Introduction
Schrödinger symmetry in physics. The maximal group of kinematic symmetries of the free Schrödinger equation 2 i ∂ t ψ = − Δ ψ 2i\partial_t\psi=-\Delta\psi 2 i ∂ t ψ = − Δ ψ was found by Niederer [28 ] and Hagen [11 ] . Its Lie algebra, the Schrödinger algebra, contains the Galilei algebra, the dilatations t ↦ λ 2 t t\mapsto\lambda^{2}t t ↦ λ 2 t , x ↦ λ x x\mapsto\lambda x x ↦ λ x (with dynamical exponent z = 2 z=2 z = 2 ) and the special Schrödinger transformations t ↦ t / ( 1 + κ t ) t\mapsto t/(1+\kappa t) t ↦ t / ( 1 + κ t ) . In order to describe strongly anisotropic critical systems, Henkel [13 ] embedded the Schrödinger algebra (in one space dimension) into an infinite-dimensional Lie algebra, the Schrödinger–Virasoro algebra, which is the main object of the present paper (we call it the Schrödinger–Witt algebra, see below). In this algebra, the generators L m L_m L m realize the time reparametrizations t ↦ f ( t ) t\mapsto f(t) t ↦ f ( t ) , i.e., a copy of the Witt algebra, while Y n Y_n Y n generate time-dependent space translations (including the space translations and the Galilei boosts) and M m M_m M m generate time-dependent phase shifts of the wave function (M 0 M_0 M 0 corresponds to the mass, which is central). The local scale invariance generated by this algebra has been applied to dynamical scaling and ageing phenomena in systems far from equilibrium [14 , 15 , 16 ] . Non-relativistic conformal field theories with Schrödinger symmetry also describe cold atoms at unitarity [29 ] , and their holographic duals [5 , 33 ] have made Schrödinger symmetry a part of the non-relativistic AdS/CFT correspondence. A systematic mathematical study of the Schrödinger–Virasoro group and algebra, including their representation theory and cohomology, was carried out by Roger and Unterberger [30 ] ; see also the monograph [36 ] . Studying vertex algebra representations of the Schrödinger–Virasoro algebra, Unterberger [35 ] introduced the extended Schrödinger–Virasoro algebra, which contains one more family of generators N m N_m N m (denoted by e s v 1 / 2 \mathfrak{esv}_{1/2} esv 1/2 in this paper).
Algebras with the same multiplication table. Lie algebras with the multiplication table (6) (or (34)) have been studied from many algebraic points of view: representations [25 , 36 , 38 ] , automorphisms and Verma modules [34 ] , derivations and cohomology [30 ] , structure, derivations and automorphisms of the extended algebra [10 ] , biderivations and commutative post-Lie algebra structures [26 ] , 1 2 \frac{1}{2} 2 1 -derivations and transposed Poisson structures [32 , 37 ] . Since these algebras are centerless and the Virasoro algebra is the central extension of the Witt algebra, we follow [38 ] and use the name Schrödinger–Witt algebras; the name Schrödinger–Virasoro is used for the same algebras in [10 , 13 , 25 , 26 , 30 , 32 , 34 , 35 , 36 , 37 ] .
Generalized derivations. Derivations and their generalizations are among the most important tools for studying algebraic structures. Among the many generalizations of derivations we mention δ \delta δ -derivations, introduced by Filippov [9 ] and studied for many classes of algebras and superalgebras [17 , 18 , 39 , 40 ] , and generalized derivations, quasi-derivations and the quasirectroid, introduced and systematically studied by Leger and Luks [24 ] . Recall that a linear map D D D of a Lie algebra L L L is a quasi-derivation if there is a linear map D ′ D' D ′ of L L L such that
[ D ( x ) , y ] + [ x , D ( y ) ] = D ′ ( [ x , y ] ) for all x , y ∈ L . [D(x),y]+[x,D(y)]=D'([x,y]) \quad\text{for all } x,y\in L. [ D ( x ) , y ] + [ x , D ( y )] = D ′ ([ x , y ]) for all x , y ∈ L .
Every derivation is a quasi-derivation (with D ′ = D D'=D D ′ = D ) and every δ \delta δ -derivation with δ ≠ 0 \delta\ne0 δ = 0 is a quasi-derivation (with D ′ = δ − 1 D D'=\delta^{-1}D D ′ = δ − 1 D ). The 1 2 \frac{1}{2} 2 1 -derivations have attracted a lot of attention in recent years, since Ferreira, Kaygorodov and Lopatin established a relation between 1 2 \frac{1}{2} 2 1 -derivations of Lie algebras and transposed Poisson algebras [8 ] . The notion of transposed Poisson algebras was introduced by Bai, Bai, Guo and Wu [4 ] as a dual notion of Poisson algebras; transposed Poisson algebras are closely related to Novikov–Poisson algebras and 3-Lie algebras [4 ] , and Novikov algebras originate from Poisson brackets of hydrodynamic type [6 ] . Their seminal paper attracted considerable interest, and since then transposed Poisson algebras have been studied in various directions (see [1 , 2 , 7 , 19 , 23 , 27 , 31 ] and references therein). In particular, transposed Poisson structures have been described on many important Lie algebras: the Witt and Virasoro algebras [8 ] ; Galilean algebras [23 ] ; twisted Heisenberg–Virasoro, Schrödinger–Witt, and extended Schrödinger–Witt algebras [37 ] ; the original deformative Schrödinger–Witt algebras and the central extension of the extended Schrödinger–Witt algebras [32 ] ; and not-finitely graded Witt-type algebras [20 ] . Biderivations and the closely related commuting maps of Lie algebras were studied in [12 , 26 ] , and cohomological methods for Virasoro-type structures were developed, for example, in [3 , 30 ] . Some of the cited results were generalized to quasi-derivations and to the description of transposed δ \delta δ -Poisson structures: for the Witt algebra and the W ( a , b ) \mathcal{W}(a,b) W ( a , b ) algebras [21 ] , and for the Block Lie algebras B ( q ) \mathcal{B}(q) B ( q ) [22 ] . The present paper extends the known descriptions of 1 2 \frac{1}{2} 2 1 -derivations and transposed Poisson structures of the (extended) Schrödinger–Witt algebras to descriptions of their quasi-derivations, δ \delta δ -(bi)derivations and transposed δ \delta δ -Poisson structures.
Main results. The derivations of both the original and twisted Schrödinger–Witt algebras were determined in [30 ] . Further investigations in [26 ] characterized their biderivations and commutative post-Lie algebra structures. Moreover, Yuan and Hua [37 ] established that every 1 2 \frac{1}{2} 2 1 -derivation of these algebras is trivial, namely, a scalar multiple of the identity map. In Subsection 2.2 we describe all quasi-derivations of the Schrödinger–Witt algebras s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) (Theorem 19): every quasi-derivation is the sum of a derivation, a multiple of the identity map, a linear map with values in the center and a finite linear combination of maps from three new families Ω t \Omega_t Ω t , Ψ t \Psi_t Ψ t (t ≠ 0 t \ne0 t = 0 ) and Θ s \Theta_s Θ s of quasi-derivations, which are neither derivations nor δ \delta δ -derivations. As corollaries, we describe all δ \delta δ -(bi)derivations and all transposed δ \delta δ -Poisson structures on s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) for every δ ∈ C \delta\in\mathbb{C} δ ∈ C (Corollaries 21, 23 and 24); in particular, we obtain new proofs of the descriptions of derivations [30 ] and 1 2 \frac{1}{2} 2 1 -derivations [37 ] of s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) . In the last section we consider the extended Schrödinger–Witt algebras e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) . Here the situation is much more rigid: every quasi-derivation of e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) is of the form a x + λ Id a_x+\lambda\operatorname{Id} a x + λ Id (Theorem 30). Consequently, all derivations of e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) are inner (Corollary 31; for the original extended Schrödinger–Witt algebra e s v ( 1 2 ) \mathfrak{esv}\left(\frac{1}{2}\right) esv ( 2 1 ) this was proved earlier by Gao, Jiang and Pei [10 ] ), its δ \delta δ -derivations are described for all δ ∈ C \delta\in\mathbb{C} δ ∈ C (Corollary 32), every biderivation of e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) is a multiple of the Lie bracket and every δ \delta δ -biderivation with δ ≠ 1 \delta\ne1 δ = 1 is zero (Corollary 34), and all transposed δ \delta δ -Poisson structures on e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) are trivial for every δ ∈ C \delta\in\mathbb{C} δ ∈ C (Corollary 35). In particular, all transposed Poisson structures on e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) are trivial; for the original extended Schrödinger–Witt algebra e s v ( 1 2 ) \mathfrak{esv}\left(\frac{1}{2}\right) esv ( 2 1 ) these results were obtained earlier in [37 ] , and our approach gives a new proof of them.
Perspectives. From the point of view of mathematical physics, these results describe the “generalized infinitesimal symmetries” of the Schrödinger–Witt symmetry algebras and show their rigidity: neither s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) nor e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) admits a nontrivial compatible commutative associative product of transposed Poisson type (for s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) and δ = 1 \delta=1 δ = 1 the only such products are supported on the abelian ideal ⟨ M m ∣ m ∈ Z ⟩ \langle M_m \mid m \in\mathbb{Z}\rangle ⟨ M m ∣ m ∈ Z ⟩ of phase shifts), and the extended algebra admits no outer derivations at all, so that its symmetry structure cannot be deformed or enlarged within these classes of generalized derivations. In contrast, the quasi-derivations Ω t , Ψ t , Θ s \Omega_t,\Psi_t,\Theta_s Ω t , Ψ t , Θ s show that s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) carries generalized symmetries that do not come from derivations of its Witt subalgebra.
Preliminaries
All algebras and vector spaces are considered over the complex field C \mathbb{C} C . The notation ⟨ S ⟩ \langle S\rangle ⟨ S ⟩ means the C \mathbb{C} C -subspace generated by S S S . For a Lie algebra L L L and x ∈ L x\in L x ∈ L we denote by ad x \operatorname{ad}_{x} ad x the inner derivation ad x ( y ) = [ x , y ] \operatorname{ad}_{x}(y)=[x,y] ad x ( y ) = [ x , y ] . The center of L L L is denoted by Z ( L ) Z(L) Z ( L ) .
Definition 1 (see [24 ] ). Let ( L , [ − , − ] ) (L,[-,-]) ( L , [ − , − ]) be a Lie algebra and D : L → L D:L\to L D : L → L be a linear map. Then D D D is a quasi-derivation if there is a linear map D ′ : L → L D':L\to L D ′ : L → L such that
[ D ( x ) , y ] + [ x , D ( y ) ] = D ′ ( [ x , y ] ) for all x , y ∈ L . (1) [D(x),y]+[x,D(y)]=D'([x,y])\quad\text{for all }x,y\in L. \tag*{(1)} [ D ( x ) , y ] + [ x , D ( y )] = D ′ ([ x , y ]) for all x , y ∈ L . (1) The space of all quasi-derivations of L L L is denoted by QDer ( L ) \operatorname{QDer}(L) QDer ( L ) . If D D D and D ′ D' D ′ satisfy (1), we say that D ′ D' D ′ is associated with D D D .
Definition 2 (see [9 ] ). Let ( L , [ − , − ] ) (L,[-,-]) ( L , [ − , − ]) be a Lie algebra, δ ∈ C \delta\in\mathbb{C} δ ∈ C and φ : L → L \varphi:L\to L φ : L → L be a linear map. Then φ \varphi φ is a δ \delta δ -derivation if
φ ( [ x , y ] ) = δ ( [ φ ( x ) , y ] + [ x , φ ( y ) ] ) for all x , y ∈ L . (2) \varphi([x,y])=\delta\bigl([\varphi(x),y]+[x,\varphi(y)]\bigr)\quad\text{for all }x,y\in L. \tag*{(2)} φ ([ x , y ]) = δ ( [ φ ( x ) , y ] + [ x , φ ( y )] ) for all x , y ∈ L . (2) The space of δ \delta δ -derivations of L L L is denoted by Δ δ ( L ) \Delta_{\delta}(L) Δ δ ( L ) . In particular, Δ 1 ( L ) = Der L \Delta_{1}(L)=\operatorname{Der}L Δ 1 ( L ) = Der L .
If δ = 1 2 \delta=\frac{1}{2} δ = 2 1 , then φ \varphi φ is called a 1 2 \frac{1}{2} 2 1 -derivation. Multiplications by elements of the ground field are called trivial 1 2 \frac{1}{2} 2 1 -derivations.
Definition 3 (see [37 ] ). Let ( L , [ − , − ] ) (L,[-,-]) ( L , [ − , − ]) be a Lie algebra and δ ∈ C \delta\in\mathbb{C} δ ∈ C . A bilinear map ϕ : L × L → L \phi:L\times L\to L ϕ : L × L → L is a δ \delta δ -biderivation if
ϕ ( [ x , y ] , z ) = δ ( [ ϕ ( x , z ) , y ] + [ x , ϕ ( y , z ) ] ) , (3) \phi([x,y],z)=\delta\bigl([\phi(x,z),y]+[x,\phi(y,z)]\bigr), \tag*{(3)} ϕ ([ x , y ] , z ) = δ ( [ ϕ ( x , z ) , y ] + [ x , ϕ ( y , z )] ) , (3) ϕ ( x , [ y , z ] ) = δ ( [ ϕ ( x , y ) , z ] + [ y , ϕ ( x , z ) ] ) (4) \phi(x,[y,z])=\delta\bigl([\phi(x,y),z]+[y,\phi(x,z)]\bigr) \tag*{(4)} ϕ ( x , [ y , z ]) = δ ( [ ϕ ( x , y ) , z ] + [ y , ϕ ( x , z )] ) (4) for all x , y , z ∈ L x,y,z\in L x , y , z ∈ L . The 1 1 1 -biderivations are called biderivations.
In other words, ϕ \phi ϕ is a δ \delta δ -biderivation if and only if for every z ∈ L z\in L z ∈ L the maps x ↦ ϕ ( x , z ) x\mapsto\phi(x,z) x ↦ ϕ ( x , z ) and x ↦ ϕ ( z , x ) x\mapsto\phi(z,x) x ↦ ϕ ( z , x ) are δ \delta δ -derivations. For every λ ∈ C \lambda\in\mathbb{C} λ ∈ C the bilinear map ϕ ( x , y ) = λ [ x , y ] \phi(x,y)=\lambda[x,y] ϕ ( x , y ) = λ [ x , y ] is a biderivation, by the Jacobi identity. Biderivations and the closely related commuting maps have been studied extensively, for associative rings as well as for Lie algebras (see, for example, [12 ] ).
Definition 4. Let ( L , [ − , − ] ) (L,[-,-]) ( L , [ − , − ]) be a Lie algebra and δ ∈ C \delta\in\mathbb{C} δ ∈ C . A transposed δ \delta δ -Poisson structure on L L L is a commutative associative multiplication ⋅ \cdot ⋅ on L L L such that
z ⋅ [ x , y ] = δ ( [ z ⋅ x , y ] + [ x , z ⋅ y ] ) for all x , y , z ∈ L . (5) z\cdot[x,y]=\delta\bigl([z\cdot x,y]+[x,z\cdot y]\bigr)\quad\text{for all }x,y,z\in L. \tag*{(5)} z ⋅ [ x , y ] = δ ( [ z ⋅ x , y ] + [ x , z ⋅ y ] ) for all x , y , z ∈ L . (5) The transposed 1 2 \frac{1}{2} 2 1 -Poisson structures, i.e., the commutative associative multiplications ⋅ \cdot ⋅ on L L L such that $2z\cdot[x,y]=[z\cdot x,y]+[x,z\cdot y]$ for all x , y , z ∈ L x,y,z\in L x , y , z ∈ L , are called transposed Poisson structures [4 , 8 ] . A transposed δ \delta δ -Poisson structure is trivial if x ⋅ y = 0 x\cdot y=0 x ⋅ y = 0 for all $x,y\in L.
In other words, (5) means that all multiplication operators x ↦ z ⋅ x x\mapsto z\cdot x x ↦ z ⋅ x are δ \delta δ -derivations of the Lie algebra ( L , [ − , − ] ) (L,[-,-]) ( L , [ − , − ]) .
Lemma 5 (see [37 ] ). Let ⋅ \cdot ⋅ be a transposed δ \delta δ -Poisson structure on a Lie algebra L L L . Then the bilinear map ϕ ( x , y ) = x ⋅ y \phi(x,y)=x\cdot y ϕ ( x , y ) = x ⋅ y is a δ \delta δ -biderivation of L L L . In particular, if ⋅ \cdot ⋅ is a transposed Poisson structure on L L L , then ϕ \phi ϕ is a 1 2 \frac{1}{2} 2 1 -biderivation of $L.
Lemma 6. Let L L L be a perfect Lie algebra, i.e., [ L , L ] = L [L,L]=L [ L , L ] = L . Then for every D ∈ QDer ( L ) D\in\operatorname{QDer}(L) D ∈ QDer ( L ) the map D ′ D' D ′ associated with D D D is unique. In particular, the map D ↦ D ′ D\mapsto D' D ↦ D ′ is linear, D ′ = 0 D'=0 D ′ = 0 whenever D D D is a central map (i.e., a linear map with values in Z ( L ) Z(L) Z ( L ) ), and D ∈ QDer ( L ) D\in\operatorname{QDer}(L) D ∈ QDer ( L ) is a derivation if and only if D ′ = D D'=D D ′ = D .
Proof. If D 1 ′ D'_1 D 1 ′ and D 2 ′ D'_2 D 2 ′ are associated with D D D , then ( D 1 ′ − D 2 ′ ) ( [ x , y ] ) = 0 (D'_1-D'_2)([x,y])=0 ( D 1 ′ − D 2 ′ ) ([ x , y ]) = 0 for all x , y ∈ L x,y\in L x , y ∈ L , and hence D 1 ′ = D 2 ′ D'_1=D'_2 D 1 ′ = D 2 ′ on [ L , L ] = L [L,L]=L [ L , L ] = L . The remaining statements are straightforward; note that every central map D D D is a quasi-derivation with associated map 0, since [ D ( x ) , y ] + [ x , D ( y ) ] = 0 [D(x),y]+[x,D(y)]=0 [ D ( x ) , y ] + [ x , D ( y )] = 0 .
Let G G G be an abelian group and L = ⨁ g ∈ G L g L=\bigoplus_{g\in G}L_g L = ⨁ g ∈ G L g be a G G G -graded Lie algebra. For v ∈ L v\in L v ∈ L and k ∈ G k\in G k ∈ G we denote by v k v_k v k the homogeneous component of degree k k k of v v v . For a linear map D : L → L D:L\to L D : L → L and g ∈ G g\in G g ∈ G , we define the linear map D g D_g D g by
D g ( x ) = ( D ( x ) ) h + g for x ∈ L h . D_g(x)=(D(x))_{h+g}\quad\text{for }x\in L_h. D g ( x ) = ( D ( x ) ) h + g for x ∈ L h .
We call D g D_g D g the homogeneous component of degree g g g of $D . Clearly, D g ( L h ) ⊆ L h + g D_g(L_h)\subseteq L_{h+g} D g ( L h ) ⊆ L h + g and, for every x ∈ L x\in L x ∈ L , we have D ( x ) = ∑ g ∈ G D g ( x ) D(x)=\sum_{g\in G}D_g(x) D ( x ) = ∑ g ∈ G D g ( x ) , where only finitely many summands are nonzero.
In this sense we write D = ∑ g ∈ G D g D=\sum_{g\in G}D_g D = ∑ g ∈ G D g . A linear map D D D is called homogeneous of degree g g g if D = D g D=D_g D = D g .
Lemma 7. Let L = ⨁ g ∈ G L g L=\bigoplus_{g\in G}L_g L = ⨁ g ∈ G L g be a G G G -graded Lie algebra and let D ∈ QDer ( L ) D\in\operatorname{QDer}(L) D ∈ QDer ( L ) with associated map D ′ D' D ′ . Then for every g ∈ G g\in G g ∈ G the homogeneous component D g D_g D g is a quasi-derivation of L L L and D g ′ D'_g D g ′ is associated with D g D_g D g . The same holds for δ \delta δ -derivations: every homogeneous component of a δ \delta δ -derivation is a δ \delta δ -derivation.
The Schrödinger–Witt algebras
Definition 8 (see [13 ] ). Let ε ∈ { 0 , 1 2 } \varepsilon\in\left\{0,\frac{1}{2}\right\} ε ∈ { 0 , 2 1 } . The Schrödinger–Witt algebra s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) is the Lie algebra with the basis { L m , Y n , M m ∣ m ∈ Z , n ∈ ε + Z } \{L_m,Y_n,M_m\mid m\in\mathbb{Z},\ n\in\varepsilon+\mathbb{Z}\} { L m , Y n , M m ∣ m ∈ Z , n ∈ ε + Z } and the following nontrivial relations:
[ L m , L n ] = ( m − n ) L m + n , [ L m , Y n ] = ( m 2 − n ) Y m + n , [L_m,L_n]=(m-n)L_{m+n},\qquad[L_m,Y_n]=\left(\frac{m}{2}-n\right)Y_{m+n}, [ L m , L n ] = ( m − n ) L m + n , [ L m , Y n ] = ( 2 m − n ) Y m + n , [ L m , M n ] = − n M m + n , [ Y m , Y n ] = ( m − n ) M m + n . (6) [L_m,M_n]=-nM_{m+n},\qquad[Y_m,Y_n]=(m-n)M_{m+n}. \tag*{(6)} [ L m , M n ] = − n M m + n , [ Y m , Y n ] = ( m − n ) M m + n . (6) All other brackets of basis elements are zero. The algebra s v ( 1 2 ) \mathfrak{sv}\left(\frac{1}{2}\right) sv ( 2 1 ) is called the original Schrödinger–Witt algebra and s v ( 0 ) \mathfrak{sv}(0) sv ( 0 ) is called the twisted Schrödinger–Witt algebra.
The algebra s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) is graded by the abelian group G = 1 2 Z × Z 2 G=\frac{1}{2}\mathbb{Z}\times\mathbb{Z}_2 G = 2 1 Z × Z 2 :
s v ( ε ) ( t , 0 ˉ ) = ⟨ L t , M t ⟩ ( t ∈ Z ) , s v ( ε ) ( t , 1 ˉ ) = ⟨ Y t ⟩ ( t ∈ ε + Z ) , \mathfrak{sv}(\varepsilon)_{(t,\bar{0})}=\langle L_t,M_t\rangle\ (t\in\mathbb{Z}),\qquad\mathfrak{sv}(\varepsilon)_{(t,\bar{1})}=\langle Y_t\rangle\ (t\in\varepsilon+\mathbb{Z}), sv ( ε ) ( t , 0 ˉ ) = ⟨ L t , M t ⟩ ( t ∈ Z ) , sv ( ε ) ( t , 1 ˉ ) = ⟨ Y t ⟩ ( t ∈ ε + Z ) ,
and all other homogeneous components are zero; by (6),
[ s v ( ε ) ( a , x ) , s v ( ε ) ( b , y ) ] ⊆ s v ( ε ) ( a + b , x + y ) . [\mathfrak{sv}(\varepsilon)_{(a,x)},\mathfrak{sv}(\varepsilon)_{(b,y)}]\subseteq\mathfrak{sv}(\varepsilon)_{(a+b,x+y)}. [ sv ( ε ) ( a , x ) , sv ( ε ) ( b , y ) ] ⊆ sv ( ε ) ( a + b , x + y ) .
Let D D D be a linear map of s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) and ( t , x ) ∈ G (t,x)\in G ( t , x ) ∈ G . If x = 0 ˉ x=\bar{0} x = 0 ˉ and t ∉ Z t\notin\mathbb{Z} t ∈ / Z , or x = 1 ˉ x=\bar{1} x = 1 ˉ and t ∉ ε + Z t\notin\varepsilon+\mathbb{Z} t ∈ / ε + Z , then D ( t , x ) = 0 D_{(t,x)}=0 D ( t , x ) = 0 , because the corresponding homogeneous components of the image are zero. We say that a linear map is even of degree j ∈ Z j\in\mathbb{Z} j ∈ Z if it is homogeneous of degree ( j , 0 ˉ ) (j,\bar{0}) ( j , 0 ˉ ) , and odd of degree s ∈ ε + Z s\in\varepsilon+\mathbb{Z} s ∈ ε + Z if it is homogeneous of degree ( s , 1 ˉ ) (s,\bar{1}) ( s , 1 ˉ ) . Thus, every quasi-derivation of s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) is the sum of its even homogeneous components of degrees j ∈ Z j\in\mathbb{Z} j ∈ Z and its odd homogeneous components of degrees s ∈ ε + Z s\in\varepsilon+\mathbb{Z} s ∈ ε + Z , and all these components are quasi-derivations (Lemma 7).
