Graph theory and interconnection networks 1st

Edition Lih-Hsing Hsu pdf download

https://ebookgate.com/product/graph-theory-and-interconnection-
networks-1st-edition-lih-hsing-hsu/

Get Instant Ebook Downloads – Browse at https://ebookgate.com

Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

Fixed Point Theory and Graph Theory Foundations and

Integrative Approaches 1st Edition Monther Alfuraidan

https://ebookgate.com/product/fixed-point-theory-and-graph-theory-
foundations-and-integrative-approaches-1st-edition-monther-alfuraidan/

ebookgate.com

Chromatic graph theory Second Edition Gary Chartrand

https://ebookgate.com/product/chromatic-graph-theory-second-edition-
gary-chartrand/

ebookgate.com

Introductory Graph Theory with Applications 1st Edition

Fred Buckley

https://ebookgate.com/product/introductory-graph-theory-with-
applications-1st-edition-fred-buckley/

ebookgate.com

Near Rings Fuzzy Ideals and Graph Theory 1st Edition

Bhavanari Satyanarayana

https://ebookgate.com/product/near-rings-fuzzy-ideals-and-graph-
theory-1st-edition-bhavanari-satyanarayana/

ebookgate.com
Introduction to Graph Theory 5th Edition Robin J. Wilson

https://ebookgate.com/product/introduction-to-graph-theory-5th-
edition-robin-j-wilson/

ebookgate.com

Introduction to Graph Theory Solutions Manual 1st Edition

Koh Khee Meng

https://ebookgate.com/product/introduction-to-graph-theory-solutions-
manual-1st-edition-koh-khee-meng/

ebookgate.com

Discrete Mathematics with Graph Theory 2nd Edition Edgar

G. Goodaire

https://ebookgate.com/product/discrete-mathematics-with-graph-
theory-2nd-edition-edgar-g-goodaire/

ebookgate.com

Fractional Graph Theory A Rational Approach to the Theory

of Graphs 1st Edition Edward R. Scheinerman

https://ebookgate.com/product/fractional-graph-theory-a-rational-
approach-to-the-theory-of-graphs-1st-edition-edward-r-scheinerman/

ebookgate.com

Introduction to Graph Theory 2, 2002 reprint Edition

Douglas B. West

https://ebookgate.com/product/introduction-to-graph-
theory-2-2002-reprint-edition-douglas-b-west/

ebookgate.com
prelims.tex 11/8/2008 12: 54 Page i
prelims.tex 11/8/2008 12: 54 Page ii
prelims.tex 11/8/2008 12: 54 Page iii
prelims.tex 11/8/2008 12: 54 Page iv
prelims.tex 11/8/2008 12: 54 Page v

Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
Authors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

Chapter 1: Fundamental Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Graphs and Simple Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Matrices and Isomorphisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Paths and Cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4 Vertex Degrees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.5 Graph Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.6 Some Basic Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.7 Degree Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

Chapter 2: Applications on Graph Isomorphisms . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2.1 Generalized Honeycomb Tori . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2 Isomorphism between Cyclic-Cubes and Wrapped
Butterfly Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3 1-Edge Fault-Tolerant Design for Meshes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.4 Faithful 1-Edge Fault-Tolerant Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.5 k-Edge Fault-Tolerant Designs for Hypercubes . . . . . . . . . . . . . . . . . . . . . . . . . . 40

Chapter 3: Distance and Diameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.2 Diameter for Some Interconnection Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.3 Shuffle-Cubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.3.1 Route1(u, v) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.4 Moore Bound . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.5 Star Graphs and Pancake Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.6 Edge Congestion and Bisection Width . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.7 Transmitting Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

Chapter 4: Trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.1 Basic Properties. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.2 Breadth-First Search and Depth-First Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.3 Rooted Trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.4 Counting Trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.5 Counting Binary Trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.6 Number of Spanning Trees Contains a Certain Edge . . . . . . . . . . . . . . . . . . . . . 73
4.7 Embedding Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
v
prelims.tex 11/8/2008 12: 54 Page vi

vi Contents

Chapter 5: Eulerian Graphs and Digraphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

5.1 Eulerian Graphs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.2 Eulerian Digraphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.3 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.3.1 Chinese Postman Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.3.2 Street-Sweeping Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5.3.3 Drum Design Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5.3.4 Functional Cell Layout Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

Chapter 6: Matchings and Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

6.1 Matchings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.2 Bipartite Matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
6.3 Edge Cover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
6.4 Perfect Matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
6.5 Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

Chapter 7: Connectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

7.1 Cut and Connectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
7.2 2-Connected Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
7.3 Menger Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
7.4 An Application—Making a Road System One-Way . . . . . . . . . . . . . . . . . . . . . 115
7.5 Connectivity of Some Interconnection Networks . . . . . . . . . . . . . . . . . . . . . . . 117
7.6 Wide Diameters and Fault Diameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
7.7 Superconnectivity and Super-Edge-Connectivity . . . . . . . . . . . . . . . . . . . . . . . 120

Chapter 8: Graph Coloring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125

8.1 Vertex Colorings and Bounds. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
8.2 Properties of k-Critical Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
8.3 Bound for Chromatic Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
8.4 Girth and Chromatic Number . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
8.5 Hajós’ Conjecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
8.6 Enumerative Aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
8.7 Homomorphism Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
8.8 An Application—Testing on Printed Circuit Boards . . . . . . . . . . . . . . . . . . . . . 137
8.9 Edge-Colorings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138

Chapter 9: Hamiltonian Cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

9.1 Hamiltonian Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
9.2 Necessary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
9.3 Sufficient Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
9.4 Hamiltonian-Connected . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
9.5 Mutually Independent Hamiltonian Paths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
9.6 Diameter for Generalized Shuffle-Cubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
9.7 Cycles in Directed Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
prelims.tex 11/8/2008 12: 54 Page vii

Contents vii

Chapter 10: Planar Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161

10.1 Planar Embeddings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
10.2 Euler’s Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
10.3 Characterization of Planar Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
10.4 Coloring of Planar Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167

Chapter 11: Optimal k-Fault-Tolerant Hamiltonian Graphs . . . . . . . . . . . . . . . . . 171

11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
11.2 Node Expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
11.3 Other Construction Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
11.4 Fault-Tolerant Hamiltonicity and Fault-Tolerant Hamiltonian
Connectivity of the Folded Petersen Cube Networks . . . . . . . . . . . . . . . . . . 185
11.5 Fault-Tolerant Hamiltonicity and Fault-Tolerant Hamiltonian
Connectivity of the Pancake Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
11.6 Fault-Tolerant Hamiltonicity and Fault-Tolerant Hamiltonian
Connectivity of Augmented Cubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
11.7 Fault-Tolerant Hamiltonicity and Fault-Tolerant Hamiltonian
Connectivity of the WK-Recursive Networks . . . . . . . . . . . . . . . . . . . . . . . . . 206
11.8 Fault-Tolerant Hamiltonicity of the Fully Connected Cubic
Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221

Chapter 12: Optimal 1-Fault-Tolerant Hamiltonian Graphs . . . . . . . . . . . . . . . . . 227

12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
12.2 3-Join . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
12.3 Cycle Extension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
12.4 Cells for Optimal 1-Hamiltonian Regular Graphs . . . . . . . . . . . . . . . . . . . . . 241
12.5 Generalized Petersen Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
12.6 Honeycomb Rectangular Disks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
12.7 Properties with Respect to the 3-Join . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
12.8 Examples of Various Cubic Planar Hamiltonian Graphs . . . . . . . . . . . . . . . . 267
12.8.1 A ∩ B ∩ C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
12.8.2 A ∩ B ∩ C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
12.8.3 A ∩ B ∩ C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
12.8.4 A ∩ B ∩ C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
12.8.5 A ∩ B ∩ C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
12.8.6 A ∩ B ∩ C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
12.8.7 A ∩ B ∩ C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
12.8.8 A ∩ B ∩ C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
12.9 Hamiltonian Properties of Double Loop Networks . . . . . . . . . . . . . . . . . . . . 275

Chapter 13: Optimal k-Fault-Tolerant Hamiltonian-Laceable Graphs . . . . . . . . 285

13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
13.2 Super Fault-Tolerant Hamiltonian Laceability of Hypercubes . . . . . . . . . . . 287
13.3 Super Fault-Tolerant Hamiltonian Laceability of Star Graphs . . . . . . . . . . . 289
prelims.tex 11/8/2008 12: 54 Page viii

viii Contents

13.4 Construction Schemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294

13.5 Cubic Hamiltonian-Laceable Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
13.6 1p -Fault-Tolerant Hamiltonian Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
13.6.1 Spider-Web Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
13.6.2 Brother Trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
13.7 Hamiltonian Laceability of Faulty Hypercubes . . . . . . . . . . . . . . . . . . . . . . . . 318
13.8 Conditional Fault Hamiltonicity and Conditional Fault Hamiltonian
Laceability of the Star Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327

Chapter 14: Spanning Connectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339

14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
14.2 Spanning Connectivity of General Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
14.3 Spanning Connectivity and Spanning Laceability of the
Hypercube-Like Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
14.4 Spanning Connectivity of Crossed Cubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
14.5 Spanning Connectivity and Spanning Laceability of the
Enhanced Hypercube Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374
14.6 Spanning Connectivity of the Pancake Graphs . . . . . . . . . . . . . . . . . . . . . . . . . 391
14.7 Spanning Laceability of the Star Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 404
14.8 Spanning Fan-Connectivity and Spanning Pipe-Connectivity
of Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 410

