Chapter 5

Generalized Jordan triple (σ, τ )-higher

derivations in rings

5.1 Introduction
It is well known that if d : R → R is a Jordan derivation of a 2-torsion free
ring then it satisfies d(ab + ba) = d(a)b + ad(b) + d(b)a + bd(a) and d(aba) =
d(a)ba + ad(b)a + abd(a) for all a, b ∈ R. Using these identities Herstein
[45] showed that every Jordan derivation on a prime ring of characteristic
different from two is a derivation. Bresar [26], generalized this result for
semi-prime ring and termed any additive mapping d : R → R satisfying
d(aba) = d(a)ba+ad(b)a+abd(a) for all a, b ∈ R as a Jordan triple derivation.
In fact, he established that every Jordan triple derivation of a 2-torsion free
semi-prime ring is a derivation.
One natural generalization of Jordan triple derivation is generalized
Jordan triple derivation given by Jing and Lu [52]. An additive mapping
f : R → R is said to a generalized Jordan triple derivation on R if there exists
a Jordan triple derivation d on R such that f(aba) = f (a)ba+ad(b)a+abd(a)
for all a, b ∈ R. They also showed that every generalized Jordan triple deriva-
tion on a prime ring of characteristic different from two is a generalized
derivation. Further, Liu & Shiue [58] extended this result for generalized
Jordan triple (σ, τ )-derivation. Let σ, τ be the endomorphisms of R. An
additive mapping f : R → R is said to a generalized Jordan triple (σ, τ )-
derivation on R if there exists a Jordan triple (σ, τ )-derivation d on R such
that f(aba) = f(a)τ (ba) + σ(a)d(b)τ(a) + σ(ab)d(a) holds for all a, b ∈ R.
Liu & Shiue proved that on a 2-torsion free semi-prime ring every generalized
Jordan triple (σ, τ )-derivation is a generalized (σ, τ )-derivation.
In the present chapter, applying various results of Chapter 4, we shall
obtain rather more general results for generalized Jordan triple (σ, τ )-higher
derivations in prime and semi-prime rings.
In Section 5.2, we introduce the concept of generalized Jordan triple

78
(σ, τ)-higher derivation. It is shown that every generalized Jordan triple
(σ, τ)- higher derivation of a 2-torsion free semi-prime ring R is a generalized
(σ, τ)- higher derivation of R.
Further in Section 5.3, we shall consider generalized Jordan triple (σ, τ )-
higher derivations on a Lie ideal U of R and obtain the conditions on a prime
ring R under which every generalized Jordan triple (σ, τ)-higher derivation
of U into R is a generalized (σ, τ )-higher derivation of U into R.

5.2 Generalized Jordan triple (σ, τ )-higher derivations

on semi-prime rings
Inspired by the concepts of Jordan triple derivation, generalized derivation
and (σ, τ )-higher derivation we introduce generalized Jordan triple (σ, τ )-
higher derivation as follows: a family of additive mappings F = {fn }n∈N I of
R is said to be a generalized Jordan triple (σ, τ )-higher derivation if there
exist a (σ, τ)-higher
 derivation D = {dn }n∈NIn−kof R such that f0 = IR and
n−i k i
fn (aba) = fi (σ (a))dj (σ τ (b))dk (τ (a)) holds for all a, b ∈ R
i+j+k=n
and every n ∈ N.I Consequently, in the above definition for fi = di we get
the notion of Jordan triple (σ, τ )-higher derivation.

The following Lemma is crucial for developing the proof of our results.

Lemma 5.2.1 ([17], Proposition 2.2.1) Let R be a semi-prime ring. Then

Qr satisfies:

(i) R is a subring of Qr .

(ii) For all q ∈ Qr there exists a dense ideal I of R such that qI ⊆ R.

(iii) For all q ∈ Qr , qI=0 if and only if q = 0.

(iv) For any dense ideal I and d : IR → RR there exists q ∈ Qr such that
d(x) = qx for all x ∈ I.

Furthermore, these properties (i) − (iv) characterize Qr up to isomorphism.

Here, it is worth mentioning that if R is a semi-prime ring then Qr is also

semi-prime containing the identity element.

For every fixed n ∈ N I and each a, b, c ∈ R we denote [a, b, c], Ψn (a, b) and
Ψn (a, b, c) the elements of R as:

[a, b, c] = abc − cba,

79
 
Ψn (a, b) = fn (ab) − fi (σn−i (a))dj (τ n−j (b)),
i+j=n

Ψn (a, b, c) = fn (abc) − fi (σ n−i (a))dj (σk τ i (b))dk (τ n−k (c)).
i+j+k=n

It can easily be seen that Ψn (a, b) and Ψn (a, b, c) are additive in each
argument and Ψn (a, b, c) = −Ψn (c, b, a). Obviously Ψn (a, b) = 0, for all
a, b ∈ R if and only if F = {fn }n∈N I is a generalized (σ, τ)-higher derivation
and F = {fn }n∈N I is a generalized Jordan triple (σ, τ )-higher derivation
if and only if Ψn (a, b, a) = 0 for all a, b ∈ R. In view of Lemma 4.2.5 it
can be easily seen that on a 2-torsion free ring every generalized Jordan
(σ, τ)-higher derivation is a generalized Jordan triple (σ, τ )-higher derivation
but the converse need not hold in general. In the present section we explore
the converse part of this problem.

Throughout σ, τ are endomorphisms of R such that τ is one-one & onto

and στ = τσ. We begin by proving the following result:
Lemma 5.2.2 Let R be a 2-torsion free ring and F = {fn }n∈N I be a gener-
alized Jordan triple (σ, τ )- higher derivation of R. If Ψm (a, b, c) = 0 for all
m < n then Ψn (a, b, c)xτ n [g, h, k] = 0 for all a, b, c, x, g, h, k ∈ R and every
n ∈ N.
I
Proof Consider ζ = abcxcba + cbaxabc. Then,
fn (ζ) = fn (a(bcxcb)a) + fn (c(baxab)c)

= fi (σn−i (a))dj (σ k τ i (b(cxc)b))dk (τ n−k (a))
i+j+k=n

+ fi (σn−i (c))dj (σk τ i (b(axa)b))dk (τ n−k (c))
i+j+k=n


= fi (σn−i (a))dl (σ y+p+q+s+k τ i (b))dy (σp+q+s+k τ i+l (c))
i+l+y+p+q+s+k=n
dp (σ q+s+k τ i+l+y (x))dq (σs+k τ i+l+y+p (c))ds (σ k τ i+l+y+p+q (b))dk (τ n−k (a))

+ fi (σ n−i (c))dl (σy+p+q+s+k τ i (b))dy (σp+q+s+k τ i+l (a))
i+l+y+p+q+s+k=n
dp (σ q+s+k τ i+l+y (x))dq (σs+k τ i+l+y+p (a))ds (σk τ i+l+y+p+q (b))dk (τ n−k (c)).