Lemma 9 (see [30 ] ). The algebra s v ( ϵ ) \mathfrak{sv}(\epsilon) sv ( ϵ ) is perfect and Z ( s v ( ϵ ) ) = ⟨ M 0 ⟩ Z(\mathfrak{sv}(\epsilon))=\langle M_0\rangle Z ( sv ( ϵ )) = ⟨ M 0 ⟩ .
By Lemmas 6 and 9, the map D ′ D' D ′ associated with a quasi-derivation D D D of s v ( ϵ ) \mathfrak{sv}(\epsilon) sv ( ϵ ) is unique. Moreover, if D D D is a homogeneous quasi-derivation of s v ( ϵ ) \mathfrak{sv}(\epsilon) sv ( ϵ ) , then D ′ D' D ′ is homogeneous of the same degree, since by Lemma 7 the homogeneous component of D ′ D' D ′ of the degree of D D D is also associated with D D D .
The description of derivations of s v ( ϵ ) \mathfrak{sv}(\epsilon) sv ( ϵ ) is given in the following theorem.
Theorem 10 (see [30 ] ). Der s v ( ϵ ) = Inn s v ( ϵ ) ⊕ ⟨ D 1 ⟩ ⊕ ⟨ D 2 ⟩ ⊕ ⟨ D 3 ⟩ \operatorname{Der}\mathfrak{sv}(\epsilon)=\operatorname{Inn}\mathfrak{sv}(\epsilon)\oplus\langle D_1\rangle\oplus\langle D_2\rangle\oplus\langle D_3\rangle Der sv ( ϵ ) = Inn sv ( ϵ ) ⊕ ⟨ D 1 ⟩ ⊕ ⟨ D 2 ⟩ ⊕ ⟨ D 3 ⟩ , where Inn s v ( ϵ ) = { ad x ∣ x ∈ s v ( ϵ ) } \operatorname{Inn}\mathfrak{sv}(\epsilon)=\{\operatorname{ad}_x\mid x\in\mathfrak{sv}(\epsilon)\} Inn sv ( ϵ ) = { ad x ∣ x ∈ sv ( ϵ )} and D 1 , D 2 , D 3 D_1,D_2,D_3 D 1 , D 2 , D 3 are the outer derivations defined by
D 1 ( L m ) = M m , D 1 ( Y n ) = 0 , D 1 ( M m ) = 0 , D 2 ( L m ) = m M m , D 2 ( Y n ) = 0 , D 2 ( M m ) = 0 , D 3 ( L m ) = 0 , D 3 ( Y n ) = Y n , D 3 ( M m ) = 2 M m . \begin{aligned}
D_1(L_m)&=M_m, & D_1(Y_n)&=0, & D_1(M_m)&=0,\\
D_2(L_m)&=mM_m, & D_2(Y_n)&=0, & D_2(M_m)&=0,\\
D_3(L_m)&=0, & D_3(Y_n)&=Y_n, & D_3(M_m)&=2M_m.
\end{aligned} D 1 ( L m ) D 2 ( L m ) D 3 ( L m ) = M m , = m M m , = 0 , D 1 ( Y n ) D 2 ( Y n ) D 3 ( Y n ) = 0 , = 0 , = Y n , D 1 ( M m ) D 2 ( M m ) D 3 ( M m ) = 0 , = 0 , = 2 M m .
Theorem 10 will also be deduced from our results.
For every t ∈ Z t\in\mathbb{Z} t ∈ Z and s ∈ ϵ + Z s\in\epsilon+\mathbb{Z} s ∈ ϵ + Z the inner derivations act as follows:
ad L t ( L m ) = ( t − m ) L m + t , ad L t ( Y n ) = ( t 2 − n ) Y n + t , ad L t ( M m ) = − m M m + t , ad M t ( L m ) = t M m + t , ad M t ( Y n ) = 0 , ad M t ( M m ) = 0 , ad Y s ( L m ) = ( s − m 2 ) Y m + s , ad Y s ( Y n ) = ( s − n ) M n + s , ad Y s ( M m ) = 0. (7) \begin{aligned}
\operatorname{ad}_{L_t}(L_m)&=(t-m)L_{m+t}, & \operatorname{ad}_{L_t}(Y_n)&=\left(\frac{t}{2}-n\right)Y_{n+t}, & \operatorname{ad}_{L_t}(M_m)&=-mM_{m+t},\\
\operatorname{ad}_{M_t}(L_m)&=tM_{m+t}, & \operatorname{ad}_{M_t}(Y_n)&=0, & \operatorname{ad}_{M_t}(M_m)&=0,\\
\operatorname{ad}_{Y_s}(L_m)&=\left(s-\frac{m}{2}\right)Y_{m+s}, & \operatorname{ad}_{Y_s}(Y_n)&=(s-n)M_{n+s}, & \operatorname{ad}_{Y_s}(M_m)&=0. \tag*{(7)}
\end{aligned} ad L t ( L m ) ad M t ( L m ) ad Y s ( L m ) = ( t − m ) L m + t , = t M m + t , = ( s − 2 m ) Y m + s , ad L t ( Y n ) ad M t ( Y n ) ad Y s ( Y n ) = ( 2 t − n ) Y n + t , = 0 , = ( s − n ) M n + s , ad L t ( M m ) ad M t ( M m ) ad Y s ( M m ) = − m M m + t , = 0 , = 0. (7)
The descriptions of 1 2 \frac{1}{2} 2 1 -derivations and of transposed Poisson structures of s v ( ϵ ) \mathfrak{sv}(\epsilon) sv ( ϵ ) were obtained by Yuan and Hua [37 ] . Both of them will also be deduced from our results (Corollaries 21 and 24 below).
Theorem 11 (see [37 ] ). Every 1 2 \frac{1}{2} 2 1 -derivation of s v ( ϵ ) \mathfrak{sv}(\epsilon) sv ( ϵ ) is trivial, i.e., Δ 1 2 ( s v ( ϵ ) ) = ⟨ Id ⟩ \Delta_{\frac{1}{2}}(\mathfrak{sv}(\epsilon))=\langle\operatorname{Id}\rangle Δ 2 1 ( sv ( ϵ )) = ⟨ Id ⟩ . Moreover, all transposed Poisson structures on s v ( ϵ ) \mathfrak{sv}(\epsilon) sv ( ϵ ) are trivial.
Three functional equations
In this subsection we solve three functional equations arising in the study of homogeneous quasi-derivations of s v ( ϵ ) \mathfrak{sv}(\epsilon) sv ( ϵ ) ; they describe the action of the Witt part ⟨ L m ∣ m ∈ Z ⟩ \langle L_m\mid m\in\mathbb{Z}\rangle ⟨ L m ∣ m ∈ Z ⟩ on the parts ⟨ L m ⟩ \langle L_m\rangle ⟨ L m ⟩ , ⟨ Y n ⟩ \langle Y_n\rangle ⟨ Y n ⟩ and ⟨ M m ⟩ \langle M_m\rangle ⟨ M m ⟩ . For a sequence { u m } m ∈ Z \{u_m\}_{m\in\mathbb{Z}} { u m } m ∈ Z of complex numbers, let
w m = u m + 2 − 2 u m + 1 + u m w_m=u_{m+2}-2u_{m+1}+u_m w m = u m + 2 − 2 u m + 1 + u m
be its second differences. We shall use two elementary facts: w m = 0 w_m=0 w m = 0 for all m m m if and only if u m = α + β m u_m=\alpha+\beta m u m = α + β m for some α , β ∈ C \alpha,\beta\in\mathbb{C} α , β ∈ C ; and the third differences
w m + 1 − w m = u m + 3 − 3 u m + 2 + 3 u m + 1 − u m w_{m+1}-w_m=u_{m+3}-3u_{m+2}+3u_{m+1}-u_m w m + 1 − w m = u m + 3 − 3 u m + 2 + 3 u m + 1 − u m
vanish for all m m m if and only if u m = α + β m + γ m 2 u_m=\alpha+\beta m+\gamma m^2 u m = α + β m + γ m 2 for some α , β , γ ∈ C \alpha,\beta,\gamma\in\mathbb{C} α , β , γ ∈ C .
If u m + 1 − u m = c u_{m+1}-u_m=c u m + 1 − u m = c for all m m m , induction in both directions gives u m = u 0 + c m u_m=u_0+cm u m = u 0 + c m . Applying this to d m = u m + 1 − u m d_m=u_{m+1}-u_m d m = u m + 1 − u m , which satisfies d m + 1 − d m = w m d_{m+1}-d_m=w_m d m + 1 − d m = w m , proves the first fact. For the second, vanishing third differences give w m = w 0 = : 2 γ w_m=w_0=:2\gamma w m = w 0 =: 2 γ for all m m m ; since γ m 2 \gamma m^2 γ m 2 has constant second differences 2 γ 2\gamma 2 γ , the sequence u m − γ m 2 u_m-\gamma m^2 u m − γ m 2 has vanishing second differences, and the first fact applies. The converses are direct computations.
Lemma 12. Let j ∈ C j \in\mathbb{C} j ∈ C and let { a m } m ∈ Z \{a_m\}_{m\in\mathbb{Z}} { a m } m ∈ Z , { a m ′ } m ∈ Z \{a'_m\}_{m\in\mathbb{Z}} { a m ′ } m ∈ Z be complex numbers such that
( m − n + j ) a m + ( m − n − j ) a n = ( m − n ) a m + n ′ for all m , n ∈ Z . (8) (m-n+j)a_m+(m-n-j)a_n=(m-n)a'_{m+n}\quad\text{for all }m,n\in\mathbb{Z}. \tag*{(8)} ( m − n + j ) a m + ( m − n − j ) a n = ( m − n ) a m + n ′ for all m , n ∈ Z . (8) Then there are α , β ∈ C \alpha,\beta\in\mathbb{C} α , β ∈ C such that a m = α + β m a_m=\alpha+\beta m a m = α + β m and a m ′ = 2 α + β ( m + j ) a'_m=2\alpha+\beta(m+j) a m ′ = 2 α + β ( m + j ) for all m ∈ Z m\in\mathbb{Z} m ∈ Z .
Proof. Putting n = m + k n=m+k n = m + k in (8), we obtain
k a 2 m + k ′ = ( k − j ) a m + ( k + j ) a m + k for all m , k ∈ Z . (9) ka'_{2m+k}=(k-j)a_m+(k+j)a_{m+k}\quad\text{for all }m,k\in\mathbb{Z}. \tag*{(9)} k a 2 m + k ′ = ( k − j ) a m + ( k + j ) a m + k for all m , k ∈ Z . (9) Taking k = 3 k=3 k = 3 in (9), and then replacing m m m by m + 1 m+1 m + 1 and taking k = 1 k=1 k = 1 , we get 3 a 2 m + 3 ′ = ( 3 − j ) a m + ( 3 + j ) a m + 3 3a'_{2m+3}=(3-j)a_m+(3+j)a_{m+3} 3 a 2 m + 3 ′ = ( 3 − j ) a m + ( 3 + j ) a m + 3 and a 2 m + 3 ′ = ( 1 − j ) a m + 1 + ( 1 + j ) a m + 2 a'_{2m+3}=(1-j)a_{m+1}+(1+j)a_{m+2} a 2 m + 3 ′ = ( 1 − j ) a m + 1 + ( 1 + j ) a m + 2 . Eliminating a 2 m + 3 ′ a'_{2m+3} a 2 m + 3 ′ , we obtain
( 3 + j ) a m + 3 − 3 ( 1 + j ) a m + 2 − 3 ( 1 − j ) a m + 1 + ( 3 − j ) a m = 0. (3+j)a_{m+3}-3(1+j)a_{m+2}-3(1-j)a_{m+1}+(3-j)a_m=0. ( 3 + j ) a m + 3 − 3 ( 1 + j ) a m + 2 − 3 ( 1 − j ) a m + 1 + ( 3 − j ) a m = 0. In terms of the second differences w m = a m + 2 − 2 a m + 1 + a m w_m=a_{m+2}-2a_{m+1}+a_m w m = a m + 2 − 2 a m + 1 + a m this relation reads
( 3 + j ) w m + 1 + ( 3 − j ) w m = 0 for all m ∈ Z . (10) (3+j)w_{m+1}+(3-j)w_m=0\quad\text{for all }m\in\mathbb{Z}. \tag*{(10)} ( 3 + j ) w m + 1 + ( 3 − j ) w m = 0 for all m ∈ Z . (10) Similarly, taking k = 4 k=4 k = 4 in (9), and then replacing m m m by m + 1 m+1 m + 1 and taking k = 2 k=2 k = 2 , we get 4 a 2 m + 4 ′ = ( 4 − j ) a m + ( 4 + j ) a m + 4 4a'_{2m+4}=(4-j)a_m+(4+j)a_{m+4} 4 a 2 m + 4 ′ = ( 4 − j ) a m + ( 4 + j ) a m + 4 and 2 a 2 m + 4 ′ = ( 2 − j ) a m + 1 + ( 2 + j ) a m + 3 2a'_{2m+4}=(2-j)a_{m+1}+(2+j)a_{m+3} 2 a 2 m + 4 ′ = ( 2 − j ) a m + 1 + ( 2 + j ) a m + 3 , which gives ( 4 + j ) a m + 4 − 2 ( 2 + j ) a m + 3 − 2 ( 2 − j ) a m + 1 + ( 4 − j ) a m = 0 (4+j)a_{m+4}-2(2+j)a_{m+3}-2(2-j)a_{m+1}+(4-j)a_m=0 ( 4 + j ) a m + 4 − 2 ( 2 + j ) a m + 3 − 2 ( 2 − j ) a m + 1 + ( 4 − j ) a m = 0 , i.e.,
( 4 + j ) w m + 2 + 4 w m + 1 + ( 4 − j ) w m = 0 for all m ∈ Z . (11) (4+j)w_{m+2}+4w_{m+1}+(4-j)w_m=0\quad\text{for all }m\in\mathbb{Z}. \tag*{(11)} ( 4 + j ) w m + 2 + 4 w m + 1 + ( 4 − j ) w m = 0 for all m ∈ Z . (11) Now multiply (11) by ( 3 + j ) 2 (3+j)^2 ( 3 + j ) 2 and use (10) (for m m m and for m + 1 m+1 m + 1 ) in the form
( 3 + j ) w m + 1 = ( j − 3 ) w m and ( 3 + j ) 2 w m + 2 = ( j − 3 ) 2 w m . (3+j)w_{m+1}=(j-3)w_m\quad\text{and}\quad(3+j)^2w_{m+2}=(j-3)^2w_m. ( 3 + j ) w m + 1 = ( j − 3 ) w m and ( 3 + j ) 2 w m + 2 = ( j − 3 ) 2 w m . We obtain
( ( 4 + j ) ( j − 3 ) 2 + 4 ( j + 3 ) ( j − 3 ) + ( 4 − j ) ( j + 3 ) 2 ) w m = 0. \left((4+j)(j-3)^2+4(j+3)(j-3)+(4-j)(j+3)^2\right)w_m=0. ( ( 4 + j ) ( j − 3 ) 2 + 4 ( j + 3 ) ( j − 3 ) + ( 4 − j ) ( j + 3 ) 2 ) w m = 0. A direct computation shows that the expression in brackets is equal to 36 (for every j ∈ C j\in\mathbb{C} j ∈ C , including j = − 3 j=-3 j = − 3 ). Hence w m = 0 w_m=0 w m = 0 for all m ∈ Z m\in\mathbb{Z} m ∈ Z , and there are α , β ∈ C \alpha,\beta\in\mathbb{C} α , β ∈ C with a m = α + β m a_m=\alpha+\beta m a m = α + β m . Setting n = 0 n=0 n = 0 in (8), we have
m a m ′ = ( m + j ) ( α + β m ) + ( m − j ) α = m ( 2 α + β ( m + j ) ) , ma'_m=(m+j)(\alpha+\beta m)+(m-j)\alpha=m(2\alpha+\beta(m+j)), m a m ′ = ( m + j ) ( α + β m ) + ( m − j ) α = m ( 2 α + β ( m + j )) , which gives a m ′ = 2 α + β ( m + j ) a'_m=2\alpha+\beta(m+j) a m ′ = 2 α + β ( m + j ) for m ≠ 0 m\ne0 m = 0 . Setting m = 1 m=1 m = 1 , n = − 1 n=-1 n = − 1 in (8), we have
2 a 0 ′ = ( 2 + j ) ( α + β ) + ( 2 − j ) ( α − β ) = 4 α + 2 j β , 2a'_0=(2+j)(\alpha+\beta)+(2-j)(\alpha-\beta)=4\alpha+2j\beta, 2 a 0 ′ = ( 2 + j ) ( α + β ) + ( 2 − j ) ( α − β ) = 4 α + 2 j β , i.e., a 0 ′ = 2 α + β j a'_0=2\alpha+\beta j a 0 ′ = 2 α + β j .
Lemma 13. Let s ∈ C s\in\mathbb{C} s ∈ C and let { f m } m ∈ Z \{f_m\}_{m\in\mathbb{Z}} { f m } m ∈ Z , { f m ′ } m ∈ Z \{f'_m\}_{m\in\mathbb{Z}} { f m ′ } m ∈ Z be complex numbers such that
( m + s − n 2 ) f m + ( m 2 − n − s ) f n = ( m − n ) f m + n ′ for all m , n ∈ Z . (12) \left(m+s-\frac{n}{2}\right)f_m+\left(\frac{m}{2}-n-s\right)f_n=(m-n)f'_{m+n}\quad\text{for all }m,n\in\mathbb{Z}. \tag*{(12)} ( m + s − 2 n ) f m + ( 2 m − n − s ) f n = ( m − n ) f m + n ′ for all m , n ∈ Z . (12) Then there are ϕ , ψ ∈ C \phi,\psi\in\mathbb{C} ϕ , ψ ∈ C such that f m = ϕ + ψ m f_m=\phi+\psi m f m = ϕ + ψ m and f m ′ = 3 2 ϕ + ψ ( m + s ) f'_m=\frac{3}{2}\phi+\psi(m+s) f m ′ = 2 3 ϕ + ψ ( m + s ) for all m ∈ Z m\in\mathbb{Z} m ∈ Z .
Proof. For m ∈ Z m \in\mathbb{Z} m ∈ Z we set τ = τ m = m 2 + s \tau=\tau_m=\frac{m}{2}+s τ = τ m = 2 m + s . Putting n = m + k n=m+k n = m + k in (12) , we obtain
( τ m − k 2 ) f m − ( τ m + k ) f m + k = − k f 2 m + k ′ for all m , k ∈ Z . (13) \left(\tau_m-\frac{k}{2}\right)f_m-(\tau_m+k)f_{m+k}=-k f'_{2m+k}\quad\text{for all }m,k\in\mathbb{Z}. \tag*{(13)} ( τ m − 2 k ) f m − ( τ m + k ) f m + k = − k f 2 m + k ′ for all m , k ∈ Z . (13)
Taking k = 3 k=3 k = 3 in (13) , and then replacing m m m by m + 1 m+1 m + 1 (note that τ m + 1 = τ + 1 2 \tau_{m+1}=\tau+\frac{1}{2} τ m + 1 = τ + 2 1 ) and taking k = 1 k=1 k = 1 , we have
( τ − 3 2 ) f m − ( τ + 3 ) f m + 3 = − 3 f 2 m + 3 ′ , τ f m + 1 − ( τ + 3 2 ) f m + 2 = − f 2 m + 3 ′ . \left(\tau-\frac{3}{2}\right)f_m-(\tau+3)f_{m+3}=-3f'_{2m+3},\qquad\tau f_{m+1}-\left(\tau+\frac{3}{2}\right)f_{m+2}=-f'_{2m+3}. ( τ − 2 3 ) f m − ( τ + 3 ) f m + 3 = − 3 f 2 m + 3 ′ , τ f m + 1 − ( τ + 2 3 ) f m + 2 = − f 2 m + 3 ′ .
Eliminating f 2 m + 3 ′ f'_{2m+3} f 2 m + 3 ′ , we get ( τ + 3 ) f m + 3 − 3 ( τ + 3 2 ) f m + 2 + 3 τ f m + 1 − ( τ − 3 2 ) f m = 0 (\tau+3)f_{m+3}-3\left(\tau+\frac{3}{2}\right)f_{m+2}+3\tau f_{m+1}-\left(\tau-\frac{3}{2}\right)f_m=0 ( τ + 3 ) f m + 3 − 3 ( τ + 2 3 ) f m + 2 + 3 τ f m + 1 − ( τ − 2 3 ) f m = 0 . In terms of the second differences w m = f m + 2 − 2 f m + 1 + f m w_m=f_{m+2}-2f_{m+1}+f_m w m = f m + 2 − 2 f m + 1 + f m this relation reads
( τ + 3 ) w m + 1 − ( τ − 3 2 ) w m = 0 for all m ∈ Z . (14) (\tau+3)w_{m+1}-\left(\tau-\frac{3}{2}\right)w_m=0\quad\text{for all }m\in\mathbb{Z}. \tag*{(14)} ( τ + 3 ) w m + 1 − ( τ − 2 3 ) w m = 0 for all m ∈ Z . (14)
Similarly, taking k = 4 k=4 k = 4 in (13) , and then replacing m m m by m + 1 m+1 m + 1 and taking k = 2 k=2 k = 2 , we have
( τ − 2 ) f m − ( τ + 4 ) f m + 4 = − 4 f 2 m + 4 ′ , ( τ − 1 2 ) f m + 1 − ( τ + 5 2 ) f m + 3 = − 2 f 2 m + 4 ′ , (\tau-2)f_m-(\tau+4)f_{m+4}=-4f'_{2m+4},\qquad\left(\tau-\frac{1}{2}\right)f_{m+1}-\left(\tau+\frac{5}{2}\right)f_{m+3}=-2f'_{2m+4}, ( τ − 2 ) f m − ( τ + 4 ) f m + 4 = − 4 f 2 m + 4 ′ , ( τ − 2 1 ) f m + 1 − ( τ + 2 5 ) f m + 3 = − 2 f 2 m + 4 ′ ,
which gives ( τ − 2 ) f m − 2 ( τ − 1 2 ) f m + 1 + 2 ( τ + 5 2 ) f m + 3 − ( τ + 4 ) f m + 4 = 0 (\tau-2)f_m-2\left(\tau-\frac{1}{2}\right)f_{m+1}+2\left(\tau+\frac{5}{2}\right)f_{m+3}-(\tau+4)f_{m+4}=0 ( τ − 2 ) f m − 2 ( τ − 2 1 ) f m + 1 + 2 ( τ + 2 5 ) f m + 3 − ( τ + 4 ) f m + 4 = 0 , i.e.,
− ( τ + 4 ) w m + 2 − 3 w m + 1 + ( τ − 2 ) w m = 0 for all m ∈ Z . (15) -(\tau+4)w_{m+2}-3w_{m+1}+(\tau-2)w_m=0\quad\text{for all }m\in\mathbb{Z}. \tag*{(15)} − ( τ + 4 ) w m + 2 − 3 w m + 1 + ( τ − 2 ) w m = 0 for all m ∈ Z . (15)
Replacing m m m by m + 1 m+1 m + 1 in (14) , we have ( τ + 7 2 ) w m + 2 = ( τ − 1 ) w m + 1 \left(\tau+\frac{7}{2}\right)w_{m+2}=(\tau-1)w_{m+1} ( τ + 2 7 ) w m + 2 = ( τ − 1 ) w m + 1 . Now multiply (15) by ( τ + 7 2 ) ( τ + 3 ) \left(\tau+\frac{7}{2}\right)(\tau+3) ( τ + 2 7 ) ( τ + 3 ) and use ( τ + 7 2 ) ( τ + 3 ) w m + 2 = ( τ − 1 ) ( τ − 3 2 ) w m \left(\tau+\frac{7}{2}\right)(\tau+3)w_{m+2}=(\tau-1)\left(\tau-\frac{3}{2}\right)w_m ( τ + 2 7 ) ( τ + 3 ) w m + 2 = ( τ − 1 ) ( τ − 2 3 ) w m and ( τ + 3 ) w m + 1 = ( τ − 3 2 ) w m (\tau+3)w_{m+1}=\left(\tau-\frac{3}{2}\right)w_m ( τ + 3 ) w m + 1 = ( τ − 2 3 ) w m . We obtain
0 = ( ( τ + 4 ) ( τ − 1 ) ( τ − 3 2 ) + 3 ( τ + 7 2 ) ( τ − 3 2 ) − ( τ − 2 ) ( τ + 7 2 ) ( τ + 3 ) ) w m = 45 4 w m . 0=\left((\tau+4)(\tau-1)\left(\tau-\frac{3}{2}\right)+3\left(\tau+\frac{7}{2}\right)\left(\tau-\frac{3}{2}\right)-(\tau-2)\left(\tau+\frac{7}{2}\right)(\tau+3)\right)w_m=\frac{45}{4}w_m. 0 = ( ( τ + 4 ) ( τ − 1 ) ( τ − 2 3 ) + 3 ( τ + 2 7 ) ( τ − 2 3 ) − ( τ − 2 ) ( τ + 2 7 ) ( τ + 3 ) ) w m = 4 45 w m .