Chapter 15: Cubic 3∗ -Connected Graphs and Cubic 3∗ -Laceable Graphs . . . . . 417
15.1 Properties of Cubic 3∗ -Connected Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
15.2 Examples of Cubic Super 3∗ -Connected Graphs . . . . . . . . . . . . . . . . . . . . . . . 419
15.3 Counterexamples of 3∗ -Connected Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426
15.4 Properties of Cubic 3∗ -Laceable Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430
15.5 Examples of Cubic Hyper 3∗ -Laceable Graphs . . . . . . . . . . . . . . . . . . . . . . . . 431
15.6 Counterexamples of 3∗ -Laceable Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447

Chapter 16: Spanning Diameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449

16.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
16.2 Spanning Diameter for the Star Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
16.3 Spanning Diameter of Hypercubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
16.4 Spanning Diameter for Some (n, k)-Star Graphs . . . . . . . . . . . . . . . . . . . . . . . 474

Chapter 17: Pancyclic and Panconnected Property . . . . . . . . . . . . . . . . . . . . . . . . . 479

17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
17.2 Bipanconnected and Bipancyclic Properties of Hypercubes . . . . . . . . . . . . . 480
17.3 Edge Fault-Tolerant Bipancyclic Properties of Hypercubes . . . . . . . . . . . . . 485
17.4 Panconnected and Pancyclic Properties of Augmented Cubes . . . . . . . . . . . 489
17.5 Comparison between Panconnected and Panpositionable Hamiltonian . . . 500
17.6 Bipanpositionable Bipancyclic Property of Hypercube . . . . . . . . . . . . . . . . . 504
prelims.tex 11/8/2008 12: 54 Page ix

Contents ix

Chapter 18: Mutually Independent Hamiltonian Cycles . . . . . . . . . . . . . . . . . . . . 509

18.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 509
18.2 Mutually Independent Hamiltonian Cycles on Some Graphs . . . . . . . . . . . . 510
18.3 Mutually Independent Hamiltonian Cycles of Hypercubes . . . . . . . . . . . . . . 512
18.4 Mutually Independent Hamiltonian Cycles of Pancake Graphs . . . . . . . . . . 518
18.5 Mutually Independent Hamiltonian Cycles of Star Graphs . . . . . . . . . . . . . . 524
18.6 Fault-Free Mutually Independent Hamiltonian Cycles
in a Faulty Hypercube . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 526
18.7 Fault-Free Mutually Independent Hamiltonian Cycles
in Faulty Star Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 530
18.8 Orthogonality for Sets of Mutually Independent
Hamiltonian Cycles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .543

Chapter 19: Mutually Independent Hamiltonian Paths . . . . . . . . . . . . . . . . . . . . . 545

19.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
19.2 Mutually Independent Hamiltonian Laceability for Hypercubes . . . . . . . . . 545
19.3 Mutually Independent Hamiltonian Laceability for Star Graphs . . . . . . . . . 550
19.4 Mutually Independent Hamiltonian Connectivity for
(n, k)-Star Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555
19.5 Cubic 2-Independent Hamiltonian-Connected Graphs . . . . . . . . . . . . . . . . . . 575
19.5.1 Examples of Cubic 2-Independent Hamiltonian-Connected
Graphs That Are Super 3∗ -Connected . . . . . . . . . . . . . . . . . . . . . . . . 576
19.5.2 Examples of Super 3∗ -Connected Graphs That Are Not
Cubic 2-Independent Hamiltonian-Connected . . . . . . . . . . . . . . . . . 579
19.5.3 Example of a Cubic 2-Independent Hamiltonian-
Connected Graph That Is Not Super 3∗ -Connected . . . . . . . . . . . . . 580
19.5.4 Example of a Cubic 1-Fault-Tolerant Hamiltonian Graph That
Is Hamiltonian-Connected but Not Cubic 2-Independent
Hamiltonian-Connected or Super 3∗ -Connected . . . . . . . . . . . . . . . 581

Chapter 20: Topological Properties of Butterfly Graphs . . . . . . . . . . . . . . . . . . . . 585

20.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 585
20.2 Cycle Embedding in Faulty Butterfly Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . 590
20.3 Spanning Connectivity for Butterfly Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . 607
20.4 Mutually Independent Hamiltonicity for Butterfly Graphs . . . . . . . . . . . . . . 616

Chapter 21: Diagnosis of Multiprocessor Systems . . . . . . . . . . . . . . . . . . . . . . . . . 625

21.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 625
21.2 Diagnosis Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 625
21.2.1 PMC Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 626
21.2.2 Comparison Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 628
21.3 Diagnosability of the Matching Composition Networks . . . . . . . . . . . . . . . . 631
21.3.1 Diagnosability of the Matching Composition Networks
under the PMC Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
prelims.tex 11/8/2008 12: 54 Page x

x Contents

21.3.2 Diagnosability of the Matching Composition Networks

under the Comparison Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
21.4 Diagnosability of Cartesian Product Networks . . . . . . . . . . . . . . . . . . . . . . . . . 637
21.4.1 Diagnosability of Cartesian Product Networks under the
PMC Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 638
21.4.2 Diagnosability of Cartesian Product Networks under the
Comparison Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 639
21.4.3 Diagnosability of t-Connected Networks . . . . . . . . . . . . . . . . . . . . . . 639
21.4.4 Diagnosability of Homogeneous Product Networks . . . . . . . . . . . . 641
21.4.5 Diagnosability of Heterogeneous Product Networks . . . . . . . . . . . . 644
21.5 Strongly t-Diagnosable Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 645
21.5.1 Strongly t-Diagnosable Systems in the Matching
Composition Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 649
21.6 Conditional Diagnosability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 652
21.6.1 Conditional Diagnosability of Qn under the PMC Model . . . . . . . 653
21.7 Conditional Diagnosability of Qn under the Comparison Model . . . . . . . . . 658
21.7.1 Conditional Diagnosability of Cayley Graphs Generated
by Transposition Trees under the Comparison Diagnosis
Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 666
21.7.2 Cayley Graphs Generated by Transposition Trees . . . . . . . . . . . . . . 666
21.7.3 Conditional Diagnosability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 668
21.8 Local Diagnosability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .670
21.8.1 Strongly Local-Diagnosable Property . . . . . . . . . . . . . . . . . . . . . . . . . 675
21.8.2 Conditional Fault Local Diagnosability . . . . . . . . . . . . . . . . . . . . . . . 679

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 687
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 703
prelims.tex 11/8/2008 12: 54 Page xi

Preface
Since the origin of graph theory with Euler, people have liked graph theory, because
it is a delightful playground for the exploration of proof techniques without much
previous background needed. With the application of graph theory to many areas of
computing, social, and natural sciences, the importance of graph theory has been
recognized. For this reason, many good graph theoretic results have been developed
in recent years. It is almost impossible to write a book that covers all these good graph
theories.
The architecture of an interconnection network is always represented by a graph,
where vertices represent processors and edges represent links between processors.
It is almost impossible to design a network that is optimum from all aspects. One
has to design a suitable network depending on its properties and requirements. Thus,
many graphs are proposed as possible interconnection network topologies. Thus, these
graphs can be referred to as good graphs. For this reason, the theory of interconnection
networks is referred to as good-graph theory.
Thus, we need some basic background in graph theory to study interconnection
networks. The first 10 chapters cover those materials presented in most graph theory
texts and add some concepts of interconnection networks. Later, we discuss some
interesting properties of interconnection networks. Sometimes, these properties are
too good to be true—so the beginner finds them hard to believe. Such properties
can be observed from interconnection networks, diameter, connectivity, Hamiltonian
properties, and diagnosis properties. We can also study these properties in general
graphs.
Ends of proofs are marked with the symbol
. This symbol can be found directly
following a formal assertion. It means that the proof should be clear after what has
been said. There are also some theorems that are stated, without proof, via back-
ground information. In this case, such theorems are stated without proofs and the
symbol
.
The main purpose of this book is to share our research experience. Our experience
tells us that much background is not needed for doing research. Looking for a proper
theme is a difficult problem for research. We are just lucky in finding problems. For
this reason, we put some of our results at the end of each chapter but we do not put any
exercises in this book. We observed that all the material, contents, and exercises in
every book are exercises of previous books. For this reason, we hope that the readers
can find their own exercises.
It is a great pleasure to acknowledge the work of Professor Jimmy
J.M. Tan. We have worked together for a long time. Yet, at least one-third
of this book is his contribution. We also thank all the members of the Com-
puter Theory Laboratory of National Chiao-Tung University: T.Y. Sung, T. Liang,
H.M. Huang, Chiuyuan Chen, Y.C. Chuang, S. S. Kao, Y.C. Chang, T.Y. Ho,
R.S. Chou, C.N. Hung, J.J. Wang, S.Y. Wang, C.Y. Lin, C.H. Tsai, J.Y. Hwang,
C.P. Chang, J.N. Wang, J.J. Sheu, C.H. Chang, T.K. Li, W.T. Huang, Y.C. Chen,

xi
prelims.tex 11/8/2008 12: 54 Page xii

xii Preface

H.C. Hsu, K.Y. Liang, Y.L. Hsieh, M.C. Yang, P.L. Lai, L.C. Chiang, Y.H. Teng,
C.M. Sun, and K.M. Hsu.
We also thank CRC Press for offering us the opportunity of publishing this book.
Lih-Hsing Hsu also thanks his advisor, Professor Joel Spencer, for entering the area
of graph theory. We also thank Professor Frank Hsu for strongly recommending to us
the area of interconnection networks. Finally, we thank all of our teachers; without
their encouragement, this book could never have been written.
prelims.tex 11/8/2008 12: 54 Page xiii

