Bryant Surfaces: CMC‑1 in Hyperbolic Space

• Bryant surfaces are conformal CMC‑1 immersions in hyperbolic 3-space, characterized by null holomorphic curves in SL(2,C) and linked to minimal surfaces via the Lawson correspondence.
• They are constructed using diverse methods such as holomorphic representations, loop-group techniques, and Fuchsian potentials, yielding explicit examples like catenoid cousins and trinoids.
• Global existence and compactness results illustrate a duality where local holomorphic flexibility contrasts with strong topological and geometric constraints in hyperbolic 3-manifolds.

Bryant surfaces are conformal immersions $f:\Sigma\to\mathbb H^3$ with constant mean curvature $H\equiv1$. In the standard terminology of the subject, they are the hyperbolic analogue of minimal surfaces in $\mathbb R^3$, and their theory is organized around holomorphic representation formulas, null curves in $\mathrm{SL}(2,\mathbb C)$, and the same Gauss–Codazzi system that governs minimal surfaces after the appropriate change of ambient geometry (Alarcon et al., 2013, Burstall et al., 9 Mar 2026).

1. Definition and classical correspondences

In the hyperboloid model, hyperbolic 3-space is

$$\mathbb H^3=\{(x_0,x_1,x_2,x_3)\in\mathbb L^4:x_0^2=1+x_1^2+x_2^2+x_3^2,\ x_0>0\},$$

where $\mathbb L^4$ is Minkowski 4-space with Lorentz metric of signature $(-+++$) (Alarcon et al., 2013). Equivalently,

$\mathbb H^3=SL_2(\mathbb C)/SU(2),$

and a Bryant surface is a conformal immersion of a Riemann surface into this quotient with mean curvature $1$ (Alarcon et al., 2013, Alarcon et al., 2013).

Bryant’s foundational correspondence identifies simply connected Bryant surfaces with null holomorphic immersions into $\mathrm{SL}(2,\mathbb C)$. If $H\equiv1$0 is a null curve, then the projection

$H\equiv1$1

is a conformal immersion with constant mean curvature $H\equiv1$2; conversely, every simply connected Bryant surface arises in this way (Alarcon et al., 2013). In the matrix model used in the literature, the projection is

$H\equiv1$3

and null curves in $H\equiv1$4 project to Bryant immersions in $H\equiv1$5 (Alarcon et al., 2013).

A second classical bridge is the Lawson correspondence: CMC $H\equiv1$6 surfaces in a space form of curvature $H\equiv1$7 correspond to CMC $H\equiv1$8 surfaces in Euclidean space. In particular, a CMC‑1 surface in $H\equiv1$9 corresponds to a minimal surface in $\mathbb R^3$0 with the same conformal structure and Hopf differential (Jiang, 11 Sep 2025). This explains why Bryant surface theory repeatedly parallels minimal-surface theory, both locally and globally.

For closed surfaces of genus $\mathbb R^3$1 in hyperbolic 3-manifolds, the CMC‑1 case is a critical limit. Huang–Lucia–Tarantello showed that CMC $\mathbb R^3$2-immersions exist for $\mathbb R^3$3 and are parametrized by the tangent bundle of the Teichmüller space of $\mathbb R^3$4, whereas CMC $\mathbb R^3$5-immersions are attained only as limits as $\mathbb R^3$6 (Tarantello et al., 13 Jun 2025). This places Bryant surfaces at the boundary of a variational family rather than inside its uniformly coercive regime.

