You are on page 1of 12

Modal Propositional Logic on an Orthomodular Basis.

I
Author(s): L. Herman and R. Piziak
Source: The Journal of Symbolic Logic, Vol. 39, No. 3 (Sep., 1974), pp. 478-488
Published by: Association for Symbolic Logic
Stable URL: http://www.jstor.org/stable/2272890
Accessed: 05/12/2010 05:27

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=asl.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Association for Symbolic Logic is collaborating with JSTOR to digitize, preserve and extend access to The
Journal of Symbolic Logic.

http://www.jstor.org
THE JOURNALOF SYMBOLICLOGIC
Volume 39, Number 3, Sept. 1974

MODAL PROPOSITIONAL LOGIC ON AN ORTHOMODULAR


BASIS. I

L. HERMAN AND R. PIZIAK

?1. Introduction. A common method of obtaining the classical modal logics,


for example the Feys system T, the Lewis systems, the Brouwerian system etc., is
to build on a basis for the propositional calculus by adjoining a new symbol L,
specifying new axioms involving L and the symbols in the basis for PC, and
imposing one or more additional transformation rules. In the jargon of algebraic
logic, which is the point of view we shall adopt, the "necessity" symbol L may be
interpreted as an operator on the Boolean algebra of propositions of PC. For
example, the Lewis system S4 may be regarded as a Boolean algebra 2 together
with an operatorL on 2' having the properties: (1) Lp < p for allp in 2, (2)LI = 1,
(3) L(p -. q) < Lp -1Lq for all p, q in 2, and (4) Lp = L(Lp) for all p in 2. Here,
of course, -* denotes the material implication connective: p -> q = p' v q. It is
easy to verify that property (3) may be replaced by either (3') L(p A q) = Lp A Lq
for all p, q in ., or by (3") L(p - q) A Lp < Lq for all p, q in Y. In particular, it
follows from (I) through (4) above that L is a decreasing, idempotent and isotone
operator on 2. Such mappings are often called interior operators.
In a previous paper [5], we considered the problem of introducing an implication
connective into a quantum logic. This is greatly complicated by the fact that the
quantal propositions band together to form an orthocomplemented lattice which is
only "locally" distributive. Such lattices are called orthomodular.For definitions
and further discussion, the reader is referredto that paper. In it, we argued that the
Sasaki implication connective z defined by p O q = p' v (p A q) is a natural
generalization of material implication when the lattice of propositions is ortho-
modular. Indeed, if unrestricted distributivity were permitted, p = q would reduce
to the classical material implication p q. For this reason, we choose z to play the
.-*

role of material implication in an orthomodular lattice. Further properties of D


are enumerated in Example 2.2(1) and Corollary 2.4 below.
On an orthomodular basis, the conditions analogous to (3), (3') and (3") are
distinct, though (3") is derivable from either of (3) or (3') in the presence of (1) and
(2). Thus there are various modal systems generalizingS4 which may be constructed
on an orthomodular basis. However, any operator L satisfying the conditions
analogous to (1), (2), any of (3), (3') or (3"), and (4) must be an interior operator.
In our previous axiomatic treatment of implication connectives, we showed more-
over that an operator L given by Lp = I -< p, where -< is a 2-deductive binary
connective, must be an interior operator. We feel accordingly that any investigation

Received June 25, 1973.


478
0 1974, Association for Symbolic Logc
MODAL PROPOSITIONALLOGIC ON AN ORTHOMODULARBASIS 479

