A curve covers a simple curve in
if and only is a (primitive) parabolic element of fixing
Definition
The cusp region (of area ) is the quotient of
the horoball
by the subgroup
Lemma A
If projects to a simple closed curve in then it is either a vertical line or disjoint from the horoball with .
Lemma A If projects to a simple closed geodesic in then it is either:
a vertical line or
disjoint from the horoball with (in grey).
Part 1
Numbers
Boris Springborn, The hyperbolic geometry of Markov's theorem on Diophantine approximation and quadratic forms. Enseign. Math.,
rationals
simple closed curves on the torus
irrationals
quadratic irrationals eg golden ratio
other irrationals
Continued fraction of
where and for .
we write
golden ratio
Best approximation of by rationals
Truncating the continued fraction of at the -th step gives a rational approximation the n-th convergent of .
This is the best approximation of by rationals with
denominator at most .
For example, the best approximations of are
where is the -th Fibonacci number.
Visualization
Farey diagram:
vertices = rationals, edges = Farey neighbors
are Farey neighbors iff
invariant translation of
Visualization golden ratio = 1.6180339...
Visualization golden ratio = 1.6180339...
Constant
Continued Fraction Notation
First 5 Convergents (Fractions)
Golden Ratio ()
Square Root of 2 ()
Euler's Number ()
Pi ()
Cube Root of 3 ()
Can "read" the from the picture:
the number of corners of the "zig-zag" alternately on the left and the right of the vertical line ending at
The modular surface and continued fractions, Caroline Series
Visualization golden ratio = 1.6180339...
zig-zag
Best approximations alternate on the left and the right
rational numbers are poorly approximated by rationals
continued fraction is finite
quadratic irrationals are also poorly approximated by rationals
continued fraction is (pre)periodic, eg
Irrational Number (x)
Constant Cx
Numerical Value
Golden Ratio (φ)
1/
0.4472
1/
0.4472
1/(2
0.3535
1/(2
0.2886
Note: This is related to how deep a (closed) geodesic can go into the cusp of the modular surface.
Part 2
email
My name is Jia Longsong, and I am a Ph.D. student at Peking University. Recently, I have been studying your remarkable identity in detail. In your Acknowledgements, you mentioned that the motivation for the work lies in two series of seminars held at the University of Warwick. The first series concerned Thurston's unpublished work on minimal stretch maps between surfaces and the second Maskit's embedding of the Teichmuller space of the punctured torus. I am truly fascinated by the profound ideas behind your work.
This paper develops a theory of Lipschitz comparisons of hyperbolic surfaces analogous to the theory of quasi-conformal comparisons. Extremal Lipschitz maps (minimal stretch maps) and geodesics for the `Lipschitz metric' are constructed. The extremal Lipschitz constant equals the maximum ratio of lengths of measured laminations, which is attained with probability one on a simple closed curve.
Geodesic laminations = limits of simple closed curves
Definition The Birman-Series set (BS) is the union of the pointset of all compact measured laminations on a hyperbolic surface.
In fact this is just the closure of the set of the union of all
simple closed geodesics.
Theorem 10.2 (Stretch maps)
For any hyperbolic surface of finite area,
the Hausdorff dimension of BS is 0.
In particular it is:
nowhere dense
has measure zero.
Questions
nowhere dense, so where are the holes?
has measure zero, can we exploit this?
Answer for a punctured torus:
Punctured torus
modular torus obtained from a pair of ideal triangles by identification
blue arc is the unique arc disjoint from blue curve
Definition
The (Haas)cusp region is the quotient of the
horoball by
Lemma A+
If is a simple closed geodesic
then it is disjoint from the Haas cusp region.
Lemma A
If projects to a simple curve in then it is either
a vertical line (lift of arc )
or disjoint from the horoball with (lift of )
Corollary
The cusp region is a hole in the Birman-Series set.
The extended Birman-Series set (EBS) is the union of the pointset of all measured laminations on a hyperbolic surface.
In fact this is just the closure of the set
of the union of all simple closed geodesics and arcs.
Theorem 10.2 (essential stretch maps)
For any hyperbolic surface of finite area,
the Hausdorff dimension of EBS is 0.
In particular it is
nowhere dense
and has measure zero.
nowhere dense, so where are the holes in the cusp region?
Geodesics in homology classes
Definition: Let be an essential closed curve
on the punctured torus and denote its length.
Natural map:
Theorem
The shortest representative for a non trivial homology class is always a multiple of a closed simple geodesic.
If are coprime then it is a single geodesic .
Dehn twist round acting on arcs
limit geodesic on pants
Previous slide is the infinite cyclic cover of a pair of pants
obtained by cutting the torus along the geodesic .
limit geodesic on torus
Pants glued up to a torus
Shaded region is a pair of gaps in EBS
torus cut open along lamination
Theorem A
If is an arc then it is isolated in the
It sits in the center of a pair of gaps of area
Corollary
Dehn twist round acting on arcs
The sequence of arcs converges to a geodesic that is
disjoint from the lifts of the arc
asymptotic to a pair of lifts of the closed geodesic .