Hence w m = 0 w_m=0 w m = 0 for all m ∈ Z m\in\mathbb{Z} m ∈ Z , and there are ϕ , ψ ∈ C \phi,\psi\in\mathbb{C} ϕ , ψ ∈ C with f m = ϕ + ψ m f_m=\phi+\psi m f m = ϕ + ψ m . Setting n = 0 n=0 n = 0 in (12) , we get m f m ′ = ( m + s ) ( ϕ + ψ m ) + ( m 2 − s ) ϕ = m ( 3 2 ϕ + ψ ( m + s ) ) mf'_m=(m+s)(\phi+\psi m)+\left(\frac{m}{2}-s\right)\phi=m\left(\frac{3}{2}\phi+\psi(m+s)\right) m f m ′ = ( m + s ) ( ϕ + ψ m ) + ( 2 m − s ) ϕ = m ( 2 3 ϕ + ψ ( m + s ) ) , which gives the required formula for f m ′ f'_m f m ′ with m ≠ 0 m\ne0 m = 0 . Setting m = 1 m=1 m = 1 , n = − 1 n=-1 n = − 1 in (12) , we get 2 f 0 ′ = ( 3 2 + s ) ( ϕ + ψ ) + ( 3 2 − s ) ( ϕ − ψ ) = 3 ϕ + 2 s ψ 2f'_0=\left(\frac{3}{2}+s\right)(\phi+\psi)+\left(\frac{3}{2}-s\right)(\phi-\psi)=3\phi+2s\psi 2 f 0 ′ = ( 2 3 + s ) ( ϕ + ψ ) + ( 2 3 − s ) ( ϕ − ψ ) = 3 ϕ + 2 s ψ , i.e., f 0 ′ = 3 2 ϕ + s ψ f'_0=\frac{3}{2}\phi+s\psi f 0 ′ = 2 3 ϕ + s ψ .
Lemma 14. Let j ∈ Z j\in\mathbb{Z} j ∈ Z and let { b m } m ∈ Z \{b_m\}_{m\in\mathbb{Z}} { b m } m ∈ Z , { b m ′ } m ∈ Z \{b'_m\}_{m\in\mathbb{Z}} { b m ′ } m ∈ Z be complex numbers such that
( m + j ) b m − ( n + j ) b n = ( m − n ) b m + n ′ for all m , n ∈ Z . (16) (m+j)b_m-(n+j)b_n=(m-n)b'_{m+n}\quad\text{for all }m,n\in\mathbb{Z}. \tag*{(16)} ( m + j ) b m − ( n + j ) b n = ( m − n ) b m + n ′ for all m , n ∈ Z . (16) Then there are β 1 , β 2 ∈ C \beta_1,\beta_2\in\mathbb{C} β 1 , β 2 ∈ C such that b m = β 1 + β 2 m b_m=\beta_1+\beta_2m b m = β 1 + β 2 m for m ≠ − j m\ne-j m = − j and b m ′ = β 1 + β 2 ( m + j ) b'_m=\beta_1+\beta_2(m+j) b m ′ = β 1 + β 2 ( m + j ) for all m ∈ Z m\in\mathbb{Z} m ∈ Z .
Proof. Set g m = ( m + j ) b m g_m=(m+j)b_m g m = ( m + j ) b m , so that g m − g n = ( m − n ) b m + n ′ g_m-g_n=(m-n)b'_{m+n} g m − g n = ( m − n ) b m + n ′ . Taking ( m , n ) = ( m , m + 3 ) (m,n)=(m,m+3) ( m , n ) = ( m , m + 3 ) and ( m , n ) = ( m + 1 , m + 2 ) (m,n)=(m+1,m+2) ( m , n ) = ( m + 1 , m + 2 ) , we get g m − g m + 3 = − 3 b 2 m + 3 ′ g_m-g_{m+3}=-3b'_{2m+3} g m − g m + 3 = − 3 b 2 m + 3 ′ and g m + 1 − g m + 2 = − b 2 m + 3 ′ g_{m+1}-g_{m+2}=-b'_{2m+3} g m + 1 − g m + 2 = − b 2 m + 3 ′ . Hence
g m + 3 − 3 g m + 2 + 3 g m + 1 − g m = 0 for all m ∈ Z , g_{m+3}-3g_{m+2}+3g_{m+1}-g_m=0\quad\text{for all }m\in\mathbb{Z}, g m + 3 − 3 g m + 2 + 3 g m + 1 − g m = 0 for all m ∈ Z , and there are P , Q , R ∈ C P,Q,R\in\mathbb{C} P , Q , R ∈ C such that g m = P + Q m + R m 2 g_m=P+Qm+Rm^2 g m = P + Q m + R m 2 for all m ∈ Z m\in\mathbb{Z} m ∈ Z . Then
( m − n ) b m + n ′ = g m − g n = ( m − n ) ( Q + R ( m + n ) ) . (m-n)b'_{m+n}=g_m-g_n=(m-n)(Q+R(m+n)). ( m − n ) b m + n ′ = g m − g n = ( m − n ) ( Q + R ( m + n )) . For every k ∈ Z k\in\mathbb{Z} k ∈ Z we can choose m ≠ n m\ne n m = n with m + n = k m+n=k m + n = k , so b k ′ = Q + R k b'_k=Q+Rk b k ′ = Q + R k . Finally, 0 = g − j = P − Q j + R j 2 0=g_{-j}=P-Qj+Rj^2 0 = g − j = P − Q j + R j 2 , hence g m = ( m + j ) ( Q − R j + R m ) g_m=(m+j)(Q-Rj+Rm) g m = ( m + j ) ( Q − R j + R m ) . Setting β 1 = Q − R j \beta_1=Q-Rj β 1 = Q − R j and β 2 = R \beta_2=R β 2 = R , we get
( m + j ) b m = ( m + j ) ( β 1 + β 2 m ) for all m ∈ Z (m+j)b_m=(m+j)(\beta_1+\beta_2m)\quad\text{for all }m\in\mathbb{Z} ( m + j ) b m = ( m + j ) ( β 1 + β 2 m ) for all m ∈ Z and b m ′ = β 1 + β 2 ( m + j ) b'_m=\beta_1+\beta_2(m+j) b m ′ = β 1 + β 2 ( m + j ) .
Quasi-derivations of the Schrödinger–Witt algebras
Definition 15. For t ∈ Z t \in\mathbb{Z} t ∈ Z and s ∈ ε + Z s \in\varepsilon+\mathbb{Z} s ∈ ε + Z we define the linear maps Ω t \Omega_t Ω t , Ψ t \Psi_t Ψ t , Θ s \Theta_s Θ s and Ω t ′ \Omega'_t Ω t ′ , Ψ t ′ \Psi'_t Ψ t ′ , Θ s ′ \Theta'_s Θ s ′ of s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) by
Ω t ( L m ) = L m + t , Ω t ( Y n ) = 1 2 Y n + t , Ω t ( M m ) = 0 , Ω t ′ ( L m ) = 2 L m + t , Ω t ′ ( Y n ) = 3 2 Y n + t , Ω t ′ ( M m ) = M m + t , Ψ t ( L m ) = m M m + t , Ψ t ( Y n ) = 0 , Ψ t ( M m ) = 0 , Ψ t ′ ( L m ) = ( m + t ) M m + t , Ψ t ′ ( Y n ) = 0 , Ψ t ′ ( M m ) = 0 , Θ s ( L m ) = Y m + s , Θ s ( Y n ) = M n + s , Θ s ( M m ) = 0 , Θ s ′ ( L m ) = 3 2 Y m + s , Θ s ′ ( Y n ) = 2 M n + s , Θ s ′ ( M m ) = 0. \begin{aligned}
\Omega_t(L_m)&=L_{m+t}, & \Omega_t(Y_n)&=\frac{1}{2}Y_{n+t}, & \Omega_t(M_m)&=0,\\
\Omega'_t(L_m)&=2L_{m+t}, & \Omega'_t(Y_n)&=\frac{3}{2}Y_{n+t}, & \Omega'_t(M_m)&=M_{m+t},\\
\Psi_t(L_m)&=mM_{m+t}, & \Psi_t(Y_n)&=0, & \Psi_t(M_m)&=0,\\
\Psi'_t(L_m)&=(m+t)M_{m+t}, & \Psi'_t(Y_n)&=0, & \Psi'_t(M_m)&=0,\\
\Theta_s(L_m)&=Y_{m+s}, & \Theta_s(Y_n)&=M_{n+s}, & \Theta_s(M_m)&=0,\\
\Theta'_s(L_m)&=\frac{3}{2}Y_{m+s}, & \Theta'_s(Y_n)&=2M_{n+s}, & \Theta'_s(M_m)&=0.
\end{aligned} Ω t ( L m ) Ω t ′ ( L m ) Ψ t ( L m ) Ψ t ′ ( L m ) Θ s ( L m ) Θ s ′ ( L m ) = L m + t , = 2 L m + t , = m M m + t , = ( m + t ) M m + t , = Y m + s , = 2 3 Y m + s , Ω t ( Y n ) Ω t ′ ( Y n ) Ψ t ( Y n ) Ψ t ′ ( Y n ) Θ s ( Y n ) Θ s ′ ( Y n ) = 2 1 Y n + t , = 2 3 Y n + t , = 0 , = 0 , = M n + s , = 2 M n + s , Ω t ( M m ) Ω t ′ ( M m ) Ψ t ( M m ) Ψ t ′ ( M m ) Θ s ( M m ) Θ s ′ ( M m ) = 0 , = M m + t , = 0 , = 0 , = 0 , = 0. Note that Ω 0 = Id − 1 2 D 3 \Omega_0=\operatorname{Id}-\frac{1}{2}D_3 Ω 0 = Id − 2 1 D 3 and Ψ 0 = D 2 \Psi_0=D_2 Ψ 0 = D 2 . The maps Ω t \Omega_t Ω t and Ψ t \Psi_t Ψ t are even of degree t t t , and the map Θ s \Theta_s Θ s is odd of degree $s.
Lemma 16. For all t ∈ Z t \in\mathbb{Z} t ∈ Z and s ∈ ε + Z s \in\varepsilon+\mathbb{Z} s ∈ ε + Z , the maps Ω t \Omega_t Ω t , Ψ t \Psi_t Ψ t and Θ s \Theta_s Θ s are quasi-derivations of s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) with associated maps Ω t ′ \Omega'_t Ω t ′ , Ψ t ′ \Psi'_t Ψ t ′ and Θ s ′ \Theta'_s Θ s ′ , respectively. The maps Ω t \Omega_t Ω t , Θ s \Theta_s Θ s and, for t ≠ 0 t \ne0 t = 0 , Ψ t \Psi_t Ψ t are not derivations.
Proof. It is enough to verify (1) on pairs of basis elements. For Ω t \Omega_t Ω t we have
[ Ω t ( L m ) , L n ] + [ L m , Ω t ( L n ) ] = ( m + t − n ) L m + n + t + ( m − n − t ) L m + n + t = 2 ( m − n ) L m + n + t = Ω t ′ ( ( m − n ) L m + n ) , \begin{aligned}
[\Omega_t(L_m),L_n]+[L_m,\Omega_t(L_n)]
&=(m+t-n)L_{m+n+t}+(m-n-t)L_{m+n+t}\\
&=2(m-n)L_{m+n+t}=\Omega'_t((m-n)L_{m+n}),
\end{aligned} [ Ω t ( L m ) , L n ] + [ L m , Ω t ( L n )] = ( m + t − n ) L m + n + t + ( m − n − t ) L m + n + t = 2 ( m − n ) L m + n + t = Ω t ′ (( m − n ) L m + n ) ,
[ Ω t ( L m ) , Y n ] + [ L m , Ω t ( Y n ) ] = ( m + t 2 − n ) Y m + n + t + 1 2 ( m 2 − n − t ) Y m + n + t = 3 2 ( m 2 − n ) Y m + n + t = Ω t ′ ( ( m 2 − n ) Y m + n ) , \begin{aligned}
[\Omega_t(L_m),Y_n]+[L_m,\Omega_t(Y_n)]
&=\left(\frac{m+t}{2}-n\right)Y_{m+n+t}+\frac{1}{2}\left(\frac{m}{2}-n-t\right)Y_{m+n+t}\\
&=\frac{3}{2}\left(\frac{m}{2}-n\right)Y_{m+n+t}
=\Omega'_t\left(\left(\frac{m}{2}-n\right)Y_{m+n}\right),
\end{aligned} [ Ω t ( L m ) , Y n ] + [ L m , Ω t ( Y n )] = ( 2 m + t − n ) Y m + n + t + 2 1 ( 2 m − n − t ) Y m + n + t = 2 3 ( 2 m − n ) Y m + n + t = Ω t ′ ( ( 2 m − n ) Y m + n ) ,
[ Ω t ( L m ) , M n ] + [ L m , Ω t ( M n ) ] = − n M m + n + t = Ω t ′ ( − n M m + n ) , [\Omega_t(L_m),M_n]+[L_m,\Omega_t(M_n)]=-nM_{m+n+t}=\Omega'_t(-nM_{m+n}), [ Ω t ( L m ) , M n ] + [ L m , Ω t ( M n )] = − n M m + n + t = Ω t ′ ( − n M m + n ) ,
[ Ω t ( Y m ) , Y n ] + [ Y m , Ω t ( Y n ) ] = 1 2 ( m + t − n ) M m + n + t + 1 2 ( m − n − t ) M m + n + t = ( m − n ) M m + n + t = Ω t ′ ( ( m − n ) M m + n ) , \begin{aligned}
[\Omega_t(Y_m),Y_n]+[Y_m,\Omega_t(Y_n)]
&=\frac{1}{2}(m+t-n)M_{m+n+t}+\frac{1}{2}(m-n-t)M_{m+n+t}\\
&=(m-n)M_{m+n+t}=\Omega'_t((m-n)M_{m+n}),
\end{aligned} [ Ω t ( Y m ) , Y n ] + [ Y m , Ω t ( Y n )] = 2 1 ( m + t − n ) M m + n + t + 2 1 ( m − n − t ) M m + n + t = ( m − n ) M m + n + t = Ω t ′ (( m − n ) M m + n ) ,
and for the pairs ( Y m , M n ) (Y_m,M_n) ( Y m , M n ) and ( M m , M n ) (M_m,M_n) ( M m , M n ) both sides of (1) are zero. For Ψ t \Psi_t Ψ t the only nontrivial case is
[ Ψ t ( L m ) , L n ] + [ L m , Ψ t ( L n ) ] = m ( m + t ) M m + n + t − n ( n + t ) M m + n + t = ( m − n ) ( m + n + t ) M m + n + t = Ψ t ′ ( ( m − n ) L m + n ) ; \begin{aligned}
[\Psi_t(L_m),L_n]+[L_m,\Psi_t(L_n)]
&=m(m+t)M_{m+n+t}-n(n+t)M_{m+n+t}\\
&=(m-n)(m+n+t)M_{m+n+t}=\Psi'_t((m-n)L_{m+n});
\end{aligned} [ Ψ t ( L m ) , L n ] + [ L m , Ψ t ( L n )] = m ( m + t ) M m + n + t − n ( n + t ) M m + n + t = ( m − n ) ( m + n + t ) M m + n + t = Ψ t ′ (( m − n ) L m + n ) ;
for the pairs ( L m , Y n ) (L_m,Y_n) ( L m , Y n ) and ( L m , M n ) (L_m,M_n) ( L m , M n ) both sides are zero because Ψ t ′ ( Y n ) = Ψ t ′ ( M n ) = 0 \Psi'_t(Y_n)=\Psi'_t(M_n)=0 Ψ t ′ ( Y n ) = Ψ t ′ ( M n ) = 0 and [ M k , Y n ] = [ M k , M n ] = 0 [M_k,Y_n]=[M_k,M_n]=0 [ M k , Y n ] = [ M k , M n ] = 0 , and for all other pairs both sides are obviously zero. For Θ s \Theta_s Θ s we have
[ Θ s ( L m ) , L n ] + [ L m , Θ s ( L n ) ] = ( m + s − n 2 ) Y m + n + s + ( m 2 − n − s ) Y m + n + s = 3 2 ( m − n ) Y m + n + s = Θ s ′ ( ( m − n ) L m + n ) , \begin{aligned}
[\Theta_s(L_m),L_n]+[L_m,\Theta_s(L_n)]
&=\left(m+s-\frac{n}{2}\right)Y_{m+n+s}+\left(\frac{m}{2}-n-s\right)Y_{m+n+s}\\
&=\frac{3}{2}(m-n)Y_{m+n+s}=\Theta'_s((m-n)L_{m+n}),
\end{aligned} [ Θ s ( L m ) , L n ] + [ L m , Θ s ( L n )] = ( m + s − 2 n ) Y m + n + s + ( 2 m − n − s ) Y m + n + s = 2 3 ( m − n ) Y m + n + s = Θ s ′ (( m − n ) L m + n ) ,
[ Θ s ( L m ) , Y n ] + [ L m , Θ s ( Y n ) ] = ( m + s − n ) M m + n + s − ( n + s ) M m + n + s = 2 ( m 2 − n ) M m + n + s = Θ s ′ ( ( m 2 − n ) Y m + n ) , \begin{aligned}
[\Theta_s(L_m),Y_n]+[L_m,\Theta_s(Y_n)]
&=(m+s-n)M_{m+n+s}-(n+s)M_{m+n+s}\\
&=2\left(\frac{m}{2}-n\right)M_{m+n+s}
=\Theta'_s\left(\left(\frac{m}{2}-n\right)Y_{m+n}\right),
\end{aligned} [ Θ s ( L m ) , Y n ] + [ L m , Θ s ( Y n )] = ( m + s − n ) M m + n + s − ( n + s ) M m + n + s = 2 ( 2 m − n ) M m + n + s = Θ s ′ ( ( 2 m − n ) Y m + n ) ,
and for all other pairs of basis elements both sides of (1) are zero (we use Θ s ′ ( M k ) = 0 \Theta'_s(M_k)=0 Θ s ′ ( M k ) = 0 and [ M k , Y n ] = [ M k , M n ] = 0 [M_k,Y_n]=[M_k,M_n]=0 [ M k , Y n ] = [ M k , M n ] = 0 ). By Lemmas 6 and 9, a quasi-derivation D D D of s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) is a derivation if and only if D ′ = D D'=D D ′ = D . Since Ω t ′ ( M m ) ≠ Ω t ( M m ) \Omega'_t(M_m) \ne\Omega_t(M_m) Ω t ′ ( M m ) = Ω t ( M m ) , Θ s ′ ( L m ) ≠ Θ s ( L m ) \Theta'_s(L_m) \ne\Theta_s(L_m) Θ s ′ ( L m ) = Θ s ( L m ) and Ψ t ′ ( L 0 ) = m M t ≠ 0 = Ψ t ( L 0 ) \Psi'_t(L_0)=mM_t \ne0=\Psi_t(L_0) Ψ t ′ ( L 0 ) = m M t = 0 = Ψ t ( L 0 ) for t ≠ 0 t \ne0 t = 0 , the last statement follows.
Proposition 17. Let D D D be an even quasi-derivation of degree j ∈ Z j \in\mathbb{Z} j ∈ Z of s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) .
(1) If j ≠ 0 j \ne0 j = 0 , then there are u , v , μ , ν ∈ C u,v,\mu,\nu\in\mathbb{C} u , v , μ , ν ∈ C and a central map ζ \zeta ζ with ζ ( x ) = 0 \zeta(x)=0 ζ ( x ) = 0 for all basis elements x ∉ { L − j , M − j } x \notin\{L_{-j},M_{-j}\} x ∈ / { L − j , M − j } such that
D = u ad L j + v ad M j + μ Ω j + ν Ψ j + ζ . D=u\operatorname{ad}L_j+v\operatorname{ad}M_j+\mu\Omega_j+\nu\Psi_j+\zeta. D = u ad L j + v ad M j + μ Ω j + ν Ψ j + ζ . (2) If j = 0 j=0 j = 0 , then there are u , λ 1 , λ 2 , λ 3 , λ ∈ C u,\lambda_1,\lambda_2,\lambda_3,\lambda\in\mathbb{C} u , λ 1 , λ 2 , λ 3 , λ ∈ C and a central map ζ \zeta ζ with ζ ( x ) = 0 \zeta(x)=0 ζ ( x ) = 0 for all basis elements x ∉ { L 0 , M 0 } x \notin\{L_0,M_0\} x ∈ / { L 0 , M 0 } such that
D = u ad L 0 + λ 1 D 1 + λ 2 D 2 + λ 3 D 3 + λ Id + ζ . D=u\operatorname{ad}L_0+\lambda_1D_1+\lambda_2D_2+\lambda_3D_3+\lambda\operatorname{Id}+\zeta. D = u ad L 0 + λ 1 D 1 + λ 2 D 2 + λ 3 D 3 + λ Id + ζ .
Proof. Since D D D and D ′ D' D ′ are even of degree j j j , we can write
D ( L m ) = a m L m + j + b m M m + j , D ( Y n ) = c n Y n + j , D ( M m ) = d m L m + j + e m M m + j , D ′ ( L m ) = a m ′ L m + j + b m ′ M m + j , D ′ ( Y n ) = c n ′ Y n + j , D ′ ( M m ) = d m ′ L m + j + e m ′ M m + j . \begin{aligned}
D(L_m)&=a_mL_{m+j}+b_mM_{m+j},\quad D(Y_n)=c_nY_{n+j},\quad D(M_m)=d_mL_{m+j}+e_mM_{m+j},\\
D'(L_m)&=a'_mL_{m+j}+b'_mM_{m+j},\quad D'(Y_n)=c'_nY_{n+j},\quad D'(M_m)=d'_mL_{m+j}+e'_mM_{m+j}.
