Homological Classification

of Monoids by
Projectivities of Right Acts

Vom Fachbereich Mathematik der

Carl-von-Ossietzky Universitat Oldenburg
zur Erlangung des Grades einer
Doktorin der Naturwissenschaften
angenommene Dissertation, vorgelegt von

Helga Oltmanns
aus
Westrhauderfehn

Erstreferent: Prof. Dr. Dr. h.c. Ulrich Knauer

Korreferent: Prof. Dr. Mati Kilp
Tag der m
undlichen Pr
ufung: 7. Juni 2000

Homological Classication
of Monoids by
Projectivities of Right Acts

Vom Fachbereich Mathematik der

Carl-von-Ossietzky Universit
at Oldenburg
zur Erlangung des Grades einer
Doktorin der Naturwissenschaften
angenommene Dissertation, vorgelegt von

Helga Oltmanns
aus
Westrhauderfehn

Erstreferent: Prof. Dr. Dr. h.c. Ulrich Knauer

Korreferent: Prof. Dr. Mati Kilp
Tag der m
undlichen Pr
ufung: 7. Juni 2000

CONTENTS

Contents
1 Introduction

2 Preliminaries

3 (X , Y)-projectivities of acts
3.1 Notation and basic informations . . . . . . . . . . . . . . . . . .
3.2 (X, Y )-projectivities with projective second component . . . . .
3.3 Equivalences of conditions . . . . . . . . . . . . . . . . . . . . .
3.3.1 Cyclic acts and coproducts of cyclic acts as second component . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.2 Cyclic acts and coproducts of cyclic acts as rst component
3.3.3 Coproducts of (X , Y)-projective acts . . . . . . . . . . .
3.3.4 Products of (X , Y)-projective acts . . . . . . . . . . . . .
3.3.5 Rees factor acts and coproducts of Rees factor acts as
second component . . . . . . . . . . . . . . . . . . . . .
3.3.6 Rees factor acts and coproducts of Rees factor acts as
rst component . . . . . . . . . . . . . . . . . . . . . . .
3.3.7 Various rst/second components . . . . . . . . . . . . . .
3.4 Projectivities which are trivial or equivalent to other projectivities
3.5 When (S, S/)- and (S/, S/)-projective acts are (S, S/)-projective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.6 Table of results . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.7 Further implications . . . . . . . . . . . . . . . . . . . . . . . .
3.8 Implication scheme . . . . . . . . . . . . . . . . . . . . . . . . .
3.9 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

14
14
17
17

4 Summary and Supplements

4.1 Further equivalences in the case of monoids with left zero
4.1.1 Equivalences to projectivity . . . . . . . . . . . .
4.1.2 Equivalences to (S, S/)-projectivity . . . . . . .
4.1.3 Equivalences to (S, S/(x, y))-projectivity . . . .
4.1.4 Equivalences to (S, S/I)-projectivity . . . . . . .
4.1.5 Equivalences to (S, S/sS)-projectivity . . . . . . .
4.2 Summarizing table . . . . . . . . . . . . . . . . . . . . .
4.3 Further implications, implication scheme . . . . . . . . .

.
.
.
.
.
.
.
.

46
47
48
50
52
53
55
56
58

.
.
.
.

60
61
63
68
69

5 Homological classication
Condition (Moz) . . . .
S is (X , Y)-projective .
5.1 All acts are (X , Y)-projective .
5.1.1 All are (X, G)-projective

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.

19
21
22
25
26
27
28
31
33
36
39
40
43

CONTENTS

5.1.2
5.1.3

All are (X, A)-projective . . . . . . . . . . . . . . . . . . 72

All are (X,
S/m )- or (X, S/)-projective . . . . . . 74
mM

5.1.4 All are (S/(x, y), S/I)-projective . . . . . . . . . . . . . 79

5.1.5 All are (S/sS, S/I)-projective . . . . . . . . . . . . . . . 81
5.1.6 All are (S/sS, S/tS)-projective . . . . . . . . . . . . . . 90
5.2 Characterization of monoids by implications between the concepts 95
5.3 Summarizing Tables . . . . . . . . . . . . . . . . . . . . . . . . 100
6 Dierences between the concepts
6.1 Dierences between (X , Y)-projectivities . . . . . . . . .
6.1.1 Dierences between various parts of the diagram .
6.1.2 Generators as second component . . . . . . . . .
6.1.3 Second component A or C . . . . . . . . . . . . .
6.1.4 Cyclic acts and coproducts of cyclic acts as second
ponent . . . . . . . . . . . . . . . . . . . . . . . .
6.2 Dierences from other concepts . . . . . . . . . . . . . .

107
. 109
. 109
. 111
. 112

. . .
. . .
. . .
. . .
com. . . . 112
. . . . 115

Introduction

In order to examine properties of mathematical objects, it is often helpfull to

represent them by other objects, which are better known. For instance, in the
investigations of rings properties of modules over rings, i.e., of representations
of rings by endomorphisms of abelian groups, can be utilized.
The representation of monoids by transformations of sets, which leads to acts
over monoids, is the foundation of the dissertation in hand.
Like in many other categories, projective and injective objects play an important role in the category of acts. Several dierent properties of acts over
monoids can be found in the literature. They can roughly be divided in two
parts: properties around injectivity and properties around projectivity.
The rst part properties weaker than injectivity includes, for example,
weak injectivity, f g-injectivity and principally weak injectivity. These have
been considered, for example, by Gould (cf. [9]). Injective acts themselves
have been treated by Berthiaume ([2]), Burgess ([7]), Skornjakov ([20]) et al.
Properties like weak projectivities, which had been introduced in [17], as well
as projectivity itself (cf. Knauer [15]) belong to the second part (properties
weaker than projectivity). Further concepts in this part are properties like
strong atness (cf. Bulman-Fleming, [3]), equalizer atness (cf. Normak,
[19]), atness (cf. Kilp, [11]). A survey on atness properties of acts has been
given by Bulman-Fleming in [4].
A detailed exposition of these concepts can be found in Monoids, acts and
categories ([14]). Here, the authors Kilp, Knauer and Mikhalev consider
both parts. After presenting elementary properties of monoids, acts and constructions like products, pullbacks, etc., various classes of acts are described
in detail, and it is shown, that the classes are dierent. The 4. Chapter of
this book is called Homological classication. It includes a large number
of results about acts over monoids, which concern questions of homological
classication, where properties of the representations, i.e., properties of acts,
lead to internal properties of the underlying monoid.

In the following, the denition of acts, some properties of acts, special elements and subsets of monoids, results concerning at acts, etc., are needed.
This preliminaries will be given in Chapter 2.
In this dissertation, properties weaker than projectivity will be considered. Recall, that an act AS is called projective, if for arbitrary acts BS and CS every

1 INTRODUCTION

homomorphism f : A C can be lifted with respect to every epimorphism

g : B C, i.e., there exists a homomorphism f  : A B, such that the
following diagram is commutative:
A
f

C .

A question arises in a quite natural way: What happens, if we restrict the

class of epimorphisms in the denition of projectivity? I.e., what happens, if
we require f to be liftable with respect to epimorphisms g : X Y , where
X and Y are elements of particular subclasses of all acts?
This question leads to (X , Y)-projectivity. A right act AS is called (X , Y)projective, if every homomorphism from AS to an act YS Y can be lifted with
respect to every epimorphism from XS X onto YS , where various classes X
and Y will be considered, for example the class of all free acts or the class of
all generators. This denition within the respective classes is presented at the
beginning of Chapter 3.
By denition we initially get 226 formally dierent concepts of (X , Y)-projectivity. In a certain sense, the third Chapter contains a ltration of these
properties:
Some of the properties turn out to be trivial, i.e., all acts have the respective
property (Lemma 3.2). Others are equivalent or coincide with projectivity (see
Lemma 3.26). At the end, thirty six concepts remain for further investigations.
The second topic of the third Chapter is to present various implications between the remaining properties. The results are illustrated in an implication
scheme at the end of Chapter 3.
It turns out, that projectivity itself is included in the concept of (X , Y)projectivities (see for example Theorem 4.3), and it is the strongest projectivity at all.
Since zeros in monoids play an important role with respect to (X , Y)-projectivities of acts, the investigation of the concepts introduced in Chapter 3 will
be the subject matter of Chapter 4, concerning acts over monoids with left
zero. The main statements of this Chapter contain equivalences in the case
of monoids with left zero as well as a summary of the results of Chapter 3

concerning equivalences in the case of arbitrary monoids. The results of this

Chapter are tools in the derivation of results in the Chapters 5 and 6.
Note, that these results also describe the respective situation in the category
of acts with zeros.
As mentioned before, a large number of results about acts over monoids refer
to questions of homological classication, where properties of acts lead to internal properties of the underlying monoid. In Chapter 5 internal properties
of monoids will be described by properties of their representations under the
viewpoint of homological classication with respect to (X , Y)-projectivities of
acts.
The rst Section contains characterizations of monoids over which all acts are
(X , Y)-projective. For example, all acts are (S/sS, S/I)-projective if and only
if S is the disjoint union of a special submonoid R of S and a simple semigroup
or S is a group or, in particular, S is the disjoint union of a group and a zero
(Theorem 5.52, Theorem 5.53, Corollary 5.56).
Furthermore, in Section 5.2, implications between the concepts are considered.
For example, all (S/sS, S/tS)-projective acts are (S/(x, y), S/tS)-projective
if and only if S fullls condition (Moz) (Theorem 5.73), which is introduced
at the beginning of Chapter 5.
The results will be summarized in tables given at the end of Chapter 5.
The assertions of Chapter 5 will be used in Chapter 6 to point out the
dierences between the concepts of (X , Y)-projectivity. Concrete (counter-)
examples will be given.
In the same way, distinctions between (X , Y)-projectivities and other properties weaker than projectivity will be demonstrated. The considered properties
in Section 6.2 are atness properties. The needed respective results concerning
homological classication are included in Chapter 2 Preliminaries.
Acknowledgements:
Many people deserve my thanks. I am very grateful to Professor Ulrich Knauer
for supervising this work, for his support and valuable advices. I thank Professor Mati Kilp for being a reviewer of my dissertation, for his valuable hints
and for the opportunity to spend some time at the University of Tartu. A big
thank for the hospitality and the inspiring atmosphere to all members of the
Algebraic Institute of the University of Tartu.
Finally thanks to my family and my friends who accompanied me through the
ups and downs.

2 PRELIMINARIES

Preliminaries

Most of the denitions and results given in this chapter are taken from [14].
Otherwise the respective paper/book will be especially cited.
Throughout this dissertation let S be a monoid with identity 1.
Denition 2.1 A nonempty set A is called a right S-act (or right act over
S), if there exists a mapping
: A S A
(a, s) a s := (a, s),
such that
(a) a 1 = a and
(b) a (s t) = (a s) t for a A, s, t S.
A right S-acts A is denoted by AS or by A, if the context allows to omit the
index S, and is simply called an act.
Analogously, left S-acts S A are dened.
For right S-acts the following names are also used: S-sets, S-polygons, Ssystems, S-automata ([20], [12], [21] and others).
Notice, that the monoid S with the internal multiplication of elements is a
right S-act.
The one element right S-act {}S is shortly denoted by S .
Denition 2.2 If AS and BS are right S-acts, then a mapping f : AS BS
is called a homomorphism of right S-acts, or S-homomorphism, if
f (as) = f (a)s for all a A and for all s S.
An S-homomorphism will simply be called homomorphism, if the context allows
to omit S. In this case we sometimes shortly write f : A B.
Note that the composition of S-homomorphisms is an S- homomorphism.
The category of all right S-acts is denoted by Act S.
In Act S the epimorphisms are exactly the surjective homomorphisms and
the monomorphisms are the injective homomorphisms. Thus isomorphisms
are bijective homomorphisms. If f : AS BS is an isomorphism, we write
AS
= BS .
An epimorphism f : AS BS is called a retraction, if there exists g :
BS AS , such that f g = idB . In this case, BS is called a retract of AS .

Denition 2.3 Let AS be a right S-act. An equivalence relation  on A is

called an S-act congruence or congruence on AS , if a  a implies (a s)  (a s)
for a, a A, s S.
For (a, a ) A A the smallest congruence on AS , containing (a, a ), is denoted by (a, a ) and is called a principal or monocyclic congruence.
Let  be a congruence on AS . Dene a right multiplication on the factor set
A/ = {[a] | a A} by elements of S by:
[a] s = [as] for every s S.
Then A/ becomes a right S-act, which is called a factor act of A by .
The canonical surjection
: AS (A/)S
a [a]
is an S-homomorphism and is called a canonical epimorphism.
For the monoid S, every right (semigroup) congruence  on S is an S-act
congruence on SS .
A factor act of S by a right principal (monocyclic) congruence on S is called
a monocyclic act.
Note, that the kernel equivalence ker f of an S-homomorphism f is an S-act
congruence, which is called a kernel congruence of f .
Denition 2.4 Let AS be a right S-act. An element A is called a zero
of AS (a xed element, a sink) if s = for all s S.
An act could have more then one zero. If the monoid S has a left zero z, then
every element a z A is a zero of AS .
Denition 2.5 Let AS be a right S-act. A subset B A of A is called a
subact of AS , if b s B for all b B, s S.
A subact BS of AS is called a proper subact, if B = A.
Any subact BS AS denes the Rees congruence B on AS by setting
a B a i a, a B or a = a .
The resulting factor act is denoted by (A/B)S and is called the Rees factor
act of AS by the subact BS .
In particular, if I S is a right ideal of S, then (S/I)S is called the Rees
factor act of SS by the right ideal IS .

2 PRELIMINARIES

Notice, that if (A/B)S is the Rees factor act of AS by the subact BS , then
the class [b]B of an element b B is a zero in (A/B)S and the class [a]B of
a A \ B is the one-element classes {a}. Thus the Rees factor act (A/B)S
could be considered as (A/B)S = ((A \ B) {})S .

Denition 2.6 A right S-act AS is called simple, if AS has no proper subacts.

Theorem 2.7 (Homomorphism Theorem for right S-acts)
Let AS , BS Act S, let f : AS BS be an S-homomorphism and let 
be a right congruence on AS , such that a  a implies f (a) = f (a ). Then f  :
(A/)S BS with f  ([a] ) := f (a), a A, is the unique S-homomorphism,
such that the following diagram is commutative:
f

AS

f

(A/)S
If  = ker f , then f  is injective, and if f is surjective, then f  is surjective.
Note that this implies, that every epimorphic image of a right S-act AS is
isomorphic to a factor act of AS , especially epimorphic images of SS are isomorphic to factor acts (S/)S .

Denition 2.8 The right S-act AS is called cyclic, if there exists an element
a A, such that AS = aS = {as | s S}.
Proposition 2.9 An right S-act AS is cyclic if and only if there exists a right
congruence  on S, such that AS
= (S/)S .
If f : AS BS , AS a cyclic right S-act, is an S-homomorphism, then f (A)
is a cyclic subact of BS .
If f is an epimorphism and AS = aS, then BS = f (a)S.

Remark 2.10 Let I = be a set and let (Xi )iI be a family. of right S


Xi in Act S is the disjoint union
Xi with the
acts. The coproduct
iI

injections ui : Xi

.


iI

iI

Xi dened by ui = id .
iI

Xi

|Xi , (i I), i.e., for every

BS Act S and for

every family (fi )iI of homomorphisms fi : (Xi )S
.

BS the mapping f : (Xi )S BS with f (x) = fi (x) for x Xi is the unique
iI

S-homomorphism, such that f ui = fi for all i I.

f. is called the coproduct induced morphism by (fi )iI with respect to

(Xi )S and is denoted by [(fi )iI ].
iI

Denition 2.11 A right S-act is called decomposable, if there exist two

subacts BS and CS , such that AS = BS CS and BS CS = . In this case,
AS = BS CS is called a decomposition of AS . Otherwise AS is called
indecomposable.
If AS is an indecomposable right S-act and f : AS BS is an epimorphism,
then BS is indecomposable.

Theorem 2.12 Every right S-act AS has a unique decomposition into indecomposable subacts.
Denition 2.13 An element z S is called left (right) zero of S, if zt =
z (tz = z) for all t S and zero of S, if z is a left and right zero.
S is called (left) (right) zero monoid, if all elements of S \ {1} are (left)
(right) zeros of S.
Denition 2.14 An element s of S is called right (left) invertible, if there
exists s S , such that ss = 1 (s s = 1). In this case s is called a right
(left) inverse of s.
Denition 2.15 An element c of a monoid S is called right (left) cancellable, if rc = tc (cr = ct) for r, t S implies r = t.
The monoid S is called right (left) cancellative, if all its elements are right
(left) cancellable.
Denition 2.16 An element e of a monoid S is called idempotent, if e2 = e.
The set of all idempotents of S is denoted by E(S). If E(S) = S, then S is
called an idempotent monoid or a band.

10

2 PRELIMINARIES

Denition 2.17 Let S =< a >1 := {1, a, a2 , ...} be a monogenic semigroup

with generating element a and identity adjoined. If there exist natural numbers
k, l such that k = l but ak = al , then let m be the smallest natural number
with am = am+q for some q N. Then m is called the index or height of a.
Denition 2.18 A monoid is called periodic, if all its monogenic submonoids
are nite.
Denition 2.19 An element s of a monoid S is called (left) nilpotent, if
there exists n N, such that sn = z S, where z is a (left) zero of S.
A monoid is called (left) nil, if all elements of S are (left) nilpotent.
Denition 2.20 A monoid S is called left (right) reversible, if any two
right (left) ideals of S have a non-empty intersection. This is equivalent to
the following condition: for all s, t S there exist u, v S such that su =
tv (us = vt).
Note, that this condition is sometimes called left (right) Ore condition.

Denition 2.21 A monoid S is called left (right) collapsible, if for every

s, t S there exists z S such that zs = zt (sz = tz).
Denition 2.22 Let S be a monoid and let s S. The smallest right (left)
ideal of S containing s is sS (Ss). It is called the principal right (left)
ideal generated by s.
S is called a principal right (left) ideal monoid, if all its right (left) ideals
are principal right (left) ideals.
Recall, that for monoids the Greens relations R and J are dened by
sRt if and only if sS = tS and
sJ t if and only if SsS = StS.
Remark 2.23 If S contains a left zero z, then for all s S the Rees congruence sS on S, given by the principal right ideal sS, is the principal congruence
(sz, s).
Furthermore, even if in general is not a principal right congruence on S, if
S contains a zero z, then (sz, z) = (z, z) = is considered as a principal
right congruence.

11

Denition 2.24 A monoid S is called (right) simple, if S has no proper

(right) ideals (i.e., no (right) ideals I with I = S.)
A right ideal I of S is called simple, if I contains no proper right subideals.
Denition 2.25 A monoid S is called
a right (left) Rees monoid, if all right (left) congruences on S are
Rees congruences,
a right (left) principal Rees monoid, if all right (left) principal congruences on S are Rees congruences.
Denition 2.26 ([18], [10], [4]) A right ideal I of S is called left stabilizing,
if for every i I there exists k I with ki = i.
Note that the condition of Denition 2.26 is called Condition (LU) in [14].
The categorical denition of free acts is not given here. It can be found for
instance in [14]. The categorical denition of projective acts is included in the
denition of (X , Y)-projective acts given at the beginning of Chapter 3.
Result 2.27 A right S-act AS is
free , if it is isomorphic to a coproduct
non-empty set I.

S of copies of S for some

iI

([15]) projective, if it is isomorphic to a coproduct

jJ

ej S of cyclic right

S-acts, each of them generated by an idempotent ej E(S), j J = .

Proposition 2.28 For every right S-act AS there exists a free right S-act FS ,
such that AS is an epimorphic image of FS .
Proposition 2.29 Every retract of a projective act is projective.
Denition 2.30 ([13] and others) A right S-act GS is called a generator in
Act S if and only if there exists an epimorphism : GS SS (which in

SS ).
this case is a retraction with : SS GS such that = idS , i.e. GS A

Result 2.31 ([13],[17]) For every AS Act S the coproduct SS AS is a

generator in Act S (with retraction : SS AS SS such that |S =


idS , (AS ) =: z, and the natural injection SS SS AS as coretraction) if
and only if z is a left zero of S.

12

2 PRELIMINARIES

For later use, we recall the denitions of atness properties and torsion freeness.
For the denition of pullbacks and of the tensor product in Act S see [14].
Denition 2.32 A right S-act AS is called
pullback at, if the tensor functor AS S preserves pullbacks.
at, if the tensor functor AS S preserves monomorphisms.
weakly at, if the tensor functor AS S preserves all monomorphisms
from left ideals of S into S.
principally weakly at, if the tensor functor AS S preserves all
monomorphisms from principal left ideals of S into S.
torsionfree, if ac = a c implies a = a for all a, a A, c S, c right
cancellable.
Furthermore, AS satises condition (P ), if for a, a A, s, s S the
equation as = a s implies the existence of a A, u, v S, such that
a = a u, a = a v and us = vs .
Now we recall some results concerning properties of Rees factor acts and homological classication, which will be used in Chapter 6.

Proposition 2.33 Let I be a right ideal of S. The Rees factor act (S/I)S is
(a) free i |I| = 1 ([14]).
(b) projective i |I| = 1 or I = S and S has a left zero ([14]).
(c) pullback at i |I| = 1 or I = S and S is left collapsible ([14]).
(d) satises condition (P ) i |I| = 1 or I = S and S is right reversible
([18],[10]).
(e) (weakly) at i S is right reversible and I is left stabilizing ([18],[10]).
(f ) principally weakly at i I is left stabilizing ([18],[10]).
(g) torsionfree i sc I implies s I for every s, c S, c right cancellable
([14]).

13

Corollary 2.34 [14] The one element right S-act S is

(a) free i S = {1}.
(b) projective i S has a left zero.
(c) pullback at i S is left collapsible.
(d) satises condition (P ) i S is right reversible.
(e) (weakly) at i S is right reversible.
(f ) principally weakly at.
(g) torsionfree.
In [14] the assertion (b) of Proposition 2.33 is not explicitly given in this form,
but is given by 2.33 (b) for proper right ideals of S and 2.34 (b).
The same is valid for condition 2.33 (d).
Conditions (c) of Proposition 2.33 and Corollary 2.34 are proved in [14] by
using that for cyclic right S-acts pullback atness is equivalent to a so-called
condition (E).

Result 2.35 All right S-acts

(a) are free i S = {1} ([20]).
(b) are projective i S = {1} ([12]).
(c) are pullback at i S = {1} ([12]).
(d) satisfy condition (P ) i S is a group ([19]).
(e) are (weakly) at i S is regular and satises condition (R), i.e., for all
s, t S w Ss St such that w (s, t) s ([14]).
(f ) are principally weakly at i S is regular.
(g) are torsionfree i every right cancellable element of S is right invertible
([16]).
Note that these results can also be found in [14].

3 (X , Y)-PROJECTIVITIES OF ACTS

14

(X , Y)-projectivities of acts

In this Chapter we come back to the question of the introduction: What happens, if we restrict in the denition of projectivity the class of epimorphisms?
The answer is introduced at rst, namely the concepts of (X , Y)-projectivity
(Section 3.1).
During this Chapter equivalences between these properties and to projectivity
itself are proved (see for instance Proposition 3.31) as well as implications between the remaining concepts.
The results with respect to equivalences are summarized in Section 3.6, Table
1, which leads to the restricted table (Table 2). Table 2 shows the (X , Y)projectivities remaining for further investigations.
The implications between these properties are illustrated in the scheme at the
end of this chapter (Section 3.8).
Moreover, subsection 3.3.3 and 3.3.4 contain results relative to (X , Y)-projectivity
of coproducts of acts, which will be used in Chapter 5.

3.1

Notation and basic informations

Let S be a monoid, and let X , Y denote classes of right S-acts.

A right S-act AS is called (X , Y)-projective, if for every XS X , YS
Y every S-homomorphism f : AS YS can be lifted in Act S with respect
to every epimorphism g : XS YS , i.e., there exists an S-homomorphism f 
such that the following diagram is commutative:
AS
f
XS

YS

A right S-act AS is called trivially-(X , Y)-projective, if there does not exist

an epimorphism from XS onto YS or if there does not exist a homomorphism
from AS to YS , i.e., if the requirement above is empty.
A right S-act AS is called non-trivially-(X , Y)-projective, if AS is (X , Y)projective and not trivially-(X , Y)-projective.

