Geometry of Centroaffine Surfaces in R⁵
We use Cartan's method of moving frames to compute a complete set of local invariants for nondegenerate, 2-dimensional centroaffine surfaces in R⁵∖{0} with nondegenerate centroaffine metric. We then give a complete classification of all homogeneous centroaffine surfaces in this class.
Gespeichert in:
| Datum: | 2015 |
|---|---|
| Hauptverfasser: | , |
| Format: | Artikel |
| Sprache: | English |
| Veröffentlicht: |
Інститут математики НАН України
2015
|
| Schriftenreihe: | Symmetry, Integrability and Geometry: Methods and Applications |
| Online Zugang: | http://dspace.nbuv.gov.ua/handle/123456789/146863 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Назва журналу: | Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| Zitieren: | Geometry of Centroaffine Surfaces in R⁵ / N. Bushek // Symmetry, Integrability and Geometry: Methods and Applications. — 2015. — Т. 11. — Бібліогр.: 23 назв. — англ. |
Institution
Digital Library of Periodicals of National Academy of Sciences of Ukraine| id |
irk-123456789-146863 |
|---|---|
| record_format |
dspace |
| spelling |
irk-123456789-1468632019-02-12T01:24:35Z Geometry of Centroaffine Surfaces in R⁵ Bushek, N. Clelland, J.N. We use Cartan's method of moving frames to compute a complete set of local invariants for nondegenerate, 2-dimensional centroaffine surfaces in R⁵∖{0} with nondegenerate centroaffine metric. We then give a complete classification of all homogeneous centroaffine surfaces in this class. 2015 Article Geometry of Centroaffine Surfaces in R⁵ / N. Bushek // Symmetry, Integrability and Geometry: Methods and Applications. — 2015. — Т. 11. — Бібліогр.: 23 назв. — англ. 1815-0659 2010 Mathematics Subject Classification: 53A15; 58A15 DOI:10.3842/SIGMA.2015.001 http://dspace.nbuv.gov.ua/handle/123456789/146863 en Symmetry, Integrability and Geometry: Methods and Applications Інститут математики НАН України |
| institution |
Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| collection |
DSpace DC |
| language |
English |
| description |
We use Cartan's method of moving frames to compute a complete set of local invariants for nondegenerate, 2-dimensional centroaffine surfaces in R⁵∖{0} with nondegenerate centroaffine metric. We then give a complete classification of all homogeneous centroaffine surfaces in this class. |
| format |
Article |
| author |
Bushek, N. Clelland, J.N. |
| spellingShingle |
Bushek, N. Clelland, J.N. Geometry of Centroaffine Surfaces in R⁵ Symmetry, Integrability and Geometry: Methods and Applications |
| author_facet |
Bushek, N. Clelland, J.N. |
| author_sort |
Bushek, N. |
| title |
Geometry of Centroaffine Surfaces in R⁵ |
| title_short |
Geometry of Centroaffine Surfaces in R⁵ |
| title_full |
Geometry of Centroaffine Surfaces in R⁵ |
| title_fullStr |
Geometry of Centroaffine Surfaces in R⁵ |
| title_full_unstemmed |
Geometry of Centroaffine Surfaces in R⁵ |
| title_sort |
geometry of centroaffine surfaces in r⁵ |
| publisher |
Інститут математики НАН України |
| publishDate |
2015 |
| url |
http://dspace.nbuv.gov.ua/handle/123456789/146863 |
| citation_txt |
Geometry of Centroaffine Surfaces in R⁵ / N. Bushek // Symmetry, Integrability and Geometry: Methods and Applications. — 2015. — Т. 11. — Бібліогр.: 23 назв. — англ. |
| series |
Symmetry, Integrability and Geometry: Methods and Applications |
| work_keys_str_mv |
AT bushekn geometryofcentroaffinesurfacesinr5 AT clellandjn geometryofcentroaffinesurfacesinr5 |
| first_indexed |
2025-07-11T00:48:02Z |
| last_indexed |
2025-07-11T00:48:02Z |
| _version_ |
1837309744894705664 |
| fulltext |
Symmetry, Integrability and Geometry: Methods and Applications SIGMA 11 (2015), 001, 24 pages
Geometry of Centroaffine Surfaces in R5
Nathaniel BUSHEK † and Jeanne N. CLELLAND ‡
† Department of Mathematics, UNC - Chapel Hill,
CB #3250, Phillips Hall, Chapel Hill, NC 27599, USA
E-mail: bushek@unc.edu
‡ Department of Mathematics, 395 UCB, University of Colorado, Boulder, CO 80309-0395, USA
E-mail: Jeanne.Clelland@colorado.edu
Received August 23, 2014, in final form December 26, 2014; Published online January 06, 2015
http://dx.doi.org/10.3842/SIGMA.2015.001
Abstract. We use Cartan’s method of moving frames to compute a complete set of local
invariants for nondegenerate, 2-dimensional centroaffine surfaces in R5 \ {0} with nonde-
generate centroaffine metric. We then give a complete classification of all homogeneous
centroaffine surfaces in this class.
Key words: centroaffine geometry; Cartan’s method of moving frames
2010 Mathematics Subject Classification: 53A15; 58A15
1 Introduction
An immersion f̄ : M → Rn \ {0} is called a centroaffine immersion (cf. Definition 2.1) if the
position vector f̄(x) is transversal to the tangent space f̄∗(TxM) for all x ∈ M . Centroaffine
geometry is the study of those properties of centroaffine immersions that are invariant under
the action of the centroaffine group GL(n,R) on Rn \ {0}. Much attention has been given to the
study of centroaffine curves and hypersurfaces [5, 8, 9, 10, 11, 17, 18, 20], and more recently
to the study of centroaffine immersions of codimension 2 [3, 4, 13, 15, 16, 19, 22, 23]. In [21],
applications of centroaffine geometry to the problem of feedback equivalence in control theory
are discussed.
In this paper, we consider the case of a 2-dimensional centroaffine surface in R5 \ {0}. We
will use Cartan’s method of moving frames to construct a complete set of local invariants for
a large class of such surfaces under certain nondegeneracy assumptions (cf. Definition 3.1, As-
sumption 3.3). In addition, we give a complete classification of the homogeneous centroaffine
surfaces in this class – i.e., those that admit a 3-dimensional Lie group of symmetries that acts
transitively on an adapted frame bundle canonically associated to the surface (cf. Definition 6.1).
Our primary results are Theorems 4.3 and 5.3, which describe local invariants for centroaffine
surfaces with definite and indefinite centroaffine metrics, respectively, and Theorem 6.6, which
describes the homogeneous examples.
The paper is organized as follows. In Section 2, we introduce the basic concepts of centroaffine
geometry and centroaffine surfaces in R5 \ {0}, including the centroaffine frame bundle and the
Maurer–Cartan forms. In Section 3, we begin the method of moving frames and identify first-
order invariants for centroaffine surfaces. Based on these invariants, nondegenerate surfaces
may be locally classified as “space-like”, “time-like”, or “null”. In Sections 4 and 5, we continue
the method of moving frames for the space-like and time-like cases, respectively. (We do not
consider the null case here; it may be explored in a future paper.) Finally, in Section 6 we
classify the homogeneous examples in both the space-like and time-like cases.
mailto:bushek@unc.edu
mailto:Jeanne.Clelland@colorado.edu
http://dx.doi.org/10.3842/SIGMA.2015.001
2 N. Bushek and J.N. Clelland
2 Centroaffine surfaces in R5, adapted frames,
and Maurer–Cartan forms
Five-dimensional centroaffine space is the manifold R5 \ {0}, equipped with a natural GL(5,R)-
action. Specifically, GL(5,R) acts on R5 \ {0} by left multiplication: for x ∈ R5 \ {0}, g ∈
GL(5,R), we have
g · x = gx.
The group GL(5,R) may be regarded as a principal bundle over R5 \ {0}: write an arbitrary
element g ∈ GL(5,R) as
g =
[
e0 e1 e2 e3 e4
]
,
where e0, . . . , e4 ∈ R5 \ {0} are linearly independent column vectors. Then define the bundle
map π : GL(5,R)→ R5 \ {0} by
π
([
e0 e1 e2 e3 e4
])
= e0. (2.1)
The fiber group H is isomorphic to the stabilizer of the point[
1 0 0 0 0
]T ∈ R5 \ {0},
and this construction endows the manifold R5 \{0} with the structure of the homogeneous space
GL(5,R)/H.
We also think of the bundle π : GL(5,R)→ R5 \ {0} as the centroaffine frame bundle F over
R5 \ {0}. For each point e0 ∈ R5 \ {0}, the fiber over e0 consists of all frames (e0, e1, e2, e3, e4)
for the tangent space Te0(R5 \ {0}) – i.e., all frames for which the first vector in the frame is
equal to the position vector.
The Maurer–Cartan forms ωij on F are defined by the equations
dei = ejω
j
i , 0 ≤ i, j ≤ 4, (2.2)
and they satisfy the Cartan structure equations
dωij = −ωik ∧ ωkj . (2.3)
(For details, see [7, p. 18] or [1].)
We are interested in the geometry of 2-dimensional immersions f̄ : M2 → R5 \ {0}; we will
use Cartan’s method of moving frames to compute local invariants for such immersions under
the action of GL(5,R).
Definition 2.1. An immersion f̄ of a 2-dimensional manifold M into R5 \ {0} is called a cen-
troaffine immersion if the position vector f̄(x) is transversal to the tangent space f̄∗(TxM) for
all x ∈M . The image Σ = f̄(M) is called a centroaffine surface in R5 \ {0}.
