Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Minimal co-volume hyperbolic lattices Solution of Siegel’s problem in three dimensions Gaven J. Martin G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result A lattice is a discrete subgroup Γ of isometries of hyperbolic 3-space H3 with finite co-volume. Thus the orbit space O = H3 /Γ is a hyperbolic orbifold, or manifold if Γ is torsion free (no elements of finite order), of finite volume. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result A lattice is a discrete subgroup Γ of isometries of hyperbolic 3-space H3 with finite co-volume. Thus the orbit space O = H3 /Γ is a hyperbolic orbifold, or manifold if Γ is torsion free (no elements of finite order), of finite volume. An isometry Φ : H3 → H3 is uniquely determined by its boundary values on ∂H3 ≈ Ĉ (Poincaré extension) which determine a Möbius transformation and the maps Isom+ (H3 ) 3 Φ ↔ Φ|Ĉ = az + b a ↔ c cz + d b d ∈ PSL(2, C) is an isomorphism of Lie groups. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result A lattice is a discrete subgroup Γ of isometries of hyperbolic 3-space H3 with finite co-volume. Thus the orbit space O = H3 /Γ is a hyperbolic orbifold, or manifold if Γ is torsion free (no elements of finite order), of finite volume. An isometry Φ : H3 → H3 is uniquely determined by its boundary values on ∂H3 ≈ Ĉ (Poincaré extension) which determine a Möbius transformation and the maps Isom+ (H3 ) 3 Φ ↔ Φ|Ĉ = az + b a ↔ c cz + d b d ∈ PSL(2, C) is an isomorphism of Lie groups. A Kleinian group is a discrete subgroup of Isom+ (H3 ) ≈ PSL(2, C) which is not virtually abelian. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result In 1945 Siegel posed the problem of identifying the smallest co-volume hyperbolic lattices in n dimensions µ(n) = inf Γ Hn /Γ Here Γ ⊂ Isom+ (Hn ) is a discrete subgroup. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result In 1945 Siegel posed the problem of identifying the smallest co-volume hyperbolic lattices in n dimensions µ(n) = inf Γ Hn /Γ Here Γ ⊂ Isom+ (Hn ) is a discrete subgroup. For Euclidean lattices (Bieberbach or Crystallographic groups) such an infimum is obviously 0. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result In 1945 Siegel posed the problem of identifying the smallest co-volume hyperbolic lattices in n dimensions µ(n) = inf Γ Hn /Γ Here Γ ⊂ Isom+ (Hn ) is a discrete subgroup. In two dimensions Siegel shows G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result In 1945 Siegel posed the problem of identifying the smallest co-volume hyperbolic lattices in n dimensions µ(n) = inf Γ Hn /Γ Here Γ ⊂ Isom+ (Hn ) is a discrete subgroup. In two dimensions Siegel shows µ(2) = π 21 , attained for the (2, 3, 7)-triangle group. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result In 1945 Siegel posed the problem of identifying the smallest co-volume hyperbolic lattices in n dimensions µ(n) = inf Γ Hn /Γ Here Γ ⊂ Isom+ (Hn ) is a discrete subgroup. In two dimensions Siegel shows µ(2) = π 21 , attained for the (2, 3, 7)-triangle group. Poincaré-Klein refined to signature formula X (1 − 1/m) A = 2π 2g − 2 + n + G.J.Martin Minimal co-volume hyperbolic lattices . Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result In 1945 Siegel posed the problem of identifying the smallest co-volume hyperbolic lattices in n dimensions µ(n) = inf Γ Hn /Γ Here Γ ⊂ Isom+ (Hn ) is a discrete subgroup. In two dimensions Siegel shows µ(2) = π 21 , attained for the (2, 3, 7)-triangle group. Poincaré-Klein refined to signature formula X (1 − 1/m) A = 2π 2g − 2 + n + suggests a connection with Hurwitz’ 84g − 84 theorem on the symmetries of Riemann surfaces. (confirmed by McBeath in 1961) G.J.Martin Minimal co-volume hyperbolic lattices . Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result Mostow rigidity (topology = geometry in n ≥ 3). G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result Mostow rigidity (topology = geometry in n ≥ 3). Kazhdan and Margulis show µ(n) > 0 and is attained. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result Mostow rigidity (topology = geometry in n ≥ 3). Kazhdan and Margulis show µ(n) > 0 and is attained. Wang shows spectrum of volumes of lattices is discrete for n ≥ 4 G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result Mostow rigidity (topology = geometry in n ≥ 3). Kazhdan and Margulis show µ(n) > 0 and is attained. Wang shows spectrum of volumes of lattices is discrete for n ≥ 4 Selberg’s lemma gives general existence of torsion free subgroups of finite index : the most symmetric hyperbolic manifolds are quotients of smallest co-volume lattices. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result Mostow rigidity (topology = geometry in n ≥ 3). Kazhdan and Margulis show µ(n) > 0 and is attained. Wang shows spectrum of volumes of lattices is discrete for n ≥ 4 Selberg’s lemma gives general existence of torsion free subgroups of finite index : the most symmetric hyperbolic manifolds are quotients of smallest co-volume lattices. Jørgensen, Thurston show in three dimensions that the set of volumes is of type ω ω . G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result Mostow rigidity (topology = geometry in n ≥ 3). Kazhdan and Margulis show µ(n) > 0 and is attained. Wang shows spectrum of volumes of lattices is discrete for n ≥ 4 Selberg’s lemma gives general existence of torsion free subgroups of finite index : the most symmetric hyperbolic manifolds are quotients of smallest co-volume lattices. Jørgensen, Thurston show in three dimensions that the set of volumes is of type ω ω . √ Meyerhoff µ(3) > 0.00005 and identifies PGL(2, O( −3)) as smallest co-volume noncompact lattice. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result Mostow rigidity (topology = geometry in n ≥ 3). Kazhdan and Margulis show µ(n) > 0 and is attained. Wang shows spectrum of volumes of lattices is discrete for n ≥ 4 Selberg’s lemma gives general existence of torsion free subgroups of finite index : the most symmetric hyperbolic manifolds are quotients of smallest co-volume lattices. Jørgensen, Thurston show in three dimensions that the set of volumes is of type ω ω . √ Meyerhoff µ(3) > 0.00005 and identifies PGL(2, O( −3)) as smallest co-volume noncompact lattice. Borel gives co-volume formula for maximal arithmetic lattices in three dimensions. Gives useful criteria to determine arithmetity. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result Mostow rigidity (topology = geometry in n ≥ 3). Kazhdan and Margulis show µ(n) > 0 and is attained. Wang shows spectrum of volumes of lattices is discrete for n ≥ 4 Selberg’s lemma gives general existence of torsion free subgroups of finite index : the most symmetric hyperbolic manifolds are quotients of smallest co-volume lattices. Jørgensen, Thurston show in three dimensions that the set of volumes is of type ω ω . √ Meyerhoff µ(3) > 0.00005 and identifies PGL(2, O( −3)) as smallest co-volume noncompact lattice. Borel gives co-volume formula for maximal arithmetic lattices in three dimensions. Gives useful criteria to determine arithmetity. Chinburg and Friedman identify smallest arithmetic lattice G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result In a sequence of works, culminating in papers with Gehring and with Marshall we proved Theorem µ(3) = 2753/2 2−7 π −6 ζK (2) ≈ 0.03905 Here ζK is the Dedekind zeta function of the field K = Q(α) with α a complex root of α4 + 6α3 + 12α2 + 9α + 1 = 0 The extremal is uniquely achieved in the Z2 -extension of the 3-5-3 hyperbolic Coxeter group. This is an arithmetic two generator group, generated by elements of orders 2 and 5. G.J.Martin Minimal co-volume hyperbolic lattices Fred Gehring Tim Marshall Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Definitions The problem Progress and related results The main result The next three smallest co-volume lattices contain groups generated by two elements of finite orders 2 and 3 with low index. 2833/2 2−7 π −6 ζK (2) = 0.0408 . . ., K = Q(γ), γ 4 + 5γ 3 + 7γ 2 + 3γ + 1 = 0, discriminant −283 with unramified quaternion algebra. It also is a two-generator arithmetic Kleinian group. 313/2 2−6 π −4 (NP3 − 1)ζK (2) = 0.0659 . . ., K = Q(γ), γ 3 + 4γ 2 + 5γ + 3 = 0, discriminant −31 and ramified at the finite place P3 . It contains (3,0)-(3,0) Dehn sugery on the Whitehead link of index 8 and a group generated by elements of order 2 and 3 of index 4. 443/2 2−6 π −4 (NP2 − 1)ζk (2) = 0.0661 . . ., K = Q(γ), γ 3 + 4γ 2 + 4γ + 2 = 0 of discriminant −44 and ramified at the finite place P2 . It also contains a group generated by elements of order 2 and 3 of index 4. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Basically an arithmetic Kleinian group Γ is one algebraically isomorphic to something commensurable with SL(m, Z) for some m. So there’s a representation ρ : Γ → GL(m, C) so that |ρ(Γ) ∩ SL(m, Z) : SL(m, Z)| + |ρ(Γ) ∩ SL(m, Z) : ρ(Γ)| < ∞ Borel gave a nice description of these groups in three dimensions. The group will be 2 × 2 matrices defined over a finite field (likely to be of degree m) of special type with structure conditions on the associated quaternion algebra. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Basically an arithmetic Kleinian group Γ is one algebraically isomorphic to something commensurable with SL(m, Z) for some m. So there’s a representation ρ : Γ → GL(m, C) so that |ρ(Γ) ∩ SL(m, Z) : SL(m, Z)| + |ρ(Γ) ∩ SL(m, Z) : ρ(Γ)| < ∞ Borel gave a nice description of these groups in three dimensions. The group will be 2 × 2 matrices defined over a finite field (likely to be of degree m) of special type with structure conditions on the associated quaternion algebra. In higher rank, all lattices in semi-simple Lie groups are arithmetic (Margulis super-rigidity) and even in rank one only hyperbolic and complex hyperbolic lattices need not be arithmetic. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Signature formula and Takeuchi’s identification of arithmetic triangle groups shows smallest 9 lattices of H2 are arithmetic. Smallest non-arithmetic lattice is the (2, 3, 13)-triangle group. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Signature formula and Takeuchi’s identification of arithmetic triangle groups shows smallest 9 lattices of H2 are arithmetic. Smallest non-arithmetic lattice is the (2, 3, 13)-triangle group. Two smallest noncompact lattices are arithmetic - (2, 3, ∞) and (2, 4, ∞). Smallest non-arithmetic lattice is (2, 5, ∞). G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Signature formula and Takeuchi’s identification of arithmetic triangle groups shows smallest 9 lattices of H2 are arithmetic. Smallest non-arithmetic lattice is the (2, 3, 13)-triangle group. Two smallest noncompact lattices are arithmetic - (2, 3, ∞) and (2, 4, ∞). Smallest non-arithmetic lattice is (2, 5, ∞). For 3-manifolds we have the expected data (Maclachlan-Reid) G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Signature formula and Takeuchi’s identification of arithmetic triangle groups shows smallest 9 lattices of H2 are arithmetic. Smallest non-arithmetic lattice is the (2, 3, 13)-triangle group. Two smallest noncompact lattices are arithmetic - (2, 3, ∞) and (2, 4, ∞). Smallest non-arithmetic