0% found this document useful (0 votes)
120 views129 pages

Notes APL101

1) The document proves that if y1 and y2 are two linearly independent solutions of a given differential equation, then W(y1,y2) is constant, where W(y1,y2) is the Wronskian of y1 and y2. 2) It assumes the solutions to a differential equation of order 2 can be expressed as a linear combination of e^mx and finds the particular solution using the method of undetermined coefficients. 3) It introduces the method of operators to find the particular solution of a non-homogeneous differential equation, which involves applying derivative operators to the right hand side until it matches the original differential equation.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
120 views129 pages

Notes APL101

1) The document proves that if y1 and y2 are two linearly independent solutions of a given differential equation, then W(y1,y2) is constant, where W(y1,y2) is the Wronskian of y1 and y2. 2) It assumes the solutions to a differential equation of order 2 can be expressed as a linear combination of e^mx and finds the particular solution using the method of undetermined coefficients. 3) It introduces the method of operators to find the particular solution of a non-homogeneous differential equation, which involves applying derivative operators to the right hand side until it matches the original differential equation.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 129

APLIO]

-> em :

⑦suppose y2 & Ye awe two linearly


independent sol of the given
diffrential can

God
doch
+

and +

azy

Prove that
,

Leee
y
dY-32d
-

s
-> if I
and ye are linealy
inde So M SO
they 11
.
.

staisfy the solutions ,


barded and
- are
: e
& xy2 -
& xY1

a .

(d .

31
did
az(ydy y2d)
0
=
+
-

-
Z

a. d
So +

azE
-

O
,

d
d =

In z =
-

2x
+

c
Ro

I >C
-

C
z
=

2
Tu 2
3
C
z
=

Le
-

cet
32d-Bads=
#hence proved .


and +

and +aoy= F

y(x)
=

y= ( +

yp(x)

911y
and 90y =
0
+ +

dx

-> assume sol to be A eme


y =

as the order is 2 we should


get two roofs in the
basis so ,
is if roots are discrete the
I

y (x =
(zem2 +

Crea

(ii) if roots are same ,


m x mx
(x) (2 e
y= +Ce
=

(iiis if routs are


imaginary ,

2
m
=

y = e)(zei +

(e
+By

# Ending a pattlar sol


is Method of undetermind
coefficient ,

↳ (use when
you've got
s ,
Sinx ,
cos ,
Usin
N eP+
x ex , Since ,

COS +

Sinx)
exd-24-by dx
=

-> homogeneous solw


3x
-

x
e
y e y
=
=

x
Cze3x
-

&y
= =
(x) = + 2
-> non-homogeneous sol

Step :

i do the derivative of
RHS until you get
the same function
as original
(ii) make the set of
"distinct function"
-

-> then write


here get

Speak
we
Yp(x) as

the linear
F(x) = 24 combination
F((x =
4e4x of distinct
Sol
...........
so
Yp
=

Ac4x

4Ae4
Yp'
=

Yp"
=
16 Ae
*
16 Ae-2 4A-3Ae= . 2e

4x
#
yp(x
=
2 &

4
3x - x

Cze 2 &
y See
A
+
+
-

nee 5
-
y = (x) (x)
Ip

# let's
does
small
say
we
a

change - 2e3x
3x
new e e is alse ady
a soln so we should
make a other sol like

yp(x
=

A23

majick A
Z
=

->
from
->
eperator method :

FCDCy = V(x) =

ea
yp
a) V
=

FCD((ea v(x)y =
ea F(D +

a)V()

x
FCD) =
B =

↳DGea4vxy =

eax(b +
a)V(x)

F(D) =

D2

↳ -22eav(x)y =

eac a)2vx)
+

FCD)
y
=
V(x)eax

F(D)de4x 9xy Veax


-

ye
-
V(x)
a) 2ye 4x3
ax
Veea
-

2 = (b +
=

- xx
ye
=

=p a) +

yp(x
ex a
=

e4
yp(x3
+,
=

ex
-
=

+
s

432

-x
=
E

+
s

e4x2

setD]
yp(x
e23(2 x62
=
+

- 2
5
-------------

3
7/23-3

de
3dy x2ex
ex 2y
=
-
+

dx

y=x) =

Cze +

(2ex
9

Yzx e
YzGx e2x
= =

doing the steps ,

F(x) =

x2

31 x , 23 Ye
-
,

S =

Ee se , , x
↳> here one of
the independent
solm is same
as here so

we'll multiply
with one

more X .

so we'll get
S1
=

Exce ,
she ,
Be

yp(x
=

xex(a +

bx +

xx)

y(p(x =
xe(a +
bx +
(x2)

+ e"(a
2bx 3(x2) + +

-
B
y(p() =

x2"(a +
bx +
(x2)
+ ex(a +
2bx +
3(x2)

e(a 3(x2)

E
+
+
2bx +

2
+

ex(2b +
6(x)
y"pcxx =

y(x)
+
<

ce
y((x
P
+ G -

3yp3 +

2yp
=

4 3B
-x B x))
+
+ -
-

xex)
xyp
=

c -

2B =

x2ex

ze"(a
/ +
2bx +
3(x))

+ "(a +
2bx +
bxx))

((2b +
6(x) =

xiey
-

a -

2bx -

32x2 +

2b
+
6(x
=
x2

2b -

a -

2(b -
3c)x
x
=
x
-

32

-2 3
!2 a = -

E
b =

-
2

operator method

FCDCy
=

x2e FCDC =

(D -
2)
(D 1) -

ypxx e
2)
=

-
-e-
=

e 1 I x2
I 51-p>

23x23
-

= - e
1(1 -
D)

ypx =
-

e232
+
D + 12
3 x2

j
as x2 so we've

taken up to pe

e(x2 2x 2)
= - +
+

ypx
= -

e(( + x2 +

2x)