3.1 Notation and basic informations

15

In the following we will study (X , Y)-projectivity for dierent classes X and

Y.
The classes X , Y considered here are the classes of
all right S-acts,


free right S-acts FS
SS ,
=
iI

projective right S-acts PS
(ej S)S for e2j = ej S,
=
jJ

generators GS , i.e., right S-acts GS such that there exists an epimorphism

: GS SS (which in this case is a retraction with : SS GS

such that = idS , i.e., GS A SS ),

factor acts (S/)S of S by

(arbitrary) right congruences ,
right principal congruences, i.e.,  = (x, y) for some x, y S,
right Rees congruences I for right ideals I S,
right principal Rees congruences sS for s S,
coproducts

mM

(S/m )S of factor acts of S by right congruences on S,

where dierent sets of congruences will be considered,

and nally AS itself,
where the index sets I, J and M are supposed to be non-empty sets.
In the following, I, J and M denote non-empty sets. This will not be mentioned explicitly in every situation.
Instead of (X , Y)-projectivity of special classes of acts, the abbreviation (X, Y )projectivity with X X , Y Y, will be used.
In a concrete situation it will be clear by the structure of XS and YS used in
this abbreviation, which classes X and Y are considered, respectively.
The acts GS and HS will represent the class of generators in Act S, BS and
CS the class of all acts.
In this context we sometimes call XS rst and YS second component of
(X, Y )-projectivity.
Whenever an assertion is made for an arbitrary class X of acts, we simply write
XS Act S, i.e., the class X is not specied. (Analogously: YS Act S.)

3 (X , Y)-PROJECTIVITIES OF ACTS

16

Some of the properties have already been considered but named dierently.
These are:
Remark 3.1 Let S be a monoid, let AS , BS , CS Act S, let s, x, y S,
let I denote a right ideal of S and let  be a right congruence on S. Then
(B, C)-projectivity is known as projectivity (cf. [14] et al. ),
(S, S/)-projectivity is called is weak projectivity (wp) in [17],
(S, S/(x, y))-projectivity is called principal weak projectivity ( (pwp) in
[17],
(S, S/I)-projectivity is called Rees weak projectivity (Rwp) in [17],
(S, S/sS)-projectivity is called principal Rees weak projectivity (pRwp) in
[17] and
(A, C)-projectivity is called quasi projectivity (qp) in [1].

We will use these terms simultaneously to the respective (X, Y )-projectivity.

The notation x-congruence with x {, p, R, pR} will be used in the following for right, right principal, right Rees or right principal Rees congruences,
respectively, where denotes the empty symbol.
If we consider coproducts

iI

XS of copies of the same act XS we denote the

i-th copy of XS in this coproduct by XSi and the i-th natural injection by ui.
Replacing X and Y in (X , Y)-projectivity by the special acts mentioned before
leads to 226 formally dierent combinations: There are 15 dierent rst/second
components XS and YS , which leads to 225 pairs (X, Y ). Furthermore, for arbitrary acts BS we obtain (B, B)-projectivity.
Our rst aim is to exclude those (X , Y)-projectivities, which are trivially(X , Y)-projectivities, especially this means that we suppose the existence of
an epimorphism g : XS YS for some XS X , YS Y.
Then the amount of combinations is reduced to 196 by the following observations:
If the rst component is an indecomposable act then the second component is indecomposable, too, for it is the epimorphic image of the rst
act.

3.2 (X, Y )-projectivities with projective second component

17

If the rst component is S itself, the second component has to be isomorphic to a factor act of S, since it is an epimorphic image of S.
Moreover, epimorphic images of factor acts of S are isomorphic to factor
acts of S itself.
As a consequence of this observations we get for instance, that (S, C)-projectivity
has not to be considered anymore, since it is equivalent to (S, S/)-projectivity.
Equivalences of this kind cause the reduction to the amount of 196 pairs (X, Y ).
During the next sections further equivalences will be proved.

(X, Y )-projectivities with projective second component

3.2

In this section it will be shown, that some of the projectivities dened before,
are properties of all acts AS Act S, and are in this sense trivial.
Proposition 3.2 Let BS Act S and let ej E(S), j J. Then every


right S-act AS is (B,
ej S)-projective.
Since

Proof.

onto
BS

jJ

jJ

jJ

ej S is projective, every epimorphism from an act BS

(ej S)S is a retraction, i.e., there exists a homomorphism such that

jJ

(ej S)S , where = id
ej S .

jJ

Let now AS be an act.

(ej S)S is an homomorphism and if : BS
(ej S)S is an
If f : AS
jJ

jJ

epimorphism, then for f : AS BS the equality (f ) = f holds. Thus

AS is (B,
ej S)-projective.
jJ

As a consequence, all (X , Y)-projectivities with projective second component

will be left out of consideration in what follows.

3.3

Equivalences of conditions

During this section equivalences between (formally) dierent (X, Y )-projectivities

will be presented. In particular, equivalences to the concepts of weak projectivities (see Remark 3.1) will be given.

3 (X , Y)-PROJECTIVITIES OF ACTS

18

As preparation of some proofs at rst the following general results are given.
Note that general results concerning (X, Y )-projectivity of coproducts and
products will be given right before using them.
Lemma 3.3 Let X , Y denote classes of right S-acts. If for all YS Y there
exists an XS X and an epimorphism : XS YS , then (X , Z)-projectivity
implies (Y, Z)-projectivity for all classes Z of right S-acts.
Let X X , Y Y, Z Z, let AS be an (X, Z)-projective act, let
: Y Z be an epimorphism and let f : A Z be an homomorphism.
By assumption, there exists XS X and an epimorphism : X  Y .
Since AS is (X , Z)-projective, there exists f  : A X  with ()f  = f
and thus f  : A Y is an homomorphism with (f  ) = f , i.e., AS is
(Y, Z)-projective.

Proof.

Furthermore, since every act is the epimorphic image of a free act (Proposition
2.28), we obtain by Lemma 3.3:
Corollary 3.4 Let XS , YS Act S. Then (

S, Y )-projectivity implies

iI

(X, Y )-projectivity.

Since coproducts of cyclic acts appear as rst component, it is useful to get

some more information about epimorphisms from these coproducts onto S-acts
YS . The results will be used for example to prove Proposition 3.12.
Lemma 3.5 Let AS =

jJ

aj S be a coproduct of cyclic acts. If g :

iI

XSi

AS is an epimorphism, then for every j J there exists an i I such that

g |X i = gui is an epimorphism onto aj S.
Let j J and x g 1 (aj ), aj Aj . Then x X i for some i I,
and every a
aj S is the image of an element xs of X i because there exists
s S, such that the equality a
= aj s = g(x)s = g(xs) holds. Therefore g|X i
is an epimorphism from X i onto aj S.

Proof.

Corollary 3.6 Let XS , AS = aS Act S. If g :

iI

XSi AS is an

epimorphism, then there exists an i I such that g |X i = gui is an epimorphism

onto AS .

3.3 Equivalences of conditions

19

In Proposition 2.29 we saw that retracts of projective acts are projective. The
analogue is true for (X, Y )-projectivity:
Lemma 3.7 Let X , Y denote classes of right S-acts. Then retracts of (X , Y)projective right S-acts are (X , Y)-projective.
Let X X , Y Y, let AS , AS Act S let : A A be
a retraction with coretraction and let AS be (X , Y)-projective. Let g :
X Y be an epimorphism and let f : A Y be a homomorphism. Then
f : A X is a homomorphism, which by assumption can be lifted with
respect to g, i.e., there exists (f ) : A X with g(f ) = f . Thus for
(f ) : A X the equation g(f ) = f = f holds, i.e., AS is (X, Y )projective.

Proof.

After this general results with respect to (X , Y)-projectivity of acts, in the

following we consider the situation for special rst/second components.

3.3.1

Cyclic acts and coproducts of cyclic acts as second component

The following implication will be used to prove further equivalences:

Lemma 3.8 Let x {, p, R, pR}, let  be a right x-congruence on S and let
XS Act S. If AS is (S, S/)-projective, then AS is (X, S/)-projective.
Let g : X S/ be an epimorphism and let  : S S/ be the
canonical epimorphism. Since SS is projective, there exists an homomorphism
 with g =  . Since AS is (S, S/)-projective, every homomorphism f :
AS S/ can be lifted with respect to  , i.e., there exists f  with  f  = f .
This yields g( f  ) =  f  = f , and thus f can be lifted relative to g.

Proof.

Corollary 3.9 Let x {, p, R, pR} and let  be a right x-congruence on S.

Then for AS Act S the following assertions are equivalent
(i) AS is (

S, S/)-projective,

iI

(ii) AS is (S, S/)-projective.

3 (X , Y)-PROJECTIVITIES OF ACTS

20

Proof.

(i) (ii) is obvious.

(ii) (i) is Lemma 3.8.

The analogous equivalence is true even for coproducts of arbitrary acts as rst
component, but surely this can not be directly derived by Lemma 3.8:
Proposition 3.10 Let x {, p, R, pR}, let  be a right x-congruence on S,
let X be a class of right S-acts and let XS , XSi X , i I.
Then for AS Act S the following assertions are equivalent:
(i) AS is (

X i , S/)-projective,

iI

(ii) AS is (X, S/)-projective.

(i) = (ii) is obvious.

i
X S/ there
(ii) = (i). By Corollary 3.6 for every epimorphism g :
Proof.

iI

exists an i I such that g |X i = gui is an epimorphism. By (ii) for f : A

S/ there exists an homomorphism f  : AS X i with guif  = f and thus f
can be lifted with respect to g, i.e., (i) is valid.

Remark 3.11 In the following, (

X i , S/)-projectivity will be considered

iI

only by using the equivalent property of (X, S/)-projectivity.

For cyclic acts we obtain another chain of equivalent conditions:
Proposition 3.12 Let x {, p, R, pR}, let , m , m M, be right xcongruences on S, let X be a class of right S-acts and let XS , XSi X , i I.
Then for a cyclic act S/ the following assertions are equivalent:
(i) S/ is (

iI

(ii) S/ is (

X i,

mM

S/m )-projective,

X i , S/)-projective,

iI

(iii) S/ is (X, S/)-projective.

(i) = (ii) is obvious.
(ii) (iii) is Proposition 3.10.



i
S/m be a homomorphism and g :
X
(iii) = (i). Let f : S/
Proof.

mM

iI

3.3 Equivalences of conditions

mM

21

S/m an epimorphism. Since S/ is cyclic, there exists a k M with

f (S/) S/k . By Lemma 3.5, for S/k there exists a j I, such that

i
X denotes
g vj is an epimorphism from X j onto S/k , where vj : X j
iI

the j-th canonical injection. By (iii) there exists f  : S/ X j , such that

g vj f  = f . Thus f can be lifted with respect to g by using vj f  .

3.3.2

Cyclic acts and coproducts of cyclic acts as rst component

Proposition 3.13 Let m , m M, be right congruences on S, let ej

E(S), j J, and let YS Act S. Then for every AS Act S the following
assertions are equivalent:
(i) AS is (
(ii) AS is (
(iii) AS is (

S, Y )-projective,

iI

mM

jJ

S/m , Y )-projective,

ej S, Y )-projective.

(i) = (ii) is valid by Corollary 3.4.

(ii) = (iii). Every cyclic act ej S is isomorphic to a factor act S/j of S by a
right congruence (Proposition 2.9). Thus (ii) yields (iii).
(iii) = (i) since 1 E(S).

Proof.

Corollary 3.14 Let  be a right congruence on S, let e E(S) and let YS

Act S. Then for AS Act S the following assertions are equivalent:
(i) AS is (S, Y )-projective,
(ii) AS is (S/, Y )-projective,
(iii) AS is (eS, Y )-projective.

3 (X , Y)-PROJECTIVITIES OF ACTS

22

Remark 3.15 The equivalences presented in Proposition 3.13 use 1S = S =

S/.
If S does not contain a zero, then (x, y) = for all x, y S. Thus for right


principal congruences (
S/(xn , yn ), Y )-projectivity in general does not imply (

nN

S, Y )-projectivity, i.e., for coproducts of monocyclic acts as rst com-

iI

ponent, one can not expect equivalences like in Proposition 3.13.

If S is a monoid without left zero, then S/I 
= S for every right ideal I of S
and thus one also can not expect equivalences like those of Proposition 3.13 for
coproducts of Rees factor acts as rst component.
Since (S, S/)-projectivities are known as weak projectivities, we summarize
the results given in Proposition 3.13 and Corollary 3.9 for Y = S/ in the next
Corollary.
Corollary 3.16 Let x {, p, R, pR}, let , m , m M, be right congruences
on S, let be a right x-congruence on S and let let e, ej E(S), j J. Then
the following assertions are equivalent
(i) AS is (

mM

S/m , S/)-projective,

(ii) AS is (S/, S/)-projective,

(iii) AS is (

S, S/)-projective,

iI

(iii) AS is (S, S/)-projective,

(iv) AS is (

jJ

ej S, S/)-projective,

(v) AS is (eS, S/)-projective.

3.3.3

Coproducts of (X , Y)-projective acts

Since coproducts of acts appear as components in some (X, Y )-projectivities

it is useful to study properties of coproducts of acts. This will be done in this
subsection. The importance of this investigation will become clear in Lemma
5.18. For the respective properties of products see the next subsection.

3.3 Equivalences of conditions

23

Proposition 3.17 Let X, Y Act S and let Aj , j J, be indecomposable

right S-acts. If
Aj is (X, Y ) projective
for all j J, then

Aj is (X, Y ) projective.

jJ

Proof.

Let Ai Act S, i J, let ui : Ai

jJ

Aj denote the natural

injections, and let Ai be (X, Y )-projective for all i J. Consider for every
i J the following diagram:
Ai
ui

jJ

Aj
f

Then f ui is a homomorphism from Ai into Y , which can be lifted by assumption, i.e, there exists (f ui ) with g (f ui ) = f ui.
By the universal property of the coproduct f can be lifted with respect to g


by the coproduct induced homomorphism [((f ui ) )iJ ], i.e.
Aj is (X, Y )jJ

projective.

The next example shows, that in general, the converse is not true. But in the
case of monoids with left zero we nally get Proposition 3.19.
Example 3.18 [17] Let S be the three element right zero semigroup {a, b, c}


with identity adjoined. Then the coproduct S/aS S/(a, b) is (S, S/(x, y))projective but (S/(a, b))S is not.
Proposition 3.19 Let S be a monoid with left zero, let X, Y Act S and
let Aj , j J, be indecomposable right S-acts. Then
Aj is (X, Y ) projective
for all j J, i


jJ

Aj is (X, Y ) projective.

3 (X , Y)-PROJECTIVITIES OF ACTS

24

Necessity has been proved in Proposition 3.17.

Suciency. Since S contains a left zero, every act Aj has at least one zero zj .


Aj Ak with pk (a) = a, a Ak ,
Then for every k J the mapping pk :
Proof.

jJ

and pk (a) = zk , a Aj , j = k, is an epimorphism with pk uk = idAk , i.e. ,

every Aj is a retract of
Aj . Since
Aj is (X , Y)-projective, by Lemma
jJ

jJ

3.7 every Aj is (X , Y)-projective.

If Y is the class of coproducts of Rees factor acts, we obtain the analogous

equivalence in the case of arbitrary monoids because of the existence of zeros
in Rees factor acts.

Lemma 3.20 Let Ik , k K, be right ideals of S, let XS Act S and let

Aj , j J, be indecomposable right S-acts. Then


Aj is (X,

jJ

S/Ik ) projective

kK

i
Aj is (X,

S/Ik ) projective

kK

for all j J.
Suciency is included in Proposition 3.17.






Aj be (X,
S/Ik )-projective and let f : Aj
S/Ik
Necessity. Let
Proof.

jJ

kK

kK

be an homomorphism. Since Aj is an indecomposable act there exists k K

such that f (Aj ) S/Ik .




Let 0k denote the zero in S/Ik . Dene a homomorphism h :
Aj
S/Ik
jJ

0

kK

by h(a) = f (a) if a Aj and h(a) = k otherwise. Then h uj = f .

S/Ik there exists h
By assumption for every epimorphism g : X
kK

with gh = h. Therefore there exists a homomorphism h uj : Aj X with

ghuj = huj = f , i.e., Aj is (X,
S/Ik )-projective.
kK

3.3 Equivalences of conditions

3.3.4

25

Products of (X , Y)-projective acts

In this subsection we shortly consider products of (X, Y )-projective acts.


Aj of acts is the cartesian product of the acts Aj
Recall that the product
jJ

with the natural projections and componentwise multiplication.

Products are the categorically dual construction to coproducts and are therefore in some sense related to the categorically dual concepts to projectivities,
namely to injectivities. Indeed, for injective acts the analogue assertion to
Proposition 3.17 is true, i.e., products of injective acts are injective [14], which
can be proved by using the universell property of the product.
Since this property yields morphisms into the product we can not expect the
product of (X , Y)-projective acts to be (X , Y)-projective. But in the case of
monoids with left zero we obtain:
Proposition 3.21 Let S be a monoid with left zero, let X, Y Act S and
let Aj , j J, be indecomposable right S-acts. If


Aj is (X, Y ) projective,

jJ

then
Aj is (X, Y ) projective
for all j J.
Proof.

Let Aj Act S, j J, and let pj :


jJ

Aj Aj denote the

natural projections.
Since S contains a left zero, every act Aj has at least one zero. Let zj denote a

Aj with ik (a) = (xj )jJ
zero in Aj . Then for k J the mapping ik : Ak
jJ

with xk = a and xj = zj , j = k, is a homomorphism with pk ik = idAk , i.e.,


Aj .
every Ak is a retract of the product
jJ

Thus by Lemma 3.7 we obtain, that (X, Y )-projectivity of the product

implies (X, Y )-projectivity of Aj , j J.


jJ

Aj

Even if S contains a left zero, the converse is not true:

Example 3.22 Consider the monoid (Z2 , ). Then Z2Z2 is projective in Act Z2 .

The product Z2 Z2 is a four-element, indecomposable right S-act and is

therefore not isomorphic to a coproduct of cyclic acts. Thus Z2 Z2 is not
projective.

3 (X , Y)-PROJECTIVITIES OF ACTS

26

Remark 3.23 From the proof of Proposition 3.21 it became clear, that for coproducts of right S-acts (Aj )S , each of them containing a zero, the implication
holds even if S itself has no (left) zero (for instance, if every Aj is a Rees
factor act of S).
Since products of right S-acts do not occur as rst or second component of the
projectivities considered her, we do not study the analogue situations to those
in Lemma 3.20.

3.3.5

Rees factor acts and coproducts of Rees factor acts as second

component

Recall that by Proposition 3.10 we do not have to distinguish between (

X i , S/)-

iI

projectivity and (X, S/)-projectivity. In the case of coproducts of Rees factor

acts as second component and arbitrary rst ones we obtain an in some sense
analoguous assertion :

Proposition 3.24 Let I, Ik , k K, be right ideals of S and let X Act S.

Then for AS Act S the following assertions are equivalent:
(i) AS is (X,

kK

S/Ik )-projective,

(ii) AS is (X, S/I)-projective.

Proof.
(i) = (ii) is obvious.
(ii) = (i) Consider in Act S the following diagram:

A
f

X
Then X =

kK

g 1(S/Ik ) and A =

g


kK

kK

S/Ik

f 1 (S/Ik ).

Now consider for every k K the diagram:

A
fk

gk

S/Ik

3.3 Equivalences of conditions

27

with fk (a) = f (a) if a f 1 (S/Ik ), and fk (a) = 0k otherwise, where 0k denotes
the zero in S/Ik , and gk (x) = g(x), if x g 1 (S/Ik ), gk (x) = 0k otherwise.
Since by (ii) AS is (X, S/I)-projective, for every k K there exist a homomorphism fk : A X, such that fk = gk fk . This implies that fk (f 1 (S/Ik ))
g 1(S/Ik ). Therefore f  : A X dened by f  (a) = fk (a), a f 1 (S/Ik ),
is a well dened homomorphism with gf  = f . Hence, (i) is valid.

Corollary 3.25 Let sS, sl S S, l L, be principal right ideals of S, and let

XS Act S. Then for AS Act S the following assertions are equivalent:
(i) AS is (X,

lL

S/sl S)-projective,

(ii) AS is (X, S/sS)-projective.

Noticing, that the zero in Rees factor acts is an essential part of the proof of
Proposition 3.24, we get the same equivalence for arbitrary factor acts of S if
we demand S to be a monoid with left zero:
Lemma 3.26 Let S be a monoid with left zero, let , m , m M, be right
(principal) congruences on S and let XS Act S. Then for AS Act S
the following assertions are equivalent:
(i) AS is (X,

mM

S/m )-projective,

(ii) AS is (X, S/)-projective.

Proof.

Analogously to the proof of Proposition 3.24.

Note, that in Chapter 5 monoids fullling condition (Moz) will be treated and
this condition will be proved to be weaker than the existence of a left zero. By
taking a look to the proof of Lemma 3.24, it becomes clear, that this condition
is sucient for the equivalence presented in Lemma 3.26.

3.3.6

Rees factor acts and coproducts of Rees factor acts as rst

component

In the previous subsection we saw, that we do not have to distinguish between

Rees factor acts and coproduct of Rees factor acts as second component. If
these acts are considered as rst component in (X, Y )-projectivity, we do not
get an analogue assertion, but in this case we obtain:

3 (X , Y)-PROJECTIVITIES OF ACTS

28

Proposition 3.27 Let I, Ik , k K, be right ideals of S, let s, sl S, l L,

and let YS Act S. Then for AS Act S the following assertions are
equivalent:
(i) AS is (
(ii) AS is (

kK

lL

S/Ik , Y )-projective,

S/sl S, Y )-projective.

(i) = (ii) is obvious.

(ii) = (i) For every Ik let sk Ik . Then sk S Ik . Thus there exists an




epimorphism :
S/sk S
S/Ik and Lemma 3.3 completes the proof.
Proof.

kK

kK

Corollary 3.28 Let I be a right ideal of S, let s S, and let YS Act S.

Then for AS Act S the following assertions are equivalent:
(i) AS is (S/I, Y )-projective,
(ii) AS is (S/sS, Y )-projective.

Thus by Proposition 3.27 and Corollary 3.28 (S/I, Y )- and (

projectivities can be left out of consideration in what follows.

3.3.7

kK

S/Ik , Y )-

Various rst/second components

In this subsection we consider (X , Y)-projectivities for various classes X and

Y, and equivalences between them. Note that, for later use, we include even
those equivalences in the following results, which have already been given before.

Lemma 3.29 Let m , m M, be right congruences on S, let GS be a generator in Act S, let ej E(S), j J, and let BS , YS Act S. Then for
AS Act S the following assertions are equivalent:
(i) AS is (B, Y )-projective,
(ii) AS is (G, Y )-projective,
(iii) AS is (

iI

S, Y )-projective,

3.3 Equivalences of conditions

(iv) AS is (
(v) AS is (

mM

29

S/m , Y )-projective

ej S, Y )-projective.

jJ

The equivalences (iii) (iv) (v) are parts of Proposition 3.13.

The implications (i) = (ii) = (iii) are obvious.
The implication (iii) = (i), valid by Corollary 3.4, completes the proof.
Proof.

Especially for weak projectivity like dened in [17] Corollary 3.9 together with
Lemma 3.29 yields:
Corollary 3.30 Let x {, p, R, pR}, let  be a right x-congruence on S,
and let GS be a generator in Act S. Then for AS Act S the following
assertions are equivalent:
(i) AS is (G, S/)-projective,
(ii) AS is (S, S/)-projective.

In the next result the most surprising equivalence is (xi) (vi):

Proposition 3.31 Let m , m M, be right congruences on S, let BS , CS

Act S, let GS denote a generator in Act S and let ej E(S), j J. Then
for AS Act S the following assertions are equivalent:
(i) AS is (B, A)-projective,
(ii) AS is (G, A)-projective,
(iii) AS is (
(iv) AS is (
(v) AS is (

S, A)-projective,

iI

mM

jJ

S/m , A)-projective,

ej S, A)-projective,

(vi) AS is projective, i.e., AS is (B, C)-projective,

(vii) AS is (G, C)-projective,
(viii) AS is (

iI

S, C)-projective,

3 (X , Y)-PROJECTIVITIES OF ACTS

30

(ix) AS is (
(x) AS is (

mM

jJ

S/m , C)-projective,

ej S, C)-projective,

(xi) AS is (B, B)-projective.

The equivalences (i) (ii) (iii) (iv) (v) and (vi)
(vii) (viii) (ix) (x) are a part of Lemma 3.29.
The implications (vi) = (v) and (vi) = (xi) are obvious.