In order to begin the method of moving frames, consider the induced bundle of centroaffine
frames along Σ = f̄(M); this is simply the pullback bundle F0 = f̄∗F over M . A centroaffine
frame field along Σ is a section of F0 – i.e., a smooth map f : M → GL(5,R) such that π◦f = f̄ .
Throughout the remainder of this paper, we will consider the pullbacks of the Maurer–Cartan
forms on F to M via such sections f , and we will suppress the pullback notation.
We will gradually adapt our choice of centroaffine frame fields based on the geometry of Σ.
For our first adaptation, consider the subbundle F1 ⊂ F0 consisting of all frames for which
Geometry of Centroaffine Surfaces in R5 3
(e1(x), e2(x)) span the tangent space Tf̄(x)Σ for each x ∈ M . A section f : M → F1 will be
called a 1-adapted frame field along Σ. Any two 1-adapted frames (e0, . . . , e4), (ẽ0, . . . , ẽ4) based
at the same point x ∈M are related by a transformation of the form
[
ẽ0 ẽ1 ẽ2 ẽ3 ẽ4
]
=
[
e0 e1 e2 e3 e4
]
1 0 0 r03 r04
0 a11 a12 r13 r14
0 a21 a22 r23 r24
0 0 0 b33 b34
0 0 0 b43 b44
, (2.4)
where the 2× 2 submatrices
A =
[
a11 a12
a21 a22
]
, B =
[
b33 b34
b43 b44
]
are elements of GL(2,R). We will denote the group of all matrices of the form in (2.4) by G1;
then the bundle F1 is a principal bundle over M with fiber group G1.
If f, f̃ : M → F1 are two 1-adapted frame fields along Σ, then f , f̃ are related by the equation
f̃(x) = f(x) · g(x)
for some smooth function g : M → G1, as in equation (2.4). Then the corresponding gl(5,R)-
valued Maurer–Cartan forms Ω = [ωij ], Ω̃ = [ω̃ij ] on M are related as follows:
Ω̃ = g−1dg + g−1Ωg. (2.5)
3 Reduction of the structure group and first-order invariants
Now consider the pullbacks of the Maurer–Cartan forms to M via a 1-adapted frame field f .
From equation (2.2) for de0 and the fact that the image of de0 is spanned by e1 and e2, we have
ω0
0 = ω3
0 = ω4
0 = 0. (3.1)
Moreover, the 1-forms ω1
0, ω2
0 are semi-basic for the projection π : F1 → M ; in fact, they form
a basis for the semi-basic 1-forms on F1.
Differentiating equations (3.1) yields:
0 = dω0
0 = −
(
ω0
1 ∧ ω1
0 + ω0
2 ∧ ω2
0
)
,
0 = dω3
0 = −
(
ω3
1 ∧ ω1
0 + ω3
2 ∧ ω2
0
)
,
0 = dω4
0 = −
(
ω4
1 ∧ ω1
0 + ω4
2 ∧ ω2
0
)
.
Cartan’s lemma (see, e.g., [7] or [1]) then implies that there exist functions hkij = hkji, k = 0, 3, 4,
on M such that[
ωk1
ωk2
]
=
[
hk11 hk12
hk12 hk22
][
ω1
0
ω2
0
]
.
For simplicity of notation, let hk denote the matrix
hk =
[
hk11 hk12
hk12 hk22
]
, k = 0, 3, 4.
4 N. Bushek and J.N. Clelland
If f, f̃ : M → F1 are two 1-adapted frame fields related by a transformation of the form (2.4),
then we can use equation (2.5) to determine how the corresponding matrices hk, h̃k are related.
First, it follows from the fact that ẽ0 = e0 that
dẽ0 =
[
ẽ1 ẽ2
] [ω̃1
0
ω̃2
0
]
=
[
e1 e2
] [ω1
0
ω2
0
]
= de0.
Then, since we have[
ẽ1 ẽ2
]
=
[
e1 e2
]
A,
we must have[
ω̃1
0
ω̃2
0
]
= A−1
[
ω1
0
ω2
0
]
. (3.2)
Similar considerations show that[
ω̃3
1
ω̃3
2
]
=
1
(detB)
AT
(
b44
[
ω3
1
ω3
2
]
− b34
[
ω4
1
ω4
2
])
,[
ω̃4
1
ω̃4
2
]
=
1
(detB)
AT
(
−b43
[
ω3
1
ω3
2
]
+ b33
[
ω4
1
ω4
2
])
,[
ω̃0
1
ω̃0
2
]
= AT
[
ω0
1
ω0
2
]
− r03
[
ω̃3
1
ω̃3
2
]
− r04
[
ω̃4
1
ω̃4
2
]
. (3.3)
Together, equations (3.2), (3.3) imply that
h̃3 =
1
(detB)
AT
(
b44h
3 − b34h
4
)
A, h̃4 =
1
(detB)
AT
(
−b43h
3 + b33h
4
)
A,
h̃0 = ATh0A− r03h̃
3 − r04h̃
4. (3.4)
Definition 3.1. A centroaffine surface Σ = f̄(M) will be called nondegenerate if the matri-
ces h0, h3, h4 are linearly independent in Sym2(R) at every point of M .
Henceforth, we assume that Σ is nondegenerate; from the group action (3.4) it is clear that
this definition is independent of the choice of 1-adapted frame field f : M → F1 along Σ.
The next step is to use the group action (3.4) to find normal forms for the matrices h0, h3, h4.
First consider the action on h3, h4: it can be written as the composition of two separate actions
by the matrices A,B ∈ GL(2,R):
A ·
(
h3, h4
)
=
(
ATh3A,ATh4A
)
,
B ·
(
h3, h4
)
=
(
1
(detB)
(
b44h
3 − b34h
4
)
,
1
(detB)
(
−b43h
3 + b33h
4
))
.
If we let P denote the 2-dimensional subspace of Sym2(R) spanned by (h3, h4), then we see that
the action by B preserves P , while A acts on P via
A · P = ATPA. (3.5)
In order to understand the action (3.5), consider the related action
A · h = AThA (3.6)
Geometry of Centroaffine Surfaces in R5 5
on Sym2(R). It is shown in [12, p. 115] that this action preserves the indefinite quadratic form
Q(h) = −det(h) (3.7)
up to a scale factor. More precisely, Q gives Sym2(R) the structure of the Minkowski space R2,1,
and the action (3.6) gives a representation of GL(2,R) as the group CSO+(2, 1) of orientation-
preserving, orthochronous, conformal Minkowski transformations of R2,1. This action has pre-
cisely 6 orbits, represented by the matrices[
0 0
0 0
]
,
[
1 0
0 0
]
,
[
−1 0
0 0
]
,
[
1 0
0 1
]
,
[
1 0
0 −1
]
,
[
−1 0
0 −1
]
. (3.8)
In terms of the Minkowski metric on Sym2(R) determined by Q, the second and third matrices
in (3.8) represent the two orbits of null vectors (oriented in opposite time directions); the fourth
and fifth represent the two orbits of time-like vectors (oriented in opposite time directions),
while the sixth represents the single orbit of space-like vectors.
The action (3.6) induces the action (3.5) on the Grassmannian of 2-planes in Sym2(R), which
corresponds to the action of CSO+(2, 1) on planes in R2,1. This action has precisely 3 orbits,
consisting of planes that are space-like, time-like, or null with respect to the quadratic form (3.7).
These orbits are represented by the 2-planes
P1 = span
([
1 0
0 −1
]
,
[
0 1
1 0
])
(space-like),
P2 = span
([
1 0
0 1
]
,
[
1 0
0 −1
])
(time-like),
P3 = span
([
1 0
0 0
]
,
[
0 1
1 0
])
(null). (3.9)
The type of the plane P spanned by (h3, h4) (space-like, time-like, or null) is preserved by the
group action (3.4); thus the type of P at any point x ∈ M is well-defined, independent of the
choice of 1-adapted frame field f : M → F1.
Remark 3.2. We note that similar quadratic forms h3, h4 were obtained by Nomizu and
Vrancken in their study of affine surfaces in R4; see [14].
At this point, the method of moving frames dictates that we divide into cases based on the
type of P . In order to proceed, we make the following assumption:
Assumption 3.3. Assume that Σ has constant type – i.e., that the type of P is the same at
every point x ∈M .
In this paper we will consider only the space-like and time-like cases; the null case is consid-
erably more complicated and may be explored in a future paper.
4 The space-like case
First, suppose that the plane P spanned by (h3, h4) is space-like at every point of M . According
to the group action (3.5), we can find a 1-adapted frame field along Σ for which P = P1, as in
equation (3.9). Furthermore, we can then use the action by B to find a 1-adapted frame field
along Σ for which
h3 =
[
1 0
0 −1
]
, h4 =
[
0 1
1 0
]
. (4.1)
6 N. Bushek and J.N. Clelland
The next step is to determine the subgroup of G1 that preserves the conditions (4.1). To this
end, first note that the plane P1 spanned by the matrices (4.1) consists precisely of the trace-free
matrices in Sym2(R). A straightforward computation shows that, with h3, h4 as in (4.1),
tr
(
ATh3A
)
= a2
11 + a2
12 − a2
21 − a2
22, tr
(
ATh4A
)
= 2(a11a21 + a12a22).
Therefore, the action (3.5) preserves P1 if and only if
a2
11 + a2
12 − a2
21 − a2
22 = a11a21 + a12a22 = 0,
which is true if and only if
A = λA0
for some λ ∈ R∗, A0 ∈ O(2,R). Substituting this condition into equations (3.4) and imposing
the conditions h̃3 = h3, h̃4 = h4 then yields
B = λ2(A0)2.
