Abstract

Let ๐‘ be a 3-prime 2-torsion-free zero-symmetric left near-ring with multiplicative center ๐‘. We prove that if ๐‘ admits a nonzero generalized derivation ๐‘“ such that ๐‘“(๐‘)โŠ†๐‘, then ๐‘ is a commutative ring. We also discuss some related properties.

1. Introduction

Let ๐‘ be a zero-symmetric left near-ring, not necessarily with a multiplicative identity element; and let ๐‘ be its multiplicative center. Define ๐‘ to be 3-prime if for all ๐‘Ž,๐‘โˆˆ๐‘โงต{0}, ๐‘Ž๐‘๐‘โ‰ {0}; and call ๐‘ 2-torsion-free if (๐‘,+) has no elements of order 2. A derivation on ๐‘ is an additive endomorphism ๐ท of ๐‘ such that ๐ท(๐‘ฅ๐‘ฆ)=๐‘ฅ๐ท(๐‘ฆ)+๐ท(๐‘ฅ)๐‘ฆ for all ๐‘ฅ,๐‘ฆโˆˆ๐‘. A generalized derivation ๐‘“ with associated derivation ๐ท is an additive endomorphism ๐‘“โˆถ๐‘โ†’๐‘ such that ๐‘“(๐‘ฅ๐‘ฆ)=๐‘“(๐‘ฅ)๐‘ฆ+๐‘ฅ๐ท(๐‘ฆ) for all ๐‘ฅ,๐‘ฆโˆˆ๐‘. In the case of rings, generalized derivations have received significant attention in recent years.

In [1], we proved the following.

Theorem A. If N is 3-prime and 2-torsion-free and D is a derivation such that ๐ท2=0, then ๐ท=0.

Theorem B. If N is a 3-prime 2-torsion-free near-ring which admits a nonzero derivation D for which ๐ท(๐‘)โŠ†๐‘, then N is a commutative ring.

Theorem C. If N is a 3-prime 2-torsion-free near-ring admitting a nonzero derivation D such that ๐ท(๐‘ฅ)๐ท(๐‘ฆ)=๐ท(๐‘ฆ)๐ท(๐‘ฅ) for all ๐‘ฅ,๐‘ฆโˆˆ๐‘, then N is a commutative ring.

In this paper, we investigate possible analogs of these results, where ๐ท is replaced by a generalized derivation ๐‘“.

We will need three easy lemmas.

Lemma 1.1 (see [1, Lemma 3]). Let N be a 3-prime near-ring.
(i) If๐‘งโˆˆ๐‘โงต{0}, then z is not a zero divisor. (ii)If ๐‘โงต{0} contains an element z such that ๐‘ง+๐‘งโˆˆ๐‘, then (๐‘,+) is abelian. (iii)If D is a nonzero derivation and ๐‘ฅโˆˆ๐‘ is such that๐‘ฅ๐ท(๐‘)={0} or ๐ท(๐‘)๐‘ฅ={0}, then ๐‘ฅ=0.

Lemma 1.2 (see [2, Proposition 1]). If N is an arbitrary near-ring and D is a derivation on N, then D(xy) = D(x)y + xD(y) for all ๐‘ฅ,๐‘ฆโˆˆ๐‘.

Lemma 1.3. Let N be an arbitrary near-ring and let f be a generalized derivation on N with associated derivation D. Then
๎‚€๎‚๐‘“(๐‘Ž)๐‘+๐‘Ž๐ท(๐‘)๐‘=๐‘“(๐‘Ž)๐‘๐‘+๐‘Ž๐ท(๐‘)๐‘โˆ€๐‘Ž,๐‘,๐‘โˆˆ๐‘.(1.1)

Proof. Clearly ๐‘“((๐‘Ž๐‘)๐‘)=๐‘“(๐‘Ž๐‘)๐‘+๐‘Ž๐‘๐ท(๐‘)=(๐‘“(๐‘Ž)๐‘+๐‘Ž๐ท(๐‘))๐‘+๐‘Ž๐‘๐ท(๐‘); and by using Lemma 1.2, we obtain ๐‘“(๐‘Ž(๐‘๐‘))=๐‘“(๐‘Ž)๐‘๐‘+๐‘Ž๐ท(๐‘๐‘)=๐‘“(๐‘Ž)๐‘๐‘+๐‘Ž๐ท(๐‘)๐‘+๐‘Ž๐‘๐ท(๐‘).
Comparing these two expressions for ๐‘“(๐‘Ž๐‘๐‘) gives the desired conclusion.