(v) = (vi) By Proposition 2.28 there exists a free act F (A)S
S, such
=
Proof.

that there exists an epimorphism :

iI

iI

the following diagram:

S A. For this consider in Act S

A
id

iI

By (v) there exist an homomorphism : A

S such that the diagram

iI

commutes, i.e., AS is a retract of a free and thus projective act and is therefore
projective by Proposition 2.29.
(xi) = (vi) By Proposition 2.28, AS is the factor act of a free act F (A)S ,
i.e., there exists an epimorphism : F (A) A. Consider in Act S the
following diagram
A
id
(

nN

F n (A)

nN

An )

nN

F n (A)

nN

An )

with id : A A1 , id(a) = a1 denotes the identity on AS and where the

epimorphism
g:


nN

F n (A)


nN

An


nN

F n (A)

An

nN

given by

F 1 A1
g(f 1 ) = (f 1 )
n
n+1
g : A A , g(an ) = an+1 , n > 1

n
F F n1, g(f n ) = f n1, n > 1

3.4 Projectivities which are trivial or equivalent to other projectivities

31

where f i , i N, denotes the element f F (A)S in the i-th copy of F (A)S in

the coproduct
F n (A) and ai denotes the element a AS in the i-th copy
nN

of AS in the coproduct

nN

A.

By (xi) the diagram is commutative, i.e., there exists id : A

nN

F n (A)

nN

with g id = id. This implies id (A) F 1 (A). Thus g |F 1 id = id = id,
i.e., AS is a retract of a free, and therefore projective act, and is therefore
projective.

3.4

Projectivities which are trivial or equivalent to

other projectivities

In this section it turns out, that some of the (X , Y)-projectivities are equivalent
to others, whenever they are non-trivially-(X , Y)-projectivities.
Lemma 3.32 Let AS Act S.
AS is (S, A)-projective i AS is cyclic projective or trivially-(S, A)-projective.
Let g : S A be an epimorphism. Then A
= S/ker g, i.e., AS is a
cyclic act. Since AS is (S, A)-projective, for idA there exists a homomorphism
idA with g idA = idA , i.e., AS is a retract of S and is therefore projective.
If there is no epimorphism from S onto AS , AS is trivially-(S, A)-projective.

Proof.

Lemma 3.33 If AS Act S is a non-trivially-(A, G)-projective right S-act,

then AS is a generator in Act S.
Let GS be a generator in Act S, i.e., GS A SS . If there exist an

epimorphims g : A G, then g : A S is an epimorphism, i.e., AS is a

generator in Act S.

Proof.

For (X, G)-projectivity with X = A, we need some preparing results to nally

get Lemma 3.37. These are given right now.
= SS in Act S, which is an
Lemma 3.34 If there exists a generator GS
epimorphic image of S, then S is innite.

An

3 (X , Y)-PROJECTIVITIES OF ACTS

32

Proof.

Suppose S is nite. Let GS be a generator in Act S, i.e., GS

A SS ,

then |S| |G|. Let g : S G be an epimorphism. Then |G| |S|. Thus g

is an S-isomorphism, i.e. GS
= SS .

Lemma 3.35 If S is nite, then all right-S-acts AS are (S, G)-projective.

Proof.
Since S is nite, SS is the unique generator in Act S, which is an
epimorphic image of S by Lemma 3.34. Thus by the projectivity of S every
morphism from an arbitrary act AS to GS can be lifted with respect to every
epimorphism from S onto a generator.

We now change the rst component to be a Rees factor act by a principal right
ideal of S. In this case it turns out that a non-trivially-(S/sS, G)-projective
act is already (S, G)-projective.
Proposition 3.36 There exists an epimorphism from a Rees factor act
(S/sS)S of S by a principal right ideal onto a generator GS in Act S i
S contains a left zero.
Let s S, let : S/sS G be an epimorphism and let 0 denote
the zero in S/sS. Then (0) is a zero element in GS . Since GS is a generator,
there exists an epimorphism : G S. Thus ((0)) =: z is a left zero in
S.
If z is a left zero in S, then S/zS = S and S itself is a generator.

Proof.

Lemma 3.37 Let S be a monoid with left zero and let s S.

For AS Act S the following assertions are equivalent
(i) AS is (S/sS, G)-projective,
(ii) AS is (S, G)-projective.
Proof.

(i) = (ii) Let AS be (S/sS, G)-projective and let z denote the left

zero in S.
Then S/zS = S leads to AS is (S, G)-projective.
(ii) = (i) by Lemma 3.3.

Corollary 3.38 Let S be a monoid with left zero and let sl S, l L. Then
for AS Act S the following assertions are equivalent

3.5 When (S, S/)- and (S/, S/)-projective acts are (S, S/)-projective 33

(i) AS is (
(ii) AS is (

lL

S/sl S, G)-projective,
S, G)-projective.

iI

Summarizing, we get, that non-trivially-(S/sS, G)-projective acts are already

(S, G)-projective, and non-trivially-( S/sl S, G)-projective acts are ( S, G)iI

lL

projective.
Note, that therefore these projectivities will not be included in the implication
scheme at the end of this chapter.

3.5

When (S, S/)- and (S/, S/)-projective acts are

(S, S/)-projective

We start with a useful proposition concerning (S, S/)-projectivities. This result has also been formulated in another way in [17].

Proposition 3.39 ([17]) Let  be a right congruence on S and let f : AS

(S/)S be a homomorphism. Then f can be lifted with respect to every epimorphism g : SS (S/)S i f can be lifted with respect to the canonical
epimorphism  : SS (S/)S .
Proof.

For an epimorphism g : S S/ consider in Act S the following

diagram:
SS

SS

(S/)S

Since S is projective, there exists an endomorphism  of S such that g  =  .

Now let AS Act S and let f : A S/ be a homomorphism, which can
be lifted with respect to  , i.e. there exists f  : A S with  f  = f . Then
for  f  : A S the equation g  f  =  f  = f holds, i.e. f can be lifted
with respect to g.
The converse is obvious.

3 (X , Y)-PROJECTIVITIES OF ACTS

34

Remark 3.40 Let m , m M, be right congruences on S and let f : AS

(S/m )S be a homomorphism. Then f can be lifted with respect to every
mM

epimorphism g :

iI

SS

mM

(S/m )S i f can be lifted with respect to the

coproduct of the canonical epimorphism (

mM

m ) :

mM

SS

mM

(S/m )S .

This assertion can be proved by using the projectivity of coproduct of copies of

S in the same way like before.

Lemma 3.41 Let K1 and K2 denote dierent sets of right congruences on S,

such that for every  K1 there exists an K2 with . If AS Act S
is (S, S/)- and (S/, S/)-projective, then AS is (S, S/)-projective.
By Proposition 3.39 it is sucient to consider the canonical epimorphisms, respectively.
Let  K1 and let  : S S/ denote the canonical epimorphism.
Consider in Act S the following diagram:
Proof.

A
f
S

S/

By assumption there exists K2 with . Let denote the canonical epimorphism from S onto S/. By the Homomorphism Theorem for acts
(Theorem 2.7) there exists a unique homomorphism  with  =  .
Since AS is (S/, S/)-projective, there exists f  : A S/ with  f  = f .
Since AS is (S, S/)-projective, there exists f  : A S with f  = f  .
Thus there exists a homomorphism f  : A S with
 f  =  f  =  f  = f,
i.e. f can be lifted with respect to  and therefore with respect to every
epimorphism g : S S/, i.e. AS is (S, S/)-projective.

An interesting aspect of this Lemma turns out, if the consequences for weak
projectivities are explicitly given:

3.5 When (S, S/)- and (S/, S/)-projective acts are (S, S/)-projective 35

Corollary 3.42 Let s, x, y S, let  be a right congruence on S and let I be

a right ideal of S. Then for every AS Act S the following assertions hold:
(a) If AS is (S, S/(x, y))- and (S/(x, y), S/)-projective, then AS is weakly
projective.

then AS is Rees
(b) If AS is (S, S/(x, y))- and (S/(x, y), S/I)-projective,
weakly projective.
(c) If AS is (S, S/(x, y))- and (S/(x, y), S/sS)-projective, then AS is principally Rees weakly projective.

(d) If AS is (S, S/sS)- and (S/sS, S/I)-projective,

then AS is Rees weakly
projective.
Proof.
By Lemma 3.41 in each case the existence of suitable congruences
remains to be shown:
(a) If  is an arbitrary right congruence on S with x  y, then (x, y) , for
it is the smallest right congruence on S such that x and y are in the same
congruence class.
(b), (c) Let J be a (principal) right ideal of S, let J denote the Rees congruence
let j J and s S \ {1}. Then (j, js)  .
on S by J,
J
(d) Furthermore for the Rees congruence on S by the principal right ideal jS
with j J we have jS J.

In general we can not expect assertions analogue to those of Corollary 3.42.

For instance, if  is a right congruence on S, such that (S/)S does not have
a zero, there does not exist a Rees congruence I on S with I :
If I , then there exists an epimorphism g : S/I S/, which yields a
zero in (S/)S contradicting (S/)S being an act without zero.

3 (X , Y)-PROJECTIVITIES OF ACTS

36

3.6

Table of results

We tabularize the results obtained so far.

The acts in the rst column of the table denote the rst component of the pair
(X, Y ), the acts in the rst row the second one. Thus every array in the table
belongs to a pair (X, Y ) and therefore to (X, Y )-projectivity. We start with
180 pairs, omitting the pairs with projective second component (Proposition
3.2) and the pair (B, B), since (B, B)-projectivity is equivalent to projectivity
(Proposition 3.31).
Arrays belonging to (X, Y )-projectivities, which are already excluded by the
observations in section 3.1, are marked by o.
Arrays belonging to (X, Y )-projectivities, which are equivalent to others, are
marked by the number of the respective Proposition/Lemma/Corollary. If the
related property is equivalent to one of the properties projectivity (proj), quasi
projectivity (qp), wp, pwp, Rwp, pRwp, this is marked by the respective abbreviations.
By () arrays are marked, which belong to projectivities, which are equivalent
to others, whenever they are non-trivially-projectivities (see section 3.4).

Notations:
I, J, K, L, M, N are non-empty sets,
, m , m M, are right congruences on S,
, n , n N, are right principal congruences on S,
e, ej E(S), j J,
Ik , k K, are right ideals of S,
I,
s, sl S, l L.

3.6 Table of results

37

CS

AS

GS

BS

proj

proj

3.29

AS

qp

3.29
gen/
triv

GS

proj

proj

3.29

3.29

proj

proj

()

proj

proj

S/m

S/

mM

S/n

S/

nN

wp

kK

3.29

pwp

S/Ik

3.24

S/I

lL

Rwp

3.24
wp

3.29

wp

S/sl S

3.25

S/sS

pRwp

3.25

pwp

Rwp

Rwp

3.25

pRwp

pwp

Rwp

Rwp

pRwp

pRwp

iI

SS

S/m

wp

pwp

Rwp

pRwp

3.29

3.29

wp

3.29

pwp

Rwp

Rwp

pRwp

pRwp

3.14

3.14

wp

pwp

Rwp

pRwp

proj

proj

3.29

3.29

wp

3.29

pwp

Rwp

Rwp

pRwp

pRwp

3.14

3.14

wp

pwp

Rwp

pRwp

3.10

3.24

3.10

3.25

3.10

mM

S/

ej S

jJ

eS

3.10

S/n

nN

S/

kK

S/Ik

S/I

lL

3.27

3.27

3.27

3.27

3.10

3.27

3.10

3.24

3.10

3.25

3.10

3.28

3.28

3.28

3.28

3.28

3.28

3.10

3.24

3.10

3.25

3.10

()

S/sl S

S/sS

()

3.10
o

Table 1: Table of Results

For further investigations, the concepts of projectivity, weak projectivity, principal weak projectivity, Rees weak projectivity, principal Rees weak projectivity and (X, Y )-projectivities, where the pair (X, Y ) belongs to an empty place
in this table, remain.
For the aim of lucidity, we restrict our following investigations to those projectivities with rst component = A, except of Chapter 4, where further investi-

3 (X , Y)-PROJECTIVITIES OF ACTS

38

gations on projectivities of acts over monoids with left zero will be made. AS
as rst component is left out, because (A, X)-projective acts should be treated
in the context of quasi projectivity (see Remark 3.1) and, furthermore, since
all other remaining rst components are coproducts of dierent kinds of cyclic
acts and these are in some sense related to each other.
Thus, for further investigations, we can reduce Table 1 in the following way:
we delete all rows and columns, which are totally lled in Table1, taking
care, that at least one of the equivalent conditions, respectively, remains,
we delete the second row, where A is the rst component.
The resulting table is Table 2. In this table, the abbreviation xwp, x
{, p, R, pR}, is used for x weakly projective and it marks the places in the
table, which belong to the pair (X, Y ), for which (X, Y )-projectivity will be
used as dening property for x weak projectivity in the following. Note that
these are exactly the equivalences given by the original denition in [17]. Then
we obtain the following:

CS

AS

proj.

proj

()

GS

mM

S/m

S/

nN

S/n

3.9

S/

S/I

S/sS

3.9

3.9

3.9

pwp

Rwp

pRwp

3.10

3.10

3.10

3.10

3.10

3.10

iI

SS


nN

S/n

S/


lL

3.10

S/sl S

S/sS

wp

()

()

3.10

Table 2: Remaining pairs (X, Y )

3.7 Further implications

3.7

39

Further implications

Some implications are quiet clear by denition or by the results obtained so

far. Following, implications which are not obvious will be proved, for giving
an implication scheme at the end of this chapter.
Lemma 3.43 Let  be a right congruence on S and let GS denote a generator
in Act S. Then (S, S/)-projectivity implies (S, G)-projectivity.
Proof.
If there exists an epimorphism g from S onto GS , then GS is a factor
act of S. Thus, since AS is (S, S/)-projective, every homomorphism from AS
to GS can be lifted with respect to g. Thus AS is (S, G)-projective.
Otherwise, AS is trivially-(S, G)-projective and is therefore (S, G)-projective.

Lemma 3.44 Let XS Act S and let s, x, y S. Then (S/(x, y), X)projectivity implies (S/sS, X)-projectivity.
If there exist an s S with |sS| = 1, then s is a left zero in S. Then
every right principal Rees congruence s S is the principal congruence (s s, s )
and the implication holds.
Let s S with |sS| 2. Then for u S we get: (s, su) sS is valid.
Thus there exists an epimorphism : S/(s, su) S/sS and Lemma 3.3
completes the proof.

Proof.

Corollary 3.45 Let XS Act S, let sl S, l L, xn , yn S, n N.

Then (
S/(xn , yn ), X)-projectivity implies ( S/sl S, X)-projectivity.
nN

Remark 3.46 In the following, (

lL

nN

S/(xn , yn ), G)-projectivity and

(S/(x, y), G)-projectivity will left out of consideration (see Lemma 3.37 and
Corollary 3.38 together with Lemma 3.44 and Corollary 3.45).
Lemma 3.47 Let s S, let AS Act S and let  be a right congruence on
S. If AS is (S/sS, S/)-projective, then AS is (S/sS, A)-projective.
If there exists an epimorphism g : S/sS A, then AS
= S/ for
a right congruence  on S and thus the implications holds. Otherwise AS is
trivially-(S/sS, A)-projective and there is nothing to show.

Proof.

3 (X , Y)-PROJECTIVITIES OF ACTS

40

3.8

Implication scheme

The implications between the projectivities introduced before are presented

in the following scheme, where (X, Y )-projectivity is abbreviated by the pair
(X, Y ). The notations are equal to those in Table 1 (2), except in the case
of monocyclic acts (S/)S , which will be denoted by (S/(x, y))S , x, y S.
Additionally, let x , y  , t S and let j , j J be principal right congruences
on S.
By reason of lucidity, the implication scheme is divided in part 1 and part
2. Projectivities included in both parts are marked by double framed boxes.
These are the places, where part 1 an part 2 can be combined to one scheme.
Remind, that concepts like (S/sS, G)-projectivity, which are trivially or equi

valent to other concepts as well as (
S/(xn , yn ), G)- and (S/(x, y), G)nN

projectivity, are left out. Furthermore, projectivities with rst component A,

which in some sense belong to quasi projectivity, will be left out, for they wont
be considered any further, except of Chapter 4.
Then we obtain:

3.8 Implication scheme

41

projective

q
S/n ,C)

nN

S,G)

iI

(S,S/)

(S,G)

S/n ,

nN

lL

S/m )

mM

S/sl S,

S/n ,A)

nN

S/m )

mM

s

lL

(S/(x,y),A)

S/sl S,C)

lL

S/sl S,A)

S/sS,S/

Implication scheme part 1

(S/sS,A)

3 (X , Y)-PROJECTIVITIES OF ACTS

42

projective

S,

iI

iI

S,

S/m )

mM

S/n )

(S,S/)

nN

(S,S/(x,y))

s

S/j ,

jJ

S/m )

mM

S/n )

(S/(x,y),S/(x ,y  ))

lL

S/sl S,

S/n )

nN

(S/sS,S/(x,y))

(S/(x,y),S/I)

s (S/(x,y),S/tS)

(S/sS,S/I)

(S/(x,y),S/)

nN

(S,S/I)

(S,S/tS)

S/n ,

nN

(S/sS,S/tS)

Implication scheme part 2

lL

S/sl S,

S/m )

mM

(S/sS,S/)

3.9 Comments

3.9

43

Comments

The representation of monoids by endomorphisms of sets has lead to acts and

vice versa. It is well known, that the representations of rings by endomorphisms of abelian groups leads to modules over rings and vice versa. Thus, in
some sense they are related to acts over monoids.
In this Chapter various projectivity properties have been introduced for acts
over monoids. What is the situation like in the category of (right) modules?
Also in the theory of modules over rings some projectivities have been studied.
Projective modules are dened analogously to projective acts by the categorical denition, i.e, P is called projective, if the functor MorM od (P, ) preserves
epimorphisms.
In 1966, Y. Miyashita ([24]) introduced a generalization of projective modules,
so called quasi-projective modules and many authors followed him. Some of
the weaker concepts studied by now are given by the following denitions (cf.
[26], [23]):
Let M, N, A be modules. A module P is called
(1) M-projective, if for every submodule A of M, for every epimorphism
g : M M/A and for every homomorphism f : P M/A there
exists f  , such that the following diagram commutes:
P
f
M

M/A

(2) self-projective (quasi projective), if P is P -projective.

(3) epi-projective relative to M, if P is M-projective with respect to all
epimorphisms f : P M/A.
(4) self-epi-projective, if P is epi-projective relative to P .
(5) semi-projective, if P is self-projective with respect to epimorphisms g :
P N, N a submodule of P .
(6) direct-projective, if every direct summand of a decomposition P = A1 A2
is a retract of of P .

3 (X , Y)-PROJECTIVITIES OF ACTS

44

For these properties we get the following implication scheme:

(1)

(2)

(3)

(4)

(5)

(6)

Most of these concepts could be considered as (X , Y)-projectivities with X

consisting of one element. A more general investigation with respect to modules, comparable with (X , Y)-projectivities with dierent classes X and Y, is
not known to me.
In [23] the notation weak projectivity occurs, which is called kostetig in [26].
There is no connection to weak projectivity of acts, as dened in [17].
The same is true for weak projective covers of a module, introduced by Park
and Kim in [25] in 1991. Here the adjective weak is related to the whole
expression projective cover of (the respective module P is supposed to be
projective), weak projective modules do not occur in this article.
Another concept of projectivity occurs in [8] for arbitrary categories:
In a category C a class P Ob (C) is called (E, M)-pseudoprojective, if for
arbitrary objects A, B P and for every D C, for which there exists C P
and a monomorphism m : D C, every homomorphism f : A D can
be lifted with respect to every epimorphism e : B D, i.e., there exists f  ,
such that the following diagram is commutative:


f
f

In the category Act S we see, that (E, M)-pseudoprojectivity is not directly

comparable with (X , Y)-projectivity of acts, since it is a property of a class of
acts, and even if we suppose the class to consist of one element only, we do
not obtain a concept like the one introduced in Chapter 3.

3.9 Comments

45

Nevertheless, the following equivalence holds:

A class P of S-acts is (E, M)-pseudoprojective if and only if every P P is
(P, P  )-projective, where the class P  is given by P  = {D Act S | Q
P such that there exists a monomorphism m : D Q}.
As indicated by the respective properties of modules, it is a quite natural way
to ask for weaker concepts of projectivity by restricting the denition to special
homomorphisms f , i.e., to look at properties like epi-projectivity.
In the category of acts, epi-(X , Y)-projectivity, i.e., (X , Y)-projectivity with
respect to epimorphisms f , leads to weaker concepts than considered in this
dissertation.
Another interesting property is (B, B/A)-projectivity for arbitrary acts BS
and Rees factor acts (B/A)S by subacts AS of BS (see the denition of Mprojectivity of modules).
Moreover, (E, M)-pseudoprojective classes of acts could be studied, where the
results concerning homological classication of monoids by (X , Y)-projectivities of acts (Chapter 5) could be used because of the equivalence presented
before.
A further interesting idea is, to consider (X , Y)-projectivity of acts, where Y
is a class of amalgamated coproducts of acts.
These are not considered here.

46

4 SUMMARY AND SUPPLEMENTS

Summary and Supplements

As mentioned at the end of Chapter 3, page 38, in this Chapter we consider

(X, Y )-projectivities, where AS itself as rst component is permitted. Thus,
in this situation the restriction of Table 1 is given by the following table (Table
3). (For the used abbreviations and for informations see the explanations with
respect to Table 1 in Section 3.6. (pages 36, 37)

CS
AS

AS

mM

S/m

S/

nN

S/n

S/

S/I

S/sS

3.9

3.9

3.9

pwp

Rwp

pRwp

3.10

3.10

3.10

3.10

3.10

3.10

gen
or
triv

qp

GS

proj

proj

()

3.9

iI

SS


nN

S/n

S/


lL

3.10

S/sl S

S/sS

wp

()

()

3.10

Table 3: Remaining pairs (X, Y ) for monoids with left zero

In the following we will summarize the results of Chapter 3 with special attention to projectivities with rst component S. Recall, that these are called
weak, principally weak, Rees weak and principally Rees weak projectivity in
[17].
Nevertheless, the main aspect is, to investigate (X , Y)-projectivities of acts
over a monoid with left zero. For (X , Y)-projectivities of acts the presence of

4.1 Further equivalences in the case of monoids with left zero

47

a left zero plays an important role, among other things, since in this case every
act contains a zero. We obtain further equivalences, which are tabularized in
Table 4 of Section 4.2.
Questions concerning homological classication with respect to acts over monoids
with left zero are discussed in the general context in Chapter 5.
It will turn out, that in the case of a monoid with left zero, the remaining properties are mainly the weak projectivities and those projectivities with
rst component A, i.e., (X, Y )-projectivities which are related to the concept
of quasi-projectivity.
After proving further implications, these are collected in an implication scheme
in Section 4.3.

4.1

Further equivalences in the case of monoids with

left zero

Recall, that if S has a left zero z, then every right principal Rees congruence
by the right ideal sS is the right principal congruence (sz, s) by Remark 2.23.
First of all, equivalences of projectivities with arbitrary second component Y
will be given. They will be used in the proofs of the following (ve) theorems.
Lemma 4.1 Let S be a monoid with left zero z, and let sl , xn , yn S (l
L, n N). Then for all YS Act S the following assertions are equivalent:
(i) AS is (
(ii) AS is (
(iii) AS is (

S, Y )-projective,

iI

nN

lL

S/(xn , yn ), Y )-projective,

S/sl S, Y )-projective.

(i) = (ii) is a part of Corollary 3.4.

(ii) = (iii) is valid, since every Rees factor act by a right principal ideal is a
factor act by a right principal congruence by the existence of a left zero in S
(Remark 2.23).
(iii) = (i). For the left zero z the Rees factor act S/zS is S itself.

Proof.

48

4 SUMMARY AND SUPPLEMENTS

Corollary 4.2 Let S be a monoid with left zero z, and let s, x, y S. Then
for all YS Act S the following assertions are equivalent:
(i) AS is (S, Y )-projective,
(ii) AS is (S/(x, y), Y )-projective,
(iii) AS is (S/sS, Y )-projective.
In the following, we consider equivalences to projectivity and to (S, X)-projectivity, starting with the strongest property in 4.1.1, i.e. with projectivity.

