Indian J. pure appl. Math., 39(6): 459-465, December 2008 °c Printed in India.

ANNIHILATOR CONDITIONS WITH DERIVATIONS IN PRIME RINGS OF 2

YU WANG

College of Mathematics, Jilin Normal University, Siping, Jilin 136000, Peoples’ Republic of China e-mail: [email protected]

(Received 1 August 2006; after final revision 11 August 2008; accepted 22 August 2008)

Let R be a prime of characteristic 2 with d a nonzero derivation of R,L a noncentral Lie of R, a ∈ R such that a[d(u), u]n is central, for all u ∈ L, where n is a fixed positive

. Then a = 0 unless R satisfies S4, the standard identity in 4 variables.

Keywords:Primering;derivation;generalizedpolynomialidentity;centraldifferential identity

Let R be a prime ring with center Z. For x, y ∈ R, we set [x, y] = xy − yx and [x, y]2 = [[x, y], y]. 2 2 Note that [x, y]2 = xy + y x when char R = 2. Let n be a fixed positive integer. We denote by d a nonzero derivation of R. As usual, we denote by S4 the standard identity in 4 variables. A well known result of Posner [15] states that if [d(x), x] ∈ Z, for all x ∈ R, then R is commutative. Posner’s theorem was extended to the case of Lie ideals of prime rings by Lee and Lee [13] and Lanski [10], respectively. Carini and De Filippis [2] studied a more generalized situation when [d(u), u]n ∈ Z for all u in a noncentral Lie ideal of R of characteristic 6= 2. Furthermore, De Filippis [6] investigated the left annihilator of power values of commutators in prime rings of characteristic 6= 2 and obtained the following:

Theorem A [6] — Let R be a prime ring of characteristic different from 2, d a nonzero derivation of R, L a noncentral Lie ideal of R, n ≥ 1 a fixed integer and a ∈ R. If a[d(u), u]n ∈ Z for all u ∈ L, then either a = 0 or R satisfies S4, the standard identity in 4 variables. 460 YU WANG

The main purpose of this paper is to remove the restriction of characteristic in Theorem A. Explicitly we prove the following:

Theorem 1 — Let R be a prime ring of characteristic 2, Z the center of R, d a nonzero derivation of R, L a noncentral Lie ideal of R and a ∈ R. If a[d(u), u]n ∈ Z for all u ∈ L, then either a = 0 or R satisfies S4, the standard identity in 4 variables.

As a direct consequence of the above two results we have the following general result.

Theorem 2 — Let R be a prime ring, Z the center of R, d a nonzero derivation of R, L a noncentral Lie ideal of R and a ∈ R. If a[d(u), u]n ∈ Z for all u ∈ L, then either a = 0 or R satisfies S4, the standard identity in 4 variables.

In all that follows, R will be a prime ring with its center Z, its extended centroid C and its two-sided Martindale quotient ring Q. All these notions are explained in detail in a book of Beidar et al. [1]. It is well known that every derivation d of R can be uniquely extended to a derivation of Q, which will be also denoted by d. Recall that a derivation d of R is called Q-inner induced by some element b ∈ Q if d(x) = [b, x] for all x ∈ R. Otherwise, d is called Q-outer. Let

T = Q ∗C C{X,Y } be the free product over C of C-algebra Q and C{X,Y }, the free C-algebra in two noncommuting indeterminates X and Y .

In order to prove our main result we need some lemmas as follows.

Lemma 1 — Let R be a prime ring of characteristic 2 and a, b ∈ Q with b ∈ C such that n a([b, [x, y]]2) = 0 for all x, y ∈ R, where n ≥ 0 is a fixed integer. If R does not satisfy any nontrivial generalized polynomial identity, then a = 0.

n PROOF : By a theorem due to Chuang [4, Theorem 2] we have a([b, [x, y]]2) = 0 for all x, y ∈ Q. Since R does not satisfy any nontrivial generalized polynomial identity, we see that n a([b, [X,Y ]]2) is the zero element in the free product T = Q ∗C C{X,Y }, that is

n−1 2 n−1 2 a([b, [X,Y ]]2) b[X,Y ] + a([b, [X,Y ]]2) [X,Y ] b = 0.

