A study on some combinatorial sets in Partial Semigroups
Arpita Ghosh
[email protected]
[email protected]
Department of Mathematics, University of Kalyani,
Kalyani, Nadia-741235
West Bengal, India
Abstract.
In this article, we investigate the image and preimage of the important combinatorial sets such as central sets, C-sets, and Jδ-sets which play an important
role in the study of combinatorics under certain partial semigroup homomorphism. Using that we prove certain results which deal with the existence of C-set which are not central in partial semigroup framework.
Key words and phrases:
Central sets theorem, Partial semigroups, Algebraic structure of Stone-Čech
compactification
1. introduction
The notion of the central subset of N was originally introduced by Furstenberg [3] in
terms of a topological dynamical system. Before defining central sets let us start with
original Central Sets Theorem due to Furstenberg
Theorem 1.1**.**
(Original Central Sets Theorem)
Let l∈N and for each i∈{1,2,…,l}, and
let ⟨yi,n⟩n=1∞ be a sequence
in Z. Let C be a central subset of N. Then
there exist sequences ⟨an⟩n=1∞
in N and ⟨Hn⟩n=1∞
in Pf(N) such that
for all n, maxHn<minHn+1 and
for all F∈Pf(N) and all
i∈{1,2,…,l},
[TABLE]
.
Proof.
[3, Proposition 8.21].
∎
This theorem has several combinatorial consequences such as Rado’s theorem which deals with the regularity of the system of integral equations. Central sets in natural numbers are known to have substantial combinatorial structure. For example, any central set contains arbitrary long arithmatic progressions, all finite sums of distinct terms of an infinite sequence.
In [1], Vitaly Bergelson and Neil Hindman came with an algebraic characterization of the central subsets of natural numbers in terms of the algebra of βN. Analogously, the notion of a central set has been defined in an arbitrary discrete semigroup S. To define the central subsets in a semigroup, we recall a brief introduction of the algebraic structure of βS for a discrete semigroup (S,⋅).
We take the points of βS to be
the ultrafilters on S, identifying the principal ultrafilters with
the points of S and thus pretending that S⊆βS.
Given A⊆S let us set, A={p∈βS∣A∈p}.
Then the set {A∣A⊆S} is a basis for a topology
on βS. The operation ⋅ on S can be extended to the Stone-Čech
compactification βS of S so that (βS,⋅) is a compact
right topological semigroup (meaning that for any p∈βS,
the function ρp:βS→βS defined by ρp(q)=q⋅p
is continuous) with S contained in its topological center (meaning
that for any x∈S, the function λx:βS→βS
defined by λx(q)=x⋅q is continuous). Given p,q∈βS
and A⊆S, A∈p⋅q if and only if {x∈S∣x−1⋅A∈q}∈p,
where x−1⋅A={y∈S∣x⋅y∈A}.
A nonempty subset I of a semigroup (T,⋅) is called a left ideal
of T if T⋅I⊂I, a right ideal if I⋅T⊂I,
and a two-sided ideal (or simply an ideal) if it is both a left and
a right ideal. A minimal left ideal is a left ideal that does not
contain any proper left ideal. Similarly, we can define a minimal right
ideal.
Any compact Hausdorff right topological semigroup (T,⋅) has a smallest
two sided ideal
[TABLE]
Given a minimal left ideal L and a minimal right ideal R, L∩R
is a group, and in particular, contains an idempotent. An idempotent
in K(T) is called a minimal idempotent. If p and q are idempotents
in T, we write p≤q if and only if p⋅q=q⋅p=p. An idempotent
is minimal with respect to this relation if and only if it is a member
of the smallest ideal. See [9] for an elementary introduction
to the algebra of βS and for any unfamiliar
details.
Now we recall the central sets of a semigroup as
Definition 1.2**.**
Consider a discrete semigroup S and a subset A of S. Then A is central if and only if there is an idempotent in K(βS)∩A.
In [9], the Central Sets Theorem was extended to arbitrary semigroups by allowing the choice of countably many sequences at a time. More extended version of the Central Sets Theorem considering all sequences at one time has been established in [2]. The sets which satisfy the conclusion of the above Central Sets Theorem are the objects that matter. They have several applications in the study of all non-trivial partition regular systems of homogeneous integral equations. Hindman and Strauss understand their importance and named them as C-sets. Also, they found similar kind of algebraic characterizations of C-sets in terms of the algebra of βS and J-sets, which was also introduced by them. We recall the definition of C-set which will be useful later.