\end{aligned} D ( L m ) D ′ ( L m ) = a m L m + j + b m M m + j , D ( Y n ) = c n Y n + j , D ( M m ) = d m L m + j + e m M m + j , = a m ′ L m + j + b m ′ M m + j , D ′ ( Y n ) = c n ′ Y n + j , D ′ ( M m ) = d m ′ L m + j + e m ′ M m + j .
where all coefficients are complex numbers. Let us apply (1) to the pairs of basis elements. For example, for the pair ( L m , Y n ) (L_m,Y_n) ( L m , Y n ) we have
[ D ( L m ) , Y n ] + [ L m , D ( Y n ) ] = ( ( m + j 2 − n ) a m + ( m 2 − n − j ) c n ) Y m + n + j [D(L_m),Y_n]+[L_m,D(Y_n)]=\left(\left(\frac{m+j}{2}-n\right)a_m+\left(\frac{m}{2}-n-j\right)c_n\right)Y_{m+n+j} [ D ( L m ) , Y n ] + [ L m , D ( Y n )] = ( ( 2 m + j − n ) a m + ( 2 m − n − j ) c n ) Y m + n + j
and D ′ ( [ L m , Y n ] ) = ( m 2 − n ) c m + n ′ Y m + n + j D'([L_m,Y_n])=\left(\frac{m}{2}-n\right)c'_{m+n}Y_{m+n+j} D ′ ([ L m , Y n ]) = ( 2 m − n ) c m + n ′ Y m + n + j . Considering in the same way the pairs ( L m , L n ) (L_m,L_n) ( L m , L n ) , ( L m , Y n ) (L_m,Y_n) ( L m , Y n ) , ( L m , M n ) (L_m,M_n) ( L m , M n ) , ( Y m , Y n ) (Y_m,Y_n) ( Y m , Y n ) , ( Y m , M n ) (Y_m,M_n) ( Y m , M n ) and ( M m , M n ) (M_m,M_n) ( M m , M n ) and comparing the coefficients at the basis elements, we obtain the following relations:
( m − n + j ) a m + ( m − n − j ) a n = ( m − n ) a m + n ′ , m , n ∈ Z , (17) (m-n+j)a_m+(m-n-j)a_n=(m-n)a'_{m+n},\qquad m,n\in\mathbb{Z}, \tag*{(17)} ( m − n + j ) a m + ( m − n − j ) a n = ( m − n ) a m + n ′ , m , n ∈ Z , (17)
( m + j ) b m − ( n + j ) b n = ( m − n ) b m + n ′ , m , n ∈ Z , (18) (m+j)b_m-(n+j)b_n=(m-n)b'_{m+n},\qquad m,n\in\mathbb{Z}, \tag*{(18)} ( m + j ) b m − ( n + j ) b n = ( m − n ) b m + n ′ , m , n ∈ Z , (18)
( m + j 2 − n ) a m + ( m 2 − n − j ) c n = ( m 2 − n ) c m + n ′ , m ∈ Z , n ∈ ε + Z , (19) \left(\frac{m+j}{2}-n\right)a_m+\left(\frac{m}{2}-n-j\right)c_n=\left(\frac{m}{2}-n\right)c'_{m+n},\qquad m\in\mathbb{Z},\ n\in\varepsilon+\mathbb{Z}, \tag*{(19)} ( 2 m + j − n ) a m + ( 2 m − n − j ) c n = ( 2 m − n ) c m + n ′ , m ∈ Z , n ∈ ε + Z , (19)
( m − n − j ) d n + n d m + n ′ = 0 , m , n ∈ Z , (20) (m-n-j)d_n+nd'_{m+n}=0,\qquad m,n\in\mathbb{Z}, \tag*{(20)} ( m − n − j ) d n + n d m + n ′ = 0 , m , n ∈ Z , (20)
( n + j ) e n + n a m = n e m + n ′ , m , n ∈ Z , (21) (n+j)e_n+na_m=ne'_{m+n},\qquad m,n\in\mathbb{Z}, \tag*{(21)} ( n + j ) e n + n a m = n e m + n ′ , m , n ∈ Z , (21)
( m − n ) d m + n ′ = 0 , ( m − n + j ) c m + ( m − n − j ) c n = ( m − n ) e m + n ′ , m , n ∈ ε + Z , (22) (m-n)d'_{m+n}=0,\qquad(m-n+j)c_m+(m-n-j)c_n=(m-n)e'_{m+n},\qquad m,n\in\varepsilon+\mathbb{Z}, \tag*{(22)} ( m − n ) d m + n ′ = 0 , ( m − n + j ) c m + ( m − n − j ) c n = ( m − n ) e m + n ′ , m , n ∈ ε + Z , (22)
( m − n + j 2 ) d n = 0 , m ∈ ε + Z , n ∈ Z , (23) \left(m-\frac{n+j}{2}\right)d_n=0,\qquad m\in\varepsilon+\mathbb{Z},\ n\in\mathbb{Z}, \tag*{(23)} ( m − 2 n + j ) d n = 0 , m ∈ ε + Z , n ∈ Z , (23)
m d n − n d m = 0 , m , n ∈ Z . (24) md_n-nd_m=0,\qquad m,n\in\mathbb{Z}. \tag*{(24)} m d n − n d m = 0 , m , n ∈ Z . (24)
Step 1. Taking two different values of m ∈ ε + Z m\in\varepsilon+\mathbb{Z} m ∈ ε + Z in (23), we get d n = 0 d_n=0 d n = 0 for all n ∈ Z n\in\mathbb{Z} n ∈ Z . Then (20) with n = 1 n=1 n = 1 gives d k ′ = 0 d'_k=0 d k ′ = 0 for all k ∈ Z k\in\mathbb{Z} k ∈ Z .
Step 2. By Lemma 12 and (17), there are α , β ∈ C \alpha,\beta\in\mathbb{C} α , β ∈ C such that
a m = α + β m , a m ′ = 2 α + β ( m + j ) for all m ∈ Z . a_m=\alpha+\beta m,\qquad a'_m=2\alpha+\beta(m+j)\quad\text{for all }m\in\mathbb{Z}. a m = α + β m , a m ′ = 2 α + β ( m + j ) for all m ∈ Z .
By Lemma 14 and (18), there are β 1 , β 2 ∈ C \beta_1,\beta_2\in\mathbb{C} β 1 , β 2 ∈ C such that b m = β 1 + β 2 m b_m=\beta_1+\beta_2m b m = β 1 + β 2 m for m ≠ − j m\ne-j m = − j and
b m ′ = β 1 + β 2 ( m + j ) for all m ∈ Z . b'_m=\beta_1+\beta_2(m+j)\quad\text{for all }m\in\mathbb{Z}. b m ′ = β 1 + β 2 ( m + j ) for all m ∈ Z .
Step 3. Setting n = 1 n=1 n = 1 in (21), we have e m + 1 ′ = α + β m + ( 1 + j ) e 1 e'_{m+1}=\alpha+\beta m+(1+j)e_1 e m + 1 ′ = α + β m + ( 1 + j ) e 1 . Hence, if we define ρ ∈ C \rho\in\mathbb{C} ρ ∈ C by 2 ρ = ( 1 + j ) ( e 1 − β ) 2\rho=(1+j)(e_1-\beta) 2 ρ = ( 1 + j ) ( e 1 − β ) , then
e k ′ = α + β ( k + j ) + 2 ρ for all k ∈ Z . (25) e'_k=\alpha+\beta(k+j)+2\rho\quad\text{for all }k\in\mathbb{Z}. \tag*{(25)} e k ′ = α + β ( k + j ) + 2 ρ for all k ∈ Z . (25)
Setting n = − j n=-j n = − j in (21) and using (25), we get 0 = j ( e m − j ′ − a m ) = 2 j ρ 0=j(e'_{m-j}-a_m)=2j\rho 0 = j ( e m − j ′ − a m ) = 2 j ρ , i.e.,
j ρ = 0. (26) j\rho=0. \tag*{(26)} j ρ = 0. (26)
Setting m = 0 m=0 m = 0 in (21) and using (25), (26), we get
( n + j ) e n = n ( β ( n + j ) + 2 ρ ) = ( n + j ) ( β n + 2 ρ ) − 2 j ρ = ( n + j ) ( β n + 2 ρ ) . (n+j)e_n=n\bigl(\beta(n+j)+2\rho\bigr)=(n+j)(\beta n+2\rho)-2j\rho=(n+j)(\beta n+2\rho). ( n + j ) e n = n ( β ( n + j ) + 2 ρ ) = ( n + j ) ( β n + 2 ρ ) − 2 j ρ = ( n + j ) ( β n + 2 ρ ) .
Hence e n = β n + 2 ρ e_n=\beta n+2\rho e n = β n + 2 ρ for all n ≠ − j n\ne-j n = − j .
Step 4. Let us set c ~ n = c n − α 2 − β n \widetilde{c}_n=c_n-\frac{\alpha}{2}-\beta n c n = c n − 2 α − β n for n ∈ ε + Z n\in\varepsilon+\mathbb{Z} n ∈ ε + Z . A direct computation shows that
( m − n + j ) ( α 2 + β m ) + ( m − n − j ) ( α 2 + β n ) = ( m − n ) ( α + β ( m + n + j ) ) . (m-n+j)\left(\frac{\alpha}{2}+\beta m\right)+(m-n-j)\left(\frac{\alpha}{2}+\beta n\right)=(m-n)(\alpha+\beta(m+n+j)). ( m − n + j ) ( 2 α + β m ) + ( m − n − j ) ( 2 α + β n ) = ( m − n ) ( α + β ( m + n + j )) .
Hence, by the second relation of (22) and (25),
( m − n + j ) c ~ m + ( m − n − j ) c ~ n = 2 ρ ( m − n ) for all m , n ∈ ε + Z . (m-n+j)\widetilde{c}_m+(m-n-j)\widetilde{c}_n=2\rho(m-n)\quad\text{for all }m,n\in\varepsilon+\mathbb{Z}. ( m − n + j ) c m + ( m − n − j ) c n = 2 ρ ( m − n ) for all m , n ∈ ε + Z .
Writing m = p + ε m=p+\varepsilon m = p + ε , n = q + ε n=q+\varepsilon n = q + ε with p , q ∈ Z p,q\in\mathbb{Z} p , q ∈ Z and applying Lemma 12 to the sequences { c ~ p + ε } p ∈ Z \{\widetilde{c}_{p+\varepsilon}\}_{p\in\mathbb{Z}} { c p + ε } p ∈ Z and { 2 ρ } k ∈ Z \{2\rho\}_{k\in\mathbb{Z}} { 2 ρ } k ∈ Z , we obtain α 0 , β 0 ∈ C \alpha_0,\beta_0\in\mathbb{C} α 0 , β 0 ∈ C such that c ~ p + ε = α 0 + β 0 p \widetilde{c}_{p+\varepsilon}=\alpha_0+\beta_0p c p + ε = α 0 + β 0 p and 2 ρ = 2 α 0 + β 0 ( k + j ) 2\rho=2\alpha_0+\beta_0(k+j) 2 ρ = 2 α 0 + β 0 ( k + j ) for all p , k ∈ Z p,k\in\mathbb{Z} p , k ∈ Z . Therefore β 0 = 0 \beta_0=0 β 0 = 0 and α 0 = ρ \alpha_0=\rho α 0 = ρ , that is,
c n = α 2 + β n + ρ for all n ∈ ε + Z . c_n=\frac{\alpha}{2}+\beta n+\rho\quad\text{for all }n\in\varepsilon+\mathbb{Z}. c n = 2 α + β n + ρ for all n ∈ ε + Z .
Now substituting a m a_m a m and c n c_n c n in (19), a direct computation gives
( n − m 2 ) ( c m + n ′ − 3 2 α − β ( m + n + j ) − ρ ) = j ρ = 0. \left(n-\frac{m}{2}\right)\left(c'_{m+n}-\frac{3}{2}\alpha-\beta(m+n+j)-\rho\right)=j\rho=0. ( n − 2 m ) ( c m + n ′ − 2 3 α − β ( m + n + j ) − ρ ) = j ρ = 0.
For every r ∈ ε + Z r\in\varepsilon+\mathbb{Z} r ∈ ε + Z we can choose n ∈ ε + Z n\in\varepsilon+\mathbb{Z} n ∈ ε + Z with r ≠ 3 n r\ne3n r = 3 n and put m = r − n m=r-n m = r − n , so that m 2 − n = r − 3 n 2 ≠ 0 \frac{m}{2}-n=\frac{r-3n}{2}\ne0 2 m − n = 2 r − 3 n = 0 . Hence c r ′ = 3 2 α + β ( r + j ) + ρ c'_r=\frac{3}{2}\alpha+\beta(r+j)+\rho c r ′ = 2 3 α + β ( r + j ) + ρ for all r ∈ ε + Z r\in\varepsilon+\mathbb{Z} r ∈ ε + Z .
Step 5. Now we compare D D D with known quasi-derivations, using (7).
Let j ≠ 0 j\ne0 j = 0 . Then ρ = 0 \rho=0 ρ = 0 by (26). Consider the quasi-derivation
D ^ = ( α + β j ) Ω j − β ad L j + β 1 j ad M j + β 2 Ψ j . \widehat{D}=(\alpha+\beta j)\Omega_j-\beta\operatorname{ad}_{L_j}+\frac{\beta_1}{j}\operatorname{ad}_{M_j}+\beta_2\Psi_j. D = ( α + β j ) Ω j − β ad L j + j β 1 ad M j + β 2 Ψ j .
Then
D ^ ( L m ) = ( α + β m ) L m + j + ( β 1 + β 2 m ) M m + j , \widehat{D}(L_m)=(\alpha+\beta m)L_{m+j}+(\beta_1+\beta_2m)M_{m+j}, D ( L m ) = ( α + β m ) L m + j + ( β 1 + β 2 m ) M m + j ,
D ^ ( Y n ) = ( α 2 + β n ) Y n + j , D ^ ( M m ) = β m M m + j . \widehat{D}(Y_n)=\left(\frac{\alpha}{2}+\beta n\right)Y_{n+j},\qquad\widehat{D}(M_m)=\beta mM_{m+j}. D ( Y n ) = ( 2 α + β n ) Y n + j , D ( M m ) = β m M m + j .
By Steps 1–4, the map ζ = D − D ^ \zeta=D-\widehat{D} ζ = D − D vanishes on all basis elements except L − j L_{-j} L − j and M − j M_{-j} M − j , and ζ ( L − j ) , ζ ( M − j ) ∈ ⟨ M 0 ⟩ \zeta(L_{-j}),\zeta(M_{-j})\in\langle M_0\rangle ζ ( L − j ) , ζ ( M − j ) ∈ ⟨ M 0 ⟩ . Hence ζ \zeta ζ is a central map and the first statement holds with u = − β u=-\beta u = − β , v = β 1 j v=\frac{\beta_1}{j} v = j β 1 , μ = α + β j \mu=\alpha+\beta j μ = α + β j , ν = β 2 \nu=\beta_2 ν = β 2 .
Let j = 0 j=0 j = 0 . Consider the quasi-derivation D ^ = α Ω 0 − β ad L 0 + β 1 D 1 + β 2 D 2 + ρ D 3 \widehat{D}=\alpha\Omega_0-\beta\operatorname{ad}_{L_0}+\beta_1D_1+\beta_2D_2+\rho D_3 D = α Ω 0 − β ad L 0 + β 1 D 1 + β 2 D 2 + ρ D 3 . Then
D ^ ( L m ) = ( α + β m ) L m + ( β 1 + β 2 m ) M m , \widehat{D}(L_m)=(\alpha+\beta m)L_m+(\beta_1+\beta_2m)M_m, D ( L m ) = ( α + β m ) L m + ( β 1 + β 2 m ) M m ,
D ^ ( Y n ) = ( α 2 + β n + ρ ) Y n and D ^ ( M m ) = ( β m + 2 ρ ) M m . \widehat{D}(Y_n)=\left(\frac{\alpha}{2}+\beta n+\rho\right)Y_n\quad\text{and}\quad\widehat{D}(M_m)=(\beta m+2\rho)M_m. D ( Y n ) = ( 2 α + β n + ρ ) Y n and D ( M m ) = ( β m + 2 ρ ) M m .
As above, ζ = D − D ^ \zeta=D-\widehat{D} ζ = D − D is a central map vanishing on all basis elements except L 0 L_0 L 0 and M 0 M_0 M 0 . Since Ω 0 = Id − 1 2 D 3 \Omega_0=\operatorname{Id}-\frac{1}{2}D_3 Ω 0 = Id − 2 1 D 3 , we get D = − β ad L 0 + β 1 D 1 + β 2 D 2 + ( ρ − α 2 ) D 3 + α Id + ζ D=-\beta\operatorname{ad}_{L_0}+\beta_1D_1+\beta_2D_2+\left(\rho-\frac{\alpha}{2}\right)D_3+\alpha\operatorname{Id}+\zeta D = − β ad L 0 + β 1 D 1 + β 2 D 2 + ( ρ − 2 α ) D 3 + α Id + ζ .
Proposition 18. Let D D D be an odd quasi-derivation of degree s ∈ ε + Z s\in\varepsilon+\mathbb{Z} s ∈ ε + Z of s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) . Then there are w , γ ∈ C w,\gamma\in\mathbb{C} w , γ ∈ C and a central map ζ \zeta ζ with ζ ( x ) = 0 \zeta(x)=0 ζ ( x ) = 0 for all basis elements x ≠ Y − s x\ne Y_{-s} x = Y − s such that
D = w ad Y s + γ Θ s + ζ . D=w\operatorname{ad}_{Y_s}+\gamma\Theta_s+\zeta. D = w ad Y s + γ Θ s + ζ .
Proof. Since D D D and D ′ D' D ′ are odd of degree s s s , we can write
D ( L m ) = f m Y m + s , D ( M m ) = g m Y m + s , D ( Y n ) = h n L n + s + k n M n + s , D(L_m)=f_mY_{m+s},\quad D(M_m)=g_mY_{m+s},\quad D(Y_n)=h_nL_{n+s}+k_nM_{n+s}, D ( L m ) = f m Y m + s , D ( M m ) = g m Y m + s , D ( Y n ) = h n L n + s + k n M n + s ,
D ′ ( L m ) = f m ′ Y m + s , D ′ ( M m ) = g m ′ Y m + s , D ′ ( Y n ) = h n ′ L n + s + k n ′ M n + s . D'(L_m)=f'_mY_{m+s},\quad D'(M_m)=g'_mY_{m+s},\quad D'(Y_n)=h'_nL_{n+s}+k'_nM_{n+s}. D ′ ( L m ) = f m ′ Y m + s , D ′ ( M m ) = g m ′ Y m + s , D ′ ( Y n ) = h n ′ L n + s + k n ′ M n + s .
Applying (1) to the pairs ( L m , L n ) (L_m,L_n) ( L m , L n ) , ( L m , Y n ) (L_m,Y_n) ( L m , Y n ) , ( L m , M n ) (L_m,M_n) ( L m , M n ) , ( Y m , Y n ) (Y_m,Y_n) ( Y m , Y n ) and ( Y m , M n ) (Y_m,M_n) ( Y m , M n ) and comparing the coefficients at the basis elements, we obtain the following relations (the pair ( M m , M n ) (M_m,M_n) ( M m , M n ) gives no relations):
( m + s − n 2 ) f m + ( m 2 − n − s ) f n = ( m − n ) f m + n ′ , m , n ∈ Z , (27) \left(m+s-\frac{n}{2}\right)f_m+\left(\frac{m}{2}-n-s\right)f_n=(m-n)f'_{m+n},\qquad m,n\in\mathbb{Z}, \tag*{(27)} ( m + s − 2 n ) f m + ( 2 m − n − s ) f n = ( m − n ) f m + n ′ , m , n ∈ Z , (27)
( m − n − s ) h n = ( m 2 − n ) h m + n ′ , m ∈ Z , n ∈ ε + Z , (28) (m-n-s)h_n=\left(\frac{m}{2}-n\right)h'_{m+n},\qquad m\in\mathbb{Z},\ n\in\varepsilon+\mathbb{Z}, \tag*{(28)} ( m − n − s ) h n = ( 2 m − n ) h m + n ′ , m ∈ Z , n ∈ ε + Z , (28)
( m + s − n ) f m − ( n + s ) k n = ( m 2 − n ) k m + n ′ , m ∈ Z , n ∈ ε + Z , (29) (m+s-n)f_m-(n+s)k_n=\left(\frac{m}{2}-n\right)k'_{m+n},\qquad m\in\mathbb{Z},\ n\in\varepsilon+\mathbb{Z}, \tag*{(29)} ( m + s − n ) f m − ( n + s ) k n = ( 2 m − n ) k m + n ′ , m ∈ Z , n ∈ ε + Z , (29)
( m 2 − n − s ) g n = − n g m + n ′ , m , n ∈ Z , (30) \left(\frac{m}{2}-n-s\right)g_n=-ng'_{m+n},\qquad m,n\in\mathbb{Z}, \tag*{(30)} ( 2 m − n − s ) g n = − n g m + n ′ , m , n ∈ Z , (30)
( m + s 2 − n ) h m + ( m − n + s 2 ) h n = ( m − n ) g m + n ′ , m , n ∈ ε + Z , (31) \left(\frac{m+s}{2}-n\right)h_m+\left(m-\frac{n+s}{2}\right)h_n=(m-n)g'_{m+n},\qquad m,n\in\varepsilon+\mathbb{Z}, \tag*{(31)} ( 2 m + s − n ) h m + ( m − 2 n + s ) h n = ( m − n ) g m + n ′ , m , n ∈ ε + Z , (31)
( m − n − s ) g n − n h m = 0 , m ∈ ε + Z , n ∈ Z . (32) (m-n-s)g_n-nh_m=0,\qquad m\in\varepsilon+\mathbb{Z},\ n\in\mathbb{Z}. \tag*{(32)} ( m − n − s ) g n − n h m = 0 , m ∈ ε + Z , n ∈ Z . (32)
Step 1. Setting n = 0 n=0 n = 0 in (32), we have ( m − s ) g 0 = 0 (m-s)g_0=0 ( m − s ) g 0 = 0 for all m ∈ ε + Z m\in\varepsilon+\mathbb{Z} m ∈ ε + Z , so g 0 = 0 g_0=0 g 0 = 0 . Setting n = 1 n=1 n = 1 and n = 2 n=2 n = 2 in (32), we have h m = ( m − 1 − s ) g 1 h_m=(m-1-s)g_1 h m = ( m − 1 − s ) g 1 and 2 h m = ( m − 2 − s ) g 2 2h_m=(m-2-s)g_2 2 h m = ( m − 2 − s ) g 2 . Hence
2 ( m − 1 − s ) g 1 = ( m − 2 − s ) g 2 for all m ∈ ε + Z . 2(m-1-s)g_1=(m-2-s)g_2\quad\text{for all }m\in\varepsilon+\mathbb{Z}. 2 ( m − 1 − s ) g 1 = ( m − 2 − s ) g 2 for all m ∈ ε + Z .
Comparing the coefficients at m m m and the constant terms of these polynomials of degree at most one in m m m (which coincide at infinitely many points), we get 2 g 1 = g 2 2g_1=g_2 2 g 1 = g 2 and 2 ( 1 + s ) g 1 = ( 2 − s ) g 2 = 2 ( 2 + s ) g 1 2(1+s)g_1=(2-s)g_2=2(2+s)g_1 2 ( 1 + s ) g 1 = ( 2 − s ) g 2 = 2 ( 2 + s ) g 1 , so g 1 = 0 g_1=0 g 1 = 0 . Therefore h m = 0 h_m=0 h m = 0 for all m ∈ ε + Z m\in\varepsilon+\mathbb{Z} m ∈ ε + Z . Now (32) gives ( m − n − s ) g n = 0 (m-n-s)g_n=0 ( m − n − s ) g n = 0 for all m ∈ ε + Z m\in\varepsilon+\mathbb{Z} m ∈ ε + Z , whence g n = 0 g_n=0 g n = 0 for all n ∈ Z n\in\mathbb{Z} n ∈ Z .
Step 2. Since h n = 0 h_n=0 h n = 0 , the relation (28) gives ( m 2 − n ) h m + n ′ = 0 \left(\frac{m}{2}-n\right)h'_{m+n}=0 ( 2 m − n ) h m + n ′ = 0 . For every r ∈ ε + Z r\in\varepsilon+\mathbb{Z} r ∈ ε + Z we choose n ∈ ε + Z n\in\varepsilon+\mathbb{Z} n ∈ ε + Z with r ≠ 3 n r\ne3n r = 3 n and put m = r − n m=r-n m = r − n ; then m 2 − n ≠ 0 \frac{m}{2}-n\ne0 2 m − n = 0 and h r ′ = 0 h'_r=0 h r ′ = 0 . Similarly, (31) gives ( m − n ) g m + n ′ = 0 (m-n)g'_{m+n}=0 ( m − n ) g m + n ′ = 0 for all m , n ∈ ε + Z m,n\in\varepsilon+\mathbb{Z} m , n ∈ ε + Z , and since every k ∈ Z k\in\mathbb{Z} k ∈ Z can be written as k = m + n k=m+n k = m + n with m ≠ n m\ne n m = n , m , n ∈ ε + Z m,n\in\varepsilon+\mathbb{Z} m , n ∈ ε + Z , we get g k ′ = 0 g'_k=0 g k ′ = 0 for all k ∈ Z k\in\mathbb{Z} k ∈ Z . Now (30) holds trivially. Step 3. By Lemma 13 and (27), there are ϕ , ψ ∈ C \phi,\psi\in\mathbb{C} ϕ , ψ ∈ C such that
f m = ϕ + ψ m , f m ′ = 3 2 ϕ + ψ ( m + s ) for all m ∈ Z . f_m=\phi+\psi m,\qquad f'_m=\frac{3}{2}\phi+\psi(m+s)\quad\text{for all }m\in\mathbb{Z}. f m = ϕ + ψ m , f m ′ = 2 3 ϕ + ψ ( m + s ) for all m ∈ Z .
Step 4. A direct computation shows that
( m + s − n ) ( ϕ + ψ m ) − ( n + s ) ( ϕ + 2 ψ n ) = ( m 2 − n ) ( 2 ϕ + 2 ψ ( m + n + s ) ) . (m+s-n)(\phi+\psi m)-(n+s)(\phi+2\psi n)=\left(\frac{m}{2}-n\right)(2\phi+2\psi(m+n+s)). ( m + s − n ) ( ϕ + ψ m ) − ( n + s ) ( ϕ + 2 ψ n ) = ( 2 m − n ) ( 2 ϕ + 2 ψ ( m + n + s )) .
Hence (29) can be written as
( n + s ) ( k n − ϕ − 2 ψ n ) = ( n − m 2 ) ( k m + n ′ − 2 ϕ − 2 ψ ( m + n + s ) ) . (n+s)(k_n-\phi-2\psi n)=\left(n-\frac{m}{2}\right)(k'_{m+n}-2\phi-2\psi(m+n+s)). ( n + s ) ( k n − ϕ − 2 ψ n ) = ( n − 2 m ) ( k m + n ′ − 2 ϕ − 2 ψ ( m + n + s )) .
