Professional Documents
Culture Documents
A New Look at Lie Algebras: Alina Dobrogowska, Grzegorz Jakimowicz
A New Look at Lie Algebras: Alina Dobrogowska, Grzegorz Jakimowicz
Abstract
We present a new look at description of real finite-dimensional Lie
algebras. The basic element turns out to be a pair (F, v) consisting of
a linear mapping F ∈ End(V ) and its eigenvector v. This pair allows
to build a Lie bracket on a dual space to a linear space V . This algebra
is solvable. In particular, when F is nilpotent, the Lie algebra is also
nilpotent. We show that these solvable algebras are the basic bricks
of the construction of all other Lie algebras. Using relations between
the Lie algebra, the Lie–Poisson structure and the Nambu bracket, we
show that the algebra invariants (Casimir functions) are solutions of
an equation which has a geometric sense. Several examples illustrate
the importance of these constructions.
Keywords: Lie algebras, nilpotent and solvable Lie algebras, Lie brack-
ets, Poisson brackets, Nambu brackets
1 Introduction
There is a well-known isomorphism between special orthogonal Lie algebra
so(3) and R3 . For the first structure, the Lie bracket is given by the matrix
commutator [X, Y ] = XY − Y X for X, Y ∈ so(3), and for the second by the
cross product × for vectors from R3 . The mapping
0 −z y x
X= z 0 −x 7−→ v = y
(1)
−y x 0 z
1
The main goal of the paper is to show that one can construct a similar
isomorphism for any Lie algebra. We will show that Lie algebras have a lot
in common with linear maps, and more precisely with linear maps with a
fixed eigenvector. We will present an easy construction of a Lie bracket on a
linear space V ∗ from a pair (F, v), where F ∈ End(V ) and v is an eigenvector
of F . In our considerations, we will restrict ourselves to the linear space V
over a field R. It means that we will analyze in detail only real Lie algebras.
However, we want to emphasize that the presented formulas also work for
vector spaces over the field of complex numbers.
The presented construction seems to us new and at the same time very
simple. Moreover, it generates natural connections between algebra, analysis
and geometry. To be precise, we have relationship between objects such
as: Lie algebras, eigenproblem for operators, Lie–Poisson brackets, Nambu
brackets and finally invariants (Casimir operators) for initial Lie algebras.
The idea of constructing these Lie brackets has its origin in the algebroid
structure. The concept of Lie algebroid come from Lie groupoids [32]. A Lie
algebroid (A, [·, ·], a) is a vector bundle A −→ M over manifold M, together
with a vector bundle map (anchor map) a : A −→ T M and Lie bracket
[·, ·] : Γ(A) × Γ(A) −→ Γ(A). The anchor and the Lie bracket satisfy the
Leibniz rule
[α, f β] = f [α, β] + a(α)(f )β (2)
for all α, β ∈ Γ(A), f ∈ C ∞ (M). It is a well-known fact that there are one-
to-one correspondences between vector bundles with linear Poisson struc-
tures and Lie algebroids as well as between vector spaces with linear Poisson
structures and Lie algebras. There are many books and articles about Lie
albebroids or related structures [6, 9, 13, 17, 19, 20, 22, 23, 24, 25, 28, 36, 37].
The idea of Lie brackets construction has exactly its source in our work
[8] (see also [10, 11]), in which we presented new ideas how to construct
algebroid structures on cotangent bundles starting from classical algebroid
structures on T M and vector fields satisfying certain assumptions. We also
found in this paper constructions of families of algebroids on A∗ when we
have an algebroid structure on A. It is also possible to look at these brackets
in the language torision-free and flat connections as described by Balcerzak
[2].
It also seems that linking the Lie bracket to eigenproblems of linear oper-
ators may be a step towards the classification of Lie algebras. The problem
of classifying real Lie algebras is completely solved for Lie algebras up to
2
dimension six, see [3, 5, 14, 15, 21, 26, 31] and the references therein.
The paper is organized as follows. In the beginning of Section 2 we recall
basic facts about Lie algebras, Poisson brackets and Nambu brackets. Next
sections contain the main results of the article. In Section 3 we define the
Lie bracket on V ∗ starting from a pair (F, v), where F ∈ End(V ) and v ∈ V
(Theorem 1). Moreover, we study the properties of Lie algebras obtained
in this way. We present some examples in Section 4. The culmination of
our research are Theorem 8 and 9 in Section 5, where we show that every
Lie algebra and its invariants can be associated with a collection of linear
mappings and their eigenvectors.
for all x, y, z ∈ g.