of modal logics on orthomodular bases which resemble S4 requiresa close examina-


tion of interior operators on orthomodular lattices.
Throughout this paper, the symbol . shall denote an orthomodular lattice.
We have already referred the reader to [5] for definitions. However, the notion of
compatibility is so important that we remind the reader of a few of its salient
properties necessary to this paper. Elements p, q in . are called compatible, in
symbols p C q, when p A q = p A (q V p'). The conventional interpretation of two
propositions being compatible is that they are simultaneously testable. It is a fact
that C is a symmetric relation. One of the characterizations of compatibility is that
p C q if and only if p = (p A q) v (p A q'). The essential result for performing
computations in orthomodular lattices is the Foulis-Holland theorem [1, p. 53];
this fundamental theorem states that every distributive law holds among any three
elements as long as one is compatible with the other two. In the sequel, we use this
result almost without comment. Also, we shall use "iff" as an abbreviation for
"if and only if."
With each a E 2, it is useful to associate the mappings 4pa : - .? defined by
'Pa(x)= a A (a' v x) and A+:.2' defined by +(x) = a' v (a A x) = a : x.
The mapping s is called the Sasaki projectiononto a. The basic properties of these
mappings that we shall need are as follows (see (2]):
(1) pa and pa+are isotone mappings; that is, if x < y, then 7a(x) < (pa(y)and
Spa+Wx < ON+(Y).
(2) pa(q (x)) < x < p+(qoa(x)) for all x in Y.
(3) 'PG(x)< b iffx < 97+(b).
(4) a C b iffPa(b) = a A b.
(5) aIbiff 97.(b)= 0.
(6) b < a iffpa(b) = b.
(7) (a preserves any existing suprema.
(8) (Pa(O)= 0 and pa(l) = a.
(9) 9+(0) =a' and Ta+(l) = 1.
(10) a A b < Ta(b) < a.
(11) a Cb iff JPamb = TPbzPa iff PaAb = (Pa(Pb.
In this paper, which is meant to be the first of a series, we consider four kinds of
objects:
(1) Binary connectives -< such that (Pa(P -< q) < b iff p -< q < a < b. These
will be termed "strict implications."
(2) Subsets / of 2' such that V0 [0,e] = V{xE Y I x e f and x < e} exists
and belongs to f for each e in .2' These shall be called "implicative sets." Such sets
have been called dual relatively complete in [4].
(3) Interior operators on .; that is, mappings L: Y --?. . where
(i) Le < e for all e (decreasing),
(ii) if e < f, then Le < Lf (isotone),
(iii) LLe = Le for all e (idempotent).
(4) Closure operators on 2; that is, mappings M: Y -> Y where
(i) e < Me for all e (increasing),
(ii) if e < f, then Me < Mf (isotone),
(iii) MMe = Me for all e (idempotent).
480 L. HERMAN AND R. PIZIAK

The purpose of this paper is to establish one-to-one correspondences between the


above four kinds of objects. In a subsequent paper, we shall use these results to
investigate modal logics on orthomodular bases which generalize systems that
include S4.

?2. Strict implication. We begin our study with the notion of a "strict implica-
tion."
2.1. DEFINITION. A binary operation <: Y x . -* ' such that
(SI) p-q < a<b iff 9)a(p -<q) < b
is called a strict implication connective on J.2 The image of -< is called the set of
"<-implications" and an element of this set is naturally called a "<-implication."
Note that if < is a strict implication and if (p -< q) C a, then (p < q) A a < b
iff(p < q) < (a < b). If . is Boolean, then (SI) states thatx A a < b iffx < a < b,
whenever x is an implication. This is a weakened version of the pure strict
implicational fragment of S4.
2.2. ExAmPLs. (1) Define a < b = a-b = a' v (a A b). This is the "Sa-
saki hook" introduced in [5], and referred to in the introduction. Note that
a Z' b = +(b). Thusx < a - biffpa(x) < b. Hence,p - q ? a = biff pa(p = q)
< b and we see that = is a strict implication. In a sense, this is the most funda-
mental of all the strict implications, asp < q < a < b iff p < q < a : b. We note
then also that every orthomodular lattice carries at least one nontrivial strict
implication connective.
(2) Define a < b = a =t b = 1 if a < b and 0 otherwise. This is the "trivial
hook." First supposep =t q < a Dt b. Show 9Pa(p=ztq) < b. If p =t q = 0, then
9a(P tq)=a(?) =O < b. If p tq= 1, then a- b= 1 so a < b. Also
Pa(P =t q) = TPa(l)= a < b. Conversely, suppose TPa(p t q) < b. Show p =t q <
a =t b. If p t q = 0, there is nothing to prove. So, supposep t q = 1. Then
b > 9(P =t q) = 'Pa(l) = a, so a < b; whence a =t b = 1 and p =t q < a =t b.
Thus :t is a strict implication. Hence we see an orthomodular lattice can carry
more than one strict implication.
(3) Let . be complete. Let C(.) denote the center of 2, that is, the set of all x
in . such that x Cy for all y in Y. For a, b in 2, define a <b = a b=
V{c e C() I c A a < b}. Suppose p zc q ? a z'b. Then Pa(P zq) =a A
(P =, q) < a A (a =, b) < b. Conversely, suppose cPa(p =c q) < b. Now p =, q e
C(4), so that (p z? q) A a = 9Pa(pz>cq) < b. Thus p '0 q belongs to
{C e C(Z) I C A a < b}
and so
p 'c q < V{Ce C(2) I c A a < b} = a b.
(4) Define a < b = a' V b. The reader will verify that a <b is a strict implica-
tion iff .2 is a Boolean orthomodular lattice.
Throughout the rest of this section, let -< be a fixed strict implication connective
on Y.
2.3. THEOR. (1) a A (a < b) < b.
(2) a A (a -<b) < opa(a-<b) < a A\.b < b.
MODALPROPOSITIONALLOGIC ON AN ORTHOMODULARBASIS 481