Authors
Lih-Hsing Hsu received his BS in mathematics from Chung Yuan Christian Univer-
sity, Taiwan, Republic of China, in 1975, and his PhD in mathematics from the State
University of New York at Stony Brook in 1981. From 1981 to 1985, he was an asso-
ciate professor at the Department of Applied Mathematics at National Chiao Tung
University in Taiwan. From 1985 to 2006, he was a professor at National Chiao Tung
University. From 1985 to 1988, he was the chairman of the Department of Applied
Mathematics at National Chiao Tung University. After 1988, he joined the Depart-
ment of Computer and Information Science of National Chiao Tung University. In
2004, he retired from National Chiao Tung University, holding a title as an honorary
scholar of that university. He is currently the chairman of the Department of Computer
Science and Information Engineering, Providence University, Taichung, Taiwan. His
research interests include interconnection networks, algorithms, graph theory, and
VLSI layout.

Cheng-Kuan Lin received his BS in applied mathematics from Chinese Culture

University, Taiwan, Republic of China, in 2000, and his MS in mathematics from
the National Central University in Taiwan in 2002. His research interests include
interconnection networks, algorithms, and graph theory.

xiii
prelims.tex 11/8/2008 12: 54 Page xiv
ch001.tex 30/6/2008 9: 37 Page 1

1 Fundamental Concepts

1.1 GRAPHS AND SIMPLE GRAPHS

A graph G with n vertices and m edges consists of the vertex set V (G) = {v1 , v2 , . . . , vn }
and edge set E(G) = {e1 , e2 , . . . , em }, where each edge consists of two (possibly equal)
vertices called, endpoints. An element in V (G) is called a vertex of G. An element in
E(G) is called an edge of G. Because graph theory has a variety of applications, we
may also use a node for a vertex and a link for an edge to fit the common terminology
in that area. We use the unordered pair (u, v) for an edge e = {u, v}. If (u, v) ∈ E(G),
then u and v are adjacent. We write u ↔ v to mean u is adjacent to v. A loop is an edge
whose endpoints are equal. Parallel edges or multiple edges are edges that have the
same pair of endpoints. A simple graph is a graph having no loops or multiple edges.
For a graph G = (V , E), the underlying simple graph UG is the simple graph with
vertex V and (x, y) ∈ E(UG) if and only if x  = y and (x, y) ∈ E. A graph is finite if its
vertex set and edge set are finite. We adopt the convention that every graph mentioned
is finite.
We illustrate a graph on paper by assigning a point to each vertex and drawing
a curve for each edge between the points representing its endpoints, sometimes omit-
ting the name of the vertices or edges. In Figure 1.1a, e is a loop and f and g are
parallel edges. Figures 1.1b and 1.1c illustrate two ways of drawing one simple graph.
The order of a graph G, written n(G), is the number of vertices in G. An n-vertex
graph is a graph of order n. The size of a graph G, written e(G), is the number of
edges in G, even though we also use e by itself to denote an edge. The degree of
a vertex v in a graph, written degG (v), or deg(v), is the number of nonloop edges
containing v plus twice the number of loops containing v. The maximum degree of G,
denoted by
(G), is max{deg(v) | v ∈ V (G)} and the minimum degree of G, denoted
by δ(G), is min{deg(v) | v ∈ V (G)}. A graph G is regular if
(G) = δ(G), and k-regular
if
(G) = δ(G) = k. A vertex of degree k is k-valent. The neighborhood of v, written
NG (v) or N(v), is {x ∈ V (G) | x ↔ v}; x is a neighbor of v if x ∈ N(v). An isolated
vertex has a degree of 0.
A directed graph or digraph G consists of a vertex set V (G) and an edge set E(G),
where each edge is an ordered pair of vertices. We use the ordered pair (u, v) for the
edge uv, where u is the tail and v is the head. (Note that the notation (u, v) is used both
as an unordered pair in an unordered graph and as an ordered pair in a directed graph.
However, it is easy to distinguish each from the other from the context.) Sometimes,
we may use the term arc for an edge of a digraph. We write u → v when (u, v) ∈ E(G),
meaning there is an edge from u to v. A simple digraph is a digraph in which each
ordered pair of vertices occur at most once as an edge. For a digraph G = (V , E), the

1
ch001.tex 30/6/2008 9: 37 Page 2

2 Graph Theory and Interconnection Networks

x v w
f
d y
g y
u
e h i x
w v u
(a) (b) (c)

FIGURE 1.1 Examples of graphs.

y
w

a b c

d
z
x

FIGURE 1.2 A digraph.

underlying graph UG is the simple graph with the vertex set V and (x, y) ∈ E(UG) if
and only if x  = y and either (x, y) ∈ E(G) or (y, x) ∈ E(G).
The choice of head and tail assigns a direction to an edge, which we illustrate by
assigning edges as arrows. See Figure 1.2 for an example of a digraph. Sometimes,
weights are assigned to the edges of graphs or digraphs. Thus, we have weighted
graphs, weighted digraphs, (unweighted) graphs, and (unweighted) digraphs.
Graphs arise in many settings, which suggests useful concepts and terminologies
about the structure of graphs.

Example 1.1
A famous brain-teaser asks: Does every set of six people have three mutual acquaintances
or three mutual strangers?

Because an acquaintance is symmetric, we can model it by a simple graph having

a vertex for each person and an edge for each acquainted pair. The nonacquaintance
relation on the same set yields another graph. The complement G of a simple graph G
with the same vertex set V (G) is defined by (u, v) ∈ E(G) if and only if (u, v) ∈
/ E(G).
In Figure 1.3, we draw a graph and its complement.
Let two graphs equal G = (V (G), E(G)) and H = (V (H), E(H)). We say that H is a
subgraph of G or G is a supergraph of H if V (H) ⊆ V (G) and E(H) ⊆ E(G); we write
H ⊆ G and say that G contains H. In particular, H is called a spanning subgraph of
G or G is a spanning supergraph of H if H is a subgraph of G and V (H) = V (G).
Let G = (V , E) be a graph and S be a subset of V . The subgraph of G induced by
S, denoted by G[S], is the subgraph with vertex set S and those edges of G with both
ends in S. In particular, the graph G[V − S] is denoted by G − S. Let v be a vertex of
G. We use G − v to denote G − {v}. Let F be a subset of E. The subgraph generated
ch001.tex 30/6/2008 9: 37 Page 3

Fundamental Concepts 3

u u

y y v
v

x w x w
G G

FIGURE 1.3 A graph and its complement.

by F, denoted by GF , is the subgraph of G with F as its edge set and {v | v is incident

with some edge of F} as its vertex set. The subgraph G − F denotes the subgraph of
G with V as its vertex set and E − F as its edge set. Let e be an edge of E. The graph
G − {e} is denoted by G − e.
A complete graph or clique is a simple graph in which every pair of vertices is an
edge. A complete graph has many subgraphs that are not cliques, but every induced
subgraph of a complete graph is a clique. The complement of a complete graph has
no edges.
An independent set in a graph G is a vertex set S ⊆ V (G) that contains no edge of
G (that is, G[S] has no edges). In the graph G shown in Figure 1.3, the largest clique
and the largest independent set have cardinalities three and two, respectively. These
values reverse in the complement G, because cliques become independent sets (and
vice versa) under complementation. Our six-person brain-teaser asks whether every
six-vertex graph has a clique or an independent set of size 3. It is worthwhile to verify
this statement. The generalization of the six-person brain-teaser leads the Ramsey
Theory [128].

Example 1.2
Suppose that we have m jobs and n people, and each person can do some of the jobs. Can
we make assignments to fill the jobs? We model the available assignments by a graph H
having a vertex for each job and each person, putting job j adjacent to person p if p can do j.

A graph G is bipartite if V (G) is the union of two disjoint sets such that each edge
consists of one vertex from each set. The graph H of available assignments is bipartite,
with the two sets being the people and the jobs. Because a person can do only one job
and we assign a job to only one person, we seek m pairwise disjoint edges in H.
A complete bipartite graph, illustrated in Figure 1.4, is a bipartite graph whose
edge set consists of all pairs having a vertex set from each of two disjoint sets covering
the vertices. When the assignment graph H is a complete bipartite graph, pairing up
vertices is easy, so we seek the best way to do it. We can assign numerical weights on
the edges to measure desirability. The best way to match up the vertices is often the
one where the selected edges have maximum total weight.
ch001.tex 30/6/2008 9: 37 Page 4

4 Graph Theory and Interconnection Networks

FIGURE 1.4 A complete bipartite graph.