We say that a linear subspace h is an ideal of a Lie algebra g when
[g, h] ⊆ h. Of course the set [h, h] is also an ideal. In particular, g is a
trivial ideal of g. Then we define a sequence of ideals (the derived series
g(0) ⊇ g(1) ⊇ · · · ⊇ g(i) ⊇ . . . )
g(0) = g, g(1) = [g, g], g(2) = [g(1) , g(1) ], . . . , g(i) = [g(i−1) , g(i−1) ], . . . (4)
A Lie algebra g is called solvable if, for some positive integer i, g(i) = 0. In
addition, if we introduce the following sequence of ideals (the lower central
series g(0) ⊇ g(1) ⊇ · · · ⊇ g(i) ⊇ . . . )
g(0) = g, g(1) = [g(0) , g], g(2) = [g(1) , g], . . . , g(i) = [g(i−1) , g], . . . , (5)
we say that algebra g is called nilpotent if the lower central series terminates
g(i) = 0 for some i ∈ N. Obviously, a nilpotent Lie algebra is also solvable.
3
Let M be a finite-dimensional smooth manifold. A Poisson bracket {·, ·} :
C (M) × C ∞ (M) −→ C ∞ (M) is a bilinear, antisymmetric map, which
∞
In the linear case for the Poisson bracket, these functions correspond to
invariants (Casimir operators) of the Lie algebra.
In 1973, Nambu [27] proposed a generalization of the Poisson bracket on
R3 to the Nambu bracket in the form
3
∂(f1 , f2 , f3 ) X ∂f1 ∂f2 ∂f3
{f1 , f2 , f3 }(x) = = ǫijk (x) (x) (x), (10)
∂(x1 , x2 , x3 ) i,j,k=1 ∂xi ∂xj ∂xk
4
and Leibniz rule (derivation law)
{f1 , . . . , fn−1 , f g} = f {f1 , . . . , fn−1 , g} + {f1 , . . . , fn−1 , f }g. (12)
More information about Nambu structure can be found in [1, 4, 7, 16, 18,
33, 35].
5
If we have two linear mappings F and G that have a common eigenvector
v then the following property holds [ψ, φ]F,v + [ψ, φ]G,v = [ψ, φ]F +G,v .
By introducing the notion of the internal multiplication operator ιv :
V −→ R given by ιv ψ = ψ(v), we can rewrite the Lie bracket (13) in the
∗
form
[ψ, φ]F,v = (ιv ∧ F ∗ ) (φ, ψ) . (16)
Since we have the canonical isomorphism V ∗∗ ∼ = V , we will identify the
vector v with ιv . Moreover, we will use isomorphisms End(V ) ∼ = V ∗ ⊗ V or
∗ ∼
End(V ) = V ⊗ V defined by (ψ ⊗ v)(w) = ψ(w)v, (v ⊗ ψ)(φ) = φ(v)ψ for
∗
or
2. F ∗ = ιv ⊗ ρ + ιF v ⊗ η, where ρ, η ∈ V ∗ .
Proof. For F = 0 it is obvious. Suppose now that F is not equal zero. After
simple calculation we obtain
[[ψ, φ]F,v , ζ]F,v + [[ζ, ψ]F,v , φ]F,v + [[φ, ζ]F,v , ψ]F,v = (17)
6
It implies that F ∗ (e∗i ) = 0 for i = 3, ..., n. Therefore, the mapping in this
basis has the form
0 a12 a13 . . . a1n
1 a22 a23 . . . a2n
F = 0 0 0 ... 0 (19)
,
.. .. .. . . ..
. . . . .
0 0 0 ... 0
We recognize that it is the Lie algebra structure related to the Lie algebra
g2,1 (more exactly it is isomorphic with the direct sum g2,1 ⊕ he∗3 , ..., e∗n i). An
isomorphism is given by the mapping {e∗1 , e∗2 , e∗3 , ..., e∗n } 7→ {e∗2 , e∗1 + a22 e∗2 +
. . . + a2n e∗n , e∗3 , ..., e∗n }.
Even though this algebra was obtained using vector v which was not an
eigenvector of F , it can also be obtained from the mapping F = diag(0, −1, 0 . . . , 0)
and the eigenvector v = e1 . Thus without the loss of generality from now on
we will always assume that v is an eigenvector for F .
It is easy to prove that this construction of Lie brackets gives solvable Lie
algebras.
Theorem 3. Let [·, ·]F,v be given by (13), then the Lie algebra (V ∗ , [·, ·]F,v )
is solvable.
7
Additionally, if we assume that the mapping F giving the Lie bracket is
nilpotent, then we obtain a nilpotent Lie algebra.
..
.
Finally, since (F ∗ )n = 0 for some n, then we have g(n) = 0.
Now, we show when two different linear mappings give to isomorphic Lie
algebras.
