11
This article was downloaded by: [University of Regina] On: 02 May 2013, At: 00:27 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Communications in Algebra Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/lagb20 Self-adjoint ideals in baer *-rings Gary F. Birkenmeier a & Jae Keol Park b a Department of Mathematics, University of Southwestern Louisiana, Lafayette, LA, 70504-1010, U.S.A E-mail: b Department of Mathematics, Busan National University, Busan, 609-735, South Korea E-mail: Published online: 27 Jun 2007. To cite this article: Gary F. Birkenmeier & Jae Keol Park (2000): Self-adjoint ideals in baer *-rings, Communications in Algebra, 28:9, 4259-4268 To link to this article: http://dx.doi.org/10.1080/00927870008827088 PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

Self-adjoint ideals in baer *-rings

  • Upload
    jae

  • View
    221

  • Download
    4

Embed Size (px)

Citation preview

Page 1: Self-adjoint ideals in baer *-rings

This article was downloaded by: [University of Regina]On: 02 May 2013, At: 00:27Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Communications in AlgebraPublication details, including instructions for authors and subscriptioninformation:http://www.tandfonline.com/loi/lagb20

Self-adjoint ideals in baer *-ringsGary F. Birkenmeier a & Jae Keol Park ba Department of Mathematics, University of Southwestern Louisiana,Lafayette, LA, 70504-1010, U.S.A E-mail:b Department of Mathematics, Busan National University, Busan,609-735, South Korea E-mail:Published online: 27 Jun 2007.

To cite this article: Gary F. Birkenmeier & Jae Keol Park (2000): Self-adjoint ideals in baer *-rings,Communications in Algebra, 28:9, 4259-4268

To link to this article: http://dx.doi.org/10.1080/00927870008827088

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Any substantialor systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply,or distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representationthat the contents will be complete or accurate or up to date. The accuracy of anyinstructions, formulae, and drug doses should be independently verified with primarysources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand,or costs or damages whatsoever or howsoever caused arising directly or indirectly inconnection with or arising out of the use of this material.

Page 2: Self-adjoint ideals in baer *-rings

COMMUNICATIONS IN ALGEBRA, 28(9), 42594268 (2000)

SELF-ADJOINT IDEALS I N B A E R *-RINGS

Gary F. Birkenmeier Department of Mathematics

University of Southwestern Louisiana Lafayette, LA 70504-1010

U. S. A. E-mail: gf bll27Qusl. edu

and

Jae Keol Park Department of Mathematics Busan National University

Busan 609-735, South Korea E-mail: j kparkQhyowon . cc . pusan. ac . kr

0. Introduct ion. Throughout this paper R denotes an associative ring with unity. If an involution * is defined on R, then R is called a *-ring. If X is a nonempty subset of R, then X is said to be self-adjoint if X = X*. We use M,(R) for the n x n matrix ring over R. The complex number field, the rational number field, and the ring of integers are denoted by @, Q, and Z, respectively.

Recall from [14], R is a Baer ring if the right annihilator of every nonempty subset is generated, as a right ideal, by an idempotent. If R is a *-ring, then R is called a Baer *-ring if the right annihilator of every nonernpty subset is generated, as a right ideal, by a projection (i.e., e is a projection if e = e2 and e* = e). The study of Baer and Baer *-rings has its roots in functional analysis [14] and [I]. The class of Baer *-rings includes the von Neumann algebras (e.g., the algebra of all bounded operators on a Hilbert space), the commutative Cc-algebra C(T) of continuous complex valued functions on a Stonian space T, and the complete *-regular rings. Also note that the Sherman-Takeda theorem in [18] and [19] shows that every CL-algebra has a universal enveloping von Neumann algebra (hence a Baer *-ring). Kaplansky shows in [14] that the definitions of a Baer ring and a Baer *-ring are left-right symmetric.

Copyright O 2000 by Marcel Dekker, Inc.

