Gary
GaryVasco
Posts: 3,352
|
Post by Gary on May 24, 2016 2:03:12 GMT
Vasco
I am hoping that someone can see where I have gone wrong in calculating $S_E(w)$ or in some fundamental misinterpretation of the these pages:
(A link to a version revised on May 29, 2016 appears in later in the thread.)
Gary
|
|
|
Post by Admin on May 24, 2016 9:08:47 GMT
Gary
On page 1 of your document, the second equation from the bottom has the second term involving $|w|$, which you have left in place and solved for $\bar{w}$ on the bottom line.
However, just as in the equations on page 256 for the Schwarz function and the Schwarzian reflection of $z$ across $K$, $z\bar{z}=|z|^2$ is only true for $z$ on $K$, so for the ellipse $E$ the Schwarz function is only equal to $\bar{z}$ if $z$ is on the ellipse.
So again $|w|=(w\bar{w})^{1/2}$ is only true on the ellipse. This means that generally (i.e. for points which could be on or off the ellipse), we should write $\tilde{w}=\mathfrak{R}_E(w)$ as on page 255, noting that if $w$ is on the curve then $\tilde{w}=\bar{w}$, and so we should also write your second equation as $\tilde{w}+\alpha(w\tilde{w})^{1/2}+w+\beta=0$, to avoid confusion, where $\alpha$ and $\beta$ are constants depending only on $a$ and $b$, and you need to re-arrange this to solve for $(\tilde{w})^{1/2}$.
We can write $w\tilde{w}=w\bar{w}=|w|^2$ only if $w$ is on the ellipse.
I think the easiest way to do this is to substitute $q=(\tilde{w})^{1/2}$ and $\tilde{q}=(w)^{1/2}$ which gives a quadratic in $q$. Solving this quadratic for $q$ we can then calculate $\tilde{w}$ as $q^2$.
As before you will probably need to be careful choosing the sign of the square root!
Vasco
|
|
Gary
GaryVasco
Posts: 3,352
|
Post by Gary on May 24, 2016 19:52:59 GMT
Gary On page 1 of your document, the second equation from the bottom has the second term involving $|w|$, which you have left in place and solved for $\bar{w}$ on the bottom line. However, just as in the equations on page 256 for the Schwarz function and the Schwarzian reflection of $z$ across $K$, $z\bar{z}=|z|^2$ is only true for $z$ on $K$, so for the ellipse $E$ the Schwarz function is only equal to $\bar{z}$ if $z$ is on the ellipse. So again $|w|=(w\bar{w})^{1/2}$ is only true on the ellipse. This means that generally (i.e. for points which could be on or off the ellipse), we should write $\tilde{w}=\mathfrak{R}_E(w)$ as on page 255, noting that if $w$ is on the curve then $\tilde{w}=\bar{w}$, and so we should also write your second equation as $\tilde{w}+\alpha(w\tilde{w})^{1/2}+w+\beta=0$, to avoid confusion, where $\alpha$ and $\beta$ are constants depending only on $a$ and $b$, and you need to re-arrange this to solve for $(\tilde{w})^{1/2}$. We can write $w\tilde{w}=w\bar{w}=|w|^2$ only if $w$ is on the ellipse. I think the easiest way to do this is to substitute $q=(\tilde{w})^{1/2}$ and $\tilde{q}=(w)^{1/2}$ which gives a quadratic in $q$. Solving this quadratic for $q$ we can then calculate $\tilde{w}$ as $q^2$. As before you will probably need to be careful choosing the sign of the square root! Vasco Vasco, Thank you for the reply. I need further clarification. In sentence 3, you say It does not seem so to me, because w is not on the ellipse E, though it is on the ellipse f(E), where $f = z^2$, z in E. But aside from that, w is just a complex number, so shouldn't the statement be true for all w? To put it another way, all z under consideration are on E, but a number $z^2$ is actually on $\hat{E} = f(z)$, z in E. Again, in sentence 3: The discussion on p. 255 does not take into account the second ellipse created by f(E). My view is that w is on or near f(E) on the RHS of the plot. That being the case, it seems wrong to write $\mathfrak{R}_E(w)$ under the given conditions. Any w should be reflected in $\hat{E}$ rather than in $E$. This requires a second Schwarzian function $S_\hat{E}$ and Schwarzian reflection $\mathfrak{R}_\hat{E}$ for the derived curve. I have used $\mathfrak{R}_f$ in place of $\mathfrak{R}_\hat{E}$. Gary
|
|
|
Post by Admin on May 25, 2016 9:13:10 GMT
Gary
I agree with your comment about $|w|=(w\bar{w})^{1/2}$, and your other comments. I meant to write $|w|=(w\widetilde{w})^{1/2}$ ($w$ tilde not $w$ bar). However this correction is of no real consequence since as I understand from reading your post, my comments in general were not relevant to the problem as you see it.