Then Lie algebras (V ∗ , [·, ·]F,v ), (W ∗ , [·, ·]G,w ) are isomorphic if and only if
there exists a linear bijection Φ : V −→ W such that
8
On the other hand if we suppose that (V ∗ , [·, ·]F,v ) ∼
= (W ∗ , [·, ·]G,w ), which
means there is a linear bijection Ψ : W → V such that
∗ ∗
9
Proof. Using Jacobi identity we calculate
1
[[ψ, φ]ǫF,v,G,w , ζ]ǫF,v,G,w + [[ζ, ψ]ǫF,v,G,w , φ]ǫF,v,G,w + [[φ, ζ]ǫF,v,G,w , ψ]ǫF,v,G,w
ǫ
(31)
∗ ∗
= ζ(w) (φ(v) ([F, G]) (ψ) − ψ(v) ([F, G]) (φ))
+φ(w) (ψ(v) ([F, G])∗ (ζ) − ζ(v) ([F, G])∗ (ψ))
+ψ(w) (ζ(v) ([F, G])∗ (φ) − φ(v) ([F, G])∗ (ζ))
+ (ψ(w)φ(G(v)) − φ(w)ψ(G(v))) F ∗ (ζ)+(ζ(w)ψ(G(v)) − ψ(w)ζ(G(v))) F ∗ (φ)
+ (φ(w)ζ(G(v)) − ζ(w)φ(G(v))) F ∗ (ψ)+(ψ(v)φ(F (w)) − φ(v)ψ(F (w))) G∗ (ζ)
+ (ζ(v)ψ(F (w)) − ψ(v)ζ(F (w))) G∗ (φ)+(φ(v)ζ(F (w)) − ζ(v)φ(F (w))) G∗ (ψ) = 0
for ψ, φ, ζ ∈ V ∗ . It can be rewritten as
∗ ∗ ∗
ιw ∧ [F, G] ∧ ιv + ιw ∧ ιGv ∧ F + ιv ∧ ιF w ∧ G (ψ, φ, ζ) = 0. (32)
[F, G] = 0, (33)
∃λ ∈ R F v = λv, (34)
∃γ ∈ R Gw = γw, (35)
F w = 0, (36)
Gv = 0, (37)
10
Proof. Let ψ1 , ..., ψk+1 ∈ V ∗ and (F1 , v1 ), ..., (Fk , vk ) be pairs such that Fi ∈
End(V ), vi ∈ V and Fi vi = λi vi , where λi ∈ R for all i = 1, . . . , k. We put
ǫ = 1 which does not reduce generality. We assume that the pairs (Fi , vi ),
(Fj , vj ) for all i, j ∈ {1, , , , k} satisfy all the assumptions of previous Corollary
1. Then
[[[[ψ1 , ψ2 ]1F1 ,v1 , ψ3 ]1F2 ,v2 , ...]1Fk−1 ,vk−1 , ψk+1 ]1Fk ,vk (38)
= ((F1 ◦ ... ◦ Fk )∗ ⊗ ιv1 ⊗ ... ⊗ ιvk − ιv1 ⊗ (F1 ◦ ... ◦ Fk )∗ ⊗ ιv2 ⊗ ... ⊗ ιvk )
(ψ1 , ..., ψk+1 ).
In our case (Fi , vi ) are either (F, v) or (G, w). From above and the assumption
that F, G are nilpotent the theorem is proved.
4 Examples
In this section we study the case of dimensions three and four. We will now
show that using the above theorems we can easily obtain low dimensional
Lie algebras. We use the classification from the paper [30], see also [34].
We will start with the three-dimensional Lie algebras.
1. If we take
λ1 0 0
F = 0 λ2 0 , (39)
0 0 0
where λ1 , λ2 ∈ R, we obtain the Lie bracket of the form
11
(a) For λ1 = λ2 = 1, we recognize the Lie structure related to the
Lie algebra g3,3 . The commutator rules for g3,3 are [e∗1 , e∗3 ] = e∗1 ,
[e∗2 , e∗3 ] = e∗2 .
(b) For λ1 = −λ2 = 1, we recognize the Lie structure related to the
Lie algebra g3,4 . The commutator rules for g3,4 are [e∗1 , e∗3 ] = e∗1 ,
[e∗2 , e∗3 ] = −e∗2 .
(c) For λ1 = 1, λ2 = a, we recognize the Lie structure related to the
Lie algebra ga3,5 . The commutator rules for ga3,5 are [e∗1 , e∗3 ] = e∗1 ,
[e∗2 , e∗3 ] = ae∗2 .
(d) For λ1 = 1, λ2 = 0, we recognize the Lie structure related to the
Lie algebra g2,1 ⊕ he∗2 i.
