Processing math: 59%
Research article

Triangular algebras with nonlinear higher Lie n-derivation by local actions

  • Received: 14 November 2023 Revised: 07 December 2023 Accepted: 10 December 2023 Published: 26 December 2023
  • MSC : 15A78, 16W25, 17B40

  • This paper was devoted to the study of the so-called nonlinear higher Lie n-derivation of triangular algebras T, where n is a nonnegative integer greater than two. Under some mild conditions, we proved that every nonlinear higher Lie n-derivation by local actions on the triangular algebras is of a standard form. As an application, we gave a characterization of higher Lie n-derivation by local actions on upper triangular matrix algebras, block upper triangular matrix algebras and nest algebras, respectively.

    Citation: Xinfeng Liang, Mengya Zhang. Triangular algebras with nonlinear higher Lie n-derivation by local actions[J]. AIMS Mathematics, 2024, 9(2): 2549-2583. doi: 10.3934/math.2024126

    Related Papers:

    [1] He Yuan, Qian Zhang, Zhendi Gu . Characterizations of generalized Lie $ n $-higher derivations on certain triangular algebras. AIMS Mathematics, 2024, 9(11): 29916-29941. doi: 10.3934/math.20241446
    [2] Junyuan Huang, Xueqing Chen, Zhiqi Chen, Ming Ding . On a conjecture on transposed Poisson $ n $-Lie algebras. AIMS Mathematics, 2024, 9(3): 6709-6733. doi: 10.3934/math.2024327
    [3] Xiuhai Fei, Zhonghua Wang, Cuixian Lu, Haifang Zhang . Higher Jordan triple derivations on $ * $-type trivial extension algebras. AIMS Mathematics, 2024, 9(3): 6933-6950. doi: 10.3934/math.2024338
    [4] Guangyu An, Xueli Zhang, Jun He, Wenhua Qian . Characterizations of local Lie derivations on von Neumann algebras. AIMS Mathematics, 2022, 7(5): 7519-7527. doi: 10.3934/math.2022422
    [5] Khalil Hadi Hakami, Junaid Nisar, Kholood Alnefaie, Moin A. Ansari . Characterizing N-type derivations on standard operator algebras by local actions. AIMS Mathematics, 2024, 9(9): 25319-25332. doi: 10.3934/math.20241236
    [6] Shan Li, Kaijia Luo, Jiankui Li . Generalized Lie $ n $-derivations on generalized matrix algebras. AIMS Mathematics, 2024, 9(10): 29386-29403. doi: 10.3934/math.20241424
    [7] He Yuan, Zhuo Liu . Lie $ n $-centralizers of generalized matrix algebras. AIMS Mathematics, 2023, 8(6): 14609-14622. doi: 10.3934/math.2023747
    [8] Dan Liu, Jianhua Zhang, Mingliang Song . Local Lie derivations of generalized matrix algebras. AIMS Mathematics, 2023, 8(3): 6900-6912. doi: 10.3934/math.2023349
    [9] Xinfeng Liang, Lingling Zhao . Bi-Lie n-derivations on triangular rings. AIMS Mathematics, 2023, 8(7): 15411-15426. doi: 10.3934/math.2023787
    [10] Senrong Xu, Wei Wang, Jia Zhao . Twisted Rota-Baxter operators on Hom-Lie algebras. AIMS Mathematics, 2024, 9(2): 2619-2640. doi: 10.3934/math.2024129
  • This paper was devoted to the study of the so-called nonlinear higher Lie n-derivation of triangular algebras T, where n is a nonnegative integer greater than two. Under some mild conditions, we proved that every nonlinear higher Lie n-derivation by local actions on the triangular algebras is of a standard form. As an application, we gave a characterization of higher Lie n-derivation by local actions on upper triangular matrix algebras, block upper triangular matrix algebras and nest algebras, respectively.



    In this paper, we assume that A is an unital algebra over commutative ring R and Z(A) is the center of A. The main purpose of this paper is to study the structure of nonlinear higher Lie n-derivation by local actions on triangular algebras. To achieve this goal, we first introduce some definitions related to nonlinear higher Lie n-derivation by local actions.

    Let N be the set of all nonnegative integers and Δ={δm}mN be a family of R-linear (resp. nonlinear) mapping δm:AA on A such that δ0=idA. Δ is called:

    a) a (resp. nonlinear) higher derivation if

    δm(xy)=i+j=mδi(x)δj(y) (1.1)

    for all x,yA and for each nN;

    b) a (resp. nonlinear) higher Lie n-derivation if

    δm(Pn(x1,,xn))=i1++in=mPn(δi1(x1),δi2(x2),,δin(xn)), (1.2)

    where Pn(x1,,xn)=[Pn1(x1,,xn1),xn] is a polynomial defined by induction with variables x1,,xn for all x1,,xnA; the symbol [x1,x2]=x1x2x2x1 is called the Lie product. If Eq (1.2) holds only under condition x1xn=0 for all x1,,xnA and for each m,nN, Δ={δm}mN is said to be higher Lie n-derivation by local actions. That is, Δ is said to be higher Lie n-derivation by local actions if every mapping δm satisfies the equation

    δm(Pn(x1,,xn))=i1++in=mPn(δi1(x1),δi2(x2),,δin(xn)) (1.3)

    for all x1,,xnA with x1xn=0 and m,nN. It should be noted that the study of derivatives satisfying local properties originated from papers[1,2]. For n=2,3 and any positive integer m in (1.2) (resp. (1.3)), Δ is referred to as Lie high derivation (resp. by local actions) and Lie triple high derivation (resp. by local actions), respectively. If m=1 and any positive integer n in Eq (1.2) (resp. (1.3)), then the mapping δ1 is called Lie n-derivation (resp. by local actions). Therefore, in the sense of Herstein Lie type mapping, higher Lie n-derivation is a natural extension of Herstein Lie type mappings. From the structure of maps (1.1)–(1.3), it can be seen that the sum of the higher derivations and the central map of annihilation Pn(x1,,xn) is still higher Lie n-derivation (resp. by local actions), for all x1,,xnA (resp. with x1xn=0). If every higher Lie n-derivation has this decomposition form, it is said that higher Lie n-derivation has a standard form. Under this framework, some special situations of (nonlinear) higher Lie n-derivation have been studied by many scholars, (nonlinear) Lie triple derivation in paper [3,4], (nonlinear) Lie higher derivations in papers [5,6,7], Lie higher derivations studied in paper[8], higher derivations [9,10] etc.

    In the author's knowledge system, the study of the structure of higher Lie n-derivation by local actions satisfying Eq (1.3) over rings or algebras has attracted many scholars to study among the many extensions of Lie-n derivation (see [11,12,13,14,15,16,17,18,19]. In 2011, Ji and Qi[11] studied the structural form of linear Lie derivation by local actions on triangular algebras (in (1.3), where (m,n)=(1,2)), and proved that each Lie derivation by local action has a standard form. Subsequently, Lin[12] extended this to three Lie higher derivation by local actions (in (1.3), m was an arbitrary positive integer and n = 2) and obtained that each Lie derivation by local action has a standard form. At the same time, in recent years, the authors and collaborators have found that many scholars have studied the structural problems of some special cases of nonlinear higher Lie n-derivation by local actions defined by equation (1.3) on rings or algebras. Liu in [15] worked the structure of Lie triple derivations by local actions satisfying the condition x1x2x3Ω={0,p} (in Relation (1.3) with n=3 and m=1), where p is a fixed nontrivial projection of factor von Neumann algebra M with dimension greater than one. He showed that every Lie triple derivations by local actions be of standard form, for von Neumann algebra with no central abelian projections M. In 2021, Zhao[16] considered the structure of nonlinear Lie triple derivations by local actions (in Relation (1.3) with n=3 and m=1) on triangular algebra. He confirmed that every nonlinear Lie triple derivations by local actions be of standard form. On the basis of his work[16], the first authors and collaborators extended the structure of nonlinear Lie triple derivations [16] to nonlinear Lie triple higher derivations by local actions[17] (condition: For arbitrary mN and n=3 in (1.3)), and proved that each nonlinear Lie triple higher derivation by local actions has a standard form (see Eq (1.4)) under the same conditions as Zhao[16]. Inspired by the above results[16,17], it is natural to consider the structure form of the higher Lie n-derivation by local actions (the case: mN and n>2 in Eq (1.3)) on triangular algebras. The results of this paper generalize Zhao[16] and Liang[17] to a more general form: For arbitrary mN and n3. It should be noted that we are temporarily unable to find a method to prove the structural form of nonlinear Lie higher derivations by local actions on triangular algebras. This is also an open problem left over in this article.

    In this paper we established the higher Lie n-derivation by local actions (n>2) on triangular algebras. Let T be a triangular algebra over a commutative ring R. Under some mild conditions, we prove that if a family Δ={δm}mN of nonlinear mappings δm on T satisfies the condition (1.3), then there exists an additive higher derivation D={χm}mN and a nonlinear mapping hm:TT on T vanishing all Pn(x1,,xn) for all x1,,xnT with x1xn=0, such that

    δm(x)=χm(x)+hm(x).

    Next, we immediately apply our results to typical examples of triangular algebra: Upper triangular matrix algebras, block upper triangular matrix algebras and nest algebras. At the same time, our conclusion generalizes the conclusions of papers [16,Theorems 2.1 and 2.2] and [17,Theorems 1 and 3].

    In this part, we introduce some basic theories of triangular algebra. In 2001, triangular algebra was first introduced by Cheung[20].

    Based on algebra A with identity 1A and algebra B with identity 1B, defined on ring R and faithful (A,B)-bimodule M, for aA, aM={0} implies a=0 and for bB, Mb={0} implies b=0. Cheung introduced a set

    T=[AM0B]={[am0b]| aA,mM,bB}.

    According to the addition and multiplication properties of matrices, set T is an associative and noncommutative R-algebra. This algebra is called triangular algebra. The most classical examples of triangular algebras are upper triangular matrix algebras, nest algebras and block upper triangular matrix algebras (see [7,20,21] for details). Furthermore, the center Z(T) of T is (see [20,22])

    Z(T)={[a00b]|am=mb,  mM}. ()

    Let us define two natural R-linear projections πA:TA and πB:TB by

    πA:[am0b]aandπB:[am0b]b.

    It follows from simple calculation that πA(Z(T)) is a subalgebra of Z(A) and that πB(Z(T)) is a subalgebra of Z(B). Additionally, there exists a unique algebraic isomorphism τ:πA(Z(T))πB(Z(T)) such that am=mτ(a) for all aπA(Z(T)) and for all mM.

    Regarding the center of algebra A, we need to make the following notes. In 2012, Benkovic and Eremita[23] introduced the following useful condition: For an arbitrary R-algebra A:

    [x,A]Z(A)xZ(A),   xA. ()

    This amounts to saying that

    [[x,A],A]=0[x,A]=0Z(A),   xA.

    Note that () is equivalent to the condition that there does not exist nonzero central inner derivations on A. The usual examples of algebras satisfying () are upper triangular matrix algebras, block upper triangular matrix algebras and nest algebras (see [24]).

    This section examines the structure of Lie n-derivations on triangular algebras by local operations at zero product. To put it more specifically, we demonstrate that any nonlinear Lie n-derivations by local actions at zero product have a standard form under mild conditions after first demonstrating that the nonlinear Lie n-derivations by local actions at zero product are an additive mapping of module Z(T). The information will provide the generalized version matching to [16,Theorems 2.1 and 2.2].

    Theorem 3.1. Let T=[AMOB] be a triangular algebra satisfying πA(Z(T))=Z(A) and πB(Z(T))=Z(B)). Suppose that a mapping Ln:TT (n3) is a nonlinear map satisfying

    Ln(Pn(x1,x2,,xn))=ni=1Pn(x1,x2,,δ1(xi),,xn)

    for all x1,x2,,xnT with x1x2xn=0, then for every nN,

    Ln(x+y)Ln(x)Ln(y)Z(T)

    for all x,yT.

    For convenience, let us write A11=A, A22=B and A12=M, then triangular algebra T=[AMOB] can be rewritten by T=[A11A12OA22].

    In order to facilitate readers' understanding, we will divide the proof process into the following lemmas.

    Lemma 3.1. [16,Claim 1] Let aiiAii,i{1,2}. If a11m12=m12a22 for all m12A12, then a11a22Z(T).

    Lemma 3.2. Ln(0)=0.

    In particular, take xi=0 in formula (1.1) for i{1,2,,n}.

    Lemma 3.3. Let aijAij, for 1ij2, then

    1) Ln(a11+a12)Ln(a11)Ln(a12)Z(T);

    2) Ln(a22+a12)Ln(a22)Ln(a12)Z(T).

    Proof. 1) Let a11A11 and a12,c12A12. Denote T=Ln(a11+a12)Ln(a11)Ln(a12). It is clear that the elements a11,a12,c12 and idempotents p1,p2 satisfy the relations a12(a11+c12)p1p2 p2n3 copies=0=a12a11p1p2p2n3 copies=0=a12c12p1p2p2n3 copies, then we have

    Ln(a11a12)=Ln(Pn(a12,a11+c12,p1,p2p2n3 copies))=Pn(Ln(a12),a11+c12,p1,p2p2n3 copies)+Pn(a12,Ln(a11+c12),p1,p2p2n3 copies)+Pn(a12,a11+c12,Ln(p1),p2p2n3 copies)+ni=4Pn(a12,a11+c12,p1,p2,,Ln(p2)ith component,,p2),

    and on the other hand, we obtain

    Ln(a11a12)=Ln(Pn(a12,a11,p1,p2,,p2n3 copies))+Ln(Pn(a12,c12,p1,p2,,p2n3 copies))=Pn(Ln(a12),a11+c12,p1,p2p2n3 copies)+Pn(a12,Ln(a11)+Ln(c12),p1,p2,,p2n3 copies)+Pn(a12,a11+c12,Ln(p1),p2,,p2n3 copies)+ni=4Pn(a12,a11+c12,p1,p2,,Ln(p2)ith component,,p2).

