
The generalized ℓ-connectivity κℓ(G) of a graph G is a generalization of classical connectivity κ(G) with κ2(G)=κ(G). It serves to measure the capability of connection for any ℓ vertices. The folded Petersen cube network FPQn,k can be used to model the topological structure of a communication-efficient multiprocessor. This paper shows that the generalized 4-connectivity of the folded Petersen cube network FPQn,k is n+3k−1. As a corollary, the generalized 3-connectivity of FPQn,k also is obtained and the results on the generalized 4-connectivity of hypercube Qn and folded Petersen graph FPk can be verified. These conclusions provide a foundation for studying the generalized 4-connectivity of Cartesian product graphs.
Citation: Huifen Ge, Shumin Zhang, Chengfu Ye, Rongxia Hao. The generalized 4-connectivity of folded Petersen cube networks[J]. AIMS Mathematics, 2022, 7(8): 14718-14737. doi: 10.3934/math.2022809
[1] | Ali Raza, Mobeen Munir, Tasawar Abbas, Sayed M Eldin, Ilyas Khan . Spectrum of prism graph and relation with network related quantities. AIMS Mathematics, 2023, 8(2): 2634-2647. doi: 10.3934/math.2023137 |
[2] | Huifeng Zhang, Xirong Xu, Ziming Wang, Qiang Zhang, Yuansheng Yang . (2n−3)-fault-tolerant Hamiltonian connectivity of augmented cubes AQn. AIMS Mathematics, 2021, 6(4): 3486-3511. doi: 10.3934/math.2021208 |
[3] | Shabbar Naqvi, Muhammad Salman, Muhammad Ehtisham, Muhammad Fazil, Masood Ur Rehman . On the neighbor-distinguishing in generalized Petersen graphs. AIMS Mathematics, 2021, 6(12): 13734-13745. doi: 10.3934/math.2021797 |
[4] | Kai An Sim, Kok Bin Wong . On the cooling number of the generalized Petersen graphs. AIMS Mathematics, 2024, 9(12): 36351-36370. doi: 10.3934/math.20241724 |
[5] | Ganesh Gandal, R Mary Jeya Jothi, Narayan Phadatare . On very strongly perfect Cartesian product graphs. AIMS Mathematics, 2022, 7(2): 2634-2645. doi: 10.3934/math.2022148 |
[6] | Yanyi Li, Lily Chen . Injective edge coloring of generalized Petersen graphs. AIMS Mathematics, 2021, 6(8): 7929-7943. doi: 10.3934/math.2021460 |
[7] | Najmeddine Attia . Advances on fractal measures of Cartesian product sets in Euclidean space. AIMS Mathematics, 2025, 10(3): 5971-6001. doi: 10.3934/math.2025273 |
[8] | Bo Zhu, Shumin Zhang, Jinyu Zou, Chengfu Ye . Structure connectivity and substructure connectivity of data center network. AIMS Mathematics, 2023, 8(4): 9877-9889. doi: 10.3934/math.2023499 |
[9] | Li Zhu . On pairs of equations with unequal powers of primes and powers of 2. AIMS Mathematics, 2025, 10(2): 4153-4172. doi: 10.3934/math.2025193 |
[10] | Anas Al-Masarwah, Mohammed Alqahtani . Operational algebraic properties and subsemigroups of semigroups in view of k-folded N-structures. AIMS Mathematics, 2023, 8(9): 22081-22096. doi: 10.3934/math.20231125 |
The generalized ℓ-connectivity κℓ(G) of a graph G is a generalization of classical connectivity κ(G) with κ2(G)=κ(G). It serves to measure the capability of connection for any ℓ vertices. The folded Petersen cube network FPQn,k can be used to model the topological structure of a communication-efficient multiprocessor. This paper shows that the generalized 4-connectivity of the folded Petersen cube network FPQn,k is n+3k−1. As a corollary, the generalized 3-connectivity of FPQn,k also is obtained and the results on the generalized 4-connectivity of hypercube Qn and folded Petersen graph FPk can be verified. These conclusions provide a foundation for studying the generalized 4-connectivity of Cartesian product graphs.
As usual, the topological structure of an interconnection network is regarded as a graph G=(V,E), in which vertices correspond to processors and edges represent communication links between processors. The fault tolerance is one of the most important factors in the design and analysis of an interconnection network and it can be measured by the connectivity of a graph. If the connectivity of a network is larger, then its fault tolerance is higher. The traditional connectivity κ(G) of a graph G is defined as the minimum number of vertices whose deletion results in a disconnected graph. An excellent theorem of Whitney [32] provided an equivalent statement about the definition of the connectivity. That is, for any 2-subset S={u,v}⊆V(G), if κ(S) denotes the maximum number of internally disjoint paths between u and v in G, then κ(G)=min{κ(S):S⊆V(G),|S|=2}. Clearly, κ(G) reflects the connectivity between any two processors. To measure the connectivity capability of more processors, Chartrand et al. [6] and Hager et al. [11] introduced independently the concept of the generalized connectivity of a graph by generalizing the equivalent definition of connectivity.
Let G be a connected graph with order n and ℓ be an integer such that 2≤ℓ≤n. For S⊆V(G), a tree T in G is called an S-tree if S⊆V(T). Let κ(S) denote the maximum number r of edge-disjoint S-trees T1,…,Tr satisfying V(Ti)∩V(Tj)=S for any two distinct integers i,j∈{1,…,r}. The generalized ℓ-connectivity κℓ(G) of G is defined as min{κ(S):S⊆V(G),|S|=ℓ}. Actually, κ2(G) is exactly κ(G).
Though there are numerous results about the generalized ℓ-connectivity over the past years, for general integer ℓ, the exact values of κℓ(G) are known for only a small class of graphs: complete graph [7], complete bipartite graph [16] and complete equipartition 3-partite graph [18]. Meanwhile, for a given graph G, any fixed integer k≥2 and a subset S of V(G), the decision problem whether κ(S)≥k is NP-complete [19]. The upper and lower bounds of the generalized 3-connectivity of a graph [21,25] and of Cartesian (Lexicographic) product of two graphs [13,14,26] were investigated, and extremal problems were studied in [17,22]. The generalized 3-connectivity of some graph classes are known, including Cayley graphs [20,31], star graphs and bubble-sort graphs [23], alternating group graphs and (n,k)-star graphs [34], k-ary n-cubes, split-star graphs and bubble-sort star graphs [37], (n,k)-bubble sort graphs [38], etc. We refer the readers to [15,27] for more details.
Unfortunately, the results of the generalized 4-connectivity are less known. Only hypercubes [24], dual cubes [35], exchanged hypercubes [33], (n,k)-star graphs [12], hierarchical cubic networks [36] have been studied.
The main focus of this paper is to determine the generalized 4-connectivity of the folded Petersen cube networks FPQn,k. The following result is obtained.
Theorem 1.1. Let k,n be two integers. Then κ4(FPQn,k)=n+3k−1.
Theorem 1.1 implies that if k=0, then κ4(FPQn,0)=n−1, which coincides the value of κ4(Qn). The key to prove Theorem 1.1 is Theorem 1.2.
Theorem 1.2. Let FPk be a k-dimensional folded Petersen graph. Then κ4(FPk)=3k−1.
For a regular graph, the following lemma is useful.
Lemma 1.1. [24] Let G be an r-regular graph. If κk(G)=r−1, then κk−1(G)=r−1, where k≥4.
Combining Theorem 1.1 and Lemma 1.1, the following corollary is an immediate consequence.
Corollary 1.1. Let k,n be two integers. Then κ3(FPQn,k)=n+3k−1.
For helping readers to understand the proof process, a flow chart in Figure 1 is presented to illustrate the relationship between different lemmas, theorems and corollaries.
This section introduces some basic notations and results that will be used throughout the paper. All graphs considered in this paper are connected, simple, undirected, finite and for the notation and terminology not defined refer to [5].