-enex
ecd-2dY-by 2e-10sic a
=

33C
32 3
-
>C

y xx =
2
,
=

3x
Cze
-

(x) (ze
y= = -

F(x) =

ze -
10 sinx

-
3 x Yp
=

Ae +
Bsinc
+ [COS>

B (osx
Yp'(x) Ae
=
+

2x
- Csin

2 x Yp" (x) = Ae -
Bsinx
-

[COSx]

-
4Ae +

(-210S -

2B(OSx
-
3[(0Sx]

+
(-3Bsinx +

2(sinx
BSinx)
-
4 Ae
-
-

(4 +
2B)COS

-
(4B -
2C) Since
= 2e
-
20 Sin

42 = 2B

⑰2 =-
4 B -
2) =
10

2B -
C =
S

-22
(D -
a)
y
=

F(x)

E
Ie4xye 4x3
-

... (D -

a) =

F(x)

... ed 3y(xe -

4x3 =

F(x)

D3y(x) 4x3 D
-
-

... e =
F(x)e

e9x)F(xe adx
-

y(xx
=

by this twick

(D 3) (D 2) ze- zosinx
y
+
-
=

[10sinx

I
=le
y
( ) (D +
1) -D +
3




D
=
1

=-
ex
-
[
=35z]zusince
I

=
=
4
[ex) - 3 Iosinsdx
ezosind]
-
-

↳de +

y =
losx

y =
(x) =

2y(0Sx
ce sina
+

accosx
yp(x
=

-
b'xsin

operator method ,
-

(D2 +

1) y =

COSx
3a
-
0

y
i) 0
=

, -

-( :
i)
COS>
#method of variation
-

of Parameter

-
a
.
d
dach
+

azd +

93y =
F(x)

↳ y(x (1yz (eYe

.
+

always
-
-> we'll have this
two
condy

a. d1Yz
97dyz 03y1
+ +

=
F(x)
dece-dx
-
&

ardly
92dy
+ 9342 FCx
a
+ =

-
03/08/23/ Lecture -

yp
=

Vy(xy1 +

Ve(x)y2 -

y p'(x) Vy(yj Yexxyc


=

+ + V
! xy2
we'll choose

E
V2xxy , +

V , xy2v2(x3 & Vy(x)


in such a

that
8 way
this termill
be zewo .

yp"(xx =
Vy(xyy" +

Vy(x)yz"
+
vy'(xy2
+

vz(xxyy
-
if we also take this
term zewo then
& Vecas will
ve(x
be constant which
is undesired .

them into
- putting eqm
we'll get ,

a. (Vjcx) ye +

Ve '(xye)
=

F(x)

=>
Vy(x)yj +

vex)y2 =

F)
do

b

from

ve'Yz +
Ve 'Ye =
0

VI'Ya'
+

Ve'Ye= F

[ 333223"=[Exsal
using law
Cramers'
/Fx/a

-
VI =

ve
Eaa
de
+

y secx
eq
=

y (e) =

(z(0) x +

(esinx
=

yp(xx
=

Vy(x)(0sx +

Vy(xysimx
vi =
exe e
ne
I I COS>2

VI
=
-

tanc

vz(x)
=

log(cosx)
similarly

Ve(C) x
=

Whoa
operator
-

+
FCD =

02 +

is

-Eile:] Be
sec

- Sesecue -
-

is
is

J
-

-
e secce

x
Iz =

Je- secdx

I
=

Je secx dx

I = (20sx sec xdx =

(dx =
x

Iz i) sinc secode= Standa


-
=

log(sex)
=

Iz =

x +

ilog(se(x)
Iz =
x -

ilog (se(x)

Yp((
=

cos xlog)cosx) +
xsinx
* Elew auchy dential ea

doxdy2 Af azy Free


=
+
+

x
= et - by using
-
his transfor
we can

convert the
zan in linear

equ with
constant
Coefficient .

et
ge =
= x

&
dx
- -
xdy =
-


d32
= +
&
d+2
I
-

x
dt
des

x
2
-
=
- +

d = -

Y -

0
2

dx2 dt

-
from

God
-
+(ay
-40)d +
acy =
Get i

G(t) =
F(et)

Note assume the homogeneou sol


=

y xx = Accm
=

and by which we can get


soi without converting
it into constan coefficient
and
.

fornon-homogeneous
do exactly as before .

if me F
Me,

YF(x =

(yx2z +

CexNe

if ma
=

me ,

y= (xx =

(2 +

(log(c)) x
2x2
e x y -

xy' -

3y
=

Ist method

y
= Axm

-2 -

m - 3 =

m =
z -523
222)
m

= =

1 - 53

2
-

2
(e)
y (1x 1 x
=

#method
x
=
et
feat
&R (2
2)dy by
=
+ +
-

Am2emt+ 2 Ament
-

3 Aemt = O

m2 + 2m -
3 = 0

(m -
3)(m +

2) =
0

m = 3
-

,
2
y (x =

(ze
- 3t +
(ze2t
=

i
type

-
special
diy
-
-

F(y)