For simplicity, we will restrict to transformations with A0 ∈ SO(2,R), λ > 0; this has the
advantage of producing a frame bundle whose fiber is a connected Lie group. Thus we will
assume that
A =
[
λ cos(θ) −λ sin(θ)
λ sin(θ) λ cos(θ)
]
, B =
[
λ2 cos(2θ) −λ2 sin(2θ)
λ2 sin(2θ) λ2 cos(2θ)
]
, (4.2)
where λ > 0, θ ∈ R.
Next, consider the effect of the action (3.4) on h0. With A = I2 and r03, r04 chosen ap-
propriately, we can add any linear combination of h3, h4 to h0. Thus we can find a 1-adapted
frame field for which h0 is a multiple (nonzero by the nondegeneracy assumption) of the identity
matrix I2. Then under the action (3.4) with A, B as in (4.2), we have
h̃0 = λ2h0 +
[
−r03 −r04
−r04 r03
]
.
Therefore, we can find a 1-adapted frame field along Σ satisfying the additional condition that
h0 = ±I2, and this condition is preserved by transformations of the form (3.4) with A, B as
in (4.2), λ = 1, and and r03 = r04 = 0.
Definition 4.1. Let Σ = f̄(M) be a nondegenerate, space-like centroaffine surface in R5 \ {0}.
A 1-adapted frame field f : M → F1 will be called 2-adapted if it satisfies the conditions
h3 =
[
1 0
0 −1
]
, h4 =
[
0 1
1 0
]
, h0 =
[
ε 0
0 ε
]
, (4.3)
with ε = ±1, at every point of M .
Any two 2-adapted frames (e0, . . . , e4), (ẽ0, . . . , ẽ4) based at the same point x ∈M are related
by a transformation of the form[
ẽ0 ẽ1 ẽ2 ẽ3 ẽ4
]
=
[
e0 e1 e2 e3 e4
]
1 0 0 0 0
0 cos(θ) − sin(θ) r13 r14
0 sin(θ) cos(θ) r23 r24
0 0 0 cos(2θ) − sin(2θ)
0 0 0 sin(2θ) cos(2θ)
. (4.4)
Geometry of Centroaffine Surfaces in R5 7
We will denote the group of all matrices of the form in (4.4) by G2; then the 2-adapted frame
fields along Σ are the smooth sections of a principal bundle F2 ⊂ F1 over M with fiber group G2.
The equations (4.3) are equivalent to the condition that the Maurer–Cartan forms associated
to a 2-adapted frame field satisfy the conditions
ω3
1 = ω1
0, ω3
2 = −ω2
0, ω4
1 = ω2
0, ω4
2 = ω1
0, ω0
1 = εω1
0, ω0
2 = εω2
0. (4.5)
Differentiating equations (4.5) yields(
2ω1
1 − ω3
3
)
∧ ω1
0 +
(
ω1
2 − ω2
1 − ω3
4
)
∧ ω2
0 = 0,(
ω1
2 − ω2
1 − ω3
4
)
∧ ω1
0 +
(
ω3
3 − 2ω2
2
)
∧ ω2
0 = 0,(
2ω2
1 − ω4
3
)
∧ ω1
0 +
(
ω1
1 + ω2
2 − ω4
4
)
∧ ω2
0 = 0,(
ω1
1 + ω2
2 − ω4
4
)
∧ ω1
0 +
(
2ω1
2 + ω4
3
)
∧ ω2
0 = 0,(
2εω1
1 − ω0
3
)
∧ ω1
0 +
(
εω1
2 + εω2
1 − ω0
4
)
∧ ω2
0 = 0,(
εω1
2 + εω2
1 − ω0
4
)
∧ ω1
0 +
(
2εω2
2 + ω0
3
)
∧ ω2
0 = 0.
Applying Cartan’s lemma to these equations shows that there exists a 1-form α and functions hijk
on F2 such that
ω0
3 = h0
31ω
1
0 + h0
32ω
2
0, ω0
4 = h0
41ω
1
0 + h0
42ω
2
0,
ω1
1 = h1
11ω
1
0 + h1
12ω
2
0, ω1
2 = α+ h1
21ω
1
0 + h1
22ω
2
0,
ω2
1 = −α+ h1
21ω
1
0 + h1
22ω
2
0, ω2
2 = h2
21ω
1
0 + h2
22ω
2
0,
ω3
3 = h3
31ω
1
0 + h3
32ω
2
0, ω3
4 = 2α+ h3
41ω
1
0 + h3
42ω
2
0,
ω4
3 = −2α+ h4
31ω
1
0 + h4
32ω
2
0, ω4
4 = h4
41ω
1
0 + h4
42ω
2
0. (4.6)
Moreover, the functions hijk satisfy the relations
2h1
12 − h3
32 + h3
41 = 0, 2h2
21 − h3
31 − h3
42 = 0,
h1
11 − 2h1
22 + h2
21 + h4
32 − h4
41 = 0, h1
12 − 2h1
21 + h2
22 − h4
31 − h4
42 = 0,
h0
32 − h0
41 + 2ε
(
h1
21 − h1
12
)
= 0, h0
31 + h0
42 + 2ε
(
h2
21 − h1
22
)
= 0. (4.7)
If f, f̃ : M → F2 are two 2-adapted frame fields related by a transformation of the form (4.4),
then we can once again use equation (2.5) to determine how the corresponding functions hijk, h̃
i
jk
are related. Some of these relationships are more complicated than others; the most straight-
forward to compute are those corresponding to the forms ω̃0
3, ω̃0
4. These forms appear as the
coefficients of ẽ0 = e0 in the equations (2.2) for dẽ3, dẽ4. By applying equations (4.4) and (4.6),
one can show that[
h̃0
31
h̃0
41
]
=
[
cos(2θ) sin(2θ)
− sin(2θ) cos(2θ)
][
h0
31
h0
41
]
+ ε
[
r13
r14
]
,[
h̃0
32
h̃0
42
]
=
[
cos(2θ) sin(2θ)
− sin(2θ) cos(2θ)
][
h0
32
h0
42
]
+ ε
[
r23
r24
]
.
Thus we can find a 2-adapted frame field along Σ satisfying the conditions that
h0
31 = h0
32 = h0
41 = h0
42 = 0,
and these conditions are preserved by transformations of the form (4.4) with r13 = r14 = r23 =
r24 = 0.
8 N. Bushek and J.N. Clelland
Definition 4.2. Let Σ = f̄(M) be a nondegenerate, space-like centroaffine surface in R5 \ {0}.
A 2-adapted frame field f : M → F2 will be called 3-adapted if it satisfies the conditions
h0
31 = h0
32 = h0
41 = h0
42 = 0 (4.8)
at every point of M .
Any two 3-adapted frames (e0, . . . , e4), (ẽ0, . . . , ẽ4) based at the same point x ∈M are related
by a transformation of the form[
ẽ0 ẽ1 ẽ2 ẽ3 ẽ4
]
=
[
e0 e1 e2 e3 e4
]
1 0 0 0 0
0 cos(θ) − sin(θ) 0 0
0 sin(θ) cos(θ) 0 0
0 0 0 cos(2θ) − sin(2θ)
0 0 0 sin(2θ) cos(2θ)
. (4.9)
We will denote the group of all matrices of the form in (4.9) by G3; note that G3
∼= SO(2,R).
Then the 3-adapted frame fields are the smooth sections of a principal bundle F3 ⊂ F2 over M
with fiber group G3.
The equations (4.8) are equivalent to the condition that the Maurer–Cartan forms associated
to a 3-adapted frame field satisfy the conditions
ω0
3 = ω0
4 = 0. (4.10)
Differentiating equations (4.10) yields
ω1
3 ∧ ω1
0 + ω2
3 ∧ ω2
0 = ω1
4 ∧ ω1
0 + ω2
4 ∧ ω2
0 = 0,
and applying Cartan’s lemma shows that there exist functions hijk on F3 such that
ω1
3 = h1
31ω
1
0 + h1
32ω
2
0, ω2
3 = h1
32ω
1
0 + h2
32ω
2
0,
ω1
4 = h1
41ω
1
0 + h1
42ω
2
0, ω2
4 = h1
42ω
1
0 + h2
42ω
2
0.
Moreover, on F3, the last two relations in equations (4.7) simplify to
h1
21 − h1
12 = h2
21 − h1
22 = 0.
At this point, we have canonically associated to any nondegenerate, space-like centroaffine
surface in R5 \ {0} a frame bundle F3 over M with fiber group isomorphic to SO(2,R). Thus
we have the following theorem:
Theorem 4.3. Let f̄ : M → R5 \ {0} be a centroaffine immersion whose image Σ = f̄(M) is
a nondegenerate, space-like centroaffine surface. Then the pullbacks of the Maurer–Cartan forms
on GL(5,R) to the bundle F3 of 3-adapted frames on Σ determine a well-defined Riemannian
metric
I =
(
ω1
0
)2
+
(
ω2
0
)2
on Σ, called the centroaffine metric. Moreover, there is a well-defined “centroaffine normal
bundle” NΣ whose fiber NxΣ at each point x ∈ M is spanned by the vectors (e3(x), e4(x)) of
any 3-adapted frame at x, together with a well-defined Riemannian metric
Inormal =
(
ω3
0
)2
+
(
ω4
0
)2
on NΣ.
Geometry of Centroaffine Surfaces in R5 9
In order to obtain more information about the centroaffine metric, consider the structure
equations (2.3) for the semi-basic forms ω1
0, ω2
0 on M . Based on our adaptations, it is straight-
forward to compute that
dω1
0 = −α ∧ ω2
0, dω2
0 = α ∧ ω1
0.
Therefore, α is the Levi-Civita connection form associated to the centroaffine metric on Σ, and
the Gauss curvature K of this metric is determined by the equation
dα = Kω1
0 ∧ ω2
0.
