos x
Assume u = e x dv = cos x dx
∴ du = e x dx and v = sin x
I = ∫ e x sin x dx =
1
2
(
− e x sin x + e x sin x + c )
5x e 5 x sin 3 x 5
I = ∫ e cos 3 x dx = − ∫ e 5 x sin 3 x dx
3 3
u = e5 x dv = sin 3 x dx
− cos 3x
∴ du = 5e5 x dx and v=
3
From this formula ∫ u dv = u v − ∫ v du we get:
e 5 x sin 3 x 5 e 5 x cos 3 x 5
I= ∫ e 5 x cos 3 x dx = − * − + ∫ e 5 x cos 3 x dx
3 3 3 3
Chapter Three 97
2
5x e5 x sin 3 x 5 5 x 5
I=∫ e cos 3 x dx = + e cos 3 x − ∫ e5 x cos 3 x dx
3 9 3 1442443
I
5x
25 25 5x e sin 3x 5 5 x
1 + * I = 1 + * ∫ e cos 3 x dx = + e cos 3x
9 9 3 9
34 34 5x e 5 x sin 3 x 5 5 x
∴ * I = * ∫ e cos 3 x dx = + e cos 3 x
9 9 3 9
9 e sin 3 x 5 5 x
5x
∴I =∫ e 5x
cos 3 x dx = + e cos 3 x + c
34 3 9
1
( )
∴ I = ∫ e 5 x cos 3 x dx = 3e 5 x sin 3 x + 5e 5 x cos 3 x + c
34
1
( )
∴ I = ∫ e 5 x cos 3 x dx = 3e 5 x sin 3 x + 5e 5 x cos 3 x + c
34
3 sin 3 x + 5 cos 3 x
∴ I = ∫ e 5 x cos 3 x dx = e 5 x +c
34
dx
Example 31 Find the results of the following: ∫ (x + 1)(x + 3)
Chapter Three 99
Solution : Breaking into partial fractions we get:
1 A B
= +
(x + 1)(x + 3) (x + 1) (x + 3)
Multiply ( x + 1)( x + 3) to both sides of the equation, Then:
1 = A( x + 3) + B( x + 1) = Ax + 3 A + Bx + b
1 = Ax + Bx + 3 A + b
1 = ( A + B )x + 3 A + B
The coefficients on both sides of the equation must be the same that
is that the coefficient of x on the left side of the equation must equal
the coefficient of x on the right side of the equation.
A + B = 0 and 3 A + B = 1
Solving the above equations for A and B,
1 1
∴ A= , and B = −
2 2
So the integral is equal to:
dx 1/ 2 (− 1 / 2)
∴ ∫ (x + 1)(x + 3) = ∫ (x + 1) + (x + 3) dx
dx 1 1 1
∴∫ = ∫ dx − ∫ dx
(x + 1)(x + 3) 2 (x + 1) ( x + 3)
dx 1
∴∫ = (ln( x + 1) − ln( x + 3)) + c
(x + 1)(x + 3) 2
100Calculus
Problems:
1- Find the slope of the tangent line to the following function at
(a) y = 3 x + 4 , at x = 2 (b) y = x + 1 at x = −1
dy
2- Find for the following functions:
dx
2
a) y = (1 + tan 2 x ) 3 (b) y = sin 1 − x ( )
Find the following integrals
π
x3 + 2 x 2 − x 2
1
(a) ∫ 3 x
dx (b) ∫ + sin x dx
0
3 cos x
C=−4 C=4
C=−3 C=3
C=−2 C =2
C=−1 C=1
C=1 C=−1
C =2 C=−2
C=3 C=−3
C=4 C=−4
∫ xdx + ∫ 2 y dy = c1
x2
∴ + y 2 = c1 or x 2 + 2 y 2 = c
2
Where c = 2c1 and c, and c1 are constants.
Example 3 Solve the following differential equation:
xydx + (2 xy 2 + 4 y 2 − x − 2)dy = 0
Solution: The above differential equation can be reduced to be as
x (2 y 2 − 1) 2 1
dx + dy = 0 Or 1 − dx + 2 y − dy = 0
( x + 2) y ( x + 2) y
Integrating both sides we have:
2 1
∫ ( x + 2) ∫
1 − dx + 2 y − dy = C
y
∴ x − Ln( x + 2) 2 + y 2 − Ln y = LnC
( )
∴ x + y 2 = − Ln cy ( x + 2) 2 Or
2
e x + y = cy ( x + 2) 2
Example 4 Solve the following differential equation:
4 xy 2 dx + ( x 2 + 1)dy = 0
x dx − y e − x dy = 0 where, y (0) = 1
Solution: x dx − y e − x dy = 0 can be changed to be in the
following form: x dx = y e − x dy
By integrating both sides we get the following:
∫ ydy = ∫ xe x dx + c
∴
y2
2
( ) (
= − xe x − e x + c ∴ y 2 = 2 e x − xe x + c1)
But y (0 ) = 1 , then by substituting in the above equation we can get
2
the value of c1 , 1 = 2(1 − 0 ) + c1 ∴ c1 = − = −1
2
( ) (
∴ y 2 = 2 e x − xe x − 1 ∴ y = 2 e x − xe x − 1 )
4.3.3 Equation Reducible To Separable Form
In the case of the differential equations not separable but it can be
made separable by a simple change of variable, the following
procedures can be used to solve this kind of differential equations:
Let the differential equation can be has the following form for
example:
Chapter Four 109
y
y′ = g (8)
x
Where g is any given function of y / x .The form of the equation
suggests that we set:
y
=u (9)
x
∴ y = ux , by differentiation
∴ y′ = u + u ′x (10)
By substituting (10) into (8) we have:
u + u ′x = g (u ) (11)
Now we separate the variables u and x we get the following:
du dx
= (12)
g (u ) − u x
If we integrate (12) as we make with normal separable equation then
replace u by y / x , we get the general solution of (8)
xu ′ − u
Assume u = xy Then, y = u / x and y ′ =
x2
Substitute the values of u , y, y ′ in the differential equation (13) we
u xu ′ − u
get: (1 − u ) − x(1 + u ) =0
x x2
110 Ordinary Differential Equations
du
∴ 2u − xu ′(1 + u ) = 0 ∴x (1 + u ) = 2u
dx
∴ 2∫
dx
=∫
(1 + u ) du + c
1
x u
∴ 2 Ln ( x ) = Ln (u ) + u + Ln (c)
x2
∴ Ln =u
uc
∴ x 2 = xyce xy
∴ x = yce xy
y − 4x − 2
∴ = Ce 4 x
y + 4x + 2
Chapter Four 111
Example 10 Solve the following differential equation:
dy du
∴ = x+u
dx dx
( )
du
dx
∴ 1 + u 3 − 3u 2 x + u = 0
du
∴ 1 + u 3 − 3u 3 − 3u 2 x=0
dx
du
∴ 1 − 2u 3 − 3u 2 x=0
dx
du
∴ 1 − 2u 3 = 3u 2 x
dx
3u 2 dx
∴∫ 3
du = ∫ + c1
1 − 2u x
1 2 * 3u 2 dx
∴ ∫
2 1 − 2u 3
du = ∫ x
+ c1
∴
1
2
( )
ln 1 − 2u 3 = ln x + ln c = ln (cx )
( )
∴ ln 1 − 2u 3 = 2 ln (cx ) = ln (cx )2
112 Ordinary Differential Equations
∴ 1 − 2u 3 = (cx )2 ∴ u3 =
1
2
(
1 − kx 2 )
Substitute in the above equation for u = y / x
( )
3
y 1
∴ = 1 − kx 2
x 2
∴ y3 =
1 3
2
(
x 1 − kx 2 )
∴y=x 3 (1 − kx 2 )
2
∂F ∂F
∴ dF = dx + dy
∂x ∂y
∂F ∂F
∴M = and N =
∂x ∂y
By differentiating the first equation with respect to y and the second
one with respect to x, these two equations lead to:
∂M ∂ 2 F ∂N ∂ 2 F
= and =
∂y ∂y∂x ∂x ∂y∂x
∂M ∂N
∴ = (17)
∂y ∂x
Thus for (14) to be exact it is necessary that (17) be satisfied.
Let us now show that, if condition (17) is satisfied, then (14) is an
exact equation.
114 Ordinary Differential Equations
If (14) is exact, the function F ( x, y ) can be found by the
following way:
F = ∫ M ( x, y ) dx + k ( y ) (18)
Solution: ∴ M = 3 x( xy − 2) and N = ( x 3 + 2 y )
∂M ∂N
= = 3x 2 (21)
∂y ∂x
Then (20) is exact. Therefore, its solution is F = c , where,
∂F
= M = 3x( xy − 2) (22)
∂x
∂F
And = N = ( x3 + 2 y) (23)
∂y
Let us attempt to determine F from (22)
( )
∴ F = ∫ 3 x 2 y − 6 x dx + k ( y )
∴ F = x 3 y − 3x 2 + k ( y ) (24)
∂k ( y )
∴ = 2y
∂y
∴ k ( y ) = ∫ 2 y dy ∴ k ( y) = y 2
∴ F = x3 y − 3x 2 + y 2 = c
116 Ordinary Differential Equations
Solving this example by using the other method:
Let us attempt to determine F from (23)
( )
∴ F = ∫ x 3 + 2 y dy + k ( x )
∴ F = x3 y + y 2 + k (x ) (25)
result: ∴ F = x 3 y − y 2 − 3 x 2 = c
It is clear that we get the same solution as we get in the above
solution.
Example12 Solve the following differential equation:
2 2
( y 2 e x y + 4 x 3 )dx + (2 xy e x y − 3 y 2 )dy = 0 (26)
1442443 1442443
M N
∂M ∂N 2 2
Solution: = = 2 y e x y + 2 xy 3e x y
∂y ∂x
Then according to the condition (17), the equation is exact.
∂F 2
∴ = M = y 2 e x y + 4 x3 (27)
∂x
∂F 2
And = N = 2 xy e x y − 3 y 2 (28)
∂y
Chapter Four 117
∴ F = ∫ y 2 e x y + 4 x 3 dx + k ( y )
2
2
∴ F = e x y + x 4 + k ( y) (29)
∂F 2 ∂ k ( y)
∴ = N = 2 xy e x y +
∂y ∂y
Compare the above value with the value of N in (26) we get the
∂ k ( y)
value of .
∂y
∂ k ( y)
∴ = −3 y 2
∂y
∴ k ( y) = − y 3
2
∴ F = e x y + x4 − y3 = C
Example 13 Solve the following differential equation:
dy 4 − 2 x cos y − 2 y 3 sec 2 2 x
= (30)
dx 3 y 2 tan 2 x − x 2 sin y
dy 4 − 2 x cos y − 2 y 3 sec 2 2 x
Solution: =
dx 3 y 2 tan 2 x − x 2 sin y
(
14444244443
) (
144424443
)
∴ 4 − 2 x cos y − 2 y 3 sec 2 2 x dx = 3 y 2 tan 2 x − x 2 sin y dy (31)
M N
∂M
∴ = −2 x sin y + 6 y 2 sec 2 2 x (32)
∂y
∂N
∴ = 6 y 2 sec 2 2 x − 2 x sin y (33)
∂x
118 Ordinary Differential Equations
∂M ∂N
It is clear that, =
∂y ∂x
Then, the differential equation is exact. Integrate (33) with
( )
respect to y. ∴ F = ∫ 3 y 2 tan 2 x − x 2 sin y dy + k ( x)
144424443
N
∴ F = y 3 tan 2 x + x 2 cos y + k ( x)
By differentiating the above equation with respect to x and then
∂ k ( x)
equate it with M as in (31) we can obtain . Then by
∂x
integration with respect to x we can obtain k ( x ) as follows:
∂F
∂x
= 2 y 3 sec 2 2 x + 2 x cos y +
∂k ( x)
∂x
( )
= 4 − 2 x cos y − 2 y 3 sec 2 2 x
14444244443
M
∂ k ( x)
∴ = −4, ∴k ( x) = −4 x
∂x
∴ F = y 3 tan 2 x + x 2 cos y − 4 x = constant
Example 14 Solve the following differential equation
y 1
3 yx 2 − 2 dx + x 3 + cos y + dy = 0 (34)
x x
Solution:
y 1
∴ M = 3 yx 2 − 2 and N = x 3 + cos y + (35)
x x
∂M ∂N 1
∴ = = 3x 2 − 2
∂y ∂x x
Chapter Four 119
Then (34) is exact. Therefore, its solution is F = c where,
∂F y
= M = 3 yx 2 − 2 (36)
∂x x
∂F 1
And = N = x 3 + cos y + (37)
∂y x
Let us attempt to determine F by integrating (36) with respect to x.
y
∴ F = x3 y + + k ( y)
x
∂F 1 ∂k ( y ) 3 1
∴ = x3 + + = x + cos y +
∂y x ∂y 1442443 x
N
∂k ( y )
∴ = cos y
∂y
∴ k ( y ) = ∫ cos y dy
∴ k ( y ) = sin y
y
∴ F = x3 y + + sin y = C
x
∂M ∂N
∴ = = 3e3 x
∂y ∂x
Then (38) is exact. Therefore, its solution is F = c where,
120 Ordinary Differential Equations
∂F
∂x
(
= M = 3e3 x y − 2 x ) (40)
∂F
And = N = e3 x (41)
∂y
Let us attempt to determine F by integrating (40) with respect to x.
( )
∴ F = ∫ 3e3 x y − 2 x dx +k ( y ) = y e3 x − x 2 + k ( y ) (42)
∴ F = y e3 x − x 2 = c
Example 16 Solve the following differential equation:
sinh x cos y dx − cosh x sin y dy = 0 (43)
Solution: ∴ M = sinh x cos y and N = cosh x sin y (44)
∂M ∂N
= = − sinh x sin y (45)
∂y ∂x
Then (43) is exact. Therefore, its solution is F = c where,
∂F
= M = sinh x cos y (46)
∂x
∂F
And = N = − cosh x sin y (47)
∂y
Chapter Four 121
Let us attempt to determine F from (46)
∴ F = cosh x cos y + k ( y ) (48)
∂k ( y )
∴ =0 ∴ k ( y ) = c1
∂y
∴ F = cosh x cos y + c1 = C
[ ]
y ( x ) = e − h ∫ e h r dx + C Where h = ∫ f ( x) dx (52)
x y′ + y − x 4 = 0
Solution: The above differential equation can be written as:
1
y′ + y = x 3 This equation in the form of (52)
x
1 1
∴ f ( x) = , r ( x) = x 3 and h = ∫ dx = Ln( x)
x x
Substitute in (52) we get the following result.
[
∴ y ( x) = e − Ln ( x ) ∫ e Ln ( x ) x 3 dx + C ]
∴ y ( x) =
1
x
[ 4
]
1 x5 x4 C
∫ x dx + C = x 5 + C = 5 + x
Example 19 Solve the following differential equation:
y′ − 2 y = x 2 e5 x
Solution: This equation is linear first order differential equation
from (50). Then we can use the general solution (52) to solve it.
∴ f ( x) = −2 , r ( x) = x 2 e5 x and h = ∫ − 2 dx = −2 x
[
∴ y ( x) = e 2 x ∫ e − 2 x x 2 e 5 x dx + C ]
∴ y ( x) = e 2 x [∫ x 2
e3 x dx + C ]
124 Ordinary Differential Equations
e5 x 2 2 2 2x
∴ y ( x) = x − 3 x + 9 + C e
3
Example 20 Solve the following differential equation:
xy′ = y + ( x + 1)2
Solution: The above equation can be reduced to be as the following
form: y′ −
1
y=
( x + 1)2
x x
This equation is linear first order differential equation. Then we
can use the general solution, (52) to solve it.
1
∴ f ( x) = − , r ( x) =
( x + 1)2 1 1
And h = ∫ − dx = − ln x = ln
x x x x
− ln
1 1
e x
ln
e x
(x + 1)2 dx + C
∫
∴ y ( x) =
x
1 ( x + 1)2
∴ y ( x) = x ∫ dx + C
x x
( x + 1)2
∴ y ( x) = x ∫ dx + C
x2
2 1
∴ y ( x) = x ∫ 1 + + 2 dx + C
x x
1
∴ y ( x) = x x + 2 ln x − + C
x
∴ y ( x) = x 2 + x ln x 2 − 1 + Cx
Chapter Four 125
4.4 Engineering Applications
4.4.1 Newton’s Second Low Of Motion
“The product of the mass and the acceleration equal to the external
force”. In symbols,
F = ma (53)
Where F is the external force, m is the mass of the body, and a is
its acceleration in the direction of F. Equation (53) can be put in the
following from:
dv
F =m (54)
dt
Where v is the velocity of the body.
method we get: T (t ) = 20 + Ce kt
Substitute the initial conditions [T(0)=100, T(20)=70] in the
1 y
Q Ln = −0.02877t − 0.0693
10 y + 10
1 2
Q Ln = −0.02877t − 0.0693
10 2 + 10
∴ t = 3.82 hours
r=10 m
− π * r 2 dh = πro2 * 0.6 2 gh dt
The above differential equation can be solved by separation of
dh ro2
variables. ∴ = −0.6 2 gh 2 dt
h r
By integrating both sides of the above equation we get the
dh r2
following: ∫ = ∫ − 0.6 2 gh o2 dt + c
h r
Where c is the constant of integration,
ro2
∴2 h = −0.6 2 g 2
t+c (60)
r
130 Ordinary Differential Equations
So, at time t = 0 h = H o . Substitute this condition in the above
equation we get the following values for c. 2 H o = 0 + c
∴ c = 2 H o Substitute this value in (60) we get the final
ro2
results: 2 h = −0.6 2 g t + 2 Ho
2
r
To get the time required for tank to get empty we can substitute in
the above equation for h = 0 ,
2 Ho
∴ te = (61)
0.6 2 g ro2 / r 2
So, we can substitute the data in our example in the above equation
to obtain the time for the tank to get empty, t e .
2 25
te = 2 2
= 1505 sec . = 25.08 min .
0.6 2 * 9.81 * 0.5 / 10
dV = −π r 2 dh (62)
Now by Torricelli’s law the velocity with which a liquid issues
from an orifice is: v = 0.6 2 gh where g is the acceleration of
∴ r 2 + (h − R )2 = R 2 or r 2 = 2hR − h 2 (65)
With this, the differential equation (65) can be written as:
( )
∴π 2hR − h 2 dh = −π ro2 * 0.6 2 gh dt (66)
132 Ordinary Differential Equations
This is a simple separable equation.
( )
∴ 2 Rh1 / 2 − h 3 / 2 dh = −ro2 * 0.6 2 g dt
4 2
∴ Rh3 / 2 − h5 / 2 = −ro2 * 0.6 2 g t + c (67)
3 5
14
Since h = R at t = 0 , ∴ R 5 / 2 = c
15
4 2 14
∴ Rh3 / 2 − h 5 / 2 = − ro2 * 0.6 2 g t + R 5 / 2
3 5 15
To fiend how long it will take the tank to empty, we must determine
14
the value of t when h=0. ∴ 0 = −ro2 * 0.6 2 g t + R 5 / 2
15
5/ 2
14 R
∴t = (68)
15 ro2 * 0.6 2 g
So in our example R = 1m and ro = 0.05m also g = 9.81 m / sec 2 .
Substitute these values in the above equation we get:
14 15 / 2
∴ t= = 140.5 sec .
15 0.05 2 * 0.6 2 * 9.81
gravity.
Chapter Four 133
Solution: Let the origin be chosen at the lowest point of the tank
and let h be the instantaneous depth, V the instantaneous volume,
and r the instantaneous radius of the free surface of the water as
shown in Fig.4.
Let us start first with finding the time required for the cylindrical
portion to get empty. Then in an infinitesimal time dt, the water
level will fall by the amount dh, and the resultant decrease in
volume of the water in the tank
will be:
dV = −π 10 2 dh
Fig.4 20m
Now by torricelli’s law the
velocity with which a liquid
issues from an orifice is:
v = 0.6 2 gh where g is the 10m
acceleration of gravity and h is the
instantaneous height, or head, of the
liquid above the orifice. In the interval dt.
Area = 50 cm 2
The unit volume escape from the orifice at v = 0.6 2 gh
time dt is:
dV = 0.6 2 gh * 50 * 10 − 4 dt
So, by equating both volumes we get the following equation:
dh
0.6 2 gh * 50 * 10 − 4 dt = −π 10 2 dh , ∴ = −4.23 * 10 − 5 dt
h
134 Ordinary Differential Equations
It is clear that the above equation is first order differential
equation in separable form, so it is easy to solve this equation by
integrating both sides.
dh
∫ = ∫ − 4.23 *10 −5 dt + c ∴ 2 h = −4.23 * 10 −5 t + c
h
It is clear from the initial condition that at t = 0 , h = 20 + 10 = 30m
Substitute this initial condition in the above equation we get:
2 30 = −4.23 * 10 − 5 * 0 + c ∴ c = 2 30 = 10.954
So, the solution of differential equation becomes,
2 h = −4.23 * 10 − 5 t + 10.954
At the bottom of the cylinder portion h = 10m so we can obtain
the time required for the cylindrical portion to get empty if we
substitute in the solution of differential equation for h = 10m . Then,
2 10 = −4.23 * 10 − 5 t + 10.954
∴ tcylinder = 30.4 hours
In the same way we can obtain the time required for spherical
portion to get empty. We use also unit element in spherical portion
(
∴ dV = −π 20h − h 2 dh )
This volume will take time dt to escape from the orifice of the tank.
∴ dV = 0.6 2 gh * 50 * 10 − 4 dt
Chapter Four 135
( )
∴ −π 20h − h 2 dh = 0.6 2 gh * 50 * 10 − 4 dt
20 1.5 h 2.5
∴ h − = −0.0042298 t + k
1.5 2.5
At t = 30.4 * 3600 sec, h = 10m
20 1.5 10 2.5
∴ 10 − = −0.0042298 * 30.4 * 3600 + k
1.5 2.5
Then, k = 758.055
20 1.5 10 2.5
∴ 10 − = −0.0042298 * 30.4 * 3600 + 758.055
1.5 2.5
The total time required for the total tank to get empty can be
obtained by applying h = 0 in the above equation. Then,
∴ t = 179218 sec = 49.7827hours
Another Solution
For cylindrical portion we can use (60) to get the time required for
this portion to get empty
ro2
∴2 h = −0.6 2 g 2
t+c (60)
r
So, at time t = 0 h = 30m . Substitute this condition in the above
equation we get the following values for c.
∴2 30 = 0 + c ∴c = 2 30
ro2
∴2 h = −0.6 2 g t + 2 30
r2
136 Ordinary Differential Equations
To get the time required for cylindrical portion of the tank to get
empty we can substitute in the above equation for h = 10 m . Then,
ro2
∴2 10 = −0.6 2 g t + 2 30
10 2
Q 50 *10 − 4 = πro2 ∴ ro = 3.99cm
.0399 2
∴2 10 = −0.6 2 * 9.81 2
t + 2 30
10
tcylinder = 109427 sec . = 30.4 hours
In the same way we can obtain the time required for spherical
portion to get empty. We use (67) as explained before.
4 3/ 2 2 5/ 2
∴ Rh − h = −ro2 * 0.6 2 g t + c
3 5
Since h = R = 10 at t = 109427 sec . ,
4 2
∴ 10 * 103 / 2 − 105 / 2 = −0.03992 * 0.6 2 * 9.81 * 109427 + c
3 5
∴ c = 758.135
4 3/ 2 2 5/ 2
∴ Rh − h = −ro2 * 0.6 2 g t + 758.135
3 5
Then the time required for the hole tank to get empty is at h = 0 .
Then substitute h = 0 in the above equation to get tTotal
r
dh
h-R R
Fig.6
138 Ordinary Differential Equations
dV = −2 * 2 Rh − h 2 * L dh (69)
Also, this volume dV can be calculated at the orifice of the tank as
∴ − 2 * 2 Rh − h 2 * L dh = πro2 * 0.6 2 gh dt
−π 2
∴ 2 R − h dh = ro * 2 g dt
2L
By integrating both sides of the above equation we can get the
following equation:
∴
2
(2 R − h )3 / 2 dh = − π ro2 * 2 g t + c (71)
3 2L
Where c is the constant of integration.
At time t = 0, h = 2 R Substitute this condition in the above
equation we get the value of the constant c,
∴c =0 (72)
Substitute from (72) into (71).
∴
2
(2 R − h )3 / 2 = − π ro2 * 2 g t (73)
3 2L
To obtain the time required for the upper half to get empty we have
to put h = R in the above equation, (73).
4 LR 3 / 2
∴ tu = (74)
3π ro2 2 g
Chapter Four 139
For the lower half
The horizontal incremental volume dV can be taken in the upper
half of the tank. It is clear from geometry of the tank, the width of
the incremental volume, dV is 2r and the length L of the cylinder.
dV = −2 * 2 Rh − h 2 * L dh (75)
Also, this volume dV can be calculated at the orifice of the tank as
− 2 * 2 Rh − h 2 * L dh = πro2 * 0.6 2 gh dt
−π 2
∴ 2 R − h dh = ro * 2 g dt
2L
L
dh
R-h R
h r
Fig.7
following equation: ∴
2
(2 R − h )3 / 2 = − π ro2 * 2 g t + c (77)
3 2L
140 Ordinary Differential Equations
Where c is the constant of integration. At time
4 LR 3 / 2
t = tu = , h = R substitute this condition in the above
3π ro2 2 g
equation we get the value of the constant c,
2 3/ 2 − π 2 4 LR 3 / 2
∴ (2 R − R ) = ro * 2 g +c (78)
3 2L 3π ro2 2 g
∴ c = 0 , substitute for c = 0 in (78) we get the following results
∴
2
(2 R − h )3 / 2 = − π ro2 * 2 g t
3 2L
The total time required for the whole tank to get empty can be
obtained by Substituting in the above equation for h = 0 .
4 L(2 R )3 / 2
∴ ttotal = (79)
3π ro2 2g
So the time required for the lower portion only to get empty is given
by the following equation: t L = ttotal − tu (80)
4 * 16 * 83 / 2
tu = = 3468.92 sec . = 57.82 min .,
3 * π * 0.12 * 2 * 9.81
4 * 16 * (2 * 8)3 / 2
ttotal = = 9811.59 sec . = 163.52 min ., and,
3 * π * 0.12 * 2 * 9.81
∴t L = ttotal − tu = 163.52 − 57.82 = 105.7 min .
Chapter Four 141
Example 28 A conical tank shown in Fig.8, filled with water and
10 m
has outlet of radius ro = 20 cm
at the bottom. At time t=0 the Fig.8
following: dV = −π r 2 dh (81)
r R R
But = ∴r = h (82)
h L L
Substitute the value of r in (82) into (81) we get the following result:
2
R
dV = −π * h dh (83)
L
142 Ordinary Differential Equations
In the same way dV can be obtained at the bottom of the tank as
following: dV = πro2 * 0.6 2 gh dt (84)
By equating (83) and (84) we get the following result:
2
R
− π * h dh = πro2 * 0.6 2 gh dt
L
Lro2
∴ h dh = −0.6 2 g
3/ 2
dt
R
By integrating both sides of the above equation we get the
2 Lr 2
following: ∴ h 5 / 2 = 0.6 2 g o t + c (85)
5 R
where c is the constant of integration
At time t = 0 , h = L . Substitute this initial condition into (85):
2 2 5/ 2
∴ L5 / 2 = 0 + c ∴c = L (86)
5 5
Substitute the value of c in (86) into (85) we get the following:
2 5/ 2 Lro2 2 5 / 2
∴ h = 0.6 2 g t + L
5
R 5
So the time required for the tank to get empty is given by
applying for h = 0 in the above equation. So,
2
2 L5 / 2 R
∴t = * (87)
5 * 0. 6 2 g Lr
o
Substitute the data obtained in (87) where L = 20m , R = 5m R,
ro = 0.2m
2
2 * 205 / 2 5
∴t = * = 420.68 sec . = 7.01 min .
5 * 0.6 2 * 9.81 20 * 0.2
Chapter Four 143
4.4.5 Half Life Of Nuclear Materials
Example 29 Experiment shows that radium disintegrates at a rate
proportional to the amount of radium instantly present. Its half life
(that is the time in which 50% of a given amount will disappear),
is 1590 years. What percent will disappear in one year?
Solution: Assume the instantaneous amount and starting amount of
radium exists are y and yo respectively.
dy
∴ = ky (88)
dt
Where k is the proportional constant. Equation (88) can be easily
∴ y = y0 e kt (90)
The second initial condition is y = y0 / 2 at t = 1590 years .
Substitute this condition into (90)
y0
∴ = y0 e k *1590 ∴ k = −4.36 *10 − 4
2
−4
∴ y = y0 e − 4.36 *10 t (91)
To obtain the percent disappear after one year we can do the
following:
_
+
− 3t
(a) E (t ) = 40 V , (b) E (t ) = 20 e E(t)
_ I(t)
and (c) E (t ) = 50 sin 5 t .
+
Solution: From Kirchoff’s laws vL(t)
dI Fig.9
RI + L = E (t )
dt
dI R E (t )
∴ + I=
dt L L
10 40
(a) I& + I=
2 2
∴ I& + 5 I = 20
By using the general solution of the first order linear differential
equations, (52) we can solve it.
∴ f (t ) = 5 r (t ) = 20 ∴ h = ∫ 5dt = 5t
∴ I (t ) = e − 5t [∫ e 5t
]
* 20dt + k ∴ I (t ) = 4 + ke − 5t
where k is a constant of integration.
But I (0) = 0 ∴ k = −4 ∴ I (t ) = 4 * 1 − e − 5t ( )
(b) For E (t ) = 20 e − 3t
I (t ) = e − 5t [∫ e 5t
*10 e − 3t dt + k ]
Chapter Four 145
[ ]
I (t ) = e − 5t 5 e 2 t + k ∴ I (t ) = 5 e − 3 t + ke − 5t
I (t ) = e − 5t [∫ e 5t
* 25 sin 5t dt + k ]
5
I (t ) = e − 5t e5t (sin 5t − cos 5t ) + k But I (0) = 0 ∴ k = 5 / 2
2
∴ I (t ) =
5
2
[(sin 5t − cos 5t ) + e −5t ]
Example 31
Find the current in the RC circuit in
Fig.10. Assume I (0) = 0 R=5 Ω, vR(t)
C=.02F and E (t ) = 50 sin 20 t .
+
_
+ +
Assume (a) vc (0) = 0 and E(t)
_ I(t) _ vC(t)
(b) vc (0) = 100
Solution:
From Kerchief’s laws Fig.10
1
C∫
RI + Idt = E (t )
dI 1 1 dE (t )
∴ + I=
dt RC R dt
1 1d
∴ I& + I= 50 sin 20 t
RC R dt
146 Ordinary Differential Equations
1 1
∴ I& + I = * 50 * 20 cos 20 t ∴ I& + 10 I = 200 cos 20t
5 * .02 5
By using the general solution of the first order linear differential
equations we can solve it.
f (t ) = 10 r (t ) = 200 cos 20t ∴ h = ∫ 10 dt = 10t
1
C∫
(a) For vc (0) = Idt =0
t =0
k −10t
∴ 0.2 sin( 20 t ) − .4 cos (20t ) − e =0 ∴ k = −4
10 t =0
1 k −10t
∴ 0 . 2 sin( 20 t ) − .4 cos ( 20t ) − e = 100
0.02 10 t =0
∴ k = −24
5) y′ = x + 1, y ( 0) = 1 6) y ′ − y = 1 − e − x , y ( 0) = 0
7) y ′ + y 2 = 0, y (5) = 0.25 8) 9 yy ′ + 4 x = 0, y (3) = −4
[III] Find the general Solution of the following differential
equations:
dy x 2 dy x2
9) = 10) =
dx y dx y (1 + x 3 )
11)
dy
dx
+ y 2 sin x = 0 12)
dy
dx
( )(
= cos 2 x cos 2 2 y )
dy 1 − y2 dy x − e − x
13) = 14) =
dx x dx y + e y
dy
15) = y tanh x 16) ( xLn( x) )dy − ydx = 0
dx
21) y ′ = 3 x 2 e − y , y (−1) = 0
2
22) yy ′ = xe y , y (1) = 0
23) xyy ′ = y + 2, y (2) = 0
( )
25) x 3 + y 3 dx − 3 xy 2 dy = 0 26) xdy − y + x 2 − y 2 dx = 0
Equation reducible to separable form
٢٧) (2 x + 3 y )dx + ( y − x)dy = 0
٢٨) ( x + y )dx + (3x + 3 y − 4)dy = 0
٢٩) (1 − xy − x 2 y 2 )dx + (x3 y − x 2 )dy = 0
٣٠) tan 2 ( x + y )dx − dy = 0
٣١) (2 + 2 x 2 y1 / 2 )ydx + (x 2 y1 / 2 + 2)xdy = 0
٣٢) dx + ((1 − x 2 )cot y )dy = 0
٣٣) (x 2 + y 2 )dx + (x 2 − xy )dy = 0
٣٤) 2 x 3 ydx + (x 4 + y 4 )dy = 0
٣٥) ( xy ′ − y )cos (2 y / x) = −3x 4
Chapter Four 149
Exact differential equations
[VI] Solve the following differential equations by prove that it
is Exact, and then solve it.
45) y ′ = ( y − 1) cot x ( )
46) 1 − x 2 y ′ + xy = x
[VIII] Solve the following initial value problem
47) y ′ + y = ( x + 1) 2 , y (0) = 0
48) y ′ − 2 y = 2 cosh 2 x + 4, y (0) = −125
49) xy′ − 3 y = x 4 (e x + cos x) − 2 x 2 , y (π ) = eπ + 2 / π
2
50) 2( y + 1) y ′ − ( y + 1) 2 = x 4 , y (1) = 2 / 3 − 1
x
150 Ordinary Differential Equations
[VIV] Applications
51) After two days, 10 grams of a radioactive chemical is present.
Three days later 5 grams is present. How much of the chemical was
present initially assuming the rate of disintegration is proportional to
the instantaneous amount which is present.
+ +
5.1 Definition
The general linear differential equations of order n has the form
shown in (1) or (2).
that: C1 2e 3 x + C 2 5e 3 x + C3e 3 x = 0
Identically, for instance, C3 = 0,.C2 = 2, and C1 = 5
λ2 − 3λ + 2 = 0 ∴ λ1 = 1 and λ2 = 2
Then, λ1 and λ2 are real distinct roots. So, the general solution is
given by (5). y ( x) = C1e x + C 2 e 2 x
154 Linear Differential Equations Of Higher Order
Example 4 Solve the following differential equation:-
y ′′′ − 5 y ′′ + 8 y ′ − 4 y = 0
Solution:- The auxiliary equation is:∴ λ3 − 5λ2 + 8λ − 4 = 0
∴ y = C1e x + (C2 x + C3 ) e 2 x
solution of (1).