Definition 1.3**.**
Let (S,+) be a commutative semigroup and let A⊆S and
T=SN, the set of sequences in S. The
set A is a C-set if and only if there exist functions α:Pf(T)→S
and H:Pf(T)→Pf(N)
such that
(1) if F,G∈Pf(T) and F⊊G,
then maxH(F)<minH(G) and
(2) whenever m∈N,G1,G2,…,Gm∈Pf(T),G1⊊G2⊊…⊊Gm
and for each i∈{1,2,…,m} , fi∈Gi, one has
∑i=1m(α(Gi)+∑t∈H(Gi)fi(t))∈A.
In this article, the main object of study is partial semigroups. In [4], the author extended the Central Sets Theorem obtained by taking all possible sequences in commutative adequate partial semigroup. It states as
Theorem 1.4**.**
Let (S,∗) be a commutative adequate partial semigroup and let C
be a central subset of S. Let TS be the set of all adequate sequences in S. There exist functions α:Pf(TS)→S
and H:Pf(TS)→Pf(N)
such that
F,G∈Pf(TS)* and F⊊G,
then maxH(F)<minH(G) and*
whenever m∈N, G1,G2,…,Gm∈Pf(TS),
G1⊊G2⊊…⊊Gm, and for each
i∈{1,2,…,m}, fi∈Gi, one
has ∏i=1m(α(Gi)∗∏t∈H(Gi)fi(t))∈C.
Proof.
[4, Theorem 2.4].
∎
In the same article, the author also characterized those sets in any adequate commutative partial semigroups which satisfy the new version of the Central Sets Theorem [4, Theorem 2.4] and introduced the analogous notion of C-set and J-set, namely respectively C-set and Jδ-set in adequate partial semigroup as
Definition 1.5**.**
Let (S,∗) be a commutative adequate partial semigroup and A be a subset of S. Let TS be the set of all adequate sequences in S.
a) A is said to be a C-set if and only if there exist functions
α:Pf(TS)→S and H:Pf(TS)→Pf(N)
such that
if F,G∈Pf(TS) and F⊊G,
then maxH(F)<minH(G) and
whenever m∈N, G1,G2,…,Gm∈Pf(TS),
G1⊊G2⊊…⊊Gm
and for each i∈{1,2,…,m}, fi∈Gi,
one has ∏i=1m(α(Gi)∗∏t∈H(Gi)fi(t)∈A.
b) A is a Jδ-set if and only if whenever
F∈Pf(TS), W∈Pf(S), there exist
a∈σ(W) and H∈Pf(N) such
that for each f∈F, ∏t∈Hf(t)∈σ(W∗a) and
a∗∏t∈Hf(t)∈A.
c) Jδ(S)={p∈δS: for all A∈p,A is a Jδ-set}.
For a semigroup S, the C-sets, defined in purely combinatorial terms, are characterized as members of idempotents in J(S). In [4], the author studied a similar kind of properties in the context of adequate partial semigroups and introduced the set Jδ(S). In this article, we first show that for a commutative adequate partial semigroup sets are ideals in δS (cf. Theorem 3.4). Next, we study the behaviour of these combinatorial sets under a surjective partial semigroup homomorphism (cf. Theorem 5.5., Theorem 5.6., Theorem 5.7., and Theorem 5.9.). With the help of these results we able to construct a way to construct enormous amount of C-sets which are not central.
Notation 1.6**.**
TS= The set of all adequate sequences in S.
2. Preliminaries
In this section, we recall some definitions and results from partial semigroup context. For more details, the readers are referred to [9]. We start with the following definition
Definition 2.1**.**
(Partial semigroup) A partial semigroup is defined as a pair (G,∗) where ∗ is an operation
defined on a subset X of G×G and satisfies the statement that for all
x, y, z in G, (x∗y)∗z=x∗(y∗z) in the sense that if either side
is defined, so is the other and they are equal. A partial semigroup is commutative if x∗y=y∗x
for every (x,y)∈X.
Partial semigroups which arise naturally our minds.
Example 2.2**.**
Any semigroup is an obvious example of a partial semigroup.
Example 2.3**.**
Let us consider G=Pf(N)={F∣∅=F⊆N and F is finite} and let X={(α,β)∈G×G∣α∩β=∅}
be the family of all pairs of disjoint sets, and let ∗:X→G
be the union. It is easy to check that this is a commutative partial
semigroup. We shall denote this partial semigroup as (Pf(N),⊎).
Next, we define homomorphisms between partial semigroups.
Definition 2.4**.**
Let (S,∗) and (T,∗′) be partial semigroups and let h:S→T. Then h is a partial semigroup homomorphism if and only if whenever y∈ϕS(x), one has that h(y)∈ϕT(h(x)) and h(x∗y)=h(x)∗′h(y). Here ϕS(s) stands for the set {t∈S∣s∗t is defined in S}.
For a semigroup, it is known that there is a notion of compactification, namely the Stone-Cˇech compactification, which is a compact right topological semigroup. One can do the same thing for any partial semigroup but that won’t be a semigroup. To get a compact topological semigroup out of a partial semigroup we recall the following definitions.