FIGURE 1.5 A 3-partite graph.

Example 1.3
Suppose that we must schedule for Senate committee meetings into designated weekly
time periods. We cannot assign two committees to the same time if they have a common
member. How many different time periods do we need?

We create a vertex for each committee, with two vertices adjacent when their
committees have a common member. We must assign labels (time periods) to vertices
so that the endpoints of each edge receive different labels. The labels have no numer-
ical value, so we call them colors, and the vertices receiving a particular label form
a color class. The minimum number of colors needed is the chromatic number of G,
written χ(G). Because vertices of the same color must form an independent set, χ(G)
equals the minimum number of independent sets that partition V (G). This generalizes
bipartite graphs. A graph G is k-partite if V (G) is the union of k (possibly empty) inde-
pendent sets. When they are pairwise disjoint, these sets that partition V (G) are called
partite sets (or color classes). The graph in Figure 1.5 has chromatic number 3 and is
3-partite (there could be graphs that are also 4-partite, 5-partite, etc.). We will study
the chromatic number of graphs in Chapter 10.
The most (in)famous problem in graph theory involves coloring. A map is a
partition of the plane into connected regions. Can we color the regions of every
map, using at most four colors, so that neighboring regions have different colors? In
each map, we introduce a vertex for each region and an edge for regions sharing a
boundary. We obtain a planar graph. A graph is planar if it can be drawn in the plane
without line crossings. In other words, the map coloring problem asks whether each
planar graph has a chromatic number of at most 4. We will study planar graphs in
Chapter 12.
ch001.tex 30/6/2008 9: 37 Page 5

Fundamental Concepts 5

FIGURE 1.6 A disconnected graph.

Example 1.4
We can model a road network by a graph having edges that correspond to road segments
between intersections. We can assign edge weights to measure distance or travel time.
We may want to know the shortest route from x to y.

Informally, we think of a path in a graph as a single vertex or an ordered list of

distinct vertices v1 , v2 , . . . , vn such that (vi−1 , vi ) is an edge for 2 ≤ i ≤ n. The first and
the last vertices of a path are its endpoints. A path from u to v is a path with endpoints
u and v. Similarly, we think of a cycle as an ordered list v1 , v2 , . . . , vn such that all
(vi−1 , vi ) for 2 ≤ i ≤ n and also (vn , v1 ) are edges.
To find the shortest route from x to y, we want to find the path from x to y having
the least total weight among all paths from x to y in G. Similarly, in a network of n
cities, we may want to visit all the cities and return home at the minimum total cost.
Using the cost that we weight on the edges of the complete graph, we seek the n-vertex
cycle with minimum total cost. This is the Traveling Salesman Problem.
In a road network or communication network, every site should be reachable
from each other. A graph G is connected if it has a path from u to v for each pair
u, v ∈ V (G). A graph is disconnected if it is not connected. The graph in Figure 1.6 is
disconnected.

1.2 MATRICES AND ISOMORPHISMS

How can we specify a graph? We can list the vertices and edges, but there are other
useful ways to encode this information.
Given a graph or digraph G with vertices indexed as V (G) = {v1 , v2 , . . . , vn }, the
adjacency matrix of G, written A(G), is the matrix in which entry aij is the number
of copies of the edges (vi , vj ) in G. If vertex v belongs to edge e, then v and e are
ch001.tex 30/6/2008 9: 37 Page 6

6 Graph Theory and Interconnection Networks

w y
w x y z a b c
w 0 1 0 0 w 0 0 1
c b a A(G) x 1 0 1 0 M(G ) x 0 1 1
G
y 0 1 0 1 y 1 1 0
z 0 0 1 0 z 1 0 0
x z

w y
w x y z a b c
w 0 0 0 0 w 0 0 ⫺1
H c b a A(H ) x 1 0 0 0 M(H ) x 0 ⫺1 1
y 0 1 0 1 y 1 1 0
z 0 0 0 0 z ⫺1 0 0
x z

FIGURE 1.7 Adjacency matrices and incidence matrices.

incident. The incidence matrix M(G) of a loopless graph G has rows indexed by
V (G) and columns indexed by E(G), with mij = 1 if vertex vi belongs to ej ; otherwise
mij = 0. For a loopless digraph, mij = +1 if vi is the tail of ej , mij = −1 if vi is the
head of ej , and mij = 0 if otherwise.
A graph may have many adjacency matrices, depending on the ordering of the
vertices. Each ordering of the vertices determines an adjacency matrix. The adjacency
matrix of G is symmetric if G is a graph (not a digraph). Moreover, every entry in
A(G) is 0 or 1, with 0’s on the diagonal if G is a simple graph.

Example 1.5
In Figure 1.7, we draw a simple graph G and a digraph H, together with the adjacency
matrices and incidence matrices that result from the vertex ordering w, x, y, z and the
edges ordering a, b, c. The adjacency matrix for the graph having two copies of each of
these edges would obtained by changing each 1 to 2.

When we present an adjacency matrix for a graph, we are implicitly naming the
vertices by the order of its rows. The ith vertex corresponds to the ith row and column.
This provides names for the vertices. We cannot store a graph in a computer without
naming the vertices. Nevertheless, we want to study properties (like chromatic number
or connected) that do not depend on the names of the vertices. If we can find a one-to-
one correspondence (bijection) between V (G) and V (H) that preserves the adjacency
relation, then G and H have the same structural properties.
An isomorphism from G to H is a bijection f : V (G) → V (H) such that (u, v) ∈
E(G) if and only if ( f (u), f (v)) ∈ E(H). We say “G is isomorphic to H,” written G ∼
= H,
if there is an isomorphism from G to H.
When G is isomorphic to H and H is also isomorphic to G, we may say G and H
are isomorphic (to each other). Because an adjacency matrix encodes the adjacency
ch001.tex 30/6/2008 9: 37 Page 7

Fundamental Concepts 7

p q p r q s a b a b c d
p 0 0 0 1 a 0 0 1 1
A(G) r 0 0 1 0 A(H) b 0 0 1 0

s r q 0 1 0 1 d c c 1 1 0 0
G s 1 0 1 0 H d 1 0 0 0

FIGURE 1.8 Two isomorphic graphs.

relation, we can also describe isomorphism using adjacency matrices. The graphs G
and H are isomorphic if and only if we can apply a permutation to the rows of A(G)
and the same permutation to the columns of A(G) to obtain A(H).

Example 1.6
The graphs G and H in Figure 1.8 are 4-vertex paths. They are isomorphic by an isomor-
phism that maps p, q, r, s to b, a, d, c, respectively. Rewriting A(G) by placing the rows
in the order p, q, r, s and the columns also in order yields A(H). Another isomorphism
maps s, r, q, p to a, b, c, d, respectively.

The set of the pair (G, H) such that G is isomorphic to H is the isomorphism
relation on the set of graphs. Obviously, isomorphism is an equivalence relation. An
isomorphism class of graphs is an equivalence class of graphs under the isomorphism
relation.
When discussing the structure of a graph G, we consider a fixed vertex set, but our
comments apply to every graph isomorphic to G. Therefore, we sometimes use the
informal expression unlabeled graph to mean an isomorphic class of graphs. When we
draw a graph, its vertices are named by their physical locations, even if we give them
no other names. We often use the name graph for the drawing of a graph. When we
redraw a graph to display some structural aspect, we have chosen a more convenient
member of the isomorphism class.
We often discuss two isomorphic graphs using the same name. Our statement
applies to all graphs in their isomorphism class. Hence, we usually write G = H instead
of G ∼= H. Similarly, when we say that H is a subgraph of G, we mean technically that
H is isomorphic to a subgraph of G, or G contains a copy of H.
This treatment of isomorphism classes leads us to using the notation Kn , Pn , and
Cn , respectively, to denote a clique, a path, or a cycle with n vertices. Similarly, Kr,s
denotes the complete bipartite graph with partite sets of cardinalities r and s.

Example 1.7
Suppose that X is a set of size n. Obviously, X contains C(n; 2) = n(n − 1)/2 unordered
pairs of (u, v) with u = v. We may include or omit each pair as an edge. Thus, there are
2C(n;2) simple graphs with vertex set X. For example, there are sixty four simple graphs
having four vertices. However, these fall into eleven isomorphism classes as illustrated in
Figure 1.9 in complementary pairs. Only P4 is isomorphic to its complement. Note that
ch001.tex 30/6/2008 9: 37 Page 8

8 Graph Theory and Interconnection Networks

FIGURE 1.9 Graphs with four vertices.

FIGURE 1.10 Graphs for Example 1.8.

the isomorphism classes have different sizes. We cannot count the isomorphism classes
of n-vertex simple graphs, though, by dividing 2C(n;2) by the size of a class.

An isomorphism from G into H preserves the adjacency relation. As structural

properties of graphs are determined by their adjacency relations, we can prove that
each G and H are not isomorphic by finding some structural properties that are not
true of the other. If they have different vertex degrees, different sizes of the largest
clique or smallest cycle, and so on, then they cannot be isomorphic, because these
properties are preserved by isomorphism. On the other hand, no known list of common
structural properties implies that G ∼
= H. We must present a bijection f : V (G) → V (H)
that preserves the adjacency relation.

