00footnotetext: Corresponding author: Ruipu Bai, E-mail: [email protected].
Semi-Associative 3-Algebras
RuiPu Bai
College of Mathematics and Information Science,
Hebei University
[email protected]
and
Yan Zhang
College of Mathematics and Information Science,
Hebei University, Baoding 071002, China
[email protected]
Abstract.
A new 3-ary non-associative algebra, which is called a semi-associative 3-algebra, is introduced, and the double modules and double extensions by cocycles are provided. Every semi-associative 3-algebra (A,{,,}) has an adjacent 3-Lie algebra (A,[,,]c). From a semi-associative 3-algebra (A,{,,}), a double module (ϕ,ψ,M) and a cocycle θ, a semi-direct product semi-associative 3-algebra A⋉ϕψM and a double extension (A+˙A∗,{,,}θ) are constructed, and structures are studied.
††2010 Mathematics Subject Classification: 15A69; 17D99.††Key words and phrases: semi-associative 3-algebra, 3-Lie algebra, double module, extension
1. Introduction
3-ary algebraic systems are applied in
mathematics and mathematical physics. For example, 3-Lie algebras
provided by Filippov in 1985 [1, 2], have close relation with
the theory of integrable-systems and Nambu mechanical-systems
[3, 4]. Bagger and Lambert [5, 6, 7] proposed a
supersymmetric fields theory model for multiple M2-branes based
on the metric 3-Lie algebras [8].
C. Bai and coauthors [9], defined pre-3-Lie algebras and local cocycle 3-Lie bialgebras and generalized Yang-Baxter equation in 3-Lie algebras, and proved that tensor form solutions of 3-Lie Yang-Baxter equation can be constructed by local cocycle 3-Lie bialgebras. Three classes of 3-algebras are constructed in [10]:
(1) The classical Nambu bracket based on three variables ( say x,y,z ) is the simplest to
compute, in most situations. It involves the Jacobian-like determinant of partial derivatives
of three functions A,B,C,
[TABLE]
and it satisfies Filippove identity:
[TABLE]
(2) The sum of single operators producted with commutators of the remaining two,
or as anticommutators acting on the commutators,
[TABLE]
The multiplication does not satisfy Filippov identity, and the 3-algebra can by
realized by operators [10].
(3) The re-packaging
commutators of any Lie algebra to define
[TABLE]
This is again a totally antisymmetric trilinear combination, but it is a singular construction
for finite dimensional realizations in the sense that Tr⟨A,B,C⟩=0, and authors in [11] proved that all the m-dimensional 3-Lie algebras with m≤5, except for the unique simple 3-Lie algebra,
can be constructed by the method from Lie algebras.
Likewise, in [9], the paper provides an algebra which is called the 3-pre-Lie algebra, and induces a sub-adjacent 3-Lie algebra. However,
the 3-pre-Lie algebra does not satisfy associative property. Motivated by this, in this paper, we define a new 3-algebra, the 3-ary multiplication is none completely antisymmetry, but it has close relation with 3-Lie algebras.
In the following we assume that Z is the set of integers, F is an algebraic closed field of characteristic zero, and for a subset S of a linear space V, ⟨S⟩ denotes the subspace spanned by S.
And we usually omit zero products of basis vectors when we list the multiplication of 3-algebras.
2. Natures of semi-associative 3-algebras
Definition 2.1**.**
A semi-associative 3-algebra (A,{,,}) is a vector space A with a 3-ary linear multiplication {,,}:A⊗A⊗A→A satisfying that for all
xi∈A,1≤i≤5,
[TABLE]
[TABLE]
[TABLE]
For subspaces B1,B2,B3 of A,
denotes {B1,B2,B3} the subspace spanned by vectors {x1,x2,x3}, ∀xi∈Bi, i=1,2,3; {A,A,A} is denoted by A1, which is called the derived algebra of A. If A1=0, then A is called abelian.
For example, let A be a 3-dimensional vector space with a basis v1,v2,v3. Then A is a semi-associative 3-algebra with the multiplication
[TABLE]
Definition 2.2**.**
A subalgebra of a semi-associative 3-algebra A is a subspace B satisfying {B,B,B}⊆B. If B satisfies that {A,A,B}⊆B and (A,B,A)⊆B, then B is called an ideal of A.
Obviously, {0} and A are ideals of A, which are called trivial ideals. If A has non proper ideals, then A is called a simple semi-associative 3-algebra.
For a given subspace V≤A, the subalgebra
[TABLE]
is called the centralizer of V in A. ZA(A) is called the center of A, and is simply denoted by Z(A), that is,
Z(A)={ x∣x∈A,{x,A,A}={A,A,x}=0 }. It is clear that Z(A) is an ideal.
Definition 2.3**.**
Let A and A1 be semi-associative 3-algebras. If a linear mapping (linear isomorphism) f:A→A1
satisfies
[TABLE]
then f is called an algebra homomorphism (an algebra isomorphism).
Let I be an ideal of A. Then quotient space A/I={x+I∣ x∈A} is a semi-associative 3-algebra in the multiplication
[TABLE]
which is called the quotient algebra of A by I, and
π:A→A/I, π(x)=x+I, ∀x∈A, is a surjective algebra homomorphism.