    The above two equations can be lead to Pn(a12,T,p1,p2p2n3 copies)=0, which is p1Tp1a12=a12p2Tp2. It follows from Lemma 3.1 that

    p1Tp1p2Tp2Z(T). (3.1)

    Let's prove that p1Tp2=0. Since the elements a11,a12 and idempotents p1,p2 satisfy the relation p2(a11+a12)p1p2 p2n3 copies=0=p2a11p1p2p2n3 copies=0=p2a12p1p2p2n3 copies, we study the form of elements Ln(a12) form two perspectives, namely,

    Ln(a12)=Ln(Pn(p2,a11+a12,p1,p2p2n3 copies))=Pn(Ln(p2),a11+a12,p1,p2p2n3 copies)+Pn(p2,Ln(a11+a12),p1,p2p2n3 copies)+Pn(p2,a11+a12,Ln(p1),p2p2n3 copies)+ni=4Pn(p2,a11+a12,p1,p2,,Ln(p2)ith component,,p2)

    and

    Ln(a12)=Ln(Pn(p2,a11,p1,p2p2n3 copies))+Ln(Pn(p2,a12,p1,p2p2n3 copies))=Pn(Ln(p2),a11+a12,p1,p2p2n3 copies)+Pn(p2,Ln(a11)+Ln(a12),p1,p2p2n3 copies)+Pn(p2,a11+a12,Ln(p1),p2p2n3 copies)+ni=4Pn(p2,a11+a12,p1,p2,,Ln(p2)ith component,,p2)

    for all a11A11,a12A12. It follows from the above two equations that Pn(p2,T,p1,p2p2n3 copies)=0, which is p1Tp2=0. Combining with relation (3.1), we have

    Ln(a11+a12)Ln(a11)Ln(a12)Z(T)

    for all aijAij,1ij2.

    Through a similar calculation process, we can conclude that (2) holds.

    Lemma 3.4. Let a12,c12A12, then Ln(a12+c12)=Ln(a12)+Ln(c12).

    Proof. Combining (a12p1)(p2+c12)p1p2p2n3 copies=0 with Lemmas 3.2 and 3.3, we have

    Ln(a12+c12)=Ln(Pn(a12p1,p2+c12,p1,p2,,p2))=Pn(Ln(a12)+Ln(p1),p2+c12,p1,p2,,p2)+Pn(a12p1,Ln(p2)+Ln(c12),p1,p2,,p2)+Pn(a12p1,p2+c12,Ln(p1),p2,,p2)+ni=4Pn(a12p1,p2+c12,p1,p2,,Ln(p2)ith component,,p2)=Ln(Pn(a12,p2,p1,p2,,p2))+Ln(Pn(a12,c12,p1,p2,,p2))+Ln(Pn(p1,p2,p1,p2,,p2))+Ln(Pn(p1,c12,p1,p2,p2,,p2))=Ln(a12)+Ln(c12)

    for all a12,c12A12.

    Lemma 3.5. Let aiiAii,i{1,2}, then δ1(aii+bii)=δ1(aii)+δ1(bii)+Zaii,bii for some central element Zaii,biiZ(T).

    Proof. We will only prove the case with i=1. The proof of the case i=2 can be proved through a similar process.

    Let a11,b11A11,w12A12. Denote U=δ1(a11+b11)δ1(a11)δ1(b11). It follows from w12p1(a11+b11)p2p2n3 copies=w12p1a11p2p2n3 copies=w12p1b11p2p2n3 copies=0 that

    Ln((a11+b11)w12)=Ln(Pn(w12,p1,(a11+b11),p2,,p2))=Pn(Ln(w12),p1,(a11+b11),p2,,p2)+Pn(w12,Ln(p1),(a11+b11),p2,,p2)+Pn(w12,p1,Ln(a11+b11),p2,,p2)+ni=4Pn(w12,p1,(a11+b11),p2,,Ln(p2)ith component,,p2)

    and

    Ln((a11+b11)w12)=Ln(a11w12)+Ln(b11w12)=Ln(Pn(w12,p1,a11,p2,,p2))+Ln(Pn(w12,p1,b11),p2,,p2))=Pn(Ln(w12),p1,(a11+b11),p2,,p2)+Pn(w12,Ln(p1),(a11+b11),p2,,p2)+Pn(w12,p1,Ln(a11)+Ln(b11),p2,,p2)+ni=4Pn(w12,p1,(a11+b11),p2,,Ln(p2)ith component,,p2).

    By observing the two equations above, we have Pn(w12,p1,U,p2,,p2)=0, which is w12U=Uw12. It follows from Lemma 3.1 that

    p1Up1p2Up2Z(T). (3.2)

    In the rest of the lemma, we prove that the equation p1Up2=0 holds.

    Since the equations (a11+b11)p2p2n1 copies=a11p2p2n1 copies=b11p2p2n1 copies=0 hold, then we have

    0=Ln(Pn(a11+b11,p2,,p2n1 copies))=Pn(Ln(a11+b11),p2,,p2n1 copies)+ni=2Pn(a11+b11,p2,,Ln(p2)ith component,,p2n1 copies)

    and

    0=Ln(Pn(a11,p2,,p2n1 copies))+Ln(Pn(b11,p2,,p2n1 copies))=Pn(Ln(a11)+Ln(b11),p2,,p2n1 copies)+ni=2Pn(a11+b11,p2,,Ln(p2)ith component,,p2n1 copies).

    By observing the two equations above, we have Pn(U,p2,,p2n1 copies)=0, which is p1Up2=0. Combining with Eq (3.2), we have

    Ln(a11+b11)Ln(a11)Ln(b11)Z(T)

    for all a11,b11A11.

    Lemma 3.6. Let aijAij,1ij2, then Ln(a11+a12+a22)=Ln(a11)+Ln(a12)+Ln(a22)+Za11,a12,a22.

    Proof. Let aijAij,1ij2. Denote U=Ln(a11+a12+a22)Ln(a11)Ln(a12)Ln(a22). With the help of the fact that equation (a11+a12+a22)c12p1p2p2n3 copies=0 holds, we have

    Ln(c12a22a11c12)=Ln(Pn(a11+a12+a22,c12,p1,p2,p2n3 copies))=Pn(Ln(a11+a12+a22),c12,p1,p2,,p2n3 copies)+Pn(a11+a12+a22,Ln(c12),p1,p2,,p2n3 copies)+Pn(a11+a12+a22,c12,Ln(p1),p2,,p2n3 copies)+ni=4Pn(a11+a12+a22,c12,p1,p2,,Ln(p2)ith component,,p2n3 copies)

    and

    Ln(c12a22a11c12)=Ln(c12a22)+Ln(a11c12)=Ln(Pn(a11,c12,p1,p2,p2n3 copies))+Ln(Pn(a12,c12,p1,p2,p2n3 copies))+Ln(Pn(a22,c12,p1,p2,p2n3 copies))=Pn(Ln(a11)+Ln(a12)+Ln(a22),c12,p1,p2,,p2n3 copies)+Pn(a11+a12+a22,Ln(c12),p1,p2,,p2n3 copies)+Pn(a11+a12+a22,c12,Ln(p1),p2,,p2n3 copies)+ni=4Pn(a11+a12+a22,c12,p1,p2,,Ln(p2)ith component,,p2n3 copies).

    By observing the two equations above, we have Pn(U,c12,p1,p2,,p2n3 copies)=0, which is c12p2Up2=p1Up1c12. We have

    p1Up1p2Up2Z(T). (3.3)

    For the rest, we prove the conclusion: p1Up2=0. It is clear that (a11+a12+a22)(p1)p2p2n2 copies=0. We then obtain

    Ln(a12)=Ln(Pn(a11+a12+a22,p1,p2,,p2))=Pn(Ln(a11+a12+a22),p1,p2,,p2)+Pn(a11+a12+a22,Ln(p1),p2,,p2)+ni=3Pn(a11+a12+a22,p1,p2,,Ln(p2)ith component,,p2)

    and

    Ln(a12)=Ln(Pn(a11,p1,p2,,p2))+Ln(Pn(a12,p1,p2,,p2))+Ln(Pn(a22,p1,p2,,p2))=Pn(Ln(a11)+Ln(a12)+Ln(a22)),p1,p2,,p2)+Pn(a11+a12+a22,Ln(p1),p2,,p2)+ni=3Pn(a11+a12+a22,p1,p2,,Ln(p2)ith component,,p2).

    By observing the two equations above, we have Pn(U,p1,p2,,p2n2 copies)=0, which is p1Up2=0. Combining with Eq (3.3), we have

    Ln(a11+a12+a22)=Ln(a11)+Ln(a12)+Ln(a22)+Za11,a12,a22

    for all aijAij,1ij2.

    Proof of Theorem 3.1: For arbitrary x,yT, elements x and y have decomposition form x=a11+a12+a22 and y=b11+b12+b22, where aij,bijAij,1ij2. It follows from Lemmas 3.2–3.6 that there exists ClZ(T), l{1,2,3,4,5}.

    Ln(x+y)=Ln(a11+a12+a22+b11+b12+b22)=Ln(a11+b11)+Ln(a12+b12)+Ln(a22+b22)+C1=Ln(a11)+Ln(b11)+Ln(a12)+Ln(b12)+Ln(a22)+Ln(b22)+C1+C2+C3=Ln(a11+a12+a22)+Ln(b11+b12+b22)+C1+C2+C3+C4+C5.

    Therefore, we have

    Ln(x+y)=Ln(x)+Ln(y)+C0

    for some C0=C1+C2+C3+C4+C5Z(T).

    Based on the almost additivity of Ln, we present the main theorem of this part to the readers as follows.

    Theorem 3.2. Let T=[A11A12OA22] be a triangular algebra satisfying

    i) πA11(Z(T))=Z(A11) and πA22(Z(T))=Z(A22);

    ii) For any a11A11, if [a11,A11]Z(A11), then a11Z(A11) or for any a22A22, if [a22,A22]Z(A22), then a22Z(A22).

    Suppose that a mapping Ln:TT(n3) is a nonlinear mapping satisfying

    Ln(Pn(x1,x2,,xn))=ni=1Pn(x1,x2,,Ln(xi)ith component,,xn)

    for all x1,x2,,xnT with ni=1xi=0, then for every nN,

    Ln(x)=dn(x)+fn(x)

    for all xT, where dn:TT is an additive derivation, and fn:TZ(T) is a nonlinear mapping such that fn(Pn(x1,x2,,xn))=0 for any x1,x2,,xnT with x1x2xn=0.

    In order to facilitate readers' understanding, we will divide the proof process into the following lemmas for explanation.

    Lemma 3.7. With notations as above, we have

    1) Ln(a12)A12;

    2) p1Ln(p1)p1p2Ln(p1)p2Z(T) and p1Ln(p2)p1p2Ln(p2)p2Z(T);

    3) Ln(p1)A12+Z(T) and Ln(p2)A12+Z(T).

    Proof. Since the equation a12p1p1p2p2=0 holds, we have

    Ln(a12)=Ln(Pn(a12,p1,p1,p2,,p2))=Pn(Ln(a12),p1,p1,p2,,p2)+Pn(a12,Ln(p1),p1,p2,,p2)+Pn(a12,p1,Ln(p1),p2,,p2)+ni=4Pn(a12,p1,p1,p2,,Ln(p2)ith component,,p2)=p1Ln(a12)p2+2p1Ln(p1)a122a12Ln(p1)p2+(n3)p1[a12,Ln(p2)]p2. (3.4)

    According to above relation (3.4), we have Ln(a12)A12. Multiplying by p1 on the left side and p2 on the right side of the above Eq (3.4), we can obtain

    2p1[a12,Ln(p1)]+(n3)p1[a12,Ln(p2)]p2=0 (3.5)

    for all a12A12.

    On the other hand, with the help of p2a12p1p2p2n3 copies=0, we have

    Ln(a12)=Ln(Pn(p2,a12,p1,p2,,p2))=Pn(Ln(p2),a12,p1,p2,,p2)+Pn(p2,Ln(a12),p1,p2,,p2)+Pn(p2,a12,Ln(p1),p2,,p2)+ni=4Pn(p2,a12,p1,p2,,Ln(p2)ith component,,p2)=p1[Ln(p2),a12]p2+p1Ln(a12)p2p1[a12,Ln(p1)]p2+(n3)p1[a12,Ln(p2)]p2. (3.6)

    According to (3.6), we have

    (n4)p1[a12,Ln(p2)]p2p1[a12,Ln(p1)]p2=0. (3.7)

    Combining (3.5) with (3.6), we have

    p1[a12,Ln(p2)]p2=p1[a12,Ln(p1)]p2. (3.8)

    Using equalities (3.7) with (3.8) and considering the (n1)-torsion-free properties of rings R, we conclude that

    p1[a12,Ln(p2)]p2=p1[a12,Ln(p1)]p2=0

    for all a12A12.

    With the help of Lemma 3.1 that

    p1Ln(p1)p1p2Ln(p1)p2Z(T) and p1Ln(p2)p1p2Ln(p2)p2Z(T),

    the structural characteristics of triangular algebra lead to conclusions

    Ln(p1)=p1Ln(p1)p1+p1Ln(p1)p2+p2Ln(p1)p2A12+Z(T).

    Furthermore, we claim that this lemma holds.

    Lemma 3.8. With notations as above, for all aiiAii,i{1,2}, we have

    i) Ln(a11)A11+A12+Z(T), where p2Ln(a11)p2Z(A22);

    ii) Ln(a22)A22+A12+Z(T), where p1Ln(a22)p1Z(A11);

    iii) p1Ln(p1)p2+p1Ln(p2)p2=0.

    Proof. In fact, it is clear that if a11p2p2=0, we have

    0=Ln(Pn(a11,p2,,p2))=Pn(Ln(a11),p2,,p2)+Pn(a11,Ln(p2),p2,,p2)=p1Ln(a11)p2+a11Ln(p2)p2,

    which implies that

    p1Ln(a11)p2+a11Ln(p2)p2=0 (3.9)

    for all a11A11. Through similar calculations, combining with relation that if a22p1p1p2,p2=0, we can show that

    0=Ln(Pn(a22,p1,p1,p2,,p2))=Pn(Ln(a22),p1,p1,p2,,p2)+Pn(a22,Ln(p1),p1,p2,,p2)=p1Ln(a22)p2+p1Ln(p2)a22,

    which implies

    p1Ln(a22)p2+p1Ln(p2)a22=0 (3.10)

    for all a22A22.