2. Holomorphic representations and integrable formulations

The basic Bryant representation can be written in several equivalent ways. In the $\mathbb R^3$7 formulation, a holomorphic null curve $\mathbb R^3$8 gives a Bryant surface by

$\mathbb R^3$9

with $\mathrm{SL}(2,\mathbb C)$0 realized as the positive-definite Hermitian matrices of determinant $\mathrm{SL}(2,\mathbb C)$1 (Raujouan et al., 2024). A global meromorphic version uses a meromorphic function $\mathrm{SL}(2,\mathbb C)$2 and a holomorphic 1-form $\mathrm{SL}(2,\mathbb C)$3, assembled into

$\mathrm{SL}(2,\mathbb C)$4

and then solves

$\mathrm{SL}(2,\mathbb C)$5

If the monodromy lies in $\mathrm{SL}(2,\mathbb C)$6, the map $\mathrm{SL}(2,\mathbb C)$7 is a conformal CMC‑1 immersion with Gauss map $\mathrm{SL}(2,\mathbb C)$8 and Hopf differential $\mathrm{SL}(2,\mathbb C)$9 (Traizet, 2014).

A closely related formulation arises from null curves in $$\mathbb H^3=\{(x_0,x_1,x_2,x_3)\in\mathbb L^4:x_0^2=1+x_1^2+x_2^2+x_3^2,\ x_0>0\},$$0. Martín–Umehara–Yamada discovered a biholomorphism

$$\mathbb H^3=\{(x_0,x_1,x_2,x_3)\in\mathbb L^4:x_0^2=1+x_1^2+x_2^2+x_3^2,\ x_0>0\},$$1

that carries null curves in $$\mathbb H^3=\{(x_0,x_1,x_2,x_3)\in\mathbb L^4:x_0^2=1+x_1^2+x_2^2+x_3^2,\ x_0>0\},$$2 into null curves in $$\mathbb H^3=\{(x_0,x_1,x_2,x_3)\in\mathbb L^4:x_0^2=1+x_1^2+x_2^2+x_3^2,\ x_0>0\},$$3 (Alarcon et al., 2013, Alarcon et al., 2013). This gives an efficient transfer mechanism from the Oka–Runge theory of null curves in $$\mathbb H^3=\{(x_0,x_1,x_2,x_3)\in\mathbb L^4:x_0^2=1+x_1^2+x_2^2+x_3^2,\ x_0>0\},$$4 to the Bryant setting.

Integrable-systems formulations make this parallel even more explicit. In the soliton-surface approach, CMC–$$\mathbb H^3=\{(x_0,x_1,x_2,x_3)\in\mathbb L^4:x_0^2=1+x_1^2+x_2^2+x_3^2,\ x_0>0\},$$5 surfaces in $$\mathbb H^3=\{(x_0,x_1,x_2,x_3)\in\mathbb L^4:x_0^2=1+x_1^2+x_2^2+x_3^2,\ x_0>0\},$$6 satisfy the same Gauss–Codazzi system as minimal surfaces in $$\mathbb H^3=\{(x_0,x_1,x_2,x_3)\in\mathbb L^4:x_0^2=1+x_1^2+x_2^2+x_3^2,\ x_0>0\},$$7 when $$\mathbb H^3=\{(x_0,x_1,x_2,x_3)\in\mathbb L^4:x_0^2=1+x_1^2+x_2^2+x_3^2,\ x_0>0\},$$8, and the immersion is recovered from a $$\mathbb H^3=\{(x_0,x_1,x_2,x_3)\in\mathbb L^4:x_0^2=1+x_1^2+x_2^2+x_3^2,\ x_0>0\},$$9 linear problem

$\mathbb L^4$0

The limit

$\mathbb L^4$1

recovers the Enneper–Weierstrass representation of minimal surfaces in $\mathbb L^4$2 from the Bryant-type representation in hyperbolic space (Doliwa et al., 2015).

The loop-group version, the Loop Weierstrass Representation, uses an LWR frame $\mathbb L^4$3 and two evaluation points $\mathbb L^4$4. The hyperbolic null curve is

$\mathbb L^4$5

and the corresponding Bryant surface is

$\mathbb L^4$6

Within this framework, associated families arise by varying evaluation points, dual surfaces arise by swapping them, Goursat transformations are realized by holomorphic dressing, and simple factor dressing preserves the Hopf differential (Raujouan et al., 2024).

