0

Let $L1$ be the $x$-axis, let $L2$ be the $y$-axis and let $L3$ be the vertical line $x = 1$. For each $k ∈ Z$ let $C_{k}$ denote the circle of radius $r = 1/2$ with centre $z = 1/2 + ki$. Let $f(z) = {2z}/{(z+1)}$.

1) Sketch $L1, L2, L3$ and the circles $C_{k}$.

2) Show that $f(z)$ maps $L1$ to itself and find the images $f(L2)$, $f(L3)$ and $f(C_{0})$ under the transformation $f(z)$.

I have attempted to sketch the points, I believe I should end up with circles one above each other but not sure?

I'm struggling with part 2) too. I'm told as a hint to consider the angles between $L1, L2, L3$ and $C_{0}$ but don't know how to use this. In my attempt I have just subbed say $z=iy$ for $L2$ to represent the y axis. I then did the same for 3 points on the circle to end up with $f(1/2)=2/3, f(-1/2) = -2$ and $f(i/2) = 2i/5+1/5$ but I have no idea whether this is right or where I would go from here as finally I am asked to produce another sketch of all of the images of the points without any further calculations.

harry55
  • 449

2 Answers2

1

For the circle $C_0$ I think that the approach you have taken is correct, by looking at $3$ points on the circle and studying where they get mapped.
But note that the points you chose are not on $C_0$.

$C_0$ is the circle around $z=1/2$ with radius $r=1/2$.
Let's look at $$z_1 = 1,\ \ z_2=\frac{1}{2} + \frac{i}{2}, \ \ z_3 = 0$$ under $f(z) = \frac{2z}{z+1}$ we have: $$ f(z_1) = 1, \ \ f(z_2) = \frac{4}{5}-\frac{2}{5}i, \ \ f(z_3) = 0$$

Since we know that Mobius transformations map circles and lines to circles or lines, we can check that these $3$ points lie on the same $C_0$ circle around $z=1/2$ with radius $r=1/2$.

I hope this helps.

Update:
To show that $f(z_1),\ f(z_2),\ f(z_3)$ lie on $C_0$ we can follow the steps in https://math.stackexchange.com/a/213670/201454

We get the following $3$ equations: $$(1-x_c)^2 + y_c^2 -r^2 = 0$$ $$(\frac{4}{5}-x_c)^2 + (-\frac{2}{5} - y_c)^2 -r^2 = 0$$ $$x_c^2 + y_c^2 -r^2 = 0$$

Subtracting the third equation from the first gives: $$1 - 2x_c = 0 \implies x_c=\frac{1}{2}$$

Subtracting the third equation from the second gives: $$\frac{16}{25}-\frac{8}{5}x_c + \frac{4}{25} + \frac{4}{5}y_c = 0 \implies \frac{4}{5} - \frac{8}{5}x_c + \frac{4}{5}y_c = 0$$