    Now, we consider the following equation

    0=Ln(Pn(p1,p2,,p2))=Pn(Ln(p1),p2,,p2)+Pn(p1,Ln(p2),,p2)=p1Ln(p1)p2+p1Ln(p2)p2.

    That is,

    p1Ln(p1)p2+p1Ln(p2)p2=0. (3.11)

    In the light of equations a22a11a12p2p2=0, we have

    0=Ln(Pn(a22,a11,a12,p2,,p2))=Pn(Ln(a22),a11,a12,p2,,p2)+Pn(a22,Ln(a11),a12,p2,,p2)=p1[[Ln(a22),a11],a12]p2+p1[[a22,Ln(a11)],a12]p2=p1[[Ln(a22),a11]p2+p1[a22,Ln(a11)],a12]p2.

    According to Lemma 3.1, we have

    [Ln(a22),a11]+[a22,Ln(a11)]Z(T).

    Furthermore, we have

    [p1Ln(a22)p1,a11]Z(A11)  and  [a22,p2Ln(a11)p2]Z(A22)

    for all a11A11,a22A22. Thanks to the assumption (ii), we have

    p1Ln(a22)p1Z(A11) and p2Ln(a11)p2Z(A22)

    for all a11A11,a22A22. Therefore, this lemma holds.

    Now, we define two mappings rn1:A11Z(A11) and rn2:A22Z(A22), as following

    rn1(a11)=τ1(p2Ln(a11)p2)+p2Ln(a11)p2

    and

    rn2(a22)=τ(p1Ln(a22)p1)+p1Ln(a22)p1

    for all a11A11,a22A22, respectively. It follows from Lemma 3.11 that rn1:A11Z(A11) satisfies the relation rn1(Pn(a111,a211,,an11))=0 for all a111,a211,,an11A11, with a111a211an11=0 and rn2:A22Z(A22) satisfying the relation rn2(Pn(a122,a222,,an22))=0 for all a122,a222,,an22A22 with a122a222an22=0. Now, setting

    hn(x)=rn1+rn2=τ1(p2Ln(a11)p2)+p2Ln(a11)p2+τ(p1Ln(a22)p1)+p1Ln(a22)p1

    for all xT, which satisfies the form x=a11+a12+a22, it is clear that hn(x)Z(T) and hn(Pn(x1,x2,,xn))=0 for all x1,x2,,xnT with x1x2xn=0.

    Let's define an important mapping:

    Ψn(x)=Ln(x)hn(x)

    for all xT.

    Now, we can easily obtain the following lemmas.

    Lemma 3.9. With notations as above, for all aijAij,1ij2, we have

    i) Ψn(0)=0;

    ii) Ψn(A11)A11+A12, Ψn(A22)A22+A12;

    iii) Ψn(A12)=Ln(A12); Ψn(pi)A12.

    Lemma 3.10. With notations as above, for all aijAij,1ij2, we have

    i) Ψn(a11a12)=Ψn(a11)a12+a11Ψn(b12);

    ii) Ψn(a12a22)=Ψn(a12)a22+a12Ψn(a22).

    Proof. We now only prove the conclusion (i) and the proof of conclusion (ii) can be obtained by similar methods.

    (i) It follows from a12a11p1p2p2=0 that

    Ψn(a11a12)=Φn(Pn(a12,a11,p1,p2,,p2))=Pn(Φn(a12),a11,p1,p2,,p2)+Pn(a12,Φn(a11),p1,p2,,p2)=Pn(Ψn(a12),a11,p1,p2,,p2)+Pn(a12,Ψn(a11),p1,p2,,p2)=a11Ψn(a12)+a11Ψn(a12).

    By a similar method, the conclusion (ii) holds.

    Lemma 3.11. With notations as above, for all aii,biiAii,i{1,2}, we have Ψn(aiibii)=Ψn(aii)bii+aiiΨn(bii).

    Proof. By Lemma 3.9, for all c12A12, we have

    Ψn(a11b11c12)=Ψn(a11)b11c12+a11Ψn(b11c12)=Ψn(a11)b11c12+a11Ψn(b11)c12+a11b11Ψn(c12)

    and

    Ψn(a11b11c12)=Ψn(a11b11)c12+a11b11Ψn(c12).

    Combining the two equations above, we have

    (Ψn(a11b11)Ψn(a11)b11a11Ψn(b11))c12=0

    for all a11,b11A11,c12A12. Since A12 is a faithful (A11,A22)-bimodule, we have

    Ψn(a11b11)=Ψn(a11)b11+a11Ψn(b11)

    for all a11,b11A11.

    After a similar calculation process, we can get

    Ψn(a22b22)=Ψn(a22)b22+a22Ψn(b22)

    for all a22,b22A22.

    Lemma 3.12. With notations as above, for all aii,biiAii, we have

    i) Ψn(a12+b12)=Ψn(a12)+Ψn(a12);

    ii) Ψn(a11+a12)Ψn(a11)Ψn(a12)Z(T) and Ψn(a22+a12)Ψn(a22)Ψn(a12)Z(T);

    iii) Ψn(aii+bii)=Ψn(aii)+Ψn(bii).

    Proof. According to the above symbols, we can prove the lemma by using the relationship between mapping Ψn and mapping Ln. The details are as follows:

    i). It is the direct result of Lemmas 3.4 and 3.9;

    ii). It is the direct result of Lemma 3.3;

    iii). According to the above Lemma 3.5, we have

    Ψn(aii+bii)=Ψn(aii)+Ψn(bii)+Z(aii,bii)

    for some Z(aii,bii)Z(T). Let us consider the center element Z(aii,bii).

    On the one hand,

    Ψn((a11+b11)c12)=Ψn(a11c12)+Ψn(b11c12)=Ψn(a11)c12+a11Ψn(c12)+Ψn(b11)c12+b11Ψn(c12)=(Ψn(a11)+Ψn(b11))c12+(a11+b11)Ψn(c12),

    and on the other hand, we have

    Ψn((a11+b11)c12)=Ψn(a11+b11)c12+(a11+b11)Ψn(c12)

    for all a11,b11A11,c12A12.

    By observing the two equations above, we can obtain Z(a11,b11)c12=p1Z(a11,b11)p1c12=0 for all c12A12. Since A12 is a faithful (A11,A22)-bimodule, then p1Z(a11,b11)p1=0. Note that there is an algebra isomorphism τ:Z(A11)Z(A22), and we can obtain p2Z(a11,b11)p2=τ(p1Z(a11,b11)p1)=0 and Z(a11,b11)c12=p1Z(a11,b11)p1+p2Z(a11,b11)p2=0. Therefore,

    Ψn(a11+b11)=Ψn(a11)+Ψn(b11)

    for all a11,b11A11.

    After a similar calculation process, we can get

    Ψn(a22+b22)=Ψn(a22)+Ψn(b22)

    for all a22,b22A22.

    According to the definition of Ψn and Theorem 3.1, we can get the following lemma immediately.

    Lemma 3.13. Let aijAij,1ij2, then Ψn(a11+a12+a22)=Ψn(a11)+Ψn(a12)+Ψn(a22)+Ca11,a12,a22.

    Remark 3.1. Lemma 3.13 enables us to establish a mapping gn:TT

    gn(x)=Ψn(x)Ψn(p1xp1)Ψn(p1xp2)Ψn(p2xp2)

    for all xT, then define a mapping dn:TT by

    dn(x)=Ψn(x)gn(x)

    for all xT. It is easy to verify that each dn satisfies the following property:

    1) dn(aij)=Ψn(aij);

    2) dn(a11+a12+a22)=dn(a11)+dn(a12)+dn(a22)

    for all aijAij,ij{1,2}.

    Now, we are in a position to prove our main theorem.

    Proof of Theorem 3.1: It follows from the definitions of dn and gn that

    Ln(x)=Φn(x)+hn(x)=dn(x)+gn(x)+hn(x)=dn(x)+fn(x),

    where fn(x)=gn(x)+hn(x) is a mapping from T into its center Z(T) for all xT.

    For arbitrary x,yT, elements x and y have decomposition form x=a11+a12+a22 and y=b11+b12+b22, where aij,bijAij,1ij2. It follows from Remark 3.1 that

    dn(x+y)=dn(a11+a12+a22+b11+b12+b22)=dn(a11+b11)+dn(a12+b12)+dn(a22+b22)=dn(a11)+dn(b11)+dn(a12)+dn(b12)+dn(a22)+dn(b22)=dn(a11+a12+a22)+dn(b11+b12+b22).

    Therefore, we have

    dn(x+y)=dn(x)+dn(y).

    Next, we prove that dn satisfies the Leibniz formula.

    First, we should prove that

    p1dn(a11)a22+a11dn(a22)p2=0

    for all a11A11,a22A22.

    In fact, because of a11a22p2p2=0 and Eqs (3.9)–(3.10), we have

    p1dn(a11)a22+a11dn(a22)p2=p1(Ψn(a11)+gn(a11))a22+a11(Ψn(a22)+gn(a22))p2=p1Ψn(a11)a22+a11Ψn(a22)p2=p1Ln(a11)a22+a11Ln(a22)p2=p1a11Ln(p2)a22+a11Ln(p1)a22=p1a11(p1Ln(p2)p2+p1Ln(p1)p2)a22=0

    for all a11A11,a22A22.

    In view of Lemmas 3.8–3.11, we have

    dn(xy)=dn(a11b11+a11b12+a12b22+a22b22)=Ψn(a11b11)+Ψn(a11b12)+Ψn(a12b22)+Ψn(a22b22)=Ψn(a11)b11+a11Ψn(b11)+Ψn(a11)b12+a11Ψn(b12)+Ψn(a12)b22+a12Ψn(b22)+Ψn(a22)b22+a22Ψn(b22)+Ψn(a11b22)+a11Ψn(b22)=dn(a11)b11+a11dn(b11)+dn(a11)b12+a11dn(b12)+dn(a12)b22+a12dn(b22)+dn(a22)b22+a22dn(b22)+dn(a11)b22+a11dn(b22)=(dn(a11)+dn(a12)+dn(a22))(b11+b12+b22)+(a11+a12+a22)(dn(b11)+dn(b12)+dn(b22))=dn(a11+a12+a22)(b11+b12+b22)+(a11+a12+a22)dn(b11+b12+b22)=dn(x)y+xdn(y)

    for all x,yT.

    Now, we consider the properties of the mapping fn:TZ(T). For arbitrary x1,,xnT with x1xn=0, note that dn is an additive derivation of T, and we have

    fn(Pn(x1,,xn))=(Lndn)(Pn(x1,,xn))=Ln(Pn(x1,,xn))dn(Pn(x1,,xn))=nk=1Pn(x1,,(Lndn)(xk),,xn)=nk=1Pn(x1,,fn(xk),,xn)=0.

    Therefore, Theorem 3.1 holds.

    This section focuses on researching higher Lie n-derivation at zero product on triangular algebras. We will provide the more advanced version (Theorems 4.1 and 4.2) that corresponds to Theorems 3.1 and 3.2. These conclusions generalize a few previous conclusions under the same presumptions, such as [16,Theorems 2.1 and 2.2] and [17,Theorem 1 and 3].

    Theorem 4.1. Let T=[A11A12OA22] be a triangular algebra over (n1)-torsion-free commutative ring R. Suppose that a sequence Δ={δm}mN of mappings δm:TT is a nonlinear mapping

    δm(Pn(x1,x2,,xn))=i1++in=mPn(δi1(x1),δi2(x2),,δin(xn)) (4.1)

    for all x1,x2,,xnT with x1x2xn=0 and n3. If πA11(Z(T))=Z(A11) and πA22(Z(T))=Z(A22), then every nonlinear mapping δm is almost additive on T, which is

    δm(x+y)δm(x)δm(y)Z(T)

    for all x,yT.

    It is worth noting that the mapping δ1 in Theorem 4.1 is equal to the mapping Ln in Theorem 3.1. Based on this, we begin to prove theorem 4.1.

    Assume that a sequence Δ={δm}mN of nonlinear mappings δm:TT is a Lie-n higher derivation by local actions on triangular algebras T=[A11A12OA22]. We will use an induction method for the component index m. For m=1, δ1=Ln:TT is a Lie n-derivation by local actions. According to Theorems 3.2 and 3.1, we obtain that a nonlinear Lie n-derivation δ1 by local actions meet the following attributes:

    C1={δ1(0)=0,δ1(a11+a12)δ1(a11)δ1(a12)Z(T),δ1(a22+a12)δ1(a22)δ1(a12)Z(T);δ1(a12+a12)=δ1(a12)+δ1(a12);δ1(aii+aii)δ1(aii)δ1(aii)Z(T);δ1(a11+a12+a22)δ1(a11)δ1(a12)δ1(a22)Z(T)

    for all aijA with ij{1,2}.

    We assume that the result holds for all 1<s<m, mN, then nonlinear Lie n-derivation {δl}l=sl=0 satisfies the following

    Cs={δs(0)=0,δs(a11+a12)δs(a11)δs(a12)Z(T),δs(a22+a12)δs(a22)δs(a12)Z(T);δs(a12+a12)=δs(a12)+δs(a12);δs(aii+aii)δs(aii)δs(aii)Z(T);δs(a11+a12+a22)δs(a11)δs(a12)δs(a22)Z(T)

    for all aijA with ij{1,2}.

    Our goal is to prove that the above conditions Cs also hold for m. The process of induction can be achieved through a series of lemmas.

    Lemma 4.1. With notations as above, we have δm(0)=0.

    Proof. With the help of condition Cs, we find that

    δm(0)=δm(Pn(0,0,,0))=i1++in=mPn(δi1(0),δi2(0),,δin(0))=0.

    Lemma 4.2. With notations as above, we have

    i) δm(a11+a12)δm(a11)δm(a12)Z(T);

    ii) δm(a22+a12)δm(a22)δm(a12)Z(T)

    for all aijA with ij{1,2}.

    Proof. Here, we only prove that the conclusion i) holds and that the proof of conclusion ii) can be similarly obtained.

