
Given two digital images (Xi,ki),i∈{1,2}, first of all we establish a new PNk-adjacency relation in a digital product X1×X2 to obtain a relation set (X1×X2,PNk), where the term ″PN" means ″pseudo-normal". Indeed, a PN-k-adjacency is softer or broader than a normal k-adjacency. Next, the present paper initially develops both PN-k-continuity and PN-k-isomorphism. Furthermore, it proves that these new concepts, the PN-k-continuity and PN-k-isomorphism, need not be equal to the typical k-continuity and a k-isomorphism, respectively. Precisely, we prove that none of the typical k-continuity (resp. typical k-isomorphism) and the PN-k-continuity (resp. PN-k-isomorphism) implies the other. Then we prove that for each i∈{1,2}, the typical projection map Pi:X1×X2→Xi preserves a PNk-adjacency relation in X1×X2 to the ki-adjacency relation in (Xi,ki). In particular, using a PN-k-isomorphism, we can classify digital products with PNk-adjacencies. Furthermore, in the category of digital products with PNk-adjacencies and PN-k-continuous maps between two digital products with PNk-adjacencies, denoted by DTC▴k, we finally study the (almost) fixed point property of (X1×X2,PNk).
Citation: Jeong Min Kang, Sang-Eon Han, Sik Lee. Digital products with PNk-adjacencies and the almost fixed point property in DTC▴k[J]. AIMS Mathematics, 2021, 6(10): 11550-11567. doi: 10.3934/math.2021670
[1] | Jeong Min Kang, Sang-Eon Han . The product property of the almost fixed point property for digital spaces. AIMS Mathematics, 2021, 6(7): 7215-7228. doi: 10.3934/math.2021423 |
[2] | Sang-Eon Han . Adjacency relations induced by some Alexandroff topologies on $ {\mathbb Z}^n $. AIMS Mathematics, 2022, 7(7): 11581-11596. doi: 10.3934/math.2022645 |
[3] | Yaoqiang Wu . On partial fuzzy k-(pseudo-)metric spaces. AIMS Mathematics, 2021, 6(11): 11642-11654. doi: 10.3934/math.2021677 |
[4] | Yan Han, Shaoyuan Xu, Jin Chen, Huijuan Yang . Fixed point theorems for $ b $-generalized contractive mappings with weak continuity conditions. AIMS Mathematics, 2024, 9(6): 15024-15039. doi: 10.3934/math.2024728 |
[5] | Mohd Danish Siddiqi, Fatemah Mofarreh . Schur-type inequality for solitonic hypersurfaces in $ (k, \mu) $-contact metric manifolds. AIMS Mathematics, 2024, 9(12): 36069-36081. doi: 10.3934/math.20241711 |
[6] | Velusamy Kavitha, Dumitru Baleanu, Jeyakumar Grayna . Measure pseudo almost automorphic solution to second order fractional impulsive neutral differential equation. AIMS Mathematics, 2021, 6(8): 8352-8366. doi: 10.3934/math.2021484 |
[7] | Gonca Durmaz Güngör, Ishak Altun . Fixed point results for almost ($ \zeta -\theta _{\rho } $)-contractions on quasi metric spaces and an application. AIMS Mathematics, 2024, 9(1): 763-774. doi: 10.3934/math.2024039 |
[8] | Zhen Lin . On the sum of the largest $ A_{\alpha} $-eigenvalues of graphs. AIMS Mathematics, 2022, 7(8): 15064-15074. doi: 10.3934/math.2022825 |
[9] | Ramazan Yazgan . An analysis for a special class of solution of a Duffing system with variable delays. AIMS Mathematics, 2021, 6(10): 11187-11199. doi: 10.3934/math.2021649 |
[10] | Fatimah Alshahrani, Wahiba Bouabsa, Ibrahim M. Almanjahie, Mohammed Kadi Attouch . Robust kernel regression function with uncertain scale parameter for high dimensional ergodic data using $ k $-nearest neighbor estimation. AIMS Mathematics, 2023, 8(6): 13000-13023. doi: 10.3934/math.2023655 |
Given two digital images (Xi,ki),i∈{1,2}, first of all we establish a new PNk-adjacency relation in a digital product X1×X2 to obtain a relation set (X1×X2,PNk), where the term ″PN" means ″pseudo-normal". Indeed, a PN-k-adjacency is softer or broader than a normal k-adjacency. Next, the present paper initially develops both PN-k-continuity and PN-k-isomorphism. Furthermore, it proves that these new concepts, the PN-k-continuity and PN-k-isomorphism, need not be equal to the typical k-continuity and a k-isomorphism, respectively. Precisely, we prove that none of the typical k-continuity (resp. typical k-isomorphism) and the PN-k-continuity (resp. PN-k-isomorphism) implies the other. Then we prove that for each i∈{1,2}, the typical projection map Pi:X1×X2→Xi preserves a PNk-adjacency relation in X1×X2 to the ki-adjacency relation in (Xi,ki). In particular, using a PN-k-isomorphism, we can classify digital products with PNk-adjacencies. Furthermore, in the category of digital products with PNk-adjacencies and PN-k-continuous maps between two digital products with PNk-adjacencies, denoted by DTC▴k, we finally study the (almost) fixed point property of (X1×X2,PNk).
Given two digital images, the study of a Cartesian product with a normal k-adjacency plays an important role in applied topology including digital topology and digital geometry [3,5,6]. Indeed, the paper [3] initially developed the notion of a normal k-adjacency of a digital product [3] which is a digital topological version of the strong adjacency of a graph product in typical graph theory [1]. However, it is clear that these two versions have their own features which need not be equivalent to each other (see Remarks 3.2, 3.5, 3.8, and 3.11, and Example 3.1). A normal k-adjacency of a digital product is essential for studying the multiplicative property of a digital k-fundamental group (see [3]). Besides, it has been often used to examine some product properties of digital topological invariants [10]. Given two digital images (Xi,ki) in Zni,i∈{1,2}, consider a Cartesian product X1×X2. Then, not every X1×X2⊂Zn1+n2 has a normal k-adjacency (see Example 3.2 and Remark 3.11). As a matter of fact, a normal k-adjacency relation between two elements in X1×X2 is very rigid so that we alternatively have the following research goals.
∙ Is there a certain relation in a digital product X1×X2 such that
(1) it is softer than a normal k-adjacency relation,
(2) for each i∈{1,2}, the typical projection map Pi:X1×X2→Xi preserves the established relation in X1×X2 to ki-adjacency relation in (Xi,ki)?
In case this relation is formulated, it can certainly make the earlier works in the literature more generalized and vivid from the viewpoints of digital topology and digital geometry. Hence, in this paper we will develop the so-called PNk-adjacency on X1×X2, denoted by (X1×X2,PNk), supporting this initiative. To be precise, this new PNk-adjacency need not be a typical k-adjacency of Zn, i.e., it is a new relation in a digital product associated with the typical k-adjacency (see Remark 3.5).
Next, given two digital products with PNk- and PNk′-adjacency, (X1×X2,PNk) and (Y1×Y2,PNk′), we naturally pose the following queries.
∙ How to introduce the notion of PN-(k,k′)-continuity of a map or a PN-(k,k′)-isomorphism between (X1×X2,PNk) and (Y1×Y2,PNk′)?
