1. Introduction
We give a section by section summary of the contents of this paper.
In §1 we define the Schwarz–Christoffel conformal map (2) of the complex plane less onto a quadrilateral Q, which is formed by reflecting a rational triangle in the real axis.
In §2, following Aurell and Itzykson [
1] we associate to the map
the affine Riemann surface
in
defined by
, where
has coordinates
and
. Thinking of
as a branched covering
with branch points at
,
and
∞ corresponding to the branch values 0, 1, and
∞, respectively, we show that
has genus
, where
for
. Let
be the set of nonsingular points of
. The map
is a holomorphic
n-fold covering map with covering group the cyclic group generated by
In §3 we build a model
of the affine Riemann surface
. The quadrilateral
Q is holomorphically diffeomorphic to a fundamental domain
of the action of the covering group on
. Rotating
Q by
gives a regular stellated
n-gon
, which is invariant under the dihedral group
G generated by the mappings
R and
. We study the group theoretic properties of
. We show that
is invariant under the reflection
in the ray
for
. To construct the model
of the affine Riemann surface
from the regular stellated
n-gon
we follow Richens and Berry [
2]. We identify two nonadjacent closed edges of
, the closure of
, if one edge is obtained from the other by a reflection
for some
. The identification space
, where
is the center of
, is a complex manifold except at points corresponding to
or a vertex of
, where it has a conical singularity. The action of
G on
induces a free and proper action on the identification space
, whose orbit space
is a complex manifold with compact closure in
, with genus
. Moreover
is holomorphically diffeomorphic to the affine Riemann surface
.
In §4, we construct an affine model of the Riemann surface . In other words, we find a discrete subgroup of the 2-dimensional Euclidean group , which acts freely and properly on such that after forming an identification space the orbit space is holomorphically diffeomorphic to . We now describe the group . Reflect the regular stellated n-gon in its edges, and then in the edges of the reflected regular stellated n-gons, et cetera. We obtain a group generated by translations so that sends the center of to the center of a repeatedly reflected reflected n-gon. The set is the union of the image under of a vertex of and its center for every . Let be the semi-direct product . The fundamental domain of the action on is less its vertices and center. Identifying equivalent open edges of and then taking G orbits, it follows that the affine model of the affine Riemann surface is the orbit space .
In §5 we show that the mapping
straightens the nowhere vanishing holomorphic vector field
X (11) on
, that is,
for every
. We pull back the flat metric
on
by
to the metric
on
. So
is a local developing map. Since
is the geodesic vector field on
, it follows that
X is a holomorphic geodesic vector field on
.
In §6 we study the geometry of the developing map . The dihedral group generated by and is a group of isometries of . The group G generated by R and is a group of isometries of . Extend the holomorphic map to a holomorphic map map by requiring that on . This works since is a fundamental domain of the action of the covering group on , which implies . Thus, the local holomorphic diffeomorphism intertwines the action on with the G action on and intertwines the local geodesic flow of the holomorphic geodesic vector field X with the local geodesic flow of the holomorphic vector field .
Following Richens and Berry [
2] we impose the condition: when a geodesic, starting at a point in
, meets
it undergoes a reflection in the edge of
that it meets. Such geodesics never meet a vertex of
. Thus, this type of geodesic becomes a billiard motion in
, which is defined for all time. Billiard motions in polygons have been extensively studied. For a nice overview see Berger ([
3], chpt. XI) and references therein. An argument shows that
invariant geodesics on
correspond under the map
to billiard motions on
.
Repeatedly reflecting a billiard motion in an edge of
and suitable edges of suitable
translations of
gives a straight line motion
on
. The image of the segment of a billiard motion, where
intersects
, in the orbit space
, is a geodesic. Here we use the flat Riemannian metric
on
, which is induced by the
invariant Euclidean metric
on
restricted to
. Consequently,
is an affine analogue of the affine Riemann surface
thought of as the orbit space of a discrete subgroup of
acting on
with the Poincaré metric, see Weyl [
4].
3. The Geometry of an Affine Riemann Surface
Let
and
be coordinate functions on
. Consider the affine Riemann surface
in
, associated to the holomorphic mapping
, defined by
see Aurell and Itzykson [
1]. We determine the singular points of
by solving
For , we have if and only if or . Thus, the set of singular points of is . So the affine Riemann surface is a complex submanifold of . Actually, , for if and , then either or .
Lemma 1. Topologically is a compact Riemann surface of genus less three points: , , and . Here for .