Dow

nloa

ded

by [

Uni

vers

ity o

f R

egin

a] a

t 00:

27 0

2 M

ay 2

013

Page 3: Self-adjoint ideals in baer *-rings

4260 BIRKENMEIER AND PARK

From [a], a ring R is quasi-Baer if the right annihilator of every right ideal is generated, as a right ideal, by an idempotent. This is a nontrivial general- ization of the class of Baer rings. For example prime rings with nonzero right singular ideal [15] are quasi-Baer and not Baer. If R is a non-Priifer com- mutative domain, then Mn(R) (n > 1) is a prime P I quasi-Baer ring which is not a Baer ring [16] and [14, p.171. The n x n (n > 1) upper triangular matrix ring over a domain which is not a division ring is a quasi-Baer ring but not Baer [16] and [14, p.161. Every semiprime right FPF ring [lo, P. 1681 and every piecewise domain [ll] and [5, Corollary 4.131 are quasi-Baer rings.

We define R to be a quasi-Baer *-ring if the right annihilator of every ideal is generated, as a right ideal, by a projection. In [a], Clark proves that the definition of a quasi-Baer ring is left-right symmetric. This is shown in the sequel for quasi-Baer *-rings. There are numerous examples of quasi- Baer *-rings which are not Baer *-rings. For example, from [6] Mz(C)[x] is a quasi-Baer *-ring (* is the conjugate transpose involution) which is not a Baer *-ring. Other examples will be provided in the sequel.

It is well known that every ideal in a von Neumann algebra is self-adjoint. This follows from the polar decomposition of elements [9, p.91. In [I , p.1341, an abstract polar decomposition is defined (also see [14]) for elements of a *- ring. If a *-ring has this polar decomposition property (PD) for all elements, then every ideal is self-adjoint [l, Exercise 5A, p.1351. In [I, pp.135-1361 several conditions on Baer *-rings are given which imply PD. Handelman [12, Theorem 3.61 shows that if R is a commutative *-ring such that Mn(R) is a Baer *-ring for all n, then every ideal of R is self-adjoint. Also a routine argument shows that in a *-regular ring (i.e., a regular ring with a proper involution) every ideal is self-adjoint. At this point, one might conjecture that at least in a commutative Baer *-ring, every ideal is self-adjoint. However we will show by example that this is not the case. This motivates one to ask: How closely is the ideal structure of a (quasi-) Baer *-ring connected to its *-ideal (i.e., self-adjoint ideal) structure?

In Section 1, our main result (Theorem 1.8) shows that every ideal in a quasi-Baer *-ring is "essentially sandwiched" between self-adjoint ideals. This result is applied in Section 2 to show that minimal left ideals of a quasi- Baer *-ring have either a "Hermitian" form or an "alternating" form. Since not every ideal in a (quasi-) Baer *-ring is self-adjoint, it is natural to ask if the (quasi-) Baer *-condition can be characterized in terms of self-adjoint ideals. An affirmative answer is provided in Section 3.

Recall from [2] an idempotent e E R is left (resp. right) semicentral in R if eRe = Re (resp. eRe = eR). Equivalently, an idempotent e is left (resp, right) semicentral in R if eR (resp. Re) is an ideal of R. Note if R is semiprime then every left (resp. right) semicentral idempotent is central. For a nonempty subset X of R , 1 ( X ) and r ( X ) denote the left and right annihilators of X, respectively. Also Ze(R) and Z,(R) denote the left and

Dow

nloa

ded

by [

Uni

vers

ity o

f R

egin

a] a

t 00:

27 0

2 M

ay 2

013

Page 4: Self-adjoint ideals in baer *-rings

SELF-ADJOINT IDEALS IN BAER *-RINGS 426 1

right singular ideals of R, respectively. The right socle of R will be symbolized by Soc(R). Observe that if R is semiprime, then Soc(R) coincides with the left socle of R. If X is a nonempty subset of R , then X Ce R (resp. X C, R) denote that X is a left (resp. right) ideal of R , respectively. For two left (resp. right) ideals X and Y of R, X C;SS Y (resp. X CyS Y) denote that X is left (resp, right) essential in Y, respectively. Recall 11, p.101 an involution * is called proper if x'x = 0 implies x = 0. Other terminology can be found in [I].

1. Essentially self-adjoint. In this section, we present our main result which shows that every ideal in a quasi-Baer *-ring is an essential extension of a self-adjoint ideal and is also essential in a self-adjoint ideal (i.e., "essentially sandwiched").

Proposition 1.1. Let R be a *-ring. Then the following conditions are equivalent:

(i) R is a quasi-Baer *-ring. (ii) R is a quasi-Baer ring in which each left semicentral idempotent is a

projection. (iii) R is a semiprime quasi-Baer ring in which each central idempotent is

a projection (iv) The left annihilator of every left ideal is generated, as a left ideal, by

a projection.

