Gehring-Hayman Inequality

• Gehring-Hayman inequality is a set of comparison principles that bound the length of intrinsic hyperbolic geodesics by a constant multiple of the length of any competing curve with the same endpoints.
• It underpins studies in quasihyperbolic geometry, Gromov hyperbolicity, uniformization theory, and invariant metrics in several complex variables.
• Its extensions include dimension-free and infinite-dimensional generalizations, providing a robust framework for rigidity and analytic approximation across diverse settings.

The Gehring-Hayman inequality is a family of comparison principles asserting that geodesics, or suitably controlled quasi-geodesics, for an intrinsic hyperbolic-type metric have ambient length bounded by a constant multiple of the length of arbitrary competing curves with the same endpoints. It originated in the 1962 work of Gehring and Hayman for simply connected planar domains and has since become a structural tool in quasihyperbolic geometry, Gromov hyperbolicity, uniformization theory, several complex variables, and analysis on metric measure spaces (Guo et al., 5 Feb 2025, Koskela et al., 2012, Liu et al., 2020).

1. Classical formulation and representative forms

In its classical form, the inequality states that in a suitable domain—classically a simply connected planar domain, and more generally a uniform domain or a domain quasiconformally equivalent to one—the hyperbolic or quasihyperbolic geodesic between two points has Euclidean length at most a constant times the Euclidean length of any other curve joining the same endpoints. A standard formulation is

$\ell(\gamma) \leq b\,\ell(\sigma),$

where $\gamma$ is the distinguished geodesic and $\sigma$ is an arbitrary competitor (Guo et al., 5 Feb 2025).

A closely related formulation replaces arbitrary competitor length by the inner distance $\sigma_\Omega(x,y)$, yielding

$\ell_d(\gamma_{xy}) \leq C\,\sigma_\Omega(x,y),$

for a quasihyperbolic geodesic $\gamma_{xy}$ in a domain $\Omega$ (Guo et al., 12 Sep 2025).

The literature now contains several inequivalent but closely related “Gehring-Hayman type” estimates, depending on the ambient category.

Setting	Distinguished curves	Representative estimate
Classical planar / quasihyperbolic	hyperbolic or quasihyperbolic geodesics	$\ell(\gamma)\le b\,\ell(\sigma)$
Quasihyperbolic domains	quasihyperbolic geodesics	$\ell_d(\gamma_{xy})\le C\,\sigma_\Omega(x,y)$
Strongly pseudoconvex domains	Kobayashi geodesics	$\ell(\gamma_{z,w})\le C|z-w|$
Finite-type pseudoconvex domains in $\gamma$0	Kobayashi $\gamma$1-quasi-geodesics	$\gamma$2
Conformally deformed Gromov hyperbolic spaces	$\gamma$3-short arcs	$\gamma$4
Diameter variant	quasihyperbolic geodesics	$\gamma$5

These variants share a common principle: intrinsic geodesic structure constrains extrinsic geometry. This suggests a unifying interpretation of the inequality as a rigidity statement for hyperbolic-type metrics, although the precise ambient metric and the admissible competitor class depend strongly on context.

2. Quasihyperbolic geometry, ball separation, and Gromov hyperbolicity

A central modern setting is the quasihyperbolic metric

$\gamma$6

where $\gamma$7 and the infimum runs over rectifiable curves in $\gamma$8 (Guo et al., 5 Feb 2025). In this setting, the Gehring-Hayman condition and the ball-separation condition jointly characterize Gromov hyperbolicity in broad classes of domains.

For Ahlfors regular length metric measure spaces, Koskela, Lammi, and Manojlović showed that Gromov hyperbolicity of $\gamma$9 is equivalent to the combination of the Gehring-Hayman condition and ball separation (Koskela et al., 2012). In the notation used later for Euclidean and metric spaces, the two conditions take the form

$\sigma$0

for quasihyperbolic geodesics $\sigma$1, and

$\sigma$2

for every point $\sigma$3 on a quasihyperbolic geodesic and every competing curve $\sigma$4 joining the same endpoints (Guo et al., 12 Sep 2025).