Example 1.8
In Figure 1.10, we have four graphs with each vertex having a degree of 3. However,
these graphs are not pairwise isomorphic. Several are drawings of K3,3 , as shown by their
exhibited isomorphisms. One contains C3 and hence cannot be a drawing of K3,3 .
Consider the graphs in Figure 1.11. Because they have many edges, we prefer to
consider their complements. Note that graphs G and H are isomorphic if and only if
G and H are isomorphic. It is observed that the complement of one of these graphs is
connected, but the complement of the other is not connected. These two graphs cannot
be isomorphic.

An automorphism of G is a permutation of V (G), that is, an isomorphism from

G into G. A graph G is vertex-transitive if for every pair u, v ∈ V (G) there is an
automorphism that maps u to v. A graph G is edge-transitive if for every pair
e, f ∈ E(G) there is an automorphism that maps e to f . Obviously, every cycle is
not only vertex-transitive but also edge-transitive. Suppose that n > 2. Then Pn is not
ch001.tex 30/6/2008 9: 37 Page 9

Fundamental Concepts 9

FIGURE 1.11 Two nonisomorphic graphs.

d
x y z
H
e f

b c
G

FIGURE 1.12 A vertex-transitive graph and an edge-transitive graph.

vertex-transitive, because no automorphism can map a vertex of degree 1 into a vertex

of degree 2.

Example 1.9
Let G be the path with a vertex set {1, 2, 3, 4} and an edge set {(1, 2), (2, 3), (3, 4)}. This
graph has two isomorphisms: an identity permutation and the permutation that switches
1 with 4 and switches 2 with 3. Interchanging vertices 1 and 2 is not an automorphism of
G, although G is isomorphic to the graph H with the vertex set {1, 2, 3, 4} and the edge
set {(1, 2), (1, 3), (3, 4)}.
The automorphisms of G can be viewed as the permutations that can be applied
simultaneously to the rows and columns of A(G) without changing A(G). For example,
permuting the vertices of one independent set of Kr,s does not change the adjacency
matrix. Thus, Kr,s has r!s! automorphism if r  = s. Moreover, Kt,t has 2(t!)2 auto-
morphisms. Thus, the complete bipartite graph Kr,s is vertex-transitive if and only
if r = s.

The graph G in Figure 1.12 is vertex-transitive but not edge-transitive. Yet the
graph H in Figure 1.12 is edge-transitive but not vertex-transitive.

1.3 PATHS AND CYCLES

A walk of length k is a sequence v0 , e1 , v1 , e2 , . . . , ek , vk of vertices such that
ei = (vi−1 , vi ) for all i. The length of a walk W is denoted by l(W ). A trail is a
ch001.tex 30/6/2008 9: 37 Page 10

10 Graph Theory and Interconnection Networks

walk with no repeated edge. A path is a walk with no repeated vertex. A walk from
u to v has first vertex u and last vertex v; these are its endpoints. A walk (or trail) is
closed if it has a length of at least 1 and its endpoints are equal. A cycle is a closed
trail in which first = last is the only vertex repetition (a loop is a cycle of length 1).
Paths and trails are walks and hence have endpoints. A path or trail may have
length 0, but a cycle or closed walk cannot. We use the words path and cycle in three
closely related contexts: as a graph or subgraph, as the special case of a walk, and as
a set of edges. In a simple graph, a walk is completely specified by its sequence of
vertices. Hence, we usually describe a path or a cycle (or walk) in a simple graph by
its ordered list of vertices. Although we may list only vertices, the walk still consists
of both vertices and edges. We may start a cycle at any vertex. To emphasize the cyclic
aspect of the ordering, we often list each vertex only once when naming a cycle.

Example 1.10
The graph in Figure 1.13 has a closed walk of length 11 that visits vertices in the order
a, x, a, x, u, y, c, d, y, v, b, a . Omitting the first two steps yields a closed trail (no edge
repetition). The edge set of this trail is the the union of the edge sets of three pairwise
edge-disjoint cycles. The trail u, y, c, d, y, v contains the edge of the path u, y, v .

With the previous example, we have the following observation: suppose that u
and v are distinct vertices in G. Obviously, every walk from u to v in G contains a
path from u to v. Moreover, every closed odd walk contains an odd cycle.
The definitions and remarks hold for digraphs, with each ei being an ordered pair
(vi−1 , vi ). In a path or cycle of a directed graph, successive edges follow the arrows.
For example, the digraph consisting only of the edge (x, y) has a path from x to y but
no path from y to x.
A graph G is connected if it has a path from u to v for each pair u, v ∈ V (G).
Otherwise, it is disconnected. The vertex u is connected to the vertex v in G if G has a
path from u to v. The connection relation in a graph consists of the vertex pairs (u, v)
such that u is connected to v.
Connectedness is a property of graphs. However, the connection relation on ver-
tices and the phrase u is connected to v are convenient when writing proofs. To specify
the stronger condition (u, v) ∈ E(G), we say u and v are adjacent or u and v are joined
by an edge. We use connected to for the existence of a path from u to v, and we use
joined to for the existence of an edge (u, v).
a c

x y

b v d

FIGURE 1.13 Graph for Example 1.10.

ch001.tex 30/6/2008 9: 37 Page 11

Fundamental Concepts 11

r u y

s
v w z

q t x

FIGURE 1.14 Graph for Example 1.11.

The components of a graph G are its maximal connected subgraphs. A component

(or graph) is nontrivial if it contains an edge. A cut vertex of a graph is a vertex whose
deletion increases the number of components. Similarly, a cut edge of a graph is an
edge whose deletion increases the number of components.

Example 1.11
The graph in Figure 1.14 has three components. The vertex sets of components are {q, r},
{s, t, u, v, w}, and {x, y, z}. The vertex v is a cut vertex. The edges (q, r) and (v, w) are the
cut edges.

The connection relation is an equivalence relation. The equivalence classes of this

relation are the vertex sets of the components.

PROPOSITION 1.1 A graph is connected if and only if there is an edge with

endpoints in both sets for every partition of its vertices into two nonempty sets.
Proof: Suppose that G is connected. Let S and T be any partition of V (G); that is,
V (G) = S ∪ T , S ∩ T = ∅, min{|S|, |T |}  = 0. We choose u ∈ S and v ∈ T . Because G is
connected, G has a path P from u to v. After its last vertex in S, P has an edge from
S to T .
Suppose that G is disconnected. Let H be a component of G. Then no edge has
exactly one endpoint in H. This means that the partition S and T with S = V (H) has
no edge with endpoints in both sets. We have proved that if G is disconnected, then
the partition condition fails. By the contrapositive, the partition condition implies that
G is connected.
Adding an edge e to G reduces the number of components by at most one. Similarly,
deleting a cut edge increases the number of components by exactly one.

THEOREM 1.1 An edge of a graph is a cut edge if and only if it belongs to no cycle.
Proof: Suppose that edge e = (x, y) belongs to a component H of G. Note that the
deletion of e affects no other components. It suffices to prove that H − e is connected
if and only if e belongs to a cycle.
Suppose that H − e is connected. Then H − e contains a path P from x to y.
Obviously, P ∪ {e} forms a cycle containing e.
ch001.tex 30/6/2008 9: 37 Page 12

12 Graph Theory and Interconnection Networks

Conversely, suppose that e belongs to a cycle C. Let u and v be any two different
vertices in V (H). Because H is connected, H has a path P from u to v. Suppose that
P does not contain e. Then P exists in H − e. Thus, H − e has a path from u to v.
Suppose that P contains e. By the symmetry role of x and y, we can assume that x is
between u and y on P. Since H − e contains a path from u to x along P, a path from x
to y along C, and a path from y to v along P. The transitivity of the connection relation
implies that H − e has a path from u to v. Since u and v were chosen arbitrarily from
V (G), we have proved that H − e is connected.

PROPOSITION 1.2 Let G be a graph with V (G) = {v1 , v2 , . . . , vn } with n ≥ 3.

Suppose that at least two of the subgraphs G − v1 , G − v2 , . . . , G − vn are connected.
Then G is connected.
Proof: Suppose that G is not connected and has components H1 , H2 , . . . , Hk . When
we delete a vertex from Hi , we still have at least k components unless Hi = K1 , in
which case k − 1 components remain. Thus, having at least two of G − v1 , G − v2 , . . . ,
G − vn be connected requires that k = 2 and that each of these two components is a
single vertex. This contradicts the hypothesis that G has at least three vertices.

COROLLARY 1.1 Every nontrivial graph has at least two vertices that are not
cut-vertices.

THEOREM 1.2 A graph is bipartite if and only if it has no odd cycle.

Proof: Suppose that G is a bipartite graph. Every walk in G alternates between the
two sets of bipartition of G. Thus, every return to the original class (including the
original vertex) happens after an even number of steps. Hence, G has no odd cycle.
Suppose that graph G has no odd cycle. We prove that G is bipartite by constructing
a bipartition of each nontrivial component. Let u be a vertex in a nontrivial component
H of G. For each v ∈ V (G), we claim that all walks from u to v have the same parity.
If not, the concatenation of two walks from u to v of different parity (reversing the
second) is a closed walk of odd length. Then G has an odd cycle. This contradicts the
assumption of no odd cycle.
We can thus partition V (G) into disjoint sets X, Y by letting X = {v ∈ V (G) | walks
from u to v have even length} and Y = {v ∈ V (G) | walks from u to v have odd length}.
Because an edge joining two vertices in X will create a closed odd walk, X is an
independent set. Similarly, Y is also an independent set. Thus, G is a bipartite
graph.