Chapter Five 155
The particular solution can be assumed depending on the form of
r (x) . The trial solution to be assumed in each form of r (x) is
shown in the following table. The trial solutions in this table hold in
case no terms in the assumed trial solution appear in the
complementary solution. If any term of the assumed trial solution
does appear in the complementary solution, we multiply this trial
solution by the small positive integer power of x which is large
enough so that none of the terms which are then present appear in
the complementary solution.
ce px Ae px
c cos px + k sin px A cos px + B sin px
cn x n + cn −1 x n −1 + ......c0 An x n + An −1 x n −1 + ......A0
(
e px cn x n + cn −1 x n −1 + ......c0 ) (
e px An x n + An −1 x n −1 + ......A0 )
(
cos px * cn x n + cn −1 x n −1 + ......c0 + ) ( )
cos px * An x n + An −1 x n −1 + ...... A0 +
sin px * (k n x n + k n −1 x n −1 + ......k 0 ) sin px * (Bn x n + Bn −1 x n −1 + ......B0 )
( )
e qx cos px * c n x n + c n −1 x n −1 + ......c0 + ( )
e qx cos px * An x n + An −1 x n −1 + ...... A0 +
e sin px * (k n x + k n −1 x
qx n n −1
+ ......k 0 ) e qx sin px * (Bn x n + Bn −1 x n −1 + ......B0 )
y ′′ + y ′ − 2 y = 2 e 3 x
∴ y h ( x) = C1e x + C 2 e − 2 x
The particular solution takes the following form:
y p ( x) = ke3 x
∴ y ′p ( x) = 3ke3 x
∴ y ′p′ ( x) = 9ke 3 x
∴ e 3 x (9k + 3k − 2k ) = 2e 3 x
∴k = 0.2 , ∴ y p ( x) = 0.2 e 3 x
∴ y ( x) = y h ( x) + y p ( x) = C1e x + C 2 e − 2 x + 0.2 e 3 x
y ′′ + y ′ − 2 y = 2 e x + e − 2 x
y p ( x) = k1 xe x + k 2 xe − 2 x
∴ y ′p ( x) = k1 xe x + k1e x + k 2 e − 2 x − 2k 2 xe − 2 x
∴ y ′p′ ( x) = 2k1e x + k1 xe x − 2k 2 e − 2 x + 4k 2 xe − 2 x − 2k 2 e − 2 x
2 1
∴ k1 = and k 2 = −
3 3
2 1
∴ y p ( x) = xe x − xe − 2 x
3 3
2 x 1 −2x
∴ y ( x) = y h ( x) + y p ( x) = C1e x + C 2 e − 2 x + xe − xe ]
3 3
the above equation and substitute the results into the differential
equation we get that the final result for the particular solution is as
the following:
y p = − x sin( 2 x)
∴ y ( x) = y h + y p = A cos(2 x) + B sin(2 x) − x sin(2 x)
y ′′ − 4 y ′ + 4 y = e 2 x
Solution:- The auxiliary equation is λ2 − 4λ + 4 = 0
∴ λ1 = λ2 = 2
Because of λ1 = λ2 , so the homogeneous solution takes the form in
∴ y ′p ( x) = 2kx 2 e 2 x + 2kxe 2 x
∴ y p ( x ) = 0. 5 x 2 e 2 x
∴ y ( x) = y h ( x) + y p ( x) = (C1 + C 2 x )e 2 x + 0.5 x 2 e 2 x
y ′′ − 5 y ′ + 6 y = x 2 e 5 x
Solution:- The auxiliary equation is λ2 − 5λ + 6 = 0
Then, λ1 = 2, and λ2 = 3
∴ y h ( x) = C1e 2 x + C2 e3 x
( )
y p ( x) = k 2 x 2 + k1 x + k 0 e 5 x
( )
∴ y ′p ( x) = 5 k 2 x 2 + k1 x + k 0 e 5 x + (2k 2 x + k1 ) e 5 x
( )
∴ y ′p′ ( x) = 25 k 2 x 2 + k1 x + k 0 e 5 x + 5(2k 2 x + k1 ) e 5 x
+5(2k 2 x + k1 ) e 5 x + (2k 2 ) e 5 x
160 Linear Differential Equations Of Higher Order
By substituting the above values into the differential equation and
by comparing both sides of the resultant equation we get the values
of the constants k 2 , k1 , and k 0 .
1 5 19
∴ k 2 = , k1 = − , and k 0 =
6 18 108
1 5 19 5 x
∴ y p ( x) = x 2 − x + e
6 18 108
1 5 19 5 x
∴ y( x) = yh ( x) + y p ( x) = C1e 2 x + C2e3x + x 2 − x + e
6 18 108
Example 11 Solve the following differential equation:-
y ′′ − 5 y ′ + 6 y = x 2 e 2 x
Solution:- The auxiliary equation is λ2 − 5λ + 6 = 0
∴ λ1 = 2, λ2 = 3
∴ y h ( x) = C1e 2 x + C2 e3 x
The particular solution corresponding to r ( x) = x 2 e 2 x is
( )
y p ( x) = k 2 x 2 + k1 x + k 0 e 2 x . But there is a term in the
( )
y p ( x) = k 2 x 2 + k1 x + k 0 e 2 x .
So we have to multiply it by x to be as following:
( )
y p ( x) = k 2 x 2 + k1 x + k 0 x e 2 x
( ) ( )
∴ y ′p ( x) = 2 k 2 x 3 + k1 x 2 + k 0 x e 2 x + 3k 2 x 2 + k1 x + k 0 e 2 x
y ′′′ − 6 y ′′ + 12 y ′ − 8 y = x 2 + e 2 x
Solution:- The auxiliary equation is: λ3 − 6λ2 + 12λ − 8 = 0
∴ λ1 = λ2 = λ3 = 2 ( )
∴ y h ( x) = C1 + C2 x + C3 x 2 e 2 x
The particular solution corresponding to r ( x) = x 2 + e 2 x is
( )
y p ( x) = k 2 x 2 + k1 x + k 0 + k3e 2 x . But there is a term in
( )
y p ( x) = k 2 x 2 + k1 x + k 0 + k3 xe 2 x . But, still there is another term
( )
y p ( x) = k 2 x 2 + k1 x + k 0 + k3 xe 2 x . Then we have to multiply
again till we become sure that there are no any dependant terms
between homogeneous and particular solutions. Then the particular
solution takes the following form:
162 Linear Differential Equations Of Higher Order
( )
y p ( x) = k 2 x 2 + k1 x + k 0 + k3 x 3e 2 x
3 y ′′ − 6 y ′ + 36 y = e x sin (3 x )
[
∴ y h =e x A cos 11x + B sin 11x ]
By inspecting the right hand side of the differential equation we can
Coefficient of e x cos 3 x = 0
∴ 3 * −9k1 + 9k 2 + 9k 2 + 9k 2 + 3k1 − 18k 2 − 6k1 + 36k1 = 0
∴ k1 = 0
Coefficient of e x sin 3 x = 1
∴ −27 k 2 − 9k1 − 9k1 + 3k 2 + 18k1 − 6k 2 + 36k 2 = 1
1 e x sin 3x
∴ k2 = ∴ yp =
6 6
e x sin 3 x
[
∴ y ( x ) = yh + y p =e A cos 11x + B sin 11x +
x
6
]
Example 14 Solve the following differential equation:-
y ′′ + y = − 2 sin x + 4 x cos x
[− k x3
3
+ (6k5 − k 4 ) x 2 + (4k 6 + 6k3 − k1 )x + 2(k 2 + k 4 ) cos x ]
[− k x 5
3
− (k 6 + 6k3 )x 2 + (6k5 − 4k 4 )x + (2k 6 − 2k1 ) sin x]
( ) (
+ k3 x 3 + k 4 x 2 + k1 x cos x + k5 x 3 + k 6 x 2 + k 2 x sin x )
= −2 sin x + 4 x cos x
Chapter Five 165
Coefficient of cos x = 0
∴ 0 = 2(k 2 + k 4 ), ∴ k 2 = −k 4 (8)
Coefficient of x 2 cos x = 0 ∴ 6k 5 − k 4 + k 4 = 0
∴k 5 = 0 (10)
By substituting (20), (22) and (24) into (18) and collecting terms
containing u and terms containing v we obtain the following:
u ( y1′′ + f y1′ + g y1 ) + v ( y 2′′ + f y 2′ + g y 2 ) + u ′y1′ + v′y 2′ = r (25)
e2x
y ′′ − 4 y ′ + 4 y =
x
Solution:- The auxiliary equation is ∴ λ2 − 4λ + 4 = 0
∴ λ1 = λ2 = 2, ∴ y h = (c1+ c2 x ) e 2 x
170 Linear Differential Equations Of Higher Order
The particular solution can be obtained from (29)
y2 r yr
y p ( x) = − y1 ∫ dx + y 2 ∫ 1 dx
W W
e2x xe 2 x 4x
∴W = 2 x = e
2e (2 x + 1)e 2 x
xe 2 x e 2 x / x e2xe2x / x
∴ y p ( x ) = −e 2x
∫ dx + xe 2x
∫ dx
e4x e4x
∴ y p ( x ) = −e 2 x x + x e 2 x ln ( x )
= e 2 x (ln x − x )
y ( x ) = (c1+ c2 x ) e 2 x + e 2 x (ln x − x )
∴ y ( x ) = (c1+ c2 x + ln x − x ) e 2 x
x cos x sin x
∴ yp = +
8 32
x sin x sin x
∴ y ( x ) = A cos 3x + B sin 3 x + + (36)
8 32
Another solution for obtaining y p by using Wronskian determinant.
(
y p = k 2 x 2 + k1 x + k o xe 2 x)
(
∴ y p = k 2 x 3 + k1 x 2 + k o x e 2 x )
( ) ( )
∴ y ′p = 2 k 2 x 3 + k1 x 2 + k o x e 2 x + 3k 2 x 2 + k1 x + k o e 2 x
Coefficient of e 2 x = 0 ∴ 2k o + 2k o + 2k1 − 5k o = 0
Then, 2k1 − k o = 0 (38)
Coefficient of xe 2 x = 0
∴ 4k o + 4k1 + 4k1 + 6k 2 − 10k o − 10k1 + 6k o = 0
∴ 6k 2 − 2k1 = 0 (39)
Coefficient of x 2 e 2 x = 1
∴ 4k1 + 6k 2 + 6k 2 − 10k1 − 15k 2 + 6k1 = 1
1
∴ − 3k 2 = 1 , ∴ k 2 = − (40)
3
Substitute from (40) into (39) we get:
Chapter Five 175
6
∴ k1 = − , ∴ k1 = −1 (41)
(3 * 2)
Substitute from (41) into (38) we get: ko = −2
x2
∴ y p = − + x + 2 x e2x
3
x2
∴ y ( x ) = c1e + c2 e −
2x
+ x + 2 x e2x
3x
(42)
3
Another solution to obtain y p by using the Wronskian theory
e2x e3x
y1 = e 2x
, and y 2 = e 3x
∴W = 2x 3x
= e5 x
2e 3e
2x e3 x x 2e 2 x 3x e 2 x x 2e 2 x
∴ y p = −e ∫ e 5x
dx + e ∫ e 5x
dx
( )
3
x
∴ y p = −e 2 x * − e3 x x 2 + 2 x + 2 e − x
3
2 x x
3
∴ y p = −e + x2 + 2x + 2
3
x3
∴ y ( x ) = c1e + c2 e − + x 2 + 2 x + 2 e 2 x
2x 3x
3
3 x x
3
∴ y ( x ) = (c1 − 2 )e + c2 e −
2x
+ x2 + 2x e2x
3
3 x x
3
∴ y ( x ) = c1 e + c2 e −
* 2x
+ x2 + 2x e2x (43)
3
Where c1* = c1 − 2
As we see the general solution (43) is the same as we get before
in (42).
176 Linear Differential Equations Of Higher Order
5.5 Cauchy Equations
The following equations are so called second order and third order
Cauchy differential equations.
m 2 + (a − 1)m + b = 0 (46)
We have two roots of (46). So, we have three different cases,
Case 1 If m1 and m2 are two different roots of (46), then the two
( )
∴ y h ( x) = xα C1Cos ( Ln( x β )) + C 2 sin ( Ln ( x β )) (50)
The same way can be used with the third order Cauchy differential
equation (45). So the auxiliary equation will be as follows:-
( )
∴ y h ( x) = xα C1Cos ( Ln( x β )) + C 2 sin ( Ln ( x β )) + C3 x m3 (54)
x 2 y ′′ + 3 x y ′ + y = 0
Solution:- From (46) the auxiliary equation is
m 2 + 2m + 1 = 0 Then m1 = m2 = −1
∴ y ( x) = (C1 + C 2 Ln x )x −1
( x − 2 ) 2 y ′′ + 5 ( x − 2 ) y ′ + 3 y = 0
Solution:- Put t = x − 2, ∴ dt = dx
∴ t 2 y ′′ + 5t y ′ + 3 y = 0
From (46) the auxiliary equation is
m 2 + 4m + 3 = 0 ∴ m1 = −1, m2 = −3 ,
∴ y ( x) = C1t −1 + C2t − 3 But, t = x − 2
C C2
∴ y ( x) = 1 +
x − 2 ( x − 2 )3
178 Linear Differential Equations Of Higher Order
Example 22 Solve the following differential equation:-
x 2 y ′′ + 9 x y ′ + 25 y = 125
Solutions:- From (46) the auxiliary equation is
m 2 + 8m + 25 = 0 ∴ m1 , m2 = −4 ± j 3
(
∴ y h ( x) = x 4 C1Cos ( Ln( x 3 )) + C2 sin ( Ln ( x 3 )) )
Assume y p = C3 .Substitute in the equation, we get
C3 = 5 ∴ y p = 5 . Then the total solution is
( )
∴ y ( x) = y h ( x) + y p ( x) = x 4 C1Cos ( Ln( x 3 )) + C 2 sin ( Ln ( x 3 )) + 5
m 3 − 12m 2 + 45m − 50 = 0 ∴ m1 = 2, m2 = m3 = 5
∴ y h ( x) = C1 x 2 + (C2 + C3 Ln ( x) )x 5
5.6 Applications
5.6.1 Free Oscillation
Fig.1
Example 24 A weight W suspended from the
end of a vertical spring stretches it y
meters. Let A and B represents the y
m
position of the end of the spring
before and after the weight W is put on. B is called
the equilibrium position. Call y the displacement of W at any
position C from the equilibrium position. Assume that y is positive
in the down ward direction. If we pull the body down a certain
Chapter Five 179
distance and then release it, it undergoes a motion. We want to
determine the motion of this mechanical system. For this purpose
we consider the forces acting on the body during the motion. This
will lead to a differential equation, and by solving this differential
equation we shall then obtain the displacement as a function of time.
The first force acting on the weight is the attraction of gravity.
F1 = mg (55)
Where m is the mass of the body and (g=9.81 m/sec.) is the
acceleration of gravity.
The next force to be considered is the spring force extended by
the spring. From Hook s law
F2 = ky (56)
Where y is the stretch, the constant of proportionality k is called the
spring modulus.
The last force acting on the weight is the weight upward due to
the weight stretch the spring by an amount y0 .
Then also from Hook s law:-
F3 = − ky0 (57)
It is clear that from equilibrium
mg = ky0 (58)
So that the total force acting on the weight at any position y is
FT = mg − ky0 − ky (59)
Substitute (58) into (59).
∴ FT = − ky (60)
By Newton s law, “Mass*acceleration=net force on the weight”
180 Linear Differential Equations Of Higher Order
m&y& = − ky , or m&y& + ky = 0 (61)
Is the differential equation represents the system shown in Fig.1.
The roots of (61) are complex. So, the oscillation of the system is
called harmonic oscillation.
If we connect the mass m to the dashpot, then we have to take the
corresponding various damping into account. The damping force has
direction opposite to the instantaneous motion. So, the damping
force is of the form
F4 = −cy& (62)
Where c is the damping constant. So the resulting force acting on
the body when stretched and released is
FT = − ky − cy& (63)
So the motion of the system is governed by the following
differential equation
m&y& + cy& + ky = 0 (64)
c k
The auxiliary equation is: λ2 + λ + = 0 (65)
m m
c 1
∴ λ1, 2 = − ± c 2 − 4mk (66)
2m 2m
c 1
∴α = − and β = c 2 − 4mk (67)
2m 2m
Where, λ1,2 = −α ± β (68)
Case 2 c 2 < 4mk , Then the roots of auxiliary equation are complex
conjugate roots (under damping).
_
Farads and E (t ) = 481cos10 t . + +
di 1
L + RI + ∫ idt = 481cos10 t vL(t)
dt c
Differentiate both sides of the Fig.2
above differential equation.
1
∴ LI&& + RI& + I = −4810 sin 10t
c
∴ 20 I&& + 160 I& + 500 I = −4810 sin 10t
x L-x
A B
x
Deflection, y(x)
y
Fig.3
Solution:- The total weight of the beam is WL, so each end supports
1
weight is WL . Let x be the distance from the left end A of the
2
beam. To find the bending moment M at x, consider forces to the left
of x
184 Linear Differential Equations Of Higher Order
1 1
(١) Force WL at A has moment − WLx
2 2
(٢) Force due to weight of the beam to left of x has magnitude Wx
1
and moment Wx( x / 2) = Wx 2
2
1 2 1
Then the total bending moment at x is Wx − WLx .
2 2
1 1
∴ EIy ′′ = Wx 2 − WLx (75)
2 2
By solving this equation for y (0) = 0 and y ′(0) = 0 we find,
W
y ( x) = ( x 4 − 2 Lx 3 + L3 x) (76)
24 EI
(b) The maximum deflection occurs at x = L / 2 , then substitute
5WL4
in (76) for x = L / 2 we get the maximum deflection as
384 EI
x L-x
A C
B
5-x 30000 N
x
y Deflection, y(x)
Fig.4
Chapter Five 185
Solution:-(a) The total weight of the beam is 3000*10=30000
Newton then the total weight is 30000+30000=60000 Newton, so
each end supports weight is 60000 / 2 = 30000 Newton. Let x be the
distance from the left end A of the beam. To find the bending
moment M at x, consider forces to the left of x
(١) Force 30000 N at A has moment − 30000 x
(٢) Force due to weight of the beam to left of x has
( )
∴ y P = x 2 A1 x 2 + A2 x + A3 = A1 x 4 + A2 x 3 + A3 x 2
∴ y ′P′ = 12 A1 x 2 + 6 A2 x + 2 A3 (79)
Substitute (79) into (77)
( )
∴ EI 12 A1 x 2 + 6 A2 x + 2 A3 = 1500 x 2 − 30000 x
1500 125 30000 5000
∴ A1 = = , A2 = − =− , A3 = 0
12 EI EI 6 EI EI
Then the general solution of (77) is:
∴ y ( x) =
1
EI
( )
125 x 4 − 5000 x 3 + c2 x + c1 (80)
∴ c1 = 0, c2 =
1
EI
(
5000 L2 − 125L3 = )
375000
EI
(81)
Substitute (81) into (89) we get:
∴ y ( x) =
1
EI
[
125 x 4 − 5000 x 3 + 375000 x ] (82)
(b) The maximum deflection occurs at x = L / 2 = 5m , then
substitute in (82) for x = 5 we get the maximum
1328125
deflection as
EI
Example 28 10 m
L-x
cantilever beam has one A C x
end horizontally
B
imbedded in concrete
Deflection, y(x) Fig.5
12000 N
and a force 12000 N
acting on the other end as shown in Fig.5. Find (a) the deflection
and (b) the maximum deflection of the beam assuming its weight is
3000 Newton/meter.
Solution:-
a) Let x be the distance from the left end B of the beam. To find the
bending moment M at x, consider forces to the right of x
(١) Force 12000 N at B has moment 12000 x
(٢) Force due to weight of the beam to the right of x has
y h = c1 + c2 x (84)
( )
y P = x 2 A1 x 2 + A2 x + A3 = A1 x 4 + A2 x 3 + A3 x 2
y ′P′ = 12 A1 x 2 + 6 A2 x + 2 A3 (85)
Substitute (85) into (83) we get:
( )
EI 12 A1 x 2 + 6 A2 x + 2 A3 = 1500 x 2 + 12000 x
1500 125 12000 2000
∴ A1 = = , A2 = = , A3 = 0
12 EI EI 6 EI EI
Then the general solution of (83) is:
y ( x) =
1
EI
( )
125 x 4 + 2000 x 3 + c2 x + c1 (86)
7.75 * 10 6 1.1 * 10 6
∴ c1 = , c2 = − (87)
EI EI
Substitute (87) into (86)
∴ y ( x) =
1
EI
[
125 x 4 + 2000 x 3 − 1.1 * 10 6 x + 7.75 * 10 6 ] (88)
7.75 * 10 6
deflection as
EI
188 Linear Differential Equations Of Higher Order
Problems
Solve the following differential equations
١) y ′′′ + y ′′ − 2 y ′ = 0
٢) y ′′ + 6 y ′ + 9 y = 0
٣) y′′ − 4 y′ + 13 y = 0
٤) y′′′ + − y′′ + 9 y′ − 9 y = 0
٥) y ′′ − 6 y ′ + 9 y = e 2 x
٦) y ′′ − y = 4 xe x
٧) y′′ − y = sin 2 x
٨) y ′′ + y = csc x
٩) ( )
y ′′ − 3 y ′ + 2 y = sin e − x
١٠) y ′′ + 4 y = sec 2 2 x
١١) (
y′′′ − 2 y′′ − 5 y′ + 6 y = e 2 x + 3 )2
١٢) y′′′ − 5 y′′ + 8 y′ − 4 y = e 2 x + 2e x + 3e − x
١٣) y ′′ + 4 y = sin 2 x + cos 2 x
١٤) y ′′ − 8 y ′ + 25 y = 5 x 3e − x − 7e − x
١٥) y ′′ − 9 y ′ + 14 y = 3x 2 − 5 sin 2 x + 7 xe 6 x
١٦) y (4 )+ y ′′′ = 1 − e − x
١٧) y ′′ + 2 y ′ + y = e − x Ln( x)
Chapter Five 189
١٨) x 2 y ′′ + xy ′ − 4 y = 0
١٩) x 3 y ′′′ − 3x 2 y ′′ + 6 xy ′ − 6 y = 0
farads and E (t ) = te − t .
6.1 Definition
For any given function f (t ), t ≥ 0 we can define another function
F (s ) by the following definite integral:
∞
L{ f (t )} = F ( s ) = ∫ e − st f (t )dt (1)
0
The function F (s ) is called the Laplace transform of the function
f (t ) . Note that F (s ) is simply the total area under the curve f (t )
for t=0 to infinity, whereas F (s ) for s greater than 0 is a "weighted"
f (t ) = e at
Solution:
∞ ∞ ∞ ∞
− st − st at − ( s − a )t e − ( s − a )t 1
F ( s ) = ∫ e f (t )dt = ∫ e e dt = ∫ e dt = =
0 0 0
− ( s − a) s−a
0
1
Example3 If F ( s ) = , find f (t )
s2
1 −1 1
Solution: Since F ( s ) = , we have L 2 = t ∴ f (t ) = t
s2 s
for t >T .
∴ L{ f ′(t )} = sL{ f (t )} − f (0)
This can be extended to be as following:
L{ f ( n)
(t )} = s n L{ f (t )} − s n −1 f (0) − s n − 2 f ′(0) − s n −3 f ′′(0)... − f n −1
(0) (2)
Theorem 3
If L−1{F ( s )} = f (t ) ,
∴ L−1{F ( s − a )} = e at f (t ) (3)
Proof:
∞ ∞
−1 − ( s − a )t
L {F ( s − a )} = ∫ f (t ) e dt = ∫ f (t ) e at e − st dt =e at f (t )
0 0
Theorem 4
t
If, g (t ) = ∫ f (t ) dt
0
1
∴ L{g (t )} = L{ f (t )} (4)
s
1
Example 4 Let F ( s ) = , find f (t )
s(s 2 + w2 )
Chapter Six 193
1 1
Solution: We know that L−1 2 = sin ωt
( s + ω 2 ) ω
−1
1 1 t
∴L 2 = ∫ sin ωt dt
s ( s + ω 2 ) ω 0
1 1
∴ L−1 2 = [1 − cos ωt ]
s ( s + ω 2 ) ω 2
1
Example 5 Let F ( s ) = , find f (t )
s 2 (s 2 + ω 2 )
Solution: We know from the previous example that:
1 1
L−1 2 = [1 − cos ωt ]
s ( s + ω 2 ) ω 2
−1
1 1 t
∴L 2 2 = ∫ [1 − cosωt ]dt
s ( s + ω 2 ) ω 2 0
1 1 1
∴ L−1 2 2 = t − sin ω t
s ( s + ω 2 ) ω 2 ω
a
Example 6 Proof that L{sin at} =
(s + a 2 )
2
e jat − e − jat
Solution: We know that: sin at =
j2
It follows from the linear character of the transform that:
1 1
L{sin at} = L{e jat } − L{e − jat }
j2 j2
194 Laplace Transforms
1 1 1 1 1 j 2a a
∴ L{sin at} = − = =
j 2 ( s − ja ) j 2 ( s + ja ) j 2 ( s 2 + a 2 ) ( s 2 + a 2 )
The same procedure can be used to proof that:
s
L{cos at} =
(s + a 2 )
2
Example 7
Find the Laplace transform of the following functions
s 4s
(c) L{4 cos 5t} = 4 * =
s 2 + 25 s 2 + 25
Γ(1 / 2) −3 π π
(d) L{−3 / t } = −3L{t1 / 2 } = −3 = = −3
s1 / 2 s s
1
(e) L{sin 2 t} = L (1 − cos 2t )
2
1 1 1
L{sin 2 t} = L (1 − cos 2t ) = L(1) − L{cos 2t}
2 2 2
1 s 2
= − 2 =
( 2
2 s 2 s + 4 s ( s + 4) )
Chapter Six 195
Example 8 Find the Laplace transform of the following functions:
{
Solution: (a) L e 5t t 3 = L t 3 } { }s = s −5 = s34! =
(s − 5)4
6
s = s −5
{ }
(b) L e − 5t cos 4t = L{cos 4t} s = s + 5 =
s
s 2 + 16 s = s + 5
=
s+5
( s + 5) 2 + 16
Theorem 5
For n = 1,2,3,.....
{ n
L t f (t ) = (−1) } n dn
ds n
F (s) (6)
Where F ( s ) = L{ f (t )}
Example 9
Find Laplace transform of the following functions
{ }
(a) L te at = −
d
ds
L(e at ) = −
d 1
=
1
ds s − a (s − a )2
d d w 2 ws
(b) L {t sin wt} = − L(sin wt ) = − =
ds 2
ds s + w 2
s 2 + w2 ( )2
{ 2 d
}
(c) L t sin wt = − L(t sin wt ) = −
d 2 ws
=
6 ws 2 − 2 w3
ds ds s 2 + w 2 ( ) (s 2 + w2 )3
2
196 Laplace Transforms
{ }
L te − t cos t = −
d
ds
d
L(e − t cos t ) = − L(cos t )
ds s → s +1
(d)
=−
d s + 1
=
( s + 1)2 − 1
(
ds (s + 1)2 + 1 (s + 1)2 + 1 2 )
(e) t 3 cos ωt
s
we know that L{cos ωt} =
s2 + ω 2
{ 3
}
∴ L t cosωt = (− 1)
d3
3
2
s
3
2
ds s + ω
{ 3
∴ L t cosωt = − 2 }
d 2 s 2 + ω 2 − 2 s 2
ds s 2 + ω 2 ( )
{ 3
}
d s2 + ω 2
∴ L t cos ωt =
( )2
(− 2 s ) + (s 2 − ω 2 )(2 * (s 2 + ω 2 )* 2 s )
ds
(s 2 + ω 2 )4
=−
(s 2 + ω 2 ) (6 s 2 − 6ω 2 ) − 3 * 2 s (s 2 + ω 2 ) * (2 s 3 − 6ω 2 s )
3 2
(s 2 + ω 2 )6
6(s 2 + ω 2 ) (s 2 − ω 2 ) + 12 s 2 (s 2 − 3ω 2 )
2
∴ L{ t cos ωt }= −
3
(s 2 + ω 2 )5
Example 10 Solve the following differential equation:
y ′′ + 2 y ′ + 5 y = 0 , y (0) = 2, y ′(0) = −4
Solution:
Chapter Six 197
∴ y (t ) = 2e − t cos 2t − e − t sin 2t
∴ y (t ) = e − t (2 cos 2t − sin 2t )
G ( s)
Y ( s) = (7)
H ( s)
Where G (s ) and H (s ) are polynomials of s.
Assume G (s ) and H (s ) have no common factors and have real
coefficients. The degree of G (s ) is lower than H(s).
Let s=a be a root of H ( s ) = 0 . There are many cases for the roots of
H (s ) will be explained in the following cases:
Case 1 Unrepeated factors
G(s) A A2 A
Y (s) = = 1 + + .... n + W ( s) (8)
H ( s) s − a1 s − a2 s − an
Where W (s ) is denotes the sum of the partial fractions
corresponding to all the (unrepeated or repeated) linear factors of
H (s ) which are not under consideration.
198 Laplace Transforms
G ( s) 2s 2 − 4
Y ( s) = =
H ( s) s 3 − 4s 2 + s + 6
Chapter Six 199
Solution:
G ( s) 2s 2 − 4 A1 A2 A3
Y ( s) = = 3 = + +
H ( s ) s − 4 s 2 + s + 6 ( s − 2) ( s + 1) ( s − 3)
∴ A1 =
(s − 2) * G(s)
=
(
(s − 2) 2s 2 − 4 ) =
4
=−
4
H ( s) s = 2 (s − 2)(s + 1)(s − 3) s =2
3 * −1 3
∴ A2 =
(s + 1) * G(s)
=
( 2
(s + 1) 2s − 4 ) =
−2
=−
1
H ( s) s=−1 ( s − 2)(s + 1)(s − 3) s=−1
− 3 * −4 6
∴ A3 =
(s − 3) * G(s)
=
(s − 3) 2s 2 − 4 ( ) =
14 7
=
H ( s) s =3 ( s − 2)(s + 1)(s − 3) s =3
4 2
G ( s) − 4 / 3 − 1 / 6 7/2
∴ Y ( s) = = + +
H ( s ) ( s − 2) ( s + 1) ( s − 3)
4 1 7
∴ y (t ) = − e 2t − e − t + e3t
3 6 2
m
Case 2 Repeated factor ( s − a ) , in this case;
G(s) Am Am −1 A1
Y (s) = = + + ...... + W ( s) (11)
H ( s ) (s − a )m (s − a )m −1 (s − a )
−1 t m−1
at t m−2
L {Y(s)}= e Am + Am−1 + ....+ A2t + A1 + L−1W(s) (12)
(m −1)! (m − 2)!