Subsequent work sharpened this characterization in two directions. First, an improved, elementary, measure-independent approach established the equivalence between Gromov hyperbolicity and the conjunction of Gehring-Hayman and ball separation for proper Euclidean subdomains and, in metric form, for proper subdomains of locally compact $\sigma$5-doubling length spaces, while tracking explicit constant dependence (Guo et al., 12 Sep 2025). Second, the same work proved that ball separation alone does not imply the Gehring-Hayman inequality: for every $\sigma$6, there exists a domain in $\sigma$7 with the ball separation property that fails Gehring-Hayman. By contrast, ball separation together with an LLC-2 condition implies inner uniformity and hence the Gehring-Hayman inequality (Guo et al., 12 Sep 2025).

A diameter analogue due to Pommerenke also fits this hyperbolic framework. For a $\sigma$8-hyperbolic domain $\sigma$9, any quasihyperbolic geodesic satisfies

$\sigma_\Omega(x,y)$0

for every connecting curve $\sigma_\Omega(x,y)$1, and if $\sigma_\Omega(x,y)$2 is $\sigma_\Omega(x,y)$3-bounded turning then

$\sigma_\Omega(x,y)$4

This extends Pommerenke’s planar diameter theorem to Gromov hyperbolic domains in $\sigma_\Omega(x,y)$5 (Zhou et al., 2021).

3. Uniformization and boundary-theoretic formulations

The Gehring-Hayman property is tightly linked to uniformization procedures for Gromov hyperbolic spaces. In the Bonk-Heinonen-Koskela framework, given a basepoint $\sigma_\Omega(x,y)$6 in a $\sigma_\Omega(x,y)$7-Gromov hyperbolic space $\sigma_\Omega(x,y)$8, one defines a uniformized metric by

$\sigma_\Omega(x,y)$9

For $\ell_d(\gamma_{xy}) \leq C\,\sigma_\Omega(x,y),$0-roughly starlike $\ell_d(\gamma_{xy}) \leq C\,\sigma_\Omega(x,y),$1-Gromov hyperbolic spaces, the Gehring-Hayman theorem for $\ell_d(\gamma_{xy}) \leq C\,\sigma_\Omega(x,y),$2 is equivalent to the uniformity of $\ell_d(\gamma_{xy}) \leq C\,\sigma_\Omega(x,y),$3, and also equivalent to boundary formulations involving the canonical map from the Gromov boundary to the metric boundary of $\ell_d(\gamma_{xy}) \leq C\,\sigma_\Omega(x,y),$4 and the behavior of Gromov sequences. The boundary, Gromov-sequence, and original Gehring-Hayman statements remain equivalent even without rough starlikeness (Rogovin et al., 2022).

This boundary perspective also yields sharp thresholds. For hyperbolic space $\ell_d(\gamma_{xy}) \leq C\,\sigma_\Omega(x,y),$5, the critical exponent for the uniformized space to remain uniform is $\ell_d(\gamma_{xy}) \leq C\,\sigma_\Omega(x,y),$6; for the model spaces $\ell_d(\gamma_{xy}) \leq C\,\sigma_\Omega(x,y),$7 of constant curvature $\ell_d(\gamma_{xy}) \leq C\,\sigma_\Omega(x,y),$8, it is $\ell_d(\gamma_{xy}) \leq C\,\sigma_\Omega(x,y),$9; for hyperbolic fillings with parameter $\gamma_{xy}$0, it is $\gamma_{xy}$1; and for metric trees the Gehring-Hayman theorem holds for all $\gamma_{xy}$2 (Rogovin et al., 2022).

A parallel intrinsic theory replaces geodesics by $\gamma_{xy}$3-short arcs. For intrinsic Gromov hyperbolic spaces, a continuous density $\gamma_{xy}$4 satisfying

$\gamma_{xy}$5

induces weighted length

$\gamma_{xy}$6

For sufficiently small $\gamma_{xy}$7, every admissible $\gamma_{xy}$8-short arc $\gamma_{xy}$9 between $\Omega$0 and $\Omega$1 satisfies

$\Omega$2

for every rectifiable competitor $\Omega$3. With densities $\Omega$4 coming from Busemann functions, any complete intrinsic hyperbolic space with at least two points in its Gromov boundary can be uniformized, and the Gromov boundary admits a natural identification with the metric boundary of the deformed space (Allu et al., 2024).

The boundary viewpoint extends further to unbounded domains. For unbounded uniform domains in $\Omega$5, Gromov hyperbolicity together with a natural quasisymmetric identification between the Euclidean boundary and the punctured Gromov boundary equipped with a Hamenstädt metric gives a characterization of uniformity; the proof uses Busemann functions and the diameter Gehring-Hayman theorem (Zhou et al., 2021).