Let G=(V,E) be a graph with vertex set V(G) and edge set E(G). For a vertex v of G, we use NG(v) to denote the set of neighbors of v in G, and the degree of v is dG(v)=|NG(v)|. The minimum degree and the maximum degree of G are denoted by δ(G) and Δ(G), respectively. A graph G is r-regular if δ(G)=Δ(G)=r. For two vertices u,v∈V(G), a (u,v)-path is denoted by Puv and the length of a shortest (u,v)-path is called the distance between u and v, denoted by dG(u,v). A subgraph of G is a graph H=(V′,E′) such that V′(H)⊆V(G) and E′(H)⊆E(G). If V′(H)=V(G), then H is called a spanning subgraph of G. The subgraph of G induced by V′ is denoted by G[V′]. Let [b]={1,…,b} for a given integer b.
Lemma 2.1. [21] If there are two adjacent vertices of degree δ, then κℓ(G)≤δ−1 for 3≤ℓ≤|V(G)|.
Theorem 2.1. (Menger's theorem [5]) A graph G is r-connected if and only if any two distinct vertices of G are connected by at least r internally disjoint paths.
Lemma 2.2. (Fan Lemma [5]) Let G be an r-connected graph, x be a vertex of G, and let Y⊆V(G)∖{x} be a set of at least r vertices of G. Then there exists an r-fan in G from x to Y, that is, there exists a family of r internally disjoint (x,Y)-paths whose terminal vertices are distinct in Y.
The Cartesian product of graphs is an important tool to construct a bigger network. Recall that the Cartesian product of two graphs G and H, denoted by G◻H, is a graph with the vertex set V(G)×V(H) such that (g,h) and (g′,h′) are adjacent if and only if either g=g′ and hh′∈E(H), or h=h′ and gg′∈E(G).
The Petersen graph P with a vertex set {0,1,2,…,9} has an outer 5-cycle and an inner 5-cycle are joined by a perfect matching (Figure 2(a) depicts P with decimal vertex-labeling). It is a 3-regular 3-connected graph with diameter 2. The k-dimensional folded Petersen graph FPk is constructed by the k-th iteration of Cartesian product on Petersen graph P, defined as FPk=P◻⋯◻P=(V,E), where V={x1⋯xk:xi∈{0,1,...,9},1≤i≤k} and E={(x1⋯xi−1xixi+1⋯xk,x1⋯xi−1yxi+1⋯xk):y∈{0,1,…,9}, dP(y,xi)=1 and 1≤i≤k}. As depicted in Figure 1(b), FP2 is obtained from P by replacing each of its vertices by P, denoted by P0,P1,…,P9, respectively, moreover, between Pi and Pj have a perfect matching for dP(i,j)=1. In fact, for k≥2, FPk is obtained from P by replacing each of its vertices by FPk−1, and we denote the 10 connected subgraphs by FPi:0k,FPi:1k,…,FPi:9k, respectively, where FPi:jk=FPk[{x1⋯xk∈V(FPk),xi=j}]. Clearly, FPi:jk≅FPk−1. For convenience, FPj denotes FP1:jk for j∈{0,1,…,9} in the rest of this paper. It is easy to see that the subgraph induced by V(FPi)∪V(FPj) is isomorphic to FPk−1◻K2 for dP(i,j)=1, and the edges between V(FPi) and V(FPj) are called the crossed edges.
For any x∈V(FPi), we call that xij is a corresponding vertex of x in FPij if x and xij differ in exactly the first digit. In particular, a corresponding vertex xij of x is an outside neighbor of x if dP(i,ij)=1. Apparently, there are exactly three outside neighbors of x. Two graphs G′ and G″ are corresponding if their vertices correspond one to one and G′≅G″. Furthermore, FPk is a 3k-regular 3k-connected graph with diameter 2k and order 10k, also vertex and edge symmetric. More details see [8,10].
The folded Petersen cube network FPQn,k=FPk◻Qn with 10k×2n vertices is introduced by Öhring and Das [28], where Qn is an n-dimensional hypercube and it can be regarded as n-th iteration of Cartesian product on K2. In particular, FPQ0,k=FPk and FPQn,0=Qn. There are a lot of topological structures like linear arrays, rings, meshes, hypercubes can be embedded into it. Some research findings on the folded Petersen cube networks have been published for the past several years, see [9,28,30].
The following result will be used to prove Theorem 1.2.
Lemma 2.3. Let S be a vertex subset of the Petersen graph P with |S|=4. Then there exist two internally disjoint S-trees in P.
Proof. Let P be a Petersen graph, S={x,y,z,w}⊆V(P) and G=P[S]. Then dG(v)≤3 for arbitrary v∈V(G). Since P is vertex symmetric, assume that dG(x)=Δ(G). If dG(x)=3, then NG(x)={y,z,w}. Two internally disjoint S-trees are demonstrated in Figure 3(a). If dG(x)=2, without loss of generality, let NG(x)={y,z}. It suffices to consider two cases and the required S-trees are demonstrated in Figure 3(b) and Figure 3(c), respectively. If dG(x)=1, assume that NG(x)={y}, it suffices to consider two cases and the required S-trees are demonstrated in Figure 3(d) and Figure 3(e), respectively. If dG(x)=0, Figure 2(f) demonstrates two internally disjoint S-trees.
In this section, we mainly determine κ4(FPk). First, propose the general idea of the proof κ4(FPk)≥3k−1. From the definition of κ4(FPk), it suffices to show that κFPk(S)≥3k−1 for arbitrary S⊆V(FPk) with |S|=4. Furthermore, according to the definition of κFPk(S), find out 3k−1 internally disjoint S-trees in FPk. Let S be a set of arbitrary four distinct vertices in FPk. Recall that FPk can be decomposed into 10 disjoint sub-folded Petersen graphs FP0,FP1,…,FP9, each of which is isomorphic to FPk−1 by removing all crossed edges.
For arbitrary four vertices v0,v1,v2,v3 in P, by Lemma 2.3, there are two internally disjoint {v0,v1, v2, v3}-trees T∗1 and T∗2 in P. At least one of the trees T∗1 and T∗2 contains a vertex u different from v0,v1,v2,v3. Without loss of generality, let T∗2 be such a tree. Clearly, (T∗1∪T∗2)◻FPk−1 is a subgraph of FPk. If we can prove that κT∗1◻FPk−1(S)≥3k−2 and find out an S-tree in T∗2◻FPk−1, then κFPk(S)≥κT∗1◻FPk−1(S)+1≥3k−1. In this case, the problem is converted into finding out 3k−2 internally disjoint S-trees in T∗1◻FPk−1 and an S-tree in T∗2◻FPk−1 such that all of the S-trees are internally disjoint.
Observation 3.1. For arbitrary three vertices v0,v1,v2 in P, one of the following holds.
(i) If {v0,v1,v2} is an isolated set, then there are three internally disjoint {v0,v1,v2}-trees T∗1,T∗2,T∗3 in P. Moreover, there exists a vertex ui with degree 3 different from v0,v1,v2 in T∗i for i=1,2,3.
(ii) If {v0,v1,v2} is not an isolated set, then there are a C5 and a {v0,v1,v2}-tree T∗3 in P such that v0,v1,v2∈V(C5) and V(C5)∩V(T∗3)={v0,v1,v2}. Moreover, there exists a vertex u with degree 3 different from v0,v1,v2 in T∗3.
For Observation 3.1(i), it is easy to see that (⋃3i=1T∗i)◻FPk−1 is a subgraph of FPk (see Figure 4). Notice that T∗1,T∗2 and T∗3 are internally disjoint in P, if we can prove that κT∗1◻FPk−1(S)≥3k−3, find out an S-tree in T∗2◻FPk−1 and another S-tree in T∗3◻FPk−1, then κFPk(S)≥κT∗1◻FPk−1(S)+2≥3k−1. In this case, the problem is converted into finding out 3k−3 internally disjoint S-trees in T∗1◻FPk−1, an S-tree in T∗2◻FPk−1 and another S-tree in T∗3◻FPk−1 such that all of the S-trees are internally disjoint.
