Basic Hoops: An Algebraic Study of Continuous t-Norms

Author(s): P. Aglianó, I. M. A. Ferreirim and F. Montagna

Source: Studia Logica: An International Journal for Symbolic Logic, Vol. 87, No. 1 (Oct.,
2007), pp. 73-98
Published by: Springer
Stable URL: http://www.jstor.org/stable/40210800
Accessed: 17-08-2016 03:13 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/40210800?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

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 [email protected].

Springer is collaborating with JSTOR to digitize, preserve and extend access to Studia Logica: An
International Journal for Symbolic Logic

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
 P. Agliano Basic Hoops: an Algebraic
I. M.A. Ferreirim Study
F. MONTAGNA
of Continuous t-norms

Abstract. A continuous t-norm is a continuous map * from [0, 1]2 into [0, 1] suc
([0, 1], *, 1) is a commutative totally ordered monoid. Since the natural ordering o
is a complete lattice ordering, each continuous t-norm induces naturally a residu
- > and ([0, 1], *, - >, 1) becomes a commutative naturally ordered residuated mono
called a hoop. The variety of basic hoops is precisely the variety generated by all a
([0, 1], *, - >, 1), where * is a continuous t-norm. In this paper we investigate the st
of the variety of basic hoops and some of its subvarieties. In particular we prov
complete description of the finite subdirectly irreducible basic hoops, and we show
the variety of basic hoops is generated as a quasivariety by its finite algebras. We
these results to Hajek's BL-algebras, and we give an alternative proof of the fact
the variety of BL-algebras is generated by all algebras arising from continuous t-no
[0, 1] and their residua. The last part of the paper is devoted to the investigation
subreducts of BL-algebras, of Godel algebras and of product algebras.

Keywords: basic hoops, continuous t-norms, subreducts of BL-algebras.

A continuous t-norm is a continuous map * from [0, 1]2 into [0, 1] suc
([0, 1], *, 1) is a commutative totally ordered monoid. There are thre
damental continuous t-norms: the Lukasiewicz t-norm defined by x *
max(x + y - 1, 0), the Godel (or lattice) norm x *g y = x A y and the prod
norm x*pj/ = xy. Indeed it is known ([24, 35]) that, up to isomorph
every continuous t-norm behaves locally as one of the above.
Since the natural ordering on [0, 1] is a complete lattice ordering,
t-norm induces naturally a residuation, or an implication in more l
terms, by x - > y = sup{z : z * x < y}. The implications associated
three fundamental norms are:

x-^l!/ = min(y - x + 1, 1)
if x < y
y otherwise

Research partly supported by research projects Praxis 2/ 2.1/ MAT/73/94 "Algeb

e Matematicas Discretas" and POCTI/0143/2003 "Fundamental and Applied Algebr

Presented by Daniele Mundici; Received November 14, 2006

Studia Logica (2007) 87: 73-98

DOI: 10.1007/S11225-007-9078-1 ©Springer 2007

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
74 P. Agliano, I. M.A. Ferreirim and F. Montagna

fl ifx<y
and x ^P y = < . .
\y/x otherwise .

It is clear that the residual - > of a continuou

x - » x = 1

a; -+1 = 1

1 -> rr = x

x - » y = 1 and y - ► x = 1 imply x = y.

Hence the variety V% generated by any class X of algebras of the form

([0, 1],*,- ►,()), where * is a continuous t-norm and - > is its residual, is
ideal- determined [23] and therefore it is the equivalent algebraic semantics
of its assertional logic (see [3], Section 3). In particular, there is a propo-
sitional calculus naturally associated to V% that is strongly complete with
respect to Vx-
In his important monograph [24], Hajek considers in detail the three
relevant cases and provides an axiomatization for the corresponding va-
rieties of algebras and logics. Algebras in the variety WA generated by
([0, 1], -l, - ►LjO, 1) are known as Wajsberg algebras [22] or MV-algebras [13]
and the propositional calculus of which they constitute a complete semantics
is the Lukasiewicz many- valued logic [13]. Algebras in the variety %A gener-
ated by ([0, 1], «g, - ►g? 0, 1) are called Godel algebras and form an equivalent
algebraic semantics for Dummett's Logic [17], also called Godel Logic, the
infinite- valued version of the infinitely many finitely- valued systems, which
Godel considered in his proof that Intuitionist Logic is not finitely- valued.
The variety 7A of product algebras is generated by ([0, 1], -p, - »p, 0, 1), with
product logic [25] as its associated propositional calculus. Last but not least,
Hajek introduces the variety ££ of BL-algebras and calls basic logic the as-
sociated propositional calculus; then he formulates the conjecture that this
variety is in fact generated by all algebras of the form ([0, 1], *, - >, 0, 1), where
* is a continuous t-norm on [0, 1]. This conjecture has been verified in [14];
we give a simpler proof of a stronger statement.
One of the relevant algebraic aspects of a continuous t-norm on [0, 1]
is the fact that the associated monoid is residuated. Residuated (partially
ordered) monoids have long been considered of interest by algebraists, start-
ing from the classical example of the lattice-ordered monoid of the ideals of
a ring with unit. In particular Bosbach [12] devoted several papers to the

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
Basic Hoops: an Algebraic Study of Continuous t-norms 75

study of left-complemented monoids, i.e. the residuated monoids in w

the underlying ordering is natural:

x < y if and only if 3 z (x = zy).

Bosbach's work seems to have been the main source of inspiration for Biichi
and Owens' research on commutative complemented monoids, which they
called hoops. They prepared a manuscript entitled "Complemented Monoids
and Hoops" in the middle seventies but, mainly because the manuscript
was never published, their ideas caught on slowly. Blok and Pigozzi in
[10] applied these ideas in the study of hoops with dual normal operators,
which are a generalization of Boolean algebras with operators, but the first
systematic study of the structural properties of hoops appeared in Ferreirim's
thesis [19]. Some of the results obtained there can be found in two joint
papers with Blok [5] and [6]; in particular the description of subdirectly
irreducible hoops ([6, Theorem 2.9]) will play a crucial role in this paper.
Since the ordering induced by the residual of any continuous t-norm is the
natural ordering on [0, 1], any algebra of the form ([0, 1], *, - ►, 0, 1) is a hoop
that is also bounded, i.e., has a smallest element 0. Hence all the varieties we
have considered so far are varieties of (bounded) hoops. This suggests the
possibility that the structure theory of hoops can be used to achieve a better
understanding of these varieties (and of the logics involved); conversely the
class of varieties arising from these logics might shed more light on the be-
havior of other classes of hoops (and their implicative subreducts). The aim
of this paper is to show that this enterprise can be successful. In Section 1
we investigate the variety of basic hoops, i.e., the variety of hoops naturally
associated with basic logic; we clarify its relationship with the variety of
basic BL-algebras and we characterize completely its finite subdirectly irre-
ducible members in terms of ordinal sums of hoops. Then we do the same for
its implicative subreducts. In Sections 2 and 3 we proceed to show that the
variety of basic hoops is generated as a quasivariety by its finite algebras and
the same holds for its implicative subreducts. As a by-product we are able
to give a new proof and a slight improvement of the completeness result in
[14]. In Section 4 we point out that the variety of G-hoops, naturally associ-
ated with Godel logic, consists of well-known objects and we characterize its
implicative subreducts. Sections 5 and 6 are devoted to product hoops, i.e.,
hoops coming from the product t-norm. In Section 5 we study the structure
of product hoops and their implicative subreducts, while in Section 6 we
describe completely the lattice of subvarieties of product hoops.

