TOPIC: NUMERICAL ANALYSIS MASTER CADRE MATHS SINGLA MATHS POINT
NUMERICAL DIFFERENTIATION
FORWARD DIFFERENCE OPERATOR
∆f(x)=f(x+h)-f(x)
is called first order forward difference operator
∆2 f(x)= ∆[∆f(x)]= ∆ [f(x+h)-f(x)]= ∆f(x+h)- ∆f(x)
⟹ f(x+2h)- f(x+h)-f(x+h)+ f(x)
⟹ f(x+2h)- 2f(x+h)+ f(x)
is called second order forward difference operator
or
∆yk=yk+1- yk
BACKWARD DIFFERENCE OPERATOR
∇f(x)= f(x) - f(x-h)
CENTRAL DIFFERENCE OPERATOR
ℎ ℎ
𝛿f(x)=f(x+ 2)- f(x - 2)
SHIFT OPERATOR
Ef(x)=f(x+h)
E-1f(x)=f(x-h)
E2f(x)=f(x+2h)
E3f(x)=f(x+3h)
.
Enf(x)=f(x+nh)
AVERAGE/MEAN OPERATOR
ℎ ℎ
f(x+ )+ f(x − )
2 2
𝜇f(x)= 2
RELATIONS BETWEEN OPERATORS Δ, ∇ AND E
1. Δ ≡ E −1 (Relations between the operators Δ and E)
Proof: From the definition of Δ we know that
Δ f (x) = f (x + h) − f (x) and E [ f (x)] = f (x + h)
where h is the interval of difference.
Δ f (x) = f (x + h) − f (x)
Δ f (x) = Ef (x) − f (x)
⇒ Δ f (x) = (E −1) f (x)
Δ ≡ E −1
∴ E≡1+Δ
2. EΔ ≡ Δ E
Proof:
E(Δf(x)) = E[f(x+h)- f(x)]
= E f(x+h) - E f(x)
= f(x+2h) - f(x+h)
=Δf(x+h)
=ΔEf(x)
EΔ ≡ ΔE
3. ∇≡ 1-E-1 (Relations between the operators ∇ and E )
∇f (x) = f (x) - f (x-h)
= f (x) – E-1f (x)
= (1- E-1) f (x)
∇ ≡ (1- E-1)
∇ ≡ 1 – 1/E
Hence ∇ ≡ [E – 1]/E
4. 𝜹=E1/2 – E-1/2 (Relations between the operators 𝛿 and E)
ℎ ℎ
𝛿f(x)=f(x+ 2)- f (x - 2)
𝛿f(x)=E1/2f (x) – E-1/2 f (x)
𝛿=E1/2 – E-1/2
5. 𝜇=1/2[E1/2 + E-1/2] (Relations between the operators 𝜇and E)
ℎ ℎ
f(x+ )+ f(x − )
2 2
𝜇f(x)= 2
𝜇f(x)=1/2[E1/2f (x) + E-1/2 f (x)]
𝜇=1/2[E1/2 + E-1/2]
RELATIONS BETWEEN OPERATORS Δ, ∇ AND E
Properties
Few common properties of E operator are given below:
(i) Ec = c, where c is a constant.
(ii) E{cf(x)} = cEf(x).
(iii) EmEnf(x) = EnEmf(x) = Em+nf(x).
(iv) EnE−nf(x) = f(x).
In particular, EE−1 ≡ I, I is the identity operator and it is sometimes
denoted by 1.
(v) (En)mf(x)=Enmf(x).
(vi)E{f(x)g(x)}=Ef(x)Eg(x).
(vii)E∆f(x)=∆Ef(x).
fundamental theorem of difference calculus?It states that nth
difference of a polynomial of n degree is constant thus (n+1) th
difference will be zero.
BISECTION METHOD
Assume that f(x) is continuous on a given interval [a, b] and that is
also satisfies f(a)>0, f(b) < 0 with f(a) ≠ 0 and f(b) ≠ 0. Using the
intermediate value theorem, we can see that the function f(x) has
atleast one root in [a, b]. We assume that there is only one root for
the equation in the interval [a, b]. The Bisection includes the
following steps:
Y=f(x)
f(a)=+ve i.e. f(a)>0
f(b)=-ve i.e. f(b) < 0
then root lies in (a,b) if sign is opposite.
(a + b)
Root = x1 = 2
(a + b)
Let the approximate value of the root be x1 = 2
i.e. the point of
bisection of the interval (a, b), now if we evaluate f(x1) there are
three possibilities:
i. f(x1)=0, in which case x1 is the root.
ii. f(x1)<0, in which case the root lies in the interval (a, x₁).
iii. f(x1)>0, in which case the root lies in the interval (x1, b).
The process of bisection of the interval containing the root can be
repeated until the root is obtained to the desired accuracy.