Proposition 2.4**.**
Let A be a semi-associative algebra, I1,I2,I3 be ideals of A. Then
1) I1+I2, I1∩I2 and {I1,I2,I3} are ideals of A.
2) If I1⊆I2, then I2/I1 is an ideal of A/I1.
Proof.
It can be verified by a direct computation according to Definition 2.2.
∎
Proposition 2.5**.**
Let A and A1 be semi-associative algebras, f:A→A1 be an algebra homomorphism. Then
1) K=Kerf={x∈A∣f(x)=0} is an ideal of A, and f(A) is a subalgebra of A1.
2) If f(A)=A1, then linear mapping fˉ:A/K→A1, fˉ(x+K)=f(x), for all x∈A, is an algebra isomorphism, and there is one to one correspondence between subalgebras of A containing K, with the subalgebras of A1, and ideals correspond to ideals.
Proof.
The result is easily verified by a direct computation.
∎
Theorem 2.6**.**
Let A be a semi-associative 3-algebra.
1) If there exist nonzero vectors e1,e2,e3,e4∈A such that {e1,e2,e3}=e4, then e4=λes, where λ∈F,λ=0, s=1,2,3.
2) If there exist nonzero vectors e1,e2,e3∈A such that {e1,e2,e3}=0, then {e1,e2,e3}=λe1+μe2,∀λ,μ∈F. Therefore, if
{e1,e2,e3}=ae1+be2+ce4=0, then c=0, and e1,e2,e4 are linearly independent.
Proof.
If there exist nonzero vectors e1,e2,e3,e4∈A such that {e1,e2,e3}=e4=λe1, λ∈F,λ=0, then by Eqs (1) and (2)
e4={e1,e2,e3}=λ1{e2,{e2,e1,e3},e3}=λ1{e2,e2,{e1,e3,e3}}=0.
Contradiction. Therefore, e4=λe1.
By the similar discussion to the above, e4=μe2, μ∈F.
If e4=αe3, α∈F,α=0, then
e4={e1,e2,e3}=α1{e1,e2,{e1,e2,e3}}=−α1{e1,e1,{e2,e2,e3}}=0.
Contradiction. The result 1) follows.
If there exist nonzero vectors e1,e2,e3∈A such that {e1,e2,e3}=λe1+μe2=0,λ,μ∈F.
Without loss of generality, we can suppose μ=0. Then
[TABLE]
Contradiction. Therefore, {e1,e2,e3}=λe1+μe2,∀λ,μ∈F.
The proof is complete.
∎
Theorem 2.7**.**
Let A be a non-abelian semi-associative 3-algebra with dimA=m≥3. Then there exist linearly independent vectors ei,ej,ek∈A such that
[TABLE]
Proof.
Since A is non-abelian, by (1), there are linearly independent vectors e1,e2∈A, such that {e1,e2,A}=0.
Suppose {e1,e2,⋯,em} is a basis of A. If {e1,e2,el}=0, for all l≥3, then there exist a,b∈F, such that
{e1,e2,ae1+be2}=0, we get {e1,e2,ae1+be2+e3}=0, and vectors e1,e2,ae1+be2+e3 are linearly independent. The proof is complete.
∎
Theorem 2.8**.**
Let A be an s-dimensional semi-associative 3-algebra with s≤6. Then A1⊆Z(A).
Proof.
If A is abelian, then the result is trivial.
If A1=0, then we will prove {A,A1,A}={A,A,A1}=0.
Since A1=0, then there are nonzero vectors ei,ej,ek,el∈A such that
[TABLE]
∙ {ei,ej,ek} are linearly dependent.
Then we can suppose ek=aei+bej, a,b∈F, and
el=a{ei,ej,ei}+b{ei,ej,ej}.
By (2) and (3), for all em,en∈A, we have
{em,el,en}=a{em,{ei,ej,ei},en}+b{em,{ei,ej,ej},en}
=−a{em,ej,{ei,ei,en}}+b{em,ei,{ej,ej,en}}=0,
{em,en,el}=a{em,en,{ei,ej,ei}}+b{em,en,{ei,ej,ej}}
=a{ei,ei,{en,ej,em}}+b{ej,ej,{en,ei,em}=0.
Therefore, {A,el,A}={A,A,el}=0.
∙∙ {ei,ej,ek} are linearly independent.
∙∙1 If there a,b,c∈F, such that el=aei+bej+cek, then
[TABLE]
By Theorem 2.6, ac=0, or bc=0. Without loss of generality, suppose ac=0, then
[TABLE]
Therefore, for all em,en∈A,
{em,el,en}={em,ac{ej,ek,ek},en}={em,acej,{ek,ek,en}}=0, and
{em,en,el}={em,en,ac{ej,ek,ek}}=ac{ek,{ek,en,ej},em}=0.
It follows {A,A,el}={A,el,A}=0.
∙∙2 {ei,ej,ek,el} are linearly independent.
Then in the case dimA=4, {ei,ej,ek,el} is a
basis of A. For all em,en∈A, set
em=a1ei+b1ej+c1ek+d1el, en=a2ei+b2ej+c2ek+d2el, ai,bi,ci,di∈F,i=1,2.
