I nternat. J. Math. & Mh. Sci. (1982)21-30 Vol. 5 No. 21 SOME REMARKS ON CERTAIN CLASSES OF SEMILATTICES P.V. RAMANA MURTY and M. KRISHNA MURTY Department of Mathematics Andhra University Waltair 530003 India. (received June 28, 1979 and in revised form March 7, 1980) ABSTRACT. In this paper the concept of a ,-semilattice is introduced as a generalization to distributive *-lattice first introduced by Speed [i]. shown that almost all the results of Speed can be extended to a more of distributive -cmilattices. It is eneral class In pseudocomplemented semilattices and distri- butive semilattices the set of annihilators of an element is an ideal in the sense of Gr’tzer [2]. But it is not so in general and thus we are led to the definition of a weakly distributive semilattice. In 2 we actually obtain the interesting corollary that a modular ,-semilattice is weakly distributive if and only if its dense filter is neutral. In 3 the concept of a sectionally pseudo- complemented semilattice is introduced in a natural way. It is proved that iven a sectionally pseudocomplemented semilattice there is a smallest quotient of it which is a sectionally Boolean algebra. Further as a corollary to one of the theorems it is obtained that a sectionally pseudocomplemented semilattice with a dense element becomes a ,-semilatticeo Finally a necessary and sufficient con- dition for a *-semilattice to be a pseudocomplemented semilattice is obtained. 22 P.V.R. MURTY AND M.K. MURTY KEY WORDS AND. PHRASES. Pseudocomplemented semilattice, -semllattice, weakly dis’tt@tive semiatce, Nal filter, Sectionally pseudocomplemented semilattice. 1980 i. MATHEATICS SUBJECT CLASSIFICATION CODES: 06 INTRODUCTION. Speed [i] has introduced the concept of a distributive *-lattice. In this paper we observe that the condition he has placed on a distributive lattice to become a distributive ,-lattice can be placed on any meet semilattice with 0 an@ accordingly in this paper we call such semilatticesas ,-semiltices (see definition I). It is shown that almost all the results of Speed can be extended to a more general class of distributive ,-semilattices. The concept of a filter in join semilattice and hence (dually) that of an ideal in a meet semilattice is due to Gr’tzer [2]. In pseudocomplemented semilattices and distributive semilattices the set of annihilators of an element is an ideal in the sense of Gtzer. But it is not so in general and thus we are led to the definition of a weakly distributive semilattice. In 2 we prove as a corollary to theorem 4 that a mod- ular @-semilattice is weakly distributive if and only if its dense filter is neutral. Further we show in Theorem 3 that the filter congruence induced by the dense filter on a modular R on L defined by R ,-.qemilattice {(x.y) L is equal to the well known congruence (x)* ( L x L (y)*}. In 3 we study sectionally pseudocomplemented semilattices the concept of which is introduced in [3] and [4]. Actually we prove in Theorem 8 that given a sectionally pseudocomplemented semilattice there is a smallest quotient of it which is a sectionally Boolean algebra. Further we show in Theorem i0 that given a sectionally pseudocomplemented semilattice L there is a Boolean algebra S such that if L has a dense element then L/R is a Boolean algebra isomorphic to S, and obtain as a corollary that a sectionally pseudocomplemented semilattice with a dense element becomes a ,-milattice. Finally we obtain a necessary and suffi- cient condition for a ,-semilattice to be a pseudocomplemented semilattice. 