->
multiply both side by
2dI
dx

(* ) =

2 *

(5)] = ,
eFcya
dx2

d) (2)) =
2 F(y)dy

) 2) F(y) dy
=
dy
Ecy) dy
=
dy I dx
Feycy=
e
d sing
=

2dI =

2 sing

d()] esing dy
=

(2) =

esingdy
2fcosy
1)
2 =
+

C]
dy
=
= -
2[- 20sy+J
da


xcm , z =

Jame
-
#eliptic integral form

dy =
dx

osy
cosy
->
2cosly -
1

cos

cyde (8)
+ 2
ec

d
assume w =

#
07108123/ Lecture 5
(not attended)
-

#
Eigen
I= value
-
Podem :

40k3d2y +
42(x3dy +

42(x)y
=

F(x)
dx2 dx

-> where all the coefficients


aux , Af(x) , 92(x) are contr
function of x Over the
interval (a b] ,

=>
↳y =

F(x)
L
90(x2 az(x) as i
+
+
=

=> above differential equ is


called "mas" if the

operator (satisfies
following cond

(is((yz +

Yz) =

L(yz) +

L(yz)

(ii) (Ccys= CLCYS ,


C is
constant

(I) if the solution yaxs satisfies


the
boundary condr
it

[
(i) y(ab
=
A
-
Them

y(b) = B called
linear

boundary
value
problem
(I) y(a =
A , y'(a) =
B
-> Then it's initial value
problem consider a linear

boundary value problem

Ly xy
=

-> with boundary condu


Gye4(a)

E
G22y(a)
+ +

byzy(b>
+ bez Y'(b) =
A

general - Ozzy(a) +

G
,2 Y'(a)
boundary
bzzy(b) bec Y'(b)
conde B
=
+
+

-> for a scalar I ,


if there
exists a non-trivial
sol ↑ cas
that satisfies
14 =

L4 and also
satisfies the given
conde then 1
boundary
is called am
eigen
value and Y is

called as an associated
eigen function of the
given eigen value problem
LEVP]

T21
:

If Ya & Ye awe

two distinct eigen


function cossosponding
to zigen value X
then
(y4z (eYz
+

also
952 eigen function
corresponding to

eigen value X .

La L NI =

4
142 Ye
=

=>

((C4z +

(Yed
2

(((Yz))
+

((44))
=(2x4z +

Ce 4
-
x([z4z +

(2Ye)

=> xy 0y(x) 0
=
=
+

y(51) 0
=

= -

Ly = -

xy 1 =

de
dy
case-I x =

0 =
0
=

*
y(st)
=
0 =

C2iy(0) = 0 =

x y
=

(zx +

(2

(2 xL0 x =
-

y2 ,
M +
0

Ly =

My - (Ly =
-

xy)

e M2y 0
=

-
M
yCc = Ae +
Be

y(0) =

A +

B =

0
, y(st) =
0

y(ST) =

A =

0
,
B =
0

(a53 x 30 ,
x =

M2

M2y
a
+ =

mc +

y2 =
0

y
A eiMx
-
xx
B
y =
+

AeiM -
Be-ib =

=> 2 As in x = 0 M F G

=>
M
=

xJT = n F
O

-iMC
y()
=

AeiMx -
Ae
=

2 As in x

n2 =
2 A sin
x njTs
=

↳A sin En

#
separated
condr :
boundary

aey(a)
azy(a)
y
A
=
+

B
byy(b) bcy(b)
=
+

->
periodic boundary con

y(a) y(b) ; y'(a) y'(b)


=
=

# Lionville
Sturm -

Boundary
value problem
dY
-
+

P(x)1y +
[a(x +

xRxx)
dec d2
I
0
=
is P(c ,
Q and RG) are

continous over interval


(a ,
b]

(ii) PG2)<0
,
RK) >0 Xcf(a ,
b)
(P(x)dx JP(x)d 32

z
e +
e dy
-

dx

+ [Q(x) +

xR(x]
d
e/PC 0
y =

(ePad]
+
[Q() + xRxx)]
e(PxdxI
0
=

p(x) =
e)P6xdx
q(x) =
e(P2(d x x
QC)

(P(x)dx
WCx27 =
2 R(xC)

(etPad-)
IPada
=
e

P(x)

(b)(dy) (9(x) xxxx))


R + +

dec
y
=
0

-
(0x() (94x xxx)
+ +

is culled strum-lionville
operator .

-> If P(x
=
1 q(x
,
=

0
,
5(x) =

1
the it becomes
eigen
value problem
Inner-poduct
#

-
:

suppose
V is a vector
space over the real
number
assigned to each
pair of vectors
belonging
to V . The inner product
must
satisfy following
condition

L
Assume all
vector

lineat .
in

varef
-

(i) <4 V)= .


<V US ,

(ii) <Cu ,
>= << 4 ,
V >

(iii) <u ,
v +w>
=

<u v> +

(4 ,
w>
,

(iv) <u ,
u > >0

(v) < k ,
u
=

0 then u =

V
(vi) Def <U V) U
=

.
.
cu , v =

jurdx
↑ a

->
Thissatisfies all the
properties mentioned
above .

Adjoint Operator
# :

Adjoint operator of
·
an
operator
denoted and
↳ is
by +
define as

<U ,
> =
<U ,
>

suppose L-> matrice operatur


(A)
CAU , r>
=

< U ,
Ato>

Hermitian
matrice
~

Al (A 2)
-

=
Godly
e -

da
-

azd +azy

y(u)
=

y(b)

y'(a) =

y'(b)
Here d ,A2 , 2 Gre constan .