2. If we take
λ1 0 0
F = 1 λ1 0 , (42)
0 0 0
where λ1 ∈ R, we obtain the Lie bracket of the form
[ψ, φ]F,v = (λ1 (ψ1 φ3 − ψ3 φ1 ) + ψ2 φ3 − ψ3 φ2 ) e∗1 (43)
+ λ1 (ψ2 φ3 − ψ3 φ2 ) e∗2 .
The commutator rules are following
[e∗1 , e∗2 ] = 0, [e∗1 , e∗3 ] = λ1 e∗1 , [e∗2 , e∗3 ] = e∗1 + λ1 e∗2 . (44)
(a) For λ1 = 0, we recognize the Lie structure related to the Lie algebra
g3,1 . The commutator rule for g3,1 is [e∗2 , e∗3 ] = e∗1 .
(b) For λ1 = 1, we recognize the Lie structure related to the Lie algebra
g3,2 . The commutator rules for g3,2 are [e∗1 , e∗3 ] = e∗1 , [e∗2 , e∗3 ] =
e∗1 + e∗2 .
3. If we take
a −1 0
F = 1 a 0 , (45)
0 0 0
where a ∈ R, we obtain the Lie bracket of the form
[ψ, φ]F,v = (a (ψ1 φ3 − ψ3 φ1 ) + ψ2 φ3 − ψ3 φ2 ) e∗1 (46)
+ (−ψ1 φ3 + ψ3 φ1 + a (ψ2 φ3 − ψ3 φ2 )) e∗2 .
12
The commutator rules are following
[e∗1 , e∗2 ] = 0, [e∗1 , e∗3 ] = ae∗1 − e∗2 , [e∗2 , e∗3 ] = e∗1 + ae∗2 . (47)
(a) For a 6= 0, we recognize the Lie structure related to the Lie algebra
ga3,7 . The commutator rules for ga3,7 are [e∗1 , e∗3 ] = ae∗1 −e∗2 , [e∗2 , e∗3 ] =
e∗1 + ae∗2 .
(b) For a = 0, we recognize the Lie structure related to the Lie algebra
g3,6 . The commutator rule for g3,6 are [e∗1 , e∗3 ] = −e∗2 , [e∗2 , e∗3 ] = e∗1 .
4. If we take
0 −1 0
F = 1 0 0 , (48)
0 0 0
and as the second linear mapping and its eigenvector we take
0 0 1 0
G = 0 0 0 , w = 1 , (49)
0 0 0 0
then we obtain the Lie bracket of the form
[e∗1 , e∗2 ] = e∗3 , [e∗1 , e∗3 ] = −e∗2 , [e∗2 , e∗3 ] = e∗1 . (51)
Above we recognize the Lie structure related to the Lie algebra g3,9 =
so(3).
5. If we take
0 −2 0
F = 0 0 0 , (52)
0 0 0
and as the second linear mapping and its eigenvector we take
1 0 0 0
G = 0 0 0 , w = 1 , (53)
0 0 −1 0
13
then we obtain the Lie bracket of the form
[e∗1 , e∗2 ] = e∗1 , [e∗1 , e∗3 ] = −2e∗2 , [e∗2 , e∗3 ] = e∗3 . (55)
Above we recognize the Lie structure related to the Lie algebra g3,8 =
sl(2, R).
We have obtained all three dimensional Lie algebras, see Table 1. We got
seven of them using one linear mapping and its eigenvector. To describe the
other two algebras we needed two linear mappings and their eigenvectors.
In this case, after identifying V ∼ = V∗ ∼ = R3 , the bracket (13) can be
written as
[ψ, φ]F,v = F ⊤ (v × (ψ × φ)) , (56)
where we used the vector triple product expansion
14
where l ∈ C ∞ (R3 )), then we recognize in this the structure of the Nambu
bracket
∂(f, g, c)
{f, g, c}(x) = . (62)
∂(x1 , x2 , x3 )
If we treat a function c as a fixed parameter we obtain from Nambu bracket
a Poisson bracket for which c is a Casimir function. We illustrate it by some
examples.
Example 2. Let us take the Lie algebra g3,3 and let us consider the equation
(61) for Casimir function in this case
∂c
(x1 , x2 , x3 ) = x2 l(x1 , x2 , x3 )
∂x
1
∂c
(x1 , x2 , x3 ) = −x1 l(x1 , x2 , x3 ) . (63)
∂x2
∂c
(x1 , x2 , x3 ) = 0
∂x3
Determining from the first equation the function l and substituting into the
second equation we obtain the linear first order partial differential equation
∂c ∂c
− x1 (x1 , x2 , x3 ) + x2 (x1 , x2 , x3 ) = 0. (64)
∂x1 ∂x2
After solving, we have Casimir functions c and the auxiliary function l
x2
c(x1 , x2 , x3 ) = , (65)
x1
1
l(x1 , x2 , x3 ) = 2 , (66)
x1
see also Table 1 (whether the integral factor appears significantly has to do
with the compatibility of the Lie brackets, see [12, 29]).
