zelmanowitz1976

PROCEEDINGS OF THE
AMERICAN MATHEMATICAL SOCIETY

Volume 57, Number 2, lune 1976
THE FINITE INTERSECTION PROPERTY ON

ANNIHILATOR RIGHT IDEALS
J. M. ZELMANOWITZ1
Abstract. In this article the finite intersection property on annihilator right

ideals will be shown to be an adequate substitute for more stringent chain
conditions on such ideals. One application of the investigation will produce
a new characterization of orders in semisimple artinian rings, another will
generate new classes of absolutely torsion-free rings.
Until otherwise indicated all rings are arbitrary associative rings not
necessarily possessing an identity element. To simplify the statements of the
results, we expand the usual definition of a prime ideal to include the ring
itself. For any subset A of a ring R we set /-(/I) = (0: A) = {r G R\Ar = 0},
the right annihilator of A; (0: x) being written for (0: {x}). More generally,
for A and B subsets of R, (5: A) will denote {r G R\Ar C B). We let ¡iA)
denote the left annihilator of A.
A ring R is said to have the finite intersection property on right annihilators
provided that whenever riA) = 0 for a right ideal A C R there exists
xv ..., xn G A with n"=i(0: x¡) = 0. It is readily apparent that a ring which
satisfies the descending chain condition on right annihilators possesses this
property; for choosing xx, . . . , xn G A with n"=1(0: x¡) minimal among all
such intersections forces n"=1(0: xt) = 0. The converse is false however. For
instance a commutative subdirectly irreducible (i.e., having a unique minimal
ideal) nil ring which is not nilpotent has the finite intersection property; in
fact it satisfies the stronger requirement that DxSAi0: x) = 0 implies that
(0: x) = 0 for some x G A. But such a ring cannot satisfy the descending
chain condition on annihilators, else by well-known theorem [3, Theorem 1] it
would be nilpotent. For a specific example of such a ring one may take any
subdirectly irreducible homomorphic image of ®0<a<xFxa where F is a field
and multiplication is defined by xaXß = xa + /8 if a + ß < 1 and 0 otherwise
(see Example 3 of [1]). This example also demonstrates that finite intersection
properties on annihilators cannot force the nilpotence of nil rings.
A ring will be called nonsingular if its right singular ideal Z(Ä) is zero,
where Z(/\) = {a G /?|(0: a) is an essential right ideal).
Proposition 1. Assume (1) riP) = Ofor every prime ideal P of R; and
(2) R has the finite intersection property on right annihilators.
Received by the editors December 18, 1974 and, in revised form, March 27, 1975.
AMS (MOS) subject classifications (1970). Primary 16A34; Secondary 16A12, 16A18, 16A46.
Key words and phrases. Finite intersection property on right annihilators, descending chain
condition on right annihilators, right Goldie ring, absolutely torsion-free ring, hereditary kernel
functor.
1 This research was supported in part by NSF Grant GP 34098.
© American Mathematical Society 1976
213
License or copyright restrictions may apply to redistribution; see http://www.ams.org/journal-terms-of-use
214 J. M. ZELMANOWITZ
Then given any nonzero right ideal I of R there exist xx, . . . , xn G / with
n;=1(o:*,.) = o.
Proof. It suffices to show that r(I) = 0 for all nonzero right ideals /. If this
is not the case, then use the finite intersection property together with Zorn's
lemma to choose a right ideal P maximal with respect to r(P) i= 0. Now
r(P + RP) = r(P), so P = P + RP by the maximality of P, and thus P is an
ideal of R. In fact P is a prime ideal. For if A and B are ideals of R properly
containing P then r(A) = r(B) = 0. Hence r(AB) = 0, and it follows that
P C¿ AB. Thus P is a prime ideal with r(P) =£ 0, a contradiction which
establishes the conclusion.
Corollary. The rings satisfying the hypotheses of Proposition 1 are prime