Definition 2.5**.**
Let (S,∗) be a partial semigroup.
(a) For s∈S, ϕS(s)={t∈S∣s∗t is defined in S}.
(b) For H∈Pf(S), σS(H)=⋂s∈HϕS(s).
(c) (S,∗) is adequate if and only if σS(H)=∅
for all H∈Pf(S).
(d) δS=⋂x∈SclβS(ϕS(x))=⋂H∈Pf(S)clβS(σS(H)).
So, the partial semigroup (Pf(N),⊎) is adequate. We are specifically interested in
adequate partial semigroups as they lead to an interesting subsemigroup δS of βS,
the Stone-Čech compactification of S which is itself
a compact right topological semigroup. Notice that adequacy of S
is exactly what is required to guarantee that δS=∅.
If S is, in fact, a semigroup, then δS=βS.
Now we recall some of the basic properties of the operation ∗
in δS.
Definition 2.6**.**
Let (S,∗) be a partial semigroup. For s∈S and A⊆S,
s−1A={t∈ϕS(s)∣s∗t∈A}.
Lemma 2.7**.**
Let (S,∗) be a partial semigroup, let A⊆S and let
a,b,c∈S. Then c∈b−1(a−1A) if and only if both b∈ϕS(a)
and c∈(a∗b)−1A.
In particular, if b∈ϕS(a), then b−1(a−1A)=(a∗b)−1A.
Proof.
[6, Lemma 2.3]
∎
Definition 2.8**.**
Let (S,∗) be an adequate partial semigroup.
(a) For a∈S and q∈ϕS(a), a∗q={A⊆S∣a−1A∈q}.
(b) For p∈βS and q∈δS, p∗q={A⊆S∣{a−1A∈q}∈p}.
Lemma 2.9**.**
*Let (S,∗) be an adequate partial semigroup.
(a) If a∈S and q∈ϕS(a), then a∗q∈βS.
(b) If p∈βS and q∈δS, then p∗q∈βS.
(c) Let p∈βS,q∈δS, and a∈S. Then ϕS(a)∈p∗q
if and only if ϕS(a)∈p.
(d) If p,q∈δS, then p∗q∈δS.*
Proof.
[6, Lemma 2.7].
∎
Lemma 2.10**.**
Let (S,∗) be an adequate partial semigroup and let q∈δS.
Then the function ρq:βS→βS defined by
ρq(p)=p∗q is continuous.
Proof.
[6, Lemma 2.8].
∎
Lemma 2.11**.**
Let p∈βS and let q,r∈δS. Then p∗(q∗r)=(p∗q)∗r.
Proof.
[6, Lemma 2.9].
∎
Definition 2.12**.**
Let p=p∗p∈δS and let A∈p. Then A∗={x∈A∣x−1A∈p}.
Given an idempotent p∈δS and A∈p, it is immediate that A∗∈p.
Lemma 2.13**.**
Let p=p∗p∈δS, let A∈p, and let x∈A∗. Then x−1A∗∈p.
Proof.
[6, Lemma 2.12].
∎
As a consequence of the above results, we have that if (S,∗) is an adequate
partial semigroup, then (δS,∗) is a compact right topological semigroup.
Being a compact right topological semigroup, δS contains
idempotents, left ideals, a smallest two-sided ideal, and minimal idempotents.
Thus δS provides a suitable environment for considering the notion of central sets and it defines as
Definition 2.14**.**
Let (S,∗) be an adequate partial semigroup. A set
C⊆S is central if and only if there is an idempotent
p∈C∩K(δS).
In the Central sets Theorem for semigroup, we have studied that a necessary condition for the central sets along the line of sequences. Here, in the partial semigroup setting we can expect the similar thing for an adequate sequences instead of the normal one.
The notion of adequate sequence plays an important role in the study of central sets theorem for the partial semigroups. So, we recall the definition as
Definition 2.15**.**
Let (S,∗) be an adequate partial semigroup and let⟨yn⟩n=1∞
be a sequence in S. Then ⟨yn⟩n=1∞
is adequate if and only if ∏n∈Fyn is defined for each
F∈Pf(N) and for every K∈Pf(S),
there exists m∈N such that FP(⟨yn⟩n=m∞)⊆σS(K).
Definition 2.16**.**
Let W1,W2∈Pf(S), then define W1∗W2={w1∗w2∣w1∈W1,w2∈W2andw1∗w2is defined}.
The sets which satisfy the new version of Central Sets Theorem 1.4 in adequate partial semigroup are said to be C-sets.
3. Properties of Jδ-sets
This section concerns with the close look up on the Jδ-sets and comes with some essential properties. We start with two essential lemmas.
