Solve an ODE using Fourier series

Click For Summary
The discussion focuses on solving the ordinary differential equation (ODE) y''(t) + ay(t) = y(t + π) using Fourier series. It establishes that for the equality to hold, a must equal (-1)^n + n^2, leading to the condition a = m^2 + (-1)^m for non-trivial solutions. The linear independence of the functions e^(int) is highlighted, necessitating that the coefficients c_n must not all be zero. Specific cases for m, such as m=1 and m=2, are explored, revealing that only certain coefficients can be non-zero, leading to solutions of the form c_me^(imt) + c_{-m}e^(-imt) for m ≠ 0. The analysis concludes that for m=0, the solution simplifies to a constant term c_0.
psie
Messages
315
Reaction score
40
Homework Statement
Find the values of the constant ##a## for which the problem ##y''(t)+ay(t)=y(t+\pi), \ t\in\mathbb R##, has a solution with period ##2\pi## which is not identically zero. Determine all such solutions.
Relevant Equations
The complex Fourier series of a ##2\pi## periodic function, namely ##\sum_{n\in\mathbb Z} c_ne^{int}##.
I've assumed ##y(t)## to be the sum of a complex Fourier series, and we get $$\sum (-n^2)c_ne^{int}+\sum ac_ne^{int}=\sum c_ne^{int}e^{in\pi},$$ which we can write as $$\sum ((-n^2)+a)c_ne^{int}=\sum (-1)^n c_ne^{int}.$$ We see here that equality holds if ##a=(-1)^n+n^2##. But how do I solve ##y''(t)+ay(t)=y(t+\pi)## when ##a=(-1)^n+n^2##. I don't think I understand the problem fully.
 
Physics news on Phys.org
The functions ##e^{int}## are linearly independent, so you must have
$$\sum_{n=-\infty}^\infty \underbrace{[a-(n^2+(-1)^n)]c_n}_0 e^{int} = 0.$$ To get a non-trivial solution, ##a=m^2+(-1)^m## for some ##m \in \mathbb{Z}##. (Don't use ##n## because that's the dummy variable.)

What do you get if ##m=1##? ##m=2##?
 
vela said:
What do you get if ##m=1##? ##m=2##?
Hmm, we need ##[a-(n^2+(-1)^n)]c_n## to be zero for all ##n## and we want a non-trivial solution, so ##c_n## can’t be zero for all ##n##. It can be non-zero when ##n=m##. I guess we can allow for the coefficient ##n=-m## also to be non-zero, since ##a(m)## is even.

For example, if ##m=1##, we should get ##c_1e^{it}+c_{-1}e^{-it}## being the only terms that remain. The solution takes the form ##c_me^{imt}+c_{-m}e^{-imt}## for ##m\neq 0##. For ##m=0##, we get that the only coefficient that can be nonzero is ##c_0##, a constant solution.
 
Last edited:
First, I tried to show that ##f_n## converges uniformly on ##[0,2\pi]##, which is true since ##f_n \rightarrow 0## for ##n \rightarrow \infty## and ##\sigma_n=\mathrm{sup}\left| \frac{\sin\left(\frac{n^2}{n+\frac 15}x\right)}{n^{x^2-3x+3}} \right| \leq \frac{1}{|n^{x^2-3x+3}|} \leq \frac{1}{n^{\frac 34}}\rightarrow 0##. I can't use neither Leibnitz's test nor Abel's test. For Dirichlet's test I would need to show, that ##\sin\left(\frac{n^2}{n+\frac 15}x \right)## has partialy bounded sums...