4. Several complex variables and invariant metrics

In several complex variables, Gehring-Hayman type estimates are formulated for invariant metrics such as the Kobayashi, Bergman, Carathéodory, and Kähler-Einstein metrics. A foundational result concerns bounded $\Omega$6-convex domains with Dini-smooth boundary and bounded strongly pseudoconvex domains with $\Omega$7-smooth boundary. In these settings, the Euclidean length of a Kobayashi geodesic $\Omega$8 satisfies

$\Omega$9

with $\ell(\gamma)\le b\,\ell(\sigma)$0 in the $\ell(\gamma)\le b\,\ell(\sigma)$1-convex case and $\ell(\gamma)\le b\,\ell(\sigma)$2 in the strongly pseudoconvex case (Liu et al., 2020).

Here $\ell(\gamma)\le b\,\ell(\sigma)$3-convexity means that for some $\ell(\gamma)\le b\,\ell(\sigma)$4,

$\ell(\gamma)\le b\,\ell(\sigma)$5

where $\ell(\gamma)\le b\,\ell(\sigma)$6 is Euclidean distance to the boundary and $\ell(\gamma)\le b\,\ell(\sigma)$7 is distance to the boundary along the complex line through $\ell(\gamma)\le b\,\ell(\sigma)$8 in direction $\ell(\gamma)\le b\,\ell(\sigma)$9. Dini-smoothness is expressed by the Dini condition

$\ell_d(\gamma_{xy})\le C\,\sigma_\Omega(x,y)$0

for the modulus of continuity of the inner normal vector (Liu et al., 2020).

When $\ell_d(\gamma_{xy})\le C\,\sigma_\Omega(x,y)$1 is Gromov hyperbolic, the same estimate extends to $\ell_d(\gamma_{xy})\le C\,\sigma_\Omega(x,y)$2-quasi-geodesics. On uniformly squeezing domains—defined by the existence, for every $\ell_d(\gamma_{xy})\le C\,\sigma_\Omega(x,y)$3, of a holomorphic embedding $\ell_d(\gamma_{xy})\le C\,\sigma_\Omega(x,y)$4 with $\ell_d(\gamma_{xy})\le C\,\sigma_\Omega(x,y)$5 and

$\ell_d(\gamma_{xy})\le C\,\sigma_\Omega(x,y)$6

for some fixed $\ell_d(\gamma_{xy})\le C\,\sigma_\Omega(x,y)$7—the Bergman, Carathéodory, and Kähler-Einstein metrics are bilipschitz equivalent to the Kobayashi metric, so the same Gehring-Hayman type bound holds for quasi-geodesics in those metrics as well (Liu et al., 2020).

These estimates have boundary-theoretic consequences. Under the same hypotheses, the identity map extends to a bi-Hölder homeomorphism

$\ell_d(\gamma_{xy})\le C\,\sigma_\Omega(x,y)$8

where $\ell_d(\gamma_{xy})\le C\,\sigma_\Omega(x,y)$9 is a visual metric on the Gromov boundary, and biholomorphisms or more general rough quasi-isometries with respect to the Kobayashi metrics extend continuously to the boundary with bi-Hölder boundary maps (Liu et al., 2020).

Later work strengthened these results in important special cases. For bounded strongly pseudoconvex domains with $\ell(\gamma_{z,w})\le C|z-w|$0-smooth boundary, every Kobayashi geodesic $\ell(\gamma_{z,w})\le C|z-w|$1 satisfies the linear estimate

$\ell(\gamma_{z,w})\le C|z-w|$2

established using scaling methods, Lempert theory, and visibility arguments stemming from Gromov hyperbolicity (Kosiński et al., 2023). For smoothly bounded pseudoconvex domains of finite type $\ell(\gamma_{z,w})\le C|z-w|$3 in $\ell(\gamma_{z,w})\le C|z-w|$4, every Kobayashi $\ell(\gamma_{z,w})\le C|z-w|$5-quasi-geodesic $\ell(\gamma_{z,w})\le C|z-w|$6 satisfies

$\ell(\gamma_{z,w})\le C|z-w|$7

and this is applied to a quantitative comparison between global and local Kobayashi distances near boundary points (Li et al., 2023).