Proof. Consider the (projective) Riemann surface
specified by the condition
if and only if
Thinking of
G as a polynomial in
with coefficients which are polynomials in
and
, we may view
as the branched covering
When
we get the affine branched covering
From (
3) it follows that
, where
for
is an
nth root of unity with and
is the complex
nth root used in the definition of the mapping
(
1). Thus, the branched covering mapping
(
6) has
n “sheets” except at its branch points. Since
and
it follows that
and
are branch points of the mapping
of degree
and
, since
for
, see McKean and Moll ([
5], p. 39). Because
∞ is a branch point of the mapping
of degree
, where
. Hence the ramification index of 0, 1,
∞ is
,
, and
, respectively. Thus, the map
has
fewer sheets at 0,
fewer at 1, and
fewer at
∞ than an
n-fold covering of
. Thus, the total ramification index
r of the mapping
is
. By the Riemann–Hurwitz formula, the genus
g of
is
. In other words,
Consequently, the affine Riemann surface
is the compact connected surface
less the point at
∞, namely,
. So
is the compact connected surface
less three points:
,
, and
. □
Examples of
, , ; . So , , . Hence . So .
, , ; . So . Hence . So .
Table 1 below lists all the partitions
of
n, which give a low genus Riemann surface
Corollary 1. If n is an odd prime number and is a partition of n into three parts, then the genus of is .
Proof. Because n is prime, we get . Using the formula we obtain . □
Corollary 2. The singular points of the Riemann surface are , , and if then also .
Proof. A point
if and only if
, that is,
and
We need only check the points
,
and
. Since the first two points are singular points of
, they are singular points of
. Thus, we need to see if
is a singular point of
. Substituting
into the right hand side of (
10b) we get
Thus,
is a singular point of
only if
. □
Lemma 2. The mappingis a surjective holomorphic local diffeomorphism. Proof. Let
and let
By hypothesis
implies that
. The vector
is defined and is nonzero. From
and
, it follows that
. Using the definition of
(
12) and the definition of the mapping
(
7), we see that the tangent of the mapping
(
11) at
is given by
Since and are nonzero vectors, they form a complex basis for and , respectively. Thus, the complex linear mapping is an isomorphism. Hence is a local holomorphic diffeomorphism. □
Corollary 3. (11) is a surjective holomorphic n to 1 covering map. Proof. We only need to show that
is a covering map. First we note that every fiber of
is a finite set with
n elements, since for each fixed
we have
. Here
for
, is an
root of 1 and
is the complex
root used in the definition of the Schwarz–Christoffel map
(
2). Hence the map
is a proper surjective holomorphic submersion, because each fiber is compact. Thus, the mapping
is a presentation of a locally trivial fiber bundle with fiber consisting of
n distinct points. In other words, the map
is a
n to 1 covering mapping. □
Consider the group
of linear transformations of
generated by
Clearly
, the identity element of
and
. For each
we have
So
. Thus, we have an action of
on the affine Riemann surface
given by
Since is finite, and hence is compact, the action is proper. For every we have and . So maps into itself. At the isotropy group is , that is, the -action on is free. Thus, the orbit space is a complex manifold.
Corollary 4. Consider the holomorphic mappingwhere is the -orbit of in . The principal bundle presented by the mapping ρ is isomorphic to the bundle presented by the mapping (11). Proof. We use invariant theory to determine the orbit space
. The algebra of polynomials on
, which are invariant under the
-action
, is generated by
. Since
, these polynomials are subject to the relation
Equation (
15) defines the orbit space
as a complex subvariety of
. This subvariety is homeomorphic to
, because it is the graph of the function
. Consequently, the orbit space
is holomorphically diffeomorphic to
.
It remains to show that
is the group of covering transformations of the bundle presented by the mapping
(
11). For each
look at the fiber
. If
, then
, since
or
and
. Thus,
. So
. Hence
. Thus,
is a covering transformation for the bundle presented by the mapping
. So
is a subgroup of the group of covering transformations. These groups are equal because
acts transitively on each fiber of the mapping
. □
4. Another Model for
We construct another model
for the smooth part
of the affine Riemann surface
(
3) as follows. Let
be a fundamental domain for the
action
(
14) on
. So
if and only if for
we have
. Here
is the
root used in the definition of the mapping
(
2). The domain
is a connected subset of
with nonempty interior. Its image under the map
(
11) is
. Thus,
is one “sheet” of the covering map
. So
is one to one.