2. The Main Theorem

Our best result is an extension of Theorem B.

Theorem 2.1. Let N be a 3-prime 2-torsion-free near-ring. If N admits a nonzero generalized derivation f such that ๐‘“(๐‘)โŠ†๐‘, then N is a commutative ring.

In the proof of this theorem, as well as in a later proof, we make use of a further lemma.

Lemma 2.2. Let R be a 3-prime near-ring, and let f be a generalized derivation with associated derivation ๐ทโ‰ 0. If ๐ท(๐‘“(๐‘))={0}, then ๐‘“(๐ท(๐‘))={0}.

Proof. We are assuming that ๐ท(๐‘“(๐‘ฅ))=0 for all ๐‘ฅโˆˆ๐‘. It follows that ๐ท(๐‘“(๐‘ฅ๐‘ฆ))=๐ท(๐‘“(๐‘ฅ)๐‘ฆ)+๐ท(๐‘ฅ๐ท(๐‘ฆ))=0 for all ๐‘ฅ,๐‘ฆโˆˆ๐‘, that is,
๐‘“(๐‘ฅ)๐ท(๐‘ฆ)+๐ท(๐‘ฅ)๐ท(๐‘ฆ)+๐‘ฅ๐ท2(๐‘ฆ)=0โˆ€๐‘ฅ,๐‘ฆโˆˆ๐‘.(2.1) Applying ๐ท again, we get
๐‘“(๐‘ฅ)๐ท2(๐‘ฆ)+๐ท2(๐‘ฅ)๐ท(๐‘ฆ)+๐ท(๐‘ฅ)๐ท2(๐‘ฆ)+๐ท(๐‘ฅ)๐ท2(๐‘ฆ)+๐‘ฅ๐ท3(๐‘ฆ)=0โˆ€๐‘ฅ,๐‘ฆโˆˆ๐‘.(2.2) Taking ๐ท(๐‘ฆ) instead of ๐‘ฆ in (2.1) gives ๐‘“(๐‘ฅ)๐ท2(๐‘ฆ)+๐ท(๐‘ฅ)๐ท2(๐‘ฆ)+๐‘ฅ๐ท3(๐‘ฆ)=0, hence (2.2) yields
๐ท2(๐‘ฅ)๐ท(๐‘ฆ)+๐ท(๐‘ฅ)๐ท2(๐‘ฆ)=0โˆ€๐‘ฅ,๐‘ฆโˆˆ๐‘.(2.3) Now, substitute ๐ท(๐‘ฅ) for ๐‘ฅ in (2.1), obtaining ๐‘“(๐ท(๐‘ฅ))๐ท(๐‘ฆ)+๐ท2(๐‘ฅ)๐ท(๐‘ฆ)+๐ท(๐‘ฅ)๐ท2(๐‘ฆ)=0; and use (2.3) to conclude that ๐‘“(๐ท(๐‘ฅ))๐ท(๐‘ฆ)=0 for all ๐‘ฅ,๐‘ฆโˆˆ๐‘. Thus, by Lemma 1.1(iii), ๐‘“(๐ท(๐‘ฅ))=0 for all ๐‘ฅโˆˆ๐‘.