(3) a< b is the largest -<-implicationundera Z) b.


(4) p <q < a <aforallp,q,ain .
(5) a < a = b -< b for all a, b in .; thus there is a largest <-implication. Denote
this element by I < 1.
(6) If a ? b, then a<b = 1 <1 and (a <b) <c < cfor all c; thus a < b
implies p < q < a -< b for all p, q in Y. Also we have (p <p) A q < q for all
p, q e Y
(7) p <q < biff (p-<q)-<b = 1 < 1.
(8) a 0 < a'foralla in Y.
(9) If b < c, thena <b < a<cforallamin .
(10) If a C (b A c) and a < b, then b < c < a-< c.
(11) a-<b=a<<(qA b).
(12) Ifb < a <b, thenaCb.
(13) (a < b) Cc iff a b < c < (a b).
(14) a <(a < b) < a -<b.
(15) (a -b)Caifa b=a-<(a<b).
(16) (a -< b) v )a < c) < a < (b v c).
(17) a<(b A c) < (a<b) A (a'<c).
(18) 1 a < aforalla.
(19) If aA b=0,thena<b < a';inparticulara~a' < a'.
(20) b'F< a=(a.v b) < bfor alla, b.
(21) If a Cb, then a <b = b' <a'.
(22) a-< 0 I <a7'or all a.
(23) p <q < a' iffp <q < a .< 0.
(24) If aCb,thena (b<c) < (aA b)<cforallc.
(25) If a < b, then a<(b<c) < a < cforallc.
PROOF. (1) and (2). We always have a A (a < b) < 9p.(a -< b) < a. But
a < b < a < b means pa(a-< b) < b; so we can strengthen the above to
pa(a < b) < a A b.
(3) By (2),qa(a -< b) < b; so a < b < q+(b) = a : b. Let x be any implication
under a D b. ThenTx) < Tpa(a= b) < b. Thus x < a < b.
(4) Now TpA(x) < a for any x; sO tpa(p-< q) < a. Thus, by (SI), p < q < a < a.
(5) is clear from (4).
(6) Let a < b. Then TAx) < a < b for any x. Thus Tp(1 -< 1) < b; so, by (SI),
1 < 1 <.a -< b. Thus 1 -< 1=a < b in view of (5). Also, (a -< b)< c
((a < b) < c) A (a -< b) ? C.
(7) Suppose x -< b =1 < 1 and assume x is an implication. Then x < 1 < 1,
so xC (I < 1) andso x-=x A (1<1-Xl<1 XX<b) < b-
(8) Let b = 0 in (3). Then a <0 < a 0 = a'.
(9) Letb < c.Thenq94(a-<b) < b < c; soa-<b < a< c.
(10) Suppose a C (b A c) and a < b. Then

a(b< c) < Tpa(b' c)-C


= (bTV (b A c)) = p(b') V pa(bA C)
= 0 V PA
p(b c) = pa(b A c) = a A b A C < C.

So, by (SI), b< c < a Kc.


