1
$\begingroup$

Suppose the following function $f(z)$

$$ f(z)=\frac{\sin^{2}z}{z^{2}},\qquad z\in[-\pi,0)\land(0,\pi]. $$

The function has a removable singularity at $z=0$, since

$$ \lim f(z)_{z\rightarrow0}\equiv\lim f(z)_{z\rightarrow0}\frac{\sin2z}{2z}\equiv\lim f(z)_{z\rightarrow0}\frac{2\cos2z}{2}=1, $$

by L'Hospital rule, and $f(0)=1$. We want to expand $f(z)$ at $z=0$. The function has a simple pole of order $m=2$ at $0$. The Laurent series has the form of

$$ h(z)=\frac{b_{2}}{z^{2}}+\frac{b_{1}}{z}+\sum_{k=0}^{\infty}a_{k}z^{k}, $$

where \begin{align*} b_{2} & =\frac{1}{2\pi i}\oint_{C}\frac{f(\theta)}{\theta^{1}}\mathrm{d\theta},\\ b_{1} & =\frac{1}{2\pi i}\oint_{C}\frac{f(\theta)}{\theta^{2}}\mathrm{d\theta},\\ a_{1} & =\frac{1}{2\pi i}\oint_{C}\frac{f(\theta)}{\theta^{3}}\mathrm{d\theta},\\ a_{2} & =\frac{1}{2\pi i}\oint_{C}\frac{f(\theta)}{\theta^{4}}\mathrm{d\theta}, \end{align*} I can determine $b_{1}$ easily as \begin{align*} b_{1} & =\frac{1}{\left(m-1\right)!}\lim_{z\rightarrow z_{0}}\left\{ \frac{d^{m-1}}{dz^{m-1}}[(z-z_{0})^{m}f(z)]\right\} ,\\ & =\lim_{z\rightarrow0}\frac{d}{dz}\left[z^{2}\frac{\sin^{2}z}{z^{2}}\right]=\frac{d}{dz}\sin^{2}z=\left[\sin(2z)\right]_{z=}0=0. \end{align*} How to determine $b_{2}$, $a_{1}$, $a_{2}$? Is the following identity held \begin{align*} a_{1} & =\lim_{z\rightarrow0}(f^{\prime}(z)),\\ a_{2} & =\frac{\lim_{z\rightarrow0}(f^{(2)}(z))}{2!}, \end{align*} where $a_{1},a_{2}$ represent the limits (instead of function values) in the Taylor series? May I ask you for the explanation using the above mentioned formulas?

The solution based on the multiplication of $\sin^{2}z/z^{2}=(2z^{2}/2!-8z^{4}/4!+32z^{6}/6!+O(7))/z^{2}$ is known form me.

The Mathematica solution is

Normal[Series[Sin[z]^2/z^2, {z, 0, 4}]]

$$ h(z)=1-\frac{z^{2}}{3}+\frac{2z^{4}}{45}+O(4). $$

Thanks for your help.

$\endgroup$
  • 1
    $\begingroup$ Since it looks like z is constrained to be real, and the function is even, I would expect all the coefficients of odd powers of z to be 0. So by inspection, $a_1 = 0$. $\endgroup$ – Andy Walls Oct 4 '17 at 18:55
  • $\begingroup$ Why are you saying $z\in [-\pi,\pi]$ when we're in the realm of complex analysis? And why do you include $z=0$ in the domain at the outset, when the expression is not defined there. $\endgroup$ – zhw. Oct 4 '17 at 19:25
  • $\begingroup$ @ zhw.: Thanks for your comment, it was only a typo... Corrected. Could you clarify, please, your first comment? $\endgroup$ – justik Oct 4 '17 at 20:49
  • $\begingroup$ @ Andy Walls: You are right. I spent some time solving the integrals, see the posted solution, please. $\endgroup$ – justik Oct 4 '17 at 20:57
  • 1
    $\begingroup$ The function does not have a pole at $z=0$. As you say, the singularity is removable. That is enough to tell you that $b_2=b_1=0$ $\endgroup$ – Ross Millikan Oct 6 '17 at 14:06
3
$\begingroup$

Using:

$$\cos(2z) = 1-2\sin^2(z)$$

we can write:

$$\frac{\sin^2(z)}{z^2}=\frac{1-\cos(2z)}{2 z^2} = 2\sum_{k=0}^{\infty}(-1)^k 2^{2k}\frac{z^{2k}}{(2k+2)!}$$

$\endgroup$
  • 1
    $\begingroup$ @ Count Iblis: thanks for your post.However, my question refers to the above-mentioned integrals and their solutions... $\endgroup$ – justik Oct 5 '17 at 7:42
1
$\begingroup$

I spent some time solving the integrals. For anyone interested in the solution:

Suppose the disc $C=\left\Vert z\right\Vert =1$ and the substitution

$$ z=e^{i\theta},\qquad\mathrm{dz}=ie^{i\theta}\mathrm{d\theta.} $$

Then \begin{align*} b_{2} & =\frac{1}{2\pi i}\int_{-\pi}^{\pi}\frac{\sin^{2}e^{i\theta}}{e^{i\theta}}ie^{i\theta}\mathrm{d\theta}=\frac{1}{2\pi}\int_{-\pi}^{\pi}\sin^{2}e^{i\theta}d\theta,\\ b_{1} & =\frac{1}{2\pi i}\int_{-\pi}^{\pi}\frac{\sin^{2}e^{i\theta}}{e^{2i\theta}}ie^{i\theta}\mathrm{d\theta}=\frac{1}{2\pi}\int_{-\pi}^{\pi}\frac{\sin^{2}e^{i\theta}}{e^{i\theta}}d\theta,\\ a_{1} & =\frac{1}{2\pi i}\int_{-\pi}^{\pi}\frac{\sin^{2}e^{i\theta}}{e^{2i\theta}}ie^{i\theta}\mathrm{d\theta}=\frac{1}{2\pi}\int_{-\pi}^{\pi}\frac{\sin^{2}e^{i\theta}}{e^{2i\theta}}d\theta,\\ a_{2} & =\frac{1}{2\pi i}\int_{-\pi}^{\pi}\frac{\sin^{2}e^{i\theta}}{e^{2i\theta}}ie^{i\theta}\mathrm{d\theta}=\frac{1}{2\pi}\int_{-\pi}^{\pi}\frac{\sin^{2}e^{i\theta}}{e^{3i\theta}}d\theta. \end{align*} Taking into account that \begin{align*} \frac{1}{2\pi}\int\sin^{2}e^{i\theta}d\theta & =\frac{1}{4\pi}\left(\theta-\int\cos(2e^{i\theta})\mathrm{d\theta}\right), \end{align*} the principal part is

\begin{align*} b_{2} & =\frac{1}{2\pi}\int_{-\pi}^{\pi}\sin^{2}e^{i\theta}d\theta\\ & =\frac{1}{4\pi}\left[\theta\right]_{-\pi}^{\pi}-2\pi,\\ & =0.\\ b_{1} & =-\frac{1}{4\pi}ie^{-i\theta}\left[-1+\cos2e^{i\theta}-2e^{i\theta}\int_{-\pi}^{\pi}\sin(2e^{i\theta})\mathrm{d\theta}\right]_{-\pi}^{\pi},\\ & =\frac{1}{4\pi}\left[ie^{-i\theta}-ie^{-i\theta}\cos2e^{i\theta}\right]_{-\pi}^{\pi}+\frac{i}{2\pi}\int_{-\pi}^{\pi}\sin(2e^{i\theta})\mathrm{d\theta,}\\ & =\frac{1}{4\pi}\left(ie^{-i\pi}-ie^{-i\pi}\cos2e^{-i\pi}-ie^{i\pi}+ie^{i\pi}\cos2e^{i\pi}+0\right),\\ & =\frac{1}{4\pi}\left(-i+i\cos2+i-i\cos2+0\right),\\ & =0. \end{align*} Analogously, the remaining coefficients are \begin{align*} a_{0} & =-\frac{1}{8\pi}ie^{-2i\theta}\left[-1+\cos2e^{i\theta}-4e^{2i\theta}\int_{-\pi}^{\pi}\cos(2e^{i\theta})\mathrm{d\theta}-2e^{i\theta}\sin(2e^{i\theta})\right]_{-\pi}^{\pi}=1,\\ a_{1} & =\frac{1}{12\pi}ie^{-3i\theta}\left[1+\left(-1+2e^{i\theta}\right)\cos2e^{i\theta}+e^{i\theta}\sin2e^{i\theta}-4e^{3i\theta}\int_{-\pi}^{\pi}\sin(2e^{i\theta})\mathrm{d\theta}\right]_{-\pi}^{\pi}=0\\ a_{2} & =\frac{1}{12\pi}\left[-\frac{1}{24}ie^{-2i\theta}\left(-2+3e^{-2i\theta}\right)\cos2e^{i\theta}+\frac{1}{24}i\left(3e^{-4i\theta}-8\int_{-\pi}^{\pi}\cos(2e^{i\theta})\mathrm{d\theta}\right)+\right.\\ & =\left. +\frac{1}{12}ie^{-i\theta}\left(-2+e^{-2i\theta}\right)\sin2e^{i\theta}\right]_{-\pi}^{\pi},\\ & =-\frac{1}{3}. \end{align*}

Finally, the Laurent series lacks its principal part $$ h(z)=1-\frac{z^{3}}{3}+O(4). $$

$\endgroup$
1
$\begingroup$

This one is quite obvious , you don't even have to calculate b2 and b1 as they will surely be zero , since function has removable singularity it will not contain any bi's .

Now to find any ak it should be clear from the Laurent's series expansion that one must differentiate whole series k times to remove the (kth and lesser) degree monomial which is present with it and then divide by the factor k! Which has appeared after differentiating the series . Then finally tend the limit to zero so that all the terms after ak vanishes. Then finally we can calculate any coefficient ak

$\endgroup$
  • 1
    $\begingroup$ @ Devendra Singh Rana: Thanks for your helpful comment and explanation. I was not sure whether the removable singularities affecting $b_i$ or not. Does it mean that all non-principal members represent the above-mentioned limits or not? $\endgroup$ – justik Oct 4 '17 at 21:05
  • 1
    $\begingroup$ Ya you can calculate any coefficient by this method $\endgroup$ – Devendra Singh Rana Oct 5 '17 at 4:32
  • $\begingroup$ @Justik if you like the explanation please provide upvote $\endgroup$ – Devendra Singh Rana Oct 5 '17 at 19:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.