MATHEMATICS-II

 Prepared by : Ms. K. Rama Jyothi,

 : Mr. G Nagendra Kumar
CONTENTS
 UNIT-I:Vector Calculus
 UNIT-II: Fourier series and Fourier transforms
 UNIT-III: Interpolation and Curve Fitting
 UNIT-IV: Solutions of Algebraic and transcendental
equations and Linear system of equations
 UNIT-V: Numerical Integration and Numerical
solutions of First order differential equations
TEXT BOOKS
 Advanced Engineering Mathematics by Kreyszig,
John Wiley & Sons.
 Higher Engineering Mathematics by Dr. B.S.
Grewal, Khanna Publishers
REFERENCES
 Mathematical Methods by T.K.V. Iyengar,
B.Krishna Gandhi & Others, S. Chand.
 Introductory Methods by Numerical Analysis by
S.S. Sastry, PHI Learning Pvt. Ltd.
 Mathematical Methods by G.ShankarRao, I.K.
International Publications, N.Delhi
 Mathematical Methods by V. Ravindranath, Etl,
Himalaya Publications.
REFERENCES
 Advanced Engineering Mathematics with
MATLAB, Dean G. Duffy, 3rd Edi, 2013, CRC Press
Taylor & Francis Group.
 6. Mathematics for Engineers and Scientists, Alan
Jeffrey, 6ht Edi, 2013, Chapman & Hall/ CRC
 7. Advanced Engineering Mathematics, Michael
Greenberg, Second Edition. Pearson Education
Vector Calculus
INTRODUCTION
 In this chapter, vector differential calculus is considered,
which extends the basic concepts of differential calculus,
such as, continuity and differentiability to vector functions
in a simple and natural way. Also, the new concepts of
gradient, divergence and curl are introducted.
 Example: i,j,k are unit vectors.
VECTOR DIFFERENTIAL OPERATOR
 The vector differential operator ∆ is defined as ∆=i ∂/∂x +
j ∂/∂y + k ∂/∂z. This operator possesses properties
analogous to those of ordinary vectors as well as
differentiation operator. We will
∂/∂define now some
quantities known as gradient, divergence and curl
involving this operator.
GRADIENT
 Let f(x,y,z) be a scalar point function of position defined
in some region of space. Then gradient of f is denoted by
grad f or ∆f and is defined as
grad f=i∂f/∂x + j ∂f/∂y + k ∂f/∂z
 Example: If f=2x+3y+5z then gradf= 2i+3j+5k
DIRECTIONAL DERIVATIVE
 The directional derivative of a scalar point function f at a
point P(x,y,z) in the direction of g at P and is defined as
grad g/|grad g|.grad f
 Example: The directional derivative of f=xy+yz+zx in the
direction of the vector i+2j+2k at the point (1,2,0) is 10/3
DIVERGENCE OF A VECTOR
 Let f be any continuously differentiable vector point
function. Then divergence of f and is written as div f and
is defined as
Div f = ∂f1/∂x + j ∂f2/∂y + k ∂f3/∂z
 Example 1: The divergence of a vector 2xi+3yj+5zk is 10
 Example 2: The divergence of a vector f=xy2i+2x2yzj-
3yz2k at (1,-1,1) is 9
SOLENOIDAL VECTOR
 A vector point function f is said to be solenoidal vector if
its divergent is equal to zero i.e., div f=0
 Example 1: The vector f=(x+3y)i+(y-
2z)j+(x-2z)k is solenoidal vector.
 Example 2: The vector f=3y4z2i+z3x2j-3x2y2k is solenoidl
vector.
CURL OF A VECTOR
 Let f be any continuously differentiable vector point
function. Then the vector function curl of f is denoted by
curl f and is defined as curl f= ix∂f/∂x + jx∂f/∂y +
kx∂f/∂z
 Example 1: If f=xy2i +2x2yzj-3yz2k then curl f at (1,-1,1)
is –i-2k
 Example 2: If r=xi+yj+zk then curl r is 0
IRROTATIONAL VECTOR
 Any motion in which curl of the velocity vector is a null
vector i.e., curl v=0 is said to be irrotational. If f is
irrotational, there will always exist a scalar function
f(x,y,z) such that f=grad g. This g is called scalar potential
of f.
 Example: The vector
f=(2x+3y+2z)i+(3x+2y+3z)j+(2x+3y+3z)k is irrotational
vector.
VECTOR INTEGRATION
 INTRODUCTION: In this chapter we shall define line,
surface and volume integrals which occur frequently in
connection with physical and engineering problems. The
concept of a line integral is a natural generalization of the
concept of a definite integral of f(x) exists for all x in the
interval [a,b]
WORK DONE BY A FORCE
 If F represents the force vector acting on a particle moving
along an arc AB, then the work done during a small
displacement F.dr. Hence the total work done by F during
displacement from A to B is given by the line integral ∫F.dr
 Example: If f=(3x2+6y)i-14yzj+20xz2k along the lines
from (0,0,0) to (1,0,0) then to (1,1,0) and then to (1,1,1) is
23/3
SURFACE INTEGRALS
The surface integral of a vector point function F
expresses the normal flux through a surface. If F
represents the velocity vector of a fluid then the
surface integral ∫F.n dS over a closed surface S
represents the rate of flow of fluid through the
surface.
Example:The value of ∫F.n dS where F=18zi-12j+3yk
and S is the part of the surface of the plane
2x+3y+6z=12 located in the first octant is 24.
VOLUME INTEGRAL
 Let f (r) = f1i+f2j+f3k where f1,f2,f3 are functions of x,y,z.
We know that dv=dxdydz. The volume integral is given by
∫f dv=∫∫∫(f1i+f2j+f3k)dxdydz
 Example: If F=2xzi-xj+y2k then the value of ∫f dv where
v is the region bounded by the surfaces
x=0,x=2,y=0,y=6,z=x2,z=4 is128i-24j-384k
VECTOR INTEGRAL THEOREMS
 In this chapter we discuss three important vector integral
theorems.
 1)Gauss divergence theorem
 2)Green’s theorem
 3)Stokes theorem
GAUSS DIVERGENCE THEOREM
 This theorem is the transformation between surface
integral and volume integral. Let S be a closed surface
enclosing a volume v. If f is a continuously differentiable
vector point function, then
 ∫div f dv=∫f.n dS
 Where n is the outward drawn normal vector at any point
of S.
GREEN’S THEOREM
 This theorem is transformation between line integral and
double integral. If S is a closed region in xy plane
bounded by a simple closed curve C and in M and N are
continuous functions of x and y having continuous
derivatives in R, then
 ∫Mdx+Ndy=∫∫(∂N/∂x - ∂M/∂y)dxdy
STOKES THEOREM
 This theorem is the transformation between line integral
and surface integral. Let S be a open surface bounded by a
closed, non-intersecting curve C. If F is any differentiable
vector point function then
 ∫F.dr=∫Curl F.n ds
Fourier Series and
Fourier Transform
INTRODUCTION
 Suppose that a given function f(x) defined in (-π,π) or
(0, 2π) or in any other interval can be expressed as a
trigonometric series as
f(x)=a0/2 + (a1cosx + a2cos2x +a3cos3x +…+ancosnx)+
(b1sinx +
b2sin 2x+……..bnsinnx)+……..
f(x) = a0/2+ ∑(ancosnx + bnsinnx)
 Where a and b are constants with in a desired range of
values of the variable such series is known as the
fourier series for f(x) and the constants a0, an,bn are
called fourier coefficients of f(x)