1.4 VERTEX DEGREES

We begin with an essential tool of graph theory, sometimes called the First Theorem
of Graph Theory or the Handshaking Lemma.


THEOREM 1.3 (Degree-Sum Formula). If G is a graph, then v∈V (G) deg(v) =
2e(G).
ch001.tex 30/6/2008 9: 37 Page 13

Fundamental Concepts 13

Proof: Summing the degrees counts each edge twice, as each edge has two ends
and contributes to the degree at each endpoint.

By the degree-sum formula, the average vertex degree is 2e(G)/n(G); hence

δ(G) ≤ 2e(G)/n(G) ≤
(G). We list two other immediate corollaries.

COROLLARY 1.2 Every graph has an even number of vertices of odd degree. No
graph of odd order is regular with odd degree.

COROLLARY 1.3 A k-regular graph with n vertices has nk/2 edges.

Example 1.12
The Petersen graph is drawn in Figure 1.15. The graph is useful enough to have an
entire book devoted to it (Holten and Sheehan [156]). The Petersen graph has a simple
description using the set S of all two-element subsets of a 5-element set. Let G be the
graph with V (G) = S such that two vertices join an edge if and only if they are disjoint
as sets. Hence, the Petersen graph is vertex-transitive. Obviously, the Petersen graph is
a 3-regular graph with 10 vertices. Thus, it has 15 = 3 × 10/2 edges.

Example 1.13
Let S be the set of all n-tuples in which each position is 0 or 1. Obviously, S contains 2n
elements. The n-dimensional hypercube is the graph Qn with vertex set S in which two
n-tuples are adjacent if and only if they differ in exactly one position. In Figure 1.16, we
illustrate Q3 . The hypercube is an architecture for parallel computers [220]. Processing
units can communicate directly if they correspond to adjacent vertices in Qn .
The Hamming weight of a 0, 1-vector is the number of 1’s. Every edge of Qn consists
of a vector of even weight and a vertex of odd weight. Hence, the vectors of all even
weights form an independent set, and the vectors of all odd weights form an independent

FIGURE 1.15 Some drawings of the Petersen graph.

010 110
011 111

001 101
000 100

FIGURE 1.16 Q3 .
ch001.tex 30/6/2008 9: 37 Page 14

14 Graph Theory and Interconnection Networks

set. Thus, Qn is bipartite. Because each vector can be changed in n places, Qn is n-regular.
By Corollary 3.3, Qn has n2n−1 edges.
Deleting the dashed edges in the illustration leaves 2Q2 . This suggests an inductive
description of Qn . The basis is the 1-vertex graph Q0 , which consists of the unique binary
vector of length 0. Given Qn−1 , we construct Qn in two steps: (1) Take two disjoint copies
of Qn−1 . Call them Q0 and Q1 ; (2) For each v ∈ V (Qn−1 ), append a 0 to the end of the
vector for v in Q0 and a 1 to the end of the vector for v in Q1 , and add an edge consisting
of these two vertices.

1.5 GRAPH OPERATIONS

We need more methods to generate graphs and discuss their properties. For this reason,
we present some graph operations.
Let G1 = (V1 , E1 ) and G2 = (V2 , E2 ) be two graphs. The union of G1 and G2 ,
written G1 ∪ G2 , has vertex set V1 ∪ V2 and edge set E1 ∪ E2 . To specific the disjoint
union with V1 ∩ V2 = ∅, we write G1 + G2 . More generally, mG is the graph consisting
of m pairwise disjoint copies of G. The join of G1 and G2 , written G1 ∨ G2 , is obtained
from G1 + G2 by adding the edges {(x, y) | x ∈ V1 , y ∈ V2 }. We illustrate C3 + C4 in
Figure 1.17a and C3 ∨ C4 in Figure 1.17b.
Note that the adjacency matrix of G1 + G2 is the direct sum of the adjacency
matrix of G1 and the adjacency matrix of G2 .
Let G1 = (V1 , E1 ) and G2 = (V2 , E2 ) be two graphs. The Cartesian product of G1
and G2 , denoted by G1 × G2 , is the graph with vertex set V1 × V2 such that (u1 , v1 ) is
joined to (u2 , v2 ) k times if and only if either u1 = u2 and v1 is joined to v2 k times in
G2 or v1 = v2 and u1 is joined to u2 k times in G1 . We illustrate the Cartesian product
of G1 in Figure 1.18a and G2 in Figure 1.18b as the graph in Figure 1.18c.
Many interconnection networks are constructed by Cartesian product [6,220].
An n-dimensional mesh (abbreviated as mesh) M(m1 , m2 , . . . , mn ) is defined as the
Cartesian product of n paths Pm1 × Pm2 × · · · × Pmn , where Pmi is the path graph with
mi vertices, and an n-dimensional torus (abbreviated as torus) C(m1 , m2 , . . . , mn ) is
defined as the Cartesian product of n cycles Cm1 × Cm2 × · · · × Cmn where Cmi is the
cycle graph with mi vertices. Meshes and Tori are widely used computer architec-
tures [220]. In particular, the hypercube Qn is M(m1 , m2 , . . . , mn ) where mi = 2 for
1 ≤ i ≤ n.
We introduce another family of such interconnection networks based on the Carte-
sian product. The Petersen graph is a 3-regular graph with ten vertices such that the

(a) (b)

FIGURE 1.17 (a) C3 + C4 and (b) C3 ∨ C3 .

ch001.tex 30/6/2008 9: 37 Page 15

Fundamental Concepts 15

y3 (x1,y3 ) (x2,y3) (x3,y3)

x2 x3 y2 (x1,y2) (x2,y2) (x3,y2)

x1

y1 (x1,y1) (x2,y1) (x3,y1)

(a) (b) (c)

FIGURE 1.18 (a) G1 , (b) G2 , and (c) G1 × G2 .

(x 1,y3) (x 2,y3) (x3,y3 ) (x 1,y3) (x2 ,y3 ) (x3 ,y3)

(x1 ,y2) (x2,y2 ) (x3,y2) (x 1,y2) (x 2,y2) (x3,y2 )

(x1,y1) (x 2,y1) (x 3,y1) (x 1,y1) (x 2,y1) (x 3,y1)

(a) (b)

FIGURE 1.19 (a) G1 · G2 and (b) G1 ∗ G2 .

length of the shortest path between any two vertices is at most 2. Compared to this
graph, the three-dimensional hypercube is a 3-regular graph with eight vertices such
that the length of the shortest path between any two vertices is at most 3. It has more ver-
tices as compared to the three-dimensional hypercube and a shorter distance between
any two vertices. We call it the simple Petersen graph. As an extension, Öhring and Das
[260] introduce the k-dimensional folded Petersen graph, FPk , to be Pk . It is observed
that FPk possesses qualities of a good network topology for distributed systems with
a large number of sites, because it accommodates 10k vertices and is a symmetric,
3k-regular graph such that the distance between any two vertices is at most 2k. Being
an iterative Cartesian product on the Petersen graph, it is scalable. Moreover, Öhring
and Das [260] define the folded Petersen cube networks FPQn,k as Qn × Pk . It turns
out that FPQn,k and its special derivations FPQ0,k = FPk and FPQn−3,1 = HPn , called
the n-dimensional hyper Petersen network (originally proposed by Das and Banerjee
[79]), are better than the comparable-size hypercubes and several other networks with
respect to the usual metrics of a multicomputer architecture.
Let G1 = (V1 , E1 ) and G2 = (V2 , E2 ) be two graphs. The tensor product of G1 and
G2 , denoted by G1 · G2 , is the graph with vertex set V1 × V2 such that (u1 , v1 ) is joined
to (u2 , v2 ) k times if and only if u1 is joined to u2 m times in G1 and v1 is joined to v2
n times in G2 with k = mn. The strong product of G1 and G2 , denoted by G1 ∗ G2 , is
the graph (G1 × G2 ) ∪ (G1 · G2 ). We illustrate G1 · G2 in Figure 1.19a and G1 ∗ G2 in
Figure 1.19b where G1 and G2 are shown in Figure 1.18.
ch001.tex 30/6/2008 9: 37 Page 16

16 Graph Theory and Interconnection Networks

Note that the adjacency matrix of G1 · G2 is the tensor product of the adjacency
matrix of G1 and the adjacency matrix of G2 . Let m and n be positive integers. It is
apparent that Km ∗ Kn = Kmn .
Let G be any graph and k be any positive integer. We use k G to denote the graph
obtained by duplicating k times to each edge in G. Obviously, 1 G = G. The graph 2 K2
is denoted by C2 .

1.6 SOME BASIC TECHNIQUES

We can count a set by finding a one-to-one correspondence between it and a set of
known size.