2. RESULTS ON MODULAR,-SEMILATTICES. REMARKS ON CLASSES OF SEMILATTICES DEFINITION I. Let S be a meet semilattice with 0" then for any subset A of S, A * stands for {x E S (a) * 23 An element s E S x ^a 0 for all a 6A}. If A {a} then we denote A is called a dense element of S if and only if (a) by {0}. D denotes the set of all dense elements of S and it can be seen that D is a filter of S (if D is nonempty). The concept of a distributive ,-lattice is due to Speed [i]. However the condition he has placed on a distributive lattice in his definition may as well be placed on any meet semilattice with 0 as in the following" DEFINITION 2. A meet semilattice with 0 is said to be a only if for any aE L there exists DEFINITION 3. a’ L (a’) such that (a) (Gratzer [2] and Katrinak [5]). ,-semilattice if and A nonempty subset I of a meet semilattice S is called an ideal of S if and only if (i) xI and a <x (a 6 S) implies a I (ii) x,yE I implies the existence of z in I such that DEFINITION 4. I(a,b S) implies either and x DEFINITION 6. semilattice if (a) a or b I. A meet semilattice (S, ^) is called a distributive semilattice x>_y^ z imply the existence of yl,Zl if and only if x,y,z 6S and Yl--> ’ Zl_>Z <z and V < z. An ideal I of a meet semilattice S is said to be a prime ideal if and only if a ^b DEFINITION 5. " such that Yl ^Zl" A meet semilattice L with O is called a weakly distributive is an ideal for every a6L. The notion of weak distibutivity in semilattices is a natural generalization of the notion of 0-distributivity introduced by Varlet [6]. Clearly every pseudocomplemented semilattice and every distributive meet semilattice with 0 is weakly distributive. Also one can oh- serve that in a weakly distributive semilattice S, a proper filter U is an ultra filter of S if and only if S-U is a minimal prime ideal of S. It will be shown in this article that almost all the results of Speed [I] for distributive *-lattices (see Theorems 1,2 and 6 of this paper) can be ex- tended to weakly distributive *-semilattices and proofs are given only for those results which can not be obtained from those of Speed. 24 P.V.R. MURTY AND M.K. MIfRTY From now onwards in this article, L denotes a weakly distributive semilattice. M Let - For any subset A of L write be the set of all minimal prime ideals of L. {M h(A) A M} and M (A) {M6 MI 4 ,M}. Then {h (A) A_c called the hull-kernel topology and {, {x} of hl is a topology on a base for the above topology. Also {h(a) la6.L} L} of subsets = X x 6L} is is a base for some topology on called the dual hull-kernel topology. LEMMA i. (i) In L the following hold: A prime ideal M is minimal if and only if (x) (ii) MX (iii) h(x) h((x) (iv) (z) (x) i (Y) (v) (x (vi) (x) h((x) ^ y) ** (x) ** I topology on C (y) h((y) for the hull-kernel topology and ]d for the dual hull-kernel- In L, the following conditions are equivalent. L is a /-semilattice (2) T h Td" hl is compact in the hull-kernel topology. (i)(2) and (2)(3) are straightforward. PROOF. I h(x) i h(y) then we have the Theorem" THEOREM i. (3) nT" h(z) h(x) (y) If we write (I) for every x EM M # Now h(x) is compact. set of t 6 (x) I. that h(xlh(tl), ...I-h(tn (z) Thus THEOREM 2. In L the following are equivalent. (2) For any x AX’ for some t l,...,t (3),(i). L is a ,-semilattice. x so that by compactness of h(x) it follows such that t.