scieee Science in your language
[en] (orig)

Constant mean curvature surfaces in 3-dimensional Thurston geometries

Abstract

This is a survey on the global theory of constant mean curvature surfaces in Riemannian homogeneous 3-manifolds. These ambient 3-manifolds include the eight canonical Thurston 3-dimensional geometries, i.e. R3, H3, S3, H2 × R, S2 × R, the Heisenberg space Nil3, the universal cover of PSL2(R) and the Lie group Sol3. We will focus on the problems of classifying compact CMC surfaces and entire CMC graphs in these spaces. A collection of important open problems of the theory is also presented.

Read accessible full text

Constant mean curvature surfaces in 3-dimensional Thurston geometries

Author: Fernández Delgado, Isabel; Mira, Pablo
Publisher: World Scientific
Year: 2011
DOI: 10.1142/9789814324359_0076
Source: https://idus.us.es/bitstreams/bbc05757-48a0-4151-bd88-3ade4ca250ff/download
P oceedings o he In e na ional Cong ess o Ma hema icians
Hyde abad, India, 2010
Cons an mean cu a u e su aces in
3-dimensional Thu s on geome ies
Isabel Fe n´andez and Pablo Mi a
Abs ac . This is a su ey on he global heo y o cons an mean cu a u e su aces
in Riemannian homogeneous 3-mani olds. These ambien 3-mani olds include he eigh
canonical Thu s on 3-dimensional geome ies, i.e. R3,H3,S3,H2×R,S2×R, he
Heisenbe g space Nil3, he uni e sal co e o PSL2(R) and he Lie g oup Sol3. We will
ocus on he p oblems o classi ying compac CMC su aces and en i e CMC g aphs in
hese spaces. A collec ion o impo an open p oblems o he heo y is also p esen ed.
Keywo ds. Cons an mean cu a u e su aces, homogeneous spaces, Thu s on geome-
ies, ha monic maps, minimal su aces, en i e g aphs.
1. In oduc ion
Cons an mean cu a u e (CMC) su aces appea as c i ical poin s o a na u al geo-
me ic a ia ional p oblem: o minimize su ace a ea wi h o wi hou a olume cons ain
( he uncons ained case co esponds o ze o mean cu a u e, i.e. o minimal su aces). A
undamen al p oblem o his discipline is he geome ic s udy and classifica ion o CMC
su aces unde global hypo heses like compac ness, comple eness, p ope ness o embed-
dedness. The s udy o his p oblem o CMC su aces in he model spaces R3,S3and
H3has p oduced a e y ich heo y, in which geome ic a gumen s in e ac wi h com-
plex analysis, ha monic maps, in eg able sys ems, maximum p inciples, ellip ic PDEs,
geome ic measu e heo y and so on.
One o he mos ema kable achie emen s o his field in he las decade has been
he ex ension o his classical heo y o he case o CMC su aces in simply connec ed
homogeneous 3-dimensional ambien spaces. Apa om R3,S3and H3, hese spaces a e
he emaining fi e Thu s on 3-dimensional geome ies (i.e. H2×R,S2×R, he Heisenbe g
g oup Nil3, he uni e sal co e ing o PSL2(R) and he Lie g oup Sol3), oge he wi h 3-
dimensional Be ge sphe es and some o he Lie g oups wi h le -in a ian me ics (see
Sec ion 2).
I mus be said he e ha he e is an impo an numbe o con ibu ions ega ding
CMC su aces in gene al Riemannian 3-mani olds (no e en homogeneous), many o which
deal o ins ance wi h isope ime ic ques ions o wi h geome ic consequences de i ed om
he s abili y ope a o associa ed o he second a ia ion o he su ace. The achie emen in
he case o homogeneous ambien 3-spaces has been he cons uc ion o a e y ich global
heo y o CMC su aces, analogous o he case o R3,S3and H3, wi h an emphasis on
he geome ic classifica ion (up o ambien isome ies) o p ope ly imme sed o p ope ly
2Isabel Fe n´andez and Pablo Mi a
embedded CMC su aces. The ac ha he ambien space is homogeneous, i.e. i has
he same local geome y a all poin s, makes his p oblem ex emely na u al.
Ou aim he e is o p esen a su ey on some undamen al aspec s o he global heo y
o CMC su aces in homogeneous 3-mani olds. We do no plan, howe e , o gi e a sys-
ema ic accoun o all impo an esul s o his al eady b oad heo y, bu o discuss some
specific p oblems a he co e o i . Hence, he e will be many impo an esul s omi ed,
and we apologize in ad ance o ha .
In o de o explain he p oblems we shall be dealing wi h, le us dis inguish be ween
compac and non-compac CMC su aces in hese spaces.
In he case o compac CMC su aces, h ee undamen al p oblems a e he Alexand o
p oblem (i.e. o classi y compac embedded CMC su aces), he Hop p oblem (i.e. o
classi y CMC sphe es), and he isope ime ic p oblem ( ecall ha isope ime ic egions
on a Riemannian 3-mani old a e bounded by compac embedded CMC su aces, bu he
con e se is no always ue). By classical esul s, ound sphe es cons i u e he solu ion
o each o hese h ee p oblems in he case o CMC su aces in R3. One o ou main
objec i es will be o explain wha is known (and wha is no known) o hese p oblems
in he b oade con ex o CMC su aces in homogeneous 3-mani olds.
In he case o non-compac CMC su aces, one o he basic p oblems is o s udy he
p ope ly embedded CMC su aces o fini e opology. A classical esul in ha di ec ion
is gi en by Be ns ein’s heo em: planes a e he only en i e minimal g aphs in R3. As in
all Thu s on 3-dimensional geome ies he e is a na u al no ion o en i e g aph, i is an
impo an p oblem o he discipline o sol e he Be ns ein p oblem o CMC g aphs, i.e.
o classi y all en i e CMC g aphs in hese 3-dimensional ambien spaces. This will be ou
o he main objec i e.
The heo y o CMC su aces in Thu s on 3-dimensional geome ies s a ed o de elop
as a consis en unified heo y a e some pionee wo ks by Ha old Rosenbe g, join ly wi h
William H. Meeks [MeRo1, MeRo2, Ros] o he case o minimal su aces in p oduc
spaces, and join ly wi h Uwe Ab esch [AbRo1, AbRo2] o he case o CMC su aces in
homogeneous spaces wi h a 4-dimensional isome y g oup.
On one hand, Meeks and Rosenbe g es ablished many esul s on comple e minimal
su aces in M2×R, wha has guided a la ge numbe o subsequen wo ks in he field. A
ecen majo con ibu ion in his sense is he Collin-Rosenbe g heo em [CoRo] on he
exis ence o ha monic diffeomo phims om Con o he hype bolic plane H2, ob ained by
cons uc ing an en i e minimal g aph o pa abolic con o mal ype in H2×R.
On he o he hand, Ab esch and Rosenbe g disco e ed a holomo phic quad a ic di -
e en ial o CMC su aces in hese homogeneous spaces wi h 4-dimensional isome y
g oup ( he E3(κ, τ) spaces), and sol ed he Hop p oblem o hem. The gene al in eg a-
bili y heo y o CMC su aces in he homogeneous E3(κ, τ) spaces was hen es ablished
by B. Daniel [Dan1]. The disco e y by he au ho s o a ha monic Gauss map in o H2
o H= 1/2 su aces in H2×R u ned in o a se ies o pape s by Daniel, Fe n´andez,
Hauswi h, Mi a, Rosenbe g, Sp uck [FeMi1, Dan2, FeMi2, HRS, DaHa] in which he
Be ns ein p oblem o CMC g aphs o c i ical mean cu a u e (including minimal g aphs
in Heisenbe g space Nil3, see Sec ion 6) was sol ed. Ve y ecen ly, he Hop and Alexan-
d o p oblems o CMC su aces ha e been sol ed by Daniel-Mi a and Meeks [DaMi, Mee]
in he emaining Thu s on 3-dimensional geome y: he Lie g oup Sol3, whose isome y
g oup is only 3-dimensional.
We ha e o ganized his exposi ion as ollows. In Sec ion 2 we will in oduce he
3-dimensional homogeneous ambien spaces. In Sec ion 3 we will p esen he basic in-
eg abili y equa ions by Daniel o CMC su aces in he homogeneous spaces E3(κ, τ),
Cons an mean cu a u e su aces in 3-dimensional Thu s on geome ies 3
oge he wi h he holomo phic Ab esch-Rosenbe g diffe en ial, and wi h some basic defi-
ni ions on s abili y o CMC su aces. In Sec ion 4 we will discuss he Hop , Alexand o
and isope ime ic p oblems in he homogeneous spaces E3(κ, τ). Sec ion 5 will be de o ed
o sol ing he Hop and Alexand o p oblems in he eigh h Thu s on geome y, i.e. he
Lie g oup Sol3. In Sec ion 6 we will p esen he solu ion o he Be ns ein p oblem o
en i e g aphs o c i ical CMC in he homogeneous E3(κ, τ) spaces. Finally, in Sec ion
7 we shall expose he Collin-Rosenbe g heo em on pa abolic en i e minimal g aphs in
H2×R, oge he wi h some de elopmen s on he heo y o comple e minimal su aces o
fini e o al cu a u e in H2×R. Mos sec ions finish wi h a selec ion o impo an open
p oblems. See [Mee, DHM] o mo e open p oblems in he heo y.
