Dehn surgery and negatively curved 3-manifolds

Daryl Cooper, Marc Lackenby

This paper, titled "Dehn surgery and negatively curved 3-manifolds," is a significant contribution to low-dimensional topology. It explores the deep relationship between the topological process of Dehn surgery and the geometric properties of negative curvature in 3-manifolds.

TL;DR

The authors prove that any given 3-manifold can be constructed in only finitely many ways via "long" Dehn surgery on hyperbolic manifolds. This provides a partial solution to a long-standing mystery in Kirby’s problem list by showing that a fixed manifold can be obtained from $S^3$ by surgery on only finitely many hyperbolic knots, provided the surgery coefficient is sufficiently large ($|q| > 22$).

What Problem Does This Paper Address?

Dehn surgery is the primary method for constructing 3-manifolds: you drill out a neighborhood of a knot (a solid torus) and glue it back in according to a specific "slope" (a fraction $p/q$). Despite its ubiquity, fundamental questions remain. Specifically, Kirby’s Problem 3.6 (D) asks: Does there exist a 3-manifold that can be obtained from the 3-sphere ($S^3$) by surgery along an infinite number of distinct knots?

Before this paper, it was unclear if the "shape" of the knot's complement (its hyperbolic geometry) could be used to bound the number of such constructions. The authors tackle this by investigating how the geometry of the original manifold (the "knot complement") constrains the topology of the manifold resulting from surgery.

Methods

The authors bridge topology and geometry using the following key techniques:

  • The ‘2π’ Theorem: They build upon a well-known result by Gromov and Thurston, which states that if you perform Dehn filling on a hyperbolic manifold along slopes with length $> 2\pi$, the resulting manifold admits a negatively curved metric.
  • Geometric Estimates: The core of the paper involves establishing "controlled" versions of the 2π theorem. The authors derive precise estimates relating the volume, sectional curvature, and the Gromov norm (a topological invariant) of the manifold before and after surgery.
  • Geometric Convergence: They use the theory of convergence for hyperbolic manifolds. By showing that the volumes of potential "parent" manifolds are bounded, they can apply compactness results to prove that only finitely many such manifolds can exist for a given target $M$.
  • Slope Length Analysis: They establish lower bounds on the geometric length of surgery slopes based on topological data, such as the intersection number (distance) between slopes.

Datasets

Not discussed in the paper. As a theoretical mathematics paper, it relies on logical proofs and existing theorems rather than empirical datasets.

Key Insights and Ideas

  • Geometry as a Finiteness Filter: The most profound insight is that the existence of a negatively curved metric with sectional curvatures bounded away from zero (the $2\pi + \epsilon$ condition) strictly limits the possible underlying hyperbolic manifolds.
  • Volume and Gromov Norm Parity: The authors show that the Gromov norm of a manifold obtained by Dehn filling is very close to the Gromov norm of the original hyperbolic manifold when the slopes are long. Since the Gromov norm is a topological invariant, this "locks" the volume of the original manifold.
  • The "Almost Hyperbolic" Characterization: In Section 6, the authors introduce the concept of "almost hyperbolic" manifolds—those admitting metrics with sectional curvatures arbitrarily close to -1. They prove a "rigidity" result: any such manifold must actually possess a complete, finite-volume hyperbolic structure.
  • Intersection Distance vs. Length: They prove a lemma (Lemma 2.1) showing that if the topological distance between two slopes is large, at least one of them must be geometrically "long," providing a bridge between discrete topology and continuous geometry.

Key Results

  • Theorem 4.1 (The Main Result): For a fixed 3-manifold $M$ and a constant $\epsilon > 0$, there are only finitely many hyperbolic manifolds $X$ and sets of slopes $s_i$ (each with length $\ge 2\pi + \epsilon$) such that Dehn filling $X$ along $s_i$ yields $M$.
  • Corollary 4.5 (Knot Surgery): For any closed orientable 3-manifold $M$, there are at most finitely many hyperbolic knots $K$ in $S^3$ and fractions $p/q$ (with $|q| > 22$) such that $M$ is obtained by $p/q$-surgery on $K$.
  • Theorem 6.2: A compact orientable 3-manifold is hyperbolic if and only if it is "almost hyperbolic."

Limitations and Future Work

The primary limitation is the $2\pi + \epsilon$ condition. The finiteness results only apply to "long" surgeries. While it is known that only a finite number of slopes on any given torus are shorter than $2\pi$, the paper does not rule out the possibility of an infinite number of "short" surgeries yielding the same manifold (though this is generally considered unlikely).

The specific bound of $|q| > 22$ in the knot surgery corollary is derived from the geometric estimates in the paper. Future work might aim to lower this bound or remove the $\epsilon$ dependency entirely by further refining the relationship between volume and curvature in the "pinched" negative curvature setting.

Summary updated

Abstract

We show that, for any given 3-manifold M, there are at most finitely many hyperbolic knots K in the 3-sphere and fractions p/q (with q > 22), such that M is obtained by p/q surgery along K. This is a corollary of the following result. If M is obtained by Dehn filling the cusps of a hyperbolic 3-manifold X, where each filling slope has length more than 2 \pi + \epsilon, then, for any given M and \epsilon > 0, there are only finitely many possibilities for X and for the filling slopes. In this paper, we also investigate the length of boundary slopes, and sequences of negatively curved metrics on a given 3-manifold.
BibTeX
@article{cooper1998dehn,
  author = {Daryl Cooper and Marc Lackenby},
  title = {{Dehn surgery and negatively curved 3-manifolds}},
  journal = {Journal of Differential Geometry},
  year = {1998},
  doi = {10.4310/jdg/1214424971},
}

Loading related content...

Figure