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
76 P. Agliano, I. M.A. Ferreirim and F. Montagna

1. Basic hoops and BL-algebras

As mentioned in the introduction, a thorough algebraic study of the variety

of hoops may be found in [6]. We start this section by recalling some known
definitions and results. Then we study in detail the variety of basic hoops,
which is generated by all totally ordered hoops.
A hoop is an algebra A = (A, •, - ►, 1) such that (A, •, 1) is a commutative
monoid and for all x, y, z 6 A

1. x-*x = l

2. x(x -► y) = y{y -► x)

3. x -► (y -+ z) = xy -► z.

If A is a hoop, define a0 = 1, a - ► b = b and a ^i b = a

any natural number n. Then an - > 6 = a A 6 for all n.
If A = (A, •, - >, 1) is a hoop then the binary relation de
and only if a - ► 6 = 1 is a partial order on A with respect to
is a naturally ordered residuated commutative monoid, or
pocrim. The residuation is given by ab < c if and only if a
A bounded hoop is an algebra A = (A, -, - ►, 0, 1) such that
a hoop and 0 < a for all a G -A.
A Wajsberg hoop is a hoop satisfying the identity

(x -> y) -> y « (y -> x) -> x. (T)

Bounded Wajsberg hoops are term-equivalent to Wajsberg alg

[10]; they are also term-equivalent to Chang's MV-algebras [1
CN algebras [33]. It follows from the fact that every W
a {-,-►, l}-subreduct of a Wajsberg algebra [6, Propositio
theory of algebraizable logics [9, Corollary 2.12] that the v
Wajsberg hoops is the equivalent algebraic semantics of the posit
of Lukasiewicz's infinite- valued logic.
Finite Wajsberg hoops will play a crucial role in the se
natural number n, Cn denotes the finite totally ordered W
whose universe is Cn = {1 = a0, a, a2, ..., an} and akam = amin
am = amax(m-fc,o) for 0 < fc, m < n; similarly Wan denotes the
ordered Wajsberg algebra (Cn, •, - ►, an, 1).
A filter of a hoop A is a subset F containing 1 and closed
ment: if a, a - ► 6 G F, then b G F. Given a subset X of a ho

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
Basic Hoops: an Algebraic Study of Continuous t-norms 77

generated by X in A, denoted FgA(X), is

FgA(X) = {beA\3au...,aneX a1-+(a2-+(...(an^b)...)) = l}

= {beA\3ai,...,OneX aia2...On < 6}.

In particular, the principal filter generated by an element a is

FgA(a) = {beA\3n with a A b = 1} = {b € A | 3n with an < 6}.

If 0 is a congruence on A, then 1/0 is a filter. Moreover, the map 9 i - ► 1/0

determines a lattice isomorphism between the congruence lattice and the
filter lattice of a hoop, with inverse map F i - ► Op, where Op = {(a>,b) :
a - ► 6, 6 -» a € F} [12]. The variety of hoops is therefore congruence
regular at 1, with witness term (x - * y) A (y - ► x).
In the next proposition we collect two useful facts concerning the Wajs-
berg hoop ([0, 1], *l, -+L, 1) (resp. the Wajsberg algebra ([0, 1], *L, ->l, 0, 1)).

Proposition 1.1. l.Ifa<b€R, then it is possible to define • and - ► on

[a, 6] in swc/i a way £/&a£ ([a, 6], •, - ►, b) is isomorphic to ([0, 1], *l, -+L? !)•
2. Each Cn is embeddable in ([0, 1], *l? - >l» !)•
5. i. and £. /ioW /or the Wajsberg algebras ([0, 1], *l, - >l, 0, 1) and Wan.

Proof. 1. Define for all u, v e [a, b]

uv = max(tz + v - 6, a) u - ► t; = min(6 - ?x + v, 6).

Then the maps

/(x) = a + (b - a)x for x G [0, 1] g{u) = ^- ^ for u G [a, 6]

are mutually inverse isomorphisms between [0, 1] and [a, 6].

2. The map defined by ak i - ► (n - fc)/n, for A: < n, is an embedding of Cn
into ([0,1], *L,^L,1).
3. Clear. ■

We now tu
known tha
BCK-algeb
Meredith's
form a qua
is any vari

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
78 P. Agliano, I. M.A. Ferreirim and F. Montagna

subreducts of algebras in V, is a variety of BCK-algeb

stated in [11, Section 4, Example III]; it also follows from
We give here a direct proof, based on the observation that,
of hoops V, the class S~*(V) has the filter extension pro
B e S"*(V), A < B and F is a filter of A, then there exis
such that GDA = F (namely G = FgB(F)).
Proposition 1.2. IfV is any variety of hoops, then th
variety of BCK-algebras.

Proof. Recall that the variety of hoops satisfies the follo

troduced by Cornish [16]:

(((X _+ y) -+ y) _+ X) _+ X « (((y ^ X) -+ X)

[5, Corollary 4.8].

Let V be a variety of hoops; it is sufficient to show th
under homomorphisms. Let A be a {- », l}-subreduct
denotes the {- ►, l}-reduct of B, we may write A < B
and let G be the filter of B generated by 1/0; it is easy
FgB(l/0) = FgB^(l/0). Clearly 0 C QG n A2. On th
(a, 6) e 6Gn A2, then ((a -» b) -► b) -+ a 0G ((a -►
and similarly ((6 - ► a) - » a) - » b Oq 1. Thus both expr
l/^flA = GO A = 1/9 (by the filter extension property)

a = 1 -> a 9 (((a -► b) -> b) -► a) -► a

= (((6 -+ a) ^ a) ^ 6) -+ 6 0 1 -+ b = b.

Thus 9 = 9GC\ A2, and therefore A/0 is a subreduct

Identity (J) determines a variety of BCK-algebras [16

the variety 'K'B&X, consisting of all implicative subreduc
WBGX is the variety of BCK-algebras satisfying identiti

(x ^ y) -+ (x -* z) « (y ^ x) -+ (y -^ z) (H)

[5, 20]. A syntactic derivation of (J) from axioms for BCK-algebras tog
with (H) may be found in [34].
It is known that for BCK-algebras identity (T) implies (J). Theref
(T) defines a subvariety of WSeOC, the variety L'hQX of Lukasiewicz
algebras. Lukasiewicz BCK-algebras were introduced by Komori [32],
the name C algebras, to study the implicative fragment of Lukasiewicz ma
valued logic. The variety L'BQX coincides with the variety of implic

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
Basic Hoops: an Algebraic Study of Continuous t-norms 79

subreducts of Wajsberg hoops (algebras) [5] and it plays a relevant r

characterizing subdirectly irreducible members of WBGX [20, Theorem
It is also known that if A is a BCK-algebra that belongs to a varie
of BCK-algebras, then the filter lattice and the congruence lattice of
isomorphic.
The underlying order of a hoop is always a A-semilattice order (w
a A b = a(a -►&)), but not necessarily a lattice order.
When the join of two elements exists, it reflects in the lattice of f
as follows:

Proposition 1.3. Let A be a hoop and a, b £ A. If aV b exists, then

FgA(a V 6) = FgA(a) D FgA(b).

This property is shared by many ordered structures, for example Heyt

algebras and Wajsberg algebras. In a more general context, it was
established for BCK-algebras whose underlying poset is a V-semilattic
Corollary 2], [41, Corollary 3]. However, its proof depends only o
existence of the required join; it appears in full generality in [40].
Next we investigate a class of hoops with a (term-definable) lattice ord
closely related to the logical system introduced by Hajek in his mono
[24]. This propositional calculus, which Hajek called basic (many-v
logic, is proposed as the "most general" many- valued logic with truth val
in [0,1]. The algebraic semantics for basic logic given in [24] consi
BL-algebras.

Definition 1.4. A BL-algebra is an algebra (A, V, A, •, - ►, 0, 1) such t

1. (A, V, A, •, -+, 0, 1) is a bounded residuated lattice;

2. xAy = x(x^>y);
3. {x -+ y) V (y -* x) = 1.

The class of BL-algebras is denoted by !B£.

The binary connective V defined in basic logic (BL) by

ct> V </> := {{<t> -+ i>) -+ V>) A ((V -> <t>) -> 0)

corresponds, via the completeness theorem [24, Theorem 2.3.19], to th

operation on BL-algebras. Thus, join is term-definable using only mee
implication
x V y = ((x -► y) -> y) A ((y -> x) -> x).

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
 80 P. Agliano, I. M.A. Feneirim and F. Montagna

In [24] it was shown that BL-algebras satisfy all axioms of h

a BL-algebra A = (A, V, A, •, - ►, 0, 1), it is appropriate to c
hoop reduct of A.

Definition 1.5. A hoop is called a basic hoop if it is a ho

a subhoop of the hoop reduct) of a BL-algebra. The class
denoted by 3 *H.

The algebraic translation of axiom (A6) in [24, Definit

identity
(a _> y) _> z < ((„ _> x) _> z) -> z. (B)
which holds in every BL-algebra. It follows that every basic hoop also
fies (B). Next we show that identity (B) plays a crucial role in characte
the class of basic hoops. Our main tools are the notion of ordinal sum
well as the characterization of subdirectly irreducible hoops described
Theorem 2.9(iii)]. Recall that, given two hoops A and B, A © B de
their ordinal sum. If A is a subdirectly irreducible hoop then it is of the f
F © S, where F and S are subhoops of A, and S is a subdirectly irred
Wajsberg hoop.

Theorem 1.6. Let A be a hoop. The following are equivalent:1

(i) A is a basic hoop.

(ii) A satisfies the identity

(x^y)^z<((y-+x)-+z)-+ z. (B)

(iii) A is isomorphic to a subdirect product of linea

PROOF. (i)=»(ii). Clear from the remarks following

hoop.
(ii)=> (iii). Without loss of generality, we may assume that A is sub-
directly irreducible. Then, by [6, Theorem 2.9] A decomposes as F © S,
where S is a non-trivial subdirectly irreducible (hence linearly ordered) Wa-
jsberg hoop. By way of contradiction, let a, b 6 A be such that a ^ b and
b j£ a. Then necessarily a, b € F, since S is totally ordered and if x G 5 and
y € F\ {1} then y < x. Let c € 5 \ {1} (such c exists since \S\ > 1). Since
a - > b and b - ► a are in F \ {1}, one has a - ► b < c and b -+ a < c, but
c ^ 1, and (B) fails in A, yielding a contradiction.

1The equivalence between (ii) and (iii) has been considered by Palasinski in [38], with
respect to BCK-algebras.

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
Basic Hoops: an Algebraic Study of Continuous t-norms 81

(iii)^(i). It is enough to prove that every linearly ordered hoop is

If A is a linearly ordered hoop, then 2 © A is a bounded linearly or
hoop in which the lattice join is definable by x V y = ((x - > y) - ► y) A (
x) - ► #)(= max(x,y)), and (x - > y) V (y - > x) = 1. Then, 2 © A
BL-algebra, of which A is a hoop subreduct. ■

Note that if A = (A, •, ->, 0, 1) is a bounded hoop, its co

is the same as the congruence lattice of its hoop reduct (A, •,
consequence of Theorem 1.6 is the following characterizati

Theorem 1.7. The variety ££, of all BL-algebras, is term

variety of bounded hoops satisfying

(x->y)->z<((y^x)-^z)-+ z. (B)

Proof. Given an arbitrary bounded hoop A, satisf

product of subdirectly irreducible factors, each of w
by Theorem 1.6. In A, the operations A and V are
x(x - > y) and x V y = ((x - ► y) - ► y) A ((y
These are the lattice operations corresponding t
(A, •, ->, A, V, 0, 1) a BL-algebra. ■

Recall that an algebra A is finitely subdirectly

family of nontrivial congruences of A has a nont
an algebra is finitely subdirectly irreducible if and
principal congruences have a nontrivial intersecti
In view of the lattice isomorphism between congr
which principal congruences correspond to principal
a hoop A is finitely subdirectly irreducible if and on
principal filters has a non-trivial intersection. In
ordered hoop, it is necessarily finitely subdirec
implies FgA(b) Q FgA(a). Moreover we have the f

Proposition 1.8. A basic hoop is finitely subdirect

if it is totally ordered.

Proof. Let A be a basic hoop and a, b 6 A. Then

a BL-algebra and therefore (a - ► b) V (6 - > a) =
6) fl Fg\(b - ► a) = {1}, by Proposition 1.3. If
irreducible, this implies that either Fg\(a - ► 6)
Hence either a < b or b < a and A is totally ordered. ■

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
82 P. Agliano, I. M.A. Ferrekim and F. Montagna

Ordinal sums can be defined for an arbitrary number

summands. Clearly, the class of totally ordered hoops is c
sums.

Since, for finite algebras, being subdirectly irreducibl

being finitely subdirectly irreducible, and every totally ord
essarily finitely subdirectly irreducible, then every finite to
hoop is subdirectly irreducible. In particular, it follows f
3.13] that each finite totally ordered Wajsberg hoop (alge
isomorphic to some Cn (Wan).
This yields an interesting classification of finite subdi
basic hoops.

Corollary 1.9. A is a finite subdirectly irreducible basic

if there are fc, ni, . . . , n& € N such that

A = Cni 0 Cn2 0 • • • © Cnfc .

Proof. First note that Cni 0 Cn2 0 • • • 0 Cnfc is a (finit

subdirectly irreducible basic hoop. Conversely, let A be a
irreducible basic hoop. We prove the claim by induction
of A. If \A\ = 1, then there is nothing to prove. Suppo
then by [6, Theorem 2.9 (iii)] there is a basic hoop F and
Wajsberg hoop S, with A = F 0 S and \S\ > 1. Since A is
so is F and hence it is finitely subdirectly irreducible (by
Since |F| < r, by the induction hypothesis there exist
such that F = Cni 0 • • • 0 Cnfc_1. Moreover S is a finit
Wajsberg hoop, hence it is isomorphic to some Cnfc. In c

A^Cni0.-.0Cnfc,
as claimed. ■

Since every
see that the
the same as th
bounded hoop
correspond to
to BL-algebra
It follows fr
hoop subredu
isfying (B). T
BL-algebras, l

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
Basic Hoops: an Algebraic Study of Continuous t-norms 83

Proposition 1.10. If V is a variety of BL-algebras, then the clas

of all hoop subreducts of members of V, is a variety of basic hoops.

Proof. If A is a basic hoop and B is a BL-algebra, let us write

to mean that A is isomorphic to a hoop subreduct of B. The class
subreducts of V is clearly closed under subalgebras and direct pro
To show that Sh(V) is closed under homomorphisms, let A € Sh(V
a G Con(A). Then I/a = F is a hoop filter and hence it has the deta
property (i.e., if a, a - » b £ F then b € F). Let B be a BL-algeb
that A <h B, let G be the filter of B generated by F and let 6q
congruence associated to G in B. Then (u, v) € 0q if and only if th
ai, . . . , an € F such that

a>i -> (02 -*...-> (an -> (u -> v)) . . . ) = 1

ai -> (fl2 -♦...-> (an -> (v -> u)) . . . ) = 1

If u, v £ A, then ?z- > v, v - > w € F by the detachment property

(u, v) G a. Therefore a = 0c(^Ax A and hence A/a <h B

Corollary 1.11. [24] The variety of BL- algebras consists entirely o

direct products of totally ordered BL-algebras. Any subdirectly irr
BL-algebra is totally ordered. A BL-algebra is finitely subdirectly ir
if and only if it is totally ordered, hence any finite totally ordered BL
is subdirectly irreducible.

Corollary 1.12. A finite BL-algebra A is subdirectly irreducible if a

if there are fe, ni, . . . , n^ € N such that

A S Wa^ 0 Wan2 0 - - • 0 Wa^ .

2. Generation by finite algebras

In this section we shall show that both "BJi and ££ are generated as qua-
sivarieties by their finite algebras. As a by-product we will obtain a new
proof and a slight improvement of the completeness result [14] mentioned
in the introduction. The results obtained in this section are a refinement
for the varieties £ % and ££ of similar results obtained for a large class of
varieties of hoops - see [6, section 3]. An algebra A in a class % has the
finite embeddability property (FEP) with respect to X if for any finite partial
subalgebra A; of A there exists a finite algebra B € % such that A' is em-
beddable in B. A class X has the FEP if each of its members has the FEP

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
84 P. Agliano, L M.A. Ferreirim and F. Montagna

with respect to X. It is well known that if a variety V h

V = HSP(Vfin), where V/;n is the class of all finite memb
Theorem 4]). This argument was extended to show that,
then V = SPPu(Vfin), i.e., V is generated, as a quasivariet
members (see [6]).
In [6, Lemma 3.7] the authors proved that if the class
irreducible members of V has the FEP, so does V. Then t
that both the variety of hoops and the variety of Wajsb
Wajsberg algebras) have the FEP, hence WM = SPPu{C
WA = SPPu(Wain : n 6 N). In particular they showed t
totally ordered Wajsberg hoops (algebras) has the FEP (see
orem 3.9]). This fact is crucial to the following:

Theorem 2.1. The class of totally ordered basic hoops has

PROOF. Let C be a totally ordered basic hoop and let C b

subhoop of C. Let C = {ci, . . . , Cn}, where c\ > c<i > • • •
by induction on n. If n = 1, then trivially C embeds into
If n > 1, then the finite set {q - » Cj : 1 < i < j < n} has a
say Ci0 - ► Cj0. We claim that any maximal congruence <p se
Cj0 also separates any pair (q , Cj) with i < j. Suppose that
(q --> Cj, 1)6<^ and, since F = 1/tp is a filter and q - > Cj <
(cioiCjo) € (p, a contradiction. Therefore (p separates any
i < j and C' embeds into C/</?, which is a subdirectly irr
ordered hoop. By [6, Theorem 2.9] C/(p = D 0 T, whe
totally ordered subhoops of C/<p and T is a Wajsberg ho
Cj/V; by definition of y?, any nontrivial congruence of C/ip
Cj0, hence if C = {ci, . . . , c^}, then C' C\ D has fewer el
Applying the induction hypothesis, there exists a finite total
Di such that C' flD embeds into Di. On the other hand, C
ordered partial Wajsberg hoop, hence by the result in [6
can be embedded into a finite totally ordered Wajsberg h
embeds into Di © Ti and since it is isomorphic as a parti
result follows. ■

Since it is clear
a totally ordered
proof goes throu

Corollary 2.2. T

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
Basic Hoops: an Algebraic Study of Continuous t-norms 85

Combining the results above with the description of finite subdir

irreducible basic hoops and BL-algebras in Corollary 1.9 and Corollar
we get:

Corollary 2.3. The varieties 'BDi and B£ are generated as quasivarieties

by their finite members. In fact

<BM = SPPu(Cni © • - - 0 Cnk : fc,m, . . . ,nfc 6 N)

££ = SPPu(Wani 0 • • - 0 Wanfc : fc,m, . . . ,nk € N)

A well-known result on logical systems due to Harrop [26], applied to

quasivarieties [5, Lemma 3.13] and combined with FEP yields the following:

Corollary 2.4. The quasi- equational theories of'B'K and'BH are decidable.

Since the variety of BL-algebras is the equivalent algebraic semantics of

basic logic, the quasi-equational theory of ££ is (equivalent to) propositional
basic logic, which is therefore decidable. It follows that the consequence
relation in basic logic is decidable. In [24] the author conjectured that basic
logic is "the logic of continuous t-norms" ; in algebraic terms this is equivalent
to showing that ££ is generated as a variety by all algebras of the form
([0,1],*,-*, 1,0), where * is a continuous t-norm on [0,1] and - ► is the
associated residual. This conjecture has been verified in [14]. Using a simple
algebraic argument, we extend the result by showing that ££ is generated
as a quasivariety by all algebras of the form ([0, 1], *, - >, 1, 0).

Theorem 2.5. The variety of basic hoops is generated as a quasivariety by

all algebras of the form ([0, 1],*, - ►, 1), where * is a continuous t-norm on
[0, 1] and - ► is its residual

Proof. By Corollary 2.3 it is sufficient to show that any finite subdirectly

irreducible basic hoop A can be embedded in ([0, 1], *, - >, 1) for some con-
tinuous t-norm * on [0, 1]. By Corollary 1.9 there are fc, ni, . . . , n& € N with
A = Cni © • • • © Cnfc. Moreover, by Proposition 1.1.2, each Cnk is embed-
dable in ([0, 1], *l, ->l, 1); let B* be the copy of ([0, 1], *l, ->l, 1) in which
Cni is embedded. By Proposition 1.1.1, Bf is isomorphic to the algebra
Di = ([oi,k],-,->,&i) where
t-1 . i
ai = - bi=k .
and the operations are defined accordingly. It follows that Cni © • • • ©
is embeddable in D = Di © • • • © D^. The universe of D is clear

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
86 P. Agliano, I. M.A. Ferreirim and F. Montagna

moreover, the binary operation * defined on D, the ordinal su

is clearly a t-norm. Continuity of * is easy to verify: given a
a, 6 € A for some i ^ fc; or a G A> & € Dj for some i, j.
a * 6 = a -i 6 in Z?f and •$ is continuous. In the latter case
with loss of generality, that i > j and 6^1; hence a * 6 =
preserves continuity. ■

Remark 2.6. By Theorem 2.5

basic logic is complete with re
i.e. for every finite set F of
formula <^, one has F h <j> if a
V, A, 0, 1), where * is a contin
evaluation e in B such that e
it follows from Corollary 2.
logic is decidable.

3. The implicative reducts: basic BCK-algebras

In this section we shall characterize the class of implicative subreducts of

basic hoops. By Proposition 1.2, the class of such subreducts is a variety of
BCK-algebras, contained in W&QX.

Definition 3.1. A BCK-algebra is called a basic BCK-algebra if it is a

{- >, l}-subreduct of a basic hoop. We denote the variety of basic BCK-
algebras by SBeOC.

Note that every subdirectly irreducible basic BCK-algebra satisfies (B)

and (H) and, hence, is a HBCK-algebra satisfying (B). By the proof of The-
orem 1.6, (ii) =$> (iii), every subdirectly irreducible HBCK-algebra satisfying
(B) and therefore every subdirectly irreducible basic BCK-algebra, is totally
ordered.

Remark 3.2. It is known that BCK-algebras that are subdirect products of

totally ordered BCK-algebras coincide with BCK-algebras satisfying (B) [38,
proof of Theorem 3]. However, the quasivariety of BCK-algebras satisfying
(B) turns out to be strictly larger than BBC3C. Consider the algebra A =
([0, 1], - >, 1), where for a, b G A

a->bL=<
L fl ifa<6
^max{l - a, 6} otherwise

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
Basic Hoops: an Algebraic Study of Continuous t-norms 87

Note that - > is the residual of the nilpotent minimum t-norm [21], de
by

_ J min{a, b} if a > 1 - b
yO otherwise
Thus, A is the implicative reduct of a commutative integral residuated lat-
tice, hence a BCK-algebra. However A does not satisfy (H). This can be
seen by taking x = 2/3, y = 1/3 and z = 0. It follows that A is a (totally
ordered) BCK-algebra that is not a subreduct of any hoop, hence does not
belong to SSeOC.

Using [32, Theorem 3.13] and [5, Theorem 4.3] we can characterize the
finite subdirectly irreducible HBCK-algebras satisfying (B). Here the build-
ing blocks are the implicative reducts of the hoops Cn, which will be denoted
henceforth by Ln.

Proposition 3.3. A finite HBCK-algebra A, satisfying (B), is subdirectly

irreducible if and only if there are fc, ni, . . . , n& € N with

A^Lnie--.0Lnfc.

Our next goal is to show that the variety BBC9C is generated by its finite
algebras. Since every subdirectly irreducible basic BCK-algebra is a tota
ordered HBCK-algebra, it suffices to show that the class of totally order
HBCK-algebras has the FEP; using [5, Theorem 4.3], the proof of Theor
2.1 will go through for totally ordered HBCK-algebras. It is only necessa
to show that the class of totally ordered Lukasiewicz BCK-algebras has t
FEP, so we proceed to prove it.

Lemma 3.4. The class of totally ordered Lukasiewicz BCK-algebras has t

FEP.

Proof. Let L be a totally ordered Lukasiewicz BCK-algebra and let P

a partial subalgebra of L. By [5, Lemma 4.1], it is known that L may b
embedded in a totally ordered Wajsberg hoop A; hence P may be regar
as a partial subalgebra of A. It follows that there is a finite totally ord
Wajsberg algebra A' in which P is embeddable. But P is also embedd
in the {- >, l}-reduct L' of A7 and L' is a totally ordered Lukasiewicz B
algebra. ■

Proposition 3

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
88 P. Agliano, I. M.A. Ferreirim and F. Montagna

Theorem 3.6. The variety BBC3C is generated as a quasivar

members. In particular

SSC3C = SPPu(Lni 0 • • • 0 Lnfc : fc, m , . . . , nfc € N

Finally, we obtain an equational description of 'B'BGX.

Theorem 3.7. The variety SSC3C is the class of BCK-algeb

the identities (B) and (H).

Proof. As observed before, every basic BCK-algebra satis

Conversely, observe that the class of BCK-algebras whic
(H) is a variety, since it is the subvariety of HBCK-algebr
Now, every algebra in a variety is a subalgebra of an ult
finitely generated subalgebras, hence it suffices to show t
generated BCK-algebra satisfying (B) and (H) is a {- »,
basic hoop.
Using [5, Theorem 4.3] and the fact that subdirectly i
algebras satisfying (B) and (H) are totally ordered HBC
by induction on the number of generators, as in [5, The
Theorem 5.1]) to prove that every totally ordered finitely
algebra is a {->, l}-subreduct of a totally ordered (hen

4. G-algebras and their reducts

In [24] a G-algebra is defined to be a BL-algebra satisfying the equation

x2 = x; the variety %A of G-algebras is proposed as an equivalent algebraic
semantics for Godel logic, i.e. the propositional calculus naturally associated
with the Godel t-norm. It is clear that the variety of hoop-subreducts of G-
algebras is the variety SM of basic hoops that satisfy x2 = #, i.e. idempotent
basic hoops. It is almost immediate to see that in an idempotent hoop the
meet and the product coincide: an idempotent hoop is always a subreduct
of a Heyting algebra [29] and it is better known as a Brouwerian semilattice
[31], [8] and [7]. It follows that the variety SJC coincides with the variety of
Brouwerian semilattices that are subdirect products of chains.
The algebras in SJ{ are term-equivalent to relative Stone algebras [27]
or semi-Boolean lattices [37]; their properties have been thoroughly inves-
tigated. In particular, it is well-known that this variety and the variety of
Heyting algebras generated by chains are locally finite and so the quasi-
equational theories of S^£ and $A are decidable. Thus, Godel logic is decid-
able, a fact already known, since Godel logic is coNP-complete [24].

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
Basic Hoops: an Algebraic Study of Continuous t-norms 89

We consider now the variety $15QX consisting of implicative subre

of algebras in S^C Since algebras in 93< are Brouwerian semilattices,
variety 923 C3C consists entirely of Hilbert algebras. It is well-known tha
variety of Hilbert algebras is the variety of BCK-algebras satisfying

x -> (x -> y) = x -> y (G)

hence the algebras in S2JCK are basic BCK-algebras

see that the converse also holds.

Theorem 4.1. 9BC3C is the variety of basic BCK-algebras satisfying (G).

Proof. It is enough to show that every subdirectly irreducible basic BCK

algebra satisfying (G) is a reduct of basic hoop. If A is a subdirectly irre
ducible basic BCK-algebra satisfying (G) then it is totally ordered and, by
(G), a Hilbert algebra. But a linearly ordered Hilbert algebra is the reduc
of a (linearly ordered) Brouwerian semilattice, hence of a basic hoop. ■

5. Product algebras and their reduct s

Product algebras were introduced in [24] as an algebraic semantics of the

logical system naturally associated with the product i-norm. In any BL-
algebra it is possible to define a unary operation of negation (-i) by ->x =
x - ♦ 0; a product algebra is a BL-algebra satisfying the equations

xAnx = 0 (PAl)
-i-* -> {{xz -> yz) ->(*-+ y)) = 1. (PA2)

The variety 7 A of product algebras seems to be the most interesting

associated with t-norms and has attracted a lot of attention recentl
particular, in [1] it is shown that product algebras are term equivalen
class of bounded hoops and that 7 A is the equivalent algebraic seman
product logic. In [15] it is shown that 7 A is generated by any infinite to
ordered product algebra and the finitely generated free product algeb
characterized. Moreover it is observed that for any product algebra
set A \ {0} is a filter, hence the only simple product algebra is th
element bounded hoop that is term-equivalent to the two element B
algebra. Since any product algebra has the two element Boolean a
as a subalgebra, there are exactly three varieties of product algebra
trivial variety, the variety of Boolean algebras and the entire variety.
the following lemma, necessary in the sequel.

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
90 P. Agliano, I. M.A. Ferreirim and F. Montagna

Lemma 5.1. [24] Let A be a totally ordered product algebra a

Then:

1. if a/ 0, then -»a = 0.;

2. if a/ 0, then ba = ca implies b = c;
5. if a ^ 0, tfien 6a < ca implies b < c.

A hoop A is cancellative if its underlying monoid is cancellative. This

is equivalent to saying that for all a, b € A a - ► ba = 6. Cancellative hoops
form a variety C of Wajsberg hoops [6]; hence they satisfy (T).

Proposition 5.2. A is a subdirectly irreducible product algebra if and only

if either A = 2 or A = 2 © C, where 2 is the two element Boolean algebra
and C is a subdirectly irreducible cancellative hoop.

PROOF. If A is subdirectly irreducible and different from 2, then it follows

from Lemma 5.1.2 and [6, Theorem 2.9] that A = 2©C for some subdirectly
irreducible hoop C. Hence, by Lemma 5.1.2, C is a totally ordered hoop and
its underlying monoid is cancellative. Conversely, let C be any cancellative
subdirectly irreducible hoop. It is easy to check that 2 © C is a subdirectly
irreducible product algebra. ■

Now, we want to describe the variety

of hoop subreducts of product algebras.
as the previous cases, since the consta
the defining axioms. Hence we adopt
It is well-known [4] that the lattice
two atoms: the variety C of cancellative
Boolean algebras, i.e. the variety gener
varieties consist of basic hoops, so the
subvarieties of BIK. The interesting fa
varieties are both Wajsberg hoops and pr
known facts about Wajsberg hoops an
W!K and 7Di both lie above the join of
finite Wajsberg chain other than 2 is
A hoop is semi-cancellative if it satisfi

Vxy(3zz<xy) =*> x = y-*xy;

a hoop is quasi- cancellative if it satisfies

Vxyz (3w < z) => (xz - ► yz) - * (x - ► y) = 1.

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
Basic Hoops: an Algebraic Study of Continuous t-norms 91

Any cancellative hoop is quasi-cancellative and any quasi-cancellative

is semi-cancellative; moreover an unbounded quasi-cancellative hoop is
cellative and if A is a bounded (by 0) quasi-cancellative hoop, then A
is cancellative.

Proposition 5.3. Any totally ordered Wajsberg hoop is semi-cancellative.

No finite Wajsberg chain with more than two elements is quasi-cancellative.

Proof. The second claim is easily proved by inspection, using the above
description of finite Wajsberg chains. Let A be a totally ordered Wajsberg
hoop and a,b,c€ A with c < ab. Then ab £ c and by residuation a£b - ► c.
Since A is totally ordered, 6 - ► c < a. By a repeated use of equation (T) we
get
a = a V (b -> c)
= (a->(6-c))->(6->c)
= (ab - ♦ c) - > (b - > c)
= b - > ((ab -> c) -* c)
= 6->(a&Vc)
= 6 -► aft. m

Corollary 5.4. Let A be a nontriv

cancellative if and only if it is an u

Proof. If A is cancellative, then no

potent. However in a bounded hoop
potent. It follows that a cancellativ
Conversely, let A be an unbounde
for all ayb € A there is a, c € A
a = b - > ab and A is cancellative. ■

We axe now ready to characterize

with a lemma.

Lemma 5.5. Let A be a subdirectly irreducible basic hoop satisfying the equa-
tion

(x-+y)->y<((y->z)-+ ((y -> x) -> x)) -> ((y -+ x) -> x). (PB)
Then either A is a Wajsberg hoop or A = 2 ©B for some Wajsberg hoop B.

Proof. Let A be a subdirectly irreducible basic hoop satisfying (PB). Then

A = F©B, where B is a subdirectly irreducible Wajsberg hoop. The thesis

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
92 P. Agliano, I. M.A. Ferreirim and F. Montagna

will be proved if we show that either F is trivial or it is e

that b € F, b < 1 and b is not the minimum element of
a,c G A with c < b < a < 1 and a € B. However, A is a b
which b = a - > 6 = a(a - > 6) = ab. Since 6 - > c £ F \ {
have 6 - ► c < a.
Hence

a = 1 - ► a = ((6 - * c) - > a) - ► a since b - > c < a

= ((6 - > c) - * ((6 - > a) -» a)) -► ((6 - > a) - ► a) since b < a
> (a -> 6) -+ 6 = b -> 6 = 1 by (PB)
and, since a - > 6 = 6, that is a contradiction. This concludes the proof. ■

Theorem 5.6.2 The variety 3*K consists exactly of basic hoops satisfying

(V^z)V((y^xy)-*x) = l. (PH)

Proof. We prove first that (PH) holds in all prod

verify it in any totally ordered product hoop H. N
of H then y - > z = 1; otherwise (y - > xy) - > x =
Conversely, we show that if H is a subdirectly i
ordered) basic hoop satisfying (PH), then H is
that y is not the minimum of H. Take z < y. The
(y - * xy) - ► x = 1 for every x. Thus if H has n
a cancellative hoop, hence a product hoop. If H
by the above argument, H \ {0} is the domain o
and H = 2 © C is (the hoop reduct of) a product a
hoop. ■

The descr
1.9 is har
rectly irr
uct hoops

Corollary
if it is th
irreducib
subdirectl

Moreover
subdirect

2This axiom

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
Basic Hoops: an Algebraic Study of Continuous t-norms 93

Proposition 5.7. A subdirectly irreducible hoop is quasi-cancellative

only if it belongs to 7Ji.

Proof. It is easily seen that any subdirectly irreducible hoop in V'K i

cancellative. Let now A be a quasi-cancellative subdirectly irreducibl
Then A = B © C, where C is a subdirectly irreducible Wajsberg h
A is unbounded, then it is a cancellative hoop, therefore is in TJ{, b
hoop subreduct of the product algebra 2 © A. If A is bounded by 0
0 e B. If there is a b € B with 0 < b < 1, then for c € C \ {1} w
cb = b = 16. Since b > 0 and A is quasi-cancellative we must have 1 - ♦
i.e. c = 1, a contradiction. Hence B = 2 and A = 2©C. Moreover, s
is quasi-cancellative and c > 0 for all c € C, then C must be cancell

We now turn to implicative subreducts of product hoops (algebra

call that the variety H'BCX of Lukasiewicz BCK-algebras coincid
implicative subreducts of Wajsberg hoops. Since the variety 6 of c
tive hoops is contained in the variety W3i of Wajsberg hoops it is cl
S^(G) C JliiEX. The fact that the converse also holds will be crucial
sequel.
Let G be an abelian ^-group; the positive cone of G is the hoop P(G) =
({x € G : x > 0}, •, -►, lp(G)) where

xy = x + y
x->y = 0V(y-x)
lp(G) = 0g-

It is easily seen that P(G) is cancellative. The converse is true as well [6,
Theorem 1.17]: if A is a cancellative hoop then there is an abelian ^-group
G such that A^P(G).
Lemma 5.8. Any Lukasiewicz BCK-algebra is isomorphic to a subreduct of
a cancellative hoop; hence S^(G) = L^QX.

Proof. Let A e L'BQX. We may assume that A is subdirectly irreducible,

hence totally ordered. Therefore there exists a bounded Wajsberg hoop
B of which A is a subreduct [20, Theorem 3.3]. Being bounded, B is
(equivalent to) a Wajsberg algebra. By [36] there is an abelian ^-group
G = (G, +, - , V, A, 0g) and u e G such that B is isomorphic to the Wajs-
berg algebra
({x e G : 0 < x < u}, -►, •, 0B, 1b)

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
94 P. Agliand, I. M.A. Ferreirim and F. Montagna

where:

xy = u A (x + y)
x->y = OV(y-:r)
U = 0G
Oa = u.

Now it is easily checked that A is an implicative subreduct of the cancellative

hoop P(G). ■
An implicative
BCK-algebra; we

Theorem 5.9. Th
satisfying the eq

PROOF. Since (PB

involves only im
Conversely supp
satisfying (PB). A
BCK-algebra or
the first case, by
hoop C. Then 2©C
5.6) of which A i
hoop of which B
that is again a product hoop. ■

Corollary 5.10. An algebra in 3*2630 is

if it is 2, or it is a subdirectly irreduc
equal to 2 © C for some subdirectly ir

6. The lattice of subvarieties of product hoops

The link between abelian ^-groups, cancellative hoops and product algebras
has been explored in [6] and [15]. A similar result holds for product hoops.
It is well-known [4] that the variety C of cancellative hoops is generated
as a quasivariety by the cancellative hoop C^ = (C^, - >, •, 1) (i.e. 6 =
SPPU{CU)), where Cu = {an : n £ uj} is the free monogenerated monoid
and
anam = an+m an -► am = amax(m"n'°) .

It is also clear that C^ = P(Z). Recall that if A is a cancellative hoop,

then 2 © A is a product algebra. So there is an abelian ^-group G such that

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
Basic Hoops: an Algebraic Study of Continuous t-norms 95

A = P(G). Since ^-groups are torsion-free, then Z (regarded as a ^-gr

is embeddable in G and since P(Z) = C^ we have:

Lemma 6.1. The product algebra 2©Cu; is embeddable in any product a

of the form 2 © A, where A is a cancellative hoop.

Theorem 6.2. The variety VOi is generated as a quasivariety by 2

i.e. <S><H = SPPU{2®C(JJ).

Proof. First we show that any subdirectly irreducible algebra A i

belongs to SPPU(2 © C^). This is obvious if A = 2; if A is a cancel
hoop, then A € SPPu(Cu,) C 5PPu(2©Cu;). Let now A = 2©C, wh
is cancellative, and let Cf^/U the ultrapower in which C embeds. It is
seen that A embeds into (2 © C^)1 /U. By Corollary 5.6 there is noth
else to check.
Since any algebra in ^Ji is a subdirect product of subdirectly irreducible
product hoops, from the above we get that VK C SPSPPU{2 © Cw) C
SPPU(2 © C^). This is enough to prove the result. ■

We can now describe completely the lattice of subvarieties of pro

hoops.

Theorem 6.3. The lattice of subvarieties of product hoops (depicted in Fig-

ure 1) consists exactly of five subvarieties: the trivial variety 7, the variety
7%, the variety 6 of cancellative* hoops, the variety 3 of generalized Boolean
algebras and the variety 3 V C.

Proof. We know that 3 = " V(2), 6 = V(CW) and 7% = V{2 © Cw).

Moreover Sve < 3*K, since it consists entirely of Wajsberg hoops. Hence, by
Corollary 5.6, it is enough to show that VK is generated by any subdirectly
irreducible product algebra 2 © A, where A is a subdirectly irreducible
cancellative hoop. But by Lemma 6.1 and Theorem 6.2

7% = V(2 © Cw) C V(2 © A) C KK,

hence equality holds. ■

It is easy to check that the

CP3K) by x - ► x2 « 1 and x -

Proposition 6.4. The variety

equation (T).

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
96 P. Agliano, I. M.A. Ferreirim and F. Montagna

s\>e T

Figure 1. The lattice of subvarieties of WH

PROOF. It is clear that the only subdirectly irreducible product hoops in

S V 6 axe 2 and the subdirectly irreducible cancellative hoops. Since they
are all Wajsberg hoops, equation (T) holds in them.
Conversely suppose that V is a variety of product hoops satisfying (T)
and suppose, by way of contradiction, that 2 © A € V for some subdirectly
irreducible cancellative hoop A. If a € A with a < 1 from (T) we obtain

l = (a->0)->0=(0->a)->a=l-»a = a

a contradiction. It follows that no subdirectly irreducible produ

the form 2 © A belongs to V and hence V = S V Q. ■

References

[1] Adillon, R.J., and V. Verdu, 'On product logic', Soft Computing 2 (19
146.

[2] Agliano, P., 'Ternary deductive terms in residuated structures', Ada Sc

(Szeged) 64 (1998), 397-429.
[3] Agliano, P., and A. Ursini, 'On subtractive varieties III: Prom ideals to
ences', Algebra Universalis 37 (1997), 296-333.
[4] Amer, K., 'Equationally complete classes of commutative monoids with
Algebra Universalis 18 (1984), 129-131.
[5] Blok, W.J., and I.M.A. Ferreirim, 'Hoops and their implicational reduc
stract), Algebraic Methods in Logic and Computer Science, Banach Center P
tions 28 (1993), 219-230.
[6] Blok, W.J., and I.M.A. Ferreirim, 'On the structure of hoops', Algebra Uni
43 (2000), 233-257.
[7] Blok, W.J., P. Kohler, and D. Pigozzi, 'On the structure of varieties wi
tionally definable principal congruences IF, Algebra Universalis 18 (1984), 3
[8] Blok, W.J., and D. Pigozzi, 'On the structure of varieties with equationally
able principal congruences F, Algebra Universalis 15 (1982), 195-227.

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
Basic Hoops: an Algebraic Study of Continuous t-norms 97

[9] Blok, W.J., and D. Pigozzi, 'Algebraizable logics', vol. 77 (396), Mem. A
Math. Soc, Providence, 1989.
[10] Blok, W.J., and D. Pigozzi, 'On the structure of varieties with equationally d
able principal congruences III , Algebra Universalis 32 (1994), 545-608.
[11] Blok, W.J., and J. Raftery, 'Varieties of commutative residuated inte
pomonoids and their residuation subreducts', J. Algebra 190 (1997), 280-328.
[12] BOSBACH, B., 'Komplementare Halbgruppen. Axiomatik und Arithmetik',
Math. 64 (1969), 257-287.
[13] Chang, C.C., 'Algebraic analysis of many valued logics', Trans. Amer. Mat
88 (1958), 467-490.
[14] ClGNOLl, R., F. Esteva, L. Godo, and A. TORRENS, 'Basic fuzzy logic is th
of continuous t-norms and their residua', Soft Computing 4 (2000), 106-112.
[15] Cignoli, R., and A. Torrens, 'An algebraic analysis of product logic', Mult-Val
Log. 5 (2000), 45-65.
[16] Cornish, W.H., 'A large variety of BCK-algebras', Math. Japon. 26 (1981), 339
[17] Dummett, M., 'A propositional calculus with denumerable matrix', J. Symb.
24 (1959), 97-106.
[18] Evans, T., 'Some connections between residual finiteness, finite embeddabilit
the word problem', J. London Math. Soc. 1 (1969), 399-403.
[19] Ferreirim, I.M.A., On varieties and quasivarieties of hoops and their reducts,
thesis, University of Illinois at Chicago, 1992.
[20] Ferreirim, I.M.A., 'On a conjecture by Andrzej Wronski for BCK-algebr
subreducts of hoops', Sci. Math. Jpn. 53 (2001).
[21] Fodor, J. C, 'Nilpotent minimum and related connectives for fuzzy logic', Pr
IEEE International Conference on Fuzzy Systems 4 (1995), 2077-2082.
[22] Font, J.M., A.J. Rodriguez, and A. Torrens, 'Wajsberg algebras', Stochas
(1984), 5-31.
[23] Gumm H.P., and A. Ursini, 'Ideals in universal algebra', Algebra Universalis 19
(1984), 45-54.
[24] Hajek, P., Metamathematics of fuzzy logic, Trends in Logic- Studia Logica Library,
no. 4, Kluwer Academic Publ., Dordrecht/ Boston/ London, 1998.
[25] Hajek, P., L. Godo, and F. Esteva, 'A complete many- valued logic with product
conjunction', Arch. Math. Logic 35 (1996), 191-208.
[26] Harrop, R., 'On the existence of finite models and decision procedures for proposi-
tional calculi', Proc. Cambridge Philos. Soc. 58 (1958), 1-13.
[27] Hecht, T., and T. Katrinak, 'Equational classes of relative Stone algebras', Notre
Dame J. Formal Logic 13 (1972), 248-254.
[28] HiGGS, D., 'Dually residuated commutative monoids with identity element do not
form an equational class', Math. Japon. 29 (1984), 69-75.
[29] HORN, A., 'The separation theorem of Intuicionistic Propositional Calculus', J. Symb.
Log. 27 (1962), 391-399.
[30] Iseki, K., 'An algebra related with a propositional calculus', Proc. Japan Acad. 42
(1966), 26-29.
[31] Kohler, P., 'Brouweriansemilattices', Trans. Amer. Math. Soc. 268 (1981), 103-126.

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms
98 P. Agliano, I. M.A. Ferreirim and F. Montagna

[32] KOMORI, Y., 'Super-Lukasiewicz implication^ logics', Nagoya

127-133.

[33] Komori, Y., 'Super-Lukasiewicz propositional logics', Nagoya Math. J. 84 (1981),

119-133.

[34] Kowalski, T., 'A syntactic proof of a conjecture of Andrzej Wronski', Rep. Math.
Logic 28 (1994), 81-86.
[35] Mostert, P.S., and A.L. Shields, 'On the structure of semigroups on a compact
manifold with boundary', Ann. of Math. 65 (1957), 117-143.
[36] Mundici, D., 'Interpretation of AFC* -algebras in Lukasiewicz sentential calculus',
J. Functional Analysis 65 (1986), 15-63.
[37] Nemitz, W., 'Semi-boolean lattices', Notre Dame J. Formal Logic 10 (1969), 235-
238.

[38] Palasinski, M., 'Some remarks on BCK-algebras', Math. Seminar Notes, Kobe Univ.
8 (1980), 137-144.
[39] Palasinski, M., 'Ideals in BCK-algebras which are lower semilattices', Math. Japon.
26 (1981), 245-250.
[40] Raftery, J.G., On ideals, congruences and extensions of BCK-algebras, Ph.D. the-
sis, University of Natal, Durban, 1987.
[41] Raftery, J.G., and T. Sturm, 'On ideal and congruence lattices of BCK-
semilattices', Math. Japon. 32 (1987), 465-474.
[42] Wronski, A., 'BCK-algebras do not form a variety', Math. Japon. 28 (1983), 211-
213.

P. Agliano
Dipartimento di Matematica
Via del Capitano 15
53100 Siena, Italy
aglianoQunisi . it

I.M.A. Ferreirim
Departamento de Matemdtica
FCUL
Campo Grande, Ed. C6, Piso 2
1749-016 Lisboa, Portugal
mimaf erQptmat .fc.ul.pt

F. Montagna
Dipartimento di Matematica
Via del Capitano 15
53100 Siena, Italy
montagnaQunisi . it

This content downloaded from 178.250.250.21 on Wed, 17 Aug 2016 03:13:34 UTC
All use subject to http://about.jstor.org/terms