I have looked again and the only thing I noticed is that at the bottom of page 1 of your document you seem to be wanting to map the original ellipse $E$ using the function $w=f(z)=z^2$. However in the subsequent maths you seem to have taken the equation of the ellipse $E$ and substituted for $z,\bar{z}$ to obtain an equation in $w$. This new equation in $w$ is not the equation of the ellipse under the mapping $f=z^2$ and so I think that as you had concluded yourself, your equation for $\mathcal{S}_f(w)$ is not correct.
Vasco
|
|
Gary
GaryVasco
Posts: 3,352
|
Post by Gary on May 25, 2016 17:55:58 GMT
Vasco,
Thanks for looking at this. The ellipse E on the LHS was plotted using the equation in paragraph 2, p. 256. The ellipse (call it $\tilde{E}$) on the RHS was plotted by applying $z^2$ to the points of E.
I agree. I wasn't sure about how to go about it. I suppose I should solve the original ellipse equation for z and then apply f.
Gary
|
|
|
Post by Admin on May 25, 2016 18:04:17 GMT
Vasco, Thanks for looking at this. The ellipse E on the LHS was plotted using the equation in paragraph 2, p. 256. The ellipse (call it $\tilde{E}$) on the RHS was plotted by applying $z^2$ to the points of E. I agree. I wasn't sure about how to go about it. I suppose I should solve the original ellipse equation for z and then apply f. Gary Gary So the plot of the ellipse on the right should be correct, but if you are using the equation for any of the other calculations then it could obviously be a problem. I have derived the equation but I haven't checked it out yet. Vasco
|
|
|
Post by Admin on May 25, 2016 18:32:54 GMT
Gary
I have checked my $z ^2$ ellipse out, but it looks different from yours in the sense that in your plot the "egg" is lying in a stable position whereas mine is standing on its end in an unstable position. It has a centre at the point $(0,1.5)$ with a horizontal semi-axis of 2.5 and a vertical semi-axis of 3. Does that seem right to you? I also have an equation for it in the form $((x-c)/A)^2+(y/B)^2=1$, where the centre is $(c,0)$, and expressions for $A$ and $B$ in terms of $a$ and $b$.
Vasco
|
|
|
Post by Admin on May 26, 2016 8:53:02 GMT
Gary I have checked my $z ^2$ ellipse out, but it looks different from yours in the sense that in your plot the "egg" is lying in a stable position whereas mine is standing on its end in an unstable position. It has a centre at the point $(0,1.5)$ with a horizontal semi-axis of 2.5 and a vertical semi-axis of 3. Does that seem right to you? I also have an equation for it in the form $((x-c)/A)^2+(y/B)^2=1$, where the centre is $(c,0)$, and expressions for $A$ and $B$ in terms of $a$ and $b$. Vasco Gary I have written a short document which is here showing how I derived the equation of the transformed ellipse. I'm not sure if it will be any use to you. In the course of producing this document and plotting the transformed ellipse, I realised that the vertical semi-axis is not 3 but 2 which puts the ellipse in a stable position the way yours is, with the horizontal axis longer than the vertical axis. So my statements in red in the above quote are incorrect. I certainly agree with your comment in a recent post that drawing a diagram is a good investment. This is in the spirit of Needham's book of course - Visual Complex Analysis! Vasco
|
|
Gary
GaryVasco
Posts: 3,352
|
Post by Gary on May 26, 2016 15:52:13 GMT
Vasco,
Thank you for the document. I think that will get me back on track.
