Complete spacelike hypersurfaces in a Robertson-Walker spacetime
Alma L. Albujer
(joint work with Fernanda E. C. Camargo and Henrique F. de Lima)
University of C´ordoba
GELOSP’ 2013 S˜ao Paulo, July 22∼26, 2013
Partially supported by MICINN/FEDER project MTM2009-10418 and Fundaci´on S´eneca project 04540/GERM/06, Spain
Calabi-Bernstein theorem (1970)
Non-parametric version
The only entire maximal graphs inL3are the spacelike planes.
Equivalently, the only entire solutions to the maximal surface equation
Div Du
p1− |Du|2
!
= 0, |Du|2<1
inR2are the affine functions.
Parametric version
The only complete maximal surfaces inL3 are the spacelike planes.
? Cheng and Yau (1976) extended this theorem to general dimension.
? Since then several authors have approached this result from different points of view: McNertey (1980), Kobayashi (1983), Estudillo and Romero (1991, ’92 y ’94), Romero (1996), Al´ıas and Palmer (2001), Albujer and Al´ıas (2009), Romero and Rubio (2010), Caballero, Romero and Rubio (2010).
Calabi-Bernstein theorem (1970)
Non-parametric version
The only entire maximal graphs inL3are the spacelike planes.
Equivalently, the only entire solutions to the maximal surface equation
Div Du
p1− |Du|2
!
= 0, |Du|2<1
inR2are the affine functions.
Parametric version
The only complete maximal surfaces inL3 are the spacelike planes.
? Cheng and Yau (1976) extended this theorem to general dimension.
? Since then several authors have approached this result from different points of view: McNertey (1980), Kobayashi (1983), Estudillo and Romero (1991, ’92 y ’94), Romero (1996), Al´ıas and Palmer (2001), Albujer and Al´ıas (2009), Romero and Rubio (2010), Caballero, Romero and Rubio (2010).
Calabi-Bernstein theorem (1970)
Non-parametric version
The only entire maximal graphs inL3are the spacelike planes.
Equivalently, the only entire solutions to the maximal surface equation
Div Du
p1− |Du|2
!
= 0, |Du|2<1
inR2are the affine functions.
Parametric version
The only complete maximal surfaces inL3 are the spacelike planes.
? Cheng and Yau (1976) extended this theorem to general dimension.
? Since then several authors have approached this result from different points of view: McNertey (1980), Kobayashi (1983), Estudillo and Romero (1991, ’92 y ’94), Romero (1996), Al´ıas and Palmer (2001), Albujer and Al´ıas (2009), Romero and Rubio (2010), Caballero, Romero and Rubio (2010).
Calabi-Bernstein theorem (1970)
Non-parametric version
The only entire maximal graphs inL3are the spacelike planes.
Equivalently, the only entire solutions to the maximal surface equation
Div Du
p1− |Du|2
!
= 0, |Du|2<1
inR2are the affine functions.
Parametric version
The only complete maximal surfaces inL3 are the spacelike planes.
? Cheng and Yau (1976) extended this theorem to general dimension.
? Since then several authors have approached this result from different points of view: McNertey (1980), Kobayashi (1983), Estudillo and Romero (1991, ’92 y ’94), Romero (1996), Al´ıas and Palmer (2001), Albujer and Al´ıas (2009), Romero and Rubio (2010), Caballero,
Two generalizations of the Calabi-Bernstein theorem
Aiyama (1992), Xin (1991)
Let Σn be a complete spacelike hypersurface with constant mean curvature immersed intoLn+1. Suppose thatthe hyperbolic image of Σn is contained in a geodesic ball ofHn. Then Σn is a spacelike hyperplane.
? Hn is a spacelike hypersurface inLn+1with constant mean curvature which is not a spacelike hyperplane.
Albujer and Al´ıas (2009)
LetM2be a (necessarily complete) Riemannian surface with
non-negative Gaussian curvature,KM≥0. Then, any complete maximal surface Σ2in−R×M2 is totally geodesic. Moreover, ifKM>0 at some point onM, then Σ is a slice{t0} ×M, {t0} ∈R.
? This result is no longer true in−R×H2.
Two generalizations of the Calabi-Bernstein theorem
Aiyama (1992), Xin (1991)
Let Σn be a complete spacelike hypersurface with constant mean curvature immersed intoLn+1. Suppose thatthe hyperbolic image of Σn is contained in a geodesic ball ofHn. Then Σn is a spacelike hyperplane.
? Hnis a spacelike hypersurface inLn+1with constant mean curvature which is not a spacelike hyperplane.
Albujer and Al´ıas (2009)
LetM2be a (necessarily complete) Riemannian surface with
non-negative Gaussian curvature,KM≥0. Then, any complete maximal surface Σ2in−R×M2 is totally geodesic. Moreover, ifKM>0 at some point onM, then Σ is a slice{t0} ×M, {t0} ∈R.
? This result is no longer true in−R×H2.
Two generalizations of the Calabi-Bernstein theorem
Aiyama (1992), Xin (1991)
Let Σn be a complete spacelike hypersurface with constant mean curvature immersed intoLn+1. Suppose thatthe hyperbolic image of Σn is contained in a geodesic ball ofHn. Then Σn is a spacelike hyperplane.
? Hnis a spacelike hypersurface inLn+1with constant mean curvature which is not a spacelike hyperplane.
Albujer and Al´ıas (2009)
LetM2be a (necessarily complete) Riemannian surface with
non-negative Gaussian curvature,KM≥0. Then, any complete maximal surface Σ2in−R×M2 is totally geodesic. Moreover, ifKM >0 at some point onM, then Σ is a slice{t0} ×M, {t0} ∈R.
? This result is no longer true in−R×H2.