Lemma 3.1**.**
Let (S,∗) be an adequate partial semigroup. Let f be an adequate sequence in S and let ⟨Hn⟩n=1∞ be a sequence in Pf(N) such that maxHn<minHn+1 for each n∈N. Define g:N→S such that for each n∈N, g(n)=∏t∈Hnf(t). Then g is an adequate sequence in S.
Proof.
To see that g is an adequate sequence, let G∈Pf(N) and let H=⋃n∈GHn. Then H∈Pf(N). Since f is adequate then ∏t∈Hf(t) is defined. Now ∏n∈Gg(n)=∏n∈G∏t∈Hnf(t)=∏t∈Hf(t), then ∏n∈Gg(n) is defined. Let K∈Pf(S) be given. Then there exist m∈N such that FP(⟨f(t)⟩t=m∞)⊆σS(K). Now we want to show that for K∈Pf(S), there exists m∈N such that FP(⟨g(t)⟩t=m∞)⊆σS(K). Let N∈Pf({n∈N:n⩾m}). Then for each n∈N, minHm⩽minHn. Let H′=⋃n∈NHn. Then H′∈Pf({n∈N:n⩾m}) and hence ∏t∈H′f(t)⊆σS(K). Now ∏n∈Ng(n)=∏n∈N(∏t∈Hnf(t))=∏t∈H′f(t)∈σS(K). Therefore, g is an adequate sequence.
∎
Lemma 3.2**.**
Let S be an adequate commutative partial semigroup, let A be a Jδ-set in S. Let F∈Pf(TS), and let ⟨Hn⟩n=1∞ be a sequence in Pf(N) such that for each n, maxHn<minHn+1. Then for all W∈Pf(S), there exist a∈σS(W) and G∈Pf(N) such that for all f∈F, ∏k∈G∏t∈Hk∈σS(W∗a) and a∗∏k∈G∏t∈Hk∈A.
Proof.
For f∈F, define gf:N→S by gf(k)=∏t∈Hkf(t). By Lemma 3.1, gf is an adequate sequence in S. Now since A is a Jδ-set, then for W∈Pf(S) there exist a∈σS(W) and G∈Pf(N) such that for each, f∈F, ∏k∈Ggf(k)∈σS(W∗a) and a∗∏k∈Ggf(k)∈A. These imply ∏k∈G∏t∈Hk∈σS(W∗a) and a∗∏k∈G∏t∈Hk∈A.
∎
Theorem 3.3**.**
Let S be an adequate commutative partial semigroup. Let A be a Jδ-set in S, and assume that A=A1∪A2. Either A1 is a Jδ-set in S or A2 is a Jδ-set in S.
Proof.
Suppose the conclusion of the statement of the theorem is false. That is both A1 and A2 are not Jδ-sets. Then, we can pick F1,F2∈Pf(TS) and W1,W2∈Pf(S) such that for all i∈{1,2} and for all d∈σS(Wi) and K∈Pf(N) such that there exist f∈Fi such that d∗∏t∈Kf(t)∈/Ai∩σS(Wi). Now, set W=W1∪W2 and F=F1∪F2. Assume that F={f1,⋯,fp}.
Next, consider B={w∣wis a word of lengthnover{1,⋯,p}}. For each w=b1⋯bn∈B define a function gw:N→S given by
[TABLE]
where y∈N. Then, Lemma 3.1 guarantees that for each w∈B, gw is an adequate sequence. Define G={gw∣w∈B}⊆Pf(TS). Since, A is a Jδ-set then for this G and the set W chosen earlier we get a∈σS(W) and H∈Pf(N) such that for each w∈B
[TABLE]
By [9, Lemma 14.8.1], we can pick n∈N such that whenever the set B is 2-colored there is a variable word w(v) beginning and ending with a constant and without successive occurrences of v such that {w(l):l∈{1,⋯,p}} is monochromatic.
Define a 2-coloring ϕ:B→{1,2} on the set B given by
[TABLE]
So, we can pick a variable word w(v) beginning and ending with a constant and without successive occurrences of v such that {w(l):l∈{1,⋯,p}} is monochromatic. Without loss of generality, we assume that ϕ(w(l))=1. In other words, we assume that for each l∈{1,⋯,p},
[TABLE]
Suppose w(v)=b1⋯bn and r is the total number of occurrence of v in the word w(v). Then we can write down the set {j∈{1,⋯,n}∣bj∈{1,⋯,p}}=⋃i=1r+1L(i) and {j∈{1,⋯,n}∣bj=v}={s(1),⋯,s(r)} for some function s:{1,⋯,r}→N such that maxL(x)<s(x)<minL(x+1). Precisely, the function s refers the position function for v in w(v). The construction of L(i)’s can be understand by the following example. Let w(v)=2v31v12. Then r=2, L(1)={1}, L(2)={3,4}, and L(3)={6,7}. So, L(i) is the set of j∈{1,⋯,n} such bj∈{1,⋯,p} and bj’s are in between (i−1)th and ith occurrences of v in w(v).