 It has period 2π and hence any function represented

by a series of the above form will also be periodic with
period 2π
POINTS OF DISCONTINUITY
 In deriving the Euler’s formulae for a0,an,bn it was
assumed that f(x) is continuous. Instead a function
may have a finite number of discontinuities. Even then
such a function is expressable as a fourier series.
DISCONTINUITY FUNCTION
 For instance, let the function f(x) be defined by
 f(x) = ø (x), c< x< x0
 = Ψ(x), xo<x<c+2π
 where x0 is thepoint of discontinuity in (c,c+2π).
DISCONTINUITY FUNCTION
In such cases also we obtain the fourier series for f(x) in
the usual way. The values of a0,an,bn are given by
a0 = 1/π [ ∫ ø(x) dx + ∫ Ψ(x) dx ]
an =1/π [ ∫ ø(x)cosnx dx + ∫ Ψ(x)cosnx dx
bn = 1/π [ ∫ ø(x)sinnx dx + ∫ Ψ(x)sinnx dx]
EULER’S FORMULAE
 The fourier series for the function f(x) in the interval
C≤ x ≤ C+2π is given by
f(x) = a0/2+ ∑(ancosnx + bnsinnx)
where a0 = 1/ π ∫ f(x) dx
an = 1/ π ∫ f(x)cosnx dx
bn = 1/ π ∫ f(x)sinnxdx
These values of ao , an, bn are known as Euler’s formulae
EVEN AND ODD FUNCTIONS
 A function f(x) is said to be even if f(-x)=f(x) and
odd if f(-x) = - f(x).
 If a function f(x) is even in (-π, π ), its fourier series
expansion contains only cosine terms, and their
coefficients are ao
 and an.
 f(x)= a0/2 + ∑ an cosnx
 where ao= 2/ π ∫ f(x) dx
 an= 2/ π ∫ f(x) cosnx dx
ODD FUNCTION
 When f(x) is an odd function in (-π, π) its fourier
expansion contains only sine terms.
 And their coefficient is bn
 f(x) = ∑ bn sinnx
 where bn = 2/ π ∫ f(x) sinnx dx
HALF RANGE FOURIER SERIES