15
F v l(x) Casimir Name
G w
0 0 0 0
1
F = 1 v = 0 x1
0 0 g3,1
x1
0 0 0 1
1 0 0 0 x2 x2
1 −
x1
−
x1
F = 1 v = 0 e x1 e
1 0 g3,2
x1
0 0 0 1
1 0 0 0
1 x2
F = 0 v = 0
1 0 − g3,3
x2
1 x1
0 0 0 1
1 0 0 0
F = 0 v = 0 x1 x2
−1 0
−1 g3,4
0 0 0 1
1 0 0 0
1 x1
F = 0 v = 0 ga
a 0 − a+1 3,5
x1 xa
1
0 0 0 1
0 −1 0 0
F = 1 v = 0 2 x2 2
1 + x2
0 0
g3,6
0 0 0 1
a −1 0 0 x1 x1
2aarctg 2aarctg
F = 1 v = 0 2e x2 (x2 2
1 + x2 )e
x2 ga
a 0 3,7
0 0 0 1
0 −2 0 0
F = 0 v = 0 1 x1 x3 + x2
0 0
2 g3,8
0 0 0 1
1 0 0 0
G = 0 w = 1
0 0
0 0 −1 0
0 −1 0 0
F = 1 v = 0 2 x2 2 2
1 + x2 + x3
0 0
g3,9
0 0 0 1
0 0 1 0
G = 0 w = 1
0 0
0 0 0 0
16
1. If we take
λ1 0 0 0
0 λ2 0 0
F =
0 0 λ3
, (67)
0
0 0 0 0
where λ1 , λ2 , λ3 ∈ R, we obtain the Lie bracket of the form
where ψ = ψ1 e∗1 + ψ2 e∗2 + ψ3 e∗3 + ψ4 e∗4 and φ = φ1 e∗1 + φ2 e∗2 + φ3 e∗3 + φ4 e∗4 .
The nonzero commutator rules are following
[e∗1 , e∗4 ] = λ1 e∗1 , [e∗2 , e∗4 ] = λ2 e∗2 , [e∗3 , e∗4 ] = λ3 e∗3 . (69)
2. If we take
λ1 0 0 0
1 λ1 0 0
F =
0 1 λ1
, (70)
0
0 0 0 0
where λ1 ∈ R, we obtain the Lie bracket of the form
[e∗1 , e∗4 ] = λ1 e∗1 , [e∗2 , e∗4 ] = e∗1 + λ1 e∗2 , [e∗3 , e∗4 ] = e∗2 + λ1 e∗3 . (72)
(a) For λ1 = 0, we recognize the Lie structure related to the Lie algebra
g4,1 . The commutator rules for g4,1 are [e∗2 , e∗4 ] = e∗1 , [e∗3 , e∗4 ] = e∗2 .
(b) For λ1 = 1, we recognize the Lie structure related to the Lie algebra
g4,4 . The commutator rules for g4,4 are [e∗1 , e∗4 ] = e∗1 , [e∗2 , e∗4 ] =
e∗1 + e∗2 , [e∗3 , e∗4 ] = e∗2 + e∗3 .
17
3. If we take
λ1 0 0 0
0 λ2 0 0
F =
0 1 λ2
, (73)
0
0 0 0 0
where λ1 , λ2 ∈ R, we obtain the Lie bracket of the form
[e∗1 , e∗4 ] = λ1 e∗1 , [e∗2 , e∗4 ] = λ2 e∗2 , [e∗3 , e∗4 ] = e∗2 + λ2 e∗3 . (75)
4. If we take
λ1 0 0 0
0 b −1 0
F =
0
, (76)
1 b 0
0 0 0 0
where λ1 , b ∈ R, we obtain the Lie bracket of the form
[e∗1 , e∗4 ] = λ1 e∗1 , [e∗2 , e∗4 ] = be∗2 − e∗3 , [e∗3 , e∗4 ] = e∗2 + be∗3 . (79)
(a) For λ1 = a, we recognize the Lie structure related to the Lie algebra
ga,b a,b ∗ ∗ ∗ ∗ ∗
4,6 . The commutator rules for g4,6 are [e1 , e4 ] = ae1 , [e2 , e4 ] =
∗ ∗ ∗ ∗ ∗ ∗
be2 − e3 , [e3 , e4 ] = e2 + be3 .
18
(b) For λ1 = 0, see Example 1.