Then {em,el,en}={a1ei+b1ej+c1ek+d1el,el,en}
=a1{ei,ei,{ej,ek,en}}−b1{ej,ej,{ei,ek,en}}−c1{ek,ei,{ek,ej,en}}
=c1{ek,ek,{ei,ej,en}}=0.
By the similar to the above, {em,en,el}=0.
In the case dimA=5, then we can suppose that {ei,ej,ek,el,et} is a basis of A.
From the discussion of the case dimA=4, we only need to prove that {et,el,et}=0.
Thanks to (1), (2) and (3),
{et,el,et}={et,{ei,ej,ek},et}=−{et,et,{ej,ek,ei}}=0.
In the case dimA=6, then we can suppose e1,e2,e3,e4,e5,e6 is a basis of A, where e1=ei, e2=ej, e3=ek, and e4=el. Then
{e1,e2,e3}=e4.
By the above discussion,
[TABLE]
where B=⟨e1,e2,e3,e4,e5⟩, C=⟨e1,e2,e3,e4,e6⟩. Therefore, we only need to discuss the products
{e5,e4,e6},
{e6,e4,e5} and {e5,e6,e4}. Suppose
[TABLE]
Thanks to (4),
λ3{e5,e4,e6}={{e5,e4,e6}−a3e1−b3e2−c3e3−d3e4−μ3e6,e4,e6}={{e5,e4,e6},e4,e6}
={e4,{e4,e5,e6},e6}={e4,e4,{e5,e6,e6}}=0.
Therefore, λ3{e5,e4,e6}=λ3ae1+λ3b3e2+λ3c3e3+λ3d3e4+λ32e5+λ3μ−3λ3e6=0,
λ3=0.
By the complete similar discussion, we have a3=b3=c3=d3=μ3=0.
Therefore,
{e5,e4,e6}={e5,e6,e4} ={e6,e4,e5}=0, and
A1⊆Z(A).
∎
3. Derivations and centroid of semi-associative 3-algebras
3.1. Derivations of semi-associative 3-algebras
Definition 3.1**.**
Let A be a semi-associative 3-algebra, and D:A→A be a linear mapping. If D satisfies that
[TABLE]
then D is called a derivation of A. Der(A) denotes the set of all derivations of A.
Proposition 3.2**.**
Der(A)* is a subalgebra of the general linear algebra gl(A).*
Proof.
The result follows from a direct computation.
∎
For D∈Der(A), if D satisfies D(A)⊆Z(A) and D(A1)=0, then D is called a central derivation of A.
And DerC(A) denotes the set of central derivations. It is clear that DerC(A) is a subalgebra of Der(A).
For all x1,x2∈A, define linear mappings,
L(x1,x2), R(x1,x2), S(x1,x2):A×A⟶A, by
[TABLE]
[TABLE]
L(x1,x2) and R(x1,x2) are called the left multiplication and the right multiplication, respectively.
Lemma 3.3**.**
Let A be a semi-associative 3-algebra. Then for all x1,x2,x3,x4∈A,
[TABLE]
[TABLE]
[TABLE]
[TABLE]
[TABLE]
Proof.
Eqs (8), (9) and (10) follow from (1), (2) and (3), directly.
For ∀xi∈A, 1≤i≤5, by (2) and (3), we have
R(x3,x4)R(x1,x2)(x5)+R(x3,x4)R(x2,x1)(x5)
={{x5,x1,x2},x3,x4}+{{x5,x2,x1},x3,x4}
=−{x3,{x5,x1,x2},x4}−{x3,{x5,x2,x1},x4}
=−{x3,{x5,x1,x2},x4}+{x3,{x5,x1,x2},x4}=0,
Similarly, we have
L(x1,x2)L(x3,x4)(x5)−L(x3,x1)L(x2,x4)(x5)
={x1,x2,{x3,x4,x5}}−{x3,x1,{x2,x4,x5}}=0, and
L(x1,x2)L(x3,x4)(x5)−(L(x4,x2)R(x3,x1)(x5)+L(x3,x1)R(x4,x2)(x5))=0.
It follows (11) and (12).
∎
Let L(A), R(A) and S(A) be subspaces of End(A) spanned by linear mappings L(x1,x2), R(x,x2) and S(x1,x2), respectively, ∀x1,x2∈A, that is,
[TABLE]
[TABLE]
Then L(A)⊆T(A) and R(A)⊆T(A).
Theorem 3.4**.**
Let A be a semi-associative 3-algebra. Then T(A) is an abelian subalgebra of gl(A).
Proof.
By (1) and (2), for all xi∈A, 1≤i≤5,
[L(x1,x2),L(x3,x4)](x5)=(L(x1,x2)L(x3,x4)−L(x3,x4)L(x1,x2))(x5)
={x1,{x2,x3,x4},x5}−{x3,{x4,x1,x2},x5}
={x1,x3,{x4,x2,x5}}−{x3,{x4,x1,x2},x5}
=({x3,{x4,x1,x2},x5}−{x3,{x4,x1,x2},x5}=0.