A mo e de ailed in oduc ion o he global heo y o CMC su aces in homogeneous
3-spaces can be ound in he Lec u e No es by Daniel, Hauswi h and Mi a [DHM].
The au ho s a e g a e ul o H. Rosenbe g, B. Daniel and J.A. G´al ez o use ul ob-
se a ions abou his manusc ip .
2. Homogeneous 3-spaces and Thu s on geome ies
Homogeneous spaces a e he na u al gene aliza ion o space o ms. By defini ion,
a mani old is said o be homogeneous i he isome y g oup ac s ansi i ely on he
mani old. Roughly speaking, he mani old looks he same a all he poin s, e en hough,
s anding a one poin , he mani old can look diffe en in diffe en di ec ions. In he simply
connec ed case, he classifica ion o he 3-dimensional homogeneous spaces is well-known.
I u ns ou ha any simply connec ed homogeneous 3-space mus ha e isome y g oup
o dimension 6, 4 o 3. The comple e lis o hese spaces is he ollowing (see subsec ions
below o mo e de ails):
•The spaces wi h 6-dimensional isome y g oup a e he space o ms: he Euclidean
space R3, he hype bolic space H3(κ), and he s anda d sphe e S3(κ). Fo simplici y
we will assume ha κ=±1 and w i e H3=H3(−1) and S3=S3(1).
•The spaces wi h 4-dimensional isome y g oup a e fib a ions o e he 2-dimensional
space o ms. They a e he p oduc spaces H2×Rand S2×R, he Be ge sphe es,
he Heisenbe g space Nil3and he uni e sal co e ing o he Lie g oup PSL(2,R).
•The spaces wi h 3-dimensional isome y g oup a e a ce ain class o Lie g oups;
among hem we specially quo e he space Sol3.
These spaces a e closely ela ed wi h Thu s on’s Geome iza ion Conjec u e. This
ecen ly p o ed conjec u e s a es ha any compac o ien able 3-mani old can be cu by
disjoin embedded 2-sphe es o o i in o pieces, each one o hem, a e gluing 2-balls o
solid o i along i s bounda y componen s, admi s a geome ic s uc u e. A 3-mani old
wi hou bounda y is said o admi a geome ic s uc u e i i can be endowed wi h a
comple e locally homogeneous me ic. In his case, by conside ing i s uni e sal co e -
ing we ob ain a comple e simply-connec ed locally homogeneous space and hence, by a
esul o Singe , homogeneous. Thus, a 3-mani old admi ing a geome ic s uc u e can
be ealized as he quo ien o a homogeneous simply connec ed 3-space unde he ac ion
o a subg oup o a Lie g oup ac ing ansi i ely by isome ies. The lis o he maximal
geome ic s uc u es ha gi e compac quo ien s consis s o eigh o he p e iously de-
sc ibed spaces: he h ee space o ms, he wo p oduc spaces, Nil3, he uni e sal co e ing
o PSL(2,R) and Sol3(Be ge sphe es mus be excluded om his lis because hey a e
4Isabel Fe n´andez and Pablo Mi a
no maximal, hei isome y g oup a e con ained in he one o he s anda d sphe e S3).
We e e o [Sco, Bon] o mo e de ails.
2.1. Homogeneous spaces wi h 4-dimensional isome y g oup.
Deno e by M2(κ) he 2-dimensional space o m o cons an cu a u e κ( o example,
M2(κ) = R2,H2,S2 o κ= 0,−1,1 espec i ely). As commen ed abo e, any simply
connec ed homogeneous 3-space wi h 4-dimensional isome y g oup admi s a fib a ion
o e M2(κ), o some κ∈R. Mo eo e , hese spaces can be pa ame e ized in e ms o
he base cu a u e κand he bundle cu a u e τ, ha sa is y κ−4τ26= 0. We will use
he no a ion E3(κ, τ) o hese homogeneous spaces.
1. When τ= 0, we ha e he p oduc spaces M2(κ)×R, i.e. up o scaling, he spaces
S2×Rwhen κ > 0, and H2×Rwhen κ < 0.
2. When τ6= 0 and κ > 0, he co esponding spaces a e he Be ge sphe es, a amily
o 2-pa ame e (1-pa ame e a e a homo he ical change o coo dina es) me ics
on he sphe e, ob ained by de o ming he s anda d me ic in such a way ha he
Hop fib a ion is s ill a Riemannian fib a ion. They can also be seen as he Lie
g oup SU(2) endowed wi h a 1-pa ame e amily o le -in a ian me ics.
3. When τ6= 0 and κ= 0, E3(κ, τ) is he Heisenbe g g oup Nil3, he nilpo en Lie
g oup 



1a b
0 1 c
0 0 1
;a, b, c ∈R

,
endowed wi h a 1-pa ame e amily o le -in a ian me ics, all o hem isome i-
cally equi alen a e a homo he ical change o coo dina es.
4. When τ6= 0 and κ < 0, we ob ain he uni e sal co e ing o he Lie g oup PSL(2,R),
endowed wi h a 2-pa ame e (again 1-pa ame e a e homo he ies) amily o le -
in a ian me ics.
The e exis s a common se ing o all hese spaces. Indeed, label D(ρ) = {(x1, x2)∈
R2;x2
1+x2
2< ρ2}. Then, i κ= 0 ( esp. κ < 0), he space E3(κ, τ) can be iewed as R3
( esp. D2/√−κ×R) endowed wi h he me ic
ds2=λ2(dx2
1+dx2
2) + τλ(x2dx1−x1dx2) + dx32, λ =1
1 + κ
4(x2
1+x2
2).(1)
Also, o κ > 0, (R3, ds2) co esponds o he uni e sal co e o E3(κ, τ) minus one fibe .
In all cases, up o a homo he ical change o coo dina es we can suppose wi hou loss o
gene ali y ha κ−4τ2=±1.
The co esponding Riemannian fib a ion π:E3(κ, τ)→ M2(κ) is gi en he e by he
p ojec ion on he fi s wo coo dina es. The uni a y ec o field
ξ=∂
∂x3
is a Killing field angen o he fibe s o π, and will be e e ed o as he e ical field o
he space E3(κ, τ). I sa isfies he equa ion
b
∇Xξ=τX ×ξ
o all ec o fields Xin E3(κ, τ). He e b
∇is he Le i-Ci i a connec ion, × he c oss
p oduc and τ he bundle cu a u e ( his is basically he defini ion o τ).
Cons an mean cu a u e su aces in 3-dimensional Thu s on geome ies 5
A ema kable diffe ence be ween he spaces E3(κ, τ) is ha hei isome y g oup has
ou connec ed componen s in he case τ= 0, and only wo when τ6= 0. This ollows
om he ac ha any isome y in he p oduc spaces can ei he p ese e o e e se he
o ien a ion o he base and he fibe s independen ly, while in he case τ6= 0 i can only
ei he p ese e o e e se bo h o ien a ions. In pa icula , eflec ions only exis in p oduc
spaces.
Also, when τ6= 0 he spaces E3(κ, τ) a e Lie g oups, and i we se σ:= κ
2τ, an
o hono mal ame o le -in a ian ec o fields (called he canonical ame) is gi en by
E1=λ−1cos(σx3)∂
∂x1
+ sin(σx3)∂
∂x2+τ(x1sin(σx3)−x2cos(σx3)) ∂
∂x3
,
E2=λ−1−sin(σx3)∂
∂x1
+ cos(σx3)∂
∂x2+τ(x1cos(σx3) + x2sin(σx3)) ∂
∂x3
,
E3=ξ=∂
∂x3
.
2.2. Homogeneous spaces wi h 3-dimensional isome y g oup.
O all homogeneous spaces wi h 3-dimensional isome y g oup, Sol3is specially impo -
an , since i is he only Thu s on geome y among hem. We will now desc ibe some
aspec s o his space.
A use ul ep esen a ion o Sol3is he space R3wi h he me ic
ds2=e2x3dx2
1+e−2x3dx2
2+dx2
3,
ha is le -in a ian o he s uc u e o Lie g oup gi en by
(x1, x2, x3)·(y1, y2, y3) = (x1+e−x3y1, x2+ex3y2, x3+y3).