The remaining structure equations (2.3) determine relations between the functions hijk on F3
and their covariant derivatives with respect to ω1
0, ω2
0. These relations may be viewed as analogs
of the Gauss and Codazzi equations for Riemannian surfaces in Euclidean space. In particular,
the analog of the Gauss equation is
K = 1
2
(
−h3
32h
4
31 − h4
41h
3
42 + h3
41h
4
42 − h3
32h
4
42 + h4
32h
3
31
+ h4
32h
3
42 − h4
41h
3
31 + h3
41h
4
31 + h1
31 − h2
32 + 2h1
42
)
− ε, (4.11)
while the remainder of the relations are partial differential equations involving the functions hijk.
An analog of Bonnet’s theorem (see [6]) guarantees that, at least locally, any solution of this
PDE system gives rise to a nondegenerate, space-like centroaffine surface, and that this surface
is unique up to the action of GL(5,R) on R5 \ {0}. In particular, the functions hijk on F3 form
a complete set of local invariants for such surfaces.
5 The time-like case
Now, suppose that the plane P spanned by (h3, h4) is time-like at every point of M . According
to the group action (3.5), we can find a 1-adapted frame field along Σ for which P = P2, as in
equation (3.9). Furthermore, we can then use the action by B to find a 1-adapted frame field
along Σ for which
h3 =
[
1 0
0 0
]
, h4 =
[
0 0
0 1
]
. (5.1)
(This choice of h3, h4 represents a null basis for P2 with respect to the indefinite quadratic
form (3.7) on P2.)
The next step is to determine the subgroup of G1 that preserves the conditions (5.1). To this
end, first note that the plane P2 spanned by the matrices (5.1) consists precisely of the diagonal
matrices in Sym2(R). A straightforward computation shows that, with h3, h4 as in (5.1), the off-
diagonal components of tr(ATh3A) and tr(ATh4A) are equal to a11a12 and a21a22, respectively.
Therefore, the action (3.5) preserves P2 if and only if
a11a12 = a21a22 = 0.
Substituting this condition into equations (3.4) and imposing the conditions h̃3 = h3, h̃4 = h4
then shows that we must have either
A =
[
a11 0
0 a22
]
, B =
[
a2
11 0
0 a2
22
]
, a11, a22 6= 0, (5.2)
or
A =
[
0 a12
a21 0
]
, B =
[
0 a2
12
a2
21 0
]
, a12, a21 6= 0.
10 N. Bushek and J.N. Clelland
Since the latter transformation may be obtained from the former simply by interchanging (e1, e2)
and (e3, e4), we will restrict our attention to transformations of the form (5.2), where A, B are
diagonal matrices and B = A2.
Next, consider the effect of the action (3.4) on h0. With A = I2 and r03, r04 chosen ap-
propriately, we can add any linear combination of h3, h4 to h0. Thus we can find a 1-adapted
frame field for which h0 is a multiple (nonzero by the nondegeneracy assumption) of the matrix[
0 1
1 0
]
. Then under the action (3.4) with A, B as in (5.2), we have
h̃0 = a11a22h
0 +
[
−r03 0
0 −r04
]
.
Thus we can find a 1-adapted frame field along Σ satisfying the additional condition that
h0 =
[
0 1
1 0
]
,
and this condition is preserved by transformations of the form (3.4) with A, B as in (5.2), such
that a11a22 = 1 and r03 = r04 = 0. For simplicity, we will assume that a11 > 0; then we can set
A =
[
eλ 0
0 e−λ
]
, B =
[
e2λ 0
0 e−2λ
]
, λ ∈ R.
Definition 5.1. Let Σ = f̄(M) be a nondegenerate, time-like centroaffine surface in R5 \ {0}.
A 1-adapted frame field f : M → F1 will be called 2-adapted if it satisfies the conditions
h3 =
[
1 0
0 0
]
, h4 =
[
0 0
0 1
]
, h0 =
[
0 1
1 0
]
(5.3)
at every point of M .
Any two 2-adapted frames (e0, . . . , e4), (ẽ0, . . . , ẽ4) based at the same point x ∈M are related
by a transformation of the form
[
ẽ0 ẽ1 ẽ2 ẽ3 ẽ4
]
=
[
e0 e1 e2 e3 e4
]
1 0 0 0 0
0 eλ 0 r13 r14
0 0 e−λ r23 r24
0 0 0 e2λ 0
0 0 0 0 e−2λ
. (5.4)
We will denote the group of all matrices of the form in (5.4) by G2; then the 2-adapted frame
fields along Σ are the smooth sections of a principal bundle F2 ⊂ F1 over M with fiber group G2.
The equations (5.3) are equivalent to the condition that the Maurer–Cartan forms associated
to a 2-adapted frame field satisfy the conditions
ω3
1 = ω1
0, ω3
2 = 0, ω4
1 = 0, ω4
2 = ω2
0, ω0
1 = ω2
0, ω0
2 = ω1
0. (5.5)
Differentiating equations (5.5) yields(
2ω1
1 − ω3
3
)
∧ ω1
0 + ω1
2 ∧ ω2
0 = 0, ω1
2 ∧ ω1
0 − ω3
4 ∧ ω2
0 = 0,
− ω4
3 ∧ ω1
0 + ω2
1 ∧ ω2
0 = 0, ω2
1 ∧ ω1
0 +
(
2ω2
2 − ω4
4
)
∧ ω2
0 = 0,(
2ω2
1 − ω0
3
)
∧ ω1
0 +
(
ω1
1 + ω2
2
)
∧ ω2
0 = 0,
(
ω1
1 + ω2
2
)
∧ ω1
0 +
(
2ω1
2 − ω0
4
)
∧ ω2
0 = 0.
Geometry of Centroaffine Surfaces in R5 11
Applying Cartan’s lemma to these equations shows that there exists a 1-form α and functions hijk
on F2 such that
ω0
3 = h0
31ω
1
0 + h0
32ω
2
0, ω0
4 = h0
41ω
1
0 + h0
42ω
2
0,
ω1
1 = α+ h1
11ω
1
0 + h2
22ω
2
0, ω1
2 = h1
21ω
1
0 + h1
22ω
2
0,
ω2
1 = h2
11ω
1
0 + h2
12ω
2
0, ω2
2 = −α+ h1
11ω
1
0 + h2
22ω
2
0,
ω3
3 = 2α+ h3
31ω
1
0 + h3
32ω
2
0, ω3
4 = h3
41ω
1
0 + h3
42ω
2
0,
ω4
3 = h4
31ω
1
0 + h4
32ω
2
0, ω4
4 = −2α+ h4
41ω
1
0 + h4
42ω
2
0. (5.6)
Moreover, the functions hijk satisfy the relations
2h2
22 − h1
21 − h3
32 = 0, h1
22 + h3
41 = 0,
h2
11 + h4
32 = 0, 2h1
11 − h2
12 − h4
41 = 0,
2h1
11 − 2h2
12 + h0
32 = 0, 2h2
22 − 2h1
21 + h0
41 = 0. (5.7)
If f, f̃ : M → F2 are two 2-adapted frame fields related by a transformation of the form (5.4),
then we can once again use equation (2.5) to determine how the corresponding functions hijk, h̃
i
jk
are related. As in the space-like case, the most straightforward to compute are those correspon-
ding to the forms ω̃0
3, ω̃0
4. These forms appear as the coefficients of ẽ0 = e0 in the equations (2.2)
for dẽ3, dẽ4. By applying equations (5.4) and (5.6), one can show that[
h̃0
31
h̃0
41
]
=
[
e2λ 0
0 e−2λ
][
h0
31
h0
41
]
+
[
r23
r24
]
,
[
h̃0
32
h̃0
42
]
=
[
e2λ 0
0 e−2λ
][
h0
32
h0
42
]
+
[
r13
r14
]
.
Thus we can find a 2-adapted frame field along Σ satisfying the conditions that
h0
31 = h0
32 = h0
41 = h0
42 = 0,
and these conditions are preserved by transformations of the form (5.4) with r13 = r14 = r23 =
r24 = 0.
Definition 5.2. Let Σ = f̄(M) be a nondegenerate, time-like centroaffine surface in R5 \ {0}.
A 2-adapted frame field f : M → F2 will be called 3-adapted if it satisfies the conditions
h0
31 = h0
32 = h0
41 = h0
42 = 0 (5.8)
at every point of M .
Any two 3-adapted frames (e0, . . . , e4), (ẽ0, . . . , ẽ4) based at the same point x ∈M are related
by a transformation of the form
[
ẽ0 ẽ1 ẽ2 ẽ3 ẽ4
]
=
[
e0 e1 e2 e3 e4
]
1 0 0 0 0
0 eλ 0 0 0
0 0 e−λ 0 0
0 0 0 e2λ 0
0 0 0 0 e−2λ
. (5.9)
We will denote the group of all matrices of the form in (5.9) by G3; note that G3
∼= SO+(1, 1).
Then the 3-adapted frame fields are the smooth sections of a principal bundle F3 ⊂ F2 over M
with fiber group G3.
12 N. Bushek and J.N. Clelland
The equations (5.8) are equivalent to the condition that the Maurer–Cartan forms associated
to a 3-adapted frame field satisfy the conditions
ω0
3 = ω0
4 = 0. (5.10)
Differentiating equations (5.10) yields
ω2
3 ∧ ω1
0 + ω1
3 ∧ ω2
0 = ω2
4 ∧ ω1
0 + ω1
4 ∧ ω2
0 = 0,
and applying Cartan’s lemma shows that there exist functions hijk on F3 such that
ω1
3 = h1
31ω
1
0 + h1
32ω
2
0, ω2
3 = h2
31ω
1
0 + h1
31ω
2
0,
ω1
4 = h1
41ω
1
0 + h1
42ω
2
0, ω2
4 = h2
41ω
1
0 + h1
41ω
2
0.