Proof. (i)=+(ii) Clearly R is a quasi-Baer ring. Let e be a left semicentral idempotent. Then r( (1 - e)R) = eR. There exists a projection f such that eR = f R. Hence eR is a self-adjoint ideal. So Re* = eR. Thus e := ee* = e*. Consequently e is a projection.

(ii)$(iii) Since every central idempotent is a left semicentral idempotent, we only need to show that R is semiprime. Assume x E R such that xRx = 0. Then x E r(xR) = eR, where e is a left semicentral idempotent. Then eR is a self-adjoint ideal of R. Hence ex* = x*. So xe = x. But xe == 0. Thus R is semiprime.

(iii)+(iv) Let Y be a left ideal of R. Since R is quasi-Baer, there exists a right semicentral idempotent such that l (Y) = Re. Since Re is an ideal of R , er = ere for every r E R and (1 - e)Re is an ideal of R. The semiprimeness of R implies that (1 - e)Re = 0. Thus re = ere for every r E R. So we have er = ere = re for every r E R. Therefore e is central and hence e is a projection.

(iv)=+(i) Let X be a right ideal of R. Observe that ( r (X))* = e(X*). There exists a projection e such that C(X*) = Re. Hence r ( X ) = eR. Therefore R is a quasi-Baer *-ring.

Corollary 1.2. If R is a prime ring with an involution *, then R is a quasi-Baer *-ring.

Dow

nloa

ded

by [

Uni

vers

ity o

f R

egin

a] a

t 00:

27 0

2 M

ay 2

013

Page 5: Self-adjoint ideals in baer *-rings

4262 BIRKENMEIER AND PARK

Proof. Since a prime ring is a quasi-Baer ring, this result follows from Proposition 1.1 (iii).

Example 1.3. There exists a semiprime Baer ring R with an involution * which is not a quasi-Baer *-ring. Take R = F $ F, where F is a field and * is defined by (a, b)* = (b, a) , for all a, b E F.

The next two examples show that, in general, not every ideal of a Baer *-ring or a quasi-Baer *-ring is self-adjoint.

Example 1.4. There exists a commutative Baer *-ring R which has an ideal I such that I is not self-adjoint. Let * be the involution on C[x] induced by the conjugate involution on C. Take cu E C which is not a real. Then the difference between x - a and its conjugate is a nonzero element of C. Hence the ideal I = (x - a)C[x] is not self-adjoint. Note that @[XI is a Banach @-algebra.

Surprisingly, by [12, Theorem 3.61, the n x n matrix ring, M,(@[[x]]), is a Baer *-ring (the involution is conjugate transpose) in which every ideal is self-adjoint for all n. The next example is a prime subring of M2(@[[x]]) which further illustrates the delicate nature of the condition: every ideal is self-adjoint.

Example 1.5. Let R = ( @[[xll xc [ [x l l ) mom 14, Example A] and x@[[xIl @[[XI]

Corollary 1.2, R is a quasi-Baer *-ring which is not a Baer ring. But the

ideal I = ( x@[[xll xc[[xll) is not self-adjoint, x2@@[[41 x@I[xIl

However all is not lost as our main theorem will demonstrate. Let X, Y be ideals of R such that X c Y. We say X is ideal essential in Y if any nonzero ideal of R which has nonzero intersection with Y also has nonzero intersection with X .

Lemma 1.6. Let R be a semiprime ring with ideals X and Y. If X is ideal essential in Y, then r (X) = r(Y) = 1(X) = C(Y).

Proof. Clearly, r(Y) 5 r (X) . Observe, if Ynr (X) # 0, then X n r ( X ) # 0, a contradiction. Hence Y fl r (X) = 0. So r (X) E r(Y). Thus r (X) = r (Y). Since R is semiprime, r (X) = C(X) = C(Y).

Lemma 1.7. Let R be a semiprime ring with an involution * and I a nonzero ideal of R. Then the following conditions are equivalent:

(i) r ( I ) is a self-adjoint ideal of R. (ii) 11* # 0. (iii) 11* Ce,sS I. (iv) 11* zys I. Proof. (i)+(ii) Assume 11* = 0. Then I* C r ( I ) . Hence I c r ( I ) , a

contradiction.

