The spanning laceability on the faulty bipartite hypercube-like
networks
Cheng-Kuan Lin
a, Yuan-Hsiang Teng
b,⇑, Jimmy J.M. Tan
c, Lih-Hsing Hsu
d, Dragan Marušicˇ
e aInstitute of Information Science, Academia Sinica, Taipei City 11529, Taiwan, ROC
b
Department of Computer Science and Information Engineering, Hungkuang University, Taichung City 433, Taiwan, ROC
c
Department of Computer Science, National Chiao Tung University, Hsinchu 30010, Taiwan, ROC
d
Department of Computer Science and Information Engineering, Providence University, Taichung City 433, Taiwan, ROC
e
IMFM, University of Ljubljana, 1111 Ljubljana, Slovenia
a r t i c l e
i n f o
Keywords: Hamiltonian Hamiltonian laceable Hypercube networks Hypercube-like network Spanning laceabilitya b s t r a c t
A w-container Cðu;
v
Þ of a graph G is a set of w-disjoint paths joining u tov
. A w-container of G is a w-container if it contains all the nodes of VðGÞ. A bipartite graph G is w-laceable if there exists a w-container between any two nodes from different parts of G. Let n and k be any two positive integers with n P 2 and k 6 n. In this paper, we prove that n-dimensional bipartite hypercube-like graphs are f-edge fault k-laceable for every f 6 n 2 and f þ k 6 n.Ó 2013 Elsevier Inc. All rights reserved.
1. Introduction
1.1. Basic graph definitions and notations
The research about interconnection networks is important for parallel and distributed computer systems. The layouts of processors and links in distributed computer systems are usually represented by a network structure. Computer network topologies are usually represented by graphs where nodes represent processors and edges represent links between proces-sors. The containers of graphs do exist in information engineering design, telecommunication networks, and biological neu-ral systems ([1,2]and its references). The study of w-container, w-wide distance, and their w-versions play a pivotal role in
the design and the implementation of parallel routing and efficient information transmission in large scale networking sys-tems. In bioinformatics and neuroinformatics, the existence as well as the structure of a w-container signifies the cascade
effect in the signal transduction system and the reaction in a metabolic pathway.
For graph definitions and notations, we follow[3,4]. Let G ¼ ðV; EÞ be a graph where V is a finite set and E is a subset of fðu;
v
Þjðu;v
Þ is an unordered pair of Vg. We say that V is the node set and E is the edge set. We use nðGÞ to denote jVj. Two nodes u andv
are adjacent if ðu;v
Þ 2 E. For a node u, we use NGðuÞ to denote the neighborhood of u which is the setf
v
jðu;v
Þ 2 Eg. For any node u of V, we denote the degree of u by degGðuÞ ¼ jNGðuÞj. A graph G is k-regular if degGðuÞ ¼ k forevery node u in G. A path P between nodes
v
1andv
kis a sequence of adjacent nodes, hv
1;v
2; . . . ;v
ki, in which the nodesv
1;v
2; . . . ;v
k are distinct except that possiblyv
1¼v
k. We use P1 to denote the path hv
k;v
k1; . . . ;v
1i. The length ofP; lðPÞ, is the number of edges in P. We also write the path P as h
v
1;v
2; . . . ;v
i;Q ;v
j;v
jþ1; . . . ;v
ki, where Q is the pathh
v
i;v
iþ1; . . . ;v
ji. Hence, it is possible to write a path as hv
1;v
2;Q ;v
2;v
3; . . . ;v
ki if lðQ Þ ¼ 0. Let IðPÞ ¼ VðPÞ fv
1;v
kg be theset of the internal nodes of P. A set of paths fP1;P2; . . . ;Pkg are internally node-disjoint (abbreviated as disjoint) if
0096-3003/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.amc.2013.02.027
⇑Corresponding author.
E-mail address:yhteng@sunrise.hk.edu.tw(Y.-H. Teng).
Contents lists available atSciVerse ScienceDirect
Applied Mathematics and Computation
IðPiÞ \ IðPjÞ ¼ ; for any i – j. A path is a hamiltonian path if it contains all nodes of G. A graph G is hamiltonian connected if
there exists a hamiltonian path joining any two distinct nodes of G[5]. A cycle is a path with at least three nodes such that the first node is the same as the last one. A hamiltonian cycle of G is a cycle that traverses every node of G. A graph is ham-iltonian if it has a hamham-iltonian cycle. A graph G is bipartite if its node set can be partitioned into two subsets V1and V2such
that every edge connects nodes between V1and V2. A bipartite graph G is hamiltonian laceable if there is a hamiltonian path
of G joining any two nodes from distinct bipartition[6]. A bipartite graph G is k-edge fault hamiltonian laceable if G F is ham-iltonian laceable for any edge subset F of G with jFj 6 k.
A graph G is k-connected if there exists a set of k internally disjoint paths fP1;P2; . . . ;Pkg between any two distinct nodes u
and
v
. A subset S of VðGÞ is a cut set if G S is disconnected. A container Cðu;v
Þ between two distinct nodes u andv
in G is a set of disjoint paths between u andv
. A w-container Cwðu;v
Þ in a graph G is a set of w internally node-disjoint paths between uand
v
. The concepts of a container and of a wide distance were proposed by Hsu[2]to evaluate the performance of commu-nication for an interconnection network. The connectivity of G,j
ðGÞ, is the minimum number of nodes whose removal leaves the remaining graph disconnected or trivial. Hence, a graph G is k-connected ifj
ðGÞ P k. It follows from Menger’s Theorem[7]that there is a w-container for w 6 k between any two distinct nodes of G if G is k-connected. 1.2. w-connected graphs and w-laceable graphs
In this paper, we are interested in a specific type of container. A w-container C
wðu;
v
Þ in a graph G is a w-container suchthat every node of G is on some path in Cwðu;
v
Þ. A graph G is w-connected if there exists a w-container between any twodistinct nodes in G. Obviously, we have the following remark.
Remark 1. ð1:aÞ a graph G is 1-connected if and only if it is hamiltonian connected[5], ð1:bÞ a graph G is 2-connected if it is hamiltonian, and ð1:cÞ an 1-connected graph except K1and K2is 2-connected.
The study of w-connected graph is motivated by the 3-connected graphs proposed by Albert et al.[8]. Some related
works have appeared in[8,9]. Assume that the graph G is w-connected with w 6
j
ðGÞ. The spanning connectivity of a graphG;
j
ðGÞ, is the largest integer k such that G is i-connected for every i with 1 6 i 6 k. A graph G is super spanning connected ifj
ðGÞ ¼j
ðGÞ. In such case, the numberj
ðGÞ ¼j
ðGÞ is called the super spanning connectivity of G. In[10–13], some familiesof graphs are proved to be super spanning connected.
