Kyoto 2

The geometry and topology of arithmetic hyperbolic
3-manifolds
Alan W. Reid∗
May 12, 2007
1 Introduction
This paper is based on three lectures given by the author at the RIMS Symposium, “Topology, Com-
plex Analysis and Arithmetic of Hyperbolic Spaces” held at the Research Institute for Mathematical
Sciences, Kyoto University, in December 2006. The goal of the lectures was to describe recent work
on understanding topological and geometric properties of arithmetic hyperbolic 3-manifolds, and
the connections with number theory. This is the theme of the paper.
Our discussion of topological aspects of arithmetic hyperbolic 3-orbifolds is motivated by the
following central conjectures from 3-manifold topology:
Conjecture 1.1. Let M be a closed hyperbolic 3-manifold, then M is virtually Haken; ie M has a
finite sheeted cover N which contains embedded incompressible surface (necessarily of genus at least
2).
Conjecture 1.2. Let M be a closed hyperbolic 3-manifold, then M has a finite sheeted cover N for
which b1 (N ) > 0 (where b1 (N ) is the first Betti number of N ).
We can also rephrase Conjecture 1.2 to say that vb1 (M) > 0 where vb1 (M) is defined by
vb1 (M) = sup{b1 (N) : N is a finite cover of M},

and is called the virtual first Betti number of M .
Conjecture 1.3. Let M be a closed hyperbolic 3-manifold, then vb1 (M) = ∞.
Conjecture 1.4. Let M be a closed hyperbolic 3-manifold, then π1 (M ) is large; that is to say, some
finite index subgroup of π1 (M ) admits a surjective homomorphism onto a non-abelian free group.
Now it is clear that Conjecture 1.4 implies Conjecture 1.3 implies Conjecture 1.2, and standard
3-manifold topology shows that Conjecture 1.2 implies Conjecture 1.1. Our interest here is in recent
work towards reversing these implications.
Our geometric discussion is centered around the set of lengths of closed geodesics, as well as the
set of geodesics themselves, and in particular on how these force a certain rigidity on commensura-
bility classes. For example, the length spectrum L(M ) of a hyperbolic 3-manifold M is the set of all
lengths of closed geodesics on M counted with multiplicities. A question that has attracted some
attention is whether hyperbolic 3-manifolds with the same length spectra are commensurable. We
discuss this and other related questions in this paper.
∗ supported in part by the NSF
1
Given these preliminary remarks, the paper is organized as follows. In §2 we recall the definition
of arithmetic Fuchsian and Kleinian groups, arithmetic hyperbolic 2 and 3-orbifolds together with
some other terminology and notation that we will use. In §3, we discuss a geometric characterization
of arithmetic hyperbolic 3-manifolds which will be used in §4 to show that for arithmetic hyperbolic
3-manifolds, Conjecture 1.2 implies Conjecture 1.4. In §4, we also discuss how certain conjectures
from automorphic forms implies Conjectures 1.1–1.4. This is well-known to the experts, and our
purpose is simply to sketch some of the ideas involved. In the opposite direction, in §6 we discuss how
widely distributed arithmetic rational homology 3-spheres are. We motivate this in §5 by discussing
a similar question in dimension 2 that offers useful comparisons for dimension 3. Finally, in §7 we
discuss the geometric properties of arithmetic hyperbolic 2 and 3-orbifolds.
Acknowledgements: The author would like to thank Michihiko Fuji for organizing the symposium
and his hospitality during the symposium. He also wishes to thank his various co-authors of the
results discussed here, particularly D. D. Long. In addition, the author would like to thank F.
Voloch for a helpful conversation, and he is very grateful to F. Calegari for helpful discussion and
correspondence regarding the proof of Theorem 4.14.
2 Preliminaries
For more details on the topics covered in this section see [29] for number theoretic background and
[39] for details on quaternion algebras and arithmetic groups.
2.1
By a number field k we mean a finite extension of Q. The ring of integers of k will be denoted
Rk . A place ν of k will be one of the canonical absolute values of k. The finite places of k
correspond bijectively to the prime ideals of Rk . An infinite place of k is either real, corresponding
to an embedding of k into R, or complex, corresponding to a pair of distinct complex conjugate
embeddings of k into C. We denote by kν the completion of of k at a place ν. When ν is an infinite
place, kν is isomorphic to R or C depending on whether ν is real or complex.
If A is an ideal of Rk , the norm of A is the cardinality of the quotient ring Rk /A and will be
denoted NA.
2.2
Let k be a field of characteristic different from 2. The standard notation
for a quaternion algebra
over k is the following. Let a and b be non-zero elements of k, then a,b k (known as the Hilbert
Symbol) denotes the quaternion algebra over k with basis {1, i, j, ij} subject to i2 = a, j 2 = b and
ij = −ji.
Let k be a number field, and ν a place of k. If B is a quaternion algebra defined over k, the
classification of quaternion algebras Bν = B ⊗k kν over the local fields kν is quite simple. If ν is
complex then Bν is isomorphic to M (2, kν ) over kν . Otherwise there is, up to isomorphism over kν ,
a unique quaternion division algebra over kν , and Bν is isomorphic over kν to either this division
algebra or to M (2, kν ).
Let B be a quaternion algebra over the number field k. B is ramified at a place ν of k if Bν
is a division algebra. Otherwise we say B is unramified at ν. We shall denote the set of places
(resp. finite
Q places) at which B is ramified by Ram B (resp. Ramf B). The discriminant of B is the
Rk -ideal ν∈Ramf B Pν where Pν is the prime ideal associated to the place ν.
2
We summarize for convenience the classification theorem for quaternion algebras over number
fields (see [39] Chapter 7).
Theorem 2.1. • The set Ram B is finite, of even cardinality and contains no complex places.
• Conversely, suppose S is a finite set of places of k which has even cardinality and which
contains no complex places. Then there is a quaternion algebra B over k with Ram B = S,
and this B is unique up to isomorphism over k.
• B is a division algebra of quaternions if and only if Ram B 6= ∅. t
u
2.3
We next recall the definition of arithmetic Fuchsian and Kleinian groups.
Let k be a totally real number field, and let B be a quaternion algebra defined over k which is
ramified at all infinite places except one. Let ρ : B → M (2, R) be an embedding, O be an order of
B, and O1 the elements of norm one in O. Then Pρ(O1 ) < PSL(2, R) is a finite co-area Fuchsian
group, which is co-compact if and only if B is not isomorphic to M (2, Q). A Fuchsian group Γ is
defined to be arithmetic if and only if Γ is commensurable with some such Pρ(O1 ).
Notation: Let B/k be as above and O be an order of B. We will denote the group Pρ(O1 ) by Γ1O .
Arithmetic Kleinian groups are obtained in a similar way. In this case we let k be a number field
having exactly one complex place, and B a quaternion algebra over k which is ramified at all real
places of k. As above, if O is an order of B and ρ : O1 ,→ SL(2, C), then Γ1O is a Kleinian group of
finite co-volume. An arithmetic Kleinian group Γ is a subgroup of PSL(2, C) commensurable with
a group of the type Γ1O . An arithmetic Kleinian group is cocompact if and only if the quaternion
algebra B as above is a division algebra.
In both the Fuchsian and Kleinian cases, the isomorphism class of the quaternion algebra B/k
determines a wide commensurability class of groups in PSL(2, R) and PSL(2, C) respectively (see
[39] Chapter 8). By Theorem 2.1 the isomorphism classes of such quaternion algebras will be
completely determined by the finite set of places of k at which B is ramified.
A hyperbolic orbifold H2 /Γ or H3 /Γ will be called arithmetic if Γ is an arithmetic Fuchsian or
Kleinian group.
Recall that if Γ is a Kleinian group, the invariant trace-field of Γ is the field
kΓ = Q({tr2 γ : γ ∈ Γ})
and the invariant quaternion algebra
AΓ = {Σaj γj : aj ∈ kΓ, γj ∈ Γ(2) }.
As discussed in [39] these are invariants of commensurability. When Γ is arithmetic, the field k
and algebra B coincide with kΓ and AΓ
Arithmetic Fuchsian or Kleinian groups form a small but interesting subclass of Fuchsian or
Kleinian groups. For example it is known that there are only finitely many conjugacy classes of
arithmetic Kleinian groups whose volume is bounded above by some constant K. A similar statement
holds for arithmetic Fuchsian groups and this also shows that there are only finitely many conjugacy
classes of arithmetic Fuchsian groups of a fixed signature (see [39] Chapter 11).
Notation: We shall call a group G is derived from a quaternion algebra if G is a subgroup of some
Γ1O , regardless of whether G has finite index in Γ1O .
3
2.4
We set up some notation and collect some information on 3-orbifolds that we will make use of.
Let Q be a compact orientable 3-orbifold. We shall denote by sing(O) its singular locus, and |Q|
the underlying 3-manifold. Let sing0 (Q) and sing− (Q) denote the components of the singular locus
with, respectively, zero and negative Euler characteristic. For any prime p, let singp (Q) denote the
union of the arcs and circles in sing(Q) with singularity order that is a multiple of p. Let sing0p (Q)
and sing−p (Q) denote those components of singp (Q) with zero and negative Euler characteristic.
When Q = H3 /Γ is a closed orientable hyperbolic 3-orbifold the nature of the singular locus Σ
is completely understood. Σ is a link or graph in |Q| and the local groups are either cyclic, dihedral,
A4 , S4 or A5 . Note the dihedral group of order 4 is Z/2Z × Z/2Z. The cases of the dihedral groups,
A4 , S4 and A5 arise when there are vertices in the singular locus.
It can be shown that one can always pass to a finite cover that has no vertices but does have
non-empty singular locus (cf. the proof of Proposition 4.4 of [28]). On the other hand, it is shown
(and used in a crucial way) in [28] that for certain orbifolds with vertices in the singular locus, then
the number of vertices can be increased without bound in passage to finite sheeted covers.
2.5
Let Γ be a non-cocompact Kleinian (resp. Fuchsian) group acting on H3 (resp. H2 ) with finite
co-volume. Let U(Γ) denote the subgroup of Γ generated by parabolic elements of Γ. Note that
U(Γ) is visibly a normal subgroup of Γ, and we may define:
V (Γ) = (Γ/U(Γ))ab ⊗Z Q.
Setting r(Γ) = dimQ (V (Γ)), then it follows from standard arguments that r(Γ) denotes the
dimension of the space of non-peripheral homology.