5. If we take
0 0 0 0
0 0 −1 0
F =
0
, (80)
1 0 0
0 0 0 0
and as the second linear mapping and its eigenvector we choose
0 0 0 0 0
1 0 0 0 0
G= 0 0 0
, w = , (81)
0 1
0 0 0 0 0
then we obtain the Lie bracket of the form
[ψ, φ]F,v,G,w = [ψ, φ]F,v + [ψ, φ]G,w = (ψ2 φ3 − ψ3 φ2 ) e∗1 (82)
+ (ψ3 φ4 − ψ4 φ3 ) e∗2 − (ψ2 φ4 + ψ4 φ2 ) e∗3 .
The nonzero commutator rules are following
[e∗2 , e∗3 ] = e∗1 , [e∗2 , e∗4 ] = −e∗3 , [e∗3 , e∗4 ] = e∗2 . (83)
Above we recognize the Lie structure related to the Lie algebra g4,10 .
6. If we take
2 0 0 0
0 1 0 0
F =
0
, (84)
1 1 0
0 0 0 0
and as the second linear mapping and its eigenvector we choose
0 0 0 0 0
1 0 0 0 0 ,
G= 0 0 0 , w = (85)
0 1
0 0 0 0 0
then we obtain the Lie bracket of the form
[ψ, φ]F,v,G,w = [ψ, φ]F,v + [ψ, φ]G,w (86)
∗
= (ψ2 φ3 − ψ3 φ2 + 2 (ψ1 φ4 − ψ4 φ1 )) e1
+ (ψ2 φ4 − ψ4 φ2 + ψ3 φ4 − ψ4 φ3 ) e∗2 + (ψ3 φ4 + ψ4 φ3 ) e∗3 .
19
The nonzero commutator rules are following
[e∗2 , e∗3 ] = e∗1 , [e∗1 , e∗4 ] = 2e∗3 , [e∗2 , e∗4 ] = e∗2 , [e∗3 , e∗4 ] = e∗2 + e∗3 . (87)
Above we recognize the Lie structure related to the Lie algebra g4,7 .
7. If we take
0 0 0 0
0 1 0 0
F =
0
, (88)
0 −1 0
0 0 0 0
and as the second linear mapping and its eigenvector we choose
0 0 0 0 0
1 0 0 0 0 ,
G= 0 0 0 , w = (89)
0 1
0 0 0 0 0
[e∗2 , e∗3 ] = e∗1 , [e∗2 , e∗4 ] = e∗2 , [e∗3 , e∗4 ] = −e∗3 . (91)
Above we recognize the Lie structure related to the Lie algebra g4,8 .
8. If we take
1+b 0 0 0
0 1 0 0
F =
0
, (92)
0 b 0
0 0 0 0
and as the second linear mapping and its eigenvector we choose
0 0 0 0 0
1 0 0 0 0 ,
G= 0 0 0 , w = (93)
0 1
0 0 0 0 0
20
then we obtain the Lie bracket of the form
[ψ, φ]F,v,G,w = [ψ, φ]F,v + [ψ, φ]G,w (94)
= (ψ2 φ3 − ψ3 φ2 + (1 + b) (ψ1 φ4 − ψ4 φ1 )) e∗1
+ (ψ2 φ4 − ψ4 φ2 ) e∗2 + b (ψ3 φ4 + ψ4 φ3 ) e∗3 .
The nonzero commutator rules are following
[e∗2 , e∗3 ] = e∗1 , [e∗1 , e∗4 ] = (1 + b)e∗1 , [e∗2 , e∗4 ] = e∗2 , [e∗3 , e∗4 ] = be∗3 . (95)
Above we recognize the Lie structure related to the Lie algebra gb4,9 .
9. If we take
2a 0 0 0
0 a −1 0
F =
0
, (96)
1 a 0
0 0 0 0
and as the second linear mapping and its eigenvector we choose
0 0 0 0 0
1 0 0 0 0
G= 0 0 0
, w = , (97)
0 1
0 0 0 0 0
then we obtain the Lie bracket of the form
[ψ, φ]F,v,G,w = [ψ, φ]F,v + [ψ, φ]G,w (98)
= (ψ2 φ3 − ψ3 φ2 + 2a (ψ1 φ4 − ψ4 φ1 )) e∗1
+ (a (ψ2 φ4 − ψ4 φ2 ) + ψ3 φ4 − ψ4 φ3 ) e∗2
+ (−ψ2 φ4 + ψ4 φ2 + a (ψ3 φ4 + ψ4 φ3 )) e∗3 .
The nonzero commutator rules are following
[e∗2 , e∗3 ] = e∗1 , [e∗1 , e∗4 ] = 2ae∗1 , [e∗2 , e∗4 ] = ae∗2 − e∗3 ,
[e∗3 , e∗4 ] = e∗2 + ae∗3 .