A bipartite graph is said to be w-laceable if there exists a w-container between any two nodes from different partite sets
for some w with 1 6 w 6
j
ðGÞ. Any bipartite w-laceable graph with w P 2 has the equal size of bipartition. We have thefollowing remark.
Remark 2. ð2:aÞ an 1-laceable graph is also known as hamiltonian laceable graph[6], ð2:bÞ a graph G is 2-laceable if and only if it is hamiltonian, and ð2:cÞ an 1-laceable graph except K1and K2are 2-laceable.
The spanning laceability of a bipartite graph G;
j
LðGÞ, is the largest integer k such that G is i-laceable for every i with1 6 i 6 k. A graph G is super spanning laceable if
j
LðGÞ ¼j
ðGÞ. Recently, Chang et al.[14]proved that the n-dimensionalhypercube Qnis super spanning laceable for every positive integer n. It was proved in[11]that the n-dimensional star graph
Snis super spanning laceable if and only if n – 3.
1.3. Hypercube-like graphs H0n
Among all interconnection networks proposed in the literature, the hypercube Qnis one of the most popular topologies
[14–17]. However, the hypercube does not have the smallest diameter for its resources. Various networks are proposed by twisting some pairs of links in hypercubes[18–21]. Because of the lack of the unified perspective on these variants, results of one topology are hard to be extended to others. To make a unified study of these variants, Vaidya et al. introduced the class of hypercube-like graphs[22]. We denote these graphs as H0-graphs. The class of H0-graphs, consisting of simple, connected,
and undirected graphs, contains most of the hypercube variants.
Let G0¼ ðV0;E0Þ and G1¼ ðV1;E1Þ be two disjoint graphs with the same number of nodes. A 1–1 connection between G0
and G1 is defined as E ¼ fð
v
;/ðv
ÞÞjv
2 V0;/ðv
Þ 2 V1, and / : V0! V1 is a bijectiong. We use G0 G1 to denoteG ¼ ðV0[ V1;E0[ E1[ EÞ. The operation ‘‘’’ may generate different graphs depending on the bijection /. There are some
studies on the operation ‘‘’’[23,24]. Let G ¼ G0 G1, and let x be any node in G. We use x to denote the unique node
matched under /.
Now, we can define the set of n-dimensional H0-graph, H0n, as follows:
(1) H0
1¼ fK2g, where K2is the complete graph with two nodes.
(2) Assume that G0;G12 H0n. Then G ¼ G0 G1is a graph in H0nþ1.
We can define the set of bipartite n-dimensional H0-graph, B0
(1) B0
1¼ fK2g, where K2is the complete graph defined on fa; bg with bipartition V0¼ fag and V1¼ fbg.
(2) For i ¼ 0; 1, let Gibe a graph in B0nwith bipartition V i 0and V
i
1. Let / be a bijection between V 0 0[ V 0 1and V 1 0[ V 1 1such that /ð
v
Þ 2 V11iifv
2 V 0 i. Then G ¼ G0 G1is a graph in B0nþ1. Every graph in H0nis an n-regular graph with 2
nnodes, and every graph in B0
ncontains 2
n1nodes in each bipartition. Note
that the n-dimensional hypercube Qn2 B0n.
Let G be a graph in H0nþ1. Then G ¼ G0 G1with both G0and G1in H0n. Let u be a node in VðGÞ. Then u is a node in VðGiÞ for
some i ¼ 0; 1. We use u to denote the node in VðG1iÞ matched under /. So u ¼
v
if u ¼v
.In the following section, we give some properties about the bipartite n-dimensional hypercube-like graphs B0
n. Let n and k
be any two positive integers with n P 2 and k 6 n. In Section3and Section4, we prove that every B0
nis f-edge fault k
-lace-able for every f 6 n 2 and f þ k 6 n. We give our conclusion in the final section.
2. Preliminaries
Park and Chwa[25]studied the hamiltonian laceability properties of the bipartite hypercube-like networks. Some results are listed as follows.
Theorem 1 [25]. Every graph in B0
nis hamiltonian laceable, and every graph in B0nis hamiltonian if n P 2. Theorem 2 [25]. Suppose that n P 2; i 2 f0; 1g, and G is a graph in B0
nwith bipartition G0and G1. Let fu1;u2g # VðGiÞ with
u1–u2, and f
v
1;v
2g # VðG1iÞ withv
1–v
2. Then there are two disjoint paths P1and P2of G such that (1) P1joins u1tov
1,(2) P2joins u2to
v
2, and (3) P1[ P2spans G.The fault-tolerance hamiltonian laceability of the bipartite hypercube-like networks is studied by Lin et al. in[26]. Theorem 3 [26]. Let n P 2. Every graph in B0
nis ðn 2Þ-edge fault hamiltonian laceable. Theorem 4 [26]. Suppose that n P 2; i 2 f0; 1g, and G is a graph in B0
n with bipartition G0 and G1. Let z 2 VðGiÞ, and
fu;
v
g # VðG1iÞ with u –v
. Then there is a hamiltonian path of G fzg joining u to v.3. The super spanning laceability of the graph in B0 n
Let n and k be any two positive integers with n P 2 and k 6 n. In this section, we show that every graph in B0
nis f-edge
fault k-laceable for every f 6 n 2 and f þ k 6 n. We give the concept of the spanning fan first. We note that there is another
Menger-type Theorem. Let u be a node of G and S ¼ f
v
1;v
2; . . . ;v
kg be a subset of VðGÞ not including u. An ðu; SÞ-fan is a set ofdisjoint paths fP1;P2; . . . ;Pkg of G such that Pijoins u to
v
ifor every 1 6 i 6 k[27]. It is proved that a graph G is k-connected ifand only if there exists an ðu; SÞ-fan between any node u and any k-subset S of VðGÞ such that u R S. With this observation, we define a spanning fan is a fan that spans a graph G. Naturally, we can study
j
fanðGÞ as the largest integer k such that there
exists a spanning ðu; SÞ-fan between any node u and any k-node subset S with u R S. However, we defer such a study for the following reasons.