Proof of Theorem 2.1. Since ๐‘“โ‰ 0, there exists ๐‘ฅโˆˆ๐‘ such that0โ‰ ๐‘“(๐‘ฅ)โˆˆ๐‘. Since ๐‘“(๐‘ฅ)+๐‘“(๐‘ฅ)=๐‘“(๐‘ฅ+๐‘ฅ)โˆˆ๐‘,(๐‘,+) is abelian by Lemma 1.1(ii). To complete the proof, we show that ๐‘ is multiplicatively commutative.
First, consider the case ๐ท = 0, so that ๐‘“(๐‘ฅ๐‘ฆ)=๐‘“(๐‘ฅ)๐‘ฆโˆˆ๐‘ for all ๐‘ฅ,๐‘ฆโˆˆ๐‘. Then ๐‘“(๐‘ฅ)๐‘ฆ๐‘ค=๐‘ค๐‘“(๐‘ฅ)๐‘ฆ, hence ๐‘“(๐‘ฅ)(๐‘ฆ๐‘คโˆ’๐‘ค๐‘ฆ)=0 for all ๐‘ฅ,๐‘ฆ,๐‘คโˆˆ๐‘. Choosing ๐‘ฅ such that ๐‘“(๐‘ฅ)โ‰ 0 and invoking Lemma 1.1(i), we get ๐‘ฆ๐‘ค โˆ’ ๐‘ค๐‘ฆ = 0 for all ๐‘ฆ,๐‘คโˆˆ๐‘.
Now assume that ๐ทโ‰ 0, and let ๐‘โˆˆ๐‘โงต{0}. Then ๐‘“(๐‘ฅ๐‘)=๐‘“(๐‘ฅ)๐‘+๐‘ฅ๐ท(๐‘)โˆˆ๐‘; therefore, (๐‘“(๐‘ฅ)๐‘+๐‘ฅ๐ท(๐‘))๐‘ฆ=๐‘ฆ(๐‘“(๐‘ฅ)๐‘+๐‘ฅ๐ท(๐‘)) for all ๐‘ฅ,๐‘ฆโˆˆ๐‘, and by Lemma 1.3, we see that ๐‘“(๐‘ฅ)๐‘๐‘ฆ+๐‘ฅ๐ท(๐‘)๐‘ฆ=๐‘ฆ๐‘“(๐‘ฅ)๐‘+๐‘ฆ๐‘ฅ๐ท(๐‘). Since both ๐‘“(๐‘ฅ) and ๐ท(๐‘) are in ๐‘, we have ๐ท(๐‘)(๐‘ฅ๐‘ฆโˆ’๐‘ฆ๐‘ฅ)=0 for all ๐‘ฅ,๐‘ฆโˆˆ๐‘, and provided that ๐ท(๐‘)โ‰ {0}, we can conclude that ๐‘ is commutative.
Assume now that ๐ทโ‰ 0 and ๐ท(๐‘)={0}. In particular, ๐ท(๐‘“(๐‘ฅ))=0 for all ๐‘ฅโˆˆ๐‘. Note that for ๐‘โˆˆ๐‘ such that ๐‘“(๐‘)=0, ๐‘“(๐‘๐‘ฅ)=๐‘๐ท(๐‘ฅ)โˆˆ๐‘; hence by Lemma 2.2, ๐ท(๐‘ฅ)๐ท(๐‘ฆ)โˆˆ๐‘ and ๐ท(๐‘ฆ)๐ท(๐‘ฅ)โˆˆ๐‘ for each ๐‘ฅ,๐‘ฆโˆˆ๐‘. If one of these is 0, the other is a central element squaring to 0, hence is also 0. The remaining possibility is that ๐ท(๐‘ฅ)๐ท(๐‘ฆ) and ๐ท(๐‘ฆ)๐ท(๐‘ฅ) are nonzero central elements, in which case ๐ท(๐‘ฅ) is not a zero divisor. Thus ๐ท(๐‘ฅ)๐ท(๐‘ฅ)๐ท(๐‘ฆ)=๐ท(๐‘ฅ)๐ท(๐‘ฆ)๐ท(๐‘ฅ) yields ๐ท(๐‘ฅ)(๐ท(๐‘ฅ)๐ท(๐‘ฆ)โˆ’๐ท(๐‘ฆ)๐ท(๐‘ฅ))=0=๐ท(๐‘ฅ)๐ท(๐‘ฆ)โˆ’๐ท(๐‘ฆ)๐ท(๐‘ฅ). Consequently, ๐‘ is commutative by Theorem C.