G (s )
Where: Am = (s − a )m (13)
H (s )
1 d m − k (s − a )m G ( s)
Ak = , (14)
(m − k ) ! ds m − k H ( s)
s =a
Where, k = 1,2,...., m − 1
200 Laplace Transforms
Example 12 Solve the following initial value problem:
y ′′ − 3 y ′ + 2 y = 4 t + e 3t , y ( 0 ) = 1, y ′( 0 ) = − 1
Solution:
4 1
s 2Y ( s ) − sy (0) − y′(0) − 3[ sY ( s ) − y (0)] + 2Y ( s ) = +
s2 s −3
G ( s ) s 4 − 7 s 3 + 13s 2 + 4s − 12
Y (s) = =
H (s) s 2 ( s − 3)( s 2 − 3s + 2)
A A B C D
= 22 + 1 + + +
s s ( s − 3) ( s − 2) ( s − 1)
2G (s ) s 4 − 7 s 3 + 13s 2 + 4 s − 12 − 12
A2 = s * = = =2
H (s ) s = 0 ( s − 3)( s 2 − 3s + 2) −6
s =0
1 d s 4 − 7 s 3 + 13s 2 + 4 s − 12
A1 = =3
( 2 − 1)! ds ( s − 3)( s 2 − 3s + 2) s =0
G (s ) s 4 − 7 s 3 + 13s 2 + 4 s − 12 1
B = (s − 3) * = =
H (s ) s = 3 s 2 ( s 2 − 3s + 2) 2
s =3
s 4 − 7 s 3 + 13s 2 + 4 s − 12
C= = −2
s 2 ( s − 3)( s − 1) s=2
G (s ) s 4 − 7 s 3 + 13s 2 + 4 s − 12 1
D = (s − 1) * = = −
H (s ) s =1 s 2 ( s − 3)( s − 2) 2
s =1
2 3 1/ 2 2 1/ 2
∴ Y (s) = + + − −
s 2 s ( s − 3) ( s − 2) ( s − 1)
1 1
∴ y (t ) = 2t + 3 + e3t − 2e 2t − et
2 2
Chapter Six 201
Case 3 Unrepeated complex factors (s − a )
2 2
Where a = α + jβ , a = α − jβ , ( s − a )( s − a ) = ( s − α ) + β
G(s) As + B
Y ( s) = = + W (s) (15)
H ( s ) (s − α )2 + β 2
1 αt
L−1{Y ( s)} = e [Ta cos β t + S a sin β t ] + L−1{W ( s)} (16)
β
[
Where Ra ( s) = S a + jTa = ( s − α ) 2 + β 2 ] HG((ss)) (17)
s =a
∴ S a = −2, Ta = 4
1
∴ L−1{Y ( s )} = eαt [Ta cos β t + S a sin β t ] + L−1{W ( s)}
β
1
= e − t [4 cos 2 t − 2 sin 2 t ]
2
= e − t [2 cos 2 t − sin 2 t ]
202 Laplace Transforms
Example 14 Solve the initial value problem
y ′′ − 3 y ′ + 2 y = e t , y ( 0 ) = 1, y ′( 0 ) = 1
1
Solution: s 2Y ( s ) − sy (0) − y ′(0) − 3[ sY ( s ) − y (0)] + 2Y ( s ) =
s −1
G ( s) s 2 − 3s + 3 A2 A1 B
Y (s) = = = + +
H ( s ) (s − 1)2 ( s − 2) (s − 1)2 (s − 1) ( s − 2)
s 2 − 3s + 3
∴ A2 = = −1
( s − 2)
s =1
d s 2 − 3s + 3
∴ A1 =
ds ( s − 2)
s =1
’
2
( s − 3s + 3) − ( s − 2)(2 s − 3)
= =0
( s − 2) 2 s =1
s 2 − 3s + 3
∴ B= =1
2
( s − 1) s=2
−1 1
∴Y ( s) = +
(s − 1)2 ( s − 2)
∴ y (t ) = −tet + e 2t
s 3 − 7 s 2 + 14s − 9
F (s) =
( s − 1) 2 ( s − 2)3
Solution:
Chapter Six 203
A2 A1 B3 B2 B1
F (s) = + + + +
( s − 1) 2 ( s − 1) ( s − 2)3 ( s − 2) 2 ( s − 2)
s 3 − 7 s 2 + 14 s − 9
A2 = =1
( s − 2) 3 s =1
d s 3 − 7 s 2 + 14s − 9
A1 = =0
ds ( s − 2) 3 s =1
s 3 − 7 s 2 + 14s − 9
B3 = = −1
( s − 1) 2 s=2
d s 3 − 7 s 2 + 14s − 9
B2 = =0
ds ( s − 1) 2 s =2
1 d 2 s 3 − 7 s 2 + 14s − 9
B1 = =0
2! ds 2 ( s − 1) 2 s=2
1 1
∴ F (s) = 2
−
( s − 1) ( s − 2) 3
1
∴ f (t ) = te t − t 2 e 2t
2
s +1
Example 16 Find f (t ) if F (s ) equals: F (s) =
s ( s 2 + 2) 3
2
Solution:
A2 A1 B3 B2 B1
F (s ) =
) ( )
+ + + +
s2 s (
s2 + 2
3
) (
s2 + 2
2
s2 + 2
204 Laplace Transforms
s +1 1
∴ A2 = =
( s 2 + 2) 3 s =0
8
d s +1 1
∴ A1 = =
ds ( s 2 + 2) 3 8
s =0
(
Q s2 + 2 = 0 )3 ∴ s = ± j 2 = α + jβ
∴ α = 0, β = 2
s +1 1 1
∴ B3 = =− − j
s2 s= j 2 2 2
d s +1 s 2 − 2s (s + 1) 1 1
∴ B2 = = =− − j
ds s 2 s = j 2 s4 2 2 2
s= j 2
1 d − s 2 − 2s
∴ B1 =
2! ds s4 s= j 2
1 s 4 (− 2 s − 2 ) + 4 s 3 s 2 + 25
=
( ) = −0.75 − j 0.3535
2 s8 s = j 2
t 1 1 −1 1 2 1
∴ y(t ) = + + cos 2t − sin 2 t − cos 2t − sin 2t
8 8 2 2 2 4 2
− 0.35 cos 2t − 0.75sin 2t ]
∴ y (t ) =
(t + 1) − (cos 2t + 1.24 sin 2t )
8
Chapter Six 205
Theorem 6
f (t ) ∞
If L{ f (t )} = F ( s ) ∴ L = ∫ F ( s ) ds (18)
t s
Proof:
∞
By definition: L{ f (t )} = F ( s ) = ∫ f (t )e − st dt
0
∞
∞ s −1 − 1
f (t )= tL−1 ∫ 2 2 ds = tL 2 s 2 − 1
s
s − 1( )
( )s
1 sinh t
∴ f (t ) = tL−1 2
2 s(− 1
=t
) 2
uc(t)
Fig.1
t
t=c
e − cs
∴ L{uc (t )} =
s
Example 21 Find Laplace transform of 8 * u3 (t )
Solution: This function has the following shape Fig.2
8e − 3s
∴ L{8 * u3 (t )} =
s
208 Laplace Transforms
8u3(t)
t
t=3
Fig.2
f (t )
Example 22 Find
Laplace transform of the
following function 2
2 2≤t≤3
f (t ) =
0 Otherwise 2 3
Solution: f 1 (t )
The function f (t ) is u 2 (t )
shown in Fig.3. This 2
function can expressed in f 2 (t ) 3
∴ F (s ) = L{ f (t )} = L{2u 2 (t ) − 2u3 (t )} =
2e − 2 s 2e − 3s 2 e − 2 s − e − 3s
− =
( )
s s s
We can check the above solution by using the elementary Laplace
transform principles as following:
L{ f (t )} = 2 * ∫
3 − st
e dt = 2*
e − st
3
= 2 = (
e − 3s − e − 2 s 2 e − 2 s − e − 3s )
2 −s − s s
2
Chapter Six 209
Example 23 Find Laplace transform of the following function
3 0≤t ≤2 4
f (t ) = − 2 2≤t ≤6 3
A B
0 Otherwise
2
Solution:
1 Fig.4
In this example we will try to
E
get general method to convert
O 1 2 3 4 5 6
this discontinuous function in -1
terms of unit step function. If
-2 C
we look deeply to the function D
shown in Fig.4 we can see that, this function consists of four unit
step functions as following: 3uo (t ), −5u 2 (t ) , and, 2u6 (t ) ,
where f (t ) = 3uo (t ) − 5u 2 (t ) + 2u6 (t )
If we look deeply to Fig.4 we can see that it is easy to get that by
starting from the origin and take each jump (arrow) as a unit step
function. It is clear from Fig.4 that we have three jumps (arrows)
OA, BC, and DE. For OA its direction is up so we will take +ve sign
for the unit step function and its length is 3 so the amplitude of unit
step function is 3 and it happened at t = 0 , then the suffix for unit
step function will be zero. Then the first arrow (OA) equivalent to
OA = 3uo (t ) . In the same way we can deal with the jump (arrow) BC
and DE and they equivalent to − 5u 2 (t ) and 2u6 (t ) respectively.
The following table clarifies the jump method.
210 Laplace Transforms
Jump Direction Sign Amplitude time Equivalent
OA UP +ve 3 t =0 3uo (t )
BC DOWN -ve 5 t=2 − 5u 2 (t )
DE UP +ve 2 t =6 2u6 (t )
∴ f (t ) = 3uo (t )− 5u 2 (t ) + 2u6 (t )
3 − 5e − 2 s + 2e − 6 s
∴ L{ f (t )} = {3uo (t ) − 5u 2 (t ) + 2u6 (t )} =
s
t2 0<t <3
Solution: The function f (t ) Is show inFig.6.
∴ F (s ) =
[ ]
s 2 − 9se − 3s + 3e − 3s − 3e − 3s − 2s (3se − 3s + e − 3s − 1)
s4
− 9s 2 e − 3s − 6se − 3s − 2e − 3s + 2
∴ F (s ) =
s3
∴ F (s ) =
− e − 3s
s3
(9s 2 + 6s + 2)+ s23
t2
Fig.6
3
uo (t ) − u3 (t )
t 2 (uo (t ) − u3 (t ))
Another solution:
{ ( ) 2
∴ L t 2uo (t ) − (t − 3)2 + 6(t − 3) + 9 u3 (t ) =
− }+
6 9 − 3s
2
+ e
s3 s3 s 2 s
Which is clear, the same result that we got from previous solution.
et 0 < t <1
Solution: As explained in the previous example the above
1 e − (s −1)
∴ F (s ) = L{ f (t )} = −
s − 1 (s − 1)
Another solution:
This equation, f (t ) = et [uo (t ) − u1 (t )] can be modified to be in the
et −1
form of (21). ∴ e [uo (t ) − u1 (t )] = e uo (t ) − −1 u1 (t )
t t
e
It is clear the above equation is in the form of (21).
t e t −1 1 e− s 1 e − ( s −1)
∴ L e uo (t ) − −1 u1 (t ) = − * e = −
e
s − 1 s − 1 s − 1 s − 1
Which is clear, the same result that we got from previous solution.
f (t ) = sin t + u 2π (t ) cos (t )
To put the above equation in the form of (21), we have to replace
cos(t ) with cos(t − 2π ) . It is clear that cos(t ) = cos(t − 2π ) .
f (t ) = sin t + u 2π (t ) cos (t ) = sin t + u 2π (t ) cos (t − 2π )
∴ L{ f (t )} = L{sin t} + L{u 2π (t ) cos (t − 2π )}
∴ L{ f (t )} = L{sin t} + e − 2πs L{cos (t )}
1 − 2πs s 1 + s e − 2πs
∴ L{ f (t )} = +e =
s2 + 1 s2 + 1 s2 + 1
Example 29 Find Laplace transform of the following function:
f (t ) = t 2 u 2 (t )
Chapter Six 215
Solution: Laplace transform of the above function can be
obtained by two different methods as following:
First method :We First find L{u2 (t )} . Then, by using (6) we can
{ }
obtain L t 2 u 2 (t ) as following:
e − 2s
Q L{u 2 (t )} =
s
{ }
∴ L t u 2 (t ) = (− 1)
2 d 2 e − 2 s
2
ds 2 s
= e − 2s 2
3
s
+
4 4
+
s2 s
The second method: We can modify t 2 u2 (t ) to take the form in
{ } {( ) }
∴ L t 2u 2 (t ) = L (t − 2 )2 + 4(t − 2) + 4 u 2 (t )
∴ L{t 2u 2 (t )}=
2e − 2 s 4e − 2 s 4e − 2 s
3
+ 2 +
s s s
∴ L{t 2u 2 (t )}= 3 + 2 + e − 2 s
2 4 4
s s s
1 − e− s
Example 30 Find the inverse transform of
s2
Solution: From the linearity of the transform (Theorem 7) we have:
1 − e − s −1 1 −1 e − s
−1
f (t ) = L 2 = L 2− L 2
s s s
∴ f (t ) = t − (t − 1)u1 (t )
Hence f (t ) may also be given in the form:
0 ≤ t < 1,
f (t ) = 1
t,
t ≥1
216 Laplace Transforms
Example 31 Solve the following initial value problem
y ′′ + 4 y = r (t ) y (0) = 1, y ′(0) = 0 (23)
π ≤ t < 2π ,
Where, r (t ) =
1,
(24)
0 otherwise
2 e −πs e − 2πs
( s + 4)Y ( s ) − s = −
s s
−πs
s e e − 2πs
Y ( s) = 2 + − (26)
( s + 4) s ( s 2 + 4) s ( s 2 + 4)
To compute y (t ) we must obtain the inverse transform of each term
on the right side of equation (26). The inverse transform of the first
term of (26) is:
s
y1 (t ) = L−1 = cos 2t (27)
2
( s + 4)
The inverse transform of the second term of (26) is:
−1
e − πs
−1 −πs A1 A2 s + A3
L = L e +
s ( s 2 + 4) s ( s 2
+ 4)
e −πs 1
∴ A1 = =
( s 2 + 4) s = 0 4
Chapter Six 217
u (t ) u (t )
∴ y2 (t ) = π + π [Ta cos 2 (t − π ) + S a sin 2 (t − π )] (28)
4 2
[
Where Ra ( s ) = S a + jTa = ( s − α ) 2 + β 2 ]HG((ss)) = S1 =
1
j2
s= j2
1 1
∴ S a + jTa = − j , then Ta = − (29)
2 2
Substitute (29) into (28) we get:
u (t ) uπ (t ) 1
∴ y 2 (t ) = π + − cos 2 (t − π )
4 2 2
u (t )
∴ y2 (t ) = π [1 − cos 2 (t − π )] (30)
4
Similarly, the inverse transform of the third term of (26) is:
u 2π (t )
∴ y3 (t ) = [1 − cos 2 (t − 2π )] (31)
4
Combining (27), (30) and (31) we obtain finally:
y (t ) = y1 (t ) + y 2 (t ) + y3 (t )
u (t ) u (t ) (32)
= cos 2t + π [1 − cos 2 (t − π )] − 2π [1 − cos 2 (t − 2π )]
4 4
f (t )
1
t
b 2b
The period of the given function is 2b. Hence from (33) we can get
the Laplace transform as following:
1 2b
L{ f (t )} = ∫
− 2bs 0
f (t )e − st dt
1− e
1 b1 * e − st dt + 2b (− 1)e − st dt
∴ L{ f (t )} = ∫ ∫b
1 − e − 2bs 0
− st b 2b
1 e e − st
∴ L{ f (t )} = −
1 − e − 2bs − s −s
0 b
∴ L{ f (t )} =
1 1 − 2e − bs + e − 2bs
=
(
1 − e − bs )2
1 − e − 2bs s ( )(
s 1 − e − bs 1 + e − bs
)
∴ L{ f (t )} =
1 − e − bs (ebs / 2 − e −bs / 2 ) = 1 tanh bs
s (1 + e − bs ) s (e bs / 2 + e − bs / 2 ) s
=
2
Chapter Six 219
Example 33 Find Laplace transform of the saw tooth wave shown in
Fig.8
f(x)
k k
x
k 2k 3k
Fig.8 Saw tooth waveform.
Solution: In this case, the period of the function is k, hence;
k
1 k − st 1 e − st
L{ f (t )} = ∫
− ks 0
te dt = 2 (− st − 1)
1− e 1 − e − ks s 0
1 − (1 + ks )e − ks (1 + ks ) − (1 + ks )e − ks − ks
∴ { f (t )} =
(
s 2 1 − e − ks ) =
(
s 2 1 − e − ks )
(1 + ks )(1 − e − ks ) − ks (1 + ks ) k
∴ L{ f (t )} = = −
s 1− e 2
(
− ks
s )
s 1 − e − ks
2
( )
Example 34 What is the Laplace transform of the staircase function
shown in Fig.9?
f(t)
4 Fig.9
3
2
1
k 2k 3k 4k t
220 Laplace Transforms
f(t)
4 Fig.10
3
2
1
k 2k 3k 4k t
t+k
k
k 2k
3k 4k
The required transform can easily be found by direct calculation.
However, it is even simpler to obtain it by considering f (t ) to be
the difference between the two functions shown in Fig.10. The
transform of the sawtooth function ( f1 (t )) can be obtained as in the
previous example as following:
k
1 k t − st 1/ k e − st
L{ f1 (t )} = ∫
− ks 0 k
e dt = 2 (− st − 1)
1− e 1 − e − ks s 0
Chapter Six 221
1 − (1 + ks )e − ks (1 + ks ) − (1 + ks )e − ks − ks
∴ L{ f (t )} = =
(
ks 2 1 − e − ks ) (
ks 2 1 − e − ks )
∴ L{ f (t )} =
(1 + ks )(1 − e − ks ) − ks = (1 + ks ) − 1
(
ks 2 1 − e − ks ) ks 2 (
s 1 − e − ks )
t + k
The transform of the linear function f 2 (t ) = can be found
k
(t + k ) 1 1 k
as following: L = +
k k s2 s
Then Laplace transform of the staircase function is:
1 1 k 1 (1 + ks ) k 1
L{ f (t )} = 2 + − 2 −
k s s k s s 1− e − ks =
(
s 1− e
− ks
) ( )
6.6 Pulse Functions
In some applications it is necessary to deal with functions with
pulse nature, for example, voltages or forces of large magnitude,
which act over very short time intervals. Such problems often lead
to differential equation in the following form:
ay ′′ + by ′ + cy = r (t ) (34)
Where r (t ) is very high during the short interval 0 < t < τ , and
otherwise zero. In particular let us suppose that r (t ) is given by:
1 , 0 ≤ t <τ,
r (t ) = τ = δ (t ) (35)
0 otherwise
222 Laplace Transforms
∞
Where ∫ δ (t )dt = 1 (36)
−∞
∞ τ
− st 1 − st 1 − e −τs
∴ L{δ (t )} = ∫ e δ (t )dt = lim ∫e dt = lim =1 (37)
τ →0 τ τ →0 τs
0 0
equation: ( s 2 + 2 s + 2) Y ( s ) = e −πs
e −πs 1
∴ Y (s) = = e −πs
( s 2 + 2 + 2) ( s + 1) 2 + 1
∴ y (t ) = uπ (t ) e − (t −π ) sin(t − π )
Chapter Six 223
6.7 Applications
6.7.1 Electric circuits
Example 36 By using Laplace transform, Find the current I (t ) in
Fig.11. Assume I (0) = 1 A and v L (0) = 1 V for R=20 Ω, L=5H,
C=0.04 farads and E (t ) = 100 cos 5 t .
vR(t)
Solution:
_
From KVL + +
VL + VR + VC = E (t )
E(t)
_ I(t) _ vC(t)
di 1
+ RI + ∫ idt = 100 cos 5t
+
L
dt c vL(t)
Differentiate both sides of the above
Fig.11
differential equation. Then.
1
LI&& + RI& + I = −500 sin 5t
c
∴ I&& + 4 I& + 5 I = −100 sin 5t
I ( 0 ) = 1, V L ( 0 ) = L I& ( 0 ) = 1 ∴ I& ( 0 ) = 0 .2
− 100 * 5
∴ s 2 I ( s ) − s − 0.2 + 4[ sI ( s ) − 1] + 5 I ( s ) =
( s 2 + 25)
[
∴ Ra ( s ) = S a + jTa = ( s − α1 ) 2 + β12 ]HG((ss)) = [s + 5]HG((ss))
2
− 500
= = 12.5 + j12.5
(−25 + j 20 + 5)
∴ S a = 12.5, Ta = 12.5
1 αt
∴ L−1{I1 ( s)} = e [Ta cos β1 t + S a sin β1 t ] + L−1{W ( s )}
β1
1
= [12.5 cos 5 t + 12.5 sin 5 t ] + L−1{W ( s)}
5
= 2.5[cos 5 t + sin 5 t ] + L−1{W ( s )}
a2 = −2 + j1, a2 = −2 − j1 , ∴α 2 = −2, β 2 = 1
[
∴ Ra ( s ) = S a + jTa = ( s − α 2 ) 2 + β 2 2 ]HG((ss))
− 500 + ( s + 4.2)( s 2 + 25)
=
( s 2 + 25) s = −2 + j1
From KVL
t,sec
0.1 0.2 0.3
VL + VR + VC = E (t )
di 1
L + RI + ∫ idt = E (t )
dt c
Differentiate both sides of the above differential equation. Then.
1
∴ LI&& + RI& + I = E (t )
c
R 1 1 d
∴ I&& + I& + I= (E (t ))
L Lc L dt
1
∴ I&& + 3I& + 2 I = * 1000
50
1
∴ I&& + 3I& + 2 I = * 1000
50
Q v L = LI&, vL (0 ) = 5
∴ 50 I&(0 ) = 5
∴I&(0 ) = 0.1
20
∴ s 2 I ( s ) − 0.1 + 3sI ( s ) + 2 I (s ) =
s
[
∴ I ( s ) s 2 + 3s + 2 = ] 20
s
+ 0.1 =
20 + 0.1s
s
226 Laplace Transforms
G ( s) 20 + 0.1s 0.1s + 20
∴ I (s) = = =
H ( s ) s ( s 2 + 3s + 2) s ( s + 1)(s + 2 )
G(s) A B C
Let I ( s ) = = + +
H ( s ) s ( s + 1) (s + 2 )
0.1s + 20
∴ A= = 10
( s + 1)(s + 2 ) s = 0
0.1s + 20
∴ B= = −19.9
s (s + 2 ) s = −1
0.1s + 20
∴C= = 9.9
s ( s + 1) s = −2
10 19.9 9.9
∴ I (s ) = − +
s s +1 s + 2
∴ I (t ) = 10 − 19.9e − t + 9.9e − 2t
2π
500 sin 5t ⇒ 0≤t ≤
5
E (t ) =
500 2π
⇒ t>
5
Solution: It is clear that E (t ) can be expressed as following:
E (t ) = 500 sin 5t − u2π / 5 (500 sin 5t − 500)
di 1
From KVL , L + RI + ∫ Idt = E (t )
dt c
Chapter Six 227
R 1 1
∴ I& + I + ∫ Idt = E (t )
L LC L
1 u
∴ I&& + 3I& + 2 I = 2500 cos 5t − 2π (2500 cos 5t )
50 5
u
∴ I&& + 3I& + 2 I = 50 cos 5t − 2π (50 cos 5t )
5
2π
− s
50 s 50.s.e 5
∴ s 2 I (s ) − sI (0) − I&(0 ) + 3[sI (s ) − I (0 )] + 2 I (s ) = −
s 2 + 25 s 2 + 25
I (0) = 0, VL = LI&, VL (0) = 50 * I&(0 ) = 5
∴ I&(0 ) = 0.1 A / sec .
2π
− s
50s 1 − e 5
[ ]
∴ I (s ) s + 3s + 2 − 0.1 =
2
s 2 + 25
2π
− s
2
0.1s + 50s + 2.5 − 50s e 5
∴ I (s ) =
(s + 1)(s + 2 )(s 2 + 25)
A B C
∴ I (s ) = + + 2
(s + 1) (s + 2 ) s + 25( )
2π
− s
2
0.1s + 50s + 2.5 − 50s e 5
∴A= = 4.934
(s + 2)(s 2 + 25)
s = −1
228 Laplace Transforms
2π
− s
0.1s 2 + 50 s + 2.5 − 50 s e 5
∴B = = −39.2217
(s + 1)(s 2
+ 25 )
s = −2
2
For s + 25 , α = 0, β = 5
2π
− s
2
0.1s + 50s + 2.5 − 50s e 5
∴ S a + jTa = = 0 + j0
(s + 1)(s + 2 )
s = j5
4.934 39.2217
∴ I (s ) = −
(s + 1) (s + 2 )
∴ i (t ) = 4.934 e − t − 39.2217 e − 2t
E (t ) = 100 sin t
t
-π 2π
E ′(t ) π
100
t
-π π 2π
∴I&(0 ) = 0.1
1
∴ s 2 I ( s ) − 0.1 + 3sI ( s ) + 2 I (s ) = L{E ′(t )}
50
E ′(t ) is shown in Fig.13. This function is periodical and Laplace
transform for it can be obtained from (33) as following:
π
− st
∫ 100 * cos t e dt
∴ L{E ′(t )} = 0
1 − e −πs
πs
(
100 s 1 + e
∴ L{E ′(t )} = 2
−πs
)
100s / tanh
2
(
s +1 1− e )(
−πs
= 2
s +1 ) ( )
πs
100s / tanh
1 2
∴ s 2 I ( s ) − 0.1 + 3sI ( s ) + 2 I (s ) =
50 2
s +1 ( )
230 Laplace Transforms
πs
2 s / tanh
( )
∴ I ( s ) s 2 + 3s + 2 = 0.1 + 2
s2 + 1 ( )
πs
2 s / tanh
0.1 2
∴ I (s) = 2
(
s + 3s + 2
+ 2
) (
s + 1 s 2 + 3s + 2 )( )
πs
2 s / tanh + 0.1 s 2 + 1
2
( )
∴ I (s) =
( )(
s 2 + 1 s 2 + 3s + 2 )
A B Cs + D
∴ I ( s) = + + 2
(s + 1) (s + 2) s + 1 ( )
πs
2 s / tanh + 0.1 s 2 + 1
2
( )
∴A=
( )
s + 1 (s + 2 )
2
= 1.19
s = −1
πs
2s / tanh + 0.1 s 2 + 1
2
( )
∴B =
( )
s + 1 (s + 2 )
2
= −0.903
s = −2
πs
2s / tanh + 0.1 s 2 + 1
2
( )
S a + jTa =
(
s 2 + 3s + 2 ) =0
s = j1
1.19 0.903
∴ I (s ) = −
s +1 s + 2
∴ I (t ) = 1.19e − t − 0.903e − 2t
Chapter Six 231
6.7.2 Newton s law
Example 40 The mass m of Fig.2 is suspended from the end of a
vertical spring of constant k [force required to produce unite
stretch]. An external force
F (t ) acts on the mass as well as
a resistive force proportional to
the instantaneous velocity.
Assuming that y is the
displacement of the mass at time y
m
t and that the mass starts from
rest at y=0, (a) set up a
differentia equation for the motion and (b) find y at any time t.
Solution:
dy
(a) The resistive force is given by − B . The restoring force is
dt
given by − ky . Then by Newton’s law,
d2y dy
m = − B − ky + F (t )
dt dt
d2y dy
∴m + B + ky = F (t ) (42)
dt dt
Where, y (0) = 0, , y& (0) = 0 (43)
(b) Taking the Laplace transform of (42), we obtain
wm 0∫
y (t ) = F (u )e 2m sin w(t − u )du
Case 2 R=0
Bt
−1 1 −
In this case L = te 2 m
B
2
m s +
2 m
Chapter Six 233
And by convolution form in (44) yields:
B (t − u )
1 t −
wm 0∫
y (t ) = F (u )(t − u )e 2m sin w(t − u )du
αm 0∫
y (t ) = F (u ) e 2m sinh α (t − u )du
x L-x
A C
B
5-x 30000 N
x
y Deflection, y(x)
Fig.3
234 Laplace Transforms
Solution: (a) The total weight of the beam is 3000*10=30000
Newton then the total weight is 30000+30000=60000 Newton, so
60000
each end supports weight is = 30000 Newton. Let x be the
2
distance from the left end A of the beam. To find the bending
moment M at x, consider forces to the left of x
(١) Force 30000 N at A has moment
− 30000 x
(٢) Force due to weight of the beam to left of
x has magnitude 3000x and moment
3000 x( x / 2) = 1500 x 2
[
EI s 2Y ( s ) − sy (0) − y ′(0) = ] 1500 * 2
s3
−
30000
s2
1 3000 30000 y ′(0)
∴ Y ( s) = − + 2
EI s 5 4
s s
∴ y (t ) =
1
EI
[ ]
125 x 4 − 5000 x 3 + y ′(0) x (46)
But y (L = 10 ) = 0 , then substitute in (46), we get:
y ′(0) = 375000 / (EI )
Chapter Six 235
Substitute that in (46) we get the general solution of (45)
∴ y (t ) =
1
EI
[
125 x 4 − 5000 x 3 + 375000 x ]
Problems
[I] Find the Lapalce transform of the following functions f(t) where
c,k, and w are constants
١) t 2 + 3t + 4
٢) t 2 e − 2t
٣) e − t cos 2t
٤) sin( wt + k )
٥) cosh 2 (3t )
٦) 3t sin wt + wt 2 cos wt
٧) t 3 sin wt
٨) t 2 sinh wt
٩) cos wt sinh wt + sin et cosh wt
١٠)
t
(
1 − ct
e − e − kt )
[II] Solve the following initial value problems by using Laplace
transformation
١١) 4 y ′′ − 8 y ′ + 5 y = 0, y (0) = 0, y ′(0) = 1
١٢) y′′ + y′ = 3 cos 2t , y (0) = 0, y′(0) = 0
236 Laplace Transforms
٢٢) y ′′ + 4 y = δ (t − π ) − δ (t − 2π ), y ( 0) = 0, y ′( 0 ) = 0
٢٣) y′′ + 2 y′ + y = δ (t ) + u 2π (t ), y ( 0) = 0, y ′( 0 ) = 1
Chapter Six 237
٢٤) y′′ − y = 2δ (t − 1), y ( 0) = 1, y ′( 0 ) = 0
٢٥) y′′ + ω 2 y = δ (t − π / ω ), y ( 0) = 1, y ′( 0 ) = 0
٢٦) y′′ + 2 y ′ + 3 y = sin t + δ (t − π ), y ( 0) = 0, y ′( 0 ) = 1
5s 2 − 15s + 7
٢٢)
( s + 1)( s − 2) 3
s2 + 2
٢٣)
( s 2 + 10)( s 2 + 20)
[IV] In each case graph the given function, which is assumed to
be zero outside the given interval, and find Laplace transform
٢٤) t 0<t<2
٢٥) t2 0<t <3
٢٦) et 0 < t <1
٢٧) k sin wt 0 < t <π /w
٢٨) k cos wt 0 < t <π /w
٢٩) k cos wt 0 < t < 2π / w
٣٠) 1 − e −t 0<t <π
Find and graph the inverse Laplace transform of the following
functions:
238 Laplace Transforms
2e − 0.5s
٣١)
s
−πs
se
٣٢)
s2 + 4
e −πs
٣٣)
s 2 + 2s + 2
s (1 + e −πs )
٣٤)
s2 +1
٣٥) Find the current in the RLC circuit in Fig.1. Assume
I (0) = 0 and v L (0) = 0 for R=160 , L=20H,
7.1 Introduction
An important mathematical question raised by Joseph Fourier in
1807, arising from his practical work on heat conduction, is whether
an arbitrary function f (x) with period 2 L can be represented in the
form of a Fourier series:
∞ nπx nπx
f ( x ) = a0 + ∑ an cos + bn sin (1)
n =1 L L
The above representation is good for all x,−∞ < x < ∞ . If f (x) is
not periodic outside the interval (− L < x < L) or if f (x) is not
defined beyond this interval, the representation is good only in the
restricted interval.
A second question is: suppose we can indeed represent f (x) by a
Fourier series of the form (1), how do we calculate Fourier
Coefficients a0 , an , bn ?
Before we start finding the Fourier coefficients we have to state
some important notes
Theorem 1 A function f ( x ) is said to be even on the interval
(− L, L) if f (− x) = f ( x) .
This means that this function is symmetric about the y axis.
Chapter Seven 239
Theorem 2 A function f ( x ) is said to be odd on the interval
(− L, L ) if f (− x) = − f ( x) . This means that this function is
symmetric about the origin in xy plan.
Theorem 3 On computing the Fourier coefficients we found that;
• Odd* odd =even
• Even* even=even
• Odd* even =odd
• Even* odd =odd
L L for m = n
m πx n πx
1- ∫ cos cos dx =
− L 144L42 44 4L 3 0 for m ≠ n
even
L
m πx n πx
2- ∫ cos sin dx {= 0 for all m and n
L
− L 144424443 L
odd
L L for m = n
mπx n πx
3- ∫ sin * sin dx
− L 1444
L
424444 L 3 0 for m ≠ n
even
The above relations known as orthognality relations
240 Fourier Series
1 L 1 2L
2L −∫L 2L 0∫
∴ a0 = f ( x) dx, or a0 = f ( x) dx
(3)
1 L nπx 1 2L nπx
∴ an = ∫ f ( x) cos dx, or an = ∫ f ( x) cos dx
L −L L L0 L
So that Fourier coefficients can be obtained as shown in (2) and (3).
f(x)
-L L
Fig.1
Solution:-
1 L k L k
a0 = ∫
2L − L
f ( x) dx = ∫
2L 0
dx =
2
1 L nπx k L nπx
an = ∫ f ( x) cos dx = ∫ cos dx = 0
L −L L L0 L
242 Fourier Series
1 L nπx k L nπx
bn = ∫ f ( x) sin dx = ∫ sin dx
L −L L L0 L
0 , n even
kL
= (1 − cos nπ )
Lπn 2 k , n odd
πn
k 2k ∞
1 nπx
f ( x) = +
2 Lπ
∑
n =1, 3, 5,....... n
sin
L
k 2 k πx 1 3πx 1 5πx
f ( x) = + sin + sin + sin + ...........
2 π L 3 L 5 L
Fig.2 shows that the partial sum of Fourier series do approach f (x)
as n increases.
1 1
0.8 0.8
0.6
terms=2 0.6
terms=3
0.4 0.4
0.2 0.2
0 0
-0.2 -0.2
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
1
1
0.8
terms=4
0.8
0.6
0.6
terms=10
0.4
0.4
0.2
0.2
0
0
-0.2
-3 -2 -1 0 1 2 3 -0.2
-3 -2 -1 0 1 2 3
0.8
0.6 terms=10
0.4
0.2
-0.2
-10 -8 -6 -4 -2 0 2 4 6 8 10
Fig.2 continue
Theorem 5 If f ( x ) is an odd function of x , i.e. f ( x ) = − f ( − x )
∴ an = 0 , n = 0,1,2,3,4,....... (4)
2L nπx
and bn = ∫ f ( x) sin dx (5)
L0 L
∴ bn = 0, n = 1, 2, 3, 4,....... (6)
2L nπx 1L
and an = ∫ f ( x) cos dx, a0 = ∫ f ( x) dx (7)
L0 L L0
244 Fourier Series
L
x
-L L
Fig.3
Solution:- Since f (x) is an odd periodic function, its Fourier
coefficients are: an = 0, n = 0,1,2,............................
2L nπx 2L
bn = ∫ f ( x) sin dx = (−1) n +1, n = 1,2,3,.............
L0 L nπ
2L ∞(−1) n +1 nπ x
∴ f ( x) = ∑
π n =1 n
sin
L
−2π 2π
Fig.4
Chapter Seven 245
Solution:- This function is not odd or even. So, the Fourier
coefficients can be determined as follows,
1 L 1 2π 2 4π 2
2 L −∫L 2π 0∫
∴ a0 = f ( x) dx = x dx =
3
1 2L nπx 1 2π 2 nπx
∴ an = ∫ f ( x) cos dx = ∫ x cos dx
L 0 L π 0 π
2π
1 sin nx − cos nx − sin nx 4
∴ an = x 2 − 2x + 2 =
π n n2 n3 0 n2
1 L nπx 1 2π 2 nπx
∴ bn = ∫ f ( x) sin dx = ∫ x sin dx
L −L L π 0 π
2π
1 − cos nx − sin nx cos nx − 4π
∴ bn = x 2 − 2x + 2 =
π n n2 n3 0 n
2 4π 2 ∞ 4 4π
∴ f ( x) = x = + ∑ 2 cos nx − sin nx
3 n =1 n n
x
-π 2π
π
Fig.5
246 Fourier Series
2π 1π
=
π ∫ sin x cos (n x ) dx = π ∫ sin ( x + nx) + cos (x − n x ) dx
0 0
π
1 cos(n + 1) x cos(n − 1) x
= − +
π n +1 n −1 0
π
1 cos(n + 1) x cos(n − 1) x
= − +
π n +1 n −1 0
2 2 ∞ (1 + cos(nπ ))
∴ f ( x) = ∑−
π π n=2 (n 2 − 1) cos nx
f(x)
2
Fig.6
-2 2
bn = 0 . Also,
1L 12
a0 = ∫ f ( x) dx = ∫ x dx = 1
L0 20
2L nπx 22 nπx
an = ∫ f ( x) cos dx = ∫ x cos dx
L0 L 20 2
2
2 n πx − 4 n πx
= x sin − 2 2 cos
nπ 2 n π L 0
4
= 2 2
(cos nπ − 1)
n π
4 nπx
f ( x) = 1 + ∑ (cos nπ − 1) cos
n 2π 2 2
8 πx 1 3πx 1 5πx
f ( x) = 1 − cos + cos + cos + ......
π2 2 32 2 52 2
248 Fourier Series
J s = Jumps for f at xs
J s′ = Jumps for f ′ at xs (s = 1,2,3,........m) (9)
J s′′ = Jumps for f ′′ at xs
So that in case of function f (x) having period 2 L , the Fourier
transform can be obtained from the following function:
1 m n πx s L m nπ xs
an = − ∑ J s sin − ∑ J s′ cos +
nπ s =1 L nπ s =1 L
(10)
L
2 m
nπ x s L
3 m
nπ x s
∑ s J ′′ sin + ∑ s J ′′′ cos − − + + ...
nπ s =1 L nπ s =1 L
1 m nπ x s L m nπ x s
bn = ∑ J s cos − ∑ J s′ sin −
nπ s =1 L nπ s =1 L
(11)
L
2 m
nπ x s L
3 m
nπ x s
∑ s J ′
′ cos + ∑ s J ′
′′ sin − − + +.....
nπ s =1 L nπ s =1 L
Chapter Seven 249
For n = 1, 2, 3, 4,...., ∞ and n = 0 can be obtained from equation (3)
as usual.