Next, we try to understand the term a∗∏y∈Hgw(l)(y). Using the previous discussion so far we have
[TABLE]
Where
[TABLE]
. Then definitely K is a finite subset of N. Also, (1) tells us W∗a∗∏y∈Hgw(l)(y) is defined i.e. W∗d∗∏t∈Kfl(t) is defined. Using associativity, this implies W∗d is defined. Hence d∈σS(W) and d∗∏t∈Kfl(t)∈A1∩σS(W)⊂A1∩σS(W1). This a contradiction to our assumption.
∎
Theorem 3.4**.**
Let (S,∗) be an adequate commutative partial semigroup. Then Jδ(S) is a closed two sided ideal of δS.
Proof.
Let A be a Jδ-set in S. Then Theorem 3.3 yields the Jδ-sets are partition regular. Next, we claim that Jδ(S)=∅. To prove the claim, we consider the sets R={A⊆S∣Aa is Jδ-set} and A={ϕS(s)∣s∈S}. Observe that for σS(F)=∩s∈FϕS(s) we always can find a Jδ-set A such that A⊆σS(F). This gives that for any finite intersections of the members of A lies inside R↑. Therefore, using Theorem [9, Theorem 3.11] we get a ultrafilter p of S such that A⊆p⊆R↑. The first inclusion implies that for each s∈S, ϕS(s)∈p, therefore, p∈δS. This forces p∈Jδ(S). Hence, the claim is established. Next, we want to show that Jδ(S) is closed. Let p∈δS∖Jδ(S), then there exists B∈p such that B is not a Jδ-set, therefore, B∩Jδ(S)=∅, i.e., Jδ(S) is closed.
We want to prove that Jδ(S) is a two sided ideal of δS, i.e., for all p∈δS, q∈Jδ(S) imply that p∗q∈Jδ(S) and q∗p∈Jδ(S).
Let A∈p∗q. We want to show that A is a Jδ-set.
Since A∈p∗q, then {a∈S∣a−1A∈q}∈p. Pick a∈S such that a−1A∈q. Now q∈Jδ(S), then a−1A is a Jδ-set. Therefore, for F∈Pf(TS) and W∈Pf(S), there exist b∈σS(W∗a) and H∈Pf(N) such that ∏t∈Hf(t)∈σS(W∗a∗b) and b∗∏t∈Hf(t)∈a−1A={d∈ϕS(a)∣a∗d∈A}. This implies a∗b∗∏t∈Hf(t)∈A. Now define c=a∗b. Then for F∈Pf(TS) and W∈Pf(S), there exist c∈σS(W) and H∈Pf(N) such that for each f∈F, ∏t∈Hf(t)∈σS(W∗c) and c∗∏t∈Hf(t)∈A. Therefore, A∈p∗q.
Now let A∈q∗p. We want to show that A is a Jδ-set.
Let B={a∈S∣a−1A∈p}. Since A∈q∗p, then {a∈S∣a−1A∈p}∈q. This implies B∈q. Now q∈Jδ(S), then B is a Jδ-set. Let F∈Pf(TS) and F={f1,f2,⋯,fk}. Let W0∈Pf(S), then there exist b∈σS(W0) and H∈Pf(N) such that for each f∈F, ∏t∈Hf(t)∈σS(W0∗b) and b∗∏t∈Hf(t)∈B. Therefore, (b∗∏t∈Hf(t))−1A∈p for all f∈F and this yields f∈F⋂(b∗∏t∈Hf(t))−1A∈p. Now since p∈δS, then σS(W)∈p for all W∈Pf(S). Therefore, σS(W)∩f∈F⋂(b∗∏t∈Hf(t))−1A∈p for all W∈Pf(S). So it must be nonempty. Take W=W0∗F0, where F0={b∗∏t∈Hf1(t),b∗∏t∈Hf2(t),⋯,b∗∏t∈Hfk(t)} and pick
[TABLE]
Now choose a=b∗c. Then for each f∈F, ∏t∈Hf(t)∈σS(W0∗a) and a∗∏t∈Hf(t)∈A. This suffices A is a Jδ-set.
∎
Corollary 3.5**.**
Let (S,∗) be an adequate commutative partial semigroup. Let A⊆S. If A is a central set in S, then A is a C-set in S.
Proof.
Since A is a central set in S, then there is an idempotent p such that p∈K(δS)∩Aˉ. By lemma 3.4, Jδ(S) is a two sided ideal of δS. Then K(δS)⊆Jδ(S). Therefore, by [4, Theorem 3.4], A is a C-set in S.
∎
4. Construction of Jδ-sets
This section deals with some technical construction of Jδ-sets. We start with some definitions.