The ollowing ec o fields o m an o hono mal le -in a ian ame
E1=e−x3∂
∂x1
, E2=ex3∂
∂x2
, E3=∂
∂x3
.
The isome ies in Sol3a e gene a ed by he h ee 1-pa ame e g oups o ansla ions
(x1, x2, x3)7→ (x1+c, x2, x3),(x1, x2, x3)7→ (x1, x2+c, x3),
(x1, x2, x3)7→ (e−cx1, ecx2, x3+c),
and by he o ien a ion e e sing isome ies fixing he o igin
(x1, x2, x3)7→ (−x1, x2, x3),(x1, x2, x3)7→ (x2,−x1,−x3).
A ema kable ac is he exis ence o wo canonical olia ions, namely
F1={x1= cons an },F2={x2= cons an },
whose lea es a e o ally geodesic su aces isome ic o he hype bolic plane H2. Reflec ions
ac oss any o hese lea es a e o ien a ion e e sing isome ies o Sol3.

6Isabel Fe n´andez and Pablo Mi a
3. CMC su aces: basic equa ions
In his sec ion we p esen h ee impo an ools o ou s udy. One is he se o
in eg abili y equa ions o CMC su aces in E3(κ, τ) by Daniel [Dan1]. Ano he one he
Ab esch-Rosenbe g diffe en ial, a holomo phic quad a ic diffe en ial geome ically defined
on any CMC su ace in E3(κ, τ). The hi d one is a local isome ic co espondence o
CMC su aces in E3(κ, τ) ia which one can pass om one homogeneous space in o ano he
when s udying CMC su aces [Dan1]. Some no ions abou he s abili y ope a o o CMC
su aces a e also gi en.
3.1. In eg abili y equa ions in E3(κ, τ). I is well known ha he Gauss-
Codazzi equa ions a e he in eg abili y condi ions o su ace heo y in R3,S3and H3. In
o he homogeneous spaces, he si ua ion is mo e complica ed.
Le ψ: Σ →E3(κ, τ) be an isome ic imme sion wi h uni no mal map η, and conside
on Σ he con o mal s uc u e gi en by i s induced me ic ia ψ. Associa ed o a con o mal
pa ame e z=s+i on Σ, we will conside he usual ope a o s ∂z= (∂s−i∂ )/2 and
∂¯z= (∂s+i∂ )/2. Also deno e by ξ he e ical Killing field o E3(κ, τ).
De ini ion 3.1. We call he undamen al da a o ψ he 5- uple (λ|dz|2, u, H, p dz2, A dz)
whe e His he mean cu a u e and
λ= 2hψz, ψ¯zi, u =hN, ξi, p =−hψz, Nzi, A =hξ, ψzi.
The unc ion uis commonly called he angle unc ion o he su ace.
Once he e, a se o necessa y and sufficien condi ions o he in eg abili y o CMC
su aces in E3(κ, τ) can be w i en in e ms o hese undamen al da a. This is a esul
by B. Daniel [Dan1], al hough he o mula ion ha we expose he e (i.e. in e ms o a
con o mal pa ame e on he su ace) comes om [FeMi2].
Theo em 3.2 ([Dan1, FeMi2]). The undamen al da a o an imme sed su ace ψ: Σ →
E3(κ, τ)sa is y he ollowing in eg abili y condi ions:

















(C.1)p¯z=λ
2(Hz+uA(κ−4τ2)).
(C.2)A¯z=uλ
2(H+iτ).
(C.3)uz=−(H−iτ)A−2p
λ¯
A.
(C.4)4|A|2
λ= 1 −u2.
(2)
Con e sely, i Σis simply connec ed, hese equa ions a e also sufficien o he exis-
ence o a su ace ψ: Σ →E3(κ, τ)wi h undamen al da a (λ|dz|2, u, H, p dz2, A dz). This
su ace is unique up o ambien isome ies p ese ing he o ien a ions o base and fibe
o E3(κ, τ).
We see hen ha , in he spaces E3(κ, τ), mo e equa ions apa om he Gauss-
Codazzi ones a e needed, due o he loss o symme ies. As a ma e o ac , (C.1) is
he Codazzi equa ion, while he Gauss equa ion does no appea (i is deduced om he
es ). These new equa ions e idence he special cha ac e o he e ical di ec ion in he
E3(κ, τ) spaces.
Cons an mean cu a u e su aces in 3-dimensional Thu s on geome ies 7
De ini ion 3.3. The Ab esch-Rosenbe g diffe en ial o he imme sion is defined as he
quad a ic diffe en ial on Σgi en by
Qdz2=2(H+iτ)p−(κ−4τ2)A2dz2.
I is hen easy o see by means o (C.2) ha he Codazzi equa ion (C.1) can be
eph ased in e ms o Qas
Q¯z=λHz+ (κ−4τ2)H¯zA2
(H+iτ)2.(3)
Consequen ly, one has he ollowing heo em, which gene alized he classical ac ha he
Hop diffe en ial is holomo phic o CMC su aces in R3,S3and H3.
Theo em 3.4 ([AbRo1, AbRo2]). Qdz2is a holomo phic quad a ic diffe en ial on any
CMC su ace in E3(κ, τ).
This is a c ucial esul o he heo y, since i allows he use o holomo phic unc ions
in he geome ic classifica ion o CMC su aces in E3(κ, τ) (see Sec ion 4 and Sec ion 6,
o ins ance).
An impo an ool in he desc ip ion o CMC su aces in R3,S3and H3is he classi-
cal Lawson co espondence. I es ablishes an isome ic one- o-one local co espondence
be ween CMC su aces in diffe en space o ms ha allows o pass, o ins ance, om
minimal su aces in R3 o H= 1 su aces in H3.
The Lawson co espondence was gene alized by B. Daniel o he con ex o homoge-
neous spaces. Indeed, Daniel disco e ed in [Dan1] an isome ic local co espondence o
CMC su aces in all he homogeneous spaces E3(κ, τ), which can be desc ibed as ollows
in e ms o he undamen al da a defined abo e.
Theo em 3.5 (Sis e co espondence, [Dan1]). Le (λ|dz|2, u, H1, p1dz2, A1dz)be he
undamen al da a o a simply connec ed H1-CMC su ace in E(κ1, τ1), and conside
κ2, τ2, H2∈Rso ha
κ2−4τ2
2=κ1−4τ2
1, H2
2+τ2
2=H2
1+τ2
1.
Then i we se θ∈Rgi en by H2−iτ2=eiθ(H1−iτ1), he undamen al da a gi en by
(λ|dz|2, u, H2, p2dz2=e−iθp1dz2, A2dz =e−iθA1dz) (4)
gi e ise o a (simply connec ed) H2-CMC su ace in E3(κ2, τ2), which is locally isome ic
o he o iginal one.
Two su aces ela ed by he abo e co espondence a e called sis e su aces wi h phase
θ. In pa icula , he co esponding Ab esch-Rosenbe g diffe en ials o sis e su aces a e
ela ed by Q2=e−2iθQ1. As special cases o his co espondence we ob ain he associa e
amily o minimal su aces in M2(κ)×R, and a co espondence be ween minimal su aces
in Nil3and CMC 1
2su aces in H2×R. Gene ically, and up o ambien isome ies and
dila ions, he amily o sis e su aces o a gi en choice o (H, κ, τ) is a con inuous 1-
pa ame e amily.
The e is a na u al no ion o g aph in hese spaces. Since E3(κ, τ) has a canonical
fib a ion o e M2(κ) (see Sec ion 2), we will say ha an imme sed su ace Σ in E3(κ, τ)
8Isabel Fe n´andez and Pablo Mi a
is a (local) g aph i he p ojec ion o he base is a (local) diffeomo phism. The CMC-
equa ion o he g aph o a unc ion u=u(x, y) is he PDE (see [Lee])
2H
δ2=∂
∂x α
ω+∂
∂y β
ω,(5)
whe e
δ= 1 + κ
4(x2+y2), ω =p1 + δ2(x2+y2),
α=ux+τy
δ, β =uy−τx
δ.
Fo ins ance, a g aph u=u(x, y) in Nil3≡E3(0, τ) is minimal i and only i i sa isfies
he ellip ic PDE
(1 + β2)uxx −2αβ uxy + (1 + α2)uyy = 0,(6)
whe e α:= ux+y/2 and β:= uy−x/2.