5. Dimension-free and infinite-dimensional generalizations

A major recent development is the removal of dimension dependence from quasihyperbolic Gehring-Hayman estimates. Heinonen and Rohde had proved in 1993 that if $\ell(\gamma_{z,w})\le C|z-w|$8 is quasiconformally equivalent to a uniform domain, then quasihyperbolic geodesics satisfy a Gehring-Hayman inequality with a multiplicative constant depending on the dimension. A 2025 result replaced this with a dimension-free constant, generalized geodesics to quasigeodesics, and weakened quasiconformal equivalence to coarsely quasihyperbolic equivalence (Guo et al., 5 Feb 2025).

In this setting, a $\ell(\gamma_{z,w})\le C|z-w|$9-quasigeodesic is a curve $\gamma$00 such that for all $\gamma$01 on $\gamma$02,

$\gamma$03

where $\gamma$04 is quasihyperbolic length. If $\gamma$05 is $\gamma$06-CQH-homeomorphic to a uniform domain, then any $\gamma$07-quasigeodesic $\gamma$08 and any competing curve $\gamma$09 with the same endpoints satisfy

$\gamma$10

and the constant $\gamma$11 is independent of the ambient dimension (Guo et al., 5 Feb 2025).

The relevant map class is that of $\gamma$12-coarsely quasihyperbolic homeomorphisms: $\gamma$13 This strictly contains the quasiconformal class, so the domain class is correspondingly broader (Guo et al., 5 Feb 2025).

The same paper extends the theory to Banach spaces, where strict quasihyperbolic geodesics may fail to exist but quasigeodesics still do. In a Banach space $\gamma$14, if $\gamma$15 is homeomorphic to a uniform domain via a CQH mapping, then the dimension-free Gehring-Hayman inequality holds for quasigeodesics in $\gamma$16. The proof uses a contradiction-compactness argument, recursive construction of good subcurves, and a compactness theorem for $\gamma$17-pairs, while avoiding Whitney-cube and modulus arguments that are inherently finite-dimensional (Guo et al., 5 Feb 2025). This gives an affirmative answer to the open problem raised by Heinonen and Rohde in 1993 and reformulated by Väisälä in 2005 (Guo et al., 5 Feb 2025).

6. Analytic consequences and further extensions

The Gehring-Hayman condition also functions as a geometric hypothesis for analytic approximation theorems. In PI spaces, a bounded domain $\gamma$18 is called a $\gamma$19-GHS domain if it satisfies ball separation, the Gehring-Hayman condition

$\gamma$20

for quasihyperbolic geodesics $\gamma$21, and $\gamma$22 is a locally compact length space. For such domains, the Newtonian Sobolev space $\gamma$23 is dense in $\gamma$24 for $\gamma$25. The Gehring-Hayman condition enters through Whitney-type decompositions, chain lemmas, and control of the length of quasihyperbolic geodesics used in partition-of-unity constructions (Koivu, 25 Nov 2025).

There are also function-theoretic extensions closer to the original conformal-mapping setting. For meromorphic univalent functions $\gamma$26 on the unit disk with a simple pole at $\gamma$27 and continuous extension to the left half of the unit circle, the ratio of the length of the image of the vertical diameter $\gamma$28 to the length of the image of $\gamma$29 is bounded by a constant depending only on $\gamma$30. Writing $\gamma$31 for the best such constant, one has explicit bounds

$\gamma$32

and the result extends to arbitrary hyperbolic geodesics and Jordan arcs with a bound controlled by the pole’s hyperbolic position relative to the geodesic. This proves a conjecture of Bhowmik and Maity and extends the classical Gehring-Hayman inequality to meromorphic univalent mappings (Bhowmik et al., 10 Dec 2025).

Across these settings, the Gehring-Hayman inequality serves less as a single theorem than as a robust paradigm: it links intrinsic hyperbolic metrics to ambient geometry, characterizes hyperbolicity in tandem with ball separation, interacts with boundary identification and uniformization, and supplies the geometric input for extension and density theorems. The current literature shows that this paradigm remains effective in Euclidean domains, complex domains, intrinsic Gromov hyperbolic spaces, metric measure spaces, and Banach spaces, with the main distinctions arising from the choice of intrinsic metric, the admissible curve class, and the regularity assumptions imposed on the ambient space (Guo et al., 12 Sep 2025, Liu et al., 2020, Guo et al., 5 Feb 2025).