MIT differential equations---23,24,25

unit impulse

suppose f(t) is the force to time

for an undamped system y'' + y = f(t), which means you give it a kick

now let f(t) be 1/h Uoh(t)

The L.T of f(t) is 1/h  ( U(t) - U(t-h)  )  = 1/h  (1/s  -  e^-hs /s)

Let h goes to 0, the limit will equals 1.

We call this divac function although it's not a function. But it has many good characteristic.

delta(t) -------L------->  1

u(t) f(t) *  delta(t)  -------L ---------->  F(s) 1           L.T of convolution is the product of each one's L.T

u(t) * f(t) ------------L--------->F(s) 

delta(t) is acting as a identity of convolution function, unit element.

u'(t) = delta(t)

general solution to ODE via L.T

y'' + ay' + by = f(t)   y(0) = 0; y'(0) = 0;

s^2 y + as y + by = F(s)

Y = F(s) / ( s^2 + as + b) 

1/(s^2 + as +b ) is called transfer function    W(s). Its inverse L.T is called weight function of the system  W(t).

y(t) = f(t) * w(t)

What's w(t) really?

y'' + ay' + by = delta(t) , kick of To unit impulse

s^2 Y + as Y + bY = 1,  Y = w(t)

Lesson 24

System of first order ODE

x' = f(x,y,t)

y'= g(x,y,t)         x, y are the dependent variables, and t is the independent variable

Linear system 

{  x' = ax + by + R1(t)

{  y' = cx + dy + R2(t)        

if a,b,c,d are constants, it's called const-coefficient system

a,b,c,d can also be function of t

If R1 = R2 = 0, it's called linear homogeneous system

The number of initial conditions = the sum of orders

In terms of modeling, what' the meaning of linear system ODE?

We started discussion of ODE by mixing, radioactive decay, temperature and defusion problems. So when two or more of them occur, we 

get a system.

For example, there's a water bath and you coddle the egg in the water bath.

T1-------temperature of yolk

T2-------temperature of egg water

Te-------temperature of water bath

T1' = a(T2 - T1)

T2' = a(T1 -T2) + b(Te - T2)

The easiest way is to eliminate one of dependent variables. 

Suppose we eliminate T2, then you can get a second order ODE in terms of T1.

So empirically, we are always trading complexity for another.  " conservation of mathematical difficulties"

x' = f(x,y)

y' = g(x,y)  autonomous system

The solution, say , is x = x(t) , y = y(t). this is known a s the parametrised curve

Then x'(t), y'(t) will be the velocity field. 

Geometric understanding of ODE system.

We didn't begin the class by discussing how to solve ODE but instead, we talk about the geometrichen we can get  understanding.

ODEs can be solved by computer and especially system. The key is to learn how to interpret it. Know why instead of how.

 

LESSON 25

Solving autonomous ODE system by matrix

recall we use trial solution to solve 2nd order ODE

so we discover the solution to autonomous ODE system by this method

the solution shall be in the form

[x;y] = [a1;a2] e^lambda t

****plug in the original ODE system, we can get rid of the exponential and reduce the calculus problem to algebra

In more general form, this procedure can be written as:

[x;y]' = [a b; c d] [x; y]

X' = A X

Then we can get the characteristic equation:

lambda^2 -trace(A) lambda + det(A) = 0

For higher order of X, lambda is the eigenvalue of the matrix A