∙ What differences are there between the typical (k,k′)-continuity and the PN-(k,k′)-continuity and further, between a typical (k,k′)-isomorphism and a PN-(k,k′)-isomorphism?
∙ Given digital products with PNk-adjacencies, how to classify them by using a PN-k-isomorphism?
∙ Let DTC▴k be the category of digital products (X1×X2,PNk) and PN-k-continuous maps (for details, see Section 3). Then, how can we establish the (almost) fixed point property of a given (X1×X2,PNk) in DTC▴k?
∙ Under what condition do we have the almost fixed point property of (X1×X2,PNk) in DTC▴k?
To address these issues, first of all we will introduce the notion of a PNk-neighborhood of a point (see the property of (3.3)) in a given digital product with a PNk-adjacency (see Definition 3.4 in the present paper). Besides, for (X1×X2,PNk), we will investigate various properties of a PN-k-continuous self-map and a PN-k-isomorphism.
This paper is organized as follows. Section 2 provides some basic notions which will be used in the paper. Section 3 establishes a pseudo-normal k-adjacency (PNk-adjacency, for short) of a digital product and further, intensively studies some properties of the PN-k-continuity. Section 4 studies various properties of a PN-k-isomorphism. Using a PN-k-isomorphism, we can classify digital products with PNk-adjacencies. Section 5 refers to the (almost) fixed point property of (X1×X2,PNk) in the category of digital products with PNk-adjacencies and PN-k-continuous maps, denoted by DTC▴k. Finally, after establishing the notion of the almost fixed point property (AFPP for brevity) in DTC▴k, we establish some conditions supporting the AFPP in DTC▴k. Section 6 concludes the paper with some remarks and a further work. In the paper we will deal with digital products X×Y such that each of the cardinalities of the sets X and Y is greater than or equal to 2, i.e., |X×Y|≥4.
This section recalls basic notions of the graph-theoretical approach of digital topology, i.e., Rosenfeld model [12,13]. In relation to the study of digital images in Zn, in the case we follow the Rosenfeld model, a digital picture is usually represented as a quadruple (Zn,k,ˉk,X), where n∈N, a black points set X⊂Zn is the set of points we regard as belonging to the image depicted, k represents as an adjacency relation for X and ˉk is used for the elements in Zn∖X. We say that the pair (X,k) is a digital image in a quadruple (Zn,k,ˉk,X). Owing to the digital k-connectivity paradox of a digital image (X,k) [11], we remind the reader that k≠ˉk except the case (Z,2,2,X). However, the present paper is not concerned with the ˉk-adjacency of Zn∖X.
Motivated by the digital k-connectivity for low dimensional digital image in Z3 [11,12], the k-adjacency relations of Zn were initially established to study a high dimensional digital image, as follows [3]: For a natural number m with 1≤m≤n, the two distinct points in Zn
p=(pi)i∈[1,n]Zandq=(qi)i∈[1,n]Z, |
are k(m,n)-adjacent if at most m of their coordinates differ by ±1 and the others coincide. According to this statement, the k(m,n)-adjacency relations of Zn,n∈N, were formulated [3,7]), as follows:
k:=k(m,n)=m∑i=12iCni,whereCni=n!(n−i)! i!. | (2.1) |
For instance,
(n,m,k)∈{(1,1,2),(2,1,4),(2,2,8),(3,1,6),(3,2,18),(3,3,26),(4,1,8),(4,2,32),(4,3,64),(4,4,80),and(5,1,10),(5,2,50),(5,3,130),(5,4,210),(5,5,242).} | (2.2) |
Hereinafter, for our purposes, for {a,b}⊂Z with a⪇b, the set [a,b]Z is assumed to be the set {m∈Z|a≤m≤b}. For a digital image (X,k), two points x,y∈X are k-connected (or k-path connected) if there is a finite k-path from x to y in X⊂Zn [11]. We say that a digital image (X,k) is k-connected (or k-path connected) if any two points x,y∈X is k-connected (or k-path connected). Indeed, in a digital image (X,k) the two notions, k-connectedness and k-path connectedness are equivalent to each other. Besides, a digital image (X,k) with a singleton is assumed to be k-connected for any k-adjacency. Given a k-adjacency relation of (2.1), a simple k-path from x to y on X⊂Zn is assumed to be the sequence (xi)i∈[0,l]Z⊂X⊂Zn such that xi and xj are k-adjacent if and only if either j=i+1 or i=j+1 [11] and further, x0=x and xl=y. The length of this simple k-path, denoted by lk(x,y), is the number l. To be precise, lk(x0,x) is the length of a shortest simple k-path from x0 to x. Besides, a simple closed k-curve with l elements in Zn, denoted by SCn,lk [3,11], is a sequence (xi)i∈[0,l−1]Z in Zn, where xi and xj are k-adjacent if and only if |i−j|=±1(modl) [11].
As a matter of fact, this lk(x,y) induces a certain metric function dk on a digital image (X,k) [5,8]. To be specific, assume a function on a digital image (X,k) which need not be k-connected, as follows:
dk:(X,k)×(X,k)→N∪{0,∞} | (2.3) |
such that
dk(x,x′):={0,ifx=x′,lk(x,x′),ifx≠x′andxisk-connected withx′and∞,ifxis notk-connected withx′.} | (2.4) |
Owing to (2.3) and (2.4), the map dk is obviously a metric function [5,8] and further, we can see that dk(x,x′)≥1 if x≠x′. In addition, in (2.3) and (2.4), in the case (X,k) is k-connected, the codomain of the map dk of (2.3) can be replaced with N∪{0}. Thus, we can represent a digital k-neighborhood of the point x0 with radius 1 [5,8] in the following way [8]
Nk(x0,1)={x∈X|dk(x0,x)≤1}. | (2.5) |
Besides, we can also generalize it to formulate a digital k-neighborhood of the point x0 with radius ϵ∈N, as follows:
Nk(x0,ε)={x∈X|dk(x0,x)≤ε}. | (2.6) |
Indeed, we can represent the typical digital (k0,k1)-continuity of a map by using the above digital k-neighborhood (see Proposition 2.1). Let us investigate some properties of maps between digital images. To map every k0-connected subset of (X,k0) into a k1-connected subset of (Y,k1), the paper [13] established the notion of digital continuity. To represent the digital continuity more conveniently and mathematically, we have the following.
Proposition 2.1. [3,4] Let (X,k0) and (Y,k1) be digital images on Zn0 and Zn1, respectively. A function f:X→Y is (k0,k1)-continuous if and only if for every x∈X, f(Nk0(x,1))⊂Nk1(f(x),1).
In Proposition 2.1, in the case n0=n1 and k0=k1, we call it k0-continuous.
Based on these concepts, let us consider a digital topological category, denoted by DTC, consisting of the following two data [3] (see also [5]):
∙ the set of (X,k) on Zn as objects;
∙ for every ordered pair of objects (X,k0) and (Y,k1), the set of (k0,k1)-continuous maps as morphisms.
In DTC, in the case n0=n1 and k0=k1:=k, we will particularly use the notation DTC(k) [5].