Taking m = 2 n ∈ Z m=2n\in\mathbb{Z} m = 2 n ∈ Z , we get ( n + s ) ( k n − ϕ − 2 ψ n ) = 0 (n+s)(k_n-\phi-2\psi n)=0 ( n + s ) ( k n − ϕ − 2 ψ n ) = 0 , i.e., k n = ϕ + 2 ψ n k_n=\phi+2\psi n k n = ϕ + 2 ψ n for all n ≠ − s n\ne-s n = − s . Then the right hand side vanishes for all m ∈ Z m\in\mathbb{Z} m ∈ Z , n ∈ ε + Z n\in\varepsilon+\mathbb{Z} n ∈ ε + Z , and, as in Step 2, we obtain k r ′ = 2 ϕ + 2 ψ ( r + s ) k'_r=2\phi+2\psi(r+s) k r ′ = 2 ϕ + 2 ψ ( r + s ) for all r ∈ ε + Z r\in\varepsilon+\mathbb{Z} r ∈ ε + Z .
Step 5. Let γ = ϕ + 2 ψ s \gamma=\phi+2\psi s γ = ϕ + 2 ψ s and w = − 2 ψ w=-2\psi w = − 2 ψ . By (7),
( w ad Y s + γ Θ s ) ( L m ) = ( w ( s − m 2 ) + γ ) Y m + s = ( ϕ + ψ m ) Y m + s , (w\operatorname{ad}Y_s+\gamma\Theta_s)(L_m)=\left(w\left(s-\frac{m}{2}\right)+\gamma\right)Y_{m+s}=(\phi+\psi m)Y_{m+s}, ( w ad Y s + γ Θ s ) ( L m ) = ( w ( s − 2 m ) + γ ) Y m + s = ( ϕ + ψ m ) Y m + s ,
( w ad Y s + γ Θ s ) ( Y n ) = ( w ( s − n ) + γ ) M n + s = ( ϕ + 2 ψ n ) M n + s , (w\operatorname{ad}Y_s+\gamma\Theta_s)(Y_n)=(w(s-n)+\gamma)M_{n+s}=(\phi+2\psi n)M_{n+s}, ( w ad Y s + γ Θ s ) ( Y n ) = ( w ( s − n ) + γ ) M n + s = ( ϕ + 2 ψ n ) M n + s ,
and ( w ad Y s + γ Θ s ) ( M m ) = 0 (w\operatorname{ad}Y_s+\gamma\Theta_s)(M_m)=0 ( w ad Y s + γ Θ s ) ( M m ) = 0 . By Steps 1–4, the map ζ = D − w ad Y s − γ Θ s \zeta=D-w\operatorname{ad}Y_s-\gamma\Theta_s ζ = D − w ad Y s − γ Θ s vanishes on all basis elements except Y − s Y_{-s} Y − s , and ζ ( Y − s ) = ( k − s − ϕ + 2 ψ s ) M 0 ∈ ⟨ M 0 ⟩ \zeta(Y_{-s})=(k_{-s}-\phi+2\psi s)M_0\in\langle M_0\rangle ζ ( Y − s ) = ( k − s − ϕ + 2 ψ s ) M 0 ∈ ⟨ M 0 ⟩ . Hence ζ \zeta ζ is a central map.
Now we are ready to formulate and prove the main result of the paper.
Theorem 19. A linear map D D D of s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) is a quasi-derivation if and only if there are a derivation d ∈ Der ( s v ( ε ) ) d\in\operatorname{Der}(\mathfrak{sv}(\varepsilon)) d ∈ Der ( sv ( ε )) , a complex number λ \lambda λ , three finitely supported families of complex numbers { α t } t ∈ Z ∖ { 0 } \{\alpha_t\}_{t\in\mathbb{Z}\setminus\{0\}} { α t } t ∈ Z ∖ { 0 } , { β t } t ∈ Z ∖ { 0 } \{\beta_t\}_{t\in\mathbb{Z}\setminus\{0\}} { β t } t ∈ Z ∖ { 0 } , { γ s } s ∈ ε + Z \{\gamma_s\}_{s\in\varepsilon+\mathbb{Z}} { γ s } s ∈ ε + Z and a central map ζ : s v ( ε ) → ⟨ M 0 ⟩ \zeta:\mathfrak{sv}(\varepsilon)\to\langle M_0\rangle ζ : sv ( ε ) → ⟨ M 0 ⟩ such that
D = d + λ Id + ∑ t ≠ 0 α t Ω t + ∑ t ≠ 0 β t Ψ t + ∑ s ∈ ε + Z γ s Θ s + ζ . (33) D=d+\lambda\operatorname{Id}+\sum_{t\ne0}\alpha_t\Omega_t+\sum_{t\ne0}\beta_t\Psi_t+\sum_{s\in\varepsilon+\mathbb{Z}}\gamma_s\Theta_s+\zeta. \tag*{(33)} D = d + λ Id + t = 0 ∑ α t Ω t + t = 0 ∑ β t Ψ t + s ∈ ε + Z ∑ γ s Θ s + ζ . (33) Explicitly, this means that
D ( L m ) = d ( L m ) + λ L m + ∑ t ≠ 0 α t L m + t + ∑ t ≠ 0 β t m M m + t + ∑ s ∈ ε + Z γ s Y m + s + ζ ( L m ) , D(L_m)=d(L_m)+\lambda L_m+\sum_{t\ne0}\alpha_tL_{m+t}+\sum_{t\ne0}\beta_t mM_{m+t}+\sum_{s\in\varepsilon+\mathbb{Z}}\gamma_sY_{m+s}+\zeta(L_m), D ( L m ) = d ( L m ) + λ L m + t = 0 ∑ α t L m + t + t = 0 ∑ β t m M m + t + s ∈ ε + Z ∑ γ s Y m + s + ζ ( L m ) , D ( Y n ) = d ( Y n ) + λ Y n + 1 2 ∑ t ≠ 0 α t Y n + t + ∑ s ∈ ε + Z γ s M n + s + ζ ( Y n ) , D(Y_n)=d(Y_n)+\lambda Y_n+\frac{1}{2}\sum_{t\ne0}\alpha_tY_{n+t}+\sum_{s\in\varepsilon+\mathbb{Z}}\gamma_sM_{n+s}+\zeta(Y_n), D ( Y n ) = d ( Y n ) + λ Y n + 2 1 t = 0 ∑ α t Y n + t + s ∈ ε + Z ∑ γ s M n + s + ζ ( Y n ) , D ( M m ) = d ( M m ) + λ M m + ζ ( M m ) . D(M_m)=d(M_m)+\lambda M_m+\zeta(M_m). D ( M m ) = d ( M m ) + λ M m + ζ ( M m ) . The decomposition (33) is unique, and the map associated with D D D is
D ′ = d + 2 λ Id + ∑ t ≠ 0 α t Ω t ′ + ∑ t ≠ 0 β t Ψ t ′ + ∑ s ∈ ε + Z γ s Θ s ′ . D'=d+2\lambda\operatorname{Id}+\sum_{t\ne0}\alpha_t\Omega'_t+\sum_{t\ne0}\beta_t\Psi'_t+\sum_{s\in\varepsilon+\mathbb{Z}}\gamma_s\Theta'_s. D ′ = d + 2 λ Id + t = 0 ∑ α t Ω t ′ + t = 0 ∑ β t Ψ t ′ + s ∈ ε + Z ∑ γ s Θ s ′ .
Proof. If D D D is of the form (33), then D D D is a quasi-derivation with the indicated associated map by Lemmas 6 and 16 (all sums are finite).
Conversely, let D D D be a quasi-derivation. By Proposition 17, for every j ∈ Z ∖ { 0 } j \in\mathbb{Z} \setminus\{0\} j ∈ Z ∖ { 0 } there are u j , v j , μ j , ν j ∈ C u_j,v_j,\mu_j,\nu_j \in\mathbb{C} u j , v j , μ j , ν j ∈ C and a central map ζ j \zeta_j ζ j such that the even component of degree j j j of D D D is equal to u j ad L j + v j ad M j + μ j Ω j + ν j Ψ j + ζ j u_j\operatorname{ad}L_j+v_j\operatorname{ad}M_j+\mu_j\Omega_j+\nu_j\Psi_j+\zeta_j u j ad L j + v j ad M j + μ j Ω j + ν j Ψ j + ζ j , and there are u 0 , λ 1 , λ 2 , λ 3 , λ ∈ C u_0,\lambda_1,\lambda_2,\lambda_3,\lambda\in\mathbb{C} u 0 , λ 1 , λ 2 , λ 3 , λ ∈ C and a central map ζ 0 \zeta_0 ζ 0 such that the even component of degree 0 of D D D is equal to
u 0 ad L 0 + λ 1 D 1 + λ 2 D 2 + λ 3 D 3 + λ Id + ζ 0 . u_0\operatorname{ad}L_0+\lambda_1D_1+\lambda_2D_2+\lambda_3D_3+\lambda\operatorname{Id}+\zeta_0. u 0 ad L 0 + λ 1 D 1 + λ 2 D 2 + λ 3 D 3 + λ Id + ζ 0 .
By Proposition 18, for every s ∈ ε + Z s \in\varepsilon+\mathbb{Z} s ∈ ε + Z there are w s , γ s ∈ C w_s,\gamma_s \in\mathbb{C} w s , γ s ∈ C and a central map ζ s ′ \zeta'_s ζ s ′ such that the odd component of degree s s s of D D D is equal to w s ad Y s + γ s Θ s + ζ s ′ w_s\operatorname{ad}Y_s+\gamma_s\Theta_s+\zeta'_s w s ad Y s + γ s Θ s + ζ s ′ .
Let us show that only finitely many of the numbers u j , v j , μ j , ν j , w s , γ s u_j,v_j,\mu_j,\nu_j,w_s,\gamma_s u j , v j , μ j , ν j , w s , γ s are nonzero. By (7), for j ≠ 0 j \ne0 j = 0 and m + j ≠ 0 m+j \ne0 m + j = 0 the coefficients of D ( L m ) D(L_m) D ( L m ) at L m + j L_{m+j} L m + j and at M m + j M_{m+j} M m + j are equal to
u j ( j − m ) + μ j and v j j + ν j m , u_j(j-m)+\mu_j \qquad\text{and}\qquad v_jj+\nu_jm, u j ( j − m ) + μ j and v j j + ν j m ,
respectively, and the coefficient of D ( L m ) D(L_m) D ( L m ) at Y m + s Y_{m+s} Y m + s is equal to w s ( s − m 2 ) + γ s w_s\left(s-\frac{m}{2}\right)+\gamma_s w s ( s − 2 m ) + γ s . Since D ( L 0 ) D(L_0) D ( L 0 ) , D ( L 1 ) D(L_1) D ( L 1 ) and D ( L 2 ) D(L_2) D ( L 2 ) are finite linear combinations of basis elements, for all but finitely many j j j and s s s we have
u j j + μ j = u j ( j − 1 ) + μ j = 0 , v j j + ν j = v j j + 2 ν j = 0 and w s s + γ s = w s ( s − 1 ) + γ s = 0 u_jj+\mu_j=u_j(j-1)+\mu_j=0,\quad v_jj+\nu_j=v_jj+2\nu_j=0\quad\text{and}\quad w_ss+\gamma_s=w_s(s-1)+\gamma_s=0 u j j + μ j = u j ( j − 1 ) + μ j = 0 , v j j + ν j = v j j + 2 ν j = 0 and w s s + γ s = w s ( s − 1 ) + γ s = 0
(we use m = 0 , 1 m=0,1 m = 0 , 1 for the first and the last pair of equalities and m = 1 , 2 m=1,2 m = 1 , 2 for the second one), which gives u j = μ j = v j = ν j = 0 u_j=\mu_j=v_j=\nu_j=0 u j = μ j = v j = ν j = 0 and w s = γ s = 0 w_s=\gamma_s=0 w s = γ s = 0 for all but finitely many j j j and s s s .
Consequently, x = ∑ j ( u j L j + v j M j ) + ∑ s w s Y s x=\sum_j(u_jL_j+v_jM_j)+\sum_s w_sY_s x = ∑ j ( u j L j + v j M j ) + ∑ s w s Y s (with v 0 = 0 v_0=0 v 0 = 0 ) is an element of s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) , and d = ad x + λ 1 D 1 + λ 2 D 2 + λ 3 D 3 d=\operatorname{ad}x+\lambda_1D_1+\lambda_2D_2+\lambda_3D_3 d = ad x + λ 1 D 1 + λ 2 D 2 + λ 3 D 3 is a derivation. The map
ζ = D − d − λ Id − ∑ t ≠ 0 μ t Ω t − ∑ t ≠ 0 ν t Ψ t − ∑ s γ s Θ s \zeta=D-d-\lambda\operatorname{Id}-\sum_{t\ne0}\mu_t\Omega_t-\sum_{t\ne0}\nu_t\Psi_t-\sum_s\gamma_s\Theta_s ζ = D − d − λ Id − t = 0 ∑ μ t Ω t − t = 0 ∑ ν t Ψ t − s ∑ γ s Θ s
is linear, and for every basis element y y y we have ζ ( y ) = ∑ j ζ j ( y ) + ∑ s ζ s ′ ( y ) ∈ ⟨ M 0 ⟩ \zeta(y)=\sum_j\zeta_j(y)+\sum_s\zeta'_s(y)\in\langle M_0\rangle ζ ( y ) = ∑ j ζ j ( y ) + ∑ s ζ s ′ ( y ) ∈ ⟨ M 0 ⟩ . Hence ζ \zeta ζ is a central map and (33) holds with α t = μ t \alpha_t=\mu_t α t = μ t and β t = ν t \beta_t=\nu_t β t = ν t .
Let us prove the uniqueness. Assume that
d + λ Id + ∑ t ≠ 0 α t Ω t + ∑ t ≠ 0 β t Ψ t + ∑ s γ s Θ s + ζ = 0. d+\lambda\operatorname{Id}+\sum_{t\ne0}\alpha_t\Omega_t+\sum_{t\ne0}\beta_t\Psi_t+\sum_s\gamma_s\Theta_s+\zeta=0. d + λ Id + t = 0 ∑ α t Ω t + t = 0 ∑ β t Ψ t + s ∑ γ s Θ s + ζ = 0.
By Lemmas 6 and 9, the map associated with the left hand side is zero, i.e.,
d + 2 λ Id + ∑ t ≠ 0 α t Ω t ′ + ∑ t ≠ 0 β t Ψ t ′ + ∑ s γ s Θ s ′ = 0. d+2\lambda\operatorname{Id}+\sum_{t\ne0}\alpha_t\Omega'_t+\sum_{t\ne0}\beta_t\Psi'_t+\sum_s\gamma_s\Theta'_s=0. d + 2 λ Id + t = 0 ∑ α t Ω t ′ + t = 0 ∑ β t Ψ t ′ + s ∑ γ s Θ s ′ = 0.
Subtracting the first equality from the second one, we get
λ Id + ∑ t ≠ 0 α t ( Ω t ′ − Ω t ) + ∑ t ≠ 0 β t ( Ψ t ′ − Ψ t ) + ∑ s γ s ( Θ s ′ − Θ s ) − ζ = 0. \lambda\operatorname{Id}+\sum_{t\ne0}\alpha_t(\Omega'_t-\Omega_t)+\sum_{t\ne0}\beta_t(\Psi'_t-\Psi_t)+\sum_s\gamma_s(\Theta'_s-\Theta_s)-\zeta=0. λ Id + t = 0 ∑ α t ( Ω t ′ − Ω t ) + t = 0 ∑ β t ( Ψ t ′ − Ψ t ) + s ∑ γ s ( Θ s ′ − Θ s ) − ζ = 0.
Applying this map to M n M_n M n with n ≠ 0 n\ne0 n = 0 and n ≠ − t n\ne-t n = − t for all t t t with α t ≠ 0 \alpha_t\ne0 α t = 0 , we obtain λ M n + ∑ t ≠ 0 α t M n + t ∈ ⟨ M 0 ⟩ \lambda M_n+\sum_{t\ne0}\alpha_tM_{n+t}\in\langle M_0\rangle λ M n + ∑ t = 0 α t M n + t ∈ ⟨ M 0 ⟩ , whence λ = 0 \lambda=0 λ = 0 and α t = 0 \alpha_t=0 α t = 0 for all t t t . Applying it to L m L_m L m , we obtain ∑ t ≠ 0 β t t M m + t + 1 2 ∑ s γ s Y m + s ∈ ⟨ M 0 ⟩ \sum_{t\ne0}\beta_t tM_{m+t}+\frac{1}{2}\sum_s\gamma_sY_{m+s}\in\langle M_0\rangle ∑ t = 0 β t t M m + t + 2 1 ∑ s γ s Y m + s ∈ ⟨ M 0 ⟩ for all m ∈ Z m\in\mathbb{Z} m ∈ Z , whence β t = 0 \beta_t=0 β t = 0 and γ s = 0 \gamma_s=0 γ s = 0 for all t t t and s s s . Then d + ζ = 0 d+\zeta=0 d + ζ = 0 and, by the second equality, d = 0 d=0 d = 0 ; hence ζ = 0 \zeta=0 ζ = 0 .
Corollary 20. Let Q ( s v ( ϵ ) ) \mathcal{Q}(\mathfrak{sv}(\epsilon)) Q ( sv ( ϵ )) be the space of all finite linear combinations of the maps Ω t , Ψ t \Omega_t,\Psi_t Ω t , Ψ t (t ∈ Z ∖ { 0 } t\in\mathbb{Z}\setminus\{0\} t ∈ Z ∖ { 0 } ) and Θ s \Theta_s Θ s (s ∈ ϵ + Z s\in\epsilon+\mathbb{Z} s ∈ ϵ + Z ). Then
QDer ( s v ( ϵ ) ) = Der ( s v ( ϵ ) ) ⊕ Δ 1 2 ( s v ( ϵ ) ) ⊕ Q ( s v ( ϵ ) ) ⊕ Hom ( s v ( ϵ ) , Z ( s v ( ϵ ) ) ) . \operatorname{QDer}(\mathfrak{sv}(\epsilon))=\operatorname{Der}(\mathfrak{sv}(\epsilon))\oplus\Delta_{\frac{1}{2}}(\mathfrak{sv}(\epsilon))\oplus\mathcal{Q}(\mathfrak{sv}(\epsilon))\oplus\operatorname{Hom}(\mathfrak{sv}(\epsilon),Z(\mathfrak{sv}(\epsilon))). QDer ( sv ( ϵ )) = Der ( sv ( ϵ )) ⊕ Δ 2 1 ( sv ( ϵ )) ⊕ Q ( sv ( ϵ )) ⊕ Hom ( sv ( ϵ ) , Z ( sv ( ϵ ))) . In particular, QDer ( s v ( ϵ ) ) \operatorname{QDer}(\mathfrak{sv}(\epsilon)) QDer ( sv ( ϵ )) is strictly larger than Der ( s v ( ϵ ) ) + Δ 1 2 ( s v ( ϵ ) ) + Hom ( s v ( ϵ ) , Z ( s v ( ϵ ) ) ) \operatorname{Der}(\mathfrak{sv}(\epsilon))+\Delta_{\frac{1}{2}}(\mathfrak{sv}(\epsilon))+\operatorname{Hom}(\mathfrak{sv}(\epsilon),Z(\mathfrak{sv}(\epsilon))) Der ( sv ( ϵ )) + Δ 2 1 ( sv ( ϵ )) + Hom ( sv ( ϵ ) , Z ( sv ( ϵ ))) .
Proof. This follows from Theorems 19 and 11 , Lemma 9 and the uniqueness of the decomposition (33).
Corollaries
Corollary 21. Let ϵ ∈ { 0 , 1 2 } \epsilon\in\{0,\frac{1}{2}\} ϵ ∈ { 0 , 2 1 } and δ ∈ C \delta\in\mathbb{C} δ ∈ C . Then
Δ δ ( s v ( ϵ ) ) = { Der ( s v ( ϵ ) ) , δ = 1 , ⟨ Id ⟩ , δ = 1 2 , 0 , δ ≠ 1 2 , 1. \Delta_\delta(\mathfrak{sv}(\epsilon))=
\begin{cases}
\operatorname{Der}(\mathfrak{sv}(\epsilon)), & \delta=1,\\
\langle\operatorname{Id}\rangle, & \delta=\frac{1}{2},\\
0, & \delta\ne\frac{1}{2},1.
\end{cases} Δ δ ( sv ( ϵ )) = ⎩ ⎨ ⎧ Der ( sv ( ϵ )) , ⟨ Id ⟩ , 0 , δ = 1 , δ = 2 1 , δ = 2 1 , 1. In other words, s v ( ϵ ) \mathfrak{sv}(\epsilon) sv ( ϵ ) has no nonzero δ \delta δ -derivations for δ ∉ { 1 2 , 1 } \delta\notin\{\frac{1}{2},1\} δ ∈ / { 2 1 , 1 } , every 1 2 \frac{1}{2} 2 1 -derivation of s v ( ϵ ) \mathfrak{sv}(\epsilon) sv ( ϵ ) is trivial (cf. Theorem 11 and [37 ] ), and the 1 1 1 -derivations are the derivations described in Theorem 10 .
Proof. The case δ = 1 \delta=1 δ = 1 is the definition. If δ = 0 \delta=0 δ = 0 , then a 0 0 0 -derivation vanishes on [ s v ( ϵ ) , s v ( ϵ ) ] = s v ( ϵ ) [\mathfrak{sv}(\epsilon),\mathfrak{sv}(\epsilon)]=\mathfrak{sv}(\epsilon) [ sv ( ϵ ) , sv ( ϵ )] = sv ( ϵ ) (Lemma 9 ). Let δ ∉ { 0 , 1 } \delta\notin\{0,1\} δ ∈ / { 0 , 1 } and put c = δ − 1 ≠ 1 c=\delta^{-1}\ne1 c = δ − 1 = 1 . A linear map φ \varphi φ is a δ \delta δ -derivation if and only if it is a quasi-derivation with φ ′ = c φ \varphi'=c\varphi φ ′ = c φ , and by Lemma 7 every homogeneous component D D D of φ \varphi φ is a quasi-derivation with D ′ = c D D'=cD D ′ = cD . We use Propositions 17 and 18 together with Lemmas 6 and 16 .
Let D D D be even of degree j ≠ 0 j\ne0 j = 0 , so that
D = u ad L j + v ad M j + μ Ω j + ν Ψ j + ζ and D ′ = u ad L j + v ad M j + μ Ω j ′ + ν Ψ j ′ . D=u\operatorname{ad}L_j+v\operatorname{ad}M_j+\mu\Omega_j+\nu\Psi_j+\zeta
\quad\text{and}\quad
D'=u\operatorname{ad}L_j+v\operatorname{ad}M_j+\mu\Omega'_j+\nu\Psi'_j. D = u ad L j + v ad M j + μ Ω j + ν Ψ j + ζ and D ′ = u ad L j + v ad M j + μ Ω j ′ + ν Ψ j ′ . Comparing D ′ ( M n ) D'(M_n) D ′ ( M n ) and c D ( M n ) cD(M_n) cD ( M n ) for n ≠ − j n\ne-j n = − j , we get − u n + μ = − c u n -un+\mu=-cun − u n + μ = − c u n for all n ≠ − j n\ne-j n = − j , so u = μ = 0 u=\mu=0 u = μ = 0 . Comparing the coefficients at M m + j M_{m+j} M m + j in D ′ ( L m ) D'(L_m) D ′ ( L m ) and c D ( L m ) cD(L_m) cD ( L m ) for m ≠ − j m\ne-j m = − j , we get v j + ν ( m + j ) = c ( v j + ν m ) vj+\nu(m+j)=c(vj+\nu m) v j + ν ( m + j ) = c ( v j + ν m ) for all m ≠ − j m\ne-j m = − j , so ν = 0 \nu=0 ν = 0 and v = 0 v=0 v = 0 . Then D = ζ D=\zeta D = ζ and 0 = D ′ = c ζ 0=D'=c\zeta 0 = D ′ = c ζ , so D = 0 D=0 D = 0 .