lattice is (2, 5, ∞). For 3-manifolds we have the expected data (Maclachlan-Reid) Five smallest closed manifolds arithmetic but only 20 of first 50 G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Signature formula and Takeuchi’s identification of arithmetic triangle groups shows smallest 9 lattices of H2 are arithmetic. Smallest non-arithmetic lattice is the (2, 3, 13)-triangle group. Two smallest noncompact lattices are arithmetic - (2, 3, ∞) and (2, 4, ∞). Smallest non-arithmetic lattice is (2, 5, ∞). For 3-manifolds we have the expected data (Maclachlan-Reid) Five smallest closed manifolds arithmetic but only 20 of first 50 Smallest noncompact manifold arithmetic but only 2 of first 50 G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Signature formula and Takeuchi’s identification of arithmetic triangle groups shows smallest 9 lattices of H2 are arithmetic. Smallest non-arithmetic lattice is the (2, 3, 13)-triangle group. Two smallest noncompact lattices are arithmetic - (2, 3, ∞) and (2, 4, ∞). Smallest non-arithmetic lattice is (2, 5, ∞). For 3-manifolds we have the expected data (Maclachlan-Reid) Five smallest closed manifolds arithmetic but only 20 of first 50 Smallest noncompact manifold arithmetic but only 2 of first 50 In general not a lot of data - in joint work with Gehring, Maclachlan and Reid we expect the dozen smallest lattices arithmetic, G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Signature formula and Takeuchi’s identification of arithmetic triangle groups shows smallest 9 lattices of H2 are arithmetic. Smallest non-arithmetic lattice is the (2, 3, 13)-triangle group. Two smallest noncompact lattices are arithmetic - (2, 3, ∞) and (2, 4, ∞). Smallest non-arithmetic lattice is (2, 5, ∞). For 3-manifolds we have the expected data (Maclachlan-Reid) Five smallest closed manifolds arithmetic but only 20 of first 50 Smallest noncompact manifold arithmetic but only 2 of first 50 In general not a lot of data - in joint work with Gehring, Maclachlan and Reid we expect the dozen smallest lattices arithmetic, .0390, .0408, .0659, .0661, .0717, .0785, .0845, .0933, .1028, .1268, .1274, .1374 Smallest non-arithmetic commensurable with a tetrahedral group. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Signature formula and Takeuchi’s identification of arithmetic triangle groups shows smallest 9 lattices of H2 are arithmetic. Smallest non-arithmetic lattice is the (2, 3, 13)-triangle group. Two smallest noncompact lattices are arithmetic - (2, 3, ∞) and (2, 4, ∞). Smallest non-arithmetic lattice is (2, 5, ∞). For 3-manifolds we have the expected data (Maclachlan-Reid) Five smallest closed manifolds arithmetic but only 20 of first 50 Smallest noncompact manifold arithmetic but only 2 of first 50 In general not a lot of data - in joint work with Gehring, Maclachlan and Reid we expect the dozen smallest lattices arithmetic, .0390, .0408, .0659, .0661, .0717, .0785, .0845, .0933, .1028, .1268, .1274, .1374 Smallest non-arithmetic commensurable with a tetrahedral group. Prevalence of two generator arithmetic groups G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups simple and non-simple elements spherical points Any nonparabolic f element of a Kleinian group has two fixed points on Ĉ and setwise fixes the closed hyperbolic line connecting these points the axis. Up to conjugacy these points may be {0, ∞} and the map is conjugate to α 0 z 7→ α2 z ∼ 0 1/α Ç The map translates along the axis distance τ (f ) = 2 log |α|, C (translation length) and rotates about the axis an axis in H3 θ(f ) = 2 arg α, the holonomy. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups simple and non-simple elements spherical points An axis α in Γ is simple if for each g ∈ Γ g (α) = α, or G.J.Martin g (α) ∩ α = ∅ Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups simple and non-simple elements spherical points An axis α in Γ is simple if for each g ∈ Γ g (α) = α, or g (α) ∩ α = ∅ The axis with shortest translation length is always simple (cut and paste near the crossing). G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups simple and non-simple elements spherical points An axis α in Γ is simple if for each g ∈ Γ g (α) = α, or g (α) ∩ α = ∅ The axis with shortest translation length is always simple (cut and paste near the crossing). Discreteness shows if an axis α is simple, then r= n o 1 inf ρH (α, g (α)) : g ∈ Γ, g (α) 6= α > 0 2 so α lies in a precisely invariant solid hyperbolic cylinder C(α, r ). G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups simple and non-simple elements spherical points An axis α in Γ is simple if for each g ∈ Γ g (α) = α, or g (α) ∩ α = ∅ The axis with shortest translation length is always simple (cut and paste near the crossing). Discreteness shows if an axis α is simple, then r= n o 1 inf ρH (α, g (α)) : g ∈ Γ, g (α) 6= α > 0 2 so α lies in a precisely invariant solid hyperbolic cylinder C(α, r ). Clearly volH (H3 /Γ) ≥ volH (C/Γα ) = πτ sinh2 (r ) p as the (set) stabiliser Γα of α has a particularly simple structure - at worst its Z × Zp o Z2 . G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups simple and non-simple elements spherical points An axis α in Γ is simple if for each g ∈ Γ g (α) = α, or g (α) ∩ α = ∅ The axis with shortest translation length is always simple (cut and paste near the crossing). Discreteness shows if an axis α is simple, then r= n o 1 inf ρH (α, g (α)) : g ∈ Γ, g (α) 6= α > 0 2 so α lies in a precisely invariant solid hyperbolic cylinder C(α, r ). Clearly volH (H3 /Γ) ≥ volH (C/Γα ) = πτ sinh2 (r ) p as the (set) stabiliser Γα of α has a particularly simple structure - at worst its Z × Zp o Z2 . Must bound r , τ from below and p from above. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups simple and non-simple elements spherical points An axis α in Γ is simple if for each g ∈ Γ g (α) = α, or g (α) ∩ α = ∅ The axis with shortest translation length is always simple (cut and paste near the crossing). Discreteness shows if an axis α is simple, then r= n o 1 inf ρH (α, g (α)) : g ∈ Γ, g (α) 6= α > 0 2 so α lies in a precisely invariant solid hyperbolic cylinder C(α, r ). Clearly volH (H3 /Γ) ≥ volH (C/Γα ) = πτ sinh2 (r ) p as the (set) stabiliser Γα of α has a particularly simple structure - at worst its Z × Zp o Z2 . Must bound r , τ from below and p from above. Never sharp ! G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups simple and non-simple elements spherical points If α is the axis of an elliptic f and g (α) ∩ α = x0 ∈ H3 , then f and g ◦ f ◦ g −1 share a fixed point x0 and hf , g ◦ f ◦ g −1 i is conjugate to a group of rotations - spherical triangle group. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups simple and non-simple elements spherical points If α is the axis of an elliptic f and g (α) ∩ α = x0 ∈ H3 , then f and g ◦ f ◦ g −1 share a fixed point x0 and hf , g ◦ f ◦ g −1 i is conjugate to a group of rotations - spherical triangle group. These are classified as A4 , A5 and S4 and the dihedral groups Dn , n ≥ 2. We call such a point x0 a spherical fixed point. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups simple and non-simple elements spherical points If α is the axis of an elliptic f and g (α) ∩ α = x0 ∈ H3 , then f and g ◦ f ◦ g −1 share a fixed point x0 and hf , g ◦ f ◦ g −1 i is conjugate to a group of rotations - spherical triangle group. These are classified as A4 , A5 and S4 and the dihedral groups Dn , n ≥ 2. We call such a point x0 a spherical fixed point. Again discreteness gives r= n o 1 inf ρH (x0 , g (x0 )) : g ∈ Γ \ Γx0 > 0 2 Thus BH (x0 , r ) is a precisely invariant ball and volH (H3 /Γ) ≥ volH (B/Γx0 ) = π(sinh2 (2r ) − 2r )/|Γx0 | so again we seek to bound r G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups simple and non-simple elements spherical points If α is the axis of an elliptic f and g (α) ∩ α = x0 ∈ H3 , then f and g ◦ f ◦ g −1 share a fixed point x0 and hf , g ◦ f ◦ g −1 i is conjugate to a group of rotations - spherical triangle group. These are classified as A4 , A5 and S4 and the dihedral groups Dn , n ≥ 2. We call such a point x0 a spherical fixed point. Again discreteness gives r= n o 1 inf ρH (x0 , g (x0 )) : g ∈ Γ \ Γx0 > 0 2 Thus BH (x0 , r ) is a precisely invariant ball and volH (H3 /Γ) ≥ volH (B/Γx0 ) = π(sinh2 (2r ) − 2r )/|Γx0 | so again we seek to bound r G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups simple and non-simple elements spherical points If α is the axis of an elliptic f and g (α) ∩ α = x0 ∈ H3 , then f and g ◦ f ◦ g −1 share a fixed point x0 and hf , g ◦ f ◦ g −1 i is conjugate to a group of rotations - spherical triangle group. These are classified as A4 , A5 and S4 and the dihedral groups Dn , n ≥ 2. We call such a point x0 a spherical fixed point. Again discreteness gives r= n o 1 inf ρH (x0 , g (x0 )) : g ∈ Γ \ Γx0 > 0 2 Thus BH (x0 , r ) is a precisely invariant ball and volH (H3 /Γ) ≥ volH (B/Γx0 ) = π(sinh2 (2r ) − 2r )/|Γx0 | so again we seek to bound r Never sharp ! G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. The question of bounding the quantities in the estimates above reduce to questions about two generator discrete groups. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. The question of bounding the quantities in the estimates above reduce to questions about two generator discrete groups. Let Γ be a Kleinian group. The trace of f ∈ Γ is tr (f ) = ±tr (Af ) where Af ∈ PSL(2, C) represents f . Then for f , g ∈ Isom+ (H) β(f ) = tr 2 (f ) − 4, β(g ) = tr 2 (g ) − 4 and γ(f , g ) = tr [f , g ] − 2 These parameters determine hf , g i uniquely up to conjugacy and encode other geometric quantities. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. γ(f , g ) = 0 if and only if f and g share a common fixed point in Ĉ G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. γ(f , g ) = 0 if and only if f and g share a common fixed point in Ĉ f and g are elliptic or loxodromic with translation lengths τf and τg and holonomies θf and θg , then τf + iθf 2 β(f ) = 4 sinh 2 τ + iθg g 2 β(g ) = 4 sinh 2 β(f )β(g ) γ(f , g ) = sinh2 (δ + iφ) 4 where δ + iφ is the complex distance between the two axes. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. View the space of all two-generator Kleinian groups modulo conjugacy as a subset of C3 via the map hf , g i → (γ(f , g ), β(f ), β(g )) G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. View the space of all two-generator Kleinian groups modulo conjugacy as a subset of C3 via the map hf , g i → (γ(f , g ), β(f ), β(g )) The fundamental problem is to find the points in C3 which correspond to discrete groups - or at least give a good description of this space. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. View the space of all two-generator Kleinian groups modulo conjugacy as a subset of C3 via the map hf , g i → (γ(f , g ), β(f ), β(g )) The fundamental problem is to find the points in C3 which correspond to discrete groups - or at least give a good description of this space. This space has very complicated structure, but those points corresponding to two-generator lattices will be isolated. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. View the space of all two-generator Kleinian groups modulo conjugacy as a subset of C3 via the map hf , g i → (γ(f , g ), β(f ), β(g )) The fundamental problem is to find the points in C3 which correspond to discrete groups - or at least give a good description of this space. This space has very complicated structure, but those points corresponding to two-generator lattices will be isolated. Lemma (Projection) (γ(f , g ), β(f ), β(g )) discrete ⇒ (γ, β, −4) discrete This map is a contraction on complex distances. Away from a finite set of exceptional parameters this preserves “non-elementary”. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Let ha, bi be the free group on the two letters a and b. Say w ∈ ha, bi is a good word if w can be written as w = b s1 ar1 b s2 ar2 . . . b sm−1 arm−1 b sm where s1 ∈ {±1}, sj = (−1)j+1 s1 and rj 6= 0 G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Let ha, bi be the free group on the two letters a and b. Say w ∈ ha, bi is a good word if w can be written as w = b s1 ar1 b s2 ar2 . . . b sm−1 arm−1 b sm where s1 ∈ {±1}, sj = (−1)j+1 s1 and rj 6= 0. Good words start with b and end in b ±1 depending on whether m is even or odd - the exponents of b alternate in sign G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Let ha, bi be the free group on the two letters a and b. Say w ∈ ha, bi is a good word if w can be written as w = b s1 ar1 b s2 ar2 . . . b sm−1 arm−1 b sm where s1 ∈ {±1}, sj = (−1)j+1 s1 and rj 6= 0. Good words start with b and end in b ±1 depending on whether m is even or odd - the exponents of b alternate in sign. Next we introduce a key tool used in our study of parameter spaces of discrete groups. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Let ha, bi be the free group on the two letters a and b. Say w ∈ ha, bi is a good word if w can be written as w = b s1 ar1 b s2 ar2 . . . b sm−1 arm−1 b sm where s1 ∈ {±1}, sj = (−1)j+1 s1 and rj 6= 0. Good words start with b and end in b ±1 depending on whether m is even or odd - the exponents of b alternate in sign. Next we introduce a key tool used in our study of parameter spaces of discrete groups. Theorem Let a, b ∈ PSL(2, C) and w = w (a, b) ∈ ha, bi be a good word. Set β = β(f ) and γ = γ(f , g ). Then there is a monic polynomial with integer coefficients such that γ(f , w (f , g )) = pw (γ, β). G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. There are three things to note. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. There are three things to note. ◦ If b 2 = 1 : alternating sign is redundant and every w (a, b) is good. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. There are three things to note. ◦ If b 2 = 1 : alternating sign is redundant and every w (a, b) is good. ◦ A semigroup operation on good words: If w1 = w1 (a, b) and w2 = w2 (a, b) are good , then so is w1 ∗ w2 = w1 (a, w2 (a, b)) For example (bab −1 ab) ∗ (bab −1 ) = bab −1 a(bab −1 )−1 abab −1 = bab −1 aba−1 b −1 abab −1 G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. There are three things to note. ◦ If b 2 = 1 : alternating sign is redundant and every w (a, b) is good. ◦ A semigroup operation on good words: If w1 = w1 (a, b) and w2 = w2 (a, b) are good , then so is w1 ∗ w2 = w1 (a, w2 (a, b)) For example (bab −1 ab) ∗ (bab −1 ) = bab −1 a(bab −1 )−1 abab −1 = bab −1 aba−1 b −1 abab −1 ◦ pw1 ∗w2 (γ, β) = pw1 (pw2 (γ, β), β) corresponding to polynomial composition in the first slot. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. There are three things to note. ◦ If b 2 = 1 : alternating sign is redundant and every w (a, b) is good. ◦ A semigroup operation on good words: If w1 = w1 (a, b) and w2 = w2 (a, b) are good , then so is w1 ∗ w2 = w1 (a, w2 (a, b)) For example (bab −1 ab) ∗ (bab −1 ) = bab −1 a(bab −1 )−1 abab −1 = bab −1 aba−1 b −1 abab −1 ◦ pw1 ∗w2 (γ, β) = pw1 (pw2 (γ, β), β) corresponding to polynomial composition in the first slot. Generalised Chebychev polynomials ?? G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Notice the obvious facts ◦ hf , g i Kleinian, implies hf , w (f , g )i discrete. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Notice the obvious facts ◦ hf , g i Kleinian, implies hf , w (f , g )i discrete. ◦ for any word w = w (f , g ) and m, n ∈ Z, γ(f , f m wf n ) = γ(f , w ) - the requirement the word start and end in a nontrivial power of b is simply to avoid obvious redundancy. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Notice the obvious facts ◦ hf , g i Kleinian, implies hf , w (f , g )i discrete. ◦ for any word w = w (f , g ) and m, n ∈ Z, γ(f , f m wf n ) = γ(f , w ) - the requirement the word start and end in a nontrivial power of b is simply to avoid obvious redundancy. Two simple examples of word polynomials and how they generate inequalities. (later computer assisted searches amount in large part to mechanising these arguments). We fix β and write z for γ(f , g ) and use this as a variable. Then w = bab −1 w = bab −1 ab w = bab −1 a−1 b pw (z) = z(z − β) pw (z) = z(1 + β − z)2 pw (z) = z(1 − 2β + 2z − βz + z 2 ) We indicate how these words are used to describe parts of the parameter space for two-generator Kleinian groups. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Notice the obvious facts ◦ hf , g i Kleinian, implies hf , w (f , g )i discrete. ◦ for any word w = w (f , g ) and m, n ∈ Z, γ(f , f m wf n ) = γ(f , w ) - the requirement the word start and end in a nontrivial power of b is simply to avoid obvious redundancy. Two simple examples of word polynomials and how they generate inequalities. (later computer assisted searches amount in large part to mechanising these arguments). We fix β and write z for γ(f , g ) and use this as a variable. Then w = bab −1 w = bab −1 ab w = bab −1 a−1 b pw (z) = z(z − β) pw (z) = z(1 + β − z)2 pw (z) = z(1 − 2β + 2z − βz + z 2 ) We indicate how these words are used to describe parts of the parameter space for two-generator Kleinian groups. We take for granted the well known fact that the space of discrete non-elementary groups is closed (a general fact concerning groups of isometries of negative curvature). G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Let’s recover Jørgensen’s inequality. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Let’s recover Jørgensen’s inequality. Consider J = inf{ |γ| + |β| : (γ, β, β 0 ) parameters of a Kleinian group}. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Let’s recover Jørgensen’s inequality. Consider J = inf{ |γ| + |β| : (γ, β, β 0 ) parameters of a Kleinian group}. The minimum is attained by Γ = hf , g i. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Let’s recover Jørgensen’s inequality. Consider J = inf{ |γ| + |β| : (γ, β, β 0 ) parameters of a Kleinian group}. The minimum is attained by Γ = hf , g i. Consider Γ0 = hf , gfg −1 i. By minimality, |γ| + |β| ≤ |γ(γ − β)| + |β| 1 ≤ |γ − β| ≤ |γ| + |β| G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Let’s recover Jørgensen’s inequality. Consider J = inf{ |γ| + |β| : (γ, β, β 0 ) parameters of a Kleinian group}. The minimum is attained by Γ = hf , g i. Consider Γ0 = hf , gfg −1 i. By minimality, |γ| + |β| ≤ |γ(γ − β)| + |β| 1 ≤ |γ − β| ≤ |γ| + |β| We need to analyse the zero locus of the polynomial (in this case the variety γ = β) Theorem (Jørgensen’s inequality) Let G = hf , g i be a Kleinian group. Then |γ(f , g )| + |β(f )| ≥ 1 This inequality holds with equality for (2, 3, p)-triangle groups, p ≥ 7. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Jørgensen’s inequality implies that geodesics cannot be too short without making a specific volume contribution. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Jørgensen’s inequality implies that geodesics cannot be too short without making a specific volume contribution. If αf is the axis of f and α̃f its nearest translate (by g ) at complex distance δ + iθ, we compute G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Jørgensen’s inequality implies that geodesics cannot be too short without making a specific volume contribution. If αf is the axis of f and α̃f its nearest translate (by g ) at complex distance δ + iθ, we compute | sinh2 (δ + iθ)| ≥ sinh2 (2r ) ≥ πτ sinh2 (r ) ≥ γ 1 − |β| ≥ β |β| √ 1 − 2|β| 3 − 8πτ ≈ |β| 4πτ √ 3 − 8πτ ≈ 0.108 . . . 16 ≈ follows from a diaphantine analysis on the holonomy (replace f by f n ) and n τ n θ 4πτ min β(f n ) = min 4 sinh2 +i ≈ √ n n 2 2 3 G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. As far as our search for Kleinian group parameters in C3 is concerned, Jørgensen’s inequality tell us that the region {(γ, β, β 0 ) : |γ| + |β| < 1 or |γ| + |β 0 | < 1} contains no parameters for Kleinian groups - eliminating a quantitative neighbourhood of (0, 0, 0) the identity group. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. As far as our search for Kleinian group parameters in C3 is concerned, Jørgensen’s inequality tell us that the region {(γ, β, β 0 ) : |γ| + |β| < 1 or |γ| + |β 0 | < 1} contains no parameters for Kleinian groups - eliminating a quantitative neighbourhood of (0, 0, 0) the identity group. A point to observe here is that in more general situations we must examine and eliminate, for some geometric reason, the zero locus of pw (arithmeticity !) G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. As far as our search for Kleinian group parameters in C3 is concerned, Jørgensen’s inequality tell us that the region {(γ, β, β 0 ) : |γ| + |β| < 1 or |γ| + |β 0 | < 1} contains no parameters for Kleinian groups - eliminating a quantitative neighbourhood of (0, 0, 0) the identity group. A point to observe here is that in more general situations we must examine and eliminate, for some geometric reason, the zero locus of pw (arithmeticity !) Next, if we minimize |γ| + |1 + β| and use the second polynomial z(1 + β − z)2 we see that at the minimum |γ| + |1 + β| ≤ 1 |γ(1 + β − γ)2 | + |1 + β| ≤ |1 + β − γ| G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Lemma Let hf , g i be a Kleinian group. Then |γ| + |1 + β| ≥ 1 unless γ = 1 + β (in which case fg or fg −1 is elliptic of order 3). G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Lemma Let hf , g i be a Kleinian group. Then |γ| + |1 + β| ≥ 1 unless γ = 1 + β (in which case fg or fg −1 is elliptic of order 3). The zero locus we consider is {γ = 1 + β}; groups are Nielsen equivalent to those generated by elliptics of order 2 and 3. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Lemma Let hf , g i be a Kleinian group. Then |γ| + |1 + β| ≥ 1 unless γ = 1 + β (in which case fg or fg −1 is elliptic of order 3). The zero locus we consider is {γ = 1 + β}; groups are Nielsen equivalent to those generated by elliptics of order 2 and 3. As a consequence, if f has order 6, then β = −1 and we have |γ(f , g )| ≥ 1 analogous to Shimitzu-Leutbecher. This quickly gives us another useful result G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Theorem Let Γ be a Kleinian group with an elliptic of order p ≥ 6, then √ vol(H3 /Γ) ≥ vol(H3 /PGL(2, O( −3)) = 0.0846 . . . G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Theorem Let Γ be a Kleinian group with an elliptic of order p ≥ 6, then √ vol(H3 /Γ) ≥ vol(H3 /PGL(2, O( −3)) = 0.0846 . . . It is relatively easy to get a bound bigger than this for torsion free lattices (the manifold case). However, recently Gabai, Meyerhoff and Milley identified the Weeks manifold - (5, 2) and (5, 1) Dehn surgery on the Whitehead link (vol = 0.9427 . . .) - as having minimal volume. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Theorem Let Γ be a Kleinian group with an elliptic of order p ≥ 6, then √ vol(H3 /Γ) ≥ vol(H3 /PGL(2, O( −3)) = 0.0846 . . . It is relatively easy to get a bound bigger than this for torsion free lattices (the manifold case). However, recently Gabai, Meyerhoff and Milley identified the Weeks manifold - (5, 2) and (5, 1) Dehn surgery on the Whitehead link (vol = 0.9427 . . .) - as having minimal volume. So we can focus on lattices which are not torsion free and for which all the elements of finite order lie are 2, 3, 4 or 5. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. We want to consider groups generated by two elliptic elements p and q. In this case β(f ) = −4 sin2 (π/p) and β(g ) = −4 sin2 (π/q) and the free parameter is γ = γ(f , g ) ∈ C. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. We want to consider groups generated by two elliptic elements p and q. In this case β(f ) = −4 sin2 (π/p) and β(g ) = −4 sin2 (π/q) and the free parameter is γ = γ(f , g ) ∈ C. f ∼ cos(π/p) − sin(π/p) sin(π/p) , cos(π/p) g∼ cos(π/q) − sin(π/q)/ω ω sin(π/q) cos(π/q) with |ω| ≤ 1 and γ = −4 sin2 (π/p) sin2 (π/q)(ω − 1/ω)2 G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. We want to consider groups generated by two elliptic elements p and q. In this case β(f ) = −4 sin2 (π/p) and β(g ) = −4 sin2 (π/q) and the free parameter is γ = γ(f , g ) ∈ C. f ∼ cos(π/p) − sin(π/p) sin(π/p) , cos(π/p) g∼ cos(π/q) − sin(π/q)/ω ω sin(π/q) cos(π/q) with |ω| ≤ 1 and γ = −4 sin2 (π/p) sin2 (π/q)(ω − 1/ω)2 Isometric circles of f are |z ± cot(π/p)| = 1/ sin(π/p) and g are |z ± ω cot(π/p)| = |ω|/ sin(π/p) and these are disjoint as soon as |γ| > 4. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. We want to consider groups generated by two elliptic elements p and q. In this case β(f ) = −4 sin2 (π/p) and β(g ) = −4 sin2 (π/q) and the free parameter is γ = γ(f , g ) ∈ C. f ∼ cos(π/p) − sin(π/p) sin(π/p) , cos(π/p) g∼ cos(π/q) − sin(π/q)/ω ω sin(π/q) cos(π/q) with |ω| ≤ 1 and γ = −4 sin2 (π/p) sin2 (π/q)(ω − 1/ω)2 Isometric circles of f are |z ± cot(π/p)| = 1/ sin(π/p) and g are |z ± ω cot(π/p)| = |ω|/ sin(π/p) and these are disjoint as soon as |γ| > 4. Thus hf , g i is free on generators and not a lattice. Also |γ| < 1 − 4 sin2 (π/p) gives hf , g i not discrete. G.J.Martin Minimal co-volume hyperbolic lattices Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups The (2, 3)-commutator plane. COMMUTATOR PARAMETER FOR n= 3, m= 2 γ= 1 1 3 sinh2 (δ + iθ) 0.5 -4 -3 -2 -1 0 1 -0.5 -1 G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Theorem (Gehring Maclachlan Martin Reid - after Borel) Γ generated by elliptics of order p and q with γ = γ(f , g ). Γ is a discrete subgroup of an arithmetic group if and only if γ is the root of a monic polynomial with exactly one complex conjugate pair of roots and all the real roots lie in the interval [−4 sin2 (π/p) sin2 (π/q), 0]. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Theorem (Gehring Maclachlan Martin Reid - after Borel) Γ generated by elliptics of order p and q with γ = γ(f , g ). Γ is a discrete subgroup of an arithmetic group if and only if γ is the root of a monic polynomial with exactly one complex conjugate pair of roots and all the real roots lie in the interval [−4 sin2 (π/p) sin2 (π/q), 0]. Theorem (Maclachlan Martin) There are only finitely many arithmetic Kleinian groups generated by two elements of orders p and q, 2 ≤ p, q ≤ ∞. In particular: only finitely many arithmetic generalised triangle groups. This is contrary to a conjecture of Hilden and Montesinos. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Theorem (Gehring Maclachlan Martin Reid - after Borel) Γ generated by elliptics of order p and q with γ = γ(f , g ). Γ is a discrete subgroup of an arithmetic group if and only if γ is the root of a monic polynomial with exactly one complex conjugate pair of roots and all the real roots lie in the interval [−4 sin2 (π/p) sin2 (π/q), 0]. Theorem (Maclachlan Martin) There are only finitely many arithmetic Kleinian groups generated by two elements of orders p and q, 2 ≤ p, q ≤ ∞. In particular: only finitely many arithmetic generalised triangle groups. This is contrary to a conjecture of Hilden and Montesinos. What are they ? Serious computational number theory, using work of Stark, Odlyzko, Diaz Y Diaz and Olivier on discriminant bounds for fields as well as computer searches, then obtaining co-volume bounds and a topological description. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Theorem There are 41 non-uniform arithmetic lattices p, q-generated. (∞, ∞) 4 two bridge knot & link complements (2, ∞) 6 groups, (4, 6) 1 group (3, ∞), (4, ∞), (2, 3), (2, 4), (2, 6), (3, 6) 3 groups each (6, ∞) (3, 3), (3, 4), (4, 4), (6, 6) 2 groups each There are 18 uniform arithmetic groups with p, q ≥ 6: (6, 6) 14, (8, 8) & (10, 10), 1 each, (12, 12) - 2 G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Theorem There are 41 non-uniform arithmetic lattices p, q-generated. (∞, ∞) 4 two bridge knot & link complements (2, ∞) 6 groups, (4, 6) 1 group (3, ∞), (4, ∞), (2, 3), (2, 4), (2, 6), (3, 6) 3 groups each (6, ∞) (3, 3), (3, 4), (4, 4), (6, 6) 2 groups each There are 18 uniform arithmetic groups with p, q ≥ 6: (6, 6) 14, (8, 8) & (10, 10), 1 each, (12, 12) - 2 Corollary (a conjecture of Montesinos) Let K be (p, 0), (q, 0) orbifold Dehn surgery on a two bridge knot or link (p = q for knots) and K