To classify digital images, we have often used the terminology a ″(k0,k1)-isomorphism" as used in [4] rather than a ″(k0,k1)-homeomorphism" as proposed in [2]. For two digital images (X,k0) on Zn0 and (Y,k1) in Zn1, a map h:X→Y is called a (k0,k1)-isomorphism if h is a (k0,k1)-continuous bijection and further, h−1:Y→X is (k1,k0)-continuous. In particular, in the case n0=n1 and k0=k1, we can call it a k0-isomorphism.
For two digital images (X,k1) on Zn1 and (Y,k2) on Zn2, this section formulates the so-called PNk-adjacency relation in a digital product X×Y derived from the given (X,k1) and (Y,k2) so that we obtain a new relation set (X×Y,PNk). Hereinafter, each digital image (X,k) is assumed to be k-connected. Then, we initially establish the notion of PN-k-continuity of a map between two digital products with PNk-adjacencies and compare it with the typical k-continuity. Since this work is associated with a normal k-adjacency of a digital product in [3], let us recall it as follows: Motivated by the strong adjacency in [1] from the viewpoint of graph theory, the following notion was initially established in [3].
Definition 3.1. [3] For two digital images (X,k1) on Zn1 and (Y,k2) on Zn2, a certain k-adjacency on the Cartesian product X×Y⊂Zn1+n2 is defined as follows: For two points (x,y),(x′,y′)∈X×Y, (x,y) is k-adjacent to (x′,y′) if and only if
(1) y is equal to y′ and x is k1-adjacent to x′, or
(2) x is equal to x′ and y is k2-adjacent to y′, or
(3) x is k1-adjacent to x′ and y is k2-adjacent to y′.
Then, we say that the relation set (X×Y,k) is a digital product with a normal k-adjacency derived from the given two digital images (X,k1) and (Y,k2).
Remark 3.2. (1) One of the important things is that the number k for a normal k-adjacency of Definition 3.1 is one of the number k in (2.1).
(2) A normal k-adjacency is a little bit different from the strong adjacency in graph theory in [1]. More precisely, given any two graphs G1,G2, we always have a strong adjacency in [1] for a graph product G1×G2. However, as stated in Definition 3.1, for two digital images (X,k1) and (Y,k2), not every normal k-adjacency exists on the Cartesian product X×Y (see also Proposition 3.3 below). For instance, consider the case SC2,84×SC2,68. Then any k-adjacency of Z4 is not eligible to be a normal k-adjacency for SC2,84×SC2,68⊂Z4.
Based on the product adjacency relation in X×Y⊂Zn1+n2 stated in Definition 3.1, the papers [3,5,6] studied various properties of digital products with normal k-adjacencies. Indeed, a digital product with a normal k-adjacency (X×Y,k) is kind of relation set and further, the normal adjacency relation is symmetric in X×Y. Thus (X×Y,k) is a kind of digital space [9], where we say that a digital space is a nonempty, π-connected, symmetric relation set, denoted by (X,π), if it is k-connected [9]. Thus a digital product with a normal k-adjacency can be represented as a relation set as follows:
Proposition 3.3. [5,6] Assume two digital images (X,k1) on Zn1 and (Y,k2) on Zn2. For the Cartesian product X×Y⊂Zn1+n2, the following are equivalent.
(1) The relation set (X×Y,k) is a digital product with a normal k-adjacency.
(2) For two distinct points p:=(x,y),q:=(x′,y′)∈X×Y,
q∈Nk(p,1)if and only ifbothx′∈Nk1(x,1)andy′∈Nk2(y,1). | (3.1) |
(3) For any point p:=(x,y)∈X×Y,
Nk(p,1)=Nk1(x,1)×Nk2(y,1). | (3.2) |
Indeed, the paper [5] called the normal k-adjacency using the property of (3.2) an S-compatible k-adjacency. As for the normal k-adjacency of a Cartesian product, we have many examples [5,6] including ([a,b]Z×[c,d]Z,8),(SC2,68×[a,b]Z,26),(SC2,68×SC2,68,80),(SC3,618×SC2,48,242), and so on. Furthermore, neither in SC2,84×SC2,68 nor in SC2,84×SC2,84 there exists any k-adjacency in Z4 satisfying Definition 3.1. Namely, given two digital images (X,k1) and (Y,k2), not every Cartesian product X×Y⊂Zn1+n2 has a normal k-adjacency. As mentioned in Remark 3.2, the notion of a normal k-adjacency is rigid. However, its utilities are very huge when studying some product properties of digital topological invariants. Hence, motivated by the several types of equivalent representations of the normal k-adjacency in Proposition 3.3, we may establish the following relation to address the first query posed in Section 1.
Definition 3.4. For two digital images (X,k1:=k1(m1,n1)) on Zn1 and (Y,k2:=k2(m2,n2)) on Zn2, we say that a point (x,y)∈X×Y is related to (x′,y′)∈X×Y on Zn1+n2 if
they are k:=k(m,n1+n2)-adjacent using a certain k-adjacency of Zn1+n2, where m is a certain number with m∈[m1+m2,n1+n2]Z, such that only
(1) in the case y=y′, x is k1-adjacent to x′, and
(2) in the case x=x′, y is k2-adjacent to y′, and
(3) in the case of neither y=y′ nor x=x′, x is k1-adjacent to x′ and y is k2-adjacent to y′.
Then we say that these two related points (x,y) and (x′,y′) are pseudo-normally k-adjacent (PNk-adjacent, for brevity).
After comparing the two adjacencies in Definitions 3.1 and 3.4, we can observe some distinctions between them.
Remark 3.5. (1) The relation of Definition 3.4 is softer or broader than that of Definition 3.1. Indeed, Definition 3.1 is used the condition ″if and only if" for establishing a normal adjacency. More precisely, as mentioned above, given two digital images (X,k1) on Zn1 and (Y,k2) on Zn2 (see Definition 3.1), not every normal k-adjacency exists on the digital product X×Y. However, according to Definition 3.4, we always have a certain PNk-adjacency relation in a digital product as a subset of Zn1+n2.
(2) Two typical k-adjacent points in X×Y need not be PNk-adjacent. For instance, see the point p3 in SC2,44×[0,1]Z (see Figure 1(c)). To be specific, while the point p1 is typically 18-adjacent to p3, it is not PN18-adjacent to p3.
(3) Given the hypothesis of Definition 3.4, not every m∈[m1+m2,n1+n2]Z is used for formulating a certain PNk-adjacency. The number m is taken in [m1+m2,n1+n2]Z depending on the situation. Indeed, there is at least a number m:=m1+m2 determining a PNk-adjacency, where k:=k(m1+m2,n1+n2).
In view of Definitions 3.1 and 3.4, comparing this approach with the normal k-adjacency relation of Definition 3.1, we clearly observe that a PNk-adjacency relation is relatively weak and soft, as follows:
Remark 3.6. A normal k-adjacency implies a PNk-adjacency. However, the converse does not hold.