(5.2.1)

Again, fn (ζ) = fn (abc)x(cba) + (cba)x(abc) . Applying Lemma 4.2.5(iii) we
get;

fn (ζ) = (fα (σn−α (abc))dp (σ γ τ α(x))dγ (τ n−γ (cba))
α+p+γ=n (5.2.2)
+fα(σ n−α(cba))dp (σγ τ α (x))dγ (τ n−γ (abc))).

80
Equating (5.2.1) and (5.2.2) we have,

0 = fi (σn−i (a))dl (σ y+p+q+s+k τ i (b))dy (σp+q+s+k τ i+l (c))
i+l+y+p+q+s+k=n
dp (σ q+s+k
 τ
i+l+y
(x))dq (σ s+k τ i+l+y+p(c))ds (σ k τ i+l+y+p+q (b))dk (τ n−k (a))
− fα(σ n−α(abc))dp (σγ τ α (x))dγ (τ n−γ (cba))
α+p+γ=n
+ fi (σ n−i (c))dl (σy+p+q+s+k τ i (b))dy (σ p+q+s+k τ i+l (a))
i+l+y+p+q+s+k=n
dp (σ q+s+k
 τ
i+l+y
(x))dq (σ s+k τ i+l+y+p(a))ds (σk τ i+l+y+p+q (b))dk (τ n−k (c))
n−α
− fα(σ (cba))dp (σγ τ α (x))dγ (τ n−γ (abc)).
α+p+γ=n

(5.2.3)
Now consider the first term i.e.,

fi (σn−i (a))dl (σy+p+q+s+k τ i (b))dy (σp+q+s+k τ i+l (c))
i+l+y+p+q+s+k=n
dp (σ q+s+k τ i+l+y (x))dq (σ s+k τ i+l+y+p (c))ds (σk τ i+l+y+p+q (b))dk (τ n−k (a))


= fi (σ n−i (a))dl (σy τ i (b))dy (τ n−y (c))τ n (x)τ n (cba)
i+l+y=n


+ σ n (abc)σn (x)dq (σn−q (c))ds (σk τ q (b))dk (τ n−k (a))
q+s+k=n

0<i+l+y,q+s+k≤n−1

+ fi (σn−i (a))dl (σ y+q+s+k τ i (b))dy (σq+s+k τ i+l (c))
i+l+y+q+s+k=n
(σq+s+k τ i+l+y (x))dq (σs+k τ i+l+y (c))ds (σ k τ i+l+y+q (b))dk (τ n−k (a))

+ fi (σ n−i (a))dl (σy+q+s+k+1 τ i (b))dy (σ q+s+k+1τ i+l (c))
i+l+y+q+s+k=n−1
q+s+k i+l+y
d1 (σ τ (x))dq (σs+k τ i+l+y+1 (c))ds (σ k τ i+l+y+q+1 (b))dk (τ n−k (a))

+ · · · + f1 (σn−1 (a))(τ σ n−1 (b))(τ σn−1 (c))dn−1(τ (x))τ n (cba)

+σn (a)d1 (σ n−1 (b))(τ σn−1 (c))dn−1(τ (x))τ n (cba)
+σn (ab)d1 (σn−1 (c))dn−1 (τ(x))τ n (cba)
+σn (abc)dn−1 (σ(x))d1 (τ n−1 (c))τ n (ba)
+σn (abc)dn−1 (σ(x))(στ n−1 (c))d1 (τ n−1 (b))τ n (a)
+σn (abc)dn−1 (σ(x))(στ n−1 (c))(στ n−1(b))d1 (τ n−1(a))
+σn (abc)dn (x)τ n (cba).

81
Further, calculate the second term

fα (σn−α (abc))dp(σ γ τ α (x))dγ (τ n−γ (cba))
α+p+γ=n

= fn (abc)τ n (x)τ n (cba) + σn (abc)σ n (x)dn (cba)

0<α,γ≤n−1

+ fα (σ n−α (abc))σγ τ α (x)dγ (τ n−γ (cba))
α+γ=n

+ fα (σ n−α (abc))d1(σ γ τ α (x))dγ (τ n−γ (cba))
α+γ=n−1
+ · · · + f1 (σn−1 (abc))dn−1 (τ (x))τ n (cba)
+σ n (abc)dn−1 (σ(x))d1 (τ n−1 (cba)) + σ n (abc)dn (x)τ n (cba).

Using the hypothesis that Ψm (a, b, c) = 0 for all m < n.


fα (σn−α (abc))dp(σ γ τ α (x))dγ (τ n−γ (cba))
α+p+γ=n

= fn (abc)τ n (x)τ n (cba) + σn (abc)σ n (x)dn (cba)

0<i+l+y,q+s+k≤n−1

+ fi (σn−i (a))dl (σ y+q+s+k τ i (b))dy (τ i+l σq+s+k (c))
i+l+y+q+s+k=n
σq+s+k τ i+l+y s+k i+l+y
 (x)dq (σ n−iτ (c))ds (σ k τ i+l+y+q (b))dk (τ n−k (a))
+ fi (σ (a))dl (σy+q+s+k+1 τ i (b))dy (τ i+l σq+s+k+1 (c))
i+l+y+q+s+k=n−1
d1 (σq+s+k τ i+l+y (x))dq (σ s+k τ i+l+y+1 (c))ds (σk τ i+l+y+q+1(b))dk (τ n−k (a))
+ · · · + σ n (abc)dn−1 (σ(x))d1(τ n−1 (c))τ n (ba)
+σn (abc)dn−1 (σ(x))(στ n−1 (c))d1 (τ n−1 (b))τ n (a)
+σn (abc)dn−1 (σ(x))(στ n−1 (c))(στ n−1(b))d1 (τ n−1(a))
+f1(σ n−1(a))(τ σn−1 (b))(τ σ n−1 (c))dn−1 (τ (x))τ n (cba)
+σn (a)d1 (σ n−1 (b))(τ σn−1 (c))dn−1(τ (x))τ n (cba)
+σn (ab)d1 (σn−1 (c))dn−1 (τ(x))τ n (cba)
+σn (abc)dn (x)τ n (cba).