Dow

nloa

ded

by [

Uni

vers

ity o

f R

egin

a] a

t 00:

27 0

2 M

ay 2

013

Page 6: Self-adjoint ideals in baer *-rings

SELF-ADJOINT IDEALS IN BAER *-RINGS 4263

(ii)+(iii) First we claim that I n I* gYs I . Let Y 2, R such that I n I* n Y = 0 and Y C I . Since Y ( I n I * ) = 0 and R is semiprime, it follows that RY E r ( I n I*) . Also the right annihilator of a self-adjomt ideal in a semiprime ring is self-adjoint, thus (RY)* E r ( I n I*). So R Y + (RY)* is a self-adjoint ideal of R contained in I. Hence RY + (RY)* I n I* n r ( I n I * ) = 0. Thus Y = 0 and I n I* C Y I . Now we will show that 11* c Y s I n I * . Let K ST R such that 11* n K = 0. Then I . l*RK = 0. Observe ( I*RK)2 C I I * R K = 0. Hence I * R K = 0. But R K c I n I * , hence (RK)2 C I ( R K ) = 0. So R K = 0, hence K = 0. Therefore 11* CFSS I .

(iii)+(iv) Since 11* C_YS I, then 11* # 0. Now use a proof atlalogous to (ii) + (iii).

(iv)+(i) By Lemma 1.6, r ( I I * ) = r ( I ) . Since R is semiprime and 11* is a self-adjoint ideal of R, it follows that r ( I I * ) is a self-adjoint ideal of R.

Theorem 1.8. Let R be a quasi-Baer *-ring. (i) If I is an ideal of R, then r ( I + I * ) = r ( I ) = r ( I I* ) = eR, where e is

a central projection. (ii) If I is a nonzero ideal of R , then there is a central projection f E R

such that I T * Ce,SS I CYS I + I* STS f R , and

(iii) If X is a nonzero right ideal, then X X * # 0. (iv) If Y is a nonzero left ideal, then Y*Y # 0.

Proof. (i) If I = 0, we are finished. So assume I # 0. By Proposition 1.1 there exists a central projection e such that r ( I + I * ) = eR. Then r ( I ) = eR@Y, where Y = (1 -e )Rnr ( I ) . Observe that Y* [(I*). By Proposition 1.1, R is semiprime. So [(I*) = r ( I*) . Hence Y n Y* C eR n Y = 0. Since r(Y) is self-adjoint, Lemma 1.7 yields Y = 0. Hence r ( I + I * ) = r ( I ) . Since r ( I ) is a self-adjoint ideal, we obtain r ( I I* ) = r ( I ) from Lemrnas 1.6 and 1.7.

(ii) From Lemma 1.7, 11* EyS I . By [3, Lemma 2.21, I CpS jr R, where f is a central idempotent. From Proposition 1.1, f is a projection. Since f R is a self-adjoint ideal, I + I* CYS f R. A similar argument using the fact that [(I) = r ( I ) yields 11* CTs I Cys I + I* Cess -, f R .

(iii) Assume X X * = 0. hen X * & r ( X ) = r (RX). By part; (i) , r ( R X ) is a self-adjoint ideal. Hence X r (RX) , a contradiction.

(iv) This part is similar to part (iii).

Corollary 1.9. Let R be a quasi-Baer *-ring. The following ideals are self-adjoint: (i) annihilator ideals; (ii) minimal ideals; (iii) Soc(R).

Proof. Part (i) is a consequence of Theorem 1.8(i). From Theorem l.8(ii), each minimal ideal I = 11* = I * . From part (iii), observe that in a semiprime ring every homogeneous component is a minimal ideal and tht: right socle equals the left socle of R.

Dow

nloa

ded

by [

Uni

vers

ity o

f R

egin

a] a

t 00:

27 0

2 M

ay 2

013

Page 7: Self-adjoint ideals in baer *-rings

4264 BIRKENMEIER AND PARK

In [7, Example 4.41 Brown provides an example of a group algebra R = K G over a field K , which is a uniform primitive ring with Z,(R) # 0. Now R is a ring with involution defined by ( x a g g ) ' = xagg- ' . Therefore by Corollary 1.2, R is a quasi-Baer *-ring. Since (Z,(R))* = Zt(R), Theorem 1.8 yields that Zr(R)Ze(R) ce,8s R and Z,(R)Ze(R) Gtss R. Moreover in the introduction of [7], Brown conjectures that if R = K G is a group algebra over a field K , then Z,(R) = Ze(R). Since R has the involution indicated above, our next result supports this conjecture by showing that for group algebras which are quasi-Baer *-rings the left and right second singular ideals coincide.