Let D D D be even of degree 0 0 0 , so that
D = u ad L 0 + λ 1 D 1 + λ 2 D 2 + λ 3 D 3 + λ Id + ζ and D ′ = u ad L 0 + λ 1 D 1 + λ 2 D 2 + λ 3 D 3 + 2 λ Id . D=u\operatorname{ad}L_0+\lambda_1D_1+\lambda_2D_2+\lambda_3D_3+\lambda\operatorname{Id}+\zeta
\quad\text{and}\quad
D'=u\operatorname{ad}L_0+\lambda_1D_1+\lambda_2D_2+\lambda_3D_3+2\lambda\operatorname{Id}. D = u ad L 0 + λ 1 D 1 + λ 2 D 2 + λ 3 D 3 + λ Id + ζ and D ′ = u ad L 0 + λ 1 D 1 + λ 2 D 2 + λ 3 D 3 + 2 λ Id . Comparing D ′ ( L m ) D'(L_m) D ′ ( L m ) and c D ( L m ) cD(L_m) cD ( L m ) for m ≠ 0 m\ne0 m = 0 , we get
− u m + 2 λ = c ( − u m + λ ) and λ 1 + λ 2 m = c ( λ 1 + λ 2 m ) , -um+2\lambda=c(-um+\lambda)
\quad\text{and}\quad
\lambda_1+\lambda_2m=c(\lambda_1+\lambda_2m), − u m + 2 λ = c ( − u m + λ ) and λ 1 + λ 2 m = c ( λ 1 + λ 2 m ) , so u = λ 1 = λ 2 = 0 u=\lambda_1=\lambda_2=0 u = λ 1 = λ 2 = 0 and ( 2 − c ) λ = 0 (2-c)\lambda=0 ( 2 − c ) λ = 0 . Comparing D ′ ( Y n ) D'(Y_n) D ′ ( Y n ) and c D ( Y n ) cD(Y_n) cD ( Y n ) , we get λ 3 + 2 λ = c ( λ 3 + λ ) \lambda_3+2\lambda=c(\lambda_3+\lambda) λ 3 + 2 λ = c ( λ 3 + λ ) , whence ( 1 − c ) λ 3 = 0 (1-c)\lambda_3=0 ( 1 − c ) λ 3 = 0 and λ 3 = 0 \lambda_3=0 λ 3 = 0 . Then D = λ Id + ζ D=\lambda\operatorname{Id}+\zeta D = λ Id + ζ , and D ′ = 2 λ Id = c D D'=2\lambda\operatorname{Id}=cD D ′ = 2 λ Id = cD gives c ζ = ( 2 − c ) λ Id = 0 c\zeta=(2-c)\lambda\operatorname{Id}=0 c ζ = ( 2 − c ) λ Id = 0 , so ζ = 0 \zeta=0 ζ = 0 and D = λ Id D=\lambda\operatorname{Id} D = λ Id with λ = 0 \lambda=0 λ = 0 unless c = 2 c=2 c = 2 , i.e., δ = 1 2 \delta=\frac{1}{2} δ = 2 1 .
Let D D D be odd of degree s s s , so that
D = w ad Y s + γ Θ s + ζ and D ′ = w ad Y s + γ Θ s ′ . D=w\operatorname{ad}Y_s+\gamma\Theta_s+\zeta
\quad\text{and}\quad
D'=w\operatorname{ad}Y_s+\gamma\Theta'_s. D = w ad Y s + γ Θ s + ζ and D ′ = w ad Y s + γ Θ s ′ . Comparing D ′ ( L m ) D'(L_m) D ′ ( L m ) and c D ( L m ) cD(L_m) cD ( L m ) , we get w ( s − m 2 ) + 3 2 γ = c ( w ( s − m 2 ) + γ ) w\left(s-\frac{m}{2}\right)+\frac{3}{2}\gamma=c\left(w\left(s-\frac{m}{2}\right)+\gamma\right) w ( s − 2 m ) + 2 3 γ = c ( w ( s − 2 m ) + γ ) for all m ∈ Z m\in\mathbb{Z} m ∈ Z , and comparing D ′ ( Y n ) D'(Y_n) D ′ ( Y n ) and c D ( Y n ) cD(Y_n) cD ( Y n ) for n ≠ − s n\ne-s n = − s , we get w ( s − n ) + 2 γ = c ( w ( s − n ) + γ ) w(s-n)+2\gamma=c(w(s-n)+\gamma) w ( s − n ) + 2 γ = c ( w ( s − n ) + γ ) . Hence w = 0 w=0 w = 0 , ( 3 2 − c ) γ = ( 2 − c ) γ = 0 \left(\frac{3}{2}-c\right)\gamma=(2-c)\gamma=0 ( 2 3 − c ) γ = ( 2 − c ) γ = 0 , so γ = 0 \gamma=0 γ = 0 . Then D = ζ D=\zeta D = ζ and 0 = D ′ = c ζ 0=D'=c\zeta 0 = D ′ = c ζ , so D = 0 D=0 D = 0 .
Since φ \varphi φ is the sum of its homogeneous components, φ = 0 \varphi=0 φ = 0 if δ ≠ 1 2 \delta\ne\frac{1}{2} δ = 2 1 and φ ∈ ⟨ Id ⟩ \varphi\in\langle\operatorname{Id}\rangle φ ∈ ⟨ Id ⟩ if δ = 1 2 \delta=\frac{1}{2} δ = 2 1 . Conversely, Id \operatorname{Id} Id is a 1 2 \frac{1}{2} 2 1 -derivation.
The description of biderivations of s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) is given in the following theorem.
Theorem 22 (see [26 ] ). A bilinear map ϕ \phi ϕ of s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) is a biderivation if and only if there are λ ∈ C \lambda\in\mathbb{C} λ ∈ C and complex numbers { μ i } i ∈ Z \{\mu_i\}_{i\in\mathbb{Z}} { μ i } i ∈ Z , only finitely many of which are nonzero, such that ϕ ( x , y ) = λ [ x , y ] + χ ( x , y ) \phi(x,y)=\lambda[x,y]+\chi(x,y) ϕ ( x , y ) = λ [ x , y ] + χ ( x , y ) for all x , y ∈ s v ( ε ) x,y\in\mathfrak{sv}(\varepsilon) x , y ∈ sv ( ε ) , where χ \chi χ is the bilinear map defined by
χ ( L m , L n ) = ∑ i ∈ Z μ i M m + n + i ( m , n ∈ Z ) , \chi(L_m,L_n)=\sum_{i\in\mathbb{Z}}\mu_iM_{m+n+i}\quad(m,n\in\mathbb{Z}), χ ( L m , L n ) = i ∈ Z ∑ μ i M m + n + i ( m , n ∈ Z ) , and χ ( x , y ) = 0 \chi(x,y)=0 χ ( x , y ) = 0 if one of the basis elements x , y x,y x , y belongs to { Y n , M m ∣ m ∈ Z , n ∈ ε + Z } \{Y_n,M_m\mid m\in\mathbb{Z},\ n\in\varepsilon+\mathbb{Z}\} { Y n , M m ∣ m ∈ Z , n ∈ ε + Z } .
Corollary 23. Let ε ∈ { 0 , 1 2 } \varepsilon\in\{0,\frac{1}{2}\} ε ∈ { 0 , 2 1 } and δ ∈ C \delta\in\mathbb{C} δ ∈ C .
If δ ≠ 1 \delta\ne1 δ = 1 , then every δ \delta δ -biderivation of s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) is zero.
If δ = 1 \delta=1 δ = 1 , then the δ \delta δ -biderivations of s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) are the biderivations described in Theorem 22 .
Proof. The second statement is Theorem 22 . Let δ ≠ 1 \delta\ne1 δ = 1 and let ϕ \phi ϕ be a δ \delta δ -biderivation. For every z z z the maps x ↦ ϕ ( x , z ) x\mapsto\phi(x,z) x ↦ ϕ ( x , z ) and x ↦ ϕ ( z , x ) x\mapsto\phi(z,x) x ↦ ϕ ( z , x ) are δ \delta δ -derivations. If δ ≠ 1 2 \delta\ne\frac{1}{2} δ = 2 1 , they vanish by Corollary 21 , so ϕ = 0 \phi=0 ϕ = 0 . Let δ = 1 2 \delta=\frac{1}{2} δ = 2 1 . By Corollary 21 , there are maps λ , μ : s v ( ε ) → C \lambda,\mu:\mathfrak{sv}(\varepsilon)\to\mathbb{C} λ , μ : sv ( ε ) → C such that ϕ ( x , y ) = λ ( y ) x = μ ( x ) y \phi(x,y)=\lambda(y)x=\mu(x)y ϕ ( x , y ) = λ ( y ) x = μ ( x ) y for all x , y x,y x , y . Given y y y , choose a basis element P P P which does not occur in the decomposition of y y y with respect to the basis. Comparing the coefficients at P P P in λ ( y ) P = ϕ ( P , y ) = μ ( P ) y \lambda(y)P=\phi(P,y)=\mu(P)y λ ( y ) P = ϕ ( P , y ) = μ ( P ) y , we get λ ( y ) = 0 \lambda(y)=0 λ ( y ) = 0 . Hence ϕ ( x , y ) = λ ( y ) x = 0 \phi(x,y)=\lambda(y)x=0 ϕ ( x , y ) = λ ( y ) x = 0 for all x , y x,y x , y .
Recall that transposed δ \delta δ -Poisson structures were defined in Definition 4 ; for δ = 1 2 \delta=\frac{1}{2} δ = 2 1 these are exactly the transposed Poisson structures. By Lemma 5 , if ⋅ \cdot ⋅ is a transposed δ \delta δ -Poisson structure on a Lie algebra L L L , then the bilinear map ϕ ( x , y ) = x ⋅ y \phi(x,y)=x\cdot y ϕ ( x , y ) = x ⋅ y is a δ \delta δ -biderivation of L L L .
Corollary 24. If δ ≠ 1 \delta\ne1 δ = 1 , then every transposed δ \delta δ -Poisson structure on s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) is trivial. If δ = 1 \delta=1 δ = 1 , then the transposed 1-Poisson structures on s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) are exactly the multiplications given by
L m ⋅ L n = ∑ i ∈ Z μ i M m + n + i ( m , n ∈ Z ) , L_m\cdot L_n=\sum_{i\in\mathbb{Z}}\mu_iM_{m+n+i}\quad(m,n\in\mathbb{Z}), L m ⋅ L n = i ∈ Z ∑ μ i M m + n + i ( m , n ∈ Z ) , with all other products of basis elements equal to zero, where μ i ∈ C \mu_i\in\mathbb{C} μ i ∈ C and only finitely many of the μ i \mu_i μ i are nonzero.
Proof. Let ⋅ \cdot ⋅ be a transposed δ \delta δ -Poisson structure. By Lemma 5 , ϕ ( x , y ) = x ⋅ y \phi(x,y)=x\cdot y ϕ ( x , y ) = x ⋅ y is a δ \delta δ -biderivation, so ϕ = 0 \phi=0 ϕ = 0 if δ ≠ 1 \delta\ne1 δ = 1 by Corollary 23 . Let δ = 1 \delta=1 δ = 1 . By Theorem 22 , x ⋅ y = λ [ x , y ] + χ ( x , y ) x\cdot y=\lambda[x,y]+\chi(x,y) x ⋅ y = λ [ x , y ] + χ ( x , y ) . Since both ⋅ \cdot ⋅ and χ \chi χ are commutative, λ [ x , y ] = λ [ y , x ] \lambda[x,y]=\lambda[y,x] λ [ x , y ] = λ [ y , x ] for all x , y x,y x , y ; for x = L 1 x=L_1 x = L 1 , y = L 0 y=L_0 y = L 0 this gives 2 λ L 1 = 0 2\lambda L_1=0 2 λ L 1 = 0 , so λ = 0 \lambda=0 λ = 0 and x ⋅ y = χ ( x , y ) x\cdot y=\chi(x,y) x ⋅ y = χ ( x , y ) . Conversely, let x ⋅ y = χ ( x , y ) x\cdot y=\chi(x,y) x ⋅ y = χ ( x , y ) . This multiplication is commutative. It is associative, since all products lie in ⟨ M m ∣ m ∈ Z ⟩ \langle M_m\mid m\in\mathbb{Z}\rangle ⟨ M m ∣ m ∈ Z ⟩ and M k ⋅ y = 0 M_k\cdot y=0 M k ⋅ y = 0 for all k k k and y y y , so that ( x ⋅ y ) ⋅ z = 0 = x ⋅ ( y ⋅ z ) (x\cdot y)\cdot z=0=x\cdot(y\cdot z) ( x ⋅ y ) ⋅ z = 0 = x ⋅ ( y ⋅ z ) . Finally, (5) with δ = 1 \delta=1 δ = 1 is the identity (4) for the biderivation χ \chi χ (Theorem 22 ), with ( x , y , z ) (x,y,z) ( x , y , z ) replaced by ( z , x , y ) (z,x,y) ( z , x , y ) .
The extended Schrödinger–Witt algebras
Definition 25. Let ε ∈ { 0 , 1 2 } \varepsilon\in\{0,\frac{1}{2}\} ε ∈ { 0 , 2 1 } . The extended Schrödinger–Witt algebra e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) is the Lie algebra with the basis { L m , M m , N m , Y n ∣ m ∈ Z , n ∈ ε + Z } \{L_m,M_m,N_m,Y_n\mid m\in\mathbb{Z},\ n\in\varepsilon+\mathbb{Z}\} { L m , M m , N m , Y n ∣ m ∈ Z , n ∈ ε + Z } and the following nontrivial
relations:
[ L m , L n ] = ( m − n ) L m + n , [ L m , Y n ] = ( m 2 − n ) Y m + n , [ L m , M n ] = − n M m + n , [ Y m , Y n ] = ( m − n ) M m + n , [ L m , N n ] = − n N m + n , [ N m , Y n ] = Y m + n , [ N m , M n ] = 2 M m + n . (34) \begin{aligned}
[L_m,L_n]&=(m-n)L_{m+n},\qquad[L_m,Y_n]=\left(\frac{m}{2}-n\right)Y_{m+n},\qquad[L_m,M_n]=-nM_{m+n},\\
[Y_m,Y_n]&=(m-n)M_{m+n},\qquad[L_m,N_n]=-nN_{m+n},\qquad[N_m,Y_n]=Y_{m+n},\\
[N_m,M_n]&=2M_{m+n}. \tag*{(34)}
\end{aligned} [ L m , L n ] [ Y m , Y n ] [ N m , M n ] = ( m − n ) L m + n , [ L m , Y n ] = ( 2 m − n ) Y m + n , [ L m , M n ] = − n M m + n , = ( m − n ) M m + n , [ L m , N n ] = − n N m + n , [ N m , Y n ] = Y m + n , = 2 M m + n . (34)
All other brackets of basis elements are zero (in particular, [ N m , N n ] = 0 [N_m,N_n]=0 [ N m , N n ] = 0 ).
The algebra e s v ( 1 2 ) \mathfrak{esv}\left(\frac{1}{2}\right) esv ( 2 1 ) is the extended Schrödinger–Witt algebra (written in the conventions of the present paper; see [37 ] and the references therein), and e s v ( 0 ) \mathfrak{esv}(0) esv ( 0 ) is its twisted version. A direct verification on triples of basis elements shows that the Jacobi identity holds. Clearly, s v ( ε ) = ⟨ L m , M m , Y n ⟩ \mathfrak{sv}(\varepsilon)=\langle L_m,M_m,Y_n\rangle sv ( ε ) = ⟨ L m , M m , Y n ⟩ is a subalgebra of e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) . As for s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) , the indices of L m L_m L m , M m M_m M m , N m N_m N m belong to Z \mathbb{Z} Z and the indices of Y n Y_n Y n belong to ε + Z \varepsilon+\mathbb{Z} ε + Z .
The algebra e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) is graded by G = 1 2 Z × Z 2 G=\frac{1}{2}\mathbb{Z}\times\mathbb{Z}_2 G = 2 1 Z × Z 2 :
e s v ( ε ) ( t , 0 ˉ ) = ⟨ L t , M t , N t ⟩ ( t ∈ Z ) , e s v ( ε ) ( t , 1 ˉ ) = ⟨ Y t ⟩ ( t ∈ ε + Z ) , \mathfrak{esv}(\varepsilon)_{(t,\bar{0})}=\langle L_t,M_t,N_t\rangle\ (t\in\mathbb{Z}),\qquad\mathfrak{esv}(\varepsilon)_{(t,\bar{1})}=\langle Y_t\rangle\ (t\in\varepsilon+\mathbb{Z}), esv ( ε ) ( t , 0 ˉ ) = ⟨ L t , M t , N t ⟩ ( t ∈ Z ) , esv ( ε ) ( t , 1 ˉ ) = ⟨ Y t ⟩ ( t ∈ ε + Z ) ,
and all other homogeneous components are zero. As for s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) , we speak about even linear maps of degree j ∈ Z j\in\mathbb{Z} j ∈ Z and odd linear maps of degree s ∈ ε + Z s\in\varepsilon+\mathbb{Z} s ∈ ε + Z .
Note that the restriction of ad N 0 \operatorname{ad}_{N_0} ad N 0 to the subalgebra s v ( ε ) \mathfrak{sv}(\varepsilon) sv ( ε ) of e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) is the outer derivation D 3 D_3 D 3 of Theorem 10 . For every t ∈ Z t\in\mathbb{Z} t ∈ Z and s ∈ ε + Z s\in\varepsilon+\mathbb{Z} s ∈ ε + Z the inner derivations act as follows (we list only the values not given by (7)):
ad L t ( N m ) = − m N m + t , ad M t ( N m ) = − 2 M m + t , ad Y s ( N m ) = − Y m + s , \operatorname{ad}_{L_t}(N_m)=-mN_{m+t},\qquad\operatorname{ad}_{M_t}(N_m)=-2M_{m+t},\qquad\operatorname{ad}_{Y_s}(N_m)=-Y_{m+s}, ad L t ( N m ) = − m N m + t , ad M t ( N m ) = − 2 M m + t , ad Y s ( N m ) = − Y m + s ,
ad N t ( L m ) = t N m + t , ad N t ( Y n ) = Y n + t , ad N t ( M m ) = 2 M m + t . (35) \operatorname{ad}_{N_t}(L_m)=tN_{m+t},\qquad\operatorname{ad}_{N_t}(Y_n)=Y_{n+t},\qquad\operatorname{ad}_{N_t}(M_m)=2M_{m+t}. \tag*{(35)} ad N t ( L m ) = t N m + t , ad N t ( Y n ) = Y n + t , ad N t ( M m ) = 2 M m + t . (35)
Note also that ad N 0 \operatorname{ad}_{N_0} ad N 0 acts diagonally:
[ N 0 , L m ] = [ N 0 , N m ] = 0 , [ N 0 , Y n ] = Y n , [ N 0 , M m ] = 2 M m . [N_0,L_m]=[N_0,N_m]=0,\qquad[N_0,Y_n]=Y_n,\qquad[N_0,M_m]=2M_m. [ N 0 , L m ] = [ N 0 , N m ] = 0 , [ N 0 , Y n ] = Y n , [ N 0 , M m ] = 2 M m .
Lemma 26 (see [10 ] for ε = 1 2 \varepsilon=\frac{1}{2} ε = 2 1 ). The algebra e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) is perfect and Z ( e s v ( ε ) ) = 0 Z(\mathfrak{esv}(\varepsilon))=0 Z ( esv ( ε )) = 0 .
Proof. We have
L m = 1 m [ L m , L 0 ] for m ≠ 0 , L 0 = 1 2 [ L 1 , L − 1 ] , L_m=\frac{1}{m}[L_m,L_0]\text{ for }m\ne0,\qquad L_0=\frac{1}{2}[L_1,L_{-1}], L m = m 1 [ L m , L 0 ] for m = 0 , L 0 = 2 1 [ L 1 , L − 1 ] ,
N m = − [ L m − 1 , N 1 ] , M m = 1 2 [ N 0 , M m ] and Y n = [ N 0 , Y n ] . N_m=-[L_{m-1},N_1],\qquad M_m=\frac{1}{2}[N_0,M_m]\text{ and }Y_n=[N_0,Y_n]. N m = − [ L m − 1 , N 1 ] , M m = 2 1 [ N 0 , M m ] and Y n = [ N 0 , Y n ] .
Hence e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) is perfect. Let z z z be a central element. Since
[ L 0 , L m ] = − m L m , [ L 0 , Y n ] = − n Y n , [ L 0 , M n ] = − n M n and [ L 0 , N n ] = − n N n , [L_0,L_m]=-mL_m,\qquad[L_0,Y_n]=-nY_n,\qquad[L_0,M_n]=-nM_n\text{ and }[L_0,N_n]=-nN_n, [ L 0 , L m ] = − m L m , [ L 0 , Y n ] = − n Y n , [ L 0 , M n ] = − n M n and [ L 0 , N n ] = − n N n ,
the equality [ L 0 , z ] = 0 [L_0,z]=0 [ L 0 , z ] = 0 gives z = λ L 0 + μ Y 0 + ν M 0 + ρ N 0 z=\lambda L_0+\mu Y_0+\nu M_0+\rho N_0 z = λ L 0 + μ Y 0 + ν M 0 + ρ N 0 (with μ = 0 \mu=0 μ = 0 if ε = 1 2 \varepsilon=\frac{1}{2} ε = 2 1 ). Then 0 = [ N 1 , z ] = λ N 1 + μ Y 1 + 2 ν M 1 0=[N_1,z]=\lambda N_1+\mu Y_1+2\nu M_1 0 = [ N 1 , z ] = λ N 1 + μ Y 1 + 2 ν M 1 gives λ = μ = ν = 0 \lambda=\mu=\nu=0 λ = μ = ν = 0 , and 0 = [ z , Y n ] = ρ Y n 0=[z,Y_n]=\rho Y_n 0 = [ z , Y n ] = ρ Y n gives ρ = 0 \rho=0 ρ = 0 .
By Lemmas 6 and 26, the map D ′ D' D ′ associated with a quasi-derivation D D D of e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) is unique, and e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) has no nonzero central maps. Since every homogeneous component of a quasi-derivation of e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) is a quasi-derivation (Lemma 7 ), the associated map of a homogeneous quasi-derivation is homogeneous of the same degree.
Remark 27. The structure of the original extended Schrödinger–Witt algebra e s v ( 1 2 ) \mathfrak{esv}\left(\frac{1}{2}\right) esv ( 2 1 ) was studied by Gao, Jiang and Pei [10 ] ; in particular, they proved that every derivation of e s v ( 1 2 ) \mathfrak{esv}\left(\frac{1}{2}\right) esv ( 2 1 ) is inner. This result is recovered, for both algebras e s v ( 1 2 ) \mathfrak{esv}\left(\frac{1}{2}\right) esv ( 2 1 ) and e s v ( 0 ) \mathfrak{esv}(0) esv ( 0 ) , in Corollary 31 below. The 1 2 \frac{1}{2} 2 1 -derivations and the transposed Poisson structures of the original extended Schrödinger–Witt algebra e s v ( 1 2 ) \mathfrak{esv}\left(\frac{1}{2}\right) esv ( 2 1 ) were studied in [37 ] : every 1 2 \frac{1}{2} 2 1 -derivation of e s v ( 1 2 ) \mathfrak{esv}\left(\frac{1}{2}\right) esv ( 2 1 ) is trivial and all transposed Poisson structures on e s v ( 1 2 ) \mathfrak{esv}\left(\frac{1}{2}\right) esv ( 2 1 ) are trivial. These results are recovered, for both algebras e s v ( 1 2 ) \mathfrak{esv}\left(\frac{1}{2}\right) esv ( 2 1 ) and e s v ( 0 ) \mathfrak{esv}(0) esv ( 0 ) , in Corollaries 32 and 35 below.
Quasi-derivations of the extended Schrödinger–Witt algebras
Proposition 28. Let D D D be an even quasi-derivation of degree j ∈ Z j \in\mathbb{Z} j ∈ Z of e s v ( ε ) \mathfrak{esv}(\varepsilon) esv ( ε ) . Then there are u , v , w , λ ∈ C u,v,w,\lambda\in\mathbb{C} u , v , w , λ ∈ C , with λ = 0 \lambda=0 λ = 0 if j ≠ 0 j \ne0 j = 0 , such that
D = u ad L j + v ad M j + w ad N j + λ Id . D=u\operatorname{ad}_{L_j}+v\operatorname{ad}_{M_j}+w\operatorname{ad}_{N_j}+\lambda\operatorname{Id}. D = u ad L j + v ad M j + w ad N j + λ Id .