Now, substracting the two terms so obtained and using the hypothesis that
στ = τ σ their difference yields;

fn (abc)τ n (x)τ n (cba) − fi (σ n−i (a))dl (σ s τ i (b))ds (τ n−s (c))τ n (x)τ n (cba)
i+l+s=n


= {fn (abc) − fi (σn−i (a))dl (σs τ i (b))ds (τ n−s (c))}τ n (x)τ n (cba)
i+l+s=n

= Ψn (a, b, c)τ n (x)τ n (cba).

Similarly, the difference of the last two terms of the equation (5.2.3) yields;

82
 
fi (σ n−i (c))dl (σy+p+q+s+k τ i (b))dy (σp+q+s+k τ i+l (a))
i+l+y+p+q+s+k=n

dp (σ q+s+k τ i+l+y (x))dq (σ s+k τ i+l+y+p(a))ds (σk τ i+l+y+p+q (b))dk (τ n−k (c))

− fα (σ n−α (cba))dp (σγ τ α (x))dγ (τ n−γ (abc))
α+p+γ=n

= Ψn (c, b, a)τ n (x)τ n (abc).

Thus, equation (5.2.3) reduces to
Ψn (a, b, c)τ n (x)τ n (cba) + Ψn (c, b, a)τ n (x)τ n (abc) = 0, i.e;
Ψn (a, b, c)τ n (x)τ n [a, b, c] = 0 for all a, b, c ∈ R and each n ∈ N.
I
Since τ is one-one & onto, using Lemma 4.2.4 we obtain that
Ψn (a, b, c)xτ n [g, h, k] = 0 for all a, b, c, x, g, h, k ∈ R and each n ∈ N.
I
Lemma 5.2.3 Let R be a 2-torsion free semi-prime ring and F = {fn }n∈N I
be a generalized Jordan triple (σ, τ )-higher derivation of R. If Ψm (a, b, c) = 0
for all a, b, c ∈ R and m < n then Ψn (a, b, c) ∈ Z(R), for every n ∈ N I and
each a, b, c ∈ R.
Proof For n = 0, Ψ0 (a, b, c) = 0 ∈ Z(R). By induction let us assume that
Ψm (a, b, c) ∈ Z(R), for every a, b, c ∈ R and for all m < n. By Lemma 5.2.2,
for all a, b, c, r, g, h ∈ R we have,
τ n [τ −n (Ψn (a, b, c)), g, h]rτ n [τ −n (Ψn (a, b, c)), g, h]
= (Ψn (a, b, c)τ n (gh) − τ n (hg)Ψn (a, b, c))rτ n [τ −n Ψn (a, b, c), g, h]
= 0.
Thus [Ψn (a, b, c), g, h] = 0 for all a, b, c, g, h ∈ R. If we fix the coefficients
(a, b, c) and regard [Ψn (a, b, c), g, h] = 0 as a generalized polynomial identity
with respect to the two variables g, h and as we know that R and Qr satisfy
the same generalized polynomial identity, [Ψn (a, b, c), g, h] = 0 for all g, h ∈
Qr . Since Qr has the identity element, we obtain [Ψn (a, b, c), g] = 0 for all
g ∈ Qr . In particular, [Ψn (a, b, c), g] = 0 for all g ∈ R and hence we have
Ψn (a, b, c) ∈ Z(R).
Corollary 5.2.4 Let R be a 2-torsion free semi-prime ring and F =
{fn }n∈NI be a generalized Jordan triple (σ, τ )-higher derivation on R. If
n
Ψm (a, b, c) = 0 for all a, b, c ∈ R and m < n then Ψn (a, b, c)τ [g, h, d] = 0 for
I In particular, Ψn (a, b, c)τ n [g, h] = 0.
all a, b, c, g, h, d ∈ R and each n ∈ N.
Lemma 5.2.5 Let R be a 2-torsion free semi-prime ring. If F = {fn }n∈N I
is a generalized Jordan triple (σ, τ )-higher derivation. If Ψm (a, b, c) = 0 for
all a, b, c ∈ R and m < n then Ψn (a, b, c) = 0 for all a, b, c ∈ R and each
n ∈ N.
I

83
Proof Let us first consider the case when R is commutative. Suppose
ϑ = fn (a3 bc + cba3 )

= (fi (σ n−i (a3 ))dj (σk τ i (b))dk (τ n−k (c))
i+j+k=n
+fi (σ n−i (c))dj (σ k τ i (b))dk (τ n−k (a3 )))
Again consider
ϑ = fn ((abc)aa + aa(abc))

= (fi (σn−i (abc))dj (σk τ i (a))dk (τ n−k (a))
i+j+k=n
+fi (σ n−i (a))dj (σk τ i (a))dk (τ n−k (abc))).

Comparing the above two equations for ϑ and as D = {dn }n∈N I is (σ, τ )-
higher derivation, we have

fi (σ n−i (abc))dj (σk τ i (a))dk (τ n−k (a))
i+j+k=n
 (5.2.4)
− fi (σ n−i (c))dj (σ k τ i (b))dk (τ n−k (a3)) = 0.
i+j+k=n

Consider the first term of the equation (5.2.4)


fi (σn−i (abc))dj (σk τ i (a))dk (τ n−k (a))
i+j+k=n


= σn (abc) dj (σ n−j (a))dk (τ n−k (a)) + fn (abc)τ n (a2 )
j+k=n
0<i,j+k≤n−1

+ fi (σ n−i (abc))dj (σk τ i (a))dk (τ n−k (a)).
i+j+k=n

Now consider the second term of the equation (5.2.4)


fi (σ n−i (c))dj (σ k τ i (b))dk (τ n−k (a3))
i+j+k=n

= fi (σ n−i (c))dj (σ r+s+t τ i (b))dr (σs+t τ i+j (a))
i+j+r+s+t=n
ds (σt τ i+j+r (a))dt (τ n−t (a))

= σ n (cba) ds (σ n−s (a))dt (τ n−t (a))
 s+t=n n−i
+ fi (σ (c))dj (σr+s+tτ i (b))dr (τ n−r (a))τ n (a2 )
i+j+r=n
0<i+j+r,s+t≤n−1

+ fi (σ n−i (c))dj (σ r+s+t τ i (b))dr (σ s+tτ i+j (a))
i+j+r+s+t=n
t i+j+r
ds (σ τ (a))dt (τ n−t (a)) .

