1
$\begingroup$

Let $z=x+iy$ be a complex number.

I want to find when the solutions of the equation $$\sin z =u,u\in\mathbb{R}$$ are real.

I used this approach:

Since $\sin z=\sin x\cosh y+i \cos x\sinh y=u$, then $$\sin x\cosh y=u,\cos x \sinh y=0.$$ From the second equation we have $\cos x=0$ or $y=0$. If $y=0$, the solutions are real, (since $\cosh 0=1$ the solutions would be the solutions of $\sin x=u$). If $\cos x=0$ then $x=\frac{\pi}{2}+n\pi,n\in\mathbb{Z}$ and $\sin x=(-1)^n$.

So, if $-1<u<1$, the case $\cos x=0, y\neq 0$ is impossible since $|\pm\cosh y |\geq 1$ for all $y$, so in this case all the solutions are real.

For the case $|u|=1$ we get $\pm\cosh(y)=u$ and restricting $x$ appropriately we get $y=0$ too.

In the case that $u>1$ or $u<-1$ then it's clear that $z\notin\mathbb{R}$ so the case $y=0$ is impossible. But since $\cosh y$ doesn't change it's sign, if $u>1$ we must restrict $x=\frac{\pi}{2}+2k\pi$ and we get 2 solutions $y_1=-y_2$ of $\cosh y=u$. So the solutions are given by $y=\pm y_1$ and $x=\frac{\pi}{2}+2k\pi$.

In the case $u<-1$ it's similar, the solutions are $y=\pm y_1',x=-\frac{\pi}{2}+2k\pi$.


I might use another approach:

$$z=\arcsin u=-i\ln(iu\pm\sqrt{1-u^2})$$ If we consider $|u|<1$, then $\sqrt{1-u^2}$ is real, so $$|\pm\sqrt{1-u^2}+iu|=1$$ and $$-i\ln(iu\pm\sqrt{1-u^2})=\theta_u+2k\pi$$ where $\theta_u$ satisfies $\cos \theta_u =\pm\sqrt{1-u^2}, \sin \theta_u=u$. So the solutions are real.

In the case $|u|>1$, $\sqrt{1-u^2}=i\sqrt{u^2-1}$ so in this case $$|i\pm\sqrt{u^2-1}+iu|=|\pm\sqrt{u^2-1}+u|$$ and $$-i\ln(iu\pm\sqrt{1-u^2})=-i\ln|\pm\sqrt{u^2-1}+u|+2k\pi$$ in this case the solutions are not real, since $u^2-1\neq u^2$ for all $u$.

Is there any easier, shorter, funnier, or even a strange approach to the problem (It doesn't have to be shorter or easier, it may be a large and strange approach, or a geometric one, or something like that)?

Note that we're trying to find out when all the zeroes of $\sin z-u=0$ are real.

$\endgroup$
  • 1
    $\begingroup$ if the equation has a real solution then u must be between minus one to one.conversely, for any real number bwtee minus one to one, we must have a solution. $\endgroup$ – Ben Dec 1 '16 at 1:36
  • 1
    $\begingroup$ Yes, that's true. But how would you prove that for any real number between minus one to one, the only possible solutions are real? $\endgroup$ – David Molano Dec 1 '16 at 1:41
  • $\begingroup$ oh right. you have a point here. $\endgroup$ – Ben Dec 1 '16 at 1:48
2
$\begingroup$

Writing $$\sin z = \frac{e^{iz}-e^{-iz}}{2i} = -\frac{i}{2}\left(e^{-y}(\cos x + i\sin x)-e^y (\cos x -i\sin x)\right)\\ =\sin x + \frac{i}{2} \cos x \left( e^y-e^{-y} \right) $$ For the imaginary part of $\sin z$ to be zero, either:

  • $y=0$ so that $e^{y}-e^{-y}=0$, in which case $z$ is real and $\sin z \in [-1,1]$.

  • or $\cos x = 0$ in which case $z = \frac{2n+1}{2}\pi+i y$, and $\sin z = 0]$.

$\endgroup$
  • $\begingroup$ I got lost in the last part you say $\sin z=0$. In your formula for $\sin z$, when $\cos x=0$, $\sin z=\sin x=\pm 1$. And to say the truth, when I make the calculations i don't get that formula, but \begin{align*}-\frac{i}{2}\left( e^{-y}(\cos x+i\sin x)-e^{y}(\cos x-i\sin x)\right) &=\frac{e^{-y}+e^{y}}{2}\sin x+i\frac{e^{y}-e^{-y}}{2}\cos x\\&=\cosh y\sin x+i\sinh y \cos x\end{align*} $\endgroup$ – David Molano Dec 1 '16 at 2:06
1
$\begingroup$

Since you assume that $u$ is real, there is an alternative method: Use Pythagorean Theorem we get $\cos z=\pm\sqrt{1-u^2}$=$\pm{i}\sqrt{u^2-1}$. Now use $e^{iz}=\cos z+i\sin z$, take the $\ln$ on both sides and you are done. (assuming you know how to take the complex $\ln$). A numerical problem is worked out here, it might be helpful: Finding $z\in \mathbb C$ with $\sin z = 2$?

$\endgroup$
  • $\begingroup$ If we use $\cos z=\pm\sqrt{1-\sin^2z}$ instead $\pm\sqrt{1-\sin^2z}$, yes, that's a way. But in the end, it's a (an interesting one) way to derive the formula for the complex $\arcsin$, which I use in my second approach. $\endgroup$ – David Molano Dec 1 '16 at 2:35
  • $\begingroup$ Yes. My approach is quick if $u$ is real. In case you want to solve $sinz=a+bi$ as a general case, then you need to resort to stuff like you have posted, and the answer Mark has given to you. A general formula isn't "compact" in the sense that it is easy to remember, unfortunately... $\endgroup$ – imranfat Dec 1 '16 at 2:43

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.