Two generalizations of the Calabi-Bernstein theorem
Aiyama (1992), Xin (1991)
Let Σn be a complete spacelike hypersurface with constant mean curvature immersed intoLn+1. Suppose thatthe hyperbolic image of Σn is contained in a geodesic ball ofHn. Then Σn is a spacelike hyperplane.
? Hnis a spacelike hypersurface inLn+1with constant mean curvature which is not a spacelike hyperplane.
Albujer and Al´ıas (2009)
LetM2be a (necessarily complete) Riemannian surface with
non-negative Gaussian curvature,KM≥0. Then, any complete maximal surface Σ2in−R×M2 is totally geodesic. Moreover, ifKM >0 at some point onM, then Σ is a slice{t0} ×M, {t0} ∈R.
? This result is no longer true in−R×H2.
In this talk...
... we study constant mean curvature spacelike
hypersurfaces in a Robertson-Walker spacetime under certain energy conditions.
... Our main tool is the Omori-Yau maximum principle.
In this talk...
... we study constant mean curvature spacelike
hypersurfaces in a Robertson-Walker spacetime under certain energy conditions.
... Our main tool is the Omori-Yau maximum principle.
The Omori-Yau maximum principle
Let f : [a,b]→Rbe a continuous function. Thenf attains its maximum at some pointp0∈[a,b].
? Ifp0∈(a,b) and f has continuous second derivative in a neighbourhood ofp0, then
f0(p0) = 0 and f00(p0)≤0.
Consider now a compact Riemannian manifoldM (without boundary) and consider any smooth function f ∈ C2(M). Thenf attains its maximum at some point p0∈M and
|∇f(p0)|= 0 and ∆f(p0)≤0. When M is not compact, a given functionf ∈ C2(M) with supMf <+∞does not necessarily attains its supremum. Omori-Yau maximum principle (Omori, 1967 and Yau, 1975)
LetMn be ann-dimensional complete Riemannian manifold whose Ricci curvature is bounded from below and considerf :Mn→Ra smooth function which is bounded from above onMn. Then there is a sequence of points{pk}k∈N⊂Mn such that
k→∞lim f(pk) = sup
M
f , lim
k→∞|∇f(pk)|= 0 and lim
k→∞∆f(pk)≤0.
The Omori-Yau maximum principle
Let f : [a,b]→Rbe a continuous function. Thenf attains its maximum at some pointp0∈[a,b].
? Ifp0∈(a,b) and f has continuous second derivative in a neighbourhood ofp0, then
f0(p0) = 0 and f00(p0)≤0.
Consider now a compact Riemannian manifoldM (without boundary) and consider any smooth function f ∈ C2(M). Thenf attains its maximum at some point p0∈M and
|∇f(p0)|= 0 and ∆f(p0)≤0. When M is not compact, a given functionf ∈ C2(M) with supMf <+∞does not necessarily attains its supremum. Omori-Yau maximum principle (Omori, 1967 and Yau, 1975)
LetMn be ann-dimensional complete Riemannian manifold whose Ricci curvature is bounded from below and considerf :Mn→Ra smooth function which is bounded from above onMn. Then there is a sequence of points{pk}k∈N⊂Mn such that
k→∞lim f(pk) = sup
M
f , lim
k→∞|∇f(pk)|= 0 and lim
k→∞∆f(pk)≤0.
The Omori-Yau maximum principle
Let f : [a,b]→Rbe a continuous function. Thenf attains its maximum at some pointp0∈[a,b].
? Ifp0∈(a,b) and f has continuous second derivative in a neighbourhood ofp0, then
f0(p0) = 0 and f00(p0)≤0.
Consider now a compact Riemannian manifoldM (without boundary) and consider any smooth function f ∈ C2(M). Thenf attains its maximum at some point p0∈M and
|∇f(p0)|= 0 and ∆f(p0)≤0.
When M is not compact, a given functionf ∈ C2(M) with supMf <+∞does not necessarily attains its supremum. Omori-Yau maximum principle (Omori, 1967 and Yau, 1975)
LetMn be ann-dimensional complete Riemannian manifold whose Ricci curvature is bounded from below and considerf :Mn→Ra smooth function which is bounded from above onMn. Then there is a sequence of points{pk}k∈N⊂Mn such that
k→∞lim f(pk) = sup
M
f , lim
k→∞|∇f(pk)|= 0 and lim
k→∞∆f(pk)≤0.
The Omori-Yau maximum principle
Let f : [a,b]→Rbe a continuous function. Thenf attains its maximum at some pointp0∈[a,b].
? Ifp0∈(a,b) and f has continuous second derivative in a neighbourhood ofp0, then
f0(p0) = 0 and f00(p0)≤0.
Consider now a compact Riemannian manifoldM (without boundary) and consider any smooth function f ∈ C2(M). Thenf attains its maximum at some point p0∈M and
|∇f(p0)|= 0 and ∆f(p0)≤0.
When M is not compact, a given functionf ∈ C2(M) with supMf <+∞does not necessarily attains its supremum.
Omori-Yau maximum principle (Omori, 1967 and Yau, 1975)
LetMn be ann-dimensional complete Riemannian manifold whose Ricci curvature is bounded from below and considerf :Mn→Ra smooth function which is bounded from above onMn. Then there is a sequence of points{pk}k∈N⊂Mn such that
k→∞lim f(pk) = sup
M
f , lim
k→∞|∇f(pk)|= 0 and lim
k→∞∆f(pk)≤0.
The Omori-Yau maximum principle
Let f : [a,b]→Rbe a continuous function. Thenf attains its maximum at some pointp0∈[a,b].
? Ifp0∈(a,b) and f has continuous second derivative in a neighbourhood ofp0, then
f0(p0) = 0 and f00(p0)≤0.
Consider now a compact Riemannian manifoldM (without boundary) and consider any smooth function f ∈ C2(M). Thenf attains its maximum at some point p0∈M and
|∇f(p0)|= 0 and ∆f(p0)≤0. When M is not compact, a given functionf ∈ C2(M) with supMf <+∞does not necessarily attains its supremum.