    It is clear that x12(a11+x12)p1p2p2=x12a11p1p2p2=0=x12x12p1p2p2 for all a11A11 and x12,x12A12. On the one hand, we have

    δm(a11x12)=δm(Pn(x12,(a11+x12),p1,p2,,p2))=i1++in=mPn(δi1(x12),δi2(a11+x12),δi3(p1),δi4(p2),δin(p2)),

    and on the other hand, we have

    δn(a11x12)=δm(Pn(x12,a11,p1,p2,,p2))+δm(Pn(x12,x12,p1,p2,,p2))=i1++in=mPn(δi1(x12),δi2(a11),δi3(p1),δi4(p2),δin(p2))+i1++in=mPn(δi1(x12),δi2(x12),δi3(p1),δi4(p2),δin(p2))=i1++in=mPn(δi1(x12),δi2(a11)+δi2(x12),δi3(p1),δi4(p2),δin(p2)).

    By observing the two equations above and inductive hypothesis Cs for all 0sm1, we arrive at

    0=i1++in=mPn(δi1(x12),δi2(a11+x12)(δi2(a11)+δi2(x12)),δi3(p1),δi4(p2),δin(p2))=i1++in=m,i2mPn(δi1(x12),δi2(a11+x12)(δi2(a11)+δi2(x12)),δi3(p1),δi4(p2),δin(p2))+Pn(x12,δm(a11+x12)(δm(a11)+δm(x12)),p1,p2,,p2)=Pn(x12,δm(a11+x12)(δm(a11)+δm(x12)),p1,p2,,p2)=p1(δm(a11+x12)(δm(a11)+δm(x12)))x12x12(δm(a11+x12)(δm(a11)+δm(x12)))p2

    for all a11A11 and x12,x12A12. It then follows from the center of algebra T that

    p1(δm(a11+x12)(δm(a11)+δm(x12)))p1+p2(δm(a11+x12)(δm(a11)+δm(x12)))p2Z(T)

    for all a11A11 and x12A12.

    In the following, we prove p1(δm(a11+x12)(δm(a11)+δm(x12)))p2=0 for all a11A11 and x12A12.

    With the help of p2(a11+x12)p1p2p2n3 copies=0=p2a11p1p2p2n3 copies, we have

    δm(x12)=δm(Pn(p2,x12+a11,p1,p2,,p2n3 copies))=i1++in=mPn(δi1(p2),δi2(x12+a11),δi3(p1),δi4(p2),δin(p2))

    for all a11A11 and m12A12. On the other hand, we have

    δm(x12)=δm(Pn(p2,x12,p1,p2,,p2n3 copies))+δm(Pn(p2,a11,p1,p2,,p2n3 copies))=i1++in=mPn(δi1(p2),δi2(x12)+δi2(a11),δi3(p1),δi4(p2),δin(p2))

    for all a11A11 and x12A12. With the help of the above two equations and inductive hypothesis Cs, we have

    0=i1++in=mPn(δi1(p2),δi2(x12+a11)(δi2(x12)+δi2(a11)),δi3(p1),δi4(p2),δin(p2))=i1++in=m,i2mPn(δi1(p2),δi2(x12+a11)(δi2(x12)+δi2(a11)),δi3(p1),δi4(p2),δin(p2))+Pn(p2,δm(x12+a11)(δm(x12)+δm(a11)),p1,p2,,p2)=Pn(p2,δm(x12+a11)(δm(x12)+δm(a11)),p1,p2,,p2)=p1(δm(a11+x12)(δm(a11)+δm(x12)))p2

    for all a11A11 and x12A12. Therefore, we obtain that the conclusion (i) holds.

    For conclusion (ii), taking into accounts the relations x12(b22+x12)p1p2p2n3 copies=x12x12p1p2p2n3 copies=x12b22p1p2p2n3 copies=0, by an analogous manner one can show that the conclusion

    δm(b22+x12)(δm(b22)+δm(x12))Z(T)

    holds for all b22A22 and x12,x12A12.

    Lemma 4.3. With notations as above, we have δn(x12+x12)=δn(x12)+δn(x12) for all x12,x12M12.

    Proof. Thanks to inductive hypothesis Cs for all 1s<m and relations (x12p1)(p2+x12)p1p2p2n3 copies=0, we have

    δm(x12+x12)=δm(Pn((x12p1),(p2+x12),p1,p2,,p2n3 copies))=i1++in=mPn(δi1(x12p1),δi2(p2+x12),δi3(p1),δi4(p2),,δin(p2)n3 copies)=i1++in=mPn(δi1(x12)+δi1(p1),δi2(p2)+δi2(x12),δi3(p1),δi4(p2),,δin(p2)n3 copies)=i1++in=mPn(δi1(x12),δi2(p2),δi3(p1),δi4(p2),,δin(p2)n3 copies)+i1++in=mPn(δi1(x12),δi2(x12),δi3(p1),δi4(p2),,δin(p2)n3 copies)+i1++in=mPn(δi1(p1),δi2(p2),δi3(p1),δi4(p2),,δin(p2)n3 copies)+i1++in=mPn(δi1(p1),δi2(x12),δi3(p1),δi4(p2),,δin(p2)n3 copies)=δm(Pn(x12,p2,p1,p2,,p2n3 copies))+δm(Pn(x12,x12,p1,p2,,p2n3 copies))+δm(Pn(p1,p2,p1,p2,,p2n3 copies))+δm(Pn(p1,x12,p1,p2,,p2n3 copies))=δm(x12)+δm(x12).

    That is, δm(x12+x12)=δm(x12)+δm(x12) for all x12,x12A12.

    Lemma 4.4. With notations as above, we have

    1)δm(a11+a11)δm(a11)δm(a11)Z(T);

    2)δm(a22+a22)δm(a22)δm(a22)Z(T)

    for all aii,aiiAii with i{1,2}.

    Proof. We only prove the statement 1). The statement 2) can be proved in a similar way. Because of relations x12p1(a11+a11)p2p2n3 copies=x12p1a11p2p2n3 copies=0=x12p1a11p2p2n3 copies, we arrive at

    δm((a11+a11)x12)=δm(Pn(x12,p1,(a11+a11),p2,,p2n3 copies))=i1++in=mPn(δi1(x12),δi2(p1),δi3(a11+a11),δi4(p2),,δin(p2)n3 copies).

    On the other hand, we have

    δm((a11+a11)x12)=δm(Pn(x12,p1,(a11),p2,,p2n3 copies))+δm(Pn(x12,p1,(a11),p2,,p2n3 copies))=i1++in=mPn(δi1(x12),δi2(p1),δi3(a11),δi4(p2),,δin(p2)n3 copies)+i1++in=mPn(δi1(x12),δi2(p1),δi3(a11),δi4(p2),,δin(p2)n3 copies)=i1++in=mPn(δi1(x12),δi2(p1),δi3(a11)+δi3(a11),δi4(p2),,δin(p2)n3 copies)

    for all a11A11,x12A12. On comparing the above two relations together with inductive hypothesis Cs for all 1s<m, we see that

    0=i1++in=mPn(δi1(x12),δi2(p1),δi3(a11+a11)(δi3(a11)+δi3(a11)),δi4(p2),,δin(p2)n3 copies)=i1++in=m,i3mPn(δi1(x12),δi2(p1),δi3(a11+a11)(δi3(a11)+δi3(a11)),δi4(p2),,δin(p2)n3 copies)+Pn(x12,p1,δm(a11+a11)(δm(a11)+δm(a11)),p2,,p2n3 copies)=Pn(x12,p1,δm(a11+a11)(δm(a11)+δm(a11)),p2,,p2n3 copies),

    which is

    p1(δm(a11+a11)(δm(a11)+δm(a11)))x12=x12(δm(a11+a11)(δm(a11)+δm(a11)))p2. (4.2)

    It follows from the center of triangular algebra T and the above equation that

    p1(δm(a11+a11)(δm(a11)+δm(a11)))p1p2(δm(a11+a11)(δm(a11)+δn(a11)))p2Z(T). (4.3)

    In the following, we prove

    p1(δm(a11+a11)(δm(a11)+δm(a11)))p2=0

    for all a11,a11A11.

    Benefitting from (a11+a11)p2p2n1 copies=a11p2p2n1 copies=a11p2p2n1 copies=0, we have

    0=δm(Pn((a11+a11),p2,,p2))=i1++in=mPn(δi1(a11+a11),δi2(p2),δi3(p2),,δin(p2)n2 copies)

    and

    0=δm(Pn(a11,p2,,p2))+δm(Pn(a11,p2,,p2))=i1++in=mPn(δi1(a11),δi2(p2),δi3(p2),,δin(p2)n2 copies)+i1++in=mPn(δi1(a11),δi2(p2),δi3(p2),,δin(p2)n2 copies)=i1++in=mPn(δi1(a11)+δi1(a11),δi2(p2),δi3(p2),,δin(p2)n2 copies).

    By combining the above two equations with inductive hypothesis Cs for all 1s<m, we can get

    0=i1++in=mPn(δi1(a11+a11)(δi1(a11)+δi1(a11)),δi2(p2),δi3(p2),,δin(p2)n2 copies)=i1++in=m,i1mPn(δi1(δi1(a11+a11)(δi1(a11)+δi1(a11)),δi2(p2),δi3(p2),,δin(p2)n2 copies)+Pn(δm(a11+a11)(δm(a11)+δm(a11)),p2,p2,,p2n2 copies)=Pn(δm(a11+a11)(δm(a11)+δm(a11)),p2,p2,,p2n2 copies),

    which is

    \begin{align} p_1(\delta_m(a_{11}+a_{11}^\prime)-(\delta_m(a_{11})+\delta_m(a_{11}^\prime)))p_2 = 0. \end{align} (4.4)

    Combining (4.5) and (4.6) , this claim holds.

    Lemma 4.5. With notations as above, we have \delta_m(a_{11}+x_{12}+b_{22})-\delta_m(a_{11})-\delta_m(x_{12})-\delta_m(b_{22})\in \mathcal{Z}(\mathcal{T}) for all a_{11}\in A_{11}, x_{12}\in A_{12}, a_{22}\in A_{22} .

    Proof. For arbitrary a_{11}\in A_{11}, x_{12}\in A_{12}, a_{22}\in A_{22} , in view of (a_{11}+x_{12}+b_{22})x_{12}^\prime p_1p_2\cdots p_2 = 0 , we have

    \begin{aligned} \delta_m(x_{12^\prime}b_{22}-a_{11}x_{12}^\prime)& = \delta_m(P_n(a_{11}+x_{12}+b_{22}, x_{12}^\prime, p_1, p_2, \cdots , p_2))\\ & = \sum\limits_{i_1+\cdots +i_n = m}P_n(\delta_{i_1}(a_{11}+x_{12}+b_{22}), \delta_{i_2}(x_{12}^\prime), \delta_{i_3}(p_1), \underbrace{\delta_{i_4}(p_2), \cdots , \delta_{i_n}(p_2)}_{n-3 \ \text{copies}}) \end{aligned}

    and

    \begin{aligned} &\delta_m(x_{12}^\prime b_{22}-a_{11}x_{12}^\prime) = \delta_m(x_{12}^\prime b_{22})+\delta_m(-a_{11}x_{12}^\prime)\\ & = \delta_m(P_n(a_{11}, x_{12}^\prime, p_1, p_2, \cdots , p_2))+\delta_m(P_n(x_{12}, x_{12}^\prime, p_1, p_2, \cdots , p_2))\\ &+\delta_m(P_n(b_{22}, x_{12}^\prime, p_1, p_2, \cdots , p_2))\\ & = \sum\limits_{i_1+\cdots +i_n = m}P_n(\delta_{i_1}(a_{11})+\delta_{i_1}(x_{12})+\delta_{i_1}(b_{22}), \delta_{i_2}(x_{12}^\prime), \delta_{i_3}(p_1), \underbrace{\delta_{i_4}(p_2), \cdots , \delta_{i_n}(p_2)}_{n-3 \ \text{copies}}). \end{aligned}

    Let us set W_i = \delta_i(a_{11}+x_{12}+b_{22})-(\delta_i(a_{11})+\delta_i(x_{12})+\delta_i(b_{22})) . Taking into accounts the above two equations and using inductive hypothesis \mathfrak{C}_s for all 1\leq s\leq n , we have

    \begin{aligned} 0& = \sum\limits_{i_1+\cdots +i_n = m}P_n(W_{i_1}, \delta_{i_2}(x_{12}^\prime), \delta_{i_3}(p_1), \underbrace{\delta_{i_4}(p_2), \cdots , \delta_{i_n}(p_2)}_{n-3 \ \text{copies}})\\ & = \sum\limits_{i_1+\cdots +i_n = m, i_1\neq m}P_n(W_{i_1}, \delta_{i_2}(x_{12}^\prime), \delta_{i_3}(p_1), \underbrace{\delta_{i_4}(p_2), \cdots , \delta_{i_n}(p_2)}_{n-3 \ \text{copies}})\\ &+P_n(W_m, x_{12}^\prime, p_1, \underbrace{p_2, \cdots , p_2}_{n-3 \ \text{copies}})\\ & = P_n(W_m, x_{12}^\prime, p_1, \underbrace{p_2, \cdots , p_2}_{n-3 \ \text{copies}}), \end{aligned}

    which is p_1W_mx_{12}^\prime-x_{12}^\prime W_mp_2 = 0 , i.e.,

    \begin{align} p_1W_mp_1\oplus p_2W_mp_2\in \mathcal{Z}(\mathcal{T}). \end{align} (4.5)

    In the following part, we prove p_1W_mp_2 = 0 . It is clear that (a_{11}+x_{12}+b_{22})(-p_1) \underbrace{p_2\cdots p_2}_{n-2 \ \text{copies}} = 0 , then

    \begin{aligned} \delta_m(x_{12})& = \delta_m(P_n(a_{11}+x_{12}+b_{22}, -p_1, \underbrace{p_2, \cdots, p_2}_{n-2\ \text{copies}})\\ & = \sum\limits_{i_1+\cdots +i_n = m}P_n(\delta_{i_1}(a_{11}+x_{12}+b_{22}), \delta_{i_2}(-p_1), \underbrace{\delta_{i_3}(p_2), \cdots , \delta_{i_n}(p_2)}_{n-2 \ \text{copies}})\\ \end{aligned}

    and

    \begin{aligned} \delta_n(x_{12})& = \delta_m(P_n(a_{11}, -p_1, \underbrace{p_2, \cdots, p_2}_{n-2\ \text{copies}})+\delta_m(P_n(x_{12}, -p_1, \underbrace{p_2, \cdots, p_2}_{n-2\ \text{copies}})\\ &+\delta_m(P_n(b_{22}, -p_1, \underbrace{p_2, \cdots, p_2}_{n-2\ \text{copies}})\\ & = \sum\limits_{i_1+\cdots +i_n = m}P_n(\delta_{i_1}(a_{11})+\delta_{i_1}(x_{12})+\delta_{i_1}(b_{22}), \delta_{i_2}(-p_1), \underbrace{\delta_{i_3}(p_2), \cdots , \delta_{i_n}(p_2)}_{n-2 \ \text{copies}}). \end{aligned}

    According to the above two equations and inductive hypothesis \mathfrak{C}_s for all 1\leq s\leq m , we can get

    \begin{aligned} 0& = \sum\limits_{i_1+\cdots +i_n = m}P_n(W_{i_1}, \delta_{i_2}(-p_1), \underbrace{\delta_{i_3}(p_2), \cdots , \delta_{i_n}(p_2)}_{n-2 \ \text{copies}})\\ & = \sum\limits_{i_1+\cdots +i_n = m, i_1\neq m}P_n(W_{i_1}, \delta_{i_2}(-p_1), \underbrace{\delta_{i_3}(p_2), \cdots , \delta_{i_n}(p_2)}_{n-2 \ \text{copies}})+P_n(W_m, -p_1, \underbrace{p_2, \cdots , p_2}_{n-2 \ \text{copies}})\\ & = P_n(W_m, -p_1, \underbrace{p_2, \cdots , p_2}_{n-2 \ \text{copies}}), \end{aligned}

    which is

    \begin{align} p_1W_mp_2 = 0. \end{align} (4.6)

    It follows from relations (4.7)–(4.8) that this lemma holds.