For Observation 3.1(ii), let C=C5, it can be seen that (C∪T∗3)◻FPk−1 is a subgraph of FPk (see Figure 5). If we can prove that κC◻FPk−1(S)≥3k−2, and find out an S-tree in T∗3◻FPk−1, then κFPk(S)≥κC◻FPk−1(S)+1≥3k−1. In this case, the problem is converted into finding out 3k−2 internally disjoint S-trees in C◻FPk−1 and an S-tree in T∗3◻FPk−1 such that all of the S-trees are internally disjoint.
For arbitrary two vertices v0,v1 in P, by Theorem 2.1, there are three internally disjoint (v0,v1)-paths L∗1, L∗2 and L∗3 in P because the Petersen graph P is 3-connected. Note that ⋃3i=1L∗i is a subgraph of P and (⋃3i=1L∗i)◻FPk−1 is a subgraph of FPk (see Figure 6). If we can prove that κL∗1◻FPk−1(S)≥3k−3, find out an S-tree in L∗2◻FPk−1 and another S-tree in L∗3◻FPk−1, then κFPk(S)≥κ(∪3i=1L∗i)◻FPk−1(S)≥κL∗1◻FPk−1(S)+2≥3k−1. In this case, the problem is converted into finding out 3k−3 internally disjoint S-trees in L∗1◻FPk−1, an S-tree in L∗2◻FPk−1 and another S-tree in L∗3◻FPk−1 such that all of the S-trees are internally disjoint.
The following four lemmas will be helpful to main result.
Lemma 3.1. Let FPk be a k-dimensional folded Petersen graph and S={x,y,z,w} be a set of arbitrary four distinct vertices in FPk for k≥2. If the vertices in S belong to four sub-folded Petersen graphs of FPk, then there are 3k−1 internally disjoint S-trees in FPk.
Proof. Let FP0,FP1,…,FP9 be 10 disjoint sub-folded Petersen graphs of FPk. Since the vertices in S belong to four sub-folded Petersen graphs of FPk, there exist v0,v1,v2,v3∈{0,1,…,9} such that they are mutual distinct and x∈V(FPv0), y∈V(FPv1), z∈V(FPv2) and w∈V(FPv3).
For v0,v1,v2,v3∈V(P), there exist two internally disjoint {v0,v1,v2,v3}-trees T∗1 and T∗2 in P by Lemma 2.3. At least one of the trees T∗1 and T∗2 contains a vertex c different from v0,v1,v2,v3, say T∗2 is such a tree. Since FPc is connected, there exists an {xc,yc,zc,wc}-tree T′3k−2 in FPc. There is a unique path to connect arbitrary two vertices in T∗2. That is, we can find the corresponding path to connect arbitrary two vertices in T∗2◻FPk−1. Let T3k−2=T′3k−2∪Pxxc∪Pyyc∪Pzzc∪Pwwc.
Let x0=x. Choose 3k−2 distinct vertices x0,x1,x2,…,x3k−3 from FPv0 being X such that yv0,zv0 and wv0 belong to X and |X|=3k−2. Without loss of generality, assume that xr=yv0,xs=zv0,xt=wv0 for r,s,t∈{0,1,…,3k−3}. Let Y=⋃3k−3i=0yi, Z=⋃3k−3i=0zi and W=⋃3k−3i=0wi be the corresponding vertices of vertices of X in FPv1,FPv2 and FPv3, respectively, where yi=xv1i, zi=xv2i and wi=xv3i for each i∈{0,1,…,3k−3}. Then |Y|=|Z|=|W|=3k−2. By Lemma 2.2 and κ(FPk−1)=3k−3, there are 3k−3 internally disjoint paths A1,A2,…,A3k−3 from a to A∖{a} in FPvj for (a,A,j)=(x,X,0),(y,Y,1),(z,Z,2),(w,W,3), respectively. Since there is a unique path to connect arbitrary two vertices in T∗1, we can find the corresponding path to connect corresponding vertices in T∗1◻FPk−1. Construct 3k−2 internally disjoint S-trees in T∗1◻FPk−1 as follows: Ti=Xi∪Pxiyi∪Yi∪Pyizi∪Zi∪Pziwi∪Wi for i∈{0,1,…,3k−3}, where X0={x0}={x}, Yr={yr}={y}, Zs={zs}={z} and Wt={wt}={w}. Then, T0,T1,…,T3k−2 are 3k−1 internally disjoint S-trees in FPk. See Figure 7.
Lemma 3.2. Let FPk be a k-dimensional folded Petersen graph and S={x,y,z,w} be a set of arbitrary four distinct vertices in FPk for k≥2. If the vertices in S belong to three sub-folded Petersen graphs of FPk, then there are 3k−1 internally disjoint S-trees in FPk.
Proof. Let FP0,FP1,…,FP9 be 10 disjoint sub-folded Petersen graphs of FPk. Since the vertices in S belong to three sub-folded Petersen graphs of FPk, there exist v0,v1,v2∈{0,1,…,9} such that x,y∈V(FPv0), z∈V(FPv1) and w∈V(FPv2) by the symmetry of FPk.
Since FPk−1 is (3k−3)-connected, there exist 3k−3 internally disjoint (x,y)-paths P1,P2,…,P3k−3 in FPv0. Choose xi∈V(Pi) such that xi≠x. Let zi=xv1i and wi=xv2i for i∈[3k−3]. By Lemma 2.2, there are 3k−3 paths Pzz1,…,Pzz3k−3 and 3k−3 paths Pww1,…,Pww3k−3 in FPv1 and FPv2, respectively. It is possible that one of the paths Pzzi (resp. Pwwi) is a single vertex.
Case 1. {v0,v1,v2} is an isolated set.
By Observation 3.1(i), there exists a {v0,v1,v2}-tree T∗1 which contains a vertex a with degree 3 different from v0,v1,v2 in P. Moreover, there are a unique (v0,a)-path Pv0a, a unique (v1,a)-path Pv1a and a unique (v2,a)-path Pv2a in T∗1. It is not hard to find 3k−3 internally disjoint S-trees in T∗1◻FPk−1, that is, Ti=Pi∪Pxixai∪Pzzi∪Pzixai∪Pwwi∪Pwixai for i∈[3k−3], where Pxixai≅Pv0a, Pzixai≅Pv1a and Pwixai≅Pv2a.
Except T∗1, there are two internally disjoint {v0,v1,v2}-trees T∗2 and T∗3 in P, T∗2 and T∗3 contains a vertex with degree 3 different from v0,v1 and v2, denoted by b and c, respectively. Since FPb is connected, there exists an {xb,yb,zb,wb}-tree T′3k−2 in FPb (resp. {xc,yc,zc,wc}-tree T′3k−1 in FPc). By the definition of FPk, there exist the paths Pxxb,Pyyb,Pzzb,Pwwb (resp. Pxxc,Pyyc,Pzzc,Pwwc) in FPk such that Pxxb≅Pyyb≅Pv0b,Pzzb≅Pv1b,Pwwb≅Pv2b (resp. Pxxc≅Pyyc≅Pv0c,Pzzc≅Pv1c,Pwwc≅Pv2c) where Pv0b,Pv1b,Pv2b belong to T∗2 (resp. Pv0c,Pv1c,Pv2c belong to T∗3). Let T3k−2=T′3k−2∪Pxxb∪Pyyb∪Pzzb∪Pwwb (resp. T3k−1=T′3k−1∪Pxxc∪Pyyc∪Pzzc∪Pwwc). Then, T1,T2,…,T3k−1 are 3k−1 internally disjoint S-trees in FPk. See Figure 8.
Case 2. {v0,v1,v2} is not an isolated set.
