5
$\begingroup$

I want to find an integral representation of the conformal map that sends the upper half plane $\text {Im}z>0$ onto the infinite L-shaped region $$ \Omega = \{z = x+iy; \ x > 0, \ y > 0, \ \min(a,\frac yb) < 1 \} $$ enter image description here

which is a slight modification of this question, where we change the vertex of the turn from the point $1+i$ to the more general point $1+bi$ in the first quadrant. (This is an exercise in Gamelin's text)

By the Schwarz-Christoffel formula, the solution is of the form $$ F(w) = K\int_{z_1}^w (\zeta-z_0)^{-1}(\zeta-z_1)^{-1/2}(\zeta-z_2)^{-1} \ d\zeta $$ where $z_0, z_1, z_2$ are the numbers that get mapped by the conformal mapping to the vertices at $i \infty, 0, +\infty$ respectively. I do not know what are these preimages of the vertices.

In the linked question they are given as $+1, -1, 0, \infty$ for the particular case.

Ahlfors says in his book that in general there is no formula for determining the prevertices, and only $3$ of them can be arbitrarily chosen. Since we have $4$ points to determine in this case, we may assume that $3$ of the prevertices are $z_0=-1, z_1=0, z_2=\infty$.

I was told that the right decision for the remaining vertex is $b^2$. How can I get this? Is there an intuitive explanation for why this is the right choice of the prevertex $z_2$, perhaps using the symmetry of the domain along the line $y=bx$?

$\endgroup$
2
+50
$\begingroup$

The L-shaped domain has four vertices $0,\infty, 1+bi$ and $i\infty.$ As Ahlfors says, we can choose $3$ of prevertices aibitrarily, so we may assume$$ z=-1\longleftrightarrow w=i\infty,\quad z=0\longleftrightarrow w=0,\quad z=\infty \longleftrightarrow w=1+bi.$$ Let the remaining prevertex be $z=a$ (it's value is unknown at the present time). Then the mapping function has the form
$$ F(z) = K\int_{0}^z (\zeta+1)^{-1}\zeta^{-1/2}(\zeta-a)^{-1} \ d\zeta.\tag{1} $$ We have to determine the value of $a$ so that $F(\infty)=1+bi.$ \begin{align} F(\infty)&=F(i\infty)= K\int_0^\infty \frac{idt}{(it+1)\sqrt{it}(it-a)} \\ &= -\frac{K}{\sqrt{2}}\int_0^\infty \frac{(1+i)(t^2+a+it(1-a))}{(t^2+1)\sqrt{t}(t^2+a^2)}dt \\&=I+J+i(I-J), \end{align} where $$ I=-\frac{K}{\sqrt{2}}\int_0^\infty \frac{t^2+a}{(t^2+1)\sqrt{t}(t^2+a^2)}dt,\quad J=-\frac{K}{\sqrt{2}}\int_0^\infty \frac{(a-1)\sqrt{t}}{(t^2+1)(t^2+a^2)}dt.$$ Since $F(i\infty)$ should be $1+bi,$ $K,$ $a$ should be determined so that $$ I+J=1,\quad I-J=b$$ holds. I could not get $a=b^2$, however $(1)$ is an integral expression for $w=F(z)$.
EDIT: $$a=\frac{1}{b^2}$$ will be true, not $a=b^2.$

In the case $b=1$, we can find $F(z)$ explicitly. When $b=1$, we have $a=1.$ Thus $$ -\frac{K}{\sqrt{2}}\int_0^\infty \frac{dt}{(t^2+1)\sqrt{t}}=-\frac{\pi K}{2}=1 $$ gives $K=-\frac{2}{\pi}$ and we get explicitly $$F(z)=-\frac{2}{\pi}\int_0^z \frac{d\zeta }{(\zeta^2 -1)\sqrt{\zeta }}=\frac{1}{\pi}\left(\log \frac{1+\sqrt{z}}{1-\sqrt{z}}+\tan^{-1}\sqrt{z}\right).$$

$\endgroup$
  • $\begingroup$ Thank you. Did you get $a=\frac{1}{b^2}$ by directly doing the the long integral $I$? Perhaps this result for the value of $a$ can be interpreted somehow, heuristically? (There is some calculation considering the residue of the integrand at the prevertices in the linked question that I do not understand but might be relevant. Calculating $F(z)$ explicitly seems not much harder than finding the integral representation, which makes it strange if that's Gamelin's intention) $\endgroup$ – Emolga Apr 20 '17 at 23:39
  • $\begingroup$ Directly doing the looong integral...,using resdue theorem. I'm now looking for easier ways to prove $$ \int_0^\infty \frac{t^2+a+(a-1)t}{(t^2+1)\sqrt{t}(t^2+a^2)}dt=\sqrt{a}\int_0^\infty \frac{t^2+a-(a-1)t}{(t^2+1)\sqrt{t}(t^2+a^2)}dt.$$ $\endgroup$ – ts375_zk26 Apr 21 '17 at 2:31

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.