Omori-Yau maximum principle (Omori, 1967 and Yau, 1975)
LetMn be ann-dimensional complete Riemannian manifold whose Ricci curvature is bounded from below and considerf :Mn→Ra smooth function which is bounded from above onMn. Then there is a sequence of points{pk}k∈N⊂Mn such that
k→∞lim f(pk) = sup
M
f , lim
k→∞|∇f(pk)|= 0 and lim
k→∞∆f(pk)≤0.
Robertson-Walker spacetimes
Let (Mn,h,iM) be a Riemannian manifold of constant sectional curvature κ, I ⊆Ra real interval and f >0 a positive smooth function onI:
ARobertson-Walker spacetimeis the product manifold I ×Mnendowed with the Lorentzian metric
h,i=−dt2+f2h,iM.
? We denote it by−I ×f Mn.
−I ×f Mn has constant sectional curvatureκif and only ifMn has constant sectional curvatureκand
f00
f =κ=(f0)2+κ f2 .
Robertson-Walker spacetimes
Let (Mn,h,iM) be a Riemannian manifold of constant sectional curvature κ, I ⊆Ra real interval and f >0 a positive smooth function onI:
ARobertson-Walker spacetimeis the product manifold I ×Mnendowed with the Lorentzian metric
h,i=−dt2+f2h,iM.
?We denote it by −I ×f Mn.
−I ×f Mn has constant sectional curvatureκif and only ifMn has constant sectional curvatureκand
f00
f =κ=(f0)2+κ f2 .
Robertson-Walker spacetimes
Let (Mn,h,iM) be a Riemannian manifold of constant sectional curvature κ, I ⊆Ra real interval and f >0 a positive smooth function onI:
ARobertson-Walker spacetimeis the product manifold I ×Mnendowed with the Lorentzian metric
h,i=−dt2+f2h,iM.
?We denote it by −I ×f Mn.
−I ×f Mn has constant sectional curvatureκif and only ifMn has constant sectional curvatureκand
f00
f =κ=(f0)2+κ f2 .
Robertson-Walker spacetimes
Some energy conditions
A spacetime Mn+1 satisfies thetimelike convergence condition (TCC)ifRic(Z,Z)≥0, for all timelike vector Z.
A spacetime Mn+1 satisfies thenull convergence condition (NCC)ifRic(Z,Z)≥0, for all lightlike vectorZ.
? For the case of a Robertson-Walker spacetime:
TCC ⇔
f00≤0 κ≥sup
I
(ff00−f02)
NCC ⇔ κ≥sup
I
(ff00−f02)
If−I ×f Mn has constant sectional curvature it satisfiesNCC.
Robertson-Walker spacetimes
Some energy conditions
A spacetime Mn+1 satisfies thetimelike convergence condition (TCC)ifRic(Z,Z)≥0, for all timelike vector Z.
A spacetime Mn+1 satisfies thenull convergence condition (NCC)ifRic(Z,Z)≥0, for all lightlike vectorZ.
? For the case of a Robertson-Walker spacetime:
TCC ⇔
f00≤0 κ≥sup
I
(ff00−f02)
NCC ⇔ κ≥sup
I
(ff00−f02)
If−I ×f Mn has constant sectional curvature it satisfiesNCC.
Robertson-Walker spacetimes
Some energy conditions
A spacetime Mn+1 satisfies thetimelike convergence condition (TCC)ifRic(Z,Z)≥0, for all timelike vector Z.
A spacetime Mn+1 satisfies thenull convergence condition (NCC)ifRic(Z,Z)≥0, for all lightlike vectorZ.
? For the case of a Robertson-Walker spacetime:
TCC ⇔
f00≤0 κ≥sup
I
(ff00−f02)
NCC ⇔ κ≥sup
I
(ff00−f02)
If−I ×f Mn has constant sectional curvature it satisfiesNCC.
Robertson-Walker spacetimes
Some energy conditions
A spacetime Mn+1 satisfies thetimelike convergence condition (TCC)ifRic(Z,Z)≥0, for all timelike vector Z.
A spacetime Mn+1 satisfies thenull convergence condition (NCC)ifRic(Z,Z)≥0, for all lightlike vectorZ.
? For the case of a Robertson-Walker spacetime:
TCC ⇔
f00≤0 κ≥sup
I
(ff00−f02)
NCC ⇔ κ≥sup
I
(ff00−f02)
If−I ×f Mn has constant sectional curvature it satisfiesNCC.
Robertson-Walker spacetimes
Some energy conditions
A spacetime Mn+1 satisfies thetimelike convergence condition (TCC)ifRic(Z,Z)≥0, for all timelike vector Z.
A spacetime Mn+1 satisfies thenull convergence condition (NCC)ifRic(Z,Z)≥0, for all lightlike vectorZ.
? For the case of a Robertson-Walker spacetime:
TCC ⇔
f00≤0 κ≥sup
I
(ff00−f02)
NCC ⇔ κ≥sup
I
(ff00−f02)
If−I ×f Mn has constant sectional curvature it satisfiesNCC.
Spacelike hypersurfaces in −I ×
fM
nLet Σn be aspacelike hypersurfaceimmersed into−I ×f Mn.
? Since∂t = (∂/∂t)(t,x) is a unitary timelike vector field globally defined on−I×f Mn, then there exists a unique timelike unitary normal vector field Nglobally defined on Σn such that
hN, ∂ti ≤ −1<0 on Σn. N is called the future-pointing Gauss map ofΣnand
coshθ=−hN, ∂timeasures thenormal hyperbolic angleof Σn. Let hdenote theheight functionof Σn,h= (πI)>. It is not difficult to see that
∇h=−∂t>=−∂t+ coshθN and k∇hk2= cosh2θ−1. Givent0∈R, the spacelike hypersurface{t0} ×Hn is called aslice. Slices are characterized byθ≡0. Or equivalently,h=c∈R.