 THE SINE SERIES: If it be required to express f(x) as a

sine series in (0,π), we define an odd function f(x) in
(-π, π ) ,identical with f(x) in (0,π).
 Hence the half range sine series (0,π) is given by
 f(x) = ∑ bn sinnx
 Where bn = 2/ π ∫ f(x) sinnx dx
HALF RANGE SERIES
 The cosine series: If it be required to express f(x) as a
cosine series, we define an even function f(x) in (- (-π,
π ) , identical with f(x) in (0, π ) , i.e we extend the
function reflecting it with respect to the y-axis, so that
f(-x)=f(x).
HALF RANGE COSINE SERIES
 Hence the half range series in (o,π) is given by
 f(x) = a0/2 + ∑ an cosnx
 where a0= 2/ π∫f(x)dx
 an= 2/ π∫f(x)cosnxdx
CHANGE OF INTERVAL
 So far we have expanded a given function in a Fourier
series over the interval (-π,π)and (0,2π) of length 2π.
In most engineering problems the period of the
function to be expanded is not 2π but some other
quantity say 2l. In order to apply earlier discussions to
functions of period 2l, this interval must be converted
to the length 2π.
PERIODIC FUNCTION
 Let f(x) be a periodic function with period 2l defined
in the interval c<x<c+2l. We must introduce a new
variable z such that the period becomes 2π.
CHANGE OF INTERVAL
 The fourier expansion in the change of interval is given
by
 f(x) = a0/2+∑ancos nπx/l +∑ bn sin nπx/l
 Where a0 = 1/l ∫f(x)dx
 an = 1/l ∫f(x)cos nπx/l dx
 bn = 1/l ∫f(x)sin nπx/l dx
EVEN AND ODD FUNCTION
 Fourier cosine series : Let f(x) be even function in (-l,l)
then
 f(x) = a0/2 + ∑ ancos nπx/l
 where a0 = 2/l ∫f(x)dx
 an =2/l ∫f(x)cos nπx/l dx
FOURIER SINE SERIES
 Fourier sine series : Let f(x) be an odd function in (-l,l)
then
 f(x) = ∑ bn sin nπx/l
 where bn = 2/l ∫f(x)sin nπx/l dx
 Once ,again here we remarks that the even nature or
odd nature of the function is to be considered only
when we deal with the interval (-l,l).
HALF-RANGE EXPANSION
 Cosine series: If it is required to expand f(x) in the
interval (0,l) then we extend the function reflecting in
the y-axis, so that f(-x)=f(x).We can define a new
function g(x) such that f(x)= a0/2 + ∑ ancos nπx/l
 where a0= 2/l ∫f(x)dx
 an= 2/l ∫f(x)cos nπx/l dx
HALF RANGE SINE SERIES
 Sine series : If it be required to expand f(x)as a sine
series in (0,l), we extend the function reflecting it in
the origin so that f(-x) = f(x).we can define the fourier
series in (-l,l) then,
 f(x) = a0/2 + ∑ bn sin nπx/l
 where bn = 2/l ∫f(x)sin nπx/l dx
FOURIER INTEGRAL TRANSFORMS
 INTRODUCTION:A transformation is a
mathematical device which converts or changes
one function into another function. For example,
differentiation and integration are
transformations.
 In this we discuss the application of finite and
infinite fourier integral transforms which are
mathematical devices from which we obtain the
solutions of boundary value.
 We obtain the solutions of boundary value problems
related toengineering. For example conduction of heat,
free and forced vibrations of a membrane, transverse
vibrations of a string, transverse oscillations of an
elastic beam etc.
 DEFINITION: The integral transforms of a function
f(t) is defined by
 F(p)=I[f(t) = ∫ f(t) k(p,t )dt
 Where k(p,t) is called the kernel of the integral
transform and is a function of p and t.
FOURIER COSINE AND SINE INTEGRAL
 When f(t) is an odd function cospt,f(t) is an odd
function and sinpt f(t) is an even function. So the first
integral in the right side becomes zero. Therefore we
get
 f(x) = 2/π∫sinpx
FOURIER COSINE AND SINE INTEGRAL
 When f(t) is an odd function cospt,f(t) is an odd
function and sinpt f(t) is an even function. So the first
integral in the right side becomes zero. Therefore we
get
 f(x) = 2/π∫sinpx ∫f(t) sin pt dt dp
 which is known as FOURIER SINE INTEGRAL.
 When f(t) is an even function, the second integral in
the right side becomes zero.Therefore we get
 f(x) = 2/π∫cospx ∫f(t) cos pt dt dp
 which is known as FOURIER COSINEINTEGRAL.
FOURIER INTEGRAL IN COMPLEX FORM
 Since cos p(t-x) is an even functionof p, we have
 f(x) = 1/2π∫∫eip(t-x) f(t) dt dp
 which is the required complex form.
INFINITE FOURIER TRANSFORM
 The fourier transform of a function f(x) is given by
 F{f(x)} = F(p) = ∫f(x) eipx dx
 The inverse fourier transform of F(p) is given by
 f(x) = 1/2π ∫ F (p) e-ipx dp
FOURIER SINE TRANSFORM
 The finite Fourier sine transform of f(x) when 0<x<l, is
definedas
 Fs{f(x) = Fs (n) sin (nπx)/l dx where n is an integer and
the function f(x) is given by
 f(x) = 2/l∑ Fs (n) sin (nπx)/l is called the Inverse
finite Fourier sine transform Fs(n)
FOURIER COSINE TRANSFORM
 We have f(x) = 2/π∫cospx ∫f(t) cos pt dt dp
 Which is the fourier cosine integral .Now
 Fc(p) = ∫ f(x) cos px dx then
 f(x) becomes f(x) =2/π∫ Fc(p) cos px dp which is the
fourier cosine transform.
PROPERTIES
 Linear property of Fourier transform
 Change of Scale property
 Shifting property
 Modulation property
Interpolation and
Curve Fitting
Finite difference methods
Let (xi,yi),i=0,1,2…………..n be the equally spaced data of the
unknown function y=f(x) then much of the f(x) can be extracted by analyzing
the differences of f(x).
Let x1 = x0+h
x2 = x0+2h
.
.
.
xn = x0+nh be equally spaced points where the function value of f(x)
be y0, y1, y2…………….. yn
 Symbolic operators
Forward shift operator(E) :
It is defined as Ef(x)=f(x+h) (or) Eyx = yx+h
The second and higher order forward shift operators are defined
in similar manner as follows