Example 3.1. Assume a digital image represented via the sequence MSC18:=(x0,x1,x2,x3,x4,x5) which is called a minimal simple closed 18-curve with six elements [5], where x0=(0,0,0),x1=(1,−1,0),x2=(1,−1,1),x3=(2,0,1),x4=(1,1,1),x5=(1,1,0). Furthermore, consider the digital image as a sequence SC3,526:=(y0,y1,y2,y3,y4) [8], where y0=(0,0,0),y1=(−1,−1,1),y2=(−1,0,2),y3=(0,1,2),y4=(1,1,1). Then assume the digital product MSC18×SC3,526⊂Z6 as the following (6×5)-matrix
[ci,j]i∈[0,5]Z,j∈[0,4]Z, |
where ci,j:=(xi,yj)∈Z6. Let us now take the point c3,3=(x3,y3). Then we obtain
N▴k(5,6)(c3,3,1)={ci,j|i,j∈[2,4]Z} |
which is different from Nk(5,6)(c3,3,1), where k(5,6) is the adjacency of (2.1).
In particular, we need to remind that no normal k-adjacency exists on MSC18×SC3,526 (see Proposition 3.3).
Let us now intensively characterize the relation of Definition 3.4 using a certain neighborhood of a point of X×Y. Based on the relation of Definition 3.4, we now establish the following notation.
Definition 3.7. For a point p∈X×Y,
N▴k(p):={q∈X×Y|qisPNk-adjacent top} | (3.3) |
and further,
N▴k(p,1):=N▴k(p)∪{p}. | (3.4) |
Then we call N▴k(p,1) a PNk-neighborhood of p.
Owing to this feature, given a PNk-adjacency relation in a Cartesian product of Definition 3.4, unlike the approach of Definition 3.1, we also have a certain digital space [9] derived from a PNk-adjacency relation.
Remark 3.8. Owing to this structure of (3.3), we obtain the following:
(1) N▴k(p,1) always and uniquely exists in X×Y.
(2) N▴k(p) need not be equal to N∗k(p), where N∗k(p):={q∈X×Y|qisk-adjacent top}.
(3) Not every Nk(p,1) is equal to N▴k(p,1),p∈X×Y.
(4) The adjacency k for establishing an N▴k(p) need not be unique. For instance, given SC2,48 and SC3,618, we can obtain the k-adjacency for establishing an N▴k(p,1)⊂SC2,48×SC3,618⊂Z5,k∈{210,242} (see (2.2)). More precisely, assume a digital product X1×X2 with a normal k-adjacency derived from two digital images (Xi,ki),i∈{1,2}. Then, for each point p of X1×X2, we can establish Nk(p,1) such that N▴k(p,1)=Nk(p,1).
Let us now characterize N▴k(p) with some examples more precisely.
Example 3.2. (1) Let us consider the digital images X:=SC2,44 and (Y:=[0,1]Z,2) (see Figure 1 (1), (a) and (b)). Then we can consider an N▴18(p) in the digital product (X×Y,PN18) (see Figure 1 (2)), where X:=SC2,44:={x1=(0,0),x2=(1,0),x3=(1,1),x4=(0,1)}. Let us now assume the set P:=X×Y={p1=(0,0,0),p2=(1,0,0),p3=(1,1,0),p4=(0,1,0),p5=(0,0,1),p6=(1,0,1),p7=(1,1,1),p8=(0,1,1)}. Then, for the point p1:=(0,0,0), we obtain the following (see Figure 1(2)):
N▴18(p1,1)=(X×Y)∖{p3,p7}. |
Naively, we strongly need to observe that the point p1 is not PN18-adjacent to each of p3 and p7.
(2) Unlike the situation of (1) above, let us consider the set X in (1) above with 8-adjacency, (X,8), instead of (X:=SC2,44,4) and (Y,2). Then we obtain a digital product X×Y with a normal 26-adjacency, i.e., (X×Y,26) such that for any point pi∈X×Y, we obtain N▴26(pi,1)=X×Y=N26(pi,1),i∈[1,8]Z.
In view of Definition 3.4, using N▴k(p) of (3.3), we can observe the following.
Remark 3.9. Given two digital images (Xi,ki) on Zni, i∈{1,2}, for two distinct points p,q∈X1×X2⊂Zn1+n2, the following are equivalent because the PNk-adjacency relation is symmetric.
(1) p and q are PNk-adjacent.
(2) q∈N▴k(p).
As mentioned in Remark 3.9, owing to the symmetric relation of a PNk-adjacency, we can clearly obtain the following:
Based on the PNk-adjacency of a digital product, motivated by the classical notions in a typical digital image in [11] (see the previous part in Section 2), we now define some terminology used in DTC▴k. For a digital product with a certain PNk-adjacency, (X×Y,PNk), we say that two points z,w∈X×Y are PNk-connected (or PNk-path connected) if there is a finite PNk-path (z0,z1,⋯,zm)⊂X×Y from z to w on X×Y such that z0=z and zm=w, where we say that a PNk-path from z to w in X×Y means a finite sequence (z0,z1,⋯,zm)⊂X×Y such that zi is PNk-adjacent to zj if j=i+1 or i=j+1. We say that a digital product (X×Y,PNk) is PNk-connected (or PNk-path connected) if any two points z,w∈X×Y are PNk-connected (or PNk-path connected). In DTC▴k, a singleton as a subset of X×Y is assumed to be PNk-connected. Indeed, in a digital product (X×Y,PNk) the two notions, PNk-connectedness and PNk-path connectedness are equivalent to each other. Given a PNk-adjacency relation in X×Y, a simple PNk-path from z to w in X×Y is assumed to be the PNk-path (zi)i∈[0,l]Z⊂X×Y such that zi and zj are PNk-adjacent if and only if either j=i+1 or i=j+1 and further, z0=x and zl=y. Besides, a simple closed PNk-curve with l elements in X×Y⊂Zn, denoted by SCn,lPNk, is a sequence (zi)i∈[0,l−1]Z in X×Y, where zi and zj are PNk-adjacent if and only if |i−j|=±1(modl).
Proposition 3.10. The relation set (X1×X2,PNk) is a digital space, where (Xi,ki) is ki-connected, i∈{1,2}.
Proof: By Remark 3.9, since the relation PNk in X1×X2 is symmetric, we need to examine if (X1×X2,PNk) is PNk-connected. Take any two distinct points p:=(x1,x2) and q:=(x′1,x′2) in X1×X2. Then, without loss of generality, we may assume the case x1⪇x′1 and x2≤x′2 or the case x1≤x′1 and x2⪇x′2. For our purposes, we may take the first case, i.e., x1⪇x′1 and x2≤x′2. Then consider the differences |x1−x2|⪈0 and |y1−y2|≥0. Depending on these finite differences, we can take a finite set
{p:=p1,p2,p3,⋯,pn:=q}⊂X1×X2 | (3.5) |
such that pi is PNk-adjacent to pi+1 in (X1×X2,PNk),i∈[1,n−1]Z and
p,q∈⋃i∈[1,n]ZN▴k(pi,1)⊂X1×X2. | (3.6) |
Owing to (3.5) and (3.6), we can conclude that (X1×X2,PNk) is PNk-connected.