In case of the function f (x) is periodical function having period
2π , the Fourier transform can be obtained from the following
formula:
1 m 1 m
an = − ∑ J s sin n xs − ∑ J s′ cos n xs +
nπ n s =1
s =1
(12)
1 m 1 m
∑ J s′′ sin n xs + ∑ J s′′′ cos n xs − − + +.....
n2 s =1 n3 s =1
1 m 1 m
bn = ∑ J s cos n xs − ∑ J s′ sin n xs −
nπ n s =1
s =1
(13)
1 m 1 m
2 ∑ s 3 ∑ s
J ′′ cos n x + J ′′′ sin n x + − − + + .....
n s =1
s
n s =1
s
For n = 1, 2, 3, 4,...., ∞ and n=0 can be obtained from equation (3).
f(x)
π
J2
J1
x
-π π
Fig.7
250 Fourier Series
Solution:-
Jumps at x1 = 0 Jumps at x2 = π
f J1 = −π J 2 = −π
1π 1π π
π 0∫
a0 = f ( x) dx = ∫ x dx =
π0 2
1 m
an = − ∑ J s sin n x s = 1 (− J1 sin n x1 − J 2 sin n x2 )
nπ nπ
s =1
1
an = (sin n 0 + sin n π ) = 0 for all n.
n
bn can be obtained from equation (13) and the above table as
follows:
1 m 1
bn = ∑ J s cos n x s =
(J cos n x1 + J 2 cos n x2 )
nπ 1
nπ s =1
1
∴ bn = (cos n 0 + cos n π ) = 2 , n = 2, 4, 6, 8,.....
n n
Otherwise bn = 0 .
1 1 1
∴ f ( x) = 2 sin 2 x + sin 4 x + sin 6 x + ......
2 4 6
Chapter Seven 251
Example 7 Find Fourier transform without integration for the even
waveform f ( x) = x, 0 ≤ x < π as shown in Fig.8.
Solution:-
Jumps at x1 = 0 Jumps at x2 = π
f J1 = 0 J2 = 0
f′ J1′ = 2 J 2′ = −2
1π 1π π
π 0∫
a0 = f ( x) dx = ∫ x dx =
π0 2
As we see it is even function then bn = 0 , for n = 1, 2, 3
an can be obtained from equation (12) as follows :
f(x)
π
-π π
f(x)
π
-π
-1
Fig.8
252 Fourier Series
1 m m
an = − ∑ J s sin n x s − 1 ∑ J s′ cos n x s
nπ n s =1
s =1
1 1
an = − J1 sin n x1 − J 2 sin n x2 − ( J1′ cos n x1 + J 2′ cos n x2 )
nπ n
Substitute in this equation from the above table we get:
1 1 −4
an = − (2 cos n 0 − 2 cos n π ) = 2 , n = 1, 3,5,7,...
nπ n n π
π 4 1 1 1
f ( x) = − cos x + cos 3x + cos 5 x + cos 7 x + ......
2 π 9 25 49
Example 8 Find Fourier transform without integration for the odd
waveform f ( x) = x, 0 ≤ x < π as shown in Fig.9.
Solution:-
Jumps at x1 = 0 Jumps at x2 = π
f J1 = 0 J 2 = −2π
f′ J1′ = 0 J 2′ = 0
1 m 1 n +1 2
bn = ∑ J s cos n xs =
nπ
(− 2π cos n π ) = (− 1)
nπ s =1 n
1 1 1
f ( x) = 2 sin x − sin 2 x + sin 3x − sin 4 x + ......
2 3 4
Chapter Seven 253
Example 9 Find Fourier transform of the waveform in Fig.10:
f(x)
Fig.10
Io
7π 11π
π 6 6 2π
π 5π
6 6
− Io
2 3 Io 1 1 1 1
∴ f ( x) = sin x − sin 5 x − sin 7 x + sin 11x + sin 13 x − − + +....
π 5 7 11 13
254 Fourier Series
Fig.11
−π −π / 2 π /2 π 3π / 2 2π
x π /6 5π / 6 7π / 6 11π / 6
f (x ) J1 = I o J 2 = −Io J3 = −Io J 4 = Io
ao = a n = 0
1 m
bn = ∑ J s cos nxs
nπ s =1
Io nπ 5nπ 7 nπ 11nπ
= cos − cos − cos + cos ....
nπ 6 6 6 6
2 3
∴ bn = I o , n = 1, 11, 13, 23, 25,
nπ
2 3
∴ bn = − I o , n = 5, 7, 17, 19,....
nπ
∴ bn = 0 n = else where(2, 3, 4, 6, 8, ....
2 3I o 1 1 1 1
∴ f (x ) = sin x − sin 5 x − sin 7 x + sin 11x + sin 13x....
π 5 7 11 13
Chapter Seven 255
7.4 Complex Fourier Series
In this section we will discuss other form of Fourier series, which is
complex Fourier series.
As we know, the Fourier coefficients are:
1 L
2 L −∫L
a0 = f ( x) dx
1 L nπx
an = ∫ f ( x) cos dx. (14)
L −L L
1 L nπx
bn = ∫ f ( x) sin dx
L −L L
If we change n to -n in the above definition for an and bn , we will
The form (18) is equivalent to normal Fourier series form but this
form is Sometimes preferred it is coefficients are easier to
remember.
1
cn = (an + jbn )
2
-L L
Fig.12
Chapter Seven 257
1 L k L k
2 L −∫L 2 L 0∫
Solution:- a0 = f ( x ) dx = dx =
2
nπ x L
nπ x nπ x j
1 L j kL j ke L
2 L −∫L L 0∫
cn = f ( x) e L dx = e L dx =
L j nπ
L x =0
k
cn = j
k
2nπ
(
1− e jnπ
) =j
2nπ
1 − (cos nπ + j sin
1 2 n3π )
=0, for all n
2k
cn = j , Otherwise cn = 0
nπ
Substitute in equation (18) we get:
nπ x
∞ − j
L
∴ f ( x ) = ∑ cn e
n = −∞
πx 3π x 5π x
2k − j L 1 − j L 1 − j L
= j e + e + e +
π 3 5
nπ x
∞
L = j 2k cos π x − j sin π x
−j
∴ f ( x ) = ∑ cn e
n = −∞
π L L
1 3π x 3π x 1 5π x 5π x
+ cos − j sin + cos − j sin + ...
3 L L 5 L L
k 2 k πx 1 3πx 1 5πx
∴ f ( x) = + sin + sin + sin + +.......
2 π L 3 L 5 L
Compare the above result with the result of Example 1 in this
chapter.
258 Fourier Series
Problems
Find Fourier series of the function f(x) which is assumed to have
the period of 2 by two different techniques
− 1 if −π / 2 < x < 0
١) f ( x) = 1 if 0< x <π /2
0 if π /2< x <π
0 if −π < x < 0
٢) f ( x) = − 1 if 0< x <π /2
1 if π /2< x <π
٣) f ( x) = x −π < x < π
٤) f ( x) = x 2 −π < x < π
٥) f ( x) = x 3 −π < x < π
٦) f ( x) = x −π < x < π
π + x if −π / 2 < x < 0
٧) f ( x) =
π −x if 0< x <π
x if −π / 2 < x < π / 2
٨) f ( x) =
π − x if π / 2 < x < 3π / 2
x 2 if −π / 2 < x < π / 2
٩) f ( x) = 2
π / 4 if π / 2 < x < 3π / 2
sin x if 0 < x <π
١٠) f ( x) =
0 if π < x < 2π
١١) f ( x) = sin x −π < x < π
cos x if 0< x <π
١٢) f ( x) =
0 if π < x < 2π
١٣) f ( x) = cos x −π < x < π
Chapter 8
Least Square Technique
8.1 Introduction
Suppose you are in a science class and you receive these
instructions:
Find the temperature of the water at the times 1, 2, 3, 4, and 5
seconds after you have applied heat to the container. Conduct your
experiment carefully. Graph each data point with time on the x-axis
and temperature on the y-axis. Your data should follow a straight
line. Find the equation of this line. The data from the experiment
looks like this when charted and graphed:
260 Least Square Technique
Notice that our data points don't fall exactly on a straight line as
they were supposed to, so how are we going to find the slope and
intercept of the line?
This is a common problem with experimental sciences because
the data points that we measure seldom fall on a straight line.
Therefore, scientists try to find an approximation. In this case, they
would try to find the line that best fits the data in some sense. The
first problem is to define "best fit." It is convenient to define an error
as a distance from the actual value of y for x (the value that was
measured in the experiment) to the predicted value of y for x.
Therefore, it seems reasonable that the "best fit" line would
somehow minimize the errors, but how? You could minimize the
sum of the absolute values of the errors; this is called the l1 fit. It
would also be reasonable to find the biggest error for each line and
choose the line that minimizes this quantity; this is called the l∞ fit.
However, the fit that is used most often is the l2 fit which is called
the least squares fit. This method is called the least squares fit
because it finds the line that minimizes the sum of the squares of the
errors. Gauss developed this method to solve a problem when he
was a young man (15 years old!!) to help his friend solve a
chemistry problem. This is the fit that is most often used because it
is the only one that can be found by solving a system of linear
equations.
Chapter Eight 261
8.2 Least Squares Fit (Second Technique)
You have just read a lot of new information, so let's illustrate the
concepts with our example. We have the graph of the data above.
Now we need to guess which line best fits our data. If we assume
that the first two points are correct and choose the line that goes
through them, we get the line y = 1 + x . If we substitute our points
into this equation, we get the following chart. The points and line
are graphed below.
Therefore, the sum of the squares of the errors is 27. Do you think
that we can do better than this?
If we choose the line that goes through the points when x = 3 and
4, we get the line y = 4 + x . Will we get a better fit? Let's look at it.
262 Least Square Technique
The sum of the squares of the error is 18. That is a better fit, but can
we do even better?
Let's try the line that is half way between these two lines. The
equation would be y = 2.5 + x . It looks like this:
Chapter Eight 263
The sum of the squares of the error is 11.25 with this line, so this is
the best line yet. Can we do better? It doesn't seem very scientific or
efficient to keep guessing at which line would give the best fit.
Surely there is a methodical way to determine the best fit line. Let's
think about what we want.
A line in slope-intercept form looks like c0 + c1 x = y where c0 is
c0 + 1c1 = 2
c0 + 2c1 = 3
c0 + 3c1 = 7
c0 + 4c1 = 8
c0 + 5c1 = 9
We know that there may not exist c0 and c1 that fit all these
equations, so we try to find the best fit. We can write these
equations in the form X c = y (these are just new letters for our
familiar equation Ax = b ) where
1 1 2
1 2 3
c0
X = 1 3 , c = , and, y = 7
c1
1 4 8
1 5 9
264 Least Square Technique
1 1 2
1 2 3
1 1 1 1 1 c0 1 1 1 1 1
1 2 3 4 5 1 3 c = 1 2 3 4 5 7
1
1 4 8
1 5 9
5 15 c0 29
15 55 c = 106
1
Using Cramer s rule or Gauss elimination can easily solve the above
equation. ∴ c0 = 0.1 , and, c1 = 1.9
When we graph and chart the line y = 0.1 + 1.9 x , we get:
We can find the curve that best fits our data in a similar manner. The
X T Xc = X T y
1 −1 1
1 1 1 1 1 1 0 0 c0
− 1 0 1 2
3 1 1 1 c1
1 0 1 4 9 1 2 4 c2
1 3 9
5 5 15 c0 7
5 15 35 c1 = 7
15 35
99 c2 33
268 Least Square Technique
X T X c = X T y with
1 x1 x12 y1
y
1 x2 x22 c0
2
X = M M M c = c1 y= M
1 xn −1 xn2−1 c2 y
n −1
1 xn xn2 yn
Notice that the normal equations used to find the best fit line and
the best fit parabola have the same form. Do you think that we could
270 Least Square Technique
This strategy has many advantages including the fact that it gives a
unique line for any given data.
∂S r n
∴ = −2 ∑ ( yi − a0 − a1 xi ) = 0 (5)
∂a0 i =1
∂S r n
∴ = −2 ∑ [( yi − a0 − a1 xi )xi ] = 0 (6)
∂a1 i =1
♠
Steven C. Chapman and Raymond P. Canale Numerical Method of
Engineering , book, McGraw-Hill international edition, third edition,1998
Chapter Eight 273
By simplifying the above equations we can obtain the following:
n n n
∑ yi − ∑ a0 − ∑ a1xi = 0 (7)
i =1 i =1 i =1
n n n
∑ yi xi − ∑ a0 xi − ∑ a1xi2 = 0 (8)
i =1 i =1 i =1
The above two equations (7) and (8) can be further simplified to be
as following:
n n
na0 + ∑ xi a1 = ∑ yi
(9)
i =1 i =1
n n n
∑ xi a0 + ∑ xi2 a1 = ∑ xi yi (10)
i =1 i =1 i =1
n n n
n ∑ xi yi − ∑ xi * ∑ yi
a1 = i =1 i =1 i =1 (11)
2
n n
n ∑ xi2 − ∑ xi
i =1 i =1
a0 = y − a1 x (12)
y = a0 + a1x + a2 x 2 + e (13)
For this case the sum of squares of the residuals is:
274 Least Square Technique
( )
∂S r n
= −2 ∑ yi − a0 − a1 xi − a2 xi 2 = 0 (15)
∂a0 i =1
∂S r
∂a1
n
[( )]
= −2 ∑ yi − a0 − a0 xi − a2 xi 2 xi = 0 (16)
i =1
∂S r
∂a2
n
[( ) ]
= −2 ∑ yi − a0 − a0 xi − a2 xi 2 xi2 = 0 (17)
i =1
( )
n n n
na0 + ∑ ( xi ) a1 2
+ ∑ x a2 = ∑ yi
i =1 i =1 i =1
Example 1
For the following data determine a second order equation fits this
data
xi 1 2 3 4 5 6 7
yi 2 3 5 6 8 7 9
Solution:-
xi yi xi * yi
xi2 xi3 xi4 xi2 yi
1 2 2 1 1 1 2
2 3 6 4 8 16 12
3 5 15 9 27 81 45
4 6 24 16 64 256 96
5 8 40 25 125 625 200
6 7 42 36 216 1296 252
7 9 63 49 343 2401 441
∑ 28 40 192 140 784 4676 1048
Then,
276 Least Square Technique
7 a0 28a1 140a2 = 40
28a0 140a1 140a2 = 192
140a0 784a1 467a2 = 1048
From the above equation we can get:
2 n
b j = ∑ yi * sin( jω t ) (27)
n i =1
278 Least Square Technique
2 L 0∫ .02 0.∫02
a0 = f ( x ) dx = dt − ∫ dt = 0
0.12
2L πt 200 0.08 πt
a1 = ∫ f (t ) cos dt = ∫ cos dt = 0
L0 L .01 0.02 0.01
Chapter Eight 279
2L πx 200 0.08 π t
b`1 = ∫ f ( x) sin dx =
.01 0.∫02 0.01
sin dt = 127.324
L0 L
The following computer program can be used to get m terms of
Fourier series, where we divided the number of points to n=200
points then m = (200 − 1) / 2 = 99 . Then we can get the following
coefficients.
xi y i
y i cos x i y i sin x i
1 0 0 0 0
2 0.001 0 0 0
3 0.002 100 80.90161308 58.7786441
4 0.003 100 58.77834693 80.90182898
5 0.004 100 30.90141996 95.10574244
6 0.005 100 -0.000367321 100
7 0.006 100 -30.90211865 95.10551542
8 0.007 100 -58.77894126 80.90139717
9 0.008 0 0 0
10 0.009 0 0 0
11 0.01 0 0 0
12 0.011 0 0 0
13 0.012 -100 80.90118126 58.77923843
14 0.013 -100 58.77775259 80.90226079
15 0.014 -100 30.90072128 95.10596945
16 0.015 -100 -0.001101962 99.99999999
17 0.016 -100 -30.90281733 95.1052884
18 0.017 -100 -58.7795356 80.90096535
19 0.018 0 0 0
20 0.019 0 0 0
sum 0.19 0 161.796153 1021.586851
280 Least Square Technique
For I=1 to N
Is Abs(A(J)) > 0.1
SUMY=SUMY+Y(I) Print A(J)
For j=1 to m
1 End
∑
10.50 0.50 -3.16
284 Least Square Technique
n yi 10.5
∴ a0 = ∑ n
=
20
= 0.525
i =1
2 n 2
∴ a1 = ∑
n i =1
yi * cos(ω t ) =
20
* 0.5 = 0.05
2 n 2
∴ b1 = ∑
n i =1
yi * sin(ω t ) =
20
* (−3.16) = −0.316
If we use the exact Fourier transform will get the following results
1 2L
a0 =
2L
∫ f ( x) dx
0
0.001
1 0.001 1 1000t 2
.001 0∫
= 1000t dt = = 0.5
.001 2 0
1 2L πt 1 0.001 πt
a1 = ∫ f (t ) cos dt =
0.05 0∫
1000 t * cos dt = 0
L 0 L 0.0005
1 2L πt
b1 = ∫ f (t ) sin dt =
L 0 L
1 0.001 πt
= ∫
0.05 0
1000 t * sin dt = −0.16
0.0005
Problems
Solve problems of Chapter 7 numerically and compare the results.
Chapter 9
Power Series Solution Of Differential Equations
9.1 Introduction
In chapter five we consider solving the linear differential equation
with constant coefficients. Caushy equation was the only equation in
that chapter has non-constant coefficients and has been solved by
special manner. In this chapter, we shall consider solving
differential equations by so-called power series method, which
yields solution in the form of power series. This is a very efficient
standard procedure in connection with linear differential equations
whose variable coefficients.
We start by showing some examples of power series of famous
1 ∞
functions: = ∑ x m = 1 + x + x 2 + x 3 + x 4 + ......
1 − x m=0
∞
xm x 2 x3 x 4
x
e = ∑ =1+ x + + + + ......
m=0 m! 2 ! 3! 4!
∞ (−1) m x 2m x2 x4
cos x = ∑ (2m)! = 1 −
2!
+
4!
− +.........
m=0
(−1) m x 2m +1
∞ x3 x5
sin x = ∑ = x− + − +.........
m=0 ( 2 m + 1)! 3! 5!
See Appendix 1 for more functions.
286 Power Series Solution Of Differential Equations
The power series (in power of ( x − a ) ) is an infinite series of the
following form:
∞
∑ cm ( x − a) m = c0 + c1 ( x − a) + c2 ( x − a) 2 + c3 ( x − a)3 + ....... (1)
m=0
d2y dy
2
+ P( x) + Q( x) y = 0 (3)
dx dx
Where P ( x) and Q ( x) are functions in x only.
We will look for solution of (3) in power series of x (or in powers
of x − a ), if a = 0 , then the solution has the following form:
∞
y= ∑ cm x m = c0 + c1x + c2 x 2 + c3 x 3 + ....... (4)
m =0
∞
∴ y′ = ∑ mcm x m −1 = c1 + 2c2 x + 3c3 x 2 + 4c4 x 3 ....... (5)
m=0
Chapter Nine 287
∞
∴ y′′ = ∑ m(m −1)cm xm−2 = 2c2 + 3* 2 c3 x + 4 * 3c4 x2 + 5 * 4c5 x3. + ..(6)
m=0
Insert (4), (5) and (6) into (3) and collecting like powers of x, we
may write the relating equation in the following form:
k0 + k1 x + k 2 x 2 + k3 x 3 + ....... = 0 (7)
Substitute the values of c0 , c1 , c2 ,...... into (9) we get the final form
for the solution of the differential equation as following:
∞
m x 2 x 4 x 6 x8
∴ y= ∑ cm x = c0 1 − + − +
2! 4! 6! 8!
− +........
m=0
x3 x5 x 7 x9
+ c1 x − + − + − +........
3! 5! 7! 9!
∴ y = c0 cos x + c1 sin x
Coefficients of x 0 = 0 , ∴ c2 = 0 ,
2c
Coefficients of x1 = 0 , ∴ 3 * 2 c3 − 2c0 = 0 , ∴ c3 = 0 ,
3* 2
2c
Coefficients of x 2 = 0 , ∴ 4 * 3 c4 − 2c1 = 0 , ∴ c4 = 1 ,
4*3
3 2c2
Coefficients of x = 0 , 5 * 4 c5 − 2c2 = 0 , c5 = = 0,
5*4
Coefficients of x 4 = 0 , ∴ 6 * 5 c6 − 2c3 = 0 ,
2c3 2 * 2c0
∴ c6 = = ,
6 *5 6 *5*3* 2
Coefficients of x 5 = 0 , ∴ 7 * 6 c7 − 2c4 = 0 ,
2c4 2 * 2c1
∴ c7 = = ,
7*6 7*6*4*3
2c
Coefficients of x 6 = 0 , ∴ 8 * 7 c8 − 2c5 = 0 , ∴ c8 = 5 = 0 ,
8*7
Coefficients of x 7 = 0 , ∴ 9 * 8 c9 − 2c6 = 0 ,
290 Power Series Solution Of Differential Equations
2 2 2*2 23
∴ c9 = c6 = * c0 = c0 ,
9*8 9*8 6*5*3* 2 9 *8* 6 *5*3* 2
Coefficients of x 8 = 0 , ∴ 10 * 9 c10 − 2c7 = 0 ,
2 2 2*2 23
∴ c10 = c7 = * c1 = c1 ,
10 * 9 10 * 9 7 * 6 * 5 * 3 10 * 9 * 7 * 6 * 5 * 3
Coefficients of x 9 = 0 ,
2c8
∴ 11 * 10 c11 − 2c8 = 0 , and ∴ c11 = = 0,
11 * 10
Substitute the values of c0 , c1 , c2 ,...... into (4) we get the final
solution as following:
2 3 22 6 23
∴ y = c0 1 + x + x + x 9 + ...
3 * 2 6*5*3* 2 9 *8* 6 *5*3* 2
2 4 22 7 23
+ c1 x + x + x + x10 + ..
4*3 7*6*4*3 10 * 9 * 7 * 6 * 5 * 3
m−2
∞ x 2 x 4 x 6 x8 ∞
∑ m ( m − 1) c m x + 1 − + −
2! 4! 6! 8!
+ − +..... ∑ cm x m = 0
m=0 m=0
c
Coefficients of x 0 = 0 , ∴ 2c2 + c0 = 0, ∴ c2 = − 0 ,
2
c
Coefficients of x1 = 0 , ∴ 6 c3 + c1 = 0 , ∴ c3 = − 1 ,
6
c c
Coefficients of x 2 = 0 , ∴ 12c4 + c2 − 0 = 0 , ∴ c4 = 0 ,
2 4*3
c 2c1
Coefficients of x 3 = 0 , ∴ 5 * 4 c 5 + c 3 − 1 = 0 ∴ c5 = ,
2 5*4*3
Substitute the values of c0 , c1 , c2 ,...... into the assumed solution.
∞
m x2 1 4 1 6
∴ y= ∑ cm x = c0 1 − +
2 4*3
x −
6*5
x + −.......
m=0
x3 1
+ c1 x − + x 5 − +.......
3* 2 5*3* 2
∴ y ( x) = c0 y1 ( x) + c1 y 2 ( x)
cn −1
∴ cn + 2 =
(n + 2)(n + 1) n = 2,3,4,....
c1 c2 c c0
∴ c4 = , ∴ c5 = = 0 , ∴ c6 = 3 = ,
4*3 5*4 6 * 5 6 * 5 * 3!
c4 c1 c1 c5
∴ c7 = = = , ∴ c8 = = 0,
7 * 6 7 * 6 * 4 * 3 7 * 3 * 4! 8*7
c6 co co
∴ c9 = = = ,
9 * 8 9 * 8 * 6 * 5 * 3! 9 * 2 * 6!
c7 c1 c1
∴ c10 = = = ,
10 * 9 10 * 9 * 7 * 3 * 4! 9 * 7 * 6!
Chapter Nine 293
c8 c9 co
∴ c11 = = 0 , ∴ c12 = =
11 * 10 12 * 11 12 * 11 * 9 * 2 * 6!
x3 x6 x9 x12
∴ y ( x ) = co 1 + + + + + ......
3! 180 12960 12 * 11 * 9 * 2 * 6!
x4 x7 x10
+ c1 x + + + + ........
12 7 * 3 * 4! 9 * 7 * 6!
(n + 1)cn cn
∴ cn + 2 = ∴ cn + 2 =
(n + 2)(n + 1) n = 2,3,4,.... (n + 2) n = 2,3,4,....
c2 c c3 c c c0
∴ c4 = = o , ∴ c5 = = 1 , ∴ c6 = 4 =
4 4*2 5 5*3 6 6*4*2
c5 c1 c1 c6 co
∴ c7 = = = , ∴ c8 = =
7 7 * 5 * 3 7 * 3 * 4! 8 8*6*4*2
c7 c1 c8 co
∴ c9 = = , ∴ c10 = =
9 9*7*5*3 10 10 * 8 * 5 * 4 * 2
x2 x4 x6 x8 x10
∴ y ( x ) = co 1 + + 2 + 3 + 4 + 5 + ......
2! 2 * 2! 2 * 3! 2 * 4! 2 * 5!
x3 x5 x7 x9
+ c1 x + + + + ........
3 5*3 7*5*3 9*7*5*3
c
∴ 2c2 + c0 = 0 , ∴ c2 = − 0 ∴ 3 * 2 * c3 − 2c2 + c1 = 0 ,
2!
co c1
∴ c3 = − −
6 6
∞
∴ ∑ [(n + 2)(n + 1)cn + 2 − n(n + 1)cn +1 + cn ]x n = 0
n=2
∴ cn + 2 =
(n + 1) c − cn
(n + 2)(n + 1) n +1 (n + 2)(n + 1) n = 2,3,4,....
n cn
∴ cn + 2 = cn +1 −
(n + 2) (n + 2)(n + 1) n = 2,3,4,....
2c3 c c c c c c
∴ c4 = − 2 =− o − 1 + o =− o − 1,
4 4*3 12 12 24 24 12
3c4 3c3 − 3c0 3c1 co c1
∴ c5 = − = − + + ,
5 5 * 4 5 * 24 5 * 12 6 * 5 * 4 6 * 5 * 4
co c
∴ c5 = − − 1,
5 * 4 * 3 24
296 Power Series Solution Of Differential Equations
4c5 c − 2c 0 2c1 co c1
c6 = − 4 = − + + ,
6 6 * 5 3 * 5 * 4 * 3 3 * 24 6 * 5 * 24 6 * 5 * 12
− 7c0 11c1
∴ c6 = −
720 420
x 2 x3 x 4 x5 7 x 7
∴ y ( x ) = co 1 − − − − − + ......
2 6 24 60 720
x 3 x 4 x 5 11x 6
+ c1 x − − − − ........
6 12 24 420
∑ (m 2 − 2m + 4)c m x m
∞
=0
1=4
m 2 44
424444
3
assume n = m
Chapter Nine 297
4c 2 4c c
∴ c4 = − = o = o,
2*4*3 4! 3!
∴ c5 =
− 7c3
=
7c1
, ∴ c6 = −
(4 * 2 + 4)c4 =−
co
,
2 * 5 * 4 10 * 16 2*6*5 30
∴ c7 = −
(5 * 3 + 4)c5 =−
19c1
2*7*6 1920
x4 x6
∴ y ( x ) = co 1 − x +
2
− + ......
3! 30
x 3 7 x 5 19 x 7
+ c1 x − + − ........
4 160 1920
h
(Charles Hermit (1822-1901). A French mathematician, held the chair of
higher algebra at the university of Paris from 1876 to 1901. This equation is
important in many branches in mathematical physics; for example, in quantum
mechanics.
298 Power Series Solution Of Differential Equations
Then y ′ and y ′′ can be obtained as in (5) and (6) respectively.
Substitute these power series in the differential equation we get:
∞ ∞ ∞
∴ ∑ m(m − 1)cm x m − 2 − 2 x ∑ mcm x m −1 +λ ∑ cm x m = 0
m=0 m=0 m=0
∞
∴ (2c2 + λc0 ) + (3 * 2c3 − 2c1 + λc1 )x + ∑ m(m − 1)cm x m − 2
1=4
m 4 4424443
assume m = n + 2
∞ ∞
m −1
− 2 x ∑ mcm x +λ ∑ cm x m = 0
14 m =4
2 44 = 244
4244m4 3
assume m = n
∞
∴ (2c2 + λc0 ) + (3 * 2c3 − 2c1 + λc1 )x + ∑ (n + 2)(n + 1)cn + 2 x n
n=2
∞ ∞
n
− 2 ∑ ncn x +λ ∑ cn x n = 0
n=2 n=2
λc
Coefficients of x 0 = 0 , then, 2c2 + λc0 = 0, ∴ c2 = − 0
2
Coefficients of x1 = 0 , ∴ 3 * 2 c3 − 2c1 + λc1 = 0 ,
(2 − λ )
∴ c3 = c1 ,
3* 2
Coefficients of x n = 0 , ∴ (n + 2 )(n + 1)cn + 2 − 2ncn + λcn = 0 ,
Chapter Nine 299
2n − λ
∴ cn + 2 = c
(n + 2)(n + 1) n
Substitute the values of c0 , c1 , c2 ,...... into (4) we get:
∞ λ (4 − λ )(−λ ) 4 (8 − λ )(4 − λ ) 6
∴ y= ∑ cm x m = c0 1 − 2! x 2 + 4!
x +
6!
x + .... +
m=0
(2 − λ ) 3 (6 − λ )(2 − λ ) 5 (10 − λ )(6 − λ )(2 − λ ) 7
c1 x + x + x + x + ..
3! 5! 7!
∴ y ( x) = c0 y1 ( x) + c1 y 2 ( x)
Coefficient of x o = 0 ∴ c2 = c1 / 9 * 2
c1 2c 2 5c
Coefficient of x1 = 0 ∴ c3 = + = 51
9 *3* 2 9 *3* 2 3
∞ ∞
∑ [n(n − 2)cn + 9(n + 2)(n + 1)cn + 2 − (n + 1)cn +1 ]x n
+ ∑ cn − 4 x n = 0
n=2 n=4
∴ x [0co + 9 * 4 * 3c4 − 3c3 ] + x [3 * 1c3 + 9 * 5 * 4c5 − 4c4 ] +
2 3
∞ ∞
∑ [n(n − 2)cn + 9(n + 2)(n + 1)cn + 2 − (n + 1)cn +1 ]x n
+ ∑ cn − 4 x n = 0
n=4 n=4
∞
∑ [n(n − 2)cn + 9(n + 2)(n + 1)cn + 2 − (n + 1)cn +1 + cn − 4 ]x n = 0
n=4
3c3 5
Coefficient of x 2 = 0 ∴ c4 = = 2 7 c1
(9 * 4 * 3) 2 * 3
4c − 3c3 13
Coefficient of x 3 = 0 ∴ c5 = 4 =− c
(9 * 5 * 4) 2*3 9 1
Coefficient of x n = 0
∴ cn + 2 =
[(n + 1)cn +1 − n(n − 2)cn − cn − 4 ]
9(n + 2)(n + 1) 4,5,6....
Chapter Nine 301
5c5 − 4 * 2c4 − co 72 1
∴ c6 = = − 2 12 c1 − co
9*6*5 2 *3 2 * 5 * 32
6c − 5 * 3c5 − c1 181 * 487 1
∴ c7 = 6 = − 14 c1 − co
9*7*6 3 *7 2 * 5 * 7 * 34
7 c − 6 * 4c 6 − c 2 23 * 4363 43
∴ c8 = 7 = − 14 4 c1 + 4 6 o
c
9 *8* 7 3 *2 2 *7*3
x 6
x 7
x8
∴ y ( x ) = co 1 − 2
− + + ......
2 * 5 * 3 2 * 5 * 7 * 34 2 4 * 7 * 36
x2 5x3 5x 4 13 x 5 72 x6
+ c1 x + + 5 + 2 7− 9
− 2 12
9 * 2 3 2 * 3 2 * 3 2 *3
181 * 487 x 7 23 * 4363 x 8
− − .......
314 * 7 2 4 * 316
∑ (m 2 − 5m + 6) cm x m = 0
∞
1=4
m 2 44
424444
3
assume n = m
c1
∴ 2c2 + 6co = 0, ∴ c2 = −3co , and c3 = −
3
n=2
∴ cn + 2 = −
(n 2 − 5n + 6 )
cn
(n + 2)(n + 1) n = 2,3, 4,....
∴ c4 = 0 , ∴ c5 = 0 , and, cn + 2 = 0
[ 2
] x3
∴ y ( x ) = co 1 − 3x + c1 x −
3
( x 2 + 1) y ′′ + x y ′ − y = 0
∞
Solution: Assume, y = ∑ cm x m (4)
m=0
Then y ′ and y ′′ can be obtained as in (5) and (6) respectively.
Substitute these power series in the differential equation we get:
∞ ∞ ∞
2 m−2 m −1
( x + 1) ∑ m(m − 1)cm x +x ∑ mcm x − ∑ cm x m = 0
m=0 m=0 m=0
Chapter Nine 303
∞ ∞
∴ ∑ m(m − 1)cm x m
+ ∑ m(m − 1)cm x m − 2
m=0 m=0
∞ ∞
+ ∑ mcm x m − ∑ cm x m = 0
m=0 m=0
∞
∴ (2c2 − c0 ) + (3 * 2c3 + c1 − c1 )x + ∑ m(m − 1)cm x m
1=4
m 2 4
42444
3
assume m = n
∞ ∞ ∞
+ ∑ m(m − 1)cm x m − 2 + ∑ mcm x m − ∑ cm x m = 0
1=4
m 4 442444 3 m 1=4
2 44 424 m =4
2 44
3
assume m = n + 2 assume m = n
∞
∴ (2c2 − c0 ) + 6c3 x + ∑[n(n − 1)cn + (n + 2)(n + 1)cn + 2 + ncn − cn ]x n = 0
n=2
∞
∴ (2c2 − c0 ) + 6c3 x + ∑ [(n + 1)(n − 1)cn + (n + 2)(n + 1)cn + 2 ]x n = 0
n=2
c
Coefficients of x 0 = 0 , ∴ 2c2 − c0 = 0, ∴ c2 = 0 and
2
1
Coefficients of x = 0 , ∴ 6 c3 = 0 , ∴ c3 = 0 ,
1− n
∴ cn + 2 = cn
2+n
Substitute the values of c0 , c1 , c2 ,...... into (4) we get:
∞
∴y= ∑ cm x m = c1x +
m=0
x 2
1 1 * 3 6 1 * 3 * 5 8 1 * 3 * 5 * 7 10
c0 1 + − 2 x4 + 3 x − 4 x + 5
x ....
2! 2 * 2! 2 * 3! 2 * 4! 2 * 5!
∴ y ( x) = c0 y1 ( x) + c1 y 2 ( x)
304 Power Series Solution Of Differential Equations
Example 12 Solve the following differential equation about the
d2y dy
+x + ( x 2 + 4) y = 0
dx 2 dx
d2y dy
+ ex + sin( x) y = 0
dx 2 dx
308 Power Series Solution Of Differential Equations
We have P ( x) = e x and Q( x) = sin( x) which are analytic
functions for all x. Thus all points x are ordinary points of the
differential equation.
Example 15 Consider the following differential equation:
d2y
+ ( Ln( x)) y = 0
dx 2
This differential equation has a singular point at x = 0 . Because
Q ( x) = Ln( x) possesses no power series in x.
