Evolving Systems
Was listening to strangs lecture on diagonalization Well , let me ask you one thing ……. why to study it at all
Whats the point of knowing the decompostion of a matrix in to S D Sinv… Remember it always is S times D
BECOZ
the derivation goes like
AS = DS
where D is the diagnol matrix with eigen values on the diagonal Ok, whats the big deal about diagnolization
WELL , once I can diagonalize, I can calulate the powers of A very easily….Eigen vectors dont change, but eigen values gets multiplied
Thats the key..
Another brilliant fact is that : You can always solve a recurrence relation
For example, a fibonacci … you basically figure out a way to change the recurrence relation to a matrix equation
then you diagnolize it and you can then write the equation for a growing system
Thats a wonderful idea
This means that for an evolving system, the eigen values and eigen vectors of the first order difference equation is the key.
I am certain that hamilton would have dealt it that way
Let me check .
But this 20 min talk by strang in lecture 16 was too too superb..I somehow really am surprised at what all new things one can learn from revisiting profs lectures
I heard this lecture in the cab ride back to my home. I am blogging about it while in the cab.
Why ? Becoz …well, its a marathon..you got to run , irrespective of wherever you are..
Takeaway
- Evolving System == Diagnolization
- Diagonlization == eigen values and eigen vectors
- Repeated eigen values means basically you are screwed as the matrix cannot be diagonlized
- Det A – Lambda I = 0 gives the characteristic equation which will help you solve the equation
I CHECKED HAMILTON JUST NOW AND HE WAS USING EIGEN VECTORS AND Using DIAGNOLIZATION…. wow!!! I AM SUPER HAPPY that I could immediately make the connection between strangs lecture and solving ARMA equations. Strang’s words — Now you can solve dynamic evolving systems…actually helped me make the connection