Let ω denotes the first infinite ordinal and each ordinal is the set of its predecessors. In particular, [0]=∅ and for n∈N, [n]={0,1,⋯,n−1}.
Definition 4.1**.**
If f is a function and domain(f)=[n]∈ω, then for all x∈S, f⌢x=f∪{(n,x)}. Precisely, it means we extend the domain of f to [n+1] by defining f(n)=x.
Definition 4.2**.**
Let T={f∣f:[n]→S}, i.e., T is a set of functions whose domains are members of ω. For each f∈T, define Bf(T)={x∣f⌢x∈T}.
Using these notions in hand we can stated the following
Lemma 4.3**.**
Let S be an adequate partial semigroup and let p∈δS. Then p is an idempotent if and only if for each A∈p there is a nonempty set T of functions such that
For all f∈T, domain(f)∈ω* and range*(f)⊆A.**
For all f∈T, Bf(T)∈p.
For all f∈T and all x∈Bf(T), Bf⌢x(T)⊆x−1Bf(T).
Proof.
We claim that p is an idempotent element in δS, i.e., {x∈S∣x−1A∈p}∈p. Now there is given A∈p and T which satisfies the above conditions. Pick f∈T. Then Bf(T)∈p. Therefore, if we will prove that Bf(T)⊆{x∈S∣x−1A∈p}, then our claim will be proved. Let x∈Bf(T), then f⌢x∈T. Therefore, by (b) Bf⌢x(T)∈p. Now by (c), Bf⌢x(T)⊆x−1Bf(T)⊆x−1A. Thus x−1A∈p, and so p is an idempotent element.
Conversely, let p be an idempotent element in δS and let A∈p. For any B∈p, let define B∗={x∈B∣x−1B∈p}. Then B∗∈p and by [6, Lemma 2.12], for x∈B∗, x−1B∗∈p. To prove the result we inductively construct a filtration of T=[n]∈ω⋃Tn where Tn={f∈T∣domain(f)=[n]} and for each f∈Tn, define Bf=Bf(T). Note that T∅={0}. Now set B∅=A∗.
Now let [n]∈ω and assume that we have defined Tk for k≤n and defined Bf for f∈Tk such that
Tk is a set of functions with domain [k] and range contained in A.
If f∈Tk and x∈Bf, then Bf∈p and x−1Bf∈p.
If k<n, f∈Tk, and x∈Bf, then Bf⌢x=(x−1Bf)∗.
If n=0, T0={∅}. Then (a) is trivially true.
For (b), B∅=A∗∈p and if x∈A∗, then x−1A∗∈p. Again (c) is vacuously true.
Now, we will check the hypotheses for Tn+1.
Tn+1={f∈T∣domain(f)=[n+1]}. So, for any f∈Tn+1 can be written as f=f∣[n]⌢x where f(n)=x. So, we can write Tn+1={f⌢x∣f∈Tnandx∈Bf}. Now let g∈Tn+1, then we can take g=f⌢x for f∈Tn and x=g(n). Now by (b), for f∈Tn and x∈Bf, x−1Bf∈p. Let Bg=(x−1Bf)∗. Now let y∈Bg=(x−1Bf)∗, then y−1Bg∈p.
Therefore, the hypotheses are true for Tn+1.
∎
Now we are in a position to construct a huge amount of Jδ-sets using the set Bf(T) and some particular idempotent elements in the algebra δS. Precisely, we have
Theorem 4.4**.**
Let (S,∗) be an adequate partial semigroup and let A⊆S. If there is an idempotent p∈Aˉ∩Jδ(S). Then there is a non-empty set T of functions such that:
For all f∈T, domain(f)∈ω* and range*(f)⊆A.**
For all f∈T and all x∈Bf(T), Bf⌢x(T)⊆x−1Bf(T).
For all F∈Pf(T), ⋂f∈FBf(T) is a Jδ-set.
Moreover, there is a downward directed family ⟨AF⟩F∈I of subsets of A such that:
For all F∈I and all x∈AF, there exists G∈I such that AG⊆x−1AF.
For each F∈Pf(I),
⋂F∈FAF is a Jδ-set.
Proof.
Since p is an idempotent element in δS and A∈p, then by Lemma 4.3, (i) and (ii) are true and for each f∈F, Bf(T)∈p. Therefore, ⋂f∈FBf(T)∈p. Now since p∈Jδ(S), then ⋂f∈FBf(T) is a Jδ-set. This concludes part (iii).
For the existence of a downward directed family we do the following construction. Let I∈Pf(T). For F∈I, define AF=⋂f∈FBf(T). For part a), consider G={f⌢x∣f∈F}. Now by Lemma 4.3, for each f∈F, Bf⌢x(T)⊆x−1Bf(T). This implies that AG⊆x−1AF. Next, for part b), using the previous arguments, we obtain AF is a Jδ-set. Given F∈Pf(I), if H=⋃F, then ⋂F∈FAF=AH. Therefore, AH is a Jδ-set.
