Ifadu Esfais

Garss Qudrature Rule:

wi
Review on
previous approaches:

from Reimann sum, we observed that numerical solutionfor

the

integration can be written in

quadraticform (i.e. weighted ↑

values at specified pointswithin domain ofintegrations

Sum
offruction

⑪ Newton-Cotes andrature formutus use

specified stined) points T
the interpolate a curve

usingpolynomial interpolation techniques (e.g. Lagrangium). Then Determine

The weights by enactsolving the integral for example polynomials.

-dal
Rule:

1-
specify two points ↓
those for point. approach
Lagrangian
interpolate line from -
2- a

Switch- Wifen, twefue

Interpolation polynomial
fat xo! fon)
i
u? (x, -40) i0
=

3 - Determine the exactsolution for a Firstorder polynomial(line)

y for
(m)(fens+frn,) =>(*)faro + (*

is exact for itorder since weight determined for isorder poly-

Trapezoidal in exactly
Sets
Rute:

specify
1- three points -
Equally spaced
int
2 -

interpolate a 2nd order polynomial from three points ->

Lagrangian approach
(n-noix-n)

Interpolation polynomial fan)-- fonz

m-ne fast a
:
(r -20) (4,-42) (x2-20) (12-2,
(Ko-x,)(4-42)

3 - Determine the exactsolution for a and order

polynomial (quadratic function.

Simpson exact foreorder since weight determined "order

by is

-
in exactly for poly-

/Rute:
Equally spaced

specify
1- four its
poin
/Moe, a de
->
↑

2 -

interpolate a
rd
order polynomial from three points ->
Lagrangian approach
Interpolation polynomial : Locnsfon+ Hicnfal + tamfan+tyfund

3 - Determine the exactsolution for a 2nd order

polynomial (quadratic function.

is exact for order since weight determined for order poly-

Trapezoidal in exactly
 Therefore, Rule No, of requiredpoint order ofpoly to have enact solution
impezoidal 2 I

Simpson 113 3 2

Simpson 3/8
4 3
&

Question: Are we able to use tower number ofpoints to have enact value for higher

polynomial?
② Garss. Quadrature Rule:

we don't have prespecifiedpoints. The

points should the determined such that

we recievedexact value for a high order

polynomial
for example order
2
points - > should I exact for
poly-
3
points -a 4th order poly-
There five: b n
points -> s2n-19th order
poly.
Ifraidn=2wifm
??
one-point Gauss quadrature rule:
No. ofpts.
Find the
point, and w, so thatthe resultisexactfor
up
to 21=1.
P(x) 90
P,
=

d. a,x

-
M) =
+

p() a,x =

representative polynomials fal=1; fes =2

so the
integral cam on written

Sdn 2
=
asj'fousda-e, for

Sndn-win
I

1, de-w,;
i

inde o
=
two-point Gauss quadrature rule: No. ofpts.
Find two pointmine and w, we so thatthe resultisexactfor
up to
241=3
representative polynomials fal=1; fas=n; faxl-x2; fm3=x'

So the
integral can be written as I

-
jdn
I
=

I, dx w, W2
2

= +

Linde =
Juss-win,
-

i
I
we
areafee e

I nde 0
Guide s
-

=wx? +we
=

~I

E
W, Nz 2
syst.
+ =

of W,x,+W2kz
i YW W2=
0
=

x -
-D =
=
=

monline -

We, + Wx?=s
equations we? Wx?o
I For up to sitpoly enact
is

(adn =6(
so
-55)+f(-) ** Since we

x,
exactly
and
end
or
determine
for
up to poly.
 -

↑
 ofintegral:
-range
In Gauss
quadrature rule, range ofintegral is
alwayI,). I
the
Using
relation,
following we can
alway convert any finite integral to
1,4
--
Linear
fee.......... I-----
·

x
.u
.