First, let S be a cut set of a graph G. Let u be any node of VðGÞ S. It is easy to see that there is no spanning ðu; SÞ-fan in G. Thus,
j
fanðGÞ <
j
ðGÞ if G is not a complete graph.Second, let G be a bipartite graph with bipartition G0¼ ðV0;E0Þ and G1¼ ðV1;E1Þ such that jV0j ¼ jV1j. Let u be a node in Vi
with i 2 f0; 1g; S ¼ f
v
1;v
2; . . . ;v
kg # VðGÞ fug, and k 6j
ðGÞ. Suppose that jS \ V1ij ¼ r. Without loss of generality, weas-sume that f
v
1;v
2; . . . ;v
rg V1i. Let fP1;P2; . . . ;Pkg be any spanning ðu; SÞ-fan of G. Then lðPiÞ is odd if i 6 r, and lðPiÞ is even ifr < i 6 k. Let lðPiÞ ¼ 2tiþ 1 if i 6 r and lðPiÞ ¼ 2tiif i > r. For i 6 r, there are ti 1 nodes of Piin Viother than u, and there are ti
nodes of Pi in V1i. For i > r, there are tinodes of Piin Viother than u, and there are tinodes of Pi in V1i. Thus, we have
jVij ¼ 1 r þPki¼1tiand jV1ij ¼Pki¼1ti. Since jVij ¼ jV1ij; r ¼ 1. Thus, r ¼ 1 is a natural requirement as we study the
span-ning fan of bipartite graphs with equal size of bipartition.
Theorem 5 [12]. Suppose that n and k are two positive integers with k 6 n. Let G be a graph in B0
nwith bipartition G0and G1. There exists a spanning ðu; SÞ-fan in G for any node u in VðGiÞ and any node subset S with jSj ¼ k 6 n such that u R S, and jS \ VðG1iÞj ¼ 1 with i 2 f0; 1g.
Lemma 1. Suppose that
1. n P 2; f ¼ n 2, and i 2 f0; 1g,
2. G is a graph in B0nwith bipartition G0and G1, and
3. F EðGÞ with jFj ¼ f .
Proof. ByTheorem 3, there is a hamiltonian path P ¼ hx; P1;u; P2;yi of G F joining x to y. Then fP1;P2g is the spanning
ðu; fx; ygÞ-fan of G F. h The following are the main results. Theorem 6. Suppose that
1. n P 2; k 6 n, and i 2 f0; 1g,
2. G is a graph in B0nwith bipartition G0¼ ðV0;E0Þ and G1¼ ðV1;E1Þ, and
3. F EðGÞ with jFj þ k 6 n and jFj 6 n 2.
Then, for any u 2 Viand S # VðGÞ fug with jSj ¼ k and jS \ V1ij ¼ 1, there exists a spanning ðu; SÞ-fan in G F.
We prove the theorem by induction. However, the proof of the theorem is rather long. We prove it in the following section.
Theorem 7. The bipartite n-dimensional hypercube-like graph B0
nis f-edge fault k
-laceable for f 6 n 2 and f þ k 6 n. Proof. Let G be a graph in B0
nwith bipartition G0and G1. Assume that x 2 VðGiÞ and y 2 VðG1iÞ for some i 2 f0; 1g. Suppose
that F EðGÞ with jFj ¼ f and f 6 n 2. Let S # VðGiÞ fxg adjacent to y in G F with jSj ¼ k 1 and k 6 n f . We assume
that S ¼ fy1;y2; . . . ;yk1g. ByTheorem 6, there exists a spanning ðx; S [ fygÞ-fan fP1;P2; . . . ;Pkg in G F such that Pkjoins x to
y, and Pijoins x to yifor 1 6 i 6 k 1. Let Qi¼ hx; Pi;yi;yi for 1 6 i 6 k 1. Thus, fPk;Q1;Q2; . . . ;Qk1g forms a k-container
between x and y in G F. The theorem is proved. h
4. Proof of Theorem 6
Let G ¼ G0 G1in B0nwith bipartition V j 0and V j 1for j 2 f0; 1g. Thus, V 0 0[ V 1 0and V 0 1[ V 1
1form the bipartition of G. Assume
that jFj ¼ f . Let u be any node in V00[ V 1
0and S ¼ f
v
1;v
2; . . . ;v
kg be any node subset in G fug withv
1being the unique nodein ðV01[ V 1
1Þ \ S. Without loss of generality, we assume that u 2 V 0
0. For n ¼ 2, we have G is isomorphic to a cycle with four
nodes. Thus, this statement holds on n ¼ 2. ByLemma 1, Theorem 3, and Theorem 5, this statement holds on n ¼ 3. Thus, we assume that n P 4. ByLemma 1 and Theorem 3, this statement holds on k 2 f1; 2g and f ¼ n 2. ByTheorem 5, this statement holds on k 6 n and f ¼ 0. Thus, we assume that k P 3 and 1 6 f 6 n 3 with k þ f 6 n. We set T ¼ S f
v
1g; Fj¼ F \ EðGjÞ for j 2 f0; 1g, and F2¼ F ðF0[ F1Þ. Note that jFj ¼ jF0j þ jF1j þ jF2j and jFj0j 6 n 3 for everyj0 2 f0; 1; 2g. Now we have the following cases. Case 1. jT \ V0 0j ¼ jTj.
v
1v
2u
v
3v
4v
5v
6x
G
1v
2v
3u
v
4v
5v
1v
6x
G
1u
x
x
u
v
2v
3u
v
4v
5v
1v
6x
G
y
1y
u
x
(a)
(b)
(c)
(d)
v
2v
3u
v
4v
5v
1v
6x
y
y
x
G
1G
0- F
0G
0- F
0G
0- F
0G
0- F
0Case 1.1. jF0j ¼ jFj and
v
12 V01. Let H ¼ S fv
kg. We have H G0;jH \ V01j ¼ 1, and jHj ¼ k 1. By induction, there is aspan-ning ðu; HÞ-fan fP1;P2; . . . ;Pk1g of G0 F0. Without loss of generality, we assume that Pijoins u to
v
ifor every 1 6 i 6 k 1.Suppose that
v
k2 VðP1Þ. Without loss of generality, we write P1 as hu; Q1;v
k;x; Q2;v
1i. Sincev
k2 V00;x 2 V 01. (Note that
x ¼
v
1 if lðQ2Þ ¼ 0.) By Theorem 1, there is a hamiltonian path R of G1 joining node u 2 V11 to node x 2 V 10. We set
W1¼ hu; u; R; x; x; Q2;
v
1i; Wi¼ Pifor every 2 6 i 6 k 1, and Wk¼ hu; Q1;v
ki. Then fW1;W2; . . . ;Wkg forms the spanningðu; SÞ-fan of G F. SeeFig. 1(a) for an illustration. Suppose that
v
k2 VðPiÞ for some 2 6 i 6 k 1. Without loss of generality,we assume that
v
k2 VðPk1Þ and write Pk1as hu; Q1;v
k;x; Q2;v
k1i. Sincev
k2 V00;x 2 V 01. ByTheorem 1, there is a
hamilto-nian path R of G1joining node u 2 V11to node x 2 V 1
0. We set Wi¼ Pifor every 1 6 i 6 k 2; Wk1¼ hu; u; R; x; x; Q2;
v
k1i, andWk¼ hu; Q1;
v
ki. Then fW1;W2; . . . ;Wkg forms the spanning ðu; SÞ-fan of G F. SeeFig. 1(b) for an illustration.Case 1.2. jF0j ¼ jFj and
v
12 V11. We choose a node x in V01. Let H ¼ ðT [ fxgÞ fv
kg. We have H G0;jH \ V01j ¼ 1, and jHj ¼ k 1. By induction, there is a spanning ðu; HÞ-fan fP1;P2; . . . ;Pk1g of G0 F0such that P1joins u to x, and Pijoins u tov
i for every 2 6 i 6 k 1. Note that u 2 V11 and x 2 V 10. Without loss of generality, we assume that
v
k2 VðP1Þ. Let P1¼ hu; Q1;y;v
k;Q2;xi. Sincev
k2 V00, we have y 2 V01and y 2 V10.Suppose that
v
1– u. ByTheorem 2, there are two disjoint paths R1and R2in G1such that ð1Þ R1joins y tov
1;ð2Þ R2joins uto x, and ð3Þ R1[ R2spans G1. We set W1¼ hu; Q1;y; y; R1;
v
1i; Wi¼ Pifor every 2 6 i 6 k 1, and Wk¼ hu; u; R2; x; x; Q12 ;v
ki.Then fW1;W2; . . . ;Wkg forms the spanning ðu; SÞ-fan of G F. SeeFig. 1(c) for an illustration.