Using the extended Schwarz–Christoffel mapping
(
2), we give a more geometric description of the fundamental domain
. Consider the mapping
where the map
is given by Equation (
11). The map
is a holomorphic diffeomorphism of
onto
, which sends
homeomorphically onto
. Look at
, which is a closed quadrilateral with vertices
O,
D,
C, and
. The set
contains the open edges
,
, and
but
not the open edge
of
, see
Figure 3 above.
Let
be the region in
formed by repeatedly rotating
through an angle
. Here
R is the rotation
. We say that the quadrilateral
forms, see
Figure 4 above.
Theorem 1. The connected set is a regular stellated n-gon with its vertices omitted, which is formed from the quadrilateral , see Figure 5. Proof. By construction the quadrilateral
is contained in the quadrilateral
. Note that
. Thus,
So . Thus, is the regular stellated n-gon less its vertices, one of whose open sides is the diagonal of . □
We would like to extend the mapping
(
16) to a mapping of
onto
. Let
where
is the
action defined in Equation (
14). So we have a mapping
defined by
. The mapping
is defined on
, because
, since
is a fundamental domain for the
-action
(
14) on
. Because
, the mapping
is surjective. Hence
is holomorphic, since it is continuous on
and is holomorphic on the dense open subset
of
. Let
and let
G be the group generated by the rotation
R and the reflection
U subject to the relations
and
. Shorthand
. Then
. The group
G is the dihedral group
. The closure
of
in
is invariant under
, the subgroup of
G generated by the rotation
R. Because the quadrilateral
Q is invariant under the reflection
, and
, it follows that
is invariant under the reflection
U. So
is invariant under the group
G.
We now look at some group theoretic properties of .
Lemma 3. If F is a closed edge of the polygon and for some , then .
Proof. Suppose that . Then for some and some , . Let and suppose that F is an edge of such that , where . Then , but . So . Now suppose that . Then . So . Hence . Finally, suppose that with . Then . So . □
Lemma 4. For put . Then is a reflection in the closed ray . The ray is the closure of the side of the quadrilateral in Figure 5. Proof. fixes every point on the closed ray
, because
Since , it follows that is a reflection in the closed ray . □
Corollary 5. For every and every let . Here , because . Then is a reflection in the closed ray .
Proof. This follows because
and
fixes every point on the closed ray
, for
□
Corollary 6. For every , every with , and every , we have for a unique .
Proof. We compute. For every
we have
and
where
. Since
R and
U generate the group
G, the corollary follows. □
Corollary 7. For let be the group generated by the reflections for . Then is a normal subgroup of G.
Proof. Clearly
is a subgroup of
G. From Equations (
18) and (
19) it follows that
for every
and every
, since
G is generated by
R and
U. However,
is generated by the reflections
for
, that is, every
may be written as
, where for
we have
. So
for every
, that is,
is a normal subgroup of
G. □
As a first step in constructing the model
of
from the regular stellated
n-gon
we look at certain pairs of edges of
. For each
we say two distinct closed edges
E and
of
are
adjacent if and only if they intersect at a vertex of
. For
let
be the set of unordered pairs of
equivalent closed edges
E and
of
, that is, the edges
E and
are not adjacent and
for some generator
of
. Recall that for
x and
y in some set, the unordered pair
is precisely one of the ordered pairs
or
. Note that
is the set of all unordered pairs of nonadjacent edges of
. Geometrically, two nonadjacent closed edges
and
E of
are equivalent if and only if
is obtained from
E by reflection in the line
for some
and some
. In
Figure 6, where
, parallel edges of
, which are labeled with the same letter, are
-equivalent. This is no coincidence.
Lemma 5. Let be formed from the quadrilateral , where T is the isosceles rational triangle less its vertices. Then nonadjacent edges of are -equivalent if and only if they are parallel, see Figure 7. Proof. In
Figure 7, let
be the triangle
T with
,
, and
. Let
be the quadrilateral formed by reflecting the triangle
in its edge
. The quadrilateral
reflected it its edge
is the quadrilateral
. Let
be perpendicular to
and
be perpendicular to
, see
Figure 7. Then
is a straight line if and only if
. By construction
and
. So
if and only if
. Hence the edges
and
are parallel if and only if the triangle
is isosceles. □
Theorem 2. Let be the regular stellated n-gon formed from the rational quadrilateral with for . The G orbit space formed by first identifiying equivalent edges of the regular stellated n-gon formed from Q less O and then acting on the identification space by the group G is , which is a smooth 2-sphere with g handles, where , less some points corresponding to the image of the vertices of .