84
Since R is commutative, equation (5.2.4) reduces to

{fn (abc) − fi (σ n−i (a))dj (σk τ i (b))dk (τ n−k (c))}τ n (a2 ) = 0,
i+j+k=n
i.e., Ψn (a, b, c)τ n (a2) = 0 for all a, b, c ∈ R and each n ∈ N.
I

Using Lemma 5.2.3, we obtain that (Ψn (a, b, c)τ n (a))R(Ψn (a, b, c)τ n (a)) =
Ψn (a, b, c)τ n (a2 )RΨn (a, b, c) = 0. Implementing the semiprimeness of
R we get Ψn (a, b, c)τ n (a) = 0 for all a, b, c ∈ R and each n ∈ N. I
n
Linearizing the above equation on a we obtain Ψn (a, b, c)τ (s) +
Ψn (s, b, c)τ n (a) = 0, for all a, b, c, s ∈ R and each n ∈ N. I Thus we
have, Ψn (a, b, c)τ (s)RΨn (a, b, c)τ (s)=−Ψn (s, b, c)τ (a)RΨn (a, b, c)τ n (s)
n n n

=−Ψn (s, b, c)τ n (s)RΨn (a, b, c)τ n (a) =0 for all a, b, c, s ∈ R and each n ∈ N.
I
As R is semi-prime, Ψn (a, b, c)τ n (s) = 0 for all a, b, c, s ∈ R and each n ∈ N.
I
Finally, Ψn (a, b, c) = 0 for all a, b, c ∈ R and each n ∈ N I.

Now let us consider the case when R is non-commutative. From Lemma

5.2.3 and Corollary 5.2.4 we can easily obtain that for all a, b, c, r, k, l ∈ R
and each n ∈ N,
I

Ψn (a, b, c)rτ n [k, l] + τ n [k, l]rΨn (a, b, c) = 0.

As the existence of prime ideal is ensured by the fact that every ring R is
trivially prime ideal in itself therefore using Lemma 4.2.3 if P is a prime ideal
of R and if Ψn (a, b, c)
∈ P then τ n [k, l] ∈ P . As this holds for every prime
ideal in R. Hence τ n [k, l] ∈ L, where L is the intersection of all prime ideals
and so is a nil ideal (see Theorem 4.21 of [59]) so there exists a positive
integer m1 such that (τ n [k, l])m1 = 0. Since τ is one-one & onto, we get
[k, l] = 0 for all k, l ∈ R a contradiction. Therefore, Ψn (a, b, c) ∈ P and so
Ψn (a, b, c) ∈ L. Thus (Ψn (a, b, c))m2 = 0, for some positive integer m2 . Since
the center of a semi-prime ring contains no non-zero nilpotent elements, this
yields that Ψn (a, b, c) = 0 for all a, b, c ∈ R.

Theorem 5.2.6 Let R be a 2-torsion free semi-prime ring. Then every gen-
eralized Jordan triple (σ, τ)-higher derivation of R is generalized (σ, τ)-higher
derivation of R.

Proof Let F = {fn }n∈N I be a generalized Jordan triple (σ, τ )-higher deriva-
tion of R. It can be easily seen that Ψ0(a, b) = 0. By induction assume that
Ψm (a, b) = 0 for all a, b ∈ R and each m < n. For a, b, x ∈ R, take ξ = abxab
and using Lemma 5.2.5, we get

85
 fn (ξ) = fn (a(bxa)b)

= fi (σ n−i (a))dj (σk τ i (bxa))dk (τ n−k (b))
i+j+k=n


= fi (σn−i (a))ds (τ n−s (b))(τ n (xab))
i+s=n

+ (σ n (abx))dq (σn−q (a))dk (τ n−k (b))
q+k=n

i+s,q+k≤n−1
+ fi (σn−i (a))ds (σ q+k τ i (b))(σ q+k τ i+s (x))
i+s+q+k=n
dq (σ k τ 
s+i
(a))dk (τ n−k (b))
+ fi (σ n−i (a))ds (σ1+q+k τ i (b))d1 (σq+k τ i+s (x))
i+s+q+k=n−1
dq (σ k τ 1+s+i (a))dk (τ n−k (b))
+ · · · + f1 (σn−1 (a))(σ n−1 τ(b))dn−1 (τ (x))(τ n (ab))
+σn (a)d1(σ n−1(b))dn−1 (τ (x))(τ n (ab))
+σn (ab)dn−1 (σ(x))d1 (τ n−1(a))(τ n (b))
+σn (ab)dn−1 (σ(x))(στ n−1 (a))d1(τ n−1 (b))
+σn (ab)dn (x)τ n (ab).

On the other hand, using the fact that Ψm (a, b) = 0 for all m < n, we obtain

fn (ξ) = fn ((ab)x(ab))

= fi (σn−i (ab))dj (σk τ i (x))dk (τ n−k (ab))
i+j+k=n

= σn (ab)σ n (x)dn (ab) + fn (ab)τ n (x)τ n (ab)


l+s,q+p≤n−1
+ fl (σ n−l (a))ds (τ l σq+p(b))(σ q+k τ l+s (x))
l+s+p+q=n
dq (σ τ  (a))dp (τ n−p (b))
p l+s

+ fl (σn−l (a))ds (τ l σ q+p+1 (b))d1 (σq+k τ l+s (x))

l+s+p+q=n−1
dq (σ p τ l+s+1(a))dp (τ n−p (b))
+ · · · + f1 (σ n−1 (a))(τσ n−1(b))dn−1 (τ (x))(τ n (ab))
+σn (a)d1(σ n−1 (b))dn−1 (τ (x))(τ n (ab))
+(σn (ab))dn−1 (σ(x))d1 (τ n−1(a))τ n (b)
+(σn (ab))dn−1 (σ(x))(στ n−1 (a))d1(τ n−1 (b))
+(σn (ab))dn (x)dk (τ n (ab)).