Similarly, [L(x1,x2),R(x3,x4)]x5=0. Thanks to (3),
[R(x1,x2),R(x3,x4)](x5)=(R(x1,x2)R(x3,x4)−R(x3,x4)R(x1,x2))(x5)
={{x5,x3,x4},x1,x2}−{{x5,x1,x2},x3,x4}
=−{x1,{x5,x3,x4},x2}+{x3,{x5,x1,x2},x4}
=−{x1,{x5,x3,x4},x2}+{x5,{x1,x3,x2},x4}
=−{x2,{x5,x3,x4},x1}−{x1,{x2,x3,x4},x5}
+{x4,{x1,x3,x2},x5}
+{x5,{x4,x3,x2},x1}=0.
Therefore,
[L(A),T(A)]=[R(A),T(A)]=0,[T(A),T(A)]=0. The proof is complete.
∎
Theorem 3.5**.**
Let A be a semi-associative 3-algebra. Then
1) for any x1,x2∈A, S(x1,x2) is a derivation of A;
2) S(A) is an ideal of Der(A), and [S(A),S(A)]=0.
Proof.
For ∀x1,x2,x,y,z∈A, by (6) and (7),
S(x_{1},x_{2})\{x,y,z\}$$=\{x_{1},x_{2},\{x,y,z\}\}-\{\{x,y,z\},x_{1},x_{2}\}
={x1,{x2,x,y},z}+{x1,{x,y,z},x2},
{S(x1,x2)(x),y,z}+{x,S(x1,x2)(y),z}+{x,y,S(x1,x2)(z)}
={{x1,x2,x},y,z}−{{x,x1,x2},y,z}+{x,{x1,x2,y},z}
−{x,{y,x1,x2},z}+{x,y,{x1,x2,z}}−{x,y,{z,x1,x2}}
={x1,{x2,x,y},z}+{x1,{x,y,z},x2}.
Therefore,
S(x1,x2){x,y,z}={S(x1,x2)(x),y,z}+{x,S(x1,x2)(y),z}+{x,y,S(x1,x2)(z)}.
The result 1) follows.
For any S(x1,x2)∈S(A),D∈Der(A), and x∈A, since
[S(x1,x2),D](x)
=S(x1,x2)D(x)−DS(x1,x2)(x)
={x1,x2,D(x)}−{D(x),x1,x2}−D{x1,x2,x}+D{x,x1,x2}
={x1,x2,D(x)}−{D(x),x1,x2}−{D(x1),x2,x}−{x1,D(x2),x}
−{x1,x2,D(x)}+{D(x),x1,x2}+{x,D(x1),x2}+{x,x1,D(x2)}
=(S{x1,D(x2)}−S{D(x1),x2})(x).
Then [S(A),Der(A)]⊆S(A), that is, S(A) is an ideal of Der(A).
Thanks to Theorem 3.4, [S(A),S(A)]=[L(A)−R(A),L(A)−R(A)]=0.
It follows 2).
∎
For any x1,x2∈A, S(x1,x2) is called an inner derivation of A.
3.2. The centroid of semi-associative 3-algebras
Definition 3.6**.**
Let A be a semi-associative 3-algebra. The vector space
[TABLE]
is called the centroid of A.
For any φ∈End(A), by (1), φ∈Γ(A) if and only if for all x1,x2,x3∈A,
[TABLE]
Theorem 3.7**.**
Let A be a semi-associative 3-algebra. Then
1) Γ(A) is a subalgebra of the general linear Lie algebra gl(A).
2) For any φ∈Γ(A), if φ(A)⊆Z(A) and φ(A1)=0, then φ is a central derivation.
3) For any φ∈Γ(A), and D∈Der(A). Then φD∈Der(A).
4) DerC(A)=Γ(A)∩Der(A).
Proof.
For all φ1,φ2∈Γ(A), and x1,x2,x3∈A, by Definition 3.6,
[φ1,φ2]{x1,x2,x3}=(φ1φ2−φ2φ1){x1,x2,x3}
={φ1φ2(x1),x2,x3}−{φ2φ1(x1),x2,x3}
={[φ1,φ2](x1),x2,x3},
[φ1,φ2]{x1,x2,x3}=(φ1φ2−φ2φ1){x1,x2,x3}
={x1,x2,φ1φ2(x3)}−{x1,x2,φ2φ1(x3)}
={x1,x2,[φ1,φ2](x3)}.
Therefore, [φ1,φ2]∈gl(A), the result 1) follows.
For any φ∈Γ(A), if φ(A)⊆Z(A) and φ(A1)=0, then by Definition 3.6 and (5), φ∈DerA. It follows 2).
For any φ∈Γ(A), D∈DerA, and x1,x2,x3∈A,
φD{x1,x2,x3}=φ{D(x1),x2,x3}+φ{x1,D(x2),x3}+φ{x1,x2,D(x3)}
={φD(x1),x2,x3}+{x1,φD(x2),x3}+{x1,x2,φD(x3)},
φD∈DerA. It follows 3).
For any φ∈Γ(A)∩Der(A), by Definition 3.6, and (5),
∀x1,x2,x3∈A, we have
φ{x1,x2,x3}={φ(x1),x2,x3}+{x1,φ(x2),x3}+{x1,x2,φ(x3)}=3φ{x1,x2,x3}.
Therefore, φ(A1)=0. By φ{x1,x2,x3}={φ(x1),x2,x3}={x1,x2,φ(x3)}=0, we have φ(A)⊆Z(A), it follows Γ(A)∩Der(A)⊆DerC(A).
