In the last of the twenty-four questions that closed his 1982 survey Three-dimensional manifolds, Kleinian groups and hyperbolic geometry (Bull. Amer. Math. Soc. 6 (1982), 357–381), Thurston asked to "show that volumes of hyperbolic -manifolds are not all rationally related" (p. 380). Twenty-two of the twenty-four have since been answered — geometrization by Perelman, tameness by Agol and by Calegari–Gabai, the ending lamination conjecture by Brock–Canary–Minsky, virtual fibering by Agol — and this one is among the two that remain open.
Some rational relations are forced, and for a trivial reason: a degree cover of a hyperbolic -manifold has times its volume, so any two commensurable manifolds have rationally related volumes. The question, which remains open, is whether every rational relation arises that way — equivalently, whether some two hyperbolic -manifolds have irrational volume ratio. Remarkably, not a single such pair is known.
The bundle fixes the meaning of every term. Hyperbolic -space is the upper half-space . Its volume is Lebesgue measure with density — the Riemannian volume of the metric written out, so that no Riemannian machinery is required. The hyperbolic distance is given by its closed formula
A Kleinian action is a free, properly discontinuous action by hyperbolic
isometries; the quotient is a complete hyperbolic -manifold, discreteness and
torsion freeness being consequences rather than hypotheses. The volume of the
quotient is the measure of a fundamental domain, in the sense of Mathlib's
MeasureTheory.IsFundamentalDomain, and the set of volumes collects those that
are finite and positive.
Two conventions are stated rather than derived, and are worth flagging. Isometries are not required to preserve orientation, so the set of volumes also contains those of non-orientable quotients; this enlarges the set but not its -span, so neither goal is affected. And preservation of the hyperbolic volume is a field of the structure rather than a consequence of preserving the distance: it holds for every hyperbolic isometry, but deriving it amounts to classifying , which is not the subject of this mission.
The goal is that the volumes are not all rationally related: there are two of them, and , with for every rational .
Two milestones support it. The first is that passing to a subgroup of index multiplies the volume by , a fundamental domain for the subgroup being the union of translates of one for the whole group; this is the source of every known rational relation, and it is why the question is phrased as it is. The second is that the set of volumes is nonempty — that some finite-volume hyperbolic -manifold exists at all — without which the goal would be vacuously false rather than open.
A stronger form of the question, that the -span of the set of volumes is infinite dimensional, is also stated.
The question is a geometric statement whose difficulty is arithmetic. For the Bianchi groups of an imaginary quadratic field , Humbert's formula gives the covolume as , so the ratio of two such volumes is, up to explicit algebraic factors, a ratio of Dedekind zeta values at ; and Neumann and Yang showed that the Bloch invariant of a hyperbolic -manifold lies in a subgroup of finite -rank determined by its invariant trace field, so that manifolds sharing an invariant trace field with a single complex place, such as an imaginary quadratic one, have rationally related volumes. Producing one irrational ratio therefore means separating two such transcendentals — a statement of the same order of difficulty as the irrationality of . The value of formalizing the question is not that it will be closed, but that its statement, and the elementary relations that make its naive form false, are pinned down exactly.
namespace Thurston23
open MeasureTheory
/-- **Thurston's Question 23.** The volumes of hyperbolic `3`-manifolds are not
all rationally related: some two of them have irrational ratio.
Read literally, "the volumes are rationally independent" is false, since a
degree `n` cover has `n` times the volume (Milestone 1); the open question is
whether every rational relation arises that way. It is not known how to exhibit
a single pair of hyperbolic `3`-manifolds whose volumes have irrational ratio.
For arithmetic examples Humbert's formula expresses the volume of the Bianchi
quotient of an imaginary quadratic field `F` as `|δ_F|^{3/2} ζ_F(2) / 4π²`, so
the question reduces to the irrationality of a ratio of Dedekind zeta values at
`2`, a transcendence statement of the same difficulty as the irrationality of
`ζ(5)`. -/
theorem thurston_question_23 :
∃ v ∈ hyperbolicVolumes, ∃ w ∈ hyperbolicVolumes,
∀ q : ℚ, v ≠ (q : ℝ) * w := by
sorry
end Thurston23
There are two hyperbolic 3-manifolds whose volumes have irrational ratio. Read literally, 'the volumes are rationally independent' is false, since a degree n cover has n times the volume; the open question is whether every rational relation arises that way. No such pair is known: for arithmetic examples Humbert's formula makes the ratio a ratio of Dedekind zeta values at 2, and separating those is a transcendence statement of the difficulty of the irrationality of zeta(5).