Dehn twist acting on arcs
grey region is gap in EBS
dotted red lines are lifts of arcs
To promote this to an equality we need to show :
show every gap is "adjacent" to an arc
then apply Theorem 10.2 to show that the union of the gaps has
full measure in the cusp region.
Show every gap is "adjacent" to an arc:
extend the segment bounding a gap to a complete geodesic
look at it in the homology cover, it's essentially a line of slope .
if is irrational use the continued fraction expansion of to find an infinite sequence of arcs converging to this line alternately from left and right. contradiction
so is rational and the geodesic is asymptotic to the lift of a
simple closed geodesic.
Part 3
Another problem
Markoff numbers are integers that appear a Markoff triple
which are solutions of a Diophantine equation the so-called Markoff cubic
Odd index Fibonacci numbers are Markoff numbers
Odd index Pell numbers are Markoff numbers
Frobenius uniqueness conjecture
The largest integer in a Markoff triple
determines the two other numbers.
Uniqueness conjecture: tree structure
The largest integer in a triple determines the two other numbers.
The multiplicity of any number in the complementary regions to the tree is at most 6
Partial results
m = Markoff number then there are at most 6 triples containing m.
transitive mapping class group action on essential simple curves on
Tree structure
comes from Bass-Serre tree of
Role of the character variety
H. Cohn Approach to Markoff’s Minimal Forms Through Modular Functions (1955)
modular torus = quotient of upper half plane by commutator subgroup of , acting by Mobius transformations
relates Markoff numbers to lengths of simple closed geodesics
modular torus = quotient of upper half plane by commutator subgroup of
obtained from a pair of ideal triangles by identification
elliptic involution swaps triangles fixes midpoint of diagonal
Character variety
modular torus =
fundamental group of the torus.
any hyperbolic torus = ,
discrete faithful representation
lifts to
generators of
Definitioncharacter map
Definitioncharacter map
Theorem: (Fricke, Cohn and many others) the semi-algebraic set:
can be identified with the Teichmueller space
of the punctured torus.
there is a finite index subgroup of the automorphisms induced by the action of the mapping class group
the permutation
the involution (the flip)
pairing arcs and curves
simple closed curve coprime integer
simple closed geodesic
hyperbolic length
arc disjoint from the closed geodesic
-length of
traces= lengths
Lemma B
For an appropriate normalization:
-length of
appropriate normalization
= choice of cusp region
Proof (by recurrence):
Let be a pair of simple closed geodesics
on the punctured torus that intersect once.
The corresponding traces together
with some other z>0$ satisfythe Markoff equation:
Call the solutions . These satisfy Vieta's relations:
(Trace relation in )
(Ptomley relation for -lengths)
Bugeaud, Reutenauer, Siksek; A Sturmian sequence related to the uniqueness conjecture for Markoff numbers:
Odd index Fibonacci Odd index Pell =
Up to a multiplicative factor these are sequence of
traces of simple closed geodesics ( lengths of arcs)
obtained by doing Dehn twists on the modular torus.
Traces
Theorem
Let be a simple closed geodesic on the punctured torus and the unique arc disjoint from . and a pair of arcs such that are the sides of an embedded ideal triangle.
Then
1.the (normalised) -lengths of is a Markoff triple.
any power of Dehn twist along gives a new Markoff
triple containing the -length of trace of .
Unicity conjecture
Then
1.the (normalised) -lengths of is a Markoff triple.
any power of Dehn twist along gives a new Markoff
triple containing the -length of trace of .
Unicity conjecture is equivalent to the statement that
no fractional Fenchel-Nielsen twist along
gives integer -lengths to the images of .
Bugeaud, Reutenauer, Siksek
This is equivalent to a statement like:
no "fractional" twist along one of the
takes the arc of length that meets
to an arc of -length .
Bugeaud, Reutenauer, Siksek
How was I going to prove it?
is the intersection of
the line and the axis of
use Dehn twists to construct a sequence of points "too quickly" for infinitely many to be rational.
Dehn twist of the axes
dotted red lines are lifts of arcs
is the axis of
Where did I go wrong? Reciprocal geodesics
is conjugate to it's own inverse by
the associated (simple) closed geodesic is invariant by the elliptic involution
any closed simple geodesic on the punctured torus is invariant
under the elliptic involution
so I picked a lifts of a sequence of simple closed geodesics
with traces that all went through the fixed point of ie .
Where did I go wrong? Reciprocal geodesics
any closed simple geodesic on the punctured torus is invariant
under the elliptic involution
so I picked a lifts of a sequence of simple closed geodesics
with traces that all went through the fixed point of .
These converge to the axis of and the intersection points give me the result
Unfortunately not
Dehn twist of the axes
dotted red lines are lifts of arcs
is the axis of
is the axis of
limit of the is intersection of with black semicircle
black semicircle projects to a leaf of the lamination containing
= "limit" of the Dehn twists
THE END
#
<!-1- _transition: cube -1->
- slides : google **greg mcshane github**
- click on **serfest**
- if there is a bug in my slides blame [this guy](https://github.com/yhatt)