4.1.1

Equivalences to projectivity

Theorem 4.3 Let m , m M, be right congruences on S, let GS denote a

generator in Act S, let BS , CS Act S and let ej E(S), j J. Then
for AS Act S the following assertions are equivalent:
(i) AS is (B, C)-projective (i.e., AS is projective),
(ii) AS is (B, A)-projective,
(iii) AS is (G, A)-projective,
(iv) AS is (
(v) AS is (
(vi) AS is (

S, A)-projective,

iI

mM

S/m , A)-projective,

ej S, A)-projective,

jJ

(vii) AS is (G, C)-projective,

(viii) AS is (
(ix) AS is (
(x) AS is (

S, C)-projective,

iI

mM

jJ

S/m , C)-projective,

ej S, C)-projective,

(xi) AS is (B, B)-projective.

Furthermore, let Ik , k K, be right ideals of S and let sl , xn , yn S (l
L, n N). If S contains a left zero, the assertions (i) to (xi) are equivalent
to:

4.1 Further equivalences in the case of monoids with left zero

(xii) AS is (
(xiii) AS is (
(xiv) AS is (
(xv) AS is (
(xvi) AS is (
(xvii) AS is (

kK

lL

nN

kK

lL

S/Ik , A)-projective,

S/sl S, A)-projective,

49

S/(xn , yn ), A)-projective,
S/Ik , C)-projective,

S/sl S, C)-projective,

nN

S/(xn , yn ), C)-projective,

(xviii) AS is (B, G)-projective,

(xix) AS is (
(xx) AS is (
(xxi) AS is (
(xxii) AS is (

S, G)-projective,

iI

kK

lL

S/Ik , G)-projective,

S/sl S, G)-projective,

nN

S/(xn , yn ), G)-projective.

Proof.
The equivalences (i) to (xi) are given in Proposition 3.31.
(i) = (xix) is obvious.


(xix) = (i) Let g :
S A be an epimorphism. Consider in Act S the

following diagram:

iI

A
idA

iI

A

Since S contains a left zero, by Result 2.31 the coproduct A S is a generator

in Act S. Consider now in Act S the following diagram
A
idA
(

iI

S)

g idS

50

4 SUMMARY AND SUPPLEMENTS

where g idS : (

iI

S)

S A

S with g idS (s) = g(s) if s

S and

iI

g idS (s) = s if s S is the coproduct of the morphisms idS and g. By (xix)

there exists idA : A ( S) S, such that the diagram is commutative,
iI

i.e., (g idS ) idA = idA . By the denition of g idS , this implies idA (A)
Thus idA is a homomorphism from A to
retract of

iI

S.

iI

S with g idA = idA , i.e., AS is a

S and is therefore projective by Proposition 2.29.

iI

The equivalences (xii) (xiii), (xv) (xvi) and (xx) (xxi) are parts of
Proposition 3.27.
The equivalences (xiii) (xiv), (xvi) (xvii) and (xxi) (xxii) are parts
of Lemma 4.1.
The implications (xix) = (xv) and (xix) = (xviii) follow from (xix) =
(i).
The implications (xv) = (xii), (xvii) = (xxii) and (xviii) = (xix) are
obvious.
(xiii) = (xix) and (xxi) = (xix): Since S contains a left zero z, by
Lemma 4.1 (xxi) implies (xix) and (xiii) = (iv) = projective, which has
been proved to be equivalent to (xix).

4.1.2

Equivalences to (S, S/)-projectivity

Theorem 4.4 Let , m , m M, be right congruences on S, let GS denote

a generator in Act S, let BS Act S and let e, ej E(S), j J. Then
for AS Act S the following assertions are equivalent
(i) AS is (S, S/)-projective (i.e., AS is weakly projective),
(ii) AS is (B, S/)-projective,
(iii) AS is (G, S/)-projective,
(iv) AS is (
(v) AS is (

S, S/)-projective,

iI

mM

S/m , S/)-projective,

(vi) AS is (S/, S/)-projective,

(vii) AS is (

jJ

ej S, S/)-projective,

4.1 Further equivalences in the case of monoids with left zero

51

(viii) AS is (eS, S/)-projective.

Furthermore, let R = be a set, let r , r R, be right congruences on S, let
Ik , k K, be right ideals of S, and let s, sl , x, y, xn , yn S (l L, n
I,
N). If S contains a left zero, the assertions (i) to (viii) are equivalent to:
(ix) AS is (B,
(x) AS is (G,
(xi) AS is (
(xii) AS is (
(xiii) AS is (
(xiv) AS is (
(xv) AS is (

rR

rR

S,

iI

mM

jJ

S/r )-projective,

rR

S/r )-projective,

S/m ,

ej S,

kK

kK

S/r )-projective,

rR

S/Ik ,

rR

S/r )-projective,

S/r )-projective,

rR

S/r )-projective,

S/Ik , S/)-projective,

S/)-projective,
(xvi) AS is (S/I,
(xvii) AS is (
(xviii) AS is (

lL

lL

S/sl S,

rR

S/r )-projective,

S/sl S, S/)-projective,

(xix) AS is (S/sS, S/)-projective,

(xx) AS is (
(xxi) AS is (

nN

nN

S/(xn , yn ),

rR

S/r )-projective,

S/(xn , yn ), S/)-projective,

(xxii) AS is (S/(x, y), S/)-projective,

Proof.

The equivalences (i) (vi) (viii) are parts of Corollary 3.14.

The equivalences (i) (iv), (v) (vi), (vii) (viii), (xv) (xvi), (xviii)
(xix) and (xxi) (xxii) are parts of Proposition 3.10.
The equivalences (ii) (iii) (iv) are given by Lemma 3.29.
Thus we get that for an arbitrary monoid S the equivalence of the assertions
(i) to (viii) holds.
The equivalences (ix) (x) (xi) (xii) (xiii) are a part of Lemma
3.29.

52

4 SUMMARY AND SUPPLEMENTS

The equivalences (xiv) (xvii) and (xv) (xviii) are given by Proposition
3.27.
(xviii) (xxi) and (xvii) (xx) (xi) are given by Lemma 4.1, since S
has a left zero.
(xx) (xxi) is valid by Lemma 3.26, since S contains a left zero.
Thus for a monoid S containing a left zero, the assertions (ix) to (xxii) are
equivalent.
The equivalence (iv) (xviii), given by Lemma 4.1, completes the proof.

4.1.3

Equivalences to (S, S/(x, y))-projectivity

Theorem 4.5 Let , m , m M, be right congruences on S, let GS denote

a generator in Act S, let BS Act S, let x, y S, and let e, ej
E(S), j J. Then for AS Act S the following assertions are equivalent
(i) AS is (S, S/(x, y))-projective (i.e., AS is principally weakly projective),
(ii) AS is (B, S/(x, y))-projective,
(iii) AS is (G, S/(x, y))-projective,
(iv) AS is (
(v) AS is (

S, S/(x, y))-projective,

iI

mM

S/m , S/(x, y))-projective,

(vi) AS is (S/, S/(x, y))-projective,

(vii) AS is (

jJ

ej S, S/(x, y))-projective,

(viii) AS is (eS, S/(x, y))-projective.

Furthermore, let R = be a set, let xr , yr , xn , yn , s, sl S (r R, n
Ik , k K, be right ideals of S. If S contains a left zero,
N, l L) and let I,
the assertions (i) to (viii) are equivalent to:
(ix) AS is (B,
(x) AS is (G,
(xi) AS is (

iI

nN

nN

S,

S/(xn , yn ))-projective,
S/(xn , yn ))-projective,

nN

S/(xn , yn ))-projective,

4.1 Further equivalences in the case of monoids with left zero

(xii) AS is (
(xiii) AS is (
(xiv) AS is (
(xv) AS is (

mM

S/m ,

ej S,

jJ

kK

kK

nN

nN

S/Ik ,

53

S/(xn , yn ))-projective,

S/(xn , yn ))-projective,

nN

S/(xn , yn ))-projective,

S/Ik , S/(x, y))-projective,

S/(x, y))-projective,
(xvi) AS is (S/I,
(xvii) AS is (
(xviii) AS is (

lL

lL

S/sl S,

nN

S/(xn , yn ))-projective,

S/sl S, S/(x, y))-projective,

(xix) AS is (S/sS, S/(x, y))-projective,

(xx) AS is (
(xxi) AS is (

rR

rR

S/(xr , yr ),

nN

S/(xn , yn ))-projective,

S/(xr , yr ), S/(x, y))-projective,

(xxii) AS is (S/(xr , yr ), S/(x, y))-projective,

In analogy to the proof of Theorem 4.4, using Corollary 3.14, Proposition 3.27, Lemma 3.29 and Lemma 4.1 for Y = S/(x, y) and Proposition
3.10 and Lemma 3.26 for right principal congruences  and m , m M.

Proof.

4.1.4

Equivalences to (S, S/I)-projectivity

Theorem 4.6 Let , m , m M, be right congruences on S, let GS denote a

Ik , k K, be right ideals of S,
generator in Act S, let BS Act S, let I,
and let e, ej E(S), j J. Then for AS Act S the following assertions
are equivalent

(i.e., AS is Rees weakly projective),

(i) AS is (S, S/I)-projective
(ii) AS is (B,

kK

S/Ik )-projective,

(iii) AS is (B, S/I)-projective,

(iv) AS is (G,

kK

S/Ik )-projective,

54

4 SUMMARY AND SUPPLEMENTS

(v) AS is (G, S/I)-projective,

(vi) AS is (
(vii) AS is (
(viii) AS is (
(ix) AS is (

S,

iI

kK

S/Ik )-projective,

S, S/I)-projective,

iI

mM

mM

S/m ,

kK

S/Ik )-projective,

S/m , S/I)-projective,

(x) AS is (S/, S/I)-projective,

(xi) AS is (
(xii) AS is (

ej S,

jJ

kK

S/Ik )-projective,

ej S, S/I)-projective,

jJ

(xiii) AS is (eS, S/I)-projective.

Furthermore, let R = be a set, let s, sl , x, y, xn , yn S (l L, n N),
and let O, Ir , r R, be right ideals of S. If S contains a left zero, the
assertions (i) to (xiii) are equivalent to:
(xiv) AS is (
(xv) AS is (

rR

rR

S/Ir ,

kK

S/Ik )-projective,

S/Ir , S/I)-projective,

(xvi) AS is (S/O, S/I)-projective,

(xvii) AS is (
(xviii) AS is (

lL

lL

S/sl S,

kK

S/Ik )-projective,

S/sl S, S/I)-projective,

(xix) AS is (S/sS, S/I)-projective,

(xx) AS is (
(xxi) AS is (

nN

nN

S/(xn , yn ),

kK

S/Ik )-projective,

S/(xn , yn ), S/I)-projective,

(xxii) AS is (S/(x, y), S/I)-projective.

4.1 Further equivalences in the case of monoids with left zero

Proof.

55

The equivalences (i) (x) (xiii) are parts of Corollary 3.14.

The equivalences (ii) (iii), (iv) (v), (vi) (vii), (viii) (ix), (xi)
(xii), (xiv) (xv), (xvii) (xviii) and (xx) (xxi) are included in
Proposition 3.24.
The equivalences (i) (vii), (ix) (x), (xv) (xvi), (xviii) (xix) and
(xxi) (xxii) are given by Proposition 3.10.
The equivalences (iii) (v) (vii) and (viii) (xi) are parts of Lemma
3.29.
So far for an arbitrary monoid S the assertions (i) to (xiii) are equivalent.
The equivalence (xv) (xviii) is given by Proposition 3.27.
(xviii) (xxi) is given by Lemma 4.1.
The last equivalence (xx) (vi), given by Lemma 4.1, completes the proof.

4.1.5

Equivalences to (S, S/sS)-projectivity

Theorem 4.7 Let , m , m M, be right congruences on S, let GS denote

a generator in Act S, let e, ej E(S), j J, let BS Act S, and
let s, sl S, l L. Then for AS Act S the following assertions are
equivalent
(i) AS is (S, S/sS)-projective (i.e., AS is principally Rees weakly projective),
(ii) AS is (B,

lL

S/sl S)-projective,

(iii) AS is (B, S/sS)-projective,

(iv) AS is (G,

lL

S/sl S)-projective,

(v) AS is (G, S/sS)-projective,

(vi) AS is (
(vii) AS is (
(viii) AS is (
(ix) AS is (

S,

iI

lL

S/sl S)-projective,

S, S/sS)-projective,

iI

mM

mM

S/m ,

lL

S/sl S)-projective,

S/m , S/sS)-projective,

(x) AS is (S/, S/sS)-projective,

56

4 SUMMARY AND SUPPLEMENTS

(xi) AS is (
(xii) AS is (

jJ

jJ

ej S,

lL

S/sl S)-projective,

ej S, S/sS)-projective,

(xiii) AS is (eS, S/sS)-projective.

Furthermore, let R = be a set, let x, y, xn , yn , t, sr S (n N, r R),
Ik , k K, be right ideals of S. If S contains a left zero, the
and let I,
assertions (i) to (xiii) are equivalent to:
(xiv) AS is (
(xv) AS is (

rR

rR

S/sr S,

lL

S/sl S)-projective,

S/sr S, S/sS)-projective,

(xvi) AS is (S/tS, S/sS)-projective,

(xvii) AS is (
(xviii) AS is (

kK

kK

S/Ik ,

lL

S/sl S)-projective,

S/Ik , S/sS)-projective,

S/sS)-projective,
(xix) AS is (S/I,
(xx) AS is (
(xxi) AS is (

nN

nN

S/(xn , yn ),

lL

S/sl S)-projective,

S/(xn , yn ), S/sS)-projective,

(xxii) AS is (S/(x, y), S/sS)-projective.

Using Corollary 3.14, Proposition 3.27, Lemma 3.29 and Lemma 4.1


S/sl S, respectively, for right Rees congruences 
with Y = S/sS or Y =
Proof.

lL

Xi
and m , m M, Proposition 3.10 with X
= S/Ik , and Corollary
= S/I,
3.25, one gets the equivalences in analogy to the proof of Theorem 4.6.

4.2

Summarizing table

By the results of section 4.1, in the case of S being a monoid with left zero,
Table 3 of page 46 turns into the following table (Table 4).
Arrays belonging to a pair (X, Y ), for which (X, Y )-projectivity is equivalent
to quasi-projectivity (qp) or x weak projectivity with x {, p, R, pR}, are
marked by the respective abbreviation (see informations to Table 1).

4.2 Summarizing table

57

If (X, Y )-projectivity has been proved to be equivalent to another projectivity,

this is marked by the number of the respective Proposition/Lemma/Corollary.
Like before, an array is marked by (), if it belongs to a pair (X, Y ), for which
(X, Y )-projectivity is equivalent to another, whenever it is a non-trivially(X, Y )-projectivity.

CS
AS

AS

mM

S/m

S/

nN

S/n

S/

S/I

S/sS

gen
or
triv

qp

GS

proj

proj

()

proj

proj

proj

wp

wp

pwp

pwp

Rwp

pRwp

wp

pwp

Rwp

pRwp

proj

wp

wp

pwp

pwp

Rwp

pRwp

()

4.2

wp

pwp

Rwp

pRwp

proj

proj

proj

wp

wp

pwp

pwp

Rwp

pRwp

()

4.2

wp

pwp

Rwp

pRwp

iI

SS


nN

S/n

S/


lL

S/sl S

S/sS

Table 4: Table of results for monoids with left zero

As indicated in the introduction of Chapter 4, in the case of S being a monoid

with left zero, the concepts of (A, Y )-projectivity with YS Act S, the
x-weak projectivities and (S, G)-projectivity remain for further investigations.

58

4.3

4 SUMMARY AND SUPPLEMENTS

Further implications, implication scheme

To summarize the remaining properties in the case of monoids with left zero
in an implication scheme, we prove further implications:
Lemma 4.8 Let S be a monoid with left zero, let XS Act S and let
s, x, y S. Then AS is (X, S/(x, y))-projective implies AS is (X, S/sS)projective.
Since S has a left zero z, every right principal Rees congruence sS
is the right principal congruence (s, sz) by Remark 2.23.

Proof.

Lemma 4.9 Let S be a monoid with left zero and let , m , m M, be right


S/m )congruences on S. If AS is (S, S/)-projective then AS is (A,
mM

projective.
Proof.

Let g : A

mM

S/m be an epimorphism, let f : A

be a homomorphism and let :=

mM

m :

mM

mM

coproduct of the canonical epimorphisms. By the projectivity of

exists  :

mM

mM

S/m

S/m be the

S, there

mM

S A, such that g  = . Since AS is (S, S/)-projective,

which by Theorem 4.4 is equivalent to AS being (

there exists f  : A

mM

iI

S,

mM
 

S/m )-projective,

S with f  = f . Thus for f : A A we get

g( f  ) = (g  )f  = f  = f and thus AS is (A,

mM

S/m )-projective.

Lemma 4.10 Let S be a monoid with left zero, let x {, p, R, pR} and
let  be a right x-congruence on S. If AS is (S, S/)-projective then AS is
(A, S/)-projective.
This implication is obtained by using the projectivity of S in the
same way like in the proof of the previous Lemma.

Proof.

4.3 Further implications, implication scheme

59

Thus, if S is a monoid with left zero z, the implication scheme part 1 and part
2, giving at the end of Chapter 3, turns into:

projective

qp

(A,A)

(S,G)

(A,

nN

S/n )

(A,

(A,G)

(S,S/)

S/m )

mM

(A,S/)

(S,S/I)

(S,S/(x,y))

(S,S/sS)

(A,S/(x,y))

(A,S/I)

s
(A,S/sS)

Implication scheme for S with left zero

60

5 HOMOLOGICAL CLASSIFICATION

Homological classication

In this Chapter questions concerning homological classication will be discussed, i.e., questions like When do all right S-acts have property A? or
When does property A imply property B? will be answered by characterizing internal properties of the monoid S.
The properties A and B considered here are all (X , Y)-projectivities as dened
in Chapter 3.
Whenever studying homological classication of monoids, it is useful to have
a look at the one element act S at rst. For instance, in Proposition 5.8 it
will be proved, that S is (S, S/sS)-projective if and only if S contains a left
zero. Since all right S-acts are ... implies S is ... , by the observations
concerning S we get necessary conditions on monoids over which all acts
have a special property. Therefore the second paragraph deals with (X , Y)projectivities of S .
In this context a new kind of monoids occurs: Monoids fullling condition
(Moz). They are introduced in the rst part and some basic results concerning
these monoids are given (see for instance Proposition 5.4).
In the rst section we begin with studying the question of all acts being (X , Y)projective, starting with the highest properties of the implication scheme part
1 (page 41), going to the highest one of part 2 (page 42) and nally ending at
the bottom, i.e., at all acts being (S/sS, S/tS)-projective.
For instance, in turns out, that all right S-acts are (S, S/sS)-projective if and
only if S is the disjoint union of a group and a left zero semigroup or S = {1}
(Corollary 5.68).
The next step is to study the conditions, under which further implications
between the concepts are valid. We will see, for example, that (S/sS, S/tS)projectivity implies (S/(x, y), S/tS)-projectivity if and only if S fullls condition (Moz) (Theorem 5.73). The respective results are the main content of
Section 5.2.
The results will be summarized in tables at the end of this chapter. They
will be used in Chapter 6 to make sure, that there are dierences between
the concepts of (X , Y)-projectivities and to prove, that there are no implications between (X , Y)-projectivities and atness properties as introduced in
Denition 2.32. Flatness properties are considered in this context, since like
(X, Y )-projectivities, they are properties of acts, which are weaker then projectivity.
A more detailed clue will be given right at the beginning of each subsection.

61

Condition (Moz)
In this paragraph monoids fullling condition (Moz) are introduced and some
results concerning these monoids are presented. In the case of monoids with at
least two idempotents a characterization of these monoids is given in Proposition 5.4. For monoids S with E(S) = {1}, there is no general characterization
so far. In the case of the groups (Zn , +), n N, we get a necessary condition
for fullling (Moz) in Lemma 5.5.

Denition 5.1 A monoid S is said to fulll condition (Moz), if every

monocyclic act has a zero.
Remark 5.2 If  = is a right congruence on S, then for (x, y)  we get
(x, y) . Thus there exists an epimorphism from (S/(x, y))S onto (S/)S .
Therefore, if S is a monoid, which fullls (Moz), then every factor act of S
by a right congruence  = has a zero.

Note, that if S is a monoid with left zero, then all acts have zeros and thus S
fullls condition (Moz).

Examples 5.3 (a) Every left zero semigroup with identity adjoined fullls
condition (Moz).
(b) The group (Z3 , +) fullls condition (Moz), since every monocyclic act
(Z3 /(x, y))Z3 is isomorphic to Z3 .
(c) The group (Z4 , +) does not fulll condition (Moz), since the monocyclic
act (Z4 /(1, 3))Z4 has no zero.
Proposition 5.4 For a monoid S the following assertions are equivalent:
(i) |E(S)| > 1 and S fullls condition (Moz),
(ii) S contains a left zero.
(i) = (ii) For e E(S), e = 1 consider the right principal congruence (e, 1). By Lemma 2.2 of [5], for x, y S and any e S the relation
x (e, 1) y is equivalent to em x = en y for some m, n 0. Since e E(S) this
yields:
x (e, 1) y ex = ey.

Proof.

62

5 HOMOLOGICAL CLASSIFICATION

Let now [z](e,1) denote the zero in S/(e, 1), i.e., [z](e,1) = [z](e,1) s = [zs](e,1)
for every s S. Thus for every s S we get z (e, 1) zs, which is equivalent
to ez = ezs for every s S. Therefore ez is a left zero in S.
(ii) = (i) If z is a left zero in S, then z 2 = z, i.e., z E(S) and thus
|E(S)| > 1. Furthermore, if S contains a left zero, then every right S-act has
a zero, which implies condition (Moz) for the monoid S.

Lemma 5.5 Let n N, n 5. If (Zn , +) fullls condition (Moz) then

n = 2k + 1 for k N.
Let {0, 1, 2, . . . , n 1} denote a representing system of the residue
classes of Z by n and let x, y Zn . Without loss of generality let y x.
Proof.

By Lemma 2.2 of [5] we get the following chain of equivalences:

x (0, 2) y

i, j N0 : i2 + x = j2 + y
i, j N0 : x y = 2(j i)
k1 , k2 , k3 , k4 N0 :
(x = 2k1 y = 2k2 )
(x = 2k3 + 1 y = 2k4 + 1)

Let now [z] be a zero in Zn /(0, 2). Then [z] = [z + (n 1)] and thus:
z (0, 2) (z + (n 1))

k1 , k2 , k3 , k4 N0 :
(z = 2k1 z + (n 1) = 2k2 )
(z = 2k3 + 1 z + (n 1) = 2k4 + 1)
k N : n 1 = 2k
k N : n = 2k + 1

(As I presume, even the converse is true, but it has not been proved so far.)
A more general result in view to monoids without idempotents e, e = 1, fullling condition (Moz) remains to be proved.
A useful property of monocyclic acts over monoids fullling condition (Moz)
is proved in the following Lemma, which will be used in the second paragraph
of subsection 5.1.3.
Lemma 5.6 Let S be a monoid, which fullls condition (Moz) and let u, x, y
S. Then every monocyclic act (S/(x, y))S is the epimorphic image of a Rees
factor act (S/uS)S of SS by a principal right ideal uS of S.

63

Let [u](x,y) be the zero in S/(x, y). Then we get us [u](x,y) for
every s S, i.e., I = uS [u](x,y) . Therefore uS (x, y) and thus by
the Homomorphism Theorem for acts (2.7) there exists an epimorphism from
S/uS onto S/(x, y).

Proof.

S is (X , Y)-projective
Like indicated before, in some of the proofs concerning results with respect
to homological classication of monoids, the one element act S , having the
respective property, will be used. Therefore, in this paragraph we study S
with respect to dierent (X , Y)-projectivities.
Lemma 5.7 Let X , Y be classes of right S-acts. If S Y, then S is
(X , Y)-projective if for all XS X the zero morphism z : XS S is a
retraction.
Moreover, this implies that every XS X has a zero.
Proof.
If S Y, then (X , Y)-projectivity of S implies, that for every
X X the identity id can be lifted with respect to the zero morphism
z : X , i.e., there exists id : X, such that z id = id , i.e, z is a
retraction and id () is a zero in X.

Ik , k
Proposition 5.8 Let , m , m M, be right congruences on S, let I,
K, be right ideals of S and let s, sl S, l L. Then the following assertions
are equivalent:
(i) S is (

S,

iI

mM

S/m )-projective,

(ii) S is (S, S/)-projective,

(iii) S is (

S,

iI

kK

S/Ik )-projective,

(iv) S is (S, S/I)-projective,

(v) S is (

iI

S,

lL

S/sl S)-projective,

(vi) S is (S, S/sS)-projective,

(vii) S contains a left zero,
(viii) S is (S, )-projective,
(ix) S is projective

64

5 HOMOLOGICAL CLASSIFICATION

Since S
= S/ = S/1S,  = S S, is a cyclic act, by Proposition
3.12 the equivalences (i) (ii), (iii) (iv) and (v) (vi) are valid.