Gary
|
|
Gary
GaryVasco
Posts: 3,352
|
Post by Gary on May 27, 2016 8:37:24 GMT
Vasco, I followed up on your equation for the squared ellipse. It seems to work out well. Two new equations derived from it produced the same ellipse in the plot as the former, but there was a great difference in the Schwarzian function. The composite function $\mathfrak{R}_f \circ f \circ \mathfrak{R}_E(z)$ now seems to work as intended. See if you think it is what Needham had in mind. I know the plot is overly busy, but there were quite a few interdependencies in the problem. nh.ch5.pp.256-257.pdf (241.9 KB) (Revised May 29, 2016) By the way, there is a typo in your document, paragraph 3: Should be $1 - 2 sin^2 t = cos 2t$. It wasn't carried through. Gary
|
|
|
Post by Admin on May 27, 2016 9:23:28 GMT
By the way, there is a typo in your document, paragraph 3: Should be $1 - 2 sin^2 t = cos 2t$. It wasn't carried through. Gary Thanks. I have corrected this, and the link in my original post now points to the corrected version. I will get back to you about the other points as soon as I have studied your document. Vasco
|
|
|
Post by Admin on May 27, 2016 15:26:44 GMT
Vasco, I followed up on your equation for the squared ellipse. It seems to work out well. Two new equations derived from it produced the same ellipse in the plot as the former, but there was a great difference in the Schwarzian function. The composite function $\mathfrak{R}_f \circ f \circ \mathfrak{R}_E(z)$ now seems to work as intended. See if you think it is what Needham had in mind. I know the plot is overly busy, but there were quite a few interdependencies in the problem. View AttachmentGary Gary The way I see the ideas on page 257 is that they are the full generalisation of Schwarz's Symmetry Principle, which subsection 5 of section XI slowly introduces, building slowly from 'simple'(?) ideas to the general Schwarz Principle. The way I understand your document is as follows: The curves $L$ and $\widehat{L}$ in the particular case you have investigated are the segments of the two ellipses. The function $f$ is $z^2$ and $\mathfrak{R}_L$ and $\mathfrak{R}_{\widehat{L}}$ are the Schwarzian reflections in the two ellipses. The function $f^{\ddagger}$ is the analytic continuation of $f=z^2$ from one side of $E$ to the other, and so should turn out to be $z^2$ on this extended region. The example you have chosen is an example like the one described on page 254 in the second paragraph which begins: "Of course if $f$ is already defined...". I would be interested to know if $f^{\ddagger}$ is indeed behaving like $z^2$. In your document, at the end, where you list problems you have identified, it seems to me that problem 2 is not really a problem. Generally speaking the Schwarzian Reflection function $\mathfrak{R}$ only approximates to normal reflection. The closer the point is to the curve (in your case the ellipse), the more accurately the Schwarzian reflection coincides with normal reflection. So I don't think it is surprising that $f(a)$ and $f\circ\mathfrak{R}_E(a)$ do not quite lie on the circle. Vasco
|
|
Gary
GaryVasco
Posts: 3,352
|
Post by Gary on May 27, 2016 16:13:55 GMT
Vasco,
Thank you for looking it over.
The best indicator on this plot would be the green points, which are plotted with $f\ddagger = \mathfrak{R}_f \circ f \circ \mathfrak{R}_E$, as you can see in the parens after v@ in this line of Mathematica code:
diskPtsOnfE = {EdgeForm @ Black, FaceForm @ Green, Table[Disk[v @ (REf @ f @ RE @ pt), rDisk + .05], {pt, ptsOnE}]};
Also notice that f(a) and $\mathfrak{R}_f \circ f \circ \mathfrak{R}_E(a)$ fall on the same point of the plot.
I agree about the locations of $f(a)$ and $\mathfrak{R}_E(a)$. The explanation can be seen in the distortion in Figure [31], p. 256.
I will rewrite a bit later on.
Gary
|
|
|
Post by Admin on May 28, 2016 14:49:51 GMT
Gary
I have looked again at your document and apart from the typo of "a=1 and b=1" on page 3 just before your formula for $\mathcal{S}_f(w)$, it all seems to tie together very well. I have used a spreadsheet to check the formula for $\mathcal{S}_f(w)$ and it seems spot on. Do you still have any concerns about it?
Vasco
|
|
Gary
GaryVasco
Posts: 3,352
|
Post by Gary on May 28, 2016 15:20:53 GMT
Vasco,
Thank you very much for checking it. I'll make the correction and small revisions at the first opportunity.
Gary
The document has been fixed and revised a bit as of May 29, 2016.
|
|