Comparing both the equations and reordering the indices, we have

86
 
0 = {fn (ab) − fi (σn−i (a))dj (τ n−j (b))}τ n (x)τ n (ab)
i+j=n
n n
= Ψn (a, b)τ (x)τ (ab) for all a, b ∈ R.

Using Lemma 4.2.4 we have Ψn (a, b)xτ n (wz) = 0 for all a, b, x, w, z ∈ R.

This yields that wτ −n (Ψn (a, b))xwτ −n (Ψn (a, b)) = 0 for all a, b, x, w ∈ R
and the semiprimeness of R gives wτ −n (Ψn (a, b)) = 0 for all a, b, w ∈ R.
Hence τ −n (Ψn (a, b))Rτ −n (Ψn (a, b)) = 0 for all a, b ∈ R. Finally we obtain
that Ψn (a, b) = 0 for all a, b ∈ R which completes our proof.

Following are the immediate corollaries of our result;

Corollary 5.2.7 ([26], Theorem 4.3) Every Jordan triple derivation of a

2-torsion free semi-prime ring is a derivation.

Corollary 5.2.8 [([38], Theorem 1.2) Let R be a 2-torsion free semi-prime

ring. Then every Jordan triple higher derivation on R is a higher derivation.

Corollary 5.2.9 ([74], Theorem 2.7) Let R be a 2-torsion free semi-prime

ring. Then every generalized Jordan triple higher derivation on R is a gen-
eralized higher derivation.

5.3 Generalized Jordan triple (σ, τ )-higher

derivations on Lie ideals of prime rings
In this section we define generalized Jordan triple (σ, τ )-higher derivation on
a Lie ideal U of R. A family of additive mappings F = {fn }n∈N I of R is
said to be a generalized Jordan triple (σ, τ )-higher derivation of U into R if
there exist a (σ, τ)-higherderivation D = {dn }n∈N I of U into R such that
f0 = IR and fn (aba) = fi (σ n−i (a))dj (σk τ i (b))dk (τ n−k (a)) holds for all
i+j+k=n
a, b ∈ U and every n ∈ N. I
It can be easily seen that on a 2-torsion free ring R, every generalized
(σ, τ)-higher derivation of a square closed Lie ideal U into R is a generalized
Jordan triple (σ, τ )-higher derivation of U into R. In fact, if F = {fn }n∈N
I is
generalized 
(σ, τ)-higher derivation of a square closed Lie ideal U into R then
fn (uv) = fi (σn−i (u))dj (τ n−j (v)) for all u, v ∈ U and for each n ∈ N.
I
i+j=n
Replacing v by
 2vu and using the fact that R is 2-torsion free, we obtain
fn (uvu) = fi (σn−i (u))dj (σ k τ i (v))dk (τ n−k (u)) for all u, v ∈ U and for
i+j+k=n
each n ∈ N I i.e., F = {fn }n∈NI is generalized Jordan triple (σ, τ )-higher
derivation of U into R but the converse need not be true in general.

87
 In the present section our objective is to find the conditions on R under
which every generalized Jordan triple (σ, τ)-higher derivation of U into R
becomes a generalized (σ, τ )-higher derivation of U into R. In fact our
result generalizes the main theorem obtained by Ferrero and Haetinger ([38],
Theorem 1.2).

For every fixed n ∈ N I and each u, v, w ∈ U we denote by [u, v, w],

Φn (u, v) and Φn (u, v, w) the elements of R as:

[u, v, w] = uvw − wvu,


Φn (u, v) = fn (uv) − fi (σn−i (u))dj (τ n−j (v)),
i+j=n

Φn (u, v, w) = fn (uvw) − fi (σn−i (u))dj (σ k τ i (v))dk (τ n−k (w)).
i+j+k=n

It can easily be seen that Φn (u, v) and Φn (u, v, w) are additive in each
argument. Also, Φn (u, v, w) = −Φn (w, v, u).

Suppose that σ, τ are endomorphisms of R such that τ is one-one &

onto and στ = τ σ.

We begin by stating the following lemma which can be obtained in[38];

Lemma 5.3.1 Let R be a prime ring of characteristic different from two

and U be anon-central Lie ideal of R. Then there exist elements u, v, w ∈ U
such that [u, v, w]
= 0.

Now we shall prove the following;

Lemma 5.3.2 Let R be a 2-torsion free ring and U be a square closed Lie
ideal of R. If F = {fn }n∈N I is generalized Jordan triple (σ, τ)-higher deriva-
tion of U into R and Φm (u, v, w) = 0 for all u, v, w ∈ U and every m < n,
then Φn (u, v, w)τ n (x)τ n [w, v, u] = 0 for all u, v, w, x ∈ U and every n ∈ N.
I

Proof Suppose ζ = 4(uvwxwvu + wvuxuvw). Then

fn (ζ) = fn (u(4vwxwv)u) + fn (w(4vuxuv)w)


= 4 fi (σ n−i (u))dj (σ k τ i (v(wxw)v))dk (τ n−k (u))
i+j+k=n

+4 fi (σn−i (w))dj (σk τ i (v(uxu)v))dk (τ n−k (w))
i+j+k=n

88
 
= 4 fi (σ n−i (u))dl (σ y+p+q+s+k τ i (v))dy (σp+q+s+k τ i+l (w))
i+l+y+p+q+s+k=n
dp (σq+s+k τ i+l+y (x))dq (σs+k τ i+l+y+p (w))ds (σk τ i+l+y+p+q (v))dk (τ n−k (u))

+4 fi (σ n−i (w))dl (σy+p+q+s+k τ i (v))dy (σ p+q+s+k τ i+l (u))
i+l+y+p+q+s+k=n
dp (σq+s+k τ i+l+y (x))dq (σs+k τ i+l+y+p (u))ds (σ k τ i+l+y+p+q (v))dk (τ n−k (w)).