Proof.

The implications (ii) = (iv) = (vi) are obvious.

(vi) = (vii) and (viii) = (vii). Since S
= S/1S, by (vi) (or (viii),
respectively) S contains a left zero by Lemma 5.7.
(vii) (ix) is a part of Corollary 2.34.
(ix) = (i) and (ix) = (viii) are obvious.

Proposition 5.9 Let x, y, xn , yn S, n N. Then the following assertions

are equivalent:
(i) S is (

S,

iI

nN

S/(xn , yn ))-projective,

(ii) S is (S, S/(x, y))-projective,

(iii) S is a monoid with left zero or no monocyclic act has a zero.
Since S
= S/ = S/1S,  = S S, is a cyclic act, by Proposition
3.10 the equivalence (i) (ii) is valid.

Proof.

(ii) = (iii). If there exists a monocyclic act S/(x, y) with zero z, then
f : S/(x, y) with f () = z is a homomorphism. By (ii) there exists a
homomorphism f  : S, which implies, that f  () is a left zero in S.
(iii) = (i). If S contains a left zero z, then S
= zS, which is projective and
thus (i) is valid.
Otherwise, there does not exist a homomorphism from S into a coproduct




S/(xn , yn ))-projective and
of monocyclic acts, i.e., S is trivially-( S,
therefore (

S,

iI

nN

iI

nN

S/(xn , yn ))-projective.

Note, that (iii) of Proposition 5.9 implies that S contains a left zero or S does
not fulll condition (Moz).

Proposition 5.10 Let s, sl S, l L, let x, y, xn , yn S, n N, and let

I, Ik , k K, be right ideals of S. Then the following assertions are equivalent:
(i) S is (

nN

S/(xn , yn ), ) -projective,

(ii) S is (S/(x, y), )-projective,

(iii) S is (

nN

S/(xn , yn ),

kK

S/Ik )-projective,

65

(iv) S is (S/(x, y), S/I)-projective,

(v) S is (

nN

S/(xn , yn ),

lL

S/sl S)-projective,

(vi) S is (S/(x, y), S/sS)-projective,

(vii) S fullls condition (Moz).
Since S
= S/(S S) = S/1S is a cyclic act, by Proposition 3.12
the equivalences (i) (ii), (iii) (iv) and (v) (vi) are valid.
Proof.

The implication (iv) = (vi) is obvious.

(ii) = (vii) and (vi) = (vii). Since S
= S/1S, for x, y S by (vi) or
(ii), respectively, Lemma 5.7 gives the existence of a zero in S/(x, y), i.e.,
(vii).
(vii) = (ii) and (vii) = (iv). If S contains a left zero z, then S
= zS,
which is projective, since z E(S). Thus (vii) = (ii) and (vii) = (iv).
If S is a monoid without left zero fullling condition (Moz), then for x, y S
the zero morphism z : S/(x, y) is a retraction, i.e., (vii) = (ii) by
Lemma 5.7.
Furthermore, since in this case every Rees factor act S/I has a unique zero
0, then f : S/I being a homomorphism implies f () = 0. Let g :
S/(x, y) S/I be an epimorphism, let z be a zero in S/(x, y). Then
g(
z ) = 0 and the mapping f  : S/(x, y) dened by f  () = z is a
homomorphism with g f  = f . Thus (vii) = (iv).

Note that the assertions made in the previous Proposition are not equivalent to S being (S/(x, y), S/)- and (S/sS, S/)-projective. The case of
(S/sS, S/)-projectivity will be discussed later on (Lemma 5.12 and Proposition 5.14).

Proposition 5.11 Let s, t, sl S, l L, and let I be a right ideal of S.

The one element right S-act S is
(a) (

lL

S/sl S, )-projective,

(b) (S/sS, )-projective,

(c) (S/sS, S/I)-projective,
(d) (S/sS, S/tS)-projective.

66

5 HOMOLOGICAL CLASSIFICATION

Since every coproduct of Rees factor acts has a zero, the zero mor

phism z :
S/sl S is a retraction. By Lemma 5.7 this yields (a), which
Proof.

lL

implies (b).
If S contains a left zero z, then S
= zS and is therefore projective, which
implies (c) and (d).
If S is a monoid without left zero, then every Rees factor act contains a unique
zero. Let 0sS denote the zero in S/sS and 0I the zero in S/I. Then for
an epimorphism g : S/sS S/I we get g(0sS ) = 0I . Furthermore, for a
homomorphism f : S/I the equation f () = 0I holds. Then f can be
lifted with respect to g by f  : S/sS with f  () = 0sS , i.e., (c), which
implies (d).

If the rst component is the coproduct of Rees factor acts by principal right
ideals, one could not use the same argumentation like in the proof of Proposition 5.11, if the second component is a factor act of S by a (principal) right
congruence  (or a coproduct of these), because there could exist two zeros
in (S/)S , one of them the image of S by the zero morphism, the other the
image of the zero element of (S/sS)S under an epimorphism. Then this morphism could not be lifted, i.e., S is not (S/sS, S/)-projective.
In the following we will see, that for left reversible monoids this case can not
arise.

Lemma 5.12 Let S be a monoid without left zero. Then every epimorphic
image of a Rees factor act by a proper right ideal contains a unique zero i S
is left reversible.
Suciency. Let I be a proper right ideal of S, let 0 denote the zero
in S/I and let g : S/I S/ be an epimorphism. Then g(0) =: z is a zero
of S/. Suppose z is a zero in S/. Since g is an epimorphism, there exists
s S/I such that g(
s) = z . Then for all t S we get:
Proof.

= g(
g(st)
s t) = g(
s) t = z t = z .
Thus g(
sS) = z .
If S is left reversible, there exists j sS I. Thus z = g(j) = z, i.e., the zero
in S/I is unique.
Necessity. Suppose S not to be left reversible, i.e., there exist right ideals I, J
with I J = . Then I J = S, since this would imply 1 J, contradicting
I J = . Then J = {j S/I | j J} is a subact of (S/I)S and the Rees
S is an epimorphic image of (S/I)S , which contains two
factor act ((S/I)/J)

67

zeros.
Thus the uniqueness of the zero implies left reversibility of the monoid.

Together with Lemma 3.5 this yields:

Corollary 5.13 Let S be a monoid without left zero, let Im , m M, be
right ideals of S, let m , m M, be right congruences on S and let g :




(S/Im )S
(S/m )S be an epimorphism. Then every factor act
mM

mM

(S/m )S has a unique zero i S is left reversible.

Proposition 5.14 Let S be a monoid, which fullls condition (Moz), let

s, sl S, l L and let , m , m M, be right congruences on S. Then the
following assertions are equivalent:
(i) S is (

lL

S/sl S,

mM

S/m )-projective,

(ii) S is (S/sS, S/)-projective,

(iii) S is left reversible or S contains a left zero.
(i) (ii) is valid by Proposition 3.12, since S is a cyclic act.
(ii) = (iii). Let S be a monoid without left zero. For every proper right ideal
I of S there exists a right congruence  on S, such that I . Thus there
exists an epimorphism g : S/I S/. For i I there exists an epimorphism
h : S/iS S/I. Thus gh : S/iS S/ is an epimorphism, too.
Let 0iS denote the zero in S/iS, which is unique, since S is a monoid without
left zero. Then gh(0iS ) =: z is a zero in S/.
For every zero z  in S/ the mapping f : S/ with f () = z  is a
homomorphism. By (ii), there exists a homomorphism f  : S/iS, such
that (gh)f  = f . By the uniqueness of the zero in S/iS we get f  () = 0iS .
This implies:
z  = f () = (gh)f () = (gh)(0iS ) = z,
Proof.

i.e., the zero in S/ is unique. Thus by Proposition 5.12, S is left reversible.
(iii) = (ii). If S is a monoid with left zero, then by Proposition 5.8 S is
projective and thus (S/sS, S/)-projective.
Let S be a left reversible monoid without left zero, let g : S/sS S/ be
an epimorphism and let 0sS denote the zero in S/sS. Then, by Lemma 5.12,
S/ contains a unique zero z. If f : S/ is a homomorphism, then
f () = z = g(0sS ). Thus for f  : S/sS with f  () = 0sS the equality
gf  = f holds, i.e., (ii).

68

5 HOMOLOGICAL CLASSIFICATION

Remark 5.15 Note, that by Corollary 2.34 the one-element right S-act fullls
condition (P ) if and only if S is at, which is the case if and only if S is
right reversible.

An important part of the proof of Proposition 5.14 was the fact, that epimorphic images of Rees factor acts are factor acts of S by some right congruences
on S. These congruences need not to be right principal congruences. Thus for
factor acts by right principal congruences as second component, one could not
use Lemma 5.12.

5.1

All acts are (X , Y)-projective

In this section properties of monoids over which all acts are (X , Y)-projective
are studied. The results concerning (X , Y)-projectivity of the one-element act
S will be used in various proofs.
At the beginning, we recall some results concerning homological classication
of monoids by projectivity, which are also important for the following proofs
For details, see ([14], [5] et al). For the same reason, the equivalences of Lemma
5.18 are given.
Result 5.16 (cf. [14]) All right S-acts are projective if and only if S = {1}.
Result 5.17 ([5]) All monocyclic right S-acts, i.e., all factor acts of the form
(S/(x, y))S for x, y S, are projective if and only if S = {1} or S = {0, 1}.
We start with a general result, which will be used in the next proofs:
Lemma 5.18 Let X , Y denote classes of right S-acts and let X X , Y Y.
Then the following assertions are equivalent:
(i) All right S-acts are (X, Y )-projective,
(ii) all Z Y are (X, Y )-projective,
(iii) for all Z Y every epimorphism g : XS ZS is a retraction.
Proof.
(i) = (ii) is obvious.
(ii) = (iii) Let Z Y and g : X Z be an epimorphism. Then by
(ii) the identity idZ on Z can be lifted with respect to g, i.e., there exists a
homomorphism with g = idZ . This yields (iii).
(iii) = (i) Let AS Act S. Consider in Act S the following diagram:

5.1 All acts are (X , Y)-projective

69

A
f

By (iii) there exists a homomorphism with g = idY , which implies g ( f ) =

f . Thus AS is (X, Y )-projective.

5.1.1

All are (X, G)-projective

This subsection deals with (X , Y)-projectivities, where the second component

is the class of all generators in Act S. Starting with rst components being
coproducts of copies of S, we shortly consider copies of Rees factor acts of
S by principal right ideals as rst components in Corollary 5.20 again, even
if they were left out in the implication scheme (part 1). At the end, (S, G)projectivity of all acts is studied.
Theorem 5.19 Let GS be a generator in Act S. Then the following assertions are equivalent:
(i) All right S-acts are (

S, G)-projective,

iI

(ii) all generators in Act S are (

S, G)-projective,

iI

(iii) all generators in Act S are projective,

(iv) S is a group,
(v) all right S-acts fulll condition (P ).
Proof.
(i) (ii) is a part of Lemma 5.18.
(ii) = (iii) Let HS be a generator in Act S. By Proposition 2.28, there




exists a free act
S and an epimorphism :
S HS .
iI

iI

By Lemma 5.18, is a retraction. Thus HS is a retract of a free and therefore

projective act and is therefore projective by Proposition 2.29.
(iii) = (ii) is obvious.
(iii) (iv) has been proved in [13], Theorem 3.8.
(iv) (v) by Result 2.35.

70

5 HOMOLOGICAL CLASSIFICATION

Remember that by Proposition 3.30 there exists a non-trivially-(

projective act if and only if S has a left zero. This leads to:

lL

S/sl S, G)-

Corollary 5.20 Let S be a monoid with left zero and let sl S, l L. Then
the following assertions are equivalent:
(i) All right S-acts are (

lL

S/sl S, G)-projective,

(ii) S = {1}.
(i) = (ii). Since S is a monoid with left zero, by Corollary 3.38 we




obtain that ( S/sl S, G)-projectivity is equivalent to ( S, G)-projectivity.

Proof.

lL

Thus (i) implies that all acts are (

iI

S, G)-projective. By Theorem 5.19, this

iI

is equivalent to S being a group. Since S has a left zero, this yields S = {1}.
(ii) = (i). If S = {1} then all acts are projective by Result 5.16 and thus


( S/sl S, G)-projective.
lL

Although it will turn out, that (S, G)-projectivity of all acts does not give
much informations about the internal properties of S, for the aim of completeness the respective result is given in Theorem 5.21.
In Lemma 3.35 it has been proved, that for a nite monoid S all acts are
(S, G)-projective. Thus studying all acts being (S, G)-projective is mainly interesting in the case of innite monoids.
By Lemma 5.18 and Proposition 3.18.6 of [14] we get:
Theorem 5.21 Let GS be a generator in Act S. Then the following assertions are equivalent:
(i) All right S-acts are (S, G)-projective,
(ii) all generators in Act S are (S, G)-projective,
(iii) every epimorphism : SS HS onto a generator HS in Act S is a
retraction,
(iv) all cyclic generators in Act S are projective,
(v) all cyclic generators in Act S are isomorphic to cyclic acts generated
by an idempotent e with eJ 1.

5.1 All acts are (X , Y)-projective

71

The equivalences (i) (ii) (iii) are parts of Lemma 5.18.

The equivalence (iv) (v) follows from Proposition 3.18.6 of [14].
(iii) = (iv). Let HS = (S/)S be a cyclic generator. Then by (iii) the
canonical epimorphism is a retraction. Thus HS is a retract of a projective act
and thus HS is projective by Proposition 2.29.
(iv) = (iii). Let : S H be an epimorphism onto a generator in
Act S. Then HS is cyclic and therefore by (iv) projective. Thus the identity
idH can be lifted with respect to , i.e., is a retraction.

Proof.

Recall that by Lemma 3.34 for nite monoids S every cyclic generator is isomorphic to SS , and that by Theorem 5.21 in this case all acts are obviously
(S, G)-projective.
By the following example it becomes clear, that in general condition (v) of
Theorem 5.21 is not only satised in a trivial way if S is innite, i.e., that eJ 1
does not imply (eS)S
= SS .
Example 5.22 Cyclic projective generators need not to be isomorphic to S
itself:
Let S be the monoid generated by the elements e, k, k  and the relations
e2 = e, ek = k, k  k = 1. Then eS is a cyclic projective generator in Act S
but eS and S are not isomorphic as right S-acts ([15]).
(Another example, due to B. M. Schein, can also be found in [15]).
Notes:
In Proposition 3.18.9 of [14] it has been proved, that if S is a group,
then every projective generator in Act S is isomorphic to S. Thus,
condition (iv) of Theorem 5.19 implies condition (v) of Theorem 5.21,


which is related to the implication ( S, G)-projectivity = (S, G)iI

projectivity.
The converse is not true (see Lemma 3.35).
Example 5.22 has already been used in Chapter 3 (Example 18.11) of
[14] to make sure, that there exist cyclic projective generators, which are
not isomorphic to S.
Furthermore, in Chapter 5 (Example 3.23) of [14] the same example gives
rise to the monoids S and eSe, which are Morita equivalent (i.e., the
categories Act S and Act eSe are equivalent) but neither isomorphic
nor anti-isomorphic.
In general, condition (v) of Theorem 5.21 is hard to handle with respect to the
decision, if (i) is true. If S is periodic, then by Proposition 1.3.26 of [14] for
e E(S) the relation eJ 1 implies e = 1. Thus in this case we obtain:

72

5 HOMOLOGICAL CLASSIFICATION

Corollary 5.23 Let S be a periodic semigroup. Then all right S-acts are
(S, G)-projective i all cyclic generators are isomorphic to SS .
Note, that Lemma 3.35 could also be proved by using the fact that every nite
monoid is periodic.

5.1.2

All are (X, A)-projective

This part deals with projectivities with second component A. The concepts of
(S, S/)- and (S, S/(x, y))-projectivity are also included, since all acts having
one of these properties is equivalent to all acts being (S, A)-projective.
Even if (S, A)-projectivity has been left out in the implication scheme because
of Lemma 3.32, it yields an interesting result:
Theorem 5.24 Let x, y S and let  be a right congruence on S. Then the
following assertions are equivalent:
(i) All right S-acts AS are (S, A)-projective,
(ii) all cyclic acts are (S, S/)-projective,
(iii) all right S-acts are (S, S/)-projective,
(iv) all cyclic acts are projective,
(v) S = {1} or S = {0, 1},
(vi) all monocyclic acts are (S, S/(x, y))-projective,
(vii) all right S-acts are (S, S/(x, y))-projective.
(i) = (iv). Let (S/ )S be a cyclic act. (S/ )S is (S, S/ )-projective
by (i). Thus by Lemma 5.18 the canonical epimorphism is a retraction, i.e.,
(S/ )S is a retract of a projective act and is therefore projective, i.e., (iv) is
valid.
Proof.

(iv) = (ii) is obvious.

(ii) (iii) is included in Lemma 5.18.
(ii) = (iv). If (S/ )S is (S, S/)-projective, then by Lemma 5.18 the canonical epimorphism is a retraction. Since S is projective, this implies the
projectivity of (S/ )S .

5.1 All acts are (X , Y)-projective

73

(iv) = (v). If all cyclic acts are projective, then all monocyclic acts are
projective and Result 5.17 yields (v).
(v) = (i). If S = {1}, then by Result 5.16 all acts are projective and therefore
(S, A)-projective.
Let S = {0, 1}. If g : S A is an epimorphism, then A is isomorphic to a
factor act of S. Thus A
= S or A
= . SS is projective and therefore (S, A)projective. Since S contains a left zero, S is (S, A)-projective by Proposition
5.8.
(v) (vi) is Result 5.17.
(vi) (vii) is a part of Lemma 5.18.

Remark 5.25 The equivalences (iii) (v) (vii) have already been proved
in [17] in terms of principally weak and weak projectivity.
Changing the rst component to be a Rees factor act by a principal right ideal
of S yields:

Proposition 5.26 Let i, s S and let AS Act S.

If all right S-acts are (S/sS, A)-projective, then all Rees factor acts (S/I)S by
right ideals I of S are isomorphic to Rees factor acts by principal right ideals
iS, i.e., S is a principal right ideal monoid.
Proof.
Let I be a right ideal of S.
If |I| = 1, then I is generated by a left zero, i.e., I is a principal right ideal.
If I = S, then I is generated by the identity.

Let |I| > 1, I = S.

All acts are (S/sS, A)-projective implies that all Rees factor acts S/I are
(S/sS, S/I)-projective. For i I, iS I holds. Thus the mapping :
S/iS S/I with (1S/iS ) = 1S/I is an epimorphism, which is a retraction
by Lemma 5.18, i.e., there exists a homomorphism : S/I S/iS with
= idS/I . Since 1S/I = {1} = 1S/iS , this implies (1S/I ) = 1S/iS . Thus for
u S/iS we get:
(
u) = (1S/iS u) = ((1S/iS )u) = (1S/I u) = (1S/I )u = 1S/iS u = u,
i.e., = idS/iS . Thus is an isomorphism of acts.
Since iS I, this implies iS = I, i.e., every right ideal of S is a principal right
ideal.

74

5 HOMOLOGICAL CLASSIFICATION

Corollary 5.27 All right S-acts are (S/sS, A)-projective implies the equivalence of (S, S/tS)- and (S, S/I)-projectivity.

Example 5.28 All right S-acts are (S/sS, A)-projective does not imply, that
all factor acts (S/)S by (principal) right congruences of S are isomorphic to
Rees factor acts by principal right ideals:
Let S be the two element right zero semigroup {a, b} with identity 1 adjoined.
Then all Rees factor acts (S/xS)S of S by right congruences are (S/1S)S = S
and (S/aS)S = (S/bS)S . If there exists an epimorphism g : (S/xS)S AS ,
then AS = S or AS = (S/aS)S and thus every right S-act AS is (S/sS, A)projective.
Now consider the right congruence (1, a). Since (S/(1, a))S has no zero, it
can not be isomorphic to a Rees factor act of S.
Especially, this means, that S need not to be a right (principal) Rees monoid
in this case.

5.1.3

All are (X,

mM

S/m )- or (X, S/)-projective

Now we start studying properties, which are included in the implication scheme
part 2, i.e., (X , Y)-projectivities, where Y is the class of cyclic S-acts or of
coproducts of cyclic S-acts, with regard to the question of all acts having this
property.
The rst lemma shows, that the question of (

kK

S/k ,

mM

S/m )-projectivi-

ty of all acts, can be reduced to the question of all acts being (S/, S/)projective. Nevertheless, the main statements with respect to (X , Y)-projectivity for various classes X will be giving in detail, i.e., in each case they will
include both kinds of projectivity. The respective titles of the paragraphs will
use the coproduct variant.
Lemma 5.29 Let x, y {, p, R, pR} and let , k , k K, be right xcongruences and , j , j J, , m , m M, be right y-congruences on S.
Then the following assertions are equivalent:
(i) All right S-acts are (

kK

S/k ,

mM

S/m )-projective,

5.1 All acts are (X , Y)-projective

(ii) all coproducts

(

kK

S/k ,

jJ

mM

75

(S/j )S of factor acts of S by right y-congruences are

S/m )-projective,

(iii) every epimorphism g :

kK

(S/k )S

(iv) all cyclic acts (S/)S are (

kK

S/k ,

jJ

mM

(S/j )S is a retraction,

S/m )-projective,

(v) all cyclic acts (S/)S are (S/, S/)-projective,

(vi) all right S-acts are (S/, S/)-projective,
(vii) every epimorphism g : (S/ )S (S/)S is a retraction.
Proof.
(i) (ii) (iii) and (v) (vi) (vii) are parts of Lemma 5.18.
(ii) = (iv) and (iv) = (v) are obvious.
(v) = (iv) is Proposition 3.12.
(iv) = (ii) is Proposition 3.17.

The previous Lemma will be used in various proofs of the next subsections,
where special sets of congruences and , respectively, are considered.

all are (

iI

S,

mM

S/m )-projective

In this part the second component is supposed to be a factor act of S by a

right (principal) congruence or a coproduct of those and the rst a free act. It
turns out, that in this case all acts have the respective property if and only if
S is trivial or S = {0, 1} (Theorem 5.30).
Rees congruences will be considered later on, when (S/sS, S/I)- and (S/sS, S/tS)projectivity will be studied.
Theorem 5.30 Let , j , j J, , m , m M, be right (principal) congruences on S.
Then the following assertions are equivalent:
(i) All right S-acts are (

iI

(ii) all coproducts

ences are (

iI

jJ

S,

S,

mM

S/m )-projective,

(S/j )S of factor acts of S by right (principal) congru

mM

S/m )-projective,

76

5 HOMOLOGICAL CLASSIFICATION

(iii) all coproducts

jJ

(S/j )S of factor acts of S by right (principal) congru-

ences are projective,

(iv) all cyclic acts (S/)S are projective,
(v) S = {1} or S = {0, 1},
(vi) all right S-acts are (S, S/)-projective.
Proof.

(i) (ii) is Lemma 5.18.

(ii) = (iii). Let pj : S S/j denote the j-th canonical epimorphism.

Then
pj :
S
S/j is an epimorphism, which is a retraction by
jJ

jJ

jJ

(ii) and Lemma 5.18. Thus

jJ

S/j is a retract of a projective act and is

therefore projective.
(i) (vi) is a part of Lemma 5.29.
(iii) = (ii) is obvious.
(iv) = (iii) is a part of Lemma 5.18.
(iv) (v) (vi) have been proved in Theorem 5.24.

Like mentioned before, the equivalence (v) (vi) has already been proved in
[17] in terms of (principally) weak projectivity.

all are (
all are (

nN


lL

S/(xn , yn ),

S/sl S,

mM

mM

S/m )-projective

S/m )-projective

In this paragraph, at rst we consider (X, S/)-projectivities with monocyclic

acts as rst component. We already know, that these are equivalent to the
respective coproduct version (Lemma 5.29).
It turns out that all acts being (S/(x, y), S/)-projective implies that S fullls condition (Moz), like dened in Denition 5.1. Moreover, we will see, that
all acts being (S/(x, y), S/)-projective is equivalent to S fullling condition
(Moz) together with all acts being (S/sS, S/)-projective, i.e., the question is
reduced to (S/sS, S/)-projectivity of all acts.
For monoids with left zero a complete answer will be given in Corollary 5.36
for factor acts of S by right (principal) congruences on S as second component.
For Rees factor acts as second component see the following subsections.