It is clear that if φ∈DerC(A), then φ∈Γ(A). This implies DerC(A)=Γ(A)∩Der(A).
∎
Theorem 3.8**.**
Let A be a semi-associative 3-algebra. Then ∀D∈Der(A),φ∈Γ(A)
1) [D,φ]⊆Γ(A),
2) Dφ∈Γ(A) if and only if φD∈DerC(A),
3) Dφ∈Der(A) if and only if [D,φ]∈DerC(A).
Proof For any D∈Der(A),φ∈Γ(A), and x1,x2,x3∈A, since
Dφ{x1,x2,x3}=D{φ(x1),x2,x3}
={Dφ(x1),x2,x3}+{φ(x1),D(x2),x3}+{φ(x1),x2,D(x3)}
={Dφ(x1),x2,x3}+φD{x1,x2,x3}−{φD(x1),x2,x3}, and
Dφ{x1,x2,x3}=D{x1,x2,φ(x3)}
={D(x1),x2,φ(x3)}+{x1,D(x2),φ(x3)}+{x1,x2,Dφ(x3)}
=φD{x1,x2,x3}−{x1,x2,φD(x3)}+{x1,x2,Dφ(x3)},
we have (Dφ−φD){x1,x2,x3}={(Dφ−φD)(x1),x2,x3}, and
(Dφ−φD){x1,x2,x3}={x1,x2,(Dφ−φD}(x3)}. Therefore, [D,φ]∈Γ(A).
Thanks to Theorem 3.7, φD∈DerA. If Dφ∈Γ(A), then by [D,φ]∈Γ(A), we have φD∈Γ(A). Therefore,
φD∈Der(A)∩Γ(A).
Conversely, if φD∈Γ(A), then by [D,φ]∈Γ(A), we have Dφ∈Γ(A).
We get 2). The result 3) follows from 1) and 2).
4. Sub-adjacent 3-Lie algebras and double modules of semi-associative 3-algebras
4.1. Sub-adjacent 3-Lie algebras of semi-associative 3-algebras
3-Lie algebras have close relationships with Lie algebras, pre-Lie algebras, associative algebras, commutative associative algebras and et al. Now we discuss
the relation between 3-Lie algebras with semi-associative 3-algebras.
A 3-Lie algebra (L,[,,]) is a vector space L with a linear multiplication [,,]:L∧L∧L→L which satisfies that for all xi∈L, 1≤i≤5,
[TABLE]
Theorem 4.1**.**
Let A be a semi-association 3-algebra. Then (A,[,,]c) is a 3-Lie algebra, where for all x1,x2,x3∈A,
[TABLE]
Proof.
By (1), (2) and (3), the multiplication [,,]C given by
(14) is skew-symmetric, and for all xi∈A,1≤i≤5,
[[x1,x2,x3]c,x4,x5]c−[[x1,x4,x5]c,x2,x3]c−[x1,[x2,x4,x5]c,x3]c
−[x1,x2,[x3,x4,x5]c]c
={{x1,x2,x3},x4,x5}+{{x2,x3,x1},x4,x5}+{{x3,x1,x2},x4,x5}
+{x4,x5,{x1,x2,x3}}
+{x4,x5,{x2,x3,x1}}+{x4,x5,{x3,x1,x2}}
+{x5,{x1,x2,x3},x4}
+{x5,{x2,x3,x1},x4}
+{x5,x3,x1,x2},x4}
−{{x1,x4,x5},x2,x3}−{{x4,x5,x1},x2,x3}
−{{x5,x1,x4},x2,x3}
−{x2,x3,{x1,x4,x5}}
−{x2,x3,{x4,x5,x1}}−{x2,x3,{x5,x1,x4}}
−{x3,{x1,x4,x5},x2}
−{x3,{x4,x5,x1},x2}
−{x3,{x5,x1,x4},x2}
−{x1,{x2,x4,x5},x3}
−{x1,{x4,x5,x2},x3}
−{x1,{x5,x2,x4},x3}
−{{x2,x4,x5},x3,x1}−{{x4,x5,x2},x3,x1}−{{x5,x2,x4},x3,x1}
−{x3,x1,{x2,x4,x5}}−{x3,x1,{x4,x5,x2}}−{x3,x1,{x5,x2,x4}}
−{x1,x2,{x3,x4,x5}}
−{x1,x2,{x4,x5,x3}}−{x1,x2,{x5,x3,x4}}
−{x2,{x3,x4,x5},x1}−{x2,{x4,x5,x3},x1}
−{x2,{x5,x3,x4},x1}
−{{x3,x4,x5},x1,x2}−{{x4,x5,x3},x1,x2}−{{x5,x3,x4},x1,x2}
={x4,x5,{x1,x2,x3}}+{x5,{x4,x2,x3},x1}+{x4,x5,{x2,x3,x1}}
+{x5,{x4,x3,x1},x2}
+{x4,x5,{x3,x1,x2}}+{x5,{x4,x1,x2},x3}
−{x2,x3,{x1,x4,x5}}−{x3,{x2,x4,x5},x1}
−{x2,x3,{x4,x5,x1}}
−{x3,{x2,x5,x1},x4}−{x2,x3,{x5,x1,x4}}−{x3,{x2,x1,x4},x5}
−{x3,x1,{x2,x4,x5}}−{x1,{x3,x4,x5},x2}−{x3,x1,{x4,x5,x2}}
−{x1,{x3,x5,x2},x4}
−{x3,x1,{x5,x2,x4}}−{x1,{x3,x2,x4},x5}
−{x1,x2,{x3,x4,x5}}−{x2,{x1,x4,x5},x3}
−{x1,x2,{x4,x5,x3}}
−{x2,{x1,x5,x3},x4}−{x1,x2,{x5,x3,x4}}−{x2,{x1,x3,x4},x5}
=0.