E2f(x)= E(Ef(x))= E(f(x+h)= f(x+2h)= yx+2h

E3f(x)= f(x+3h)
.
.
Ekf(x)= f(x+kh)
Backward shift operator(E-1) :
It is defined as E-1f(x)=f(x-h) (or) Eyx = yx-h

The second and higher order backward shift

operators are defined in similar manner as follows

E-2f(x)= E-1(E-1f(x))= E-1(f(x-h)= f(x-2h)= yx-2h

E-3f(x)= f(x-3h)
.
.
E-kf(x)= f(x-kh)
Forward difference operator (∆) :
The first order forward difference operator of a function
f(x) with increment h in x is given by
∆f(x)=f(x+h)-f(x) (or) ∆fk =fk+1-fk ; k=0,1,2………
∆2f(x)= ∆[∆f(x)]= ∆[f(x+h)-f(x)]= ∆fk+1- ∆fk ; k=0,1,2…………
…………………………….
…………………………….
Relation between E and ∆ :
∆f(x)=f(x+h)-f(x)
=Ef(x)-f(x) [Ef(x)=f(x+h)]
=(E-1)f(x)

∆=E-1 E=1+ ∆
Backward difference operator (nabla) :
The first order backward difference operator of a function
f(x) with increment h in x is given by
 f(x)=f(x)-f(x-h) (or) fk =fk+1-fk ; k=0,1,2………
f(x)= [f(x)]= [f(x+h)-f(x)]=fk+1- fk ; k=0,1,2…………
…………………………….
…………………………….
Relation between E and nabla :
nabla f(x)=f(x+h)-f(x)
=Ef(x)-f(x) [Ef(x)=f(x+h)]
=(E-1)f(x)

nabla=E-1 E=1+ nabla

 Central difference operator ( δ) :
The central difference operator is defined as
δf(x)= f(x+h/2)-f(x-h/2)
δf(x)= E1/2f(x)-E-1/2f(x)
= [E1/2-E-1/2]f(x)
δ= E1/2-E-1/2
INTERPOLATION : The process of finding a
missed value in the given table values of X, Y.

FINITE DIFFERENCES : We have three finite

differences
1. Forward Difference
2. Backward Difference
3. Central Difference
 RELATIONS BETWEEN THE OPERATORS
IDENTITIES:

1. ∆=E-1 or E=1+ ∆
2.  = 1- E -1
3. δ = E 1/2 – E -1/2
4. µ= ½ (E1/2-E-1/2)
5. ∆=E  = E= δE1/2
6. (1+ ∆)(1-  )=1
Newtons Forward interpolation formula :
y=f(x)=f(x0+ph)= y0 + p∆y0 + p(p-1)/2! ∆2y0 + p(p-1)(p-
2)/3! ∆3y0+…….. ………………+p(p-1)(p2) …… [p-
(n-1)]/n! ∆ny0 .