3. On Theorems A and C

Theorem C does not extend to generalized derivations, even if ๐‘ is a ring. As in [3], consider the ring ๐ป of real quaternions, and define ๐‘“โˆถ๐ปโ†’๐ป by ๐‘“(๐‘ฅ)=๐‘–๐‘ฅ+๐‘ฅ๐‘–. It is easy to check that ๐‘“ is a generalized derivation with associated derivation given by ๐ท(๐‘ฅ)=๐‘ฅ๐‘–โˆ’๐‘–๐‘ฅ, and that ๐‘“(๐‘ฅ)๐‘“(๐‘ฆ)=๐‘“(๐‘ฆ)๐‘“(๐‘ฅ) for all ๐‘ฅ,๐‘ฆโˆˆ๐ป.

Theorem A also does not extend to generalized derivations, as we see by letting ๐‘ be the ring ๐‘€2(๐น) of 2 ร— 2 matrices over a field ๐น and letting ๐‘“ be defined by ๐‘“(๐‘ฅ)=๐‘’12๐‘ฅ. However, we do have the following results.

Theorem 3.1. Let N be a 3-prime near-ring, and let f be a generalized derivation on N with associated derivation D. If ๐‘“2=0, then๐ท3=0. Moreover, if N is 2-torsion-free, then ๐ท(๐‘)={0}.

Proof. We have ๐‘“2(๐‘ฅ๐‘ฆ)=๐‘“(๐‘“(๐‘ฅ)๐‘ฆ+๐‘ฅ๐ท(๐‘ฆ))=๐‘“(๐‘ฅ)๐ท(๐‘ฆ)+๐‘“(๐‘ฅ)๐ท(๐‘ฆ)+๐‘ฅ๐ท2(๐‘ฆ)=0โˆ€๐‘ฅ,๐‘ฆโˆˆ๐‘.(3.1) Applying ๐‘“ to (3.1) gives
๐‘“(๐‘ฅ)๐ท2(๐‘ฆ)+๐‘“(๐‘ฅ)๐ท2(๐‘ฆ)+๐‘“(๐‘ฅ)๐ท2(๐‘ฆ)+๐‘ฅ๐ท3(๐‘ฆ)=0โˆ€๐‘ฅ,๐‘ฆโˆˆ๐‘.(3.2) Substituting ๐ท(๐‘ฆ) for ๐‘ฆ in (3.1) gives
๐‘“(๐‘ฅ)๐ท2(๐‘ฆ)+๐‘“(๐‘ฅ)๐ท2(๐‘ฆ)+๐‘ฅ๐ท3(๐‘ฆ)=0;(3.3) Therefore, by (3.2) and (3.3),
๐‘“(๐‘ฅ)๐ท2(๐‘ฆ)=0โˆ€๐‘ฅ,๐‘ฆโˆˆ๐‘.(3.4) It now follows from (3.3) that ๐‘ฅ๐ท3(๐‘ฆ)=0 for all ๐‘ฅ,๐‘ฆโˆˆ๐‘; and since ๐‘ is 3-prime, ๐ท3=0.
Suppose now that ๐‘ is 2-torsion-free and that ๐ท(๐‘)โ‰ {0}, and let ๐‘งโˆˆ๐‘ be such that ๐ท(๐‘ง)โ‰ 0. Then if ๐‘ฅ,๐‘ฆโˆˆ๐‘ and ๐‘“(๐‘)๐‘ฅ={0}, then ๐‘“(๐‘ฆ๐‘ง)๐‘ฅ=๐‘“(๐‘ฆ)๐‘ง๐‘ฅ+๐‘ฆ๐ท(๐‘ง)๐‘ฅ=0=๐‘ฆ๐ท(๐‘ง)๐‘ฅ; and since ๐‘ is 3-prime and ๐ท(๐‘ง) is not a zero divisor, ๐‘ฅ = 0. It now follows from (3.4) that ๐ท2=0 and hence by Theorem A that ๐ท = 0. But this contradicts our assumption that ๐ท(๐‘)โ‰ {0}, hence ๐ท(๐‘)={0} as claimed.

Theorem 3.2. Let N be a 3-prime and 2-torsion-free near-ring with 1. If f is a generalized derivation on N such that ๐‘“2=0 and ๐‘“(1)โˆˆ๐‘, then ๐‘“=0.