5.1 All acts are (X , Y)-projective

77

As indicated before, we have to study monoids over which all acts are (S/(x, y), S/)projective. As preparation of the proof of the main theorem in this context,
we need:

Lemma 5.31 Let u {, R, pR} and let  be a right u-congruence on S. If

all right S-acts are (S/(x, y), S/)-projective, then S fullls condition (Moz).
Consider the right congruence  = S S. Then (S/)S
= S and
for all x, y S the zero morphism g : S/(x, y) is an epimorphism,
which is a retraction by Lemma 5.18. Thus there exists a monomorphism
: S/(x, y), which implies, that () is a zero in (S/(x, y))S .
For right (principal) Rees congruences take  = I with I = 1S = S.

Proof.

The converse is not true:

Example 5.32 Let S be the monoid {1, a} with a1 = 1a = a and a2 = a and
zero 0 externally adjoined.
Since S contains a left zero, every (cyclic) right S-act has a zero, i.e., S fullls
condition (Moz).
Furthermore, since S contains a left zero, (S/(x, y), S/)- projectivity is equivalent to (S, S/)-projectivity (Lemma 4.4, 4.5, 4.6, 4.7). Since (S/aS)S is not
a retract of S, it is not (S, S/)-projective, i.e., not (S/(x, y), S/)-projective.
Note that monocyclic acts as second component are not included in Lemma
5.31. The reason turns out if we take a look at the proof: The main aspect of
this proof is, that S is isomorphic to a factor act of S by a right u-congruence.
Since in general this is not true for right principal congruences, one can not
use S in the same way for monocyclic acts.
Proposition 5.33 Let s, x, y S, let S be a monoid, which fullls condition
(Moz), let u {, p, R, pR} and let  be a right u-congruence on S. Then the
following assertions are equivalent:
(i) All right S-acts are (S/(x, y), S/)-projective,
(ii) all right S-acts are (S/sS, S/)-projective.
Proof.

(i) = (ii) is Lemma 3.44.

(ii) = (i). Let (S/(x, y))S be a monocyclic act. By assumption, there

exists a zero [z](x,y) in (S/(x, y))S . Since [z](x,y) is a zero, we obtain zS

78

5 HOMOLOGICAL CLASSIFICATION

[z](x,y) and thus zS (x, y). Therefore every factor act (S/(x, y))S is the
epimorphic image of a Rees factor act (S/zS)S and Lemma 3.3 completes the
proof.

Now by Lemma 5.29 and Proposition 5.33 we obtain:

Corollary 5.34 Let S be a monoid, which fullls condition (Moz), let u
{, p, R, pR}, let , j , j J, , m , m M, be right u-congruences on S
and let s, sl , x, y, xn , yn S, l L, n N. Then the following assertions
are equivalent:
(i) All right S-acts are (

nN

S/(xn , yn ),

mM

S/m )-projective,

(ii) all right S-acts are (S/(x, y), S/)-projective,

(iii) all right S-acts are (

lL

S/sl S,

mM

S/m ) -projective,

(iv) all right S-acts are (S/sS, S/)-projective.

Thus by Lemma 5.31 and Proposition 5.33, we have:

Theorem 5.35 Let s, x, y S, let u {, R, pR} and let  be a right
u-congruence on S. Then the following assertions are equivalent:
(i) All right S-acts are (S/(x, y), S/)-projective,
(ii) S is a monoid, which fullls condition (Moz), and all right S-acts are
(S/sS, S/)-projective.

At rst, we saw that the question of all acts being (

nN

S/(xn , yn ),

mM

S/m )-

projective can be reduced to the one of all acts being (S/(x, y), S/)-projective.
Now we obtained, that for characterizing monoids S over which all acts have
the respective property, we have to investigate monoids over which all acts are
(S/sS, S/)-projective.
For monoids with left zero, we can give a complete answer by Theorem 5.30
and Lemma 4.4 to 4.5, for  not being a Rees congruence, right now; for Rees
congruences I refer to the next subsection.

5.1 All acts are (X , Y)-projective

79

Corollary 5.36 Let S be a monoid with left zero, let  be a right congruence
on S, and let s, x, y, x, y S. Then the following assertions are equivalent:
(i) All right S-acts are (S/sS, S/))-projective,
(ii) all right S-acts are (S/sS, S/(x, y))-projective,
(iii) all right S-acts are (S/(x, y), S/)-projective,
(iv) all right S-acts are (S/(
x, y), S/(x, y))-projective,
(v) all right S-acts are (S, S/)-projective,
(vi) all right S-acts are (S, S/(x, y))-projective,
(vii) S = {1} or S = {0, 1},
(viii) all cyclic acts are projective.

From now on, we change the second component to be a Rees factor act of S
by a (principal) right ideal.

5.1.4

All are (S/(x, y), S/I)-projective

In this subsection the question, whether all acts are (S/(x, y), S/I)- projective, will be discussed. Starting with a useful Proposition, we get a necessary
condition on S in Proposition 5.40, which leads to Proposition 5.41, which
nally gives Theorem 5.57.
We will use the notation S is the disjoint union of the semigroup T and
the monoid R . This means, that T = is a subsemigroup of S, R = is
a submonoid of S and the set S is the disjoint union of the sets T and R, the
products tt , t, t T and rr , r, r  R are the respective products in R and
T , whereas the products t r and r t with t T and r R may be in R or in T .
(Although in some situations it would be more comfortable to permit T and
R to be empty, we follow the tradition to suppose semigroups to be not empty
(cf. [22].)
Since the set of all right invertible elements will be used in the following, we
introduce it explicitly by:

80

5 HOMOLOGICAL CLASSIFICATION

Denition 5.37 For every monoid S let R = {s S | s S : s s = 1}

denote the submonoid of S, consisting of all right invertible elements
of S.

Remark 5.38 The monoid R, introduced in Denition 5.37, is right cancellative (since R consists of all right invertible (and thus right cancellable) elements
of S).
The elements of the submonoid R need not to be right invertible in R, since
the right inverse s of an element s R is in general not right invertible, i.e.,
s  R is possible. Thus R need not to be a group.
Lemma 5.39 Let x, y S and let J be a right ideal of S. If all right S-acts
are (S/(x, y), S/J)-projective, then all elements u, v S satisfy one of the
following conditions:
(a) uS vS = S, i.e., u or v is right invertible,
(b) the principal congruence (u, v) generated by u and v is the Rees congruence I by the right ideal I = uS vS.
Proof.
Let u, v S and let I = uS vS = S. Then (u, v) I . Thus g :
S/(u, v) S/I with g(1S/(u,v) ) = 1S/I is an epimorphism. By Lemma 5.18
, g is a retraction, i.e., there exists id : S/I S/(u, v) with g id = idS/I .

Since 1S/(u,v) = {1} = 1S/I , we get id (1S/I ) = 1S/(u,v) . Thus, for every i I
we obtain the following chain of equations:
id ([i]I ) = id (1S/I i) = id (1S/I ) i = 1S/(u,v) i = [i](u,v) .
Since [i]I = [u]I for every i I, this implies [i](u,v) = [u](u,v) , i.e., i [u](u,v)
for all i I and thus I (u, v). This yields I = (u, v), i.e., (b).
If uS vS = S, then 1 uS vS, i.e., there exists u S with uu = 1 or
v  S with vv  = 1, i.e., u or v is right invertible.

This Proposition leads to the following necessary condition on S with all right
S-acts being (S/(x, y), S/I)-projective.

Lemma 5.40 Let x, y S and let I be a right ideal of S. If all right S-acts
are (S/(x, y), S/I)-projective, then S is the disjoint union of a right principal
Rees semigroup and the submonoid R of S, consisting of all right invertible
elements of S, and fullls condition (Moz).

5.1 All acts are (X , Y)-projective

81

The subsemigroup of S, consisting of all non right invertible elements

of S, is a right principal Rees semigroup by Lemma 5.39.
S fullls condition (Moz) by Proposition 5.31.

Proof.

Together with Theorem 5.36 this yields:

Proposition 5.41 For a monoid S the following assertions are equivalent:

(i) All right S-acts are (S/(x, y), S/I)-projective,
(ii) S is the disjoint union of a right principal Rees semigroup and the submonoid R of S, consisting of all right invertible elements of S, S fullls
condition (Moz), and all right S-acts are (S/sS, S/I)-projective.

Thus we have to consider (S/sS, S/I)-projectivity next. We will come back to

(S/(x, y), S/I)-projectivity in Theorem 5.57.

5.1.5

All are (S/sS, S/I)-projective

We already know, that (S/sS, X)-projectivity is equivalent to (S, X)-projectivity, if S has a left zero (see Lemma 4.4, 4.5, 4.6, 4.7). Thus in the following
subsections by considering (S/sS, X)-projectivity of all acts, we also get characterizations of monoids S with all acts being (S, X)-projective. The respective
results will be given separately.
By Lemma 5.18 we know, that all acts are (S/sS, S/I)-projective if and only
if all Rees factor acts (S/J)S are. Therefore, we start with some useful propositions concerning Rees factor acts of S, which prepare the main statements of
this subsection.
By doing this, we obtain for instance Lemma 5.44, which indicates, that it
is useful to divide the following investigations in those for monoids with and
those for monoids without left zero. We will obtain necessary conditions on S,
over which all acts are (S/sS, S/I)-projective, and we will see, that in this situations S has to be left reversible (Proposition 5.47, Proposition 5.48). Thus
the next step is to consider left reversible monoids in view to (S/sS, S/I)projectivity of all acts and it turns out, that in this case the necessary conditions on S developed before (Lemma 5.44), are already sucient for all acts

82

5 HOMOLOGICAL CLASSIFICATION

being (S/sS, S/I)-projective (Theorem 5.51).

Finally, this leads to the characterizations of monoids, over which all acts are
(S/sS, S/I)-projective in Theorem 5.52 for S without left zero and in Theorem
5.53 for monoids with left zero. They will be summarized in a general characterization in Corollary 5.56. Note, that the characterization with respect to all
acts being (S, S/I)-projective, which has been proved in [17] in terms of Rees
weak projectivity, is obtained as a corollary of Theorem 5.53.
Like promised before, at the end of this subsection we get an answer to the
question of all acts being (S/(x, y), S/I)-projective in Theorem 5.57.
Now we are starting by considering Rees factor acts with respect to (S/sS, S/I)projectivity.

Lemma 5.42 Let I, J be right ideals of S. If S is left reversible or a monoid

without left zero then the mapping : (S/I)S (S/J)S with (1S/I ) = 1S/J
is an S-epimorphism i I J.
Proof.
Suciency is clear, since I J implies I J .
Necessity. Let S be a monoid without left zero. Then every Rees factor act
contains a unique zero. Let 0 denote the zero in S/I, i.e., 0 = I, and let 0
denote the zero in S/J, i.e., 0 = J. By being an S-homomorphism we get
(0) = 0.
Thus for every i I we have:

0 = (0) = ([i]I ) = (1S/I i) = 1S/J i = [i]J ,

i.e., [i]J = 0 and thus i J.
If S is left reversible, then there exists i I J. Then (0) = ([i]I ) =
[i]J = 0, since i J. Thus for every i I we obtain: ([i ]I ) = ([i]I ) = 0
and therefore I J.

Note, that if J is a one-element right ideal of S, then (S/J)S

= SS is projective
and thus (S/sS, S/I)-projective.

Lemma 5.43 Let J be a right ideal of S with |J| =

 1. If the Rees factor act
(S/J)S is (S/sS, S/I)-projective, then J satises one of the following conditions:
(a) J = S,

5.1 All acts are (X , Y)-projective

83

(b) J has no proper subideals, i.e., jS = J for all j J.

Let J be a right ideal of S with |J| =
 1. Since for all j J the
relation jS J is valid, : S/jS S/I with (1S/jS ) = 1S/J is an epimorphism. Now consider the following diagram:
Proof.

S/J
id
S/jS

S/J

By assumption there exists id with id = id.

If J = S, then every congruence class of an element s  J is a one-element
class, especially 1S/jS = {1} = 1S/J . This implies id (1S/J ) = 1S/jS . Thus for
i J we get the following chain of equations:
id ([i]J ) = id (1S/J i) = id (1S/J ) i = 1S/jS i = [i]jS .
Since [i]J = [j]J for every i J, we get [i]jS = [j]jS .
Thus i jS j which is equivalent to i jS. This implies J jS and thus
J = jS, i.e., J has no proper subideals.

Lemma 5.44 Let s S and let I S denote a right ideal of S. Then

all right S-acts are (S/sS, S/I)-projective, i all Rees factor acts (S/J)S are
(S/sS, S/I)-projective.
Furthermore, this implies that every proper right ideal J of S is simple (i.e.,
J has no proper subideals, i.e., J = jS for every j J).
In this case, S is a principal right ideal monoid.
The equivalence is given by Lemma 5.18.
If all Rees factor acts (S/J)S are (S/sS, S/I)-projective, then by Lemma 5.43
every right ideal of S fullls (a), (b) or (c).
By denition, this implies that S is a principal right ideal monoid.

Proof.

Remark 5.45 If there exists a right ideal I = S with |I| =

 1, which is simple,
then S is a monoid without left zero z, since otherwise for all i I the implication izS = {iz} iS = I is valid, i.e., izS would be a proper subideal of I.
Thus, if S is a monoid with left zero, then for I = S the right ideal I is simple
if and only if |I| = 1.

84

5 HOMOLOGICAL CLASSIFICATION

Lemma 5.46 Let S be a monoid with left zero. If all Rees factor acts are
(S/sS, S/I)-projective, then the left zero in S is unique and is therefore a zero.
If z, z  are left zeros in S, then J = {z, z  } is a right ideal of S,
which does not fulll one of the conditions of Lemma 5.44, contradicting S/J
is (S/sS, S/I)-projective. Thus S contains a unique left zero z.
Let s S. Then for all t S the equation (sz)t = s(zt) = sz implies, that sz
is a left zero in S. By the uniqueness of the left zero, this yields sz = z, i.e., z
is a right zero of S and is therefore a zero.

Proof.

Conditions (a) to (c) of Lemma 5.44 in general do not give the (S/sS, S/I)projectivity for all acts. In the following, we dierentiate between monoids
with left zero and those without a left zero (as indicated by Remark 5.45).
First we will see, that in both cases the (S/sS, S/I)-projectivity of all acts
leads to a special class of monoids, namely to left reversible monoids.

Proposition 5.47 Let S be a monoid without left zero, let s S and let I
be a right ideal of S. If all right S-acts are (S/sS, S/I)-projective, then S
is the disjoint union of a right simple semigroup and the submonoid R of S,
consisting of all right invertible elements of S or S is a group.
Furthermore, this implies that S is left reversible.
Let S be a monoid without left zero. If all acts are (S/sS, S/I)projective, then by Lemma 5.44 every proper right ideal of S is simple. Thus
every element of S is right invertible or generates a simple right ideal of S.

Proof.

Let I = {s | sS is a proper simple right ideal of S}.

If I = , then S = R, which implies, that S is a group and is therefore left
reversible.
Let I = . Then obviously R I = .
I is a right ideal of S, since s I, u S implies suS = sS and is therefore
simple, i.e., I is a right simple subsemigroup of S. Thus S is the disjoint union
of R and a right simple semigroup.
Furthermore, for s, t R we get: sS tS = S = . For s R, t
I, sS tS = tS = and for s, t I we nally get sS = tS = I = . Thus S
is left reversible.

Proposition 5.48 Let S be a monoid with left zero z, let s S and let I be
a right ideal of S. If all right S-acts are (S/sS, S/I)-projective, then S is the
disjoint union of a group and a zero or S = {1}.
Furthermore, this implies, that S is left reversible.

5.1 All acts are (X , Y)-projective

85

By Lemma 5.46 the left zero z in S is a zero.

Without loss of generality, let |S| 2. Since all acts are (S/sS, S/I)-projective,
we get by Lemma 5.44 for every s S \ {z}, that sS = S, i.e., s is right
invertible, i.e., R = S \ {z}. Let r R and let r  S with rr  = 1. Then
r  R and thus there exists r  R such that r  r  = 1. This implies r =
r(r r  ) = (rr )r  = r  ,, i.e., r  r = 1. Thus every r R is invertible and
therefore R is a group.
Proof.

Since z is a zero in S, we get z I for every right ideal I of S. Thus every

two right ideals of S have a non empty intersection, i.e., S is left reversible.

Note, that the left reversibility of S is also sucient for the uniqueness of a
left zero in S.
Propositions 5.47 and 5.48 imply, that to get more informations about internal properties of the sublying monoid S, we have to study (S/sS, S/I)projectivities of acts over a left reversible monoid S.
The respective results in Lemma 5.49 and Lemma 5.50 will be needed to prove
Theorem 5.51.

Lemma 5.49 Let S be a left reversible monoid. If every proper right ideal of
S is simple, then S has at least one proper right ideal I = S \ R.
Let I, J S be proper right ideals of S. Since S is left reversible,
there exists i I J. This implies iS I and iS J. Since I and J are
simple, this yields I = iS = J, i.e., S has at least one proper right ideal I = iS.
Since I = iS = S, for all s S we get is = 1, i.e., I = iS S \ R.
Let s S \ R. By the uniqueness of the proper right ideal, we get sS = iS.
Thus there exists s S, such that s = i s , i.e., S \ R iS = I and thus
S \ R = I.

Proof.

Notice, that if S is a left reversible monoid with left zero z, then by Lemma
5.49 the right ideal S \ R = {z} is the unique proper subideal of S.
Lemma 5.50 Let S be a monoid and let I = S \ R be the proper right ideal of
S, consisting of all non right invertible elements of S. Then every epimorphism
g : (S/I)S (S/I)S is an isomorphism.
Proof.
For s S \ I, s = 1, the class of s in S/I is a one element class and
will be denoted by s itself, the class of the identity will be denoted by 1S/I ,
the class of elements of I will be denoted by 0.

Since 1  I, we get 1S/I = {1} and |S/I| 2.

Suppose, there exists v S \ I = R such that g(v) = 0. Then g(1S/I ) =

86

5 HOMOLOGICAL CLASSIFICATION

g(v v 1 ) = g(v) v 1 = 0 v 1 = 0. This implies g(S/I) = 0, i.e., g is not

surjective contradicting g being an epimorphism.
Thus the image of an element of S \ I is an element of S \ I.
Moreover, this implies g(0) = 0, since g is surjective.
Since g is an epimorphism, there exists s S \ I, such that g(s) = 1S/I . Let
v, w S/I. Then
g(v) = g(w) = 0 v = 0 = w,
g(v) = g(w) = 0 v, w R.
Then g(v) = g(s s1 v) = g(s) s1 v = 1S/I s1 v = s1 v and analogously
g(w) = s1 w. Thus g(v) = g(w) s1 v = s1 w s1 = s1 w v 1 ,
which implies 1S/I = s s1 = s s1 w v 1 = 1S/I w v 1 which is equivalent
to v = w.
Therefore g is injective, i.e., g is an S-isomorphism.

For left reversible monoids S we are now able to give a characterization with
respect to all acts being (S/sS, S/I)-projective.

Theorem 5.51 Let S be a left reversible monoid, let s S and let I be a

right ideal of S. Then the following assertions are equivalent:
(i) All right S-acts are (S/sS, S/I)-projective,
(ii) all Rees factor acts (S/J)S are (S/sS, S/I)-projective,
(iii) every proper right ideal J of S is simple (i.e., J has no proper subideals,
i.e., J = jS for every j J).
(i) (ii) is a part of Lemma 5.18.
(ii) = (iii) follows from Lemma 5.44.
(iii) = (ii) If S contains a left zero z, then by (iii) and Remark 5.45, for
all right ideals I we get |I| = 1 or I = S. Thus all Rees factor acts of S are
isomorphic to SS or to S . The epimorphisms, which have to be considered
in this case, are: the zeromorphism g1 : S , the identity id and epimorphisms g2 : S S.
The identity is obviously a retraction, g1 is a retraction, since S contains a left
zero and g2 is a retraction, since S is projective.
Thus every epimorphism from a Rees factor act by a principal right ideal onto
a Rees factor act by a right ideal is a retraction and Lemma 5.18 gives (ii).
Proof.

If S is a monoid without left zero, then for every right ideal I of S we have
I = S or I is simple. By Lemma 5.49 we obtain I = S or I = S \ R. Thus

5.1 All acts are (X , Y)-projective

87

S/I
= (R {0}) or S/I = S/S = .
The zero morphism g1 : (R 0}) is a retraction, since (R 0}) contains
a zero. The identity id is obviously a retraction. By Lemma 5.50, every
epimorphism g2 : (R
0}) (R
0}) is an isomorphism, which is a retraction.
Thus we get, that even in this case the respective epimorphisms are retractions
and Lemma 5.18 completes the proof.

Now we can characterize monoids, over which all acts are (S/sS, S/I)-projective,
where we start with monoids without left zero. By Lemma 5.44, Remark 5.45,
Proposition 5.47, Lemma 5.49 and Theorem 5.51 we get:

Theorem 5.52 Let S be a monoid without left zero, let s S and let I be a
right ideal of S. Then the following assertions are equivalent:
(i) All right S-acts are (S/sS, S/I)-projective,
(ii) all Rees factor acts (S/J)S are (S/sS, S/I)-projective,
(iii) S is left reversible and every proper right ideal J of S is simple (i.e., J
has no proper subideals, i.e., J = jS for every j J),
(iv) if I is a proper right ideal of S, then I = S \ R,
(v) S is the disjoint union of the submonoid R of S, consisting of all right
invertible elements of S, and a right simple semigroup, or S is a group.
(i) (ii) is given by Lemma 5.18.
(ii) = (iii). By Proposition 5.47, the monoid S is left reversible, and thus
Theorem 5.51 yields (iii).
(iii) = (ii) is Theorem 5.51.
(iii) = (iv) is Lemma 5.49.
(iv) = (iii). Without loss of generality, suppose that S \ R = I is a proper
right ideal of S and is unique by (iv). Then obviously S is left reversible.
Furthermore, the uniqueness of I yields that I is simple and thus (iii) holds.
(ii) = (v) is Proposition 5.47.
(v) = (iii). If S is a group, then S does not contain a proper right ideal.
Thus (iii) is obviously true.
Let S be the disjoint union of R and a simple semigroup H and let I be a right
ideal of S. If I R = , then I = S, i.e., (iii)(a).
Let I R = , i.e., I H. Since H is simple, this implies I = H, i.e., I has
no subideals, i.e., (iii)(b).
The left reversibility is obvious now.

Proof.

88

5 HOMOLOGICAL CLASSIFICATION

Note, that if S is the disjoint union of a monoid T and a zero, we also talk
about S as a monoid with zero adjoined. For monoids with left zero, we obtain
the following theorem, where the analogous assertions to (ii), (iii) and (iv) of
Theorem 5.52 are left out, because they will not be used anymore.

Theorem 5.53 Let S be a monoid with left zero z, let s S and let I be a
right ideal of S. Then the following assertions are equivalent:
(i) All right S-acts are (S/sS, S/I)-projective,
(ii) S is a group with zero adjoined or S = {1}.
Proof.

(i) = (ii) is Proposition 5.48.

(ii) = (i) If S is a group with zero adjoined, then all Rees factor acts of S
are isomorphic to S or to . S is projective and thus (S, S/I)-projective and
so is S
= 0S.
If S = {1}, then by Result 2.35 all acts are projective and thus (S/sS, S/I)projective.

Recall, that if all acts are (S, S/I)-projective, then the respective projectivity
of S yields a left zero in S by Proposition 5.8. Moreover, if S contains a left
zero, then by Theorem 4.6 (S, S/I)-projectivity is equivalent to (S/sS, S/I)projectivity. Together with Theorem 5.53 we now obtain the respective results
in view to Rees weak projectivity as:

Corollary 5.54 Let s S and let I denote a right ideal of S. Then the
following assertions are equivalent:
(i) All right S-acts are (S, S/I)-projective (Rees weakly projective),
(ii) S is a group with zero adjoined or S = {1}.

The previous result has been proved directly in [17] in terms of Rees weak
projectivity. Remark, that this characterization is given by a corollary in this
context.

Remark 5.55 Each of the conditions of Theorem 5.53 implies that all cyclic
acts fulll condition (P ) (cf [14]).

5.1 All acts are (X , Y)-projective

89

Summarizing, for arbitrary monoids S we get:

Corollary 5.56 Let s S and let I be a right ideal of S. Then the following
assertions are equivalent:
(i) All right S-acts are (S/sS, S/I)-projective,
(ii) S satises one of the following conditions:
(a) S is the disjoint union of the submonoid R of S, consisting of all
right invertible elements of S, and a right simple semigroup.
(b) S is a group.
(c) S is a group with zero adjoined.