Example 1. Before we begin proving Theorem 2 we consider the following special case. Let be a regular stellated hexagon formed by repeatedly rotating the quadrilateral by R through an angle , see Figure 6. Let
be the group generated by the reflections
for
. Here
is the reflection which leaves the closed ray
fixed. Define an equivalence relation on
by saying that two points
x and
y in
are
equivalent,
, if and only if 1)
x and
y lie on
with
x on the closed edge
E and
for some reflection
or 2) if
x and
y lie in the interior of
and
. Let
be the space of equivalence classes and let
be the identification map which sends a point
to the equivalence class
, which contains
p. Give
the topology induced from
. Placing the quotient topology on
turns it into a connected topological manifold without boundary, whose closure is compact. Let
be
less its vertices. The identification space
is a connected 2-dimensional smooth manifold without boundary.
Let
. The usual
G-action
preserves equivalent edges of
and is free on
. Hence it induces a
G action on
, which is free and proper. Thus, its orbit map
is surjective, smooth, and open. The orbit space
is a connected 2-dimensional smooth manifold. The identification space
has the orientation induced from an orientation of
, which comes from
. So
has a complex structure, since each element of
G is a conformal mapping of
into itself.
Our aim is to specify the topology of
. The regular stellated hexagon
less the origin has a triangulation
made up of 12 open triangles
and
for
; 24 open edges
,
,
, and
for
; and 12 vertices
and
for
, see
Figure 6.
Consider the set
of unordered pairs of equivalent closed edges of
, that is,
is the set
for
, where
E is a closed edge of
.
Table 2 lists the elements of
.
G acts on
, namely,
, for
. Since
is the group generated by the reflections
,
, it is a normal subgroup of
G. Hence the action of
G on
restricts to an action of
on
and the
G action permutes
-orbits in
. Thus, the set of
-orbits in
is
G-invariant.
We now look at the
-orbits on
. We compute the
-orbit of
as follows. We have
Since
and
the
orbit
of
is
. Here
, since
. Similarly, the
-orbit
of
is
. Since
, we have found all
-orbits on
. The
G-orbit of
is
for
, since
; while the
G-orbit of
is
,
for
, since
.
Suppose that B is an end point of the closed edge E of . Then E lies in a unique of . Let be the -orbit of . Then is an end point of the closed edge of for every . So the -orbit of the vertex B. It follows from the classification of -orbits on that and are -orbits of the vertices of , which are permuted by the action of G on . Furthermore, and are G-orbits of vertices of , which are end points of the G-orbit of the rays and , respectively.
To determine the topology of the
G orbit space
we find a triangulation of
. Note that the triangulation
of
, illustrated in
Figure 6, is
G-invariant. Its image under the identification map
is a
G-invariant triangulation
of
. After identification of equivalent edges, each vertex
, each open edge
, having
as an end point, or each open edge
, where
is a pair of equivalent edges of
, and each open triangle
in
lies in a unique
G orbit. It follows that
,
or
, and
is a vertex, an open edge, and an open triangle, respectively, of a triangulation
of
. The triangulation
has 4 vertices, corresponding to the
G orbits
,
,
, and
; 18 open edges corresponding to
,
, and
for
; and 12 open triangles
and
for
. Thus, the Euler characteristic
of
is
. Since
is a 2-dimensional smooth real manifold,
, where
g is the genus of
. Hence
. So
is a smooth 2-sphere with 2 handles, less a finite number of points, which lies in a compact topological space
, that is its closure, see
Figure 8. This completes the example.
Proof of Theorem 2. We now begin the construction of
by identifying equivalent edges of
. For each
let
be an unordered pair of equivalent closed edges of
. We say that
x and
y in
are
equivalent,
, if 1)
x and
y lie in
with
and
for some
and some
or 2)
x and
y lie in
and
. The relation ∼ is an equivalence relation on
. Let
be the set of equivalence classes and let
be the map which sends
p to the equivalence class
, that contains
p. Compare this argument with that of Richens and Berry [
2]. Give
the topology induced from
and put the quotient topology on
. □
Theorem 3. Let be less its vertices. Then is a smooth manifold. Furthermore, is a topological manifold.
Proof. To show that is a smooth manifold, let be an open edge of . For let be a disk in with center at , which does not contain a vertex of . Set . For each let be an open edge of , which is equivalent to via the reflection , that is, is an unordered pair of equivalent edges. Let and set . Then is an open neighborhood of in , which is a smooth 2-disk, since the identification mapping is the identity on . It follows that is a smooth 2-dimensional manifold without boundary.