    Next, we give the proof of this theorem. For arbitrary x = a_{11}+x_{12}+b_{22} and y = a_{11}^\prime+x_{12}^\prime+b_{22}^\prime , we have

    \begin{aligned} \delta_m(x+y)& = \delta_m(a_{11}+a_{11}^\prime+x_{12}+x_{12}^\prime+b_{22}+b_{22}^\prime)\\ & = \delta_m(a_{11}+a_{11}^\prime)+\delta_m(x_{12}+x_{12}^\prime)+\delta_m(b_{22}+b_{22}^\prime)+Z_1\\ & = \delta_m(a_{11})+\delta_m(a_{11}^\prime)+\delta_m(x_{12})+\delta_m(x_{12}^\prime)+\delta_m(b_{22})+\delta_m(b_{22}^\prime)+Z_1+Z_2+Z_3\\ & = \delta_m(x)+\delta_m(y)+Z_1+Z_2+Z_3+Z_4+Z_5, \end{aligned}

    which implies that \delta_m(x+y)-\delta_m(x)-\delta_m(y)\in \mathcal{Z}(\mathcal{T}) .

    Based on the additive of \delta_m on \mathcal{T} , we give the main result in this section reading as follows.

    Theorem 4.2. Let \mathcal{T} = \left[\begin{array}{cc} A_{11} & A_{12}\\ O & A_{22} \end{array}\right] be a triangular algebra satisfying

    \text{i}) \pi_{A_{11}}(\mathcal{Z}(\mathcal{T})) = \mathcal{Z}(\mathcal{A}_{11}) and \pi_{A_{22}}(\mathcal{Z}(\mathcal{T})) = \mathcal{Z}(\mathcal{A}_{22}) ;

    \text{ii}) For any a_{11}\in A_{11} , if [a_{11}, A_{11}]\in \mathcal{Z}(\mathcal{A}_{11}) , then a_{11}\in \mathcal{Z}(\mathcal{A}_{11}) , or for any a_{22}\in A_{22} , if [a_{22}, A_{22}]\in \mathcal{Z}(\mathcal{A}_{22}) , then a_{22}\in \mathcal{Z}(\mathcal{A}_{22}) .

    Suppose that a sequence \Delta = \{\delta_m\}_{m\in \mathcal{N}} of mappings \delta_m:\mathcal{T}\rightarrow \mathcal{T} is a nonlinear map satisfying

    \delta_m(P_n(x_1, x_2, \cdots, x_n)) = \sum\limits_{i_1+\cdots +i_n = m}P_n(\delta_{i_1}(x_1), \delta_{i_2}(x_2), \cdots, \delta_{i_n}(x_n))

    for all x_1, x_2, \cdots, x_n \in \mathcal{T} with x_1x_2\cdots x_n = 0 and n\geq 3 . For every m\in \mathcal{N} ,

    \delta_m(x) = \chi_m(x)+f_m(x)

    for all x\in \mathcal{T} , where a sequence \Upsilon = \{\chi_m \}_{m\in \mathcal{N}} of additive mapping \chi_m:\mathcal{T}\rightarrow \mathcal{T} is a higher derivation, and f_m:\mathcal{T}\rightarrow \mathcal{Z}(\mathcal{T}) is a nonlinear mapping such that f_m(P_n(x_1, x_2, \cdots, x_n)) = 0 for any x_1, x_2, \cdots, x_n \in \mathcal{T} with x_1x_2\cdots x_n = 0 .

    In the process of proof, we will use mathematical induction for index m . For m = 1 , \delta_1 = L_n is a Lie-n derivation on \mathcal{T} by local action at zero. By Theorem 3.2, it follows from Theorem 3.2 that there exists an additive derivation d_1 and a nonlinear center mapping f_1 , satisfying f_1(P_n(x_1, x_2, \cdots, x_n)) = 0 for any x_1, x_2, \cdots, x_n \in \mathcal{T} with x_1x_2\cdots x_n = 0 and n\geq 3 , such that \delta_1(x) = d_1(x)+f_1(x) for all x\in \mathcal{T} . Furthermore, \delta_1 and d_1 satisfy the following properties

    \mathfrak{F}_1 = \begin{cases} \delta_1(0) = 0, \delta_1(A_{11})\subseteq A_{11}+A_{12}+\mathcal{Z}(\mathcal{T}), \delta_1(A_{22})\subseteq A_{22}+A_{12}+\mathcal{Z}(\mathcal{T});\\ \delta_1(A_{12})\subseteq A_{12}, \delta_1(p_i)\subseteq A_{12}+\mathcal{Z}(\mathcal{T});\\ d_1(A_{ii})\subseteq A_{ii}+A_{ij}, d_1(A_{12})\subseteq A_{12}; f_1(P_n(x_1, x_2, \cdots, x_n)) = 0 \end{cases}

    for i\leq j\in \{1, 2\} and for any x_1, \cdots, x_n\in \mathcal{T} with x_1\cdots x_n = 0 .

    We assume that the result holds for s for all 1 < s < m , m\in \mathcal{N} , then there exists an additive derivation d_s and a nonlinear center mapping f_s , satisfying f_s(P_n(x_1, x_2, \cdots, x_n)) = 0 for any x_1, x_2, \cdots, x_n \in \mathcal{T} with x_1x_2\cdots x_n = 0 , such that \delta_s(x) = d_s(x)+f_s(x) for all x\in \mathcal{T} . Moreover, \delta_s and d_s satisfy the following properties

    \mathfrak{F}_s = \begin{cases} \delta_s(0) = 0, \delta_1(A_{11})\subseteq A_{11}+A_{12}+\mathcal{Z}(\mathcal{T}), \delta_1(A_{22})\subseteq A_{22}+A_{12}+\mathcal{Z}(\mathcal{T});\\ \delta_s(A_{12})\subseteq A_{12}, \delta_s(p_i)\subseteq A_{12}+\mathcal{Z}(\mathcal{T});\\ d_s(A_{ii})\subseteq A_{ii}+A_{ij}, d_s(A_{12})\subseteq A_{12};f_s(P_n(x_1, x_2, \cdots, x_n)) = 0 \end{cases}

    for i\leq j\in \{1, 2\} and for any x_1, x_2, \cdots, x_n \in \mathcal{T} with x_1x_2\cdots x_n = 0 .

    The induction process can be realized through a series of lemmas.

    Lemma 4.6. With notations as above, we have

    1) \delta_m(A_{12})\subseteq A_{12} ;

    2) p_1\delta_m(p_1)p_1\oplus p_2\delta_m(p_1)p_2\in \mathcal{Z}(\mathcal{T}) and p_1\delta_m(p_2)p_1\oplus p_2\delta_m(p_2)p_2\in \mathcal{Z}(\mathcal{T}) ;

    3) \delta_m(p_1)\in M_{12}+\mathcal{Z}(\mathcal{T}) .

    Proof. Because of x_{12}p_1p_1\underbrace{p_2\cdots p_2}_{n-3 \ \text{copies}} = 0 for x_{12}\in A_{12} , with the help of condition \mathfrak{F}_s for all 1 < s < m , we have

    \begin{aligned} \delta_m(x_{12})& = \delta_m(P_n(x_{12}, p_1, p_1, p_2, \cdots , p_2))\\ & = \sum\limits_{i_1+\cdots+i_n = m}P_n(\delta_{i_1}(x_{12}), \delta_{i_2}(p_1), \delta_{i_3}(p_1), \delta_{i_4}(p_2), \cdots , \delta_{i_n}(p_2))\\ & = \sum\limits_{i_1+\cdots+i_n = m, 0 < i_1, \cdots, i_n < m}P_n(\delta_{i_1}(x_{12}), \delta_{i_2}(p_1), \delta_{i_3}(p_1), \delta_{i_4}(p_2), \cdots , \delta_{i_n}(p_2))\\ &+P_n(\delta_m(x_{12}), p_1, p_1, p_2, \cdots , p_2)+P_n(x_{12}, \delta_m(p_1), p_1, p_2, \cdots , p_2)\\ &+P_n(x_{12}, p_1, \delta_m(p_1), p_2, \cdots , p_2)+\sum\limits_{s = 4}^nP_n(x_{12}, p_1, p_1, p_2, \cdots , p_2, \underbrace{\delta_m(p_2)}_{s-th \ component}, \cdots , p_2)\\ & = P_n(\delta_m(x_{12}), p_1, p_1, p_2, \cdots , p_2)+P_n(x_{12}, \delta_m(p_1), p_1, p_2, \cdots , p_2)\\ &+P_n(x_{12}, p_1, \delta_m(p_1), p_2, \cdots , p_2)\\ & = p_1\delta_m(x_{12})p_2+p_1\delta_m(p_1)x_{12}+\delta_m(p_1)x_{12}-2x_{12}\delta_m(p_1), \end{aligned}

    then we can obtain that \delta_m(x_{12})\in A_{12} . Multiplying by p_1 on the left side and p_2 on the right side of the above equation, we can obtain that p_1\delta_m(p_1)x_{12} = x_{12}\delta_m(p_1)p_2 for all x_{12}\in A_{12} . It follows from definition of center that

    p_1\delta_m(p_1)p_1\oplus p_2\delta_m(p_1)p_2\in \mathcal{Z}(\mathcal{T}).

    Because of p_2x_{12}p_1p_2\cdots p_2 = 0 , we adopt the same discussion as relations

    \begin{aligned} \delta_m(x_{12})& = \delta_m(P_n(p_2, x_{12}, p_1, p_2\cdots, p_2))\\ & = \sum\limits_{i_1+\cdots+i_n = m}P_n(\delta_{i_1}(p_{2}), \delta_{i_2}(x_{12}), \delta_{i_3}(p_1), \delta_{i_4}(p_2), \cdots , \delta_{i_n}(p_2))\\ & = \sum\limits_{i_1+\cdots+i_n = m, 0 < i_1, \cdots, i_n < m}P_n(\delta_{i_1}(p_{2}), \delta_{i_2}(x_{12}), \delta_{i_3}(p_1), \delta_{i_4}(p_2), \cdots , \delta_{i_n}(p_2))\\ &+P_n(\delta_m(p_{2}), x_{12}, p_1, p_2, \cdots , p_2)+P_n(p_2, \delta_m(x_{12}), p_1, p_2, \cdots , p_2)\\ &+P_n(p_2, x_{12}, \delta_m(p_1), p_2, \cdots , p_2)+\sum\limits_{s = 4}^nP_n(p_2, x_{12}, p_1, p_2, \cdots , p_2, \underbrace{\delta_m(p_2)}_{s-th \ component}, \cdots , p_2)\\ & = P_n(\delta_m(p_2), x_{12}, p_1, p_2, \cdots , p_2)+P_n(p_2, \delta_m(x_{12}), p_1, p_2, \cdots , p_2)\\ &+P_n(p_2, x_{12}, \delta_m(p_1), p_2, \cdots , p_2)\\ & = -p_1\delta_m(p_2)x_{12}+x_{12}\delta_m(p_2)p_2+p_1\delta_m(x_{12})p_2-x_{12}\delta_m(p_1)p_2+\delta_m(p_1)x_{12}\\ & = -p_1\delta_m(p_2)x_{12}+x_{12}\delta_m(p_2)p_2+p_1\delta_m(x_{12})p_2 \end{aligned}

    Multiplying by p_1 on the left side and p_2 on the right side of the above equation, we can obtain that p_1\delta_m(p_2)x_{12} = x_{12}\delta_m(p_2)p_2 for all x_{12}\in A_{12} . It follows from definition of center and we can prove that p_1\delta_m(p_2)p_1\oplus p_2\delta_m(p_2)p_2\in \mathcal{Z}(\mathcal{T}) holds.

    Lemma 4.7. With notations as above, we have

    1) \delta_m(a_{11})\in A_{11}+A_{12}+\mathcal{Z}(\mathcal{T}) , where p_2\delta_m(a_{11})p_2\in \mathcal{Z}(A_{11}) ;

    2) \delta_m(a_{22})\in A_{22}+A_{12}+\mathcal{Z}(\mathcal{T}) , where p_1\delta_m(a_{22})p_1\in \mathcal{Z}(A_{22})

    for all a_{ii}\in A_{ii} with i\in \{1, 2\} .

