Bi-John Domains: Two-Sided Geometric Control

• Bi-John domains are domains defined by bilateral (two-sided) John-type control, where every curve between points satisfies both interior thickness and distance conditions.
• They integrate techniques such as carrot/cigar constructions and quasihyperbolic metrics to bridge the gap between one-sided John domains and uniform or chord‐arc domains.
• This framework has analytic implications, supporting mapping theory, Poincaré inequalities, and boundary prime-end analyses in geometric function theory.

Searching arXiv for recent and relevant papers on John domains, quasihyperbolic geometry, and related “Bi-John” interpretations. Bi-John domains do not appear as a single standardized class in the cited literature. Instead, the term is typically understood through bilateral or two-sided John-type control: within the usual geometric-analysis literature, “bi-John” (or “two-sided John”) normally refers to domains that are John in a bilateral sense, while closely related papers work with uniform, diameter John, distance John, carrot John, cigar John, inner uniform, and quasihyperbolic $(b,\lambda)$-uniform domains (García, 2016, Huang et al., 2011). A natural synthesis suggested by these works is that a Bi-John domain is a domain in which the one-sided John condition is supplemented by two-sided control along curves between arbitrary pairs of points, often together with quasihyperbolic or inner-uniform control (Huang et al., 2011).

1. Foundational definitions and the emergence of a bilateral viewpoint

A bounded domain $\Omega \subset \mathbb{R}^n$ is a John domain with parameter $\beta>1$ if there exists a point $x_0\in \Omega$ such that for every $y\in\Omega$ there is a rectifiable curve $\gamma:[0,\ell]\to\Omega$, parametrized by arc length, with $\gamma(0)=y$, $\gamma(\ell)=x_0$, and

$$\operatorname{dist}\bigl(\gamma(t), \partial\Omega\bigr) \ge \frac{t}{\beta} \quad\text{for all } t\in[0,\ell].$$

Equivalent formulations replace the distinguished center by a pairwise cone condition: a rectifiable arc $\gamma$ joining $\Omega \subset \mathbb{R}^n$0 is a $\Omega \subset \mathbb{R}^n$1-cone arc if

$\Omega \subset \mathbb{R}^n$2

where $\Omega \subset \mathbb{R}^n$3 (García, 2016, Huang et al., 2011).

A uniform domain strengthens this by adding global length control. In the formulation used by Huang–Wang, $\Omega \subset \mathbb{R}^n$4 is $\Omega \subset \mathbb{R}^n$5-uniform if every pair $\Omega \subset \mathbb{R}^n$6 can be joined by a rectifiable arc $\Omega \subset \mathbb{R}^n$7 satisfying the same double-cone condition and also

$\Omega \subset \mathbb{R}^n$8

The heuristic recorded there is especially important for the present topic: $\Omega \subset \mathbb{R}^n$9 whereas

$\beta>1$0

This heuristic is one of the main sources for the modern interpretation of “Bi-John” behavior (Huang et al., 2011).

The literature also isolates three pairwise variants. In Banach spaces, a length $\beta>1$1-John domain is defined by the cone condition in terms of arc length, a diameter $\beta>1$2-John domain replaces $\beta>1$3 by $\beta>1$4, and a distance $\beta>1$5-John domain replaces it by $\beta>1$6. The basic implication chain is

$\beta>1$7

(Allu et al., 2023).

For simply connected planar domains, an intrinsic two-sided formulation is already standard. A simply connected planar domain $\beta>1$8 is a $\beta>1$9-John disk if any pair $x_0\in \Omega$0 can be joined by a rectifiable curve $x_0\in \Omega$1 such that

$x_0\in \Omega$2

In that setting, the basepoint formulation and the two-sided formulation are equivalent up to a change of constant (Koskela et al., 2020).

