PDE: Proving that a set is an orthogonal bases for L2

  • #1
RJLiberator
Gold Member
1,095
63

Homework Statement


Show that the set {sin(nx)} from n=1 to n=∞ is orthogonal bases for L^2(0, π).

Homework Equations




The Attempt at a Solution


Proof: Let f(x)= sin(nx), consider scalar product in L^2(0, π)

[itex]
(ƒ_n , ƒ_m) = \int_{0}^π ƒ_n (x) ƒ_m (x) \, dx = \int_{0}^π sin(nx)sin(mx) \, dx = \frac{1}{2} \int_{0}^π cos(n-m)x - cos(n+m)x \, dx = \frac{1}{2} ( \int_{0}^π cos(n-m)x - \int_{0}^π cos(n+m)x \, dx = \frac{1}{2} ( {\frac{\sin(n-m)x}{(n-m)x}} - {\frac{\sin(n+m)x}{(n+m)x}} ) = 0
[/itex]

Of course the last result is from 0 to pi for both parts, but didn't know how to latex it :p.

My question is, is this everything to show that the set is orthogonal bases for L2 (0, π)?

I think it is, but definitions in the crazy world of PDE are confusing to me. I do not believe I need to show completeness per the question.

Thank you.
 

Answers and Replies

  • #2
andrewkirk
Science Advisor
Homework Helper
Insights Author
Gold Member
3,902
1,466
My question is, is this everything to show that the set is orthogonal bases for L2 (0, π)?
That's part of the answer, as it proves the elements of the set are pairwise orthogonal.
But now you need to prove it's a basis, which involves proving:
  1. That the set spans ##\mathscr{L}^2(0,\pi)##, ie that any element of ##\mathscr{L}^2(0,\pi)## can be written as the limit of a sequence of linear combinations of elements of the set.
  2. That the set is linearly independent, ie that there is no nontrivial finite linear combination of elements that is equal to zero.
 
  • Like
Likes RJLiberator
  • #3
RJLiberator
Gold Member
1,095
63
@andrewkirk I see what you are saying.

I only proved that the system was orthogonal, but did not show that it was a bases.

I'm having a little trouble going further.

I guess I can set d = sin(nx)/||sin(nx)||, where {d} from 1 to infinity is a complete orthogonal set in L^2.

I can expand this in terms of {sin(nx)} for any f in L^2[0,pi]

f = sum from 1 to infinity <f,sin(nx)>sin(nx)/||sin(nx)||^2

So I can show that it is complete orthogonal set in L^2
 
  • #4
andrewkirk
Science Advisor
Homework Helper
Insights Author
Gold Member
3,902
1,466
If you've shown it's complete then all that's left to prove is linear independence, and that follows easily from the orthogonality. If we have a finite set of nonzero functions in the set, for which there is a linear combination that is zero, how can we derive a contradiction from that using the fact that they're all orthogonal?
 
  • Like
Likes RJLiberator
  • #5
RJLiberator
Gold Member
1,095
63
Could we take two elements and take the inner product and by orthogonality we show that element*element = 0. thus we see linear independence?
 
  • #6
andrewkirk
Science Advisor
Homework Helper
Insights Author
Gold Member
3,902
1,466
That's the broad idea. But it's not necessarily just two elements.

All that's needed for the set to not be linearly independent is for there to exist some integer ##N>0## and set of complex numbers ##c_1,....,c_N##, such that ##c_N\neq 0## and
$$\sum_{k=1}^N c_k\sin kx=0$$

Following your suggestion of taking an inner product, what could you take an inner product of with both sides of this equation, to obtain (via a few more steps) a contradiction?
 
  • Like
Likes RJLiberator
  • #7
RJLiberator
Gold Member
1,095
63
Ugh, I understand exactly what you are trying to do, but my mind is blown.