    Proof. In fact, it is clear that a_{22}a_{11}a_{12}\underbrace{p_2\cdots p_2}_{n-3 \ \text{copies}} = 0 for all a_{ij}\in A_{ij} and for all i, j \in \{1, 2\} , then according to inductive hypothesis \mathfrak{F}_s , we have

    \begin{aligned} 0 = &\delta_m(P_n(a_{22}, a_{11}, a_{12}, p_2, \cdots , p_2))\\ = &\sum\limits_{i_1+\cdots +i_n = m}P_n(\delta_{i_1}(a_{22}), \delta_{i_2}(a_{11}), \delta_{i_3}(a_{12}), \delta_{i_4}(p_2), \cdots , \delta_{i_n}(p_2))\\ = &\sum\limits_{i_1+\cdots +i_n = m, i_1, \cdots , i_n < m}P_n(\delta_{i_1}(a_{22}), \delta_{i_2}(a_{11}), \delta_{i_3}(a_{12}), \delta_{i_4}(p_2), \cdots , \delta_{i_n}(p_2))\\ &+P_n(\delta_m(a_{22}), a_{11}, a_{12}, p_2, \cdots , p_2)+P_n(a_{22}, \delta_m(a_{11}), a_{12}, p_2, \cdots , p_2)\\ = &P_n(\delta_m(a_{22}), a_{11}, a_{12}, p_2, \cdots , p_2)+P_n(a_{22}, \delta_m(a_{11}), a_{12}, p_2, \cdots , p_2)\\ = &P_n(p_1\delta_m(a_{22})p_1, a_{11}, a_{12}, p_2, \cdots , p_2)+P_n(a_{22}, p_2\delta_m(a_{11})p_2, a_{12}, p_2, \cdots , p_2)\\ = &P_{n-1}([p_1\delta_m(a_{22})p_1, a_{11}]+[a_{22}, p_2\delta_m(a_{11})p_2], a_{12}, p_2, \cdots , p_2) \end{aligned}

    for all a_{ij}\in A_{ij} and for all i\leq j \in \{1, 2\} . In light of Lemma 3.1, we obtain

    \begin{align} [p_1\delta_m(a_{22})p_1, a_{11}]\oplus [a_{22}, p_2\delta_m(a_{11})p_2]\in \mathcal{Z}(\mathcal{T}) \end{align} (4.7)

    for all a_{ii}\in A_{ii} and or all i \in \{1, 2\} . With the help of characterization of algebraic center, we have

    [p_1\delta_m(a_{22})p_1, a_{11}]\in \mathcal{Z}(A_{11}) \ \text{and} \ [a_{22}, p_2\delta_m(a_{11})p_2]\in \mathcal{Z}(A_{22}),

    then

    \begin{align} p_1\delta_m(a_{22})p_1\in \mathcal{Z}(A_{11}) \ \text{and} \ p_2\delta_m(a_{11})p_2\in \mathcal{Z}(A_{22}) \end{align} (4.8)

    for all a_{ii}\in A_{ii} and for all i \in \{1, 2\} . Further based on theorem hypothesis (\text{ii}) and above Eqs (4.9) and (4.10) , we arrive at

    \begin{aligned} \delta_m(a_{11})& = p_1\delta_m(a_{11})p_1-\tau^{-1}(p_2\delta_m(a_{11})p_2)+p_1\delta_m(a_{11})p_2\\ &+\tau^{-1}(p_2\delta_m(a_{11})p_2)+p_2\delta_m(a_{11})p_2\in A_{11}+A_{12}+\mathcal{Z}(\mathcal{T}) \end{aligned}

    and

    \begin{aligned} \delta_m(a_{22})& = p_2\delta_m(a_{22})p_2-\tau(p_1\delta_m(a_{22})p_1)+p_1\delta_m(a_{22})p_2\\ &+p_1\delta_m(a_{22})p_1+\tau(p_1\delta_m(a_{22})p_1)\in A_{22}+A_{12}+\mathcal{Z}(\mathcal{T}) \end{aligned}

    for all a_{ij}\in A_{ij} with i\leq j\in \{1, 2\} . We can conclude that this claim can be established.

    Now, we define mapping f_{m1}(a_{11}) = \tau^{-1}(p_2\delta_m(a_{11})p_2)+p_2\delta_m(a_{11})p_2 and f_{m2}(a_{22}) = p_1\delta_m(a_{22})p_1+\tau(p_1\delta_m(a_{22})p_1) for all a_{11}\in A_{11} and a_{22}\in A_{22} . It follows from Lemma 4.7 that f_{m1}:A_{11}\rightarrow \mathcal{Z}(A_{11}) such that f_{n1}(P_n(a_{11}^1, \cdots, a_{11}^n)) = 0 for all a_{11}^1, \cdots, a_{11}^n\in A_{11} with a_{11}^1a_{11}^2\cdots a_{11}^n = 0 and f_{m2}:A_{22}\rightarrow \mathcal{Z}(A_{22}) , such that f_{m2}(P_n(a_{22}^1, \cdots, a_{22}^n)) = 0 for all a_{22}^1, \cdots, a_{22}^n\in A_{22} with a_{22}^1a_{22}^2\cdots a_{22}^n = 0 . Now, set

    \begin{align} \begin{aligned} f_m(x)& = f_{m1}(a_{11})+f_{m2}(a_{22}) = \tau^{-1}(p_2\delta_m(a_{11})p_2)+p_2\delta_m(a_{11})p_2\\ &+p_1\delta_m(a_{22})p_1+\tau(p_1\delta_m(a_{22})p_1)\\ \end{aligned} \end{align} (4.9)

    for all x = a_{11}+a_{12}+a_{22}\in \mathcal{T} . It is clear that f_m(x)\in \mathcal{Z}(\mathcal{T}) and f_m(P_n(x_1, x_2, \cdots, x_n)) = 0 with x_1x_2\cdots x_n = 0 for all x_1, x_2, \cdots, x_n\in \mathcal{T} . Define a new mapping

    \begin{align} \varpi_m(x) = \delta_m(x)-f_m(x) \end{align} (4.10)

    for all x\in \mathcal{T} .

    Taking into account Lemmas 4.6 and 4.7 together with (4.11) and (4.12) , we can easily get the following Lemma 4.8.

    Lemma 4.8. With notations as above, we have

    1) \varpi_m(0) = 0 , \varpi_m(a_{12}) = \delta_m(a_{12})\in A_{12} , \varpi_m(p_i)\in A_{12} ;

    2) \varpi_m(a_{11})\in A_{11}+A_{12} and \varpi_m(a_{22})\in A_{22}+A_{12} ,

    for all a_{ij}\in A_{ij} with i\leq j\in \{1, 2\} .

    Lemma 4.9. With notations as above, we have

    1) \varpi_m(a_{11}a_{12}) = a_{11}\varpi_m(a_{12})+\varpi_m(a_{11})a_{12}+\sum_{i+j = m, 0 < i, j < m}d_i(a_{11})d_{j}(a_{12}) ;

    2) \varpi_m(a_{12}a_{22}) = a_{12}\varpi_m(a_{22})+\varpi_m(a_{12})a_{22}+\sum_{i+j = n, 0 < i, j < m}d_i(a_{12})d_{j}(a_{22})

    for all a_{ij}\in A_{ij} with i\leq j\in \{1, 2\} .

    Proof. Now, we only prove the conclusion 1) , and conclusion 2) can be proved by similar methods. It follows from a_{12}a_{11}p_1\underbrace{p_2\cdots p_2}_{n-3 \ \text{copies}} = 0 and the induction hypothesis \mathfrak{F}_s for all 1\leq s\leq m-1 that

    \begin{aligned} \varpi_m(a_{11}a_{12})& = \delta_m(a_{11}a_{12}) = \delta_m(P_n(a_{12}, a_{11}, p_1, p_2, \cdots, p_2))\\ & = \sum\limits_{i_1+\cdots +i_n = m}P_n(\delta_{i_1}(a_{12}), \delta_{i_2}(a_{11}), \delta_{i_3}(p_1), \delta_{i_4}(p_2), \cdots, \delta_{i_n}(p_2))\\ & = \sum\limits_{i_1+\cdots +i_n = m, i_1, \cdots , i_n < m}P_n(\delta_{i_1}(a_{12}), \delta_{i_2}(a_{11}), \delta_{i_3}(p_1), \delta_{i_4}(p_2), \cdots, \delta_{i_n}(p_2))\\ &+P_n(\delta_m(a_{12}), a_{11}, p_1, p_2, \cdots, p_2)+P_n(a_{12}, \delta_m(a_{11}), p_1, p_2, \cdots, p_2)\\ &+P_n(a_{12}, a_{11}, \delta_m(p_1), p_2, \cdots, p_2)\\ & = \sum\limits_{i_1+\cdots +i_n = m, i_1, \cdots , i_n < m}P_n(d_{i_1}(a_{12}), d_{i_2}(a_{11}), d_{i_3}(p_1), d_{i_4}(p_2), \cdots, d_{i_n}(p_2))\\ &+a_{11}\varpi_n(a_{12})+\varpi_n(a_{11})a_{12}\\ & = \sum\limits_{i_1+i_2 = m, 0 < i_1, i_2 < m}P_n(d_{i_1}(a_{12}), d_{i_2}(a_{11}), p_1, p_2, \cdots, p_2) +a_{11}\varpi_n(a_{12})+\varpi_n(a_{11})a_{12}\\ & = a_{11}\varpi_n(a_{12})+\varpi_n(a_{11})a_{12}+\sum\limits_{i_1+i_2 = m, 0 < i_1, i_2 < m}d_{i_2}(a_{11})d_{i_1}(a_{12})\\ \end{aligned}

    for all a_{st}\in A_{st} with s\leq t\in \{1, 2\} .

    Adopt the same discussion as relations \varpi_m(a_{12}a_{22}) = \delta_m(a_{12}a_{22}) = \delta_m(P_n(a_{22}, a_{12}, p_1, p_2, \cdots, p_2)) with a_{22}a_{12}p_1p_2\cdots p_2 = 0 , and we can prove

    \varpi_n(a_{12}a_{22}) = a_{12}\varpi_n(a_{22})+\varpi_n(a_{12})a_{22}+\sum\limits_{i+j = n, 0 < i, j < n}d_i(a_{12})d_{j}(a_{22})

    for all a_{st}\in A_{st} with s\leq t\in \{1, 2\} .

    Lemma 4.10. With notations as above, we have

    1) \varpi_m(a_{11}a^\prime_{11}) = \varpi_m(a_{11})a^\prime_{11}+a_{11}\varpi_m(a^\prime_{11})p_2+\sum_{i+j = m, 0 < i, j < m}d_i(a_{11})d_{j}(a^\prime_{11}) ;

    2) \varpi_m(b_{22}b^\prime_{22}) = \varpi_m(b_{22})b^\prime_{22}+b_{22}\varpi_m(b^\prime_{22})p_2+\sum_{i+j = n, 0 < i, j < m}d_i(b_{22})d_{j}(b^\prime_{22})

    for all a_{ii}, a_{ii}^\prime\in A_{ii} with i\in \{1, 2\} .

    Proof. For conclusion 1) , arbitrary a_{11}, a_{11}^\prime\in A_{11} and a_{12}\in A_{12} and by conclusion 1) in Lemma 4.9, we have

    \begin{align} \begin{aligned} \varpi_m(a_{11}a_{11}^\prime a_{12})& = a_{11}a_{11}^\prime\varpi_m(a_{12})+\varpi_m(a_{11}a_{11}^\prime)a_{12}\\ &+\sum\limits_{i+j = m, 0 < i, j < m}d_i(a_{11}a_{11}^\prime)d_{j}(a_{12})\\ & = a_{11}a_{11}^\prime\varpi_m(a_{12})+\varpi_m(a_{11}a_{11}^\prime)a_{12}\\ &+\sum\limits_{i+j = m, 0 < i, j < m}(\sum\limits_{i_1+i_2 = i, 0 < i < m}d_{i_1}(a_{11})d_{i_2}(a_{11}^\prime))d_{j}(a_{12})\\ & = a_{11}a_{11}^\prime\varpi_m(a_{12})+\varpi_m(a_{11}a_{11}^\prime)a_{12}\\ &+\sum\limits_{i_1+i_2+j = m, 0 < i_1, i_2, j < m}d_{i_1}(a_{11})d_{i_2}(a_{11}^\prime)d_{j}(a_{12}) \end{aligned} \end{align} (4.11)

    and

    \begin{align} \begin{aligned} \varpi_m(a_{11}a_{11}^\prime a_{12})& = a_{11}\varpi_m(a_{11}^\prime a_{12})+\varpi_m(a_{11})a_{11}^\prime a_{12}\\ &+\sum\limits_{i+j = m, 0 < i, j < m}d_i(a_{11})d_{j}(a_{11}^\prime a_{12})\\ & = a_{11}a^\prime_{11}\varpi_m(a_{12})+a_{11}\varpi_m(a^\prime_{11})a_{12}+\varpi_m(a_{11})a_{11}^\prime a_{12}\\ &+\sum\limits_{i+j = n, 0 < i, j < m}a_{11}d_i(a^\prime_{11})d_{j}(a_{12})+\sum\limits_{i+j = m, 0 < i, j < m}d_i(a_{11})d_{j}(a^\prime_{11} a_{12})\\ & = a_{11}a^\prime_{11}\varpi_m(a_{12})+a_{11}\varpi_m(a^\prime_{11})a_{12}+\varpi_m(a_{11})a^\prime_{11} a_{12}\\ &+\sum\limits_{i+j = m, 0 < i, j < m}a_{11}d_i(a^\prime_{11})d_{j}(a_{12})+\sum\limits_{i+j = m, 0 < i, j < m}d_i(a_{11})(\sum\limits_{j_1+j_2 = j, 0 < j < m}d_{j_1}(a^\prime_{11})d_{j_2}( a_{12}))\\ & = a_{11}a^\prime_{11}\varpi_m(a_{12})+a_{11}\varpi_m(a^\prime_{11})a_{12}+\varpi_m(a_{11})a_{11}^\prime a_{12}\\ &+\sum\limits_{i+j = m, 0 < i, j < m}a_{11}d_i(a^\prime_{11})d_{j}(a_{12})+\sum\limits_{i+j_1+j_2 = m, 0 < i, j_1, j_2 < m}d_i(a_{11})d_{j_1}(a^\prime_{11})d_{j_2}( a_{12})\\ & = a_{11}a^\prime_{11}\varpi_m(a_{12})+a_{11}\varpi_m(a^\prime_{11})a_{12}+\varpi_m(a_{11})a^\prime_{11} a_{12}\\ &+\sum\limits_{i+j = m, 0 < i, j < m}d_i(a_{11})d_{j}(a^\prime_{11})a_{12}+\sum\limits_{i+j_1+j_2 = m, 0 < i, j_1, j_2 < m}d_i(a_{11})d_{j_1}(a^\prime_{11})d_{j_2}(a_{12}) \end{aligned} \end{align} (4.12)

    for all a_{tt}, a_{tt}^\prime\in A_{tt} with t\in \{1, 2\} .