Therefore, the multiplication [ , , ]c satisfies (13), and Ac is a 3-Lie algebra. The proof is complete.
∎
The 3-Lie algebra (A,[,,]c) is called the sub-adjacent 3-Lie algebra of the semi-association 3-algebra A, and is simply denoted by Ac.
Der(Ac) denotes the derivation algebras of the sub-adjacent 3-Lie algebra Ac, and
Theorem 4.2**.**
Let A be a semi-associative 3-algebra. Then
1) derivation algebra Der(A) is a subalgebra of Der(Ac);
2) for all xi∈A, 1≤i≤4, and S(x1,x2), S(x3,x4)∈Der(A),
[TABLE]
3) if I is an ideal ( a subalgebra ) of semi-associative algebra A, then I is also an ideal ( a subalgebra ) of the 3-Lie algebra Ac.
Proof.
For any D∈Der(A), and x,y,z∈A,
D[x,y,z]c=D({x,y,z}+{y,z,x}+{z,x,y})
={Dx,y,z}+{x,Dy,z}+{x,y,Dz}
+{Dy,z,x}+{y,Dz,x}+{y,z,Dx}
+{Dz,x,y}+{z,Dx,y}
+{z,x,Dy}
=[Dx,y,z]C+[x,Dy,z]C+[x,y,Dz]c,
therefore, D∈Der(Ac). It follows 1).
By the similar discussion to the above, we get 2) and 3).
∎
Theorem 4.3**.**
Let A be a semi-associative 3-algebra. Then L(A),R(A)⊆ Der(Ac).
Proof.
By (1)-(3), (6) and (14), for any x,y,u,v,w∈A,
[TABLE]
[TABLE]
Therefore, L(x,y)[u,v,w]c=[L(x,y)(u),v,w]c+[u,L(x,y)(v),w]c+[u,v,L(x,y)(w)]c.
By the similar discussion to the above, we have
[TABLE]
The proof is complete.
∎
4.2. Double modules of the semi-associative 3-algebras
In this section, we discuss double modules of semi-associative 3-algebras.
A representation (or a module) (V,ρ) of a 3-Lie algebra (G,[,,]) is a vector space V and a linear transformation ρ: G∧G⟶End(V),
satisfying, ∀x1,x2,x3,x4∈G,
[TABLE]
[TABLE]
We know that (V,ρ) is a 3-Lie algebra (G,[,,])-module if and only if G+˙V is a 3-Lie algebra in the multiplication [,,]ρ,
where ∀x1,x2,x3∈A and v1,v2,v3∈V,
[TABLE]
Definition 4.4**.**
Let A be a semi-associative 3-algebra, V be a vector space, and ϕ,ψ: A×A⟶End(V) be linear mappings. If ϕ and ψ satisfy the following properties, ∀x1,x2,x3,x4∈A,
[TABLE]
[TABLE]
[TABLE]
[TABLE]
[TABLE]
[TABLE]
then (ϕ,ψ,V) is called a double representation of A, or is simply called a double module of A.
Theorem 4.5**.**
Let A be a semi-associative 3-algebra, and l,r: A×A⟶End(V) be linear mappings.
Then triple (ϕ,ψ,V) is a double module of A if and only if (A+˙V,{,,}ϕψ) is a
semi-associative 3-algebra, where ∀xi∈A,vi∈V,i=1,2,3,
[TABLE]
Proof.
If (ϕ,ψ,V) is a double mode of the semi-associative algebra A, we will prove that (A+˙V,{,,}ϕψ) is a semi-associative 3-algebra.
Thanks to (24), for all xi∈A,vi∈V,1≤i≤5,
{x1+v1,x2+v2,x3+v3}ϕψ=−{x2+v2,x1+v1,x3+v3}ϕψ, (1) holds.
By (24), (19), (20) and (23),
{x1+v1,{x2+v2,x3+v3,x4+v4}ϕψ,x5+v5}ϕψ
={x1,x2,{x3,x4,x5})+ψ(x2,{x3,x4,x5})v1−ψ(x1,{x3,x4,x5})v2
+ϕ(x1,x2)ψ(x4,x5)v3−ϕ(x1,x2)ψ(x3,x5)v4+ϕ(x1,x2)ϕ(x3,x4)v5
={x1+v1,x2+v2,{x3+v3,x4+v4,x5+v5}ϕψ}ϕψ,
(2) holds.