PROPOSITION 1.3 For n ≥ 2, there are 2C(n−1;2) simple graphs with the vertex
set {v1 , v2 , . . . , vn } such that every vertex degree is even.
Proof: Let A be the set of simple graphs with the vertex set {v1 , v2 , . . . , vn−1 } and
let B be the set of simple graphs with the vertex set {v1 , v2 , . . . , vn } such that every
vertex degree is even. In Example 1.7, we observed that |A| = 2C(n−1;2) . To prove this
proposition, we set a bijection from A to B as follows.
Let G be a simple graph with vertices v1 , v2 , . . . , vn−1 . We form a new graph G
by adding a vertex vn and making it adjacent to each vertex that has an odd degree in
G, as illustrated in Figure 1.20.
The vertices with an odd degree in G have an even degree in G. Also, vn itself
has even degree because the number of vertices of odd degree in G is even. Con-
versely, deleting the vertex vn from any graph on {v1 , v2 , . . . , vn } with even degrees
produces a graph on {v1 , v2 , . . . , vn−1 }, and this is the inverse of the first proce-
dure. We have established a one-to-one correspondence between the sets. Hence,
|B| = 2C(n−1;2) .

The pigeonhole principle is a simple notion that leads to elegant proofs and may
reduce case analysis.

PROPOSITION 1.4 Every simple graph with at least two vertices has two vertices
of equal degree.
Proof: In a simple graph with n vertices, every vertex degree belongs to the set
{0, 1, . . . , n − 1}. Suppose that fewer than n values occur. By the pigeonhole principle,
the proposition holds. Otherwise, both (n−1) and 0 occur as vertex degrees. However,

vn

G⬘
G

FIGURE 1.20 Illustration of Proposition 1.3.

ch001.tex 30/6/2008 9: 37 Page 17

Fundamental Concepts 17

this is impossible, because the vertex of degree (n − 1) is adjacent to all others, but
the vertex of degree 0 is an isolated vertex. We get a contradiction.

PROPOSITION 1.5 Suppose that G is a simple n-vertex graph with δ(G) ≥

(n − 1)/2. Then G is connected. Moreover, the bound (n − 1)/2 is tight.
Proof: Let G be a simple n-vertex graph with δ(G) ≥ (n − 1)/2. Suppose that u and
v are any two different vertices of G. Suppose that u is not adjacent with v. Because
δ(G) ≥ (n − 1)/2, at least (n − 1) edges join {u, v} to the remaining vertices. However,
there are (n − 2) other vertices. By the pigeonhole principle, one of them receives two
of these edges. Because G is simple, this vertex is a common neighbor of u and v.
In other words, every pair of vertices is either adjacent or has a common neighbor.
Therefore, G is connected.
 graph Kn/2 + Kn/2 has two components. Note that it has a minimum degree
The
of n2 − 1 and is yet disconnected. Thus, the bound is tight.

We can prove that something exists by building it. Such proofs can be implemented
as computer algorithms. A constructive proof requires more than stating an algorithm.
We must also prove that the algorithm terminates and yields result.

THEOREM 1.4 Every loopless graph G has a bipartite subgraph with at least
e(G)/2 edges.
Proof: We start with an arbitrary partition of V (G) into two sets, X and Y . By
including the edges having one endpoint in each set, we obtain a bipartite subgraph
H with bipartition X and Y . Suppose that H contains fewer than half the edges of G
incident to a vertex v. Then v has more neighbors in its own class than in the other
class, as illustrated in Figure 1.21. By moving v to the other class, we gain more edges
of G than we lose.
We make such a local switch in the bipartition as long as the current bipartition
subgraph has a vertex that contributes fewer than half its edges. Each such switch
increases the number of edges in the subgraph. Obviously, the process must termi-
nate. When it terminates, we have degH (v) ≥ degG (v)/2 for every v ∈ V (G). Hence,
e(H) ≥ e(G)/2 by the degree-sum formula.

Example 1.14
The algorithm used in Theorem 3.4 does not necessarily produce a bipartite subgraph
with the most edges—only a bipartite graph with at least half the edges. For example, the
graph in Figure 1.22 is 5-regular with 8 vertices. Hence, it has 20 edges. The bipartition
X = {a, b, c, d} and Y = {e, f , g, h} yields a bipartite subgraph with 12 edges. In this

X Y
v

FIGURE 1.21 Illustration of Theorem 1.4.

ch001.tex 30/6/2008 9: 37 Page 18

18 Graph Theory and Interconnection Networks

h a

g b

f c

e d

FIGURE 1.22 Illustration of Example 1.14.

bipartite graph, each vertex has degree 3. The algorithm terminates here. Any switching
of one vertex would pick up two edges but lose three. Nevertheless, the bipartition
X = {a, b, g, h} and Y = {c, d, e, f } produces a 4-regular bipartite subgraph with 16 edges.
Thus, an algorithm seeking a maximum example by local changes. It may get stuck in a
local maximum.

The proof in Theorem 1.4 illustrates one way to prove the existence of the desired
configuration. Define a sequence of changes to an arbitrary configuration that must
terminate but can terminate only when the desired property occurs.

1.7 DEGREE SEQUENCES

The degree sequence of a graph is the list of vertices’ degrees, usually written in non-
increasing order, as d1 ≥ d2 ≥ · · · ≥ dn . Some applications use nondecreasing order.
Let d1 , d2 , . . . , dn be a sequence of nonnegative integers. Can we determine whether
there is a graph with the required sequence?

PROPOSITION 1.6 The nonnegative

integers d1 , d2 , . . . , dn are the vertex degrees
of some graph if and only if ni=1 di is even.
Proof: Obviously, the
degree-sum formula makes the condition necessary. Con-
versely, suppose that ni=1 di is even. We need to construct a graph with the vertex
, vn and deg(vi ) = di for all i.
set v1 , v2 , . . .

Because ni=1 di is even, the number of odd values is even. We can form an arbi-
trary pairing of {vi | di is odd} and establish an edge for each pair. Now the remaining
degree needed  at each vertex is even and nonnegative. To satisfy this demand for each
i, we put d2i loops at vi . We find the required graph.

The availability of loops makes the construction
easy. If a loop is forbidden,

(2, 0, 0) is not realizable. Thus, that the condition ni=1 di is even is no longer sufficient.
A graphic sequence is a list of nonnegative numbers that is the degree sequence
of a certain simple graph. A simple graph with degree sequence d realizes d.
ch001.tex 30/6/2008 9: 37 Page 19

Fundamental Concepts 19

FIGURE 1.23 Graphs with degree sequences 1, 0, 1 and 2, 2, 1, 1.

FIGURE 1.24 Graph with degree sequence 33333322.

Example 1.15
The list 2, 0, 0 is not graphic. However, the list 2, 2, 1, 1 and the list 1, 0, 1 are graphic.
Obviously, the graph K2 + K1 realizes 1, 0, 1. Suppose that we add a new vertex adjacent
to the isolated vertex and to one vertex of degree 1. We obtain a graph with degree
sequence 2, 2, 1, 1 as illustrated in Figure 1.23. Conversely, if we have a graph realizing
the list 2, 2, 1, 1 in which some vertex w of maximum degree is adjacent to vertices of
degrees 2 and 1, we can delete w to obtain a graph with degree list 1, 0, 1.

Example 1.16
These observations suggest a recursive test for graphic sequences. To test the sequence
33333322, we can seek a realization with a vertex y of degree 3 that has three neighbors of
degree 3. Such graph exists if and only if 2223322 is graphic by deleting y. We rearrange
this as 3322222 and seek a realization having a vertex x of degree 3 with one neighbor
of degree 3 and two neighbors of degree 2. Such a graph exists if and only if 211222 is
graphic by deleting x. We rearrange this as 222211 and seek a realization having a vertex
w of degree 2, with neighbors of degree 2. Such a graph exists if and only if 11211 is
graphic. Perhaps you recognize that this is indeed graphic. Beginning with a realization
11211, we can insert w, y, x with the properties desired to obtain a realization of the
original sequence 33333322 as shown in Figure 1.24. The realization is not unique.

THEOREM 1.5 (Havel [141], Hakimi [135]) For n > 1, the nonnegative integer
list d of size n is graphic if and only if d  is graphic, where d  is the list of size n − 1
obtained from d by deleting its largest element
and subtracting 1 from its
next
largest elements. The only 1-element graph is sequence is d1 = 0.
ch001.tex 30/6/2008 9: 37 Page 20

20 Graph Theory and Interconnection Networks

S
x

y
w

FIGURE 1.25 Illustration of Theorem 1.5.