Remark 3.11. Comparing the two relations of Definitions 3.1 and 3.4, we can confirm some differences between them using a PNk-neighborhood. For instance, for the given digital images (SC2,44,4) and (Y,2) as mentioned in Example 3.2, we can follow Definition 3.4 to obtain a PN18-adjacency relation in SC2,44×Y. Then, while each point p∈SC2,44×Y has an N▴18(p,1) in SC2,44×Y, the Cartesian product SC2,44×Y does not have a normal 18-adjacency because the 18-adjacency does not satisfy Definition 3.1. Thus (SC2,44×Y,PN18) is a digital space.
In view of these notions, we can take the following:
Remark 3.12. Given an (X×Y,PNk) in Zn, we have the following:
(1) A PNk-path need not be equal to a k-path.
(2) SCn,lPNk need not be equal to SCn,lk.
(3) A PN-(k,k′)-continuous map f:(X×Y,PNk)→(X′×Y′,PNk′) implies that any PNk-connected subset of (X×Y,PNk) is preserved onto a PNk′-connected subset of (X′×Y′,PNk′) by the map f.
Based on the relation set (X×Y,PNk) established in Definition 3.4, we obtain the following:
Lemma 3.13. Assume a digital product with a PNk-adjacency, (X1×X2,PNk), derived from two digital images (X,ki),i∈{1,2}. Then, given a point p∈X1×X2, we always obtain N▴k(p,1)⊂Nk(p,1). However, N▴k(p,1) need not be equal to Nk(p,1), i.e., |N▴k(p,1)|≤ |Nk(p,1)|.
Proof: It is clear that for any point q∈N▴k(p,1), according to the property (1) of Definition 3.4, we obtain q∈Nk(p,1). However, in view of Example 3.2(1) as a counterexample, we can disprove Nk(p,1)⊂N▴k(p,1). Naively, consider the digital product (SC2,44×Y,PN18) in Example 3.2. Then we obviously obtain the following property.
N▴18(p1,1)=(SC2,44×Y)∖{p3,p7}andN18(p1,1)=(SC2,44×Y)∖{p7} |
so that N18(p1,1)⊊N▴18(p1,1), which implies that N▴k(p,1) need not be equal to Nk(p,1).
Motivated by a relation preserving mapping, we introduce the following map between two digital products with PNk- and PNk′-adjacency.
Definition 3.14. Consider two digital products (X1×X2,PNk) and (Y1×Y2,PNk′). A function f:(X1×X2,k)→(Y1×Y2,k′) is PN-(k,k′)-continuous at a point p:=(x1,x2) if for any point q∈X1×X2 such that q∈N▴k(p) (denoted by p↔k▴q), we obtain f(q)∈N▴k′(f(p),1) (denoted by f(p)⇔▴k′f(q)). In the case the map f is PN-(k,k′)-continuous at each point p∈X1×X2, we call the map f is PN-(k,k′)-continuous.
In Definition 3.14, in the case k=k′, we say that the map f is a PN-k-continuous map. Besides, it is clear that the property ″f(q)∈N▴k′(f(p),1)" implies that f(p) is equal to f(q) or f(p) is PNk′-adjacent to f(q).
Using both PNk- and PNk′-neighborhood of (X1×X2,PNk) and (Y1×Y2,PNk′), we can mathematically represent the PN-(k,k′)-continuity of a map, as follows:
Proposition 3.15. Assume (X1×X2,PNk) and (Y1×Y2,PNk′). A function f:(X1×X2,k)→(Y1×Y2,k′) is PN-(k,k′)-continuous if and only if for every p∈X1×X2, f(N▴k(p,1))⊂N▴k′(f(p),1).
Proof: Assume any point p∈X1×X2 and take any point q∈X1×X2 such that q∈N▴k(p). By the hypothesis, we obtain f(p)⇔k▴f(q) which implies that f(N▴k(p,1))⊂N▴k′(f(p),1). The converse clearly holds.
Lemma 3.16. The PN-(k,k′)-continuity has the transitive property.
Proof: Assume f:(X1×X2,k)→(Y1×Y2,k′) which is PN-(k,k′)-continuous and g:(Y1×Y2,k′)→(Z1×Z2,k′′) which is PN-(k′,k′′)-continuous. For any point p∈X1×X2 since f(N▴k(p,1))⊂N▴k′(f(p),1)) and g(N▴k′(f(p),1))⊂N▴k′′(g(f(p)),1)), the proof is completed.
Theorem 3.17. Assume (X1×X2,PNk) derived from two digital images (Xi,ki),i∈{1,2}, where ki:=k(mi,ni). Then, each of the projection maps P1 and P2 preserve the PNk-adjacency of X1×X2 onto the k1- and k2-adjacency, respectively.
Proof: Assume a digital product with a PNk-adjacency, (X1×X2,PNk) derived from two digital images (X,ki),i∈{1,2}. For each point p:=(x1,x2)∈X1×X2, consider the PNk-neighborhood of p, N▴k(p,1). Next, consider each of the projection maps P1 and P2 as follows:
{P1:X1×X2→X1given byP1(p)=x1,P2:X1×X2→X2given byP2(p)=x2} | (3.7) |
Then, since N▴k(p,1)=Nk1(x1,1)×Nk2(x2,1), we obtain
P1(N▴k(p,1))=Nk1(x1,1)andP2(N▴k(p,1))=Nk2(x2,1), |
which implies the preservation of the PNk-adjacency of X1×X2 onto the k1- and k2-adjacency by each projection map.
Let us now compare the typical (k,k′)-continuity and the PN-(k,k′)-continuity
Theorem 3.18. None of the typical (k,k′)-continuity and the PN-(k,k′)-continuity implies the other.
Proof: For the sake of contradiction, we will use counterexamples for the proof. Consider the two digital images (X,4) and (Y,2), where X={x1=(0,0),x2=(1,0),x3=(1,1),x4=(0,1)}. Then we may assume a PN18-adjacency of digital product X×Y with such as (P:=X×Y,PN18) and further, (Q:=SC2,44×Y,PN18) as in Figure 2. Let us now consider the following two cases.
(1) Let us assume the map f:P→Q given by
f(pi)=qi,i∈[1,8]Z∖{3,7}andf(p3)=q7=f(p7). |
Naively, we have f(P)=Q∖{q3}.
Then, for the points pi∈X×Y,i∈{1,5} we obtain f(N▴18(pi,1))=N▴18(f(pi),1). In particular, we see
f(N▴18(p3,1))=Q∖{q1,q3,q5}⊊N▴18(f(p3),1)=Q∖{q1,q5}=N▴18(q7,1). |
This implies the PN-18-continuity of f.
However, the map f is not an 18-continuous map in DTC(18) because while p3∈N18(p1,1), we have
f(p3)=q7∉N18(f(p1,1))=N18(q1,1), |
which implies the typical non-18-continuity of f (see Proposition 2.1).
(2) Let us consider the map g:P→Q given by (see Figure 2(2))
g(pi)=qi,i∈[1,8]Z∖{3,5}andg(p3)=q5,g(p5)=q3, | (3.8) |
where Q:=SC2,44×[0,1]Z. In DTC(18), since g(N18(pi,1)=N18(g(pi),1),i∈[1,8]Z, g is a typically 18-continuous map in DTC(18). However, consider the points p1,p5∈X×Y. For the point p5∈N▴18(p1,1), we obtain g(p5)∉g(N▴18(p1,1)), which implies the non-PN-18-continuity of the map g at the point p1.