? Slices are totally umbilical spacelike hypersurfaces withH= ff0(t(t0)
0).
Spacelike hypersurfaces in −I ×
fM
nLet Σn be aspacelike hypersurfaceimmersed into−I ×f Mn.
? Since∂t = (∂/∂t)(t,x) is a unitary timelike vector field globally defined on−I×f Mn, then there exists a unique timelike unitary normal vector field Nglobally defined on Σn such that
hN, ∂ti ≤ −1<0 on Σn.
N is called the future-pointing Gauss map ofΣnand
coshθ=−hN, ∂timeasures thenormal hyperbolic angleof Σn. Let hdenote theheight functionof Σn,h= (πI)>. It is not difficult to see that
∇h=−∂t>=−∂t+ coshθN and k∇hk2= cosh2θ−1. Givent0∈R, the spacelike hypersurface{t0} ×Hn is called aslice. Slices are characterized byθ≡0. Or equivalently,h=c∈R.
? Slices are totally umbilical spacelike hypersurfaces withH= ff0(t(t0)
0).
Spacelike hypersurfaces in −I ×
fM
nLet Σn be aspacelike hypersurfaceimmersed into−I ×f Mn.
? Since∂t = (∂/∂t)(t,x) is a unitary timelike vector field globally defined on−I×f Mn, then there exists a unique timelike unitary normal vector field Nglobally defined on Σn such that
hN, ∂ti ≤ −1<0 on Σn. N is called the future-pointing Gauss map ofΣnand
coshθ=−hN, ∂timeasures thenormal hyperbolic angleof Σn.
Let hdenote theheight functionof Σn,h= (πI)>. It is not difficult to see that
∇h=−∂t>=−∂t+ coshθN and k∇hk2= cosh2θ−1. Givent0∈R, the spacelike hypersurface{t0} ×Hn is called aslice. Slices are characterized byθ≡0. Or equivalently,h=c∈R.
? Slices are totally umbilical spacelike hypersurfaces withH= ff0(t(t0)
0).
Spacelike hypersurfaces in −I ×
fM
nLet Σn be aspacelike hypersurfaceimmersed into−I ×f Mn.
? Since∂t = (∂/∂t)(t,x) is a unitary timelike vector field globally defined on−I×f Mn, then there exists a unique timelike unitary normal vector field Nglobally defined on Σn such that
hN, ∂ti ≤ −1<0 on Σn. N is called the future-pointing Gauss map ofΣnand
coshθ=−hN, ∂timeasures thenormal hyperbolic angleof Σn. Let hdenote the height functionof Σn,h= (πI)>. It is not difficult to see that
∇h=−∂t>=−∂t+ coshθN and k∇hk2= cosh2θ−1.
Givent0∈R, the spacelike hypersurface{t0} ×Hn is called aslice. Slices are characterized byθ≡0. Or equivalently,h=c∈R.
? Slices are totally umbilical spacelike hypersurfaces withH= ff0(t(t0)
0).
Spacelike hypersurfaces in −I ×
fM
nLet Σn be aspacelike hypersurfaceimmersed into−I ×f Mn.
? Since∂t = (∂/∂t)(t,x) is a unitary timelike vector field globally defined on−I×f Mn, then there exists a unique timelike unitary normal vector field Nglobally defined on Σn such that
hN, ∂ti ≤ −1<0 on Σn. N is called the future-pointing Gauss map ofΣnand
coshθ=−hN, ∂timeasures thenormal hyperbolic angleof Σn. Let hdenote the height functionof Σn,h= (πI)>. It is not difficult to see that
∇h=−∂t>=−∂t+ coshθN and k∇hk2= cosh2θ−1.
Givent0∈R, the spacelike hypersurface{t0} ×Hn is called aslice.
Slices are characterized byθ≡0. Or equivalently, h=c ∈R.
? Slices are totally umbilical spacelike hypersurfaces withH= ff0(t(t0)
0).
Spacelike hypersurfaces in −I ×
fM
nLet Σn be aspacelike hypersurfaceimmersed into−I ×f Mn.
? Since∂t = (∂/∂t)(t,x) is a unitary timelike vector field globally defined on−I×f Mn, then there exists a unique timelike unitary normal vector field Nglobally defined on Σn such that
hN, ∂ti ≤ −1<0 on Σn. N is called the future-pointing Gauss map ofΣnand
coshθ=−hN, ∂timeasures thenormal hyperbolic angleof Σn. Let hdenote the height functionof Σn,h= (πI)>. It is not difficult to see that
∇h=−∂t>=−∂t+ coshθN and k∇hk2= cosh2θ−1.
Givent0∈R, the spacelike hypersurface{t0} ×Hn is called aslice.
Slices are characterized byθ≡0. Or equivalently, h=c ∈R.
? Slices are totally umbilical spacelike hypersurfaces withH= ff0(t(t0)
0).
Spacelike hypersurfaces in −I ×
fM
nLet A:X(Σ)→X(Σ) be theshape operatorof Σn with respect to N and letκ1, ..., κnbe the principal curvatures of Σn.
Ther-mean curvature functionHr of the spacelike hypersurface Σn is defined by
n r
Hr(p) = (−1)r X
i1<...<ir
κi1(p)· · ·κir(p), 1≤r ≤n
? In the particular case when r = 1,H1=H=−1ntr(A) is the mean curvature of Σn.
? In the caser = 2,H2defines a geometric quantity related to the scalar curvature of the hypersurface.
Σn is said to be contained in atimelike bounded region if it is contained in a slab
Σ⊂[t1,t2]×Mn={(t,x)∈ −I×f Mn:t1≤t≤t2}.