⑦ determine the adjoint


operator .

sol

adde as
-
-
= +
+
2

<1u ,
0) = < u
,
+r>
b

<u 0
Juwd
=

a
b

(auden and ackju


-
e
tace
b

-so Vdx

Fee
-
b

-as ! auf a de e -

0
b

+ as -
[urSa as] -

ude
b

-an( unde

here , y'(a) =

u'(a) =
u(b) =

u(b)
and same for
ucas =

4(b)

SO
b

--

au) Y dodx-as Judee


b

I
+

ar
unde

=-
! auudew
a
Esc
des

b

as] under
-
e we're to
choose the
b condu for
+

an nor a

that
such

vicas =

vb>

n jard-and +acbjda
-and -And+ az
Ans I
- L
#Self-Adjoint operator :

An adj oint is self-adjuinl


operator if L =Lt
&

a
ex + e
y
=

y(ab
=

y(b)

y'(a) =

y'(b)

S
a
1 =

<1u ,
r) =
<u ,
(r)
Them-2 Prove that the
eigen
:

value of a real
symmetric matrice is
real

Proof ->
given
for " A = AT
-

matrice A (ii) A : A

- x =
E to be poured .

Ax eigen function
*
=

eigen
value

(complex) ... Ax =

Ex

Ax EX
=

...

transpose) : AT =
Ext

... A =

Ext
side
(both
by X
S ...
T

x 1 Ax =

ExYx)

: xxx = exx
... - Exx =

I nonzero

*
N x x 0
N A
- =

e
x =
E


Hence proved .

.T = 5 TTa
.

# < +5 ,
> =
<5 ,
+ >
Prove that
Tem-2 the
eigen
:

functions of a real
symmetric matrice awe

Orthogonal
Proof ->
given matrice A ii) A : At
- ,

& Xe tao
->
suppose xz are

different eigen functions


corrosponding to
1 2
* &
.

distinct
Axy =

xzxz

At xzt
+

transpose)
...
x2 x
=

... xzTA =

xy xzT

XzT Axe
(bothby side) x2x
... =

x2
Xe
+
x2 +2xc x2 x2 +
=

...
+

-
xc
(X2 x2) xy 0
-
=
-

e
N

So
,
x2 Xe =
0

>AX2 ,
Xe) =
<X2 ATX2> ,

... (X X2
,
,
Xe) =
<X1 , texex

Xy<Xz , X2 > =
xe(X2 X2) ,

...(x1 -

xe)(X1 x2)) ,
=

0
ne
#
0
<X2 ,
Xe) =

orthogonal .
#
14/08/23 Lecture -
7
En : The eigenvalue of self-adjoint
operator are real .

(ie .
x =

E)
-

given L =
It

-> suppose PI and Ye are two

eigenfunction
<LYz Y ,
=

<42 , Yz)

<LYz , 42 =
<
Ny ,
L4e>

suppose By =

Ye

<242 , 42) : < Y ,


(4y>

z eigenvalue ->
Pa
x2 11 ->
Pz
(42 =

x
42

<X142 42) =
<Yx ,
xz4z)
.

<42 Y ,
=
Padx
so
b
;

a 24 dx
! d
-
-

- =

= x Y
I 2

1E -
z)(42/dx
= =

e
#
xz =

x1
-> hence proved .
Noum V is vector
: a space
over the field # ,

then the norm of a


vector u is dinoted
as 11 Ull and defined
as

11 Mr
Ull=
2/2

-(d)
Am
:

The distinct eigenfunction


of self-adjoint
A

operator are orthogonal

Roof :

suppose VI and Yz
are two distinct
eignenfuctions
to
corrosponding
xz and * 2
L4z =
x142

L42 Ye Ye
=

<LY2 , Ye =
<42 , L Ye

=<Yz .
142

<X24z , Ye =

<4z . xzYe)

+242 ,
Ye =
x(4z 4) ,

(2 -

x2) (<42 42) ,


=

11
-
<42 , 42) = 0

hence
,
proved .
Segies
#

Station

a P(xd Qaxy O
+
+

where ,
Po and QCx are

analytic function
of .

=> Analytic function : All order


-

derivative
of f(x)
a82 exist
then it's
called
analytic
function .

in Other words
-> Fxs can be represented
as tay los series expansion
about a point .
i
f(x cx -c.
=
=

n =

-
point Co is
called as
co
ordinary
point .

z
0

=

x(x 2) -

nere PCC = &

acx
= 5x

-> the =

1 & x
=

0 are

singular point as

P(x) & Q(x) awe

not
analytic func tion .
exa xd ( 2)y
+
+
+

·
-> PC & Q(x) are

analytic function
So , can use series
SOM ,

y(x
=
5 In
n = 0

2
S (nncw
-

y
=

n = 2

2
E 2) x
-

y -
(nn(n -

n =
2

-> substitue in the


diffrential ear ,

n 2 1
E + E(nnx
-

n -

(nn(n -

73x C

n =

2 n =
2

(12 + e) Cr x =

0
n =
0

Expand the teams und


match the coefficients

c
:

22 +
20 =

Ce
= -

10

x :
6(z +
2z
+

2)z =
0

B)
(3
z(I
-
=

2
x
:

12(p +

2(z +
(- +

2(z
B)
I
[4 -

Co
I

x3 (2 5(z
20(5 3(z 0
=
+

: + +
=>
( = (2
48

y(
=

( +

(2x +

(x2 +

(yx3
+(44 +

25xS

(0(2 2x4 ---)


x2
= - +

(z(x 2x3 3) ) +
+ -
-
#
Patial Differential
Equation
efintion A mathematical
:

relation b/w
independent
variables .

dependent variable
and it's partial
desivative is
called a partial
diffrential ear .

Notation :

Auc +

BUsy +CUyy + DU +Edy


+

Fu =
G

U
- ucy i "gy=
a
E E
Ux
up
= =

-> classification of first-or des


partial differential ear .

P(x ,
y ,
u) 2 +
Q( ,
y u)2
,

-x
by
=

R( ,
y u)u
,

(i) Linear
-
order pde
If P Q & R , are

functions of
only xRy
Then the above pole is
called linear first
order pe

3
+
=U

by
(ii) Semi Linear first-order
pole :

-> If P and Q are

function of and I
x
then the above equ is
called semi-lineat
poe .