Proof. Note that ๐‘“(๐‘ฅ)=๐‘“(1๐‘ฅ)=๐‘“(1)๐‘ฅ+1๐ท(๐‘ฅ), so
๐‘“(๐‘ฅ)=๐‘๐‘ฅ+๐ท(๐‘ฅ),๐‘โˆˆ๐‘.(3.5) If ๐‘ = 0, then ๐‘“=๐ท and ๐ท2=0, so ๐ท = 0 by Theorem A and therefore ๐‘“ = 0.
If ๐‘โ‰ 0, then ๐‘ is not a zero divisor, hence by (3.4) ๐ท2=0 and ๐ท = 0. But then ๐‘“(๐‘ฅ)=๐‘๐‘ฅ and ๐‘“2(๐‘ฅ)=๐‘2๐‘ฅ=0 for all ๐‘ฅโˆˆ๐‘. Since ๐‘2 is not a zero divisor, we get ๐‘={0}โ€”a contradiction. Thus, ๐‘ = 0 and we are finished.

4. More on Theorem C

In [4], the author studied generalized derivations ๐‘“ with associated derivation ๐ท which have the additional property that

๐‘“(๐‘ฅ๐‘ฆ)=๐ท(๐‘ฅ)๐‘ฆ+๐‘ฅ๐‘“(๐‘ฆ)โˆ€๐‘ฅ,๐‘ฆโˆˆ๐‘.(โˆ—) Our final theorem, a weak generalization of Theorem C, was stated in [4]; but the proof given was not correct. (At one point, both left and right distributivity were assumed.) We now have all the results required for a proof.

Theorem 4.1. Let N be a 3-prime 2-torsion-free near-ring which admits a generalized derivation ๐‘“ with nonzero associated derivation D such that f satisfies (โˆ—). If ๐‘“(๐‘ฅ)๐‘“(๐‘ฆ)=๐‘“(๐‘ฆ)๐‘“(๐‘ฅ) for all ๐‘ฅ,๐‘ฆโˆˆ๐‘, then N is a commutative ring.

Proof. It is correctly shown in [4] that (๐‘,+) is abelian and either ๐‘“(๐‘)โŠ†๐‘ or ๐ท(๐‘“(๐‘))={0}. Hence, in view of Theorem 2.1, we may assume that ๐ท(๐‘“(๐‘))=0 and therefore, by Lemma 2.2, that ๐‘“(๐ท(๐‘))={0}. We calculate ๐‘“(๐ท(๐‘ฅ)๐ท(๐‘ฆ)) in two ways. Using the defining property of ๐‘“, we obtain ๐‘“(๐ท(๐‘ฅ)๐ท(๐‘ฆ))=๐‘“(๐ท(๐‘ฅ))๐ท(๐‘ฆ)+๐ท(๐‘ฅ)๐ท2(๐‘ฆ)=๐ท(๐‘ฅ)๐ท2(๐‘ฆ); and using (โˆ—), we obtain ๐‘“(๐ท(๐‘ฅ)๐ท(๐‘ฆ))=๐ท2(๐‘ฅ)๐ท(๐‘ฆ)+๐ท(๐‘ฅ)๐‘“(๐ท(๐‘ฆ))=๐ท2(๐‘ฅ)๐ท(๐‘ฆ). Thus, ๐ท2(๐‘ฅ)๐ท(๐‘ฆ)=๐ท(๐‘ฅ)๐ท2(๐‘ฆ) for all ๐‘ฅ,๐‘ฆโˆˆ๐‘. But since ๐ท(๐‘“(๐‘))={0}, (2.3) holds in this case as well; therefore ๐ท2(๐‘ฅ)๐ท(๐‘ฆ)=0 for all ๐‘ฅ,๐‘ฆโˆˆ๐‘, hence by Lemma 1.1(iii) ๐ท2=0. Thus, ๐ท = 0, contrary to our original hypothesis, so that the case ๐ท(๐‘“(๐‘))={0} does not in fact occur.

Acknowledgment

This research is supported by the Natural Sciences and Engineering Research Council of Canada, Grant no. 3961.