Suppose that
v
1¼ u. By Theorem 4, there is a hamiltonian path R of G1 fv
1g joining y to x. We setW1¼ hu; u ¼
v
1i; Wi¼ Pifor every 2 6 i 6 k 1, and Wk¼ hu; Q1;y; y; R; x; x; Q12 ;v
ki. Then fW1;W2; . . . ;Wkg forms thespan-ning ðu; SÞ-fan of G F. SeeFig. 1(d) for an illustration.
Case 1.3. jF0j < jFj and
v
12 V01. Since jF0j < jFj ¼ f , we have k þ jF0j 6 k þ f 1 6 n 1. By induction, there is a spanning ðu; SÞ-fan fP1;P2; . . . ;Pkg of G0 F0. Without loss of generality, we assume that Pi joins u tov
i for every 1 6 i 6 k. Since jVðG0Þj ¼ 2n1and [ki¼1Pispan G0, we have Pk
i¼1jEðPiÞj ¼ 2n1 1. Since 2n1 1 > 3n 8 > 2ðf 1Þ þ k if n P 3, there exists an edge ðx; yÞ in [k
i¼1EðPiÞ such that ðx; xÞ R F2and ðy; yÞ R F2. Without loss of generality, we assume that ðx; yÞ 2 EðPjÞ for some 1 6 j 6 k. Let Pj¼ hu; R1;x; y; R2;
v
ji. Note that u ¼ x if lðR1Þ ¼ 0 and y ¼v
jif lðR2Þ ¼ 0. Since x and y are adjacent, x and y are in distinct bipartition of G0. Moreover, x and y are in distinct bipartition of G1. ByTheorem 3, there is a hamiltonian path W of G1 F1 joining x to y. We set Wi¼ Pi for every i 2 f1; 2; . . . ; kg fjg and set Wj¼ hu; R1;x; x; W; y; y; R2;v
ji. Then fW1;W2; . . . ;Wkg forms the spanning ðu; SÞ-fan of G F.Case 1.4. jF0j < jFj and
v
12 V11. Since jV 0 1j ¼ 2n2>n 3 if n P 3, there exists a node x 2 V0
1 such that ðx; xÞ R F2. Let
H ¼ T [ fxg. Since jF0j < jFj ¼ f , we have k þ jF0j 6 k þ f 1 6 n 1. By induction, there is a spanning ðu; HÞ-fan
fP1;P2; . . . ;Pkg of G0 F0. Without loss of generality, we assume that P1is joining u to x and Piis joining u to
v
ifor every2 6 i 6 k. Since x 2 V0
1, we have x 2 V 1
0. By Theorem 3, there is a hamiltonian path R of G1 F1joining x to
v
1. We setW1¼ hu; P1;x; x; R;
v
1i and Wi¼ Pifor every 2 6 i 6 k. Then fW1;W2; . . . ;Wkg forms the spanning ðu; SÞ-fan of G F.Case 2. jT \ V10j ¼ 1. We assume that
v
k2 V10. Note that u 2 V 1 1.Case 2.1. jF0j ¼ jFj and
v
12 V01. Let H ¼ S fv
kg. We have H G0;jH \ V01j ¼ 1, and jHj ¼ k 1. By induction, there is aspan-ning ðu; HÞ-fan fW1;W2; . . . ;Wk1g of G0 F0. ByTheorem 1, there is a hamiltonian path R of G1joining u to
v
k. We setWk¼ hu; u; R;
v
ki. Then fW1;W2; . . . ;Wkg forms the spanning ðu; SÞ-fan of G F. SeeFig. 2(a) for an illustration.Case 2.2. jF0j ¼ jFj and
v
12 V11. By Theorem 1, there is a hamiltonian path R of G1 joiningv
1 tov
k. We write R ash
v
1;R1; u; x; R2;v
ki. Note thatv
1¼ u if lðR1Þ ¼ 0 and x ¼v
k if lðR2Þ ¼ 0. Since u 2 V11, we have x 2 V 10 and x 2 V 0 1. Let
H ¼ ðT [ fxgÞ f
v
kg. Thus H G0;jH \ V10j ¼ 1, and jHj ¼ k 1. By induction, there is a spanning ðu; HÞ-fanfP1;P2; . . . ;Pk1g of G0 F0 such that P1 joins u to x, and Pi joins u to
v
i for every 2 6 i 6 k 1. We setv
1v
2u
v
3v
4v
5v
6G
v
2v
3u
v
4v
5v
6v
1x
G
(a)
(b)
x
u
u
1 1G
0- F
0G
0- F
0W1¼ hu; u; R11 ;
v
1i; Wi¼ Pifor every 2 6 i 6 k 1, and Wk¼ hu; P1; x; x; R2;v
ki. Then fW1;W2; . . . ;Wkg forms the ðu; SÞ-fan ofG F. SeeFig. 2(b) for an illustration. Case 2.3. jF0j < jFj and
v
12 V01. Since jV0 0j ¼ 2
n2>n > k þ f 1 if n P 4, there exists a node x in V0
0 ðT [ fugÞ such that
ðx; xÞ R F2. Let H ¼ ðS [ fxgÞ f
v