Example 16 Consider the following differential equation
d2y dy
( x 2 − 1) + 2x + 6y = 0
dx 2 dx
d2y dy2x 6
∴ + + y=0
dx 2 ( x 2 − 1) dx ( x 2 − 1)
2x 6
We have P ( x) = and Q ( x ) = which are
( x 2 − 1) ( x 2 − 1)
analytic except at x = 1 and x = −1. Thus x = 1 and x = −1 are
singular points of the differential equation, and all other finite value
of x are ordinary points.
We now test P( x ) and Q( x ) at each singular point
For x = −1
2x 2x
(x + 1) P(x ) = (x + 1) * = = 1 Also,
x 2 − 1 x − 1 x = −1
6 6( x + 1)
(x + 1)2 Q( x ) = ( x + 1)2 * 2 = = −1
x −1 x − 1 x = −1
Chapter Nine 309
It is clear that the point x = −1 is regular singular point.
For x = 1
2x 2x
(x − 1) P(x ) = (x − 1) * 2
= = 1 Also,
x −1 x + 1 x =1
( x − 1)2 Q( x ) = ( x − 1)2 * 26 = 6( x − 1) = 1
x −1 x + 1 x =1
It is clear that the point x = 1 is regular singular point.
So, the two points x = 1, and x = −1 are regular singular points
for the differential equation.
2 2 d2y dy
( x − 4) + ( x − 2) +y=0
dx 2 dx
d2y 1 dy 1
∴ 2
+ 2 dx
+ 2 2
y=0
dx ( x − 2)( x + 2) ( x − 2) ( x + 2)
1 1
We have P ( x) = and Q ( x ) =
( x − 2)( x + 2) 2 ( x − 2) 2 ( x + 2) 2
It is clear that x = −2 and x = 2 are singular points. We now
test P (x) and Q (x) at each singular point.
¾ For x = −2
1 1
( x + 2) P ( x ) = and ( x + 2) 2 Q ( x) =
( x − 2)( x + 2) ( x − 2) 2
It is clear that ( x + 2) P ( x) is not analytic at x = −2 and hence
x = −2 is an irregular single point of the differential equation.
310 Power Series Solution Of Differential Equations
¾ For x = 2
1 1
( x − 2) P ( x ) = and ( x − 2) 2 Q( x) = are analytic
( x + 2) 2 ( x + 2) 2
function at x = 2 is a regular singular point of the differential
equation.
Chapter Nine 311
9.4 Solutions Around Singular Points
To solve (3) around the regular singular point we employ the
following theorem:
Theorem 1 If x = a is a regular singular point of equation (3) then
there exists at least one series solution of the form
∞ ∞
y ( x) = ( x − a ) r
∑ cm ( x − a ) m
= ∑ cm ( x − a ) m + r (13)
m=0 m =0
∞
∴ y ′′ = ∑ (m + r )(m + r − 1)cm x m + r − 2 (17)
m =0
∞ ∞
m + r −1
∑ (m + r )(3m + 3r − 3 + 1)cm x − ∑ cm x m + r =0
m=0 m =0
∞
∴ r (3r − 2)c0 x r −1 + x r ∑ (m + r )(3m + 3r − 2)cm x m −1
1=4
m 1 44442444443
n = m −1
∞
− xr ∑ cm x m = 0
m
14= 024
3
n=m
∞
∴ x r r (3r − 2)c0 x −1 + x n ∑ [(n + r + 1)(3n + 3r + 1)cn +1 − cn ] = 0
n =0
Which implies that r (3r − 2)co = 0 , (18)
(3) then the function xP(x) and x 2Q ( x) obtained from (3) are
analytic at x = 0 . That is, the following expressions are valid on
intervals that have a positive radius of convergence.
xP( x) = p0 + p1 x + p2 x 2 + ..........
r1 = 0, r2 = −2 , respectively.
Since (n + r + 1)(n + r + 3)cn +1 − cn = 0 , n = 0, 1, 2, ........ (25)
cn
It follows that, when r1 = 0 , cn +1 =
(n + 1)(n + 3)
c0 c 2c0 c 2c
∴ c1 = , ∴ c2 = 1 = , ∴ c3 = 2 = 0
1* 3 2 * 4 2!*4! 3 * 5 3!*5!
c3 2c
∴ c4 = = 0
4 * 6 4!*6!
2c 0
∴ cn = , n = 1, 2, 3,......
n!*(n + 1)!
Thus one series solution is as following:
∞
0 1
∴ y1 ( x) = c0 x 1 + ∑ xn
n =1n!*(n + 2)!
∞ 1
Or y1 ( x) = c0 ∑ n!*(n + 2)! x n x <∞ (26)
n=0
Chapter Nine 317
Now when r2 = −2 ,we get the following equation:
Case III If r1 = r2
The above three cases are analyzed in details in the following:
Case I If r1 and r2 are distinct and do not differ be an integer, then
there exist two linearly independent solutions of (3) of the form (30)
and (31) and as explained in Example 18 and the following example.
∞
y1 = ∑ cm x m + r1 , c0 ≠ 0, (30)
m =0
∞
y2 = ∑ bm x m + r2 , b0 ≠ 0, (31)
m=0
∞
∴ x r r(2r −1)c0 x −1 + xn ∑[(n + r + 1)(2n + 2r + 1)cn+1 + (k + r +1)cn ] = 0
n=0
1
Which implies that r (2r − 1) = 0 , ∴ r1 = , and r2 = 0
2
(n + r + 1)(2n + 2r + 1)cn +1 + (n + r + 1)cn = 0 n = 0,1, 2, 3,......
1 − cn
For r1 = : ∴ cn +1 = n = 2, 3, .....
2 2(n + 1)
− c0 − c1 c
∴ c1 = , ∴ c2 = = 20 ,
2 *1 2*2 2 *2
− c2 −c
∴ c3 = = 3 0 ,
2 * 3 2 * 3!
320 Power Series Solution Of Differential Equations
(−1) n c0
∴ cn = n = 1,2,3,......
2 n * n!
Thus from (30) and (31) we can write the final solution as:
1/ 2 n
∞ (−1) n ∞ (−1) n
∴ y1 ( x) = c0 x 1 + ∑ n x = c0 ∑ n x n +1 / 2
n = 02 * n! n = 02 * n!
∞
∴ x r r (2r − 1)c0 x −1 + x n ∑[(n + r + 1)(2n + 2r + 1)cn+1 + (n + r − 2)cn ] = 0
n=0
Which implies that r (2r − 1) = 0 ,
322 Power Series Solution Of Differential Equations
∴ (n + r + 1)(2n + 2r + 1)cn +1 + (n + r + 1)cn = 0 n = 0,1, 2, 3,......
1
∴ r1 = , r2 = 0
2
3
− n c n
1
∴ cn +1 =
2
For r1 = :
2 3
n + ( 2 n + 2)
2 n = 2, 3, .....
c c c
At n = 0 ∴ c1 = 0 , At n = 1∴ c2 = 1 = 0
2 20 40
− c2 − c0 − c3 co
At n = 2 ∴ c3 = = , At n = 3 ∴ c4 = =
42 1680 24 40320
− c4 co
At n = 4 ∴ c5 = =−
22 887040
Thus we can write
x x2 x3 x4 x5
∴ y1 ( x) = c0 x1 / 2 1 + + − + − + −.......
2 40 1680 40320 887040
Now for r2 = 0
− (n − 2 )cn (2 − n )cn
∴ cn +1 = = n = 0, 1, 2, 3, .....
(2n + 1)(n + 1) (2n + 1)(n + 1)
At n = 0 c1 = 2co ,
c c
At n = 1 ∴ c2 = 1 = o , At n = 2 c3 = 0
6 3
At n = 2 c4 = 0
Chapter Nine 323
∴ cn = 0, for n = 3,4,5,6,....
x2
∴ y2 ( x) = c0 1 + 2 x +
3
∴ cn +1 =
(3 − n ) c , n = 0, 1, 2, 3, 4,.......
(n + 1)(n − 6) n
3 c c
∴ c1 = co , ∴ c2 = − 1 = 0 ,
1 * (− 6) 5 10
c c
∴ c3 = − 2 = − 0 , ∴ c4 = c5 = c6 = 0 ,
12 120
c7 =
(− 3) c = 0 = Undefined number.
6
7*0 0
Chapter Nine 325
This implies that co and c7 can be chosen arbitrarily.
−4
And for n ≥ 7 , ∴ c8 = c7 ,
8 *1
5 4*5
∴ c9 = − c8 = c7 ,
9*2 2! 8 * 9
6 − 4*5*6
∴ c10 = − c9 = c7 ,
10 * 3 3! 8 * 9 * 10
(−) n +1 4 * 5 * 6......(n − 4)
∴ cn = c7 n = 8, 9, 10, ..... ,
(n − 7)! 8 * 9 * 10...n
If we choose c7 = 0 and c0 ≠ 0 we obtain the polynomial solution:
1 1 1 3
∴ y1 ( x) = c0 1 − x + x 2 − x ,
2 10 120
But when c7 ≠ 0 and c0 = 0 , it follows that a second, through
infinite series, solution is
∞ ( −1) n +1 4 * 5 * 6.....( n − 4)
∴ y2 ( x) = c7 x 7 + ∑ xn
n = 8 ( n − 7)! 8 * 9 * 10.....n
Finally, for x > 0 the general solution of the differential equation is
y ( x) = C1 y1 ( x) + C2 y 2 ( x)
1 1 1 3
y ( x) = C1 1 − x + x 2 − x
2 10 120
7 ∞ (−1) n +1 4 * 5 * 6.....(n − 4) n
+ C2 x + ∑ x
n =8 ( n − 7 )! 8 * 9 * 10 .....n
It is interesting to observe that in the proceeding example the
larger root r1 = 7 was not used.
326 Power Series Solution Of Differential Equations
Example 23 Solve the following differential equation:
x y ′′ − ( 4 + x ) y ′ + 2 y = 0
∞
Solution: If y = ∑ cm x m + r (15)
m =0
Coefficient of x n = 0
∴ (n + r + 1)(n + r − 4)cn +1 − (n + r − 2)cn = 0 n = 0,1, 2, 3,......
Chapter Nine 327
(n + r − 2)
∴ cn +1 = cn n = 0, 1, 2, 3,......
(n + r + 1)(n + r − 4)
( n − 2)
At r2 = 0 ∴ cn +1 = cn n = 0, 1, 2, 3,......
(n + 1)(n − 4)
− 2co co − c1 c
At n = 0 , ∴ c1 = = , At n = 1 , ∴ c2 = = o
−4 2 2 * −3 4 * 3
At n = 2 , ∴ c3 =
(2 − 2)c2 = 0, At n = 3 , ∴ c4 =
c3
=0
3 * (− 2) 5 *1
2c4 0
At n = 4 , c5 = = Undefined number
5 * (4 − 4 ) 0
Assume c5 = c5*
different methods. x 2 y ′′ + 5 x y ′ + 3 y = 0
Solution: It is clear the above differential equation in the form of
2
Cauchy differential equation: x y′′ + axy′ + by = 0
∴ r (r + 4 ) + 3 = 0, or , r 2 + 4r + 3 = 0
∴ r1 = −1, and, r2 = −3 , It is clear that r1 − r2 = 2
This is the second case of the indicial equation, so,
∞
At r1 = −3 , ∴ x −3
∑ ((n + (− 3))(n + (− 3) + 4) + 3)cn x n = 0
n =1
∞
∴ x −3 ∑ ((n − 3)(n + 1) + 3)cn x n = 0
m =1
∴ ((n − 3)(n + 1) + 3)cn n =1, 2,3, 4,....
=0
( )
∴ y ( x ) = x − 3 co + c2 x 2 , or y ( x ) = co x − 3 + c2 x −1
It is clear we get the same solution in both methods.
330 Power Series Solution Of Differential Equations
Case III If r1 = r2 there always exist two linearity independent
solutions of equation (3) of the following form:
∞
y1 ( x ) = ∑ cm x m + r1 , c0 ≠ 0, and y2 ( x ) = y1 ( x) Ln x (32)
m=0
∞
∴ x r ∑ ((m + r )(m + r − 1 + 3) + 1)cm x m
m = 0
∞
− ∑ (m + r )(m + r − 1 + 1)cm x m −1 = 0
m =0
Chapter Nine 331
∞
∴ x ∑ ((m + r )(m + r + 2) + 1)cm x m
r
m 1=4
0 44442444443
n=m
∞
2 −1 m −1
− r co x − ∑ (m + r )(m + r − 1 + 1)cm x =0
=14444
1m4 42444444 3
n = m −1, or , m = n +1
The indicial equation is the coefficient of x r −1 which is
∞ ∞
x r ∑ ((n + r )(n + r + 2) + 1)cn x n − ∑ (n + r + 1) 2 cn +1 x n = 0
n = 0 n =0
Substitute in the above equation for r = 0 we get:
∑ [(n(n + 2) + 1)cn − (n + 1) 2 cn +1 ]x n = 0
∞
r
x
n=0
∑ [(n + 1) 2 cn − (n + 1) 2 cn +1 ]x n = 0
∞
r
∴x
n=0
∴ (n + 1) 2 cn − (n + 1) 2 cn +1 = 0
∴ cn +1 = cn , or co = c1 = c2 = c3 = c4 = c5 = c6 = c6 = ....cn
[
∴ y1 ( x ) = co 1 + x + x 2 + x 3 + x 4 + x 5 + ........ ]
1
But, = 1 + x + x 2 + x 3 + x 4 + x 5 + .......
1− x
c c ln ( x )
So, y1 ( x ) = o , and y2 ( x ) = y1 ( x ) ln ( x ) = o
1− x 1− x
∴ y ( x ) = C1 y1 ( x ) + C2 y 2 ( x ) =
1
1− x
( )
C1* + C2* ln ( x )
332 Power Series Solution Of Differential Equations
Example 26 Solve the following differential equation by two
different methods x 2 y ′′ + 3 x y ′ + y = 0
Solution: It is clear the above differential equation in the form of
m 2 + 2m + 1 = 0 , ∴ m1 = m2 = −1 ∴ y ( x) = (C1 + C2 Ln x )x −1
∞
∴ x r (r + 2) + 1 + ∑ ((m + r )(m + r + 2) + 1)cm x = 0
r m
1=4
m 1 44442444443
n=m
∞
∴x r
∑ ((n + r )(n + r + 2) + 1)cn x n = 0
n =1
∴ r (r + 2 ) + 1 = 0, or r 2 + 2r + 1 = 0
∴ r1 = r2 = −1
It is clear that this is the third case of the indicial equation, so,
∞
At r = −1 , ∴ x r
∑ ((n − 1)(n + 1) + 1)cn x n = 0
n =1
∴ y 2 ( x ) = c y1 ( x ) ln ( x ) = c* x −1 ln ( x )
( )
∴ y ( x ) = C1 y1 ( x ) + C2 y 2 ( x ) = C1* + C2* ln ( x ) x −1
334 Power Series Solution Of Differential Equations
9.7 Bessel Differential Equation
The following equation is called Bessel’s differential equation,
x 2 y ′′ + x y ′ + ( x 2 − v 2 ) y = 0 (33)
The above equation is so important differential equation used in
applied mathematics, physics and engineering.
In solving (33) we shall assume v ≥ 0 , where in. Since we seek
series solution of each equation about x = 0 , we observe that the
origin is a regular singular point of Bessel s equation.
(r 2 − r + r − v 2 )c0 x r +
∑ [(m + r )(m + r − 1) + (m + r ) − v ]c
∞ ∞
x r 2
mx
m
+x r
∑ cm x m + 2 = 0
m =1 m=0
∞
( )
∞
∴ (r 2 − v 2 )c0 x r + x r ∑cm (m + r ) 2 − v 2 x m + ∑cm x m+ 2 = 0
m=1 m=0
From above equation, we see that the indicial equation is:
Chapter Nine 335
r 2 − v2 = 0 (34)
So that the indicial roots are r1 = v and r2 = −v
∞ ∞
∴ x v ∑ c m m ( m + 2v ) x m + ∑ c m x m + 2 = 0
m =1 m=0
∞ ∞
v +1 v m m+2
∴ (1 + 2v)c1 x + x ∑ c m m ( m + 2v ) x + ∑ c m x =0
m 1=4
2 442444 3 m1=42
0 43
n=m n=m+2
( )
∞
v
n
∴ x (1 + 2v)c1 x + x ∑ (n + v) 2 − v 2 cn + cn − 2 = 0
n=2
Therefore, by the usual argument we can write: (1 + 2v ) c1 = 0 ,
( )
∴ (n + v) 2 − v 2 cn + cn − 2 = 0 n = 2, 3,4,5,......
− cn − 2
∴ cn = 2 2
= 0 n = 2, 3,4,5,...... (35)
(n + v) − v
The choice c1 = 0 in (35) implies c3 = c5 = c7 = .... = 0 , so for
n = 2, 4,6,8 .... , we fined, after letting n=2k, k=1,2,3,…..,
− c2 k − 2
∴ c2 k = , k = 1,2,3,4,......
2 2 k (k + v)
− co − c2 c0
∴ c2 = , c4 = = ,
2 2 4
2 1(1 + v) 2 * 2(2 + v) 2 * 1 * 2(1 + v)(2 + v)
− c4 c0
c6 = 2 =− 6 ,
2 * 3(3 + v) 2 * 1 * 2 * 3(1 + v)(2 + v)(3 + v)
336 Power Series Solution Of Differential Equations
(−1) k co
∴ c2 k = , k = 1,2,3,.... (36)
2 2k k! (1 + v)(2 + v)(3 + v)....(k + v)
Since this latter function possesses the convenient property
Γ(1 + v) = vΓ(v) , we can reduce the indicated product in the
demonstrator of (36) to one term.
For example:
Γ(1 + v + 1) = (1 + v)Γ(1 + v)
Γ(1 + v + 2) = (2 + v)Γ(2 + v)
= (2 + v)(1 + v)Γ(1 + v)
Γ(1 + v + k ) = (k + v)(k − 1 + v).....(3 + v)(2 + v)(1 + v)Γ(1 + v)
Γ(1 + v + k )
(k + v)(k − 1 + v).....(3 + v)(2 + v)(1 + v) =
Γ(1 + v)
Hence we can write (37) as:
(−1) k
So that finally: c2 k = , k = 0,1,2,3,.... (37)
2 2k + v k! Γ(1 + v + k )
So substituting the above results into assumed solution for v ≥ 0 :
Chapter Nine 337
∞
∴ y ( x) = ∑ c2 k x 2 k + v
k =0
∞ 2k + v
(−1) k x
∴ y ( x) = ∑ (38)
k =0 k ! Γ (1 + v + k ) 2
For each v , the function y ( x ) is called a Bessel function of the first
∞ 2 j+m
(−1) j * (−1) m x
∴ J − m ( x) = ∑
j =0( j + m )! Γ (1 + j ) 2
∞ 2 j+m
m (−1) j x
∴ J − m ( x) = (−1) ∑
( j
j = 0 123 + m )! Γ
1
(1
424
+
3
j ) 2
= Γ ( j + m +1) !j
∞ 2 j+m
m (−1) j
x
∴ J − m ( x) = (−1) ∑
j =0 j! Γ( j + m + 1)
2
In the above summation, assume the index j = k .
∞ 2k + m
m (−1) k x
∴ J − m ( x) = (−1) ∑ k! Γ(k + m + 1)
k =0 2
∞ 2k + v
(−1) k x
But we know that, J v ( x) = ∑
k =0 k ! Γ (1 + v + k ) 2
Chapter Nine 341
2 x x
n ∞ (− 1)m −1 (hm + hm + n ) 2m
Yn ( x ) = J n ( x ) ln + γ + ∑ 2m + n x
π 2 π m=0 2 m! (m + n )!
(49)
x − n n −1
(n − m − 1)! x 2m
−
π
∑ 2m − n
m=0 2 m!
342 Power Series Solution Of Differential Equations
Where x > 0, n = 0,1, 2,.... and γ is the so called Euler constant and
γ = 0.577 215 664 9 and
1 1 1
ho = 0, and hs = 1 + + + .... + , (s = 1,2,....) (50)
2 3 s
And when n = 0 the last sum in (49) is to be replaced by 0. For n = 0
the representation (49) takes the form (51). Furthermore, it can be
∞ (− 1) hm 2m
m −1
2 x
Y0 ( x ) = J 0 ( x ) ln + γ + ∑ x (51)
π 2 m =1 2 2m (m!)2
x 2 y ′′ + x y ′ + ( x 2 − 4 ) y = 0 on 0<x<∞
٢) x 4 y′′ − xy ′ + 2 y = 0
٣) (1 + x 2 ) y′′ − 2 y = 0
٤) (1 + 2 x 2 ) y ′′ + 3 xy ′ − 6 y = 0
٥) (1 − x 2 ) y ′′ − 10 xy ′ − 18 y = 0
٦) 3 x 2 y′′ + xy′ − (1 + x) y = 0
٧) 2 xy′′ + (1 + 2 x 2 ) y′ − xy = 0
٨) 2 xy′′ + (1 + 2 x) y′ + 4 y = 0
١٣) xy′′ + (1 − x 2 ) y′ − xy = 0
١٤) 4 x 2 y ′′ + (3 x + 1) y = 0
١٥) x 2 y′′ + x( x − 1) y′ + (1 − x) y = 0
٢٠) ( )
2 x 2 y ′′ + 2 x 2 + x y ′ − y = 0
٢١) 2 x 2 y ′′ + (2 x 2 − 3 x )y ′ + ( x + 2 ) y = 0
٢٢) 4 x 2 y ′′ + (4 x + 1) y = 0
٢٣) xy ′′ + (1 + x ) y ′ + y = 0
٢٤) x 2 y ′′ − xy ′ + (1 + x ) y = 0
٢٥) ( )
9 x 2 y ′′ + 9 x 2 + x y ′ + (12 x − 1) y = 0
٢٦) ( )
3 x 2 y ′′ + 3 x 2 + 5 x y ′ + (6 x − 1) y = 0
٢٧) ( )
x 2 y ′′ + x 2 − x y ′ + y = 0
٢٨) xy′′ + y′ + xy = 0
٢٩) x 2 y ′′ + 3 xy ′ + (1 + x ) y = 0
٣٠) x 2 y ′′ − x(2 − x ) y ′ + (2 − x ) y = 0
٣١) (
x 2 y ′′ + xy ′ − 2 − x 2 y = 0 )
٣٢) x 2 y ′′ − x(4 − x ) y ′ + (6 − 2 x ) y = 0
٣٣) (
x 2 y ′′ + xy ′ − 6 − x 2 y = 0 )
٣٤) x 2 y ′′ + 3 xy ′ + (1 + x ) y = 0
٣٥) (
x 2 y ′′ − 2 x 2 y ′ + x 4 + x 2 − 6 y = 0 )
346 Power Series Solution Of Differential Equations
٣٦) xy ′′ − y ′ + 4 x 5 y = 0
٣٧) ( )
x 2 y ′′ + xy ′ − 6 − x 2 y = 0
٣٨) x y ′′ + xy ′ + (x + 9 )y = 0
2 2
10.1 Definition
“A partial differential equation (PDE) is any equation involving a
function of more than one independent variable and at least one
partial derivative of that function. The order of a PDE is the order
of the highest order derivative that appears in the PDE”
∂f f ( x + ∆x, y ) − f ( x, y )
fx = = lim =
∂x ∆x → 0 ∆x
(1)
∂f f ( x, y + ∆y ) − f ( x, y )
fy = = lim =
∂y ∆y → 0 ∆y
That is to say, to find f x , we keep y constant and differentiate with
f ( x, y ) = cosh ( x − 6 y ) + e − xy
Solution:
∂f
(a) f x = = sinh ( x − 6 y ) − ye − xy
∂x
∂f
∴ fx = = sinh (2 − 6) − e − 2 = − sinh 4 − e − 2
∂x ( 2,1)
∂f
fy = = −6 sinh ( x − 6 y ) − xe − xy
∂y
∂f
∴ fy = = −6 sinh (−4) − 2e − 2 = 6 sinh 4 − 2e − 2
∂y ( 2,1)
x ∂u 1 ∂u
Example 3 If u = x y Show that + = 2u
y ∂x Ln x ∂y
Solution:
348 Partial Differential Equations
∂u x ∂u
= y x y −1 , ∴ = xy = u (2)
∂x y ∂x
∂u 1 ∂u
= x y Ln x , ∴ = xy = u (3)
∂y Ln x ∂y
x ∂u 1 ∂u
Add (2) and (3) we get: + = 2u
y ∂x Ln x ∂y
∂ ∂f ∂ 2 f
f xx = = 2 (6)
∂x ∂x ∂x
∂ ∂f ∂ 2 f
f xy = = (7)
∂y ∂x ∂y∂x
∂ ∂f ∂ 2 f
f yx = = (8)
∂x ∂y ∂x∂y
∂ ∂f ∂ 2 f
f yy = = 2 (9)
∂y ∂y ∂y
∂ ∂f ∂2 f
Thus the notation f xy or or means that we
∂y ∂x ∂y∂x
differentiate with respect to x then with respect to y. If the function
f xy and f yx are both continuous, then f xy = f yx , a similar
350 Partial Differential Equations
statement holds for functions of more than two variables. This
result, known as cumulative property of partial derivatives, may be
assumed to be true for all functions encountered at this stage.
f ( x, y, z ) = xe y cos z
Solution:
dy
Example 6 If y = 2 x 2 , and , x = 5t 2 Find
dt
Solution: From (10) we can get the following results
dy dy dx
= * = 4 x * 10t = 40 xt
dt dx dt
dy
But, x = 5t 2 ∴ = 200t 3
dx
Chapter Ten 351
Case1 Suppose that z = f ( x, y ) is a differentiable function of x and
y, where x = X (t ) , y = Y (t ) are both differentiable function of t,
then z is a differentiable function of t and,
dz ∂ z d x ∂ z d y
∴ = * + * (11)
dt ∂x dt ∂y dt
∂z ∂z ∂x ∂z ∂y
= + (12)
∂s ∂x ∂s ∂y ∂s
∂z ∂z ∂x ∂z ∂y
= + (13)
∂t ∂x ∂t ∂y ∂t
352 Partial Differential Equations
∂z ∂z
Example 8 If z = e x sin y , x = st 2 , y = s 2 t find and
∂s ∂t
Solution: From case 2 of Chain rule
∂z ∂z ∂x ∂z ∂y
= +
∂s ∂x ∂s ∂y ∂s
= (e x sin y ) t 2 + (e x cos y ) 2 st
∂z ∂z ∂x ∂z ∂y
= +
∂t ∂x ∂t ∂y ∂t
= (e x sin y ) 2 st + (e x cos y ) s 2
It can be easily to extend case 2 of Chain rules to the function f
which contains three variables f = f ( x, y, z )
Where x = x( s, t ), y = y ( s, t ), z = z ( s, t ) .
Then we have the extension of case 2 of Chain rules
∂f ∂f ∂x ∂f ∂y ∂f ∂z
= + + (14)
∂s ∂x ∂s ∂y ∂s ∂z ∂s
∂f ∂f ∂x ∂f ∂y ∂f ∂z
= + + (15)
∂t ∂x ∂t ∂y ∂t ∂z ∂s
∴
∂f
∂s
( )( ) ( )( )(
= 4 x 3 y ret + x 4 + 2 y z 3 2rse − t + 3 y 2 z 2 r 2 sin t)( )
When r = 2, s = 1, t = 0 we have x = 2, y = 2, z = 0
∂f
∴ = 64 * 2 + 16 * 4 + 0 * 0 =192
∂s
x2 y2 z2 ∂z ∂z
Example 10 If + + = 1, Find and
a2 b2 c2 ∂x ∂y
Solution:
x2 y2 z2
Let
2
+ 2
+ 2
−1 = 0
a b c
∂z Fx 2x / a2 c2 x
∴ =− =− =− 2
∂x Fz 2z / c2 a z
∂z Fy 2 y / b2 c2 y
∴ =− =− =− 2
∂y Fz 2z / c 2
b z
354 Partial Differential Equations
10.8 Applications
In this section we will drive the differential equations which governs
the conduction of heat in solids.
y
In the following sections some
basic problems associated with x
this equation are solved. In the L
case of heat conduction in two x0 x 0 + ∆x
parallel plates of the same area A and different constant
temperatures T1 and T2 respectively are separated by a small
distance d, an amount of heat H per unit time will pass from the
warmer to the cooler. Moreover, to a high degree of approximation,
H is proportional to the area A, the temperature difference
T2 − T1 , and inversely proportional to the separation distance, d.
Thus we can write:
kA T2 − T1
H= (17)
d
Where the positive proportionality factor k is called the thermal
conductivity and depends only on the material between the two
plates. The physical law expressed by equation (17) is known as
Newton’s law of cooling.
Now consider a straight rod of uniform cross section and
homogeneous material, oriented so x-axis lies along the axis of the
rod. Let x = 0 and x = L designate the ends of the bar. We will
assume that the sides of the bar are perfectly isolated so that there is
Chapter Ten 355
no passage of heat through them. We will also assume that the
temperature u depends only on the axial position x and time t, and
not on the lateral coordination y and z; hence u equals u ( x, t ) .
Consider an element of the bar lying between the cross section
x = x0 and x = x0 + ∆x , where x = x0 is arbitrary and ∆x is so
small. The instantaneous rate of heat transfer from left to the right
across the cross section x = x0 is obtained by taking the limit in
H ( x0 , t ) = −kA u x ( xo , t ) (18)
The minus sign appears in this equation since there will be positive
flow of heat from left to right only if the temperature is greater to
the left of x = x0 than to the right; in this case u x ( xo , t ) will be
negative. In a similar manner the rate at which heat passes from left
to right through the cross section x = x0 + ∆x is given by:
1 Q∆t Q∆t
∆u = = (21)
s ∆m sρ A∆x
Where the constant of proportionality s is known as the specific heat
of the material of the bar, and ρ is its density. The average
temperature change ∆u in the bar element under consideration is the
temperature change at some intermediate point x = x0 + θ ∆x ,
where 0 < θ < 1. Thus (21) can be written as:
Q ∆t
u ( x0 + θ ∆x, t + ∆t ) − u ( x0 + θ ∆x, t ) = (22)
s ρ A ∆x
Solving equation (22) for Q and equating the resulting expression
with the given by equation (20) yields the following equation:
u x ( x0 + ∆x, t ) − u x ( x0 , t )
Ak =
∆x
(23)
u ( x0 + θ ∆x, t + ∆t ) − u ( x0 + θ ∆x, t )
sρ A
∆t
If we let ∆x and ∆t approach zero in equation (23), we obtain
the heat conduction of diffusion equation:
α 2u xx = ut (24)
∂ 2u ∂ 2u ∂ 2u ∂u ∂u
A 2 +B +C 2 +D +E + Fu = G (28)
∂x ∂x∂y ∂y ∂x ∂y
Where A, B, ........, G may depend on x and y but not u. A second
order equation with independent variable x and y which does not
have the form (28) is called nonlinear. Some terms of the above
equation can be eliminated depending on the physical application
used. In the following there is some easy examples showing how to
solve second order partial differential equations.
∂2z
= x 2 y where, z ( x,0) = x 2 , z (1, y ) = cos y
∂x∂y
Chapter Ten 359
Solution: The above partial differential equation can be written
∂ ∂z
in the following form: = x 2 y
∂x ∂y
By integrating the above equation with respect to x we fiend:
∂z 1 3
= x y + F (y) where F ( y ) is arbitrary.
∂y 3
Integrating the above equation with respect to y we get:
1 3 2
z= x y + ∫ F ( y )dy + G ( x ) where G ( x ) is arbitrary.
6
The above results can be written in the following form:
1 3 2
z= x y + H ( y ) + G(x )
6
The above equation has two arbitrary functions and is therefore a
general solution.
x 2 = 0 + H (0 ) + G ( x ) Or, G ( x ) = x 2 − H (0 )
1 3 2
∴z= x y + H ( y ) + x 2 − H (0)
6
Since z (1, y ) = cos y , we have from above equation,
1 2
∴ cos y = y + H ( y ) + 1 − H (0)
6
1 2
∴ H ( y ) = cos y − y − 1 + H (0 )
6
1 1
∴ z = x 3 y 2 + cos y − y 2 + x 2 − 1
6 6
360 Partial Differential Equations
∂ 2u ∂u
Example 13 Solve t +2 = x2
∂x∂t ∂x
∂ ∂u
Solution: Write the equation as t + 2u = x 2
∂x ∂t
∂u 1
Integrating with respect to x, ∴ t + 2u = x 3 + F (t )
∂t 3
∂u 2 1 x 3 F (t )
∴ + u= +
∂t t 3 t t
This is a linear first order ordinary differential equation
ln t 2 1 x 3 F (t )
∴u =e − ln t 2
*∫ e + dt + H ( x )
3 t t
2 1 x 3 F (t )
∴ t u = ∫ t
2
+ dt + H ( x )
3 t t
1
∴ t 2u = t 2 x 3 + G (t ) + H ( x )
6
X ′′( x ) + λ2 X ( x ) = 0 (36)
λ2 = 0
Case 1 If λ2 > 0
2 2
∴ X ( x) = k1 sin λ x + k 2 cos λ x , and T (t ) = e −α λ t
2 2
∴ u ( x, t ) = X ( x) T (t ) = e −α λ t (k1 sin λ x + k 2 cos λ x) (38)
Is a solution of the partial differential equation (29) regardless of the
values of λ and the arbitrary constants k1 and k 2 .
4π x
Example 14 Let f ( x) = 3 sin
L
nπ x
∴ u n ( x,0) = cn sin = f ( x)
L
Provided n=4 and cn =3.
The solution to the complete boundary value problem is:
2 2 2 4πx
u ( x, t ) = 3e −16 π α t / L sin = 0, n = 1,2,3,......
L
Chapter Ten 365
Example 15: Use the method of separation of variables to solve the
following partial differential equation:
∂u ∂ 2u
= ,
∂t ∂x 2
3πx 9πx
where u x (0, t ) = 0, u (2, t ) = 0 , and, u ( x,0 ) = 8 cos − 6 cos
4 4
Solution:
Assume u ( x, t ) = X ( x) T (t )
∂u ∂ 2u
Q = ∴ X ( x )T ′(t ) = X ′′( x )T (t )
∂t ∂x 2
X ′′( x ) T ′(t )
∴ = = −λ2
X ( x ) T (t )
∴ X ′′( x ) + λ2 X ( x ) = 0 (40)
T (t ) = ce − λ
2
t
Q u ( x, t ) = X ( x) T (t )
∴ u x ( x, t ) = λe − λ (K 2 cos λx − K1 sin λx )
2
t
Q u x (0, t ) = 0
∴ u x (0, t ) = λe − λ (K 2 ) = 0
2
t
∴ K2 = 0
Q u (2, t ) = 0
∴ u (2, t ) = e − λ ( K1 cos 2λ ) = 0
2
t
nπ
∴ 2λ = for n=1, 3, 5,….
2
nπ
∴λ = for n=1, 3, 5,….