Corollary 1.10. Let R be a quasi-Baer *-ring. Then the left and the right second singular ideals coincide. In particular R is left nonsingular if and only if R is right nonsingular.

Proof. Since (Z,(R))* = Ze(R), the result is a consequence of Theorem 1.8(ii).

We close this section with the following open problem: Characterize the Baer *-rings (quasi-Baer *-rings) in which every ideal is self-adjoint.

2. Minimal left ideals. In this section we show that a semiprime quasi- Baer ring R has a ring decomposition R = A $ B , where Soc(A) is essential in A and Soc(B) = 0. Moreover each homogeneous component of A is a primitive ring and a direct summand of R. Next by considering quasi- Baer *-rings, we are able to show that the minimal left ideals have either a "Hermitian" form or an "alternating" form. Although the results of this section are left-right symmetric, we consider left ideals to conform to the usage in [17] which is the main reference for this section.

Lemma 2.1. Let R be a quasi-Baer ring with a prime ideal P. Then exactly one of the following conditions is satisfied:

(i) P is left essential in R. (ii) P = Re for some idempotent e E R.

Proof. Assume that there exists 0 # x E R such that P n Rx = 0. Then Rx C r ( P ) = dR, where d = d2 E R. Then P C R ( l - d). Also (1 - d)Rd = 0 E P. Since d @ P, then 1 - d E P. Let e = 1 - d. Then P = Re.

Observe that from Corollary 1.9(i), if R is a quasi-Baer *-ring and P is a nonself-adjoint prime ideal, then P is left and right essential in R.

Lemma 2.2. Let R be a quasi-Baer ring with a minimal left ideal of the form Re, where e = e2. Then R = A @ B (left ideal direct sum), where A is a primitive ring with unity, Soc(A) # 0, and B is a primitive ideal of R.

Proof. From the hypothesis R = Re $ R ( l - e), where R ( l - e) is a maximal left ideal. If R ( l - e) contains no nonzero ideal, then R is a left primitive ring with nonzero socle, so take R = A. Otherwise there exists a

Dow

nloa

ded

by [

Uni

vers

ity o

f R

egin

a] a

t 00:

27 0

2 M

ay 2

013

Page 8: Self-adjoint ideals in baer *-rings

SELF-ADJOINT IDEALS IN BAER *-RINGS 4265

nonzero left primitive ideal B of R such that B 2 R ( l - e). From Lemma 2.1, B = Rb for some idempotent b E R. Let a = 1 - b. Then A = Ra is a left primitive ring with unity a. Since A is ring isomorphic to R / B , then A has an isomorphic copy of Re. Hence Soc(A) # 0. Note that A has nonzero socle, then A is both left and right primitive.

Proposition 2.3. Let R be a semiprime quasi-Baer ring. Then R = A $ B (ring direct sum), where Soc(A) is right and left essential in A and Soc(B) = 0. Moreover, if H is a homogeneous component of A, there exists a central idempotent c E A such that H is in cA and cA is a primitive ring.

Proof. From (3, Lemma 2.21, there exists a central idempotent (1. E R such that Soc(R) is right and left essential in aR. Let A = a R and B - (1 - a)R. Then Soc(A) = Soc(R) and Soc(B) = 0. The remainder of the p:roof follows from the semiprimeness of A and Lemma 2.2.

Note that in Proposition 2.3 if there are only finitely many nonisomorphic minimal left ideals in R , then A is a finite direct sum of primitive rings.

In the remaining results of this section, Proposition 2.3 allows us to con- sider each minimal left ideal of a quasi-Baer *-ring R as being contained in a primitive *-ring which is a direct summand of R. Primitive *:-rings with minimal one sided ideals are discussed in detail in [13]. For the terminology in the remainder of this section see [17, pp.302-3031.

Proposition 2.4. Let R be a quasi-Baer *-ring. If L is a minim.al left ideal of R , then either:

(i) L is generated by a projection; or (ii) EndR(,!,) is a field and xx* = 0 for all x E L.