A different representation is integration-free. For a Bryant type linear Weingarten surface with parameter $\mathbb L^4$7, the Bianchi–Calò method starts from a holomorphic hyperbolic Gauss map $\mathbb L^4$8 and radius function

$\mathbb L^4$9

In the special case $(-+++$0, this yields the classical Bryant surfaces, now reconstructed from the Euclidean center surface $(-+++$1 without solving a differential equation (Burstall et al., 9 Mar 2026).

3. Ends, auxiliary flat metrics, and local singularity models

Bryant surfaces carry several auxiliary conformal metrics. In the Bryant representation one has a meromorphic function $(-+++$2 and a holomorphic 1-form $(-+++$3, and the paper on isolated singularities of flat metrics notes that Bryant surfaces admit a Bryant representation with data $(-+++$4 and $(-+++$5, while the associated flat metric

$(-+++$6

is a flat conformal metric on the underlying Riemann surface (Li et al., 2019). Ends of a Bryant surface correspond to punctures of this surface, so the local behavior of flat metrics near isolated singularities becomes relevant to the analysis of Bryant ends.

Under a polynomial area growth condition, Li–Xu classify isolated singularities of flat conformal metrics on punctured disks into exactly three local models: $(-+++$7 These are, respectively, a conical model, a cylindrical cusp model, and a logarithmic type model (Li et al., 2019). The same paper states that in Bryant surface theory such flat metrics may arise from $(-+++$8, $(-+++$9, or related metrics near punctures, and that the polynomial area growth condition rules out essential singularities of the developing map. In particular, finite area forces the conical model (Li et al., 2019).

This suggests a precise analytic mechanism behind the familiar distinction between regular and irregular ends: under geometric finiteness assumptions strong enough to imply polynomial area growth or finite area for the relevant flat metric, the end cannot exhibit essential-singularity behavior in the developing map. A plausible implication is that local end asymptotics in Bryant theory can often be read from the exponent or logarithmic term in an auxiliary flat metric rather than only from the induced hyperbolic metric (Li et al., 2019).

Traizet’s construction of Bryant surfaces from horosphere packings makes this local picture geometric. Starting from a finite connected horosphere packing with $\mathbb H^3=SL_2(\mathbb C)/SU(2),$0 horospheres and $\mathbb H^3=SL_2(\mathbb C)/SU(2),$1 tangency points, he uses Bryant representation and the technique of opening nodes to construct a smooth family of complete embedded CMC‑1 surfaces converging to the packing. The resulting surfaces have genus $\mathbb H^3=SL_2(\mathbb C)/SU(2),$2, $\mathbb H^3=SL_2(\mathbb C)/SU(2),$3 catenoid-cousin-type ends, and finite total curvature (Traizet, 2014). Here the neck regions are modeled on catenoid cousins, while the original horospheres supply the end data at infinity.

4. Construction theory and explicit examples

Several modern constructions make Bryant surfaces as explicit as minimal surfaces in $\mathbb H^3=SL_2(\mathbb C)/SU(2),$4. In the Loop Weierstrass framework, catenoid cousins and trinoids come from Fuchsian potentials. For catenoids and catenoid cousins one uses

$\mathbb H^3=SL_2(\mathbb C)/SU(2),$5

and the same potential produces either a minimal catenoid in $\mathbb H^3=SL_2(\mathbb C)/SU(2),$6 or a CMC‑1 catenoid cousin in $\mathbb H^3=SL_2(\mathbb C)/SU(2),$7, depending on the evaluation data in the loop frame (Raujouan et al., 2024). For trinoids, the same paper writes the potential in Schwarz form on $\mathbb H^3=SL_2(\mathbb C)/SU(2),$8 and describes irreducible trinoids in terms of weights and monodromy unitarization (Raujouan et al., 2024).