Spacelike hypersurfaces in −I ×
fM
nLet A:X(Σ)→X(Σ) be theshape operatorof Σn with respect to N and letκ1, ..., κnbe the principal curvatures of Σn.
Ther-mean curvature functionHr of the spacelike hypersurface Σn is defined by
n r
Hr(p) = (−1)r X
i1<...<ir
κi1(p)· · ·κir(p), 1≤r ≤n
? In the particular case when r = 1,H1=H=−1ntr(A) is the mean curvature of Σn.
? In the caser = 2,H2defines a geometric quantity related to the scalar curvature of the hypersurface.
Σn is said to be contained in atimelike bounded region if it is contained in a slab
Σ⊂[t1,t2]×Mn={(t,x)∈ −I×f Mn:t1≤t≤t2}.
Spacelike hypersurfaces in −I ×
fM
nLet A:X(Σ)→X(Σ) be theshape operatorof Σn with respect to N and letκ1, ..., κnbe the principal curvatures of Σn.
Ther-mean curvature functionHr of the spacelike hypersurface Σn is defined by
n r
Hr(p) = (−1)r X
i1<...<ir
κi1(p)· · ·κir(p), 1≤r ≤n
? In the particular case when r = 1,H1=H=−1ntr(A) is the mean curvature of Σn.
? In the caser = 2,H2defines a geometric quantity related to the scalar curvature of the hypersurface.
Σn is said to be contained in atimelike bounded region if it is contained in a slab
Σ⊂[t1,t2]×Mn={(t,x)∈ −I×f Mn:t1≤t≤t2}.
Spacelike hypersurfaces in −I ×
fM
nLet A:X(Σ)→X(Σ) be theshape operatorof Σn with respect to N and letκ1, ..., κnbe the principal curvatures of Σn.
Ther-mean curvature functionHr of the spacelike hypersurface Σn is defined by
n r
Hr(p) = (−1)r X
i1<...<ir
κi1(p)· · ·κir(p), 1≤r ≤n
? In the particular case when r = 1,H1=H=−1ntr(A) is the mean curvature of Σn.
? In the caser = 2,H2defines a geometric quantity related to the scalar curvature of the hypersurface.
Σn is said to be contained in atimelike bounded region if it is contained in a slab
Σ⊂[t1,t2]×Mn={(t,x)∈ −I×f Mn:t1≤t≤t2}.
Calabi-Bernstein type results in −I ×
fM
nTheorem (Parametric version)
Let−I ×f Mn be a Robertson-Walker spacetime whose Riemannian fiber Mnhas constant sectional curvature κand such that it obeysNCC. Let Σn be a complete spacelike hypersurface of−I ×f Mn with constant mean curvatureH6= 0 andcontained in a timelike bounded region. If
0≤Hsup
Σ
f0 f ◦h
≤H2
and
k∇hk2≤α
H−sup
Σ
f0 f ◦h
β
for some positive constantsαandβ, then Σn is a slice.
Sketch of the proof:
• MAIN IDEA: To apply the O-Y maximum principle to f(h) coshθ.
? f(h) coshθ is bounded from abovesince Σnis contained in a timelike bounded region andk∇hk2≤α
H−supΣ
f0 f ◦h
β
.
? Let {E1, ...En}be a local orthonormal frame of X(Σ). Then, by the Gauss equation, for any X ∈X(Σ) it holds that
Ric(X,X) =X
i
hR(X,Ei)X,Ei)i+
AX +nH 2 X
2
−n2H2 4 kXk2 Ric(X,X) is bounded⇔P
ih,R(X,Ei)X,Eiiis bounded.
Sketch of the proof:
• MAIN IDEA: To apply the O-Y maximum principle to f(h) coshθ.
? f(h) coshθ is bounded from abovesince Σnis contained in a timelike bounded region andk∇hk2≤α
H−supΣ
f0 f ◦h
β
.
? Let {E1, ...En}be a local orthonormal frame of X(Σ). Then, by the Gauss equation, for any X ∈X(Σ) it holds that
Ric(X,X) =X
i
hR(X,Ei)X,Ei)i+
AX +nH 2 X
2
−n2H2 4 kXk2 Ric(X,X) is bounded⇔P
ih,R(X,Ei)X,Eiiis bounded.
Sketch of the proof:
• MAIN IDEA: To apply the O-Y maximum principle to f(h) coshθ.
? f(h) coshθ is bounded from abovesince Σnis contained in a timelike bounded region andk∇hk2≤α
H−supΣ
f0 f ◦h
β
.
? Let {E1, ...En}be a local orthonormal frame of X(Σ). Then, by the Gauss equation, for any X ∈X(Σ) it holds that
Ric(X,X) =X
i
hR(X,Ei)X,Ei)i+
AX +nH 2 X
2
−n2H2 4 kXk2
Ric(X,X) is bounded⇔P
ih,R(X,Ei)X,Eiiis bounded.
Sketch of the proof:
• MAIN IDEA: To apply the O-Y maximum principle to f(h) coshθ.
? f(h) coshθ is bounded from abovesince Σnis contained in a timelike bounded region andk∇hk2≤α
H−supΣ
f0 f ◦h
β
.
? Let {E1, ...En}be a local orthonormal frame of X(Σ). Then, by the Gauss equation, for any X ∈X(Σ) it holds that
Ric(X,X) =X
i
hR(X,Ei)X,Ei)i+
AX +nH 2 X
2
−n2H2 4 kXk2 Ric(X,X) is bounded⇔P
ih,R(X,Ei)X,Eiiis bounded.
