3
$\begingroup$

I am given the function $w(z)=\int_0^z \frac1{\sqrt{1-t^2} \sqrt{1-k^2t^2}}dz$ and shall show that this is mapping the upper half-plane onto a rectangle. We just discussed the Schwarz-Christoffel integral, and we can rewrite this as $\int_0^z \frac1{\sqrt{t-1} \sqrt{t+1} \sqrt{kt-1} \sqrt{kt+1}}dt$, and since the exponents of the four factors are all $-1/2$, we have $\alpha_i-1=-1/2$ which tells us that we are dealing with four right angles.

But, isn't the map actually mapping the unit circle to the rectangle, and not the upper half-plane onto the rectangle? Where is my mistake?

Also, I want to show that the inverse function extends to a meromorphic function on $\mathbb C$. What is the trick here? I don't have any idea on it.

Best regards,

$\endgroup$
  • 2
    $\begingroup$ Note: $w(z)$ is one of the canonical examples of an elliptic integral, and the inverse function of $w(z)$ is in fact one of Jacobi's elliptic functions. There should be a discussion of this particular Schwarz-Christoffel mapping in McKean and Moll's book. $\endgroup$ – J. M. is a poor mathematician Dec 2 '11 at 0:14
  • $\begingroup$ See also the discussion here. $\endgroup$ – J. M. is a poor mathematician Dec 5 '11 at 2:23
  • 1
    $\begingroup$ Hi @Guesswhoitis., I studied the section in McKean's book just now, and I understand how the real line, traces out the perimeter of a rectangle. However, how do we know that the UHP is necessarily mapped to the interior of the rectangle? This doesn't seem obvious to me, for some reason. Is it just simply testing a point from the UHP, e.g., integrating the above integral from 0 to +i, and try to show that +i actually gets mapped to the interior of the rectangle? Thanks, $\endgroup$ – User001 Jul 18 '15 at 1:23
2
$\begingroup$

As pointed out by J.M., this is an elliptic integral, whose inverse function is Jacobi's elliptic function $\text{sn}$.

As $z$ travels along the real line, the argument of $\sqrt{(1-t^2)(1-k^2t^2)}$ undergoes sudden $\pi/2$ changes as $z$ crosses the points $1,-1,1/k,-1/k$, provided $k$ is real. Hence the real line is mapped to the boundary of a rectangle of side lengths $\int_{-1}^1 \frac{dt}{\sqrt{(1-t^2)(1-k^2t^2)}}$ and $\int_1^{1/k}\frac{dt}{\sqrt{(1-t^2)(1-k^2t^2)}}$ (provided here that $|1/k|>1$). Hence no, it is not the unit disc which is mapped to a rectangle. However, the upper-half plane and unit disc are conformally equivalent and by a change of variables (eg. $z \mapsto (z-i)/(z+i)$) you can obtain a conformal map which does take the unit disc to the rectangle.

Proving that the inverse function extends to a meromorphic function on $\mathbb{C}$ can be done using the Lagrange inversion theorem.

A very detailed study of this integral can be found in Markushevish's centennial Theory of functions.

$\endgroup$
  • $\begingroup$ Conventionally, the modulus $k$ is taken to be $0 < k < 1$ in most applications, so in that case $1/k$ is definitely bigger than $1$. For other real values of $k$, things are a bit more complicated, but there are ways to reduce to the $0 < k < 1$ case. $\endgroup$ – J. M. is a poor mathematician Dec 5 '11 at 4:55
  • $\begingroup$ This shows the meromorphicity and doubly periodic character of the inverse of the elliptic integral, and more. $\endgroup$ – J. M. is a poor mathematician Dec 5 '11 at 5:00
  • $\begingroup$ Thank you for the added precision J.M. :) $\endgroup$ – Bruno Joyal Dec 5 '11 at 5:59
  • $\begingroup$ Which value $k$ does sn corrspond to? $\endgroup$ – PyRulez Dec 22 '17 at 1:12

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.