(99)
a
Above we recognize the Lie structure related to the Lie algebra g4,11 .
10. If we take
0 −1 0 0
1 0 0 0
F =
0 0 0 0 ,
(100)
0 0 0 0
21
and as the second linear mapping and its eigenvector we choose
1 0 0 0 0
0 1 0 0 0 ,
G= 0 0 0 , w = (101)
0 1
0 0 0 0 0
[e∗1 , e∗3 ] = e∗1 , [e∗2 , e∗3 ] = e∗2 , [e∗1 , e∗4 ] = −e∗2 , [e∗2 , e∗4 ] = e∗1 . (103)
Above we recognize the Lie structure related to the Lie algebra g4,12 .
In conclusion, we have obtained all four-dimensional Lie algebras, see Ta-
ble 2. To generate six of them, we had to use one linear mapping and its
eigenvector. The remaining six algebras required the use of two linear map-
pings and their eigenvectors.
We see that in the case of building three- and four-dimensional Lie alge-
bras (their classification), the first "layer" of linear mapping is closely related
to the classification of equilibrium points for a linear homogeneous system
of differential equations with constant coefficients. We can assume that the
first "layer" is related to the Jordan form of the map F (more precisely with
the class of such maps).
5 General structure
In this section, we study a general situation of arbitrary finite dimension.
We answer the question how a linear mapping and its eigenvector or linear
mappings and their eigenvectors can be assigned to a Lie algebra, when
its structure constants are known. We also find the geometric formula for
Casimir functions in the language of pairs (Fi , vi ), where Fi ∈ End(V ) and
vi is the eigenvector of the mapping Fi .
22
F v Name F v G w Name
0 0 0 0 0 2 0 0 0 0 0 0 0 0 0
1 0 0 0 0 0 1 0 0 0 1 0 0 0 0
F = v = F = v = G= w =
g4,1 g4,7
0 1 0 0 0 0 1 1 0 0
0 0 0 0 1
0 0 0 0 1 0 0 0 0 1 0 0 0 0 0
a 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 1 0 0 0 0 1 0 0 0 1 0 0 0 0
F = v = ga F = v = G= w =
4,2 g4,8
0 1 1 0 0 0 0 −1 0 0 0 0 0 0 1
0 0 0 0 1 0 0 0 0 1 0 0 0 0 0
1 0 0 0 0 1 + b 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 1 0 0 0 1 0 0 0 0
F = v = F = v = G= w = gb
g4,3 4,9
0 1 0 0 0 0 0 b 0 0 0 0 0 0 1
0 0 0 0 1 0 0 0 0 1 0 0 0 0 0
1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 0 0 0 0 0 −1 0 0 1 0 0 0 0
F = v = F = v = G= w =
g4,4 g4,10
0 1 1 0 0 0 1 0 0 0 0 0 0 0 1
0 0 0 0 1 0 0 0 0 1 0 0 0 0 0
1 0 0 0 0 2a 0 0 0 0 0 0 0 0 0
0 a 0 0 0 0 a −1 0 0 1 0 0 0 0
F = v = gab F = v = G= w =
4,5 g4,11
0 0 b 0 0 0 1 a 0 0 0 0 0 0 1
0 0 0 0 1 0 0 0 0 1 0 0 0 0 0
a 0 0 0 0 0 −1 0 0 0 1 0 0 0 0
0 b −1 0 0 1 0 0 0 0 0 1 0 0 0
F = v = gab F = v = G= w =
4,6 g4,12
0 1 b 0 0 0 0 0 0 0 0 0 0 0 1
0 0 0 0 1 0 0 0 0 1 0 0 0 0 0
Table 2: Linear mappings and their eigenvectors giving four dimensional Lie
algebras.
23
Let {e1 , . . . , en } be a basis of g. As is well
Pn known, Lie algebra structure is
given by commutation relations [ei , ej ] = k=1 cij ek , where ckij are structure
k
24
mapping
c11 i c21 i ... ci−1
1i 0 ci+1
1i ... cn1 i
c12 i c22 i ... ci−1
2i 0 ci+1
2i ... cn2 i
.. .. .. .. .. .. .. ..
. . . . . . . .
. . . ci−1 0 ci+1
1
c2
c . . . cni−1 i
(108)
Fn−i+1 = i−1 i i−1 i i−1 i i−1 i
0 0 ... 0 0 0 ... 0
0 0 ... 0 0 −ci+1
i i+1 ... 0
. .. .. .. .. .. .. ..
.. . . . . . . .
0 0 ... 0 0 0 0 −cnin
[ψ, φ]F1 ,v1 ,...,Fn ,vn = [ψ, φ]F1,v1 + [ψ, φ]F2 ,v2 + . . . + [ψ, φ]Fn,vn . (110)
Theorem 8. Every Lie algebra (g, [·, ·]) is isomorphic to the corresponding
Lie algebra (Rn , [·, ·]F1,v1 ,...,Fn ,vn ).