3.2. S abili y and index o CMC su aces. As i is well known, CMC
su aces in Riemannian 3-mani olds appea as he c i ical poin s o he a ea unc ional
associa ed o a ia ions o he su ace wi h compac suppo and cons an enclosed ol-
ume. Equi alen ly, an imme sed su ace Shas cons an mean cu a u e Hi and only i
i is a c i ical poin o he unc ional A ea −2HVol. The second a ia ion o mula o
his unc ional is gi en by
Q( , ) = −ZS
L( ),
whe e Lis he Jacobi ope a o (o s abili y ope a o ) o he su ace:
L= ∆ + ||B||2+ Ric(η).
He e ∆ is he Laplacian o he induced me ic on he su ace, Bis he second undamen al
o m, ηis he uni no mal ec o field, and Ric is he Ricci cu a u e in he ambien
mani old. As a pa icula case, he Jacobi ope a o o CMC su aces in he spaces
E3(κ, τ) can be ew i en (see [Dan1]) as
L= ∆ −2K+ 4H2+ 4τ2+ (κ−4τ2)(1 + u2),
being K he Gaussian cu a u e o he su ace and u he angle unc ion (see Defini ion
3.1). A Jacobi unc ion is a unc ion o which L( ) = 0.
A CMC su ace Sis said o be s able ( esp. weakly s able) i
Q( , ) = −ZS
L( )≥0
holds o any smoo h unc ion on Swi h compac suppo ( esp. wi h compac suppo
and RS = 0). Fo ins ance, CMC g aphs in E3(κ, τ) a e s able, and compac CMC
su aces bounding isope ime ic egions a e weakly s able (bu no necessa ily s able, as
ound sphe es in R3show).
An impo an concep ela ed o s abili y is he index o a CMC su ace. The index
o a compac CMC su ace is defined as he numbe o nega i e eigen alues o i s Jacobi
ope a o . Thus, s able CMC su aces (in pa icula CMC g aphs) ha e index ze o. Round
sphe es in R3ha e index one.
We e e o [MPR] o mo e de ails abou s abili y o CMC su aces.
Cons an mean cu a u e su aces in 3-dimensional Thu s on geome ies 9
4. Compac CMC su aces in E3(κ, τ)
In his sec ion we explain he mos impo an esul s ha a e known ega ding he
exis ence and uniqueness o compac CMC su aces in he homogeneous 3-spaces E3(κ, τ).
The undamen al examples a e he o a ional CMC sphe es, and we shall be in e es ed in
hei uniqueness among compac embedded CMC su aces, and among imme sed CMC
su aces. These p oblems a e called, espec i ely, he Alexand o and Hop p oblems.
4.1. Ro a ional compac CMC su aces. Al hough ound sphe es in he
model spaces R3,S3,H3a e CMC sphe es, his does no hold o he es o homogeneous
spaces. Howe e , in all he spaces E3(κ, τ) he e exis o a ions wi h espec o he
e ical axis, and so he e is a na u al no ion o o a ional su ace. I is hence na u al
o seek CMC sphe es (and CMC o i) in E3(κ, τ) among he class o o a ional su aces.
This can be done by ODE analysis, and he esul o his can be summa ized as ollows:
Theo em 4.1. (S uc u e o o a ional CMC sphe es in E3(κ, τ)).
1. I κ−4τ2>0, hen o e e y H∈R he e exis s a unique o a ional CMC Hsphe e
(up o isome ies) in E3(κ, τ). These sphe es a e embedded i τ= 0, i.e. in S2×R,
and also o mos Be ge sphe es. Howe e , o some Be ge sphe es wi h small
bundle cu a u e τ(wi h espec o a fixed κ) he e is a ce ain egion o a ia ion
o he pa ame e s (H, τ)whe e he sphe es a e non-embedded. This egion can be
explici ly desc ibed, see [To ].
2. I κ−4τ2<0, hen
•i H2⩽−κ
4, hen he e exis s no o a ional CMC Hsphe e in E3(κ, τ),
•i H2>−κ
4, hen he e exis s a unique o a ional CMC Hsphe e (up o
isome ies) in E3(κ, τ). All hese sphe es a e embedded.
Le us ema k ha all hese CMC sphe es can be cons uc ed explici ly. We shall call
hem canonical o a ional CMC sphe es. Fo example, he o a ional CMC Hsphe es in
S2×R⊂R4a e gi en by he o mula
ψ(u, ) = (−cos k(u),sin k(u) cos , sin k(u) sin , h(u)),
whe e −1≤u≤1, H∈Rand
k(u) := 2 a c an 2H
√1−u2, h(u) := 4H
√4H2+ 1 a csinh u
√1−u2+ 4H2.
Besides hese o a ional CMC sphe es, he e also exis o a ional CMC o i in E3(κ, τ)
when (and only when) κ−4τ2>0 (excluding minimal su aces in S2×R). Fo S2×R,
hey a e all embedded (see Ped osa [Ped]). Fo Be ge sphe es he si ua ion is explained
by To albo and U bano in [To , ToU ]; one has o e e y H o a ional embedded CMC
o i gi en by he Hop li o a ci cle in S2, bu he e also exis some o he non-fla
o a ional CMC o i. The embeddedness p oblem o such o i is open in gene al, bu
o he minimal case he e a e embedded o a ional o i o he han Cliffo d o i. This
con as s wi h he case o embedded minimal o i in S3.
A gene al s udy o CMC su aces in H2×Rand S2×Rin a ian by a con inuous
1-pa ame e subg oup o ambien isome ies can be ound in [SaE, SaTo].
16 Isabel Fe n´andez and Pablo Mi a
De ini ion 6.3. We will say ha a ha monic map Gin o H2admi s Weie s ass da a
{Q0, τ0}i he pullback me ic induced by Gcan be w i en as
hdG, dGi=Q0dz2+µ|dz|2+¯
Q0d¯z2, µ =τ0
4+4|Q0|2
τ0
,
τ0being a posi i e smoo h unc ion.
6.1.1. H= 1/2 su aces in H2×R.We will ega d H2×R=E3(−1,0) in i s
Minkowski model, i.e.
H2×R={(x0, x1, x2, x3) : x0>0,−x2
0+x2
1+x2
2=−1} ⊂ L3×R= L4.
Using his model, he uni no mal ec o ηo an imme sed su ace ψ= (N, h) : Σ →
H2×R akes alues in he de Si e 3-space, and {η, N}is an o hono mal ame o he
Lo en zian no mal bundle o ψin L4. Mo eo e , i uis he angle unc ion o he su ace
( ha is, he las coo dina e o η) and we assume ha u6= 0 ( ha is, ha ψis nowhe e
e ical, o equi alen ly, ha i is a mul ig aph), hen we can w i e
1
u(η+N) = (G, 1),(12)
o a ce ain map G: Σ →H2.
De ini ion 6.4 ([FeMi1]). The map Ggi en by (12) will be called he hype bolic Gauss
map o an imme sed (nowhe e e ical) su ace in H2×R.
The main p ope y o he hype bolic Gauss map is he ollowing [FeMi1]:
Theo em 6.5 (Fe n´andez-Mi a). The hype bolic Gauss map o a CMC su ace wi h
H= 1/2in H2×Ris a ha monic map in o H2, and admi s Weie s ass da a {−Q, λu2},
whe e Qdz2,λ|dz|2and ua e, espec i ely, he Ab esch-Rosenbe g diffe en ial, he me ic,
and he angle unc ion o he su ace.
Con e sely, i Σis simply connec ed, any ha monic map G: Σ →H2admi ing
Weie s ass da a is he hype bolic Gauss map o some H= 1/2su ace in H2×R.
Mo eo e , he space o H= 1/2su aces in H2×Rwi h he same hype bolic Gauss
map Gis gene ically wo-dimensional, and i can be eco e ed om Gby a ep esen a ion
o mula.
The p oo o he di ec pa o he abo e esul ollows om equa ions (2) and he
e y defini ion o G. The con e se pa is an in eg abili y a gumen . This esul is o
g ea impo ance o he es o his sec ion, since i allows he use o ha monic maps in
he desc ip ion o su aces o c i ical CMC.
6.1.2. Minimal su aces in Nil3.The exis ence o his ha monic Gauss map o
H= 1/2 su aces in H2×Rwas ex ended by B. Daniel [Dan2] o he case o minimal
su aces in Nil3=E3(0,1
2).
This ime, he ha monic Gauss map is gi en by he Lie g oup Gauss map o he
su ace. Indeed, i we iden i y he Lie algeb a o Nil3wi h he angen space a a poin
by le mul iplica ion, we can s e eog aphically p ojec he uni no mal ec o field o
ob ain a map aking alues in he ex ended complex plane. Mo e specifically, we will
conside he model o Nil3gi en in Sec ion 2 and i s canonical ame o le -in a ian

Cons an mean cu a u e su aces in 3-dimensional Thu s on geome ies 17
fields {E1, E2, E3}. I N=PNiEiis he uni no mal o X: Σ →Nil3, hen he Gauss
map o Xis gi en by
g=N1+iN2
1 + N3
: Σ →C.