arithmetic. Then p, q ∈ {2, 3, 4, 6, 8, 10, 12} G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Here is the sort of data we pick up: Γ = hf , g i be discrete generated by elliptics of orders p and q and let with δ + iθ be the complex distance between the axes. Then (p = 3, q = 2) (p = 4, q = 2) 0.19707 + i0.78539 arithmetic 0.41572 + i0.59803 arithmetic 0.21084 + i0.33189 arithmetic 0.42698 + i0.44303 arithmetic 0.23371 + i0.49318 arithmetic 0.44068 + i0.78539 arithmetic 0.24486 + i0.67233 arithmetic 0.50495 + i0.67478 arithmetic 0.24809 + i0.40575 arithmetic 0.52254 + i0.34470 arithmetic 0.27407 + i0.61657 arithmetic 0.52979 + i0.24899 arithmetic 0.27465 + i0.78539 arithmetic 0.52979 + i0.53640 arithmetic 0.27702 + i0.56753 arithmetic 0.53063 arithmetic 0.27884 + i0.22832 arithmetic 0.53063 + i0.45227 arithmetic δ > 0.28088 G.J.Martin δ > 0.53264 Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. (p = 3, q = 3) (p = 2, q = 5) 0.39415 + i1.57079 arithmetic 0.4568 arithmetic 0.42168 + i0.66379 arithmetic 0.5306 arithmetic 0.46742 + i0.98637 arithmetic 0.6097 arithmetic 0.48973 + i1.34468 arithmetic 0.6268 arithmetic 0.49619 + i0.81150 arithmetic 0.6514 notarithmetic 0.54814 + i1.23135 arithmetic 0.6717 notarithmetic 0.54930 + i1.57079 arithmetic 0.6949 arithmetic 0.55404 + i1.13507 arithmetic 0.7195 arithmetic 0.55769 + i0.45665 arithmetic 0.7273 arithmetic δ > 0.56177 G.J.Martin δ > 0.73 Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. From this data general position arguments on the placement of elliptic axes a3 b yP SF Q P g SG g g(Q) and then an anlysis of what happens when the axes meet or coincide yields data such as G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Let P and Q be spherical points in a Kleinian group (points stabilised by a finite spherical triangle subgroup) G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Let P and Q be spherical points in a Kleinian group (points stabilised by a finite spherical triangle subgroup) P, Q tetrahedral points. If ρH (P, Q) < 1.026, then P and Q lie on a common axis of order 3 and ρH (P, Q) = 0.64244 or 0.6931 or ρH (P, Q) > 0.7209. Extremals are arithmetic G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Let P and Q be spherical points in a Kleinian group (points stabilised by a finite spherical triangle subgroup) P, Q tetrahedral points. If ρH (P, Q) < 1.026, then P and Q lie on a common axis of order 3 and ρH (P, Q) = 0.64244 or 0.6931 or ρH (P, Q) > 0.7209. Extremals are arithmetic P, Q octahedral points. If ρH (P, Q) < 1.6140, then P and Q lie on a common axis of order 4 and ρH (P, Q) = 1.0595 or 1.0612 or 1.1283 or ρH (P, Q) > 1.14. Extremals are arithmetic G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. Let P and Q be spherical points in a Kleinian group (points stabilised by a finite spherical triangle subgroup) P, Q tetrahedral points. If ρH (P, Q) < 1.026, then P and Q lie on a common axis of order 3 and ρH (P, Q) = 0.64244 or 0.6931 or ρH (P, Q) > 0.7209. Extremals are arithmetic P, Q octahedral points. If ρH (P, Q) < 1.6140, then P and Q lie on a common axis of order 4 and ρH (P, Q) = 1.0595 or 1.0612 or 1.1283 or ρH (P, Q) > 1.14. Extremals are arithmetic P, Q icosahedral points. If ρH (P, Q) < 2.1225, then P and Q lie on a common axis. If this axis has order 5, then ρH (P, Q) = 1.3825 or 1.6169 or ρH (P, Q) > 1.98968. If this axis has order 3, then ρH (P, Q) ≥ 1.9028. Extremals are arithmetic G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. We are now left with the case that there are only Klein 4-groups, these are wound up by discrete groups generated by two loxodromics whose axes meet orthogonally. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. We are now left with the case that there are only Klein 4-groups, these are wound up by discrete groups generated by two loxodromics whose axes meet orthogonally. Then Theorem Let f and g be loxodromic generating a discrete group, the axes of f and g perpendicular and τf and τg translation lengths. Then ! √ 3+1 max{τf , τg } ≥ λ⊥ = arccosh = 0.831446 . . . 2 Equality holds for an arithmetic four fold cover of (4,0) & (2,0) Dehn surgery on the 2 bridge link complement 622 of Rolfsen’s tables. G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. We are now left with the case that there are only Klein 4-groups, these are wound up by discrete groups generated by two loxodromics whose axes meet orthogonally. Then Theorem Let f and g be loxodromic generating a discrete group, the axes of f and g perpendicular and τf and τg translation lengths. Then ! √ 3+1 max{τf , τg } ≥ λ⊥ = arccosh = 0.831446 . . . 2 Equality holds for an arithmetic four fold cover of (4,0) & (2,0) Dehn surgery on the 2 bridge link complement 622 of Rolfsen’s tables. We expect that this represents the extreme case independently of the angle at which the axes of loxodromics meet and that λ⊥ is the Margulis number (for torsion free lattices) G.J.Martin Minimal co-volume hyperbolic lattices Introduction Arithmetic hyperbolic geometry Geometry of discrete groups Two generator groups Parameters for two generator groups Polynomial trace identities Short geodesics Eliminating large torsion 2-torsion and Klein 4 groups. What next ? Extend to identify next few lattices. Identify all the arithmetic generalised triable groups with 2 ≤ p, q ≤ 5 (mostly done). Use polynomial trace identities to reprove (log 3)/2 theorem. Say a lot more about the commutator plane boundary a circle ? relations between different planes? metrics ? Conway notation ↔ pleating rays. Scattering ? identify Margulis constant G.J.Martin Minimal co-volume hyperbolic lattices