1 1 1 0
Example 1: Since PSL(2, Z) = P < , >, it follows that r(PSL(2, Z)) = 0.
0 1 1 1
More generally if Γ is a non-cocompact Fuchsian group such that H2 /Γ has underlying space of
genus g, then r(Γ) = 2g.
Example 2: If L is a link in S 3 with S 3 \ L = H3 /Γ then r(Γ) = 0, since Γ is generated by

meridians.
2.6
Let p be a prime, and let Fp denote the field of order p. If X is a group, space or orbifold, let dp (X)
be the dimension of H1 (X; Fp ). With this we make the following definition that will be useful in
what follows.
Definition: Let X be a group, space or orbifold and let p be a prime. Then a collection {Xi } of
finite index subgroups or finite-sheeted covers of X with index or degree [X : Xi ] is said to have
linear growth of mod p homology if
dp (Xi )
inf i > 0.
[X : Xi ]
Much of the discussion in §4 is motivated by the extra structure that having non-empty singular
locus provides. An indication of this is the following result proved in [26] (see also [28] in the case
of p = 2 for a proof that does not use the Golod-Shafarevich inequality).
4
Theorem 2.2. Let Q be a compact orientable 3-orbifold with non-empty singular locus and a finite-
volume hyperbolic structure. Let p be a prime that divides the order of an element of π1orb (Q). Then
Q has a tower of finite-sheeted covers {Qi } that has linear growth of mod p homology.
3 A geometric characterization of arithmeticity

A fundamental dichotomy of Margulis [40] (which holds much more generally) asserts that a Kleinian
group Γ of finite co-volume is arithmetic if and only if the commensurator
Comm(Γ) = {x ∈ PSL(2, C) : xΓx−1 is commensurable with Γ}
is dense in PSL(2, C). Moreover, when Γ is non-arithmetic the commensurator is also a Kleinian
group of finite co-volume, which is the unique maximal element in the commensurability class of Γ.
In this section we interpret this geometrically. In particular we will prove:
Theorem 3.1. Let M = H3 /Γ be an orientable finite volume hyperbolic 3-manifold. Then M is
arithmetic if and only if the following condition holds:
Let γ ⊂ M be a closed geodesic. Then there exists a finite sheeted cover Mγ → M such that Mγ
admits an orientation-preserving involution τ such that the fixed point set of τ image of γ.
Proof: That this condition is satisfied when Γ is arithmetic is proved in [12] (and implicit in [28]).
We give the proof for completeness. We can assume without loss of generality that Γ is derived
from a quaternion algebra B/k. Let a ∈ Γ be a hyperbolic element whose axis Aa projects to γ.
Let b ∈ Γ be chosen so that its axis Ab is disjoint from Aa . Now the Lie product ab − ba defines
an involution τa,b for which the axis of rotation is the perpendicular bisector of Aa and Ab in H3 .
Denote this geodesic by δ. As shown in [28] (see Proposition 2.4), there is an order O of B for which
τa,b lies in the image in PSL(2, C) of the normalizer of O in B. This is an arithmetic Kleinian group
commensurable with Γ (see [28] or [39] Chapter 6). Hence there is a hyperbolic element g ∈ Γ whose
axis Ag is the geodesic δ.
It follows that Aa is now the perpendicular bisector of the axes Ag and aAg . Repeating the
argument of the previous paragraph provides an involution fixing Aa (namely arising from the Lie
product of the elements g and aga−1 ) and lies in an arithmetic Kleinian group ∆ commensurable
with Γ. To complete the proof, take the core of Γ ∩ ∆ in ∆, and let Mγ be the corresponding cover
of M
For the converse, assume the manifold M = H3 /Γ is non-arithmetic and satisfies the condition.
Then by the result of Margulis mentioned above, Comm(Γ) is the unique maximal element in the
commensurability class of Γ. We can assume without loss of generality that Γ = Comm(Γ) (note
that we now allow the existence of elements of finite order Γ = Comm(Γ)).
Let {δn } be a collection of hyperbolic elements in Γ that lie in distinct cyclic subgroups of Γ up
to conjugacy in Γ. We are assuming that the condition holds, and so for each n we can construct
an involution τn that normalizes a finite index subgroup of Γ and which has axis of rotation in H3
the axis of δn . Let ∆n be the subgroups constructed above which are normalized by τn . Since ∆n
is commensurable with Γ and Γ is the unique maximal element it follows that τn ∈ Γ for all n.
Now Γ contains only a finite number of cyclic groups of order 2 up to conjugacy. Hence infinitely
many of these are conjugate in Γ, and in particular there is a Γ conjugacy that takes the axis of
τn to that of τ1 (say). Hence there is a Γ conjugacy that takes the axis of γn to that of a fixed γ1 .
However, this contradicts how these elements were chosen. t u
A consequence of Theorem 3.1 is that every arithmetic Kleinian group is commensurable with one
containing an element of order 2. Indeed more is true, the following is shown in [28].
5
Theorem 3.2. Let Γ be an arithmetic Kleinian group. Then Γ is commensurable with a Kleinian
group Γ0 such that Γ0 contains a group isomorphic to Z/2Z × Z/2Z.
Remarks:(1) The Z/2Z × Z/2Z constructed in Theorem 3.2 is related to a Hilbert Symbol for
the invariant quaternion algebra. More precisely, if a,b
k is a Hilbert Symbol, then notice that
the projection of the group generated by i and j to PGL(2, C) ∼
= PSL(2, C) is simply a copy of
Z/2Z × Z/2Z.
(2) In §7 we mention (see the proof of Theorem 7.9) another geometric characterization of Comm(Γ).
In this case, it is the axes themselves that are used rather than the involutions.
4 Towards Conjectures 1.1–1.4 for arithmetic hyperbolic 3-

manifolds
Conjectures 1.1, 1.2, 1.3 and 1.4 still seem rather intractible for an arbitrary closed hyperbolic 3-
manifold. However, recent evidence points to the situation being somewhat better for those closed
hyperbolic 3-manifolds commensurable with a hyperbolic 3-orbifold with non-empty singular locus.
This is evident in recent work of M. Lackenby, [26] and [27] as well as in [28]. A philosophical
reason that this case may be more amenable to study is that it is known from [11] that if M is
a finite volume non-compact hyperbolic 3-manifold, then π1 (M ) is large, so the strongest of the
conjectures discussed in §1 holds in this setting. Having non-trivial singular locus seems to be an
aid that helps replace the existence of the cusp (which is the crucial thing in the methods [11]) in
the cusped setting. As shown in §3, arithmetic Kleinian groups are always commensurable with
groups containing non-trivial elements of finite order, and so fit into this picture. On the other
hand, “most” closed hyperbolic 3-manifolds are never commensurable with an orbifold as above.
4.1
Here we sketch the proof of the following theorem.
Theorem 4.1. Conjecture 1.2 implies Conjecture 1.4 for arithmetic hyperbolic 3-manifolds.
This theorem will follow immediately from Theorem 3.2 and the next two results proved in [12]
and [28] respectively. We can assume the manifold is closed by [11].
Theorem 4.2. Conjecture 1.2 implies Conjecture 1.3 for closed arithmetic hyperbolic 3-manifolds.
Theorem 4.3. Let Q = H3 /Γ be a 3-orbifold (with possibly empty singular locus) commensurable
with a closed orientable hyperbolic 3-orbifold that contains Z/2Z × Z/2Z in its orbifold fundamental
group. Suppose that vb1 (Q) ≥ 4. Then Γ is large.
We will now sketch some of the ideas in the proofs of these results.
Sketch proof of Theorem 4.2:
Let M be an arithmetic hyperbolic 3-manifold which is assumed to have positive first Betti number.
An important ingredient in the proof of Theorem 4.2 is the following general geometric result
proved in [12].
Theorem 4.4. Let N be a closed hyperbolic 3-manifold. Then there is a closed geodesic η with the
property that it has a non-right angle transverse intersection with every least area surface in N .
6
Given the geodesic η ⊂ M provided by the above theorem, Theorem 3.1 provides a finite cover
M
f of M such that M f admits an orientation-preserving involution τ such that the fixed point set of
τ contains a component of the preimage of η. The key point now is to show that b1 (M f) > b1 (M ).
Suppose to the contrary they have the same rank, so that p∗ : H2 (M ) −→ H2 (M ) is a rational
f
isomorphism. Pick a connected embedded surface F , whose homology class represents an eigenvalue
±1 for the action of τ∗ on H2 (Mf). Consider the class p∗ [F ] ∈ H2 (M ), this might not be primitive,
so take a least area embedded surface G in M representing the primitive class. Hence we may write
p∗ [F ] = a[G] for some integer a. The following two lemmas are proved in [12].
Lemma 4.5. Let p : M f −→ M be a finite sheeted covering and suppose that H2 (M
f) has the same
rank as H2 (M ).
Fix a connected embedded surface F in M representing some nonzero class in H2 (M ) and let
fk be the components of p−1 (F ).
F1 , . . . , F
f
Then for every i, j, [Fei ] = ±[F
fj ] in H2 (M
f).
Lemma 4.6. In the notation of Lemma 4.5, suppose in addition that F is least area for [F ] .
Then every Fei has the same area and this is least area for the class [Fei ].
Given these lemmas, the sketch of the proof is finished as follows. First, it can be argued that
the surface G is connected (otherwise π1 (M ) can be shown to be large). Now p−1 (G) consists
of components each of which is an embedded surface and therefore a primitive class in H2 (M f).
Furthermore, since p∗ is a rational isomorphism, [F ] = ±[G∗ ], where G∗ is any choice of a component
of p−1 (G). It follows we have that τ∗ [G∗ ] = ±[G∗ ]. Moreover, by Lemma 4.6, any such G∗ is a least
area surface in the homology class [G∗ ] = ±[F ].
By choice of η, there is at least one component of the preimage of G which has a nonright
angle transverse intersection with ηe. Make this choice for G∗ . However, this surface can be used to
violate a result of Hass on intersections of least area surfaces [21] to obtain a contradiction. Briefly,
the surfaces G∗ and τ (G∗ ) are homologous up to orientation, least area, yet they meet without
coinciding for angle reasons using the involution. Repeated application of this increases the first
Betti number without bound. t u
Remarks:(1) An interesting feature of this argument is that although it uses arithmetic in an

essential way, it is geometric, in that arithmeticity is used to produce involutions.
(2) Theorem 4.2 was proved in the case of congruence arithmetic hyperbolic 3-manifolds by Borel
in [2]. This method of proof was recently generalized to give different proofs of Theorem 4.2 in [1]
and [53].
Sketch proof of Theorem 4.3:
Key to this proof is work of Lackenby, [25] and [26]. In particular the following result of Lackenby
gives a method of proving largeness of an arbitrary finitely presented group (which is a consequence
of a stronger result in [25]).
Proposition 4.7. Let G be a finitely presented group, and let φ : G → Z be a surjective homomor-
phism. Let Gi = φ−1 (iZ), and suppose that, for some prime p, {Gi } has linear growth of mod p
homology. Then G is large.
The crucial hypothesis in Proposition 4.7 is the linear growth of mod p homology. In the context
of orbifolds, as noted in §2, Theorem 2.2 provides a tower with linear growth in mod p homology
for some prime p. However, Proposition 4.7 uses a very specific tower. This can also be established,
for in [28] it is shown that:
7
Proposition 4.8. Let Q be a compact orientable 3-orbifold, and let C be a component of sing0p (Q)
for some prime p. Let pi : |Qi | → |Q| (i ∈ N) be distinct finite sheeted covering spaces of |Q| such
that the restriction of pi to each component of p−1
i (C) is a homeomorphism onto C. Let Qi be the
corresponding covering spaces of Q. Then {Qi } has linear growth of mod p homology.
Proof. A standard 3-manifold argument (see [28] Proposition 3.1) shows that
dp (Qi ) ≥ |sing0p (Qi )|.
By assumption p−1
i (C) is a homeomorphism onto C and so it follows that
|sing0p (Qi )| ≥ [Qi : Q].