We now handle the vertices of
. Let
be a vertex of
and set
, where
is a disk in
with center at the vertex
. The map
with
and
is a homeomorphism, which sends the wedge with angle
to the wedge with angle
. The latter wedge is formed by the closed edges
and
of
, which are adjacent at the vertex
such that
with the edge
being swept out through
during its rotation to the edge
. Because
is a rational regular stellated
n-gon, the value of
s is a rational number for each vertex of
. For each
let
be an edge of
, which is equivalent to
and set
. Then
is a vertex of
, which is the center of the disk
. Set
. Then
is a disk in
. The map
, where
and
, is a homeomorphism of
D into a neighborhood
of
in
. Consequently, the identification space
is a topological manifold. □
We now describe a triangulation of
. Let
be the open rational triangle
with vertex at the origin
O, longest side
on the real axis, and interior angles
,
, and
. Let
be the quadrilateral
. Then
is a subset of the quadrilateral
, see
Figure 5. Moreover
. The
triangles
and
with
form a triangulation
of
with
vertices
and
for
;
open edges
,
,
, and
for
; and
open triangles
,
with
. The image of the triangulation
under the identification map
(
21) is a triangulation
of the identification space
.
The action of G on preserves the set of unordered pairs of equivalent edges of for each . Hence G induces an action on , which is proper, since G is finite. The G action is free on and thus on by Lemma A2. We have proved
Lemma 6. The G-orbit space is a compact connected topological manifold with being a smooth manifold. LetThen σ is the G orbit map, which is surjective, continuous, and open. The restriction of σ to has image and is a smooth open mapping. We now determine the topology of the orbit space
. For each
and
let
be an end point of a closed edge
E of
, which lies on the unordered pair
. Then
is an end point of the edge
of the unordered pair
of
. See
Appendix A for the definition of the group
. The sets
with
are permuted by
G. The action of
G on
preserves the set of open edges of the triangulation
. There are
-orbits:
;
, since
; and
, since
for
. So the image of the triangulation
under the continuous open map
is a triangulation
of the
G-orbit space
with
vertices
, where
and
;
open edges
,
, and
for
; and
open triangles
and
for
. Thus, the Euler characteristic
of
is
. However,
is a smooth manifold. So
, where
g is the genus of
. Hence
. Compare this argument with that of Weyl ([
4], p. 174). This proves Theorem 2.
Since the quadrilateral Q is a fundamental domain for the action of G on , the G orbit map restricted to Q is a bijective continuous open mapping. However, is a bijective continous open mapping of the fundamental domain of the action on . Consequently, the orbit space is homeomorphic to the G orbit space . The mapping is holomorphic except possibly at 0 and the vertices of . So the mapping is a holomorphic diffeomorphism.
5. An Affine Model of
We construct an affine model of the affine Riemann surface as follows. Return to the regular stellated n-gon , which is formed from the quadrilateral less its vertices. Repeatedly reflecting in the edges of and then in the edges of the resulting reflections of et cetera, we obtain a covering of by certain translations of . Here is the union of the translates of the vertices of and its center O. Let be the group generated by these translations. The semidirect product acts freely, properly and transitively on . It preserves equivalent edges of and it acts freely and properly on , the space formed by identifying equivalent edges in . The orbit space is holomorphically diffeomorphic to and is the desired affine model of . We now justify these assertions.
First we determine the group of translations.
Lemma 7. Each of the sides of the regular stellated n-gon is perpendicular to exactly one of the directionsfor . Proof. From
Figure 9 we have
. So
. Hence the line
, containing the edge
of
, is perpendicular to the direction
. Since
is the reflection of
in the line segment
, the line
, containing the edge
of
, is perpendicular to the direction
. Because the regular stellated
n-gon
is formed by repeatedly rotating the quadrilateral
through an angle
, we find that Equation (
23) holds. □
Since
, it follows that
is the distance from the center
O of
to the line
containing the side
, or to the line
containing the side
. So
is the closest point
H on
to
O and
is the closest point
on
to
O. Since the regular stellated
n-gon
is formed by repeatedly rotating the quadrilateral
through an angle
, the point
lies on the line
, which contains the edge
of
; while
lies on the line
, which contains the edge
of
for every
. Furthermore, the line segments
and
are perpendicular to the line
and
, respectively, for
.
Corollary 8. For we have Proof. We compute. From (
24) it follows that
while from (
25) we get
□
Corollary 9. For k, we have Proof. If
, then
, by definition. So
If
, then
, by definition. So
□
For
let
be the translation
Corollary 10. For k, we have Proof. For every
, we have
□
Reflecting the regular stellated n-gon in its edge contained in gives a congruent regular stellated n-gon with the center O of becoming the center of .