2. Candidate meanings of “Bi-John” and the role of carrot and cigar structures

The cited papers repeatedly state that the term itself is not standard. One formulation given explicitly is that, within the usual geometric-analysis literature, “bi-John” normally refers to domains that are John in a bilateral sense: the domain is John with respect to every point, or satisfies a John condition in both directions, or is both John and reverse John; in many settings such domains coincide, up to constants, with uniform domains or chord-arc domains (García, 2016).

A second, more geometric formulation emerges from carrot and cigar constructions. For a curve $x_0\in \Omega$3 joining $x_0\in \Omega$4 to a center $x_0\in \Omega$5, the carrot set is

$x_0\in \Omega$6

and a domain is carrot John if each point admits such a curve with the carrot contained in the domain. For a curve $x_0\in \Omega$7 joining two points $x_0\in \Omega$8, the cigar set is

$x_0\in \Omega$9

with

$y\in\Omega$0

In bounded domains, carrot and cigar definitions are equivalent up to constants. The cigar condition is intrinsically two-sided and is therefore especially close to a Bi-John interpretation (Su et al., 2024).

The unbounded case is more delicate. Carrot John with center at $y\in\Omega$1 is not equivalent to cigar John, but an unbounded carrot John domain can be covered by a uniformly finite number of unbounded John domains defined conventionally through cigars. Moreover, within each such component, any two points admit two carrot curves meeting a common interior ball: $y\in\Omega$2 with equal terminal radii

$y\in\Omega$3

This is an explicit two-sided internal thickness statement and provides one of the clearest geometric models of Bi-John behavior in the unbounded setting (Su et al., 2024).

A third model comes from the planar John disk theory. Since a John disk is defined there by a pairwise two-sided inequality, it may be regarded as a one-sided quasidisk, but at the level of internal curves it already exhibits the bilateral control that motivates the term Bi-John (Koskela et al., 2020).

3. Quasihyperbolic and inner-uniform formulations of two-sided John behavior

The quasihyperbolic metric provides the main metric reformulation of two-sided John geometry. For a rectifiable curve $y\in\Omega$4,

$y\in\Omega$5

and

$y\in\Omega$6

The inner distance is

$y\in\Omega$7

A domain is quasihyperbolic $y\in\Omega$8-uniform if

$y\in\Omega$9

Väisälä’s theorem identifies this with inner uniformity, while Gehring–Osgood characterize uniform domains by a corresponding quasihyperbolic inequality (Huang et al., 2011).

Huang–Wang prove the central two-sided geodesic theorem in this direction. If $\gamma:[0,\ell]\to\Omega$0 is an $\gamma:[0,\ell]\to\Omega$1-John domain which is homeomorphic to a $\gamma:[0,\ell]\to\Omega$2-uniform domain $\gamma:[0,\ell]\to\Omega$3 via a $\gamma:[0,\ell]\to\Omega$4-quasiconformal mapping, then every quasihyperbolic geodesic $\gamma:[0,\ell]\to\Omega$5 in $\gamma:[0,\ell]\to\Omega$6 joining arbitrary points $\gamma:[0,\ell]\to\Omega$7 is an $\gamma:[0,\ell]\to\Omega$8-cone arc, where $\gamma:[0,\ell]\to\Omega$9 depends only on $\gamma(0)=y$0. Under the same hypotheses, $\gamma(0)=y$1 is a quasihyperbolic $\gamma(0)=y$2-uniform domain, and hence an inner uniform domain (Huang et al., 2011).

These results are the most explicit justification in the cited literature for interpreting Bi-John domains quasihyperbolically. A natural interpretation proposed there is that a Bi-John domain should exhibit John-type control “in both directions” between arbitrary pairs of points: every quasihyperbolic geodesic is a cone arc, and the quasihyperbolic metric is controlled by inner distance. In that sense,

$\gamma(0)=y$3

which is “very close” to a Bi-John condition (Huang et al., 2011).