It follows from b∈ / C that

n−1 2 n−1 2 a([b, [X,Y ]]2) b[X,Y ] = 0 = a([b, [X,Y ]]2) [X,Y ] b

n−1 and so a([b, [X,Y ]]2) = 0 ∈ T . Continuing this process, we obtain that a[b, [X,Y ]]2 = 0, that is ab[X,Y ]2 + a[X,Y ]2b = 0. Since b∈ / C, we get that a[X,Y ]2b = 0 and so a = 0 as desired.

Lemma 2 — Let R be a prime ring of characteristic 2, I a nonzero ideal of R, d a nonzero derivation of R, a ∈ R and n ≥ 1 a fixed integer. If a[d([x, y]), [x, y]]n = 0 for all x, y ∈ I, then a = 0 unless R satisfies S4, the standard identity in 4 variables. DERIVATIONS IN PRIME RINGS OF CHARACTERISTIC 2 461

PROOF : Suppose on the contrary that both a 6= 0 and R does not satisfy S4. By assumption we have

a[[d(x), y] + [x, d(y)], [x, y]]n = 0 for all x, y ∈ I. If d is not Q-inner, by Kharchenko’s theorem in [9], we get

a[[z, y] + [x, t], [x, y]]n = 0 for all x, y, t, z ∈ I. In particular, for z = 0, we obtain that a[[x, t], [x, y]]n = 0 for all x, y, t ∈ I. By a theorem due to Chuang [4, Theorem 2], a[[x, t], [x, y]]n = 0 for all x, y, t ∈ Q and so in R. Let v = [[x, t], [x, y]]n. Then av = 0. Now we can write a[[w, v], [ua, v]]n = 0 for all w, u ∈ R. Since av = 0, it reduces to a(wv2ua)n = 0. This can be written as (awv2u)n+1 = 0 for all w, u ∈ R. By Levitzki’s lemma [7, Lemma 1.1] awv2u = 0 for all w, u ∈ R. Since R is prime, v2 = 0 that is [[x, t], [x, y]]2n = 0 for all x, y, t ∈ R. This is a polynomial identity and hence there exists a field

F such that R ⊆ Mm(F ) with m > 1 and R and Mm(F ) satisfy the same polynomial identities. 2n But choosing x = e11, y = e21 and t = e12 we get 0 = [[x, t], [x, y]] = e11 + e22 which is a contradiction.

Thus we assume that d is Q-inner. That is, there exists a noncentral element b ∈ Q such that n d(x) = [b, x] for all x ∈ R. So a([b, [x, y]]2) = 0 for all x, y ∈ I. Since I and Q satisfy the same generalized polynomial identities (or, GPIs in brief), we have

n a([b, [x, y]]2) = 0 for all x, y ∈ Q. (1)

In case, C is infinite, the GPI (1) is also satisfied by Q ⊗C C where C is the algebraic closure of

C. Since both Q and Q⊗C C are prime and centrally closed [5], we may replace R by Q or Q⊗C C according as C is finite or infinite. Thus we may assume that R is centrally closed over Z which is n either finite or algebraically closed such that a([b, [x, y]]2) = 0 for all x, y ∈ R. By Martindale’s theorem in [14] R is a having nonzero socle and the commuting division D is a finite dimensional central division algebra over Z. Since Z is either finite or algebraically closed, D must coincide with Z. Thus R is isomorphic to a dense of EndZ V for some vector space V over

Z. Since R does not satisfy S4, it is obvious that dimZ V ≥ 3. We will show that, for any given v ∈ V, v and bv are Z-dependent. Assume on the contrary that v and bv are Z-independent and set

W = Zv + Zbv. Since dimZ V ≥ 3, there exists u ∈ V such that v, bv, u are also Z-independent.

If av 6= 0, by the density of R in End Z V there exist two elements r1 and r2 in R such that

r1v = 0, r1bv = u, r1u = v; r2v = 0, r2bv = 0, r2u = u.

n It follows from (1) that 0 = a([b, [r1, r2]]2) v = av, a contradiction. 462 YU WANG

Suppose that av = 0. Since a 6= 0, there exists w ∈ V such that aw 6= 0 and so a(v − w) 6= 0. As above we get that there exist β, γ ∈ Z such that

bw = βw and b(w − v) = γ(w − v).

This yields that (β − γ)w ∈ W . Now β = γ implies the contradiction that bv = βv. Thus β 6= γ and so w ∈ W . But if u ∈ V with au = 0, then a(w + u) 6= 0. So w + u ∈ W forcing u ∈ W . Thus V = W and so dimZ V = 2, a contradiction.