482 L. HERMANAND R. PIZIAK

( 1) Now a Ab < b; so a < (a A b) < a < b. But T(a < b) < a A b; so


a < b < a < (a A b). Hence we have equality.
(12) Suppose b < a-< b. Then b < a z b, so a A b?a(b) ' ' Ta(a = b)-
a A (a z' b) < a A b. Thus pa(b) = a A band hence a Cb.
(13) Suppose (a < b) C c. Then Tp,(a-< b) = c A (a < b) < a < b, so a < b <
c < (a -< b). The converse follows from (12).
(14) Now (pa(a < (a -< b)) < a A (a < b) < b, so, by (SI), a < (a -. b) <
a < b.
(15) follows from (13) and (14).
(16) rpa((a<b) v (a <c)) -a(a
= = b) v 9'a(a < c) < b v c, so, by (SI),
v
(a- b) (a <c) < a< (b v c).
(17) By(9),a<(b A c) ? a-.<b and a<(b A c) < a<c, so a-< (b A C) <
(a.< b) A (a-< c).
(18) is clear from (1).
(19) Suppose a A b-0. Then a< b < a D b=a' v (a A b) = a'.
(20) First 9P)vb(b < a') a 9eavb(b = a') = 9Pavb(b V (b' A a')) = 9)avb(b) v
9avb(b v a)' = b v 0 = b. Thus b'< a' < (a v b) < b.
Conversely, 9)b'((a V b) < b) < pb,((a V b) D b) = 4pb,((a' A b') v b)-
9b'(a' A b') V 9b'(b) = (a' A b') v 0 = a' A b' < a'. Thus (a v b) -< b < b' < a'.
(21) Suppose a C b. Then a' C b'. It suffices to show a.< b < b' < a'. But
9Pb'(a < b) < 9Pb'(a = b) = 9b'(a V (a A b)) = 9Pb'(a') V 9pb,(a A b) = 9b'(a') = b A
a' < a'. Thus, by (SD),a -< b < b' < a'.
(22) is clear from (21).
(23) If p -< q < a', then TAP -< q) < Ta(a') = 0; sop -< q < a< 0. Conversely,
if p-<q < a<0,thenp.<q < a.<0 < a : 0 = a'.
(24) Suppose a C b. Then paAb(a < (b < c)) = pba(a < (b < c)) < rPb(b < c)
< c using (2). So, by (SI), a < (b < c) < (a A b) < c.
(25) is clear from (24).
Among the strict implication connectives, the Sasaki hook may be characterized
in numerous ways as the next result shows.
2.4. COROLLARY. Each of the following conditions is equivalent to.< being the
Sasaki hook:
(1) a <0 = a'for all a.
(2) 1 <b=bforallb.
(3) Each member of ? is a -<-implication.
(4) If aCb, thenb < a.<b.
(5) a' < a -< bfor all a, b.
(6) If b < a, then a<b = a' v b.
(7) Ifa l b then a b = a'.
(8) If a < b, thenb = a'-< b.
(9) b = (a A b)'-< bfor all a, b.
(10) a v b = a'-< (a v b) for all a, b.
(11) a = a' <afor all a.
(12) a A (a-< b) = a A b.
PROOF. Simple computations using the Foulis-Holland theorem show that the
Sasaki hook has all these properties; the reader may consult [5] for some of the
MODAL PROPOSITIONALLOGIC ON AN ORTHOMODULARBASIS 483