-u
+
zu
-

by

(iii) Quasi-lineat first order


pde :

If
->
only R is a
function of 3C and
I then above Poe
called quasi-linear
pale .

u
-
=

ex
Live Full Non-linear first-order
pole :

-> the pde not lying in


above three linear ,

semi-line as arc
quasi-line as .

j) + =
a

B
:

Introduction to

partial differential
Shankar
eun by K .

RuO .
#
17/08/23 Lecture -
8

u)24
P(x y
u)be
, ,
+
ace) , y ,

R(x ,
y u),

->
suppose I
U
=

UCx , y) is

defined on C domain
D bounded by curve
↳ 4
y R

- u =

U
,
on curve
-
0=
,
D

, ,
in
(given) .
>C

Then the curve I is

called chauchy data


curve .
->
suppose ,
I variable is

time and x-axis is


is curve .


ta then U

!
u =

+
is called
2 initial
t =
X X T

x >
condition
t =
0 x x
T o
-
Chachy
condition

u -u(x , y)
=

f(x , y ,
u)

f(c , y ,
u) - is called
integral surface
#mod of teristic

Y (4 0
+ =


Linear

hyperbolic
pde

u( ,
+ =

f(x) -
ct)
↳ D'Alembert
solution
u(x , 0) =
f(x)
t
t t(x)

:
=

R
-> if you diffrentiat
bo the
- = Te >
at
along
chracteristic curve
0 x t =

+ (x)

= ** **I
** -

*** -

I =c
, 8
=

B
c =
c + stant u
=

cont on

the
x =
2t +

x characteristi
* curve
if
you choose
different
chatater istic
curve you'll
get different
value .

x -
+ =
CC0

u =
f(x -
ct ) =

f(x0)
2u
8 3
ex +
=

8x

U(( , 0) =
ex

ga
3
-
the answer should be

t
e
-

UC , t) =

de -

3
8U +
=
zu

of

by comparing ,

a I
= zu = 3
(n(u) =

2t +

(1x =
3t +

(2

2t +

(1
e x 3t Ce
+
=

u
=


e2t
+
x.
u
=

C
=

C2

& &x0 =

x -
3t

x -
t
u e
=

gd high
[x oh +

e
- -
-

Ot

h(c ,
0 =
f(x)

Or
-

+ (9) usee
ucx ,
0
=

f(x) Quasi-linear
differentiate along the
S
characteristic curve

e
a E g
O =

U
=

constant on the
characteristic curve .

=
f(c(0)

x -

[Sinf(scostI
R
+

2/2

I x0)gndt)I
x f(xu) UC t)
=
-
=
,
#
21-08-23 Lecture - 9

-> The curve on which the


so is given has A
form
paramatric .

P(x y u) bu k(x
u)
=
y,
+

, , ,
-

3x

R(x , y , 4)

u =

4 4 .
=

[x =
x .
(5)
, y
=

y . (s)
-
< s[a]
y

8xis
a
M(t) X -> is (r) is an
*E
X
chosen arbitary
X
>
curve such

x
that
↑Cr) 7 when
Io
=

-
=
0
= +
↳ a P
=
R =

=
a

- ⑲
Q
dy -

= dec P -

& given pF 0

&y =

& -

combining ⑦ .
② ⑬

d
dx d
=

↳ lags ange

I charpit ear

lagrange's eun

3
2x
-
22 =
x
-

by

u(x , y)
=

1 is =
Eac =

by 3

s Paramatric form
S
G
x =
=

x bs as
y
=
=

du
d -
d =

3 &I =
2
d5 d5

x 38 y 20 x2
(1
+ +
=
=

↓ AS
bS 0
C2
=

0 x
=

,
= =

r ,

-
12 bs =

C2 =
aS
x 35 b5
= +

25
+

aS
y

(x -

3y
=

(2b -

3a) S

S = a

aBy
= I

now
knowdu
.

=c =
38 +

bs
we -

d5

u 32 bs c
=
+
+

-) is -

x )
+
(3

r 0 UpC y) 1
=
=

, ,

1
[3
=

uces , y)
=

3(a) be +

I
+

+ e
e =

30 R Yo 0
=
=

U(x y) 1
iss
=

som
or
-
-
du
do
=
o
dx =
2 =

x
=

+
+

(1
do

d y
1 r
c
= +

=
=

t =
0 ,
x =

n = G =
x


#
Y n x y
=

-
-

du =
da

is
2
y C3
=
+
-

I
-1 -

y
+

(3

U 1 when
given
=

x =

n , Yo
=
0

we'll get (3 =-2

ucx =

y
of
#
Gen
function o -
·
a Stat

denoted by 4 and

difined as

d4 =

-G d .

64i
G 64j 6
= +
+

8x
by

I =
dre

f(x, y 4)
,
=
U(x , y) -
u =

#
integral
surface

x +
yz +
z1 =

1
z =
12 -
(x) +

y5 = Ex , y)

z(x , y) -

z =
0

d4 P(x d dy) P(x y)


y
-
= +
+
,
,

( .)
=
f =

- direction
goutient

-
we know .