Based on these concepts, let us establish the category for digital products with a PNk-adjacency, denoted by DTC▴, consisting of the following two data:
∙ The set of digital products (X1×X2,PNk) as objects;
∙ For every ordered pair of objects (X1×X2,PNk) and (Y1×Y2,PNk′), the set of PN-(k,k′)-continuous maps as morphisms.
In DTC▴, in the case k=k′, we will particularly use the notation DTC▴k.
By Theorem 3.18, we obtain the following:
Remark 3.19. DTC▴k need not be equal to the category consisting of digital products with a normal k-adjacency or a typical k-adjacency, and their k-continuous maps.
To classify digital products with a PNk-adjacency, we use the following notion. Assume a digital product X1×X2 with a PNk-adjacency, denoted by (X1×X2,PNk), and a digital product Y1×Y2 with a PNk′-adjacency, denoted by (Y1×Y2,PNk′). We may pose the following query: Is there a PN-(k,k′)-continuous bijection f:(X1×X2,PNk)→(Y1×Y2,PNk′) such that the inverse of f is not PN-(k′,k)-continuous?
Example 4.1. (1) Using digital images (X1:=SC2,44,4), (X2:=SC2,48,8) and digital interval (Y:=[0,1]Z,2), consider the digital products (X1×Y,PN18) and (X2×Y,PN26). Let f:(X1×Y,PN18)→(X2×Y,PN26) be a map defined by f(pi)=qi where i∈[1,8]Z. Then, f is a PN-(18,26)-continuous bijection. However, since f−1(N▴26(q1,1))⊈N▴18(f−1(q1),1)=N▴18(p1,1) so that the map f−1 is not a PN-(26,18)-continuous map.
(2) Consider digital images SC2,l4:=(xi)i∈[0,l−1]Z,l∈N0∖{2,6} and ([0,1]Z,2). Then we obtain two relation sets ([0,l−1]Z×[0,1]Z,PN8) and (SC2,l4×[0,1]Z,PN18) (see Figure 4(1)). Then consider the map f:([0,l−1]Z,2)→SC2,l4 given by f(i)=xi which is a (2,4)-continuous bijection. Next, further define the map
g:[0,l−1]Z×[0,1]Z→SC2,l4×[0,1]Z |
given by g(i,t)=(f(i),t)=(xi,t). Then the map g is clearly a PN-(8,18)-continuous bijection. However, it is clear that the inverse of g, g−1, is not PN-(18,8)-continuous because since g−1(N▴18(q,1))⊈N▴8(p,1), where p:=(0,0)∈[0,l−1]Z×[0,1]Z and q:=(x0,0)∈SC2,l4×[0,1]Z. For instance, consider the digital products [0,7]Z×[0,1]Z and SC2,84×[0,1]Z (see Figure 4(1)), where SC2,84:=(xi)i∈[0,7]Z,x0:=(0,0),x1:=(0,−1),x2:=(1,−1),x3:=(2,−1),⋯,x7:=(0,1). Define the map
g:[0,7]Z×[0,1]Z→SC2,84×[0,1]Z |
given by g(i,t)=(f(i),t)=(xi,t). Then the map g is clearly a PN-(8,18)-continuous bijection. However, it is clear that the inverse of g, g−1, is not PN-(18,8)-continuous because since g−1(N▴18(q,1))⊈N▴8(p,1), where p:=(0,0)∈[0,7]Z×[0,1]Z and q:=(x0,0)∈SC2,84×[0,l]Z.
Motivated by Example 4.1, we now establish the following notion.
Definition 4.1. Assume a digital product (X1×X2,PNk) derived from two digital images (Xi,ki),i∈{1,2} and a digital product with (Y1×Y2,PNk′) derived from two digital images (Yi,ki),i∈{1,2}. We say that a map h:(X1×X2,PNk)→(Y1×Y2,PNk′) is a PN-(k,k′)-isomorphism if h is a PN-(k,k′)-continuous bijection and the inverse of h, h−1, is a PN-(k′,k)-continuous map.
Example 4.2. Given the digital images (X,4) and (Y,2), consider two digital products X×Y with a PN18-adjacency such as (P:=X×Y,PN18) and (Q:=X×Y,PN18) as in Figure 3. Then consider the following maps h, f, and g below.
(1) Let us consider the bijection h:(P,PN18)→(Q,PN18) given by (see Figure 3(1))
h(pi)=qi,i∈[1,8]Z∖{3,7}andh(p3)=q7,h(p7)=q3. |
Since h(N▴18(pi,1))=N▴18(h(pi),1) for each point pi∈P, we have the PN-18-continuity of h. Furthermore, h is a PN-18-isomorphism.
(2) Let us now assume the bijection f:(P,PN18)→(Q,PN18) given by (see Figure 3(2))
{f(pi)=qi,i∈[1,8]Z∖{1,3,5,7}andf(p1)=q3,f(p3)=q1,f(p5)=q7,f(p7)=q5.} | (4.1) |
Then, it is clear that f is a PN-18-isomorphism.
(3) Let us recall the map g of (3.8). While it is bijection, it is not a PN-18-isomorphism.
Theorem 4.2. None of a typical (k,k′)-isomorphism and a PN-(k,k′)-isomorphism implies the other.
Proof: For the sake of contradiction, let us consider a counterexample. Let us assume (X1,4)=(Y1,4) and (X2,2)=(Y2,2), where X1:=SC2,44:=Y1 and X2=I:=[0,1]Z:=Y2(see Figure 3). Naively, each of X1×X2 and Y1×Y2 is assumed to be a kind of I3. Then consider the map
h:P:=(X1×X2,PN18)→Q:=(Y1×Y2,PN18) |
defined by (see the map h in Figure 3(1)), as follows:
h(pi)=qi,i∈[1,8]Z∖{3,7}andh(p3)=q7,h(p7)=q3. |
Then, while the bijection h is a PN-18-isomorphic map (see Example 4.2(1)), it is not an 18-isomorphism because
h(N18(p1,1))⊈N18(h(p1),1)=N18(q1,1)=Q∖{q7}. |
Conversely, let us recall the map g used in the proof of Theorem 3.18 (see the map of (3.8)). Then it is obvious that while the map g is an 18-isomorphism, it is not a PN-18-isomorphism.
Motivated by the S-compatible adjacency for a Cartesian product X1×X2 in [6], we can represent it with the following relation between two points in X1×X2 (see Proposition 3.3).
Lemma 4.3. Assume a digital product (X1×X2,PNk) derived from two digital images (X,ki),i∈{1,2}. In case each point p∈X1×X2 has the property N▴k(p,1)=Nk(p,1), the given PNk-adjacency is a normal k-adjacency.
Proof: By Proposition 3.3 and Remark 3.9, the proof is completed.
Theorem 4.4. For SCn,lk and the digital interval (I:=[0,1]Z,2), consider the digital product SCn,lk×I⊂Zn+1 with a certain k′-adjacency of Zn+1. Then we obtain the following.
(1) In the case of k=2n, no normal k′-adjacency exists.
(2) In the case of k≠2n, for 4≤l∈N, a PNk′-adjacency is a normal k′-one.