< z for i< i< n. (I) L there exists x’6 L such that 0 and Assume that is a closed subset of ,%{and hence is a compact sub- .o (h(x)h(t) there exists z 6. (x) (x) =&i’ x (3)(i)" [x), [x’) -c D n (x) Now h((x) ) and hence h(z) and hence Q.E.D. 25 REMARKS ON CLASSES OF SEMILATTICES 0 and r ^ x’ xl E (t 0 then x, 0 and hence r Assume (2). (i)(2) is obvious. PROOF. zED. Thus t c, (x)*. f for some f E F} ^ A meet semilattice (S,^) is called a (see Rhodes [7]). S such that x THEOREM 3. x,y,z x ^ Yl z ^ S and x e y z imply the existence of ^ Yl ^ Zl" I {(x,y) R where R In a modular *-semilattice S we have (y)*}. PROOF. Clearly modularity of S, y ^ implies y x 2 ^ x Y2 ^ y ^ Y2 (y)*. eD ! R. Now let (x)* x’ 0 y y implies y By Theorem 2, x 2 e x’ and x 2 e y ^ f is a congruence on S called the congruence induced by the modular semilattice if and only if x ^ Q.E.D. DEFINITION 7. (x)* t {(x,y)Ix ^ filter F. YI’ Zl x ^ ^ O. If (S,^) is a meet semilattice and F is a filter of S then y If t r)* so that there exists z -> x, x’ and z ^ r ^ Clearly (x’)** x for some ^ x x imply x ^ Y2 e x’. eD and thus ^ x2 ^ Y2 Evidently (x ^ y x x ^ 2 Similarly x 2ED. Y2 ED’ As also ^ y)* (x)*. for some x ^ we have x’ e x’. 2 0 By y ^ (x,y) x x since Q.E.D. R. It can be observed that if S is not modular then need not be equal to R In the following theorem a necessary and suffi- even if S is pseudocomplemented. cient condition for a modular *-semilattice to be weakly distributive is obtained. THEOREM 4. D If S is a *-semilattice which is directed above such that R then S is a weakly distributive semilattice if and only if for all x,y6 S Assume that S is weakly distributive. PROOF. y ^ 0. d Then there exists t for some dD. (z Now we have (z)* ^ x,y (a’)**. (a)* so that x ^ Thus we have (x [x) v D and y ^ y a ^ a’ E [y) v D. zS Let t E S and let x such that x z, y ([x) n [y)) v D. ^ Hence By (I) there exists e ED such that z (x)* and (y using ^ e _ t and z and z ^ -< t ^ e t’ Conversely suppose (I) holds 0 which implies that (x)** and (y)** a a’)* d ^ t)* so that there exists eD such that z and hence ([x) v D) N ([y) v D) and ...(I) ([x) v D) n ([y) v D) ([x) n [y)) v D ^ a)* R a’ (y)*. Therefore, x ([x) v D) n ([y) v D -< a’ for some z >- x, y. (a)* ^ a’ P.V.R. MURTY AND M.K. MIIRTY 26 Since z ^ e a ^ <- a COROLLARY I. a’ ^ O, we have z ^ z 0 and so a S is a weakly distri- Then Let S be a modular *-semilattice. Q .E .D. (a)*. butive semilattice if and only if D is neutral in the lattice of filters of S. Let L be a meet semilattice with 0 and D # THEOREM 5. , then the following are equivalent. L/ (i) x ^ is pseudocomplemented; 0 implies the existence of dD such that x z PROOF. ^ d (I)(2): (y)**. (x)* t L is a *-semilattice satisfying the condition (2) Since x Let X L and let ^ y 0 so that there exists 0. Therefore r *-semilattice. Clearly Now if x - ^ z (x)*. such that r 0 and thus r t ^ D z’ where (z)** Suppose r ^ d r _< (z)* (z’)*. We now claim that ^ (z’) ^ x 0 and d and hence r (y)**. (y)** and hence (x)* (x) 0 then y ^ d _< OD(y). ((x))* 0 we have (y)** y ^ ^ where (z)**-- (z’)*. Q.E.D. If L is a modular similattice with 0 and D *-semilattice if and only if L/ PROOF. t Thus L is a (2)(i). COROLLARY 2. ^ # then L is a is pseudocomplemented. By Theorem 3 and the above Theorem 5. Q.E.D. Theorem 2 of 5 in [i] can as well be generalized to modular *-semilattices as in the following: THEOREM 6. If L is a modular *-semilattice, then the following are equiva- lent: (i) L is a Boolean algebra (2) L is a disjunctive semilattice, that is, a,b L and a < b implies the existence of (3) cL such that a c 0 and b ^ c # 0. 