Hence inf i dp (Qi )/[Q : Qi ] ≥ 1 as required. t
u
It quickly follows from Proposition 4.7 and Proposition 4.8 that we have.
Corollary 4.9. Suppose Q be a compact orientable 3-orbifold with non-empty singular locus con-
taining a circle component. Assume that that b1 (Q) ≥ 2, then π1orb (Q) is large.
The proof of Theorem 4.3 is now completed as follows. By hypothesis, Q has a finite cover
Q0 such that b1 (Q0 ) ≥ 4. Let Q00 be the hyperbolic orbifold, commensurable with Q, containing
Z/2Z × Z/2Z in its fundamental group. Now, Q0 and Q00 are commensurable, and hence they have
a common cover Q000 , say. Since Q000 is hyperbolic, it has a manifold cover M . We may assume that
M regularly covers O00 . Now, b1 does not decrease under finite covers, and so b1 (M ) ≥ 4. Since
M → Q00 is a regular cover, it has a group of covering transformations G
Now, π1orb (Q00 ) contains Z/2Z × Z/2Z, and hence some singular point of Q00 has local group
that contains Z/2Z × Z/2Z. It follows that G contains Z/2Z × Z/2Z. Let h1 and h2 be the
commuting covering transformations of M corresponding to the generators of Z/2Z × Z/2Z. These
are involutions. Let h3 be the composition of h1 and h2 , which also is an involution. For i = 1, 2
and 3, let Qi be the quotient M/hi . Since hi has non-empty fixed point set, sing(Oi ) is a non-empty
collection of simple closed curves with order 2.
It is shown in [28] that for at least one i ∈ {1, 2, 3}, b1 (Qi ) ≥ 2. So, Corollary 4.9 now shows
π1orb (Qi ) is large for some i, and hence so is Γ. t
u
4.2
Here we summarize what us known to us at this time about when Conjecture 1.4 holds for arithmetic
hyperbolic 3-manifolds (using Theorem 4.2). We only consider closed manifolds, since as mentioned
above, the finite volume non-compact case is completely understood by [11].
Throughout the discussion M = H3 /Γ is an arithmetic hyperbolic 3-manifold with invariant
quaternion algebra B/k. Note that any proper subfield of k is totally real.
k has a real subfield of index 2: We distinguish two cases; when there are totally geodesic
surfaces and when there are not.
In general the existence of a totally geodesic surface in a closed hyperbolic 3-manifold allows
one to deduce that Conjecture 1.4 holds (see [32] and [37]). In the case when M is an arithmetic
hyperbolic 3-manifold, the existence of a totally geodesic surface forces conditions
on B/k. The
a,b
number field k satisfies [k : k ∩ R] = 2 and B has a Hilbert symbol of the form k , where a and
b are non-zero elements of k ∩ R (see [39] Chapter 9.5).
8
If one considers the case when only [k : k ∩ R] = 2, forgetting the condition on B, it is also
known that vb1 (Γ) > 0. This is proved by various authors ([24], [31] and [38]).
In some of the cases when [k : k ∩ R] = 2 but no totally geodesic surface is present, there is also
geometric/topological proofs that vb1 (Γ) > 0. This happens when there is an orientation-reversing
involution on some manifold in the commensurability class of H3 /Γ.
Other conditions on subfields: The paper [24] can be applied in some other cases as we now
describe.
Suppose ` ⊂ k is a proper subfield and there is a tower of intermediate fields:
` = `0 ⊂ `1 ⊂ . . . ⊂ `m−1 ⊂ `m = k
where each extension `i /`i−1 is either a cyclic extension of prime degree or a non-Galois cubic
extension. Then vb1 (Γ) > 0.
In a similar spirit it is shown in [45] that if k is contained inside a solvable Galois extension of
`, then vb1 (Γ) > 0. In particular a corollary of these results is the following.
Corollary 4.10. Suppose [k : Q] ≤ 4, then Γ is large.
Proof: Any such extension has solvable Galois group over Q. t
u
Clozel’s result: Perhaps the most general method known at present for proving Conjecture 1.2
for arithmetic hyperbolic 3-manifolds is [9]. In particular this applies to every field k with one
complex place, but there is a condition on the quaternion algebra. We state Clozel’s result here in
the notation of this paper. Some topological consequences are discussed below.
Theorem 4.11. Let Γ be an arithmetic Kleinian group as above. Assume that for every place
ν ∈ Ramf B, kν contains no quadratic extension of Qp where p is a rational prime and ν|p. Then
vb1 (Γ) > 0.
4.3
There are situations when Theorem 4.11 can be used together with the existence of certain finite
subgroups in arithmetic Kleinian groups to prove Conjecture 1.4. For example, we have the following
from [28].
Theorem 4.12. Let Γ be an arithmetic Kleinian group commensurable with a Kleinian group con-
taining A4 , S4 or A5 or a finite dihedral group that is derived from a quaternion algebra. Then Γ is
large.
The proof of this follows from Theorem 4.11 on noticing that the existence of these finite sub-
groups places conditions on the invariant trace-field and quaternion algebras of the arithmetic
Kleinian groups (indeed this is true arithmeticor not).
For example in the case of A4 , S4 or A5 the
−1,−1
invariant quaternion algebra is isomorphic to k , and this can be shown to be unramified at
all finite places.
The conditions forced on the algebra are reminiscent of how the existence of a totally geodesic
surface in arithmetic hyperbolic 3-manifold places conditions on the invariant trace-field and quater-
nion algebra (recall §4.2)
Remark: The discussions above, together with the discussion in §4.2, motivate the following. This
provides a uniform setting for how the existence of a totally geodesic surface or a finite group of the
9
form A4 , S4 or A5 has on the invariant quaternion algebra, and is perhaps more approachable than
the general problem.
Challenge: Let k be a field with one complex place, and ` a (necessarily) totally real subfield. Sup-
pose Γ is an arithmetic Kleinian group with invariant trace-field k and invariant quaternion algebra
B that satisfies:
B∼ = A ⊗` k
for some quaternion algebra A over ` (ie. B lies in the image of the natural map of Brauer groups
Br(`) → Br(k)). Show that Γ is large.
Note also, that removing the hypothesis in Theorem 4.12 that the dihedral subgroup is derived
from a quaternion algebra would prove that all arithmetic Kleinian groups are large. It would be
interesting to give a geometric proof of Theorem 4.12. Thus we pose as a warm-up challenge to the
general situation, and which should be more amenable.
Conjecture 4.13. Let Γ be a Kleinian group of finite co-volume containing A4 , S4 or A5 . Then Γ
is large.
Another interesting topological application of Clozel’s result is that it applies to show that if Σ
is an arithmetic integral homology 3-sphere then vb1 (Σ) > 0, and hence π1 (Σ) is large. For in this
case, the invariant trace-field of Σ has even degree over Q, and the invariant quaternion algebra of
Σ is unramified at all finite places (see for example [39] Theorem 6.4.3). So from the perspective of
Conjectures 1.1–1.4, in the arithmetic setting, integral homology spheres are easier to handle than
rational homology 3-spheres.
We close this subsection with two examples on the limitation of current techniques regarding the
arithmetic methods mentioned here.
Example 1: This is taken from [28], and is a commensurability class of arithmetic 3-orbifolds for
which none of the methods discussed applies to provide a cover with positive first Betti number.
We do this in the first possible degree, namely 5 (see Corollary 4.10).
Let p(x) = x5 − x3 − 2x2 + 1. Then p has three real roots and one pair of complex conjugate
roots. Let t be a complex root and let k = Q(t). Now k has one complex place and its Galois group
is S5 . There is a unique prime P of norm 112 in k. It follows that kP is a quadratic extension of
Q11 . Take B ramified at the real embeddings and the prime P. Then it is unknown whether any
arithmetic Kleinian group arising from B has a cover with positive first Betti number.
Briefly, if Γ is any group in the commensurability class, then since k has odd degree, the discussion
in §4.2 shows that there are no non-elementary Fuchsian subgroups. The result of Clozel does not
apply by the condition on P, and none of the other work discussed above applies since [k : Q] = 5
and the Galois group is S5 . For this final part, note that if k were contained in a solvable extension
L of Q, then the Galois closure K of k would be a subfield of L. However, this implies that the
Galois group Gal(K/k) is a quotient of Gal(L/k), which is solvable, and this is a contradiction.
One can also adjoin other primes to Ramf B that constructs quaternion algebras for which Γ1O
are torsion free.
In the next example, we construct a family of arithmetic orbifolds Qp = H3 /Γp (p a prime congruent
to 1 mod 4) where Γp contains elements of order p and for which vb1 (Γp ) is unknown to be positive.
Note that the invariant trace-fields of the orbifolds Qp have degree going to infinity (compare with
Corollary 5.6).
Example 2: Let p be a prime congruent to 1 mod 4, and let `p = Q(cos π/p). Note that `p contains
10
√
the quadratic extension Q( p).
Claim:(1) There exists a number field kp with one complex place such that [kp : `p ] = 5 and
Gal(kp /`p ) ∼
= S5 .
(2) There exist infinitely many rational primes q such that if ν is a place of kp with ν|q then kpν
contains a quadratic extension of Qq .
(3) Let ζp denote a primitive p-th root of unity. Then, the primes ν constructed in (2) do not split
in kp (ζp )/kp .
Given these three conditions we can construct orbifolds as follows. Let p and kp be as in (1) above,
let Bp /kp be a quaternion algebra ramified at all real places, and let ν be a place of kp as in (2).
Since kp has even degree we insist that Bp is ramified at another place ν 0 which we also assume is
as in (2). Since ν and ν 0 do not split in kp (ζp )/kp , it follows that kp (ζp ) embeds in Bp (see [39]
7.3.3 for example) and indeed, for any maximal order Op ⊂ Bp the group Γ1Op contains an element
of order p (see for example [39] Theorem 12.5.4). Let Qp = H3 /Γ1Op .
As in Example 1 above, it is unknown using the methods described in this paper that vb1 (Qp ) >
0. Briefly, although [kp : Q] has even degree there are no totally geodesic surfaces, since `p is the
maximal real subfield of kp which is of index 5. For if there were totally geodesic surfaces, then kp
would contain a totally real subfield of index 2 (see §4.3) which would contain `p and this is false.
As in Example 1, condition (1) above ensures that kp is not contained in any solvable extension of
`p , and the condition on the prime ν ensures that Clozel’s condition fails.
Proof of Claim:
To prove (1), note that given the existence of a field kp with [kp : `p ] = 5 and with one complex
place, then Gal(kp /`p ) ∼= S5 . This follows for example using the arguments of §3 of [8]. The
existence of such a field can be argued as follows. We seek to build a monic irreducible polynomial
p(x) = x5 + a1 x4 + a2 x3 + a3 x2 + a4 x + a5 (ai ∈ `p ), such that p(x) has 1 pair of complex conjugate
roots and 3 real roots, and furthermore, for every Galois embedding σ : `p → R the polynomial
p−1
pσ (x) obtained by applying σ to the coefficients of p has only real roots. Now `p embeds in R 2
5(p−1)
as a dense subset, so that `5p embeds in R 2 as a dense subset. Hence a polynomial p as above
5(p−1)
gives a vector in R 2 and since the condition on the roots is an open condition it follows that
we can find a a polynomial satisfying the condition on the roots. Moreover, by a similar argument,
the polynomial can be assumed irreducible.
To establish (2) and (3) we argue as follows. The Galois group of Q(ζp )/Q is cyclic of order
p − 1, so the Cebotarev Density Theorem gives infinitely many rational primes q such that q is
totally inert in Q(ζp ) (i.e. the inertial degree is precisely p − 1). As remarked above, `p contains
√ √
Q( p), and so for any prime q as above, if ω is the Q( p)-prime lying over q, and ν is a kp -prime
√ √
over q, then kν contains Q( p)ω . By properties of the inertial degree, [Q( p)ω : Qq ] = 2. This
proves (2).
Now (3) also follows, since choosing a prime q as in (2), and ν a kp -prime lying over q, then ν
cannot split in kp (ζp ). For if so, then it follows that for the unique (from the previous paragraph)
`p -prime µ with ν|µ, that µ must split in Q(ζp ), and this is false from above.
11
4.4
The methods of [9], [31], [24], and [31] are those of automorphic forms and automorphic represen-
tations (see also [48] for a survey of some of this). In particular, these methods involve applying
some type of “functoriality”; eg Base Change, or the Jacquet-Langlands correspondence (see more
on this below). Perhaps the biggest challenge from the geometric or topological perspective is to
understand geometric features of incompressible surfaces represented by duals to the cohomology
classes produced by these methods. In some cases, totally geodesic surfaces can be recognized in
the application of Base Change in quadratic extensison, but this is not always the case.
To complete the discussion about the connection between the topology of finite covers of arithmetic
hyperbolic 3-manifolds and the theory of automorphic forms, we should mention that Conjecture
1.2 is a consequence of various conjectures in automorphic forms and the Langlands Program. One
such conjecture is mentioned [9], here we briefly discuss another that relates to elliptic curves. Our
intention here is not to give a detailed exposition of all the details, but simply to indicate to a
topologist, the gist of the connection, and the powerful implications that the theory of automorphic
forms has in the setting of arithmetic groups. Because we will only give a brief discussion, we refer
the reader to [17] and [23] for more details regarding the theory of automorphic representations.
We emphasize that what we discuss below is well-known to the experts in the theory of auto-
morphic representations, our purpose is merely to advertise the connections.
We begin with Base Change for GL(2). Let k be a number field, Ak the adeles of k and denote
by A(k) the set of equivalence classes of automorphic representations of GL(2, Ak ). Let Ac denote
the set of equivalence classes of cuspidal automorphic representations of GL(2, Ak ).
Suppose that K/k is a finite extension of number fields, then part of the Langlands Program
predicts that if [π] ∈ A(k) then there is a canonically associated [π 0 ] ∈ A(K), called the the Base
Change lift of π. Furthermore if π is cuspidal, then “typically”, π 0 is cuspidal. The existence of this
Base Change lift has been established in very few cases, and are the source of the non-vanishing of
vb1 in the case of a field with one complex place having certain proper subfields as discussed in §4.3.
For our purposes, we make the following assumption:
Assumption BC: Let k be a field with one complex place, then Base Change lifts exist from A(Q)
to A(k).
Given this we sketch the proof of:

Theorem 4.14. Let M be an arithmetic hyperbolic 3-manifold. Under Assumption BC, vb1 (M) >
0, and hence π1 (M ) is large.
Sketch Proof: We can assume that M = H3 /Γ is closed. Denote the invariant trace-field of Γ
by k and B respectively. Assume that Ramf B = {ν1 , . . . , νs }. To show that vb1 (M) > 0, we shall
construct a congruence subgroup of some group Γ1O in the commensurability class of Γ and for which
the first Betti number is non-zero.
The connection with the cohomology is provided by Matsushima’s theorem [41] which says in
this setting, that if X = H3 /∆ is a closed orientable hyperbolic 3-manifold then b1 (X) can be
computed as the multiplicity of a certain irreducible unitary representation of SL(2, C) occuring in
the decomposition of L2 (SL(2, C)/∆) under the right regular representation. We shall denote this
by π0 . The goal is to show that the multiplicity of π0 can be arranged to be non-zero by passage to a
congruence subgroup. This is achieved using Assumption BC together with the Jacquet-Langlands
Correspondence, the main result of which we state below in a somewhat abbreviated form (see [17]
and [23]). To state this, we denote by BAk the adeles of the division algebra B (see [39] Chapter
∗
7 for example) and BA k
the invertible adeles. Note that irreducible unitary representations of
12
these adelic groups have the form ⊗πv (the completed restricted tensor product), where πv is an
irreducible unitary representation of the local group at the place ν.
Theorem 4.15. There is a correspondence which associates to each irreducible unitary representa-
tion π 0 of BAk (which is not 1-dimensional) an irreducible unitary representation π of GL(2, Ak ).
Indeed, if π = ⊗πv the correspondence π 0 → π is one-to-one onto the collection of cusp forms of
GL(2, Ak ) with πν a discrete series representation for each ν ∈ Ram B.
The goal is to produce a cuspidal automorphic representation π of GL(2, Ak ) such that at the
infinite places this representation is of the right type; i.e has the form π0 ⊗ C ⊗ . . . ⊗ C (the C
factors corresponding to the trivial representation of the invertible elements in the Hamiltonian
quaternions). To this end, let P1 , . . . , Ps denote the primes associated to the finite places in Ramf B,
t
and assume that NPj Q = qj = pjj .
Denoting by N = pj , let E be an elliptic curve defined over Q with j-invariant 1/N (see [50]
for details about elliptic curves). That such an elliptic curve exists can be seen explicitly from the
following construction on setting j = 1/N .
Let j ∈ Q, A = 27j/(1728 − j) and B = 2A. Then y 2 = x3 + Ax + B is an elliptic curve defined
over Q with j-invariant j. Now E/Q is modular, and the level can be shown to be N C for some
integer C. In particular, modularity implies that E is attached to a cusp form of weight 2 (which
is a newform) for Γ0 (N C), which in turn provides a cuspidal automorphic representation π with
[π] ∈ Ac (Q). We remark that the identification of this cuspidal automorphic representation from
the newform provides a cuspidal automorphic representation which is discrete series at the infinite
place of Q and at those primes dividing N C (see [17]).
Assumption BC provides a Base Change lift π 0 , which is cuspidal in this case because of the
j-invariant, which guarantees that the elliptic curve will have potentially multiplicative reduction
at primes dividing N over k. It can also be shown that, since at the infinite place of Q, π is discrete
series, then at the infinite places of k this representation has the right type. Furthermore, again
from the behavior of π, at the places in Ramf B, and k-primes lying over primes dividing C, the
representation is necessarily discrete series. Theorem 4.15 now provides a cuspidal automorphic
representation for which π0 occurs at the complex place. A more detailed analysis of the correspon-
dence actually shows that this provides a cusp form on a congruence subgroup of some Γ1O where
the level corresponds to k-primes lying over primes dividing C and k-primes lying over the primes
pj . t
u
4.5
We now discuss some other more general situations where the existence of orbifolds in a com-
mensurability class are helpful. Firstly, there are other situations where the existence of certain
2-dimensional orbifolds can be used to prove Conjecture 1.1, 1.2, 1.3 and 1.4; eg as above when
the 2-orbifold is totally geodesic by [32] (see also [38]). Here we indicate another class, which again
provides evidence that hyperbolic 3-manifolds commensurable with orbifolds do provide a more
reasonable class to work with.
Theorem 4.16. Let Q = H3 /Γ be a closed hyperbolic 3-orbifold for which π1 (|Q|) is infinite.
Assume that Q contains an essential 2-orbifold with underlying space the 2-sphere or the 2-torus.
Suppose that M is a hyperbolic 3-manifold commensurable with Q, then M is virtually Haken.
Proof: We can assume that Q is orientable, otherwise, by [34], Q has a finite sheeted cover which is
a non-orientable hyperbolic 3-manifold, and it is well known that such manifolds have positive first
Betti number (see [22] Chapter 6). Hence vb1 (M) > 0.
13
Letting N denote the normal closure of all the elliptic elements of Γ, then π1 (|Q|) = Γ/N . Note
that the solution to the Geometrization Conjecture ensures that π1 (|Q|) is residually finite. Let
φ : Γ → Γ/N be the quotient homomorphism.
We deal first with the case of the sphere. Thus suppose that S ,→ Q is an embedded essential
2-sphere with cone points. Since the orbifold group F , of S, is generated by elliptic elements it
follows that φ(F ) = 1.
Now assume that Γ0 is a torsion free subgroup of finite index in Γ. Then F0 = F ∩ Γ0 is a
surface group of genus at least 2. Restricting φ to F0 we have that φ(F0 ) = 1. Hence, since π1 (|Q|)
is infinite, and residually finite, we deduce that F is contained in infinitely many subgroups of finite
index in Γ, and hence F0 is contained in infinitely many subgroups of finite index in Γ0 . It is a
result of Jaco (see [49] Corollary 2.3) that H3 /Γ0 , and therefore M , is virtually Haken.
The argument in the case when S is the 2-torus with cone points is handled in a similar way.
However, here note that φ(F ) need not be trivial, but it is a quotient of Z × Z. Thus φ(F ) is an
Abelian subgroup of π1 (|Q|) of rank at most 2. This is either trivial or isomorphic to one of Z, Z2 ,
Z/nZ or Z × Z/2Z (see [22] Chapter 9). If φ(F ) is trivial we argue as above. Thus assume that
A = φ(F ) is one of the non-trival Abelian groups listed above. The case of A = Z ⊕ Z/2Z is ruled
out since the existence of such a subgroup forces |Q| to be non-orientable (see [22] Theorem 9.12)
contrary to the assumption.
Suppose first that π1 (|Q|) is irreducible. Note that A has infinite index in π1 (|Q|), and since
π1 (|Q|) is infinite, it follows from [22] Chapter 9 that A is isomorphic to Z or Z2 . In these cases
[20] shows that A is a separable subgroup of π1 (|Q|) (assuming the Geometrization conjecture). We
now argue as above using the result of Jaco.
If |Q| is not irreducible, then either |Q| = S 2 × S 1 or |Q| decomposes as a non-trivial connect
sum. In the former case we obtain a surjection onto Z, and we may deduce that M is virtually Haken
from this. In the latter case, the argument above will deal with all cases other than A = Z/nZ.
However, in this case, [22] Theorem 9.8 gives a decomposition π1 (|Q|) = π(X) ∗ G, where X is a
closed 3-manifold, |G| < ∞ and A is conjugate into G. Now π1 (|Q|) is infinite, so π1 (X) 6= 1. Hence
Γ surjects a non-trivial free product, and so this determines a non-trivial action of Γ on a tree. It
follows that π1 (M ) must contain a finite index subgroup that admits a non-trivial action on a tree,
and so M is virtually Haken. t u
We remark that it is shown in [28] that any arithmetic hyperbolic 3-orbifold always admits a finite
cover which is also an orbifold with non-trivial singular locus and satisfies the condition on the
fundamental group of the underlying space given in Theorem 4.16.
A well-known approach to proving Conjecture 1.1 is to show that a closed hyperbolic 3-manifold
has a finite sheeted cover for which the character variety (of (P)SL(2, C) representations) contains
a positive dimensional component. For a closed hyperbolic 3-manifold even constructing “new”
characters in a finite sheeted cover seems hard. We now address this issue exploiting orbifolds.
Note that if M = H3 /Γ with k = Q(tr Γ), then any Galois embedding σ of k defines a character
χσ obtained by applying σ to the character of a faithful discrete representation.
Definition: Let M = H3 /Γ be a closed orientable hyperbolic 3-manifold. Say that M has extra
characters if there exists an infinite irreducible (P)SL(2, C) representation of Γ that is different
from χσ for any Galois embedding of Q(tr Γ).
In addition we shall say that M virtually has extra characters if Γ contains a finite index subgroup
∆ such that H3 /∆ has extra characters.
Theorem 4.17. Let M = H3 /Γ be a closed orientable hyperbolic 3-manifold commensurable with
an orbifold commensurable that contains Z/2Z × Z/2Z in its orbifold fundamental group. Then
14
either vb1 (M) > 0 or M virtually has extra characters.
Proof: Theorem 8.1 of [28] shows that, under the hypothesis of Theorem 4.17, M is commensurable
with an orbifold Q for which π1 (|Q|) is infinite. It follows that Γ has a finite index subgroup Γ0 such
that Γ0 surjects π1 (X) where X is a closed orientable 3-manifold with π1 (X) infinite. Assuming
the Geometrization Conjecture, we can assume that X is hyperbolic, otherwise it is easily seen that
vb1 (X) > 0 and so vb1 (H3 /Γ0 ) > 0.
When X is hyperbolic, the faithful discrete representation of π1 (X) provides an extra character
(it is different from χσ because this representation is not faithful on Γ0 ). t u
Together with Theorem 4.1 we have.