a
=
x b
=
transformation = -1

x mt n
=
+

S m=B
e
a mc 1) 4
x mt+n
-

E
+
=

- D
=

b m(1) +n
=

E
xa 1

bat a xt 5a(2x
= = = -

a =
+

=
= -
(a b)
+

x b
=
- t +1 =

du=b, dt

"Fida-a fetidt
"Stre by Gauss-Quadrature rule.
Example: b
1
=

Solve
I
Faith Gauss
using quadrature rule:
at

t
ia(2x b))
= -(a +

2x
=
-
1 -x x
t;dn Edt
=
=

I
faacti=f(t)-
IT 4
-

4 +

!Enzdu= I! 4dt (f- s) 4

+
=
+
flte))

(0.96725
0.61652)
+ 0.78r
=

zdn wif wif f(

=
(n,1 +

() = -

(5). f()+

()+(is) 5 5 =
=
= +

Analytical
Sol.
Inda-bu?!" 5 -15) 3 =
=

Example;two-point GQ
Sudn=wyf(u + wif (n) f( =
-

) f() (i)(e)" +-
+ =

manualSuidn =45)
S

5 5-I):*: 5
=
-
tion
-
to
differentialEquations:
Euler's methods
Most real physical processes are
governed only by differential Equations. IV-UDEXRK's methods
-
Differental Equations Ordinary Differential Equatius CODE) BU_ODE
-
- >

I
relationship between function
A a
ofa single independant variable and

total deviatives ofthis

the
fraction art the
independant variable.

e.g. t=fit,y) FDM

-
- >
Partial Differential Equations (PDE): -

-Arm
FEM

relationship between
A fraction oftwo a or more
independant
variables and partial derivatives of its function art these

independent variables.
if at
eg.

Ordinary Differential Equations (ODE):

the general with-order ODE for gets is

any"+ any*.+azy"+ 9,y'rary=fet

Note: The
general solution of a ODE contains one or more constant ofintegration.
This a
family ofsolutions is obtained. The
number
ofintegration equal is

the order of the differential

to
equation. The particular member ofthe
conditions.
family of solution which is
ofinterest is determined
by anitary
ey,

- p Arial Displacement of
a uniform rod.

In (EAd] x=b
-
=

/EAG) =
=

(d/EAGT=) =
EA =

EA/du=c,du um)
E+ 2 (ocx 7.) =
= =

we know;

(10) = o

:(en) *
ihm Mult ·xI1
 classification ofODES:
There are two different
types (classes) ofODES, depending on the annitary
condition specified. *=f(y,t
yut' a
① Intial-value ODE -
ifall own.
Conditions are
specified
at the

vanable.
same value
ofthe independent
y,
-
⑧
& yet

*
I,

② Boundary-value ODE
-> ifthe arm, conditions are
specified
at two different value ofthe independent
variable.
yet

t,i
I, ⑧
t, to
Solution ofisorder initial-value ordinary differential equationss
Numerical Solutionto storer can he into
classified
IV-ordinary
equations
into
① single-step
/
method: use the
only one
pastvalue to march the

Solution
In+= F( an,In yuch.)
byn, hxut-tut
② multi-step methods use several ofgens & gins to construct
past values a

and entrapolate
polynomial that approximate the deviatives into the next

a.
interval. Int1 f(ansmn..... KakH'Yuth--Yurkts 1.
*

Method:
Series
order
fetis consider the nonlinear
B= fimy):
ODE
yo-yo
about
Taylor series
expansion offunction yous no
n=

yous y(0+y'ur(x-xo+I, y"nam-ro+----IY"mol(x-n"----

=

The order
higher can be
derivatives determined by successively differentiating the lower-

order derivatives, such that

y"- cy's= Ery's, y"=y'cy"s= I (3")...