Proof. Since D D D and D ′ D' D ′ are even of degree j j j , we can write
D ( L m ) = a m L m + j + b m M m + j + r m N m + j , D ( Y n ) = c n Y n + j , D(L_m)=a_mL_{m+j}+b_mM_{m+j}+r_mN_{m+j}, \qquad D(Y_n)=c_nY_{n+j}, D ( L m ) = a m L m + j + b m M m + j + r m N m + j , D ( Y n ) = c n Y n + j ,
D ( M m ) = d m L m + j + e m M m + j + p m N m + j , D ( N m ) = x m L m + j + y m M m + j + z m N m + j , D(M_m)=d_mL_{m+j}+e_mM_{m+j}+p_mN_{m+j}, \qquad D(N_m)=x_mL_{m+j}+y_mM_{m+j}+z_mN_{m+j}, D ( M m ) = d m L m + j + e m M m + j + p m N m + j , D ( N m ) = x m L m + j + y m M m + j + z m N m + j ,
and similarly for D ′ D' D ′ , with the coefficients denoted by a m ′ , b m ′ , … , z m ′ a'_m,b'_m,\ldots,z'_m a m ′ , b m ′ , … , z m ′ . Applying (1) to the pairs of basis elements indicated in brackets and comparing the coefficients at the basis elements indicated in brackets, we obtain the following relations:
( m − n + j 2 ) d n − p n = 0 [ ( Y m , M n ) , Y m + n + j ] , (36) \left(m-\frac{n+j}{2}\right)d_n-p_n=0 \qquad[(Y_m,M_n),Y_{m+n+j}], \tag*{(36)} ( m − 2 n + j ) d n − p n = 0 [( Y m , M n ) , Y m + n + j ] , (36)
y m = y n [ ( N m , N n ) , M m + n + j ] , (37) y_m=y_n \qquad[(N_m,N_n),M_{m+n+j}], \tag*{(37)} y m = y n [( N m , N n ) , M m + n + j ] , (37)
− n x m + m x n = 0 [ ( N m , N n ) , N m + n + j ] , (38) -nx_m+mx_n=0 \qquad[(N_m,N_n),N_{m+n+j}], \tag*{(38)} − n x m + m x n = 0 [( N m , N n ) , N m + n + j ] , (38)
( n + j − m ) x n = n x m + n ′ [ ( L m , N n ) , L m + n + j ] , (39) (n+j-m)x_n=nx'_{m+n} \qquad[(L_m,N_n),L_{m+n+j}], \tag*{(39)} ( n + j − m ) x n = n x m + n ′ [( L m , N n ) , L m + n + j ] , (39)
2 b m + ( n + j ) y n = n y m + n ′ [ ( L m , N n ) , M m + n + j ] , (40) 2b_m+(n+j)y_n=ny'_{m+n} \qquad[(L_m,N_n),M_{m+n+j}], \tag*{(40)} 2 b m + ( n + j ) y n = n y m + n ′ [( L m , N n ) , M m + n + j ] , (40)
n a m + ( n + j ) z n = n z m + n ′ [ ( L m , N n ) , N m + n + j ] , (41) na_m+(n+j)z_n=nz'_{m+n} \qquad[(L_m,N_n),N_{m+n+j}], \tag*{(41)} n a m + ( n + j ) z n = n z m + n ′ [( L m , N n ) , N m + n + j ] , (41)
c m + ( n + j 2 − m ) x n + z n = c m + n ′ [ ( Y m , N n ) , Y m + n + j ] , (42) c_m+\left(\frac{n+j}{2}-m\right)x_n+z_n=c'_{m+n} \qquad[(Y_m,N_n),Y_{m+n+j}], \tag*{(42)} c m + ( 2 n + j − m ) x n + z n = c m + n ′ [( Y m , N n ) , Y m + n + j ] , (42)
e m − m 2 x n + z n = e m + n ′ [ ( M m , N n ) , M m + n + j ] , (43) e_m-\frac{m}{2}x_n+z_n=e'_{m+n} \qquad[(M_m,N_n),M_{m+n+j}], \tag*{(43)} e m − 2 m x n + z n = e m + n ′ [( M m , N n ) , M m + n + j ] , (43)
− n a m + 2 r m − ( n + j ) e n = − n e m + n ′ [ ( L m , M n ) , M m + n + j ] , (44) -na_m+2r_m-(n+j)e_n=-ne'_{m+n} \qquad[(L_m,M_n),M_{m+n+j}], \tag*{(44)} − n a m + 2 r m − ( n + j ) e n = − n e m + n ′ [( L m , M n ) , M m + n + j ] , (44)
( m − n + j ) c m + ( m − n − j ) c n = ( m − n ) e m + n ′ [ ( Y m , Y n ) , M m + n + j ] . (45) (m-n+j)c_m+(m-n-j)c_n=(m-n)e'_{m+n} \qquad[(Y_m,Y_n),M_{m+n+j}]. \tag*{(45)} ( m − n + j ) c m + ( m − n − j ) c n = ( m − n ) e m + n ′ [( Y m , Y n ) , M m + n + j ] . (45)
where m , n ∈ Z m,n\in\mathbb{Z} m , n ∈ Z if the corresponding basis elements are L L L , M M M or N N N , and m , n ∈ ε + Z m,n\in\varepsilon+\mathbb{Z} m , n ∈ ε + Z if they are Y Y Y . For example, for the pair ( L m , N n ) (L_m,N_n) ( L m , N n ) we have
[ D ( L m ) , N n ] + [ L m , D ( N n ) ] = [D(L_m),N_n]+[L_m,D(N_n)]= [ D ( L m ) , N n ] + [ L m , D ( N n )] =
( m − n − j ) x n L m + n + j − ( 2 b m + ( n + j ) y n ) M m + n + j − ( n a m + ( n + j ) z n ) N m + n + j (m-n-j)x_nL_{m+n+j}-(2b_m+(n+j)y_n)M_{m+n+j}-(na_m+(n+j)z_n)N_{m+n+j} ( m − n − j ) x n L m + n + j − ( 2 b m + ( n + j ) y n ) M m + n + j − ( n a m + ( n + j ) z n ) N m + n + j
and D ′ ( [ L m , N n ] ) = − n ( x m + n ′ L m + n + j + y m + n ′ M m + n + j + z m + n ′ N m + n + j ) D'([L_m,N_n])=-n(x'_{m+n}L_{m+n+j}+y'_{m+n}M_{m+n+j}+z'_{m+n}N_{m+n+j}) D ′ ([ L m , N n ]) = − n ( x m + n ′ L m + n + j + y m + n ′ M m + n + j + z m + n ′ N m + n + j ) , which gives (39), (40) and (41); the other relations are obtained in the same way.
Step 1. Taking two different values of m ∈ ε + Z m\in\varepsilon+\mathbb{Z} m ∈ ε + Z in (36), we get d n = 0 d_n=0 d n = 0 and then p n = 0 p_n=0 p n = 0 for all n ∈ Z n\in\mathbb{Z} n ∈ Z .
Step 2. By (38) with n = 1 n=1 n = 1 , x m = m x 1 x_m=mx_1 x m = m x 1 for all m m m . By (39) with n = 1 n=1 n = 1 we have ( 1 + j − m ) x 1 = x m + 1 ′ (1+j-m)x_1=x'_{m+1} ( 1 + j − m ) x 1 = x m + 1 ′ , and by (39) with n = 2 n=2 n = 2 and m m m replaced by m − 1 m-1 m − 1 we have ( 3 + j − m ) x 2 = 2 x m + 1 ′ (3+j-m)x_2=2x'_{m+1} ( 3 + j − m ) x 2 = 2 x m + 1 ′ . Since x 2 = 2 x 1 x_2=2x_1 x 2 = 2 x 1 , we get ( 3 + j − m ) x 1 = ( 1 + j − m ) x 1 (3+j-m)x_1=(1+j-m)x_1 ( 3 + j − m ) x 1 = ( 1 + j − m ) x 1 , i.e., x 1 = 0 x_1=0 x 1 = 0 . Hence x m = 0 x_m=0 x m = 0 for all m ∈ Z m\in\mathbb{Z} m ∈ Z .
Step 3. By (37), there is y ∈ C y\in\mathbb{C} y ∈ C with y m = y y_m=y y m = y for all m m m . Then (40) with n = 0 n=0 n = 0 gives b m = − j 2 y b_m=-\frac{j}{2}y b m = − 2 j y for all $m. Step 4. Since x n = 0 x_n=0 x n = 0 , the relation (43) reads e m + z n = e m + n ′ e_m+z_n=e'_{m+n} e m + z n = e m + n ′ . Comparing this relation for the pairs ( m + 1 , n ) (m+1,n) ( m + 1 , n ) and ( m , n + 1 ) (m,n+1) ( m , n + 1 ) , we get e m + 1 − e m = z n + 1 − z n e_{m+1}-e_m=z_{n+1}-z_n e m + 1 − e m = z n + 1 − z n for all m , n ∈ Z m,n\in\mathbb{Z} m , n ∈ Z . Hence this difference is a constant, which we denote by − u -u − u , and there are λ , η ∈ C \lambda,\eta\in\mathbb{C} λ , η ∈ C such that
z n = λ − u n , e m = η − u m , e k ′ = e 0 + z k = η + λ − u k ( m , n , k ∈ Z ) . z_n=\lambda-un,\qquad e_m=\eta-um,\qquad e'_k=e_0+z_k=\eta+\lambda-uk\qquad(m,n,k\in\mathbb{Z}). z n = λ − u n , e m = η − u m , e k ′ = e 0 + z k = η + λ − u k ( m , n , k ∈ Z ) .
In the same way, (42) reads c m + z n = c m + n ′ c_m+z_n=c'_{m+n} c m + z n = c m + n ′ , whence c m + 1 − c m = z n + 1 − z n = − u c_{m+1}-c_m=z_{n+1}-z_n=-u c m + 1 − c m = z n + 1 − z n = − u , and there is γ ∈ C \gamma\in\mathbb{C} γ ∈ C with c m = γ − u m c_m=\gamma-um c m = γ − u m for all m ∈ ε + Z m\in\varepsilon+\mathbb{Z} m ∈ ε + Z .
Step 5. The relation (41) with n = 0 n=0 n = 0 gives j z 0 = 0 jz_0=0 j z 0 = 0 , i.e.,
j λ = 0. (46) j\lambda=0. \tag*{(46)} j λ = 0. (46)
By (41) with n = 1 n=1 n = 1 (and m m m replaced by m + 1 m+1 m + 1 ) and with n = 2 n=2 n = 2 , we have
a m + 1 + ( 1 + j ) z 1 = z m + 2 ′ and 2 a m + ( 2 + j ) z 2 = 2 z m + 2 ′ . a_{m+1}+(1+j)z_1=z'_{m+2}\qquad\text{and}\qquad2a_m+(2+j)z_2=2z'_{m+2}. a m + 1 + ( 1 + j ) z 1 = z m + 2 ′ and 2 a m + ( 2 + j ) z 2 = 2 z m + 2 ′ .
Hence 2 ( a m + 1 − a m ) = ( 2 + j ) z 2 − 2 ( 1 + j ) z 1 = − j λ − 2 u = − 2 u 2(a_{m+1}-a_m)=(2+j)z_2-2(1+j)z_1=-j\lambda-2u=-2u 2 ( a m + 1 − a m ) = ( 2 + j ) z 2 − 2 ( 1 + j ) z 1 = − j λ − 2 u = − 2 u , and a m = a 0 − u m a_m=a_0-um a m = a 0 − u m for all m ∈ Z m\in\mathbb{Z} m ∈ Z .
Step 6. By (44) with n = 0 n=0 n = 0 we get r m = j 2 η r_m=\frac{j}{2}\eta r m = 2 j η for all m m m . By (44) with m = 0 m=0 m = 0 and n = 1 n=1 n = 1 we get − a 0 + j η − ( 1 + j ) ( η − u ) = − ( η + λ − u ) -a_0+j\eta-(1+j)(\eta-u)=-(\eta+\lambda-u) − a 0 + j η − ( 1 + j ) ( η − u ) = − ( η + λ − u ) , i.e., a 0 = u j + λ a_0=uj+\lambda a 0 = u j + λ . Thus a m = u ( j − m ) + λ a_m=u(j-m)+\lambda a m = u ( j − m ) + λ .
Step 7. Substituting c m = γ − u m c_m=\gamma-um c m = γ − u m and e k ′ = η + λ − u k e'_k=\eta+\lambda-uk e k ′ = η + λ − u k in (45), we get ( m − n ) ( 2 γ − u ( m + n ) − u j ) = ( m − n ) ( η + λ − u ( m + n ) ) (m-n)(2\gamma-u(m+n)-uj)=(m-n)(\eta+\lambda-u(m+n)) ( m − n ) ( 2 γ − u ( m + n ) − u j ) = ( m − n ) ( η + λ − u ( m + n )) , and taking m ≠ n m\ne n m = n we obtain γ = 1 2 ( η + λ + u j ) \gamma=\frac{1}{2}(\eta+\lambda+uj) γ = 2 1 ( η + λ + u j ) .
Step 8. Let v = − y 2 v=-\frac{y}{2} v = − 2 y and w = 1 2 ( η − λ ) w=\frac{1}{2}(\eta-\lambda) w = 2 1 ( η − λ ) . By (7), (35) and (46), the map D ^ = u ad L j + v ad M j + w ad N j + λ Id \widehat{D}=u\operatorname{ad}L_j+v\operatorname{ad}M_j+w\operatorname{ad}N_j+\lambda\operatorname{Id} D = u ad L j + v ad M j + w ad N j + λ Id satisfies
D ^ ( L m ) = ( u ( j − m ) + λ ) L m + j + v j M m + j + w j N m + j = a m L m + j + b m M m + j + r m N m + j , \widehat{D}(L_m)=(u(j-m)+\lambda)L_{m+j}+vjM_{m+j}+wjN_{m+j}=a_mL_{m+j}+b_mM_{m+j}+r_mN_{m+j}, D ( L m ) = ( u ( j − m ) + λ ) L m + j + v j M m + j + w j N m + j = a m L m + j + b m M m + j + r m N m + j ,
D ^ ( Y n ) = ( u ( j 2 − n ) + w + λ ) Y n + j = c n Y n + j , \widehat{D}(Y_n)=\left(u\left(\frac{j}{2}-n\right)+w+\lambda\right)Y_{n+j}=c_nY_{n+j}, D ( Y n ) = ( u ( 2 j − n ) + w + λ ) Y n + j = c n Y n + j ,
D ^ ( M m ) = ( − u m + 2 w + λ ) M m + j = e m M m + j , \widehat{D}(M_m)=(-um+2w+\lambda)M_{m+j}=e_mM_{m+j}, D ( M m ) = ( − u m + 2 w + λ ) M m + j = e m M m + j ,
D ^ ( N m ) = − 2 v M m + j + ( λ − u m ) N m + j = y m M m + j + z m N m + j , \widehat{D}(N_m)=-2vM_{m+j}+(\lambda-um)N_{m+j}=y_mM_{m+j}+z_mN_{m+j}, D ( N m ) = − 2 v M m + j + ( λ − u m ) N m + j = y m M m + j + z m N m + j ,
where we used j w = j 2 η jw=\frac{j}{2}\eta j w = 2 j η , which holds by (46). By Steps 1–7, D = D ^ D=\widehat{D} D = D . Finally, λ = 0 \lambda=0 λ = 0 if j ≠ 0 j\ne0 j = 0 by (46).
Proposition 29. Let D D D be an odd quasi-derivation of degree s ∈ ε + Z s\in\varepsilon+\mathbb{Z} s ∈ ε + Z of esv ( ε ) \operatorname{esv}(\varepsilon) esv ( ε ) . Then there is w ∈ C w\in\mathbb{C} w ∈ C such that D = w ad Y s D=w\operatorname{ad}Y_s D = w ad Y s .
Proof. Since D D D and D ′ D' D ′ are odd of degree s s s , we can write
D ( L m ) = f m Y m + s , D ( M m ) = g m Y m + s , D ( N m ) = q m Y m + s , D(L_m)=f_mY_{m+s},\qquad D(M_m)=g_mY_{m+s},\qquad D(N_m)=q_mY_{m+s}, D ( L m ) = f m Y m + s , D ( M m ) = g m Y m + s , D ( N m ) = q m Y m + s ,
D ( Y n ) = h n L n + s + k n M n + s + l n N n + s , D(Y_n)=h_nL_{n+s}+k_nM_{n+s}+l_nN_{n+s}, D ( Y n ) = h n L n + s + k n M n + s + l n N n + s ,
and similarly for D ′ D' D ′ . As in the proof of Proposition 28 , we obtain the following relations:
g m = 2 g m + n ′ [ ( M m , N n ) , Y m + n + s ] , (47) g_m=2g'_{m+n}
\qquad[(M_m,N_n),Y_{m+n+s}], \tag*{(47)} g m = 2 g m + n ′ [( M m , N n ) , Y m + n + s ] , (47)
( m 2 − n − s ) g n = − n g m + n ′ [ ( L m , M n ) , Y m + n + s ] , (48) \left(\frac{m}{2}-n-s\right)g_n=-ng'_{m+n}
\qquad[(L_m,M_n),Y_{m+n+s}], \tag*{(48)} ( 2 m − n − s ) g n = − n g m + n ′ [( L m , M n ) , Y m + n + s ] , (48)
h m + n ′ = 0 [ ( Y m , N n ) , L m + n + s ] , (49) h'_{m+n}=0
\qquad[(Y_m,N_n),L_{m+n+s}], \tag*{(49)} h m + n ′ = 0 [( Y m , N n ) , L m + n + s ] , (49)
2 k m + ( n + s − m ) q n = k m + n ′ [ ( Y m , N n ) , M m + n + s ] , (50) 2k_m+(n+s-m)q_n=k'_{m+n}
\qquad[(Y_m,N_n),M_{m+n+s}], \tag*{(50)} 2 k m + ( n + s − m ) q n = k m + n ′ [( Y m , N n ) , M m + n + s ] , (50)
( m − n − s ) h n = ( m 2 − n ) h m + n ′ [ ( L m , Y n ) , L m + n + s ] , (51) (m-n-s)h_n=\left(\frac{m}{2}-n\right)h'_{m+n}
\qquad[(L_m,Y_n),L_{m+n+s}], \tag*{(51)} ( m − n − s ) h n = ( 2 m − n ) h m + n ′ [( L m , Y n ) , L m + n + s ] , (51)
( m + s − n ) f m − ( n + s ) k n = ( m 2 − n ) k m + n ′ [ ( L m , Y n ) , M m + n + s ] , (52) (m+s-n)f_m-(n+s)k_n=\left(\frac{m}{2}-n\right)k'_{m+n}
\qquad[(L_m,Y_n),M_{m+n+s}], \tag*{(52)} ( m + s − n ) f m − ( n + s ) k n = ( 2 m − n ) k m + n ′ [( L m , Y n ) , M m + n + s ] , (52)
− n h m + 2 l m + ( m − n − s ) g n = 0 [ ( Y m , M n ) , M m + n + s ] , (53) -nh_m+2l_m+(m-n-s)g_n=0
\qquad[(Y_m,M_n),M_{m+n+s}], \tag*{(53)} − n h m + 2 l m + ( m − n − s ) g n = 0 [( Y m , M n ) , M m + n + s ] , (53)
q m = q n [ ( N m , N n ) , Y m + n + s ] , (54) q_m=q_n
\qquad[(N_m,N_n),Y_{m+n+s}], \tag*{(54)} q m = q n [( N m , N n ) , Y m + n + s ] , (54)
− f m + ( m 2 − n − s ) q n = − n q m + n ′ [ ( L m , N n ) , Y m + n + s ] . (55) -f_m+\left(\frac{m}{2}-n-s\right)q_n=-nq'_{m+n}
\qquad[(L_m,N_n),Y_{m+n+s}]. \tag*{(55)} − f m + ( 2 m − n − s ) q n = − n q m + n ′ [( L m , N n ) , Y m + n + s ] . (55)
Step 1. By (47) for the pairs ( m , n ) (m,n) ( m , n ) and ( m + n , 0 ) (m+n,0) ( m + n , 0 ) , we get g m = 2 g m + n ′ = g m + n g_m=2g'_{m+n}=g_{m+n} g m = 2 g m + n ′ = g m + n , so g m = g 0 g_m=g_0 g m = g 0 for all m m m . Now (48) with n = 0 n=0 n = 0 gives ( m 2 − s ) g 0 = 0 \left(\frac{m}{2}-s\right)g_0=0 ( 2 m − s ) g 0 = 0 for all m ∈ Z m\in\mathbb{Z} m ∈ Z , hence g m = 0 g_m=0 g m = 0 for all m ∈ Z m\in\mathbb{Z} m ∈ Z .
Step 2. By (49) , h ′ = 0 h'=0 h ′ = 0 ; then (51) gives ( m − n − s ) h n = 0 (m-n-s)h_n=0 ( m − n − s ) h n = 0 for all m ∈ Z m\in\mathbb{Z} m ∈ Z , hence h n = 0 h_n=0 h n = 0 for all n n n . Now (53) gives l m = 0 l_m=0 l m = 0 for all m m m .
Step 3. By (54) , there is q ∈ C q\in\mathbb{C} q ∈ C with q m = q q_m=q q m = q for all m m m , and (55) with n = 0 n=0 n = 0 gives f m = ( m 2 − s ) q f_m=\left(\frac{m}{2}-s\right)q f m = ( 2 m − s ) q .
Step 4. Comparing the relation (50) for the pairs ( m + 1 , n ) (m+1,n) ( m + 1 , n ) and ( m , n + 1 ) (m,n+1) ( m , n + 1 ) , we get k m + 1 − k m = q k_{m+1}-k_m=q k m + 1 − k m = q , so there is ρ ∈ C \rho\in\mathbb{C} ρ ∈ C with k m = ρ + q m k_m=\rho+qm k m = ρ + q m for all m ∈ ϵ + Z m\in\epsilon+\mathbb{Z} m ∈ ϵ + Z . Now take m = 2 n ∈ Z m=2n\in\mathbb{Z} m = 2 n ∈ Z in (52) : we get ( n + s ) ( f 2 n − k n ) = 0 (n+s)(f_{2n}-k_n)=0 ( n + s ) ( f 2 n − k n ) = 0 , where f 2 n − k n = ( n − s ) q − ρ − q n = − ( ρ + q s ) f_{2n}-k_n=(n-s)q-\rho-qn=-(\rho+qs) f 2 n − k n = ( n − s ) q − ρ − q n = − ( ρ + q s ) . Choosing n ≠ − s n\ne-s n = − s , we obtain ρ = − q s \rho=-qs ρ = − q s , i.e., k n = q ( n − s ) k_n=q(n-s) k n = q ( n − s ) .
Step 5. Let w = − q w=-q w = − q . By (7) and (35) ,
w ad Y s ( L m ) = − q ( s − m 2 ) Y m + s = f m Y m + s , w ad Y s ( Y n ) = − q ( s − n ) M n + s = k n M n + s , w\operatorname{ad}_{Y_s}(L_m)=-q\left(s-\frac{m}{2}\right)Y_{m+s}=f_mY_{m+s},\quad w\operatorname{ad}_{Y_s}(Y_n)=-q(s-n)M_{n+s}=k_nM_{n+s}, w ad Y s ( L m ) = − q ( s − 2 m ) Y m + s = f m Y m + s , w ad Y s ( Y n ) = − q ( s − n ) M n + s = k n M n + s ,
w ad Y s ( N m ) = q Y m + s = q m Y m + s and w ad Y s ( M m ) = 0. w\operatorname{ad}_{Y_s}(N_m)=qY_{m+s}=q_mY_{m+s}\text{ and }w\operatorname{ad}_{Y_s}(M_m)=0. w ad Y s ( N m ) = q Y m + s = q m Y m + s and w ad Y s ( M m ) = 0.
By Steps 1–4, D = w ad Y s D=w\operatorname{ad}_{Y_s} D = w ad Y s .
Theorem 30. A linear map D D D of s v ( ϵ ) \mathfrak{sv}(\epsilon) sv ( ϵ ) is a quasi-derivation if and only if there are x ∈ s v ( ϵ ) x\in\mathfrak{sv}(\epsilon) x ∈ sv ( ϵ ) and λ ∈ C \lambda\in\mathbb{C} λ ∈ C such that D = ad x + λ Id D=\operatorname{ad}_x+\lambda\operatorname{Id} D = ad x + λ Id . The element x x x and the number λ \lambda λ are uniquely determined by D D D , and the map associated with D D D is D ′ = ad x + 2 λ Id D'=\operatorname{ad}_x+2\lambda\operatorname{Id} D ′ = ad x + 2 λ Id . In other words,
QDer ( s v ( ϵ ) ) = Inn ( s v ( ϵ ) ) ⊕ ⟨ Id ⟩ . \operatorname{QDer}(\mathfrak{sv}(\epsilon))=\operatorname{Inn}(\mathfrak{sv}(\epsilon))\oplus\langle\operatorname{Id}\rangle. QDer ( sv ( ϵ )) = Inn ( sv ( ϵ )) ⊕ ⟨ Id ⟩ .