(5.3.1)
From our hypothesis we have,

fn (uvu) = fi (σ n−i (u))dj (σ k τ i (v))dk (τ n−k (u)), for all u, v ∈ U, n ∈ N.
I
i+j+k=n
(5.3.2)
Linearizing the above equation we obtain,

fn (uvw + wvu) = fi (σ n−i (u))dj (σ k τ i (v))dk (τ n−k (w))
i+j+k=n

+ fi (σn−i (w))dj (σk τ i (v))dk (τ n−k (u))
i+j+k=n

for all u, v ∈ U, n ∈ N.
I
(5.3.3)
Again consider, fn (ζ) = fn ((2uvw)x(2wvu) + (2wvu)x(2uvw)). Using the
relation (5.3.3) we have,

fn (ζ) = 4 (fα (σ n−α (uvw))dp (σγ τ α (x))dγ (τ n−γ (wvu))
α+p+γ=n
+fα(σ n−α(wvu))dp (σγ τ α (x))dγ (τ n−γ (uvw)).

(5.3.4)
Combining the two equations (5.3.3) and (5.3.4)

0 = 4 fi (σ n−i (u))dl (σy+p+q+s+k τ i (v))dy (σ p+q+s+k τ i+l (w))
i+l+y+p+q+s+k=n
dp (σ q+s+k
 τ
i+l+y
(x))dq (σ s+k τ i+l+y+p(w))ds (σ k τ i+l+y+p+q (v))dk (τ n−k (u))
−4 fα (σn−α (uvw))dp (σ γ τ α(x))dγ (τ n−γ (wvu))
α+p+γ=n 
+4 fi (σ n−i (w))dl (σ y+p+q+s+k τ i (v))dy (σp+q+s+k τ i+l (u))
i+l+y+p+q+s+k=n
dp (σ q+s+k
 τ
i+l+y
(x))dq (σ s+k τ i+l+y+p(u))ds (σk τ i+l+y+p+q (v))dk (τ n−k (w))
−4 fα (σn−α (wvu))dp (σ γ τ α(x))dγ (τ n−γ (uvw))).
α+p+γ=n

(5.3.5)

89
Consider the first term,

fi (σ n−i (u))dl (σy+p+q+s+k τ i (v))dy (σ p+q+s+k τ i+l (w))
i+l+y+p+q+s+k=n
dp (σq+s+k τ i+l+y (x))dq (σs+k τ i+l+y+p (w))ds (σk τ i+l+y+p+q (v))dk (τ n−k (u))

= fi (σ n−i (u))dl (σy+q+s+k τ i (v))dy (σq+s+k τ i+l (w))
i+l+y+q+s+k=n
q+s+k i+l+y
(σ τ (x))dq (σs+k τ i+l+y (w))ds (σ k τ i+l+y+q (v))dk (τ n−k (u))

+ fi (σ n−i (u))dl (σy+1+q+s+k τ i (v))dy (σ 1+q+s+k τ i+l (w))
i+l+y+q+s+k=n−1
q+s+k i+l+y
d1 (σ τ (x))dq (σs+k τ i+l+y+1 (w))ds (σk τ i+l+y+1+q (v))dk (τ n−k (u))

+··· + fi (σn−i (u))dl (σ y+n−1+q+s+k τ i (v))dy (σ n−1+q+s+k τ i+l (w))
i+l+y+q+s+k=1
q+s+k i+l+y
dn−1 (σ τ (x))dq (σs+k τ i+l+y+n−1 (w))ds (σ k τ i+l+y+n−1+q (v))dk (τ n−k (u))

+ fi (σn−i (u))dl (σ y+n+q+s+k τ i (v))dy (σ n+q+s+k τ i+l (w))
i+l+y+q+s+k=0
dn (σ q+s+k τ i+l+y (x))dq (σs+k τ i+l+y+n (w))ds (σ k τ i+l+y+n+q (v))dk (τ n−k (u))

= fi (σ n−i (u))dl (σy τ i (v))dy (τ n−y (w))τ n (x)τ n (wvu)
i+l+y=n


+ σ n (uvw)σn (x)dq (σn−q (w))ds (σk τ q (v))dk (τ n−k (u))
q+s+k=n

0<i+l+y,q+s+k≤n−1

+ fi (σn−i (u))dl (σ y+q+s+k τ i (v))dy (σq+s+k τ i+l (w))
i+l+y+q+s+k=n
(σq+s+k τ i+l+y (x))dq (σs+k τ i+l+y (w))ds (σ k τ i+l+y+q (v))dk (τ n−k (u))


+ fi (σn−i (u))dl (σ y+q+s+k+1τ i (v))dy (τ i+l σq+s+k+1 (w))
i+l+y+q+s+k=n−1
d1 (σq+s+k τ i+l+y (x))dq (σ s+k τ i+l+y+1 (w))ds (σk τ i+l+y+q+1(v))dk (τ n−k (u))
+ · · · + f1 (σn−1 (u))(σn−1 τ(v))(σn−1 τ (w))dn−1 (τ(x))τ n (wvu)
+σn (u)d1(σ n−1 (v))(σn−1 τ (w))dn−1 (τ(x))τ n (wvu)
+σn (uv)d1 (σn−1 (w))dn−1 (τ (x))τ n (wvu)
+σn (uvw)dn−1 (σ(x))d1 (τ n−1(w))τ n (vu)
+σn (uvw)dn−1 (σ(x))(στ n−1 (w))d1 (τ n−1(v))τ n (u)
+σn (uvw)dn−1 (σ(x))(στ n−1 (w))(στ n−1 (v))d1 (τ n−1(u))
+σn (uvw)dn (x)τ n (wvu).

Calculating the second term;

90
 
fα (σ n−α (uvw))dp (σγ τ α (x))dγ (τ n−γ (wvu))
α+p+γ=n


= fα (σn−α (uvw))(σ γ τ α(x))dγ (τ n−γ (wvu))
α+γ=n
+ fα(σ n−α(uvw))d1 (σγ τ α (x))dγ (τ n−γ (wvu))
α+γ=n−1
+··· + fα (σn−α (uvw))dn−1 (σγ τ α (x))dγ (τ n−γ (wvu))
 α+γ=1
+ fα(σ n−α(uvw))dn (σγ τ α (x))dγ (τ n−γ (wvu))
α+γ=0

= fn (uvw)τ n (x)τ n (wvu) + σn (uvw)σ n (x)dn (wvu)

0<α,γ≤n−1

+ fα (σn−α (uvw))σγ τ α (x)dγ (τ n−γ (wvu))
α+γ=n

+ fα (σ n−α (uvw))d1(σ γ τ α (x))dγ (τ n−γ (wvu))
α+γ=n−1
+ · · · + f1 (σn−1 (uvw))dn−1 (τ (x))τ n (wvu)
+σn (uvw)dn−1 (σ(x))d1 (τ n−1 (wvu)) + σn (uvw)dn (x)τ n (wvu).

