
Matching preclusion originates from the measurement of interconnection network robustness in the event of edge failure. Conditional matching preclusion belongs to generalized matching preclusion. We obtain the matching preclusion number and conditional matching preclusion number for hierarchical cubic network(HCNn). Additionally, all the optimal (conditional) matching preclusion sets of HCNn are characterized, which generalize some related results of Birgham et al. and Cheng et al.
Citation: Jinyu Zou, Haizhen Ren. Matching preclusion and conditional matching preclusion for hierarchical cubic networks[J]. AIMS Mathematics, 2022, 7(7): 13225-13236. doi: 10.3934/math.2022729
[1] | Shahid Zaman, Fouad A. Abolaban, Ali Ahmad, Muhammad Ahsan Asim . Maximum $ H $-index of bipartite network with some given parameters. AIMS Mathematics, 2021, 6(5): 5165-5175. doi: 10.3934/math.2021306 |
[2] | Wenxue Sun, Huiyuan Zhao, Xiao Zhang, Yuchao Sun, Xiaoxin Liu, Xueling Lv, Di Fan . Zero-watermarking Algorithm for Audio and Video Matching Verification. AIMS Mathematics, 2022, 7(5): 8390-8407. doi: 10.3934/math.2022468 |
[3] | Jiajia He, Xiuping Zou, Tinghui Li . House price, gender spatial allocation, and the change of marriage matching. AIMS Mathematics, 2024, 9(4): 8079-8103. doi: 10.3934/math.2024393 |
[4] | Mostafa M. A. Khater, A. El-Sayed Ahmed . Strong Langmuir turbulence dynamics through the trigonometric quintic and exponential B-spline schemes. AIMS Mathematics, 2021, 6(6): 5896-5908. doi: 10.3934/math.2021349 |
[5] | Azhar Iqbal, Abdullah M. Alsharif, Sahar Albosaily . Numerical study of non-linear waves for one-dimensional planar, cylindrical and spherical flow using B-spline finite element method. AIMS Mathematics, 2022, 7(8): 15417-15435. doi: 10.3934/math.2022844 |
[6] | Anam Habib, Zareen A. Khan, Nimra Jamil, Muhammad Riaz . A decision-making strategy to combat CO$ _2 $ emissions using sine trigonometric aggregation operators with cubic bipolar fuzzy input. AIMS Mathematics, 2023, 8(7): 15092-15128. doi: 10.3934/math.2023771 |
[7] | Yixin Zhang, Yanbo Zhang, Hexuan Zhi . A proof of a conjecture on matching-path connected size Ramsey number. AIMS Mathematics, 2023, 8(4): 8027-8033. doi: 10.3934/math.2023406 |
[8] | Shenyang Zhao, Fan Wang . Hamiltonian paths passing through matchings in hypercubes with faulty edges. AIMS Mathematics, 2024, 9(12): 33692-33711. doi: 10.3934/math.20241608 |
[9] | Rui Hou, Yongqing Xu, Jinhua Fan, Yuanguo Zhu . Short time asymptotics for American maximum options with a dividend-paying asset. AIMS Mathematics, 2022, 7(8): 13977-13993. doi: 10.3934/math.2022772 |
[10] | Xia Song, Lihua Shen, Fuyang Chen . Adaptive backstepping position tracking control of quadrotor unmanned aerial vehicle system. AIMS Mathematics, 2023, 8(7): 16191-16207. doi: 10.3934/math.2023828 |
Matching preclusion originates from the measurement of interconnection network robustness in the event of edge failure. Conditional matching preclusion belongs to generalized matching preclusion. We obtain the matching preclusion number and conditional matching preclusion number for hierarchical cubic network(HCNn). Additionally, all the optimal (conditional) matching preclusion sets of HCNn are characterized, which generalize some related results of Birgham et al. and Cheng et al.
Let G=(V,E) be a simple graph with vertex set V(G) and edge set E(G). A perfect matching in a graph is a set of edges such that each vertex is incident to exactly one of them, and an almost-perfect matching is a matching covering all but one vertex of G. A graph with an even number of vertices is an even graph, otherwise it is an odd graph. We use G−F to denote the subgraph of G obtained by deleting all the vertices and/or the edges of F⊆V(G)∪E(G).
The matching preclusion number of graph G, denoted by mp(G), is the minimum number of edges whose deletion leaves the resulting graph without a perfect matching or an almost-perfect matching. Any such optimal set is called an optimal matching preclusion set. Birgham et al. [1] first introduced the definition of matching preclusion as a measure of robustness of interconnection networks in the event of edge failure. But beyond that, it also has connections to a few concepts related to theoretical topics, such as the conditional connectivity and extremal graph theory. For more on these topics see the surveys [2,3,4,5,6,7,8,9].
Let G be a graph with an even number of vertices. If it also has the property that the unique optimal matching preclusion sets are those whose elements are incident to a single vertex, then the following Proposition 1.1 holds:
Proposition 1.1. [4] Let G be a graph with an even number of vertices. Then mp(G)≤δ(G), where δ(G) is the minimum degree of G.
We call an optimal solution of the form given in Proposition 1.1 a trivial optimal matching preclusion set. If mp(G)=δ(G), then G is maximally matched, and it is super matched if mp(G)=δ(G) and every optimal matching preclusion set is trivial. Obviously, the super matching network has better robustness.
The definition of conditional matching preclusion number was given in [4]. Let G be a graph in which mp(G)>0. The conditional matching preclusion number of a graph G, denoted by mp1(G), is the minimum number of edges whose deletion leaves the resulting graph with no isolated vertices and without a perfect matching or an almost-perfect matching. Any such optimal set is called an optimal conditional matching preclusion set. If a conditional matching preclusion set does not exist in G, that is, we cannot delete edges to satisfy both conditions, we leave mp1(G) undefined.
In [4], Cheng et al. investigated a basic obstruction to a perfect matching in the resulting graph with no isolated vertices. They found that for a basic obstacle set of perfect matching there would be exist a path u−w−v, where the degree of u and the degree of v are both one. They considered any such path u−w−v in the original graph and defined that νe(G)=min{dG(u)+dG(v)−2−yG(u,v):there exists a 2-path between u and v}, where dG(.) is the degree function, and yG(u,v)=1 if u and v are adjacent and 0 otherwise. Similar to Proposition 1.1, they obtained the following Proposition 1.2:
Proposition 1.2. [4] Let G be a graph with an even number of vertices. Suppose that every vertex in G has degree at least 3. Then mp1(G)≤νe(G).
We call an optimal solution of the form induced by a 2-path with value νe(G) a trivial optimal conditional matching preclusion set. If G is a super matched graph and mp1(G)=νe(G), then it is conditionally maximally matched. If, in additional, all optimal solutions are trivial, then it is conditionally super matched.
At present, matching preclusion set(or conditional matching preclusion set) for various basic interconnection networks has attracted much attention and research interest. For instance, the matching preclusion number for balanced hypercubes was investigated in [10]. In [11], Cheng et al. considered conditional matching preclusion for the arrangement graphs. Wang et al. [12] showed that the conditional matching preclusion number for k-ary n-cube is 4n−2. For more details one can refer to [2,3,4,5,6,10,11,12,13] and reference therein.
The hierarchical cubic network HCNn was introduced by Ghose in [14]. For papers on hierarchical cubic network, we refer to the readers to [15] for a sample of results and additional references for HCNn. The HCNn has many desirable properties such as topological structure. The hierarchical cubic network HCNn(n≥2) can be decomposed into 2n clusters, where each cluster is isomorphic to an n-dimensional hypercube Qn. The matching preclusion and conditional matching preclusion of hypercube have been studied by [1] and [4] respectively. In this paper, we investigate the matching preclusion number and conditional matching preclusion number for hierarchical cubic networks. In Section 2, we introduce some notations, definitions and properties used throughout the paper. In Section 3, we discuss the matching preclusion number of HCNn and characterize all the optimal matching preclusion sets of HCNn. Finally, we consider the conditional matching preclusion number of HCNn and all the optimal conditional matching preclusion sets of HCNn are categorized in Section 4.
For a path, the length of it is the number of edges contained in it. Denote the path of length of k(k≥1) by k-path. For any vertex u∈V(G), use NG(u) to denote the neighborhood of u, that is, the set of vertices adjacent to v. The distance dG(u,v), between a pair of vertices u and v in G, is the length of a shortest path joining u and v.
The Hamming distance, denoted by dH(u,v), between any two vertices u and v, is the number of different positions between the binary strings of u and v.
Let Vn be the set of binary sequence of length n, i.e., Vn={x1x2⋯xn:xi∈{0,1},1≤i≤n}. For x=x1x2⋯xn∈Vn, the element ¯x=¯x1¯x2⋯¯xn∈Vn is called the bitwise complement of x, where ¯xi={0,1}∖{xi} for each i∈{1,2,⋯,n}.
The n-dimensional hypercube network Qn is an cube, shortly n-cube, with its vertex-set Vn, and two vertices are adjacent if and only if they differ exactly in one coordinate. Fig. 1 shows a Q3. For the sake of simplicity, we use xQn to denote the Cartesian product {x}×Qn of a vertex x and a hypercube network Qn.
Definition 2.1. [14] An n-dimensional hierarchical cubic network HCNn with vertex-set Vn×Vn is obtained from 2n n-cubes {xQn:x∈Vn} by adding edges between two n-cubes, called crossing edges, according to the following rule. A vertex (x,y) in xQn is linked to
(1) (y,x) in yQn if x≠y or
(2) (¯x,¯y) in ¯xQn if x=y.
The vertex (y,x) in yQn or (¯x,¯y) in ¯xQn is called on external neighbor of (x,y) in xQn.
The edges between any two different Qn of HCN2 are denoted by the crossing edges. And a 2-dimensional hierarchical cubic network HCN2 is shown in Fig. 2.
From Definition 2.1, it is easy to obtain the following property about crossing edges in HCNn.
(1) There are two crossing edges between two n-cubes xQn and yQn if and only if x and y are complementary, otherwise there is only.
(2) The set of crossing edges consists of a perfect matching of HCNn.
For Qn, the results on the matching preclusion and the conditional matching preclusion of it are as follows.
Lemma 2.2. [1] Let n≥2. Then mp(Qn)=n and Qn is super matched.
Lemma 2.3. [1] Let n≥3. Then mp1(Qn)=2n−2 and Qn is conditionally super matched.
Let Vn={x1,x2,⋯,x2n}, x1Qn,x2Qn,⋯,x2nQn be 2n n-cubes in HCNn. For (z,z′)∈E(HCNn), and z∈V(xiQn), z′∈V(xjQn) where 1≤i≠j≤2n, then z′ is called the external neighbor of z. For notational simplicity, suppose that x=x1. For a graph G, v,w∈V(G), if there are t(t≥2) paths joining v and w, say P1,…,Pt such that V(Pi)∩V(Pj)={v,w} for 1≤i,j≤t,i≠j, then these paths are denoted by internally vertex-disjoint paths.
Lemma 2.4. Let u be a vertex of HCNn, and u′ be its external neighbor. Then there exist n internally vertex-disjoint 7-paths joining endpoints u and u′.
Proof. Without loss of generality, suppose u∈V(xQn),u=(x,y). Let xi,yi,¯xi(1≤i≤n) be the binary sequence of length n which differ from x,y,¯x in the ith bit only respectively, and NxQn(u)={ui=(x,yi),1≤i≤n}.
Suppose d=dH(x,y), then 0≤d≤n. We consider the following two cases.
Case 1. d=0, that is x=y, u=(x,x), u′=(¯x,¯x).
So d(x,xi)=1, the external neighbor of ui=(x,xi)(1≤i≤n) is (xi,x) in xiQn respectively. For each vertex (xi,x)(1≤i≤n), NxiQn((xi,x))={(xi,x1),(xi,x2),…,(xi,xn)}. For (¯x,¯x)∈V(¯xQn), N¯xQn((¯x,¯x))={(¯x,¯x1),(¯x,¯x2),…,(¯x,¯xn)}. For each vertex (¯x,¯xi)(1≤i≤n), its external neighbors is (¯xi,¯x) in ¯xiQn, N¯xiQn((¯xi,¯x))={(¯xi,¯x1),(¯xi,¯x2),⋯,(¯xi,¯xn)}. The paths {(x,x)−(x,xi)−(xi,x)−(xi,xi)−(¯xi,¯xi)−(¯xi,¯x)−(¯x,¯xi)−(¯x,¯x),1≤i≤n} are the desired paths.
Case 2. d≥1, then u′=(y,x).
Then NxQn(u)={(x,yi),i=1,2,⋯,n}, and NyQn(u′)={(y,xi),i=1,2,⋯,n}.
If d≥2, the external neighbors of (x,yi) and (y,xi) are (yi,x) and (xi,y) respectively where 1≤i≤n. And NyiQn((yi,x))={(yi,x1),(yi,x2),⋯,(yi,xn)}, NxiQn((xi,y))={(xi,y1),(xi,y2),⋯,(xi,yn)}. The paths {(x,y)−(x,yi)−(yi,x)−(yi,xi)−(xi,yi)−(xi,y)−(y,xi)−(y,x),1≤i≤n} are the desired paths.
If d=1, there is some i such that x=yi and y=xi. The ith path joining endpoints u and u′ is (x,y)−(x,x)−(¯x,¯x)−(¯x,¯y)−(¯y,¯x)−(¯y,¯y)−(y,y)−(y,x). And we can find the remaining n−1 paths by the case d≥2, completing the proof.
Let Z⊆V(G) and Y⊆V(G)∖Z, the (Y,Z)-paths is a family of internally vertex-disjoint paths starting at a vertex y∈Y ending at a vertex z∈Z and whose internally vertices belong neither to Y nor Z. By Lemma 2.4, we can obtain the following corollary easily.
Corollary 2.5. Let u be a vertex of HCNn, and u′ is its external neighbor, U=NHCNn(u)−u′. Then there are n internally vertex-disjoint paths joining u′ and U.
Let C denote the set of all crossing edges of HCNn. Then E(HCNn)=∪2ni=1E(xiQn)∪C. Suppose F⊆E(HCNn) be a fault edge set of HCNn, Fi=F∩E(xiQn)(i=1,2,⋯,2n), Fc=F∩C, then F=∪2ni=1Fi∪Fc.
Theorem 3.1. Let n≥2. Then mp(HCNn)=n+1.
Proof. As δ(HCNn)=n+1, mp(HCNn)≤n+1. And HCNn is made up of 2n n-cubes and a perfect matching, by Lemma 2.2, mp(Qn)=n. Then mp(HCNn)≥n+1. Thus mp(HCNn)=n+1.
Theorem 3.2. Let n≥2. Then mp(HCNn)=n+1. Moreover, HCNn is super matched.
Proof. By Theorem 3.1, mp(HCNn)=n+1, we now classify the optimal solutions. Let F be any optimal matching preclusion set of HCNn, |F|=n+1, we need to prove HCNn−F has no perfect matchings. As C is a perfect matching of HCNn, |Fc|≥1 and xiQn−Fi has no perfect matching for some i(i=1,2…,n). For notational convenience, we may assume that xQn−F1(x=x1).
xQn is isomorphic to Qn, by Lemma 2.2, |F1|≥n. And |∪2ni=1Fi|≤n, thus |F1|=n, |Fi|=0(i=2,⋯,2n) and xQn−F1 has an isolated vertex, say u=(x,y). If the crossing edge (u,u′)∈F, then u is an isolated vertex in HCNn−F, thus (u,u′)∉F. If dH(x,y)≥2. By Lemma 2.4, choose one path joining u and u′, say P:u=(x,y)−(x,yi)−(yi,x)−(yi,xi)−(xi,yi)−(xi,y)−(y,xi)−(y,x)=u′ for some i∈{1,2,…,n}. And Qn is made up of n edge-disjoint perfect matchings, so xQn−{(x,y),(x,yi)}, yiQn−{(yi,x),(yi,xi)}, xiQn−{(xi,yi),(x,y)}, yQn−{(y,xi),(y,x)} have perfect matchings, say Mx, Myi, Mxi, My. And xiQn−Fi has a perfect matching for xi∈Vn∖{x,yi,xi,y}, say Mxi. Thus M=Mx∪Myi∪Mxi∪My⋃xi∈Vn∖{x,yi,xi,y}Mxi∪((x,yi),(yi,x))∪((yi,xi),(xi,yi))∪((xi,y),(y,xi))∪(u,u′) be a perfect matching of HCNn−F. If dH(x,y)≠2, the perfect matching of HCNn−F can be obtained similarly, completing the proof.
Theorem 4.1. Let n≥3. Then mp1(HCNn)=2n.
Proof. Since νe(HCNn)=2n, mp1(HCNn)≤2n. Let F be any conditional matching preclusion set of HCNn and |F|≤2n−1, it is enough to show that HCNn−F has a perfect matching if HCNn−F has no isolated vertices. By the structure of HCNn, we know that |Fc|≥1. We can claim that there is only one of {xiQn−Fi,1≤i≤2n} with no prefect matching. Otherwise, without loss of generality, suppose both xQn−F1 and x2Qn−F2 have no prefect matching, then |F1|≥n and |F2|≥n by Lemma 2.2. And |F|≥|F1|+|F2|+|Fc|≥2n+1, a contradiction with |F|≤2n−1. For notational convenience, we may assume that xQn−F1. Thus |F1|≥n. And as |F|=|F1|+⋯+|F2n|+|Fc|=2n−1, then |Fi|≤n−2 for i∈{2,3,…,2n}. We consider the following two cases.
Case 1. |Fc|≥2.
Then |F1|+⋯+|F2n|≤2n−3, |F1|≤2n−3. Let |Fc|=i(i≥2), |F1|≤2n−(i+1), and |Fc|≤n−1 for |F1|≥n. From Lemma 2.3, xQn−F1 has an isolated vertex, otherwise xQn−F1 has a perfect matching. For xQn consists of n edge-disjoint perfect matchings, then xQn−u has n edge-disjoint almost perfect matchings. And |F1∩E(xQn−u)|≤n−(i+1), there exists at least (i+1) edge-disjoint almost perfect matchings of xQn−(F1∪{u}). Without loss of generality, suppose the i+1 unmatched vertices are u1,u2,⋯,ui+1. As there is no isolated vertex in HCNn−F, (u,u′)∉F. By Corollary 2.5, there exist (i+1) internally vertex-disjoint paths joining u′ and {u1,…,ui+1}, say P′1,…,P′i+1. And |Fc|=i, thus there is at least one path of {P′1,…,P′i+1} and none of its crossing edges is in Fc, say i=1. Thus P1 is a 7-path joining endpoints u and u′, we can obtain the perfect matching of HCNn−F by the proof of Theorem 3.2.
Case 2. |Fc|=1.
Then |F1|+⋯+|F2n|=2n−2. If xQn−F1 has an isolated vertex. Similar to the proof above, we can obtain the perfect matching of HCNn−F easily.
Now suppose xQn−F1 has no isolated vertex. As xQn−F1 has no perfect matching, then F1 is a trivial conditional matching preclusion set of HCNn, namely |F1|=2n−2, |Fi|=0 for 2≤i≤2n. For notational simplicity, we may assume that it is induced by u1−u−u2. Let (x,yji)(j=2,⋯,n) be adjacent to (x,yi)(i=1,2) such that yji differs from yi in the jth bit only, and u′1,u′2 be the external neighbors of u1 and u2. By |Fc|=1, either (u1,u′1)∉Fc or (u2,u′2)∉Fc, say (u1,u′1). Let NxQn(u1)={u,v1,…,vn−1}. By Corollary 2.5, there are n−1 internally vertex-disjoint 7-paths joining u′1 and {v1,…,vn−1}, say Ri for 1≤i≤n−1. As n≥3 and |Fc|=1, there is at least one path of {R1,…,Rn−1} such that none of its crossing edges is in Fc, say R1. The perfect matching induced by R1 is denoted as M1. And xQn−(F1∪(u1,v1)) has a perfect matching, say Mx1. ∪2ni=1xiQn−(R1−{u1,v1}) also has a perfect matching, say M′, then M=Mx1∪M1∪M′ is a perfect matching of HCNn−F.
Lemma 4.2. [1] Let n≥3. Let F be a conditional matching preclusion set in Qn with |F|=2n−1. Then there exists f,f′∈F such that f and f′ are independent and both Qn−(F−{f}) and Qn−(F−{f′}) contain a perfect matching.
From the structure of Qn, we know that there is a fact: for n≥3, suppose P be a 3-path of Qn, then Qn−P has a perfect matching. Thus we can obtain the following lemmas.
Lemma 4.3. Let v,w∈V(Qn) and d(v,w)=3. Then Qn−{v,w} has a perfect matching.
Lemma 4.4. Let F′ be a fault edge set of Q3 with |F′|≤4, v,w∈V(Q3), d(v,w)=3. If v is an isolated vertex of Q3−F′, then Q3−F′ has an almost perfect matching such that w is unmatched.
Proof. Without loss of generality, let v=000, then w=111. As d(v)=3, |E(Q3−v)∩F′|=1. For any edge of E(Q3−v), we can find an almost perfect matching and w is unmatched.
Lemma 4.5. For n≥3, let F′ be a fault edge set of Qn with |F′|≤2n−2, v,w1,…,wn−1∈V(Qn), d(v,wi)=3(i=1,…,n−2). If v is an isolated vertex of Qn−F′, then Qn−F′ has an almost perfect matching and wi(i=1,…,n−2) is unmatched.
Proof. We complete the lemma by induction on n. For n=3, the conclusion holds by Lemma 4.4, now suppose n≥4 and the Lemma is true for n−1.
Qn contains two (n−1)-dimensional Qn−1, say Q0n−1 and Q1n−1, with the first bit is 0 and 1 respectively, the perfect matching between Q0n−1 and Q1n−1 is defined by M. Let F′i=F′∩Qin−1(i=0,1) and FM=F′∩M. Hence F′=F′0∪F′1∪FM. Without loss of generality, we can assume that v∈V(Q0n−1). As v is an isolated vertex of Qn−F, |FM|≥1. If |FM|≥2, hence |F′0|+|F′1|≤2n−4 and |F′i|≤2n−4 for i=0,1. By inductive assumption, the conclusion holds. Now suppose |FM|=1, |F′0|=2n−3 and |F′1|=0.
Denote the neighbors of v in Q0n−1 by vi(1≤i≤n−1). Let the external neighbors of v and vi(1≤i≤n−1) be v′ and v′i. For Q0n−1 consists of (n−1)-edge disjoint perfect matchings, Q0n−1−v must contain (n−1)-edge disjoint almost perfect matchings, and |E(Q0n−1−v)∩F|≤n−2, there exists an almost perfect matching such that vi is unmatched for some i, say v1. And Q0n−1−{v,v1} has a perfect matching, say M0. As |FM|=1, then (v1,v′1)∉FM. Let NQ1n−1(v′1)={v′,w2,…,wn−1} and w1=v′. As |F1|=0, Q1n−1−{v′1} have (n−1)-edge disjoint almost perfect matching such that wi is unmatched, and d(v,wi)=3, completing the proof.
Lemma 4.6. For n≥3 and any two vertices of V(xiQn)(1≤i≤2n) such that the distance between them is 3. Then there is a cycle L of even order and V(L) contains the two vertices.
Proof. For notational convenience, let i=1, u=(x,y), v=(x,y′). As d(u,v)=3, then dH(y,y′)=3. We complete the proof according to the value of dH(x,y).
Case 1. dH(x,y)=0 or 3.
Then dH(x,y′)=3 or 0, we only need to consider that dH(x,y)=0.
For n=3, we can assume that u=(000,000), v=(000,111). Hence L:(000,000)−(111,111)−(111,110)−(111,,100)−(111,000)−(000,111)−(000,011)−(000,001)−(000,000).
For n≥4, without loss of generality, suppose that u=(0n,0n),v=(0n,0n−3111). Then L:(0n,0n)−(1n,1n)−(1n,1n−3110)−(1n,1n−3100)−(1n,1n−3000)−(1n−3000,1n)−(1n−3000,1n−3110)−(1n−3000,1n−3100)−(1n−3000,1n−3000)−(0n−3111,0n−3111)−(0n−3111,0n−3110)−(0n−3111,0n−3100)−(0n−3111,0n−3000)−(0n,0n−3111)−(0n,0n−3110)−(0n,0n−3100)−(0n,0n).
Case 2. dH(x,y)≠0,3.
Let P:u=(x,y)−(x,z)−(x,w)−(x,y′) denote a 3-path of xQn. Then we can obtain the cycle L:(x,y)−(x,z)−(x,w)−(x,y′)−(y′,x)−(y′,x′)−(x′,y′)−(x′,w)−(x′,z)−(x′,y)−(y,x′)−(y,x)−(x,y) with dH(x,x′)=1 and x′≠y, completing the proof.
By the above lemma, we obtain that there is a path of length odd joining endpoints u and v. The following Lemmas show that there are 2 internally vertex-disjoint paths between the external neighbors of two vertices of HCNn.
Lemma 4.7. For n=3, w,q∈V(xQ3) and w is adjacent to q, w′,q′ are the external neighbors of w and q respectively. There exist 2 internally vertex-disjoint paths of length odd, joining w′ and q′.
Proof. Without loss of generality, suppose x=000, w=(x,z),q=(x,z1), and z1 differs from z in the first bit. We consider the following cases by the value of dH(x,z).
Case 1. dH(x,z)=0.
That is z=000, then dH(x,z1)=1. Let z1=001. w=(000,000),q=(000,001), w′=(111,111),q′=(001,000) are the external neighbor of w and q respectively. There are two internally vertex-disjont paths of length 5 joining w′ and q′, denoted by P1,P2, P1:(111,111)−(111,110)−(110,111)−(110,110)−(001,001)−(001,000), P2:(111,111)−(111,101)−(101,111)−(101,011)−(011,101)−(011,100)−(100,011)−(100,001)−(001,100)−(001,000), see Fig. 3.
Case 2. dH(x,z)=1.
Without loss of generality, let z=001, then dH(x,z1)=0 or dH(x,z1)=2. If dH(x,z1)=0, the result can be obtained by Case 1. Thus dH(x,z1)=2, let z1=011. w=(000,001),q=(000,011),w′=(001,000),q′=(011,000). There are two internally vertex-disjoint paths of length 5 joining (001,000) and (011,000), denoted by P1,P2, P1:(001,000)−(001,010)−(010,001)−(010,011)−(011,010)−(011,000), P2:(001,000)−(001,100)−(100,001)−(100,011)−(011,100)−(011,000), see Fig.4.
Case 3. dH(x,z)=2.
Without loss of generality, suppose z=011, then dH(x,z1)=1 or dH(x,z1)=3. If dH(x,z1)=1, we can obtain the result by case 2. Thus dH(x,z1)=3, (x,z)=(000,011), and let (x,z1)=(000,111). w=(000,011),q=(000,111), w′=(011,000) and q′=(111,000). There exist two internally vertex-disjoint 5-paths joining w′ and q′, denoted by P1,P2, P1:(011,000)−(011,010)−(010,011)−(010,111)−(111,010)−(111,000), P2:(011,000)−(011,100)−(100,011)−(100,111)−(111,100)−(111,000), see Fig. 5.
Case 4. dH(x,z)=3.
Then dH(x,z′)=2, the proof is similar to case 3.
Lemma 4.8. For n≥3, let w,q∈V(xQn) and w is adjacent to q, w′,q′ are the external neighbors of w and q respectively. There exist 2 internally vertex-disjoint paths of length odd, joining w′ and q′.
Proof. Suppose w=(x,z),q=(x,z1), z1 differs from z in the first bit. The conclusion holds clearly for n=3 by Lemma 4.7. Assume n≥4 below. We complete the proof by induction on n. Now we suppose the result holds for n−1. Then 0≤dH(x,z)≤n. If 0≤dH(x,z)≤n−2, the result can be obtained by induction hypothesis easily. Thus dH(x,z)=n−1 or n. If dH(x,z)=n, then dH(x,z1) must be n−1, , the case is the same as dH(x,z)=n−1 and dH(x,z1)=n. Now we suppose dH(x,z)=n−1, then dH(x,z1)=n−2 or n, the case dH(x,z1)=n−2 holds by induction, thus dH(x,z1)=n. Without loss of generality, suppose x=0n,z=01n−1,z1=1n. w=(x,z)=(0n,01n−1),q=(x,z1)=(0n,1n), their external neighbors are w′=(01n−1,0n) and q′=(1n,00) respectively. We can find two vertex-disjoint paths joining w′ and q′, denoted by P1 and P2, P1:(01n−1,0n)−(01n−1,0n−11)−(0n−11,01n−1)−(0n−11,1n)−(1n,0n−11)−(1n,0n), P2:(01n−1,0n)−(01n−1,10n−1)−(10n−1,01n−1)−(10n−1,1n)−(1n,10n−1)−(1n,0n), see Fig.6.
Theorem 4.9. For n≥4, mp1(HCNn)=2n, and it is conditionally super matched.
Proof. By Theorem 4.1, we know that mp1(HCNn)=2n. Now we only need to prove that HCNn is conditionally super matched. Let |F|=2n, it is enough to show that one of the following holds: (1) HCNn−F has an isolated vertex; (2) F is a conditional matching preclusion set; (3) HCNn−F has a perfect matching. We use the same notation as in the proof of Theorem 4.1. As |F1|≥n, |Fc|≥1, then |Fj|≤n−1 for j∈{2,…,2n}. We claim that |Fc|≤n. If |Fc|≥n+1, then |Fi|≤n−1 for each i(1≤i≤2n), thus xiQn−Fi has a perfect matching, say Mi, the M=∪2ni=1Mi is a perfect matching of HCNn−F. We consider the following cases.
Case 1. 2≤|Fc|≤n, then n≤|F1|≤2n−2.
If xQn−F1 has an isolated vertex, say u. Let u′i be the external neighbors of ui,(i=1,2,⋯,n). And (u,u′)∉F, otherwise u is an isolated vertex in xQn−F. Let |Fc|=ℓ(ℓ≥2), then |F1|≤2n−ℓ, and |F1∩E(xQn−u)|≤n−ℓ. As xQn−u consists of n edge-disjoint almost perfect matchings, xQn−u has at least ℓ edge-disjoint almost perfect matchings. Suppose the ℓ vertices unmatched are u1,u2,⋯,uℓ. By corollary 2.5, there are ℓ paths P′1,P′2,…,P′ℓ. If there is P′i(1≤i≤ℓ) such that none of crossing edge of E(P′i) is in Fc, say P′1, we can also find a perfect matching of HCNn−F. If for each P′i(1≤i≤ℓ), one crossing edge of E(P′i) is in Fc. From Lemma 4.5, there is an almost perfect matching such that a vertex v with d(u,v)=3 is unmatched. From Lemma 4.6, there is a cycle L of even order and V(L) contains u,v. The perfect matching induced by L is denoted by ML, and xQn−(F1∪(V(L)∩V(xQn))) has a perfect matching, say Mx1, ∪2ni=2xQn−(V(L)−(V(L)∩V(xQn))) also has a perfect matching, say M′, then M=ML∪Mx1∪M′ is a perfect matching of HCNn−F.
If xQn−F1 has no isolated vertex, then F1 is a conditional super matching preclusion set of xQn−F1 with |F1|=2n−2 and |Fc|≤2. We may assume that it is induced by u1−u−u2. Let u,v1,⋯,vn−1 be the neighbors of u1 in xQn, and u,w1,⋯,wn−1 be the neighbors of u2 in xQn, vi′, wi′ be the external neighbor of vi and wi(i=1,2,…,n−1) respectively. There is at least one of {(u1,u′1),(u2,u′2)} not in Fc, otherwise, F is a conditional super matching preclusion, say (u1,u′1)∉F. By Corollary 2.5, there are n−1 internally vertex-disjoint 7-paths joining u′1 and {v1,…,vn−1}, say Q1,…,Qn−1. As |Fc|≥2, |Fc|=2. And n≥4, there is at least one path Qi(i=1,2,…,n−1) such that none of crossing edge of E(Qi) is in Fc, say Q1. By the proof above, we know that HCNn−(V(Q1)∪{u,u2}∪F) has a perfect matching, say M0, and Q1 induces a perfect matching, say M′. Then M0∪M′∪(u,u2) is a perfect matching of HCNn−F.
Case 2. |Fc|=1.
If |F1|≤2n−2, the perfect matching of HCNn−F can be obtained as the Case 1. Thus we only consider |F1|=2n−1, and xQn−F1 has at most two isolated vertices.
If xQn−F1 has two isolated vertices. We can claim that the two isolated vertices are adjacent. Otherwise |F1|≥2n for δ(Qn)=n, a contradiction. Without loss of generality, suppose such two isolated vertices are u and u1. As both u and u1 are not isolated vertices in HCNn−F, then (u,u′)∉Fc and (u1,u′1)∉Fc. From Lemma 4.8, there are 2 internally vertex-disjoint paths of even order joining endpoints u′ and u′1, say R1,R2. As |Fc|=1, there is at least one path Ri(i=1,2) such that none of crossing edges of E(Ri) is in Fc, say R1. And HCNn−(V(R1)∪{u′,u′1}∪F) has a perfect matching, say M0, V(R1)∪{u′,u′1} induces a perfect matching, say M1, then M=M0∪M1 is a perfect matching of HCNn−F.
If xQn−F1 has an isolated vertex. As there are n edge-disjoint almost perfect matchings of xQn−u, then there exists at least one almost perfect matching of xQn−F1, say M′, and u1 is unmatched. Moreover, (u,u′)∉Fc, otherwise u is an isolated vertex of HCNn−F. If (u1,u1′)∈Fc, as u1 is not an isolated vertex of xQn−F1, there exists a vertex adjacent to u1, say v1. Let v∈V(xQn), and (v1,v)∈M′, then d(u,v)=3. From Lemma 4.6, there is a cycle L of even order and V(L) contains u and v. The perfect matching of HCNn−(V(L)∪F) is denoted by M0, and V(L) induces a perfect matching, say M1, then M=M0∪M1 is a perfect matching of HCNn−F. If (u1,u1′)∉Fc, by the proof case 1, we can find a perfect matching of HCNn−F similarly.
If xQn−F1 has no isolated vertex. Suppose f=(u1,v1),f′=(u2,w1), and f,f′ are independent, by Lemma 4.2, xQn−(F1−f) and xQn−(F1−f′) contain a perfect matching. For |Fc|=1, there is at most one of {(u1,u′1),(v1,v′1),(w,w′),(w1,w1′)} in Fc. Without loss of generality, (u1,u′1)∉Fc, (v1,v1′)∉Fc. By the proof of case 1, we can find a perfect matching of HCNn−F.
Hierarchical cube network is a very important network constructed based on 2n hypercube networks and a perfect matching between them. The matching preclusion number of a graph is the minimum number of edges whose deletion results in a graph that has neither perfect matchings nor almost-perfect matchings, and the conditional matching preclusion number of a graph is the minimum number of edges whose deletion leaves a resulting graph with no isolated vertices that has neither perfect matchings nor almost perfect matchings. In this paper, we find these two numbers for the hierarchical cubic network(HCNn), characterize all its optimal matching preclusion sets and conditional matching exclusion sets, and prove that the hierarchical cube network has the property of super matching. These results generalize some related results of Birgham et al. [1] and E. Cheng et al. [4] from hypercube network to hierarchical cubic network.
This work was partially supported by the National Natural Science Foundation of China (Nos. 12161073) and the Qinghai Natural Science Foundation of China (Nos. 2020-ZJ-924).
The authors declare no conflict of interest.
[1] | R. C. Birgham, F. Harry, E. C. Biolin, J. Yellen, Perfect matching preclusion, Congr. Numer., 174 (2005), 185–192. |
[2] |
E. Cheng, L. Lipták, Matching preclusion for some interconnection networks, Networks, 50 (2007), 173–180. http://dx.doi.org/10.1002/net.20187 doi: 10.1002/net.20187
![]() |
[3] |
E. Cheng, L. Lensik. M. Lipman, L. Lipták, Matching preclusion for alternating group graphs and their generalizations, Int. J. Found. Comp. Sci., 19 (2008), 1413–1437. http://dx.doi.org/10.1142/S0129054108006364 doi: 10.1142/S0129054108006364
![]() |
[4] |
E. Cheng, L. Lensik. M. Lipman, L. Lipták, Conditional matching preclusion sets, Inf. Sci., 179 (2009), 1092–1101. http://dx.doi.org/10.1016/j.ins.2008.10.029 doi: 10.1016/j.ins.2008.10.029
![]() |
[5] |
E. Cheng, R. Jia, D. Lu, Matching preclusion and conditional matching preclusion for augmented cubes, J. of Interconnection Networks, 11 (2010), 35–60. http://dx.doi.org/10.1142/S0219265910002726 doi: 10.1142/S0219265910002726
![]() |
[6] | W. Chang, E. Cheng, Strong matching preclusion of 2-matching composition networks, Congr. Numer., 224 (2015), 91–104. |
[7] | H. Lü, X. Li, H. Zhang, NP-completeness of anti-Kekule and matching preclusion problem, J. Amer. Math. Soc., 2017 (2017), arXiv: 1706.09321. http://dx.doi.org/10.48550/arXiv.1706.09321 |
[8] |
R. Lin, H. Zhang, W. Zhao, Matching preclusion for direct product of regular graphs, Discrete Appl. Math., 277 (2020), 221–230. http://dx.doi.org/10.1016/j.dam.2019.08.016 doi: 10.1016/j.dam.2019.08.016
![]() |
[9] |
R. Lin, Conditional matching preclusion for regular bipartite graphs and their Cartesian product, Discrete Appl. Math., 299 (2021), 17–25. http://dx.doi.org/10.1016/j.dam.2021.04.011 doi: 10.1016/j.dam.2021.04.011
![]() |
[10] |
H. Lv, X. Li, H. Zhang, Matching preclusion for balanced hypercubes, Theor. Comput. Sci., 465 (2012), 10–20. http://dx.doi.org/10.1016/j.tcs.2012.09.020 doi: 10.1016/j.tcs.2012.09.020
![]() |
[11] |
E. Cheng, M. Lipman, L. Lipták, D. Sherman, Conditional matching preclusion for the arrangement graphs, Theor. Comput. Sci., 412 (2011), 6279–6289. http://dx.doi.org/10.1016/j.tcs.2011.07.007 doi: 10.1016/j.tcs.2011.07.007
![]() |
[12] |
S. Wang, R. Wang, S. Lin, J. Li, Matching preclusion for k-ary n-cubes, Discrete Appl. Math., 158 (2010), 174–182. http://dx.doi.org/10.1016/j.topol.2011.01.008 doi: 10.1016/j.topol.2011.01.008
![]() |
[13] |
Y. Mao, E. Cheng, A concise survey of matching preclusion in interconnection networks, J. Interconnect. Networks, 19 (2019), 1940006. http://dx.doi.org/10.1142/S0219265919400061 doi: 10.1142/S0219265919400061
![]() |
[14] |
K. Ghose, K. R. Desai, Hierarchical cubic network, IEEE Trans. Parallel Distrib Syst., 6 (1995), 427–435. http://dx.doi.org/10.1109/71.372797 doi: 10.1109/71.372797
![]() |
[15] |
S. Zhou, S. Song, X. Yang, L. Chen, On the conditional fault tolerance and diagnosability of hierarchical cubic networks, Theor. Comput. Sci., 609 (2016), 421–433. http://dx.doi.org/10.1016/j.tcs.2015.10.030 doi: 10.1016/j.tcs.2015.10.030
![]() |