kg. Obviously, H G0;jH \ V10j ¼ 1, and jHj ¼ k. Since jF0j < jFj ¼ f , we havek þ jF0j 6 k þ f 1 6 n 1. By induction, there is a spanning ðu; HÞ-fan fP1;P2; . . . ;Pkg of G0 F0such that Pijoins u to
v
ifor every 1 6 i 6 k 1 and Pkjoins u to x. ByTheorem 1, there is a hamiltonian path R of G1 F1joining x to
v
k. We setWi¼ Pifor every 1 6 i 6 k 1 and Wk¼ hu; Pk;x; x; R;
v
ki. Then fW1;W2; . . . ;Wkg forms the spanning ðu; SÞ-fan of G F.Case 2.4. jF0j < jFj and
v
12 V11. Since jV 0 0j ¼ 2n2>n > k þ f 1 if n P 4, there exists a node x in V0
0 ðT [ fugÞ such that
ðx; xÞ R F2. Let F0¼ fðy; xÞjy 2 G1 and ðy; yÞ 2 F2g. We have jF1[ F0j 6 jF1j þ jF2j 6 jFj ¼ f < n 3. ByTheorem 1, there is a
hamiltonian path R of G1 ðF1[ F0Þ joining
v
1tov
k. Without loss of generality, we write R as hv
1;R1; x; z; R2;v
ki. Note thatv
1¼ x if lðR1Þ ¼ 0 and z ¼v
kif lðR2Þ ¼ 0. Since x 2 V00, we have x 2 V 1 1;z 2 V 1 0, and z 2 V 0 1. Let H ¼ ðT [ fx; zgÞ fv
kg.Obvi-ously, H G0;jH \ V01j ¼ 1, and jHj ¼ k. Since jF0j < jFj ¼ f , we have k þ jF0j 6 k þ f 1 6 n 1. By induction, there is a
span-ning ðu; HÞ-fan fP1;P2; . . . ;Pkg of G0 F0. Without loss of generality, we assume that P1joins u to x; P2joins u to z, and Pijoins
u to
v
i1 for every 3 6 i 6 k. We set W1¼ hu; P1;x; x; R11 ;v
1i; Wi¼ Piþ1for every 2 6 i 6 k 1, and Wk¼ hu; P2; z; z; R2;v
ki.Then fW1;W2; . . . ;Wkg forms the ðu; SÞ-fan of G F.
Case 3. jT \ V1
0j P 2 and jT \ V 0
0j P 1. We have n P k þ 1 ¼ jSj þ 1 P 5. Assume that A ¼ T \ V 0
0¼ f
v
2;v
3; . . . ;v
tg andB ¼ T \ V1
0¼ f
v
tþ1;v
tþ2; . . . ;v
kg for some 2 6 t 6 k 2.Case 3.1. jF0j ¼ jFj. Since ðn 1ÞjAj 6 ðn 1Þðn 3Þ < 2nn2if n P 5, there exists a node x in V10such that
v
1 R NG1ðxÞ and
v
i R NG1ðxÞ for 2 6 i 6 t. By induction, there is a spanning ðx; B [ fugÞ-fan fP1;P2; . . . ;Pktþ1g of G1 such thatP1¼ hx; x1;P01; ui joins x to u, and Pi¼ hx; xi;P0i;
v
tþi1i joins x tov
tþi1for every 2 6 i 6 k t þ 1.Case 3.1.1.
v
12 V01. We set H ¼ A [ fv
1g [ fxij2 6 i 6 k tg. Let fQ1;Q2; . . . ;Qk1g be a spanning ðu; HÞ-fan of G0 F0suchthat Qi joins u to
v
i for every 1 6 i 6 t, and Qj joins u to xjtþ1 for every t þ 1 6 j 6 k 1. We setWi¼ Qi;Wj¼ hu; Qj; xjtþ1;xjtþ1;P0jtþ1;
v
ji, and Wk¼ hu; u; P11 ;x; Pktþ1;v
ki. Then fW1;W2; . . . ;Wkg forms a spanningðu; SÞ-fan of G F.
Case 3.1.2.
v
12 V11 andv
12 VðP1Þ. We write P1¼ hx; R1;y;v
1;R2; ui. We set H ¼ A [ fxij2 6 i 6 k tg [ fyg. LetfQ1;Q2; . . . ;Qk1g be a spanning ðu; HÞ-fan of G0 F0such that Q1joins u to y; Qjjoins u to
v
jfor every 2 6 j 6 t, and Qj0joinsu to xj0tþ1 for every t þ 1 6 j0 6 k 1. We set W1¼ hu; u; R21;
v
1i; Wj¼ Qj, Wj0¼ hu; Qj0; xj0tþ1;xj0tþ1;P0j0tþ1;v
j0i, andWk¼ hu; Q1; y; y; R11 ;x; Pktþ1;
v
ki. Then fW1;W2; . . . ;Wkg forms a spanning ðu; SÞ-fan of G F.Case 3.1.3.
v
12 V11andv
12 VðPiÞ for some 2 6 i 6 k t þ 1. Without loss of generality, we assume thatv
12 VðP2Þ. LetP2¼ hx; R1;
v
1;y; R2;v
tþ1i. We set H ¼ A [ fxij3 6 i 6 k t þ 1g [ fyg. Let fQ1;Q2; . . . ;Qk1g be a spanning ðu; HÞ-fan ofG0 F0such that Q1joins u to y, Qj joins u to
v
j for every 2 6 j 6 t, and Qj0 joins u to xj0tþ1 for every t þ 2 6 j0 6 k. Weset W1¼ hu; u; P11 ;x; R1;
v
1i; Wj¼ Qj, Wtþ1¼ hu; Q1; y; y; R2;v
tþ1i, and Wj0¼ hu; Qj0; xj0tþ1;xj0tþ1;P0j0tþ1;v
j0i. ThenfW1;W2; . . . ;Wkg forms a spanning ðu; SÞ-fan of G F.