By Observation 3.1(ii), there exists a cycle C for C=C5 and a {v0,v1,v2}-tree T∗3 in P such that V(C)∩V(T∗3)={v0,v1,v2}. Let R be a shorter path containing vertices v0,v1,v2 in C, say (v0,v2)-path being such a path. Then R=Pv0v1∪Pv1v2. Let Ti=Pi∪Pxizi∪Pzzi∪Pziwi∪Pwwi for i∈[3k−3], where Pxizi≅Pv0v1 and Pziwi≅Pv1v2.
Clearly, R is a {v0,v1,v2}-tree with |V(R)|≤4 and there exists another (v0,v2)-path R′ containing a non-end vertex d, then R′=Pv0d∪Pv2d. Since FPd is connected, there exists an {xd,yd,zd,wd}-tree T′3k−2 in FPd. If zv0∉⋃3k−3i=1V(Pi) (see Figure 9(a)), then zv0∈Pzzd, where Pzzd≅Pv0v1∪Pv0d. Let T3k−2=T′3k−2∪Pxxd∪Pyyd∪Pzzd∪Pwwd, where Pxxd≅Pyyd≅Pv0d and Pwwd≅Pv2d. If zv0∈⋃3k−3i=1V(Pi) (see Figure 9(b)), without loss of generality, assume that zv0∈P1 and x1=zv0, then the path Pzz1 is a single vertex. Let z3k−2=yv1. Then there exists a path Pzz3k−2 in FPv1. Let T3k−2=T′3k−2∪Pxxd∪Pyyd∪Pyz3k−2∪Pzz3k−2∪Pwwd, where Pxxd≅Pyyd≅Pv0d, Pyz3k−2≅Pv0v1 and Pwwd≅Pv2d. The S-tree T3k−1 can be similarly constructed as T3k−1 of Case 1. Then, T1,T2,…,T3k−1 are 3k−1 internally disjoint S-trees in FPk.
Lemma 3.3. Let FPk be a k-dimensional folded Petersen graph and S={x,y,z,w} be a set of arbitrary four distinct vertices in FPk for k≥2. If the vertices in S belong to the same sub-folded Petersen graphs of FPk, then there are 3k−1 internally disjoint S-trees in FPk.
Proof. Let FP0,FP1,…,FP9 be 10 disjoint sub-folded Petersen graphs of FPk. The vertices in S belong to the same sub-folded Petersen graph, without loss of generality, suppose that x,y,z,w∈V(FP0). Since FP0≅FPk−1, by induction hypothesis, we have κ4(FPk−1)≥3k−4. Hence, there exist 3k−4 internally disjoint S-trees T1,T2,…,T3k−4 in FP0. Clearly, there exists an {x1,y1,z1,w1}-tree T′3k−3 in FP1, an {x4,y4,z4,w4}-tree T′3k−2 in FP4 and an {x5,y5,z5,w5}-tree T′3k−1 in FP5. Let T3k−3=T′3k−3∪xx1∪yy1∪zz1∪ww1, T3k−2=T′3k−2∪xx4∪yy4∪zz4∪ww4 and T3k−1=T′3k−1∪xx5∪yy5∪zz5∪ww5. Then, T1,T2,…,T3k−1 are 3k−1 internally disjoint S-trees in FPk.
Lemma 3.4. Let FPk be a k-dimensional folded Petersen graph and S={x,y,z,w} be a set of arbitrary four distinct vertices in FPk for k≥2. If the vertices in S belong to two sub-folded Petersen graphs of FPk, then there are 3k−1 internally disjoint S-trees in FPk.
Proof. Let FP0,FP1,…,FP9 be 10 disjoint sub-folded Petersen graphs of FPk. Suppose that the vertices in S belong to two distinct sub-folded Petersen graphs FPv0 and FPv1, where v0,v1∈{0,1,…,9}. For v0,v1∈V(P), since the Petersen graph P is 3-connected, by Theorem 2.1, there are three internally disjoint (v0,v1)-paths L∗1, L∗2 and L∗3 in P. For convenience, let |V(L∗1)|≤|V(L∗2)|≤|V(L∗3)|. It implies that 2≤|V(L∗1)|≤3. We just consider |V(L∗1)|=2 as the discussion for |V(L∗1)|=3 is similar. Without loss of generality, suppose that v0=0 and v1=1. By the symmetry of FPk, the following cases be considered.
Case 1. x,y,z∈V(FP0) and w∈V(FP1).
Case 1.1. w0∈{x,y,z}.
Without loss of generality, suppose that w0=z. By induction hypothesis and Lemma 1.1, we can find 3k−4 internally disjoint {x,y,z}-trees T′1,T′2,…,T′3k−4 in FP0. Let T′=⋃3k−4i=1T′i.
Case 1.1.1. |NT′(z)∩{x,y}|≤1.
First, construct two internally disjoint S-trees T3k−2 and T3k−1 in L∗2◻FPk−1 and L∗3◻FPk−1, respectively. Since FP4 and FP5 are connected, there exist an {x4,y4,z4}-tree T′3k−2 in FP4 and an {x5,y5,z5}-tree T′3k−1 in FP5, respectively. By the definition of FPk, there exist two paths Pwz4 and Pwz5 such that Pwz4≅L∗2∖{0} and Pwz5≅L∗3∖{0}. Let T3k−2=T′3k−2∪xx4∪yy4∪zz4∪Pwz4 and T3k−1=T′3k−1∪xx5∪yy5∪zz5∪Pwz5. Then T3k−2 and T3k−1 are internally disjoint S-trees in FPk.
Next, construct 3k−3 internally disjoint S-trees T1,T2,…,T3k−3 in L∗1◻FPk−1 such that T1,T2,…,T3k−1 are 3k−1 internally disjoint S-trees in FPk.
Since |NT′(z)∩{x,y}|≤1, without loss of generality, assume that |NT′i(z)∩{x,y}|=0 and dT′i(z)=1 for i=1,2,…,3k−6. Let Ti=T′i∪ziwi∪wiw, where i∈[3k−6], zi is the neighbor of z in T′i and wi=z1i. Note that if |NT′(z)∩{x,y}|=1, say y∈NT′(z). By symmetry, consider the following two cases.
Case 1.1.1.1. dT′3k−5(z)=dT′3k−4(z)=1.
If y∉NT′(z), let Ti=T′i∪ziwi∪wiw for i=3k−5 and 3k−4, where zi is the neighbor of z in T′i and wi=z1i. Otherwise, assume that y∈NT′3k−5(z). Let T3k−5=T′3k−5∪zw and T3k−4=T′3k−4∪z3k−4w3k−4∪w3k−4w. It is clear that |NT′(z)∩V(FP0)|≤3k−4, so there is a neighbor z3k−3 of z in FP0. Let T=⋃3k−4i=1Ti and W=NT(w). Then |W∩V(FP1)|≤3k−4. Since FP1 is (3k−3)-connected, FP1−W is still connected, thus we can find an {x1,y1,w,w3k−3}-tree T′3k−3 in FP1−W, where w3k−3=z13k−3. Let T3k−3=T′3k−3∪xx1∪yy1∪zz3k−3∪z3k−3w3k−3.
Case 1.1.1.2. dT′3k−5(z)=1 and dT′3k−4(z)=2.
If y∉NT′(z), let Ti=T′i∪ziwi∪wiw for i=3k−5 and 3k−4, where zi is one of the neighbors of z in T′i and wi=z1i. Let T=⋃3k−4i=1Ti and W=NT(w). Then |W∩V(FP1)|=3k−4. Since FP1 is (3k−3)-connected, FP1−W is still connected, thus there exists an {x1,y1,w}-tree T′3k−3 in FP1−W. Let T3k−3=T′3k−3∪xx1∪yy1∪zw.
Suppose that y∈NT′(z). Without loss of generality, assume that y∈NT′3k−5(z). Clearly, T′3k−5 is an (x,z)-path containing y and T′3k−4 is an (x,y)-path containing z. That is, T′3k−5=Pxy∪yz and T′3k−4=Pxz∪Pzy, where the lengths of Pxz and Pzy are least 2, see Figure 10(a). Let T″3k−5=Pxy∪Pzy and T″3k−4=Pxz∪yz, see Figure 10(b). Actually, T′3k−5 and T′3k−4 in the case y∈NT′3k−4(z) are T″3k−5 and T″3k−4, respectively. Let Ti=T″i∪ziwi∪wiw for i=3k−5,3k−4, where zi≠y and zi is the neighbor of z in T″i. Let T=⋃3k−4i=1Ti and W=NT(w), then |W∩V(FP1)|=3k−4 and FP1−W is still connected, thus there exists an {x1,y1,w}-tree T′3k−3 in FP1−W. Let T3k−3=T′3k−3∪xx1∪yy1∪zw, see Figure 10(c).
Then, T1,T2,…,T3k−1 are 3k−1 internally disjoint S-trees in FPk.
Case 1.1.2. |NT′(z)∩{x,y}|=2.
Recall the decimal vertex-labeling of the FPk. Without loss of generality, assume that w0=z=000…00. Then w=100…00. Notice that x,y,z are either in a subgraph of FPk isomorphic to C5 or C4 when k≥2.
Case 1.1.2.1. x,y,z are in a subgraph of FPk isomorphic to C5.
By symmetry, assume that x=000…01, y=000…04. If k≥3, then x,y,z,w∈FPj:0k for j∈{2,…,k−1} when dividing FPk along the jth dimension. Thus, the desired trees can be found similarly to Lemma 3.3. Suppose that k=2. Then x=01, y=04, z=00 and w=10. The desired trees can be found similarly to Lemma 3.2 when dividing FPk along the 2th dimension.
Case 1.1.2.2. x,y,z are in a subgraph of FPk isomorphic to C4.
In this case, k≥3. Without loss of generality, assume that x=000…01 and y=000…10. If k≥4, then x,y,z,w∈FPj:0 for j∈{2,…,k−2} when dividing FPk along the jth dimension. Thus, we can find out desired trees similarly to Lemma 3.3. Suppose that k=3. Then x=001, y=010, z=000 and w=110. The desired trees are shown in Figure. 11.
Case 1.2. w0∉{x,y,z}.
Case 1.2.1. |NFP0(w0)∩{x,y,z}|≤1.
First, construct 3k−3 internally disjoint S-trees T1,T2,…,T3k−3 in L∗1◻FPk−1 such that T1,T2,…,T3k−1 are 3k−1 internally disjoint S-trees in FPk.
Let S′={x,y,z,w0}. Since FP0≅FPk−1, by induction hypothesis, there exist 3k−4 internally disjoint S′-trees T′1,T′2,…,T′3k−4 in FP0. Since |NFP0(w0)∩{x,y,z}|≤1, without loss of generality, assume that |NT′i(w0)∩{x,y,z}|=0 for i∈[3k−4] and i≠1. Let T1=T′1∪ww0. Since dFP0(w0)=3k−3, we have 1≤dT′i(w0)≤2 for i∈[3k−4]. For i=2,…,3k−4, let Ti=(T′i∖w0)∪wiw1i∪w1iw if NT′i(w0)={wi} and let Ti=(T′i∖w0)∪wiw1i∪w1iw∪wjw1j∪w1jw if NT′i(w0)={wi,wj}.
Let T=⋃3k−4i=1Ti and W=NT(w). Then |W∩V(FP1)|=3k−4. Since FP1 is (3k−3)-connected, FP1−W is still connected, thus there exists an {x1,y1,z1,w}-tree T′3k−3 in FP1−W. Let T3k−3=T′3k−3∪xx1∪yy1∪zz1.
Next, construct two internally disjoint S-trees T3k−2 and T3k−1 in L∗2◻FPk−1 and L∗3◻FPk−1, respectively. Since P is a simple graph, there exist two non-end vertices 4 and 5 in L∗2 and L∗3, respectively. Clearly, FP4 is connected and there exists an {x4,y4,z4,w4}-tree T′3k−2 in FP4. Let T3k−2=T′3k−2∪xx4∪yy4∪zz4∪Pww4, where Pww4≅L∗2∖{0}. Similarly, construct T3k−1=T′3k−1∪xx5∪yy5∪zz5∪Pww5. Then, T1,T2,…,T3k−1 are 3k−1 internally disjoint S-trees in FPk.
Case 1.2.2. |NFP0(w0)∩{x,y,z}|≥2.
Without loss of generality, suppose that {x,y}⊆NFP0(w0) and w0=000⋯00. Clearly, w=100⋯00. Note that x,y,w0 are either in a subgraph of FPk isomorphic to C5 or C4 when k≥2.
Case 1.2.2.1. x,y,w0 are in a subgraph of FPk isomorphic to C5.
Without loss of generality, assume that x=000⋯01 and y=000⋯04. Let z=0z2z3⋯zk−1zk. If there exists j∈{2,…,k−1} such that zj=0, then x,y,z,w∈FPj:0k when dividing FPk along the jth dimension. Thus, we can find out desired trees similarly to Lemma 3.3. It suffices to consider the case zj≠0 for j=2,…,k−1.
If k≥3, then x,y,w∈V(FP2:0k), z∈V(FP2:z2k), z0=00z3⋯zk−1zk∉{x,y,w} and |N(z0)∩{x,y,w}|≤1 when dividing FPk along the 2th dimension. Thus, we can find out desired trees Similarly to case 1.2.1.
Suppose that k=2, there are x=01, y=04, w=10, z=0z2 and z2∉{0,1,4}. We can find out desired trees similarly to Lemma 3.1 when dividing FPk along the 2th dimension.
Case 1.2.2.2. x,y,w0 are in a subgraph of FPk isomorphic to C4.
In this case, k≥3. Without loss of generality, assume that x=000⋯01 and y=000⋯40. Let z=0z2z3…zk−2zk−1zk. If there exists j∈{2,…,k−2} such that zj=0, then x,y,z,w∈FPj:0k when dividing FPk along the jth dimension. Thus, we can find out desired trees similarly to Lemma 3.3. It suffices to consider the case z2,z3,…,zk−2≠0.
If k≥4, we divide FPk along the 2th dimension. Then x,y,w∈V(FP2:0k), z∈V(FP2:z2k), z0=00z3⋯zk−1zk∉{x,y,w} and |N(z0)∩{x,y,w}|≤1 except z0=0041. Similarly to case 1.2.1, we can find 3k−3 internally disjoint S-trees. When z0=0041, x=0001, y=0040, w=1000, assume that z=0141 (for z=0z241 and z2≠0 is similar). Clearly, |N(z0)∩{x,y,w}|=2. The desired trees are shown in Figure 12.
Suppose that k=3, we have x=001, y=040, w=100, z=0z2z3∉{x,y,w0}. If z2∉{0,4}, then x,w∈FP2:03,y∈FP2:43 and z∈FP2:z23 when dividing FP3 along the 2th dimension. Thus, we can find out desired trees similarly to Lemma 3.2. If z3∉{0,1}, then x∈FP3:13,y,w∈FP3:03 and z∈FP3:z33 when dividing FP3 along the 3th dimension. Thus, we can find out desired trees similarly to Lemma 3.2. Suppose z2∈{0,4} and z3∈{0,1}. Since z≠x,y,w, then z=041. The desired trees are shown in Figure 13.
Case 2. x,y∈V(FP0) and z,w∈V(FP1).
Notice that |V(L∗2)|≥3 and |V(L∗3)|≥3. Without loss of generality, let 4∈V(L∗2) and 5∈V(L∗3). Since FP4 is connected, there exists an {x4,y4,z4,w4}-tree T′3k−2 in FP4, similarly, there exists an {x5,y5,z5,w5}-tree T′3k−1 in FP5. By the definition of FPk, there exist four paths Pzz4, Pww4, Pzz5 and Pww5 such that Pzz4≅Pww4≅L∗2∖{0} and Pzz5≅Pww5≅L∗3∖{0}. Let T3k−2=T′3k−2∪xx4∪yy4∪Pzz4∪Pww4 and T3k−1=T′3k−1∪xx5∪yy5∪Pzz5∪Pww5. Then T3k−2 and T3k−1 are internally disjoint S-trees in FPk. Main goal is to find out 3k−3 internally disjoint S-trees in L∗1◻FPk−1. Let S′ be a set of x,y,z0 and w0. Without loss of generality, assume that d(x,z0)=mind(u,v) for u,v∈S′.
Case 2.1. |S′|≤3.
In this case, d(x,z0)=0, that is, z0=x. Since FPk−1 is (3k−3)-connected, by Theorem 2.1, there exist 3k−3 internally disjoint (x,y)-paths P1,P2,…,P3k−3 in FP0 and 3k−3 internally disjoint (z,w)-paths P′1,P′2,…,P′3k−3 in FP1. Let xi∈NPi(x) and zi∈NP′i(z). Then there exists zi∈V(P′i) such that zi is a corresponding vertex of xj in FP1 for i,j∈[3k−3], suppose that i=j. Hence, xizi∈E(FPk). Let Ti=Pi∪xizi∪P′i for i∈[3k−3]. Then, T1,T2,…,T3k−3 are 3k−3 internally disjoint S-trees in L∗1◻FPk−1. See Figure 14(a).
Case 2.2. |S′|=4.
In this case, d(x,z0)≥1. It implies that d(x,z)≥2. If d(x,z)=2. Without loss of generality, assume that x=000⋯00. Then z=1c2⋯ck such that there exist an i∈{2,…,k} satisfying dP(0,ci)=1 and others ci=0, say z=110⋯00. If there exist a dimension j such that |S∩FPj:i|≠2 for j∈[k] and i∈{0,1,…,9} when dividing FPk along the jth dimension, then we can find out the desired S-trees by the above discussion. Hence, suppose that |S∩FPj:i|=2 for arbitrary j when dividing FPk along the jth dimension. Thus, y=01b3⋯bk and w=10b3⋯bk, where bi≠0 for i∈{3,…,k}.
Let FPij=FPk[d1d2d3⋯dk∈V(FPk):d1=i,d2=j]. Then FPij≅FPk−2. Let x1 be corresponding vertex of y in FP00. Then x1≠x. Choose 3k−7 vertices x2,…,x3k−6 from NFP00(x), denote X={x1,…,x3k−6}. For i=1,…,3k−6, let zi be corresponding vertices of xi in FP11, denote Z={z1,…,z3k−6}. For i=2,…,3k−6, let yi and wi be corresponding vertices of xi in FP01 and FP10, respectively. Denote Y={y1,…,y3k−7} and W={w1,…,w3k−7}, where y1 and w1 are be corresponding vertices of x in FP01 and FP10, respectively. Since FPk−2 is (3k−6)-connected, there exist 3k−6 internally disjoint (a,A)-paths A1,…,A3k−6 in FPij for (a,A,ij)=(x,X,00),(y,Y,01),(z,Z,10),(w,W,11), respectively. Let T0=X1∪x1y∪yz1∪Z1∪z1w, T1=xy1∪Y1∪y1z∪xw1∪W1 and Ti=Xi∪xiyi∪Yi∪Zi∪ziwi∪Wi∪xiwi for i=2,…,3k−6.
Let B0 be a subgraph induced by V(FP0−FP00−FP01) and B1 be a subgraph induced by V(FP1−FP10−FP11). Then there exist two internally disjoint paths P3k−5,P3k−4 to connect x and y in FPk[V(B0)∪{x,y}] and two internally disjoint paths P′3k−5,P′3k−4 to connect z and w in FPk[V(B1)∪{z,w}]. Let Tj=Pj∪xjzj∪P′j for j=3k−5 and 3k−4, where xxj∈E(Pj) and zj is corresponding vertex of xj in P′i. Then, T0,T1,…,T3k−4 are 3k−3 internally disjoint S-trees in L∗1◻FPk−1. See Figure 14(b).
Suppose that d(x,z)≥3. There are 3k−4 internally disjoint S′-trees T′1,T′2,…,T′3k−4 in FP0 because of the induction hypothesis. Notice that |NFPk(z0)∩NFPk(w0)|≤2. Let T′1 be a tree such that |NT′1(z0)∩NT′1(w0)|=0. Let z1∈NT′1(z0) and w1∈NT′1(w0). Then z1≠w1. Let T1=T′1∪{zz0,ww0}.
Let S″={x,y,z,w0}. Since dFP0(z0)=3k−3, there are 1≤dT′i(z0)≤2 for i∈[3k−4]. Construct 3k−5 internally disjoint S″-trees as follows: T″i=(T′i∖z0)∪ziz1i∪z1iz if NT′i(z0)={zi} and T″i=(T′i∖z0)∪ziz1i∪z1iz∪zjz1j∪z1jz if NT′i(z0)={zi,zj}, where i=2,…,3k−4. Similarly, construct 3k−5 internally disjoint S-trees as follows: Ti=(T′i∖w0)∪wiw1i∪w1iw if NT″i(w0)={wi} and Ti=(T″i∖w0)∪wiw1i∪w1iw∪wjw1j∪w1jw if NT″i(w0)={wi,wj}, where i=2,…,3k−4.
Let T=⋃3k−4i=1Ti, W=NT(z)∪NT(w) and T″1 be a corresponding tree of T′1 in FP1. Then, T″1 is a tree in FP1−W. Let T3k−3=T″1∪xx1∪yy1. Then, T1,T2,…,T3k−3 are 3k−3 internally disjoint S-trees in L∗1◻FPk−1.
Therefore, T1,T2,…,T3k−1 are 3k−1 internally disjoint S-trees in FPk.
Giving an algorithm to find out 3k−1 internally disjoint S-trees in FPk for any S⊆V(FPk) with |S|=4, it means that κ4(FPk)≥3k−1.
Algorithm 1 Find out 3k−1 internally disjoint S-trees in FPk. |
Input: An k-dimensional folded Petersen network FPk and four vertices x,y,z,w of FPk. Output: 3k−1 internally disjoint {x,y,z,w}-trees T. 1: Initialization: i=0, S={x,y,z,w}, T=∅, Gi=FPk 2: While i<3k−1 and Gi is connected do 3: construct an S-tree Ti in Gi such that 1≤dTi(v)≤2 and |{v:dTi(v)=1}|≥2, where v∈S. Moreover, {u:dTi(u)=1}⊆S 4: T=T∪Ti 5: i=i+1 6: Gi=Gi−1−(V(Ti−1)∖S) 7: end while 8: return T |
Now we give the proof of the main result.
Proof of Theorem 1.2. Since FPk is a 3k-regular graph, there are κ4(FPk)≤3k−1 as Lemma 2.1. Next we will show that κ4(FPk)≥3k−1. Let S={x,y,z,w} be a set of arbitrary four distinct vertices in FPk. It suffices to show that there exist 3k−1 internally disjoint S-trees. The proof of this result by induction on k. By Lemma 2.3, the statement holds for k=1. Suppose that the statement holds in FPk−1 for k≥2. Now consider FPk. Decompose FPk into 10 disjoint sub-folded Petersen graphs FP0,…,FP9, each of which is isomorphic to FPk−1, by removing all crossed edges. We only need to take into account the following cases because of symmetry.
Case 1. x,y,z and w belong to four distinct sub-folded Petersen graphs.
By Lemma 3.1, the desired 3k−1 internally disjoint S-trees can be obtained in FPk.
Case 2. x,y,z and w belong to three distinct sub-folded Petersen graphs.
By Lemma 3.2, the desired 3k−1 internally disjoint S-trees can be obtained in FPk.
Case 3. x,y,z and w belong to two distinct sub-folded Petersen graphs.
