Summary of Ordinary Differential Equations

Download as doc, pdf, or txt
Download as doc, pdf, or txt
You are on page 1of 3

Up to Midterm 1

First order linear ODE


y p(t ) y g (t )
Exact Differential
M (t , y ) N (t , y ) y 0
M N

Exact if
t
y
Implicit solution M (t )dt h( y ) such that

M,
N
t
y

Integrating factors
Make it so M and N become related by the product rule
M Ny ' ( yO)
If p is integrable then, e p (t ) dt
N M

y
t
y
t
If it is known that is a function of only y or t, then simplify
First order linear ODE solution
( y (t ) (t )) g (t ) (t ) F
In general, M

y (t ) (t ) F
t

F (t ) F (t0 ) (t ) g (t )dt
t0

Second order homogenous


ay by cy 0
Characteristic equation (assume solution is in the form of y Ce rt
ar 2 br c 0 with roots r1 , r2
rt
rt
Solution y C1e 1 C2e 2
rt
rt
Double root: y C1e tC2 e

t
Complex conjugate roots: r i, y e (C1 cos( t ) C2 sin( t ))
Wronskian matrix (Second order)
f g
W ( f , g )(t )
f g
If f and g are linearly independent, W 0 for all t
For two solutions y1 , y2 to y py gy 0
pdt
W Ce
Higher order: Nth order: rows are f, g, h, etc, and columns are D0, D1, D2, etc.
Up to Midterm 2
Higher-order homogenous DE with constant coefficients (very similar to 2nd-order)
Find roots to characteristic equation
Complex will come in pairs, do cos/sin thing
Multiple roots, multiply by increasing powers of t
Non-homogeneous second order DE
Solution is sum of steady-state (particular) and transient (complementary) solution
Transient is solution of homogenous equation
Steady-state is the solution to the non-homogenous equation, unique to IVCs
Solving for steady-state solutions
Reduction of order
Have one solution
Guess y2 (t ) u (t ) y1 (t ) , differentiate, plug in, find u

Variation of Parameters (Second-order)


Homogenous solutions y1, y2 to y py qy g
Guess solution Y u1 y1 u2 y2 and let u1 y1 u2 y2 0 , differentiate
Y u1 y1 u2 y2 and Y u1 y1 u2 y2 u1 y1 u2 y2
Substitute and get u1 y1 u2 y2 g
t
t
y2
y1
gdt y2
gdt
Solution Y Y0 y1 t0
t
0 W(y , y )
W ( y1 , y2 )
1
2
Larger order (second order, too)
Nth order has N parameters (u1,2N)
N

Y yiui
i 1

Wi
gdt
W
Wi is the Wronskian with the ith column replaced by [0,0,1]
Undetermined coefficients
Guess solutions form, find arbitrary coefficients
t
g e t , guess y p Ae
g sin( t ) , guess y p A sin( t ) B sin( t )
ui

Non-constant coefficients
Series Solution (for a neighborhood near an ordinary point a [ p (a ) y, p( a) 0 ])

i
Solution is of the form y ai t
n 0

n 1

n2

i 1
i 2
Derive to get y iait , y i (i 1)ai t

Substitute into the equation, try to combine


Create similar exponents for t (usually i)
Shift summations: n=n+2 for instance
Ex: constant coefficients

n 0

n 0

y (i 1)ai 1t i , y (i 2)(i 1)ai 2t i


Then, sum of power series is zero, Theorem: all coefficients zero
Solve for coefficients (get a recursive formula)
Unknown starting coefficients(a0, a1, etc)
IVC coefficient

i
Find solution by plugging coefficients into y ai t
n0

Up to Final
Cauchy-Euler Equation
n

a y
i 0

(i )

x i 0, ax 2 y bxy cy 0, an 1

Assume solution is in the form y x r


Get a polynomial find roots r
ri
Real r: y ci x , multiply by ln(x) for multiple roots
i

Complex r: given x r er ln x x i x ei ln x
y | x | (c1 cos( ln | x |) c2 sin( ln | x |))
Can shift x by x0 (singular point) by x-x0 into x
Singular Points

py qy ry 0 , x0 singular if p(x0)=0
q
r
x x0 ) , lim( x x0 ) 2 finite (they are analytic), else irregular
Regular if xlim(
x0
x

x
0
p
p
Solutions near a Regular Singular Point

n 0

n 0

n
2
n
xq/p and x2r/p analytic at x=x0, so ( x x0 ) p pn ( x x0 ) , ( x x0 ) q qn ( x x0 )

Take L[y], multiply by x and divide by p to get x 2 y x( xp ) y ( x 2 q ) y 0


2
If you take p0 and q0 only: x y xp0 y q0 y 0 , an Euler equation (indicial equation)
2

r n
r n 1
, y an (r n)(r n 1) x r n 2
Assume solution y an x , y an (r n) x
n0

n 0
r+n

n 0

Get some power series in x as well as an x term, the same as the n=0 equation case
xr coefficient=0 is the indicial equation, roots are exponents at the singularity
Solve in order highest to lowest since usually lower root is more complicated
Solve lower roots only if they do not differ by an integer (else complicated)
Combine sums, coefficient=0 yields a series equation, solve for each root from the indicial equation
Laplace Transformation

L{ f (t )} F ( s ) e st f (t )dt
0

at
Exists if f (t ) Ke for t>M; a, K, M, constants, K>0, M>0 on s>a
Laplace transformation is linear
L{ f (t )} sL{ f (t )} f (0)
n

L{ f ( n ) (t )} s n L{ f (t )} s n i f ( i 1) (0)
i 1

Uses of the Laplace Transformation


L{L[y]} gives a function Y(s)=L{y} which is the Laplace transformation of the solution
Y(s) can be analyzed for solutions by doing a Laplace transformation backwards
Protips: Partial fractions for quotient polynomial Y(s)
1
a
s
L{e at }
, L{sin at} 2
, L{cos at} 2
2
sa
s a
s a2
Textbook p317 for Laplace transformations of elementary functions
equation

You might also like