0
$\begingroup$

Hi i was wondering if anyone could help me with the following problem.

In particular part c i have obtained the series i just do not know how to get to the result is it a simple case of just substituting the value of x into the Fourier series. also i am struggling with part a i don't understand how to sketch a diagram over three periods

if anyone could provide help or a website in which i could use to learn how to do these parts of the problem i would be extremely happy, Thanks

enter image description here

My Fourier series is $F(x)=\frac{\pi}{2}+\sum_{i=1}^n \frac{1-\cos(\pi n)}{n^2}\cos(nx)$

  • 0
    Please use capital letters and non run-on sentences.2017-02-18
  • 0
    What is your Fourier series.?2017-02-18
  • 0
    i have updated my post.2017-02-18
  • 0
    Sounds weird that someone manages to compute the Fourier coefficients of a given function, but is not able to draw two lines.2017-02-18
  • 0
    6 lines pal and i was unsure of the "3 periods " part, not the actual graph2017-02-18
  • 0
    @Gibberish You did not accept my answer. Is something wrong with it? Do you have any other objections than those you already pointed to me in the comments?2017-02-22
  • 0
    @NeedForHelp Apologies i thought i did but apparently i didn't it is now though. thanks for all the help2017-02-22

1 Answers 1

1

(a) You are given $f(x)$ for $-\pi\leq x\leq \pi$. Hence you can sketch the function over $[-\pi,\pi]$. The condition $f(x+2\pi)=f(x)$ means that $f$ is $2\pi$-periodic. So to sketch $f$ over three periods means to sketch $f$ on an interval of length $3\cdot2\pi=6\pi$ and you can do so by replicating the sketch on $[-\pi,\pi]$. What you get is: enter image description here

(b) To obtain the Fourier series of $f$ we must calculate its Fourier coefficients. In complex exponential form, the $k$-th coefficient is $$ \hat{f}(k) = \frac{1}{2\pi} \int_{-\pi}^{\pi}f(t)e^{-ikt} \,dt $$ Here because $f$ is given piecewise we split the integral in two: \begin{align} \hat{f}(k) &= \frac{1}{2\pi} \int_{-\pi}^{\pi}f(t)e^{-ikt} \,dt \\ &= \frac{1}{2\pi}\left( \int_{-\pi}^{0} (t-\pi)e^{-ikt} \,dt+\int_{0}^{\pi} (\pi-t)e^{-ikt} \,dt \right) \end{align} It is a simple calculus exercise to evaluate these integrals (do it by integration by parts for the $te^{-ikt}$ components). Don't forget to treat the cases $k=0$ and $k\neq0$ separately. The result is: \begin{align} \hat{f}(k) &= \begin{cases}\frac{\pi}{2}&&k=0\\ \frac{1+e^{-i\pi k}}{\pi k^2} && k\neq0\end{cases} \\ &=\begin{cases}\frac{\pi}{2}&&k=0\\ \frac{1+(-1)^{k+1}}{\pi k^2} && k\neq0\end{cases} \end{align} The Fourier series of $f$ is then $$ \sum_{k=-\infty}^{\infty}\hat{f}(k)e^{ikt} $$

(c) Since $f$ is continuous at $0$ and $C^1$ in a punctured neighborhood of $0$, we know the symmetric partial sums of the Fourier series of $f$ converge to $f(0)$. That is, $$ \lim_{n\to\infty}\sum_{k=n}^{n}\hat{f}(k) = f(0) $$ which, in this case, becomes $$ \frac{\pi}{2} + \lim_{n\to\infty}\sum_{\substack{-n\leq k \leq n \\ k\neq0}}\frac{1+(-1)^{k+1}}{\pi k^2} = \pi\tag{1} $$ Since $\displaystyle\frac{1+(-1)^{k+1}}{\pi k^2}$ is even, $(1)$ reduces to $$ \frac{\pi}{2} + 2 \lim_{n\to\infty}\sum_{k=1}^{n}\frac{1+(-1)^{k+1}}{\pi k^2} = \pi\tag{2} $$ Finally, we note that the numerator $1+(-1)^{k+1}$ equals $0$ when $k$ is even and $2$ when $k$ is odd. So $(2)$ reduces to $$ \frac{\pi}{2} + \frac{4}{\pi} \sum_{k=1}^{\infty}\frac{1}{(2k-1)^2} = \pi\tag{3} $$ The series in $(3)$ is precisely the one we are asked to evaluate. From $(3)$ we obtain $$ \sum_{k=1}^{\infty}\frac{1}{(2k-1)^2} = \frac{\pi^2}{8} $$ as was to be shown.