eg.
Be
Taylor series method to solve
-- say ·
yc01-1 and obtain y10.1)

zmy; y'c0)
=
y
- 0
=

y" =((2xy) = =
-2y- 21n--zy-zay; "w)=-2

y"= 11 27-2y')
-
= -
4y'-zy"; y'"wl--
(4)

y* (-4y'-zy") =-by"-zay";
=
y I 12

15)

y =
-

by- zy i y 0
=

y(6) =
-
yy) zxy(5)
-

;y* =

- 96

subsitutinginto the Taylor series formula, obtain we

yu 1-x - =
+ + m
15

y(0.1) 1-10.12
(*- ol =0.9200448
= +
backs:
and differentiation
require integration
- As these method analytically, they can not

be
implemented on the
computer.
methods with simple arithmetic operation be
implemented the
computer)
sonly
can on

become
-
obtaining higher-order derivative
progressively mor
complicated. Therefore .
in

practice Taylor series method can not use higher terms

deviative for most
offunctionwith
deviatives.
Complex
-> Even
by higher-order
using
terms. The is
Solution not
very accurates over the whole domain.

Notice; Senles Feature

Taylor

Let us
approximate coscu) 4

in -4[u=6.
 Recal 6

raydon series method: Yut,"Yet hersen yn)

ne
+ yicuniyy"cany)*...
y'(uniyn)
itcan the written as follow

Yut=Yuthf(an1y+Ent
↳residual or Trucation ever

Methot step size his

very
small thus, och"can he
neglected
in

Yut *InthF(un, Yn!

Explicit (Reward Euler) Methods

sigible
Euler

the general
Consider nonlinear ODE:

y'-fcary), yM0 y, =

in
aboutan y'
Truncation ofTaylor series expansion -

Inx In+h Farm. In

=

with Inti)
semplicitfuncion
Features:

1 - This method explicit

is since in does not
depend on
Inti

2. This method only

require
one known point
- Hence itisa single pointmethod.
3 -
This method
require only
one
functionevaluation.
derivative [y's per step.

Jut1 =

Y Am
+

aJane, e

eg( x-y2,
I
=
y(11 =2, An=0.1, 14

Explicit No =1, y= 2
Enter
method:
Yn+1 In+
=

Axy' =x
y, Yo+
=

Anf(x,y) y, =2 0.1(1
-4 +
-
22 2
= +

0.1) 3)-

I, 1.7
=

1 +0.1=1.)
2, No+ Ax
= =

4, 1+ 0-1 1.1, y, 67
= =

yz y,=
+

Anf(x,y,) xyz 1.710.1(1.1-1.7?)

= =

=
-
 Implicit (Backward) Enter methods

the general

↳
Consider nonlinear ODE

y'-fcary), yM0 y, =

aboutanti
Truncation ofTaylor series expansion
In Ynx
= -

hF(Rn+1/Yn+1.

->
Yuts Inth cent, Intl
=

simplicitfunction
ar t Inti)
Features:

1 - This method is
implicit since fits depends on
Inti

2. This method only

require
one known point - Hence itisa single pointmethod.
3 -
This method
require only
one
functionevaluation.
derivative [y's per step.
4 -
This method
/
require
to solve a nonlinear
equation.
-
yo f(x,y)
dy -
e.g.
in
=
x-y 1y(11 = 2, An=o.1, U14
(k., Jv) (1,
=

2)

ImplicitEnter Abynt
method:Yut -but **Y=Y.+ Anfer, y,) 2, It01=1.1
=

=>
y, 2
=
0.1(11-y,7
+ = 2+
0.11-0.1y?
->
v1y? -y, +2.11
-
0
=
y!
= -
=
11.7547X

Ax
⑯+
=
3,
1.7897-
So x, 41& y, 1.7897
=

22x,
=

An
=
+

12 y, + =

Auf(x,y) -
xyz 1.7497+
=
0.1 (1.2-y))
·32 1.7847 +0.12-0.1y?
=

-
-0.14% -y, + 1.909750
=
y -
11.6406 X

*
3=1.64056 U
⑧