Px , y , u)(u + G(x ,
y , u)24
0x
by

R(x , y u)
,

we can write it as
2
-

(i 65 2i) (p4↑ x4)


+
-
.
+

25 +

=
O

tangent to
the surface

af dy
edx c
0
=

-y

+
If du
zu

24dx
dy-du
+

-C

5 K)e(dY+
4 dy
(04
-
+

+
duk) =

- ⑧

B
we can say that 2and B
a de
parallel or
=
antiparallel

50 thies cross product


should be zero ,

i I I

P Q R =
0

des dy du

↑(Adu-Rdy) -

(Pdu-Rdx)

+> I(Pdy -
@dx) =
0

->
By equating sefficients
we'll legrange Champit
Yet
-

equ .
d
=

d A =

S(C) ,
y , ub =

C1

T(x , y ,
u) =

22

F(5 , T) =

* xy5
ex xu -
y
-
=

p y2 G R x(u (y)
xy
=
= -
- =

,
,

dx
-
=
=

y2 y x(u -

2y)

-> Don't choose relation


arbitary choose so that
solve it
you can .
-
xdx =
ydy

x2
- -

yz =
C

(2) +
yz =

xz -G

& = y

udy-eydy+ydu= 0

d(uy) Zydy 0
-

yy
-

y2 =
Cz
Homework
-
-

x) ye
E(y ux)
Y xy)6
+
- (x +
= -

fre *yz
**

tuys"
:

so we'll get

yo cdy du 0
+ =
+

d(y) +

du = 0

xy +

u
=

Cy

xdx +

ydy - udu =
0

d(x 22 2)
-
=
0
+

x2 +
y2 - u2 =
c
#
24-08-23 Lecture -
10

Second Order PDE with


# - -

Constant Coefficient :
- -

-2BUscy
(Ugy +
+DUx

# +Ely
↳ example
+
Fu =
G

Principle of a linear


part Zach order pde
compare with constant
with coefficient
sonic section
if

lat
bxy G 0 the given
a
+
=

+
dx
pole is

homogenous
if G E0 ,
1
'I

non-homogenous
T

(x +

2 bccy +

cy2 =

x Ax

xx
33/ , i)(y)
=

man
symmetric
matric

a =

[ ]

x =

QY

det(g)
b2
=
=

ac -

?
PIAl
TWIA)
x
=
-
Tra
2

*
& -
T5CAS +

det A =
0

(i) det A =
0 parabolic
(ii) det A CO
Hyperbolic
(iii) det A >O Eliptic
- =- det()= Discriminant

i if 0 =
0 =

(b) -
ac)
:

Parabolic
form

(ii) if & 30 :
b2 -ac >0
:

Hyperbolic
form

(iii if <0 : b2 -acco :


Elliptic
form


AUx +

2BKocy +
(Ugy +DUx
Fu G
+Ely
+ =

x =

x(3 n) .
2 =

5(x y) ,

④ W

y
=

y(9 , n) n =

7(xx y) ,
d =

Exdx +

Bydy
du =

4xdx +

nydy

1) 12. :

Kit
-

"

+2
4x
Ex
=

on

UgEx Unnx -
-
+

My =

UgSy +

Unhy -
Ux
=

bx(ug)3x +

Ue3xx
+5 (Un)nx
+
unx

=(455Ex Ugn]5 +

+Ug5 Me5 4nnYx]


+
+

-x

Roc
+
Un4s
-

Ug3 (Ex)+ 2
UsnEa
+
umn(nx)
+
UsEx Un"x
+

-

Kyy
=

Ug3 (Sy)+ 2Ugn3


y y

+unn(g)
+
Us Yy
+
Un
yy
-

Usy =
Ux My
x(45)5y UgEy
=
+

+ x(un) my +

Un Yoy

=[UgEx+ kgm4x]Ey+UgEy
+
SUSE +UnnYMy
+
uncy
=

455E5y +

Ugy(7 5y Exy),
+

+
Unnrhy +
UgScy +UnYxy

#Us +
25Ugn +
knn

+BUs +EUn + Fu =
E


converted/transferred
2q

A =

AEcs +
2B3xBy
+

C
By
B =

AEx4x +

B(3xhy +

nx5y)
+
Cyny

I An 2BR +
Cry
=

, y
5 AEcx 2BEccy [Syy DEx
=
+
+ +

+
EEy

E
Anxx 2BRxy C
+
+
=

YY

+
DUx +Eny

F = F T =

G
,

E
AUcx 2BKocy (Ugy +DUx
+ +

C
Fu G
+Ely
+ =

converted

E
into
#Us +
25Ugn +
knn this

+BUs +EUn + Fu =
E
5 =
B2 -
Ac =

(3xky -

2x3y)2(B2 -

Ac)

So ,
if & <O => <O

0 5
=

↓ =

= 0

↓ > 0 =

5 >0

=>
Hyperbolic
-
Pde :

HC
455 Unn
- =

Ug = H) -> rectangular
pole

AUss + 2
Ugn + Unn HCU Ug=

, , Un

E , n)

52 -
At >0

case -
F A = - I
-

B =

0
A =
S # =
= 0

B = 0

=>
Elliptic pde :

HCh us Un E h)
Us +
knn= , .
.
.

B and A I
a I
=

0
=


Police

493 =
H) 3

H) (
Unn=

Either (A. =
d and B = 0
#
28/08/23 Lecture -
11
(not attended)
* .

Hyperbolic pde
ju
che 0
-
=

-2+ 2

B AC > 0

* ex
=
Parabolic de

)
B Ac 0 Ju =
=
-

* ex
=
Elliptical pole

+
BI Acco
④.
-

Hypolic Pe :

* =

c =

0 & B +
0

B -
Ac > 0

AE , 2 BEx By [Ey2 0
+

+
=

S
A(5)2 2B(5) + +
1 =
0

An +
2BUxy +
CMy =

-
A(z) 2B(e)
0
+

+
=

· -
=
4AC

2 A

" =-
As
-> If E(x y) ,
=

constant then

di 6
63dx
0
=
=

y
+

by

dy
-=-
BA)
- -
=

dc

sit choose + for "s" then choose


you
- for and vice versa .

->
Our purpose is to find
E(x y) ,
and nx yS ,
and

they should be different

(e)
otherwise = 0

e 3Us 10
Ussy
+
3Ugy 0
=
. +

A =
3 ,
B =

3 ,
2 =

B2 - AC =

25 -9 =

16 > 0
so
-
assume E( y) 2
=

3 I d
=
* = -

0
choosen

=-
I
-
-

- -
3 E
my for

E 3x n y
3
=>
y
-
= -
=

A =

3( -
32 +

10( 3)(2)-
+

3 =
0

5
3(3)(2) 3/ 3xx
I(z)
=
+ - +

+
3(1)(1)

5
=
3 -
25 -
+
3 =
- 32

5
i
3() 20(- z)(x)
=

+ + 3 =
0

5 =
0 E =

0
,
F =

0 ,
5 =

0
,

Ugn 0
=

us
=

f (3)

u =

(f (5)d5 +

g(n)

u =
F(3) +

G(n)

u F (y x)
G(y 2)
+
-
= -

8
Alembert
sol
⑦.
④ .. olic pde.

B2 -

Ac =

0 A =

0 and
B =

G or
=

I
=

suppose A =
0 and B = 0

AE , +

2B5xEy
+

<5y2 =

= -Ac
=

-
A

A5cx +

B(Exy
+

4x5y)

(SyMy
+

= 0

-
3y(A( )4x 3)()y
+

2x) (ny)
+

3
+

=
- 2 y
x
+
=
= + =

C =

y
-

Bx
A

\() y) y
Ex
=

choose
that water time
are not pralled .

ex
-
-
Uc -

kcyUxy
+

y2yy =
0

A = x2 , B = -

xy ,
c =

y2

BY -
A2 =
=

x2yz -

xy2 =

= =
-

-
-

=
=
My C
=

=( , y) =

xy

n =

3
182 I x
Y
=
=

\(x y) ,
=

xyx
=

A =

x2y) -

2xyz +

y2x2 =
0

B =

x2(0) -

xyz +

y2 =
0

5 = 0 +
0
+

y2 =

yz

5 =

x)(0) -

xxy
+

((0) +

0
+
0

= -

2xy
E =

x(0) +
0 +
0 +
0
+
0 =
0
F =
0
i 0

2
y Unn = kcy Us
+

Unn
=

2 us
y

Unn =

25 4 :
-2
# 32108123 Lecture
#
Elliptical Pde .

BY -
AC <O
case
-
-

I
-

assume B 0 & A
=

C =

-
AEx 2BEx5y+ (5y2 AR 2BRcMy
=
+ +

(my2-Q
->

so
AE xx
+

B(5xky 4x3y) +

23y4y =
0 -

① +
2i x2

A(Ex +

inx)
+

2B((Ex + inc(sy +

ing)S
+ c(Sy +iny)2 =

0
divide by By+inyl)

A/)) 2/23] + =
0

As
a =
3 +

in
B
=

3 -

in

I=
-
As

d(x , y)
=

C and BCCC y) ,
=

& /- )
- -

= =

3 2 B B
n 2
=

=
=

2
2i
&. (1 +

x))Ucx +

(2 +

yi)Uyy
+

xUx

0
y
+

y
=

3
B =
0

A =

1 +
x
2 -> elliptic
form
c =

It y2
D =

3C

E -

F = 0 = G

t
--

(A]

Ezy
-

--
dy
dx
1 +
x2
z
logly + +y2) Filogkc +

#xx) =

log
2 =

log(y +

Ftyz) +

ilogkc + xxa)

logly +
tyz)-ilogc xx)
B +
=

3 =

2B =
logly Etyz) +

E
Ex =O

n log(x +xi)
2
= +
=

0
2
=

A
(3y2 1 y2
z
2
=

=
= =

(tyz)
I An
= =
= = =
2
2

B =

P A5x 2B3x3y 23yy


+
= +

D3x Ey
+
+

2
yzxy(y) xy)
= +
+

-Styz
⑤ - ⑧

E =

finally got
Auss +
JUnn =
0
Elliptic ⑳ Uss +Unn =
0

form

Case -

I
--
-

B =
0
,
A =
-
I

A(Ex +

n,)
+

2B(5x3y +

4xhy)
((5y2
ny2
+ +

ASMcc +BlEcy +Ychy)

Cyny
+

⑦ +
& x 2

A(5x +

xx) +

B(3x +

4x)(5y xy)
+

((5y +

hy)2 = 0
divide with
(By+hy)

A/E! 23/5)
+
+ -

In
A
=

3 = G n = d -
B
I

a =

5 +
x , B
=

S -

dy
- ( A]
-

=- -

Ex

2(x 2 B(c y) C
y)
=

=
, ,
,
e Y
-

2 sin
Ucy-cOSC Ugy
-

COS 0
y
=

A =

I B =
-
Sinx ,
=- cos
D 0 E (0Sx F G 0
=
=
= =
-

, ,

BY -AC >0- so
hyperbolic
2)
c --
[sinx =

E
de
=2-sinx

dY
--- (I + sinx)
dec

x() y) COS
y x
=

+ -

x [0Sx
B(x y) y
-

= -
,
9
B
(0x x
y n
= -
=
=

2
·
A =

A5 +

2BEccSy +
Cy2
=(1) (Sinc) +

2 (-since) (sinx) (IS

+ (-cos2) (2)

--coskc-sin=-1

· T =

Anc +

RBUxy +CMy2
=
1 +
2(- sinx) (1) (os +

( 20sx)(0)

I =I
·

(A -

= 2)
B AExx B(Exhy 2x5y)
=
+
· +

(nysy

= I . Sinx+ (-Sinx I .

B = 0

B AEx 2BExy (5yy D5x


=

+
· + +

Eby
+

(osx +
2(-sinx)6 +
0

+ (-(0Sx) I

=
0

E Anxx 2Bdxy (2yy DWe


=

+
+
+
·

+
Eny
= 0

F =
F = 0 = G =

A Us +IUnn 0
=

0
+
knn
-

Us
=

Uss-Ynn 0
=
#
↳/09/23 Lecture -
13

method of Separation
#

viation :

Ar +
2BUccy +(Ugy +

DUx+Evy
+

Fu =
G

Suppose the soM UCx y) , is


,
=

defined on a domain
I bounded by curve T.

f() y)
suppose u =

, where
DC y) E N
.
aY

~
suppose
the
.

boundary
Blus indicates
-
information of the

solution UCC y ,
. >
⑧ 3
-I Diet's bondary cond"
-

·
If B(u) =

u i e
.
. BCU) =
fx y
,

(cyc = ↑ then the

boundary condh is called


"Dirichletis boundary
condn
"
.

Case Neumann's boundary aord


-

I
-

- - -

·
if B(U) =

zu
=

Fu I .

on

· is a unit normal
vector
-> Then the boundary
conde is Neumann's

boundary cord"
- Robin's boundary Condu
- -
-

If
By
- B(u) =

au +

a
where and B are

then the boundary condn


is missed boundary
conde 02 Robin's
cond "
=

boundary

=
- Kwin
- -

0x2

U kUccx
=

f(x)

E
UC0 +) 0 u( 0
=
=

boundary e
, , ,

condition
u(L t) =

0,
intial cond?
,

0 [ x I L

#
-



>
-

Our Purpose is to separate


and convert this pole in
system of ode .

UCss .
t =

X(x) T(t)

* =
"
in this - x

example ma n
f(x) q(t)

-> The se both


a be equal only
when they
are both are

equal to same

constant value .

if x
EI 0
=

"
x =
0 T = 0
.

x =
(2 +
(2X T
=

(z
Ucx , +3 =

(y)(z +

(2X)

·
u(0 +, = 0
=

(2)z = 0

· we'll take
(2 =

0(z =

as if 23 =
0

the u'll become


Zea0 .

·
UCL , +) = 0 =
22L(z =
0

·
no matter
which We

choose as

"O "We'll get


trivial sol
for x =
0 .

-I x > 0

x =
42 (4 FOS
Mx
CzeMx
-

X = +
(ze
T =

CzeM2kT

((zeMx Mx)
-

Ucx , H =
+

(ee

Cen2kt
-> for boundary
-

onde
13 Should become

2280 and if this


happens the solution
will become trivial

#
< x <8 * FO

X =

C2COS Mx +

12 sin MxC
-

k/M2t
T
=

(3e

ucx +
,
=

[C2C0s Mc
+

[sin MC]
&
-

k
ze

Mit
usoit =>
C2Ce = 0

⑰ 0
-

+
Us .
1) =
0 => Cesin MLC be
8
=

=
y
xM2 m
=
12 ,
12. . . . .

u(x 0 ,
=
f(x)

Ces in M3s [ =
f(x)

kM2t
uxx + fxxe
-

n = 2 ,
2 , 3 .
--
. . .

-
k52
Mc , 1) = f() e 2

n =

7 , 2 ,
-
-. -

ucxit) =

n =
2
far -2
e =
0

UC0 .
y)= qcys ·
UCC 1 02 = 0

U(H , y)
=

0
,
U(x .
1) =
0

u(x , y)
=

x(x)Y(y)

x "y +
Y"x =
0

" "
x
* = =
(a-I If X 0 y (z (xy
= =
+

x = (2 +

(2X

U(x , y)
=

((z +

(2) ((3 +

(xy)

U(x , 0) =
0 =
((1 +

(2x)(3 =
0

H
(3 =

U(x , y) =

(2 +

(2x)(y

U(( . 1) = 0

=> (2 +

(2)( =
0

=>
either 22 +
(2X =
O

E = 0

-> In both cases ,


we will

have a trivial sol


we are not into ested
in x =
0 .
-# If >o

x =

42

I M2
"
x =

((z eMx Mx)


-

x =
+

(ee

y =

Cy COS My +CusinMy

(2eM MC)
-

uxx , y) =
+ (2e

CzCOSMy +
CsinMy)
·
U(x10]
([ze4 M4((z)
-

(e 0
=
+

(3 0
=
=>

( eM
Mx)csin
-

ux , y =
+

Ge
u(x ,
2)
=
0

((ze M3(ysin1
-

=>
+

Cee
O
=

Sin ML =
0 = ML =
n ST

M =
25 ,
n =
< , 2 .
--

x M
= n= 2 I2
=

--
, -

x ye
x x 7 2
= =
= , ,
---

u(0 ,
y)
=

g(y)

(Cz +

(e) C ,
Sin
My
=

g(y)

UCH 0
y)
=

MH
MH) cysin
-

(Cze
+

cee y
=
0
(eMH
-
MH
+

Cze =

MH -
MH
Ce =
-

ce

2 =
C2 e2MH

(2 -

> e 24H) Cy Sin My


=

g(y)

ce ePH) cysinMy= gy

24H
C .
-

Cy Sin My= 9

C2CySin y =
-

e(y)
esinh MH
ux , y) 2 M)
H)
CysinMy
= -

MH
e

=- -

R sinh MH

gM
e
=

You might also like