The same paper also records the coarse quasihyperbolic stability mechanism. A homeomorphism $\gamma(0)=y$4 is $\gamma(0)=y$5-CQH if

$\gamma(0)=y$6

and any $\gamma(0)=y$7-quasiconformal map in $\gamma(0)=y$8 is $\gamma(0)=y$9-CQH for $\gamma(\ell)=x_0$0 depending only on $\gamma(\ell)=x_0$1 and $\gamma(\ell)=x_0$2. This is the basic transfer principle behind quasihyperbolic Bi-John behavior (Huang et al., 2011).

4. Analytic consequences on John and stronger-than-John classes

Because every plausible Bi-John class in the cited papers contains John geometry, the analytic theory for John domains applies directly and often serves as the baseline.

For bounded Euclidean John domains, a weighted Korn inequality holds with nonnegative powers of the distance to the boundary. If $\gamma(\ell)=x_0$3, $\gamma(\ell)=x_0$4, $\gamma(\ell)=x_0$5, and $\gamma(\ell)=x_0$6 satisfies the weighted orthogonality conditions on $\gamma(\ell)=x_0$7, then

$\gamma(\ell)=x_0$8

where $\gamma(\ell)=x_0$9 is the Boman-tree geometric constant. The same paper proves weighted solvability of

$$\operatorname{dist}\bigl(\gamma(t), \partial\Omega\bigr) \ge \frac{t}{\beta} \quad\text{for all } t\in[0,\ell].$$0

in $$\operatorname{dist}\bigl(\gamma(t), \partial\Omega\bigr) \ge \frac{t}{\beta} \quad\text{for all } t\in[0,\ell].$$1 with explicit dependence on $$\operatorname{dist}\bigl(\gamma(t), \partial\Omega\bigr) \ge \frac{t}{\beta} \quad\text{for all } t\in[0,\ell].$$2. Since any bi-John domain is, at minimum, John, these inequalities apply automatically to stricter two-sided classes (García, 2016).

The Poincaré theory is similarly extensive. For fractional Orlicz–Sobolev spaces, if $$\operatorname{dist}\bigl(\gamma(t), \partial\Omega\bigr) \ge \frac{t}{\beta} \quad\text{for all } t\in[0,\ell].$$3 with $$\operatorname{dist}\bigl(\gamma(t), \partial\Omega\bigr) \ge \frac{t}{\beta} \quad\text{for all } t\in[0,\ell].$$4, then a bounded $$\operatorname{dist}\bigl(\gamma(t), \partial\Omega\bigr) \ge \frac{t}{\beta} \quad\text{for all } t\in[0,\ell].$$5-John domain supports the $$\operatorname{dist}\bigl(\gamma(t), \partial\Omega\bigr) \ge \frac{t}{\beta} \quad\text{for all } t\in[0,\ell].$$6-Poincaré inequality

$$\operatorname{dist}\bigl(\gamma(t), \partial\Omega\bigr) \ge \frac{t}{\beta} \quad\text{for all } t\in[0,\ell].$$7

Conversely, under the additional assumption that $$\operatorname{dist}\bigl(\gamma(t), \partial\Omega\bigr) \ge \frac{t}{\beta} \quad\text{for all } t\in[0,\ell].$$8 is quasiconformally equivalent to a uniform domain when $$\operatorname{dist}\bigl(\gamma(t), \partial\Omega\bigr) \ge \frac{t}{\beta} \quad\text{for all } t\in[0,\ell].$$9, or simply connected when $\gamma$0, support of the same inequality implies that $\gamma$1 is John (Feng et al., 2023). The first-order Orlicz analogue replaces $\gamma$2 by $\gamma$3 and yields

$\gamma$4

again with the same converse under the same geometric hypotheses (Feng et al., 2024).

Improved fractional Poincaré and Hardy–Sobolev inequalities also fit the same pattern. On John domains in doubling metric measure spaces, one has weighted improved fractional Poincaré inequalities of the form

$\gamma$5

together with the $\gamma$6 variant and endpoint $\gamma$7 forms under additional regularity on $\gamma$8 (Cejas et al., 2019). For unbounded John domains, weighted fractional Hardy–Sobolev inequalities with distance-to-boundary powers are proved under Assouad-dimension restrictions on $\gamma$9, and this framework yields the fractional Hardy–Sobolev–Maz’ya inequality on the half-space (Dyda et al., 2017).