Case 3.2. jF0j < jFj and n ¼ 5. We have jFj ¼ 1 and k ¼ 4. Thus, jF0j = 0 and jF1j þ jF2j ¼ 1. Moreover, A ¼ f
v
2g andB ¼ f
v
3;v
4g.Case 3.2.1. jF1j ¼ 0 and
v
12 V01. Since jV 11j ¼ 8 > 3, there exist two distinct nodes x1 and x2 in V11 fu;
v
2g such thatðx1; x1Þ R F2and ðx2; x2Þ R F2. ByTheorem 2, there are two disjoint paths P1and P2in G1such that Pijoins xi to
v
iþ2fori 2 f1; 2g, and P1[ P2spans G1. Let fQ1;Q2;Q3;Q4g be a spanning ðu; f
v
1;v
2; x1; x2gÞ-fan of G0such that Qijoins u tov
ifor1 6 i 6 2, and Qj joins u to xj2 for 3 6 j 6 4. We set Wi¼ hu; Qiþ2; xi;xi;Pi;
v
iþ2i for every 1 6 i 6 2. Then fQ1;Q2;W1;W2gforms a spanning ðu; SÞ-fan of G F. Case 3.2.2. jF1j ¼ 0 and
v
12 V11. Since jV1
0j ¼ 8, there exists a node x in V 1
0 f
v
3;v
4g such thatv
2 R NG1ðxÞ and ðx; xÞ R F2.We set F0 ¼ fðx; yÞjy 2 NG1ðxÞ and ðy; yÞ 2 F2g. We have jF0j 6 1. By induction, there is a spanning ðx; f
v
1;v
3;v
4gÞ-fanfP1;P2;P3g of G1 such that P1 joins x to
v
1 and Pi joins x tov
iþ1 for 2 6 i 6 3. We set P1¼ hx; x1;R1;v
1i andPi¼ hx; xi;Ri;
v
iþ1i for every 2 6 i 6 3. Let fQ1;Q2;Q3;Q4g be a spanning ðu; fv
2; x; x2; x3gÞ-fan of G0such that Q1joins u to
x; Q2 joins u to
v
2, and Qi joins u to xi1 for 3 6 i 6 4. We set W1¼ hu; Q1; x; x; P1;v
1i; W2¼ Q2, andCase 3.2.3. jF1j ¼ 1 and
v
12 V01. We have jF2j ¼ 0. ByTheorem 3, there is a hamiltonian path P of G1 ðF1[ fðv
3;v
2ÞgÞjoin-ing u to
v
4. We set P ¼ hu; P1;v
3;x; P2;v
4i. Let fQ1;Q2;Q3g be a spanning ðu; fv
1;v
2; xgÞ-fan of G0such that Qijoins u tov
ifor1 6 i 6 2 and Q3joins u to x. Let W1¼ hu; u; P1;
v
3i and W2¼ hu; Q3; x; x; P2;v
4i. Then fQ1;Q2;W1;W2g forms a spanningðu; SÞ-fan of G F.
Case 3.2.4. jF1j ¼ 1 and
v
12 V11. Since jV 10j ¼ 8 > 6, there exists a node x in V 1
0 f
v
3;v
4g such thatv
2 R NG1ðxÞ. By induction,there exists a ðx; f
v
1;v
3;v
4gÞ-fan fP1;P2;P3g of G1 F1such that P1joins x tov
1, and Pijoins x tov
iþ1for 2 6 i 6 3. Withoutloss of generality, we write Pi¼ hx; yi1;Ri1;
v
i1i for 2 6 i 6 3. Let fQ1;Q2;Q3;Q4g be a ðu; fx;v
2; y1; y2gÞ-fan of G0such thatQ1 joins u to x; Q2 joins u to
v
2, and Qi joins u to yi2 for 3 6 i 6 4. We set W1¼ hu; Q1; x; x; P1;v
1i; W2¼ Q2, andWi¼ hu; Qi; yi2;yi2;Ri2;
v
ii for 3 6 i 6 4. Then fW1;W2;W3;W4g forms a spanning ðu; SÞ-fan of G F.Case 3.3. jF0j < jFj and n P 6. Since ðn 1Þðf þ jAjÞ 6 ðn 1Þðf þ k 3Þ 6 ðn 1Þðn 3Þ < 2n2if n P 6, there exists a node x
in V1
0such that ðx; xÞ R F2;
v
i R NG1ðxÞ for every 2 6 i 6 t, and ðy; yÞ R F2for every y 2 NG1ðxÞ.Case 3.3.1.
v
12 V01. Since jAj þ f < 2 n2if n P 6, there exists a node y in V11such that ðy; yÞ R F2and y R A.
Suppose that x R B. By induction, there is a spanning ðx; B [ fygÞ-fan fP1;P2; . . . ;Pktþ1g of G1 F1such that Pijoins x to
v
tþifor every 1 6 i 6 k t and Pktþ1joins x to y. Without loss of generality, we set Pi¼ hx; xi;Ri;
v
tþii for every 1 6 i 6 k t 1.We set H ¼ A [ fxij1 6 i 6 k t 1g [ fyg. Let fQ1;Q2; . . . ;Qkg be a spanning ðu; HÞ-fan of G0 F0such that Qijoins u to
v
ifor every 1 6 i 6 t; Qj joins u to xit for every t þ 1 6 j 6 k 1, and Qk joins u to y. We set Wi¼ Qi for every
1 6 i 6 t; Wj¼ hu; Qj; xjt;xjt;Rjt;
v
jg for every t þ 1 6 j 6 k 1, and Wk¼ hu; Qk; y; y; Pktþ11 ;x; Pkt;v
ki. ThenfW1;W2; . . . ;Wkg forms a spanning ðu; SÞ-fan of G F.
Suppose that x 2 B. We assume that x ¼
v
k. This case is similar to the above.Case 3.3.2.
v
12 V11. Suppose that x R B. By induction, there is a spanning ðx; B [ fv
1gÞ-fan fP1;P2; . . . ;Pktþ1g of G1 F1such that Pijoins x tov
tþifor every 1 6 i 6 k t and Pktþ1joins x tov
1. Without loss of generality, we set Pi¼ hx; xi;Ri;v
tþii for every 1 6 i 6 k t. We set H ¼ A [ fxij1 6 i 6 k tg [ fxg. Let fQ1;Q2; . . . ;Qkg be a spanning ðu; HÞ-fan of G0 F0such that Q1 joins u to x; Qi joins u tov
i for every 2 6 i 6 t, and Qj joins u to xit for every t þ 1 6 j 6 k. We set W1¼ hu; Q1; x; x; Pktþ1;v
1i; Wi¼ Qi for every 2 6 i 6 t, and Wj¼ hu; Qj; xjt;xjt;Rjt;v
ji for every t þ 1 6 j 6 k. Then fW1;W2; . . . ;Wkg forms a spanning ðu; SÞ-fan of G F.Suppose that x 2 B. We assume that x ¼
v
k. The case is similar to the above.Case 4. jT \ V1
0j ¼ k 1. WithCase 2, we consider that jTj P 2.