For arbitrary monoids with or without left zero we obtained characterizations, whether all acts are (S/sS, S/I)-projective. The next section deals
with (S/sS, S/tS)-projectivity, where some properties can be directly developed from the conditions in this section.
Before going any further, we come back to monoids, over which all acts are
(S/(x, y), S/)-projective. By Proposition 5.41 and Corollary 5.56 we get :
Theorem 5.57 For a monoid S the following assertions are equivalent:
(i) All right S-acts are (S/(x, y), S/I)-projective,
(ii) S satises one of the following conditions:
(a) S fullls condition (Moz) and S is the disjoint union of the submonoid R of S, consisting of all right invertible elements of S, and
a right simple right principal Rees semigroup.
(b) S fullls condition (Moz) and S is a group.
(c) S is a group with zero adjoined.

Example 5.58 In Example 5.3 it has been shown, that the group (Z3 , +)
fullls condition (Moz). Thus every right Z3 -act is (S/(x, y), S/I)-projective.
Like said before, in the following subsection the investigations are reduced to
the case of second components being Rees factor acts by principal right ideals
of S.

90

5.1.6

5 HOMOLOGICAL CLASSIFICATION

All are (S/sS, S/tS)-projective

In this part we study monoids, over which all acts are (S/sS, S/tS)-projective,
i.e., we arrive at the bottom of the implication scheme part 2. Some of the results can be proved in analogy to the respective results concerning (S/sS, S/I)projectivity.
It will turn out, that we can not use left reversibility of S in the same way like
before, since left reversibility is not necessary for all acts being (S/sS, S/tS)projective (see Remark 5.61).
Since (S/sS, S/tS)-projectivity of all acts is equivalent to (S/sS, S/tS)-projectivity of all Rees factor acts by principal right ideals (Lemma 5.18), we start
with studying (S/rS)S with respect to this property.
Note that if r is a left zero in S, then (S/rS)S
= SS is projective and thus
(S/sS, S/tS)-projective.

Lemma 5.59 Let s, t S and let r S be not a left zero.

If the Rees factor act (S/rS)S is (S/sS, S/tS)-projective, then r satises one
of the following conditions:
(a) r is right invertible,
(b) rS has no proper subideals.
Proof.
Since r is not a left zero, |rS| =
 1. Then we obtain in analogy to
the proof of Proposition 5.43:
rS = S, i.e., there exists r  S such that rr  = 1, i.e., r is right invertible, or
rS contains no proper subideal.

Lemma 5.60 Let r, s, t S. Then all right S-acts are (S/sS, S/tS)projective, i all Rees factor acts (S/rS)S are (S/sS, S/tS)-projective.
Furthermore, this implies that every r S satises one of the following conditions:
(a) r is right invertible,
(b) rS has no proper subideals, i.e., ru R r for every u S.

5.1 All acts are (X , Y)-projective

91

The equivalence is a part of Lemma 5.18.

The implication is a consequence of Lemma 5.59.
Proof.

Remark 5.61 Remark 5.45 is valid even in this situation, i.e., if there exists
a principal right ideal rS of S with |rS| =
 1, which fullls condition (b) of
Lemma 5.60, then S is a monoid without left zero.
Note, that unlike in the case of all right S-acts being (S/sS, S/I)-projective,
(S/sS, S/tS)-projectivity of all right S-acts does not yield the uniqueness of a
left zero in S.
The reason is, that for left zeros z, z  in S the right ideal {z, z  } is in general
not principal. Thus in this situation we do not have an analogon to Lemma
5.46.
Proposition 5.62 Let s, t S. If all right S-acts are (S/sS, S/tS)-projective,
then S is the disjoint union of a semigroup I, the submonoid R of S, consisting
of all right invertible elements of S, and a left zero semigroup L, where L and
I are possibly empty.
Moreover, in this case I is the disjoint union of right simple subsemigroups i S
of S, where I  is a representing system for the classes of R on S of elements
of I.
Proof.
Let L denote the set of left zeros of S and let R be the submonoid
of right invertible elements of S.
Let I = {s S | sS is a proper principal simple right ideal of S} \ L.
Then R L = , R I = and by denition L I = . Since all acts are
(S/sS, S/tS)-projective, by Lemma 5.60, an element of S is a left zero or is
right invertible or generates a simple principal ideal of S, i.e., S = I L R.
I, L and R are closed under multiplication, i.e., are subsemigroups of S.

If i, j I, then either iS jS = or iS = jS, since both ideals are simple. Let

 
I  contain a representant i [i]R I/R for every R-class. Then I =
iS
is the disjoint union of right simple subsemigroups of S.

i I 

By Remark 5.45, we get again, that we have to distinguish between monoids

with left zero(s) and those without left zero, since in either case one of the
conditions of Lemma 5.60 is unnecessary.
By Proposition 5.62 we obtain:
Proposition 5.63 Let s, t S and let S be a monoid, such that all right
S-acts are (S/sS, S/tS)-projective.

92

5 HOMOLOGICAL CLASSIFICATION

If S contains a left zero, then S is the disjoint union a left zero semigroup and
the group R or S = {1}.
If S is a semigroup without left zero, then S is the disjoint union of right simple
semigroups of the form i S, i S, and the submonoid R of S, consisting of
all right invertible elements of S, or S is a group.
If S contains a left zero, then every simple principal right ideal of
S is a one-element right ideal, i.e., it is generated by a left zero. Thus by
Proposition 5.62, S is the disjoint union of R and the left zero semigroup L.
Without loss of generality suppose |R| 2. Let r R, r = 1. Since r
is right invertible, there exists r  S with rr  = 1. If r  is a left zero in
S, then the equality r = (rr  )r = r(r  r) = rr  = 1 yields a contradiction.
Thus r  R, i.e., there exists r  R such that r  r  = 1. This implies:
r = r(r  r  ) = (rr  )r  = r  , i.e., r  r = 1. Thus R is a monoid, such that every
element has a unique inverse, i.e., R is a group.
If |R| = 1 and if 1z = 1 for a left zero z S, then z = 1z = 1 yields S = {1}.
Proof.

The second statement follows directly from Proposition 5.62 with L = or

L = = I.

Remark 5.64 By the previous Proposition it becomes clear, that all right Sacts being (S/sS, S/tS)-projective does not imply S to be left reversible. For
example, if S is the two element left zero semigroup {a, b} with identity adjoined, then S is the disjoint union of a group and a left zero semigroup, but
since aS bS = , it is not left reversible. (Indeed, we will see later on, that
for S all right S-acts are (S/sS, S/tS)-projective.)
Nevertheless, we want to study left reversible monoids in this context in the
following.
In analogy to the proof of Proposition 5.49, we obtain:
Corollary 5.65 Let S be a left reversible monoid without left zero and let
i S. If all principal right ideals of S are simple, then S has at least one
proper principal right ideal iS = S \ R.

Theorem 5.66 Let s, t S and let S be a left reversible monoid. Then the
following assertions are equivalent:
(i) All right S-acts are (S/sS, S/tS)-projective,
(ii) all Rees factor acts by principal right ideals are (S/sS, S/tS)-projective,

5.1 All acts are (X , Y)-projective

93

(iii) every r S satises one of the following conditions:

(a) r is right invertible,
(b) rS has no proper subideals,i.e., ru R r for every u S,
(iv) S is the disjoint union of a right simple semigroup and the submonoid R
of S, consisting of all right invertible elements of S, or S is a group or
S is the disjoint union of a group and a zero,
(v) all right S-acts are (S/sS, S/I)-projective.
(i) (ii) is a part of Lemma 5.18.
(ii) = (iii) follows from Lemma 5.60.
(iii) = (ii) in analogy to the proof of Theorem 5.51.
(ii) = (iv). Let S be a monoid with left zero. By Proposition 5.63 we get
that S is the disjoint union of a group and a left zero semigroup. Since S is
left reversible, for left zeros z, z  in S we obtain: zS z  S = , which implies
z = z  . Thus the left zero in S is unique and is therefore a zero, i.e., S is the
disjoint union of a group and a zero.
Proof.

Let S be a monoid without left zero. Again by Proposition 5.63 we obtain, that
S is a group or S is the disjoint union of right simple semigroups i S, i I  , and
R. For i , i I  the left reversibility of S yields the existence of j i S i S.
Then jS i S and jS i S. Since i S and i S are right simple, this implies
i S = jS = i S. Therefore the disjoint union of the right simple semigroups
i S, i I  , is a right simple semigroup iS.
Thus S is a group or the disjoint union of R and a right simple semigroup.
(iv) (v) is Corollary 5.56.
(v) = (ii) is obvious.

Even if we can not get a more general characterization by considering left reversible monoids, in the case of S containing a left zero, we obtain:

Theorem 5.67 Let S be a monoid with left zero and let s, t S. Then the
following assertions are equivalent:
(i) All right S-acts are (S/sS, S/tS)-projective,
(ii) all elements r of S, which are not left zeros, are right invertible, i.e., S
is the disjoint union of a group and a left zero semigroup or S = {1}.

94

Proof.

5 HOMOLOGICAL CLASSIFICATION

(i) = (ii) follows from Proposition 5.63.

(ii) = (i) If S = {1}, then by Result 2.35 all acts are projective and therefore
(S/sS, S/tS)-projective.
If all non left zero elements are right invertible, then all Rees factor acts of S
by principal right ideals are isomorphic to SS or S . Since S has a left zero,
S is projective and so is S. Thus all Rees factor acts of S by principal right
ideals are (S, S/tS)-projective and Lemma 5.18 completes the proof.

By Theorem 5.67 and by the fact, that S is (S, S/tS)-projective if and only if
S contains a left zero, we obtain the following characterization of monoids S,
over which all acts are (S, S/tS)-projective. Note, that in this Corollary also
the question of all acts being (S/(x, y), S/tS)-projective is anwered.

Corollary 5.68 Let s, t, x, y S.

(i) All right S-acts are (S, S/tS)-projective,
(ii) S contains a left zero and all right S-acts are (S/(x, y), S/tS)-projective,
(iii) S contains a left zero and all right S-acts are (S/sS, S/tS)-projective,
(iv) S contains a left zero and all elements r of S, which are not left zeros,
are right invertible, i.e., S is the disjoint union of a group and a left zero
semigroup or S = {1}.
(i) = (ii). By (i), the one element act S is (S, S/I)-projective,
which gives a left zero in S by Proposition 5.8. The second part is obvious.
(ii) = (iii) is obvious.
(iii) (iv) is Theorem 5.67.
(iii) = (i) is a part of Theorem 4.6.

Proof.

Remark 5.69 The equivalence (i) (iv) has already been proved in [17] in
terms of principally Rees weak projectivity.
So far, during this chapter we studied monoids S, over which all acts have a
special property, whereas we went from the strongest to the weakest projectivity. The results will be summarized in tables at the end of this chapter
beneath those results, which will be developed in the next section.

5.2 Characterization of monoids by implications between the concepts

5.2

95

Characterization of monoids by implications between the concepts

Some implications between dierent properties of acts yield further properties

of the respective monoid. In this section we are searching for conditions, under
which a concept of projectivity implies a stronger one.
We will see, that most of those implications leads to S being a monoid with
left zero or to S fullling condition (Moz).
First we study implications (X , Y)-projectivity = (X  , Y  )-projectivity with
X being the class of all Rees factor acts by principal right ideals of S and
X  = {S}, and coproducts of these, respectively.

Theorem 5.70 Let s, t, sl S, l L, let I be a right ideal of S and let

AS Act S. Then the following assertions are equivalent:
(i) (S/sS, A)-projectivity implies (S, A)-projectivity,
(ii) (

lL

S/sl S, A)-projectivity implies (

S, A)-projectivity (which is equiva-

iI

lent to projectivity),
(iii) (

lL

(iv) (

lL

(v) (

lL

S/sl S, A)-projectivity implies (S, A)-projectivity,

S/sl S, A)-projectivity implies (S, S/I)-projectivity,

S/sl S, A)-projectivity implies (S, S/tS)-projectivity,

(vi) (S/sS, S/tS)-projectivity implies (S, S/tS)-projectivity,

(vii) (S/sS, S/I)-projectivity

implies (S, S/I)-projectivity,

(viii) (S/sS, S/I)-projectivity

implies (S, S/tS)-projectivity,
(ix) S contains a left zero.
For j {(i), (ii), (iii), (vi), (vii), (viii)} the implications (ix) = j
are obvious, recalling that for a left zero z in S the Rees factor act (S/zS)S is
isomorphic to SS .

(ix) = (iv), (v) is valid, since (ix) = (ii) and projectivity = (S, S/I)projectivity = (S, S/tS)-projectivity.
Proof.

(i) = (ix): S is (S/sS, )-projective (Proposition 5.11) and thus by (i) S

is (S, )-projective. By Proposition 5.8 the monoid S contains a left zero.

96

5 HOMOLOGICAL CLASSIFICATION

The same argument yields the implications j = (ix) for every j {(ii), (iii),
(iv), (v), (vi), (vii), (viii)}.

In general, S can not be used in the same way, if the second component is
an arbitrary cyclic act. In this case we get:
Theorem 5.71 Let S be a left reversible monoid, let s, sl S, l L, and let
, m , m M, denote right congruences on S. Then the following assertions
are equivalent:
(i) (S/sS, S/)-projectivity implies (S, S/)-projectivity,
(ii) (

lL

(iii) (

lL

S/sl S,
S/sl S,

mM

mM

S/m )-projectivity implies (

S,

iI

mM

S/m )-projectivity,

S/m )-projectivity implies (S, S/)-projectivity,

(iv) S contains a zero.

Proof.
(i) = (iv) Since S is left reversible, is (S/sS, S/)-projective by
Proposition 5.14 and thus by (i) (S, S/)-projective. By Proposition 5.8 there
exists a left zero z in S. Suppose z  is a left zero in S. Then zS = {z} =
{z  } = z  S by the left reversibility of S, i.e., z is unique and is therefore a zero
in S.
(ii) = (iv) and (iii) = (iv) can be proved in the same way.
(iv) = (ii) (which implies (iii)) and (iv) = (i) have been proved in Theorem
4.4.

To use the one element act S in the same way for monocyclic acts as second
components, we have to make sure, that there are homomorphism, which have
to be lifted, i.e., that S is not trivially-(S/sS, S/(x, y))-projective. For this
situation we obtain the following assertion, which can be proved in the same
way like Theorem 5.71 (applying Proposition 5.10 and Theorem 4.5).

Corollary 5.72 Let S be a left reversible monoid, which fullls condition

(Moz), let s, sl , x, y, xn , yn S, (l L, n N). Then the following
assertions are equivalent:
(i) (S/sS, S/(x, y))-projectivity implies (S, S/(x, y))-projectivity,
(ii) (

lL

S/sl S,

nN

projectivity,

S/(xn , yn ))-projectivity implies (

iI

S,

nN

S/(xn , yn ))-

5.2 Characterization of monoids by implications between the concepts

(iii) (

lL

S/sl S,

nN

97

S/(xn , yn ))-projectivity implies (S, S/(x, y))-projectivity,

(iv) S contains a zero.

Now we change the rst component in (X  , Y )-projectivity to the class X 

consisting of all monocyclic acts (or coproducts of these, respectively).

Theorem 5.73 Let s, sl , t, x, y, xn , yn S (l L, n N) let I denote

a right ideal of S and let AS Act S. Then the following assertions are
equivalent:
(i) (S/sS, S/tS)-projectivity implies (S/(x, y), S/tS)-projectivity,
(ii) (S/sS, S/I)-projectivity implies (S/(x, y), S/I)-projectivity,
(iii) (S/sS, S/I)-projectivity implies (S/(x, y), S/tS)-projectivity
(iv) (S/sS, A)-projectivity implies (S/(x, y), A)-projectivity,
(v) (

lL

(vi) (

lL

S/sl S, A)-projectivity implies (

nN

S/(xn , yn ), A)-projectivity,

S/sl S, A)-projectivity implies (S/(x, y), A)-projectivity,

(vii) S fullls condition (Moz).

(i) = (vii). S is (S/sS, S/tS)-projective (Proposition 5.11) and
thus by (i) S is (S/(x, y), S/tS)-projective. By Proposition 5.10 the monoid
S fullls condition (Moz).
(vii) = (i). Let u denote the zero in S/(x, y). Then for the right principal
Rees congruence uS , generated by the principal right ideal uS, we get uS
(x, y). Thus there exists an epimorphism : S/uS S/(x, y) and Lemma
3.3 yields (i).
Proof.

For j {(ii), (iii), (iv), (v), (vi)} the equivalences j (vii) can be proved in
analogy to the proof of (i) (vii).

Like Theorem 5.71, for arbitrary cyclic acts or coproducts of those as second component, we obtain the following characterization in the case of left
reversible monoids:

98

5 HOMOLOGICAL CLASSIFICATION

Theorem 5.74 Let S be a left reversible monoid, let s, x, y, sl , xn , yn

S, l L, n N, and let , m , m M, denote right congruences on S. Then
the following assertions are equivalent:
(i) (S/sS, S/)-projectivity implies (S/(x, y), S/)-projectivity,
(ii) (

lL

S/sl S,

mM

S/m )-projectivity implies (

projectivity,
(iii) (

lL

S/sl S,

mM

nN

S/(xn , yn ),

mM

S/m )-

S/m )-projectivity implies (S/(x, y), S/)-projectivity,

(iv) S fullls condition (Moz).

Proof.
(i) = (iv). Since S is left reversible, S is (S/sS, S/)-projective
and thus by (i) it is (S/(x, y), S/)-projective. Thus for x, y S the identity
on S has to be lifted with respect to the zero morphism g : S/(x, y)

= S/S 2 . Then id () is a zero in S/(x, y), i.e., S fullls condition (Moz).

(iv) = (i) can be proved analogously to (vii) = (i) in the proof of Theorem
5.73.
The remaining equivalence can be proved in the same way.

In the next theorem, we demand X to be the class of all monocyclic acts ( coproduct of monocyclic acts) and X  to consist of SS or of coproducts of copies
of S. In this case, we suppose, that S fullls condition (Moz).

Theorem 5.75 Let S be a monoid, which fullls condition (Moz), let s, x, y,

xn , yn S, n N, let I be a right ideal of S and let AS Act S. Then the
following assertions are equivalent:
(i) (S/(x, y), S/sS)-projectivity implies (S, S/sS)-projectivity,

(ii) (S/(x, y), S/I)-projectivity

implies (S, S/I)-projectivity,

(iii) (S/(x, y), S/I)-projectivity

implies (S, S/sS)-projectivity,
(iv) (S/(x, y), A)-projectivity implies (S, A)-projectivity,
(v) (

nN

(vi) (

nN

(vii) (

nN

S/(xn , yn ), A)-projectivity implies (

S, A)-projectivity,

iI

S/(xn , yn ), A)-projectivity implies (S, A)-projectivity,

S/(xn , yn ), A)-projectivity implies (S, S/I)-projectivity,

5.2 Characterization of monoids by implications between the concepts

(viii) (

nN

99

S/(xn , yn ), A)-projectivity implies (S, S/tS)-projectivity,

(ix) S contains a left zero.

(i) = (ix) Since S fullls condition (Moz), by Proposition 5.10 the
one element act S is (S/(x, y), S/sS)-projective and thus by (i) (S, S/sS)projective. Then Proposition 5.8 yields the existence of a left zero in S.
Proof.

(ix) = (i) (S/(x, y), X)-projectivity implies (S/sS, X)-projectivity. By (ix)

and Theorem 5.70 we get (i).
The remaining equivalences can be proved analogously.

We saw, that some implications between the concepts introduced in Chapter 3

lead to characterization of the monoid S. The respective results will be summarized in the tables given in section 5.3.

100

5.3

5 HOMOLOGICAL CLASSIFICATION

Summarizing Tables

The following tables summarize the results of the previous sections. The same
abbreviations as in the implication scheme part1 and part 2 are used
here. In addition, let R denote the submonoid of S, consisting of all right
invertible elements of S. The abbreviation l.z. denotes the situation, that
S contains a left zero, zero and (Moz) analogously. The respective properties appear in the same order as they came up in Chapter 5.

The rst tables include the results concerning characterizations of monoids by

properties of all acts, whereas in Table 5 the general results are summarized,
in Table 6 the results for left reversible monoids are collected and in Table 7
those for monoids with left zero, not including those characterizations, which
in general lead to S containing a left zero.
In Table 8 necessary conditions on a monoid S, over which all acts have a
special property, are collected.
The Tables 9 to 12 due to Section 5.2, where implications between dierent projectivities yielded further characterizations. Table 9 includes the general case,
Table 10 the situation for left reversible monoids, Table 11 the one for monoids
fullling condition (Moz) and nally in Table 12 the results concerning left
reversible monoids, which also fulll condition (Moz), are summarized.
Note, that in tables, which belong to Section 5.2, the empty places could be
lled by the necessary (but not sucient) property A, where in every case A
is the same property, which can be found in the same column like the array
under consideration.
Recall, that R is the submonoid of S, consisting of all right invertible elements
of S (Denition 5.37).

5.3 Summarizing Tables

All right S-acts are

S, G)-projective

101

if and only if

S is a group

iI

(S, G)-projective
(S, A)-projective
(

S,

iI

iI

mM

S,

nN

S/m )-projective

S/(xn , yn ))- projective

all cyclic generators are isomorphic to

eS with e2 = e and eJ 1
S = {1} or S = {0, 1}
S = {1} or S = {0, 1}
S = {1} or S = {0, 1}

(S, S/)-projective

S = {1} or S = {0, 1}

(S, S/(x, y))-projective

S = {1} or S = {0, 1}

(S/(x, y), S/)-projective

S fulllls condition (Moz) and

all right S-acts are (S/sS, S/)-projective
S is the disjoint union of R and a right simple
semigroup
or S is a group
or S is a group with zero adjoined

(S/sS, S/I)-projective

(S, S/I)-projective

S is a group with zero adjoined or S = {1}

(S/(x, y), S/I)-projective

S fulllls condition (Moz) and S is the disjoint

union of R and a right simple right principal
Rees semigroup
or S is a group, which fullls condition (Moz)
or S is a group with zero adjoined
S fulllls condition (Moz) and
all right S-acts are (S/sS, S/tS)-projective
S is the disjoint union of a group and a left
zero semigroup or S = {1}

(S/(x, y), S/tS)-projective

(S, S/tS)-projective

Table 5: Characterization of monoids by properties of all right S-acts

102

5 HOMOLOGICAL CLASSIFICATION

All right S-acts are

if and only if

(S/sS, S/tS)-projective

if S is the disjoint union of a right simple semigroup

and R
or S is a group
or S is a group with zero adjoined

(S/(x, y), S/tS)-projective

S fullls condition (Moz) and

(S is a group or
S is the disjoint union of a right simple semigroup
and R) or
S is group with zero adjoined

Table 6: Characterization of left reversible monoids by properties of all right

S-acts

5.3 Summarizing Tables

All right S-acts are

lL

if and only if

S/sl S, G)-projective

(S/sS, G)-projective

lL

S/sl S,

mM

S/m )-proj.

nN

S/(xn , yn ),

mM

S = {1}

All cyclic generators are isomorphic to

eS, e E(S), eJ 1

(S/sS, S/)-projective

103

S/m )-proj.

S = {1} or S = {0, 1}

S = {1} or S = {0, 1}
S = {1} or S = {0, 1}

(S/(x, y), S/))-projective

S = {1} or S = {0, 1}

(S/sS, S/(x, y))-projective

S = {1} or S = {0, 1}

(S/(x, y), S/(

x, y))-projective

S = {1} or S = {0, 1}

(S/sS, S/I)-projective

S is a group with zero adjoined or S = {1}

(S/(x, y), S/I)-projective

S is group with zero adjoined or S = {1}

(S/sS, S/tS)-projective

S is the disjoint union of a group and a left zero

semigroup or S = {1}

(S/(x, y), S/tS)-projective

S is the disjoint union of a group and a left zero

semigroup or S = {1}

Table 7: Characterization of monoids with left zero

104

5 HOMOLOGICAL CLASSIFICATION

All right S-acts are

nN

S/(xn , yn ),

implies

mM

S/m )

S fullls condition (Moz)

-projective (*)

(S/(x, y), S/)-projective (*)

S fullls condition (Moz)

(S/sS, A)-projective

S is a principal right ideal monoid

(S/sS, S/tS)-projective

S is the disjoint union of R and right simple

semigroups i S
or S is a group
or S is the disjoint union of a group and a
left zero semigroup

Table 8: Necessary conditions on a monoid S

((*) where

is a right x-congruence,

x{,R,pR})

5.3 Summarizing Tables

(S,A)

S,A)

105

(S,S/I)

(S,S/tS)

(S/,A)

iI

lL

l.z.