Hence, in any case, for all v ∈ V , v and bv are linearly Z-dependent. Thus, standard arguments show that b ∈ Z which contradicts our hypothesis. With this the proof is complete.

The following lemma is due to Chang and Lee [3, Theorem 1].

Lemma A [3] — Let R be a prime ring and ρ a nonzero right ideal of R. If ρ possesses a central differential identity, then R is a PI-ring.

The following lemma is due to Lanski [11, Lemma 2].

Lemma B [11] — Let R be a noncommutative simple algebra, finite-dimensional over its center

Z. If g(x1, . . . , xt) ∈ R∗Z Z{xj}, the free product over Z, is an identity for R that is homogeneous in {x1, . . . , xt} of degree d, then for some field F and n > 1, R ⊆ Mn(F ) and g(x1, . . . , xt) is an identity for Mn(F ).

Lemma 3 — Let R = Mm(F ), the ring of m × m matrices over a field F of characteristic 2 n n and a, b ∈ R with b∈ / F . If a([b, [x, y]]2) ∈ F for all x, y ∈ R and a([b, [r1, r2]]2) 6= 0 for some r1, r2 ∈ R, then m ≤ 2.

PROOF : Suppose on the contrary that m ≥ 3. By assumption we easily see that a is an P invertible matrix. We set b = bstest with bst ∈ F . For distinct 1 ≤ i, j, r ≤ m, if we choose [r1, r2] = [eij, ejr − eri] = eir + erj, then X X [b, [r1, r2]]2 = beij + eijb = bjieii + bjiejj + bjteit + bsiesj t6=i s6=j and by assumption we get X X n n n −1 ([b, [r1, r2]]2) = (bji) eii + (bji) ejj + αteit + βsesj ∈ F a t6=i s6=j

n with αt, βs ∈ F . Since m ≥ 3, we easily see that ([b, [r1, r2]]2) is not an invertible matrix. Thus X X n n n ([b, [r1, r2]]2) = (bji) eii + (bji) ejj + αteit + βsesj = 0. t6=i s6=j DERIVATIONS IN PRIME RINGS OF CHARACTERISTIC 2 463

n It follows that in particular, (bji) = 0 and so bji = 0. Therefore, b is a diagonal matrix, that is P b = bssess with bss ∈ F . P Next we show that b = bssess is a central matrix, that is, bii = bjj for distinct i, j. Let

ϕij ∈ AutF (R) such that ϕij(x) = (1 + eij)x(l − eij) for all x ∈ R. On the one hand, for any n x, y ∈ R, we have ϕ(a)([ϕij(b), [x, y]]2) ∈ F . Since ϕ(a) 6= 0, as above we have that ϕij(b) is also diagonal. On the other hand, ϕij(b) = b + (bjj − bii)eij, i.e., bjj = bii for any j 6= i. Hence b is a central matrix, a contradiction. The proof is complete.

We are now in a position to prove our main result.

Theorem 1 — Let R be a prime ring of characteristic 2, Z the center of R, d a nonzero derivation of R, L a noncentral Lie ideal of R and a ∈ R. If a[d(u), u]n ∈ Z for all u ∈ L, then either a = 0 or R satisfies S4, the standard identity in 4 variables.

PROOF : Suppose that R does not satisfy S4. That is dimC RC > 4. Our goal is to prove that a = 0. By a theorem of Lanski and Montgomery [12, Theorem 13], we have [L, L] 6= 0 and 0 6= [I,R] ⊆ L, where I is the ideal of R generated by [L, L]. By assumption we have a[d([x, y]), [x, y]]n ∈ Z for all x, y ∈ I. Thus

a[[d(x), y] + [x, d(y)], [x, y]]n ∈ Z for allx, y ∈ I. (2)

If a[d([x, y]), [x, y]]n = 0 for all x, y ∈ I, by Lemma 2 we get that a = 0 as desired. Otherwise, n there exist r1, r2 ∈ I such that a[d([r1, r2]), [r1, r2]] 6= 0. Thus I satisfies the central differential identity a[[d(X),Y ] + [X, d(Y )], [X,Y ]]n.

In view of Lemma A, R must be a PI-prime ring and so is Q. If d is not Q-inner, applying Kharchenko’s theorem [9] to (2), we have

[a([[t, y] + [x, z], [x, y]])n, r] = 0 for all x, y, t, z, r ∈ I.