Newtons Backward interpolatin formula :

y=f(x)=f(xn+ph)= yn+ p𝛻yn+ p(p+1)/2! 2
yn+
p(p+1)(p+2)/3!3yn+……..
………………+p(p+1)(p+2) …… [p+(n-1)]/n!nyn.
GAUSS INTERPOLATION
The Guass forward interpolation is given by yp = yo +
p ∆y0+p(p-1)/2! ∆2y-1 +(p+1)p(p-1)/3! ∆3y-
1 +(p+1)p(p-1)(p-2)/4! ∆ 4y +……
-2

The Guass backward interpolation is given by

yp = yo + p ∆y-1+p(p+1)/2! ∆2y-1 +(p+1)p(p-1)/3!
∆3y-2+ (p+2)(p+1)p(p-1)/4! ∆4y-2+……
INTERPOLATIN WITH UNEQUAL INTERVALS:

The various interpolation formulae Newton’s forward

formula, Newton’s backward formula possess can be
applied only to equal spaced values of argument. It is
therefore, desirable to develop interpolation formula for
unequally spaced values of x. We use Lagrange’s
interpolation formula.
The Lagrange’s interpolation formula is given by

Y = (X-X1)(X-X2)……..(X-Xn) Y0 +
(X0-X1)(X0-X2)……..(X0-Xn)

(X-X0)(X-X2)……..(X-Xn) Y1 +…………………..
(X1-X0)(X1-X2)……..(X1-Xn)

(X-X0)(X-X1)……..(X-Xn-1) Yn +
(Xn-X0)(Xn-X1)……..(Xn-Xn-1)
CURVE FITTING
INTRODUCTION : In interpolation, We have seen that
when exact values of the function Y=f(x) is given we fit
the function using various interpolation formulae. But
sometimes the values of the function may not be given. In
such cases, the values of the required function may be
taken experimentally. Generally these expt. Values
contain some errors. Hence by using these experimental
values . We can fit a curve just approximately which is
known as approximating curve.
Now our aim is to find this approximating curve as much
best as through minimizing errors of experimental values
this is called best fit otherwise it is a bad fit.
In brief by using experimental values the process of
establishing a mathematical relationship between two
variables is called CURVE FITTING.
METHOD OF LEAST SQUARES
Let y1, y2, y3 ….yn be the experimental values of
f(x1),f(x2),…..f(x n) be the exact values of the function
y=f(x). Corresponding to the values of x=xo,x1,x2….x n
.Now error=experimental values –exact value. If we
denote the corresponding errors of y1, y2,….yn as
e1,e2,e3,…..en , then e1=y1-f(x1), e2=y2-f(x2 )
 e3=y3-f(x3)…….en=yn-f(xn).These errors
e1,e2,e3,……en,may be either positive or negative.For our
convenient to make all errors into +ve to the square of
errors i.e e12,e22,……en2.In order to obtain the best fit of
curve we have to make the sum of the squares of the
errors as much minimum i.e e12+e22+……+en2 is
minimum.
METHOD OF LEAST SQUARES
 Let S=e12+e22+……+en2, S is minimum.When S becomes
as much as minimum.Then we obtain a best fitting of a
curve to the given data, now to make S minimum we have
to determine the coefficients involving in the curve, so
that S minimum.It will be possible when differentiating S
with respect to the coefficients involving in the curve and
equating to zero.
FITTING OF STRAIGHT LINE
Let y = a + bx be a straight line
By using the principle of least squares for solving the
straight line equations.
The normal equations are
∑ y = na + b ∑ x
∑ xy = a ∑ x + b ∑ x2
solving these two normal equations we get the values
of a & b ,substituting these values in the given straight
line equation which gives the best fit.
FITTING OF PARABOLA
Let y = a + bx + cx2 be the parabola or second degree
polynomial.
By using the principle of least squares for solving the
parabola
The normal equations are
∑ y = na + b ∑ x + c ∑ x2
∑ xy = a ∑ x + b∑ x2 + c ∑ x3
∑ x2y = a ∑ x2 + b ∑ x3 + c ∑ x4
solving these normal equations we get the values of a,b
& c, substituting these values in the given parabola
which gives the best fit.
FITTING OF AN
EXPONENTIAL CURVE
The exponential curve of the form
y = a ebx
taking log on both sides we get
logey = logea + logee bx
logey = logea + bx logee
logey = logea + bx
 Y = A+ bx
 Where Y = logey, A= logea
 This is in the form of straight line equation and this can
be solved by using the straight line normal equations we
get the values of A & b, for a =eA, substituting the values
of a & b in the given curve which gives the best fit.
EXPONENTIAL CURVE
The equation of the exponential form is of the form y =
abx
taking log on both sides we get
log ey = log ea +log ebx
Y = A + Bx
where Y= logey, A= logea , B= logeb
this is in the form of the straight line equation which
can be solved by using the normal equations we get
the values of A & B for a = eA
b = eB substituting these values in the equation
which gives the best fit.
FITTING OF POWER CURVE
Let the equation of the power curve be
y = a xb
taking log on both sides we get
log ey = logea + log exb
Y = A + Bx
this is in the form of the straight line equation which
can be solved by using the normal equations we get
the values of A & B, for a = eA b= eB, substituting
these values in the given equation which gives the
best fit.
 Solutions of Algebraic and
Transcendental equations and
Linear system of equations
Method 1: Bisection method
If a function f(x) is continuous b/w x0 and x1 and
f(x0) & f(x1) are of opposite signs, then there exsist at least one
root b/w x0 and x1
 Let f(x0) be –ve and f(x1) be +ve ,then the root lies b/w xo
and x1 and its approximate value is given by x2=(x0+x1)/2
 If f(x2)=0,we conclude that x2 is a root of the equ f(x)=0
 Otherwise the root lies either b/w x2 and x1 (or) b/w x2 and x0
depending on wheather f(x2) is +ve or –ve