4
2 nπ
∴ u ( x, t ) = e − λ t K1 cos x
4
nπ
∴ u ( x,0) = K1 cos x
4
3πx 9πx
Q u ( x,0 ) = 8 cos − 6 cos
4 4
By comparing the above equations we get the following:
nπ 3πx 9πx
∴ K1 cos x = 8 cos − 6 cos
4 4 4
From the first term we get the following:∴ K1 = 8 and n = 3
∂2 y ∂2 y
(b) =4 , where y (0, t ) = 0, y ( x,0 ) = 0, yt ( x,0 ) = 5 sin πx
∂t 2 ∂x 2
Solution: Assume u ( x, t ) = X ( x) T (t )
∂2 y ∂ 2u
Q 2
=4 2
∴ X ( x )T ′′(t ) = 4 X ′′( x )T (t )
∂t ∂x
X ′′( x ) T ′′(t )
∴ = = − λ2
X ( x ) 4T (t )
∴ X ′′( x ) + λ2 X ( x ) = 0 (42)
not valid. So k1 = 0 .
∴ k3 = 0
∴ y ( x, t ) = (k 2 sin λx ) * (k 4 sin 2λt )
∴ yt ( x, t ) = (k 2 sin λx ) * (k 4 * 2λ * cos 2λt )
Q yt ( x,0 ) = 5 sin πx
∴ yt ( x,0) = (k 2 sin λx ) * (k 4 * 2λ ) = 5 sin πx
∴ λ = π ∴ (k 2 ) * (k 4 * 2π ) = 5
5
∴ k2 k4 =
2π
5
∴ y ( x, t ) = (sin πx * sin 2π t )
2π
∂2 y ∂2 y
(c)
2
=4 2
, where y (0, t ) = 0, y ( x,0 ) = 0, and
∂t ∂x
yt ( x,0 ) = 3 sin 2πx − 2 sin 5πx
Assume u ( x, t ) = X ( x) T (t )
∂2 y ∂ 2u
Q =4
∂t 2 ∂x 2
∴ X ( x )T ′′(t ) = 4 X ′′( x )T (t )
X ′′( x ) T ′′(t )
∴ = = −λ2
X ( x ) 4T (t )
Chapter Ten 369
∴ X ′′( x ) + λ2 X ( x ) = 0 (44)
is not valid. So k1 = 0 .
Q y ( x ,0 ) = 0
∴ y ( x,0 ) = (k 2 sin λx ) * (k3 ) = 0 ∴ k3 = 0
∴ y ( x, t ) = (k 2 sin λx ) * (k 4 sin 2λt )
∴ yt ( x, t ) = (k 2 sin λx ) * (k 4 * 2λ * cos 2λt )
Q yt ( x,0 ) = 3 sin 2πx − 2 sin 5πx
∴ yt ( x,0 ) = (k 2 sin λx ) * (k 4 * 2λ ) = 3 sin 2πx − 2 sin 5πx
Then for the first term of yt ( x,0 ) , ∴ λ = 2π
3
∴ (k 2 ) * (k 4 * 4π ) = 3 ∴ k 2 k 4 =
4π
Then for the second term of yt ( x,0 ) , ∴ λ = 5π
370 Partial Differential Equations
−2
∴ (k 2 ) * (k 4 * 10π ) = −2 ∴ k 2 k 4 =
10π
3
∴ y ( x, t ) = (sin 2πx * sin 4π t ) − 1 (sin 5πx * sin 10π t )
4π 5π
2L nπx
bn = ∫ f ( x) sin dx (48)
L0 L
To summarize the solution of the heat conduction problem (29),
(30), (31);
50
L=100 cm
-L=-100 cm
50
L=100 cm
-L=-100 cm x
2L nπx
bn = ∫ f ( x) sin dx
L0 L
2 50 nπx
100
nπx
= ∫ x sin dx + ∫ (100 − x) sin dx
100 0 100 50 100
400 sin( nπ / 2)
=
n 2π 2
400 2 2 πx 1 2 2 3πx
u ( x, t ) = 2 e −π 1.14 t / 10000 sin − e − 9 π 1.14 t / 10000 sin
π L 9 L
1 2 2 5πx
+ e − 25π 1.14 t / 10000 sin + .....
25 L
∂u ∂ 2u
= u x (0, t ) = u x (π , t ) = 0, u ( x,0 ) = π (1 − x ) ,
∂t ∂x 2
where 0 < x < π , t > 0
Solution: Assume u ( x, t ) = X ( x ) * T (t )
374 Partial Differential Equations
∂u ∂ 2u
Q = 2 ∴ X ( x ) * T ′(t ) = X ′′( x ) * T (t )
∂t ∂x
X ′′( x ) T ′(t )
∴ = = −λ2
X ( x ) T (t )
∴ X ′′( x ) + λ2 X ( x ) = 0 (50)
T (t ) = ce − λ
2
t
Q u ( x, t ) = X ( x) T (t )
∴ u ( x, t ) = e − λ (K1 sin λx + K 2 cos λx )
2
t
∴ u ( x, t ) = K 2 e − λ t (cos λx )
2
(52)
It is clear from (52) that we need to compare with cosine terms, so,
assume u ( x,0 ) is even function. So, bn = 0 .
u ( x ,0 )
π
−π π
Chapter Ten 375
1π π2
∴ ao =
π ∫ π (1 − x )dx = π − 2
0
2π nπx
∴ an =
π
∫ π (1 − x ) cos π
dx
0
2π
∴ an =
π ∫ π (1 − x ) cos nxdx
0
u = (1 − x ) dv = cos nx
sin nx
du = −dx v=
n
(1 − x ) sin nx π 1 π
∴ an = 2 + ∫ sin nxdx
n 0 n0
1 π
2 4
∴ an = 2 2 (− cos nx ) = 2 (1 − cos nx ) = 2 n = 1, 3, 5,....
n 0 n n
π2 1
∴ u ( x,0 ) = π − + 4cos x + cos 3x +
2 9
1 1
+ cos 5 x + cos 7 x + ......
25 49
∞
∴ u ( x ,0 ) = ∑ k 2 cos λx
λ =0
π2 1
∴ u ( x, t ) = π − + 4e − t cos x + e − 9t cos 3x +
2 9
1 1
+ e − 25t cos 5 x + e − 49t cos 7 x + ....
25 49
376 Partial Differential Equations
Example 20 Use the method of Fourier series to solve the following
boundary value problems:
∂u ∂ 2u
(a) = 2 2 , u (0, t ) = u (4, t ) = 0, u ( x,0 ) = 25 x
∂t ∂x
where 0 < x < 4, t > 0
Solution: Assume u ( x, t ) = X ( x ) * T (t )
∂u ∂ 2u
Q = 2 2 ∴ X ( x ) * T ′(t ) = 2 X ′′( x ) * T (t )
∂t ∂x
X ′′( x ) T ′(t )
∴ = = −λ2
X ( x ) 2T (t )
∴ X ′′( x ) + λ2 X ( x ) = 0 (53)
T (t ) = ce − 2λ
2
t
Q u (0, t ) = 0 ∴ u (0, t ) = e − 2λ (K 2 ) = 0
2
t
∴ K2 = 0
nπ
Q u (4, t ) = 0 ∴ u (4, t ) = e − 2λ (K1 sin 4λ ) = 0
2
t
∴λ =
4
Chapter Ten 377
n 2π 2
− t nπ
∴ u ( x, t ) = e 8 K sin
x
1
4
nπ x
Qu ( x,0) = 25 x = K1 sin
4 (55)
It is clear from (55) that we need to compare with sine terms so we
will complete u ( x,0 ) as an odd function as shown in the figure:
u ( x ,0 )
100
x
4
4
2L nπx 50 4 nπx
∴ bn = ∫ 25 x sin
L0 L
dx ∴ bn = ∫
4 0
x sin
4
dx
25 4 2 nπx
4
nπx 4
∴bn = sin − x cos
2 n 2π 2 4 nπ 4 0
− 200
∴bn = cos nπ
nπ
− 200
∴bn = , n = 2, 4, 6, ......
nπ
200
∴bn = , n = 1, 3, 5, ......
nπ
378 Partial Differential Equations
∂u ∂ 2u
= 2 u x (0, t ) = u x (π , t ) = 0, u ( x,0) = x 2
∂t ∂x
where 0 < x < π , t > 0
Solution: Assume u ( x, t ) = X ( x ) * T (t )
∂u ∂ 2u
Q =
∂t ∂x 2
∴ X ( x ) * T ′(t ) = X ′′( x ) * T (t )
X ′′( x ) T ′(t )
∴ = = −λ2
X ( x ) T (t )
∴ X ′′( x ) + λ2 X ( x ) = 0 (56)
T (t ) = ce − λ
2
t
Q u ( x, t ) = X ( x) T (t )
Q u x (0, t ) = λe − 2λ (K1 ) = 0 ∴ K1 = 0
2
t
∴ u x (π , t ) = λe − 2λ (− K 2 sin λπ ) = 0
2
t
∴λ = n
∴ u ( x, t ) = e − 2 λ (K 2 cos nx )
2
t
(58)
It is clear from (58) that we need to compare with cosine terms so
u ( x ,0 ) = x 2
x
−π π
380 Partial Differential Equations
1 L 1π π3 π2
f ( x )dx 2
2 L −∫L
∴ ao = ∴ ao = ∫ x dx = =
π0 3π 3
1 L 2π
∴ an = ∫ f ( x ) cos nx dx 2
π 0∫
∴ an = x cos nx dx
L −L
π
2 2x 2 x2
∴ an = 2 cos nx − 3 sin nx + sin nx
π n n n 0
2 2π
∴ an = cos nx
π n 2
4
∴ an = cos nπ
n2
4
∴ an = for n = 2, 4,6,....
n2
4
∴ an = − for n = 1,3,5,....
n2
π2 1 1
∴ u ( x, o ) = + 4 − cos x + cos 2 x − cos 3x + .....
3 4 9
π2 1 1
∴ u ( x, t ) = + 4 − e − t cos x + e − 4t cos 2 x − e − 9t cos 3x + .....
3 4 9
Chapter Ten 381
10.9.3 Solution Using Laplace Transform
Example 22 Use the method of Laplace transform to solve the
∂u ∂ 2u
following boundary value problem: =4 2
∂t ∂x
Where u (0, t ) = 0, u (3, t ) = 0, u ( x,0 ) = 10 sin 2πx − 6 sin 4πx
Solution: Taking the Laplace transform of the given differential
equation with respect to t, we have:
∞ − st ∂ u
2
∞ − st ∂u
∫ e dt =∫ e 4 2 dt
0 ∂t 0
∂x
Which can be written as:
∞ − st ∂2 ∞ − st
s∫ e udt − u ( x,0 ) = 4 ∫ e u dt
0
∂x 2 0
∂ 2U
Or sU − u ( x,0 ) = 4 (59)
∂ x2
∞
where U = U ( x, s ) = L{u ( x, t )} = ∫ e − st udu
0
Using the given condition u ( x,0 ) = 10 sin 2πx − 6 sin 4πx (60)
∂ 2U
becomes: 4 − sU = 6 sin 4πx − 10 sin 2πx (61)
∂x 2
Taking the Laplace Transform of the conditions
u (0, t ) = 0, u (3, t ) = 0 , we have: L{u (0, t )} = 0, L{u (3, t )} = 0
Or U (0, s ) = 0, U (3, s ) = 0 (62)
Solving the ordinary differential equation (61) subject to condition
(62) by the usual elementary methods as following:
382 Partial Differential Equations
4λ2 − s = 0
∴ λ1 = s / 2 and λ2 = − s / 2
∴ U h ( x, s ) = C1e s /2 x
+ C2e − s /2 x
(
∴ 4 * − 42π 2 A cos4πx − 42π 2 B sin 4πx − 22π 2 D cos2πx − 22π 2 F sin 2πx )
− s( A cos4πx + B sin 4πx + D cos2πx + F sin 2πx) = 6 sin 4πx − 10sin 2πx
By comparing coefficients in both sides we get:
−6
A = 0 , D = 0 ∴ − 64π 2 B − sB = 6 ∴ B =
s + 64π 2
10
And − 16 π 2 F − sF = −10 ∴ F =
s + 16π 2
6 sin 2πx 6 sin 4πx
∴U ( x, s ) = U h + U P = C1e s /2x
+ C2 e − s /2 x
+ −
s + 16π 2 s + 64π 2
But from (62) U (0, s ) = 0,
∴ C1 + C2 = 0 And, U (3, s ) = 0
∴ C1e3 s /2
+ C2e −3 s /2
+0=0
But, C 2 = −C1 ∴ C1 e3 s /2
− e −3 s /2
=0
Chapter Ten 383
Then must be C1 = 0 ∴ C2 = 0
10 sin 2πx 6 sin 4πx
∴ U ( x, s ) = −
s + 16π 2 s + 64π 2
By taking the inverse Laplace transform we find:
u ( x, t ) = 10e −16π
2 2
t
sin 2πx − 6e − 64π t
sin 4πx
Which is the required solution:
10 sin 2πx 6 sin 4πx
∴ U ( x, s ) = −
s + 16π 2 s + 64π 2
By taking the inverse Laplace transform we find the required
solution as following:
u ( x, t ) = 10e −16π
2 2
t
sin 2πx − 6e − 64π t
sin 4πx
∂u ∂ 2u
= , where u (0, t ) = 0, u (4, t ) = 0 , and
∂t ∂x 2
πx
u ( x,0 ) = 6 sin + 3 sin πx
2
Solution: As explained in the previous example
∂ 2U
∴ sU − u ( x,0) = 2
(63)
∂ x
∞
Where U = U ( x, s ) = L{u ( x, t )} = ∫ e − st udu
0
πx
Using the given condition u ( x ,0 ) = 6 sin + 3 sin πx , (63)
2
becomes:
384 Partial Differential Equations
∂ 2U πx
− sU = 6 sin + 3 sin πxx (64)
∂x 2 2
Taking the Laplace Transform of the conditions:
u (0, t ) = 0, u (4, t ) = 0 , we have:
L{u (0, t )} = 0, L{u (4, t )} = 0
Or U (0, s ) = 0, U (4, s ) = 0 (65)
Solving the ordinary differential equation (64) subject to
condition (65) by the usual elementary methods as following:
λ2 − s = 0 ∴ λ1 = s and λ2 = − s
∴ U h ( x, s ) = C1e s x + C 2 e − s x
π π
∴ U p ( x, s ) = A cos x + B sin x + D cos πx + F sin πx
2 2
π π π π
∴U ′p ( x, s ) = − A sin x+ B cos x
2 2 2 2
− πD sin πx + πF cosπx
π2 π π2 π
∴U ′p′ ( x, s ) = − A cos x − B sin x
4 2 4 2
− π 2 D cosπx − π 2 F sinπx
π2 π π2 π
− A cos x− B sin x − π 2 D cos πx − π 2 F sin πx
4 2 4 2
π π
− s A cos x + B sin x + D cos πx + F sin πx
2 2
πx
= −6 sin − 3 sin π
2
Chapter Ten 385
π
By comparing the coefficient of cos x and cos πx we get the
2
following constants: A = 0, D = 0
π
By comparing the coefficient of sin x and sin πx we get the
2
π2
following constants: − B − sB = −6
4
6 3
∴ B= and − π 2 F − sF = −3 ∴F =
π2 s +π 2
s+
4
∴U ( x, s ) = U h + U P = C1e s /2x
+ C2 e − s /2 x
6 πx 3
+ sin + sin πx
s +π 2 4 2 s +π 2
QU (0, s ) = 0, ∴ C1 + C 2 = 0
But, C 2 = −C1 ∴ C1 e 4 s − e − 4 s = 0
Then must be C1 = 0 ∴ C2 = 0
6 πx 3
∴U ( x, s ) = U h + U P = 2
sin + 2
sin πx
s +π 4 2 s +π
By taking the inverse Laplace transform we find:
∴ u ( x, t ) = 6e − (π ) sin πx + 3e −π
2 2
4t t
sin πx
2
386 Partial Differential Equations
Problems
f ( x, y ) = x 2 sin y, and
1) If
g ( x, y ) = e y sin( x + y )
Find f x , f y , g x , g y
f ( x, y, z ) = e 2 x sinh y cosh z
4) Suppose that:
y
z = sin( xy ) cos , x = te − t , y = sinh t
x
dz
Use the Chain rule to find
dt
5) If z = x Ln y, x = st 2 , y = s 2 t
∂z ∂z
Find and
∂t ∂s
6) If sin 2 x + cos 2 y + tan 2 z = 1
∂z ∂z
Find and
∂x ∂y
7) Let a metallic rod 20cm long be heated to a uniform temperature
of 100oC. Suppose that at t=0 the ends of the bar are plunged into an
∂u ∂ 2u
(a) = , where u x (0, t ) = 0, u (2, t ) = 0 , and
∂t ∂x 2
3πx 9πx
u ( x,0 ) = 8 cos − 6 cos
4 4
∂2 y ∂2 y
(b) =4 , where
∂t 2 ∂x 2
y (0, t ) = y (5, t ) = 0, y ( x,0 ) = 0, yt ( x,0 ) = 5 sin πx
∂2 y ∂2 y
(c)
2
=4 2
, where y (0, t ) = y (5, t ) = 0, y ( x,0 ) = 0, and
∂t ∂x
yt ( x,0 ) = 3 sin 2πx − 2 sin 5πx
(9) Use the method of Fourier series to solve the following boundary
value problems:
∂u ∂ 2u
(a) = 2 2 , u (0, t ) = u (4, t ) = 0, u ( x,0 ) = 25 x
∂t ∂x
where 0 < x < 4, t > 0
388 Partial Differential Equations
∂u ∂ 2u
(b) = 2 u x (0, t ) = u x (π , t ) = 0, u ( x,0) = x 2
∂t ∂x
where 0 < x < π , t > 0 , and
∂u ∂ 2u
(a) = , where u (0, t ) = 0, u (4, t ) = 0 , and
∂t ∂x 2
πx
u ( x,0 ) = 6 sin + 3 sin πx
2
∂u ∂ 2u
(a) = , where u x (0, t ) = 0, u x (2, t ) = 0 , and
∂t ∂x 2
u ( x,0 ) = 4 cos πx − 2 cos 3πx
Chapter 11
Simultaneous Linear Differential Equations
1
An obvious solution of the last equation is X 1 = and, of course,
1
this vector multiplied by any number is also a solution. Thus, for all
1
nonzero numbers c1 , X = c1 is a characteristic vector of A
1
corresponding to λ1 = −1 .
392 Simultaneous Linear Differential Equations
If λ = 3 , the equation det (λI − A) X = 0 is equivalent to
5
− x1 + 5 x2 = 0 , which has X 2 = as an obvious solution. Thus,
1
5
for every nonzero number c2 , X = c2 is a characteristic vector
1
of A corresponding to λ2 = 3 .
As a check on the characteristic values and vectors just found, we
observe that:
− 5 5 1 0
[λ1I − A] =
1
= and that:
1 − 1 1 1 0
5 − 1 5 5 0
[λ2 I − A] = − 1 5 1 = 0
1
3
given by X = c1 2 , where c1 can be any nonzero number.
4
The equivalent component form of the matrix equation
− x1 + 2 x2 + 5 x3 = 0
(2 I − A)X = 0 is: − 4 x1 + 3x2 + 5 x3 = 0
2 x1 + x2 + 5 x3 = 0
x1 + x3 =0
This system is equivalent to
x2 +3 x3 = 0
Setting x3 = −c2 , we find that the characteristic vectors of A
1
corresponding to λ2 = 2 are given by X = c2 3 , c2 ≠ 0
− 1
394 Simultaneous Linear Differential Equations
Example 3 Find the characteristic values and the corresponding
characteristic vectors of each of the following matrix:
4 6 6
1 3 2
− 1 − 5 − 2
Solution:
λ − 4 − 6 −6
det (λI − A) = − 1 λ − 3 − 2 = 0
1 5 λ + 2
∴ (λ − 4 )(λ − 3)(λ + 2 ) + 10 + 6(− (λ + 2 ) + 2 ) − 6(− 5 − (λ − 3)) = 0
∴ λ3 − λ2 + 4λ − 4λ2 + 4λ − 16 + 12 = 0
∴ λ3 − 5λ2 + 8λ − 4 = 0
The roots of the above equations (eigen values) are:
λ1 = 1 and λ2,3 = 2
For λ1 = 1
− 3 x1 − 6 x2 − 6 x3 = 0
− x1 − 2 x2 − 2 x3 = 0
x1 + 5 x2 + 3 x3 = 0
∴ 3 x2 + x3 = 0
Let x2 = c1
∴ x3 = −3 x2 = −3c1
∴ x1 + 5c1 − 9c1 = 0
Chapter Eleven 395
x1 − 4c1 = 0
∴ x1 = 4c1
x1 4
∴ x2 = c1 1
x − 3
3
For λ = 2
− 2 x1 − 6 x2 − 6 x3 = 0
− x1 − x2 − 2 x3 = 0
x1 + 5 x2 + 4 x3 = 0
∴ 4 x2 + 2 x3 = 0
∴ x3 = −2x2
Let x2 = c2 ∴ x3 = −2c2 ∴ x1 = 3c2
The eigen vectors for λ = 2 is shown below:
x1 3
x
2 = c 2 1
x − 2
3
∴ λ3 − 4λ2 + 5λ − 2 = 0
∴ λ1,2 = 1 and λ3 = 2
− 6 x1 + 2 x2 + 4 x3 = 0
− 3x1 + x2 + 2 x3 =0
− 6 x1 + 2 x2 + 4 x3 = 0
From subtracting the last two equations we get the following
equation:
x3 = 0 ∴ −6 x1 + 2 x2 = 0
Then, if x1 = c1 ∴ x2 = 3c1
x1 1
x =
2 1 3
c
x 0
3
Eigen vectors for λ3 = 2
− 5 x1 + 2 x2 + 4 x3 = 0
− 3x1 + 2 x2 + 2 x3 = 0
− 6 x1 + 2 x2 + 5 x3 = 0
Chapter Eleven 397
From subtracting the last two equations we get the following
equation 3 x1 − 3 x3 = 0
∴ x1 = x3 = c2
1
x1
1
x 2 = c2
x 2
3 1
Example 5 Find the characteristic values and the corresponding
characteristic vectors of each of the following matrix:
2 4 −6
4 2 −6
− 6 − 6 − 15
Solution:
λ−2 −4 6
det (λI − A) = − 4 λ−2 6 =0
6 6 λ + 15
− 4 x1 − 4 x2 + 6 x3 = 0 (6)
− 4 x1 − 4 x2 + 6 x3 = 0 (7)
6 x1 + 6 x2 + 13 x3 = 0 (8)
Divide (6) by 2, we get the following equation:
398 Simultaneous Linear Differential Equations
− 2 x1 − 2 x2 + 3 x3 = 0 (9)
Multiply (9) by 3 we get the following equation:
− 6 x1 − 6 x2 + 9 x3 = 0 (10)
Add (8) to (10) we get the following:
22 x3 = 0 ∴ x3 = 0 , and x1 = − x2
Let x1 = c1 ∴ x2 = −c1
x1 1
∴ x2 = c1 − 1
x 0
3
Eigen vectors for λ2 = 9
7 x1 − 4 x2 + 6 x3 = 0 (11)
− 4 x1 + 7 x2 + 6 x3 = 0 (12)
6 x1 + 6 x2 + 24 x3 = 0 (13)
From subtracting the first two equations we get the following
equation:
11x1 − 11x2 = 0
∴ x1 = x2 = c2
∴ 7c2 − 4c2 + 6x3 = 0
−1
∴ x3 = c2
2
Chapter Eleven 399
x1 2
1
x 2 = c2 2
x 2 − 1
3
Eigen vectors for λ3 = −18
− 20 x1 − 4 x2 + 6 x3 = 0
− 4 x1 − 20 x2 + 6 x3 = 0
6 x1 + 6 x2 − 3x3 = 0
From subtracting the first two equations we get the following
equation:
− 16 x1 + 16 x2 = 0
∴ x1 = x2 = c3
∴ −20c3 − 4c3 + 6 x3 = 0
∴ x3 = 4c3
x1 1
∴ x2 = c3 1
x 4
3
∴ λ3 − 3λ2 + 2λ = 0
∴ λ1 = 0 , λ2 = 1 and λ3 = 2
Eigen vectors for λ1 = 0
− 11x1 + 4 x2 + 7 x3 = 0 (14)
− 7 x1 + 2 x2 + 5 x3 = 0 (15)
− 10 x1 + 4 x2 + 6 x3 = 0 (16)
From subtracting (14) and (16) we get the following equation:
∴ x1 − x3 = 0
∴ x1 = x3 = c1
∴ −7c1 + 2 x2 + 5c1 = 0
∴ x2 = c1
x1 1
∴ x2 = c1 1
x 1
3
Eigen vectors for λ2 = 1
− 10 x1 + 4 x2 + 7 x3 = 0 (17)
− 7 x1 + 3x2 + 5 x3 = 0 (18)
Chapter Eleven 401
− 10 x1 + 4 x2 + 7 x3 = 0 (19)
Multiply (17) by 7 and (18) by 10 we get the following two
equations:
− 70 x1 + 28 x2 + 49 x3 = 0 (20)
− 70 x1 + 30 x2 + 50 x3 = 0 (21)
Subtract (20) from (21) we get the following equation:
∴ 2 x2 + x3 = 0
Let x2 = c2 ∴ x3 = −2c2
∴ −7 x1 + 3c2 − 10c2 = 0
∴ x1 = −c2
x1 −1
∴ x 2 = c 2 1
x − 2
3
Eigen vectors for λ3 = 2
− 9 x1 + 4 x2 + 7 x3 = 0 (22)
− 7 x1 + 4 x2 + 5 x3 = 0 (23)
− 10 x1 + 4 x2 + 8 x3 = 0 (24)
Subtract (23) from (22) we get the following two equations:
∴ −2 x1 + 2 x3 = 0 (25)
∴ x1 = x3 = c3
∴ −9c3 + 4 x2 + 7c3 = 0 (26)
402 Simultaneous Linear Differential Equations
c3
∴ x2 =
2
x1 2
∴ x2 = c3 1
x 2
3
∴ λ3 − 4λ2 − λ + 4 = 0
∴ λ1 = 4 , λ2 = 1 , and λ3 = −1
Eigen vectors for λ1 = 4
−6 x2 − 6 x3 = 0 (27)
− x1 + x2 − 2 x3 = 0 (28)
x1 + 4 x2 + 7 x3 = 0 (29)
add (28) and (29) we get the following equation:
5 x2 + 5 x3 = 0 (30)
Chapter Eleven 403
∴ x2 = c1 and x3 = −c1
∴ x1 + 4c1 − 7c1 = 0 (31)
∴ x1 = 3c1
x1 3
∴ x2 = c1 1
x − 1
3
Eigen vectors for λ2 = 1
− 3 x1 − 6 x2 − 6 x3 = 0 (32)
− x1 − 2 x2 − 2 x3 = 0 (33)
x1 + 4 x2 + 4 x3 = 0 (34)
add (33) and (34) we get the following equation:
2 x2 + 2 x3 = 0 (35)
∴ x2 = c2 and x3 = −c2
∴ x1 + 4c2 − 4c2 = 0 (36)
∴ x1 = 3c1
x1 0
∴ x2 = c2 1
x − 1
3
Eigen vectors for λ3 = −1
− 5 x1 − 6 x2 − 6 x3 = 0 (37)
− x1 − 4 x2 − 2 x3 = 0 (38)
x1 + 4 x2 + 2 x3 = 0 (39)
404 Simultaneous Linear Differential Equations
Multiply (39) by 3 and add it to (37), we get the following
equation:
3 x1 + 12 x2 + 6 x3 = 0 (40)
Add (40) to (37) we get the following equation:
∴ −2 x1 + 6 x2 = 0 ∴ x1 = 3x2
Let x2 = c3 ∴ x1 = 3x2
∴ 3c3 + 4c3 + 2 x3 = 0 (41)
−7
∴ x3 = c3
2
x1 6
1
∴ x2 = c3 2
x 2 − 7
3
We first add the third row of det (λI − A) to its fourth row, which
permits us to factor λ − 4 out of the new fourth row. In the
remaining determinant, we subtract column 4 from column 3 and
then expand. This gives us the following result:
λ − 3 − 2 −1 −1
1 −6
1
λ −1 − 6
−2 λ−2
[ ]
= (λ − 4 )2 (λ − 2)2 + 1 = 0
2
−1 2 − 2 λ − 6
From the last equation, we see that λ1 = 4 , λ2 = 2 + j1 , and,
λ3 = 2 − j1 are the characteristic values of A.
Using Gauss-Jordan elimination to solve the linear system
(4 I − A)x = 0 , we find without difficulty that the characteristic
0
0
vectors of A corresponding to λ1 = 4 are given by X = c1
1
− 1
where c1 ≠ 0 .
To find the characteristic vectors which correspond to
λ2 = 2 + j1 , we reduce the following matrix to row echelon form
and then write the corresponding linear system, which we find to be
as following:
406 Simultaneous Linear Differential Equations
− 1 + j1 − 2 − 1 −1
1 1 + j1 − 6 −6
[(2 + j1)I − A] =
1 −2 j1 2
−1 2 − 2 − 4 + j1
x1 + x4 = 0
1
x2 + (− 1 + j1)x 4 = 0
2
x3 + + x4 = 0
With x4 = −2c2 , we find from these equations that the
characteristic vectors of A corresponding to λ2 = 2 + j1 are given
by the following equation:
2
− 1 + j1
X = c2 , c2 ≠ 0 .
2
−2
Having found these vectors, we know from Theorem 1 that the
characteristic vectors corresponding to the characteristic value
2
− 1 − j1
λ3 = 2 − j1 of A are the vectors x = c3 , c3 ≠ 0
2
−2
Chapter Eleven 407
11.2 Systems Of Linear First Order Differential Equations
In many problems in applied mathematics there are not one but
several dependent variables, each a function of a single independent
variable, usually time. The formulation of such a problem in
mathematical terms frequently leads to a system of simultaneous
differential equations. Often these equations are nonlinear and
exceedingly difficult, if not impossible, to solve, even with the aid
of a computer. In certain important cases, however, they are linear
in the dependent variables we can solve them easily as following:
The simultaneous system consists of m linear differential
equations in unknown functions x1 , x2 ,....xn as shown in (42). Such
a system is called a linear differential system. A linear differential
equations system (42) is homogeneous if, and only if, all the
functions f1 , f 2 ,.... f m are trivial on I, and the system is
nonhomogenous otherwise.
Li1 x1 + Li 2 x2 + ..............Lin xn = f i , 1 ≤ i ≤ m (42)
A solution of a linear differential system (42) on an interval I is a
vector function which satisfies each equation of the system for every
t in I. A linear differential system is consistent or inconsistent on I
according as it has, or does not have, a solution on I, and one system
is equivalent to another if, and only if, every solution of either
system is a solution of the other.
In particular, the constant coefficient systems we consider will all
be expressible in the following form:
408 Simultaneous Linear Differential Equations
p11 (D )x1 + p12 (D )x2 + ...... + p1n (D )xn = f1 (t )
p21 (D )x1 + p22 (D )x2 + ...... + p2 n (D )xn = f 2 (t )
(43)
...........................................................................
pn1 (D )x1 + pn 2 (D )x2 + ...... + pnn (D )xn = f n (t )
d
Where pij is a polynomial operator in D = , for 1 ≤ i, j ≤ n
dt
To solve a system like this, or to determine its inconsistency, we
try to find elementary operations that will reduce it to an equivalent
system of the following form:
q11 (D )x1 + q12 (D )x2 + ...... + q1n (D )xn = g1 (t )
q22 (D )x2 + ...... + q2n (D )xn = g 2 (t )
(44)
.........................................................................
qnn (D )xn = g n (t )
− x + (D − 4) y = −1 − 4et (47)
(D + 2 )x + (D + 6 )y = 2et (48)
Working with this new system, we operate on both members of (47)
with D + 2 and add the result to (48). Then we simplify the new
second equation and multiply the first equation through by − 1 . This
gives us the following system of equations:
x − ( D − 4 ) y = 1 + 4e t (49)
(D 2 − D − 2)y = −2 − 10et (50)
Of course, this system is equivalent to the one we started with and it
is in the form of (44). Since (50) is an equation in y only, it is a
simple matter to solve for y and we find without difficulty the
following results:
( )
x − (D − 4 ) c1e − t + c2 e 2t + 5et + 1 = 1 + 4et (52)
Solving this equation for x and simplifying, we find the following
equation:
410 Simultaneous Linear Differential Equations
− 2 2 t − 2e
2t
x2 (t ) = e =
1 2 t
e
Satisfy the homogeneous system corresponding to the given system,
which is:
dx dy
+ 2x + + 6 y = 2e t ,
dt dt
dx dy
2 + 3x + 3 + 8 y = −1
dt dt
and that the vector function is as following:
− 11 t − 3 − 11e − 3
t
v(t ) = e + 1 = satisfies the non-
5 t
5e + 1
homogeneous system itself. We shall soon see that this observation
extends to some quite general linear differential systems. Its full
significance will become clear to us once we have become familiar
with linear transformations and linear operator equations.
Chapter Eleven 411
Example 10 Find a complete solution of the differential system
( )( )
∴ D 2 − 3D − 14 c1 cos 2t + c2 sin 2t + c3et − 2 + 4e 2t = 4 + 56e 2t
Performing the indicated differentiations, collecting like terms,
simplifying, and solving for y, we have:
1
∴y =− (3c1 + c2 ) cos 2t + 1 (c1 − 3c2 ) sin 2t − 2 c3et + 1 − 5e 2t
4 4 3
Thus, in terms of the parameters k1 = c1 / 4, k 2 = c2 / 4, and
k3 = c3 / 3 a complete solution of system (59) and (60) is given by
the following:
Chapter Eleven 413
x 4 cos 2t 4 sin 2t
y = k1 − 3 cos 2t + sin 2t + k 2 − cos 2t − 3 sin 2t
(61)
3 −
2 4
+ k 3 e t + + e 2t
− 2 1 − 5
But, (55) and (56) are equivalent systems; hence (61) is also a
complete solution of the original system.
Before considering another example, let us return to system (43)
and observe that if D were a number, instead of an operator, then:
p11 (D ) p12 (D ) .... p1n (D )
p21 (D ) p22 (D ) .... p2 n (D )
(62)
.... .... .... ....
pn1 (D ) pn 2 (D ) .... pnn (D )
would be a determinant. Of course, the polynomial operators pij (D )
2 D 2 + 3D − 9 D 2 + 7 D − 14
following:
D +1 D+2
The expanded form of this determinant is the third-degree
x −y + 4z = (2 − a )et
(D − 1)x + Dy + (D − 2 )z
= aet
0 = (a − 3) et
Of course, this reduced system is equivalent to the given one.
Clearly, unless a = 3 , neither system is consistent.
On the other hand, if a = 3 , we can operate on the first equation
of the reduced system with (− D + 1) , add the result to the second
equation, simplify, and omit the third equation which is simply
0 = 0 , to obtain:
x − y + 4 z = et (63)
(2 D − 1) y − (3D − 2)z = 3et (64)
This system is in the form of (44), and it is equivalent to the original
system. From (64), y can be found in terms of z = z (t ); then, from
(63) x too can be found in terms of z. Since z is subject only to the
restriction that it be differentiable, it may contain any number of
arbitrary constants.