details. The equivalence of (1) and (2) is part (21) of the previous theorem. Clearly
(2) implies (3). If (3) holds, then a = b is itself a <-implication for each pair a and
b. So, by Theorem 2.3 (3), a -< b = a b. These remarks show that conditions
(1), (2), and (3) are individually equivalent to < = :.
(4) implies (2). For any b, 1 C b implies b 1 -< b. But 1 v< b < b always; so
I < b = b.
(5) implies (1). a' < a - < a'; soa < = a'.
(7) implies (1). a I 0 implies a< 0 = a'.
(8) implies (2). 0 < b implies b 0 ' < b I1 < b.
(8) implies (10). a < a v b implies a v b =a'-< (a v b).
(9) implies(8). a < bimpliesa'-<b = (a' v b')-<b = (a A b)'-<b = b.
(10) implies (9). b = (a A b) V (b A (a A b)')
= (a A b)' <(a A b) V (b A (a A b)') = (a A b)' -< b.
(11) implies (8) is clear.
(12) implies (2). 1 < b=1 A (I <b) = A b=b for all b.
Finally we show if (6) holds, then < = :. This is true since a A b < b implies
a < b = a -< (a A b) = a' v (a A b) = a = b.
2.5. DEFINITION.For any subset f of 2', let f(0, e] = {x e I2 xe f and
x < e}.
2.6. LENA. Let < be a strict implicationon 2. Define f/< = {x E 2 I pa(x) < b
implies x < a < b for all a, b in Y}. Then f/< is precisely the set of all <-implica-
tions and, for each e in 2, V/d<o(0,e] exists and belongs to f<.
PROOF. By the very definition of strict implication, 4p(p A< q) < b implies
p < q < a -< b for all a, b in 2, Thus every implication p.< q belongs to f<.
Conversely if x is in /<, then pw(x) = 0 < x implies x < x' < x. But we always
have x' < x < x by Theorem 2.3(19). Hence x = x' < x and so x is an implication.
Now consider any e in 2. For any x in /<[0, e], pl(x) = x < e implies x < 1 <
e. Thus 1 -< e is an upper bound for /f<[0, e] in 2. But, by Theorem 2.3(18), 1 -< e
belongs to /<[0, e]. Hence V/do<[, e] = 1 -< e belongs to ,<<since /< is the set
of -<-implications.
2.7. LMMA. Let -< be a strict implicationon 2. Define Le = 1 <e and Me =
(e < 0)'. Then
(1) Le < e for all e.
(2) e < f implies Le < Lf.
(3) LLe = Le.
(4) Me = (Le')'.
(5) e < Me for all e.
(6) e < f implies Me < Mf.
(7) MMe = Me for all e.
(8) 1.< is the set of fixed points of L and the set of orthocomplementsof fixed
points of M.
(9) L(a D b) = a< b.
PROOF. (1) By Theorem 2.3(18), (1) is clear.
(2) If e < f, then by Theorem 2.3(9), 1 < e < If.
(3) By Theorem 2.3 (14), 1 < (I -< e) < 1 -< e. Also p1(1 < e) = A (1 -< e) <
<e, so by (SI), 1 -< e < 1 -< (1 < e).
484 L. EERMAN AND R. PIZIAK

(4) Using Theorem 2.3(22), Me = (e < 0)' = (1 -< e')' = (Lel)'.


(5) through (7) follow by dual arguments of (1) through (3).
(8) It follows from (1) and Theorem 2.3(13) that L(a < b) = a < b. Thus every
implication is a fixed point of L. Conversely, if x is a fixed point of L, then x = Lx =
1 -< x is an implication. The rest is clear.
(9) Now a < b < a - b, soa-< b = L(a< b) = 1 -<A (a b) < 1 -< (a ' b) =
L(a = b). Thus a Ab < L(a v b) < a v b. However, a <b is the largest im-
plication under a v b; so a -< b = L(a = b).
We see that L is an interior operator and M is its canonically associated closure
operator.