Corollary 4.18. Let M be an arithmetic hyperbolic 3-manifold. Then either π1 (M ) is large or M
virtually has extra characters.
This also gives.
Corollary 4.19. Let M be an arithmetic hyperbolic 3-manifold. Then M has a finite sheeted cover
N for which the character variety of N contains at least one more component of characters of
irreducible representations.
Proof: By Corollary 4.18 either π1 (M ) is large, so that π1 (M ) contains a finite index subgroup
surjecting a free non-abelian group or there is a finite index subgroup with an extra character. Both
cases provide extra components in the (P)SL(2, C) character variety. t u
4.6
Another well-known conjecture in the same spirit as Conjecture 1.1–1.4 is the virtual fibering con-
jecture of Thurston.
Conjecture 4.20. Let M be a finite volume hyperbolic 3-manifold. Then M has a finite sheeted
cover that fibers over S 1 .
Unlike Conjectures 1.1–1.4 where there is a considerable amount of supporting evidence (even in
the arithmetic case), little is known about Conjecture 4.20. For the complements of knots and links
there is work of [4], [15], [30], and [57] that provides some evidence, but for closed manifolds almost
nothing is known. However, in the arithmetic setting, many well-known examples are known to be
virtually fibered (see [5] and [47]). Arithmeticity is used in that the pair (kΓ, AΓ) is a complete
invariant of the commensurability class, and so one can construct bundles (for example by surgery
on fibered knot complements) and just compare arithmetic data (see [5] and [47] for more details).
5 The 2-dimensional situation

In this section we address the following question. By saying a Fuchsian group Γ is of genus 0, we
mean that the underlying space of the orbifold H2 /Γ has genus 0.
Question 5.1. Are there infinitely many commensurability classes of arithmetic Fuchsian groups
of genus 0?
15
The motivation for this question is the situation in dimension 3 (discussed in §6). Note that if Γ
is a cocompact Fuchsian group of genus 0, then Γab is finite, and so the orbifold H2 /Γ is a “rational
homology 2-sphere”. Conjectures 1.2–1.4 predict that on passage to certain finite covers, arithmetic
hyperbolic 3-manifolds stop being rational homology 3-spheres. Thus it seems interesting to discuss
how “widely distributed” arithmetic rational homology 3-spheres are. This is address in §6 when
we discuss an analogue of Question 5.1 for arithmetic rational homology 3-spheres.
The answer to Question 5.1 is provided by the following result from [36].
Theorem 5.2. The answer to Question 5.1 is no.
In contrast, it is easy to construct infinitely many cocompact arithmetic Fuchsian groups of
genus 0 (see [36] Corollary 4.10). Before discussing the proof we recall some notation and discuss a
classical situation which Theorem 5.2 generalizes.
5.1
Recall that a congruence subgroup of PSL(2, Z) is any subgroup of PSL(2, Z) that contains a principal
congruence subgroup Γ(n) of level n for some n where
Γ(n) = P{A ∈ SL(2, Z) : A ≡ I mod n}.
More generally, a non-cocompact arithmetic Fuchsian group Γ is congruence if some conjugate of Γ

in PGL(2, R) contains some Γ(n). Any non-cocompact arithmetic Fuchsian group is conjugate into
a maximal group commensurable with PSL(2, Z), and these can be described as follows.
Any Eichler order of M (2, Q) (an interesection of two maximal orders) is conjugate to

a b
E0 (n) := { ∈ M2 (Z) | c ≡ 0 mod n}
c d
and the conjugacy classes of the maximal Fuchsian groups are the groups Γ+ E0 (n) for n square-free
(the images in PSL(2, R) of the normalizers of the Eichler orders E0 (n)). These groups have a more
familiar description as the normalizers in PSL(2, R) of the groups

a b
Γ0 (n) = P{ : c ≡ 0 mod n}.
c d
In particular these maximal arithmetic Fuchsian groups contain Γ(n), and so are congruence sub-
groups.
In the context of cusped 2-orbifolds, the analogue of a rational homology 2-sphere is when r(Q) =
0 (recall §2.5). Thus before studying the cocompact setting, an easier question is to study non-
cocompact arithmetic Fuchsian groups Γ with r(Γ) = 0 (there is after all only one commensurability
class).
Now it is well known that r(Γ0 (n)) = 0 if and only if n ∈ {1, . . . , 10, 12, 13, 16, 18, 25}, and the size
of r(Γ) carries important information regarding the theory of elliptic curves over Q. Furthermore it
is known when r(Γ+ E0 (n) ) = 0 and that this is also related to interesting number theoretic phenomena;
for example it is known that there are only finitely many n such that r(Γ+ E0 (n) ) = 0, and that these
values of n are related to the properties of the Monster simple group. Indeed, Ogg (cf. [43]) observed
that that those primes p for which the groups N (Γ0 (p)) have genus 0, coincides with those primes
p dividing the order of the Monster simple group (see [10], [36] and [43] for more details).
On the other hand, it is easy to construct infinitely many subgroups {Γi } of finite index in
PSL(2, Z) for which r(Γi ) = 0; e.g. by constructing certain cyclic covers of the the thrice-punctured
sphere.
16
5.2
The description of maximal arithmetic Fuchsian groups and congruence subgroups in terms of Eichler
orders can be extended to the cocompact setting which we briefly recall (see [36]).
Let B be a quaternion algebra over a totally real field k and let O be a maximal order of B. Let
I be an integral 2-sided O-ideal in B which means that I is a complete Rk -lattice in B such that
O = {x ∈ B | xI ⊂ I} = {x ∈ B | Ix ⊂ I}.
The principal congruence subgroup of Γ1O of level I, denoted Γ1O (I) is the image in PSL(2, R) of
O1 (I) = {α ∈ O1 : α − 1 ∈ I}.
Since O/I is a finite ring, O1 (I) has finite index in O1 .
Definition: Let Γ be an arithmetic Fuchsian group. Γ is called a congruence subgroup if there is

a quaternion algebra B defined over the totally real field k ramified at all real places except one, a
maximal order O and an ideal I of O as described above such that Γ contains Γ1O (I).
As in the case of the modular group, any maximal arithmetic Fuchsian group is a congruence
subgroup (see [36]). In addition to proving Theorem 5.2, the methods also provide a generalization
of the results when r(Γ) = 0 for Γ a maximal (or congruence) arithmetic Fuchsian group (see [36]).
Theorem 5.3. There are finitely many conjugacy classes of congruence arithmetic Fuchsian groups
of genus zero. There are thus finitely many conjugacy classes of maximal arithmetic Fuchsian groups
of genus 0.
5.3
The proof of Theorem 5.2 is completed in the following steps.
1. If a commensurability class contains an arithmetic Fuchsian group Γ of genus zero, then any
Fuchsian group containing Γ will also have genus zero, so we can assume that we have an infinite
sequence {Γi } of non-conjugate maximal arithmetic Fuchsian groups of genus 0.
Note that Area(H2 /Γi ) → ∞, since as noted in §2.3, there are only finitely many conjugacy
classes of arithmetic Fuchsian groups of bounded area.
2. We now exploit the following results that give contrasting behavior of the first non-zero eigenvalue
of the Laplacian. The first is an important result about the first non-zero eigenvalue of the Laplacian
for congruence arithmetic Fuchsian groups (see [18], [23] and [55]).
Theorem 5.4. If Γ is a congruence arithmetic Fuchsian group then
3
λ1 (Γ) ≥ ,
16
where λ1 (Γ) is the first non-zero eigenvalue of the Laplacian of Γ.
Note by the discussion in §5.2, this applies to the maximal arithmetic Fuchsian groups Γi . On
the other hand, we have the following result of P. Zograf [58].
Theorem 5.5. Let Γ be a Fuchsian group of finite co-area and let the genus of H2 /Γ be denoted by
g(Γ). If Area(H2 /Γ) ≥ 32π(g(Γ) + 1), then
8π(g(Γ) + 1)
λ1 (Γ) < .
Area(H2 /Γ)
17
3. We now have a contradiction, for with g(Γi ) = 0, and Area(H2 /Γi ) → ∞, Theorem 5.5 implies
that λ1 → 0. However this contradicts Theorem 5.4. t u
Corollary 5.6. There is an upper bound to the order of an elliptic element in an arithmetic Fuchsian
group of genus 0.
Proof: Since there are only finitely many commensurability classes of arithmetic Fuchsian groups
of genus 0, there is a bound on the degree of the defining fields k of these groups. If an arithmetic
Fuchsian group contains an element of order n ≥ 2, then k contains Q(cos 2π/n) and it is well-known
that the degree of this field goes to infinity with n. t
u
Computations suggest (see [36]) the following conjecture.