Moreover, on F3, the last two relations in equations (5.7) simplify to
h1
11 − h2
12 = h2
22 − h1
21 = 0.
At this point, we have canonically associated to any nondegenerate, time-like centroaffine
surface in R5 \ {0} a frame bundle F3 over M with fiber group isomorphic to SO+(1, 1). Thus
we have the following theorem:
Theorem 5.3. Let f̄ : M → R5 \ {0} be a centroaffine immersion whose image Σ = f̄(M) is
a nondegenerate, time-like centroaffine surface. Then the pullbacks of the Maurer–Cartan forms
on GL(5,R) to the bundle F3 of 3-adapted frames on Σ determine a well-defined Lorentzian
metric
I = 2ω1
0ω
2
0
on Σ, called the centroaffine metric. Moreoever, there is a well-defined “centroaffine normal
bundle” NΣ whose fiber NxΣ at each point x ∈ M is spanned by the vectors (e3(x), e4(x)) of
any 3-adapted frame at x, together with a well-defined Lorentzian metric
Inormal = 2ω3
0ω
4
0
on NΣ.
In order to obtain more information about the centroaffine metric, consider the structure
equations (2.3) for the semi-basic forms ω1
0, ω2
0 on M . Based on our adaptations, it is straight-
forward to compute that
dω1
0 = −α ∧ ω1
0, dω2
0 = α ∧ ω2
0.
Therefore, α is the Levi-Civita connection form associated to the centroaffine metric on Σ, and
the Gauss curvature K of this metric is determined by the equation
dα = Kω1
0 ∧ ω2
0.
As in the space-like case, the remaining structure equations (2.3) determine relations between
the functions hijk on F3 and their covariant derivatives with respect to ω1
0, ω2
0. The analog of
the Gauss equation is
K = h3
41h
4
32 − h3
32h
4
41 + 1
2
(
h1
32 + h2
41
)
− 1, (5.11)
while the remainder of the relations are partial differential equations involving the functions hijk.
As in the space-like case, any solution of this PDE system locally gives rise to a nondegenerate,
time-like centroaffine surface; this surface is unique up to the action of GL(5,R) on R5 \ {0},
and the functions hijk on F3 form a complete set of local invariants for such surfaces.
Geometry of Centroaffine Surfaces in R5 13
6 Homogeneous examples
The goal of this section is to give a complete classification (up to the GL(5,R)-action on R5\{0})
of the homogeneous examples of space-like and time-like nondegenerate centroaffine surfaces
in R5 \ {0}. First we must define precisely what we mean by the term “homogeneous”. Because
any such surface Σ = f̄(M) has a well-defined Riemannian or Lorentzian metric, any symmetry
of Σ must preserve the centroaffine metric on Σ and hence must in fact be an isometry of Σ
with its centroaffine metric. Furthermore, if the group of symmetries of Σ acts transitively, then
the centroaffine metric must have constant Gauss curvature K. As is well-known, the maximal
isometry group of any Riemannian or Lorentzian surface has dimension less than or equal to
three, and in the maximal case the isometry group acts transitively on the orthonormal frame
bundle. Thus we will use the following definition:
Definition 6.1. Let F3 be the bundle of 3-adapted frames along a nondegenerate, space-like or
time-like centroaffine surface Σ = f̄(M) in R5 \ {0}. A diffeomorphism φ : F3 → F3 is called
a symmetry of Σ if φ∗Ω = Ω; i.e., if φ preserves the Maurer–Cartan forms on F3. A nondegenerate
space-like or time-like centroaffine surface Σ = f̄(M) will be called homogeneous if the group of
symmetries of Σ is a 3-dimensional Lie group that acts transitively on F3.
Remark 6.2. It might also be of interest to consider the slightly less restrictive assumption
that Σ has a 2-dimensional group of symmetries that acts transitively on the base manifold M ,
but we will not consider this scenario here.
Our procedure for classifying the homogeneous examples is as follows. Observe that if Σ
is homogeneous, then all the structure functions hijk must be constant on the bundle F3 of
3-adapted frames on Σ. When this condition is imposed, the structure equations (2.3) become
algebraic relations among the constants hijk. Given any solution to these relations, the struc-
ture equations (2.3) imply that the corresponding Maurer–Cartan form Ω = [ωij ] takes values in
a 3-dimensional Lie algebra g which is realized explicitly as a Lie subalgebra of gl(5,R). Thus Ω
is also the Maurer–Cartan form of the connected Lie group G ⊂ GL(5,R) generated by expo-
nentiating g, and this equivalence of the Maurer–Cartan forms on F3 with those on G implies
that F3 is a homogeneous space for G; indeed, G must be precisely the symmetry group that
was assumed to act transitively on F3.
Now, choose any point f0 = (e0, e1, e2, e3, e4) ∈ F3. Recall that we can view f0 as an element
of GL(5,R). The centroaffine surface Σ is equivalent via the GL(5,R)-action to the surface
Σ̃ = f−1
0 ·Σ, and the bundle F̃3 of 3-adapted frames over Σ̃ is given by F̃3 = f−1
0 ·F3. So without
loss of generality, we may assume that f0 is the identity matrix I5. With this assumption, the
tangent space Tf0F3 is equal to the Lie algebra g, and F3 must in fact be equal to G. Finally, Σ
is given by the image of G under the projection (2.1).
In order to carry out this procedure, we consider the space-like and time-like cases separately.
6.1 Space-like homogeneous examples
From the adaptations of Section 4, the matrix Ω = [ωij ] of Maurer–Cartan forms on F3 may be
written as
Ω =
0 0 0 0 0
0 0 1 0 0
0 −1 0 0 0
0 0 0 0 2
0 0 0 −2 0
α+
0 ε 0 0 0
1 1
2(h3
31 + h3
42)− h4
32 + h4
41
1
2(h3
32 − h3
41) h1
31 h1
41
0 1
2(h3
32 − h3
41) 1
2(h3
31 + h3
42) h1
32 h1
42
0 1 0 h3
31 h3
41
0 0 1 h4
31 h4
41
ω1
0
14 N. Bushek and J.N. Clelland
+
0 0 ε 0 0
0 1
2(h3
32 − h3
41) 1
2(h3
31 + h3
42) h1
32 h1
42
1 1
2(h3
31 + h3
42) 1
2(h3
32 − h3
41) + h4
31 + h4
42 h2
32 h2
42
0 0 −1 h3
32 h3
42
0 1 0 h4
32 h4
42
ω2
0, (6.1)
while the structure equations (2.3) may be written as
dΩ = −Ω ∧ Ω. (6.2)
Substituting (6.1) into equation (6.2) and imposing the condition that all the functions hijk are
constant leads to a system of 25 algebraic equations (some linear, some quadratic) for the 14
unknown constants hijk. A somewhat tedious, but straightforward, computation shows that this
system has precisely two solutions, one for ε = 1 and one for ε = −1. These are described in the
following two examples.
Example 6.3. When ε = 1, the unique solution to equation (6.2) with all hijk constant is
Ω =
0 0 0 0 0
0 0 1 0 0
0 −1 0 0 0
0 0 0 0 2
0 0 0 −2 0
α+
0 1 0 0 0
1 0 0 1
3 0
0 0 0 0 1
3
0 1 0 0 0
0 0 1 0 0
ω1
0 +
0 0 1 0 0
0 0 0 0 1
3
1 0 0 −1
3 0
0 0 −1 0 0
0 1 0 0 0
ω2
0. (6.3)
Furthermore, the Gauss equation (4.11) implies that the centroaffine metric has Gauss curvature
K = −1
3 .
Denote the matrices in equation (6.3) by M0, M1, M2, respectively, so that
Ω = M0α+M1ω
1
0 +M2ω
2
0.
It is straightforward to compute that
[M0,M1] = −M2, [M1,M2] = 1
3M0, [M2,M0] = −M1.
These bracket relations imply that the Lie algebra g ⊂ gl(5,R) spanned by (M0,M1,M2) is
isomorphic to so(1, 2). (They also suggest that a more natural basis might be obtained by
multiplying each of M1, M2 by
√
3.) Furthermore, it is straightforward to check that g acts
irreducibly on R5 \ {0}. It is well-known (see, e.g., [2, p. 149]) that so(1, 2) has a unique
irreducible 5-dimensional representation and that this representation arises from a (unique)
irreducible representation of SO+(1, 2); it follows that the Lie group G ⊂ GL(5,R) corresponding
to the Lie algebra g is isomorphic to SO+(1, 2).
The easiest way to compute a local parametrization for G – and hence for Σ – is to compute
the 1-parameter subgroups generated by M0, M1, M2 and take products of the resulting group
elements.
Warning. This must be done carefully in order to ensure that the resulting products cover
the entire group G, which in turn guarantees that the resulting parametrization is surjective
onto Σ. The subtlety of this issue can already be seen in the standard representation for so(1, 2):
the basis
M̄0 =
0 0 0
0 0 1
0 −1 0
, M̄1 =
0 1 0
1 0 0
0 0 0
, M̄2 =
0 0 1
0 0 0
1 0 0
Geometry of Centroaffine Surfaces in R5 15
has the same bracket relations as (M0,
√
3M1,
√
3M2), and exponentiating this basis yields the
1-parameter subgroups
ḡ0(t) =
1 0 0
0 cos(t) sin(t)
0 − sin(t) cos(t)
,
ḡ1(u) =
cosh(u) sinh(u) 0
sinh(u) cosh(u) 0
0 0 1
, ḡ2(v) =
cosh(v) 0 sinh(v)
0 1 0
sinh(v) 0 cosh(v)
.