?3. Implicative sets In this section, the symbol 2' continues to denote an
orthomodular lattice.
3.1. DEFINrUON.A subset f of 2 is called an implicativeset if, for each e e 2,
the set 0f9O,e] has a largest element.
The sets f = ', / = {0}, and / = {0, 1} are easily seen to be implicative sets.
Any section [0, a] is an implicative set. A less trivial example of an implicative set
may be constructed as follows: Let a be any element of 2' and let f be the set of
elements in 2 compatible with a; then, for each e e 2, (e A a) V (e A a') is the
largest element of /[0, e]. Lemma 2.6, of course, is the motivation for studying
implicative sets.
3.2. NtoposmoN. If f is an implicativeset, then:
(1) f is subcomplete. In particular if x, y E f, then x v y E/.
(2) 0 e f.
(3) Vf exists and is the largest element of f.
(4) f is a section iff f is an order ideal.
PROOF. (1) Let M c f and suppose x = VM exists in S. We want to show
that x e f. For this, it suffices to show x = Vf[0, x]. But clearly, V/[0, x] < x,
and m in M implies m ? V/[0, x]. Hence, x = VM < Vf/[0, x] < x and so
x = V/f[o, x].
(2) is clear.
(3) By (1), Vf[0, 1] exists and is the largest member of f. Clearly, Vf/[0, 1] =
V/.
(4) Clearly, if f is a section, then it is an order ideal. Conversely, if f is an
order ideal, then Vf e f implies f = [0, Vt].
Examples of implicative subsets are particularly easy to find when 2' is itself
complete. For, in this case, the implicative subsets clearly coincide with the sub-
complete subsets. Thus, for example, the center of 2 is an implicative subset.
Moreover, the intersection of any collection of implicative subsets of 2' is again
an implicative subset, since it is easy to verify that the intersection of subcomplete
subsets is subcomplete.
Although every implicative subset of 2 is subcomplete, the converse is false, as
the following example shows: Consider the set of natural numbers N. Let
9f(N) be the lattice of finite and cofinite subsets of N. This is a Boolean and hence,
a fortiori, an orthomodular lattice. The operations, of course, are the usual power
set operations. Let E denote the set of even natural numbers and 0 the set of odd
MODAL PROPOSITIONALLOGIC ON AN ORTHOMODULARBASIS 485

natural numbers. Let f = {O. {2}, {2, 4}, {2, 4, 6}, **}. Clearly f is not an
implicative set since, for instance, f does not have a largest element. However, we
claim / is subcomplete. For, let {Je} a f and suppose J = UJAexists in gfi1'(N).
Now, J cannot be infinite, for otherwise, 0 = E' c J' and so J' is also infinite.
Hence J is finite and so there are only finitely many Jt's. Thus J is the largest of the
J. and so is itself in /.
Next, we show how to construct a strict implication from an implicative set.
3.3. LEMMA.Let / be an implicative set. For a, b in .2' define a .<,f b =
V/[O, a b]. Then -<,> is a strict implication on . and / = /v,. Moreover,
/ = {x E.2' x = Vf O,x]}.
PROOF. First we show -<e is a strict implication. Let 4p,(p<jf q) 5 b. Then
p <f <q a = b. We must show p <.q < am< b. For this, it suffices to show
f[O,pDq] < fTO,aDb]. Let xe/f[O,p=q]. Then x Vf[Op q]=
p< q < a D b, so x e f [O,a D b]. Next let p <. q < a .< b. But then
p < q < a <.<fb < a z b since a <b is a join of elements undera > b. Thus
Pa(p <.f q) < b.
Next, we show f = 1<.,. Let x e /. Consider any a, b in 2. Then 4a(x) < b
implies x < p+(b) = a : b, so xef[O,a a b]. Thus x < V/[O,a D bl =
a -<f b. Hence, by Lemma 2.6, x E f1<f. Now let x e f/S.c. Then, again by Lemma
2.6, for any a, b, we have 9Da(X) < b implies x < a -<. b. In particular, p (x)-? 0 <
x implies x < x' -<r X. Thus x < x' -< x = V/f[0, x' v xI = V/LO0,x] C x.
Hence x = V/[O, x] and so x e f and the remainder of the lemma is clear.
3.4. TEOREM. There is a one-to-one correspondencebetween implicative sets in
2' and strictimplications on 2' given by ..< f/< andf of f.
PROOF. Now a<< <b = V/<[, a Db] = a < b by Lemma 2.6. Also, by
Lemma 3.3, / =
3.5. LEMMA.Given an implicativeset f defineLie = V/TO,e]. ThenLF is an
interior operator on 2' whose range is f. Moreover, Life = 1 -<.f e for all e.
PROOF. Clearly, L is isotone and decreasing. By the definition of an implicative
set, for any e in 2, Lie is the largest member of f under e. Thus the range of L is
contained in /. Conversely if x e /, then x is itself clearly the largest member of f
under x and so x = Lifx; thus x is in the range of Li. By definition, LiLie =
Lif(Life) is the largest member of / under Lie. But Lire is, as we have just seen,
itself in f. Thus LiLife = Lie. So LI is an interior operator whose range is f.
Finally, 1 -<? e = V/[O, 1 v e] = V/f [, e] = Lie for any e.
3.6. LEMMA.Given an implicative set f define Mie = A(/' n [e, 1]). Then
MS is a closure operatoron 2' whoserange is f '. Moreover,Mie = (e -<.r 0)' and
LIe = (Mie')'.
PROOF. By the generalizedDe Morganlaw, Mie = A/'[e, 11 = (V/O[, e'])' =
(Lie')'. As is well known, this effects a correspondence between closure operators
and interior operators. The properties of Mi cited here are simply dual to those of
L proved in Lemma 3.5 above.