    Combining (4.12) with (4.13) leads to

    \varpi_m(a_{11}a_{11}^\prime)a_{12} = (\varpi_m(a_{11})a_{11}^\prime +a_{11}\varpi_m(a_{11}^\prime)+\sum\limits_{i+j = m, 0 < i, j < m}d_i(a_{11})d_{j}(a_{11}^\prime))a_{12}

    for all a_{tt}, a_{tt}^\prime\in A_{tt} with t\in \{1, 2\} .

    Since \varpi_m(A_{11})\subseteq A_{11}+A_{12} and A_{12} are faithful as a left A_{11} -module, the above relation implies that

    \begin{align} \varpi_m(a_{11}a_{11}^\prime)p_1 = \{\varpi_m(a_{11})a_{11}^\prime +a_{11}\varpi_m(a_{11}^\prime)+\sum\limits_{i+j = m, 0 < i, j < m}d_i(a_{11})d_{j}(a_{11}^\prime)\}p_1 \end{align} (4.13)

    for all a_{11}, a_{11}^\prime\in A_{11} .

    On the other hand, by a_{11}\underbrace{p_2\cdots p_2}_{n-1 \ \text{copies}} = 0 for all a_{11}\in A_{11} , we arrive at

    \begin{aligned} 0& = \delta_m(P_n(a_{11}, \underbrace{p_2, \cdots , p_2}_{n-1 \ \text{copies}}))\\ & = P_n(\delta_m(a_{11}), \underbrace{p_2, \cdots , p_2}_{n-1 \ \text{copies}})+P_n(a_{11}, \delta_m(p_2), \underbrace{p_2, \cdots , p_2}_{n-2 \ \text{copies}})\\ &+\sum\limits_{i_1+\cdots+i_n = m, i_1, \cdots, i_n < m}P_n(\delta_{i_1}(a_{11}), \delta_{i_2}(p_2), \cdots , \delta_{i_n}(p_2))\\ & = P_n(\varpi_m(a_{11}), \underbrace{p_2, \cdots , p_2}_{n-1 \ \text{copies}})+P_n(a_{11}, \varpi_m(p_2), \underbrace{p_2, \cdots , p_2}_{n-2 \ \text{copies}})\\ &+\sum\limits_{i_1+\cdots+i_n = m, i_1, \cdots, i_n < m}P_n(d_{i_1}(a_{11}), d_{i_2}(p_2), \cdots , d_{i_n}(p_2))\\ \end{aligned}

    for all a_{11}, a_{11}^\prime\in A_{11} .

    Since \varpi_n(A_{11})\subseteq A_{11}+A_{12}, \varpi_n(p_2)\in A_{12} and d_i(p_2)\in A_{12} , the above equation implies that

    0 = \varpi_m(a_{11})p_2+a_{11}\varpi_m(p_2)+\sum\limits_{i+j = m, 0\leq i, j < m}d_i(a_{11})d_j(p_2)

    for all a_{11}, a_{11}^\prime\in A_{11} .

    On substituting a_{11} by a_{11}a^\prime_{11} in above equation, we get

    \begin{aligned} 0& = \varpi_m(a_{11}a^\prime_{11})p_2+a_{11}a^\prime_{11}\varpi_m(p_2)+\sum\limits_{i+j = m, 0\leq i, j < m}d_i(a_{11}a^\prime_{11})d_j(p_2)\\ & = \varpi_m(a_{11}a^\prime_{11})p_2+a_{11}a^\prime_{11}\varpi_m(p_2)\\ &+\sum\limits_{i+j = m, 0\leq i, j < m}(\sum\limits_{i_1+i_2 = i, 0\leq i_1, i_2 < m}d_{i_1}(a_{11})d_{i_2}(a^\prime_{11}))d_j(p_2)\\ & = \varpi_m(a_{11}a^\prime_{11})p_2+a_{11}a^\prime_{11}\varpi_m(p_2)\\ &+\sum\limits_{i_1+i_2+j = m, 0\leq i_1, i_2, j < m}d_{i_1}(a_{11})d_{i_2}(a^\prime_{11})d_j(p_2) \end{aligned}

    for all a_{11}, a_{11}^\prime\in A_{11} . Therefore, we have

    \begin{align} \begin{aligned} &p_1(\varpi_m(a_{11}a^\prime_{11})p_2+a_{11}a^\prime_{11}\varpi_m(p_2)\\ &+\sum\limits_{i_1+i_2+j = m, 0\leq i_1, i_2, j < m}d_{i_1}(a_{11})d_{i_2}(a^\prime_{11})d_j(p_2))p_2 = 0. \end{aligned} \end{align} (4.14)

    Again, note that a^\prime_{11} p_2\cdots p_2 = 0 for all a^\prime_{11}\in A_{11} , and we have

    \begin{aligned} 0 = \delta_m(P_n(a^\prime_{11}, p_2, \cdots , p_2))\\ & = \sum\limits_{i_1+\cdots+i_n = m}P_n(\delta_{i_1}(a^\prime_{11}), \delta_{i_2}(p_2), \cdots , \delta_{i_n}(p_2))\\ & = P_n(\varpi_m(a^\prime_{11}), p_2, \cdots, p_2)]+P_n(a^\prime_{11}, \varpi_m(p_2), p_2, \cdots, p_2]\\ &+\sum\limits_{i_1+\cdots+i_n = m, i_1, \cdots, i_n < m}P_n(d_{i_1}(a^\prime_{11}), d_{i_2}(p_2), \cdots , d_{i_n}(p_2)). \end{aligned}

    This gives us

    \begin{align} 0 = \varpi_m(a^\prime_{11})p_2+a^\prime_{11}\varpi_m(p_2)+\sum\limits_{i+j = m, 0\leq i, j < m}d_i(a^\prime_{11})d_j(p_2). \end{align} (4.15)

    Now, left multiplying a_{11} in (4.16) and combining it with (4.15) gives us

    \varpi_m(a_{11}a^\prime_{11})p_2+\sum\limits_{i+j+k = m, 0\leq i, 0 < j}d_i(a_{11})d_j(a^\prime_{11})d_k(p_2) = a_{11}\varpi_m(a^\prime_{11})p_2.

    This implies that

    \varpi_m(a_{11}a^\prime_{11})p_2+\sum\limits_{i = 1}^md_i(a_{11})\sum\limits_{j+k = m-i, 0\leq i}d_j(a^\prime_{11})d_k(p_2) = a_{11}\varpi_m(a^\prime_{11})p_2.

    Now, using the condition \mathfrak{F}_s , we find that

    \varpi_m(a_{11}a^\prime_{11})p_2-\sum\limits_{i = 1}^{m-1}d_i(a_{11})d_{m-i}(a^\prime_{11})p_2 = a_{11}\varpi_m(a^\prime_{11})p_2,

    which gives

    \varpi_m(a_{11}a^\prime_{11})p_2 = a_{11}\varpi_m(a^\prime_{11})p_2+\sum\limits_{i = 1}^{m-1}d_i(a_{11})d_{m-i}(a^\prime_{11})p_2.

    Hence,

    \begin{align} \varpi_m(a_{11}a^\prime_{11})p_2 = \{\varpi_m(a_{11})a^\prime_{11}+a_{11}\varpi_m(a^\prime_{11})p_2+\sum\limits_{i+j = m, 0 < i, j < m}d_i(a_{11})d_{m-i}(a^\prime_{11})\}p_2. \end{align} (4.16)

    Now, adding (4.14) and (4.17) , we have

    \varpi_m(a_{11}a^\prime_{11}) = \varpi_m(a_{11})a^\prime_{11}+a_{11}\varpi_m(a^\prime_{11})p_2+\sum\limits_{i+j = m, 0 < i, j < m}d_i(a_{11})d_{j}(a^\prime_{11}).

    Adopting the same discussion, we have

    \varpi_m(b_{22}b^\prime_{22}) = \varpi_m(b_{22})b^\prime_{22}+b_{22}\varpi_m(b^\prime_{22})p_2+\sum\limits_{i+j = m, 0 < i, j < m}d_i(b_{22})d_{j}(b^\prime_{22})

    for all b_{22}, b^\prime_{22}\in A_{22} .

    Remark 4.1. Now, we establish a mapping g_m:\mathcal{T}\rightarrow \mathcal{Z}(\mathcal{T}) by

    g_m(x) = \varpi_m(x)-\varpi_m(p_1xp_1)-\varpi_m(p_1xp_2)-\varpi_m(p_2xp_2)

    and g_m(P_n(x_1, \cdots, x_n)) = 0 with x_1\cdots x_n = 0 for all x_1, \cdots, x_n\in \mathcal{T} , then define a mapping \chi_m(x) = \varpi_m(x)-g_m(x) for all x\in \mathcal{T} . It is easy to verify that

    \chi_m(a_{11}+a_{12}+a_{22}) = \chi_m(a_{11})+\chi_m(a_{12})+\chi_m(a_{22}).

    From the definition of \chi_m and g_m , we find that

    \varphi_m(x) = \varpi_m(x)+f_m(x) = \chi_m(x)+g_m(x)+f_m(x) = \chi_m(x)+h_m(x),

    where h_m(x) = g_m(x)+f_m(x) for all x\in \mathcal{T} .

    Lemma 4.11. With notations as above, we obtain that \{\chi_i\}^{i = m}_{i = 0} is an additive higher derivation on triangular algebras \mathcal{T} .

    Proof. Suppose that x, y\in \mathcal{T} , such that x = a_{11}+a_{12}+a_{22} and y = a^\prime_{11}+a^\prime_{12}+a^\prime_{22} , where a_{ij}, a^\prime_{ij}\in A_{ij} with i\leqslant j\in\{1, 2\} , then

    \begin{aligned} \chi_m(x+y)& = \chi_m((a_{11}+a_{12}+a_{22})+(a^\prime_{11}+a^\prime_{12}+a^\prime_{22}))\\ & = \chi_m((a_{11}+a^\prime_{11})+(a_{12}+a^\prime_{12})+(a_{22}+a^\prime_{22}))\\ & = \chi_m(a_{11}+a^\prime_{11})+\chi_m(a_{12}+a^\prime_{12})+\chi_m(a_{22}+a^\prime_{22})\\ & = \chi_m(a_{11})+\chi_m(a^\prime_{11})+\varpi_m(a_{12})+\chi_m(a^\prime_{12})+\chi_m(a_{22})+\chi_m(a^\prime_{22})\\ & = \chi_m(a_{11}+a_{12}+a_{22})+\chi_m(a^\prime_{11}+a^\prime_{12}+a^\prime_{22}))\\ & = \chi_m(x)+\chi_m(y). \end{aligned}

    By Lemmas 4.8 and 4.10, we have

    \begin{align} \begin{aligned} \chi_m(xy)& = \chi_m((a_{11}+a_{12}+a_{22})(a^\prime_{11}+a^\prime_{12}+a^\prime_{22}))\\ & = \chi_m(a_{11}a^\prime_{11}+a_{11}a^\prime_{12}+a_{12}a^\prime_{22}+a_{22}a^\prime_{22})\\ & = \varpi_m(a_{11})a^\prime_{11}+a_{11}\varpi_m(a^\prime_{11})+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{11})d_j(a^\prime_{11})\\ &+\varpi_m(a_{11})a^\prime_{12}+a_{11}\varpi_m(a^\prime_{12})+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{11})d_j(a^\prime_{12})\\ &+\varpi_m(a_{12})a^\prime_{22}+a_{12}\varpi_m(a^\prime_{22})+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{12})d_j(a^\prime_{22})\\ &+\varpi_m(a_{22})a^\prime_{22}+a_{22}\varpi_m(a^\prime_{22})+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{22})d_j(a^\prime_{22}). \end{aligned} \end{align} (4.17)

    On the other hand, we have

    \begin{aligned} &\chi_m(x)y+x\chi_m(y)+\sum\limits_{i+j = m, 0 < i < m}\chi_i(x)\chi_j(y)\\ & = \chi_m(a_{11}+a_{12}+a_{22})y+x\chi_m(a^\prime_{11}+a^\prime_{12}+a^\prime_{22})+\sum\limits_{i+j = m, 0 < i < m}\chi_i(x)\chi_j(y)\\ & = (\varpi_m(a_{11})+\varpi_m(a_{12})+\varpi_m(a_{22}))y+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{11})d_j(a^\prime_{11})\\ &+x(\varpi_m(a^\prime_{11})+\varpi_m(a^\prime_{12})+\varpi_m(a^\prime_{22}))+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{11})d_j(a^\prime_{12})+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{11})d_j(a^\prime_{22})\\ &+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{12})d_j(a^\prime_{11})+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{12})d_j(a^\prime_{12})+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{12})d_j(a^\prime_{22})\\ &+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{22})d_j(a^\prime_{11})+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{22})d_j(a^\prime_{12})+\sum\limits_{i+j = m, 0 < p < m}d_i(a_{22})d_j(a^\prime_{22}). \end{aligned}

    Taking into account the induction hypothesis \mathfrak{F}_s , Lemmas 4.9 and 4.10, we calculate that

    \begin{align} \begin{aligned} &\chi_m(x)y+x\chi_m(y)+\sum\limits_{i+j = m, 0 < i < m}d_i(x)d_j(y)\\ & = \varpi_m(a_{11})a^\prime_{11}+\varpi_m(a_{11})a^\prime_{12}+\varpi_m(a_{12})a^\prime_{22}+\varpi_m(a_{22})a^\prime_{22}\\ &+a_{11}\varpi_m(a^\prime_{11})+a_{11}\varpi_m(a^\prime_{12})+a_{12}\varpi_m(a^\prime_{22})+a_{22}\varpi_m(a^\prime_{22})\\ &+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{11})d_j(a^\prime_{11})+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{11})d_j(a^\prime_{12})\\ &+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{12})d_j(a^\prime_{22})+\sum\limits_{i+j = m, 0 < i < m}d_i(a_{22})d_j(a^\prime_{22}). \end{aligned} \end{align} (4.18)

    Combining (4.18) and (4.19) , we get

    \chi_m(xy) = \chi_m(x)y+x\chi_m(y)+\sum\limits_{i+j = m, 0 < i < m}\chi_i(x)\chi_j(y)

    for all x, y\in \mathcal{T} . This shows that each \chi_m satisfies the Leibniz formula of higher order on \mathcal{T} .