Then as before, we bisect the interval and repeat the

process untill the root is known to the desired accuracy
Method 2: Iteration method or successive approximation
Consider the equation f(x)=0 which can take in the form
x = ø(x) -------------(1)
where |ø1(x)|<1 for all values of x.
Taking initial approximation is x0
we put x1=ø(x0) and take x1 is the first approximation
x2=ø(x1) , x2 is the second approximation
x3=ø(x2) ,x3 is the third approximation
.
.
xn=ø(xn-1) ,xn is the nth approximation
Such a process is called an iteration process
Method 3: Newton-Raphson method or Newton iteration method
Let the given equation be f(x)=0
Find f1(x) and initial approximation x0

The first approximation is x1= x0-f(x0)/ f1(x0)

The second approximation is x2= x1-f(x1)/ f1(x1)

.
.
The nth approximation is xn= xn-1-f(xn)/ f1(xn)
 LU Decomposition Method
a11 x1  a12 x2  a13 x3  b1
a21 x1  a22 x2  a23 x3  b2
a31 x1  a32 x2  a33 x3  b3
 a11 a12 a13   x1   b1 
A   a21 a22 a23  , X   x2  , B  b2 
     
 a31 a32 a33   x3   b3 
This is in the form AX=B Where

 l11 0 0 u11 u12 u13 

Let A=LU where L  l21 l22 0  ,U   0 u22 u23 
   
l31 l32 l33   0 0 u33 
Hence LUX=B
(a)LY=B (b)UX=Y
Solve for Y from (a) then Solve for X from (b)
LY=B Can be solved by forward substitution and UX=Y can be solved by
backward substitution
Jacobi’s Iteration Method

a1 x  b1 y  c1 z  d1
a2 x  b2 y  c2 z  d 2
a3 x  b3 y  c3 z  d 3

1
x [d1  b1 y  c1 z ]  k1  l1 y  m1z
a1
1
y [d 2  a2 x  c2 z ]  k2  l2 x  m2 z
b2
1
z [d 3  a3 x  b3 y ]  k3  l3 x  m3 y
c3

Substituting these on the right hand side we get second approximations this process is
repeated till the difference between two consecutive approximations is negligible or same
Gauss-Seidel iteration Method

a1 x  b1 y  c1 z  d1
a2 x  b2 y  c2 z  d 2
a3 x  b3 y  c3 z  d 3
1
x [d1  b1 y  c1 z ]  k1  l1 y  m1z
a1
1
y [d 2  a2 x  c2 z ]  k2  l2 x  m2 z
b2
1
z [d 3  a3 x  b3 y ]  k3  l3 x  m3 y
c3