Conjecture 5.7. If Γ is an arithmetic Fuchsian group of genus 0, then the invariant trace-field has
degree at most 7.
Remarks:(1) The inequalities in Theorems 5.4 and 5.5 give an upper bound on the co-area of a
maximal (resp. congruence) arithmetic Fuchsian group of genus zero as:
128
Area(H2 /Γ) ≤ π
3
(2) The same arguments hold for arbitrary fixed g.
(3) Another feature of genus zero 2-orbifolds is that it is shown in [36] Theorem 1.5, that any Fuch-
sian group of genus 0 has a global upper bound to the length of the shortest translation length of a
hyperbolic element.
6 Arithmetic rational homology 3-spheres

As discussed in §4, the fact that arithmetic hyperbolic 3-manifolds are commensurable with a 3-
orbifold affords geometric and topological approaches to the Conjectures 1.1–1.4. The discussion of
the case of 2-dimensional rational homology spheres in §5 motivates some interesting questions in
dimension 3, that relate to Conjectures 1.1–1.4 but in the “opposite direction”.
The analogue of Question 5.1 is:
Question 6.1. Are there infinitely many commensurability classes of arithmetic rational homology
3-spheres (even allowing for orbifolds)?
Continuing with the analogy, another consequence of Theorem 5.2 is that there is an upper
bound on the degree of the invariant trace-field of an arithmetic Fuchsian group of genus 0 (from
the area bound described in §5.3). Hence we pose:
Question 6.2. Is there an upper bound to the degree of the invariant trace-field of an arithmetic
rational homology 3-sphere?
Note that a positive answer to Question 6.2 need not imply a negative answer to Question 6.1.
18
6.1
Here we provide a conjectural approach taken from [36], that provides a positive answer to Question
6.1.
As in the 2-dimensional case, we start with the non-cocompact setting which again is closely
related to the theory of elliptic curves over quadratic imaginary number fields. √
Let d be a square-free positive integer, and Od be the ring of integers in Q( −d). The ana-
logue of the modular group here is a family of groups, The Bianchi groups, PSL(2, Od ). In this
setting, The Cuspidal Cohomology Problem posed in the 1980’s asked which Bianchi groups have
r(P SL(2, Od )) = 0. In [56] it was shown that:
Theorem 6.3. r(PSL(2, Od )) = 0 if and only if
d ∈ D = {1, 2, 3, 5, 6, 7, 11, 15, 19, 23, 31, 39, 47, 71}.
As in dimension 2, there is a class of congruence subgroups which play a key role in connecting
with the arithmetic. For an ideal A ⊂ Od we define the congruence subgroup:

a b
Γ0 (A) = P{ : c ≡ 0 mod A}
c d
Set r(A) = r(Γ0 (A)). Unlike the situation in dimension 2, r(A) is far from understood at
present. On the other hand, as in dimension 2, it is easy to construct infinitely many examples of
non-cocompact arithmetic Kleinian groups Γ for which r(Γ) = 0; e.g. there are infinitely arithmetic
links in S 3 whose complements are arithmetic.
In the context of the groups Γ0 (A) work of Grunewald and Mennicke [19], and Cremona [14],
suggest the following:
Conjecture 6.4. For each d ∈ D, there exist infinitely many ideals A ⊂ Od such that r(A) = 0.
For the case of Z[i] = O1 , more precise statements can be made. For example, the following seems
reasonable.
Conjecture 6.5.
(a) There exist infinitely many pairs of prime ideals {P1 , P2 } ⊂ Z[i] such that r(P1 P2 ) = 0.
(b) In fact, let P =< 1 + i >, then there are infinitely many prime ideals Pj ⊂ Z[i] with N Pj =
1 mod 12 and r(PPj ) = 0.
As evidence of this, it is shown in [14] that of all integral ideals A ⊂ Z[i] with norm ≤ 500 only
76 give rise to groups with r(A) > 0. With reference to Conjecture 6.5(b) [14] shows that this holds
for prime ideals of norms 13, 37, 61, and 73, but fails for a prime of norm 97.
The relevance of these conjectures to the cocompact setting follows from the Jacquet-Langlands
correspondence (recall Theorem 4.15). A consequence of this is (cf. [6] Theorem 3.3).
√
Theorem 6.6. Let B be a division algebra of quaternions over Q( −d) with the set of primes
ramifying B being, {P1 , . . . , P2r }. Let O be a maximal order of B, and Γ = Γ1O . If Γab is infinite,
then r(P1 . . . P2r ) 6= 0.
A corollary of this result and Conjecture 6.5(b) is the following. We remark that the prime P
in Conjecture 6.5(b) is chosen so that the resultant groups Γ1O are torsion-free.
Corollary 6.7. If Conjecture 6.5(b) holds, then Question 6.1 has a positive answer.
19
We finish this subsection with some further discussion motivated from the 2-dimensional setting.
Corollary 5.6 provides an upper bound, say N, to the order of elliptic element in an arithmetic Fuch-
sian group of genus 0. Rephrasing this in a way that (although vacuous in dimension 2) is suggestive
for dimension 3, Corollary 5.6 shows that if an arithmetic Fuchsian group Γ is commensurable with
a group containing an element of order > N then Γ is large. This prompts in the spirit of §4 a
situation that may be more approachable than the general case—this is very much in the spirit of
some of the results in [27].
Show that for all sufficiently large integers N , any (arithmetic) Kleinian group commensurable with
one containing an element of order N is large.
Recall that if M = H3 /Γ is a closed orientable hyperbolic 3-manifold, the injectivity radius of M

(denoted by injrad(M )) is defined as the largest number r such that for all p ∈ M , the ball B(p, r)
centered at p of radius r is isometric to the r-ball in H3 . D. Cooper asked whether there is a K ∈ R,
independent of M , such that injrad(M ) ≤ K?
This question is of interest since, by residual finiteness, an affirmative answer to this question,
implies every closed hyperbolic 3-manifold has a cover with first betti number at least 1. However,
F. Calegari and N. Dunfield [6] answered this question in the negative; unlike the 2-dimensional case
discussed in §5.3.
In [6], the
√ authors construct arithmetic rational homology 3-spheres
√ arising from the division
algebra Q( −2) ramified at the two places lying above 3 in Q( −2) and with arbitrarily large
injectivity radii. The construction of [6] assumes the existence of certain Galois representations of
√ √
Gal(Q( −2)/Q( −2)) that although predicted by the general framework of the Langlands cor-
respondence are not yet known to exist. However an unconditional proof of the result of [6] was
established in [3]. It is worth remarking in light of Question 6.1 that the construction of [6] provides
commensurable rational homology 3-spheres. Also these examples show that there are infinitely
many arithmetic rational homology 3-spheres in the same commensurability classes (as in the 2-
dimensional setting with orbifolds).
Another interesting feature of the construction of [6] is that their manifolds are all Haken. Thus
it seems interesting to refine Cooper’s orginal question to.
Question 6.8. Are there non-Haken hyperbolic 3-manifolds of arbitrarily large injectivity radius
(even allowing for orbifolds)?
6.2
There are many examples of arithmetic rational homology 3-spheres. These can be found efficiently
using the census of hyperbolic 3-manifolds developed using the computer programs SnapPea, and
the exact version Snap (see [13]). We list the 10 smallest known arithmetic hyperbolic 3-manifolds.
The first two are known to be the smallest two by [7]. We list these using the notation of [13].
1: Manifold m003(−3, 1); Volume 0.94270736277692772092.

Minimum polynomial: x3 − x2 + 1; Root: -2.
Discriminant: −23; Signature (1, 1).
Real Ramification [1]; Finite Ramification ν5 .
Integral Traces; Arithmetic.

Minimum polynomial: x4 − x − 1; Root: 3.
20
Real Ramification [1, 2]; Finite Ramification ∅.

Minimum polynomial: x2 − x + 1; Root: 1.
Real Ramification ∅; Finite Ramification ν2 , ν3 .

Minimum polynomial: x4 − x3 + x2 + x − 1; Root: -3.
Real Ramification [1, 2]; Finite Ramification ∅.
5: Manifold m004(6, 1); Volume 1.2844853004683544424603370.

Minimum polynomial: x3 + 2x − 1; Root: 2.

Minimum polynomial: x3 − x2 + 1; Root: 2.

Minimum polynomial: x5 − x3 − 2x2 + 1; Root: 4.
Real Ramification [1, 2, 3]; Finite Ramification ν13 .

Minimum polynomial: x3 + x − 1; Root: 2.

Minimum polynomial: x3 − x2 + x + 1; Root: -2.

Minimum polynomial: x5 − x4 − x3 + 2x2 − x − 1; Root: -4.
Real Ramification [1, 2, 3]; Finite Ramification ν13 .
21
We also give an example of an arithmetic rational homology 3-sphere that is of the type predicted
by Theorem 6.6.
Example: Let B/Q(i) be ramified at {< 1 + i >, < 2 + 3i >}. Let O be a maximal order in B,
then Γ1O is a cocompact torsion-free Kleinian group of volume approximately 3.663862376708 . . ..
The quotient M = H3 /Γ1O is the manifold s705(1, 2) in the SnapPea census of closed hyperbolic
3-manifolds. The first homology of this manifold is Z/21Z.
7 Geodesics in arithmetic hyperbolic 3-manifolds

Recall that if M = H3 /Γ is a closed orientable hyperbolic 3-manifold then the length spectrum L(M )
of M is the set of all lengths of closed geodesics on M counted with multiplicities. The length set
L(M ) of M is the set of lengths all closed geodesics counted without multiplicities. The rational
length spectrum QL(M ) of M is the set of all rational multiples of lengths of closed geodesics of M .
We shall denote the set of axes of hyperbolic elements in Γ by A(Γ).
This section is motivated by the following questions; in these, Mj = H3 /Γj (j = 1, 2) are closed
orientable hyperbolic 3-manifolds.
Question 7.1. If L(M1 ) = L(M2 ), are M1 and M2 are commensurable?
Question 7.2. If L(M1 ) = L(M2 ), are M1 and M2 are commensurable?
Question 7.3. If QL(M1 ) = QL(M2 ), are M1 and M2 are commensurable?
Note that if M1 and M2 are commensurable then QL(M1 ) = QL(M2 ), and that a positive answer
to Question 7.2 provides positive answers to Questions 7.1 and 7.3.
In addition, since it is known (see [16] pp. 415–417) that for closed hyperbolic manifolds, the
spectrum of the Laplace-Beltrami operator action on L2 (M ), counting multiplicities, determines
L(M ), a positive answer to Question 7.2 implies a positive answer to the next question. It should be
noted that the currently known methods of producing isospectral (same spectrum of the Laplace-
Beltrami operator counted with multiplcities) closed hyperbolic 3-manifolds produce commensurable
ones (see [51] and [54]).
Question 7.4. If M1 and M2 are isospectral, are they commensurable?