PROPERTIES:
1. Order/Rate of convergence of bisection method is one.
2. Bisection method is never failed.
3. Least rate of convergence.
4. Convergence is assured in the bisection method.
5. This method is unable to find the double root.
6. Bisection method is rather slow; it takes a large number of
iterations to achieve a reasonable degree of accuracy for the root.
7. Rate of cgs is linear.
REGULA-FALSI /FALSE POSITION METHOD
f(a)=+ve i.e. f(a)>0
f(b)=-ve i.e. f(b) < 0
then root lies in (a,b) if sign is opposite.
𝑎𝑓 (𝑏 )−𝑏𝑓(𝑎)
First approximation is = 𝑓 (𝑏 )−𝑓(𝑎)
PROPERTIES:
1. Order/Rate of convergence of regula-falsi method is 1 or 1.16.
SECANT METHOD
Secant method is one of the most efficient methods among all
regula-falsi methods. Let us first explain the regula-falsi method and
given the modification in this method which leads to secant method.
The regula-falsi method is closely
related to the bisection method. Recall the bisection method is to
subdivide the interval [a, b] in which the root lies into two parts,
take the part of the interval which still holds the root and discard
the other part of the interval. Although the bisection method always
converges to the solution, the convergence is sometime very slow in
the sense that if the root is very close to one of the boundary points
(ie.,. a and b) of the interval. In such a situation, instead of taking the
midpoint of the interval, we take the weighted average of f(x) given
by
𝑎𝑓 (𝑏 )−𝑏𝑓(𝑎)
First approximation is = 𝑓 (𝑏 )−𝑓(𝑎)
PROPERTIES:
1. Order/Rate of convergence of Secant method is 1.618.
2. Convergence is not sure in this method.
3. Golden rate of convergence.
NEWTON-RAPHSON METHOD
The Newton-Raphson method is one of the fastest iterative
methods. In this method one have to take due care while selecting
the initial approximation as it is very sensitive to the initial
approximation.
If f(x) is differentiable, then nth approximation by Newton-Raphson
𝑓(𝑥𝑛)
method is Xn+1=xn − 𝑓′(𝑥
𝑛)
𝑓(𝑥0)
X1=x0 − 𝑓′(𝑥
0)
𝑓(𝑥1)
X2=x1 − 𝑓′(𝑥
1)
PROPERTIES:
1. Order/Rate of convergence of Newton-Raphson method is 2.
2. This method is fast as compare to other methods so it is the best
method for finding the roots of an equation.
3. Newton-Raphson method is failed when 𝑓′(𝑥𝑛) =0
4. This method is not applicable when f' (x) = 0.
5. Greater the numerical value of derivative f' (x), higher the speed
of convergence.
6. This method requires two evaluations per iteration so the
computation per iteration is more as compared to another methods.
7. it is very fast method.
8. Convergence is not sure in this method.
9. Quadratic rate of cgs.
FUNDAMENTAL THEOREM OF ALGEBRA
It states that every polynomial equation of degree n with complex
number coefficients has n roots, or solutions, in the complex
numbers.
The theorem can also be stated as every polynomial equation of
degree n where n ≥ 1 with complex number coefficients has at least
one root.
METHOD TO SOLVE ORDINARY DIFFERENTIAL EQUATIONS:
1. PICARD'S METHOD:
𝑑𝑦
Consider the following differential equation: =f(x,y) initial
𝑑𝑥
condition that y = y0 at x = x0 or y(x0) = y0.
Integrating the above equation between the limits x0 and x, we get'
𝑦 𝑥
∫𝑦0 𝑑𝑦=∫𝑥0 f(x, y)dx
𝑥
⟹ y-y0=∫𝑥 f(x, y)dx
0
𝑥
⟹ y = y0+∫𝑥 f(x, y)dx
0
𝑥
Now, first approximation y1 = y0+∫𝑥 f(x, 𝑦0 )dx
0
𝑥
Second approximation y2 = y0+∫𝑥 f(x, 𝑦1 )dx
0
And
𝑥
nth approximation is yn = y0+∫𝑥 f(x, 𝑦𝑛−1 )dx
0
2. EULER’S METHOD
𝑑𝑦
Consider the following differential equation: =f(x,y) initial
𝑑𝑥
condition that y = y0 at x = x0 or y(x0) = y0.
Now first approximation y1 = y0 + hf(x0, y0)
Second approximation y2 = y1 + hf(x1, y1)
And nth approximation is yn =yn -1 +hf(xn-1, yn-1)
Where h is step size
3. RUNGE-KUTTA METHOD
FIRST ORDER R-K METHOD:
Euler's method is the Runge-Kutta method of the first order.
SECOND ORDER R-K METHOD:
Modified Euler's method is the Runge-Kutta method of the second
order.