In particular, for z = 0,

[a[[t, y], [x, y]]n, r] = 0 for all x, y, t, r ∈ I. (3)

Since I and Q satisfy the same generalized polynomial identities, (3) is satisfied by Q. If a 6= 0, then (3) is a nontrivial GPI on Q. By Martindale’s theorem in [14] Q is a PI-primitive ring. By the famous Kaplansky-Amitsur’s theorem (see [8, Page 17, Theorem]) Q is finite dimensional central simple algebra over C. It follows from Lemma B that there exists a suitable field F of characteristic

2 such that Q ⊆ Mm(F ), the ring of all m × m matrices over F , and moreover Mm(F ) satisfies 464 YU WANG

P (3). Since dimC RC > 4, it implies that m > 2. We set a = astest with ast ∈ F . For distinct integers 1 ≤ i, j, s ≤ m, if we choose x = esj, y = ess, t = ejs, r = eji in (3), then X akjeki + aijejj + aisejs = 0 k

This implies that akj = 0 for any k, j, that is a = 0, a contradiction.

We next assume that d is an Q-inner derivation induced by an element b ∈ Q. Since d 6= 0, b∈ / n C. Hence a([b, [x, y]]2) ∈ Z for all x, y ∈ I and so

n [a([b, [x, y]]2) , z] = 0 for all x, y, z ∈ Q. (4)

n Since there exist r1, r2 ∈ I such that a([b, [r1, r2]]2) 6= 0 it is obvious that (4) is a nontrivial GPI on Q. By Martindale’s theorem in [14] Q is a PI-primitive ring. As above we have that Q is finite dimensional central simple algebra over C. It follows from Lemma B that there exists a suitable field F of characteristic 2 such that Q ⊆ Mm(F ), the ring of all m × m matrices over F with m ≥ 3 and moreover Mm(F ) satisfies the same (4). But it follows from Lemma 3 that m ≤ 2, a contradiction. The proof is thereby complete.

ACKNOWLEDGEMENT

The author are thankful to the referees for some helpful comments and suggestions.

REFERENCES

1. K. T. Beidar, W. S. Martindale and A. V. Mikhalev, Rings with Generalized Identities, Pure appl. Math., Dekker, New York, 1996.

2. L. Carini and De Filippis, Commutators with power central values on a Lie ideal, Pacific J. Math., 193 (2000), 269-278.

3. C. M. Chang and T. K. Lee, Annihilators of power values of derivations in prime rings, Comm. Alge- bra, 26 (1998), 2091-2113.

4. C. L. Chuang, GPIs having coefficients in Utumi quotient rings, Proc. Amer. Math. Soc., 103 (1988), 723-728.

5. J. S. Erickson, W. S. Martindale and J. M. Osborn, Prime nonassociative algebra, Pacific J. Math., 60 (1975), 49-63.

6. De Filippis, Lie ideals and annihilator conditions on power values of commutators with derivation, Indian J. pure appl. Math., 32 (2001), 649-656. DERIVATIONS IN PRIME RINGS OF CHARACTERISTIC 2 465

7. I. H. Herstein, Topics in ring theory, Univ. Chicago Press, 1996.

8. N. Jacobson, PI-algebrs, An Introduction, Lecture Notes in Math. 441, Springer-Verlag, 1975.

9. V. K. Kharchenko, Differential identities of prime rings, Algebra and Logic, 17 (1978), 155-168.

10. C. Lanski, Differential identities, Lie ideals, and Posner’s theorems, Pacific J. Math., 134 (1988), 275-297.

11. C. Lanski, An Engel condition with derivation, Proc. Amer. Math. Soc., 118 (1993), 731-734.

12. C. Lanski and S. Montgomery, Lie structure of prime rings of characteristic 2, Pacific J. Math., 42 (1972), 117-136.

13. P. H. Lee and T. K. Lee, Lie ideals of prime rings with derivations, Bull. Inst. Math. Acad. Sinica, 11 (1983), 75-80.

14. W. S. Martindale, Prime rings satisfying a generalized polynomial identity, J. Algebra, 14 (1969), 576-584.

15. E. G. Posner, Derivations in prime rings, Proc. Amer. Math. Soc., 8 (1957), 1093-1100.