“Dual” to these questions about lengths of geodesics is the following question about axes deter-
mining commensurability.
Question 7.5. If A(Γ1 ) = A(Γ2 ), are Γ1 and Γ2 commensurable?
As above, it is easy to see that if Γ1 and Γ2 are commensurable, then A(Γ1 ) = A(Γ2 ).
Addressing these questions for general closed hyperbolic 3-manifolds seems very hard at present,
but arithmetic manifolds again provide an interesting class where these questions can be answered.
Indeed, these questions can be asked for closed hyperbolic manifolds in arbitrary dimensisons, and
considerable progress has been made recently in dimensions ≥ 4 in the arithmetic case by Prasad
and Rapinchuk [44]. In particular, they show that there exist arithmetic hyperbolic 5-manifolds
such that Question 7.3 has a negative answer.
In the remainder of this section we outline proofs of positive answers to all of the above questions
for arithmetic hyperbolic 2 and 3-manifolds. This surveys work in [8], [33] and [46].
22
7.1
We begin by sketching the proof of the following result in dimension 2. This is a mild extension of
the result in [46].
Theorem 7.6. Let M1 = H2 /Γ1 be an orientable arithmetic hyperbolic 2-manifold and assume
that M2 = H2 /Γ2 is a closed orientable hyperbolic 2-manifold with L(Γ1 ) = L(Γ2 ). Then Γ2 is
arithmetic, and Γ1 and Γ2 are commensurable.
Before commencing with the sketch of the proof, we discuss, what is perhaps the key feature
that allows one to use arithmetic methods. We motivate this with an example
Example:
√
Let γ ∈ PSL(2, Z) be a hyperbolic element of trace t. Then the eigenvalues of√γ are
(−t± t2 −4) 2
2 , and these being roots of the polynomial p(x) = x2 − tx + 1. Let λγ = (−t+ 2 t −4)
be the eigenvalue satisfying |λγ | > 1. The length of the closed geodesic on the modular surface
H2 /PSL(2, Z) determined by the projection of the axis of γ is 2 ln |λγ |.
Note that λγ is a unit, and it is an elementary excercise to show that p(x) is irreducible over Z.
Thus Q(λγ ) determines a quadratic extension of Q.
Now PSL(2, Z) coincides with the projection of the image of the elements of norm 1 in M (2, Z).
Thus we can find an element u ∈ M (2, Q) such that P(u) = γ, and we deduce from this that Q(λγ )
embeds in M (2, Q) as Q(u).
Conversely, given a real quadratic extension L of Q that embeds in M (2, Q) we can constuct a
hyperbolic element γ ∈ PSL(2, Z) for which L ∼ = Q(λγ ) as follows. Since L is is a real quadratic
∗
extension of Q, the unit group RL of L has rank one. Choosing a fundamental unit u, then un ∈ /Q
∗
for all integers n 6= 0, and so L = Q(un ) for all integers n 6= 0. Now L embeds in M (2, Q), so RL
embeds in M (2, Q) The key claim is that we can construct an order O ⊂ M (2, Q) for which u ∈ O
(see [39] Chapter 12). It follows that since either u or u2 has norm 1, we can arrange that u or
u2 is an element of O1 . By commensurability, we deduce that there is an integer m 6= 0 such that
um ∈ SL(2, Z) and the image of this element in PSL(2, Z) is the required hyperbolic element.
That the quadratic extension is real is important, since otherwise, the unit group is finite, and we
cannot construct a hyperbolic element as needed.
Extending this, the following facts are proved in [39, Chapter 12]. In the statement Γ is either
an arithmetic Fuchsian or Kleinian group.
Theorem 7.7. Suppose that Γ is derived from a quaternion algebra B/k.
(i) Let γ be a hyperbolic element of Γ with eigenvalue λγ . The field k(λγ ) is a quadratic extension
field of k which embeds into B.
(ii) Let L be a quadratic extension of k. If Γ is Fuchsian assume that L is not a totally imaginary
quadratic extension of k.
Then L embeds in B/k if and only if L = k(λγ ) for some hyperbolic γ ∈ Γ. This will be true if
and only if no place of k which splits in L is ramified in B.
Sketch proof of Theorem 7.6: We begin by showing that Γ2 is arithmetic. Let the invariant
quaternion algebra of Γ1 be B1 /k. Standard considerations allow us to deduce that the sets of traces
of elements in Γ1 and Γ2 are the same up to sign, and that the sets {tr γ 2 } are the same. Since
Γ1 is arithmetic, tr γ 2 is an algebraic integer in the totally real field k. Thus all elements of Γ2
have algebraic integer trace and kΓ2 = k. To complete the proof that Γ2 is arithmetic we show
that B2 = AΓ2 is ramified at all infinite places except one, for then the characterization theorem of
Takeuchi [52] applies. To see this we argue as follows.
23
(2)
Let α1 and α2 be a pair non-commuting
hyperbolic
elements in Γ2 . Now a Hilbert Symbol
2
(2)
for AΓ2 can be computed as tr (α1 )−4,tr[α
k
1 ,α2 ]−2
(see [39] Chapter 3). Since Γ1 and Γ2 have
the same sets of traces (up to sign), we can find β1 , β2 ∈ Γ1 such that tr(β1 ) = ±tr(α1 ) and
tr(β2 ) = ±tr([α1 , α2 ]). It follows from [42] Theorem 2.2 that β1 , β2 ∈ Γ1 ∩B11 since tr(β1 ), tr(β2 ) ∈ k.
Let σ : k → R be a non-identity embedding. Since B1 is ramified at σ it follows that |σ(tr x)| < 2
for all non-trivial elements x ∈ B 1 . Hence we conclude that
σ(tr2 (β1 ) − 4) < 0 and σ(tr(β2 ) − 2) < 0.
Hence it follows that
σ(tr2 (α1 ) − 4) < 0 and σ(tr([α1 , α2 ]) − 2) < 0,

from which it follows that B2 is ramified at σ. Now σ was an arbitrary non-identity embedding,
and so we have shown that B2 is ramified at all such embeddings as required.
To establish commensurability, it suffices to show that B1 ∼
= B2 (see [39] Theorem 8.4.6). To
this end, we make the following definition.
Definition: For j = 1, 2 let
Nj = {L/k : [L : k] = 2, L emebds in Bj and is not a totally imaginary quadratic extension of k}.
Clearly B1 and B2 are isomorphic if and only if the set of quadratic extensions that embed in one
is precisely the set that embeds in the other. The key claim that needs to be shown is:
B1 and B2 are isomorphic if and only N1 = N2 .
This is established in [46]. Given this, the proof is completed as follows. Suppose L ∈ N1 . Theorem
7.7 and commensurability shows that L = k(λγ ) for some hyperbolic γ ∈ Γ1 with eigenvalue λγ .
Since L(M1 ) = L(M2 ), it follows that there exists an element γ 0 ∈ Γ2 for which λγ 0 = λγ . Hence
L = k(λγ ) = k(λγ 0 ) embeds in B2 as required. t u
In dimension 3, a similar result is established in [46] for the complex length spectrum. However,
to deal with lengths only requires considerably more work. The following is proved in [8], and in
particular gives an affirmative answer to Questions 7.1, 7.2 7.3 and 7.4 when M1 and M2 are both
arithmetic.
Theorem 7.8. If M is an arithmetic hyperbolic 3-manifold, then the rational length spectrum and
the commensurability class of M determine one another.
The proof of Theorem 7.8 is modelled on some of the ideas contained in the proof of Theorem
7.6, but more involved, since one cannot deal with traces directly. Briefly, the first step in proving
that QL(M ) determines the commensurability class is to show that the invariant trace-field is
determined by QL(M ). We then determine the commensurability class of M from QL(M ) following
ideas similar to those used in the proof of Theorem 7.6. The arguments for the determination of
kΓ splits naturally into two cases; when [k : k ∩ R] > 2 and [k : k ∩ R] = 2. Both cases require
a detailed understanding of the Galois theory of fields with one complex place and their quadratic
extensions. We refer the reader to [8] for details.
24
7.2
If we consider only the “location” of the axes of hyperbolic elements in arithmetic Fuchsian and
Kleinian groups, then again one can get extra information that allows one to prove (see [33]).
Theorem 7.9. If Γ1 and Γ2 are arithmetic Fuchsian (resp. Kleinian) groups then Question 7.5
has a positive answer.
For convenience, we shall call groups Γ1 and Γ2 isoaxial if A(Γ1 ) = A(Γ2 ).
As with the proof of Theorem 7.6 the key issue is to show that the isoaxial condition forces
the same invariant quaternion algebra. We sketch the proof for Kleinian groups that requires the
following definitions. The obvious changes can be made for Fuchsian groups.
Let Γ be a Kleinian group and define
Σ(Γ) = {γ ∈ PSL(2, C) | γ(A(Γ)) = A(Γ)}.

It is easy to check that Σ(Γ) is a subgroup of PSL(2, C), and that Comm(Γ) < Σ(Γ). The key
point now is to show that when Γ is arithmetic, then Comm(Γ) = Σ(Γ). Given this, the proof is
completed as follows.
If Γ1 and Γ2 are isoaxial Kleinian groups, then for any γ ∈ Γ2 , A(Γ1 ) = A(γΓ1 γ −1 ), and therefore
γ ∈ Σ(Γ1 ). Hence Γ2 < Σ(Γ1 ). In addition, if Γ1 is arithmetic, then the equality Comm(Γ1 ) = Σ(Γ1 )
implies that Γ2 < Comm(Γ1 ). It is now standard that, since Γ2 is also arithmetic, Γ1 and Γ2 are
commensurable. t u
Remark: In [35] we construct examples of Fuchsian groups of finite co-area that have the same set
of parabolic fixed points PSL(2, Z) but are non-arithmetic. The question of whether these Fuchsian
groups share the same set of axes as hyperbolic elements of PSL(2, Z) remains open, although we
suspect that this is not the case.
References
[1] I. Agol, Virtual Betti numbers of symmetric spaces, Preprint.
[2] A. Borel, Cohomologie de sous-groupes discrets et représentations de groupes semi-simples,
Astérisque, 32-33 (1976), 73–112.
[3] N. Boston and J. S. Ellenberg, Pro-p groups and towers of rational homology spheres, Geom-
etry and Topology 10 (2006), 331–334.
[4] S. Boyer and X. Zhang, in preparation.