Now consider the following two maps f1, f2 : R3 → SO+(1, 2), which are obtained by multiplying
the elements ḡ0(t), ḡ1(u), ḡ2(v) in different orders:
f1(u, v, t) = ḡ1(u)ḡ2(v)ḡ0(t)
=
cosh(u) cosh(v) sinh(u) cos(t)
− cosh(u) sinh(v) sin(t)
sinh(u) sin(t)
+ cosh(u) sinh(v) cos(t)
sinh(u) cosh(v) cosh(u) cos(t)
− sinh(u) sinh(v) sin(t)
cosh(u) sin(t)
+ sinh(u) sinh(v) cos(t)
sinh(v) − cosh(v) sin(t) cosh(v) cos(t)
,
f2(u, v, t) = ḡ1(u)ḡ0(t)ḡ2(v)
=
cosh(u) cosh(v)
+ sinh(u) sinh(v) sin(t) sinh(u) cos(t) cosh(u) sinh(v)
+ sinh(u) cosh(v) sin(t)
sinh(u) cosh(v)
+ cosh(u) sinh(v) sin(t) cosh(u) cos(t) sinh(u) sinh(v)
+ cosh(u) cosh(v) sin(t)
sinh(v) cos(t) − sin(t) cosh(v) cos(t)
.
It is not difficult to show that f1 is surjective onto SO+(1, 2), whereas we can see from the
middle column that f2 is not. Thus we must perform this construction with care.
Now, the obvious correspondence
M̄0 ↔M0, M̄1 ↔
√
3M1, M̄2 ↔
√
3M2,
defines a Lie algebra isomorphism between the standard representation of so(1, 2) and our Lie
algebra g ⊂ gl(5,R). Therefore, the surjectivity of the map f1 above implies that the analogous
map f : R3 → G will also be surjective onto G, and it follows that the map f̄ = π ◦ f will be
surjective onto Σ. The map f̄ will also turn out to be independent of t, and when regarded as
a function of the two variables (u, v) it will define a surjective parametrization f̄ : R2 → Σ.
With these considerations in mind, define the 1-parameter subgroups
g0(t) = exp(tM0) =
1 0 0 0 0
0 cos(t) sin(t) 0 0
0 − sin(t) cos(t) 0 0
0 0 0 cos(2t) sin(2t)
0 0 0 − sin(2t) cos(2t)
,
g1(u) = exp(u
√
3M1)
=
1
4(3 cosh(2u) + 1)
√
3
2 sinh(2u) 0 1
4(cosh(2u)− 1) 0√
3
2 sinh(2u) cosh(2u) 0 1
2
√
3
sinh(2u) 0
0 0 cosh(u) 0 1√
3
sinh(u)
3
4(cosh(2u)− 1)
√
3
2 sinh(2u) 0 1
4(cosh(2u) + 3) 0
0 0
√
3 sinh(u) 0 cosh(u)
,
16 N. Bushek and J.N. Clelland
g2(v) = exp(v
√
3M2)
=
1
4(3 cosh(2v) + 1) 0
√
3
2 sinh(2v) 1
4(1− cosh(2v)) 0
0 cosh(v) 0 0 1√
3
sinh(v)
√
3
2 sinh(2v) 0 cosh(2v) − 1
2
√
3
sinh(2v) 0
3
4(1− cosh(2v)) 0 −
√
3
2 sinh(2v) 1
4(cosh(2v) + 3) 0
0
√
3 sinh(v) 0 0 cosh(v)
.
Then set
f̄(u, v, t) = π (g1(u) · g2(v) · g0(t)) =
1
2
(
3 cosh2(u) cosh2(v)− 1
)
√
3 sinh(u) cosh(u) cosh2(v)
√
3 cosh(u) sinh(v) cosh(v)
3
2
(
cosh2(v)(cosh2(u)− 2) + 1
)
3 sinh(u) sinh(v) cosh(v)
. (6.4)
It follows from the discussion above that f̄ is a surjective map onto Σ. Moreover, it is straight-
forward to check that the tangent vectors f̄u, f̄v are linearly independent for all (u, v) ∈ R2;
therefore f̄ parametrizes a smooth surface Σ ⊂ R5 \ {0}, as expected.
We can compute the centroaffine metric on Σ explicitly as follows. Let f : R3 → G be the
map corresponding to (6.4); i.e.,
f(u, v, t) = g1(u) · g2(v) · g0(t).
Then we have
Ω = M0α+M1ω
1
0 +M2ω
2
0 = f−1df.
Comparing these two expressions for Ω shows that
ω1
0 =
√
3 (cosh(v) cos(t)du− sin(t)dv) , ω2
0 =
√
3 (cosh(v) sin(t)du+ cos(t)dv) .
Therefore, the centroaffine metric on Σ is given by
I =
(
ω1
0
)2
+
(
ω2
0
)2
= 3
(
cosh2(v)du2 + dv2
)
.
As expected, this metric has constant Gauss curvature K = −1
3 . Topologically, Σ is simply
connected and isometric to the hyperbolic plane H of “radius”
√
3, while F3 is isomorphic to
the orthonormal frame bundle of H, with π as the projection map.
In fact, we can describe Σ more intrinsically: if we denote the coordinates of a point x ∈
R5 \ {0} as (x0, . . . , x4), then the coordinates of f̄(u, v) satisfy the quadratic equations
x1(x0 − x3 − 1)− x2x4 = 0, x2(x0 + x3 − 1)− x1x4 = 0,
(4x0 − 1)2 − 12x2
1 − 12x2
2 − 9 = 0. (6.5)
Therefore, Σ is contained in the (real) algebraic variety X ⊂ R5 \{0} defined by equations (6.5).
Σ is not, however, equal to all of X; for instance, X contains an affine plane consisting of all
points of the form (1, 0, 0, x3, x4), and this plane intersects Σ only when x3 = x4 = 0. Projections
of Σ to the (x1, x2, x0), (x1, x2, x3), and (x1, x2, x4) coordinate 3-planes are shown in Fig. 1.
Geometry of Centroaffine Surfaces in R5 17
Figure 1. Projections of surface from Example 6.3 to 3-D subspaces.
Example 6.4. When ε = −1, the unique solution to equation (6.2) with all hijk constant is
Ω =
0 0 0 0 0
0 0 1 0 0
0 −1 0 0 0
0 0 0 0 2
0 0 0 −2 0
α+
0 −1 0 0 0
1 0 0 −1
3 0
0 0 0 0 −1
3
0 1 0 0 0
0 0 1 0 0
ω1
0 +
0 0 −1 0 0
0 0 0 0 −1
3
1 0 0 1
3 0
0 0 −1 0 0
0 1 0 0 0
ω2
0. (6.6)
Furthermore, the Gauss equation (4.11) implies that the centroaffine metric has Gauss curvature
K = 1
3 .
As in the previous example, denote the matrices in equation (6.6) by M0, M1, M2, respec-
tively. Then we have
[M0,M1] = −M2, [M1,M2] = −1
3M0, [M2,M0] = −M1.
These bracket relations imply that the Lie algebra g ⊂ gl(5,R) spanned by (M0,M1,M2) is iso-
morphic to so(3,R). Furthermore, it is straightforward to check that g acts irreducibly on R5 \
{0}. Similarly to the previous example, so(3,R) has a unique irreducible 5-dimensional repre-
sentation, and this representation arises from a (unique) irreducible representation of SO(3,R);
it follows that the Lie group G ⊂ GL(5,R) corresponding to the Lie algebra g is isomorphic
to SO(3,R).
We will compute a local parametrization for Σ as in the previous example: compute the
1-parameter subgroups of G generated by M0, M1, M2 and take products of the resulting group
elements. Ensuring the surjectivity of the resulting parametrization is easier than in the previous
example. First, observe that a basis (M̄0, M̄1, M̄2) for the standard representation of so(3,R)
with the same bracket relations as (M0,
√
3M1,
√
3M2) is given by
M̄0 =
0 0 0
0 0 1
0 −1 0
, M̄1 =
0 −1 0
1 0 0
0 0 0
, M̄2 =
0 0 −1
0 0 0
1 0 0
.
Exponentiating this basis yields the 1-parameter subgroups
ḡ0(t) =
1 0 0
0 cos(t) sin(t)
0 − sin(t) cos(t)
,
ḡ1(u) =
cos(u) − sin(u) 0
sin(u) cos(u) 0
0 0 1
, ḡ2(v) =
cos(v) 0 − sin(v)
0 1 0
sin(v) 0 cos(v)
.
18 N. Bushek and J.N. Clelland
Then the map f : R3 → SO(3,R) defined by
f(u, v, t) = ḡ1(u)ḡ2(v)ḡ0(t)
=
cos(u) cos(v) − sin(u) cos(t)
+ cos(u) sin(v) sin(t)
− sin(u) sin(t)
− cos(u) sin(v) cos(t)
sin(u) cos(v) cos(u) cos(t)
+ sin(u) sin(v) sin(t)
cos(u) sin(t)
− sin(u) sin(v) cos(t)
sin(v) − cos(v) sin(t) cos(v) cos(t)
is easily seen to be surjective onto SO(3,R). Thus the analogous map f : R3 → G will be
surjective onto G, and the map f̄ = π ◦ f will be surjective onto Σ.