nonsingular rings.
Proof. For in particular r(I) = 0 for every nonzero right ideal /, whence R
is prime. Also if Z(R) ^ 0 then we have the contradiction that n"=1(0: xA
= 0 for some xx, . . . , xn G Z(R). \\
We call a ring R prime (semiprime) right Goldie if R is prime (semiprime),
nonsingular, and finite dimensional. As is well known, these properties
characterize right orders in simple (semisimple) artinian rings [2].
Theorem 2. Assume (1) r(P) = 0 for every prime ideal P of R;
(2) R has the finite intersection property on right annihilators;
(3) R has a uniform right ideal.
Then R is prime right Goldie.
Proof. In view of the preceding corollary it remains only to prove that R is
finite dimensional. Let / be a uniform (i.e., 1-dimensional) right ideal of R.
By Proposition 1, there exist xv . . ., xH €■'I with n"=l(0: xA = 0. It follows
that the homomorphism $: R -> Iw defined by <f>(r)= (xxr, . . . , xnr) is an
isomorphism. Hence by [2, Theorem 1.1] R is finite dimensional. ||
With but slight modification these results can be extended to semiprime
rings. We will therefore not give complete proofs.
Proposition 3. Assume (1) P n r(P) = 0 for every prime ideal P of R;
(2) R has the finite intersection property on right annihilators.
Then given any right ideal I there exist xx, . . . , xn G / with
n
H (0: xA n / = 0.
í=i
In particular, R is a semiprime nonsingular ring.
Proof. If the conclusion fails then one can use Zorn's lemma to choose a
right ideal P maximal with respect to the property that r(F) n P =£ 0 for all
finite subsets F of P. If G is a finite subset of P + RP, then there exists a
finite subset F of P with r(F) Ç r(G), and from this it follows that P = P +
RP, so P is an ideal.
Next, from the finite intersection property one has that r(P) =£ 0.
Furthermore P n r(P) =£ 0. Else P C¿ P + r(P), whence there exists a
finite subset H C P + r(P) such that r(H) n (P + r(P)) = 0. From H one

obtains finite subsets F ç P and G C r(P) with H ç F + G. Now
THE FINITE INTERSECTION PROPERTY 215
H(r(F) n P) C F(r(F) n P) + G(r(F) n P)

= G(r(F) n P)Q r(P)P Ç r(P) n P = 0.
So r(F) n P Ç r(/7) n P = 0, contradicting the choice of P.
If A and ß are ideals properly containing P, it follows that r(A) n A = 0
= r(B) n B. Also rL4ß) n iA n ß) = 0. For if X G r(/lß) n (A n 5),
then Äx Ç r(^) n ^ = 0 whence x G r(B) n £ = 0. Finally P C¿ ^45; for
if P = /Iß then 0 ^ r(P) n P Ç rL4ß) n iA n ß) = 0. So P is a prime
ideal with P n r(P) ^ 0, contradicting (1).
Additionally, we have r(/) n / = 0 for every right ideal /, so R is
semiprime. Also if Z (R) ¥= 0 there exist z,, . . . , zn G Z (R) with
Z (P. ) n D "_ ,(0: z¡) = 0, which is impossible since D "= ,(0: z¡) is an essential
right ideal. II
As in Theorem 2 we get the following immediate consequence.
Theorem 4. Assume (1) P n r(P) = 0/or every prime ideal P of R;
(2) R has the finite intersection property on right annihilators;
(3) R has a faithful finite dimensional right ideal.
Then R is semiprime right Goldie.
Corollary. // R is a prime irespectively, semiprime) ring satisfying the
descending chain condition on right annihilators and possessing a uniform
(respectively, faithful finite dimensional) right ideal, then R is prime (respec-
tively, semiprime) right Goldie.
Right absolutely torsion-free rings (abbreviated as ATF rings) were in-
troduced in [4] as rings with identity elements in which for every kernel
functor (synonym: left exact preradical) o on right ß-modules with o(R)
¥= R, o(R) = 0. See [4] or [5] for the terminology used here. It is straightfor-
ward to see that R is a right ATF ring if and only if given any nonzero right
ideal I of R, R can be embedded in /<") = / © • • • © / for some positive
integer n (equivalently, there exist x,, . . . , xn G / with D"=1(0: x¡) = 0). See
[5] for a proof of this and other characterizations. We remark that all of what
follows is valid for rings without identity elements, provided appropriate
modifications are made. However for the sake of simplicity we henceforth
assume that rings contain identity elements.
In this terminology, Proposition 1 can be restated as follows.
Proposition 5. If R is a prime ring satisfying the finite intersection property
on right annihilators then R is right A TF.
Corollary. A prime ring which satisfies the ascending chain condition on left
annihilators is right ATF.
Proof. The ascending chain condition on left annihilators is equivalent to
the descending chain condition on right annihilators. ||
For convenience, let us call a right ideal / exceptional if R cannot be
embedded in I(n) for any positive integer n. The next result follows directly
from an examination of the proof of Proposition 1.
Proposition 6. R is right A TF if and only if every nonzero prime ideal of R