A further consequence appears in anisotropic GMT. Every $\Omega \subset \mathbb{R}^n$00-minimizer for the anisotropic perimeter $\Omega \subset \mathbb{R}^n$01 satisfies a local John property: each connected component is a $\Omega \subset \mathbb{R}^n$02-John domain for the norm $\Omega \subset \mathbb{R}^n$03, with $\Omega \subset \mathbb{R}^n$04 depending on $\Omega \subset \mathbb{R}^n$05 and $\Omega \subset \mathbb{R}^n$06, and $\Omega \subset \mathbb{R}^n$07 depending on $\Omega \subset \mathbb{R}^n$08 and $\Omega \subset \mathbb{R}^n$09. Under a global closeness assumption to the Wulff shape $\Omega \subset \mathbb{R}^n$10, the set is globally John and satisfies an anisotropic trace inequality

$\Omega \subset \mathbb{R}^n$11

This furnishes a concrete geometric input for the quantitative Wulff inequality (Su et al., 2024).

5. Mapping theory, prime ends, and metric boundary structure

The mapping-theoretic invariance results sharpen the distinction between one-sided and two-sided John conditions. In Banach spaces, a length $\Omega \subset \mathbb{R}^n$12-John domain always has the minimizing property, and curves with the minimizing property are quantitatively equivalent to diameter John arcs. A distance $\Omega \subset \mathbb{R}^n$13-John domain is characterized by the weak minimizing property (Allu et al., 2023).

These characterizations interact well with quasisymmetry relative to the boundary. If $\Omega \subset \mathbb{R}^n$14 is $\Omega \subset \mathbb{R}^n$15-quasisymmetric relative to $\Omega \subset \mathbb{R}^n$16, then the minimizing property is preserved, so a length John domain is mapped to a diameter John domain. Likewise, the weak minimizing property is preserved, so distance John domains are invariant under such maps (Allu et al., 2023). The full length John property requires an additional quasihyperbolic control: if $\Omega \subset \mathbb{R}^n$17 is $\Omega \subset \mathbb{R}^n$18-CQH and the boundary extension is $\Omega \subset \mathbb{R}^n$19-QS relative to $\Omega \subset \mathbb{R}^n$20, then the target is again a John domain (Allu et al., 2023).

For planar mapping theory, John disks are a particularly important endpoint. A John disk is a simply connected John domain and may be regarded as a one-sided quasidisk. The paper on Sobolev homeomorphic extensions proves that if $\Omega \subset \mathbb{R}^n$21 is a John disk and $\Omega \subset \mathbb{R}^n$22 is a homeomorphism, then there exists a homeomorphic extension

$\Omega \subset \mathbb{R}^n$23

with $\Omega \subset \mathbb{R}^n$24 for all $\Omega \subset \mathbb{R}^n$25. This is obtained by transferring extension theory from quasidisks to John disks via the fact that $\Omega \subset \mathbb{R}^n$26 is bi-Lipschitz equivalent to a quasidisk with Euclidean metric (Koskela et al., 2020). A plausible implication is that genuinely bilateral, or “Bi-John pair,” extension problems should admit an analogous disk-conjugation strategy, although that case is explicitly identified there as extrapolative rather than proved (Koskela et al., 2020).

The prime-end theory in metric spaces further clarifies boundary accessibility. The paper on prime ends introduces almost John domains and proves that almost John domains are finitely connected at the boundary. In such domains, every prime end has singleton impression, every boundary point is accessible and is the impression of at least one prime end, and the prime-end boundary is homeomorphic to the Mazurkiewicz boundary; for $\Omega \subset \mathbb{R}^n$27, Mod$\Omega \subset \mathbb{R}^n$28-prime ends coincide with prime ends (Adamowicz et al., 2012). Since these conclusions use only the interior John geometry, they apply directly to any Bi-John domain that retains the same interior John structure.