So, define the 1-parameter subgroups
g0(t) = exp(tM0) =
1 0 0 0 0
0 cos(t) sin(t) 0 0
0 − sin(t) cos(t) 0 0
0 0 0 cos(2t) sin(2t)
0 0 0 − sin(2t) cos(2t)
,
g1(u) = exp(u
√
3M1)
=
1
4(3 cos(2u) + 1) −
√
3
2 sin(2u) 0 1
4(1− cos(2u)) 0√
3
2 sin(2u) cos(2u) 0 − 1
2
√
3
sin(2u) 0
0 0 cos(u) 0 − 1√
3
sin(u)
3
4(1− cos(2u))
√
3
2 sin(2u) 0 1
4(cos(2u) + 3) 0
0 0
√
3 sin(u) 0 cos(u)
,
g2(v) = exp(v
√
3M2)
=
1
4(3 cos(2v) + 1) 0 −
√
3
2 sin(2v) 1
4(cos(2v)− 1) 0
0 cos(v) 0 0 − 1√
3
sin(v)
√
3
2 sin(2v) 0 cos(2v) 1
2
√
3
sin(2v) 0
3
4(cos(2v)− 1) 0 −
√
3
2 sin(2v) 1
4(cos(2v) + 3) 0
0
√
3 sin(v) 0 0 cos(v)
.
Then set
f̄(u, v, t) = π (g1(u) · g2(v) · g0(t)) =
1
2
(
3 cos2(u) cos2(v)− 1
)
√
3 sin(u) cos(u) cos2(v)
√
3 cos(u) sin(v) cos(v)
3
2
(
cos2(v)(2− cos2(u))− 1
)
3 sin(u) sin(v) cos(v)
. (6.7)
It follows from the discussion above that f̄ is a surjective map onto Σ, and we see that f̄ is also
independent of t. Unlike in the previous example, the tangent vectors f̄u, f̄v are not linearly
independent for all (u, v) ∈ R2; indeed, f̄u = 0 whenever v is an odd multiple of π2 . Nevertheless,
the restriction of f̄ to some neighborhood of the point (u, v) = (0, 0) is a smooth embedding,
and then homogeneity implies that Σ is smooth everywhere.
We can compute the centroaffine metric on Σ as in the previous example. Let f : R3 → G
be the map corresponding to (6.7); i.e.,
f(u, v, t) = g1(u) · g2(v) · g0(t).
Geometry of Centroaffine Surfaces in R5 19
Then we have
Ω = M0α+M1ω
1
0 +M2ω
2
0 = f−1df.
Comparing these two expressions for Ω shows that, whenever cos(v) 6= 0,
ω1
0 =
√
3
(
cos(v) cos(t)du− sin(t)dv
)
, ω2
0 =
√
3
(
cos(v) sin(t)du+ cos(t)dv
)
.
Therefore, the centroaffine metric on Σ is given by
I =
(
ω1
0
)2
+
(
ω2
0
)2
= 3
(
cos2(v)du2 + dv2
)
.
As expected, this metric has constant Gauss curvature K = 1
3 . Topologically, Σ is simply
connected and isometric to the sphere S of radius
√
3, while F3 is isomorphic to the orthonormal
frame bundle of S, with π as the projection map.
As in the previous example, we can show that Σ is contained in the intersection of three
quadric hypersurfaces in R5 \ {0}. The coordinates of f̄(u, v) satisfy the quadratic equations
x1(x0 + x3 − 1) + x2x4 = 0, x2(x0 − x3 − 1) + x1x4 = 0,
(4x0 − 1)2 + 12x2
1 + 12x2
2 − 9 = 0. (6.8)
Therefore, Σ is contained in the (real) algebraic variety X ⊂ R5 \{0} defined by equations (6.8).
Once again, Σ is not equal to all of X: the variety X contains affine planes consisting of all points
of the form (1, 0, 0, x3, x4) or (−1
2 , 0, 0, x3, x4); the former intersects Σ only when x3 = x4 = 0,
and the latter intersects Σ only when x2
3+x2
4 = 9
4 . Projections of Σ to the (x1, x2, x0), (x1, x2, x3),
and (x1, x2, x4) coordinate 3-planes are shown in Fig. 2.
Figure 2. Projections of surface from Example 6.4 to 3-D subspaces.
6.2 Time-like homogeneous examples
From the adaptations of Section 5, the matrix Ω = [ωij ] of Maurer–Cartan forms on F3 may be
written as
Ω =
0 0 0 0 0
0 1 0 0 0
0 0 −1 0 0
0 0 0 2 0
0 0 0 0 −2
α+
0 0 1 0 0
1 h4
41 h3
32 h1
31 h1
41
0 −h4
32 h4
41 h2
31 h2
41
0 1 0 h3
31 h3
41
0 0 0 h4
31 h4
41
ω1
0
20 N. Bushek and J.N. Clelland
+
0 1 0 0 0
0 h3
32 −h3
41 h1
32 h1
42
1 h4
41 h3
32 h1
31 h1
41
0 0 0 h3
32 h3
42
0 0 1 h4
32 h4
42
ω2
0, (6.9)
while the structure equations (2.3) may be written as
dΩ = −Ω ∧ Ω. (6.10)
Substituting (6.9) into equation (6.10) and imposing the condition that all the functions hijk are
constant leads to a system of 23 algebraic equations for the 14 unknown constants hijk. This
system has precisely one solution, which is described in the following example.
Example 6.5. The unique solution to equation (6.10) with all hijk constant is
Ω =
0 0 0 0 0
0 1 0 0 0
0 0 −1 0 0
0 0 0 2 0
0 0 0 0 −2
α+
0 0 1 0 0
1 0 0 0 0
0 0 0 0 2
3
0 1 0 0 0
0 0 0 0 0
ω1
0 +
0 1 0 0 0
0 0 0 2
3 0
1 0 0 0 0
0 0 0 0 0
0 0 1 0 0
ω2
0. (6.11)
Furthermore, the Gauss equation (5.11) implies that the centroaffine metric has Gauss curvature
K = −1
3 .
As in the space-like examples, denote the matrices in equation (6.11) by M0, M1, M2, re-
spectively, so that
Ω = M0α+M1ω
1
0 +M2ω
2
0.
Then we have
[M0,M1] = M1, [M1,M2] = 1
3M0, [M2,M0] = M2.
These bracket relations imply that the Lie algebra g ⊂ gl(5,R) spanned by (M0,M1,M2) is iso-
morphic to so(2, 1). Furthermore, it is straightforward to check that g acts irreducibly on R5\{0}.
Similarly to the previous examples, so(2, 1) has a unique irreducible 5-dimensional representa-
tion, and this representation arises from a (unique) irreducible representation of SO+(2, 1); it
follows that the Lie group G ⊂ GL(5,R) corresponding to the Lie algebra g is isomorphic
to SO+(2, 1).
We will compute a local parametrization for Σ more or less as in the space-like examples,
by computing 1-parameter subgroups of G and taking products of the resulting group elements.
Unfortunately, the basis (M0,M1,M2) is not well-suited to generating a surjective parametriza-
tion, so first we need to modify it slightly. To this end, observe that a basis (M̄0, M̄1, M̄2) for
the standard representation of so(2, 1) with the same bracket relations as (M0,
√
3M1,
√
3M2) is
given by
M̄0 =
0 0 0
0 1 0
0 0 −1
, M̄1 =
0 0 1
1 0 0
0 0 0
, M̄2 =
0 1 0
0 0 0
1 0 0
.
Now consider the modified basis
M̄ ′0 = M̄0, M̄ ′1 =
1√
2
(M̄1 − M̄2), M̄ ′2 =
1√
2
(M̄1 + M̄2).
Geometry of Centroaffine Surfaces in R5 21
Exponentiating this modified basis yields the 1-parameter subgroups
ḡ0(t) =
1 0 0
0 et 0
0 0 e−t
, ḡ1(u) =
cos(u) − 1√
2
sin(u) 1√
2
sin(u)
1√
2
sin(u) 1
2(1 + cos(u)) 1
2(1− cos(u))
− 1√
2
sin(u) 1
2(1− cos(u)) 1
2(1 + cos(u))
,
ḡ2(v) =
cosh(v) 1√
2
sinh(v) 1√
2
sinh(v)
1√
2
sinh(v) 1
2(cosh(v) + 1) 1
2(cosh(v)− 1)
1√
2
sinh(v) 1
2(cosh(v)− 1) 1
2(cosh(v) + 1)
.
Then the map f : R3 → SO+(2, 1) defined by
f(u, v, t) = ḡ1(u)ḡ2(v)ḡ0(t)
=
cos(u) cosh(v)
1√
2
et(cos(u) sinh(v)
− sin(u))
1√
2
e−t(cos(u) sinh(v)
+ sin(u))
1√
2
(sin(u) cosh(v)
+ sinh(v))
1
2e
t(sin(u) sinh(v)
+ cosh(v) + cos(u))
1
2e
−t(sin(u) sinh(v)
+ cosh(v)− cos(u))
− 1√
2
(sin(u) cosh(v)
− sinh(v))
−1
2e
t(sin(u) sinh(v)
− cosh(v) + cos(u))
−1
2e
−t(sin(u) sinh(v)
− cosh(v)− cos(u))
is surjective onto SO+(2, 1). Thus the analogous map f : R3 → G will be surjective onto G, and
the map f̄ = π ◦ f will be surjective onto Σ.
So, define the 1-parameter subgroups
g0(t) = exp(tM0), g1(u) = exp
(
u
√
3
2(M1 −M2)
)
, g2(v) = exp
(
v
√
3
2(M1 +M2)
)
.