as soon as new approximation for an unknown is found it is immediately

used in the next step
 Numerical Integration and
Numerical solution of First
order differential equations
NUMERICAL INTEGRATION
 Numerical Integration is a process of finding the value
of a definite integral.When a function y = f(x) is not
known explicity. But we give only a set of values of the
function y = f(x) corresponding to the same values of x.
This process when applied to a function of a single
variable is known as a quadrature.
NUMERICAL INTEGRATION
 For evaluating the Numerical Integration we have
three important rules i.e
 Trapezoidal Rule
 Simpsons 1\3 Rule
 Simpsons 3\8 th Rule
NUMERICAL INTEGRATION
 Trapezoidal Rule : The Trapezoidal Rule of the function y
= f(x) is given by
 f (x ) dx= h\2 ( y0 + yn ) + 2 ( y1 + y2 + y3+….+ yn-1 )
 f (x ) dx = h\2 (sum of the first and last terms ) + ( sum of
the remaining terms )
NUMERICAL INTEGRATION
 Simpson’s 1\3 rd Rule : The Simpson’s 1\3 rd Rule of the
function f ( x ) is given by
 f ( x ) dx = h/3 ( y0 + yn ) + 4 ( y1 + y3 + y5 + …..yn-1 ) + 2 (
y2 + y4 + y6 + …..)
NUMERICAL INTEGRATION
 Simpson’s 3/8 th Rule : The Simpson’s 3/8 th rule for
the function f ( x ) isgiven by
 f ( x ) dx = 3h/8 ( y0 + yn ) + 2 ( y3 + y6 + y9 + ……) + 3 ( y1
+ y2 +y4 +……)
 f (x ) dx = 3h/8 ( sum of the first and the last term) + 2 (
multiples of three ) + 3 ( sum of the remaining terms )
INTRODUCTION
 There exists large number of ordinary differential
equations, whose solution cannot be obtained by the
known analytical methods. In such cases, we use
numerical methods to get an approximate solution o f
a given differential equation with given initial
condition.
NUMERICAL DIFFERIATION
 Consider an ordinary differential equation of first
order and first degree of the form
 dy/dx = f ( x,y ) ………(1)
 with the intial condition y (x0 ) = y0 which is called
initial value problem.
 T0 find the solution of the initial value
problem of the form (1) by numerical methods, we
divide the interval (a,b) on whcich the solution is
derived in finite number of sub- intervals by the
points
TAYLOR’S SERIES METHOD
 Consider the first order differential equation
 dy/dx = f ( x, y )……..(1)
with initial conditions y (xo ) = y0 then expanding
y (x ) i.e f(x) in a Taylor’s series at the point xo we
get
y ( xo + h ) = y (x0 ) + hy1(x0) + h2/ 2! Y11(x0)+…….
Note : Taylor’s series method can be applied only
when the various derivatives of f(x,y) exist and the
value of f (x-x0) in the expansion of y = f(x) near x0
must be very small so that the series is convergent.
PICARDS METHOD OF SUCCESSIVE
APPROXIMATION
 Consider the differential equation
 dy/dx = f (x,y) with intial conditions
 y(x0) = y0 then
 The first approximation y1 is obtained by
 y1 = y0 + f( x,y0 ) dx
 The second approximation y2 is obtained by y2 = y1 + f
(x,y1 ) dx
PICARDS APPROXIMATION METHOD
 The third approximation of y3 is obtained y by y2 is
given by
 y3 = y0 + f(x,y2) dx……….and so on
 …………………………………
 ………………………………….
 yn = y0 + f (x,yn-1) dx
 The process of iteration is stopped when any two
values of iteration are approximately the same.
 Note : When x is large , the convergence is slow.
EULER’S METHOD
 Consider the differential equation
 dy/dx = f(x,y)………..(1)
 With the initial conditions y(x0) = y0
 The first approximation of y1 is given by
 y1 = y0 +h f (x0,,y0 )
 The second approximation of y2 is given by
 y2 = y1 + h f ( x0 + h,y1)
EULER’S METHOD
 The third approximation of y3 is given by
 y3 = y2 + h f ( x0 + 2h,y2 )
 ………………………………
 ……………………………..
 ……………………………...
 yn = yn-1 + h f [ x0 + (n-1)h,yn-1 ]
 This is Eulers method to find an appproximate
solution of the given differential equation.
IMPORTANT NOTE
 Note : In Euler’s method, we approximate the curve of
solution by the tangent in each interval i.e by a
sequence of short lines. Unless h is small there will be
large error in yn . The sequence of lines may also
deviate considerably from the curve of solution. The
process is very slow and the value of h must be smaller
to obtain accuracy reasonably.
MODIFIED EULER’S METHOD
 By using Euler’s method, first we have to find the value
of y1 = y0 + hf(x0 , y0)
 WORKING RULE
 Modified Euler’s formula is given by
 yik+1 = yk + h/2 [ f(xk ,yk) + f(xk+1,yk+1
 when i=1,y(0)k+1 can be calculated
 from Euler’s method.
MODIFIED EULER’S METHOD
 When k=o,1,2,3,……..gives number of iterations
 i = 1,2,3,…….gives number of times a particular
iteration k is repeated when
 i=1
 Y1k+1= yk + h/2 [ f(xk,yk) + f(xk+1,y k+1)],,,,,,
RUNGE-KUTTA METHOD
The basic advantage of using the Taylor series
method lies in the calculation of higher order total
derivatives of y. Euler’s method requires the
smallness of h for attaining reasonable accuracy. In
order to overcomes these disadvantages, the
Runge-Kutta methods are designed to obtain
greater accuracy and at the same time to avoid the
need for calculating higher order derivatives.The
advantage of these methods is that the functional
values only required at some selected points on the
subinterval.
R-K METHOD
 Consider the differential equation
 dy/dx = f ( x, y )
 With the initial conditions y (xo) = y0
 First order R-K method :
 y1= y (x0 + h )
 Expanding by Taylor’s series
 y1 = y0 +h y10 + h2/2 y110 +…….
R-K METHOD
 Also by Euler’s method
 y1 = y0 + h f (x0 ,y0)
 = y0 + h y10
 It follows that the Euler’s method agrees with the
Taylor’s series solution upto the term in h. Hence,
Euler’s method is the Runge-Kutta method of the first
order.
 The second order R-K method is given by

 y1= y0 + ½ ( k1 + k2 )
 where k1 = h f ( x0,, y0 )
 k2 = h f (x0 + h ,y0 + k1 )
 Third order R-K method is given by