    Finally, we need to prove that each h_m vanishes P_n(x_1, \cdots, x_n) with x_1\cdots x_n = 0 for all x_1, \cdots, x_n\in \mathcal{T} . Note that h_m maps into \mathcal{Z}(\mathcal{T}) , \{\chi_i\}^m_{i = 0} as an additive higher derivation of \mathcal{T} . It follows from inductive hypothesis \mathfrak{F}_s that

    h_m(P_n(x_1, \cdots, x_n)) = \delta_n(P_n(x_1, \cdots, x_n))-\chi_n(P_n(x_1, \cdots, x_n)) = 0

    with x_1\cdots x_n = 0 for all x_1, \cdots, x_n\in \mathcal{T} . We lastly complete the proof of the main theorem.

    In particular, we have the following corollaries.

    When n = 3 , we have the following corollary.

    Corollary 4.1. [17,Theorem 3.3] Let \mathcal{T} = \left[\begin{array}{cc} A_{11} & A_{12}\\ O & A_{22} \end{array}\right] be a triangular algebra satisfying

    \text{i}) \pi_{A_{11}}(\mathcal{Z}(\mathcal{T})) = \mathcal{Z}(\mathcal{A}_{11}) and \pi_{A_{22}}(\mathcal{Z}(\mathcal{T})) = \mathcal{Z}(\mathcal{A}_{22}) .

    \text{ii}) For any a_{11}\in A_{11} , if [a_{11}, A_{11}]\in \mathcal{Z}(\mathcal{A})_{11} , then a_{11}\in \mathcal{Z}(\mathcal{A}) , or for any a_{22}\in A_{22} , if [a_{22}, A_{22}]\in \mathcal{Z}(\mathcal{A}_{22}) , then a_{22}\in \mathcal{Z}(\mathcal{A}_{22}) .

    Suppose that a sequence \Delta = \{\delta_m\}_{m\in \mathcal{N}} of mappings \delta_m:\mathcal{T}\rightarrow \mathcal{T} is a nonlinear map satisfying

    \delta_m([[x, y], z]) = \sum\limits_{i+j+k = m}[[\delta_i(x), \delta_j(y)], \delta_k(z)]

    for all x, y, z \in \mathcal{T} with xyz = 0 . For every m\in \mathcal{N} ,

    \delta_m(x) = \chi_m(x)+h_m(x)

    for all x\in \mathcal{T} , where a sequence \Upsilon = \{\chi_m \}_{m\in \mathcal{N}} of additive mapping \chi_m:\mathcal{T}\rightarrow \mathcal{T} is a higher derivation and h_m:\mathcal{T}\rightarrow \mathcal{Z}(\mathcal{T}) is a nonlinear mapping, such that h_m([[x, y], z]) = 0 for any x, y, z\in \mathcal{T} with xyz = 0 .

    When n = 3 and m = 1 , we have the following corollary.

    Corollary 4.2. [16,Theorem 2.2] Let \mathcal{T} = \left[\begin{array}{cc} A_{11} & A_{12}\\ O & A_{22} \end{array}\right] be a triangular algebra satisfying

    \text{i}) \pi_{A_{11}}(\mathcal{Z}(\mathcal{T})) = \mathcal{Z}(\mathcal{A}_{11}) and \pi_{A_{22}}(\mathcal{Z}(\mathcal{T})) = \mathcal{Z}(\mathcal{A}_{22}) .

    \text{ii}) For any a_{11}\in A_{11} , if [a_{11}, A_{11}]\in \mathcal{Z}(\mathcal{A}_{11}) , then a_{11}\in \mathcal{Z}(\mathcal{A}) , or for any a_{22}\in A_{22} , if [a_{22}, A_{22}]\in \mathcal{Z}(\mathcal{A}_{22}) then a_{22}\in \mathcal{Z}(\mathcal{A}_{22}) .

    Suppose \delta_1:\mathcal{T}\rightarrow \mathcal{T} is a nonlinear map satisfying

    \delta_1([[x, y], z]) = \sum\limits_{i+j+k = 1}[[\delta_i(x), \delta_j(y)], \delta_k(z)]

    for all x, y, z \in \mathcal{T} with xyz = 0 , then there exists an additive derivation \varpi_1 of \mathcal{T} and a nonlinear map h_1:\mathcal{T}\rightarrow \mathcal{Z}(\mathcal{T}) , such that

    \delta_1(x) = \chi_1(x)+h_1(x)

    for all x\in \mathcal{T} , where \tau_1([[x, y], z]) for any x, y, z\in \mathcal{T} with xyz = 0 .

    In this section, we apply Theorem 4.2 to certain classes of triangular algebras that satisfy the hypotheses of Theorem 4.2. Some standard examples of triangular rings satisfying the hypotheses of Theorem 4.1 are: Upper triangular matrix algebras, block upper triangular matrix algebras and nest algebras (see [24,4. More applications and further topics] for details).

    According to the Theorem 4.2, we get the following corollaries.

    Corollary 5.1. Let \mathcal{R} be a 2 -torsion free commutative ring with identity and \mathcal{T}_k(\mathcal{R})(k\geq 2) be the algebra of all k\times n upper triangular matrices over \mathcal{R} . Let \{L_m\}_{m\in \mathcal{N}} be a family of nonlinear mapping L_m: \mathcal{T}_k(\mathcal{R})\longrightarrow \mathcal{T}_k(\mathcal{R}) satisfying Eq (1.3) , then every Lie- m higher derivation L_m: \mathcal{T}_k(\mathcal{R})\longrightarrow \mathcal{T}_k(\mathcal{R}) be of standard form.

    Corollary 5.2. Let \mathcal{R} be a 2 -torsion free commutative ring with identity and \mathcal{T}_s^{\overline{k}}(\mathcal{R})(s\geq 3) be a block upper triangular matrix ring with over \mathcal{T}_s^{\overline{k}}(\mathcal{R})\neq M_s(R) . Let \{L_m\}_{m\in \mathcal{N}} be a family of nonlinear mapping L_m: \mathcal{T}_s^{\overline{k}}(\mathcal{R})\longrightarrow \mathcal{T}_s^{\overline{k}}(\mathcal{R}) satisfying Eq (1.3) , then every Lie- m higher derivation L_m: \mathcal{T}_s^{\overline{k}}(\mathcal{R})\longrightarrow \mathcal{T}_s^{\overline{k}}(\mathcal{R}) be of standard form.

    Corollary 5.3. Let {\rm H} be a Hilbert space, \mathcal{N} be a nest of {\rm H} and {\rm Alg}(\mathcal{N}) be the nest algebra associated with \mathcal{N} . Let \{L_m\}_{m\in \mathcal{N}} be a family of nonlinear mapping L_m: {\rm Alg}(\mathcal{N})\longrightarrow {\rm Alg}(\mathcal{N}) satisfying Eq (1.3) , then every Lie- m higher derivation L_m: {\rm Alg}(\mathcal{N})\longrightarrow {\rm Alg}(\mathcal{N}) be of standard form.

    The purpose of this article was to prove that every nonlinear Lie-n higher derivation by local actions on the triangular algebras is of a standard form. As an application, we gave a characterization of Lie- n higher derivations by local actions on upper triangular matrix algebras, block upper triangular matrix algebras and nest algebras, respectively.

    The author declares she has not used Artificial Intelligence(AI) tools in the creation of this article.

    This work was supported by the Youth fund of Anhui Natural Science Foundation (Grant No. 2008085QA01), Key projects of University Natural Science Research Project of Anhui Province (Grant No. KJ2019A0107) and National Natural Science Foundation of China (Grant No. 11801008).

    The authors declare no conflicts of interest.



    [1] R. V. Kadison, Local derivations, J. Algebra, 130 (1990), 494–509. https://doi.org/10.1016/0021-8693(90)90095-6
    [2] D. R. Larson, A. M. Sourour, Local derivations and local automorphisms of B(X), Proc. Sympos. Pure Math., 51 (1990), 187–194.
    [3] S. Ebrahimi, Lie triple derivations on primitive rings, Asian-Eur. J. Math., 08 (2015), 1550019. https://doi.org/10.1142/S1793557115500199 doi: 10.1142/S1793557115500199
    [4] P. S. Ji, R. R. Liu, Y. Z. Zhao, Nonlinear Lie triple derivations of triangualr algebras, Linear Multilinear A., 60 (2012), 1155–1164. https://doi.org/10.1080/03081087.2011.652109 doi: 10.1080/03081087.2011.652109
    [5] X. F. Qi, Characterization of Lie higher derivations on triangular algebras, Acta Math. Sin.-English Ser., 29 (2013), 1007–1018. https://doi.org/10.1007/s10114-012-1548-3 doi: 10.1007/s10114-012-1548-3
    [6] X. F. Qi, J. C. Hou, Lie higher derivations on nest algebras, Comm. Math. Res., 26 (2010), 131–143.
    [7] Z. K. Xiao, F. Wei, Nonlinear Lie higher derivations on triangular algebras, Linear Multilinear A., 60 (2012), 979–994. https://doi.org/10.1080/03081087.2011.639373 doi: 10.1080/03081087.2011.639373
    [8] J. K. Li, Q. H. Shen, Characterizations of Lie higher and Lie triple derivations on triangular algebras, J. Korean Med. Sci., 49 (2012), 419–433. https://doi.org/10.4134/JKMS.2012.49.2.419 doi: 10.4134/JKMS.2012.49.2.419
    [9] M. Ferrero, C. Haetinger, Higher derivations and a theorem by Herstein, Quaest. Math., 25 (2002), 249–257. https://doi.org/10.2989/16073600209486012 doi: 10.2989/16073600209486012
    [10] M. Khrypchenko, F. Wei, Lie-type derivations of finitary incidence algebras, Rocky Mountain J. Math., 50 (2020), 163–175. https://doi.org/10.1216/rmj.2020.50.163 doi: 10.1216/rmj.2020.50.163
    [11] P. S. Ji, W. Q. Qi, Characterizations of Lie derivations of triangular algebras, Linear Algebra Appl., 435 (2011), 1137–1146. https://doi.org/10.1016/j.laa.2011.02.048 doi: 10.1016/j.laa.2011.02.048
    [12] W. H. Lin, Characterizations of Lie higher derivations on triangular algebras, Indian J. Pure Appl. Math., 51 (2020), 77–104. https://doi.org/10.1007/s13226-020-0386-8 doi: 10.1007/s13226-020-0386-8
    [13] M. Ashraf, A. Jabeen, Characterization of Lie type derivation on von Neumann algebra with local actions, Bull. Korean Math. Soc., 58 (2021), 1193–1208. https://doi.org/10.4134/BKMS.b200850 doi: 10.4134/BKMS.b200850
    [14] L. Liu, Lie triple derivations on factor von Neumann algebras, B. Korean Math. Soc., 52 (2015), 581–591. http://doi.org/10.4134/BKMS.2015.52.2.581 doi: 10.4134/BKMS.2015.52.2.581
    [15] L. Liu, Lie triple derivations on von Neumann algebras, Chin. Ann. Math. Ser. B., 39 (2018), 817–828. https://doi.org/10.1007/s11401-018-0098-0 doi: 10.1007/s11401-018-0098-0
    [16] X. P. Zhao, Nonlinear Lie triple derivations by local actions on triangular algebras, J. Algebra Appl., 22 (2023), 2350059. https://doi.org/10.1142/S0219498823500597. doi: 10.1142/S0219498823500597
    [17] X. F. Liang, D. D. Ren, Q. L. Li, Nonlinear Lie triple higher derivations on triangular algebras by local actions: A new perspective, Axioms, 11 (2022), 328. https://doi.org/10.3390/axioms11070328 doi: 10.3390/axioms11070328
    [18] H. Ghahramani, M. N. Ghosseiri, T. Rezaei, Characterizing Jordan derivable maps on triangular rings by local actions, J. Math., 2022 (2022), 9941760. https://doi.org/10.1155/2022/9941760 doi: 10.1155/2022/9941760
    [19] H. Ghahramani, A. H. Mokhtari, F. Wei, Lie centralizers and commutant preserving maps on generalized matrix algebras, J. Algebra Appl., 2022 (2022), 2450106. https://doi.org/10.1142/S0219498824501068 doi: 10.1142/S0219498824501068
    [20] W. S. Cheung, Commuting maps of triangular algebras, J. Lond. Math. Soc., 63 (2001), 117–127. https://doi.org/10.1112/S0024610700001642 doi: 10.1112/S0024610700001642
    [21] K. R. Davidson, Nest algebras, London: Longman Scientific and Technical Publishing Corporation, 1988.
    [22] W. S. Cheung, Lie derivations of triangular algebras, Linear Multilinear A., 51 (2003), 299–310. https://doi.org/10.1080/0308108031000096993 doi: 10.1080/0308108031000096993
    [23] D. Benkovic, D. Eremita, Multiplicative Lie n-derivations of triangular rings, Linear Algebra Appl., 436 (2012), 4223–4240. https://doi.org/10.1016/j.laa.2012.01.022 doi: 10.1016/j.laa.2012.01.022
    [24] Z. K. Xiao, F. Wei, Lie triple derivations of triangular algebras, Linear Algebra Appl., 437 (2012), 1234–1249. https://doi.org/10.1016/j.laa.2012.04.015 doi: 10.1016/j.laa.2012.04.015
  • Reader Comments
  • © 2024 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(1267) PDF downloads(176) Cited by(0)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog