Almeida : Communicated by N.R. Reilly

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Semigroup F o r u m Vol.

37 (1988) 5 3 - 5 7
9 1988 Springer-Verlag N e w York Inc.

RESEARCH ARTICLE

SOME PSEUDOVARIETY JOINS INVOLVING


THE PSEUDOVARIETY OF FINITE GROUPS

Jorge Almeida*

Communicated by N.R. Reilly

1. I N T R O D U C T I O N AND BACKGROUND

This note solves some problems of calculation of joins of pseudovarieties of semigroups


proposed by Rhodes in [10]. This type of problem is in general quite hard and in fact Rhodes
[10] announces the discovery of some rather pathological examples of fairly tame pseudovarieties
such that the resulting joins have undecidable membership problems. Thus, it appears that the
determination of joins will have to depend on ad hoc methods tailored to specific cases.
In general there are two types of approaches for the computation of joins. To obtain the
join V1 V V2 of two pseudovarieties, guess a solution W and show every member of W divides
a product $1 x $2 with Si 6 Vi (i = 1,2). This sort of approach has been used extensively
for instance to yield joins of varieties of completely regular semigroups (e.g., in [6] and [8]). It
constructively displays $1 and $2 and so we call it a structural approach. An alternate method
is based on the characterization of appropriate properties of the join. Since pseudovarieties are
defined by pseudoidentities, this may be interpreted to mean the determination of a convenient
basis for the pseudoidentities which hold simultaneously in V1 and V~. This program appears to
be carried out here for the first time for pseudovarieties, although it has been used successfully
for varieties and such type of results are sometimes transferable to pseudovarieties (cf. [2]). We
call this approach syntactic.
The success of a syntactic approach depends heavily on the knowledge of all implicit op-
erations on a suitable pseudovariety containing V1 and V~. In the present case, the effective
determination of the join of finite groups with finite commutative semigroups is based on a
canonical form for the implicit operations on the pseudovariety of all finite semigroups in which
idempotents are central which was given by Azevedo and the author [3].
In the following, we assume the reader is familiar with the basics of semigroups and pseu-

* This work was supported, in part, by INIC grant 8 5 / C E X / 4 .


ALMEIDA

dovarieties (cf. Eilenberg [4]). Also, we will use without c o m m e n t the result that pseudovarieties
are defined by pseudoidentities (i.e., formal equalities of implicit operations) due to Reiter-
man [9].
We let x ~-* x ~ denote the unary operation whose value s ~ on an element s of a finite
sernigroup S is the unique idempotent in the subsemigroup generated by s. For a set E of
pseudoidentities of semigroups, [~3] denotes the class of all finite semigroups satisfying ~.
The following is a list of pseudovarieties which will appear in the sequel.
I = Ix = y] trivial
Sl = [ x y = y x , x 2 = x] semilattices
G = [#o = 1] groups
N = Ix ~ = 0] nilpotent
A = [z ~ = x '~ aperiodic
CR = [Z w + l = ;;g] completely regular
Com= [xy : y x ] commutative
IE = ix ~ : y~] nilpotent extensions of groups
ZE = [x~'y = y x '~] all idempotents central.

This paper depends heavily on the following result.

THEOREM 1.1 (Almeida and Azevedo [3]). E v e r y i m p l i c i t o p e r a t i o n on Z E is o f t h e form

w o p l ( e y l . . . . . eyn ) w l . . . W k - l p k ( e y l . . . . , eyn)

w h e r e k >_ O, each wl is a w o r d n o t i n v o l v i n g a n y o f t h e v a r i a b l e s Y l , . . . , Y n a n d o n l y wo m a y

be the empty w o r d , e = (Yl . . . y , ) , o , each pi is an n - a r y i m p l i c i t o p e r a t i o n on G (i = 1 . . . . . k)

and pi(eyl ..... e y n ) = e m a y o n l y o c c u r f o r i = k.

In the presence of further pseudoidentities, implicit operations may assume simpler forms.
For instance, if we take an implicit operation 7r of the form given by T h e o r e m 1.1 and restrict
it to the pseudovariety Z E f'l C R then, for any of the variables x occurring in the wi, we may
replace it by x '~+1 and so, for each S E Z E f'l C R and for all sl . . . . . sr E S, 7r(sl . . . . , s~) lies
in the s u b g r o u p with idempotent ( s l , . . . ,s~)~'; hence 7r is of the form p ( e y l . . . . , e y r ) ~ where
e = ( Y l . . . Yr)~ and p is an implicit operation on G. Similarly, every implicit operation on
Z E N | z ~ y ~+1 = x~y] is either of the form j u s t described or it is explicit (i.e., it is decribed by
a word).

2. :lOINS GVCom AND GV(NVICom)

For the effective calculation of these joins, we need the following result concerning finite
groups.

LEMMA 2.1. Let


~rl = w o p l ( e y l . . . . , e y n ) W l . . . w k - l p k ( e y l ..... eyn)

'~'2 = trl(eyl . . . . , e y n ) V l . . . V l - l c r l ( e y l . . . . . e v a )

54
ALMEIDA

be two implicit operations on Z E o f the form described in T h e o r e m 1.1. S u p p o s e G ~ ~rl = ~r2


and G ~ ~1 = 1. T h e n wo = 1 (i.e., wo is the e m p t y word).

PROOF. S u b s t i t u t i n g y~ for each of the v a r i a b l e s Vl, . - . , yn, we deduce fxom G ~ ~rl = ~r~ t h a t
G ~ wowl...wk-1 = W l . . . v z - 1 , whence w o w l . . . w ~ - I = vl...vt-1 (of. E v a n s [5]). Suppose
w0 # 1 ( a n d t h u s vl # 1).
Since G ~ ~rl -- 1, t h e r e is an integer r such t h a t the g r o u p Sr of all p e r m u t a t i o n s of
{1 . . . . . r} fails the p s e u d o i d e n t i t y tr 1 -- 1. Let bl . . . . . b~ E S~ be such t h a t crl(bl . . . . . bn) ~ id,

say l a l ( b l . . . . . bn) = ct # 1.
We t a k e s = r + ] v l . . . v l _ l ] a n d consider the cycle a = (c~ r + 1 r+2 ... s) in Ss.
S u b s t i t u t i n g bi for Yi (i = 1 , . . . , n ) a n d a for every other variable in zrl and zr2, we o b t a i n
p e r m u t a t i o n s Pl,P2 E Ss, respectively. T h e n l p l < s, whereas lp2 = s. Hence S~ ~ 7rl = 7r2,
c o n t r a d i c t i n g G ~ r l = 7r2.

THEOREM 2.2. a) G V C o r n = Z E .
b) G V ( N t0 C o r n ) = G V N = I E .

PROOF. Clearly, G V C o r n _C Z E and G V ( N N C o r n ) C_ G V N C_ I E . Thus, it suffices to


show Z E C_ G V C o r n and I E C G V ( N ;3 C o r n ) .

To e s t a b l i s h the inclusion Z E C_ G V C o r n , consider two implicit o p e r a t i o n s on Z E of the


form described in T h e o r e m 1.1,

7rl -~ WOpl(eyl . . . . . e y n ) w l . . , wlc-lp~:(eyl . . . . . ey,~)

7r2 : VOCrl(f z I . . . . . f Zm)Vl . . . Vl-lCrl(f Zl . . . . . f Zm) ,

and suppose G V C o r n ~ ~r1 = ~r2. I f Yl ~ {zl . . . . . zm}, consider a cyclic aperiodic m o n o i d


M of i n d e x Iv0 ... vI-1[ + 1 w i t h g e n e r a t o r a. Then, s u b s t i t u t i n g a for Yl and 1 for e v e r y other
variable in 7rl and ~r2, we o b t a i n 0 for r l and, either a ~ # 0, or 1 (in case yl does not occur in
v 0 . . . v l - 1 ) for lr2 as their values in M . It follows t h a t {Yl . . . . . y~} = {Zl . . . . . zm} and e = f .
To conclude t h a t 7q = 7r2, we show t h a t k = l, wi = vl and pi+l = ~ri+l (i = 0 . . . . , k - 1). Since
G ~ ~r1 = 7r2, this can now be easily o b t a i n e d by induction using L e m m a 2.1. Hence G V C o r n
satisfies no m o r e p s e u d o i d e n t i t i e s t h a n Z E does, and so G V C o r n = Z E .
As t o t h e inclusion I E C_ G V ( N [3 C o r n ) , recall t h a t , in view of the r e m a r k s following
T h e o r e m 1.1, a n y implicit o p e r a t i o n on I E is either explicit or of the form p ( e y l , . . . , eyn) where
e = z ~ a n d p is an implicit o p e r a t i o n on G . Suppose t h a t 7rl and ~r2 are two implicit o p e r a t i o n s
on I E such t h a t G V ( N ~ C o r n ) ~ ~rl = :r2. T h e n ~rl and ~r~ m u s t be b o t h of the s a m e t y p e
(i.e., either b o t h explicit or none explicit) since N [3 C o r n fails any p s e u d o i d e n t i t y of the form
w = z '~ for any word w. In case ~rl and ~r~ are b o t h explicit, t h e y m u s t coincide since G satisfies
no n o n t r i v i a l identity. In the r e m a i n i n g case, ~rl and r~ m u s t also coincide since t h e y r e s t r i c t to
the s a m e o p e r a t i o n on G . Hence I E ~ ~r~ = :r2, and so I E = G Y ( N N C o r n ) .

REMARK. It is m u c h easier to show G V N = I E structurally, since, if S ~ I E h a s m i n i m a l

55
ALMEIDA

ideal K , then
S ---~S/K x K

s ~--,(~/K, ,~+1)
defines an embedding into a product of a nilpotent semigroup by a group,

For a pseudovariety V, let L V denote the class of all finite semigroups S all of whose
submonoids e S e with e ~ -- e E S lie in V.

THEORI~M 2.3. T h e diagram in Figure 1 represents a s u b l a t t i c e o f the lattice o f all pseu-


dovarieties o f semigroups.
Zg

V2

ZE I"1CR

G,

6N l;0rn~ ~ 6

I NflCom

FiQure 1

Here,
Vl = ZE N L CR = [ z ~ = yx~, z~y~+I = z%,l
V2 = Z E N L C o r n = [ x ~ = yz~, z~yz = z~zy]
V s = Vl N V2
V4 = C o r n N L C R = [ z l / = 9z, x " 9 '~+1 = zwb']
V~ -- C o r n n L SI -- [zy = ~z, x ~ g 2 = ~:~9~
V6 = [x'~y = x'~y~] n ZE,
PROOF. For the part of the diagram under Gore, see Nelson [71 (and also Almeida [1] for the
translation to pseudovarieties). To show N V ( G N C o r n ) = [:c~ = yo, x,~yz .: z,~zlt] , use the
embedding of the preceding Remark. The equalities G V ( A t~ Corn) = IE follow from the proof
~nd the s t a t e m e n t of Theorem 2.2 respectively.
The remaining relations indicated by Figure 1 are obtained in a similar way. For instance,
to prove G V Sl = [ x ' ~ yz ~, I/~+x ~ 9] recall that, in view of the pseudoidentity t/~+1 = U~all
implicit operations on the pseudovariety on the right hand side are o f the form p(etll . . . . , ey,~)

56
ALMEIDA

(with the usual meanings for p and e). If G V Sl ~ p(cyl . . . . ,ey,) ~- o ' ( / Z l , . . . , f Z r a ) , then
{Yl,-.., Yn} = {zl . . . . . z,~} using sernilattices, and then p = a using groups.
As another example,

G V N V SI = [z ~ = 1] V ([x'~ = z y " = x'~y '~] n ZE)


= [z~y = yz ~, z ~ y ~+1 = x'~y].

For the second equality, recall that, using x'~y ~+1 = z"y, every implicit operation on the
pseudovariety on the right hand side is either explicit or of the form p(eyl .... , ey,~) (as usually).
If we have G V N V S I ~ ~1 = ~ru with two such operations 7rl and 1r2, then the variables occurring
on both sides must be the same because Sl ~ 7rl = ~r2, 7rl and r~ must be described by the
same word if one of them is explicit since N ~ ~rl = ~r~, and finally ~rl = r2 if they are not
explicit since G V Sl ~ r l = 7r~.
The remaining equalities can be established by the same methods.

REFERENCES
1. Almeida, J., Some order properties of the lattice of varieties of commutative semigroups,
Canadian J. Math. 38 (1986) 19-47.
2. , Power pseudovarieties of semigroups I, Semigroup Forum 33 (1986) 357-373.
3. Almeida, J. and A. Azevedo, Implicit operations on certain classes of semigroups, in S.
Goberstein and P. Higgins (Eds.), Algebraic Theory of Semigroups and its Applications,
Proceedings 1986 Chico Conf., D. Reidel, 1987 (1-11).
4. Eilenberg, S., Automata, Languages and Machines, Vol. B, Academic Press, New York,
1976.
5. Evans, T., Varieties ofsemigroups, Semigroup Forum 2 (1971) 1-43.
6. Hall, T. E. and P. R. Jones, On the lattice of varieties of bands of groups, Pacific J. Math.
91 (1980) 327-337.
7. Nelson, E., The lattice of equational classes of commutative sen~groups, Canadian J. Math.
23 (1971) 875-895.
8. Petrich, M. and N. R. Reilly, The join of varieties of strict inverse sembgroups and rectan-
gular bands, Glasgow Math. J. 25 (1984) 59-74.
9. Reiterman, J., The Birkhofftheorem for finite algebras, Algebra Universalis 14 (1982) 1-10.
10. Rhodes, J., New techniques in global semigroup theory, in S. Goberstein and P. ttiggins
(Eds.), Algebraic Theory of Semigroups and its Applications, Proceedings 1986 Chico Conf.,
D. Reidel, 1987 (169-181).

Centro de Matems
Universidade do Minho
Braga, Portugal

Received 2 February 1987 and, in final form, 21 October 1987.

57

You might also like