Now, i he su ace is nowhe e e ical we can o ien i so ha u=hN, E3iis posi i e,
and so g akes alues in he uni disc D. By iden i ying H2wi h (D, ds2
P), whe e ds2
Pis
he Poinca ´e me ic, Daniel ob ained in [Dan2]:
Theo em 6.6 (Daniel). The Gauss map o a nowhe e e ical minimal su ace is ha -
monic in o H2.
Con e sely, le g: Σ →H2be a ha monic map defined on a simply connec ed o ien ed
Riemann su ace in o H2, and assume ha gis nowhe e an iholomo phic (i.e., gzdoes
no anish a any poin ). Take z0∈Σand X0∈Nil3.
Then he e exis s a unique con o mal nowhe e e ical minimal imme sion X: Σ →
Nil3wi h X(z0) = X0ha ing gas i s Gauss map. Mo eo e , Xcan be uniquely eco e ed
om g h ough an adequa e ep esen a ion o mula.
Fu he mo e, i can be checked ha he Weie s ass da a o gas abo e a e {−Q, λu2},
whe e Qdz2,λ|dz|2and ua e, espec i ely, he Ab esch-Rosenbe g diffe en ial, he me ic,
and he angle unc ion o he su ace defined in Sec ion 2.
As we saw in Sec ion 3, minimal su aces in Nil3and H= 1/2 su aces in H2×Ra e
ela ed by he sis e co espondence, and sis e su aces ha e he same me ic and angle
unc ion (in pa icula , he condi ion o being nowhe e e ical is p ese ed). As in his
case he sis e su aces ha e opposi e Ab esch-Rosenbe g diffe en ials, i u ns ou ha
hei espec i e ha monic Gauss maps a e conjuga e o each o he .
The ela ion be ween minimal su aces in Nil3and H= 1/2 su aces in H2×Rcan be
made mo e explici by means o he heo y o spacelike CMC su aces in L3, as ollows.
Theo em 6.7 ([FeMi3]). Le X= (F, ) : Σ →Nil3be a simply connec ed nowhe e
e ical minimal su ace wi h me ic λ|dz|2and angle unc ion u, and ψ= (N, h) : Σ →
H2×Ri s sis e su ace.
Then := (F, h) : Σ → L3is a spacelike H= 1/2su ace in he Minkowski 3-space
wi h me ic λu2|dz|2and Hop diffe en ial −Qdz2, whe e Qdz2is he Ab esch-Rosenbe g
diffe en ial o X.
6.1.3. CMC √−κ/2 su aces in ^
PSL(2,R). In a o hcoming pape [DFM], he
au ho s and B. Daniel will p o e ha he e exis s also a ha monic Gauss map o c i ical
CMC su aces in he emaining case, i.e. he uni e sal co e ing o he g oup PSL(2,R),
and will de i e a ep esen a ion o mula o hem.
6.2. Hal -space heo ems. One o he mos impo an esul s in he global
s udy o minimal su aces in R3is he classical hal -space heo em by Hoffman and Meeks
[HoMe]. This heo em says ha any p ope ly imme sed minimal su ace in R3lying in a
hal -space mus be a plane pa allel o he one de e mining he hal -space. The main ools
used he e a e he maximum p inciple and he exis ence o ca enoids, a 1-pa ame e amily
o minimal su aces con e ging o a doubly-co e ed punc u ed plane P, and in e sec ing
he planes pa allel o Pin compac cu es.
The analogous e sion o CMC one hal su aces in H2×Rwas p o ed in [HRS]. In
his se ing, ho ocylinde s play he ole o he planes in R3.
18 Isabel Fe n´andez and Pablo Mi a
Theo em 6.8 (Hauswi h-Rosenbe g-Sp uck). The only p ope ly imme sed CMC one
hal su aces in H2×R ha a e con ained in he mean con ex side o a ho ocylinde C
a e he ho ocylinde s pa allel o C.
Also, he only p ope ly embedded CMC one hal su aces in H2×Rcon aining a
ho ocylinde in i s mean con ex side a e he ho ocylinde s.
P oo . The main poin he e is o cons uc a amily o CMC one hal su aces in H2×R
o be used in he same way as ca enoids in he p oo o he hal -space heo em in R3.
This is achie ed by means o compac annuli wi h bounda ies, con ained be ween wo
ho ocylinde s.
Fo he case o Nil3, we mus dis inguish be ween ho izon al and e ical hal -spaces.
The equi alen o he hal -space heo em o su aces lying in a ho izon al hal -space
is p o ed by using he amily o o a ional annuli [AbRo2]. The co esponding e ical
e sion has been ob ained in [DaHa], by cons uc ing fi s a amily o ho izon al ca enoids,
i.e. p ope ly embedded minimal annuli (non- o a ional) wi h a geome ic beha iou good
enough o apply he Hoffman-Meeks echnique.
Theo em 6.9 (Daniel-Hauswi h). The only p ope ly imme sed minimal su aces in
Nil3 ha a e con ained in a e ical hal space a e he e ical planes pa allel o he one
de e mining he hal -space.
P oo . Using he ep esen a ion o mula o minimal su aces in Nil3(see Theo em 6.6),
i is possible o cons uc ho izon al ca enoids in Nil3. These su aces a e a 1-pa ame e
amily o p ope ly embedded minimal annuli, in e sec ing e ical planes {x2=c}in a
non-emp y closed con ex cu e. Mo eo e , he amily con e ges o a double co e ing o
{x2= 0}minus a poin . They a e ob ained by in eg a ing a amily o ha monic maps
ha belong o a mo e gene al amily used in he cons uc ion o Riemann ype minimal
su aces in H2×R[Ha]. Once we ha e hese ca enoids, we finish by using he maximum
p inciple simila ly o he Euclidean case.
6.3. The classi ica ion o en i e g aphs. In his sec ion we will desc ibe
he space o en i e g aphs o c i ical CMC in E3(κ, τ). Such a desc ip ion ollows om
he wo ks o Fe n´andez-Mi a [FeMi1, FeMi3], Hauswi h-Rosenbe g-Sp uck [HRS] and
Daniel-Hauswi h [DaHa], and is con ained in Theo ems 6.10 and 6.11. We expose he e
a unified pe spec i e o his subjec . Fi s , we ha e
Theo em 6.10 ([DaHa, FeMi3, HRS]). The ollowing condi ions a e equi alen o a
su ace o c i ical CMC in E3(κ, τ):
(1) I is an en i e g aph.
(2) I is a comple e mul ig aph.
(3) u2ds2is a comple e Riemannian me ic (whe e uis he angle unc ion and ds2 he
me ic o he su ace).
In pa icula , he sis e co espondence p ese es en i e g aphs o c i ical CMC.
Le us make some commen s on his heo em. Fi s , Hauswi h, Rosenbe g and
Sp uck p o ed (2) ⇒(1) o H= 1/2 su aces in H2×R. Second, he au ho s p o ed
in [FeMi3] ha (3) ⇒(1) ( o any su ace in E3(κ, τ), no necessa ily CMC), and ha
(1) ⇒(3) holds o minimal su aces in Nil3. Finally, Daniel and Hauswi h showed ha
(2) ⇒(1) holds o minimal su aces in Nil3. The es o he cases can be easily ob ained
om hese esul s and he sis e co espondence ( his was fi s obse ed in [DHM]).
Cons an mean cu a u e su aces in 3-dimensional Thu s on geome ies 19
P oo . I is immedia e ha (1) ⇒(2). Also, by an eigen alue es ima e, he au ho s
p o ed in [FeMi3] ha o a bi a y su aces in E3(κ, τ) i holds u2ds2≤gF, whe e
F=π◦ψis he p ojec ion on o M2(κ) o ψ. Thus, i u2ds2is comple e, Fis a local
diffeomo phism wi h comple e pullback me ic, and by s anda d opological a gumen s,
Fis a diffeomo phism, i.e. (3) ⇒(1) holds.
Tha (1) ⇒(3) holds o minimal su aces in Nil3was also p o ed in [FeMi3]: le
X= (F, ) : Σ →Nil3be an en i e minimal g aph. By Theo em 6.7, he e is an en i e
spacelike CMC g aph = (F, h) : Σ → L3, whose induced me ic is ds2
=u2ds2. Now
we can apply a heo em by Cheng and Yau [ChYa] which says ha spacelike en i e CMC
g aphs in L3ha e comple e induced me ic. Hence u2ds2is comple e, as wished.
