Sei sulla pagina 1di 4

Kap2: EULERS FORMULA: e iθ =cosθ+i sin θ .

Or a =b = a , y=( Ax+ B)e


ax
. If the roots are α+i β then the to the nearest singular point.
n
writting in complex numbers: ( ei θ) =cos nθ+i sin n θ solution is αx iβ x −i β x
y =e (A e + B e ) . or into forms like THEOREM 5: CAUCHY'S THEOREM . Let C be asmople closed
y =e (c1 sin β x +c 2 cos β x)= y =c e sin (β x+ γ) . SECOND- curve with a continuously turning tangent except possibly at a

Z =x+iy =r(cos θ+i sin θ )=r e , a b=e bln a ax ax

finite number of points (that is, we allow a finite number of


( )
2 3
n θ θ z z ORDER LINEAR EQUATION WITH CONSTANT
z 1/n= √ r cos +i sin . e z =1+z + + ... corners, but otherwise the curve be smooth). If f(z) is analytic on
n n 2! 3! COEFFICIENT AND RIGHT SIDE NOT ZERO:
If c is not equal to either a or b; C e cx , if c equals a or b, and inside C, then. ∮ f (z )dz =0
around C
Kap7:FOURIER TRANSFORM: Average of f(x) on (a, b) a ≠b ; C xe
cx
, If c=a=b; C x
2 cx
e . To find a
b THEOREM 6: CAUCHY'S INTEGRAL FORMULA . If f(z) is

=
a
{
∫ f (x)dx .The average value of sin^2 nx) and cos^2 nx particular solution of (D− a)( D−b) y = k sin α x first solve analytic on inside a simple closed curve C, the value of f(z) at a
k cos α x point z=a inside C is given by the following contour integral along
b−a (D− a)( D−b) y =k e
iα x
. and then take the real or imaginary p. 1 f (z )
1
−π
1
−π
y p of (D− a)( D −b) y =e P n (x) cx
C: f (a )= ∮
2 π i z −a
dz . This is Cauchy's integral formula.
2 2 A particular solution
2π∫ ∫ cos nx dx=1 /2
sin nx dx=
π
2 pi π where P ( x) is polynomial of degree n is: if c Is not equal to Note carefully that the point a is inside C; if a were outside C, then
n
cx ϕ (z) would be analytic everywhere inside C and the integral
The average value of sin mx cos nx over a period is: either a or b; y p=e Q n( x) , if c equal a or b, a ≠b ;
−π ec 2 cx would be zero by cauchy's theorem. If the values of f(z) are given
1 y p=x e Q n( x) if c= a=b. y p=x e Q n (x ) . Where on the boundary of a region C, then the value of f(z) at any point a
2π∫
sin mx cos nx dx =0
π Q n
(x ) is polynomial of the same degree as P n
( x) with inside C. With this interpretation you will find the cauchy's integral

{
−π 0 m≠n undetermined coefficients to be found to satisfy the given formula written with a replaced by z, and z replaced by some
1 differential equation. Note that sines and cosines are included in different dummy integration variabl, say w.
2π π
∫ sin mx sin nx dx = 1/ 2 0 m=n ≠0 cx
0 m =n = 0 e by use of complex exponentials. 1 f (w)
OTHER SECOND-ORDER EQUATIONS: In case of dependent
f ( z)= ∮
1 2 π w−z
dw . .

{
−π 0 m≠n
1 variable y missing y ' = p , y' ' = p ' . Independent variable x LAURENT SERIES: Let C_1 and C_2 be two circles with center
2π π
∫ cos mx cos nx dx = 1/ 2 0 m= n≠0 . The fourier-coeff. dp dp dy dp at z_0. Let f(z) be analytic in the region R between the circles.
1 m = n =0 missing y ' = p , y' ' = = =p . To solve
dx dy dx dy Then f(z) can be expanded in a series of the form
1 y ' ' + f ( y)=0, → y' y ' ' + f ( y) y' =0, y' dy ' + f ( y )dy=0 b b2
1: f ( x)= a 0+a 1 cos x+ a 2 cos 2x ... b1 sin x+ b2sin 2x... And 2
f ( z)=a 0+a1( z −a0 )+a2 (z − z 0) +...+ 1 + ... conver
2 1 2
−π y' +∫ f ( y)dy=const . z− z 0 (z −z 0)2
1 2
solving a n an= ∫ f ( x)cos nx dx and bn is gent in R. Such a series is called a Laurent series. The “b” series is
ππ LAPLACE TRANSFORM: called the principal part of the laurent series.