6. Stability, examples, sharpness, and open directions

Several papers emphasize that “Bi-John” is not a fixed standard term and that stronger two-sided conditions are delicate. The weighted eikonal paper proves that if $\Omega \subset \mathbb{R}^n$29 is a bounded John domain and $\Omega \subset \mathbb{R}^n$30 is continuous, bounded, and uniformly positive, then every superlevel

$\Omega \subset \mathbb{R}^n$31

of the unique viscosity solution of

$\Omega \subset \mathbb{R}^n$32

is again a John domain, with a John constant independent of $\Omega \subset \mathbb{R}^n$33. The same paper gives counterexamples showing that John regularity is sharp: one cannot upgrade the conclusion to interior-ball or even interior-cone regularity in this generality (Davoli et al., 2023). This sharply limits any attempt to replace John by much stronger bilateral classes in low-regularity PDE evolutions.

Quantitative stability of John geometry under set convergence is established for carrot John domains. If $\Omega \subset \mathbb{R}^n$34 are uniformly bounded John domains with $\Omega \subset \mathbb{R}^n$35 and a uniform lower volume bound, then after subselection their closures converge in Hausdorff distance to a set $\Omega \subset \mathbb{R}^n$36 whose interior $\Omega \subset \mathbb{R}^n$37 is connected, John, and satisfies

$\Omega \subset \mathbb{R}^n$38

The same work shows that an unbounded open set satisfying the carrot John condition with center at $\Omega \subset \mathbb{R}^n$39 can be covered by a uniformly finite number of unbounded John domains defined conventionally through cigars, and these domains support Sobolev–Poincaré inequalities (Su et al., 2024). For Bi-John theory, this is one of the strongest available stability templates.

Explicit fractal examples make the distinction between John, QHBC, and stronger two-sided control concrete. The Cantor-dust domain $\Omega \subset \mathbb{R}^n$40 and the generalized von Koch snowflake $\Omega \subset \mathbb{R}^n$41 satisfy the quasihyperbolic boundary condition with explicitly computed admissible thresholds, and they are also John domains with explicit John constants: $\Omega \subset \mathbb{R}^n$42 while

$\Omega \subset \mathbb{R}^n$43

These families provide quantitative laboratories for testing whether bilateral John behavior survives increasing fractal complexity (Harjulehto et al., 2012).

The open directions recorded in the cited papers are largely questions about how much two-sided control can be recovered from one-sided John geometry plus analytic information. Huang–Wang highlight the problem of characterizing John domains that are quasihyperbolic $\Omega \subset \mathbb{R}^n$44-uniform without assuming quasiconformal equivalence to a uniform domain (Huang et al., 2011). The Orlicz–Poincaré papers ask, in effect, how far one can remove the “quasiconformally equivalent to a uniform domain” assumption in analytic characterizations of John domains (Feng et al., 2023, Feng et al., 2024). The Sobolev-extension paper points toward a broader mapping theory for “Bi-John pairs,” but leaves that as a heuristic extension rather than a theorem (Koskela et al., 2020).

Taken together, the cited literature supports a precise encyclopedic summary. “Bi-John domain” is best regarded not as a single universally fixed definition, but as a family of two-sided John-type geometries. In bounded Euclidean settings, the strongest recurring model is “uniform domain = two-sided John + length control” (Huang et al., 2011). In pairwise curve terms, cigar John, diameter John plus naturality, and quasihyperbolic cone-geodesic behavior are the principal realizations of that bilateral control (Su et al., 2024, Allu et al., 2023, Huang et al., 2011). In analytic applications, any such class sits above John domains and therefore inherits the broad John-domain theory of Poincaré, Korn, Hardy–Sobolev, trace, extension, and prime-end results developed across these papers (García, 2016, Cejas et al., 2019, Dyda et al., 2017, Adamowicz et al., 2012).