A Lorentzian representation provides another source of examples. For spacelike conformal immersions in $\mathbb H^3=SL_2(\mathbb C)/SU(2),$9, one writes

$1$0

where $1$1 and $1$2 are complex-valued functions and $1$3 is a null vector. When $1$4 is holomorphic, the condition that the mean curvature vector be lightlike is equivalent to

$1$5

or, after $1$6,

$1$7

The paper proves that any conformal immersion in $1$8 satisfying this partial Riccati equation is congruent by a translation vector to a Bryant immersion in $1$9, and it gives explicit examples, including a catenoid cousin (Dussan et al., 2019).

Opening-nodes constructions supply high-genus embedded examples. Traizet’s surfaces are obtained by desingularizing tangency points of horosphere packings with small catenoid-cousin necks, and the genus is determined combinatorially by the number of horospheres and tangencies (Traizet, 2014). This parallels the earlier minimal-surface gluing program in $\mathrm{SL}(2,\mathbb C)$0, but the period problem is replaced by an $\mathrm{SL}(2,\mathbb C)$1-monodromy problem in the Bryant representation (Traizet, 2014).

These constructions show that Bryant surfaces are not confined to isolated classical models. They form a flexible family that supports local representation formulas, explicit Fuchsian potentials, loop-group dressings, gluing by opening nodes, and Riccati-type reductions to scalar complex ODEs or PDEs (Raujouan et al., 2024, Traizet, 2014, Dussan et al., 2019).

5. Global existence, properness, and compactness

The complex-analytic flexibility of null curves has direct consequences for Bryant surfaces. Using null curves in $\mathrm{SL}(2,\mathbb C)$2, the map $\mathrm{SL}(2,\mathbb C)$3, and Bryant’s projection to $\mathrm{SL}(2,\mathbb C)$4, Alarcón–Forstnerič and subsequent work show that every bordered Riemann surface admits proper holomorphic null embeddings into $\mathrm{SL}(2,\mathbb C)$5, and is conformally equivalent to a proper immersed Bryant surface in $\mathrm{SL}(2,\mathbb C)$6 (Alarcon et al., 2013, Alarcon et al., 2013). The same line of argument also produces complete bounded immersed Bryant surfaces with arbitrary topology (Alarcon et al., 2013).

The paper “The Calabi-Yau problem, null curves, and Bryant surfaces” states more sharply that every bordered Riemann surface admits a proper holomorphic null embedding

$\mathrm{SL}(2,\mathbb C)$7

with bounded third coordinate, which yields a proper holomorphic null embedding

$\mathrm{SL}(2,\mathbb C)$8

and hence a proper conformal Bryant immersion

$\mathrm{SL}(2,\mathbb C)$9

It also states that every bordered Riemann surface is conformally equivalent to a complete bounded immersed Bryant surface in $H\equiv1$00 (Alarcon et al., 2013). These were presented there as the first examples of proper Bryant surfaces with finite topology and of hyperbolic conformal type (Alarcon et al., 2013).

For closed Bryant surfaces in finite-volume hyperbolic 3-manifolds, much stronger rigidity is available. The paper “Area bounds for constant mean curvature surfaces in hyperbolic 3-manifolds” proves that in a closed hyperbolic 3-manifold $H\equiv1$01 there exists $H\equiv1$02 such that every closed Bryant surface $H\equiv1$03 immersed in $H\equiv1$04 satisfies

$H\equiv1$05

For finite-volume hyperbolic 3-manifolds, the same linear area bound holds for every closed Bryant surface of genus $H\equiv1$06, and again $H\equiv1$07 (Jiang, 11 Sep 2025). The same paper proves a compactness theorem for sequences of closed embedded $H\equiv1$08-surfaces with $H\equiv1$09 and bounded genus, yielding smooth convergence away from finitely many points to a strongly Alexandrov embedded limit of multiplicity one (Jiang, 11 Sep 2025).