By Lemma 3.4, the desired 3k−1 internally disjoint S-trees can be obtained in FPk.
Case 4. x,y,z and w belong to the same sub-folded Petersen graph.
By Lemma 3.3, the desired 3k−1 internally disjoint S-trees can be obtained in FPk.
Hence, κ4(FPk)≥3k−1 and the proof is completed.
Proof of Theorem 1.1. Remember that FPQn,k can be regarded as replacing every vertex of Qn by FPk. Take the Figure 15 as an example. Since FPQn,k is (n+3k)-regular, we have κ4(FPQn,k)≤n+3k−1 by Lemma 2.1. In order to prove κ4(FPQn,k)≥n+3k−1, it needs to show that there are n+3k−1 internally disjoint S-trees in FPQn,k for arbitrary S⊆V(FPQn,k) with |S|=4.
If n=1. According to Lemma 3.3 and 3.4, it is not hard to see that there exist 3k−3 internally disjoint S-trees in K2◻FPk−1. That is, κ4(FPQ1,k−1)≥3k−3. Hence, κ4(FPQ1,k)≥3k.
Suppose that n≥2. When the vertices of S distribute among one copy of FPk. Similar to Lemma 3.3, the desired n+3k−1 internally disjoint S-trees can be found in FPQn,k.
When the vertices of S distribute among two copies of FPk. Since κ(Qn)=n, there are n internally disjoint paths L∗1,L∗2,…,L∗n connecting arbitrary two vertices of Qn. Similar to Lemma 3.4, we can find 3k S-trees in L∗1◻FPk and n−1 S-trees in (⋃ni=2L∗i)◻FPk such that these n+3k−1 S-trees are internally disjoint.
When the vertices of S distribute among three copies of FPk. Since κ3(Qn)=n−1, there are n−1 internally disjoint path T∗1,T∗2,…,T∗n−1 connecting arbitrary three vertices of Qn. Similar to Lemma 3.2, we can find 3k+1 S-trees in T∗1◻FPk and n−2 S-trees in (⋃n−1i=2T∗i)◻FPk such that these n+3k−1 S-trees are internally disjoint.
When the vertices of S distribute among four copies of FPk. Since κ4(Qn)=n−1, there are n−1 internally disjoint path H∗1,H∗2,…,H∗n−1 connecting arbitrary four vertices of Qn. Similar to Lemma 3.1, we can find 3k+1 S-trees in H∗1◻FPk and n−2 S-trees in (⋃n−1i=2H∗i)◻FPk such that these n+3k−1 S-trees are internally disjoint.
Therefore, κ4(FPQn,k)=n+3k−1.
The generalized ℓ-connectivity is a natural generalization of the traditional connectivity and can serve for measuring the fault tolerance capability of a network. This paper centers on the generalized 4-connectivity of the folded Petersen cube network FPQn,k and shows that κ4(FPQn,k)=n+3k−1. As a corollary, κ3(FPQn,k)=n+3k−1 is obtained easily. Furthermore, the results κ4(Qn)=κ4(FPQn,0)=n−1 and κ4(FPk)=κ4(FPQ0,k)=3k−1 can be verified. Sabidussi [29] discussed the classical connectivity of Cartesian product graphs, in the next work, we would like to research the generalized 4-connectivity of Cartesian product graphs.
Besides, fault tolerance or connectivity is mainly to provide a data to measure the reliability of a network, but in practical, when a system failure, it is worth considering how to compensate the impact of the fault and to recover the performance of the system before the failure as far as possible such that the system runs stably and reliably. Motivated by Alhasnawi et al. [1,2,3,4], it needs to design fault tolerance control to ensure steady-state operation, enhance network' fault resilience, improve network' robust and efficient operation. In future work, we would like to apply graph theory to solve practical problems.
This work was supported by the National Natural Science Foundation of China (No. 11971054, 11731002, 11661068) and the Science Found of Qinghai Province (No. 2021-ZJ-703). The authors would like to express their thanks to the referees for their helpful comments and suggestions which improve the presentation of this paper.
The authors declare that they have no competing interests.
[1] |
B. N. Alhasnawi, B. H. Jasim, B. E. Sedhom, Distributed secondary consensus fault tolerant control method for voltage and frequency restoration and power sharing control in multi-agent microgrid, Int. J. Elec. Power, 133 (2021), 107251. https://doi.org/10.1016/j.ijepes.2021.107251 doi: 10.1016/j.ijepes.2021.107251
![]() |
[2] |
B. N. Alhasnawi, B. H. Jasim, B. E. Sedhom, J. M. Guerrero, Consensus algorithm-based coalition game theory for demand management scheme in smart microgrid, Sustain Cities Soc., 74 (2021), 103248. https://doi.org/10.1016/j.scs.2021.103248 doi: 10.1016/j.scs.2021.103248
![]() |
[3] |
B. N. Alhasnawi, B. H. Jasim, Z. A. S. A. Rahman, J. M. Guerrero, M. D. Esteban, A novel internet of energy based optimal multi-agent control scheme for microgrid including renewable energy resources, Int. J. Environ. Res. Public Health, 18 (2021), 8146. https://doi.org/10.3390/ijerph18158146 doi: 10.3390/ijerph18158146
![]() |
[4] |
B. N. Alhasnawi, B. H. Jasim, B. E. Sedhom, E. Hossain, J. M. Guerrero, A new decentralized control strategy of microgrids in the internet of energy paradigm, Energies, 14 (2021), 2183. https://doi.org/10.3390/en14082183 doi: 10.3390/en14082183
![]() |
[5] | J. A. Bondy, U. S. R. Murty, Graph theory, New York: Springer, 2008. |
[6] | G. Chartrand, S. F. Kapoor, L. Lesniak, D. R. Lick, Generalized connectivity in graphs, Bull. Bombay Math. Colloq., 2 (1984), 1–6. |
[7] |
G. Chartrand, F. Okamoto, P. Zhang, Rainbow trees in graphs and generalized connectivity, Networks, 55 (2010), 360–367. https://doi.org/10.1002/net.20339 doi: 10.1002/net.20339
![]() |
[8] | G. Chartrand, R. J. Wilson, The Petersen graph, Graph. Appl., 69 (1985), 100. |
[9] |
S. K. Das, S. Öhring, A. K. Banejee, Embeddings into hyper {P}etersen networks: Yet another hypercube-like interconnection topology, VLSl Design, 2 (1995), 335–351. https://doi.org/10.1155/1995/95759 doi: 10.1155/1995/95759
![]() |
[10] |
K. Efe, P. K. Blackwell, W. Slough, T. Shiau, Topological properties of the crossed cube architecture, Parallel Comput., 20 (1994), 1763–1775. https://doi.org/10.1016/0167-8191(94)90130-9 doi: 10.1016/0167-8191(94)90130-9
![]() |
[11] | M. Hager, Pendant tree-connectivity, J. Comb. Theory B, 38 (1985), 179–189. https: //doi.org/10.1016/0095-8956(85)90083-8 |
[12] |
C. F. Li, S. W. Li, S. J. Li, The 4-set tree connectivity of (n,k)-star networks, Theor. Comput. Sci., 884 (2020), 81–86. https://doi.org/10.1016/j.tcs.2020.08.004 doi: 10.1016/j.tcs.2020.08.004
![]() |
[13] |