Case 4.1. jF0j ¼ jFj. We have jF1j ¼ 0 and jF2j ¼ 0. Let x be a node in V10 T. By induction, there exists a spanning ðx; T [ f
ugÞ-fan fP1;P2; . . . ;Pkg of G1such that P1joins x to u, and Pijoins x to
v
ifor every 2 6 i 6 k. We set Pi¼ hx; xi;Ri;v
ii for every2 6 i 6 k.
Case 4.1.1.
v
12 V01. We set H ¼ fv
1g [ fxij2 6 i 6 k 1g. Let fQ1;Q2; . . . ;Qk1g be a spanning ðu; HÞ-fan of G0 F0such thatQ1joins u to
v
1, and Qijoins u to xifor every 2 6 i 6 k 1. We set W1¼ Q1;Wi¼ hu; Qi; xi;xi;Ri;v
ii for every 2 6 i 6 k 1,and Wk¼ hu; u; P11 ;x; Pk;
v
ki. Then fW1;W2; . . . ;Wkg forms a spanning ðu; SÞ-fan of G F.Case 4.1.2.
v
12 V11andv
12 VðP1Þ. We set P1¼ hx; Z1;y;v
1;Z2; ui. Let H ¼ fyg [ fxij2 6 i 6 k 1g. Thus, there exists aspan-ning ðu; HÞ-fan fQ1;Q2; . . . ;Qk1g in G0 F0such that Q1joins u to y, and Qi joins u to xifor every 2 6 i 6 k 1. We set
W1¼ hu; u; Z12 ;
v
1i; Wi¼ hu; Qi; xi;xi;Ri;v
ii for every 2 6 i 6 k 1, and Wk¼ hu; Q1; y; y; Z11 ;x; Pk;v
ki. ThenfW1;W2; . . . ;Wkg forms a spanning ðu; SÞ-fan of G F.
Case 4.1.3.
v
12 V11 andv
12 VðPiÞ for some 2 6 i 6 k. Without loss of generality, we assume thatv
12 VðPkÞ. We setPk¼ hx; Z1;
v
1;y; Z2;v
ki. Let H ¼ fyg [ fxij2 6 i 6 k 1g. Thus, there exists a spanning ðu; HÞ-fan fQ1;Q2; . . . ;Qk1g inG0 F0 such that Q1 joins u to y, and Qi joins u to xi for every 2 6 i 6 k 1. We set
W1¼ hu; u; P11 ;x; Z1;
v
1i; Wi¼ hu; Qi; xi;xi;Ri;v
ii for every 2 6 i 6 k 1, and Wk¼ hu; Q1; y; y; Z2;v
ki. Then fW1;W2; . . . ;Wkgforms a spanning ðu; SÞ-fan of G F.
Case 4.2. jF0j < jFj and jF1j < jFj. Since jV10j ¼ 2 n2>
ðn 1Þðn 4Þ þ ðn 2Þ P ðn 1ÞjFj þ ðk 2Þ if n P 4, there exists a node x in V1
Case 4.2.1.
v
12 V01. There exists a node y 2 V 11 fx; ug such that ðy; yÞ R F2. Let fP1;P2; . . . ;Pkg be a spanning ðx; T [ fygÞ-fan
of G1 F1such that P1joins x to y and Pijoins x to
v
ifor every 2 6 i 6 k. We set Pi¼ hx; xi;Ri;v
ii for every 2 6 i 6 k. Sincejfðz; zÞjz 2 NG1ðxÞg \ F2j 6 1, we assume that ðxi; xiÞ R F2for every 2 6 i 6 k 1. We set H ¼ f
v
1; yg [ fxij2 6 i 6 k 1g. Byinduction, there is a spanning ðu; HÞ-fan fQ1;Q2; . . . ;Qkg of G0 F0such that Q1joins u to
v
1, Qi joins u to xi for every2 6 i 6 k 1, and Qk joins u to y. Let W1¼ Q1;Wi¼ hu; Qi; xi;xi;Ri;
v
ii for every 2 6 i 6 k 1, andWk¼ hu; Qk; y; y; P11 ;x; Pk;
v
ki. Then fW1;W2; . . . ;Wkg forms a spanning ðu; SÞ-fan of G F.Case 4.2.2.
v
12 V11. Let fP1;P2; . . . ;Pkg be a spanning ðx; SÞ-fan of G1 F1such that Pijoins x tov
ifor every 1 6 i 6 k. We setPi¼ hx; xi;Ri;
v
ii for every 1 6 i 6 k. We choose an index r in f1; 2; . . . ; k 1g such that ðxi; xiÞ R F2 for everyi 2 f1; 2; . . . ; k 1g frg. We set H ¼ fxiji 2 f1; 2; . . . ; k 1g frgg [ fxg. By induction, there is a spanning ðu; HÞ-fan
fQ1;Q2; . . . ;Qkg of G0 F0 such that Qi joins u to xi for every i 2 f1; 2; . . . ; kg frg, and Qr joins u to x. Let
Wi¼ hu; Qi; xi;xi;Ri;
v
ii for every i 2 f1; 2; . . . ; kg frg and Wr¼ hu; Qr; x; x; Pr;v
ri. Then fW1;W2; . . . ;Wkg forms a spanningðu; SÞ-fan of G F.
Case 4.3. jF1j ¼ jFj. We have jF0j ¼ 0 and jF2j ¼ 0. Let x1be a node in V11 fu;
v
1g. ByTheorem 3, there is a hamiltonian path Pof G1 F1joining x1to
v
k. We set P ¼ hx1;P1;v
2;x2;P3;v
3; . . . ;xk1;Pk1;v
ki. Note that xi2 V11for every 1 6 i 6 k 1.Case 4.3.1.
v
12 V01. We set H ¼ fv
1g [ fxij1 6 i 6 k 1g. By induction, there is a spanning ðu; HÞ-fan fQ1;Q2; . . . ;Qkg of G0suck that Q1joins u to
v
1and Qijoins u to xi1for every 2 6 i 6 k. We set W1¼ Q1and Wi¼ hu; Qi; xi1;xi1;Pi1;v
ii for every2 6 i 6 k. Then fW1;W2; . . . ;Wkg forms a spanning ðu; SÞ-fan of G F.