Lemma 8. The collection of all the centers of the regular stellated n-gons, formed by reflecting in its edges and then reflecting in the edges of the reflected regular stellated n-gons et cetera, iswhere for we have Proof. For each the center of the regular stellated congruent n-gon formed by reflecting in an edge of contained in the line is . Repeating the reflecting process in each edge of gives congruent regular stellated n-gons with center at , where . Repeating this construction proves the lemma. □
The set
of vertices of the regular stellated
n-gon
is
see
Figure 5. Clearly the set
is
G invariant.
Corollary 11. The setis the collection of vertices and centers of the congruent regular stellated n-gons , , . Proof. This follows immediately from Lemma 8. □
Corollary 12. The union of , where , , and , covers , that is, Proof. This follows immediately from . □
Let
be the abelian subgroup of the 2-dimensional Euclidean group
generated by the translations
(
28) for
. It follows from Corollary 12 that the regular stellated
n-gon
with its vertices and center removed is the fundamental domain for the action of the abelian group
on
. The group
is isomorphic to the abelian subgroup
of
generated by
.
Next we define the group
and show that it acts freely, properly, and transitively on
. Consider the group
, which is the semidirect product of the dihedral group
G, generated by the rotation
R through
and the reflection
U subject to the relations
and
, and the abelian group
. An element
of
is the affine linear map
Multiplication in
is defined by
which is the composition of the affine linear map
followed by
. The mappings
and
are injective, which allows us to identify the groups
G and
with their image in
. Using (
31) we may write an element
of
as
. So
For every
we have
that is,
Hence
which is just Equation (
29). The group
acts on
as
does, namely, by affine linear orthogonal mappings. Denote this action by
Lemma 9. The set (30) is invariant under the action. Proof. Let
. Then for some
and some
where
. For
with
and
we have
where
. If
, then
while if
, then
Here see Corollary 8. So , which implies , as desired. □
Lemma 10. The action of on is free.
Proof. Suppose that for some
and some
we have
. Then
v lies in some
. So for some
we have
where
for some
. Thus,
This implies , that is, . So , that is, . Hence , which is the identity element of . □
Lemma 11. The action of (and hence ) on is transitive.
Proof. Let
and
lie in
Since and , it follows that . □
The action of on is proper because is a discrete subgroup of with no accumulation points.
We now define an edge of and what it means for an unordered pair of edges to be equivalent. We show that the group acts freely and properly on the identification space of equivalent edges.
Let
E be an open edge of
. Since
, it follows that
is an open edge of
. Let
Then
is the set of open edges of
by 12. Since
is the center of
, the element
of
is a rotation-reflection of
, which sends an edge of
to another edge of
. Thus,
sends
into itself. For
let
be the set of unordered pairs
of equivalent open edges of
, that is,
, so the open edges
and
of
are not adjacent, which implies that the open edges
E and
of
are not adjacent, and for some generator
of the group
of reflections with
we have
Let
. So
is the set of unordered pairs of equivalent edges of
. Define an action * of
on
by
where
.
Define a relation ∼ on
as follows. We say that
x and
are related,
, if 1)
and
such that
, where
with
for some
and
for some
, or 2)
x,
and
. Then ∼ is an equivalence relation on
. Let
be the set of equivalence classes and let
be the map
which assigns to every
the equivalence class
containing
p.
Lemma 12. is the map ρ (20). Proof. This follows immediately from the definition of the maps and . □
Lemma 13. The usual action of on , restricted to , is compatible with the equivalence relation ∼, that is, if x, and , then for every .
Proof. Suppose that
, where
. Then
, since
. So for some
with
, we have
. Let
. Then
So
. However,
and
. Hence
. □
Because of Lemma 13, the usual -action on induces an action of on .
Lemma 14. The action of on is free and proper.
Proof. The following argument shows that it is free. Using Lemma A2 we see that an element of , which lies in the isotropy group for , interchanges the edge F with the equivalent edge and thus fixes the equivalence class for every . Hence the action on is free. It is proper because is a discrete subgroup of the Euclidean group with no accumulation points. □
Theorem 4. The -orbit space is holomorphically diffeomorphic to the G-orbit space .
Proof. The claim follows because the fundamental domain of the -action on is is the fundamental domain of the G-action on . Thus, is a fundamental domain of the -action on , which is equal to = by Lemma 12. Hence the -orbit space is equal to the G-orbit space . So the identity map from to induces a holomorphic diffeomorphism of orbit spaces. □
Because the group is a discrete subgroup of the 2-dimensional Euclidean group , the Riemann surface is an affine model of the affine Riemann surface .