Plugging in $x_c=1/2$ gives $y_c=0$

Now plug in $x_c=1/2,\ y_c=0$ in the third equation to get $r=1/2$

benji
  • 2,436
  • Thank you! How would I show exactly that those 3 new points still lie on $C_0$? – harry55 Dec 17 '16 at 12:49
  • You can follow the steps in http://math.stackexchange.com/a/213670/201454 – benji Dec 17 '16 at 14:41
  • Okay, using what for the coordinates exactly as they are complex? Would I use say $x=4/5$ and $y= 2/5$ For $f(z2)$? – harry55 Dec 17 '16 at 15:02
  • That should be a -2/5 for y sorry – harry55 Dec 17 '16 at 15:03
  • As I followed the steps with the coordinates this way and have ended up with a centre of $(1/2 +i/2, i/4)$ but this would not be the same circle so not quite sure where I've gone wrong. – harry55 Dec 17 '16 at 15:30
  • Yes, you need to use $x=4/5,\ y=-2/5$. I will update my answer later. If you want you can double check your calculations. You can also verify that the distance between $f(z_2)$ and $1/2$ is $1/2$. – benji Dec 17 '16 at 16:00
  • I've got it now thank you! So would L3 the line x=1 be mapped to the point 2?. I'm understanding bits and pieces of this question now but am still struggling to put it all together. And why is it I have been given the hint to consider the angles between my lines and circle? As we haven't needed to use it so far. – harry55 Dec 17 '16 at 16:10
  • No, $L_3$ is a line, so under a Mobius transformation it must be mapped to a line or a circle. Try to apply the same method I used in my answer, and look at $3$ points, e.g. $1,\ 1+i,\ \infty$ and their mapping under $f$ and try to determine what is $f(L_3).$ I am honestly not sure why the hint is helpful here. I think that the method I showed is simple enough to apply. – benji Dec 17 '16 at 19:33
  • okay thank you so much for your help! Okay I will do that but I don't quite understand how $1+i$ lies on the line x=1? And when I do the map for $L1$ what would the three points be there? Is it infinity, 0 and something? – harry55 Dec 17 '16 at 20:21
  • The points on the line $x=1$ are represented as $z=1+ti$ where $t$ is real. When you map $L_1$ you can choose e.g. $0,\ 1,\ \infty$ – benji Dec 17 '16 at 23:52
  • So for L2 the y axis, do I chose the points $0, i$ and infinity? After performing the map I end up with 0, infinity and $-1-i$ so does it just map to another line? I think this is wrong because the previous answer to this question suggests it maps to a circle. – harry55 Dec 18 '16 at 19:50
  • $0,\ i,\ \infty$ are good choices for $L_2$. These points are mapped to $0,\ i+1,\ 2$ respectively. So $f(L_2)$ is the circle around $z=1$ with radius $r=1$. – benji Dec 18 '16 at 20:14
  • if infinity gets mapped to 2 how is the line L1 mapped to itself? As it would get mapped to 0,1 and 2. $\not=$ L1 as there is no point of infinity to produce the line? – harry55 Dec 18 '16 at 20:57
0

(partial answer)

Let us consider the image of $(L_2)$ by function $f$.

The idea is to describe the "generic element" $z \in (L_2)$ as being:

$$z=i t , \ \ \ t \in \mathbb{C}, \ \ \text{giving} \ \ \ f(it)=\frac{2it}{it+1}$$

Let us now convert this expression in order that the denominator doesn't contain any more a complex non-real number. This is done by multiplying the numerator and the denominator of this fraction by the conjugate expression of the denominator, i.e.,

$$f(it)=\frac{2it(-it+1)}{(it+1)(-it+1)}=\frac{2t^2+i2t}{1+t^2}=x+iy \ \ \text{with} \ \ \cases{x=\frac{2t^2}{1+t^2}\\y=\frac{2t}{1+t^2}}$$

or in a more interesting manner, by a $-1$ translation on the $x$ axis:

$$\cases{x-1=\frac{1-t^2}{1+t^2}\\y=\frac{2t}{1+t^2}}$$

These are the well known parametric equations for the unit circle (see this). All the unit circle is reached but one point. Could you find it?

Thus the looked for image of $(L_2)$ is circle is with center $(1,0)$ and radius $1$.

Indeed one checks that $(x-1)^2+y^2=1$;

But in fact there is a missing point which is $(0,2)$ which is never "reached".

Edit: Circle $(C_0)$ is parametrically described as being the set of complex numbers of the form: $$z=\frac12+\frac12 e^{it}=\frac12+\frac12(\cos(t)+i \sin(t)).$$

Thus, the image of circle $(C_0)$ by transformation $f$ is thus the set of points of the form:

$$f(z)=\dfrac{(1+\cos(t))+i \sin(t)}{(\frac32+\frac12 \cos(t))+i (\frac12 \sin(t))}$$

You just have to multiply, as before, the numerator and the denominator by the conjugate of the denominator.

Jean Marie
  • 81,803
  • thank you that's helped me see things a bit clearer but I'm still a bit stuck when it comes to the circle $C_0$. Are the points I've chosen to transform correct? If so I'm struggling to see where they go to. None of the points are sent to infinity under the mapping so I know it is not a line so does this mean the image is another circle? – harry55 Dec 17 '16 at 00:00
  • Yes, on this ground, it should be another circle. Besides I have added an Edit to my answer. – Jean Marie Dec 17 '16 at 00:28