[5] J. O. Button, Fibred and virtually fibred hyperbolic 3-manifolds in the censuses, Experiment.
Math. 14 (2005), 231–255.
[6] F. Calegari and N. Dunfield, Automorphic forms and rational homology 3-spheres, Geometry
and Topology 10 (2006), 295–329.
[7] T. Chinburg, E. Friedman, K. N. Jones and A. W. Reid, The arithmetic hyperbolic 3-manifold
of smallest volume, Ann. Scuola Norm. Sup. Pisa Cl. Sci. 30 (2001), 1–40.
[8] T. Chinburg, E. Hamilton, D. D. Long and A. W. Reid, Geodesics and commensurability
classes of arithmetic hyperbolic 3-manifolds, preprint (2006).
25
[9] L. Clozel, On the cuspidal cohomology of arithmetic subgroups of SL(2n) and the first betti
number of arithmetic 3-manifolds, Duke Math. J. 55 (1987), 475-486.
[10] J. H. Conway and S. Norton, Monstrous moonshine, Bull. L. M. S. 11 (1979), 308–339.
[11] D. Cooper, D. D. Long and A. W. Reid, Essential closed surfaces in bounded 3-manifolds, J.
A. M. S. 10 (1997), 553–563.
[12] D. Cooper, D. D. Long and A. W. Reid, On the virtual Betti numbers of arithmetic hyperbolic
3-manifolds, preprint.
[13] D. Coulson, O. A. Goodman, C. D. Hodgson and W. D. Neumann, Computing arithmetic
invariants of 3-manifolds, Experimental Math. 9 (2000), 127–152.
[14] J. E. Cremona, Hyperbolic tessellations, modular symbols and elliptic curves over complex
quadratic fields, Composito Math. 51 (1984) 275–323.
[15] D. Gabai, On 3-manifolds finitely covered by surface bundles, in Low-dimensional topology
and Kleinian groups, London Math. Soc. Lecture Note Ser. 112, 145–155, Cambridge Univ.
Press, (1986).
[16] R. Gangolli, The length spectra of some compact manifolds, J. Diff. Geom. 12 (1977), 403–424.
[17] S. S. Gelbart, Automorphic Forms on Adele Groups, Annals of Math. Studies 83, Princeton
Univ. Press, 1975
[18] S. S. Gelbart and H. Jacquet, A relation between automorphic representations of GL(2) and
GL(3), Ann. Sci. Ecole Norm. Sup. 11 (1978), 471–542.
[19] F. Grunewald and J. Mennicke, SL2 (O) and elliptic curves, Manuscript Bielefeld.
[20] E. Hamilton, Abelian subgroup separability of Haken 3-manifolds and closed hyperbolic n-
orbifolds, Proc. London Math. Soc. 83 (2001), 626–646.
[21] J. Hass, Surfaces minimizing area in their homology class and group actions on 3-manifolds,
Math. Zeit. 199 (1988), 501–509.
[22] J. Hempel, 3-Manifolds, Annals of Math. Studies 86 (1976), Princeton Univ. Press.
[23] H. Jacquet and R. P. Langlands, Automorphic Forms on GL(2), L.N.M. 114, Springer (1970).
[24] J.-P Labesse and J. Schwermer, On liftings and cusp cohomology of arithmetic groups, Invent.
Math. 83 (1986), 383-401.
[25] M. Lackenby, A characterisation of large finitely presented groups, J. Algebra 287 (2005),
458–473.
[26] M. Lackenby, Covering spaces of 3-orbifolds, Duke Math J. 136 (2007), 181–203.
[27] M. Lackenby, Some 3-manifolds and 3-orbifolds with large fundamental group, preprint.
[28] M. Lackenby, D. D. Long and A. W. Reid, Covering spaces of arithmetic 3-orbifolds, preprint.
[29] S. Lang, Algebraic Number Theory, Grad. Texts in Math. 110 Springer (1986).
[30] C. J. Leininger, Surgeries on one component of the Whitehead link are virtually fibered,
Topology 41 (2002), 307–320.
26
[31] J. S. Li and J. J. Millson, On the first betti number of a hyperbolic manifold with an arithmetic
fundamental group, Duke Math. J. 71 (1993), 365-401.
[32] D. D. Long, Immersions and embeddings of totally geodesic surfaces, Bull. London Math.
Soc. 19 (1987), 481–484.
[33] D. D. Long and A. W. Reid, On Fuchsian groups with the same set of axes, Bull. London
Math. Soc. 30 (1998), 533–538.
[34] D. D. Long and A. W. Reid, Constructing hyperbolic manifolds which bound geometrically,
Math. Research Letters 8 (2001), 443–456.
[35] D. D. Long and A. W. Reid, Pseudomodular surfaces, J. Reine Angew. Math. 552 (2002),
77–100.
[36] D. D. Long, C. Maclachlan and A. W. Reid, Arithmetic Fuchsian groups of genus zero, Pure
and Appl. Math. Quart. 2 (2006), 569–599.
[37] A. Lubotzky, Free quotients and the first Betti number of some hyperbolic manifolds, Trans-
form. Groups 1 (1996,) 71–82.
[38] A. Lubotzky, Eigenvalues of the Laplacian, the first Betti number and the congruence sub-
group problem, Annals of Math. 144 (1996), 441–452.
[39] C. Maclachlan and A. W. Reid, The Arithmetic of Hyperbolic 3-Manifolds, Graduate Texts
in Mathematics, 219, Springer-Verlag (2003).
[40] G. Margulis, Discrete Subgroups of Semi-simple Lie Groups, Ergebnisse der Mathematik und
ihr Grenzgebeite, Springer (1991).
[41] Y. Matsushima, A formula for the betti numbers of compact locally symmetric Riemannian
manifolds, J. Diff. Geom. 1 (1967), 99-109.
[42] W. D. Neumann and A. W. Reid, Arithmetic of hyperbolic manifolds, in Topology ’90, Ohio
State Univ. Math. Res. Inst. Publ. 1, 273–310, de Gruyter (1992).
[43] A.P. Ogg, Automorphismes des Courbes Modulaires, Seminaire Delange-Pisot, Poitou, 7,
(1974).
[44] G. Prasad and A. Rapinchuk, Length-commensurable locally symmetric spaces, preprint.
[45] C. S. Rajan, On the non-vanishing of the first Betti number of hyperbolic three manifolds,
Math. Ann. 330 (2004), 323–329.
[46] A. W. Reid, Isospectrality and commensurability of arithmetic hyperbolic 2- and 3-manifolds,
Duke Math. J. 65 (1992), 215–228.
[47] A. W. Reid, A non-Haken hyperbolic 3-manifold covered by a surface bundle, Pacific J. Math.
167 (1995), 163–182.
[48] J. Schwermer, Special cycles and automorphic forms on arithmetically defined hyperbolic 3-
manifolds, Asian J. Math. 8 (2004), 837–859.
[49] P. B. Shalen and P. Wagreich, Growth rates, Zp homology, and volumes of hyperbolic 3-
manifolds, Trans. Amer. Math. Soc. 331 (1992), 895–917.
27
[50] J. H. Silverman, The Arithmetic of Elliptic Curves, Graduate Texts in Mathematics, 106,
Springer-Verlag (1986).
[51] T. Sunada, Riemannian coverings and isospectral manifolds, Annals of Math. 121 (1985),
169–186.
[52] K. Takeuchi, A characterization of arithmetic Fuchsian groups, J. Math. Soc. Japan 27

(1975), 600–612.
[53] T. N. Venkataramana, Virtual Betti numbers of compact locally symmetric spaces, Preprint.
[54] M-F. Vignéras, Variétiés Riemanniennes isospectrales et non isométriques, Annals of Math.
112 (1980), 21–32.
[55] M-F Vigneras, Quelques remarques sur la conjecture λ1 ≥ 14 , Seminar on number theory,
Paris 1981–82, Progr. Math., 38 Birkhauser, (1983) 321–343.
[56] K. Vogtmann, Rational homology of Bianchi groups, Math. Annalen 272 (1985), 399–419.
[57] G. S. Walsh, Great circle links and virtually fibered knots, Topology 44 (2005), 947–958.
[58] P. Zograf, A spectral proof of Rademacher’s conjecture for congruence subgroups of the mod-
ular group, J. Reine Angew. Math. 414 (1991), 113–116.
Department of Mathematics,
University of Texas
Austin, TX 78712, U.S.A.
Email: [email protected]
28

Kyoto 2

Uploaded by

Copyright:

Available Formats

Kyoto 2

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Kyoto 2

Uploaded by

Copyright:

Available Formats

The geometry and topology of arithmetic hyperbolic

vb1 (M) = sup{b1 (N) : N is a finite cover of M},

and the invariant quaternion algebra

AΓ = {Σaj γj : aj ∈ kΓ, γj ∈ Γ(2) }.

Example 2: If L is a link in S 3 with S 3 \ L = H3 /Γ then r(Γ) = 0, since Γ is generated by

3 A geometric characterization of arithmeticity

Comm(Γ) = {x ∈ PSL(2, C) : xΓx−1 is commensurable with Γ}

4 Towards Conjectures 1.1–1.4 for arithmetic hyperbolic 3-

Sketch proof of Theorem 4.2:

Remarks:(1) An interesting feature of this argument is that although it uses arithmetic in an

Sketch proof of Theorem 4.3:

dp (Qi ) ≥ |sing0p (Qi )|.

|sing0p (Qi )| ≥ [Qi : Q].

Given this we sketch the proof of:

Together with Theorem 4.1 we have.

5 The 2-dimensional situation

Γ(n) = P{A ∈ SL(2, Z) : A ≡ I mod n}.

More generally, a non-cocompact arithmetic Fuchsian group Γ is congruence if some conjugate of Γ

Definition: Let Γ be an arithmetic Fuchsian group. Γ is called a congruence subgroup if there is

Computations suggest (see [36]) the following conjecture.

6 Arithmetic rational homology 3-spheres

d ∈ D = {1, 2, 3, 5, 6, 7, 11, 15, 19, 23, 31, 39, 47, 71}.

Recall that if M = H3 /Γ is a closed orientable hyperbolic 3-manifold, the injectivity radius of M

1: Manifold m003(−3, 1); Volume 0.94270736277692772092.

2: Manifold m003(−2, 3); Volume 0.9813688288922320880914.

3: Manifold m010(−1, 2); Volume 1.01494160640965362502120.

4: Manifold m003(−4, 3); Volume 1.263709238658043655884716.

5: Manifold m004(6, 1); Volume 1.2844853004683544424603370.

6: Manifold m003(−3, 4); Volume 1.414061044165391581381949.

7: Manifold m003(−5, 3); Volume 1.54356891147185507432847.

8: Manifold m007(4, 1); Volume 1.583166660624812836166028.

9: Manifold m006(3, 1); Volume 1.588646639300162988176913.

10: Manifold m015(5, 1); Volume 1.7571260291884513628747465.

7 Geodesics in arithmetic hyperbolic 3-manifolds

Question 7.4. If M1 and M2 are isospectral, are they commensurable?

σ(tr2 (β1 ) − 4) < 0 and σ(tr(β2 ) − 2) < 0.

Hence it follows that

σ(tr2 (α1 ) − 4) < 0 and σ(tr([α1 , α2 ]) − 2) < 0,

Definition: For j = 1, 2 let

Nj = {L/k : [L : k] = 2, L emebds in Bj and is not a totally imaginary quadratic extension of k}.

B1 and B2 are isomorphic if and only N1 = N2 .

Σ(Γ) = {γ ∈ PSL(2, C) | γ(A(Γ)) = A(Γ)}.

[4] S. Boyer and X. Zhang, in preparation.

[52] K. Takeuchi, A characterization of arithmetic Fuchsian groups, J. Math. Soc. Japan 27

You might also like