Proof: (1) In the case of k=2n, we can consider the following two cases.
(Case 1) In the case of l=4, any k′-adjacency of Zn+1 is not a normal k′-one for SCn,lk×I⊂Zn+1. As a counterexample, consider (SC2,44,4) and (I:=[0,1]Z,2) as mentioned in Example 3.2. Then every point p∈SC2,44×I⊂Z3, we observe N▴18(p,1)≠N18(p,1), which implies the assertion.
(Case 2) In the case 8≤l∈N0, the PNk′-adjacency is not a normal one either. Consider (SC2,l4,4) with 8≤l∈N0 and (I:=[0,1]Z,2). Then there is a certain point p∈SC2,l4×I⊂Z3 such that
N▴18(p,1)≠N18(p,1). |
For instance, see the points x0,x2,x4,x6,x8,x10,x12,x14 in Figure 4(1), which supports the assertion. In details, consider (SC2,84×Y,PN18) or a typical image (SC2,84×Y,18) in DTC(18). Then we observe that
N▴18(x2,1)={x1,x2,x3,x9,x10,x11},N18(x2,1)=N▴18(x2,1)∪{x0,x4} |
so that N▴18(x2,1)≠N18(x2,1).
(2) With the hypothesis, in case 4≤l∈N, let us consider any SCn,lk,k≠2n,k:=k(m,n) and take the Cartesian product SCn,lk×I. It is clear that each point p∈SCn,lk×I has N▴k′(p,1) which is equal to Nk′(p,1),k′=k′(m+1,n+1), which implies that in SCn,lk×I the PNk′-adjacency is equal to the normal k′-adjacency. For instance, with the hypothesis, consider SC2,l8,l∈N0∖{2,6} and (I:=[0,1]Z,2). Then every point p∈(SC2,l8×I)⊂Z3 (see Figure 4(2) for the case SC2,68×I), we observe N▴26(p,1)=N26(p,1). Thus we can consider a PN26-adjacency as a 26-normal one.
Example 4.3. Let us consider X:=SCn,lk and Y:=[a,b]Z. For instance, we may consider X∈{SC2,84,SC2,68,SC3,718,SC3,526} and Y:=[0,1]Z. Then we obtain the corresponding PNk-adjacency for the given digital products, k∈{18,26,64,80}.
(1) A digital product (SC2,84×Y,PN18) which does not have a normal 18-adjacency.
(2) Examples of a normal k-adjacency for the given digital products: (SC2,68×Y,26), (SC3,718×Y,64), (SC3,526×Y,80).
This section initially studies the almost fixed point property (AFPP, for brevity) and the fixed point property (FPP, for short) of a digital product with a PNk-adjacency in the category DTC▴k. To do this work, we initially establish these notions as follows:
Definition 5.1. Assume a digital product X1×X2 with a PNk-adjacency derived from (Xi,ki) on Zni,i∈{1,2}.
(1) For a digital product (X1×X2,PNk) on Zn1+n2 and every PN-k-continuous map f:X1×X2→X1×X2, if there exists a point p∈X1×X2 satisfying f(p)=p, then we say that (X1×X2,PNk) has the FPP.
(2) For a digital product (X×Y,PNk) on Zn1+n2 and every PN-k-continuous map f:X×Y→X×Y, if there exists a point p∈X×Y satisfying the property f(p)∈N▴k(p,1), then we say that (X1×X2,PNk) has the AFPP.
Lemma 5.2. [10] Only the (3m−1)-adjacency of the digital product ∏mi=1Xi⊂Zm of the digital intervals (Xi,2) is normal, i∈[1,m]Z.
Let us consider a digital product with a normal k-adjacency, X×Y⊂Zn1+n2, derived from two digital images (X,k1) on Zn1 and (Y,k2) on Zn2 as in Definition 3.1. Then, for a point p∈X×Y, the paper [10] used the notation N⋆k(p,1) (see Remark 4.2 of [10]), as follows:
{N⋆k(p):={q∈X×Y|qis normallyk-adjacent top}andN⋆k(p,1):=N⋆k(p)∪{p}.} | (5.1) |
As a matter of fact, in Theorem 4.4, Remark 4.5, Corollary 4.7, Lemma 4.8 of the paper [10], the authors just stressed on the condition ″N⋆k(p,1)=Nk(p,1)". However, we may not concern about it, as follows:
Remark 5.3. In Theorem 4.4, Remark 4.5, Corollary 4.7, Lemma 4.8 of [10], the condition ″N⋆k(p,1)=Nk(p,1)" is redundant because each of these assertions is already assumed to have a normal k-adjacency of the given digital products.
By Proposition 3.3, we obtain the following:
Lemma 5.4. Assume a digital product with a normal k-adjacency, X×Y⊂Zn1+n2 derived from two digital images (X,k1) on Zn1 and (Y,k2) on Zn2. Then, for any point p:=(x,y)∈X×Y we have
N▴k(p,1)=N⋆k(p,1)=Nk(p,1)=Nk1(x,1)×Nk1(y,1). |
By Lemma 5.2 and Lemma 5.4, we obtain the following:
Lemma 5.5. Assume a digital product (∏mi=1Xi,PNk) on Zn derived from digital intervals (Xi:=[mi,mi+pi]Z,2), i∈[1,n]Z. Then a digital product ∏ni=1Xi has the only PN3n−1-adjacency.
Hereafter, let us study the AFPP for digital n-dimensional cubes, denoted by ∏ni=1[mi,mi+pi]Z⊂Zn,mi∈Z,pi∈N, as a Cartesian product of finite digital intervals [mi,mi+pi]Z.
Corollary 5.6. Assume finite digital intervals (Xi,2),i∈[1,n]Z. If the digital product ∏mi=1Xi⊂Zn1+...+nm has a PNk-adjacency considered on Zn1+...+nm, then (∏mi=1Xi,PNk) has the AFPP in DTC▴k.
Proof: The proof is completed by Lemmas 5.2 and 5.5. More precisely, for the set X:=∏ni=1[mi,mi+pi]Z⊂Zn with PNk-adjacency, where mi∈Zn and pi∈N. Then (X,PNk) has the AFPP if k=3n−1.
Example 5.1. Let us consider a digital square such as [0,1]Z×[0,1]Z derived from the given unit digital interval ([0,1]Z,2). Using Lemma 5.4 and Corollary 5.6, we can consider a PN-8-continuous self-map f of Q:=[0,1]Z×[0,1]Z. Then the digital product (Q,PN8) has the almost fixed point property in the category DTC▴8.
Theorem 5.7. Assume two digital images (X1,k1) in Z2 and (X2,k2) in Z and consider a digital product X1×X2⊂Z3 with a certain PNk-adjacency, k∈{18,26}. If k≠26, then (X1×X2,PNk) does not have the AFPP in DTC▴k.
Before proving the assertion, we can consider PNk-adjacency of this X1×X2, k∈{18,26} (see Definition 3.4)).