Using the hypothesis that Φm (u, v, w) = 0 for all m < n.

= fn (uvw)τ n (x)τ n (wvu) + σn (uvw)σ n (x)dn (wvu)

0<i+l+y,q+s+k≤n−1

+ fi (σn−i (u))dl (σ y+q+s+k τ i (v))dy (σq+s+k τ i+l (w))
i+l+y+q+s+k=n
σq+s+k τ i+l+y s+k i+l+y
 (x)dq (σ n−iτ (w))ds (σk τ i+l+y+q (v))dk (τ n−k (u))
+ fi (σ (u))dl (σy+q+s+k+1 τ i (v))dy (σ q+s+k+1τ i+l (w))
i+l+y+q+s+k=n−1
d1 (σq+s+k i+l+y
 τ 1−i(x))d q (σ
s+k i+l+y+1
τ (w))ds (σk τ i+l+y+q+1(v))dk (τ n−k (u)) + · · ·
+( fi (σ σ n−1(u))dl (σ y τ i σ n−1(v))dy (τ 1−y σ n−1(w)))dn−1(τ (x))τ n (wvu))
i+l+y=1

+σn (uvw)dn−1 (σ(x))( dq (σ 1−q τ n−1 (w))du (σk τ q τ n−1 (v))dk (σ1−k τ n−1 (u)))
q+u+k=1
+σn (uvw)dn (x)τ n (wvu)

= fn (uvw)τ n (x)τ n (wvu) + σn (uvw)σ n (x)dn (wvu)

0<i+l+y,q+s+k≤n−1

+ fi (σn−i (u))dl (σ y+q+s+k τ i (v))dy (σq+s+k τ i+l (w))
i+l+y+q+s+k=n
σq+s+k τ i+l+y s+k i+l+y
 (x)dq (σ n−iτ (w))ds (σk τ i+l+y+q (v))dk (τ n−k (u))
+ fi (σ (u))dl (σy+q+s+k+1 τ i (v))dy (τ i+l σ q+s+k+1(w))
i+l+y+q+s+k=n−1
d1 (σq+s+k τ i+l+y (x))dq (σs+k τ i+l+y+1 (w))ds (σk τ i+l+y+q+1(v))dk (τ n−k (u))

91
 + · · · + σ n (uvw)dn−1(σ(x))d1 (τ n−1 (w))τ n (vu)
+σn (uvw)dn−1 (σ(x))(στ n−1 (w))d1 (τ n−1(v))τ n (u)
+σn (uvw)dn−1 (σ(x))(στ n−1 (w))(στ n−1 (v))d1 (τ n−1(u))
+f1(σ n−1 (u))(σ n−1τ (v))(τ σn−1 (w))dn−1 (τ (x))τ n (wvu)
+σn (u)d1(σ n−1 (v))(σn−1 τ (w))dn−1 (τ(x))τ n (wvu)
+σn (uv)d1 (σn−1 (w))dn−1 (τ (x))τ n (wvu)
+σn (uvw)dn (x)τ n (wvu).
Now, substracting the two terms so obtained we find that

fn (uvw)τ n (x)τ n (wvu)− fi (σn−i (u))dl (σ s τ i (v))ds (τ n−s (w))τ n (x)τ n (wvu)
 i+l+s=n
= {fn (uvw) − fi (σ n−i (u))dl (σs τ i (v))ds (τ n−s (w))}τ n (x)τ n (wvu)
i+l+s=n
= Φn (u, v, w)τ n (x)τ n (wvu).

Similarly, the difference of the last two terms yields


fi (σ n−i (w))dl (σ y+p+q+s+k τ i (v))dy (σ p+q+s+k τ i+l (u))
i+l+y+p+q+s+k=n
q+s+k i+l+s
dp (σ  τ (x))dq (σs+k τ i+l+y+p (u))ds (σ k τ i+l+y+p+q (v))dk (τ n−k (w))
n−α
− fα (σ (wvu))dp (σγ τ α (x))dγ (τ n−γ (uvw))
α+p+γ=n

= Φn (w, v, u)τ n (x)τ n (uvw).

Thus the equation (5.3.4) reduces to 4(Φn (u, v, w)τ n (x)τ n (wvu) +
Φn (w, v, u)τ n (x)τ n (uvw) = 0, for all u, v, w, x ∈ U. Since characteristic of R
is not two we find that Φn (u, v, w)τ n (x)τ n [w, v, u] = 0, for all u, v, w, x ∈ U .

Lemma 5.3.3 Let R be a prime ring of characteristic different from 2 and U

be a non-central square closed Lie ideal of R. If F = {fn }n∈N
I is generalized
Jordan triple (σ, τ )-higher derivation on U into R. Then Φn (u, v, w) = 0,
for all u, v, w ∈ U and each n ∈ N.
I

Proof Trivially for n = 0, Φ0 (u, v, w) = 0. By induction let us assume that

Φm (u, v, w) = 0, for every u, v, w ∈ U and for all m < n. By Lemma 5.3.2, we
have Φn (u, v, w)τ n (x)τ n [w, v, u] = 0 or τ −n (Φn (u, v, w))x[w, v, u] = 0. Again,
by Lemma 4.3.5, we have Φn (u, v, w) = 0 for all u, v, w ∈ U or [w, v, u] = 0 for
all u, v, w ∈ U. If [w, v, u] = 0 for all u, v, w ∈ U holds then this contradicts
Lemma 5.3.1. Hence, in either case Φn (u, v, w) = 0 for all u, v, w ∈ U, n ∈ N. I

Theorem 5.3.4 Let R be a prime ring of characteristic different from 2

and U be a non-central square closed Lie ideal of R. Then every generalized
Jordan triple (σ, τ)-higher derivation of U into R is a generalized (σ, τ)-higher
derivation of U into R.