S/sl S,A)

S/n ,A)

(S/,S/I)

(S/,S/tS)

(Moz)

(Moz)

nN

l.z.

(S/sS,A)

(Moz)

l.z.

(S/sS,S/I)

l.z.

l.z.

l.z.

l.z.

(Moz)

(Moz)

l.z.

(S/sS,S/tS)

(Moz)

Table 9: Characterization of monoids by implications

(S,S/)

iI

lL

S/sl S,

S/m )

S/m )

zero

(S/,S/)

mM

zero

(S/sS,S/)

S,

nN

S/n ,

S/m )

mM

(Moz)

zero

(Moz)

(Moz)

mM

Table 10: Characterization of left reversible monoids by implications

106

5 HOMOLOGICAL CLASSIFICATION

(S,A)

S,A)

(S,S/I)

(S,S/tS)

l.z.

l.z.

l.z.

l.z.

iI

l.z.

(S/,A)

l.z.

S/n ,A)

l.z.

nN

(S/,S/I)

l.z.

(S/,S/tS)

Table 11: Characterization of monoids fullling condition (Moz) by

implications

(S,S/)

iI

lL

S/sl S,

S/n )

nN

zero

(S/sS,S/)

S,

S/n )

zero

zero

nN

Table 12: Characterization of left reversible monoids, which fulll condition

(Moz), by implications

107

Dierences between the concepts

In the rst Section of this Chapter the (X , Y)-projectivities are proved to be

dierent from each other (we did not show this by now). Some distinctions
are already clear by the results concerning homological classication. In this
cases, these Results will be used for putting out the respective dierences,
sometimes concrete examples will be added. Other dierences will be proved
only by giving concrete examples.
Subsection 1 treats dierences between the various branches of the implication
scheme (part 1 and part 2), which has been presented at the end of Chapter 3.
Subsection 2 is divided in three subsubsections: the rst one deals with projectivities, where the second component is the class of all generators in Act S,
the second one with (X, A)- and (X, C)-projectivities and the third with those
projectivities, where the second component is a cyclic act or a coproduct of
cyclic acts.
The second Section of this Chapter treats dierences between (X , Y)-projectivities and other properties underneath projectivity. The considered properties
atness properties and torsionfreeness have shortly been introduced in
Chapter 2 (Preliminaries). Some results concerning homological classication
relative to these properties have also been recalled (Results 2.35), and beneath
Corollary 2.34 they will be used to prove distinctions.
Note that an implication scheme of the form
A
C

implies

D .

Consequently we only prove the rst situation in the following.

At the beginning of each example, the abbreviation A = B indicates the
situation, for which an example is given, i.e., the example shows that property
A does not imply property B.
During this Chapter let GS be a generator in Act S, let AS , BS , CS be right
S-acts, let s, t, sl , xn , yn S, let , m , m M, denote right congruences
Ik , k K, be right ideals of S.
on S and let I,

108

6 DIFFERENCES BETWEEN THE CONCEPTS

Note, that we will use the properties of the one element act S , which have
been studied in Propositions 5.8, 5.9, 5.10, 5.11 and in Corollary 5.14. Before
considering the respective situations, we treat four special monoids and their
factor acts.

Examples 6.1

1. Let S = {a, 1} be the two element group.

Then (1, a) = S S = aS is the unique principal (Rees) congruence

on S.
Thus (S/)S = (S/)S
= SS and (S/)S = (S/S 2 )S
= S are all factor
acts of S by right congruences.
2. Let S = {1, a, 0} be the two element group with zero 0 adjoined.
Then all monocyclic acts are given by (S/(1, 0))S
= (S/(a, 0))S
= S
and (S/(1, a))S with (1, a) = {(1, a), (a, 1)} .
(S/1S)S = (S/aS)S
= S and (S/0S)S
= SS are all Rees factor acts of
S.
There are no other factor acts of S by right congruences.
3. Let S = {1, a, b} be the two element right zero semigroup {a, b} with
identity adjoined.
(S/(1, a))S
= (S/(1, b))S and (S/(a, b))S are all monocyclic acts.
(S/1S)S

= (S/bS)S
= S and S/aS
= (S/{a, b})S are all Rees factor
acts.
Moreover, (S/)S
= SS is a factor act of S.
4. Let S = {1, a, b} be the two element left zero semigroup {a, b} with identity adjoined.
(S/(1, a))S
= (S/(1, b))S
= S , (S/)S
= SS and (S/(a, b))S are all
monocyclic acts.
(S/1S)S
= S and (S/aS)S
= (S/bS)S
= SS and (S/{a, b})S are all
Rees factor acts.
There are no further factor acts of S by right congruences on S.

6.1 Dierences between (X , Y)-projectivities

109

6.1

Dierences between (X , Y)-projectivities

6.1.1

Dierences between various parts of the diagram

As mentioned before, in this subsection dierences between various branches

of the implication scheme (part 1 and part 2), given in Chapter 3, will be
proved.

Example 6.2 (

S,

iI

mM

S/m ) -projective = (

S, G)-projective

iI

Let S = {0, 1} be the monoid consisting of a zero and an identity. By Theorem

5.30, all right S-acts are ( S,
S/m )-projective. Since S is not a group,
iI

mM

by Theorem 5.19, there exists a right S-act, which is not (

S, G)-projective.

iI

Example 6.3 (

iI

S,

mM

S/m ) -projective = (

lL

S/sl S, A)-projective

Let S = {0, 1} and let AS be the amalgamated coproduct of two copies of S

by the zero. By Theorem 5.30 all right S-acts are ( S,
S/m ) -projective
and so is AS . Since S has a left zero, (
to (

S, A)-projectivity. Let g : S

iI

lL

iI

mM

S/sl S, A)-projectivity is equivalent

S A where g|S is the identity. Then

the identity idA on AS can not be lifted with respect to g, i.e., AS is not


( S/sl S, A)-projective.
lL

Example 6.4 (
(

iI


iI

S, G) -projective = (S, S/tS)-projective

S, G) -projective = (S, A)-projective

Let S be a group with at least two elements. By Theorem 5.19 all right S-acts


are ( S, G)-projective. Since S has no left zero, the one-element act S is
iI

not (S, S/tS)-projective and not (S, A)-projective by Proposition 5.8.

As a concrete example take (S, ) = (Z3 , +) (or (Zn , +) for an arbitrary natural
number n 2).

110

6 DIFFERENCES BETWEEN THE CONCEPTS

Example 6.5 (

nN

nN

S/(xn , yn ), C) -projective = (S, S/(x, y))-projective

S/(xn , yn ), C) -projective = (S, S/sS)-projective

Consider Example 6.1.1. Since S has no left zero and fullls condition (Moz),
by Propositions 5.8 and 5.9 the one element act S is not (S, S/(x, y))projective and not (S, S/sS)-projective.
Since every monocyclic act is isomorphic to S , every coproduct of monocyclic




S for some set N = . Thus, if g :
(S/(xn , yn ))S
acts is isomorphic to
nN
nN


CS is an epimorphism, then CS
S for some set K = .
=
kK

If now f : S CS is a homomorphism, then there exists k K, such that

f () = k , where k denotes the element of the k-th copy of S of the coproduct. Since g is an epimorphism, there exists an n N, such that g(n ) = k


(n analogously to k ). Then f  : S
S with f  () = n is a
homomorphism with gf  = f , i.e., S is (

Example 6.6 (

lL

nN

nN

S/(xn , yn ), C)-projective.

S/sl S, C) -projective = (S/(x, y), S/tS)-projective

Consider Example 6.1.3.

The Rees factor act (S/{a, b})S = (S/aS)S is ( S/sl S, C)- projective:
lL

Every Rees factor act of S is isomorphic to (S/aS)S or to S . Thus

(S/sl S)S

lL

is the disjoint union of copies of (S/aS)S and S . Therefore an epimorphic

image CS of coproducts of Rees factor acts of S by right principal ideals is the
disjoint union of copies of (S/aS)S , copies of S and amalgamated coproducts
of copies of (S/aS)S by the zero in each copy. Thus if f : (S/aS)S CS
is a homomorphism, then f (S/aS)
= (S/aS)S or f (S/aS)
= S . Thus there


exists f with g f = f .
(S/aS)S is not (S/(x, y), S/tS)-projective:
Since a and b are right zeros of S, for every element x S/(1, b) we have
xa = a
and xb = b, which implies, that (S/(1, b))S has not a zero. Thus the
zero morphism f : (S/aS)S (S/1S)S can not be lifted with respect to the
epimorphism g : (S/(1, b))S (S/1S)S , since otherwise the zero 0 of the
Rees factor act (S/aS)S would give rise to a zero f  (0) in (S/(1, b))S .

Example 6.7 (

lL

S/sl S, C) -projective = (S/(x, y), S/(

x, y))-projective

Consider the group (S, ) = (Z4 , +). Then every coproduct of Rees factor acts
(S/sl S)S of S is isomorphic to a coproduct of copies of S and so is CS , if

6.1 Dierences between (X , Y)-projectivities

there is an epimorphism from

(S/sl S)S onto CS . Thus S is (

lL

111

lL

S/sl S, C)-

projective.
The monocyclic act (S/(1, 2))S is isomorphic to S . Thus there exists an
epimorphism g : (S/(1, 3))S (S/(1, 2))S . Since (S/(1, 3))S does not
contain a zero (see Example 5.3.(c)), the identity id can not be lifted with
respect to g, i.e., S is not (S/(x, y), S/(
x, y))-projective.

Example 6.8 (

nN

S/(xn , yn ), A) -projective = (S/sS, S/(x, y))-projective

Consider for the monoid in Example 6.1 2 the monocyclic act AS := (S/(1, a))S .
Every coproduct of monocyclic acts is isomorphic to a coproduct of copies of
(S/(1, a))S and S .


(S/(xn , yn ))S AS is an epimorphism, then by Corollary 3.6 there
If g :
nN

exists k N, such that g |S/(xk ,yk ) is an epimorphism onto AS , which implies

(S/(xk , yk ))S
= AS .
Then the identitiy idA could be lifted by idA with idA (A) = S/(xk , yk ). The
homomorphism f : AS AS with f (A) = a, where a denotes the class of a
in AS , can also be lifted by idA . Since there are no further homomorphisms


from AS into itself, this implies, that AS is (
S/(xn , yn ), A)-projective.
nN

Since S has a zero, (S/sS, S/(x, y)-projectivity is equivalent to (S, S/(x, y))projectivity by Theorem 4.4. Since AS is a monocyclic act, AS being (S, S/(x, y))projective would imply, that AS is a retract of SS and is therefore projective.
= SS and AS
= (0S)S , by Result 2.27 this leads to a contradiction.
Since AS
Thus AS is not (S, S/(x, y))-projective, i.e., not (S/sS, S/(x, y)-projective.

6.1.2

Generators as second component

Example 6.9 (S, G)-projective = (

S, G)-projective

iI

Let S be a monoid, such that all cyclic generators are isomorphic to eS with
e E(S) and eJ 1 and S be not a group. Then by Theorem 5.21, all right
S-acts are (S, G)-projective, but by Theorem 5.19, there exists at least one


right S-act, which is not ( S, G)-projective.
iI

As a concrete example, take S = {0, 1}. Then the amalgamated coproduct

GS = (S 0 S)S of two copies of S by the zero is a generator in Act S.
Since GS is an indecomposable right S-act, the identity idG can not be lifted


with respect to the canonical epimorphism g : S S G, i.e., GS is not


( S, G)-projective.
iI

112

6 DIFFERENCES BETWEEN THE CONCEPTS

Since S is nite, by Lemma 3.35 all right S acts are (S, G)-projective and so
is GS .

6.1.3

Second component A or C

Example 6.10 (

lL

S/sl S, C)-projective = (S/(x, y), A)-projective

As an example take Example 6.7.

Example 6.11 (

nN

S/(xn , yn ), A)-projective = (

(S/(x, y), A)-projective = (

lL

lL

S/sl S, C)-projective

S/sl S, A)-projective

Consider in Example 6.1.4 the monocyclic act (S/(a, b))S . Then the identity
idS/(a,b) could not be lifted with respect to g : (S/bS)S (S/(a, b))S , since


(S/(a, b))S is not a retract of (S/bS)S . Thus (S/(a, b))S is not ( S/sl S, A)projective and therefore not (

lL

lL

S/sl S, C)-projective.

Since for all x, y S the monocyclic act (S/(x, y))S is isomorphic to (S/(a, b))S
or to S , we get that (S/(a, b))S is a retract of every coproduct of mono

cyclic acts. Thus, (S/(a, b))S is (
S/(xn , yn ), A)-projective and therefore
nN

(S/(x, y), A)-projective .

6.1.4

Cyclic acts and coproducts of cyclic acts as second component

Example 6.12 (

iI

S,

nN

S/(xn , yn ))- projective = (S/sS, S/tS)-projective

Let (S, ) = (N, max) and let AN = (N/2N)N .

Since no monocyclic act contains a zero, there does not exist a homomorphism
f from a Rees factor act of N into a monocyclic act, i.e., AN is trivially







S/(xn , yn ))-projective and thus ( S,
S/(xn , yn ))-projective.
( S,
iI

nN

iI

nN

AN is not (S/sS, S/tS)-projective, since the identity on AN could not be lifted

with respect to the epimorphism g : (N/3N)N AN , g(1N/3N ) = 1N/2N :
Let idA : AN (N/3N)N be a mapping with g idA = idA , which implies
idA (1N/2N ) = 1N/3N . Let 0 denote the zero in AN , 0 the zero in (N/3N)N and
2 the class of 2 in (N/3N)N . If idA is a homomorphism, the equation 0 =
idA (0) = idA (1N/2N 2) = idA (1N/2N ) 2 = 1N/3N 2 = 2 yields the contradiction
0 = 2. Thus the identity idA could not be lifted with respect to g.

6.1 Dierences between (X , Y)-projectivities

Example 6.13 (

S/(xn , yn ),

nN

mM

113

S/m )-projective = (S, S/(x, y))-

projective




(
S/(xn , yn ),
S/m )-projective = (S, S/tS)-projective
nN

mM

Consider (Z3 , +). Then every monocyclic act is isomorphic to S . Thus, if

(S/(xn , yn ))S
(S/m )S is an epimorphism,
(S/m )S is
g :
nN

mM

mM

isomorphic to a coproduct of copies of S . Thus the one element act S is

(S/(xn , yn ))S ,
S/m )-projective.
(
nN

mM

Since S fullls condition (Moz) and S does not contain a left zero, S is not
(S, S/(x, y))-projective by Proposition 5.9. Moreover, since S has not a left
zero, by Proposition 5.8 we get that S is not (S, S/tS)-projective.

Example 6.14 (

lL

jective

S/sl S,

mM

S/m )-projective = (S/(x, y), S/tS)-pro-

Consider Example 6.1.3. Then S is (

lL

S/sl S,

mM

S/m )-projective by Corol-

lary 5.14, since S is left reversible. Since (S/(1, a))S has no zero, S does not
fulll condition (Moz). Thus by Lemma 5.10, S is not (S/(x, y), S/tS)projective.
As a further example one could consider S in Act Z4 for the group (Z4 , +).

Example 6.15 (

lL

S/sl S,

nN

S/(xn , yn ))- projective

= (S/(x, y), S/(

x, y))-projective
(S/sS, S/)-projective = (S/(x, y), S/(
x, y))-projective
Is included in Example 6.7.

Example 6.16 (
= (

jJ

lL

S/(xj , yj ),

S/sl S,

nN

mM

S/m )-projective

S/(xn , yn ))-projective

Is included in Example 6.7.

114

6 DIFFERENCES BETWEEN THE CONCEPTS

Example 6.17 (S, S/I)-projective = (S/sS, S/(x, y))-projective

Consider the monoid in Example 6.1.2. Since S is a group with zero adjoined,
by Corollary 5.54 all right S-acts are (S, S/I)-projective. Since S contains a left
zero, all right S-acts are (S/sS, S/(x, y))-projective i all are (S, S/(x, y))projective, and thus by Theorem 5.24 i S = {1} or S = {0, 1}. Therefore
there exists a non (S/sS, S/(x, y))-projective right S-act.

Example 6.18 (S/(x, y), S/I)-projective = (S, S/tS)-projective

Consider the group (Z3 , +). Since S fullls condition (Moz), by Proposition 5.10 the one element act S is (S/(x, y), S/I)-projective. Since S does
not contain a left zero, by Proposition 5.8 we get, that S is not (S, S/tS)projective.

Example 6.19 (S, S/tS)-projective = (S/sS, S/I)-projective

Consider the monoid in Example 6.1.4.
Since S has a left zero, (S/sS, S/I)-projectivity is equivalent to (S, S/I)projectivity. Consider AS = (S/{a, b})S . Since AS is not a retract of S, it
is not (S, S/I)-projective.
But AS is (S, S/tS)-projective:
Every Rees factor act of S by a principal right ideal is isomorphic to S
or to SS . f : AS S can be lifted with respect to the zero morphism
g : SS S by f  : AS SS with f  (A) = a.
f : AS SS with f (A) = a (or b, respectively) can be lifted with respect to
g = idS by itself.

Example 6.20 (S/(x, y), S/I)-projective = (S, S/tS)-projective

is included in Example 6.5.

Example 6.21 (S/sS, S/I)-projective = (S/(x, y), S/tS)-projective

For every monoid S, the one element act S is (S/sS, S/I)-projective by
Lemma 5.11.
Let S be a monoid, which does not fulll condition (Moz). Then by Lemma

6.2 Dierences from other concepts

115

5.10, S is not (S/(x, y), S/tS)-projective.

As a concrete example take the monoid in Example 6.1.3. The monocyclic act
(S/(1, a))S has no zero, i.e., S does not fulll condition (Moz) and thus by
Lemma 5.10, S is not (S/(x, y), S/tS)-projective.

6.2

Dierences from other concepts

Example 6.22 pullback at = (S, S/tS)-projective

Let S be a left collapsible monoid without left zero. Then S is pullback at
by Corollary 2.34, but by Lemma 5.8 not (S, S/tS)-projective.

Example 6.23 (see also [17]) (

iI

S,

mM

S/m )-projectivity = condition

(P )
Take Z3 = {0, 1, 2} with the usual multiplication of residue classes. Since




Z3 has a left zero, ( S,
S/m )-projectivity is equivalent to (S, S/)iI

mM

projectivity by Lemma 4.4.

Consider in Act Z3 the coproduct Z3 0 Z3 with amalgamated zero whose elements we denote as {0, 11 , 21, 12 , 22 } on which Z3 acts as follows: 1i x = (1x)i ,
2i x = (2x)i for x Z3 \{0}, 1i 0 = 2i 0 = 0, i {1, 2}, and 0 is xed.
All factor acts of Z3 are trivial or isomorphic to Z3 /(1, 2).
It is sucient to prove the lifting property in the following three cases:


For all right congruences  the zero homomorphism f : Z3 0 Z3 Z3 / with






f (Z3 0 Z3 ) = {0} can be lifted by f  : Z3 0 Z3 Z3 with f  (Z3 0 Z3 ) =
{0}.


The homomorphism f : Z3 0 Z3 Z3 /(1, 2) with f (1i ) = 1 = 2 =


f (2i), i {1, 2}, f (0) = 0 can be lifted by taking f  : Z3 0 Z3 Z3
dened by f  (1i ) = 1, f  (2i ) = 2 (i {1, 2}) and f  (0) = 0 .


The homomorphism f : Z3 0 Z3 Z3 /0Z3 with f (1i) = 1, f (2i ) = 2, i


{1, 2}, f (0) = 0 can be lifted by taking f  : Z3 0 Z3 Z3 dened by
f  (1i ) = 1, f  (2i ) = 2 (i {1, 2}) and f  (0) = 0 .


Thus Z3 0 Z3 is weakly projective.


When considering 11 0 = 12 0 one can easily verify that Z3 0 Z3 does not fulll
condition (P ).

116

REFERENCES

References
[1] Ahsan, J., Saifullah, K., Completely quasi-projective monoids, Semigroup
Forum 38, 123126, (1989)
[2] Berthiaume, P., The injective envelope of S-sets, Canad. Math. Bull. 10,
261273, (1967)
[3] Bulman-Fleming, S., Pullback at acts are strongly at Canad. Math.
Bull. 34, 456461, (1991)
[4] Bulman-Fleming, S., The classication of monoids by atness properties
of acts, Proceedings of the Conference on Semigroups and Applications
(St. Andrews), World Scientic, 18-38, (1998)
[5] Bulman-Fleming, S., Normak, P. , Flatness properties of monocyclic acts,
Monatsh. Math. 122, 307-322, (1996)
[6] Bulman-Fleming, S., Normak, P., Monoids over which all at cyclic right
acts are strongly at, Semigroup Forum 50, 233241, (1995)
[7] Burgess, W. D., The injective hull of S-sets, S a semilattice of groups,
Semigroup Forum 23, 241246, (1981)
[8] FitzGerald, D. G., Greens Relations in some Categories of Strong Graph
Homomorphisms , Semigroup Forum, to appear
[9] Gould, V. A. R., Coperfect monoids , Glasg. Math. J. 29, 7388, (1987)
[10] Kilp, M., Characterization of monoids by properties of their left Rees fac
tors, Tartu Ulikooli
Toimetised 640, 2937, (1983) (in Russian)
[11] Kilp, M., On at acts Tartu l. Toimetised 253, 6672, (1970) (in Russian)
[12] Kilp, M., On homological classication of monoids, Sib. Math. J., 396
401, (1972)
[13] Kilp, M., Knauer, U., Characterization of monoids by properties of generators, Communications in Algebra 20(7), 18411856, (1992)
[14] Kilp, M., Knauer, U., Mikhalev, A., Monoids, acts and categories, W. de
Gruyter, Berlin (2000)
[15] Knauer, U., Projectivity of acts and Morita equivalence of monoids, Semigroup Forum 3, 359370, (1972)
[16] Knauer, U., Petrich, M., Characterization of monoids by torsion-free, at,
projective and free acts, Arch. Math. 36, 289294, (1981)

REFERENCES

117

[17] Knauer, U., Oltmanns, H., Weak Projectivities for S-acts , Proceedings
of the Conference on General Algebra and Discrete Math. (Potsdam),
143-159, Aachen (1999)
[18] Laan, V., Pullbacks and atness properties of acts, Diss. Math. Univ.
Tartuensis 15, Tartu (1999)
[19] Normak, P., On equalizer-at and pullback-at acts, Semigroup Forum
Vol. 36, 293313, (1987)
[20] Skornjakov, L. On homological classication of monoids, Sib. Mat. Z. 10,
11391143, (1969)
[21] Stenstrom, B., Flatness and localizations over monoids , Math.
Nachrichten 48, 315334, (1971)

[22] Howie, J. M., Fundamentals of semigroup theory , Clarendon Press, Oxford (1995)
[23] Herrmann, P. J., Projective properties of modules , Algebra Berichte 47,
M
unchen (1984)
[24] Miyashita, Y., Quasi-projective modules, perfect modules, and a theorem
for modular lattices , J. Fac. Sc. Hokk. Univ. 19, (1966)
[25] Park, Y. S., Kim, H. S., A weak projective cover of a module , Kyungpook
Math. J. 31, No.2, 201-205, (1991)
[26] Wisbauer, R., Grundlagen der Modul- un Ringtheorie , Verlag Reinhard
Fischer, Mnchen (1988)

Lebenslauf
Name:
Vorname:
Geburtsdatum:
Geburtsort:
Beruf:
Staatsangehorigkeit:

Studium:
10.86-03.94

Oltmanns
Helga
15. Januar 1967
Westrhauderfehn
Diplom-Mathematikerin
deutsch

Studium an der Carl von Ossietzky Universitat

Oldenburg im Diplomstudiengang Mathematik
Diplomarbeit mit dem Thema
Kontinuitat geordneter Mengen
(basierend auf der Theorie Carl Adam Petris
diskrete Strukturen als Abbilder des Universums)
Betreuerin: Dr. Elke Wilkeit
Diplom: Gesamtnote sehr gut

Promotion:
03.95-06.00

Promotion am Fachbereich Mathematik der Carl

von Ossietzky Universitat Oldenburg
Betreuer: Prof. Dr. Dr. h.c. Ulrich Knauer

20.01.2000
07.06.2000

Abgabe der Dissertation

Rigorosum

1996

Stipendium der Heinz Neum

uller Stiftung

04.97-09.2000

Wissenschaftliche Mitarbeiterin am Fachbereich

Mathematik