∎
and also, we can readily reduce
Corollary 4.5**.**
Let (S,∗) be a countable adequate partial semigroup and let A⊆S. If there is an idempotent p∈Aˉ∩Jδ(S), then there is a decreasing sequence ⟨An⟩n=1∞ of subsets of A such that:
For all n∈N and all x∈An, there exists m∈N such that Am⊆x−1An.
For all n∈N, An is a Jδ-set.
Proof.
Let S be countable. Then T is also countable. So identify T as {fn∣n∈N}. For n∈N, let An=⋂k=1nBfk(T). Then each An is a Jδ-set. Let x∈An. Now pick m∈N such that {fk⌢x∣k∈{1,2,⋯,n}}⊆{f1,f2,⋯,fm}. Then Am⊆x−1An.
∎
5. Construction of C-sets which are not central
It is known that the combinatorial sets such as central sets, C-sets, and Jδ-sets have an extra importance in the study of combinatorics. Also, it is elementary from the definition that every central sets are C-sets. In this section, we try to find out a way to construct a C-set which are not central. Before go into that we start with few basic facts.
Lemma 5.1**.**
Let A and B be semigroups and let f:A→B be a surjective homomorphism. If A has a smallest ideal, show that B does as well and that K(B)=f(K(A)).
Proof.
Let b∈B and y∈f(K(A)). Since f is surjective, there exists a∈A such that f(a)=b. Now since y∈f(K(A)), then there exists x∈K(A) such that f(x)=y. As a⋅x∈K(A), then b⋅y=f(a)⋅f(x)=f(a⋅x)∈f(K(A)). Therefore, f(K(A)) is a left ideal in B. Similarly, we can show that it is a right ideal in B.
K(A)=⋃{L∣Lis a minimal left ideal ofA}. Then, it follows that
[TABLE]
Let J⊆f(L), where J is an ideal of B. Then f−1(J)⊆L. But L is minimal left ideal of A, then J=f(L). Therefore, f(L) is minimal left ideal of B. Then f(K(A))⊂K(B). But K(B) is the smallest ideal, then f(K(A))=K(B).
∎
Corollary 5.2**.**
Let f:A→B be a semigroup homomorphism. If A has a smallest ideal, then f(K(A))=K(f(A)).
Proof.
If f:A→B is a semigroup homomorphism, then f:A→f(A) is a surjective homomorphism. Then by Lemma 5.1, if A has smallest ideal, then f(A) also has smallest ideal and f(K(A))=K(f(A)).
∎
Lemma 5.3**.**
*Let S and T be adequate partial semigroups. Let h:S→T be a surjective partial semigroup homomorphism, i.e., h(S)=T. Let h~:βS→βT be the continuous extension of h. Then h~(δS)⊆δT and h~∣δS is a semigroup homomorphism.
*
Proof.
For the proof, see [9, Theorem 4.22.3].
∎
Remark 5.4**.**
By Lemma 5.3 and Corollary 5.2, h~(K(δS))=K(h~(δS)).
Theorem 5.5**.**
Let S and T be two adequate commutative partial semigroups, let h:S→T be a surjective partial semigroup homomorphism.
If A is a Jδ-set in S, then h(A) is a Jδ-set in T.
Proof.
Given A is a Jδ-set in S. Let F∈Pf(TT) and W∈Pf(T). Pick k:T→S such that h∘k(x)=x locally (pointwise). Construct G=k(F)={k∘f∣f∈F} and WS=k(W). Then G∈Pf(TS) and WS∈Pf(S). Since A is a Jδ-set in S, then there exist a∈σS(WS) and H∈Pf(N) such that for each g∈G,
[TABLE]
Choose b∈T such that b=h(a) and
[TABLE]
Now from one part of the condition (6.1), we have
[TABLE]
which implies ∏t∈Hh(g(t))∈σT(h(WS)∗h(a)), and, so that ∏t∈Hh∘k∘f(t)∈σT(W∗b). This implies ∏t∈Hf(t)∈σT(W∗b) and the other part of the condition (2) gives us that h(a∗∏t∈Hg(t))∈h(A). Since h is partial semigroup homomorphism, then we get, h(a)∗∏t∈Hh(g(t))∈h(A) which clearly implies that b∗∏t∈Hf(t)∈h(A). Therefore, h(A) is a Jδ-set in T.
∎
Theorem 5.6**.**
Let S and T be two adequate commutative partial semigroups, let h:S→T be a surjective partial semigroup homomorphism. Let h~:βS→βT be the continuous extension of h. If h~:δS→δT induces surjective map then
h~(K(δS))=K(δT).**
If A is a central set in S, then h(A) is a central set in T.