Proof. Except for minor adjustments, the proof is similar to 117, Propo- sition 2.13.19, p.3011. Observe that condition L"L = 0 in [17, Proposition 2.13.19, p.3011 is eliminated by Theorem 1.8.

Observe that in a Baer *-ring only Proposition 2.4(i) holds because every idempotent generated left ideal is generated by a projection.

Corollary 2.5. Let R be a quasi-Baer *-ring. If L is a minimal left ideal of R , then either (viewing L as a right vector space over D = Endjs(L)):

(i) L has a (*)-compatible Hermitian form which is not alternating; or (ii) L has a (*)-compatible alternating form and D is a field.

Proof. Except for minor adjustments, the proof is similar to [17, Theorem 2.13.21, p.3021.

The next example shows that for a quasi-Baer *-ring both forms may appear.

Example 2.6. [17, Example 2.13.22, p.3031 Let R = M,(F), where F is a field and n > 1. Then R is a prime ring. By Corollary 1.2, R is a quasi- Baer *-ring for any involution *. From [17, Example 2.13.22, p.3031 the

Dow

nloa

ded

by [

Uni

vers

ity o

f R

egin

a] a

t 00:

27 0

2 M

ay 2

013

Page 9: Self-adjoint ideals in baer *-rings

4266 BIRKENMEIER AND PARK

transpose involution will yield that L = Rell has a (*)-compatible Hermitian form which is not alternating. Note that for this involution if R is a Baer *-ring for all n , then F must be formally real [12, Corollary 2.51. If the symplectic involution is used for n an even positive integer, then L = Rell has a (*)-compatible alternating form and D is a field. Since this involution is improper, then R is not a Baer *-ring. Hence R is a Baer ring which is a quasi-Baer *-ring but not a Baer *-ring.

3. Characterizations in t e r m s of t h e *-structure. In this section we characterize Baer *-rings and quasi-Baer *-rings in terms of their *- structures.

Proposi t ion 3.1. Let R be a *-ring. Then the following conditions are equivalent:

(i) R is a Baer *-ring. (ii) * is a proper involution and the right annihilator of every nonempty

self-adjoint subset is generated, as a right ideal, by a projection. (iii) * is a proper involution and the left annihilator of every nonempty

self-adjoint subset is generated, as a left ideal, by a projection.

Proof. (i)+(ii) By [14, Theorem 2.11, * is a proper involution. The re- mainder of the proof of this implication follows from the definition of a Baer *-ring.

(ii)+(i) Let Y be a nonempty subset of R. Form Y*Y = { y ' y 1 y E Y). Then Y'Y is a nonempty self-adjoint subset of R. Then there exists a projection e such that r(Y*Y) = eR. Let y E Y, then we have y * y e = 0. By [I, Proposition 1, p.101, ye = 0. Hence r(Y*Y) r(Y). So r(Y) = r(Y*Y) = eR. Therefore R is a Baer *-ring.

(i)*(iii) The proof of this equivalence is analogous to the proof of (i)*(ii).

Example 3.2. There exists a Baer ring R which is a quasi-Baer *-ring with a proper involution, but R is not a Baer *-ring. Let R = M2(Z) with the involution * being transposition. By [14, p.171, R is a Baer ring. By Corollary 1.2, R is a quasi-Baer *-ring. From 112, Proposition 1.51, R is not a Baer *-ring. But since Mz(Q) is a Baer *-ring, * is a proper involution on R.

One might conjecture that Y U Y' could be used for the self-adjoint nonempty subset in the proof of (ii)+(i) in Proposition 3.1. The next exam- ple eliminates this possibility.

Example 3.3. Let R = Mz(Z3), where Z3 is the field of three elements. By

[14, p.391, R is a Baer *-ring. Let Y = { ( i)}. T h e n r ( Y ~ Y * ) = O #

r (Y) = r(Y*Y).

Proposi t ion 3.4. Let R be a *-ring. The following conditions are equivalent: (i) R is a quasi-Baer *-ring.

Dow

nloa

ded

by [

Uni

vers

ity o

f R

egin

a] a

t 00:

27 0

2 M

ay 2

013

Page 10: Self-adjoint ideals in baer *-rings

SELF-ADJOINT IDEALS IN BAER +-RINGS 4267

(ii) aRa* # 0 for all 0 # a E R and the right annihilator of every self- adjoint ideal is generated, as a right ideal, by a projection.