Thanks to (24), (20)- (22),
{x1+v1,{x2+v2,x3+v3,x4+v4}ϕψ,x5+v5}ϕψ
={x5,{x2,x3,x4},x1}+{x1,{x5,x3,x4},x2}+(ϕ(x5,{x2,x3,x4})
+ψ({x5,x3,x4},x2))v1+(ϕ(x1,{x5,x3,x4})−ψ(x5,x1)ψ(x3,x4))v2
+(ψ(x5,x1)ψ(x2,x4)+ψ(x1,x2)ψ(x5,x4))v3−(ψ(x5,x1)ϕ(x2,x3)
+ψ(x1,x2)ϕ(x5,x3))v4+(ψ({x2,x3,x4},x1)−ψ(x1,x2)ψ(x3,x4))v5
={x5+v5,{x2+v2,x3+v3,x4+v4}ϕψ,x1+v1}}ϕψ
+{x1+v1,{x5+v5,x3+v3,x4+v4}ϕψ,x2+v2}ϕψ,
(3) holds.
Therefore, (A+˙V,{,,}ϕψ) is a semi-associative 3-algebra.
Conversely, if (A+˙V,{,,}ϕψ) is a semi-associative algebra. Then by (1), ∀xi∈A,vi∈V, 1≤i≤3,
[TABLE]
Thanks to (24),
{x1,x2,x3+v3}ϕψ={x1,x2,x3}+ϕ(x1,x2)v3,
{x2,x1,x3+v3}={x2,x1,x3}+ϕ(x2,x1)v3,
therefore, ϕ(x1,x2)=−ϕ(x2,x1), (18) holds.
By (2), and (24),
{x1+v1,{x2+v2,x3+v3,x4+v4}ϕψ,x5+v5}ϕψ
={x1+v1,x2+v2,{x3+v3,x4+v4,x5+v5}ϕψ}ϕψ,
{x1+v1,{x2+v2,x3+v3,x4+v4}ϕψ,x5+v5}ϕψ
={x1,{x2,x3,x4},x5}+ψ({x2,x3,x4},x5)v1−ψ(x1,x5)ψ(x3,x4)v2
+ψ(x1,x5)ψ(x2,x4)v3−ψ(x1,x5)ϕ(x2,x3)v4+ϕ(x1,{x2,x3,x4})v5,
{x1+v1,x2+v2,{x3+v3,x4+v4,x5+v5}ϕψ}ϕψ
={x1,x2,{x3,x4,x5}}+ψ(x2,{x3,x4,x5})v1−ψ(x1,{x3,x4,x5})v2
+ϕ(x1,x2)ψ(x4,x5)v3−ϕ(x1,x2)ψ(x3,x5)v4+ϕ(x1,x2)ϕ(x3,x4)v5.
If we suppose vi=0,vj=0, for 1≤i=j≤5, then we get
[TABLE]
[TABLE]
[TABLE]
[TABLE]
[TABLE]
Thanks to (3), and (24),
{x5+v5,{x2+v2,x3+v3,x4+v4}ϕψ,x1+v1}ϕψ+{x1+v1,{x5+v5,x3+v3,x4+v4}ϕψ,x2+v2}ϕψ
=\{x_{5},\{x_{2},x_{3},x_{4}\},x_{1}\}+\{x_{1},\{x_{5},x_{3},x_{4}\},x_{2}\}+(\phi(x_{5},\{x_{2},x_{3},x_{4}\})$$+\psi(\{x_{5},x_{3},x_{4}\},x_{2}))v_{1}
+(\phi(x_{1},\{x_{5},x_{3},x_{4}\})-\psi(x_{5},x_{1})\psi(x_{3},x_{4}))v_{2}$$+(\psi(x_{5},x_{1})\psi(x_{2},x_{4})+\psi(x_{1},x_{2})\psi(x_{5},x_{4}))v_{3}
-(\psi(x_{5},x_{1})\phi(x_{2},x_{3})$$+\psi(x_{1},x_{2})\phi(x_{5},x_{3}))v_{4}+(\psi(\{x_{2},x_{3},x_{4}\},x_{1})-\psi(x_{1},x_{2})\psi(x_{3},x_{4}))v_{5},
similarly, for vi=0,vj=0, for 1≤i=j≤5, we have
[TABLE]
[TABLE]
[TABLE]
[TABLE]
[TABLE]
Now we analyze the above identities. (19) follows from (26); (22) follows from (33); and it is clear that (30) is equivalent to (28), (29) and (31); and (34) is equivalent to (25), (27) and (33); (20) follows from (25) and (26); (21) follows from (28), (29) and (31); and (23) follows from (28), (29), (31) and (32).
Therefore, (l,r,V) is a double mode of the semi-associative algebra A.
The proof is complete.
∎
Let τ: V⊗V→V⊗V be the exchange mapping, that is,
[TABLE]
Theorem 4.6**.**
Let A be a semi-associative 3-algebra, and (l,r,V) be the double module of A. Then (V,ρ) is a 3-Lie algebra Ac-module, where
[TABLE]
Proof.
Thanks to Theorem 4.5, A+˙V is a semi-associative 3-algebra in the multiplication (24).