The right side will be 0 as anything product 0 = 0.
The left side, perhaps the set of complex numbers (c1, c2,...,ck)
 
  • #8
andrewkirk
Science Advisor
Homework Helper
Insights Author
Gold Member
3,902
1,466
We want to take the inner product of the left hand side with a vector/function that will give us a nonzero left side. That will give us a contradiction because the right side is zero.

The inner product of the left side with a function f is the sum of the inner products of each term on the LHS with f.
What should we choose f to be so that we know that at least one of those N inner products is nonzero?
Bear in mind that there is only one of the ##c_k## that we know to be nonzero. Which one?
 
  • #9
Samy_A
Science Advisor
Homework Helper
1,241
510
@andrewkirk I see what you are saying.

I only proved that the system was orthogonal, but did not show that it was a bases.

I'm having a little trouble going further.

I guess I can set d = sin(nx)/||sin(nx)||, where {d} from 1 to infinity is a complete orthogonal set in L^2.

I can expand this in terms of {sin(nx)} for any f in L^2[0,pi]

f = sum from 1 to infinity <f,sin(nx)>sin(nx)/||sin(nx)||^2

So I can show that it is complete orthogonal set in L^2
Maybe I'm missing something, but how did you show the set is complete?
 
  • Like
Likes RJLiberator
  • #10
RJLiberator
Gold Member
1,095
63
We know that C_N is non zero.
So we can multiply both sides by C_N*sinkx ?

As the inner product of C_N and C_N is 1.
 
  • #11
RJLiberator
Gold Member
1,095
63
@Samy_A

From my notes, I can take say g(x) = sin(nx)/||sin(nx)|| then {g(x)} from N = 1 to infinity is an orthonormal set in L^2.

f = sum from n =1 to n=infinity of <f, sin(nx)>sin(nx) / ||sin(nx)||^2
||f||^2 = sum from n = 1 to n = infinity of |<f,sin(nx)>|^2/||sin(nx)||^2

{sin(nx)} from n = 1 to n = infinity is thus a complete orthogonal set in L^2 (0,pi).
 
  • #12
Samy_A
Science Advisor
Homework Helper
1,241
510
@Samy_A

From my notes, I can take say g(x) = sin(nx)/||sin(nx)|| then {g(x)} from N = 1 to infinity is an orthonormal set in L^2.

f = sum from n =1 to n=infinity of <f, sin(nx)>sin(nx) / ||sin(nx)||^2
||f||^2 = sum from n = 1 to n = infinity of |<f,sin(nx)>|^2/||sin(nx)||^2

{sin(nx)} from n = 1 to n = infinity is thus a complete orthogonal set in L^2 (0,pi).
Ok, if it is given (or proven) in your course that the set is complete, then that's it. :smile:
 
  • Like
Likes RJLiberator
  • #13
andrewkirk
Science Advisor
Homework Helper
Insights Author
Gold Member
3,902
1,466
We know that C_N is non zero.
So we can multiply both sides by C_N*sinkx ?
If by 'multiply' you mean 'take the inner product of' then yes. Can you do that and get to the solution?
 
  • Like
Likes RJLiberator
  • #14
RJLiberator
Gold Member
1,095
63
Yes, my apologies @andrewkirk , I mean take the inner product of it.

I'm at work now, so I'm trying to work this out in my head, but...
If we take the inner product of both sides, we would see that all the terms cancel out to be left with
Cn*Cnsin^2(kx)

where Cn*Cn is Cn multiplied by the conjugation of cn (inner product)

(I will try this out in more detail upon returning home in a number of hours)
 

Related Threads on PDE: Proving that a set is an orthogonal bases for L2

  • Last Post
Replies
2
Views
1K
  • Last Post
Replies
6
Views
1K
Replies
2
Views
863
Replies
1
Views
900
  • Last Post
Replies
3
Views
1K
  • Last Post
Replies
13
Views
2K
  • Last Post
4
Replies
79
Views
4K
Replies
1
Views
5K
  • Last Post
Replies
9
Views
2K
Top