is not exceptional.
216 J. M. zelmanowitz
Proposition 7. // P is a maximal exceptional right ideal of R, then P is a

prime ideal, either r(P) = 0 or P = lr(P), R/P is right ATF, and P = o(R)
where o is the kernel functor associated to the topologizing filter g = {/1 / is a
right ideal containing Pl"=1(0: pA for some /?,, . . . , pn G P).
Proof. This is trivial if R is right ATF, so we may assume that R is not
ATF. As in the proof of Proposition 1, a maximal exceptional right ideal of R
is a prime ideal. If P =£ lr(P). then lr(P) is not exceptional, so there exists a
monomorphism /:£.—> /r(P)(n) for some positive integer n. Since f(r(P))
= f(l)r(P) = 0, it follows that r(P) = 0.
To see that R/P is right ATF, let a right ideal / ¡? P be given. By the
choice of P there exists a finite subset F Q I with r(F) = 0. Certainly
P C (P: F). If P ¥= (P '■F), then there exists a finite set G C (P : F) with
r(G) = 0. But FG is a finite subset of P, whence r(FG) ^ 0, a contradiction.
Thus P = (P : F) which proves that Ij'P is not exceptional in R/P. So R/P
is right ATF.
For the final statement of this proposition, recall that
o(R)= {a G Ä|(0:a)GS},
from which it is obvious that P Q o(R). Note that since P is exceptional,
0 (2 g. If P ¥= o(R) then there exist xx, . . . , x„ G o(R) with n"=,(0:jc,)
= 0. So 0 6 g, a contradiction which establishes the fact that P = o(R). ||
The author gratefully acknowledges the improvements suggested by the
referee, and the assistance of L. W. Small in providing the example which
appears at the beginning of this article.
References
1. N. J. Divinsky, Rings and radicals, Mathematical Expositions, no. 14, Univ. of Toronto
Press, Toronto, Ont., 1965. MR 33 #5654.
2. A. W. Goldie, Semi-prime rings with maximum condition, Proc. London Math. Soc.
(3)10(1960),201-220. MR 22 #2627.
3. I. N. Herstein and L. W. Small, Nil rings satisfying certain chain conditions, Canad. J. Math.
16(1964),771-776; addendum, ibid. 18(1966),300-302. MR 29 #3497; 32 #5690.
4. R. A. Rubin, Absolutelytorsion-freerings, Bull. Amer. Math. Soc. 78(1972),854-856. MR 45
#8683.
5. J. Viola-Priolo, On absolutely torsion-free rings, Pacific J. Math. 56 (1975), 275-283.
Department of Mathematics, University of California, Santa Barbara, California

93106

zelmanowitz1976

Uploaded by

Copyright:

Available Formats

zelmanowitz1976

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

zelmanowitz1976

Uploaded by

Copyright:

Available Formats

PROCEEDINGS OF THE

AMERICAN MATHEMATICAL SOCIETY

THE FINITE INTERSECTION PROPERTY ON

Abstract. In this article the finite intersection property on annihilator right

Corollary. The rings satisfying the hypotheses of Proposition 1 are prime

finite subset H C P + r(P) such that r(H) n (P + r(P)) = 0. From H one

H(r(F) n P) C F(r(F) n P) + G(r(F) n P)

Proposition 6. R is right A TF if and only if every nonzero prime ideal of R

Proposition 7. // P is a maximal exceptional right ideal of R, then P is a

Department of Mathematics, University of California, Santa Barbara, California

License or copyright restrictions may apply to redistribution; see http://www.ams.org/journal-terms-of-use

You might also like