6. The Developing Map and Geodesics
In this section, we show that the mapping
straightens the holomorphic vector field
X (
12) on the fundamental domain
, see [
6] and Flaschka [
7]. We also verify that
X is the geodesic vector field for a flat Riemannian metric
on
.
First we rewrite Equation (
13) as
From the definition of the mapping
(
2) we get
where we use the same complex
nth root as in the definition of
. This implies
For each
using (
36) and (
37) we get
So the holomorphic vector field
X (
12) on
and the holomorphic vector field
on
Q are
-related. Hence
sends an integral curve of the vector field
X starting at
onto an integral curve of the vector field
starting at
. Since an integral curve of
is a horizontal line segment in
Q, we have proved
Theorem 5. The holomorphic mapping δ (35) straightens the holomorphic vector field X (12) on the fundamental domain . We can say more. Let
and
. Then
is the flat Euclidean metric on
. Its restriction
to
is invariant under the group
, which is a subgroup of the Euclidean group
.
Consider the flat Riemannian metric
on
Q, where
is the metric (
38) on
. Pulling back
by the mapping
(
2) gives a metric
on
. Pulling the metric
back by the projection mapping
gives
on
. Restricting
to the affine Riemann surface
gives
.
Lemma 15. Γ is a flat Riemannian metric on .
Proof. We compute. For every
we have
Thus, is a Riemannian metric on . It is flat by construction. □
Because
has nonempty interior and the map
(
35) is holomorphic, it can be analytically continued to the map
since
. By construction
. So the mapping
is an isometry of
onto
. In particular, the map
is an isometry of
onto
. Moreover,
is a local holomorphic diffeomorphism, because for every
, the complex linear mapping
is an isomorphism, since it sends
to
. Thus,
is a developing map in the sense of differential geometry, see Spivak ([
8], p. 97) note on §12 of Gauss [
9]. The map
is local because the integral curves of
on
Q are only defined for a finite time, since they are horizontal line segments in
Q. Thus, the integral curves of
X (
12) on
are defined for a finite time. Since the integral curves of
are geodesics on
, the image of a local integral curve of
under the local inverse of the mapping
is a local integral curve of
X. This latter local integral curve is a geodesic on
, since
is an isometry. Thus, we have proved
Theorem 6. The holomorphic vector field X (12) on the fundamental domain is the geodesic vector field for the flat Riemannian metric on . Corollary 13. The holomorphic vector field X on the affine Riemann surface is the geodesic vector field for the flat Riemannian metric Γ on .
Proof. The corollary follows by analytic continuation from the conclusion of Theorem 6, since is a nonempty open subset of and both the vector field X and the Riemannian metric are holomorphic on . □
7. Discrete Symmetries and Billiard Motions
Let
be the group of homeomorphisms of the affine Riemann surface
(
3) generated by the mappings
Clearly, the relations
hold. For every
we have
So the additional relation holds. Thus, is isomorphic to the dihedral group.
Lemma 16. is a group of isometries of .
Proof. For every
we get
and
□
Recall that the group
G, generated by the linear mappings
is isomorphic to the dihedral group.
Lemma 17. G is a group of isometries of .
Proof. This follows because R and U are Euclidean motions. □
We would like the developing map
(
39) to intertwine the actions of
and
G and the geodesic flows on
and
. There are several difficulties. The first is: the group
G does
not preserve the quadrilateral
Q. To overcome this difficulty we extend the mapping
(
39) to the mapping
(
17) of the affine Riemann surface
onto the regular stellated
n-gon
.
Lemma 18. The mapping (17) intertwines the action Φ (14) of on with the actionof G on the regular stellated n-gon . Proof. From the definition of the mapping
we see that for each
we have
for every
. By analytic continuation we see that the preceding equation holds for every
. Since
by construction and
(
11), from the definition of the mapping
(
35) we get
for every
. In other words,
for every
. By analytic continuation we see that the preceding equation holds for all
. Hence on
we have
The mapping sends the generators and of the group to the generators R and U of the group G, respectively. So it is an isomorphism. □
There is a second more serious difficulty: the integral curves of run off the quadrilateral Q in finite time. We fix this by requiring that when an integral curve reaches a point P on the boundary of Q, which is not a vertex, it undergoes a specular reflection at P. (If the integral curve reaches a vertex of Q in forward or backward time, then the motion ends). This motion can be continued as a straight line motion, which extends the motion on the original segment in Q.