Sketch of the proof:
We can compute X
i
hR(X,Ei)X,Eii= κ
f2(h)+f02(h) f2(h)
(n−1)kXk2
+ κ
f2(h)−(lnf)00(h)
(n−2)hX,∇hi2 +
κ
f2(h)−(lnf)00(h)
kXk2k∇hk2
Sketch of the proof:
We can compute X
i
hR(X,Ei)X,Eii= κ
f2(h)+f02(h) f2(h)
(n−1)kXk2
+
κ
f2(h)−(lnf)00(h)
(n−2)hX,∇hi2 +
κ
f2(h)−(lnf)00(h)
kXk2k∇hk2
Sketch of the proof:
We can compute X
i
hR(X,Ei)X,Eii= κ
f2(h)+f02(h) f2(h)
(n−1)kXk2
+ κ
f2(h)−(lnf)00(h)
(n−2)hX,∇hi2 +
κ
f2(h)−(lnf)00(h)
kXk2k∇hk2
≥ κ
f2(h)+f02(h) f2(h)
(n−1)kXk2
? Sinceκis constant and Σn is contained in a timelike bounded region P
ihR(X,Ei)X,Eiiis bounded from below.
Sketch of the proof:
We can compute X
i
hR(X,Ei)X,Eii= κ
f2(h)+f02(h) f2(h)
(n−1)kXk2
+ κ
f2(h)−(lnf)00(h)
(n−2)hX,∇hi2 +
κ
f2(h)−(lnf)00(h)
kXk2k∇hk2
≥ κ
f2(h)+f02(h) f2(h)
(n−1)kXk2
? Sinceκis constant and Σn is contained in a timelike bounded region P
ihR(X,Ei)X,Eiiis bounded from below.
Sketch of the proof:
There exists a sequence{pk}k∈N on Σnsuch that
k→∞lim (f(h(pk)) coshθ(pk)) = sup
p∈Σn
(f(h(p)) coshθ(p))
and
k→∞lim ∆ (f(h(pk)) coshθ(pk))≤0.
On the other hand,
∆(f(h) coshθ)=−nHf0(h)
+f(h) coshθ n2H2−n(n−1)H2
+ (n−1)f(h) coshθ κ
f2(h)−(lnf)00(h)
k∇hk2
Sketch of the proof:
There exists a sequence{pk}k∈N on Σnsuch that
k→∞lim (f(h(pk)) coshθ(pk)) = sup
p∈Σn
(f(h(p)) coshθ(p))
and
k→∞lim ∆ (f(h(pk)) coshθ(pk))≤0.
On the other hand,
∆(f(h) coshθ)=−nHf0(h)
+f(h) coshθ n2H2−n(n−1)H2
+ (n−1)f(h) coshθ κ
f2(h)−(lnf)00(h)
k∇hk2
Sketch of the proof:
There exists a sequence{pk}k∈N on Σnsuch that
k→∞lim (f(h(pk)) coshθ(pk)) = sup
p∈Σn
(f(h(p)) coshθ(p))
and
k→∞lim ∆ (f(h(pk)) coshθ(pk))≤0.
On the other hand,
∆(f(h) coshθ)=−nHf0(h)
+f(h) coshθ n2H2−n(n−1)H2
+ (n−1)f(h) coshθ κ
f2(h)−(lnf)00(h)
k∇hk2
Sketch of the proof:
There exists a sequence{pk}k∈N on Σnsuch that
k→∞lim (f(h(pk)) coshθ(pk)) = sup
p∈Σn
(f(h(p)) coshθ(p))
and
k→∞lim ∆ (f(h(pk)) coshθ(pk))≤0.
On the other hand,
∆(f(h) coshθ)≥ −ncoshθf(h)Hsup
Σ
f0 f ◦h
+f(h) coshθ n2H2−n(n−1)H2 + (n−1)f(h) coshθ
κ
f2(h)−(lnf)00(h)
k∇hk2
Sketch of the proof:
There exists a sequence{pk}k∈N on Σnsuch that
k→∞lim (f(h(pk)) coshθ(pk)) = sup
p∈Σn
(f(h(p)) coshθ(p))
and
k→∞lim ∆ (f(h(pk)) coshθ(pk))≤0.
On the other hand,
∆(f(h) coshθ)≥−ncoshθf(h)Hsup
Σ
f0 f ◦h
+f(h) coshθ n2H2−n(n−1)H2 + (n−1)f(h) coshθ
κ
f2(h)−(lnf)00(h)
k∇hk2
Sketch of the proof:
There exists a sequence{pk}k∈N on Σnsuch that
k→∞lim (f(h(pk)) coshθ(pk)) = sup
p∈Σn
(f(h(p)) coshθ(p))
and
k→∞lim ∆ (f(h(pk)) coshθ(pk))≤0.
On the other hand,
∆(f(h) coshθ)≥f(h) coshθnH
H−sup
Σ
f0 f ◦h
+n(n−1)f(h) coshθ H2−H2 + (n−1)f(h) coshθ
κ
f2(h)−(lnf)00(h)
k∇hk2
Sketch of the proof:
There exists a sequence{pk}k∈N on Σnsuch that
k→∞lim (f(h(pk)) coshθ(pk)) = sup
p∈Σn
(f(h(p)) coshθ(p))
and
k→∞lim ∆ (f(h(pk)) coshθ(pk))≤0.
On the other hand,
∆(f(h) coshθ)≥f(h) coshθnH
H−sup
Σ
f0 f ◦h
+n(n−1)f(h) coshθ H2−H2 + (n−1)f(h) coshθ
κ
f2(h)−(lnf)00(h)
k∇hk2
Sketch of the proof:
There exists a sequence{pk}k∈N on Σnsuch that
k→∞lim (f(h(pk)) coshθ(pk)) = sup
p∈Σn
(f(h(p)) coshθ(p))
and
k→∞lim ∆ (f(h(pk)) coshθ(pk))≤0
On the other hand,
∆(f(h) coshθ)≥f(h) coshθnH
H−sup
Σ
f0 f ◦h
+n(n−1)f(h) coshθ H2−H2 + (n−1)f(h) coshθ
κ
f2(h)−(lnf)00(h)
k∇hk2
≥0.