We will now p o e ha (2) ⇒(1) holds o minimal su aces in Nil3. Le us obse e
ha once his is done, we can also p o e he heo em o su aces o c i ical CMC in all
he spaces E3(κ, τ). Indeed, as any simply connec ed su ace o c i ical CMC is he sis e
su ace o some minimal su ace in Nil3, and as he co espondence p ese es he me ic
and he angle unc ion ( he e o e i p ese es condi ions (2) and (3) by passing o he
uni e sal co e ing), we can easily ansla e he heo em o he case o minimal su aces
in Nil3 o he es o he spaces. I is impo an he e ha we p o ed (3) ⇒(1) in all
spaces.
So, we only need o p o e (2) ⇒(1) o minimal su aces in Nil3. This was done
by Daniel and Hauswi h [DaHa]. Fo ha , hey used hei hal -space heo em in Nil3
(Theo em 6.9) and an adap a ion o Nil3o he p e ious p oo o (2) ⇒(1) o he case
o H= 1/2 su aces in H2×Rgi en by Hauswi h-Rosenbe g-Sp uck [HRS].
In o de o p o e (2) ⇒(1) o minimal su aces in Nil3, we a gue by con adic ion.
Assume ha he e exis s a comple e mul ig aph Σ ha is no en i e. Then he e exis s
an open se Σ0⊂Σ ha is a g aph o e a disc D( R2o a unc ion , and a poin
q∈∂D such ha does no ex end o q.
S ep 1: Fo any sequence o poin s {qn}in Dcon e ging o q, he sequence o no mal
ec o s a he poin s pn= (qn, (qn)) ∈Σ0con e ges o he ho izon al ec o o hogonal
o ∂D a q.
Indeed, as he su ace is a mul ig aph, i s angle unc ion u=hN, E3i, whe e N
deno es he uni no mal ec o , is a Jacobi unc ion ha does no anish. As a esul o
his, Σ is (s ongly) s able, and has bounded geome y. This means ha locally a ound
any pnwe can w i e he su ace as he g aph (in exponen ial coo dina es) o e a disc o
adius δo i s angen plane, whe e δis a uni e sal cons an depending only on Σ. This
neighbo hood o pnwill be deno ed by G(pn). The limi o he no mal ec o s {N(pn)}
mus be a ho izon al ec o since o he wise, he piece G(pn) o bounded geome y could
be ex ended as a g aph beyond q, which is impossible. Mo eo e , he limi ec o mus
be no mal o ∂D a qsince Σ0is a g aph o e D.
S ep 2: The unc ion defining he g aph Σ0di e ges a q. Mo eo e , as we app oach
q, and a e ansla ing he su ace o he o igin, he su aces con e ge o a piece o he
( ansla ed) e ical plane Ppassing h ough qand angen o ∂D.
Tha di e ges a qis a consequence o he comple eness o Σ, and he las pa can
be p o ed by ollowing he ideas o Collin and Rosenbe g in [CoRo]. We will assume ha
Pis he plane {x1=c}.
S ep 3: Σcon ains a g aph Go e a domain o he o m Uǫ= (c−ǫ, c)×R⊂R2.
Mo eo e , his g aph is disjoin om Pand asymp o ic o i as one app oaches q.
The g aph Gis ob ained by analy ical con inua ion o he su aces G(pn) used in he
fi s s ep, and a e a ca e ul s udy o he beha io o he in e sec ion cu es o hese
g aphs and he planes pa allel o P.
20 Isabel Fe n´andez and Pablo Mi a
Finally, he con adic ion ollows om he hal -space heo em (Theo em 6.9). Re-
call ha , al hough Ghas bounda y and he heo em is o mula ed o su aces wi hou
bounda y, i s p oo applies o his case, and so we a e done.
Once he e, we in es iga e he Be ns ein p oblem o en i e g aphs o c i ical CMC
in E3(κ, τ), i.e. he classifica ion o such en i e g aphs ( ecall he e ha CMC g aphs in
E3(κ, τ) sa is y he ellip ic PDE (5)). The e minology comes om he classical Be ns ein
heo em: en i e minimal g aphs in R3a e planes. Equi alen ly, any solu ion o he
minimal g aph equa ion
(1 + 2
y) xx −2 x y xy + (1 + 2
x) yy = 0 (13)
defined on he whole plane is linea .
I is in e es ing o compa e his esul wi h he Be ns ein p oblem in Nil3, i.e. he
classifica ion o en i e minimal g aphs in Nil3. This co esponds o classi ying all global
solu ions o he PDE (6). Obse e ha aking τ= 0 in (6) we ob ain he classical
equa ion (13), i.e. he classical case conside ed by Be ns ein appea s as a limi o he
Heisenbe g case.
The e exis s, howe e , a g ea diffe ence be ween bo h si ua ions. The ollowing esul
classifies he en i e g aphs o c i ical CMC in E3(κ, τ), by pa ame izing he moduli space
o such en i e g aphs in e ms o holomo phic quad a ic diffe en ials. I was ob ained fi s
o minimal g aphs in Nil3by he au ho s [FeMi3], and sho ly he ea e by Daniel and
Hauswi h [DaHa] o H= 1/2 g aphs in H2×R. The gene al case ollows easily om he
Heisenbe g case and Theo em 6.10, using he sis e co espondence ( his was obse ed
fi s in [DHM]).
Theo em 6.11 (Fe n´andez-Mi a, Daniel-Hauswi h). Le Qdz2deno e a holomo phic
quad a ic diffe en ial on Σ≡Co D, such ha Q6≡ 0i Σ≡C, and le H2=−κ/4.
The e exis s a 2-pa ame e amily o en i e CMC Hg aphs in E3(κ, τ)whose Ab esch-
Rosenbe g diffe en ial ag ees wi h Qdz2. These g aphs a e gene ically non-cong uen .
And con e sely, hese a e all he en i e g aphs o c i ical CMC in E3(κ, τ).
A his poin , he p oo o he case o minimal su aces in Nil3is a consequence o
Theo em 6.7 and he ollowing esul by Wan and Wan-Au [Wan, WaAu] on spacelike
en i e CMC g aphs in L3: o any holomo phic quad a ic diffe en ial as abo e, he e exis s
a unique (up o isome ies) spacelike en i e CMC 1/2g aph in L3wi h Hop diffe en ial
Qdz2.The 2-pa ame e amily o non-cong uen g aphs in E3(κ, τ) comes om he loss
o ambien isome ies ( om 6 dimensions o 4 dimensions) when passing om L3 o Nil3.
The emaining cases o c i ical CMC g aphs ollow since by Theo em 6.10 he sis e
co espondence p ese es en i e g aphs.
6.4. Open P oblems. As explained in Sec ion 3, en i e g aphs a e s able. I
is conjec u ed ha en i e g aphs and e ical cylinde s a e he only s able c i ical CMC
su aces ( his has been p o ed o pa abolic con o mal ype in [MaPR]). Rela ed o his
is he ques ion o non-exis ence o comple e s able H > 1/2 su aces in H2×R(p o ed
o H > 1/√3 by Nelli-Rosenbe g, [NeRo2]).
Also, no much is known abou p ope ly embedded su aces o c i ical CMC and
non- i ial opology. Can one ob ain hem by conjuga e Pla eau cons uc ions, o by
in eg able sys ems echniques? Ano he ema kable p oblem is o es ablish he s ong
hal -space heo em in Nil3: a e wo disjoin p ope ly embedded minimal su aces in Nil3
necessa ily wo pa allel e ical planes, o wo pa allel en i e minimal g aphs?
Cons an mean cu a u e su aces in 3-dimensional Thu s on geome ies 21
7. Minimal su aces in H2×Rand S2×R
Minimal su aces in p oduc spaces admi a special ea men , due o se e al easons.
One o hem is he ollowing: i ψ= (N, h) : Σ →M2×Ris a minimal su ace imme sed
in he p oduc space M2×R, whe e (M2, g) is a Riemannian su ace, hen he ho izon al
p ojec ion N: Σ →M2is a ha monic map and he heigh unc ion h: Σ →Ris a
ha monic unc ion. This implies, o ins ance, ha compac minimal su aces in M2×R
only exis i M2is compac (in pa icula , i M2=S2), and he only ones a e he slices
M2×{ 0}.
Ano he impo an ac abou minimal su aces in M2×Ris ha he e is a na u al
no ion o minimal g aph o e a domain Ω ⊂M2, and ha his g aph sa isfies a simple
ellip ic PDE in di e gence o m. This ac oge he wi h gene al exis ence esul s o
solu ions o he Pla eau p oblem in Riemannian 3-mani olds allows a good con ol on he
geome y o he su ace. Some o he mos in e es ing esul s o he heo y o minimal
su aces in p oduc spaces come om he in e play be ween he in o ma ion p o ided by
ha monic maps and by Pla eau cons uc ions and he minimal g aph equa ion.
S a ing wi h he pionee wo k o H. Rosenbe g [Ros], and W.H Meeks and H. Rosen-
be g [MeRo1, MeRo2], he heo y o minimal su aces in M2×Rhas de eloped subs an-
ially in he las decade. We will only alk he e abou a ew esul s o special ele ance
o he heo y, and no men ion many o he impo an esul s.
7.1. The Collin-Rosenbe g heo em. The classical Be ns ein heo em in
R3s a es ha planes a e he only en i e minimal g aphs in R3. This heo em can be
ex ended o he case o p oduc spaces: any en i e minimal g aph in M2×R, whe e
(M2, g)is a comple e su ace o non-nega i e cu a u e, is o ally geodesic.
In con as , in he p oduc space H2×R he e is a wide a ie y o en i e minimal
g aphs. Fo ins ance, in [NeRo1] Nelli and Rosenbe g sol ed he Di ichle p oblem a
infini y o he minimal g aph equa ion in H2×R. They p o ed ha any Jo dan cu e
a he ideal bounda y S1×R≡∂∞H2×Ro H2×Rwhich is a g aph o e S1≡∂∞H2is
he asymp o ic bounda y o a unique en i e minimal g aph in H2×R(see [GaRo] o a
p oo o his in he mo e gene al case o en i e minimal g aphs in M2×R, whe e (M2, g)
is comple e, simply connec ed and wi h KM≤c < 0).
All hese en i e minimal g aphs a e hype bolic, ha is, hey ha e he con o mal ype
o he uni disk. The p oblem o exis ence o en i e minimal g aphs o pa abolic ype
(i.e. wi h he con o mal ype o C) is much ha de , and was sol ed ecen ly by Collin and
Rosenbe g [CoRo].
Theo em 7.1 (Collin-Rosenbe g). The e exis en i e minimal g aphs in H2×Ro
pa abolic con o mal ype.
As he p ojec ion on o H2o a minimal g aph is a ha monic diffeomo phism, he abo e
heo em has he ollowing consequence, which sol es a majo p oblem in he heo y o
ha monic maps and disp o es a conjec u e by R. Schoen and S.T. Yau.
Co olla y 7.2 (Collin-Rosenbe g). The e exis ha monic diffeomo phisms om Con o
H2.
The p oo by Collin and Rosenbe g is a good example o he in e ac ion be ween he
ha monici y p ope ies o he minimal imme sion and he use o Pla eau cons uc ions
and he minimal g aph equa ion.

22 Isabel Fe n´andez and Pablo Mi a
The main idea in he p oo is o cons uc fi s (non-en i e) minimal g aphs in H2×R
o Sche k ype o e ideal geodesic polygons, ha ing al e na ing asymp o ic alues +∞
and −∞ on he sides o he polygon. This gene alizes a classical cons uc ion by Jenkins
and Se in in he case o minimal g aphs o e bounded domains in R3. This cons uc ion
is done as ollows:
Le Γ be an ideal polygon o H2, so ha all he e ices o Γ a e a he ideal bounda y
o H2and Γ has an e en numbe o sides A1, B1, A2, B2..., Ak, Bk, o de ed clockwise. A
each e ex ai, we conside a small enough ho ocycle Hiwi h Hi∩Hj=∅. Each Ai
( esp. Bi) mee s exac ly wo ho ocycles. Deno e by e
Ai( esp. e
Bi), he compac a c o
Ai( esp Bi) which is he pa o Aiou side he wo ho odisks. We deno e by |Ai| he
leng h o |e
Ai|. Define e
Biand |Bi|in he same way.
Now we can conside a(Γ) = Pk
i=1 |Ai|and b(Γ) = Pk
i=1 |Bi|. We obse e ha
a(Γ) −b(Γ) does no depend on he choice o he ho ocycle Hia ai, since ho ocycles
wi h he same poin a infini y a e equidis an . Keeping in mind hese da a, we can s a e
he ollowing heo em by Collin-Rosenbe g [CoRo] (see also Nelli-Rosenbe g [NeRo1]):
Theo em 7.3. ([NeRo1], [CoRo]) The e is a (unique up o addi i e cons an s) solu ion
o he minimal su ace equa ion in he polygonal domain P, equal o +∞on Aiand −∞
on Bi, i and only i he ollowing condi ions a e sa isfied:
1. a(Γ) = b(Γ),
2. Fo each insc ibed polygon Pin Γ,P 6= Γ, and o some choice o ho ocycles a he
e ices, one has
2a(P)<|P| and 2b(P)<|P|.
All hese examples ha e he con o mal ype o C. Once he e, Collin and Rosenbe g
designed a way o enla ging a gi en Sche k- ype g aph o e he in e io o some Γ ⊂H2
in o ano he one wi h mo e sides, and so ha : (1) he ex ended su ace is C2-close o he
o iginal one o e an a bi a y compac se in he in e io o Γ, and (2) he e is a con ol
on he con o mal adius on adequa e compac annuli on he su ace.
By passing o he limi in his sequence o minimal g aphs o e la ge and la ge
domains, hey ob ained an en i e minimal g aph in H2×Rwhich, by he con ol on he
con o mal adii o hese annuli, has he con o mal ype o C.
Rema k 7.4. The Collin-Rosenbe g heo em has been ex ended by J.A. G´al ez and H.
Rosenbe g [GaRo] o mo e gene al p oduc spaces M2×R: he e exis en i e minimal
g aphs o pa abolic con o mal ype on M2×R, whe e (M2, g) is any comple e simply
connec ed Riemannian su ace wi h Gaussian cu a u e KM≤c < 0 (KMno cons an ).
7.2. Minimal su aces o ini e o al cu a u e in H2×R.One
o he mos s udied amilies among minimal su aces in R3a e he comple e minimal
su aces o fini e o al cu a u e (FTC o sho ). A minimal su ace Σ is said o ha e
FTC i i s Gaussian cu a u e Ksa isfies
ZΣ
K dA<∞.
By classical heo ems o Hube and Osse man, comple e FTC minimal su aces in
R3a e con o mally equi alen o a compac Riemann su ace minus a fini e numbe o
poin s. Mo eo e , he Gauss map ex ends me omo phically o he punc u es, and he
o al cu a u e o he su ace is a mul iple o −4π. A key poin he e is ha he Gauss
map o a minimal su ace in R3is con o mal.
Cons an mean cu a u e su aces in 3-dimensional Thu s on geome ies 23
In H2×R he e is no con o mal Gauss map o minimal su aces. None heless, using
he global heo y o ha monic maps in o H2, L. Hauswi h and H. Rosenbe g [HaRo] we e
able o p o e ha a simila si ua ion holds in H2×R.
Theo em 7.5 (Hauswi h-Rosenbe g). Le Xbe a comple e minimal imme sion o Σin
H2×Rwi h fini e o al cu a u e. Then
1. Σis con o mally equi alen o a Riemann su ace punc u ed a a fini e numbe o
poin s, Σ≡Mg−{p1...., pk}.
2. Qdz2:= h2
zdz2is holomo phic on Mand ex ends me omo phically o each punc-
u e. I we pa ame e ize each punc u e piby he ex e io o a disk o adius , and
i Q(z)dz2=z2mi(dz)2a pi, hen mi⩾−1.
3. The hi d coo dina e uo he uni no mal ends o ze o uni o mly a each punc u e.
4. The o al cu a u e is a mul iple o 2π:
ZΣ
KdA = 2π 2−2g−2k−
k
X
i=1
mi!.
As a consequence, e e y end o a fini e o al cu a u e su ace is uni o mly asymp o ic
o a Sche k ype g aph desc ibed in Theo em 7.3.
P oo . The fi s s ep is o p o e ha locally a ound an end, Qdz2only has a mos a fini e
numbe o ze oes. Then a Hube heo em and an a gumen o Osse man gi e ha he ends
a e con o mally a punc u ed disk, and Qdz2ex ends me omo phically o he punc u e.
The final pa o he beha io o Qdz2 ollows om he ac ha Qdz2=h2
zdz2, whe e h
is he heigh unc ion o he su ace.
To p o e ha ugoes o 0 a he ends, ake an annula neighbo hood o an end whe e
Qdz2does no anish. Then epa ame e ize his annulus by w=R√Qdz. The me ic
con o mal ac o in hese coo dina es sa isfies a sinh-Go don equa ion, and he Gaussian
cu a u e mono onically dec eases o ze o. Then, es ima es on he g ow h o solu ions
o he sinh-Go don equa ion allows one o conclude ha , a a fini e o al cu a u e end,
he angen plane becomes e ical and he me ic becomes fla .
Finally, he exp ession o he o al cu a u e ollows om Gauss-Bonne o mula and
he es ima es o he sinh-Go don equa ion ob ained be o e.
In [HaRo], he ollowing ques ion was also aised: a e he e comple e non simply
connec ed minimal su aces wi h FTC in H2×R?No ice ha o a ional ca enoids ha e
infini e o al cu a u e. Ac ually, a ha ime, he only known comple e FTC minimal
su aces we e he Sche k ype g aphs.