Thus, consistent with the possibilities allowed for by Theorem 1,
we have found that, if a ≠ 3 , the original differential system has no
solution; but, if a = 3 , the system may have solutions containing
any number of arbitrary constants.
416 Simultaneous Linear Differential Equations
11.3 Linear Differential Systems with Constant Coefficients
We shall now consider, as an alternative to the use of elementary
operations, a method of solving linear differential systems with
constant coefficients which resembles the way in which we solved
single linear constant coefficient differential equations. A significant
feature of the method, which we are about to present, is that it puts
to use what we already know about characteristic value problems.
Let us first investigate the method as it applies to first order systems.
As in the case of a single equation, to solve the non-homogeneous
first order system
x′ = Ax + f (t ) (65)
where A is a constant matrix. We begin by looking for a complete
solution of the associated homogeneous system:
x′ = Ax (66)
Guided by our experience in solving single equations, we attempt
to find a solution of (66) of the form
x = ke λt (67)
where k is a constant vector and A is a scalar. Substituting (67)
into (66), dividing out the common factor e λt , and slightly altering
the resulting equation, we see that (67) yields a solution of (66) if,
and only if,
(λI − A)k = 0 (68)
Chapter Eleven 417
Thus, (67) will be a nontrivial solution of (66) if, and only if, λ is a
characteristic value of A and k is a corresponding characteristic
vector. The remaining steps required to find a complementary
function of (65) may differ somewhat according as:
١. All characteristic values of A are real and distinct.
٢. All characteristic values of A are real, but some are
repeated roots of the characteristic equation det (λI − A) = 0
٣. Complex characteristic values of A occur.
Perhaps, the best way of learning how to handle these three cases
is by means of specific examples. Our first example (Example 12)
involves a first order system whose coefficient matrix has only
distinct real characteristic values.
1 − 2
Solution: The coefficient matrix of this system is A = .
2 − 3
a
Since we are seeking solutions of (71) of the type e λt , where
b
a and b are constants, the characteristic value problem to be solved
is:
λ − 1 2 a 0
− 2 λ + 3 b = 0 (72)
and the characteristic equation is
λ − 1 2
− 2 λ + 3 = λ + 2λ + 1 = (λ + 1) = 0
2 2
Hence, λ1 = −1 is a repeated characteristic value of A. Setting
λ = −1 in (72), we obtain the linear system
− 2a + 2b = 0
which is equivalent to a −b = 0
− 2a + 2b = 0
1
Clearly, is a characteristic vector of A corresponding to
1
λ1 = −1 , and
1
X = e −t (73)
1
is a particular solution of (71)
Chapter Eleven 421
Strict analogy with a single differential equation, having -1 as a
double root of its characteristic equation, would suggest that a
second solution of (71) might be found by multiplying (73) by t.
1
But, te − t does not satisfy (73) Furthermore, the Wronskian of
1
this vector function and (73) is a trivial function over all real values
of t. This is no insurmountable problem, however for variation of
parameters, which enabled us to cope with repeated roots of
characteristic equations in the first al place, is still applicable. To
a
apply the method, we take λ = −1, as we must if e λt is to be a
b
solution of (71). Then, we vary the parameters a and b and attempt
to find a solution of (71) of the following type:
u − t
v e (74)
where u and v are functions of t. Of course, we want
e− t ue− t
to be a fundamental matrix of (71).
e− t −t
ve
k1 + k 2 + 2k1t − t
e (77)
k 2 + 2k1t
The Wronskian of this vector function and (73) has the following
value:
k1 + k 2 + 2k1t 1 − 2t
e = k1e − 2t ≠ 0
k 2 + 2k1t 1
Hence, (73) and (77) are fundamental solutions of (71), provided
k1 ≠ 0 . In particular, if k1 = 1 and k 2 = 0 , (73) and (77) give us:
e − t (1 + 2t )e − t
e − t −t
2te
as a fundamental matrix, and
1 (1 + 2t ) − t
X = c1 e − t + c2 e
1
2t
as a complete solution of the original differential system.
Chapter Eleven 423
Looking back on (63) and (64), a shorter method of finding a
fundamental matrix of (71) comes to our attention. As soon as (73)
has been found, determine constants a, b, c , and d such that:
a − t c − t (a + ct )e
−t
b e + d te =
(b + dt )e
− t
2 −3
characteristic values of the coefficient matrix A = .
3 2
Replacing λ by 2 + j 3 in the characteristic value problem (80), and
dividing out a factor of 3 from every term, we get the linear system
ja + b = 0
− a + jb = 0
which is obviously equivalent to ja + b = 0
Taking a equal to the coefficient of b, and b equal to minus the
1
coefficient of a, in the last equation, we obtain as a
− j
characteristic vector of A corresponding to the characteristic value
λ1 = 2 + j 3 .
Substituting these values of a, b, and λ into (79), gives us the
complex vector function
1 (2 + j 3)t
− e (81)
j
which formally satisfies system (78). Of course, what we want are
real fundamental solutions of (78). To find these, we use Euler's
formula to write (81) as:
2t j 3t e 2t cos 3t + j sin 3t
j 3t
1 2t
− e e = e = − j (cos 3t + j sin 3t ) e
j j
− je 3t
Chapter Eleven 425
2t j 3t e cos 3t e 2t sin 3t
2t
1
∴ e e = + j
− j 2 t
e sin 3t 2 t
− e cos 3t
It is now easy to verify that the real vector functions:
cos 3t 2t sin 3t 2t
sin 3t e and − cos 3t e
Form a basis for all real solutions of system (78). Thus, we are
able to write a complete solution:
cos 3t 2t sin 3t 2t
X = c1 e + c2 − cos 3t e of the given differential
sin 3t
system using only one of the two conjugate characteristic values of
A.
λ+2 13
det (λI − A) = =0
−1 λ −4
(λ + 2 )(λ − 4 ) + 13 = 0
λ2 − 2λ + 5 = 0 ∴ λ1,2 = 1 ± j 2
For λ1 = 1 + j 2
426 Simultaneous Linear Differential Equations
3 + j 2 13 a 0
−1 − 3 + j 2 b = 0
(3 + j 2 )a + 13b = 0
(3 + j 2)a = −13b
Let a = C1 .ﺧﻄﺄ! اﻹﺷﺎرة اﻟﻤﺮﺟﻌﻴﺔ ﻏﻴﺮ ﻣﻌﺮّﻓﺔ
1
λt 1 13 1+ j 2
X 1 = C1 − 1 e = C
− 3 − j 2 e
(3 + j 2 ) 13
1
13
13 1+ j 2
= c1 e
− 3 − j2
13 0
∴ X 1 = c1e t cos 2t − sin 2t
− 3 − 2
0 13
∴ X 2 = c2 et cos 2t + sin 2t
− 2 − 3
13 0
∴ X = X 1 + X 2 = c1et cos 2t − sin 2t
− 3 − 2
0 13
+ c2 et cos 2t + sin 2t
− 2 − 3
For λ1 = 1 + j 2
2 + j 2 4 a 0
−2 − 2 + j 2 b = 0
(2 + J 2)a + 4b = 0 and − 2a + (− 2 + j 2)b = 0
∴ a = (− 1 + j1)b
Let b = c1 ∴ a = (− 1 + j1)c1
x1 a
= c1 e (−1+ j 2 )t
x2 b
a − 1 + j1
= c1
b 1
− 1 1
θ = , and, φ =
1 0
− 1 1
∴ X 1 = c1e − t cos 2t − sin 2t
1 0
1 − 1
∴ X 2 = c2 e − t cos 2t + sin 2t
0 1
428 Simultaneous Linear Differential Equations
− cos 2t − sin 2t
∴ X = X 1 + X 2 = c1e − t
cos 2t
cos 2t − sin 2t
+ c2 e − t
sin 2t
λ +3 2
det (λI − A) = =0
−2 λ −1
(λ + 3)(λ − 1) + 4 = 0
λ2 + 2λ + 1 = 0
∴ λ1 = λ2 = −1
Eigen values For λ1 = −1
2a + 2b = 0
∴ a = −b
Let a = c1 , and b = − c1
x 1
∴ 1 = c1 e − t
x2 − 1
Chapter Eleven 429
Eigen values For λ2 = −1
x u
∴ 1 = e − t
x2 v
− ue − t = u ′e − t = (− 3u − 2v )e − t
∴ u ′ = −2(u + v )
− ve − t + v′e − t = (2u + v )e − t
v′ = 2(u + v )
Let u + v = c2
u ′ = −2c2 , ∴ u = −2c2t + c3
v = c2 − u = c2 + 2c2t − c3
− 2c2t + c3 1
W = = 2c2t − c3 − (c2 + 2c2t − c3 ) = 0
c2 + 2c2t − c3 − 1
c3 = 0
∴ u = −2c2t , and v = 2c2t + c2
1 − 2t − t
∴ X = c1 e − t + c2 e
−
1 2t + 1
(− 7 + j1)a + 2b = 0
−1
Let a = c1 ∴b = c1 (− 7 + j1)
2
1
a 1 2
= c1 7 1 = c1
b
− j 2 7 − j1
2 2
2 0
θ = , and φ=
7 − 1
2 0
X 1 = c1e − 3t cos t − sin t
7 − 1
0 2
X 2 = c2 e − 3t cos t + sin t
− 1 7
2 cos t − 3t 2 sin t
∴ X = c1e − 3t + c 2 e
7 cos t + sin t − cos t + 7 sin t
Chapter Eleven 431
10 2c
Q X (0 ) = = 1
40 7c1 − c2
∴ 2c1 = 10 and, 7c1 − c2 = 40
∴ c1 = 5 and, c2 − 5
x cos t − sin t
∴ X = 1 = 10e − 3t
x
2 4 cos t − 3 sin t
x1 1
∴ X 1 = x2 = c1 0 et
x 2
3
Eigen values for λ2 = −1
− 4a +c = 0
2a −3b −c = 0
− 8a +2c = 0
Let a = C2 ∴ c = 4C2
Substitute values of a and c into the second equation we get the
following:
3b = 2C2 − 4C2
−2
∴b = C2
3
x1 3
∴ X 2 = x2 = c2 − 2 e − t
x 12
3
Eigen values for λ3 = 2
−a +c = 0
2a −c = 0
− 8a +5c = 0
∴a = c = 0
Let b = c3
Chapter Eleven 433
x1 0
∴ X 2 = x2 = c3 1 e 2t
x 0
3
x1 1 3 0
t −t
∴ X = x2 = c1 0 e + c2 − 2 e + c3 1 e 2t
x 2 12 0
3
− 1
Q x(0) = 2
− 8
− 1 1 3 0
∴ 2 == c1 0 + c2 − 2 + c3 1
− 8 2 12 0
∴ −1 = c1 + 3c2 (82)
2= −2c2 + c3 (83)
x1 2 3
t −t
∴ X = x 2 = 0 e − − 2 e
x 4 12
3
It is not always necessary, nor advantageous, to reduce a linear
differential system with constant coefficients
434 Simultaneous Linear Differential Equations
pi1 (D )x1 + pi 2 (D )x2 + .... + pin (D )xn = f i (t ) (85)
where 1 ≤ i ≤ n
to a first-order system before solving. For a system like (85) can
usually be solved in the same manner as a first-order system. The
concepts of complementary function and particular integral carry
over and, can even be extended to far more general linear systems.
If we define a matrix operator P(D ) , and vector functions x and f
by:
p11 (D ) p12 (D ).... p1n (D )
p (D ) p ( D ).... p ( D )
P(D ) = 21 22 2 n
..............................................
pn1 (D ) pn 2 (D ).... pnn (D )
x1 f1
x f
2 2
x = .. f = ...
.. ...
xn f n
x = ke λt
( )
just like we did for first order systems. Since D r e λt = λr e λt , it
(D + 1)x + (D + 2) y + (D + 3)z = −e − t
(D + 2)x + (D + 3) y + (2 D + 3)z = e−t (90)
(4 D + 6)x + (5D + 4) y + (20 D − 12)z = 7e − t
To find a complementary function of this system, we observe that
a
solutions b e λt of the homogeneous system exist if, and only if,
c
(λ + 1)a + (λ + 2)b + (λ + 3)c =0
(λ + 2)a + (λ + 3)b + (2λ + 3)c =0 (91)
(4λ + 6)a + (5λ + 4)b + (20λ − 12)c =0
The characteristic equation of this system is:
(λ + 1) (λ + 2) (λ + 3)
(λ + 2) (λ + 3) (2λ + 3) =0
(4λ + 6) (5λ + 4) (20λ − 12)
or, expanding, collecting terms, and factoring,
− (λ − 1)(λ − 2 )(λ − 3) = 0
Substituting the roots λ1 = 1, λ2 = 2, and λ3 = 3 of this equation
into (52), one at a time, we obtain the three linear systems
2a + 3b + 4c = 0
3a + 4b + 5c = 0
10a + 9b + 8c = 0
3a + 4b + 5c = 0
4a + 5b + 7c = 0
14a + 14b + 28c = 0
Chapter Eleven 437
4a + 5b + 6c = 0
5a + 6b + 9c = 0
18a + 19b + 48c = 0
When the coefficient matrices of these systems are reduced to
row echelon form the systems themselves are transformed,
respectively, into
a−c =0 a + 3c = 0 a + 9c = 0
, ,
b + 2c = 0 b−c =0 b − 6c = 0
From these systems we see that (1,-2,1), (3,.-1,-1), and (9,-6,-1)
are nontrivial solutions of (91) corresponding to the distinct real
values 1, 2, and 3, respectively, of the parameter λ . A
complementary function of (90) is therefore
1 3 9
t
c1 − 2 e + c2 − 1 e + c3 − 6 e3t
2t
(92)
1 − 1 − 1
To complete the problem we now need to find a particular
a
v = b e − t (93)
c
Substituting (93) into (90), collecting terms, and canceling a
factor of e − t , we find:
b + 2c = −1
a+ 2b + c = 1
2a − b − 32c = 7
which is equivalent to the following:
a=3
b = −1
c=0
3
Hence, a particular integral of (51) is − 1 e − t and a complete
0
solution of the original system is:
1 3 9 3
x = c1 − 2 et + c2 − 1 e 2t + c3 − 6 e3t + − 1 e − t
1 1 − 1 0
As with first order systems and single equations, if the set of roots
{λ j } of the characteristic equation (89) includes one or more pairs
of conjugate complex numbers, it is desirable to convert all
corresponding complex exponential solutions into purely real
solutions. To see how this can be accomplished, let p ± jq be a pair
of conjugate complex roots of (89), and let k be a particular
Chapter Eleven 439
nontrivial solution vector of (88) corresponding to the root
λ = p + jq , so that P(λ )k = p( p + jq )k = 0
Then, since all the coefficients in (88) are real, it follows by
taking conjugates throughout the last equation that k is a solution
vector of (88) corresponding to λ = p − jq Therefore, in a complex
ke ( p + jq )t + k e ( p − jq )t k +k k −k
= e pt cos qt − sin qt
2 2 j2
= e pt [θ (k ) cos qt − φ (k ) sin qt ] (94a)
ke ( p + jq )t − k e ( p − jq )t k −k k −k
= e pt cos qt + sin qt
j2 j2 2
= e pt [φ (k ) cos qt + θ (k ) sin qt ] (94b)
where .θ (k ) and φ (k ) denote the column vectors whose
components are, respectively, the real parts of the components of k
and the imaginary parts of the components of k.
If p(λ ) = 0 has a double root, say λ = r , and if a is a specific
b1te rt + b2 e rt (95)
ae rt
b1te rt + b2 e rt
c1t 2 e rt + c2te rt + c3e rt (96)
.........................................................
k1t m −1e rt + k 2t m − 2 e rt + .......k m −1te rt + k m e rt
Chapter Eleven 441
Example 21 Find a complete solution of the system
(D 2 + D + 8)x1 + (D 2 + 6D + 3)x2 = 0
(D + 1)x1 + (D 2 + 1)x2 = 0
(97)
λ2 + λ + 8 λ2 + 6λ + 3
2
= λ4 + 2λ2 − 8λ + 5 = 0 (98)
λ +1 λ +1
with roots 1, 1, − 1 ± j 2 . For the root -1 + j2, (98) simplifies to:
1 + j 2
Taking k1 = 1 + j 2 and k 2 = j , we have
j
1 2
θ (k ) = and φ (k ) = .
0
1
Thus, from (94), two particular solutions of the given differential
system are:
1 2
x1 = e − t cos 2t − sin 2t
0 1
2 1
x2 = e − t cos 2t + sin 2t
1 0
a1 1
so we can take a = =
a2 − 1
0
integral, corresponding to the first term e − 2t of f (t ) , Simply
1
a1
v1 = ae − 2t = e − 2t
a2
where a1 , a2 constants
Then, substituting, we have:
2
equation P(D )x = et and using the following equations:
2
( )
If D m te λt = λm te λt + mλm −1e λt
( )
If D m t 2 e λt = λm t 2 e λt + 2mλm −1te λt + m(m − 1)λm − 2 e λt
Hence
(D + 1)x + (D + 5) y = 2 (105)
Solution: Subtract (105) from (104) we get the following
equation:
x − y = −1 (106)
Multiply (3) by (D + 1)
Chapter Eleven 449
∴ (D + 1)x − (D + 1) y = −(D + 1)
∴ (D + 1)x − (D + 1) y = −1 (107)
Subtract (107) from (105) we get the following equation:
∴ (2 D + 6 ) y = 3
dy
2 + 6y = 3
dt
This is first order ordinary differential equation. The solution of this
equation is as following:
1
y (t ) = + c1e − 3t
2
Substitute y (t ) into (106)
∴ x − y = −1
−1
∴x = + c1e − 3t
2
x 1 − 1 1
∴ = + c1 e − 3t
y 2 1 1
(D + 2 )x + (5D − 2 )y = 2t − 1 (109)
Solution: Subtract (109) from (108) we get the following
equation:
− x + (− Dy ) = −t
450 Simultaneous Linear Differential Equations
− x − Dy = −t (110)
( )
∴ −(D + 1)x + − D 2 − D y = −1 − t (111)
Add (111) to (108) we get the following equation:
(− D 2 + 3D − 2)y = −2
(D 2 − 3D + 2)y = 2
It is second order ordinary differential equation.
∴ y h (t ) = c1e t + c2 e 2t
∴ y p (t ) = 1
y (t ) = y h + y p = c1et + c2 e 2t + 1
− x − c1et − 2c2 e 2t = −t
∴ x = t − c1e t − 2c2 e 2t
x − 1 − 2 t
= c1 et + c2 e 2t +
y 1 1 1
The above example can be solved by another technique as
following:
λ + 1 4λ − 2
λ + 2 5λ − 2 = 0
Chapter Eleven 451
∴ λ2 − 3λ + 2 = 0
∴ λ1 = 1 , and λ2 = 2
For λ1 = 1
∴ a = −b
Let b = c1 , a = −c1
x − 1
∴ X 1 = = c1 et
y 1
For λ2 = 2
3a + 6b = 0
∴ a = −2b
Let b = c2 , a = −2c2
x − 2
∴ X 2 = = c2 e 2t
y 1
t − 1 0t
Q f (t ) = e
2t − 1
∴λ = 0
a − 2b = t − 1 (112)
2a − 2b = 2t − 1 (113)
Subtract (112) from (113) we get a = t ∴b = 1 / 2
x − 1 − 2 t
= c1 et + c2 e 2t +
y 1 1 1 / 2
452 Simultaneous Linear Differential Equations
Example 25 Find a complete solution the following system:
∴ λ3 + 6λ2 + 11λ + 6 = 0
λ1 = −1 , λ2 = −2 , and λ3 = −3
For λ1 = −1
9a + 2b − 3c = 0
− 3a − 2b − c = 0
−4b − 6c = 0
−3 2
Let c = c1 ∴b = c1 ∴a = c1
2 3
x 4
1 −t
∴ X 1 = y = c1 − 9 e
z 6 6
For λ2 = −2
7 a + b − 4c = 0
− 4a − 3b − 2c = 0
− a − 5b − 8c = 0
Chapter Eleven 453
After simplification of the above equation we get the following:
30 14
Let c = c2 ∴ b = − c2 and ∴ a = c2
17 17
x 14
1
∴ X 2 = y = c2 − 30
z 17 17
For λ3 = −3
5a − 5c = 0
Let a = c = c3
− 5a − 4b − 3c = 0
∴ b = −2c3
x 1
X 3 = y = c3 − 2 e − 3t
z 1
14e t
Q f (t ) = − 2et
4e
t
∴λ = 1
13a + 4b − c = 14
−a + c = −2
2a − 2b − 2c = 4
After simplification of the above equation we get the following:
a = 1, b = 0 , and c = −1
454 Simultaneous Linear Differential Equations
x 1
∴ X 4 = y = 0 e t
z − 1
X = X1 + X 2 + X 3 + X 4
Multiply (3) by (D + 2) ).
( )
∴ − 3D 2 − 8D − 4 x − (D + 2 ) y = −7 sin t − cos t (120)
Add (118) and (120).
( )
∴ − 3D 2 − 6 D − 3 x = −6 sin t − cos t
xh (t ) = (c1 + c2t )e − t
∴ x p (t ) = A cos t + B sin t
[
∴ y (t ) = 3 sin t − cos t − 3 − (c1 + c2t )e − t + c2 e − t + 1 / 6 cos t + sin t ]
[
−2 (c1 + c2t )e − t + 1 / 6 sin t − cos t ]
− sin t cos t
∴ y (t ) = + + (c1 + c2t )e − t − 3c2 e − t
3 2
456 Simultaneous Linear Differential Equations
Problems:
1- Find the characteristic values and the corresponding
characteristic vectors of each of the following matrix:
− 4 5 5
− 5 6 5
− 5 5 6
2- Find a complete solution of each of the following first order
systems. Also fiend the solution that satisfies the given condition
when one is given
x1′ = 3x1 + 2 x2
a)
x2′ = − 2 x1 − x2
−3
x1′ = x1 − 2 x2
b) 2
5
x2′ = 2 x1 + x2
2
x1′ = 3 x1 − 2 x2
c)
x2′ = 5 x1 − 3 x2
x1′ = 3 x1 + 2 x2
d)
x2′ = − 2 x1 − x2
x1′ = x1 − x2
e)
x2′ = x1 + 3 x2
x1′ = −2 x1 + x2
X (0 ) =
3
f)
x2′ = − 3x1 + 2 x2 + 2 sin t 4
Chapter Eleven 457
x1′ = 3 x1 − 2 x3 3
g) x2′ = − x1 + 2 x2 + x3 , X (0 ) = 1
3
x3′ = 4 x1 − 3 x3
x1′ = 4 x1 − 2 x2 − 10 x3 − 3
h) x2′ = x2 X (0 ) = − 2
x3′ = 2 x1 − x2 − 5 x3 − 1
x1′ = x1 − 2 x2 − x3
i) x2′ = x1 + 2 x2 + e − t
x3′ = − x1 − 3x2 − 1
b)
(2 D + 1)x + (D − 1) y = −3 cos t
(D + 2)x + (D + 3) y = 5 sin t
c)
(2 D + 1)x + (D + 2) y = 6et
(D + 2)x + (D + 4) y = 4e −t
d) (D + 5)x + (D + 7 ) y = 4e
2t
(2 D + 1)x + (3D + 1) y = 0
(2 D + 1)x + (D 2 + 6 D + 1)y = 0
(D + 2)x + (D 2 + 2 D + 5)y = 6 2t
e)
458 Simultaneous Linear Differential Equations
h) (D + 5)x + (2 D + 1) y = −e − t + et
(D + 7 )x + (3D + 1) y = 0
i)
(D + 5)x + (D + 7 ) y = 2et
(2 D + 1)x + (3D + 1) y = et
j) (2 D + 1)x + (D + 2) y = e −t
(3D − 7 )x + (3D + 1) y = 0
(2 D + 1)x + (D + 2) y = 8e −t
k)
(D 2 + D + 9)x + (D 2 − 2D + 12)y = 6
l)
(3D + 1)x + (D + 7 ) y = e − t
(2 D + 1)x + (D + 5) y = e − t
m)
(D + 1)x + (D + 2) y = −et
(3D + 1)x + (4 D + 7 ) y = −7et
n)
(D + 2)x + (D + 3) y = −4
(2 D − 6)x + (3D − 4) y = 2
o)
(D + 1)x + (D + 2) y = −t + 1
(5D + 1)x + (6 D + 3) y = −2t + 1
Chapter 12
Special Functions
Γ (7 )
Example 3 Find
2Γ(4)
Solution
Γ (7 ) 6!
= = 60
2Γ(4) 2 * 3!
q Γ(q) q Γ(q)
1.00 1.0000 1.55 0.8889
1.05 0.9735 1.60 0.8935
1.10 0.9514 1.65 0.9001
1.15 0.9330 1.70 0.9086
1.20 0.9182 1.75 0.9191
1.25 0.9064 1.80 0.9314
1.30 0.8975 1.85 0.9456
1.35 0.8912 1.90 0.9618
1.40 0.8873 1.95 0.9799
1.45 0.8857 2.00 1.0000
1.50 0.8863
462 Special Functions
1 π (2n )!
Γ n + = 2 n * (10)
2 2 n!
Chapter Twelve 463
Proof:
1 1 3 5 3 1 1
Γ n + = n − n − n − ..... * Γ
2 2 2 2 2 2 2
2n − 1 2n − 3 2n − 5 3 1
= ...... * * π
2 2 2 2 2
π
= n
(2n − 1)(2n − 3)(2n − 5).....3 *1
2
π (2n )(2n − 1)(2n − 2 )(2n − 3)(2n − 4)(2n − 5).....3 * 1
=
2n (2n )(2n − 2)(2n − 4)......4 * 2
π (2n )(2n − 1)(2n − 2 )(2n − 3)(2n − 4 )(2n − 5).....3 *1
=
22n (n )(n − 1)(n − 2)......2 *1
π (2n )!
=
2 2n (n )!
1 π (2n )!
Q Γ n + = 2 n
2 2 (n )!
1 π (6)! 15 π
∴ Γ(3.5) = Γ 3 + = 6 =
2 2 (3)! 8
∞
6 −2x
Example 10 Evaluate the following integral ∫x e dx
0
Solution: Let 2 x = y , ∴2dx = dy
Substitute in the required integral we get:
∞ ∞ 6 ∞
y dy 1 1 6! 45
∫x
6 −2x
e dx = ∫ e− y = 7 ∫ ( y )6 e− y dy = 7 Γ(7) = 7 − =
0
0
2 2 2 0 2 2 8
∞
45
∴ ∫ x 6 e − 2 x dx =
0
8
∞ 4
−x
Example 11 Evaluate the following integral ∫ e dx
0
Solution:
1
Let t = x 4 , ∴ x = t1 / 4 , ∴ dx = t − 3 / 4 dt
4
Substitute in the integral, we get
1
∞ ∞ Γ 1 +
−x 4 1 1 1 1 4
∫e dx = ∫ t − 3 / 4e − t dt = Γ =
40 4 4 4 1
0
4
= Γ(1.25) = 0.96785
∞ 4
∴ ∫ e − x dx = 0.96785
0
Chapter Twelve 465
Example 12 Evaluate the following integral
1
∫ − Ln( x ) dx
0
Solution:
Assume − Ln( x ) = t , ∴ x = e − t , dx = −e − t dt
x = 0 ⇒t = ∞
When,
x =1 ⇒t = 0
Substitute in the above integral we get:
( ) dt = ∫ (t )1 / 2 (et ) dt
1 0 ∞
∫ − Ln( x ) dx = ∫ t −e −t
0 ∞ 0
1 1 π
= Γ(3 / 2 ) = Γ =
2 2 2
Solution:
Assume u = x
∴ x = u2
∴ dx = 2udu
Then at x = 0, u = 0 and
x=∞ u=∞
466 Special Functions
∫ (u )
∞ ∞ ∞
2 1 / 4 −u
∴∫ x e 4 − x
dx = e (2udu ) = 2 ∫ u 3 / 2e − u du
0 0 0
Solution:
Assume x 3 = u ∴ x = u1 / 3
1
∴ dx = u − 2 / 3du
3
Substitute in the above integral we get the following:
∞ ∞
− x3 1
∴ ∫e dx = ∫ e − u u − 2 / 3du
0 0 3
∞
− x3 1 ∞ − 2 / 3 −u
∴ ∫e dx = ∫ u e du
0
30
Chapter Twelve 467
It is clear the above integral in the form of Gamma function, where,
q − 1 = −2 / 3 , ∴ q = 1 / 3
∞
1 1 Γ(4 / 3)
∴ ∫ e − x dx = Γ(1 / 3) =
3
= 0.8933
0
3 3 1 / 3
Solution:
−x
It is clear that 3− x = e ln 3 = e − x ln 3 = e − x e ln 3 = 3e − x
∞ ∞
−x 3
∴ ∫3 x dx = 3 ∫ e − x x 3dx
0 0
Solution:
468 Special Functions
Assume e − x = u
∴ −e − x dx = du
du − du
∴ dx = =
− e− x u
At x = ∞ u = 0 and,
At x = −∞ u = ∞ and,
∞ ∞
(x −e )
−x
0
− ln (u )− u − du − ln (u ) − u du
∴ ∫ e dx = e ∫ = e
u
∫ e
u
−∞ ∞ 0
∞ ∞
(x −e )
−x
dx = u − 2 e − u du
∴ ∫ e ∫
−∞ 0
Similarly,
∞ 2
Γ(n) = 2 ∫ y 2 n −1e − y dy (15)
0
∞ 2 m + 2 n − 2 − r 2
∴ Γ(m) Γ(n) = ∫ (r ) e 2r dr
0
π /2
*2 ∫ (cos θ )2 m −1 (sin θ )2 n −1 dθ
0
∞ 2( m + n −1) − r 2
∴ Γ(m) Γ(n) = ∫ (r ) e dr 2
0
π /2
*2 ∫ (cos θ )2 m −1 (sin θ )2 n −1 dθ
0
∞ ( m + n −1) − t
Γ ( m) Γ ( n) = ∫ t e dt
0
π /2
*2 ∫ (cos θ )2 m −1 (sin θ )2 n −1 dθ
0
∴ Γ(m) Γ(n) = Γ(m + n) * B(m, n)
Γ ( m) Γ ( n)
∴ B ( m, n ) =
Γ ( m + n)
Chapter Twelve 471
For all m and n we have:
B (m, n) = B (n, m) (16)
Proof:
Substitute in (1) by
1 − x = t , Then x = 1 − t , dx = − dt
x = 0 ⇒ t =1
x =1 ⇒ t = 0
So we have
1 0
B ( m, n ) = ∫ x m −1
(1 − x )
n −1
dx = ∫ (1 − t )m −1 (t )n −1 (− dt )
0 1
1
= ∫ (t )n −1 (1 − t )m −1 (dt ) = B (n, m)
0
∴ B(m, n) = B(n, m)
Solution:
1
Γ(5) * Γ(4) 4!*3! 1
∫ x (1 − x ) dx = B(5,4) = Γ(9) = 8! = 280
4 3
0
472 Special Functions
0
Solution
4 4 1/ 2
2t
∫ t (4 − t )
2 1/ 2
dt = 2∫ t 1 − dt
0
0
4
Assume x = t / 4 the t = 4 x, and dt = 4dx
t =0⇒ x=0
t = 4 ⇒ x =1
4 4
∴ ∫ t (4 − t )
2 1/ 2
dt = 2 ∫ (4 x )2 (1 − x )1 / 2 4dx
0 0
4
= 2 * 4 * 4 ∫ ( x )2 (1 − x )1 / 2 dx
2
0
3 Γ(3) * Γ(3 / 2)
= 2 7 * B 3, = 2 7
2 Γ(9 / 2)
2!*[1 / 2 * Γ(1 / 2)]
= 27 = 19.505
[7 / 2 * 5 / 2 * 3 / 2 * 1 / 2 * Γ(1 / 2)]
(8 − x3 ) dx
2
3
∫x
0
Solution:
(8 − x ) x 3
2 2
∫x
3 3
dx = ∫ 2 x1 − dx
2
0 0
Chapter Twelve 473
3
x
Assume that = u , then x = 2 u1 / 3
2
∴ x = 0, u = 0 And x = 2, u = 1
1
(
∴ I = 2 * ∫ 2u1 / 3 (1 − u )
0
) 1/ 3 2
u
3
−2 / 3
du
I=
3 ∫0
(
8 1 −1 / 3
u (1 − u))
1/ 3
du
2 4
Γ Γ
8 2 4 8 3 3
∴ I = B , =
3 3 3 3 Γ(2 )
4
∴ Γ = Γ(1.333)
3
0.8975
0.0063
x
0.8
912
0.02
x 0.02
from the figure shown above =
0.0063 0.05
∴ x = 0.0021
∴ Γ(1.33) = 0.8912 + x = 0.8912 + 0.0021 = 0.8933
The same way
5
Γ = Γ(1.6667 ) = 0.903
3
Also,
2 3 5 3
Γ = Γ = * 0.903 = 1.3545
3 2 3 2
(8 − x3 ) dx = 83 * 0.89331*1.3545 = 3.2266
2
∴∫x 3
0
( )
1
∫ x 3 1 − x 2 dx
0
Solution:
1 −1 / 2
Assume x 2 = u , ∴ x = u and, ∴ dx = u du
2
(1 − x )
1
1 1 −1 / 4
∴∫ x3 2
dx = ∫ u (1 − u )1 / 3 du
0
20
Chapter Twelve 475
It is clear that the above integral is in the form of Beta function,
where,
m − 1 = −1 / 4 and n − 1 = 1 / 3
∴ m = 3 / 4 and n = 4 / 3
( ) 1 3 4 1 Γ(3 / 4 ) * Γ(4 / 3)
1
∴ ∫ x 3 1 − x 2 dx = B , = = 0.527
0
2 4 3 2 Γ(25 / 12 )
Solution:
4 4
Assume u , ∴ x2 = −4
4 + x2 u
1/ 2
4
∴ x = − 4
u
−1 / 2
−2 4
∴ dx = −2u − 4 du
u
At x = 0 , u = 1 , and,
At x = ∞ , u=0
1 / 2 −1 / 2 −1 / 2
( )
∞ 0
dx 4 u −2 4
∴∫ = ∫ − 4
0 (
x 4 + x2 ) 1
u
− 2u − 4
4 u
du
476 Special Functions
∞ −3 / 4 −3 / 4
dx 2 1 −1 4 1 1 −1 4
∴∫ = ∫ u − 4 du = ∫ u (1 − u )
0 (
x 4 + x2 ) 40 u 20 u
du
∞ −3 / 4 −3 / 4
dx 21 4 11 4
∴∫ = ∫ u −1 − 4 du = ∫ u −1 (1 − u )
0 (
x 4+ x 2
) 40 u 20 u
du
∞
dx 1 − 3 / 4 1 −1 3 / 4
∴∫ ∫ u u (1 − u )− 3 / 4 du
0 (
x 4 + x2 ) = 4
2 0
∞ 1
dx 1
∴∫ = ∫u
−1 / 4
(1 − u )− 3 / 4 du
0 (
x 4+ x 2
) 2*4 3/ 4
0
( )
1 0 ∞
1
∴ ∫ Ln dx = ∫ u − eu du = ∫ ue − u du
0 x ∞ 0
Problems
(I) Evaluate the following Gamma functions
Γ(− 2 )
Γ(6.4 ) , Γ(− 4.3) , Γ(0.7 ), Γ(− 2 ),
Γ(4.5) Γ(2.3)
(II) Evaluate the following integrals
(1 − x ) dx
1 1 ∞
dx 2 q −x
1) ∫ 2) ∫ x3 3) ∫x q dx
q
0 1− x 0 0
Chapter 13
Numerical Analysis.
