MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

What is required in order to derive the expanding eigenvalues of Dr. Curt McMullen's torus orbifold bundles over the circle and the corresponding totally degenerate groups, as presented in Section 3.7 of his book, Renormalization and 3-Manifolds which Fiber over the Circle?

He provides this value for the orbifold bundle obtained from the torus with a singular point of order 2, but I cannot reproduce his method. I am eager to understand and, further, to derive the expanding eigenvalue for the orbifold bundle obtained from the torus with a singular point of order 3.

My inability to calculate these values is disconcerting because I am able to derive the values for the fixed points and the lengths of the singular closed geodesics analytically. My ambition is to have a coherent and comprehensive understanding of Dr. McMullen's work on the totally degenerate group which is isomorphic to the fundamental group of a 2-dimensional orbifold of genus 1 with a single cone point of order 3.

I believe that my fundamental problem is that I do not grasp the mathematical motivation for Dr. McMullen's reference for his work:

T. Jørgensen, "Compact 3-manifolds of constant negative curvature fibering over the circle", Ann. of Math. (2) 106: 61–72 (1977).

Thank you in advance for your assistance.

share|cite|improve this question
Since Dr. McMullen is alive, well, and in full possession of his faculties, why not ask him? – Igor Rivin Sep 1 '11 at 19:57
Dr. Rivin, you may be assured that I have, more than once and over a year ago. Dr. McMullen was kind enough to send the input file for drawing the "snowflake" limit set with his lim program. Given the constraints on his time, it is understandable that he could not respond to my inquiries for further information about this "snowflake" group. Turning your question on its head, I consider you to be a top candidate for providing a correct answer. – steve Sep 1 '11 at 20:45
Could you give some specific details in your post on what you tried and where precisely you got lost (e.g. with a short quotation or summary of the section of the book)? Your question taken literally "What is required in order to derive ..." is quite broad. – j.c. Sep 1 '11 at 22:55
I couldn't read such a dense block of text, so I added some white space. This will help people understand what you are trying to ask :) – David Roberts Sep 2 '11 at 0:22
Steve, we really need to know more of your background. What courses have you taken that relate to this? – Will Jagy Sep 2 '11 at 0:47
up vote 5 down vote accepted

Let me first make a comment on Jorgensen's work. His Annals paper that you refer to was based on computations he first made in the cusped case, but did not appear in print until recently. He then did "orbifold Dehn filling" to get a closed example, which he was then able to publish explicitly, without revealing the origin of his techniques. I believe Thurston was inspired by Jorgensen's examples to prove his geometrization theorem for fibered manifolds. One may see pictures of the Ford domains and limit sets for the cusped case using the program Opti. Unfortunately, though, you can't draw the types of "snowflake curves" with this program, unless maybe you know the coordinates of group precisely.

Given a mapping class of a punctured torus, you get a corresponding map of its representation variety. If you want the commutator to lie in a specific conjugacy class (like an elliptic of order 3), then you get a very specific equation for the traces of elements in the group related to the Markoff equation, which is given in Equation 3.3 in McMullen's book (for derivation of these types of relations in $SL(2,C)$, you could refer to Chapter 1 of Marden's book on Kleinian groups).

So now you get an action of the mapping class on the traces, which McMullen computes explicitly for the transformation $LR$, and will be the same algebraic transformation regardless of the value of $tr[A,B]$.

McMullen computes (see Marden) $$L(\alpha, \beta,\gamma)=(\gamma, \beta, \beta\gamma-\alpha),$$ $$R(\alpha,\beta,\gamma)=(\alpha,\gamma,\alpha\gamma-\beta).$$

These preserve the Markoff equation $\alpha^2+\beta^2+\gamma^2=\alpha\beta\gamma+3$ which holds in your case of an elliptic commutator of order 3. Solving for the fixed point $LR(\alpha,\beta,\gamma)=(\alpha\gamma-\beta,\gamma,\gamma^2\alpha-\beta\gamma-\alpha)=(\alpha,\beta,\gamma)$, we obtain the fixed point $(\alpha,\alpha/(\alpha-1),\alpha/(\alpha-1))$, subject to the Markov equation. Plugging in, we see that $\alpha$ is a root of the equation $\alpha^4-3\alpha^3+6\alpha-3=0$. This equation has two real roots, and two complex conjugate roots, either of which gives the desired fixed point (up to complex conjugation) giving the traces of a totally degenerate group.

Now, to compute the eigenvalue of the fixed point, we compute the characteristic polynomial of the derivative of $LR$ (which is a 3x3 matrix). I won't go through the computation, but I get the polynomial $\lambda^3-(2\alpha^2/(\alpha-1))\lambda^2 +2\alpha^2/(\alpha-1)\lambda-1$ after substituting for the fixed point. Dividing by the obvious factor $\lambda-1$ (which I think comes from the fact that the mapping class group has a conserved quantity $tr[A,B]$ on the character variety), we have your desired eigenvalue is a root of the polynomial $\lambda^2 - (2\alpha^2/(\alpha-1) +1)\lambda +1$.

share|cite|improve this answer
Dr. Agol, I am extremely grateful to you for your well-expressed answer. I already had all the references which you cite, and I have played with Opti. I have even scanned some of your papers. Still, my ignorance could be modeled by a sphere-filling, fractal curve, filling all space and time. Wonderfully, you have provided a seamless approach to some mathematics that I find very seductive. I only wish that I could add more than 15 reputation points to your current total of 11,106. You will have the opportunity to earn more points when I ask questions probing the precise meaning of "entropy". – steve Sep 6 '11 at 18:15

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.