Proof follows from writing explicitly commutation relations for [·, ·]F1,v1 ,...,Fn ,vn
and comparing them with commutation relations for g.
The isomorphism (g, [·, ·]) ∼
= (Rn , [·, ·]F1,v1 ,...,Fn ,vn ) is not canonical, we can
assign the linear mappings and their eigenvectors differently.
Each bracket (13) in the sum (110) can be rewritten in the form
25
Using the formulas (8) and (111), we can write (110) in the language of
Poisson brackets
* n +
X
{f, g}F1,v1 ,...,Fn ,vn (x) = (Fi (x) ∧ vi ) |df (x) ⊗ dg(x) , f, g ∈ C ∞ (Rn ) ,
i=1
(112)
V2
where x =V(x1 , . . . , xn ). In the above formula Fi (x) ∧ vi ∈ V and df (x) ∧
dg(x) ∈ V2 V ∗ . Of course, we can explicitly assign a two-vector ∇f (x) ∧
∇g(x) ∈ 2 V for a two-form df (x) ∧ dg(x). Since we have identified V with
Rn so we have the scalar V product which is symmetric. It extends naturally
to the space of k-vectors k V
hw|ti = det (hwi |tj i) , (113)
where i, j = 1, . . . , k andVw = w1 ∧·V· ·∧wk , t = t1 ∧· · ·∧tk . Finally, using the
Hodge star operator ∗ : 2 V −→ n−2 V , which has the following property
2
^
hw|ti = det (w ∧ ∗t) , w, t ∈ V, (114)
formula (112) can be rewritten as
n
!
X
{f, g}F1,v1 ,...,Fn ,vn (x) = det ∇f (x) ∧ ∇g(x) ∧ ∗ (Fi (x) ∧ vi ) . (115)
i=1
Proof. The proof of the theorem follows directly from the Poisson bracket
form (115) and the definition of the Casimir function.
If only the constant function satisfies the equation (116), then the Lie
algebra has no Casimir functions. In the case when the algebra has n − 2
Casimir functions, they also satisfy the following equation
n
!
X
∇c1 ∧ ∇c2 ∧ · · · ∧ ∇cn−2 = l(x) ∗ (Fi (x) ∧ vi ) (117)
i=1
26
for some function l ∈ C ∞ (Rn ). This function plays the role of an integrating
factor.
Acknowledgments
The second author G.J. was partially supported by National Science Centre,
Poland project 2020/01/Y/ST1/00123.
References
[1] H. Awata, M. Li, D. Minic, T. Yoneya, On the quanti-
zation of Nambu brackets, Journal of High Energy Physics,
JHEP02(2001)013, 2001.
[3] Boza, L.; Fedriani, E.M.; Nunez, J.; Tenorio, A.F. A Historical
review of the classifications of Lie algebras. Rev. Union Mat. Argent.
2013, 54, 75–99.
[7] F. Crespo, F.J. Molero and S. Ferrer, Poisson and integrable systems
through the Nambu bracket and its Jacobi multiplier, Journal of Ge-
ometric Mechanics, 8(2): 169-178, 2016. doi: 10.3934jgm.2016002
27
[9] A. Dobrogowska, G. Jakimowicz, Tangent lifts of bi-Hamiltonian
structures, J. Math. Phys., 58, 083505, 2017.
[13] Dufour J-P., Zung N.T., Poisson Structures and Their Normal
Forms, Birkhäuser Verlag 2005.
[18] P-M. Ho, Y. Matsuo, The Nambu bracket and M-theory, Progress of
Theoretical and Experimental Physics, Volume 2016, Issue 6, June
2016, 06A104, https://doi.org/10.1093/ptep/ptw075.
28
[19] Jóźwikowski, M., Rotkiewicz, M., Higher-order analogs of lie alge-
broids via vector bundle comorphisms, SIGMA 14 (2018), 135.
29
[30] J. Patera, R.T. Sharp, P. Winternitz, H. Zassenhaus, Invariants of
real low dimension Lie algebras, J. Math. Phys., 17, 986, 1976.
[32] Pradines, J., Théorie de Lie pour les grupoïdes différentiables. Re-
lations entre propriétés locales et globales, C. R. Acad. Sc. Paris,
Ser. A, t. 264, 245-248, 1967.
[33] A.O. Shishanin, Nambu mechanics and its applications, IOP Conf.
Ser.: Mater. Sci. Eng. 468 012029, 2018.
[37] Xu, P., Quantum grupoids, Commun. Math. Phys. 216, 539-581,
2001.
30