This ques ion was posi i ely answe ed by J. Pyo [Pyo] and also, independen ly, by
Rod ´ıguez and Mo abi o [RoMo]. Pyo cons uc ed a 1-pa ame e amily o genus ze o
p ope ly embedded minimal su aces in H2×Rwi h kends o k≥2, simila o he
k-noids in R3(al hough he fi s ones a e embedded and he k-noids in R3a e no ). They
ha e o al cu a u e 4π(1 −k), and a e asymp o ic o e ical planes a infini y. These
su aces a e ob ained as he conjuga e su aces o minimal g aphs o e infini e geodesic
iangles in H2 ha a e asymp o ic o e ical planes a infini y.
Ve y sho ly he ea e , M. Rod ´ıguez and F. Mo abi o disco e ed independen ly a
la ge amily o FTC minimal su aces, con aining he p e ious ones. I is a (2k−
2)-pa ame e o p ope ly embedded FTC minimal su aces o genus ze o wi h kends,
ob ained as he limi s o simply pe iodic minimal su aces called saddle owe s, ha a e
24 Isabel Fe n´andez and Pablo Mi a
in a ian by a e ical ansla ion o ec o (0,0,2l). Taking limi s when l→ ∞, hey
ob ain genus ze o minimal su aces wi h kends and o al cu a u e 4π(1 −k) ha a e
symme ic wi h espec o he eflec ion o e he slice H2×{0}. The su aces ound by
Pyo appea when he ends a e placed in symme ic posi ions.
7.3. Open p oblems. In [Ha], L. Hauswi h cons uc ed a amily o Riemann
ype minimal su aces in H2×Rand S2×R, cha ac e ized by he p ope y o being olia ed
by cu es o cons an cu a u e. I is a conjec u e by W. Meeks and H. Rosenbe g ha
in S2×R hey a e he only p ope ly embedded minimal annuli. An app oach o sol ing
his conjec u e using in eg able sys ems echniques has been ecen ly de eloped by L.
Hauswi h and M. Schmid . Ano he na u al p oblem is o ob ain classifica ion esul s
o p ope ly embedded minimal su aces o fini e o al cu a u e and a gi en simple
opology in R3.
Schoen and Yau p o ed he e is no ha monic diffeomo phism om he disk o a
comple e su ace o non-nega i e cu a u e. Can he e be such a ha monic diffeomo phism
on o a comple e pa abolic su ace? This is a ques ion by J.A. G´al ez.
Re e ences
[AbRo1] U. Ab esch, H. Rosenbe g, A Hop diffe en ial o cons an mean cu a u e
su aces in S2×Rand H2×R,Ac a Ma h. 193 (2004), 141–174.
[AbRo2] U. Ab esch, H. Rosenbe g, Gene alized Hop diffe en ials, Ma . Con emp. 28
(2005), 1–28.
[AEG1] J.A. Aledo, J.M.Espina , J.A. G´al ez, Heigh es ima es o su aces wi h posi i e
mean cu a u e in M×R.Illinois Jou nal o Ma h., 52 (2008), 203-211.
[Bon] F. Bonahon, Geome ic s uc u es on 3-mani olds. In Handbook o Geome ic
Topology, pages 93–164. No h-Holland, Ams e dam, 2002.
[ChYa] S.Y. Cheng, S.T. Yau, Maximal spacelike hype su aces in he Lo en z-Minkowski
spaces, Ann. o Ma h. 104 (1976), 407–419.
[CoRo] P. Collin, H. Rosenbe g, Cons uc ion o ha monic diffeomo phisms and minimal
g aphs, Ann. o Ma h., o appea (2007).
[Dan1] B. Daniel, Isome ic imme sions in o 3-dimensional homogeneous mani olds,
Commen . Ma h. Hel . 82 (2007), 87-131.
[Dan2] B. Daniel, The Gauss map o minimal su aces in he Heisenbe g g oup, p ep in ,
2006, a Xi :ma h/0606299.
[DFM] B. Daniel, I. Fe n´andez, P. Mi a, Su aces o c i ical cons an mean cu a u e.
Wo k in p og ess.
[DaHa] B. Daniel, L. Hauswi h, Hal -space heo em, embedded minimal annuli and
minimal g aphs in he Heisenbe g g oup. P oc. Lond. Ma h. Soc. (3),98 no.2 (2009),
445–470.
[DHM] B. Daniel, L. Hauswi h, P. Mi a, Cons an mean cu a u e su aces in homo-
geneous mani olds, p ep in , 2009. Published p elimina ly by he Ko ea Ins i u e o
Ad anced S udy.
[DaMi] B. Daniel, P. Mi a, Exis ence and uniqueness o cons an mean cu a u e sphe es
in Sol3. P ep in , 2008, a Xi :0812.3059
Cons an mean cu a u e su aces in 3-dimensional Thu s on geome ies 25
[dCF] M.P. do Ca mo, I. Fe n´andez, Ro a ionally in a ian CMC disks in p oduc space,
Fo um Ma h. 21 (2009), 951–963.
[EGR] J.M. Espina , J.A. G´al ez, H. Rosenbe g, Comple e su aces wi h posi i e ex in-
sic cu a u e in p oduc spaces, Commen . Ma h. Hel .,84 (2009), 351–386.
[EsRo] J.M. Espina , H. Rosenbe g, Comple e cons an mean cu a u e su aces in ho-
mogeneous spaces, Commen . Ma h. Hel ., o appea (2009).
[FeMi1] I. Fe n´andez, P. Mi a, Ha monic maps and cons an mean cu a u e su aces in
H2×R,Ame . J. Ma h. 129 (2007), 1145–1181.
[FeMi2] I. Fe n´andez, P. Mi a, A cha ac e iza ion o cons an mean cu a u e su aces in
homogeneous 3-mani olds, Diff. Geom. Appl.,25 (2007), 281–289.
[FeMi3] I. Fe n´andez, P. Mi a, Holomo phic quad a ic diffe en ials and he Be ns ein
p oblem in Heisenbe g space. T ans. Ame . Ma h. Soc.,361, no 11, (2009), 5737–
5752.
[FoWo] A.P. Fo dy, J.C. Wood. Ha monic maps and in eg able sys ems. Aspec s o Ma h-
ema ics, ol. E23, by Vieweg, B aunschweig/Wiesbaden, 1994.
[GMM] J.A. G´al ez, A. Ma ´ınez, P. Mi a, The Bonne p oblem o su aces in homoge-
neous 3-mani olds, Comm. Anal. Geom. 16 (2008), 907–935.
[GaRo] J.A. G´al ez, H. Rosenbe g, Minimal su aces and ha monic diffeomo phisms om
he complex plane on o a Hadama d su ace. P ep in , 2008, a Xi :0807.0997.
[Ha] L. Hauswi h, Minimal su aces o Riemann ype in h ee dimensional p oduc man-
i olds. Pacific J. Ma h.,224, no.1 (2006), 91–117.
[HaRo] L. Hausw i h, H. Rosenbe g. Minimal su aces o fini e o al cu a u e in H×R.
Ma . Con emp. 31 (2006), 65–80.
[HRS] L. Hauswi h, H. Rosenbe g, J. Sp uck. On comple e mean cu a u e 1
2su aces
in H2×R.Comm. Anal. Geom.,16, no.5 (2008), 989–1005.
[HoMe] D. Hoffman, W. H. Meeks III. The s ong hal space heo em o minimal su aces.
In en . Ma h. 101, no.2 (1990), 373–377.
[HsHs] W.Y. Hsiang, W.T. Hsiang, On he uniqueness o isope ime ic solu ions and
imbedded soap bubbles in noncompac symme ic spaces I, In en . Ma h. 98 (1989),
39–58.
[Lee] H. Lee. Ex ension o he duali y be ween minimal su aces and maximal su aces.
P ep in , 2009.
[MaPR] M. Manzano, J. P´e ez, M. Rod ´ıguez. Pa abolic s able su aces wi h cons an
mean cu a u e. P ep in , 2009, a Xi :0910.5373.
[MPR] W. H. Meeks III, J. P´e ez, A. Ros. S able cons an mean cu a u e su aces,
Handbook o Geome ic Analysis no.1 (2008).
[Mee] W.H. Meeks. Cons an mean cu a u e su aces in homogeneous 3-mani olds.
P ep in (2009).
[MeRo1] W.H. Meeks, H. Rosenbe g, The heo y o minimal su aces in M×R,Commen .
Ma h. Hel . 80 (2005), 811–858.
[MeRo2] W.H. Meeks, H. Rosenbe g, S able minimal su aces in M×R,J. Diffe en ial
Geom. 68 (2004), 515–534.