(iii) aRa* # 0 for all 0 # a E R and the left annihilator of every self-adjoint ideal is generated, as a left ideal, by a projection.

Proof. (i)+(ii) Theorem 1.8(iii) yields the condition aRa* # 0 for all 0 # a E R. The remainder of the proof of this implication follows from the definition of a quasi-Baer *-ring.

(ii)=+(i) First assume K is a self-adjoint ideal of R such that K2 = 0. Then K r(K) = eR for some projection e E R. Hence 0 = Ke and so 0 = e*K* = eK. Therefore K = eK = 0. By 117, Proposition 2.13.34, p.3071, R is semiprime. Now let I be a nonzero ideal of R. Then II* # 0. From Lemma 1.6 and Lemma 1.7(iii), r ( I ) = r(II*) . Since II* is self-adjoint, r ( I ) is generated by a projection.

( i ) ~ ( i i i ) The proof of this equivalence is analogous to the proof of (i)@(ii).

Example 1.3 shows that the condition aRa* # 0 for all 0 # a: E R is not superfluous in Proposition 3.4. Note that Example 2.6 and Mz(Z2) provide examples of quasi-Baer *-rings with an improper involution, where Z2 is the field of two elements.

ACKNOWLEDGEMENTS The authors wish to thank the referee for his/her helpful comments and

suggestions. The second author was supported in part by Korea Research Foundation with Research Grant Project No.1998-001-D00006. Also the sec- ond author appreciates the kind hospitality of the University of Southwestern Louisiana during his stay.

REFERENCES

1. S. K. Berberian, Baer *-Rings, Grundlehren Math. Wiss. Band 195, Springer-Verlag, Berlin-Heidelberg-New York, 1972.

2. G. F. Birkenmeier, Idempotents and completely semiprime ideals, Comm. Algebra 11 (l983), 567-580.

3. G. F. Birkenmeier, A generalization of FPF rings, Comm. Algebra 17 (1989), 855-884.

4. G. F. Birkenmeier, Decompositions of Baer-like rings, Acta Math. Hung. 59 (1992), 319-326.

5. G. F. Birkenmeier, H. E. Heatherly, J. Y. Kim and J. K. Park, Piangular matrix representations, Preprint.

6. G. F. Birkenmeier, J. Y. Kim and J. K. Park, Polynomial extensions of Baer and quasi-Baer rings, Preprint.

7. K. A. Brown, The singular ideals of group rrings, Quart. J. Math. Oxford (2)28 (1977), 41-60.

Dow

nloa

ded

by [

Uni

vers

ity o

f R

egin

a] a

t 00:

27 0

2 M

ay 2

013

Page 11: Self-adjoint ideals in baer *-rings

4268 BIRKENMEIER AND PARK

8. W. E. Clark, Twisted matrix units semigroup algebras, Duke Math. J . 34 (1967), 417-424.

9. J . Dixmier, Von Neumann Algebras, North-Holland, New York, 1969.

10. C. Faith, Injective quotient rings of commutative rings, Module Theory, Lecture Notes in Math. 700, Springer-Verlag, Berlin-Heidelberg- New York, (1979), 151-203.

11. R. Gordon and L. W. Small, Piecewise domains, J . Algebra 23 (1972), 553-564.

12. D. E. Handelman, Prufer domains and Baer *-rings, Arch. Math. 29 (1977), 241-251.

13. I. N. Herstein, Rings with Involution, Univ. of Chicago Press, Chicago, 1976.

14. I. Kaplansky, Rings of Operators, Benjamin, New York, 1968.

15. J . Lawrence, A singular primitive ring, Proc. Amer. Math. Soc. 45 (1974), 59-62.

16. A. Pollingher and A. Zaks, On Baer and quasi-Baer rings, Duke Math. J . 3 7 (1970), 127-138.

17. L. H. Rowen, Ring Theory I, Academic Press, San Diego, 1988.

18. S. Sherman, The second adjoint of a C*-algebra, Proc. Intern. Cong. Math. Cambridge 1 (1950), 470.

19. Z. Takeda, Conjugate spaces of operator algebras, Proc. Japan Acad. 30 (1954), 90-95.

Received: April 1999

Revised: August 1999

Dow

nloa

ded

by [

Uni

vers

ity o

f R

egin

a] a

t 00:

27 0

2 M

ay 2

013