?4. The modaloperators. Let L denote an interioroperator on the orthomodular


lattice 2 and M a closure operator on 2. Recall the model we have in mind is from
modal logic where L is the "necessity" and M the "possibility" operator.
486 L. HERMAN AND R. PIZIAK

4.1. PRoPOSITION. Let ef e Y. Then:


(1) L(Le v Lf) = Le v Lf; L(Le A Lf) = L(e A f).
(2) M(Me A Mf) = Me A Mf; M(Me V Mf) = M(e v f).
PROOF. Sine Lx < x for all x e 2, we clearly have L(Le v Lf) < Le v Lf
But Le < Le v Lfand L isotone and idempotent implies Le = LLe < L(Le v Lf);
similarly Lf < L(Le v Lf). Thus Le v Lf < L(Le v Lf) and we have equality.
The proof of the secondpart of (1) is similar.
Statement(2) is dual to statement(1).
As remarkedin the proof of Lemma3.5, a one-to-onecorrespondence between
closureoperatorsand interioroperatorson any orthocomplemented latticeis given
as follows: GivenL, one definesM by Me = (Le')' and checksthat Mis a closure
operator.On the other hand, given M, defineLe = (Me')'and againit is routine
to checkthat L is an interioroperator.
4.2. LEMMA.Let L be an interior operator on 2. Define a -< b = L(a = b).
Then -<L is a strict implication on 2. Moreover, Le = 1 -< e for all e in Y.
PRooF. Letrpa(p-<Lq) ? b.Thenp-<Lq < p+(b) = a D b.ThenL(p-<Lq) <
L(a D b) = soPp<Lq = L(p - q) = LL(p = q) = L(p-<Lq) < a-l<Lb.
a-<Lbb
Conversely,letp <L q < a <Lb. We mustshow(PA L q) < b. But9.(p <L q) <
(pa(a<L b) = TP,(L(a= b)) < pa(a : b) s b. Hence, <L is a strict implication.
Note 1 <L e L(l D e) = Le.
We remarkthat Lemma4.2 is the motivationfor the terminology"strict im-
plication."
4.3. THEoRM. There is a one-to-one correspondencebetween strict implications
and interioroperatorson .. More explicitly, to < we assignL< and to L we assign <z,.
PROOF. a-<Lb = L<(a D b) by Lemma4.2; L<(a > b) = a < b by Lemma
2.7(9).So a-<L- b = a < b. Also, L<Le = 1 <Le by Lemma2.7 and 1 <Le = Le
by Lemma 4.2. Hence L<Le = Le for all e in .2.
In view of Theorem4.3 and the canonical correspondencebetweenclosure
operatorsand interioroperators,we have
4.4. THEORM. There is a one-to-one correspondencebetween strict implications
and closure operators on 2. More explicitly, given < define M<e = (1 < e')' as in
Lemma2.7 andgiven M definea <M b = (M(a - b)')'. Moreover,Me = Af'[e, 1].
Finallywe tie in implicativesets.
4.5. LEmmA.Let L be an interior operatoron Y. Let fL be the range of L, which
is also the set offixedpoints of L. Then fLis an implicativeset. Infact, for each e in
., Le = VfL[O,e].
PROOF. Let x e- f 0, e]. Then x e f and x < e, so x = Lx < Le. Thus Le is
an upperboundof f[0, e]. Letf be anyupperboundof f[O, e]. NowLe e f[O, e],
so Le < f and hence Le = LLe < Lf < f.
4.6. THoNM. There is a one-to-one correspondencebetweenimplicativesets and
interior operators on .2. More explicitly, to the implicative set f, we assign Lf
definedby Lfe = \//[O, e], and to the interioroperatorL we assign the implicative
set which is its range.
PROOF. LiLe = VfL[O,e] by Lemma3.5 and V/L[O,e] = Le by Lemma4.5; so
LI~ze =Le. Also /5rLjf = {x I Lfx = x} by definitionand this is exactly f by
Lemma3.5.
LOGICON AN ORTHOMODULAR
MODALPROPOSITIONAL BASIS 487

In view of the canonical correspondence between closure operators and interior


operators we have
4.7. TBEORM. Thereis a one-to-onecorrespondence betweenimplicativesetsand
closureoperators.

