Can anyone help with eigenfunction?

  • Context: Graduate 
  • Thread starter Thread starter samleemc
  • Start date Start date
  • Tags Tags
    Eigenfunction
Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
7 replies · 3K views
samleemc
Messages
9
Reaction score
0
if y''-2y'+y=ky k=eigenvalue, y(0,pi)=0, 0<x<pi
find corresponding eigenvalues and eigenfunctions.

thx a lot!
 
Physics news on Phys.org
How would you solve the ode

y''-2y'+(1-k)y=0

Regarding k as a parameter for the moment
 
I got this

sqrt{k} = m
A(1-m)2 -2A(1-m)+A(1-k)=0
B(1+m)2 -2B(1+m)+B(1-k)=0

and what then !??

Please help! thank you very much!
 
It's a 2nd order linear ODE with constant coefficients.
Trying the solution y=e^mx (as I think you've done) gets your solutions (unless...etc, look it up).
Solve for m, then you have one function that's an eigenfunction, with eigenvalue corresponding to m.
 
samleemc said:
if y''-2y'+y=ky k=eigenvalue, y(0,pi)=0, 0<x<pi
find corresponding eigenvalues and eigenfunctions.

thx a lot!
The characteristic equation for y"- 2y'+ (1- k)y= 0 is [itex]r^2- 2r+ 1-k= 0[/itex] which is the same as [itex]r^2- 2r+ 1= (r- 1)^2= k[/itex] and has roots [itex]r= 1\pm \sqrt{k}= 1\pm m[/itex] with your choice of m as [itex]\sqrt{k}[/itex].

The general solution is [itex]y= Ae^{(1+m)t}+ Be^{(1-m)t}[/itex]
Setting that equal to 0 at x= 0 and [itex]\pi[/itex], we find that as long as 1+ m and 1- m are real, A and B must be 0.

In order for k to be an eigenvalue, k will have to be negative so that m is imaginary. Given that, and writing m= ni, [itex]y= Ae^{(1+ni)t}+ Be^{(1-nit)}= e^t(Ae^{nit}+ Be^{-nit})[/itex]. We can write that as [itex]y= e^t(C cos(nt)+ D sin(nt))[/itex].

Now we have [itex]y(0)= e^0(Ccos(0)+ D sin(0))= C= 0[/itex] and [itex]y(\pi)= e^{\pi}(Ccos(n\pi)+ Bsin(n\pi)= Be^{\pi}sin(n\pi)[/itex] (because C= 0) and that must be equal to 0. That will be true either for B= 0 or for [itex]sin(n\pi)= 0[/itex] which will be the case as long as n is an integer.

Can you find the eigenvalues and eigenvectors from there?
 
HallsofIvy said:
The characteristic equation for y"- 2y'+ (1- k)y= 0 is [itex]r^2- 2r+ 1-k= 0[/itex] which is the same as [itex]r^2- 2r+ 1= (r- 1)^2= k[/itex] and has roots [itex]r= 1\pm \sqrt{k}= 1\pm m[/itex] with your choice of m as [itex]\sqrt{k}[/itex].

The general solution is [itex]y= Ae^{(1+m)t}+ Be^{(1-m)t}[/itex]
Setting that equal to 0 at x= 0 and [itex]\pi[/itex], we find that as long as 1+ m and 1- m are real, A and B must be 0.

In order for k to be an eigenvalue, k will have to be negative so that m is imaginary. Given that, and writing m= ni, [itex]y= Ae^{(1+ni)t}+ Be^{(1-nit)}= e^t(Ae^{nit}+ Be^{-nit})[/itex]. We can write that as [itex]y= e^t(C cos(nt)+ D sin(nt))[/itex].

Now we have [itex]y(0)= e^0(Ccos(0)+ D sin(0))= C= 0[/itex] and [itex]y(\pi)= e^{\pi}(Ccos(n\pi)+ Bsin(n\pi)= Be^{\pi}sin(n\pi)[/itex] (because C= 0) and that must be equal to 0. That will be true either for B= 0 or for [itex]sin(n\pi)= 0[/itex] which will be the case as long as n is an integer.

Can you find the eigenvalues and eigenvectors from there?


k have to be negative and n have to be integer, do u mean k=0 ?!
Please answer ! Thanks !
 
samleemc said:
k have to be negative and n have to be integer, do u mean k=0 ?!
Please answer ! Thanks !

k=m2=-n2

Therefore, any k that is a square of an integer (with a minus in front) is an eigenvalue to this problem.