Sketch of the proof:
Therefore,
0≥ lim
k→∞(∆ (f(h(pk)) coshθ(pk)))≥0.
H
H−supΣ
f0
f ◦h
= 0.
? SinceH6= 0, H−supΣ
f0 f ◦h
= 0 k∇hk2= 0 Σn is a slice.
Sketch of the proof:
Therefore,
0≥ lim
k→∞(∆ (f(h(pk)) coshθ(pk)))≥0.
H
H−supΣ
f0
f ◦h
= 0.
? SinceH6= 0, H−supΣ
f0 f ◦h
= 0 k∇hk2= 0 Σn is a slice.
Sketch of the proof:
Therefore,
0≥ lim
k→∞(∆ (f(h(pk)) coshθ(pk)))≥0.
H
H−supΣ
f0
f ◦h
= 0.
? SinceH6= 0, H−supΣ
f0 f ◦h
= 0
k∇hk2= 0 Σn is a slice.
Sketch of the proof:
Therefore,
0≥ lim
k→∞(∆ (f(h(pk)) coshθ(pk)))≥0.
H
H−supΣ
f0
f ◦h
= 0.
? SinceH6= 0, H−supΣ
f0 f ◦h
= 0 k∇hk2= 0 Σn is a slice.
Calabi-Bernstein type results in −I ×
fM
nRemarks:
We do not need to ask Σn to be contained in a timelike bounded region. We just need thatf,f0 andf00 are bounded on Σnand infΣf(h(p))>0.
In order to conclude thatH= supΣ
f0 f ◦h
we only need to ask the normal hyperbolic angle to be bounded.
Therefore, for the case of a static Robertson Walker spacetime: Corollary
Let−I ×Mn be a static Robertson Walker spacetime whose Riemannian fiberMnhas non-negative constant sectional curvature. Let Σn be a complete spacelike hypersurface in−I×Mnwith constant mean curvature and bounded normal hyperbolic angle. Then, Σn is maximal.
? In this case
∆ coshθ≥ nH2+n(n−1)(H2−H2) + (n−1)κk∇hk2
coshθ≥0.
Calabi-Bernstein type results in −I ×
fM
nRemarks:
We do not need to ask Σn to be contained in a timelike bounded region. We just need thatf,f0 andf00 are bounded on Σnand infΣf(h(p))>0.
In order to conclude thatH= supΣ
f0 f ◦h
we only need to ask the normal hyperbolic angle to be bounded.
Therefore, for the case of a static Robertson Walker spacetime: Corollary
Let−I ×Mn be a static Robertson Walker spacetime whose Riemannian fiberMnhas non-negative constant sectional curvature. Let Σn be a complete spacelike hypersurface in−I×Mnwith constant mean curvature and bounded normal hyperbolic angle. Then, Σn is maximal.
? In this case
∆ coshθ≥ nH2+n(n−1)(H2−H2) + (n−1)κk∇hk2
coshθ≥0.
Calabi-Bernstein type results in −I ×
fM
nRemarks:
We do not need to ask Σn to be contained in a timelike bounded region. We just need thatf,f0 andf00 are bounded on Σnand infΣf(h(p))>0.
In order to conclude thatH= supΣ
f0 f ◦h
we only need to ask the normal hyperbolic angle to be bounded.
Therefore, for the case of a static Robertson Walker spacetime:
Corollary
Let−I ×Mn be a static Robertson Walker spacetime whose Riemannian fiberMnhas non-negative constant sectional curvature. Let Σn be a complete spacelike hypersurface in−I×Mnwith constant mean curvature and bounded normal hyperbolic angle. Then, Σn is maximal.
? In this case
∆ coshθ≥ nH2+n(n−1)(H2−H2) + (n−1)κk∇hk2
coshθ≥0.
Calabi-Bernstein type results in −I ×
fM
nRemarks:
We do not need to ask Σn to be contained in a timelike bounded region. We just need thatf,f0 andf00 are bounded on Σnand infΣf(h(p))>0.
In order to conclude thatH= supΣ
f0 f ◦h
we only need to ask the normal hyperbolic angle to be bounded.
Therefore, for the case of a static Robertson Walker spacetime:
Corollary
Let−I ×Mn be a static Robertson Walker spacetime whose Riemannian fiberMnhas non-negative constant sectional curvature. Let Σn be a complete spacelike hypersurface in−I×Mnwith constant mean curvature and bounded normal hyperbolic angle. Then, Σn is maximal.
? In this case
∆ coshθ≥ nH2+n(n−1)(H2−H2) + (n−1)κk∇hk2
coshθ≥0.
Calabi-Bernstein type results in −I ×
fM
nIn the particular case when Σn is an immersed spacelike hypersurface in Ln+1, we haveN: Σn→Hn where
Hn={x∈Ln+1:hx,xi=−1,x1≥1}.
? N(Σ) is called thehyperbolic image ofΣn.
A geodesic ball B(a, %) inHn of radius% >0 centered ata∈Hn is B(a, %) ={p∈Hn:−cosh%≤ hp,ai ≤ −1}.
The normal hyperbolic angle of Σn is bounded if and only if N(Σ)⊆B(a, %) for certaina∈Hn and% >0.
Corollary (Aiyama (1992), Xin (1991))
Let Σn be a complete spacelike hypersurface with constant mean curvature immersed intoLn+1. Suppose that the hyperbolic image of Σn is contained in a geodesic ball ofHn. Then Σn is a hyperplane.
Calabi-Bernstein type results in −I ×
fM
nIn the particular case when Σn is an immersed spacelike hypersurface in Ln+1, we haveN: Σn→Hn where
Hn={x∈Ln+1:hx,xi=−1,x1≥1}.