To make this precise, we give
Q the orientation induced from
and suppose that the incoming (and hence outgoing) straight line motion has the same orientation as
. If the incoming motion makes an angle
with respect to the inward pointing normal
N to
at
P, then the outgoing motion makes an angle
with the normal
N, see Richens and Berry [
2]. Specifically, if the incoming motion to
P is an integral curve of
, then the outgoing motion, after reflection at
P, is an integral curve of
. Thus, the outward motion makes a turn of
at
P towards the interior of
Q, see
Figure 10 (left). In
Figure 10 (right) the incoming motion has the opposite orientation from
. This extended motion on
Q is called a billiard motion. A billiard motion starting in the interior of
is defined for
all time and remains in
less its vertices, since each of the segments of the billiard motion is a straight line parallel to an edge of
and does not hit a vertex of
, see
Figure 11.
We can do more. If we apply a reflection
S in the edge of
Q in its boundary
, which contains the reflection point
P, to the initial reflected motion at
P, see
Figure 12.
The motion in
when it reaches
, et cetera, the extended motion becomes a billiard motion in the regular stellated
n-gon
, see
Figure 11. So we have verified
Theorem 7. A billiard motion in the regular stellated n-gon , which starts at a point in the interior of and does not hit a vertex of , is invariant under the action of the isometry subgroup of the isometry group G of generated by the rotation R.
Let be the subgroup of generated by the rotation . We now show
Lemma 19. The holomorphic vector field X (12) on is -invariant. Proof. We compute. For every
and for
we have
Hence for every
we get
for every
. In other words, the vector field
X is invariant under the action of
on
. □
Corollary 14. For every we have Proof. Equation (
43) is a rewrite of Equation (
42). □
Corollary 15. Every geodesic on is -invariant.
Proof. This follows immediately from the lemma. □
Lemma 20. For every and every we have Proof. From Equation (
41) we get
on
. Differentiating the preceding equation and then evaluating the result at
gives
for all
. When
, by definition
. So for every
Thus,
for every
. By analytic continuation (
45) holds for every
. Now
sends
to
. Since
for every
, it follows that
is in
. Furthermore, since
sends
to
, we get
For every
we obtain
that is, Equation (
44) holds with
. A similar calculation shows that Equation (
46) holds with
replaces by
. This verifies Equation (
44). □
We now show
Theorem 8. The image of a invariant geodesic on under the developing map (17) is a billiard motion in , see Figure 13. Proof. Because
and
are isometries of
and
, respectively, it follows from equation (
41) that the surjective map
(
17) is an isometry. Hence
is a local developing map. Using the local inverse of
and Equation (
44), it follows that a billiard motion in
is mapped onto a geodesic in
, which is possibly broken at the points
. Here
are the points where the billiard motion undergoes a reflection. However, the geodesic on
is smooth at
since the geodesic vector field
X is holomorphic on
. Thus, the image of the geodesic under the developing map
is a billiard motion. □
Theorem 9. Under the restriction of the mappingto the image of a billiard motion is a smooth geodesic on , where . Proof. Since the Riemannian metric on is invariant under the group of Euclidean motions, the Riemannian metric on is -invariant. Hence is invariant under the reflection for . So pieces together to give a Riemannian metric on the identification space . In other words, the pull back of under the map , which identifies equivalent edges of , is the metric . Since intertwines the G-action on with the G-action on , the metric is -invariant. It is flat because the metric is flat. So induces a flat Riemannian metric on the orbit space . Since the billiard motion is a -invariant broken geodesic on , it gives rise to a continuous broken geodesic on , which is -invariant. Thus, is a piecewise smooth geodesic on the smooth G-orbit space .
We need only show that
is smooth. To see this we argue as follows. Let
be a closed segment of a billiard motion
, that does not meet a vertex of
. Then
s is a horizontal straight line motion in
. Suppose that
is the edge of
, perpendicular to the direction
, which is first met by
s and let
be the meeting point. Let
be the reflection in
. The continuation of the motion
s at
is the horizontal line
in
. Recall that
is the translation of
by
. Using a suitable sequence of reflections in the edges of a suitable
each followed by a rotation
R and then a translation in
corresponding to their origins, we extend
s to a smooth straight line
in
, see
Figure 14. The line
is a geodesic in
, which in
has image
under the
-orbit map
(
47) that is a smooth geodesic on
. The geodesic
starts at
. Thus, the smooth geodesic
and the geodesic
are equal. In other words,
is a smooth geodesic. □
Thus, the affine orbit space with flat Riemannian metric is the affine analogue of the Poincaré model of the affine Riemann surface as an orbit space of a discrete subgroup of acting on the unit disk in with the Poincaré metric.