Proof. Clearly, ad x + λ Id \operatorname{ad}_x+\lambda\operatorname{Id} ad x + λ Id is a quasi-derivation with associated map ad x + 2 λ Id \operatorname{ad}_x+2\lambda\operatorname{Id} ad x + 2 λ Id . Conversely, let D D D be a quasi-derivation. By Propositions 28 and 29, there are complex numbers u j , v j , w j u_j,v_j,w_j u j , v j , w j (j ∈ Z j\in\mathbb{Z} j ∈ Z ), λ \lambda λ and w s ′ w'_s w s ′ (s ∈ ϵ + Z s\in\epsilon+\mathbb{Z} s ∈ ϵ + Z ) such that the even component of degree j j j of D D D equals u j ad L j + v j ad M j + w j ad N j u_j\operatorname{ad}_{L_j}+v_j\operatorname{ad}_{M_j}+w_j\operatorname{ad}_{N_j} u j ad L j + v j ad M j + w j ad N j for j ≠ 0 j\ne0 j = 0 and u 0 ad L 0 + v 0 ad M 0 + w 0 ad N 0 + λ Id u_0\operatorname{ad}_{L_0}+v_0\operatorname{ad}_{M_0}+w_0\operatorname{ad}_{N_0}+\lambda\operatorname{Id} u 0 ad L 0 + v 0 ad M 0 + w 0 ad N 0 + λ Id for j = 0 j=0 j = 0 , and the odd component of degree s s s of D D D equals w s ′ ad Y s w'_s\operatorname{ad}_{Y_s} w s ′ ad Y s . By (7) and (35) , for j ≠ 0 j\ne0 j = 0 the coefficient of D ( N 1 ) D(N_1) D ( N 1 ) at N 1 + j N_{1+j} N 1 + j equals − u j -u_j − u j , the coefficient of D ( N 0 ) D(N_0) D ( N 0 ) at M j M_j M j equals − 2 v j -2v_j − 2 v j and the coefficient of D ( M 0 ) D(M_0) D ( M 0 ) at M j M_j M j equals 2 w j 2w_j 2 w j ; moreover, the coefficient of D ( N 0 ) D(N_0) D ( N 0 ) at Y s Y_s Y s equals − w s ′ -w'_s − w s ′ . Since D ( N 0 ) D(N_0) D ( N 0 ) , D ( N 1 ) D(N_1) D ( N 1 ) and D ( M 0 ) D(M_0) D ( M 0 ) are finite linear combinations of basis elements, only finitely many of the numbers u j , v j , w j , w s ′ u_j,v_j,w_j,w'_s u j , v j , w j , w s ′ are nonzero. Hence
x = ∑ j ∈ Z ( u j L j + v j M j + w j N j ) + ∑ s ∈ ϵ + Z w s ′ Y s x=\sum_{j\in\mathbb{Z}}(u_jL_j+v_jM_j+w_jN_j)+\sum_{s\in\epsilon+\mathbb{Z}}w'_sY_s x = j ∈ Z ∑ ( u j L j + v j M j + w j N j ) + s ∈ ϵ + Z ∑ w s ′ Y s
is an element of e s v ( ϵ ) \mathfrak{esv}(\epsilon) esv ( ϵ ) , and D D D and ad x + λ Id \operatorname{ad}_x+\lambda\operatorname{Id} ad x + λ Id have the same homogeneous components, so D = ad x + λ Id D=\operatorname{ad}_x+\lambda\operatorname{Id} D = ad x + λ Id .
Let us prove the uniqueness. Assume that ad x + λ Id = 0 \operatorname{ad}_x+\lambda\operatorname{Id}=0 ad x + λ Id = 0 . Let ξ \xi ξ be the coefficient of x x x at L 0 L_0 L 0 . Comparing the coefficients at N 1 N_1 N 1 in [ x , N 1 ] + λ N 1 = 0 [x,N_1]+\lambda N_1=0 [ x , N 1 ] + λ N 1 = 0 and at N 2 N_2 N 2 in [ x , N 2 ] + λ N 2 = 0 [x,N_2]+\lambda N_2=0 [ x , N 2 ] + λ N 2 = 0 , we get − ξ + λ = 0 -\xi+\lambda=0 − ξ + λ = 0 and − 2 ξ + λ = 0 -2\xi+\lambda=0 − 2 ξ + λ = 0 , so λ = 0 \lambda=0 λ = 0 . Then ad x = 0 \operatorname{ad}_x=0 ad x = 0 , and x = 0 x=0 x = 0 by Lemma 26 . The formula for D ′ D' D ′ follows from the uniqueness of associated maps.
Corollaries
Corollary 31. Let ϵ ∈ { 0 , 1 2 } \epsilon\in\{0,\frac{1}{2}\} ϵ ∈ { 0 , 2 1 } . Then
Der ( e s v ( ϵ ) ) = Inn ( e s v ( ϵ ) ) \operatorname{Der}(\mathfrak{esv}(\epsilon))=\operatorname{Inn}(\mathfrak{esv}(\epsilon)) Der ( esv ( ϵ )) = Inn ( esv ( ϵ )) , i.e., all derivations of e s v ( ϵ ) \mathfrak{esv}(\epsilon) esv ( ϵ ) are inner (for the original extended Schrödinger–Witt algebra e s v ( 1 2 ) \mathfrak{esv}(\frac{1}{2}) esv ( 2 1 ) see [10 ] ).
QDer ( e s v ( ϵ ) ) = Der ( e s v ( ϵ ) ) ⊕ Δ 1 2 ( e s v ( ϵ ) ) \operatorname{QDer}(\mathfrak{esv}(\epsilon))=\operatorname{Der}(\mathfrak{esv}(\epsilon))\oplus\Delta_{\frac{1}{2}}(\mathfrak{esv}(\epsilon)) QDer ( esv ( ϵ )) = Der ( esv ( ϵ )) ⊕ Δ 2 1 ( esv ( ϵ )) .
Proof. By Theorem 30 and Lemma 6 , a quasi-derivation D = ad x + λ Id D=\operatorname{ad}_x+\lambda\operatorname{Id} D = ad x + λ Id is a derivation if and only if ad x + λ Id = D = D ′ = ad x + 2 λ Id \operatorname{ad}_x+\lambda\operatorname{Id}=D=D'=\operatorname{ad}_x+2\lambda\operatorname{Id} ad x + λ Id = D = D ′ = ad x + 2 λ Id , i.e., if and only if λ = 0 \lambda=0 λ = 0 . This proves the first statement. The second statement follows from the first one, Theorem 30 and Corollary 32 below (for δ = 1 2 \delta=\frac{1}{2} δ = 2 1 ).
Corollary 32. Let ϵ ∈ { 0 , 1 2 } \epsilon\in\{0,\frac{1}{2}\} ϵ ∈ { 0 , 2 1 } and δ ∈ C \delta\in\mathbb{C} δ ∈ C . Then
Δ δ ( e s v ( ϵ ) ) = { Inn ( e s v ( ϵ ) ) , δ = 1 , ⟨ Id ⟩ , δ = 1 2 , 0 , δ ∉ { 1 2 , 1 } . \Delta_\delta(\mathfrak{esv}(\epsilon))=
\begin{cases}
\operatorname{Inn}(\mathfrak{esv}(\epsilon)), & \delta=1,\\
\langle\operatorname{Id}\rangle, & \delta=\frac{1}{2},\\
0, & \delta\notin\{\frac{1}{2},1\}.
\end{cases} Δ δ ( esv ( ϵ )) = ⎩ ⎨ ⎧ Inn ( esv ( ϵ )) , ⟨ Id ⟩ , 0 , δ = 1 , δ = 2 1 , δ ∈ / { 2 1 , 1 } .
In particular, every 1 2 \frac{1}{2} 2 1 -derivation of e s v ( ϵ ) \mathfrak{esv}(\epsilon) esv ( ϵ ) is trivial (for the original extended Schrödinger–Witt algebra e s v ( 1 2 ) \mathfrak{esv}(\frac{1}{2}) esv ( 2 1 ) cf. [37 ] ).
Proof. If δ = 0 \delta=0 δ = 0 , then a 0 0 0 -derivation vanishes on [ e s v ( ϵ ) , e s v ( ϵ ) ] = e s v ( ϵ ) [\mathfrak{esv}(\epsilon),\mathfrak{esv}(\epsilon)]=\mathfrak{esv}(\epsilon) [ esv ( ϵ ) , esv ( ϵ )] = esv ( ϵ ) (Lemma 26). Let δ ≠ 0 \delta\ne0 δ = 0 and put c = δ − 1 c=\delta^{-1} c = δ − 1 . A δ \delta δ -derivation φ \varphi φ is a quasi-derivation with φ ′ = c φ \varphi'=c\varphi φ ′ = c φ . By Theorem 30, φ = ad x + λ Id \varphi=\operatorname{ad}_x+\lambda\operatorname{Id} φ = ad x + λ Id and ad x + 2 λ Id = φ ′ = c ad x + c λ Id \operatorname{ad}_x+2\lambda\operatorname{Id}=\varphi'=c\operatorname{ad}_x+c\lambda\operatorname{Id} ad x + 2 λ Id = φ ′ = c ad x + c λ Id , i.e., ad ( 1 − c ) x + ( 2 − c ) λ Id = 0 \operatorname{ad}_{(1-c)x}+(2-c)\lambda\operatorname{Id}=0 ad ( 1 − c ) x + ( 2 − c ) λ Id = 0 . By the uniqueness part of Theorem 30, ( 1 − c ) x = 0 (1-c)x=0 ( 1 − c ) x = 0 and ( 2 − c ) λ = 0 (2-c)\lambda=0 ( 2 − c ) λ = 0 . If δ = 1 \delta=1 δ = 1 , then λ = 0 \lambda=0 λ = 0 and φ = ad x \varphi=\operatorname{ad}_x φ = ad x ; if δ = 1 2 \delta=\frac{1}{2} δ = 2 1 , then x = 0 x=0 x = 0 and φ = λ Id \varphi=\lambda\operatorname{Id} φ = λ Id ; otherwise x = 0 x=0 x = 0 and λ = 0 \lambda=0 λ = 0 . Conversely, inner derivations are 1 1 1 -derivations and Id \operatorname{Id} Id is a 1 2 \frac{1}{2} 2 1 -derivation.
To describe the biderivations of e s v ( ϵ ) \mathfrak{esv}(\epsilon) esv ( ϵ ) we need the following lemma on its quasicentroid (in the sense of [24 ] ).
Lemma 33. Let T T T be a linear map of e s v ( ϵ ) \mathfrak{esv}(\epsilon) esv ( ϵ ) such that [ T ( x ) , y ] = [ x , T ( y ) ] [T(x),y]=[x,T(y)] [ T ( x ) , y ] = [ x , T ( y )] for all x , y ∈ e s v ( ϵ ) x,y\in\mathfrak{esv}(\epsilon) x , y ∈ esv ( ϵ ) . Then T = β Id T=\beta\operatorname{Id} T = β Id for some β ∈ C \beta\in\mathbb{C} β ∈ C .
Proof. The identity [ T ( x ) , y ] = [ x , T ( y ) ] [T(x),y]=[x,T(y)] [ T ( x ) , y ] = [ x , T ( y )] is homogeneous, so, as in Lemma 7 , every homogeneous component of T T T satisfies it. Let T T T be even of degree j j j , and write
T ( L m ) = a m L m + j + b m M m + j + r m N m + j , T ( Y n ) = c n Y n + j , T ( M m ) = d m L m + j + e m M m + j + p m N m + j , T ( N m ) = x m L m + j + y m M m + j + z m N m + j . \begin{aligned}
T(L_m)&=a_mL_{m+j}+b_mM_{m+j}+r_mN_{m+j},\quad T(Y_n)=c_nY_{n+j},\\
T(M_m)&=d_mL_{m+j}+e_mM_{m+j}+p_mN_{m+j},\quad T(N_m)=x_mL_{m+j}+y_mM_{m+j}+z_mN_{m+j}.
\end{aligned} T ( L m ) T ( M m ) = a m L m + j + b m M m + j + r m N m + j , T ( Y n ) = c n Y n + j , = d m L m + j + e m M m + j + p m N m + j , T ( N m ) = x m L m + j + y m M m + j + z m N m + j .
Comparing the coefficients at the basis elements indicated in brackets in [ T ( X ) , Z ] = [ X , T ( Z ) ] [T(X),Z]=[X,T(Z)] [ T ( X ) , Z ] = [ X , T ( Z )] for the indicated pairs ( X , Z ) (X,Z) ( X , Z ) of basis elements, we get
n x m + m x n = 0 [ ( N m , N n ) , N m + n + j ] , y m = 0 [ ( N m , N m ) , M 2 m + j ] , n a m = ( n + j ) z n [ ( L m , N n ) , N m + n + j ] , 2 b m = j y 0 [ ( L m , N 0 ) , M m + j ] , 2 r m = n a m − ( n + j ) e n [ ( L m , M n ) , M m + n + j ] , e m − z n = − m 2 x n [ ( M m , N n ) , M m + n + j ] , c m − z n = ( n + j 2 − m ) x n [ ( Y m , N n ) , Y m + n + j ] , ( m + j 2 − n ) d m + p m = 0 [ ( M m , Y n ) , Y m + n + j ] . \begin{aligned}
nx_m+mx_n&=0 &&[(N_m,N_n),N_{m+n+j}],&
y_m&=0 &&[(N_m,N_m),M_{2m+j}],\\
na_m&=(n+j)z_n &&[(L_m,N_n),N_{m+n+j}],&
2b_m&=jy_0 &&[(L_m,N_0),M_{m+j}],\\
2r_m&=na_m-(n+j)e_n &&[(L_m,M_n),M_{m+n+j}],&
e_m-z_n&=-\frac{m}{2}x_n &&[(M_m,N_n),M_{m+n+j}],\\
c_m-z_n&=\left(\frac{n+j}{2}-m\right)x_n &&[(Y_m,N_n),Y_{m+n+j}],&
\left(\frac{m+j}{2}-n\right)d_m+p_m&=0 &&[(M_m,Y_n),Y_{m+n+j}].
\end{aligned} n x m + m x n n a m 2 r m c m − z n = 0 = ( n + j ) z n = n a m − ( n + j ) e n = ( 2 n + j − m ) x n [( N m , N n ) , N m + n + j ] , [( L m , N n ) , N m + n + j ] , [( L m , M n ) , M m + n + j ] , [( Y m , N n ) , Y m + n + j ] , y m 2 b m e m − z n ( 2 m + j − n ) d m + p m = 0 = j y 0 = − 2 m x n = 0 [( N m , N m ) , M 2 m + j ] , [( L m , N 0 ) , M m + j ] , [( M m , N n ) , M m + n + j ] , [( M m , Y n ) , Y m + n + j ] .
The first relation with m = n = 1 m=n=1 m = n = 1 and then with n = 1 n=1 n = 1 gives x m = 0 x_m=0 x m = 0 for all m m m . Hence e m = z n = c k e_m=z_n=c_k e m = z n = c k for all m , n ∈ Z m,n\in\mathbb{Z} m , n ∈ Z , k ∈ ε + Z k\in\varepsilon+\mathbb{Z} k ∈ ε + Z ; we denote this number by β j \beta_j β j . Moreover, b m = 0 b_m=0 b m = 0 ; the third relation with n = 0 n=0 n = 0 and n = 1 n=1 n = 1 gives j β j = 0 j\beta_j=0 j β j = 0 and a m = ( 1 + j ) β j = β j a_m=(1+j)\beta_j=\beta_j a m = ( 1 + j ) β j = β j ; the fifth relation with n = 0 n=0 n = 0 gives 2 r m = − j β j = 0 2r_m=-j\beta_j=0 2 r m = − j β j = 0 ; and the last relation for two different values of n n n gives d m = p m = 0 d_m=p_m=0 d m = p m = 0 . Thus T = β j Id T=\beta_j\operatorname{Id} T = β j Id , and β j = 0 \beta_j=0 β j = 0 if j ≠ 0 j\ne0 j = 0 . Let T T T be odd of degree s s s , and write
T ( L m ) = f m Y m + s , T ( M m ) = g m Y m + s , T ( N m ) = q m Y m + s , T ( Y n ) = h n L n + s + k n M n + s + l n N n + s . \begin{aligned}
T(L_m)&=f_mY_{m+s},\quad T(M_m)=g_mY_{m+s},\quad T(N_m)=q_mY_{m+s},\\
T(Y_n)&=h_nL_{n+s}+k_nM_{n+s}+l_nN_{n+s}.
\end{aligned} T ( L m ) T ( Y n ) = f m Y m + s , T ( M m ) = g m Y m + s , T ( N m ) = q m Y m + s , = h n L n + s + k n M n + s + l n N n + s .
In the same way, the pairs
( N m , N m ) , ( L m , N 0 ) , ( M m , N 0 ) , ( Y n , N 0 ) , ( Y n , N 1 ) and ( Y n , M 0 ) , (N_m,N_m),\ (L_m,N_0),\ (M_m,N_0),\ (Y_n,N_0),\ (Y_n,N_1)\text{ and }(Y_n,M_0), ( N m , N m ) , ( L m , N 0 ) , ( M m , N 0 ) , ( Y n , N 0 ) , ( Y n , N 1 ) and ( Y n , M 0 ) ,
compared at Y 2 m + s Y_{2m+s} Y 2 m + s , Y m + s Y_{m+s} Y m + s , Y m + s Y_{m+s} Y m + s , M n + s M_{n+s} M n + s , N n + s + 1 N_{n+s+1} N n + s + 1 and M n + s M_{n+s} M n + s , respectively, give successively q m = 0 q_m=0 q m = 0 , f m = 0 f_m=0 f m = 0 , g m = 0 g_m=0 g m = 0 , k n = 0 k_n=0 k n = 0 , h n = 0 h_n=0 h n = 0 and l n = 0 l_n=0 l n = 0 . Hence T = 0 T=0 T = 0 . Therefore T T T is equal to its even component of degree 0 0 0 , which is a multiple of Id \operatorname{Id} Id .
Corollary 34
Corollary 34. Let ε ∈ { 0 , 1 2 } \varepsilon\in\{0,\frac{1}{2}\} ε ∈ { 0 , 2 1 } and δ ∈ C \delta\in\mathbb{C} δ ∈ C .
If δ ≠ 1 \delta\ne1 δ = 1 , then every δ \delta δ -biderivation of esv ( ε ) \operatorname{esv}(\varepsilon) esv ( ε ) is zero.
A bilinear map ϕ \phi ϕ of esv ( ε ) \operatorname{esv}(\varepsilon) esv ( ε ) is a biderivation if and only if there is β ∈ C \beta\in\mathbb{C} β ∈ C such that ϕ ( x , y ) = β [ x , y ] \phi(x,y)=\beta[x,y] ϕ ( x , y ) = β [ x , y ] for all x , y ∈ esv ( ε ) x,y\in\operatorname{esv}(\varepsilon) x , y ∈ esv ( ε ) .
Proof. The first statement is proved in the same way as the first part of Corollary 23 , using Corollary 32 instead of Corollary 21 . Let ϕ \phi ϕ be a biderivation. By Corollary 31, for every z z z there are a ( z ) , b ( z ) ∈ esv ( ε ) a(z),b(z)\in\operatorname{esv}(\varepsilon) a ( z ) , b ( z ) ∈ esv ( ε ) such that ϕ ( x , z ) = [ a ( z ) , x ] \phi(x,z)=[a(z),x] ϕ ( x , z ) = [ a ( z ) , x ] and ϕ ( z , x ) = [ b ( z ) , x ] \phi(z,x)=[b(z),x] ϕ ( z , x ) = [ b ( z ) , x ] for all x x x . Since Z ( esv ( ε ) ) = 0 Z(\operatorname{esv}(\varepsilon))=0 Z ( esv ( ε )) = 0 , the elements a ( z ) , b ( z ) a(z),b(z) a ( z ) , b ( z ) are uniquely determined, and the maps a a a and b b b are linear. For all x , y x,y x , y we have
[ ( a + b ) ( x ) , y ] + [ x , ( a + b ) ( y ) ] = ϕ ( y , x ) + ϕ ( x , y ) − ϕ ( x , y ) − ϕ ( y , x ) = 0 , [(a+b)(x),y]+[x,(a+b)(y)]=\phi(y,x)+\phi(x,y)-\phi(x,y)-\phi(y,x)=0, [( a + b ) ( x ) , y ] + [ x , ( a + b ) ( y )] = ϕ ( y , x ) + ϕ ( x , y ) − ϕ ( x , y ) − ϕ ( y , x ) = 0 ,
so a + b a+b a + b is a quasi-derivation with associated map 0 0 0 . By Theorem 30, a + b = ad w + μ Id a+b=\operatorname{ad}_w+\mu\operatorname{Id} a + b = ad w + μ Id with ad w + 2 μ Id = 0 \operatorname{ad}_w+2\mu\operatorname{Id}=0 ad w + 2 μ Id = 0 , whence w = 0 w=0 w = 0 , μ = 0 \mu=0 μ = 0 and a = − b a=-b a = − b . Therefore [ b ( x ) , y ] = ϕ ( x , y ) = [ a ( y ) , x ] = [ x , b ( y ) ] [b(x),y]=\phi(x,y)=[a(y),x]=[x,b(y)] [ b ( x ) , y ] = ϕ ( x , y ) = [ a ( y ) , x ] = [ x , b ( y )] for all x , y x,y x , y , and b = β Id b=\beta\operatorname{Id} b = β Id by Lemma 33. Hence ϕ ( x , y ) = β [ x , y ] \phi(x,y)=\beta[x,y] ϕ ( x , y ) = β [ x , y ] . Conversely, β [ x , y ] \beta[x,y] β [ x , y ] is a biderivation by the Jacobi identity.
Corollary 35
Corollary 35. Let ε ∈ { 0 , 1 2 } \varepsilon\in\{0,\frac{1}{2}\} ε ∈ { 0 , 2 1 } . For every δ ∈ C \delta\in\mathbb{C} δ ∈ C , all transposed δ \delta δ -Poisson structures on esv ( ε ) \operatorname{esv}(\varepsilon) esv ( ε ) are trivial. In particular, all transposed Poisson structures on esv ( ε ) \operatorname{esv}(\varepsilon) esv ( ε ) are trivial (for the original extended Schrödinger–Witt algebra esv ( 1 2 ) \operatorname{esv}(\frac{1}{2}) esv ( 2 1 ) cf. [37 ] ).
Proof. By Lemma 5 , ϕ ( x , y ) = x ⋅ y \phi(x,y)=x\cdot y ϕ ( x , y ) = x ⋅ y is a δ \delta δ -biderivation. If δ ≠ 1 \delta\ne1 δ = 1 , then ϕ = 0 \phi=0 ϕ = 0 by Corollary 34. If δ = 1 \delta=1 δ = 1 , then x ⋅ y = β [ x , y ] x\cdot y=\beta[x,y] x ⋅ y = β [ x , y ] , and the commutativity of ⋅ \cdot ⋅ gives β L 1 = β [ L 1 , L 0 ] = β [ L 0 , L 1 ] = − β L 1 \beta L_1=\beta[L_1,L_0]=\beta[L_0,L_1]=-\beta L_1 β L 1 = β [ L 1 , L 0 ] = β [ L 0 , L 1 ] = − β L 1 , so β = 0 \beta=0 β = 0 .
Remark 36. The identity (5) is sometimes normalized as δ z ⋅ [ x , y ] = [ z ⋅ x , y ] + [ x , z ⋅ y ] \delta z\cdot[x,y]=[z\cdot x,y]+[x,z\cdot y] δ z ⋅ [ x , y ] = [ z ⋅ x , y ] + [ x , z ⋅ y ] . For δ ≠ 0 \delta\ne0 δ = 0 this is (5) with δ \delta δ replaced by δ − 1 \delta^{-1} δ − 1 , and for δ = 0 \delta=0 δ = 0 it means that every multiplication operator D = z ⋅ ( − ) D=z\cdot(-) D = z ⋅ ( − ) is a quasi-derivation with D ′ = 0 D'=0 D ′ = 0 ; by Theorem 30, D = ad x + λ Id D=\operatorname{ad}_x+\lambda\operatorname{Id} D = ad x + λ Id with 0 = D ′ = ad x + 2 λ Id 0=D'=\operatorname{ad}_x+2\lambda\operatorname{Id} 0 = D ′ = ad x + 2 λ Id , whence x = 0 x=0 x = 0 and λ = 0 \lambda=0 λ = 0 by the uniqueness part of Theorem 30. Hence Corollary 35 also holds in this normalization, for every δ ∈ C \delta\in\mathbb{C} δ ∈ C .