Therefore, the multiplication of the subjacent 3-Lie algebra (A+˙V)c of A+˙V is (14), that is, ∀x1,x2,x3∈A, v1,v2,v3∈V,
[x1+v1,x2+v2,x3+v3]c
={x1+v1,x2+v2,x3+v3}ϕψ+{x2+v2,x3+v3,x1+v1}ϕψ+{x3+v3,x1+v1,x2+v2}ϕψ
={x1,x2,x3}+ϕ(x1,x2)v3−ψ(x1,x3)v2+ψ(x2,x3)v1+{x2,x3,x1}+ψ(x2,x3)v1
−ψ(x2,x1)v3+ψ(x3,x1)v2+{x3,x1,x2}+ϕ(x3,x1)v2−ψ(x3,x2)v1+ψ(x1,x2)v3
=[x1,x2,x3]c+(ϕ−ψτ+ψ)(x1,x2)v3+(ϕ−ψτ+ψ)(x2,x3)v1+(ϕ−ψτ+ψ)(x3,x1)v2.
Follows from (17), (V,ρ) is a 3-Lie algebra Ac, where ρ=ϕ−ψτ+ψ.
∎
Theorem 4.7**.**
Let A be a semi-associative 3-algebra, and L(x,y),R(x,y):A×A→A be left and right multiplications defined as (6), ∀x,y∈A. Then (L,R,A) is a double module of A.
Proof.
Apply Lemma 12, Theorem 4.5 and Definition 4.4.∎
The double module (L,R,A) of the semi-associative 3-algebra A is called the regular representation of A, or the adjoint module of A.
Theorem 4.8**.**
Let A be a semi-associative 3-algebra, and (l,r,V) be a double module of A.
Then (ϕ∗,ψ∗,V∗) is also a double module of A,
where V∗ is the dual space of V, and ϕ∗,ψ∗:A×A⟶End(V∗) are defined by
∀x,y∈A,v∈V,ξ∈V∗,
[TABLE]
Proof.
Apply Lemma 12, Definition 4.4 and Theorem 4.5.∎
Corollary 4.9**.**
Let A be a semi-associative 3-algebra. Then (L∗,R∗,A∗) is a double module of A.
Proof.
The result follows from Theorem 4.7 and Theorem 35, directly.
∎
4.3. Double extensions of semi-associative 3-algebras by cocycles
Definition 4.10**.**
Let A be a semi-associative 3-algebra, if linear mapping θ:A⊗A⊗A→A∗ satisfies that for all x,y,z,w,u∈A,
[TABLE]
[TABLE]
[TABLE]
[TABLE]
=θ{x,{u,z,w},y}+R∗(x,u)θ{y,z,w}**
−R∗(u,x)θ{y,z,w}−R∗(x,y)θ{u,z,w}*,
then θ is called a cocycle of the semi-associative 3-algebra A, where L∗,R∗ are defined by (35).*
Theorem 4.11**.**
Let A be a semi-associative 3-algebra, and θ:A⊗A⊗A→A∗ be a cocycle of A. Then (A+˙A∗,{,,}θ) is a semi-associative 3-algebra, where ∀xi∈A,ξi∈A∗,1≤i≤3,
[TABLE]
−R∗(x1,x3)ξ2+R∗(x2,x3)ξ1.**
Proof.
Thanks to (18) - (23), ∀xi∈A,ξi∈A∗,1≤i≤3,
\{x_{1}+\xi_{1},x_{2}+\xi_{2},x_{3}+\xi_{3}\}_{\theta}$$=-\{x_{2}+\xi_{2},x_{1}+\xi_{1},x_{3}+\xi_{3}\}_{\theta}, (1) holds.
By Definition 4.10,
{x1+ξ1,{x2+h2,x3+h3,x4+ξ4}θ,x5+ξ5}θ
={x1,x2,{x3,x4,x5}}+θ{x1,x2,{x3,x4,x5}}+L∗(x1,x2)L∗(x3,x4)ξ5
+L∗(x1,x2)θ{x3,x4,x5}−L∗(x1,x2)R∗(x3,x5)ξ4+L∗(x1,x2)R∗(x4,x5)ξ3
−R∗(x1,{x3,x4,x5})ξ2+R∗(x2,{x3,x4,x5})ξ1
={x1+ξ1,x2+ξ2,{x3+ξ3,x4+ξ4,x5+ξ5}θ}θ, (2) holds.
Thanks to Theorem 35,
{x1,{x2,x3,x4},x5}+θ{x1,{x2,x3,x4},x5}+L∗(x1,{x2,x3,x4})ξ5
−R∗(x1,x5)θ{x2,x3,x4}−R∗(x1,x5)L∗(x2,x3)ξ4+R∗(x1,x5)R∗(x2,x4)ξ3
−R∗(x1,x5)R∗(x3,x4)ξ2+R∗({x2,x3,x4},x5)ξ1
={x5+ξ5,{x2+ξ2,x3+ξ3,x4+ξ4}θ,x1+ξ1}θ+{x1+ξ1,{x5+ξ5,x3+ξ3,x4+ξ4}θ,x2+ξ2}θ,
we get
{x1+ξ1,{x2+ξ2,x3+ξ3,x4+ξ4}θ,x5+ξ5}θ
={x5+ξ5,{x2+ξ2,x3+ξ3,x4+ξ4}θ,x1+ξ1}θ+{x1+ξ1,{x5+ξ5,x3+ξ3,x4+ξ4}θ,x2+ξ2}θ, that is, (3) holds.
Therefore, (A+˙A∗,{,,}θ) is a semi-associative 3-algebra.
∎
The semi-associative 3-algebra (A+˙A∗,{,,}θ) is called a double extension of 3-semi-associative algebra by θ.