1 is the on’ly dense element of L. PROOF. Therefore L/ Clearly D (i)(2) and (2)(3). Now assume (3). L and by Corollary 2, L is pseudocomplemented. modular *-semilattice, D # 3. ^ By (3) id. Since L is a 9. SECTIONALLY PSEUDOCOMPLEMENTED SEMILATTICES. In this article a further study is made of the concept of a sectionally Q.E.D. REMARKS ON CLASSES OF SEMILATTICES 27 v Pseudocomplemented semilattice introduced by Katrink [3] and [4]. (see [3] and [4]). DEFINITION 8 A meet semilattice L with 0 is called a sectionally pseudocomplemented semilattice if and only if for every a 6 L the [0,a] interval Every is a pseudocomplemented semilattice. pseudocomplememnted semilattice is a sectionally pseudocomplemented L and X 6 [0,a] the pseudocomplement of x in this semilattice where for a terval [o,a] is given by x ,A n- a. Throughout this article L stands for a sectionally pseudocomplemented semilattice and for any x x 6 [0,a] the pseudocomplement of x LFMA 2. For any x,y,z 6 L we have , (XA y AZ) yAz Since v A (XA Z) PROOF. we have y (x in [0,a] is denoted by a y A A A Z) (x AZ) *yAZ *z (XA y < X A Z ^ 0 y *Z A < yA only if ((x ha) THEOREM 7. (y ha) If @ ,a (y) yh < a), @(y) h < <_ y h (x z) h *Z *@ (a) L, the L/@ ((x h h is a sectionally pseudo- * a) a) @((x ha) , a) (y)= @(0) then (Xhy ha,0). *a y ha) 6 (0). SO Suppose I @. Therefore O.E.D. and L LI-> are sectionally pseudocomplemented semilattices, 2 L 2 is said to be a ,-homomorphism if and only if (f(x)) *f(a) for all x, a 6 L, and x <_ a. Note that if f: LI+ L 2 is a *-homomorphism then kernel f is a ,-congruence and every filter congruence on L is a ,-congruence. DEFINITION II. , that ((xhy ha) yha a) @((x ha) If L y,z and < *yAm 0 Z L whenever (x y) ( , @(yha) A on L is said to be a *-congruence if and and hence by the above lemma (y h(Xh a) then a meet homomorphism f: * z) ^ Suppose @(x) < @(a). Clearly @(x) DEFINITION I0. f(x y for all a 6 ,y (X A , z y) (XA Z) ^ X^ y A , yAZ Z NOW (XA y AZ) is a ,-congruence on (a) such that @(x) @ Z and y A and hence (x complemented semilattice where (@(x)) PROOF. Z AZ) *yAZ DEFINITION 9. A meet congruence ,a , (XA Z) If in L every section [O,a] is a Boolean algebra for all then we call L as a sectionally Boolean algebra, also called as a (dual) P.V.R. MURTY AND M.K. MURTY 28 [8]. semi-Boolean algebra in the sense of Abbott In the following theorem we now show that given a sectionally pseudocomplemented semilattice there is a smallest quotient of it which is a sectionally Boolean alzebra. If we define @ on L by (x,y) THEOREM 8. 0 if and only if (x for all a 6 L, then @ is a ,-congruence on L such that L/ is a sectionally Boolean algebra and if for any ,-congruence L, L/ is a sectionally Boolean on algebra then R(see theorem 3, 2). First observe that @ PROOF. ,-congruence @ Also observe for any on L ((x ^a) If a6 L then S THEOREM 9. * (a) *a) ((x ^a) Q.E.D. denotes the Boolean algebra of closed elements of [0,a]. a {x If we write S ( S a6L a x(a) x(b)^ a whenever a <_ b}, then S is a Boolean algebra and there is a ,-homomorphism f" L /S (whose kernel is @) such that for each a 6 L we have S L f(x *a *b f(a))(b) are isomorphic. Let x 6 L and a 6 L with x * f((x ^a) a )(b) (since x )(b) ((xAB) Aa) **a (x by f(x)(a) /S [0,f(a)] and AaAB) **b (xAb) f(a) (f(x)) (b). _< A a(L Now define a If b _< a. (f(x)(b)) * b ((f(x))’ f(a)(b) A @. Thus f is a ,-homomorphism and ker f be seen that the map xf(x) is a homomorphism of S a into [0,f(a)]. kernel of the above map is prove that it is onto. 0 so that the map is a monomorphism. Let y 6 S such that y If b 6 L, then observe that y(a)^ b _< y(b). Now y(b) _< (y(b) ^a)**b REMARK. verse limit of f(a). _< y(b^ a) y(b) A ** b It can Now we shall Te claim that f(y(a)) y. ,,, y(b)^ a. (a Ab) A Clearly Clearlv (y(b)^ a) (y(b) Aa ^b) b **b O.E.D. Thus the map is an isomorphism. If L is also directed above then it can be seen that S is the in- {S THEOREM i0. morphic to S. y(b) Af(a)(b) L then ((xAa)*aA b) **b ((XAaAb=a Ab)= a) (a^b) **b b S r It can be seen that S is in fact a subalgebra of PROOF. f" a a ab’ L} where ab Sb /S a (a < b) defined by If L has a dense element then L/ is a Boolean ab (x) alebra x ^a. iso- REMARKS ON CLASSES OF SEMILATTICES Let d E L be dense so that L/@ PROOF. and by theorem 9 we have S S so that L/@ [0,f(d)] d _-- [O,f(d)]. 29 under the map @[x] (x Ad) Sd Since f(d)(b) **d b for all b we have _-- S. Q.E.D. If L has a dense element then L is a ,-semilattice. COROLLARY 3. Follows from Theorem i in [9]. PROOF. a sectionally Finally it is natural to ask when does a ,-semilattice become Since every section of a ,-semilattice is also pseudocomplemented semilattice. a *-semilattice we will prove the following: A necessary and sufficient condition for a ,-semilattice S to TEOREM II. x’ be pseudocomplemented is that (i) for each x E S there exists a greatest , (x) such that (x) x’’ and (2) x -< y implies _< y’’. If S is a pseudecomplemented semilattice, then PROOF. x’ complement of x as suppose that S is a ,-semilattice satisfying (i) and (2). we have x N Then (x X ^ y’ AV’. x’’ , (y) and since (x) , (x) Since takin the pseudo- it can be seen that (i) and (2) are satisfied. for any x,y 6 S we have (x) and hence (x ((xAy’)’’) * Now (x Ay’)" _< X"A y’ (by(2)). if and only if (x’’) we have Ay’ x’ (xAy’) * Hence x" Ay’ S is pseudocomplemented. x’ x’ v’. x’’’. (xVV Ay’) SO Conversely First observe that Since (x’) Suppose x Now we claim that (x x" S we have that x <_ X’’Ay’ x" <_ (x) o. ^y ^ y’ _< (xAy’)". y’ and hence 0 .E .D. ACKNOWLEDGEMENT: We thank the referee for his valuable suggestions and comments which helped in shapinz the paper into the present form. P.V.R. MURTY AND M.K. MURTY 30 REFERENCES i. SPEED, T.P. Some Remarks on a class of Distributive lattices, J. Aust. Math. Soc., IX (1969), 289-296. 2. GRTZER, G. Lattice Theory. First Concepts and Distributive lattices, W.H. Freeman and Company, 1971. v Notes on Stone lattices I, Mat. fyz. Casop SAV 16 (1966), 128-142. 3. KATRINAK, T. 4. KATRINAK, T. Charakterisierung der Verallgemeinerten Stoneschen Halbverbande, Mat. Casop. 19 (1969),235-247. 5. KATRINAK, T. 6. VARLET, J.C. A generalization of the notion of pseudocomplementedness, Bull Soc. Roy. Sci. Liege 36 (1968), 149-158. 7. RHODES, J.B. 8. ABBOTT, J.C. 9. SPEED, T.P. A Note on commutative semigroups ,J. Aust. Math. Soc., Vol. VIII (1968), 731-736. v v. Pseudokomplementare Halbverbande, Mat. Casop. 18 (1968), 121-143. Modular and distributive semilattices. Vol. 201 (1975), pp 31-41. Semi-Boolean algebra, Mat. Vestnik Trans. Amer. Math. Soc., (19) (1967), 177-198. Lattice Theory, American Mathematical Society, Colloquium Publications (Third edition 1967). i0. BIRKHOFF, G. ii. CORNISH, W.H. Pseudocomplemented modular semilattices, J. Aust. Math. Soc. Vol. XVIII, (1974), 239-251.