H. Li, X. Li, Y. Sun, The generalized 3-connectivity of {C}artesian product graphs, Discrete Math. Theor., 14 (2012), 1. https://doi.org/10.46298/dmtcs.572 doi: 10.46298/dmtcs.572
![]() |
[14] |
H. Li, Y. Ma, W. Yang, Y. Wang, The generalized 3-connectivity of graph products, Appl. Math. Comput., 295 (2017), 77–83. https://doi.org/10.1016/j.amc.2016.10.002 doi: 10.1016/j.amc.2016.10.002
![]() |
[15] | S. Li, Some topics on generalized connectivity of graphs, Nankai University, 2012. |
[16] | S. Li, W. Li, X. Li, The generalized connectivity of complete bipartite graphs, 2010, arXiv: 1012.5710v1. |
[17] |
S. Li, W. Li, Y. Shi, H. Sun, On minimally 2-connected graphs with generalized connectivity κ3=2, J. Comb. Optim., 34 (2017), 141–164. https://doi.org/10.1007/s10878-016-0075-z doi: 10.1007/s10878-016-0075-z
![]() |
[18] | S. Li, W. Li, X. Li, The generalized connectivity of complete equipartition 3-partite graphs, Bull. Malays. Math. Sci. Soc., 37 (2014), 103–121. |
[19] |
S. Li, X. Li, Note on the hardness of generalized connectivity, J. Comb. Optim., 24 (2012), 389–396. https://doi.org/10.1007/s10878-011-9399-x doi: 10.1007/s10878-011-9399-x
![]() |
[20] | S. Li, X. Li, Y. Shi, The minimal size of a graph with generalized connectivity κ3≥2, Australas. J. Comb., 51 (2011), 209–220. |
[21] |
S. Li, X. Li, W. Zhou, Sharp bounds for the generalized connectivity κ3(G), Discrete Math., 310 (2010), 2147–2165. https://doi.org/10.1016/j.disc.2010.04.011 doi: 10.1016/j.disc.2010.04.011
![]() |
[22] |
S. Li, Y. Shi, J. Tu, The generalized 3-connectivity of {C}ayley graphs on symmetric groups generated by trees and cycles, Graph. Combinator., 33 (2017), 1195–1209. https://doi.org/10.1007/s00373-017-1837-9 doi: 10.1007/s00373-017-1837-9
![]() |
[23] |
S. Li, J. Tu, C. Yu, The generalized 3-connectivity of star graphs and bubble-sort graphs, Appl. Math. Comput., 274 (2016), 41–46. https://doi.org/10.1016/j.amc.2015.11.016 doi: 10.1016/j.amc.2015.11.016
![]() |
[24] |
S. Lin, Q. Zhang, The generalized 4-connectivity of hypercubes, Discrete Appl. Math., 220 (2017), 60–67. https://doi.org/10.1016/j.dam.2016.12.003 doi: 10.1016/j.dam.2016.12.003
![]() |
[25] | X. Li, Y. Mao, Y. Sun, On the generalized (edge-)connectivity of graphs, Australas. J. Comb., 58 (2014), 304–319. |
[26] |
X. Li, Y. Mao, The generalized 3-connectivity of lexicographic product graphs, Discrete Math. Theor., 16 (2014), 339–354. https://doi.org/10.46298/dmtcs.1266 doi: 10.46298/dmtcs.1266
![]() |
[27] | X. Li, Y. Mao, Generalized connectivity of graphs, Switzerland: Springer, 2016. https://doi.org/10.1007/978-3-319-33828-6 |
[28] |
S. R. Öhring, S. K. Das, Folded {P}etersen cube networks: new competitors for the hypercubes, IEEE T. Parall. Distr., 7 (1996), 151–168. https://doi.org/10.1109/71.485505 doi: 10.1109/71.485505
![]() |
[29] |
G. Sabidussi, Graphs with given group and given graph theoretical properties, Can. J. Math., 9 (1957), 515–525. https://doi.org/10.4153/CJM-1957-060-7 doi: 10.4153/CJM-1957-060-7
![]() |
[30] |
P. C. Saxena, S. Gupta, J. Rai, A delay optimal coterie on the k-dimensional folded {P}etersen graph, J. Parallel Distr. Com., 63 (2003), 1026–1035. https://doi.org/10.1016/S0743-7315(03)00116-3 doi: 10.1016/S0743-7315(03)00116-3
![]() |
[31] |
Y. Sun, S. Zhou, Tree connectivities of {C}aylay graphs on {A}belian groups with small degrees, Bull. Malays. Math. Sci. Soc., 39 (2016), 1673–1685. https://doi.org/10.1007/s40840-015-0147-8 doi: 10.1007/s40840-015-0147-8
![]() |
[32] |
H. Whitney, Congruent graphs and connectivity of graphs, Amer. Math. Soc., 54 (1932), 150–168. https://doi.org/10.2307/2371086 doi: 10.2307/2371086
![]() |
[33] |
S. Zhao, R. Hao, The generalized 4-connectivity of exchanged hypercubes, Appl. Math. Comput., 347 (2019), 342–353. https://doi.org/10.1016/j.amc.2018.11.023 doi: 10.1016/j.amc.2018.11.023
![]() |
[34] |
S. Zhao, R. Hao, The generalized connectivity of alternating group graphs and (n,k)-star graphs, Discrete Appl. Math., 251 (2018), 310–321. https://doi.org/10.1016/j.dam.2018.05.059 doi: 10.1016/j.dam.2018.05.059
![]() |
[35] |
S. Zhao, R. Hao, E. Cheng, Two kinds of generalized connectivity of dual cubes, Discrete Appl. Math., 257 (2019), 306–316. https://doi.org/10.1016/j.dam.2018.09.025 doi: 10.1016/j.dam.2018.09.025
![]() |
[36] |
S. Zhao, R. Hao, J. Wu, The generalized 4-connectivity of hierarchical cubic networks, Discrete Appl. Math., 289 (2021), 194–206. https://doi.org/10.1016/j.dam.2020.09.026 doi: 10.1016/j.dam.2020.09.026
![]() |
[37] |
S. Zhao, R. Hao, J. Wu, The generalized 3-connectivity of some regular networks, J. Parallel Distr. Com., 133 (2019), 18–29. https://doi.org/10.1016/j.jpdc.2019.06.006 doi: 10.1016/j.jpdc.2019.06.006
![]() |
[38] |
S. Zhao, R. Hao, L. Wu, The generalized connectivity of (n,k)-bubble-sort graphs, Comput. J., 62 (2019), 1277–1283. https://doi.org/10.1093/comjnl/bxy106 doi: 10.1093/comjnl/bxy106
![]() |
1. | Mohamad Abdallah, Generalized 4-connectivity of alternating group networks, 2024, 80, 0920-8542, 12585, 10.1007/s11227-024-05922-3 | |
2. | Caixi Xue, Shuming Zhou, Hong Zhang, Qifan Zhang, The generalized 4-connectivity of complete-transposition graphs, 2024, 39, 1744-5760, 399, 10.1080/17445760.2023.2261192 |
Algorithm 1 Find out 3k−1 internally disjoint S-trees in FPk. |
Input: An k-dimensional folded Petersen network FPk and four vertices x,y,z,w of FPk. Output: 3k−1 internally disjoint {x,y,z,w}-trees T. 1: Initialization: i=0, S={x,y,z,w}, T=∅, Gi=FPk 2: While i<3k−1 and Gi is connected do 3: construct an S-tree Ti in Gi such that 1≤dTi(v)≤2 and |{v:dTi(v)=1}|≥2, where v∈S. Moreover, {u:dTi(u)=1}⊆S 4: T=T∪Ti 5: i=i+1 6: Gi=Gi−1−(V(Ti−1)∖S) 7: end while 8: return T |
Algorithm 1 Find out 3k−1 internally disjoint S-trees in FPk. |
Input: An k-dimensional folded Petersen network FPk and four vertices x,y,z,w of FPk. Output: 3k−1 internally disjoint {x,y,z,w}-trees T. 1: Initialization: i=0, S={x,y,z,w}, T=∅, Gi=FPk 2: While i<3k−1 and Gi is connected do 3: construct an S-tree Ti in Gi such that 1≤dTi(v)≤2 and |{v:dTi(v)=1}|≥2, where v∈S. Moreover, {u:dTi(u)=1}⊆S 4: T=T∪Ti 5: i=i+1 6: Gi=Gi−1−(V(Ti−1)∖S) 7: end while 8: return T |