 y1 = y0 + 1/6 ( k1 + k2+ k3 )
 where k1 = h f ( x0 , y0 )
 k2 = hf(x0 +1/2h , y0 + ½ k1 )
 k3 =hf(x0 + ½h , y0 + ½ k2)
The Fourth order R – K method :
This method is most commonly used and is often
referred to as Runge – Kutta method only. Proceeding
as mentioned above with local discretisation error in
this method being O (h5), the increment K of y
correspoding to an increment h of x by Runge – Kutta
method from
dy/dx = f(x,y),y(x0)=Y0 is given by
 K1 = h f ( x0 , y0 )
 k2 = h f ( x0 + ½ h, y0 + ½ k1 )
 k3 = h f ( x0 + ½ h, y0 + ½ k2 )
 k4 = h f ( x0 + ½ h, y0 + ½ k3 )
 and finally computing
 k = 1/6 ( k1 +2 k2 +2 k3 +k4 )
 Which gives the required approximate value as
 y1 = y0 + k
R-K METHOD
Note 1 : k is known as weighted mean of
k1, k2, k3 and k4.

Note 2 : The advantage of these methods is that the

operation is identical whether the differential
equation is linear or non-linear.
FINITE DIFFERENCE METHOD
Characteristic matrix: Let A be a square matrix of order n then
the matrix (A-λI) is called Characteristic matrix of A.where I
is the unit matrix of order n and λ is any scalar
1 2 
 
Ex: Let A= then

3 4 
A- λI= 1   2  is the characteristic matrix of A
 
3 4
Characteristic polynomial: Let A be square matrix of order n then | A- λI| is
called characteristic polynomial of A

1  2 
Ex: Let A=  
4 1 
then | A- λI| = λ2-2λ+9 is the characteristic polynomial of A
Characteristic equation: Let A be square matrix of order n then
| A- λI|=0 is called characteristic equation of A

1  2 
Ex: Let A=  
4 1

then | A- λI| = λ2-2λ+9=0 is the characteristic equation of A

Eigen values: The roots of the characteristic equation | A- λI|=0 are
called the eigen values
Ex: Let A=  2 1  then | A- λI|=λ2-6λ+5=0
 
3 4

Therefore λ=1,5 are the eigen values of the matrix A

 Eigen vector: If λ is an eigen value of the square matrix A. If
there exists a non- zero vector X such that AX=λX is said to be
eigen vector corresponding to eigen value λ of a square matrix A

 Eigen vector must be a non-zero vector

 If λ is an eigen value of matrix A if and only if there exists a non-

zero vector X such that AX=λX

 I f X is an eigen vector of a matrix A corresponding to the eigen

value λ, then kX is also an eigen vector of A corresponding to the
same eigen vector λ. K is a non zero scalar.
Properties of eigen values and eigen vectors
 The matrices A and AT have the same eigen values.

 If λ1, λ2,…..λn are the eigen values of A then

1/ λ1, 1/λ2 ……1/λn are the eigen values of A-1.

 If λ1, λ2,…..λn are the eigen values of A then λ1k, λ2k,…..λnk are
the eigen values of Ak.

 If λ is the eigen value of a non singular matrix A, then |A|/λ is

the eigen value of A.
 The sum of the eigen values of a matrix is the trace of
the matrix

 If λ is the eigen value of A then the eigen values of

B= aoA2+ a1A+a2I is aoλ2+a1λ+a2.
Similar Matrices : Two matrices A&B are said to be
similar if their exists an invertable matrix P such
that B=P-1AP.

 Eigen values of two similar matrices are same

 If A & B are square matrices and if A is invertable then the matrices A-1B
& BA-1 have the same eigen values
 Power Method
Power method for finding eigenvalues
1. Start with an initial guess for x
2. Calculate w = Ax
3. Largest value (magnitude) in w is the
estimate of eigenvalue
4. Get next x by rescaling w (to avoid
the computation of very large matrix
An )
5. Continue until converged
 Power Method
 Start with initial guess z = x0
w ( 1 )  Az ( 1 )  (k1 )  w k( 1 )
w(1) Az ( 1 )
z( 2 )  
 (1)
k (k1 ) rescaling
w ( 2 )  Az ( 2 )  (k2 )  w k( 2 )
w( 2 ) Az ( 2 )
z( 3 )   k is the
 (2)
 (2)
k k dominant
 eigenvalue
If 1  2  3    n , then
1  2  3    n
k k k k
 Power Method
1. Initial guess z (1)  x0  1,1,...,1 T

2. Calculate Az (1)  w(1)   z (2) ;  normalize z by biggest wk 

3. Calculate Az (2)  w(2)   z (3) ;  normalize z by biggest wk 

... Calculate Az ( k )  w( k )   z ( k 1)

 For large number of iterations,  should

converge to the largest eigenvalue
 The normalization make the right hand side
converge to  , rather than n