− b + b 2 − 4ac − b − b 2 − 4ac
x1 , x2
2a 2a
It is easy likely that you have seen the more intricate formula for the
solution of cubic and quadratic polynomial equations. But
unfortunately there is no any closed formula for finding roots for 5th
order polynomial or any nonlinear equations.
Our goal is to develop simple numerical methods for the
approximate solution of the equation f ( x ) = 0 . Methods of the kind
Chapter Thirteen 479
discussed here are iterative in nature and produce sequences of real
numbers, which in favorable circumstances, converge to the
required solution.
xn +1 = g ( xn ) (3)
x2 + 2
rearranged to be like this: x =
3
If we chose x0 = 0 , then we can do the following iterations:
x02 + 2
x1 = = 0.666 666 666
3
480 Numerical Analysis
x12 + 2
x2 = = 0.814 814 815
3
x22 + 2
x3 = = 0.887 974 394
3
x32 + 2
x4 = = 0.929 499 508
3
and so on till
x92 + 2
x10 = = 0.994 366 675
3
The results are very clear, where are going to the first root x = 1 .
To get the other root, assume x0 = 5 .
x02 + 2
∴ x1 = = 9.0
3
x12 + 2
∴ x2 = = 27.666 666 667
3
x22 + 2
∴ x3 = = 255.814 815
3
It is clear that, these results diverge and we can not get the solution.
If we use another value for x0 to be between 1 and 2 the results will
converge to the same root x = 1 . Then we have to try to rearrange
our equation in different manner as following:
x = 3x − 2
Chapter Thirteen 481
2
It is clear, we can not assume xo < because of the root will give
3
us imaginary part. So, chose x0 = 1.5
∴ x4 = 3 x3 − 2 = 1.780 229 56
and so on till
∴ x10 = 3 x9 − 2 = 1.954 534 710
It is very clear that the results going to the second solution x = 2 .
Let us see if we choose x0 = 3 in the previous example
∴ x1 = 3 x0 − 2 = 2.645 751 31
∴ x2 = 3 x1 − 2 = 2.436 648 09
x2 + 2
Q x = g (x ) =
3
2x
g ′( x ) =
3
The assumption x0 = 0 gives g ′( x0 ) < 1 then this assumption
sign of f (c0 ) consides either with the sign of f (a0 ) or with the
sign of f (b0 ) . Thus at the end points of one of the two intervals
484 Numerical Analysis
[a0 , c0 ] or [c0 , b0 ] the function f (x ) has the same signs and at
the end points of the other opposite signs. We retain the interval
at the end points of which f ( x ) has opposite signs and reject the
other interval since it does not contain the required root. We
denote the retained interval by [a1 , b1 ] , where
a = a0 a1 a3 x
b3
a2 b2 b = b0
b1
bisection method: f ( x ) = x 2 − 3 x + 2 = 0
Solution: If we look to the previous example
f (x ) = x 2 − 3x + 2 = 0
Assume a0 = 0 , and, b0 = 1.5
bn − an
.
2
Line joining (a, f (a )), (b, f (b )) :
x−a
y = f (a ) + ( f (b ) − f (a )) (9)
b−a
b−a af (b ) − bf (a )
y = 0 ⇒ x1 = a − f (a ) =
f (b ) − f (a ) f (b ) − f (a )
af (b ) − bf (a )
∴ c= (10)
f (b ) − f (a )
The following table (Table(2)) shows a comparison between
Bisection and False Position Methods.
impossible.
(xn − xn −1 )
≤ ε 2 but this may take a long time.
xn
How fast does it converge? In general, False position may be better,
but there are exceptions. For locally convex functions, could be
slower than the bisection method. See Fig.3 picture of locally
convex function.
Illinois method: f ( x ) = x 2 − 3 x + 2 = 0
Chapter Thirteen 493
Solution: Applying the Illinois methodology to the above equation
we get the result shown in Table(4):
Table(4) The result of Example 4.
f ( x1 ) + f ′( x1 )( x − x1 ) = 0
f ( x1 )
∴ x = x1 − this is the root of F ( x ) = 0
f ′( x1 )
f ( x1 )
Regard it as a good approximation to x* . So, let ∴ x2 = x1 −
f ′( x1 )
Repeating the process, we have
f ( xn )
∴ xn +1 = xn − (11)
f ′( xn )
496 Numerical Analysis
Notes regarding Newton’s Method:
• Need only one initial guess, whereas bisection needs a and b.
• Need to compute the derivative f ′( x )
Newton’s method: f ( x ) = x 2 − 3 x + 2 = 0
Solution:
Applying the Newton’s methodology to the above equation we get
the result shown in Table(5):
Table(5) The result of Example 5.
i xn f(xn) f '(xn)
0 0.000000000 2.000000000 -3.000000000
1 0.666666667 0.444444444 -1.666666667
2 0.933333333 0.071111111 -1.133333333
3 0.996078431 0.003936947 -1.007843137
4 0.999984741 0.000015259 -1.000030518
5 1.000000000 0.000000000 -1.000000000
Chapter Thirteen 497
13.1.7 Secant Method
In Newton’s method, f ′( x ) is needed. But f ′( x ) may be difficult to
Secant’s method: f ( x ) = x 2 − 3 x + 2 = 0
Solution: Applying the Secant’s methodology to the above
equation we get the result shown in Table(6):
Chapter Thirteen 499
Table(6) The result of Example 6.
i xn f(xn)
0 0.000000000 2.000000000
1 0.500000000 0.750000000
2 0.800000000 0.240000000
3 0.941176471 0.062283737
4 0.990654206 0.009433138
5 0.999485332 0.000514933
6 0.999995237 0.000004763
7 0.999999998 0.000000002
8 1.000000000 0.000000000
Some comments:
Two initial guesses are needed, but does not require f (a ) * f (b ) < 0
unlike bisection. But there might be problems if the root lies outside
the convergence range.
Must have f ( xn ) ≠ f ( xn −1 ) (similar to f ′( x ) ≠ 0 in Newton’s
method). I.e., a very flat function.
Another problem case might occur is a generated guess is the same
as a previous guess, resulting in the possibility of an infinite loop
that never reaches the root.
500 Numerical Analysis
13.2 Polynomial Interpolation
Suppose we have, n + 1 points
(x0 , y0 ), (x1, y1 ), (x2 , y2 ),.............. (xn , yn ) and we need to find a
Suppose we were given two points {x0 , x1} and the values at those
points. We would draw a line as shown in Fig.9.
draw a n polynomial, Pn ( x ) .
In the above, a0 , a1 , ........, an are unknowns, and {xi } and {yi } are
known values. We can find the polynomial, if we solve the above
for a0 , a1 , ........, an
502 Numerical Analysis
Example 7 Find P1 ( x ) passing through ( x0 , y0 ) and ( x1 , y1 ) .
Solution:
P1 ( x ) has the form:
P1 ( x ) = a0 + a1 x (14)
P1 ( x0 ) = a0 + a1 x0 = y0
a1 ( x0 − x1 ) = y0 − y1
P1 ( x1 ) = a0 + a1 x1 = y1
y0 − y1
a1 = if x0 ≠ x1
x0 − x1
y0 − y1 x y −x y
a0 = y0 − a1 x0 = y0 − x0 = 0 1 1 0
x0 − x1 x0 − x1
x0 y1 − x1 y0 y0 − y1
P1 ( x ) = + x (15)
x0 − x1 x0 − x1
If x0 = y0 = 0, x1 = y1 = 1, P1 ( x ) = x I.e., the polynomial P1 ( x ) = x
passes through (0, 0 ) and (1,1) . Is this the only possible solution?
Yes. Why?
Fact. For any given n + 1 points ( x0 , y0 ), ( x1 , y1 ),.............. ( xn , y n ) ,
What is Pn ( x ) ?
Solution:
Pn ( x ) = 0 i.e., a0 = a1 = ....... = an = 0 . Why? Pn ( x ) = 0
interpolates ( x0 , 0), (x1, 0), ..........., ( xn , 0), and this is a unique
interpolation of the points.
n passing through (x0 ,1), (x1 , 0), (x2 , 0),........, (xn , 0) ⇒ l0n (x ) .See
Fig.11 for picture of l0n ( x ) .
l01 ( x0 ) = 1, l01 ( x1 ) = 0
l11 ( x0 ) = 0, l11 ( x1 ) = 1
How to find l nj ( x ) ?
0 at x1 , x2 , ........, xn
l0n ( x ) degree n n =
1 at x0
Consider the following polynomial of degree n
qn ( x ) = ( x − x1 )( x − x2 ).....( x − xn )
= 0 at x1 , x2 , ........, xn
j ≠i
∴lin ( x ) = (19)
∏
n
j =0
(xi − x j )
j ≠i
2. Pn ( xi ) = yi
Pn ( xi ) = f ( xi ), i = 0, 1, 2,........, n
Then, Pn ( x ) must interpolate ( x0 , f ( x0 )), ( x1 , f ( x1 )), .., ( xn , f ( xn )) .
Use the Lagrange formula,
n
Pn ( x ) = ∑ f ( xi )lin (x )
i =0
l03 ( x ) =
(x − 1)(x − 3)(x − 4)
(− 1)(− 3)(− 4)
l13 ( x ) =
(x − 0)(x − 3)(x − 4)
(1 − 0)(1 − 3)(1 − 4)
l23 ( x ) =
(x − 0)(x − 1)(x − 4)
(3 − 0)(3 − 1)(3 − 4)
l33 ( x ) =
(x − 0)(x − 1)(x − 3)
(4 − 0)(4 − 1)(4 − 3)
Chapter Thirteen 509
2. Find interpolating polynomial.
P3 ( x ) = 3
(x3 − 8x 2 + 19x − 12) + 2 (x3 − 7 x 2 + 12x) + (x3 − 5x 2 + 4 x)
− 12 6 −6
P3 ( x ) =
(− x3 + 6 x 2 − 17 x + 36)
12
3. Use P3 (2.5) to estimate f (2.5)
P3 (2.5) =
(− (2.5)3 + 6(2.5)2 − 17(2.5) + 36 )
= 1.281 25
12
∴ f (2.5) ≈ 1.281 25
f ( x1 ) = f ( x0 ) + A1 ( x1 − x0 )
(x1 , f (x1 )) Pn ( x1 ) = f ( x1 ) ⇒ f ( x1 ) − f ( x0 ) (24)
A
1 =
x1 − x0
f ( x2 ) − f ( x1 ) f ( x1 ) − f ( x0 )
−
x2 − x1 x1 − x0
(x2 , f (x2 )) Pn ( x2 ) = f ( x2 ) ⇒ A2 = (25)
x2 − x0
New Notation:
We can note in the above expressions for A1 and A2 a relationship
in the forms of the expressions, which leads us to the following new
notation.
f [x0 ] = f ( x0 ) ⇒ A0 = f [x0 ] (26)
f [x1 ] − f [x0 ]
f [x0 , x1 ] = ⇒ A1 = f [x0 , x1 ] (27)
x1 − x0
f [x1 , x2 ] − f [x0 , x1 ]
f [x0 , x1 , x2 ] = ⇒ A2 = f [x0 , x1 , x2 ] (28)
x2 − x0
f [x1 ] − f [x0 ]
We call f [x0 , x1 ] = divided difference at [x1 , x2 ],
x1 − x0
etc. Thus, the polynomial which interpolates:
Chapter Thirteen 511
( x0 , f (x0 )), (x1 , f (x1 )), ......., (xn , f (xn ))
can be written as:
Pn ( x ) = f [x0 ] + f [x0 , x1 ]( x − x0 ) + f [x0 , x1 , x2 ]( x − x0 )( x − x1 ) + ....
(29)
........ + f [x0 , x1 ,...., xn ]( x − x0 )( x − x1 )..........( x − xn −1 )
n −1
∴ Pn ( x ) = Pn −1 ( x ) + f [x0 , x1 ,...., xn ]∏ ( x − xi ) (30)
i =0
Where,
f [ x0 ] = f ( x0 )
f [x1 ] − f [x0 ]
f [x0 , x1 ] =
x1 − x0
f [x1 , x2 ] − f [x0 , x1 ]
f [x0 , x1 , x2 ] =
x 2 − x0
f [x1 , x2 ,...., xi ] − f [x0 , x1 ,......xi −1 ]
f [x0 , x1 , ....., xi ] =
xi − x0
f [x1 , x2 ,...., xn ] − f [x0 , x1 ,......xn −1 ]
f [x0 , x1 , ....., xi ] =
xn − x0
P3 ( x ) = 3 + (− 1)( x − 0 ) +
1
(x − 0)( x − 1) + − 1 (x − 0)(x − 1)(x − 3)
6 12
∆r +1 f i = ∆r f i +1 − ∆r f i
∆2 f i = ∆(∆f i ) = ∆( f i +1 − f i ) = ∆f i +1 − ∆f i
= ( f i + 2 − f i +1 ) − ( f i +1 − f i ) = f i + 2 − 2 f i +1 + f i
Now, let us consider the form of the Newton Divided Difference
with evenly spaced points.
f1 − f 0 1
f [x0 , x1 ] = = ∆f 0
x1 − x0 h
f [x1 , x2 ] − f [x0 , x1 ]
f [x0 , x1 , x2 ] =
x2 − x0
1 1 1
= ∆f1 − ∆f 0
2h h h
1
= 2 ∆2 f 0
2h
In general, and this can be easy proved via proof by induction,
1
∴ f [x0 , x1 , ..., xk ] = ∆k f 0 (31)
k! h k
We can now modify the Newton interpolation formula to an
interpolation formula based on forward differences. Since the
polynomial is defined continuously, rather than with respect to the
discretely spaced points, we will define for the value x at which the
polynomial is defined,
x − x0
µ= (32)
h
514 Numerical Analysis
where µ is a continuous parameter.
Therefore,
x − xi = x0 + µh − x0 − ih = (µ − i )h
which leads to the following form for the interpolating formula
n u
Pn ( x ) = ∑ i ∆i f0 (33)
i = 0
where we have used the binomial coefficients
µ µ (µ − 1)........(µ − i + 1)
= , i>0 (34)
i
i !
µ
and = 1 (35)
0
For example, n = 1
P1 ( x ) = f 0 + µ∆f 0
As with Newton divided differences, we can easily construct tables
to evaluate the forward differences.
xi fi ∆f i ∆2 f i ∆3 f i L
x0 f0
x1 f1 ∆f 0
∆2 f 0
x2 f2 ∆f1 ∆3 f 0
∆2 f1 M
x3 f3 ∆f 2 ∆3 f1
∆2 f 2
x4 f4 ∆f 3
M M
Chapter Thirteen 515
Example 12. Find the interpolating function for the following table
i 0 1 2 3
xi 0 1 3 4
f ( xi ) 3 2 0 -1
P3 ( x ) = 3 + (− 1)(µ ) + (− 1)
(µ )(µ − 1) + (2) (µ )(µ − 1)(µ − 2)
2 6
Note: forward differences of order greater than three are almost
entirely the result of differencing the rounding errors in the table
entries; therefore, interpolation in this table should be limited to
polynomials of degree less than four.
As you can see, there is nothing particularly special about forward
differences.
We can equally define backward difference interpolating functions
based on
∇f i = f i − f i −1 (36)
516 Numerical Analysis
13.3 Numerical Integration
Quadrature comes from the process of “squaring”, of finding a
square equal in area to a given area, e.g., finding the area of a circle.
Now means numerical integration.
The problem is either
Easy example:
b
I = ∫ xdx =
a 2
(
1 2
b − a2 ) or more generally
b
I = ∫ Pn ( x )dx
a
b
Hard example I = ∫ e cos x dx = ?
a
Chapter Thirteen 517
b
In many applications, if f ( x ) is complicated then ∫a f ( x )dx cannot
be calculated analytically must be approximated by a numerical
value
The approach:
1. Locally interpolate f ( x ) by a simple function g ( x ) , e.g., a
xi +1 f ( xi ) + f ( xi +1 )
Therefore, ∫ f ( x )dx ≅ ( xi +1 − xi )
xi 2
So, if the step size is h, then the area of any trapezoid is
h
( f i + f i +1 )
2
The integral is thus approximately, for n + 1 points,
n −1h
I ( f ) ≅ Tn ( f ) = ∑ 2 ( fi + f i +1 )
i =0
h n −1 n −1
= ∑ ( f i ) + ∑ ( f i +1 )
2 i =0 i =0
h
= [( f 0 + f1 + f 2 + ..... + f n − 2 + f n −1 ) + ( f1 + f 2 + ..... + f n − 2 + f n )]
2
h
= [( f 0 + 2 f1 + 2 f 2 + ..... + 2 f n − 2 + 2 f n −1 + f n )]
2
h n −1
∴ I ( f ) ≅ Tn ( f ) =
0 f + 2 ∑ i n
f + f (37)
2 i =1
Where f 0 = f (a ) and f n = f (b )
Discretization Error
To perform the integration using the Trapezoid Rule, we are
approximating f ( x ) on [x0 , x1 ] by a first-order polynomial P1 ( x ).
( x − x0 )2 ( x − x0 )3
≅ P1 ( x ) + f ′′( x0 ) + f ′′′( x0 )
2 6
and in particular
h2 h3
f ( x1 ) ≅ P1 ( x ) + f ′′( x0 ) + f ′′′( x0 )
2 6
The error E (R ) in the integral I ( f ) − I (P1 ) is:
x1 (x − x0 )2
E (R ) ≈ ∫ f ′′( x0 )dx
x0 2
f ′′( x0 ) x1
≈ ∫x
(x − x0 )2 dx
2 0
x1
f ′′( x0 ) ( x − x0 )3
≈
2 3 x0
f ′′( x0 )
∴ E (R ) ≈ (x1 − x0 )3 (38)
6
Thus, the total error is the Trapezoid Rule minus the integral of
P1 ( x ) minus E (R )
h h2 h3
E (R ) ≈ ( f ( x0 ) + f ( x1 )) − hf ( x0 ) − f ′( x0 ) − f ′′( x0 ) − E (R )
2 2 6
h3 h3
∴ E (R ) ≈ f ′′( x0 ) = M2 (39)
12 12
520 Numerical Analysis
for a bound M 2 on f ′′( x ) over [x0 , x1 ]
The total possible quadrature error is the sum of all the errors for
each of the panels, [xi , xi +1 ]
n M 2 h3
Tn ( f ) − I ( f ) ≤ ∑ (40)
12
i =1
M 2 h 3n
=
12
M 2 (b − a )h 2
12
= O h2 ( )
Therefore,
M 2 (b − a )h 2
Tn ( f ) − I ( f ) ≤ (41)
12
So this is a second-order method.
π
Example 13 Evaluate I = ∫ e x cos( x )dx by composite trapezoidal
0
=
π 1
+ 1.5509 + 0 + (− 7.4605) +
(− 23.141) = −13.336
4 2 2
n=8 T8 ( f ) = −12.382
n = 64 T64 ( f ) = −12.075
n = 512 T512 ( f ) = −12.070
True solution is − 12.0703
(xi + 2 − xi ) 4( xi + 2 − xi ) x − xi
∴ I (P2 ) = f ( xi ) + f ( xi +1 ) + f ( xi + 2 ) i + 2
6 6 6
xi + 2
h
∴ ∫ f ( x )dx ≅ I (P2 ) = ( f i + 4 f i +1 + f i + 2 ) (47)
x1
3
where h = xi +1 − xi
To get the sum over the entire interval [a, b] we sum over all the
panels, noting that the end points of the panels are have even
b−a
numbered indicies, with h =
n
h
Sn ( f ) = ( f 0 + 4 f1 + f 2 ) + h ( f 2 + 4 f 3 + f 4 ) + ... + h ( f n − 2 + 4 f n −1 + f n )
3 3 3
n−2 h n / 2 −1
= ∑ ( f i + 4 f i +1 + f i + 2 ) = ∑ h ( f 2i + 4 f 2i +1 + f 2i + 2 )
i =0 3 i =0 3
i = even
h
= ( f 0 + 4 f1 + 2 f 2 + 4 f 3 + 2 f 4 + 4 f 5 + .... + 4 f n − 3 + 2 f n − 2 + 4 f n −1 + f n )
3
h n / 2 −1 n / 2−2
∴ Sn ( f ) = f 0 + 4 ∑ f 2i +1 + 2 ∑ f 2i + 2 + f n (48)
3 i =0 i =0
By using more advanced techniques, we can show that for even n
and f ( x ) four times differentiable, the local error per panel
(containing three points) is:
524 Numerical Analysis
M4
I (P2 ) − I ( f ) ≤ h 5 (49)
90
with M 4 being the bound on f (4 ) ( x ) . For the composite Simpson’s
Rule over the entire domain the upper bound on the error is
M 4 (b − a ) 1 M 4 (b − a )5
Sn ( f ) − I ( f ) ≤ h4
= 4 (50)
180 n 180
Therefore, Simpson’s Rule is fourth-order.
n
−1
p −1 2
b t x2i+2
∴ I = ∫ f ( x )dx = ∑ ∫tii+1 f (x )dx = ∑ ∫ f ( x )dx
a x
i =0 i = 0 2i
n
−1
2 h
∴I = ∑ 3 [ f (x2i ) + 4 f (x2i +1 ) + f (x2i + 2 )]
i =0
n
−1
n
−2
2 2
h
∴I = f 0 + 4 ∑ f 2i +1 + 2 ∑ f 2i + 2 + f n (53)
3 i =0 i =0
Chapter Thirteen 527
π
Example 15 Evaluate I = ∫ e π cos( x ) dx by composite trapezoidal
0
rule using 4 subintervals (panels).
Solution:
=
π 1
+ + +
− +
(− 23.141) = −13.336
1 .5509 0 7 .4605
4 2 2
n=8 CTR = −12.382
n = 64 CTR = −12.075
n = 512 CTR = −12.070
Then the true solution is ≅ −12.0703
So in a composite method, as n gets larger ⇒ the error gets smaller.
But how do we know which Y to take for a given accuracy?
b n
∫ f (x )dx ≅ ∑ ω i f (xi )
a
(54)
i =0
528 Numerical Analysis
n is fixed. xi are fixed.
In trapezoidal: n = 1, x0 = a and x1 = b .
ωi can be computed when the {xi } are given; i.e., they are
determined by xi
h
In trapezoid: ω0 = = ω1
2
Disadvantages: xi are chosen artificially how do we know they give
us the best result?
Another approach
n
I≅ ∑ ωi f (xi ) n is fixed, ωi , xi are to be determined.
i =0
n
∴I = ∑ ω i f (xi ) gives the “best” result.
i =0
∫ Pm ( x )dx = ω0 Pm ( x0 ) + ω1 Pm ( x1 ) (55)
P0 ( x ) = x 2
1 2
∫−1 xdx = ω0 x02 + ω1 x12 ⇒ ω0 x02 + ω1 x12 =
3
4. Exact for polynomial of degree 3, i.e., 55 holds for
P0 ( x ) = x 3
1 3 3 3 3
∫−1 xdx = ω0 x0 + ω1x1 ⇒ ω0 x0 + ω1x1 = 0
530 Numerical Analysis
Can we expect the method to be exact for still higher degree
polynomials? No. We have 4 unknowns, x0 , x1 , ω0 , ω1 , and if the
method is exact for polynomials of degree 3, we already have 4
equations. This is enough to determine the 4 unknowns.
By solving the four equations in boxes above, we find:
3 3
x0 = − , x1 = , ω0 = 1, ω1 = 1
3 3
1 3 3
Thus ∫−1 f ( x )dx ≅ f
3 +
− f
3
This is Gaussian Quadrature on [− 1,1] with two nodes.
3 3
From above, we know that f − + f is exact if
3 3
f = 1, x, x 2 , x 3 .
Is it exact for all polynomials of degree ≤ 3 ? Yes:
1 1 1 1 1
∫−1 f ( x )dx = a0 ∫ dx + a1 ∫ xdx + a2 ∫ x 2 dx + a3 ∫ x 3dx \
−1 −1 −1 −1
3 3 3 3
2 2 3 3
3 3
= a0 [1 + 1] + a1 − +
+ a2 −
+ 3 + a3 − 3 + 3
3 3 3
3 3
2
3
3
= a0 + a1 − + a 2 − + a3 − +
3 3 3
3 3
2
3
3
a0 + a1
3 + a 2 3 + a3 3
Chapter Thirteen 531
3 3
= f1 − + f
3 3
polynomials of degree ≤ 3
∫−1 (3 + 4 x + 8 x )
1 2
Example 17 Evaluate + 2 x 3 dx Via Gaussian
Quadrature:
2
3
3
1
( 2 3
)
3 3
∫ 3 + 4 x + 8 x + 2 x dx = 3 + 4 3 + 8 3 + 2 3
−1
− 3 − 3
2
− 3
3
+ 3 + 4 + 8 + 2
3 3 3
3 34
2
= 23 + 8 =
3 3
Compare that with straight integration:
1
∫
1
−1
(
)
3 + 4 x + 8 x 2 + 2 x 3 dx = 3x +
4 x 2 8x3
2
+
3
++
8x 4
3
−1
∴∫
1
−1
(3 + 4 x + 8x 2 + 2 x3 )dx = 23 + 83 = 343
b n
∫ f ( x )dx ≅ ∑ ω i f (xi )
a
i =0
532 Numerical Analysis
xi , ωi are chosen so that the method is exact for
1, x, x 2 ,........, x m
where m is as large as possible. What is the largest possible m for a
fixed n The number of unknowns are: 2n + 2 and also m + 1
functions m + 1 equations.
Unknown = Eqns ⇒ m + 1 = 2n + 2
i.e., m = 2n + 1
Conclusion: Gaussian quadrature with n + 1 nodes (function
evaluations) is exact for a polynomial of degree ≤ 2n + 1 . In
comparison, a Newton-Cotes rule of degree n with n + 1 nodes is
exact for polynomials of degree ≤ n .
When we have to determine ωi , xi , we have to solve a non-linear
system.
Chapter Thirteen 533
13.4 Numerical Solution of Differential Equations
y n +1 = yn + hf ( x0 + nh, y n )
Then, y n is as an approximation to y ( xn ) .
1.5 − 1
∴ xn = x0 + nh = 1 + 0.1n , N = =5
0.1
and, y n +1 = y n + 0.2(1 + 0.1n ) y n With y0 = 1
For n = 0, 1, 2, 3, 4 . The numerical results are listed in Table (7).
Note that the differential equation in (63) is separable. The (unique)
1.5 − 1
∴ xn = x0 + nh = 1 + 0.1n , N = =5
0.1
and y n +1 = yn + 2(1 + 0.1n ) yn With y0 = 1
For n = 0, 1, 2, 3, 4 . The numerical results are listed in Table(8).
The relative errors show that our approximations are not very good.
xn yn y ( xn ) Absolute Relative
error error
١٫٠٠ ١٫٠٠٠٠ ١٫٠٠٠٠ ٠٫٠٠٠٠ ٠٫٠٠
١٫١٠ ١٫٠٢٠٠ ١٫٢٣٣٧ ٠٫٠٣٣٧ ٢٫٧٣
١٫٢٠ ١٫٤٦٤٠ ١٫٥٥٢٧ ٠٫٠٨٨٧ ٥٫٧١
١٫٣٠ ١٫٨١٥٤ ١٫٩٩٣٧ ٠٫١٧٨٤ ٨٫٩٥
١٫٤٠ ٢٫٢٨٧٤ ٢٫٦١١٧ ٠٫٣٢٤٤ ١٢٫٤٢
١٫٥٠ ٢٫٩٢٧٨ ٣٫٤٩٠٤ ٠٫٥٦٢٥ ١٦٫١٢
Chapter Thirteen 537
13.4.3 Improved Euler's method.
The improved Euler's method takes the average if the slopes at the
left and right ends of each step. It is, here, formulated by means of a
predictor and a corrector:
( )
y nP+1 = ynC + hf xn , y nC , (65)
y0C = 1 ,
(
y nC+1 = y nC + 0.1 xn ynC + xn +1 y nP+1 )
for n = 0, 1, . . . , 4. The numerical results are listed in Table 5.3.
These results are much better than those listed in Table(9) for
Euler's method.
538 Numerical Analysis
Table (9) Numerical results of Example 20.
xn yn y ( xn ) Absolute Relative
error error
١٫٠٠ ١٫٠٠٠٠ ١٫٠٠٠٠ ٠٫٠٠٠٠ ٠٫٠٠
١٫١٠ ١٫٢٣٢٠ ١٫٢٣٣٧ ٠٫٠٠١٧ ٠٫١٤
١٫٢٠ ١٫٥٤٧٩ ١٫٥٥٢٧ ٠٫٠٠٤٨ ٠٫٣١
١٫٣٠ ١٫٩٨٣٢ ١٫٩٩٣٧ ٠٫٠١٠٦ ٠٫٥٣
١٫٤٠ ٢٫٥٩٠٨ ٢٫٦١١٧ ٠٫٠٢٠٩ ٠٫٨٠
١٫٥٠ ٣٫٤٥٠٩ ٣٫٤٩٠٤ ٠٫٠٣٤٤ ١٫١٣
k 2 = hf ( xn + h, yn + k1 ) (68)
1
y n +1 = y n + (k1 + k 2 ) (69)
2
Chapter Thirteen 539
Fourth-order Runge-Kutta method.
The fourth-order Runge-Kutta method is the very popular among
the explicit one-step methods. The four-stage Runge-Kutta method
of order 4 given by its formula:
k1 = hf ( xn , y n ) (70)
1 1
k 2 = hf xn + h, y n + k1 (71)
2 2
1 1
k3 = hf xn + h, y n + k 2 (72)
2 2
k 4 = hf ( xn + h, y n + k3 ) (73)
1
y n +1 = yn + (k1 + 2k 2 + 2k3 + k 4 ) (74)
6
The next example shows that the fourth-order Runge-Kutta
method yields better results for (64) than the previous methods.
xn yn y ( xn ) Absolute Relative
error error
١٫٠٠ ١٫٠٠٠٠ ١٫٠٠٠٠ ٠٫٠٠٠٠ ٠٫٠
١٫١٠ ١٫٢٣٣٧ ١٫٢٣٣٧ ٠٫٠٠٠٠ ٠٫٠
١٫٢٠ ١٫٥٥٢٧ ١٫٥٥٢٧ ٠٫٠٠٠٠ ٠٫٠
١٫٣٠ ١٫٩٩٣٧ ١٫٩٩٣٧ ٠٫٠٠٠٠ ٠٫٠
١٫٤٠ ٢٫٦١١٦ ٢٫٦١١٧ ٠٫٠٠٠١ ٠٫٠
١٫٥٠ ٣٫٤٩٠٢ ٣٫٤٩٠٤ ٠٫٠٠٠٢ ٠٫٠
y ′ = ( y − x − 1)2 + 2, y (0) = 1
Compute y 4 by means of Runge-Kutta's method of order 4 with
step size h = 0.1.
Chapter Thirteen 541
Solution: The solution is given as shown in Table(11).
Table(11). Numerical results for Example 23.
n xn yn Exact value Global error
y ( xn ) y ( xn ) − y n
0 0.0 1.000 000 000 1.000 000 000 0.000 000 000
1 0.1 1.200 334 589 1.200 334 672 0.000 000 083
2 0.2 1.402 709 878 1.402 710 0'36 0.000 000157
3 0.3 1.609 336 039 1.609 336 250 0.000 000181
4 0.4 1.822 792 993 1.822 793 219 0.000 000 226
1 1
k 2 = hf x0 + h, y0 + k1 = 0.2 * 0.5 + 0.41 + 0.09 = 0.2
2 2
1 1
k3 = hf x0 + h, y0 + k 2 = 0.2 0.5 + 0.41 + 0.1 = 0.201
2 2
k 4 = hf ( x0 + h, y0 + k3 ) = 0.2 0.5 + 0.41 + 0.09 = 0.2201
1
y1 = y0 + (k1 + 2k 2 + 2k3 + k 4 )
6
∴ y1 = 0.41 + 0.2003476 = 0.6103476
542 Numerical Analysis
Problems
Solve the following differential equations by the methods
indicated
Euler’s method:
1- y ′ = 2 x − y , x = 0, y = 1 , x = 0, 0.2
(
2- y ′ = y − x 2 )1 / 2 , x = 0, y = 1 x = 0.1, 0.2
3- y ′ = ( x + y ) / xy, x = 1, y = 1, x = 1.0, 1.2
Runge-Kutta’s method
1- y ′ = (2 x + y )1 / 2 , x = 1, y = 2, x = 1.0, 1.4
( )
2- y ′ = 1 − x 3 / y , x = 0, y = 1, x = 0, 0.2
3- y ′ = ( x − y ) / ( x + y ), x = 0, y = 1 x = 0.2, 0.4
References
[1] “Advanced Engineering Mathematics” by C. Ray Wyle and
Louis C. Barrett, Fifth Edition, 1985, by McGraw-Hill, ISBN 0-
07-Y66643-1.
[2] “Advanced Engineering Mathematics” by Erwin Kreysizig,
Third Edition, 1972 by John Wiley & Sons, Inc., ISBN 0-471-
50728-8.
[3] “ Engineering Mathematics” by K. A. Stroud, Third Edition,
1992, ELBS with Macmillan, Educational Low Priced Books
Scheme. ISBN 0-333-54454-4.
[4] “Theory and Problems of Engineering Mathematics for
Engineers and Scientists” by Murray R. Spiegel, Schaum’s
Outline Series, McGraw-Hill, 1971 ISBN 07-060216-6.
[5] “Numerical Methods for Engineers with Programing and
Software Applications” by Steven C. Chapra and Raymond P.
Canale, 3rd Edition, 1998, McGraw-Hill, ISBN 0-07-115895-2.
[6] “Engineering Mathematics” by Ian Craw, Stuart Dagger and
John Pulham, the University of Aberdeen, 2001.
[7] “ Numerical Methods” by E. A. Volkov, Translated from
Russian by L. Levant, MIR PUBLISHER MOSCOW,1986.
[8] “Mathematical Handbook of Formulas and Tables”, by Murray
R. Spiegel and John Liu, Schaum’s Outline Series, McGraw-Hill,
Second Edition, 1999, ISBN 0-07-038203-4.