If A is a central set in T, then h−1(A) is a central set in S.
Proof.
(a) Given that h~:δS→δT is surjective semigroup homomorphism. Being compact right topological semigroup δS has a smallest two sided ideal, then by Lemma 5.1, δT has also smallest two sided ideal and K(δT)=h~(K(δS)).
(b) Let A be a central set in S, then there is an idempotent p such that p∈K(δS)∩A. Since h~∣δS is a semigroup homomorphism by Lemma 5.3 and hence h~(p) is an idempotent element and by part (a), h~(K(δS))=K(δT). Therefore, h~(p) is contained in K(δT), and, so h~(p) is an idempotent element in h(A)∩K(δT).
(c) Pick an idempotent element p in K(δT)∩A. By part (a), pick q∈K(δS) such that h~(q)=p. Now pick a minimal left ideal L of δS such that q∈L. Then L∩h~−1({p}) is a compact subsemigroup of δS. So there is an idempotent r∈L∩h~−1({p}). Since A∈p and h~(r)=p, then h−1(A)∈r. Thus, h−1(A) is a central set in S.
∎
Theorem 5.7**.**
Let S and T be two adequate commutative partial semigroups, let h:S→T be a surjective partial semigroup homomorphism. Let h~:βS→βT be the continuous extension of h.
h~(Jδ(S))⊆Jδ(T)**
If there is an idempotent p∈A∩Jδ(S), then h(A) is a C-set in T.
If p∈A∩h~(Jδ(S)), then h−1(A) is a C-set in S.
Proof.
(a) Let p∈Jδ(S). We need to show that h~(p)∈Jδ(T). Now let A∈h~(p). Then h−1(A)∈p. Therefore, h−1(A) is a Jδ-set in S. Now by Lemma 5.5, A=h(h−1(A)) is a Jδ-set in T. Therefore, h~(p)∈Jδ(T).
(b) Let p∈A∩Jδ(S). Since h~∣δS is a semigroup homomorphism by Lemma 5.3 and hence h~(p) is an idempotent element in δT. Now since p∈A, then h(A)∈h~(p), and since p∈Jδ(S), then p∈δS such that for all B∈p, B is a Jδ-set in S. Therefore, h~(p)∈δT where h(B)∈h~(p) and h(B) is a Jδ-set in T by Lemma 5.5.
Now let D∈h~(p). We want to show that D is a Jδ-set in T. Now for a surjective map h, D=h(h−1(D)). Since h−1(D)∈p, then h−1(D) is a Jδ-set in S and again by Lemma 5.5 D is a Jδ-set in T. Therefore, we get an idempotent element in h(A)∩Jδ(T). Then by Theorem [4, Theorem 3.4], h(A) is a C-set in T.
(c) Let p be an idempotent element such that p∈A∩h~(Jδ(S)). By Theorem 3.4, Jδ(S) is closed two sided ideal of δS. Now being a closed subset of compact set δS, Jδ(S) is compact and Jδ(S)∩h~−1({p}) is a compact subsemigroup of δS, where h~−1({p})={r∈δS∣h~(r)=p}. Therefore, there is an idempotent q∈Jδ(S)∩h~−1({p}). Since A∈p and h~(q)=p, then h−1(A)∈q. Then by Theorem [4, Theorem 3.4], h−1(A) is a C-set in S.
∎
The following corollary is a direct consequence of Theorem 5.6 and Theorem 5.7.
Corollary 5.8**.**
Let S and T be two adequate commutative partial semigroups, let h:S→T be a surjective partial semigroup homomorphism. Let h~:βS→βT be the continuous extension of h and h~:δS→δT induces surjective map. If there is an idempotent p∈A∩h~(Jδ(S)), where A is not a central set in T, then h−1(A) is a C-set in S but not a central set in S.
Example 5.9**.**
Define h:(Pf(N),⊎)→(N,+) by h(A)=∣A∣, where ∣A∣ is cardinality of the set A. Then, h is a surjective partial semigroup homomorphism. Let h~:βPf(N)→βN be the continuous extension of h. Then, h~(δPf(N))=βN. In [7, Theorem 2.8] Hindman and Strauss gave an example which is a C-set in (N,+), but not central. Since the set A produced in [7, Theorem 2.8] is a C-set, then there is an idempotent element p∈A∩J(N,+) by [8, Theorem 2.5]. As Jδ(N,+)=J(N,+), therefore, we have that p∈A∩Jδ(N,+). So, in our case, h~(Jδ(Pf(N),⊎))=Jδ(N,+). Thus, Theorem 5.7 yields h−1(A) is a C-set, but by Corollary 5.8, it is not central set in (Pf(N),⊎).