Case 4.3.2.
v
12 V11. Suppose thatv
1 is a node in VðPtÞ for some 1 6 t 6 k. We write Pt¼ hxt;R1;v
1;xk;R2;v
ti. LetH ¼ fxij1 6 i 6 kg. By induction, there exists a spanning ðu; HÞ-fan fQ1;Q2; . . . ;Qkg of G0where Qijoins u to xi1for every
1 6 i 6 k. For every j 2 f1; 2; . . . ; k 1g ftg, we set Wj¼ hu; Qj; xj;xj;Pj;
v
jþ1i. Let Wt¼ hu; Qt; xt;xt;R1;v
1i andWk¼ hu; Qk; xk;xk;R2;
v
ti. Thus, fW1;W2; . . . ;Wkg forms a spanning ðu; SÞ-fan of G F.5. Conclusion
Computer network topologies are usually represented by graphs where nodes represent processors and edges represent links between processors[28]. In practice, the processors or links in a network may be failure. Thus the fault-tolerant prop-erty become an important issue on network topologies. Many results have been proposed in literature[29–32,26]. In this paper, we have shown that n-dimensional bipartite hypercube-like graphs are f-edge fault k-laceable for every f 6 n 2 and f þ k 6 n. Future work will try to study the fault-tolerant k-connectivity and k-laceability for some super spanning
con-nected graphs and super spanning laceable graphs, respectively. References
[1] S.B. Akers, B. Krishnamurthy, A group-theoretic model for symmetric interconnection networks, IEEE Transactions on Computers 38 (1989) 555–566. [2] D.F. Hsu, On container width and length in graphs, groups, and networks, IEICE Transactions Fundamentals E77-A (1994) 668–680.
[3] J.A. Bondy, U.S.R. Murty, Graph Theory with Applications, North Holland, New York, 1980. [4] L.H. Hsu, C.K. Lin, Graph Theory and Interconnection Networks, CRC Press, New York, 2008. [5] O. Ore, Hamiltonian connected graphs, Journal of Mathematic Pures Application 42 (1963) 21–27.
[6] G.J. Simmons, Almost all n-dimensional rectangular lattices are hamilton-laceable, in: Proceedings of the Ninth Southeastern Conference on Combinatorics, Graph Theory, and Computing, 1978, pp. 649–661.
[7] K. Menger, Zur allgemeinen kurventheorie, Fundamentale Mathematik 10 (1927) 95–115.
[8] M. Albert, R.E.L. Aldred, D. Holton, On 3-connected graphs, Australasian Journal of Combinatorics 24 (2001) 193–207.
[9] Y.H. Teng, J.M. Tan, L.H. Hsu, The globally bi-3-connected property of the honeycomb rectangular torus, Information Sciences 177 (2007) 5573–5589.
[10] C.H. Chang, C.K. Lin, J.J.M. Tan, H.M. Huang, L.H. Hsu, The super spanning connectivity and super spanning laceability of the enhanced hypercubes, The Journal of Supercomputing 48 (2009) 66–87.
[11] C.K. Lin, H.M. Huang, L.H. Hsu, The super connectivity of the pancake graphs and the super laceability of the star graphs, Theoretical Computer Science 339 (2005) 257–271.
[12] C.K. Lin, J.J.M. Tan, D.F. Hsu, L.H. Hsu, On the spanning connectivity and spanning laceability of hypercube-like networks, Theoretical Computer Science 381 (2007) 218–229.
[13] Y.H. Teng, T.L. Kung, L.H. Hsu, The 3*-connected property of pyramid networks, Computers & Mathematics with Applications 60 (2010) 2360–2363. [14] C.H. Chang, C.K. Lin, H.M. Huang, L.H. Hsu, The super laceability of the hypercube, Information Processing Letters 92 (2004) 15–21.
[15] F.T. Leighton, Introduction to parallel algorithms and architectures: arrays trees hypercubes, Morgan Kaufmann, San Mateo, CA, 1992.
[16] S.A. Mane, B.N. Waphare, Regular connected bipancyclic spanning subgraphs of hypercubes, Computers & Mathematics with Applications 62 (2011) 3551–3554.
[17] M.C. Yang, Bipanconnectivity of balanced hypercubes, Computers & Mathematics with Applications 60 (2010) 1859–1867.
[18] S. Abraham, K. Padmanabhan, The twisted cube topology for multiprocessors: a study in network asymmetry, Journal of Parallel and Distributed Computing 13 (1991) 104–110.
[19] P. Cull, S.M. Larson, The Möbius cubes, IEEE Transactions on Computers 44 (1995) 647–659.
[20] K. Efe, A variation on the hypercube with lower diameter, IEEE Transactions on Computers 40 (1991) 1312–1316.
[22] A.S. Vaidya, P.S.N. Rao, S.R. Shankar, A class of hypercube-like networks, in: Proceedings of the Fifth IEEE Symposium on Parallel and Distributed Processing, IEEE Computer Society Press, Los Alamitos, CA, 1993, pp. 800–803.
[23] Y.C. Chen, J.J.M. Tan, L.H. Hsu, S.S. Kao, Super-connectivity and super-edge-connectivity for some interconnection networks, Applied Mathematics and Computation 140 (2003) 245–254.
[24] Y.C. Chen, C.H. Tsai, L.H. Hsu, J.J.M. Tan, Super fault-tolerant hamiltonian graphs, Applied Mathematics and Computation 148 (2004) 729–741. [25] C.D. Park, K.Y. Chwa, Hamiltonian properties on the class of hypercube-like networks, Information Processing Letters 91 (2004) 11–17.
[26] C.K. Lin, T.Y. Ho, J.J.M. Tan, L.H. Hsu, Fault-tolerant hamiltonian laceability and fault-tolerant conditional hamiltonian for bipartite hypercube-like networks, Journal of Interconnection Networks 10 (2009) 243–251.
[27] G.A. Dirac, In abstrakten Graphen vorhandene vollstandige 4-Graphen und ihre Unterteilungen, Mathematische Nachrichten 22 (1960) 61–85. [28] T.J. Lin, S.Y. Hsieh, J.S.T. Juan, Embedding cycles and paths in product networks and their applications to multiprocessor systems, IEEE Transactions on
Parallel and Distributed Systems 23 (2012) 1081–1089.
[29] D.R. Duh, Y.C. Chen, R.Y. Wu, Proof that pyramid networks are 1-hamiltonian-connected with high probability, Information Sciences 177 (2007) 4188– 4199.
[30] S.Y. Hsieh, Y.R. Cian, Conditional edge-fault hamiltonicity of augmented cubes, Information Sciences 180 (2010) 2596–2617.
[31] C.W. Huang, H.L. Huang, S.Y. Hsieh, Edge-bipancyclicity of star graphs with faulty elements, Theoretical Computer Science 412 (2011) 6938–6947. [32] C.W. Lee, S.Y. Hsieh, Pancyclicity of matching composition networks under the conditional fault model, IEEE Transactions on Computers 61 (2012)