Topologically, embedded Bryant surfaces in finite-volume hyperbolic 3-manifolds are strongly constrained: they are not essential, they separate the ambient manifold, and on the mean convex side they bound an open handlebody; the induced map on fundamental groups is surjective (Jiang, 11 Sep 2025). This global picture contrasts sharply with the local flexibility of holomorphic representation theory. A plausible summary is that Bryant surfaces are locally as flexible as null-curve methods allow, but globally they are rigidly constrained by the hyperbolic geometry of the ambient 3-manifold (Jiang, 11 Sep 2025).

A complementary existence theory concerns compact surfaces of genus $H\equiv1$10 in hyperbolic 3-manifold germs. For $H\equiv1$11, CMC $H\equiv1$12-immersions are parametrized by $H\equiv1$13. At $H\equiv1$14, the paper on CMC 1-immersions proves that there is a closed analytic subvariety

$H\equiv1$15

of codimension at least $H\equiv1$16 such that, if $H\equiv1$17 and $H\equiv1$18, then there exists a unique CMC 1-immersion of $H\equiv1$19 into a germ of a hyperbolic 3-manifold with the prescribed constraint

$H\equiv1$20

Thus smooth compact Bryant surfaces exist generically in the projectivized tangent directions to Teichmüller space, while blow-up is confined to a proper analytic exceptional set (Tarantello et al., 13 Jun 2025).

6. Generalizations and broader Bryant-type geometries

The adjective “Bryant” also labels several extensions of the classical CMC‑1 theory. One direction is the class of linear Weingarten surfaces of Bryant type in $H\equiv1$21, defined by

$H\equiv1$22

The case $H\equiv1$23 is exactly $H\equiv1$24, hence the classical Bryant surfaces (Burstall et al., 9 Mar 2026). The same paper shows that Bryant type surfaces are characterized by a horosphere congruence of constant intrinsic Gauss curvature $H\equiv1$25, and that the Bianchi–Calò construction extends from the classical Bryant case $H\equiv1$26 to all Bryant type linear Weingarten surfaces (Burstall et al., 9 Mar 2026).

A second direction is gauge-theoretic. For a principal $H\equiv1$27-bundle $H\equiv1$28 with a tensorial 1-form $H\equiv1$29, a connection $H\equiv1$30, and an $H\equiv1$31-invariant bilinear form $H\equiv1$32, the pair $H\equiv1$33 defines an almost complex structure $H\equiv1$34 on $H\equiv1$35. When the integrability equations hold, there is a Bryant type correspondence between space-like, $H\equiv1$36-isotropic holomorphic immersions $H\equiv1$37 and space-like conformal immersions $H\equiv1$38 whose mean curvature vector field is

$H\equiv1$39

The classical Bryant correspondence is recovered by taking $H\equiv1$40, $H\equiv1$41, and the de Sitter analogue by taking $H\equiv1$42 (Teleman, 19 Feb 2026).

Bryant’s influence also extends well beyond $H\equiv1$43. In conformal surface theory, Bryant classified smooth Willmore spheres in $H\equiv1$44 as Möbius transforms of complete minimal surfaces in $H\equiv1$45 with planar ends, and later work extended this picture to branched Willmore spheres with low branching (Heller, 2020, Lamm et al., 2011). In twistor geometry, Bryant’s theorem identifies holomorphic Legendrian immersions $H\equiv1$46 with superminimal immersions $H\equiv1$47 of positive spin under the twistor projection $H\equiv1$48 (Alarcon et al., 2019). These are not Bryant surfaces in the strict hyperbolic sense, but they belong to the broader family of Bryant-type surface theories in which special curvature conditions are encoded by holomorphic data.

In this wider landscape, the strict term “Bryant surface” remains reserved for CMC‑1 immersions in $H\equiv1$49, while “Bryant type” typically signals a geometry obtained by extending Bryant’s holomorphic, twistor, or null-curve paradigm to a larger class of ambient spaces or curvature equations (Burstall et al., 9 Mar 2026, Teleman, 19 Feb 2026, Alarcon et al., 2019).