?5. Summary. We have now established one-to-one correspondences between


strict implications, implicative sets, interior operators and closure operators.
These correspondences may be succinctly summarized in this way:
(1) Given -<, define
/ ={p.< q Ip< q < a = b impliesp.< q < a < b for all a, b},
Le= I-< e, Me =(e -< O)'.
(2) Given /, define
a< b = V fO, a =>b],
Le= V f/O,e], Me= A'[e, 1] = A {yIy' ef andy > e}.
(3) Given L, define
f = L(Z1), as b = L(a b), Me = (Let)'.
(4) Given M, define
f = (M(Z))', a < b = (M97.(b'))', Le = (Me')'.
A pictorial summary is given by the following:
Le= V /[O, e] (OPAL)+-.(CAM) Me =(Le')'
= (Me')' I = A/Ne,
A 1]

= 1 < =e)'
e ~~~~~~~W,

/ = {x 4p(x)
I b implies a-< b = Vf/9,a a b]
x < a b for all a, b} = L(a = b)
=L(S) = (Mq9(b'))'

As a simple illustration of these correspondences, take f = [0, e]. One computes


easily that Lx = X A e and Mx = x v e', while a -< b = (a = b) A e. In the
special case where e = 1, so that/ = 2, we get that L and M are the identity maps
and a< b = a : b. If . is complete, a more interesting example is obtained by
taking / to be the center of 2. Then Lx is the central kernel of x and Mx is the
central cover of x, while a < b = a =, b. In other words, Lx is the largest central
element under x and Mx is the smallest central element above x.
As an illustration of the usefulness of these results, we offer the following easy
5.1. THEoREM. Thefollowingare all equivalent:
(1) Ie/.
(2) p < q iffp <q = for allp andq in Y.
(3) LI = 1.
(4) MO = 0.
PROOF. The equivalence of (1), (3) and (4) is clear. Now if (2) holds, then
I = 1 < 1 E /, since / is the set of < -implications. Conversely, suppose (1) holds.
488 L. HERMANAND R. PIZIAK

Then -< q = 1 impliesp = p A 1 = p A (p -< q) < q by Theorem 2.3(1); on the


otherhand, ifp < q and 1 e f,then 1 = p = qimplies I < p < q andsop < q = 1.
Thus (1) implies (2).
More delicate theorems of this type will be discussed in a subsequent paper.

BIBLIOGRAPHY

[1] G. BIRKHoFF,Lattice theory, American Mathematical Society Colloquium Publications,


vol. 25, American Mathematical Society, Providence, R.I., 1967.
[2] D. J. Fouus, A note on orthomodular lattices, Portgalae Mathemadca, vol. 21, Fasc. 1
(1962).
[3] R. J. GREEcHmand S. P. GUDDER,Quantum logics (preprint).
[4] P. HALMOS,Algebraic logic, Chelsea, New York, 1962.
[5] L. HERMAN,E. MARSDENand R. PizAx, Implication connectives in orthomodular lattices,
Notre Dame Joural of Fornum Logic (to appear).
[6] G. E. HUGHESand M. J. CRESSWELL, An inrodtion to modal logic, Methuen, London,
1968.
[7] M. F. JAowrrz, Residuated closure operators, University of New Mexico Technical
Report No. 29, 1965.
[8] J. JAY ZEMAN,Quantum logic with implication, Notre Dame Joeunal of Formal Logic (to
appear).
[9] , Modal logic: The Lewis modal systems, Clarendon Press, Oxford, 1973.

KANSAS STATE UNIVYERSrY


MANHATrAN, KANSAS 66506

UNIVERSiTY OF FLORIDA
GAINESVILL, FLORIDA 32601

You might also like