Proof: To disprove the assertion, we have the following counterexample. Assume X1:={(x1,x2)|xi∈[0,mi]Zand3≤mi∈N}, i∈[1,2]Z, in Z2. Consider (X1,4) and (X2:=[0,1]Z,2). Then we obtain the corresponding PN18-adjacency for the digital product X1×X2. Then take the two numbers m′1,m′2 such that m′1⪇m1 and m′2⪇m2. Furthermore, assume the following subsets
{A:={(x1,x2)∈X1|x1∈[0,m′1]Z,x2∈[0,m′2]Z},B:={(x1,x2)∈X1|x1∈[0,m′1]Z,x2∈[m′2+1,m2]Z},C:={(x1,x2)∈X1|x1∈[m′1+1,m1]Z,x2∈[0,m′2]Z},D:={(x1,x2)∈X1|x1∈[m′1+1,m1]Z,x2∈[m′2+1,m2]Z}.} | (5.2) |
where m′1∈[1,m1−2]Z and m′2∈[1,m2−2]Z. Then consider the self-map f of X1×X2 defined by
f(x1,x2,w)={(m′1+1,m′2+1,w),(x1,x2)∈Aandw∈X2(m′1+1,m′2,w),(x1,x2)∈Bandw∈X2(m′1,m′2+1,w),(x1,x2)∈Candw∈X2(m′1,m′2,w),(x1,x2)∈Dandw∈X2.} | (5.3) |
Then f is a PN-18-continuous map. However, there is no point (x1,x2,w)∈X1×X2 satisfying f(x1,x2,w)∈N▴18(x1,x2,w,1).
As a more general case, we have the following:
Corollary 5.8. Assume a digital product X1×X2 with a PNk-adjacency derived from two digital images (Xi,ki)⊂Zni,i∈{1,2}. If k≠3n1+n2−1, (X1×X2,PNk) does not have the AFPP in DTC▴k.
Corollary 5.9. Assume a digital product (X1×X2,PNk) derived from (Xi,ki) on Zni,i∈{1,2}. If k≠3n1+n2−1, (X1×X2,PNk) does not have the FPP in DTC▴k.
Theorem 5.10. Assume that digital products (X1×X2,PNk) on Zn derived from (Xi,ki)⊂Zni,i∈{1,2} and (Y1×Y2,PNk′) on Zn′ derived from (Yi,ki)⊂Zmi,i∈{1,2}. If (X1×X2,PNk) has the AFPP in DTC▴k, then a PN-(k,k′)-isomorphism h:(X1×X2,PNk)→(Y1×Y2,PNk′) preserves the AFPP in DTC▴k′.
Proof: Assume any PN-k′-continuous self-map g of (Y1×Y2,PNk′). Then consider the composition g:=h∘f∘h−1 as a PNk′-continuous self-map of (Y1×Y2,PNk′), where f is a PN-k-continuous self-map of (X1×X2,PNk). Owing to the AFPP of (X1×X2,PNk), there exists a point x:=(x1,x2)∈X1×X2 satisfying f(x)∈N▴k(x,1). Owing to the given PN-(k,k′)-isomorphism between (X1×X2,PNk) and (Y1×Y2,PNk′), there is a point y:=(y1,y2)∈Y1×Y2 such that h(x)=y. Let us consider the mapping
f(x)=(h−1∘g∘h)(x)=h−1(g(h(x)))=h−1(g(y)). | (5.4) |
From (5.4), it is clear that h(f(x))=g(y) and further, owing to the hypothesis of the AFPP of (X1×X2,PNk) and the given PN-(k,k′)-isomorphism h, we have
g(y)=h(f(x))∈h(N▴k(x,1))=N▴k′(h(x),1)=N▴k′(y,1), |
which implies that the point h(x) is an almost fixed point of the map g, which implies that (Y1×Y2,PNk′) has the AFPP in DTC▴k′.
Given digital images (Xi,ki),i∈{1,2} and (Yi,ki),i∈{1,2}, after establishing a PNk-adjacency of a digital product X1×X2 and a PNk′-adjacency of a digital product Y1×Y2 and further, we have developed several key notions such as PN-(k,k′)-continuity and a PN-(k,k′)-isomorphism. Using these concepts, we can classify digital products with PNk-adjacencies. Finally, after establishing the notion of AFPP for a digital product with a PNk-adjacency in the category DTC▴k. Based on this work, we have addressed several issues in the fields of digital topology and digital geometry. As a further work, we can define some other adjacency relation on a digital product which can be used in the fields of pure mathematics and applied science.
We would like to thank the referees for valuable suggestions that helped improve the original manuscript in its present form. The corresponding author (S.-E. Han) was supported by the Basic Science Research Program through the National Research Foundation of Korea(NRF) funded by the Ministry of Education, Science and Technology (2019R1I1A3A03059103). Furthemore, this research was supported by "Research Base Construction Fund Support Program" funded by Jeonbuk National University in 2021.
The authors declare no conflict of interest.
[1] | C. Berge, Graphs and Hypergraphs, 2nd ed., North-Holland, Amsterdam, 1976. |
[2] |
L. Boxer, A classical construction for the digital fundamental group, J. Math. Imaging Vis., 10 (1999), 51–62. doi: 10.1023/A:1008370600456
![]() |
[3] | S.-E. Han, Non-product property of the digital fundamental group, Information Sciences 171(1-3) (2005) 73-91. |
[4] | S.-E. Han, On the simplicial complex stemmed from a digital graph, Honam Mathematical Journal, 27 (2005), 115–129. |
[5] |
S.-E. Han, An equivalent property of a normal adjacency of a digital product, Honam Mathematical Journal, 36 (2014), 199–215. doi: 10.5831/HMJ.2014.36.1.199
![]() |
[6] |
S.-E. Han, Compatible adjacency relations for digital products, Filomat, 31 (2017), 2787–2803. doi: 10.2298/FIL1709787H
![]() |
[7] | S.-E. Han, Estimation of the complexity of a digital image form the viewpoint of fixed point theory, Appl. Math. Comput. 347 (2019), 236–248. |
[8] |
S.-E. Han, Digital k-contractibility of an n-times iterated connected sum of simple closed k-surfaces and almost fixed point property, Mathematics, 8 (2020), 345. doi: 10.3390/math8030345
![]() |
[9] |
G. T. Herman, Oriented surfaces in digital spaces, CVGIP: Graphical Models and Image Processing, 55 (1993), 381–396. doi: 10.1006/cgip.1993.1029
![]() |
[10] | J.-M. Kang, S.-E. Han, The product property of the almost fixed point property, AIMS Math. 6 (2021), 7215–7228. |
[11] | T. Y. Kong, A. Rosenfeld, Topological Algorithms for the Digital Image Processing, Elsevier Science, Amsterdam, 1996. |
[12] | A. Rosenfeld, Digital topology, Amer. Math. Monthly, 86 (1979), 76–87. |
[13] |
A. Rosenfeld, Continuous functions on digital pictures, Pattern Recogn. Lett., 4 (1986), 177–184. doi: 10.1016/0167-8655(86)90017-6
![]() |
1. | Sang-Eon Han, Semi-topological properties of the Marcus-Wyse topological spaces, 2022, 7, 2473-6988, 12742, 10.3934/math.2022705 | |
2. | Sang-Eon Han, Selma Özçağ, The semi-T3-separation axiom of Khalimsky topological spaces, 2023, 37, 0354-5180, 2539, 10.2298/FIL2308539H |