92
Proof Let F = {fn }n∈N I be generalized Jordan triple (σ, τ )-higher deriva-
tion. It can
 be easily seen that f0 = IR . By induction assume that
fm (uv) = fi (σn−i (u))dj (τ n−j (v)) holds for all m < n.
i+j=m

For u, v, x ∈ U, take ξ = 4uvxuv ∈ U and use Lemma 5.3.3, to obtain

fn (ξ) = 4fn (u(vxu)v)


= fi (σ n−i (u))dj (σ k τ i (vxu))dk (τ n−k (v))
i+j+k=n


= 4 fi (σn−i (u))ds (σp+q+k τ i (v))dp (σ q+k τ i+s (x))
i+s+p+q+k=n
dq (σ k τ s+p+i (u))dk (τ n−k (v))

= 4( fi (σ n−i (u))ds (σ q+k τ i (v))(σ q+k τ i+s (x))
i+s+q+k=n
dq (σ τ  (u))dk (τ n−k (v))
k s+i

+ fi (σ n−i (u))ds (σ 1+q+k τ i (v))d1 (σ q+k τ i+s (x))

i+s+q+k=n−1
dq (σ k τ 1+s+i
(u))dk (τ n−k (v))
+··· + fi (σn−i (u))ds (σ n−1+q+k τ i (v))dn−1 (σq+k τ i+s (x))
i+s+q+k=1
dq (σ k τ
s+n−1+i
(u))dk (τ n−k (v))
+ fi (σ n−i (u))ds (σ n+q+k τ i (v))dn (σ q+k τ i+s (x))
i+s+q+k=0
dq (σ k τ n+s+i (u))dk (τ n−k (v)))


= 4( fi (σ n−i (u))ds (τ n−s (v))(τ n (xuv))
i+s=n

+ (σ n (uvx))dq (σn−q (u))dk (τ n−k (v))
q+k=n
0<i+s,q+k≤n−1

+ fi (σn−i (u))ds (σ q+k τ i (v))(σq+k τ i+s (x))
i+s+q+k=n
dq (σ k τ 
s+i
(u))dk (τ n−k (v))
+ fi (σ n−i (u))ds (σ 1+q+k τ i (v))d1 (σ q+k τ i+s (x))
i+s+q+k=n−1
dq (σ k τ 1+s+i (u))dk (τ n−k (v))
+ · · · + f1 (σn−1 (u))(σn−1 τ(v))dn−1 (τ(x))(τ n (uv))
+σn (u)d1(σ n−1(v))dn−1 (τ(x))(τ n (uv))
+σn (uv)dn−1 (σ(x))d1 (τ n−1 (u))(τ n (v))
+σn (uv)dn−1 (σ(x))(στ n−1 (u))d1(τ n−1 (v))
+σn (uv)dn (x)τ n (uv)).

93
On the other hand,

fn (ξ) = 4fn ((uv)x(uv))


= 4 fi (σn−i (uv))dj (σk τ i (x))dk (τ n−k (uv))
i+j+k=n


= 4( fi (σn−i (uv))d0(σ k τ i (x))dk (τ n−k (uv))
i+k=n
 n−i
+ fi (σ (uv))d1(σ k τ i (x))dk (τ n−k (uv))
i+k=n−1
+··· + fi (σ n−i (uv))dn−1 (σk τ i (x))dk (τ n−k (uv))
 i+k=1n−i
+ fi (σ (uv))dn (σ k τ i (x))dk (τ n−k (uv))
i+k=0

= 4(σn (uv)σn (x)dn (uv) + fn (uv)τ n (x)τ n (uv)

0<i,k≤n−1

+ fi (σ n−i (uv))(σk τ i (x))dk (τ n−k (uv))
i+k=n

+ fi (σn−i (uv))d1(σ k τ i (x))dk (τ n−k (uv))
i+k=n−1
+ · · · + f1 (σn−i (uv))dn−1 (τ (x))(τ n (uv))
+(σ n (uv))dn−1(σ(x))d1 (τ n−1 (uv))
+(σ n (uv))dn (x)τ n (uv))

= 4(σn (uv)σn (x)dn (uv) + fn (uv)τ n (x)τ n (uv)

0<l+s,p+q≤n−1

+ fl (σ n−l (u))ds (τ l σq+p (v))(σ q+k τ l+s (x))
l+s+p+q=n
dq (σ τ  (u))dp (τ n−p (v))
p l+s

+ fl (σ n−l (u))ds (τ l σq+p+1 (v))d1 (σq+k τ l+s (x))

l+s+p+q=n−1
dq (σp τ l+s+1 (u))dp (τ n−p (v))
+ · · · + f1 (σn−1 (u))(σ n−1 τ(v))dn−1(τ (x))(τ n (uv))
+σ n (u)d1(σ n−1(v))dn−1(τ (x))(τ n (uv))
+(σ n (uv))dn−1(σ(x))d1 (τ n−1 (u))τ n (v)
+(σ n (uv))dn−1(σ(x))(στ n−1(u))d1 (τ n−1(v))
+(σ n (uv))dn (x)dk (τ n (uv))).

Comparing both the equations, reordering the indices and using the fact that
characteristic of R is different from two, we finally obtain

0 = {fn (uv) − fi (σ n−i (u))dj (τ n−j (v))}τ n (x)τ n (uv)
i+j=n
= Φn (u, v)τ n (x)τ n (uv) for all u, v ∈ U.

Since τ is one-one & onto implementing Lemma 4.3.5, it can be easily seen
that Φn (u, v) = 0 for all u, v ∈ U or u′v ′ = 0 for all u′, v ′ ∈ U. Now if

94
u′v ′ = 0 then for any r ∈ R, u′ (v′ r − rv ′) = 0. This yields that u′ Rv′ = {0}
and the primeness of R implies that U = {0} a contradiction. Therefore,
Φn (u, v) = 0 for all u, v ∈ U. This completes the proof of our theorem.

For σ = τ = IR and fn = dn for each n ∈ N I our result reduces to a

theorem given by Fererro & Haetinger for the case of prime rings:

Corollary 5.3.5 ([38], Theorem 1.2) Let R be a prime ring of characteristic

different from two and U be a square closed non-central Lie ideal of R. Then
every Jordan triple higher derivation of U into R is a higher derivation of U
into R.

95