−π − pt
1
bn= ∫ f (x) sin nx dx . Here is an example for it. F is 0 from L( f )=∫ f (t)e dt = F ( p) . 1 f ( z)dz 1 f ( z)dz
π π 0
an= ∮
2 π i (z − z 0) n+ 1
, bn= ∮
2 π i (z − z 0)−n + 1 If all the b's are
∞ t
-pi <x< 0 and 1 from 0<x<pi.and do the same for bn and put CONVOLUTION: G ( p)H (P )=
∫ ∫ e− p t g (t − τ)h (τ)d τ dt zero, f(z) is analytic at z=z_0, and we call z_0 a regular point. If
b n≠0 but all the b's after b_n are zero, f(z) is said to have a
[ ]
−π 0 t = 0 τ= 0
1
them together in [1:] an= ∫ 0⋅cos nx dx +∫ 1⋅cos nx dx
[ ]
t
π 0 pole of order n at z=z_0. If n=1, we say that f(z) has a simple pole.
π which is equal to L ∫ g (t − τ)h (τ)d τ . If there are an infinite number b's different from zero, f(z) has an
COMPLEX FORM OF FOURIER SERIES: 0
essential singularity at z=z_0 . The coefficient b_1 of 1/(z_z_0) 9s
n= ∞ DIRAC DELTA FUNCTION:
ix − ix 2i x
f ( x )=c0 + c1 e +c −1 e +c 2 e ... = ∑ cn e
in x
. or in other ∞
n (n)
called the residue of f(z) at z=z_0.
n =−∞ ∫
−∞
(n)
ϕ( x)δ ( x− a)dx =(−1) ϕ (a) . SOME FORMULAS THE RESIDUE THEOREM: Let z_0 be an isolated singular point
−π
1 of f(z). We are going to find the value of ∮ f (z)dz around a
way c n= ∫ f (x)e− inx dx do the same this as done in
2π π INVOLVING δ . u (x −a)= { 1 x> a
0 x< a
. simple closed curve C surrounding z_0 but inclosing no other
singularities. Let f(z) be expanded in laurent series about z=z_0
example 1: On OTHER INTERVALS the coefficients , they can a ) x δ(x)=0 ,b ) x δ' (x)=−δ (x), c) x δ' ' (x)=2 δ(x) .
2

1
−π
1

CONVOLUTION THEOREM: that converges near z=z_0.
be written a n= ∫ f ( x)cos nx dx b n= ∫ f ( x)sin nx dx t t ∮ f (z)dz = 2 π i⋅residue of f (z )at the singular point insideC.
ππ π 0
1
2π ∫ g (t− τ)h (τ)d τ=∫ g (τ )h(t− τ)d τ Sum of residues if we have more than one. In general we write it
− inx
and the c n= ∫ f (x)e dx . or u can write them in form 0 0
R( z 0 )=lim z→ z −(z −z 0) f (z ) when z_0 is a simple pole.
2π 0 TRANSFORMS OF DERIVATIVES OF y: 0

l L(y ' )= pY − y 0 MULTIPLE POLES: Multiply f(z) by (z-z_0)^m, where m is an


1 nπ x integer greater that or equal to the order n of the pole, differentiate
an= ∫ f ( x) cos dx and same for bn and c n .If 2
L( y ' ' )= p Y − p y0− y0 '
l −l l the result m-1 times, divide by (m-1)! And evaluate the resulting
3 2
a function is even like x^2 or cos x, whose graph for negativ x is L(y ' ' ' )= p Y − p y 0− p y 0 ' − y 0' ' expression at z=z_0.
just a reflection in y-axis of its graph for positive x KAP 9. CALCULUS OF VARIATIONS:
f ( −x)= f (x) and the opposite for odd func. d ∂ F ∂F
EULER EQUATION : − =0
f ( −x)=− f (x) . dx ∂ y ' ∂ y

{
l 0 if f (x)isodd KAP12. GENERALIZED POWER SERIES OR THE
∫ f ( x)dx= l and again if f(x) is METHOD OF2 FROBENIUS:2
−l 2 ∫ f (x)dx if f (x) iseven Example: x y ' ' +4 xy ' +(x +2) y =0 .

0
y= a 0 x s +a1 x s +1+ a2 x s +2 ...=∑ an xn+ s

{
l
2
∫ f (x) sin n πl x dx
n=0
b n= s− 1 s s +1

n + s −1
odd, l 0 if f(x) is even then. y ' =s a 0 x +( s+ 1)a a x +( s +2)a 2x ...=∑ (n + s)a n x
n=0
a n =0
y ' ' =s (s −1)a 0 xs − 2+(s +1) s a 1 x s −1+(s +2)(s +1)a2 xs

{
l ∞
2 nπ x n+s − 2
a n= ∫ f (x) cos dx = ∑ (n + s)(n+ s −1)a n x
l 0 l . PARSEVAL'S THEOREM: A n= 0

b n =0 KAP 13. PARTIAL DIFFERENTIAL EQUATIONS:


LAPLACE'S EQUATION ∇ 2 u =0 the function u maybe the
relation between the average of the square of f(x) and the
coefficients in the fourier series for f(x) that the square is finite. gravitational potential in a region containing no mass.
1 ∞ ∞ POISSON'S EQUATION ∇ 2 u= f ( x , y , z ) The function u
f ( x)= a 0+∑ a n cos nx+∑ bnsin nx .DEFINITION OF may represent the same physical quantities listed in laplac's
2 1 1

i αx
equatoin. But in a region containing mass, electric charge, or
FOURIER TRANSFORM. f ( x)=∫ g (α )e d α , sources.
−∞
1

− iα x
THE DIFFUSION OR HEAT FLOW EQUATION:
g(α )=
2 π −∞
∫ f (x)e dx . FOURIER SINE 2
∇ u= 2
1 ∂u
. Here u maybe the non-steady-state
TRANSFORMS. We define f s (x) and g s (α) a pair of α ∂t
fourier sine transforms representing odd functions. temperature, temperature varying with time, in a region with no
0 heat sources; or it may be concentration of diffusing substance.
√ 2
f s (x)= ∫ g s (α ) sin α xd α ,
π 0
0
Alpha is diffusivity constant.
WAVE EQUATION: ∇ u= 2
2 1 ∂u
.

TRANSFORM. We define
2
g s (α)= ∫ f s( s) sin α xdx . FOURIER COSINE
π 0
f x (x) and
v ∂t
KAP 14: FUNCTIONS OF COMPLEX VARIABLE:
g s (α) a pair of A function f(z) is analytic in a region of the complex plane if it has
fourier cosine transform representing even function. a derivative at every point of region. The statement “f(z) is analytic
0 at a point z=a” means that f(z) has a derivative at every point
√ 2
f s (x)= ∫ g s (α ) sin α xd α
π 0
0
inside some small circle about z=a.
THEOREM 1: If f ( z)=u (x , y)+iv (x , y ) is analytic in a

√ 2
g c (α )= ∫ f c (s) cos α x dx .
π0
KAP 8. ORDINARY DIFFERENTIAL EQUATIONS.
region, then in that region
∂u ∂v
=
∂x ∂x ∂y
called the CAUCHY-RIEMANN on conditions.
=−
∂u
. The equations are

LINEAR FIRST-ORDER EQUATIONS EEXAMPLE: THEOREM 2: If u(x,y) and v(x, y) and their partial derivatives
dy with respect to x and y are continuous and satisfy the Cauchy-
y ' +Py =Q Q=0, =− P dx → ln y=−∫ P dx
y Rienmann conditions in a region, then f(z) is analytic at all points
y =e − P dx + C
=Ae
−∫ Pdx
. SECOND-ORDER LINEAR inside the region.
EQUATIONS EXAMPLE: We have y ' ' +5y ' +4 y =0 , SOME DEFINITIONS: A regular point of f(z) is a point at which
2 f(z) is analytic. A singular point or singularity of f(z) is a point at
D =d / dx .
2
Dy=
dy
dx
= y' , D y=
2
2
( )
d dy d y
dx dx d x 2
= = y ' ' which f(z) is not analytic. It is called an isolated singular point if
f(z) is analytic everywhere else inside some small circle about the
D1 y +5D y+ 4y= 0 →(D +5 D+4) y =0 this is x^2 singular point.
expression THEOREM 3. If f(z) is analytic in a region R, then it has
(D +1)(D + 4) y= 0 →( D+ 4) y =0 &(D+1) y =0 derivatives of all orders at points inside the region and can be
−4 x −x −4 x −x
y =c1 e , y= c2 e ,→ y =c1 e +c 2 e . In general it is: expanded in a taylor series about any point z_0 inside the region.
ax bx
y =c1 e +c 2 e solution of ( D−a )( D −b) y =0 If we have The power series converges inside the circle about z_0 that extends
CAUCHY'S FORMULA: p π
nx y

Y ( y)=Y n( y)=sinh . The final result can be writter as
f ' ( z )=
1
2 πi ∮ f (w)dw
(w − z )
2
∨ f ' (a )=
2
1
π i ∮ f ( z)dz
(z −a )
2
C

dz
1+ z
2
=
−p
∫ dx
1+ x
2
+
0

pie d θ
2 2i θ
1+ p e
We know the value of
un ( x , y)= B n sin
nπ x
a
sinh
nπ y
. We will find now a solution
f (w)dw f ( z)dz the contour integral is π no matter how large p becomes since a a
n! n!
(n)
f ( z)= ∮ (n)
∨ f (a )= ∮
2 π i (w − z)n+ 1
2 π i (z − a)
12
n+ 1 there are no other singular points besides z=i in the upper half-
plane.Let p →∞ ; then the second integral on the right of the
which also satisfies the conditions u (x ,b )=
0 {
x o⩽ x< a
x =a
.
LAURENT SERIES EXAMPLE: f ( z )= equation above tends to zero since the numerator contains p and Remember that square on the x-y-axis. We call this new condition
z (2 − z)(1 +z ) a
the denominator p^2. Thus the first term on the right tends to nπb 2 nπ x
This function has singular points at z=0, z=2, and z=-1. We write it π (the value of the contour integral) as p →∞ ; and we g(x). b n sinh a = a ∫ g (x) sin a dx g(x) =x from 0 to a.
in a form. ∞
dx 0

have ∫
)[ ]
=π . THIS METHOD CAN BE USED TO −1 a
−∞ 1 + x
2

bn= sinh
nπb
a ( 2
a0
nπ x
∫ x sin a dx integrerer og får
EVALUATE ANY INTEGRAL OF THE FROM. ∫
P( x)
dx a

[ ( ) ( )
]
Q( x) 2 πn x π nx
−∞ −1 a sin a x cos
a a
. If P(x) and Q(x) are polynomials with the degree of Q at least two bn= sinh n π b 2
greater than the degree of P, and if Q(z) has no real zeros. If the a ( a ) 2 2
π n

πn 0
integrand P(x)/Q(x) is an even function, then we can also find the
) [ ]
−1 2 n+ 1

integral from 0 to infinity. COS(X) IS AN EVEN FUNCTION


AND SIN(X) IS AN ODD FUNCTION.
b n
= sinh
a (
n π b 2 a (−1)
a πn
−1
.
n+ 1
Then the solution is

( )

n π b 2 a(−1) nπ y nπ x
u (x , y)=∑ sinh ⋅sinh ⋅sin .
SOME IMPORTANT DIFFERENTIAL RELATIONS: n= 1 a πn a a
2
When we have (D +w ) y =0 We get 2
D =±i w and the HERE IS ANOTHER EXERCISE FROM OBLIG 2:
2
solution can be written in two forms: iwt
y = Ae + B e
− iwt
or ∂ u( x , y) ∂ u (x , y)
WE have a partial differential: y 2
= with
y =c1 sin w t +c 2 cos w t . ∂x ∂y
A SECOND ORDER DIFFERENTIAL EXAMPLE: the condition u (x ,o )=δ(x) . We will try to solve this
equation by use of fourier transform, with x as variable and y as
parameter. We use now the fourier transform of the derivative.
2
F [ f ' (t )] =i w F [ f (t )] , F [ f ' ' (t)]=−w F [ f (t )] . We take
the fourier transform of the equation.
2 ∂ U ( x ,t ) 2
− y w U (w ,t )= → − y w U (w , y)dy=U (w , y)
∂y
1 2 2

∫ U (w , y)dy=∫ yw 2 U (w , y)⇒U (w , y)= A(w)e− y w


∂ U (w , y) 2 −1 /2 w y
2 2

+ y w U (w , y)= 0⇒ U (w , y)= A(w) e


∂y

1
u (x , y)= ∫ U (w , y)e− iw x dx .we know that
2 π −∞

δ( x)=
∞ x =0
0 x≠a { U (w ,0)=
1
2 π −∞
∫ δ(x)e
− iw x
by

definition ∫ δ(x)

f (x)dx = f (0) hence for U(w, 0) we get
1 − iw x 1
U (w ,0)= ∫
2 π −∞
u( x ,0)e dx = = A( w) thus we have

1 − 1/ 2w y 2 2

U (w , y)= e and the final result is




1 1 i w x −1/ 2 y w 2 2

u (x , y)= ∫
√ 2 π −∞ √2 π
e e dw vi integrerer og bruker
formelen fra rottman. Resten er enkelt.

FOURIER TRANSFORM TIL Å VISE:


∫ cos w x
=
π −a∣( x )∣
a 2+w 2 2 a
0
e .

LAPLACE TRANSFORM OF DERIVATIVES:


It is for the first derivative L[ f ' (t) ] =s L[ f (t )]− f (0) For
2
the second derivative L[ f ' ' (t )] =s L [ f (t )] −s f (0)− f ' (0)
SOME OTHER METHODS FOR FINDING RESIDUES:
2π 2π

I= ∫0
dt
=
1+b cos2 t 2
1

0
dt
1 +b cos2 t
it
. → dz =i e d t =i z d t or

1 it 1 −1
dt = dz we set z =e , cos t = (z +z ) and get
iz 2
I=
1
2i ∮ b 4 b
dt
z +( +1)z 2+
b here is not so much place so ..
4 2 4
2
ib ∮ 2
zd z
2 2 2
( z − z + )( z − z - )
2
z +-=(−b−2 ±2 √ b+1) /b . Fire

enkle poler, bare polene I ±z + er innenfor C og begge


1 b π
residyene er = . Dette gir I=
2( z 2+− z 2- ) 8 √ b+1 √1+b
blir integranden 1, og I =π

RESIDUE THEOREM SIMPLER WAY:



∫−∞
dx
1+ x 2
. We

dz
consider ∫∞ 1 + z2 ∂u=∂ u
2

We have a differential equation: 2 2 . on the
Where C is closed ∂x ∂ y
boundary of the semi intervall 0⩽ x⩽ a , 0 ⩽ y⩽ b med grensebetingelser
circle shown here. For u (0, t )=u (0, y)= u(a , y)=0 . We will use separation of
any p>1, the semicircle variables to find the general equation.
incloses the singular point X ' ' ( x) Y ' ' ( y)
u (x , y)= X ( x)Y ( y)=0 Or + =0
z=i and no others; the X ( x) Y(y)
residues of the integrand at z=i is : − X ' ' (x ) Y ' ' ( y)
= =k (constant ) hence we have
1 1 X ( x) Y ( y)
R( i )= lim
z →i
( z −i ) =
( z −i )( z +i ) 2 i
. Then the value of
nπ x
X ' ' +k X =0 X(0) = X(a)=0 X ( x)= X n( x)=sin 2
the contour integral is 2 π i (1 / 2i)=π . Integral in two parts.(1) a IT was the inverse laplace transform of: .
an integral along the x-axis from -p to p; for this part z = x; (2) an 2 2
n π p3( p+2)
integral along the semicircle, where iθ
. Then we have k = k n= 2 , Y ' ' − k Y = 0 , Y(0) = 0 .
z= p e a
USING LAPLACE TRANSFORM TO OBTAIN SOLUTIONS In Case#3: the solution is given by
a a
FOR COUPLED EQUATIONS: y =c1 x cos(β ln ( x))+c 2 x sin (β ln (x))
2
ẍ(t )+2 n ẋ(t )+n x (t)= 0
we will use laplace transfrom
ÿ(t )+2n ẏ(t )+ n 2 y(t )=μ ẋ(t ) Here is an example for Green function.
to obtain the solution of x(t) and y(t). Initial conditions
x(0)=y(0)=0. ẋ(t )=λ We call L(y) = Y and L(x) = X.
2
x ' ' +2 nx ' +n x= 0
2 Transforms of derivatives og y.
y ' ' + 2 ny ' +n y =μ x '
2
L( y' )= pY − y0 , L( y ' ' )= p − p y 0− y 0' we insert this into eq
2 2 2
X ( p +2 n p + n )− λ =0 ⇒ X ( p + n) =λ
2 2 2
Y ( p + 2 n p+n )−μ p X =0 ⇒Y ( p +n) =μ p X
Y=
λ pμ
( p+ n)
4
, X=
λ
( p +n )
2
⇒ L− 1
λ
( p +n )
2
=λ L
−1 1
[
( p + n)
2
] [ ]
L−1
[ 1

1
( p + n) ( p +n)
−1

]
= L [ G ( p )⋅H ( p ) ]= g ∗h by L_34
1 1 −n t − n( t −τ )
G ( p )= ⇒ L[ g (t)] = ⇒ g (t )=e ⇒ g(t −τ )=e
p +n p +n
1 1 − nt −n τ
H ( p )= ⇒ L[ H ( p )] = ⇒e ⇒ h( τ)=e
p +n ( p+n)
t t

∫ 0
λe
−n (t − τ )
e
−n τ
d τ= ∫ 0
− nt
λe d τ⇒ x(t )= λ t e
−n t
⇒ ẋ (0)= λ

Y=
μ pλ
p +n
4
⇒ μ λ L− 1
p
2

1
( p +n ) ( p+ n)
2
[ p
]
−1 − nt
L [G ( p )⋅H ( p)] =g ∗ h⇒ G( p)= 2
⇒(1 −n t )e = g (t )
( p +n )

[ ]
2 3
1 − nt −n t t nt
H ( p )= 2
⇒h (t )=t e ⇒ y( t )=μ λ e −
( p +n) 2 6
THIS IS VERY IMPORTANT FOR DIFFERENTIAL EQ.
y ' ' + y ' −2 y =4 sin 2 x . instead of tackling this problem
directly, we are first going to solve the equation . Since exp(2ix)=
2i x
Y ' ' +Y ' −2 Y =4 e cos 2x + I sin 2x is complex, the solution
Y may be complex also. Then Y =Y R+i Y I is equivalent to
2i x
Y R ' ' +Y R ' ' −2 Y R= ℜ4 e =4 cos 2x
two equations 2i x Since the
Y I ' ' +T I ' − 2Y I = ℑ4 e =4 sin 2x
second equation above is the same as the question, we see that the
solution of the question is the imaginary part of Y. Thus to find
y p for the question we find Y p for the equation above
and take its imaginary part. We observe that 2i is not equal to
either of the roots of the auxiliary equation in Y equation.
Following the method of the last paragraph, we assume a solution DEN INHOMOGEN LIGNINGEN.
2 ix y ' ' + P (x) y ' +Q (x) y =R (x) Vi merker oss at hvis
of the form Y p= C e and subsitute it into Y equation to get:
y h( x ) er en løsning av den homogene ligningen R(x) = 0, og
2i x 2i x 4 4 (−2i −6) −1
(−4+2i −2)C e =4 e ⇒C = = = (i +3) y_p (x) en eller annen løsning av den inhomogene (partikulær
2 i −6 40 5 løsning), så er også: y ( x)= yh (x )+ y p( x) en løsning av den
−1 2ix
Y p= (i +3)e taking the imaginary part of Y_p we find inhomogene ligningen. Spørsmålet er da om denne løsningen
5 passer med gitte grenseverdier for y ( x0) og y ' (x 0) ,
−1 3
y p= cos 2x− sin 2x . slik at vi får den ønskede entydige løsningen. Med
5 5 y h( x)=c1 y 1(x)+ c2 y 2( x) får vi betingelsene
c1 y 1(x 0)+ c2 y 2( x0)= y(x 0)− y p( x0 )
krav til løsning er
c 1 y 1( x0 )+c2 y ' 2(¿0)= y' (x 0)− y p(x 0)
som kjent at determinanten til ligningssettet må være ulik null, dvs.


W= 1 0
y ( x ) y 2 ( x 0)

y ' 1 ( x0) y' 2 (x )
≠0 . Og dette vet vi fra før er ok

y 1(x ) , y 2( x) er lineært uavhengige. (Hvis høyre siden I


siden
ligningsettet ovenfor skulle være lik null, har vi direkte at
y p(x ) er en løsning som tilfresstiller kravene, og er dermed
den søkte entydige løsningen) Dermed er problemet formelt løst.
Den generelle løsningen av den inhomogene diff ligningen har
formen y ( x)= c1 y 1(x)+ c2 y 2( x)+ y p (x) Neste problem blir
imidlertid å finne en partikulær løsning y p(x ) . Her er det
flere metoder som kan brukes. Et vanlig tilfelle er diff ligninger
hvor venstre siden har konstante koeffisdienter.
y ' ' + ay ' +by = R( x) Her kan vi komme langt med lit
strategisk gjetting.
rx rx
1. R(x)= A e , prøv y p( x)=B e
2. R(x)= A sin rx+ B cos rx , prøv y p (x)=C sin rx +D cos rx
3.
R(x)= polynomav grad N , prøv y p( x)= polynom av grad N
4)
x x
R(x)=e (A sin rx + B cos rx ), prøv y p (x)=e (C sin rx + D cos rx)
rx
Eksemple: y ' ' − 2y' + y= A e . Løsning av den homogen lig
x x rx
y h( x)=c1 e +c 2 x e vi prøver y p(x)= B e og finner
A A
B= og dermed y p (x) = .
(r −1)2 (r −1)2
Another example is: y ' ' + 4 y ' + 4y= cosh x We want to find
the general solution to the differential equation, that is we want to
find the solution to the homogeneous differential equation.
y ' ' + 4 y' + 4y= 0 From the characteristic equation
2 2
λ + 4 λ +4 =(λ +2 ) = 0 ⇒ λ =−2, −2 Since we have repeated
root, we have to introduce a factor of x for one solution to ensure
THE CAUCHY-EULER EQUATION. linear independence. So we obtain u 1 =e ,u 2=x e
−2 x −2 x
The
2
x2 d y2 +a x d y +b y =0 We assume a trial solution given by wronskian of these two functions is
dx
dx
∣ ∣
− 2x −2 x
Rearranging gives : m 2+(a −1)m +b =0 we then can solve for e xe − 2x − 2x − 2 x −2 x
dy m− 1 −2 x − 2x
=−e e (2 x −1)+2x e e
y=x
m
, differentiating, we have. =m x And m. There are three particular cases of interest. −2 e −e (2 x −1)
dx −4x −4 x − 4x − 4x
2
d y m− 2
Case #1: Two distinct roots, m 1 and m 2 ¿−e (2 x −1)+2 xe =(−2 x +1 +2x)e =e Because the
=m (m−1) x . Substituting into the original equation Case #2: One real repeated root, m
2 wronskian is non zero, the two functions are linearly independent,
dx Case #3: Complex roots, a ±β i .
2 m− 2 m−1 m so this is in face the general solution for the homogeneous diff
x (m(m−1)x )+ a x(m x )+b( x )=0 In Case#1, the solution is given by:
m
y =c1 x +c2 x
m
1 2
equation. We seek functions A(x) and B(x) so
In Case#2: the solution is given by
m
y =c1 x ln( x)+c 2 x
m A(x)u 1+ B (x) u2 is a general solution of the non
homogeneous equation. We need only calculate the integrals
A(x)=−∫
1 1
u (x)B (x)dx , B (x)=∫ u1 b (x)dx i.e.
get
1 2
y
' +

f ( y )dy=const.
W 2 W CASE (d) : An equation of the form
1 2
− 2x 2x
A(x)=−∫ − 4x x e cosh x dx=−∫ xe cosh x dx a2 x 2 d y2 +a 1 x d y +a 0 y= f (x) Called an Euler Or Cauchy
e dx dx
1 x 2x equation, can be reduced to a linear equation with constant
A(x)= e ((9( x −1)+e (3 x−1))+ C)
−18 coefficients by changing the independent variable from x to z
1 − 2x 1 x dy dy
B(x)=∫
2x 2x
e cosh xdx =∫ e cosh x dx = e (3+ e )+C where. x= e
z
For the we have x = and
e− 4 x 6 dx dz
2 2
FROM CHAPTER 9: EULER EQUATION EXERCISES. x2 d y2 = d y2 − d y
x2
dx dz d z
I =∫ x √1+ y ' 2 dx We have F (x , y , y ' )= x √ 1 + y' dx
2
CASE (e) = Reduction of order. To find a second solution of
x1
y ' ' + f (x) y ' + g( x)=0 given one solution u(x), substitute
∂ F ∂ (x √1+ y' )
2
∂F 2 d √u 1 y =u (x)v(x) into equation above and solve for v(x).
=0, = ⇒u = y' +1 ⇒ =
∂y ∂ y' ∂ y' du 2 √ u A DIFFERENTIAL EXERCISE BY USE OF LAPLACE-

{
2 2
d ( y ' +1) d(y ' ) d 0 t <0 y 0= y 0' = 0
2 x x + (1) f ( t )=
∂ √ 1 + y' dy ' dy dy 1 t >0
x ⇒ =
∂y' 2 √ y ' +1
2
2 √ y ' +1
2

∂F x y'
= from Euler equation we have
partia y' √ y' 2+1
∂ ∂ F − ∂ F =0 ⇒ ∂
∂x ∂y' ∂y ( x y'
∂ x √ y' 2 +1 )
=0 ;
∂F
∂y
=0

x y' The rest of column 1 left. Fourier transform example:


=c(constant ) 2 2 2
(x y' ) =c ( y' +1)
√ y' 2+1 Uttrykk f(x) ved hjelp av uttrykket du har funnet for F(w).
2 2 2 2 2 c dy c Vi merker oss at F(w) er like funksjon I w, som ventet, slik at
x y ' = y ' c +c ⇒ y ' = 2 2 , y' = = 2 2 dx
√ x −c d x √ x −c fourierintegralet for f(x) også blir en cosinus-transformasjon
∞ ∞
c c 1 iwx 1
∫ dy=∫ 2 2 dx ⇒ y =∫ 2 dx f ( x)= ∫
√ 2 π −∞
F (w)e dw= ∫
√ 2 π −∞
F (w) cos(w x)dw
√ x −c
1
x
c√ 2
−1

π −∞

2 1 −cos (wa)
w
2
cos (w x)dw
∫ arc cosh( x)= 2 ANOTHER FOURIER TRANSFORM EXAMPLE:
√ x −1 FINN FOURIERREKKA FOR FUNKSJONEN f(x) GITT
y −k =c cosh −1( x / c )⇒
x y −k
( )
⇒ ax =cosh (ay+ b )

{
=cosh 1
c c − (L+ x) − L⩽ x <0 ∞
2 sin n x
1
a= ∧b =
−k f ( x)= tilåfind ∑ .
constants c and k are determined from given 1 n =1 n
c c (L − x) 0 < x ⩽ L
x1 ∧ x 2 . 2
points
HERE IS ANOTHER QUESTION SAME LIKE ABOVE: a ∞ nπ x ∞ nπ x
f ( x)= 0 +∑ an cos +∑ b sin Og koeffisientene
x2 2 n=1 L n= 1 n L
I =∫ (1+ y y ' )2 dx ⇒ F (x , y , y ' )=(1 + y y ' )2 then 1
L
nπ x
x1 finnes av an= ∫ f ( x) cos dx og b_n
∂F ∂F L −L L
=2y ' (1 + y y ' ) ∧ =2 y (1 + y y ' ) L
∂y ∂ y' 1 nπ x
d ∂F ∂F d
bn= ∫ f (x) sin L dx Vi merker oss at f(x) I denne
L −L
− =0 ⇒ ( 2 y (1+ y y ' ) )−2 y ' (1 + y y ' )=0
d x ∂y' ∂y dx oppgaven er en odde-funksjon, slik at vi har an=0 for alle n.
Videre finner vi for b n siden f(x) er odde funksjon.
the derivation ca n happen with product thing. L
2 x
d du dv b n= ∫ (L −x) sin n π dx For å bestemme b_n behøver vi
(u v)= v +u And at last we get. L 0 L
dx dx dx L
2 2 2
2⋅( y ⋅y ' ' + 2 y⋅ y' + y ' )− 2 y ' (1 + y⋅y ' )=0 ⇒2 y( y' ' − y )=0 nπ x −L

and we have to find the x which is equal to x=


1 2 −c'
2c
y −
2c
.
følgende integraler: og ∫
0
sin
L
dx=

n
[(−1) −1 ] og
A long rectangular metal plate has its two
L
long sides and the far end at 0 C and the
RESIDUE EXAMPLE: ∫ 2
cos pi x
2
( 1 −4 )( 1 +4 z )
dxden blir til
∫x sin
nπ x
dx=
2
−L n
(−1) (delvis integrasjon) ~til base at 100 C. The width of the plate is
i pi x
C
L nπ 10cm. Find the steady state temperature
e 0
∫ 2 2
dz Der integrasjonsveien C er en lukket L distribution inside the plate. ∇ T =0
C (1 −4 )(1 +4 z ) sammen finnes da b n= og fourierrekka for f(x) blir 2 2
n π ∂ T + ∂ T =0
kurve ra – R til R på den reelle aksen og en halvsirkel med radius Or 2 2 To solve this

∂x ∂y
R I øvre halvplan. Etter Jordans lemma vil integralet over
halvsirkelen gå mot null når R →∞ . Siden integranden har
poler på den reelle aksen for z = -1/2 og z = ½, må vi også legge
f ( x )=
L
π ∑
n =1
1
n
sin
nπx
L
setter L = π og har da summen equation,
the form.
we are going to try a solution of
T (x , y)= X (x )Y ( y) .
2 2
små halvsirkler rundt disse I øvre halvplan. Videre er det I øvre fra oppgaven. 1 d X 1 d Y 2
= =const =−k k>0 . 2
X ' ' =−k X And
halvplan en pol for z=i/2. Prinsipal verdien av integralet er da gitt DIFFERENTIAL EXERCISE: X d x 2 Y d y2
2

x y ' ' − xy ' + y = x Solve it! 2
Y ' ' = k Y The constant k^2 is called the separation constant.
ved P ∫ cos π x
dx som er lik

{ { }{ }
( 1 −4x 2) ( 1+ 4x2 ) ky ky

¿ 2 π i Res( z=i /2 )+π i [ Res( z=−1 / 2)+ Res( z =1/ 2) ] Alle


−1
π
X=
{
sin k x
cos k x e
e
, Y = − k y ⇒T = XY =
ky
sin k x
cos k x
e
e−k y
We first

polene er enkle, og vi har Res(z =i / 2)= i e 2


den andre discard the solutions containing e since we are given
8 T →0 as y → ∞ . Next discard solutions containing
−i
Res(z=−1/ 2)= Res(z = 1/ 2)= resultatet blir da. cos(kx) since T=0, when x=0. This leaves us just e− k y sin kx ,
8 but the value of k is still to be determined. When x=10, we are to
−π

∫ cos2 pi x 2 dx = π8 (1+e 2 ) have T=0, this will be tru if sin(10k)=0, that is if k =



for
C (1 −4 )(1 +4 z ) 10
FOURIER TRANSFROM EXAMPLE: n=1,2,... thus for any integral n, the solution

{
2x +a −a ⩽ x ⩽0 − nπ y / 10 nπ x
T=e sin Satisfies the given boundary conditions
f ( x)= −2x + 2a 0⩽ x⩽ a Den fourier transformerte er da. 10
0 ellers on the three T=0 sides. Finally we must have T=100 when y=0

− nπ y /10 nπ x
T =∑ b n e sin for y=0 we must have T=100
n= 1 10

nπ x
T y = 0=∑ bn sin =100 for f(x) = 100 with l=10.
n= 1 10

{
10 400
2 nπx odd n
bn= ∫
10 0
100 sin
10
dx= n π T becomes then
0 even n
T=
n
e
(
400 −π y /10 π x
sin
10
+
1 −3 π y / 10 3 π n
3
e sin
10
+...
)
OTHER SECOND ORDER EQUATIONS:
CASE (a): Dependent variable y missing: y ' = p , y' ' = p'
CASE (b) : Independent variable x missing:
dp dp dy dp
y ' = p , y' ' = = =p
dx d y d x dy
CASE ©: To solve y ' ' + f ( y)=0 multiply by y'.
y ' y ' ' + f ( y) y' =0 or y ' dy ' + f ( y )dy=0 then integrate to
ANOTHER FROBENIÜS METHOD EXAMPLE.

Potrebbero piacerti anche