  • 0
    i think this was a great solution however it is quite complicated, we have been taught to find the $a_n$ coefficents and the $b_n$ ones also, for example $a_n=\frac{1}{\pi} \int_{-\pi}^\pi f(x)\cos(nx) \,dx$ and $b_n=\frac{1}{\pi} \int_{-\pi^}\pi f(x)\sin(nx) \,dx$ where the Fourier series is given by $F(x)=\frac{a_0}{2}+\sum_{n=1}^\infty a_n\cos(nx)+b_n\sin(nx)$ i have no understanding of using complex exponentials can you explain what is going on? so i can understand how this method works, thanks for your solution2017-02-18
  • 0
    @Gibberish If you want to use the real $\cos$ / $\sin$ form of the Fourier series of $f$, then you can proceed like that yes. You don't even have to calculate $b_n$ because $f$ being even implies $b_n=0$ for all $n\geq1$. You've found $a_0=\pi/2$ and your $a_n$ are apparently off by a multiplicative factor $2/\pi$, that is, $a_n=\frac{2}{\pi}\frac{1-\cos(\pi n)}{n^2}$ for $n\geq1$. Then just note that $\cos(\pi n)=(-1)^n$ and you're directly taken to equation $(2)$ in my answer.2017-02-18
  • 0
    may i ask why are the $a_n$ terms are off by this factor of $\frac{2}{\pi}$2017-02-18
  • 0
    @Gibberish Well, I did not calculate them, but you have to get to my equation $(2)$ where there is that factor. I don't mean that $$a_n=\frac{1}{\pi} \int_{-\pi}^\pi f(x)\cos(nx) \,dx$$ is off by a factor $2/\pi$ (it is not). I mean that in this particular case, the evaluation of the integral should give $$\frac{1}{\pi} \int_{-\pi}^\pi f(x)\cos(nx) \,dx =\frac{2}{\pi}\frac{1-\cos(\pi n)}{n^2} $$2017-02-18
  • 0
    yeah thats what i got when i evaluated thanks, how but i dont see how that goes to your (2) since thats k+1 and here it would just be n?2017-02-18
  • 0
    @Gibberish $-(-1)^n=(-1)^{n+1}$2017-02-18
  • 0
    okay thanks last question, how do we know that equation (2) is equal to pi?2017-02-18
  • 0
    @Gibberish The FS of $f$ at $\color{#c00}{0}$ is $$\frac{\pi}{2}+\frac{2}{\pi}\sum_{n=1}^{\infty}\frac{1+(-1)^{n+1}}{n^2}\cos(n\cdot\color{#c00}{0})$$ and $\cos(n\cdot\color{#c00}{0})=1$ for all $n\geq1$. As I justified, this FS is equal to $f(\color{#c00}{0})$ which is $\pi$ here. By the way one can show from real analysis tools that $$\sum_{k=1}^{\infty}\frac{1}{(2k-1)^2}$$ converges and since $f$ is continuous at $0$ its FS at $0$, *if convergent* (it must not be the case for continuous functions!), **must** (by Fejér's theorem) converge to $f(0)$.This is another justification2017-02-18
  • 1
    @Gibberish By the way I made a mistake when I calculated $\hat{f}(k)$. The fact that $f$ is even does not permit us to calculate $\hat{f}(k)$ by doubling the value of the integral from $0$ to $\pi$ because the *integrand* $f(t)e^{-ikt}$ is **not** even. Actually if $f$ is even then its Fourier coefficients $\hat{f}(k)$ must be even. Then we have \begin{align}a_0&=\hat{f}(0)\\ a_k &= \hat{f}(k)+\hat{f}(-k) \\ b_k &= i(\hat{f}(k)-\hat{f}(-k))\end{align} and we refind what you got. I'll edit my answer.2017-02-18