(The explicit expressions for these group elements are each too large to fit on one line and are
not particularly enlightening.) Then set
f̄(u, v, t) = π (g1(u) · g2(v) · g0(t))
=
1
4 [3 cos2(u)(cosh(2v) + 1)− 2]
√
6
4 cos(u)[sin(u)(cosh(2v) + 1) + sinh(2v)]
−
√
6
4 cos(u)[sin(u)(cosh(2v) + 1)− sinh(2v)]
−3
8 [cos2(u)(cosh(2v) + 1)− 2(cosh(2v) + sin(u) sinh(2v))]
−3
8 [cos2(u)(cosh(2v) + 1)− 2(cosh(2v)− sin(u) sinh(2v))]
. (6.12)
It follows from the discussion above that f̄ is a surjective map onto Σ, and we see that f̄ is also
independent of t. Moreover, it is straightforward to check that the tangent vectors f̄u, f̄v are
linearly independent for all (u, v) ∈ R2; therefore f̄ parametrizes a smooth surface Σ ⊂ R5 \{0},
as expected.
We can compute the centroaffine metric on Σ as in the previous examples. Let f : R3 → G
be the map corresponding to (6.12); i.e.,
f(u, v, t) = g1(u) · g2(v) · g0(t).
Then we have
Ω = M0α+M1ω
1
0 +M2ω
2
0 = f−1df.
22 N. Bushek and J.N. Clelland
Figure 3. Projections of surface from Example 6.5 to 3-D subspaces.
Comparing these two expressions for Ω shows that
ω1
0 =
√
3
2
e−t (cosh(v)du+ dv) , ω2
0 =
√
3
2
et (− cosh(v)du+ dv) .
Therefore, the centroaffine metric on Σ is given by
I = 2ω1
0ω
2
0 = 3
(
− cosh2(v)du2 + dv2
)
.
As expected, this metric has constant Gauss curvature K = 1
3 .
In this case, F3 is isomorphic to the orthonormal frame bundle of the time-like surface S2,1
consisting of all space-like vectors of length
√
3 in R2,1. This surface is a hyperboloid of one
sheet, and so we might expect that Σ would be diffeomorphic to this hyperboloid. However, if
we regard the domain of the parametrization (6.12) as S1 × R, we see that the map (6.12) is
invariant under the transformation
(u, v)→ (u+ π,−v). (6.13)
Therefore, Σ is diffeomorphic to a Möbius band, which is precisely the quotient of the hyper-
boloid by this map. (We note, however, that the map f : S1 × R → G is one-to-one, as the
frame vectors e1(u, v), . . . , e4(u, v) are not preserved by the transformation (6.13).)
As in the space-like examples, we can show that Σ is contained in the intersection of three
quadric hypersurfaces in R5 \ {0}. The coordinates of f̄(u, v) satisfy the quadratic equations
3x2
2 − x4(4x0 + 2) = 0, 3x2
1 − x3(4x0 + 2) = 0, 2x2
0 − x0 − 3x1x2 − 1 = 0. (6.14)
Therefore, Σ is contained in the (real) algebraic variety X ⊂ R5\{0} defined by equations (6.14).
Σ is an interesting and somewhat complicated subset of X: first observe that the projection
of X to the (x0, x1, x2) coordinate 3-plane consists of the hyperboloid of one sheet defined by
Geometry of Centroaffine Surfaces in R5 23
the third equation in (6.14), minus all points of the form (−1
2 , x1, x2) except for (−1
2 , 0, 0). The
projection of X to this punctured hyperboloid is one-to-one, except over the point (−1
2 , 0, 0),
where the inverse image consists of all points of the form (−1
2 , 0, 0, x3, x4). Meanwhile, Σ consists
of that portion of X that projects to the portion of the hyperboloid with x0 > −1
2 , together
with the curve{(
−1
2 , 0, 0, x3, x4
)
|x3x4 = 9
4 , x3, x4 > 0
}
.
Projections of Σ to the (x1, x2, x0), (x1, x2, x3), (x1, x2, x4), (x1, x3, x0), and (x1, x4, x0) coordi-
nate 3-planes are shown in Fig. 3.
We collect the results of this section in the following theorem:
Theorem 6.6. Let f̄ : M → R5 \ {0} be a centroaffine immersion whose image Σ = f̄(M) is
a homogeneous, nondegenerate, space-like or time-like centroaffine surface. Then Σ is equivalent
via the GL(5,R)-action on R5 \ {0} to one of the following:
• the immersion of the hyperbolic plane H2 of Example 6.3;
• the immersion of the sphere S of Example 6.4;
• the immersion of the Lorentzian surface S2,1 of Example 6.5.
Acknowledgments
This research was supported in part by NSF grants DMS-0908456 and DMS-1206272.
References
[1] Clelland J.N., From Frenet to Cartan: the method of moving frames, in preparation.
[2] Fulton W., Harris J., Representation theory, Graduate Texts in Mathematics, Vol. 129, Springer-Verlag,
New York, 1991.
[3] Furuhata H., Minimal centroaffine immersions of codimension two, Bull. Belg. Math. Soc. Simon Stevin 7
(2000), 125–134.
[4] Furuhata H., Kurose T., Self-dual centroaffine surfaces of codimension two with constant affine mean cur-
vature, Bull. Belg. Math. Soc. Simon Stevin 9 (2002), 573–587.
[5] Gardner R.B., Wilkens G.R., The fundamental theorems of curves and hypersurfaces in centro-affine geo-
metry, Bull. Belg. Math. Soc. Simon Stevin 4 (1997), 379–401.
[6] Griffiths P., On Cartan’s method of Lie groups and moving frames as applied to uniqueness and existence
questions in differential geometry, Duke Math. J. 41 (1974), 775–814.
[7] Ivey T.A., Landsberg J.M., Cartan for beginners: differential geometry via moving frames and exterior
differential systems, Graduate Studies in Mathematics, Vol. 61, Amer. Math. Soc., Providence, RI, 2003.
[8] Laugwitz D., Differentialgeometrie in Vektorräumen, unter besonderer Berücksichtigung der unendlichdi-
mensionalen Räume, Friedr. Vieweg & Sohn, Braunschweig, 1965.
[9] Li A.M., Wang C.P., Canonical centroaffine hypersurfaces in Rn+1, Results Math. 20 (1991), 660–681.
[10] Liu H.L., Wang C.P., The centroaffine Tchebychev operator, Results Math. 27 (1995), 77–92.
[11] Mayer O., Myller A., La géométrie centroaffine des courbes planes, Ann. Sćı. de l’Universit’e de Jassy 18
(1933), 234–280.
[12] Milnor J., Husemoller D., Symmetric bilinear forms, Ergebnisse der Mathematik und ihrer Grenzgebiete,
Vol. 73, Springer-Verlag, New York – Heidelberg, 1973.
[13] Nomizu K., Sasaki T., Centroaffine immersions of codimension two and projective hypersurface theory,
Nagoya Math. J. 132 (1993), 63–90.
[14] Nomizu K., Vrancken L., A new equiaffine theory for surfaces in R4, Internat. J. Math. 4 (1993), 127–165.
http://dx.doi.org/10.1007/978-1-4612-0979-9
http://dx.doi.org/10.1215/S0012-7094-74-04180-5
http://dx.doi.org/10.1007/BF03323203
http://dx.doi.org/10.1007/BF03322271
http://dx.doi.org/10.1142/S0129167X9300008X
24 N. Bushek and J.N. Clelland
[15] Scharlach C., Centroaffine first order invariants of surfaces in R4, Results Math. 27 (1995), 141–159.
[16] Scharlach C., Centroaffine differential geometry of (positive) definite oriented surfaces in R4, in New Deve-
lopments in Differential Geometry (Budapest, 1996), Kluwer Acad. Publ., Dordrecht, 1999, 411–428.
[17] Scharlach C., Simon U., Verstraelen L., Vrancken L., A new intrinsic curvature invariant for centroaffine
hypersurfaces, Beiträge Algebra Geom. 38 (1997), 437–458.
[18] Scharlach C., Vrancken L., A curvature invariant for centroaffine hypersurfaces. II, in Geometry and Topo-
logy of Submanifolds, VIII (Brussels, 1995/Nordfjordeid, 1995), World Sci. Publ., River Edge, NJ, 1996,
341–350.
[19] Scharlach C., Vrancken L., Centroaffine surfaces in R4 with planar ∇-geodesics, Proc. Amer. Math. Soc.
126 (1998), 213–219.
[20] Wang C.P., Centroaffine minimal hypersurfaces in Rn+1, Geom. Dedicata 51 (1994), 63–74.
[21] Wilkens G.R., Centro-affine geometry in the plane and feedback invariants of two-state scalar control sys-
tems, in Differential Geometry and Control (Boulder, CO, 1997), Proc. Sympos. Pure Math., Vol. 64, Amer.
Math. Soc., Providence, RI, 1999, 319–333.
[22] Yang Y., Liu H., Minimal centroaffine immersions of codimension two, Results Math. 52 (2008), 423–437.
[23] Yang Y., Yu Y., Liu H., Flat centroaffine surfaces with the degenerate second fundamental form and vani-
shing Pick invariant in R4, J. Math. Anal. Appl. 397 (2013), 161–171.
http://dx.doi.org/10.1007/BF03322278
http://dx.doi.org/10.1090/S0002-9939-98-04408-6
http://dx.doi.org/10.1007/BF01264101
http://dx.doi.org/10.1090/pspum/064/1654544
http://dx.doi.org/10.1007/s00025-008-0321-5
http://dx.doi.org/10.1016/j.jmaa.2012.07.047
1 Introduction
2 Centroaffine surfaces in R5, adapted frames, and Maurer–Cartan forms
3 Reduction of the structure group and first-order invariants
4 The space-like case
5 The time-like case
6 Homogeneous examples
6.1 Space-like homogeneous examples
6.2 Time-like homogeneous examples
References
|