? N(Σ) is called thehyperbolic image ofΣn.
A geodesic ball B(a, %) inHn of radius% >0 centered ata∈Hn is B(a, %) ={p∈Hn:−cosh%≤ hp,ai ≤ −1}.
The normal hyperbolic angle of Σn is bounded if and only if N(Σ)⊆B(a, %) for certaina∈Hn and% >0.
Corollary (Aiyama (1992), Xin (1991))
Let Σn be a complete spacelike hypersurface with constant mean curvature immersed intoLn+1. Suppose that the hyperbolic image of Σn is contained in a geodesic ball ofHn. Then Σn is a hyperplane.
Calabi-Bernstein type results in −I ×
fM
nIn the particular case when Σn is an immersed spacelike hypersurface in Ln+1, we haveN: Σn→Hn where
Hn={x∈Ln+1:hx,xi=−1,x1≥1}.
? N(Σ) is called thehyperbolic image ofΣn.
A geodesic ball B(a, %) inHn of radius% >0 centered ata∈Hn is B(a, %) ={p∈Hn:−cosh%≤ hp,ai ≤ −1}.
The normal hyperbolic angle of Σn is bounded if and only if N(Σ)⊆B(a, %) for certaina∈Hn and% >0.
Corollary (Aiyama (1992), Xin (1991))
Let Σn be a complete spacelike hypersurface with constant mean curvature immersed intoLn+1. Suppose that the hyperbolic image of Σn is contained in a geodesic ball ofHn. Then Σn is a hyperplane.
Calabi-Bernstein type results in −I ×
fM
nIn the particular case when Σn is an immersed spacelike hypersurface in Ln+1, we haveN: Σn→Hn where
Hn={x∈Ln+1:hx,xi=−1,x1≥1}.
? N(Σ) is called thehyperbolic image ofΣn.
A geodesic ball B(a, %) inHn of radius% >0 centered ata∈Hn is B(a, %) ={p∈Hn:−cosh%≤ hp,ai ≤ −1}.
The normal hyperbolic angle of Σn is bounded if and only if N(Σ)⊆B(a, %) for certaina∈Hn and% >0.
Corollary (Aiyama (1992), Xin (1991))
Let Σn be a complete spacelike hypersurface with constant mean curvature immersed intoLn+1. Suppose that the hyperbolic image of Σn is contained in a geodesic ball ofHn. Then Σn is a hyperplane.
Calabi-Bernstein type results in −I ×
fM
nIn the particular case when Σn is an immersed spacelike hypersurface in Ln+1, we haveN: Σn→Hn where
Hn={x∈Ln+1:hx,xi=−1,x1≥1}.
? N(Σ) is called thehyperbolic image ofΣn.
A geodesic ball B(a, %) inHn of radius% >0 centered ata∈Hn is B(a, %) ={p∈Hn:−cosh%≤ hp,ai ≤ −1}.
The normal hyperbolic angle of Σn is bounded if and only if N(Σ)⊆B(a, %) for certaina∈Hn and% >0.
Corollary (Aiyama (1992), Xin (1991))
Let Σn be a complete spacelike hypersurface with constant mean curvature immersed intoLn+1. Suppose that the hyperbolic image of Σn is contained in a geodesic ball ofHn. Then Σn is a hyperplane.
Entire spacelike vertical graphs in −I ×
fM
nAnentire vertical graphin−I×f Mn is determined by a smooth functionu∈ C∞(M) and it is given by
Σn(u) ={(u(x),x) :x∈Mn} ⊂ −I ×f Mn.
? The metric induced onMn from the Lorentzian metric of−I×f Mn via the graph is
h,i=−du2+f2(u)h,iMn.
? The graph Σn(u) is a spacelike hypersurface iff|Du|2<f2(u). For any entire spacelike graph we have the following relation:
k∇hk2= |Du|2 f2(u)− |Du|2.
It is well known that an entire spacelike graph is not necessarily complete.
? In fact, there exist examples of entire non-complete graphs in
−R×Hnwhich are maximal (Albujer, 2008 and Albujer, Al´ıas, 2009) or spacelike with constant mean curvature (Alarc´on, Souam).
Entire spacelike vertical graphs in −I ×
fM
nAnentire vertical graphin−I×f Mn is determined by a smooth functionu∈ C∞(M) and it is given by
Σn(u) ={(u(x),x) :x∈Mn} ⊂ −I ×f Mn.
? The metric induced onMn from the Lorentzian metric of−I×f Mn via the graph is
h,i=−du2+f2(u)h,iMn.
? The graph Σn(u) is a spacelike hypersurface iff|Du|2<f2(u). For any entire spacelike graph we have the following relation:
k∇hk2= |Du|2 f2(u)− |Du|2.
It is well known that an entire spacelike graph is not necessarily complete.
? In fact, there exist examples of entire non-complete graphs in
−R×Hnwhich are maximal (Albujer, 2008 and Albujer, Al´ıas, 2009) or spacelike with constant mean curvature (Alarc´on, Souam).
Entire spacelike vertical graphs in −I ×
fM
nAnentire vertical graphin−I×f Mn is determined by a smooth functionu∈ C∞(M) and it is given by
Σn(u) ={(u(x),x) :x∈Mn} ⊂ −I ×f Mn.
? The metric induced onMn from the Lorentzian metric of−I×f Mn via the graph is
h,i=−du2+f2(u)h,iMn.
? The graph Σn(u) is a spacelike hypersurface iff|Du|2<f2(u).
For any entire spacelike graph we have the following relation: k∇hk2= |Du|2
f2(u)− |Du|2.
It is well known that an entire spacelike graph is not necessarily complete.
? In fact, there exist examples of entire non-complete graphs in
−R×Hnwhich are maximal (Albujer, 2008 and Albujer, Al´ıas, 2009) or spacelike with constant mean curvature (Alarc´on, Souam).