1
yn= yn-1 + 2(k1+k2)
THIRD ORDER R-K METHOD:
1
yn= yn-1 + 6(k1+4k2+k3)
FOURTH ORDER R-K METHOD:
1
yn= yn-1 + 6(k1+2k2+2k3+k4)
NEWTON-COTES QUADRATRE FORMULA:
If we put values of n in Newton-Cotes Quadrature Formula then
N=1 ⟹ Trapezoidal rule
N=2 ⟹ Simpson’s 1/3 rule
N=3 ⟹ Simpson’s 3/8 rule
N=4 ⟹ Boole’s rule
N=6 ⟹ Weddle’s rule
TRAPEZOIDAL FORMULA
One of the simple quadrature formulae is trapezoidal formula. To
obtain this formula, we substitute n = 1 to the equation
Function should be defined on [a, b] & dividing [a, b] into n equal
𝑏−𝑎
parts, each of length h= s.t.
𝑛
𝑥0 +𝑛ℎ 𝑏 ℎ
∫𝑥0 f(x)dx=∫𝑎 f(x)dx=2 [(y0+yn) +2(y1+y2+…………. +yn-1)]
ℎ
= 2 [(sum of first & last term) +2(remaining terms)]
Note:
1. it can be applied for any type of interval.
2. Here [a, b] divided into n equal parts
3. curve making in this rule is st. line
𝑏−𝑎
4. h= 𝑛
5. The error term in trapezoidal formula indicates that if the second
and higher order derivatives of the function f(x) vanish, then the
trapezoidal formula gives exact result. That is, the trapezoidal
formula gives exact result when the integrand is linear.
6. trapezoidal rule is nothing but the area of the trapezium ABCD.
7. Error in the Trapezoidal rule is of the order h2.
SIMPSON’S 1/3 RULE
We substitute n = 2 to the equation to get another quadrature
formula. Thus,
𝑥0 +𝑛ℎ 𝑏 ℎ
∫𝑥 f(x)dx=∫𝑎 f(x)dx= [(y0+yn) +2(y2+y4+ y6+…)+4(y1+y3+ y5+…)]
0 3
ℎ
= [(sum of first & last term) +2(even terms) +4(odd terms)]
3
ℎ
=3 [(y0+yn)+2E+4O]
Note:
1. it can be applied for even (multiple of 2) interval.
2. Here [a, b] divided into 2n equal parts
3. Curve making in this rule is parabola.
𝑏−𝑎
4. h= 2𝑛
5. Quadratic curve
6. Better than Simpson’s 3/8 rule.
7. Most accurate for 2 degree.
8. Error in the Simpson one-third rule is of the order h4
SIMPSON’S 3/8 RULE
This formula is obtained by substituting n = 3
𝑥0 +𝑛ℎ 𝑏 3ℎ
∫𝑥 f(x)dx=∫𝑎 f(x)dx= [(y0+yn)+2(y3+y6+y9+…)+3(remaining terms)]
0 8
Note:
1. it can be applied for multiple of 3 intervals.
2. Here [a, b] divided into 3n equal parts
Note that all these quadrature formulae are applicable for the
proper integral. If either lower or/and upper limit(s) is (are)
infinity (first type improper integral) or f(x) has infinite
discontinuity within the interval of integration [a,b] (second type
improper integral), then none of these formulae are applicable.
INTERPOLATION BY POLYNOMIALS
The following assumptions are assumed for interpolation:
(i) The function is a polynomial function.
(ii) There is no sudden jump or fall in the values.
(iii) The function is one-one.
INTERPOLATION FORMULA FOR EQUAL
DIFFERENCE
1. NEWTON'S FORWARD DIFFERENCE FORMULA
2. NEWTON'S BACKWARD DIFFERENCE FORMULA
3. GAUSS FORWARD DIFFERENCE FORMULA
4. GAUSS BACKWARD DIFFERENCE FORMULA
INTERPOLATION FORMULA FOR UNEQUAL
DIFFERENCE
1. LAGRANGE’S INTERPOLATION FORMULA
2. HERMITE'S INTERPOLATION FORMULA
3. NEWTON'S DIVIDED DIFFERENCE FORMULA
SOLUTION OF SYSTEM OF SIMULTANEOUS
LINEAR EQUATIONS:
ITERATIVE METHODS:
1. Gauss Siedel Method
DIRECT METHODS:
1. Gauss Elimination Method
2. Gauss Jordan Method
GAUSS SIEDEL METHOD
As soon as new value of variables is found by iteration it use
immediately in equation.
GAUSS ELIMINATION METHOD
1. Back substitutioJn method
2. Making matrix is upper triangular matrix.
GAUSS JORDAN METHOD
Making matrix is Diagonal matrix.
FOR JOINING JOIN OUR COURSE
CONTACT AT: 98128-25287