Proof: The statement is trivial for n = 1. Assume that n > 1. We first prove that the
condition is sufficient. Let d be the sequence d1 ≥ d2 ≥ · · · ≥ dn , and G be a simple
graph with degree sequence d  . We can add a new vertex adjacent to vertices in G hav-
ing degrees d2 − 1, d3 −1, . . . , d
+1 − 1. Note that these di are the
largest elements
of d after (only copy of)
itself. However, the numbers d2 − 1, d3 − 1, . . . , d
+1 − 1
need not be the largest numbers in d  .
To prove necessity, let G be a simple graph with degree sequence d. We want to
produce a simple graph G realizing d  . Let w be a vertex of degree
in G. Let S be
a set of
vertices in G having the desired degrees d2 , d3 , . . . , d
+1 .
Suppose that N(w) = S. We can delete w to obtain G . Suppose that N(w)  = S.
Some vertex of S is missing from N(w). In this case, we will modify G to increase
|N(w) ∩ S| without changing the degree of any vertex. Because N(w) ∩ S can increase
at most
times, repeating this procedure converts an arbitrary G that realizes d into
a graph G∗ that realizes d and has N(w) = S. From G∗ , we then delete w to obtain the
desired graph G realizing d  .
The modification of G that increases |N(w) ∩ S| without changing the degree of
any vertex is described as follows: because deg (w) =
= |S|, there exists a vertex
x ∈ S and another vertex z ∈ / S such that w is adjacent to z and w is not adjacent to x.
By the choice of S, deg(x) ≥ deg(z). Thus, there exists a vertex y ∈ / {x, z, w} such that
y is adjacent to x, but y is not adjacent to z. Now, we modify G by adding the edge set
{(w, x), (z, y)} and deleting the edge set {(w, z), (x, y)} (see Figure 1.25 for illustration).
The desired modification is obtained.

The vertex degree notation for digraphs incorporates the distinction between heads
and tails of edges. Let v be a vertex in a directed graph. The out-neighborhood or
successor set N + (v) is {x ∈ V (G) | v → x}. The in-neighborhood or predecessor set
N − (v) is {x ∈ V (G) | x → v}. The out-degree of v, deg+ (v), is the cardinality of N + (v).
Similarly, the in-degree of v, deg− (v), is the cardinality of N − (v).
ch002.tex 25/7/2008 12: 33 Page 21

2 Applications
Isomorphisms
on Graph

The main goal of this book is to share our research with the readers. We believe that
anyone can do well in research with very little background. So it is time to see what
we can do with the limited background provided in Chapter 1.

2.1 GENERALIZED HONEYCOMB TORI

Stojmenovic [288] introduced three different honeycomb tori by adding wraparound
edges on honeycomb meshes; namely, the honeycomb rectangular torus, the honey-
comb rhombic torus, and the honeycomb hexagonal torus. Recently, these honeycomb
tori have been recognized as an attractive alternative to existing torus interconnection
networks in parallel and distributed applications.
We use the brick drawing [266,288] to define the honeycomb rectangular torus.
Assume that m and n are positive even integers. The honeycomb rectangular torus
HReT(m, n) is the graph with V (HReT(m, n)) = {(i, j) | 0 ≤ i < m, 0 ≤ j < n} such that
(i, j) and (k, l) are adjacent if they satisfy one of the following conditions:

1. i = k and j = l ± 1 (mod n)
2. j = l and k = i − 1 (mod m) if i + j is even

Assume that m and n are positive integers, where n is even. The honeycomb
rhombic torus HRoT(m, n) is the graph with V (HRoT(m, n)) = {(i, j) | 0 ≤ i < m, 0 ≤
j − i < n} such that (i, j) and (k, l) are adjacent if they satisfy one of the following
conditions:

1. i = k and j = l ± 1 (mod n)
2. j = l and k = i − 1 if i + j is even
3. i = 0, k = m − 1, and l = j + m if j is even

Assume that n is a positive integer. The honeycomb hexagonal mesh HM(n)

is the graph with V (HM(n)) = {(x1 , x2 , x3 ) | −n + 1 ≤ x1 , x2 , x3 ≤ n and 1 ≤ x1 + x2 +
x3 ≤ 2}. Two vertices (x11 , x21 , x31 ) and (x12 , x22 , x32 ) are adjacent if and only |x11 − x12 | +
|x21 − x22 | + |x31 − x32 | = 1. The honeycomb hexagonal torus HT(n) is the graph with the
same vertex set as HM(n). The edge set is the union of E(HM(n)) and the wraparound
edge set
{((i, n − i + 1, 1 − n), (i − n, 1 − i, n)) | 1 ≤ i ≤ n}
∪ {((1 − n, i, n − i + 1), (n, i − n, 1 − i)) | 1 ≤ i ≤ n}
∪ {((i, 1 − n, n − i + 1), (i − n, n, 1 − i)) | 1 ≤ i ≤ n}

21
ch002.tex 25/7/2008 12: 33 Page 22

22 Graph Theory and Interconnection Networks

Chou and Hsu [62] unify these three families of honeycomb torus into gener-
alized honeycomb torus. Assume that m and n are positive integers, where n is
even. Let d be any integer such that (m − d) is an even number. The generalized
honeycomb rectangular torus GHT(m, n, d) is the graph with V (GHT(m, n, d)) =
{(i, j) | 0 ≤ i < m, 0 ≤ j < n} such that (i, j) and (k, l) are adjacent if they satisfy one of
the following conditions:

1. i = k and j = l ± 1 (mod n)
2. j = l and k = i − 1 if i + j is even
3. i = 0, k = m − 1, and l = j + d (mod n) if j is even

See Figure 2.1 for various honeycomb tori. Obviously, any GHT(m, n, d) is a 3-
regular bipartite graph. We can label those vertices (i, j) white when i + j is even or
black if otherwise.
It is easy to confirm that the honeycomb rectangular torus HReT(m, n) is isomor-
phic to GHT(m, n, 0) and the honeycomb rhombic torus HRoT(m, n) is isomorphic to
GHT(m, n, m (mod n)). With the following theorem, the honeycomb hexagonal torus
HT(n) is isomorphic to GHT(n, 6n, 3n).

THEOREM 2.1 HT(n) is isomorphic to GHT(n, 6n, 3n).

Proof: Let h be the function from the vertex set of HT(n) into the vertex
set of GHT(n, 6n, 3n) by setting h(x1 , x2 , x3 ) = (x3 , x1 − x2 + 2n) if 0 ≤ x3 < n,
h(x1 , x2 , x3 ) = (0, x1 − x2 + 5n (mod 6n)) if x3 = n, and h(x1 , x2 , x3 ) = (x3 + n, x1 −
x2 + 5n (mod 6n)) if otherwise. We need to check whether f is an isomorphism.
For any 1 − n ≤ c ≤ n, we use Xc to denote the set of those vertices (x1 , x2 , x3 ) in
HT(n) with x3 = c. We use Yc to denote the set of vertices (i, j) in GHT(n, 6n, 3n)
where (1) i = c + n and j ∈ {k | 4n − c − 3 < k < 6n} ∪ {k | 0 ≤ k < n + c} if c < 0,
(2) i = 0 and j ∈ {1 ≤ j < 4n} if c = 0, (3) i = c and { j | c ≤ j ≤ 4n − c} if 0 < c < n,
and (4) i = 0 and j ∈ {k | 4n ≤ k < 6n} ∪ {0} if c = n. Let hc denote the function of h
induced by Xc . It is easy to confirm that hc is a one-to-one function from Xc onto Yc .
Thus, h is one-to-one and onto.
We need to check that h preserves the adjacency. Let e = ((x1 , x2 , x3 ), (x1 , x2 , x3 ))
be an edge of HT(n). Without loss of generality, we assume that x1 + x2 + x3 = 2 and
x1 + x2 + x3 = 1.
Assume that e is an edge of HM(n). Then either x3 = x3 or x3 − x3 = ±1.
Case 1. x3 = x3 . Obviously, either (x1 , x2 , x3 ) = (x1 − 1, x2 , x3 ) or (x1 , x2 , x3 ) =
(x1 , x2 − 1, x3 ) holds.
Assume that 0 ≤ x3 < n. Then h(x1 , x2 , x3 ) = (x3 , x1 − x2 + 2n). Moreover,
h(x1 , x2 , x3 ) = (x3 , x1 − x2 − 1 + 2n) if (x1 , x2 , x3 ) = (x1 − 1, x2 , x3 ) and h(x1 , x2 , x3 ) =
(x3 , x1 − x2 + 1 + 2n) if (x1 , x2 , x3 ) = (x1 , x2 − 1, x3 ). Assume that x3 = n. Then
h(x1 , x2 , x3 ) = (0, x1 − x2 + 5n (mod 6n)). Moreover, h(x1 , x2 , x3 ) = (x3 , x1 − x2 − 1 +
5n (mod 6n)) if (x1 , x2 , x3 ) = (x1 − 1, x2 , x3 ) and h(x1 , x2 , x3 ) = (x3 , x1 − x2 + 1 +
5n (mod 6n)) if (x1 , x2 , x3 ) = (x1 , x2 − 1, x3 ). Assume that x3 < 0. Then h(x1 , x2 , x3 ) =
(x3 + n, x1 − x2 + 5n (mod 6n)). Moreover, h(x1 , x2 , x3 ) = (x3 , x1 − x2 − 1 + 5n
(mod 6n)) if (x1 , x2 , x3 ) = (x1 − 1, x2 , x3 ) and h(x1 , x2 , x3 ) = (x3 , x1 − x2 + 1 + 5n
ch002.tex 25/7/2008 12: 33 Page 23

Applications on Graph Isomorphisms 23

(a) (b)

x1

x3

x3 x3

x3 x3
x2
x3 x3

(c) (d)

FIGURE 2.1 (a) HReT(6, 6), (b) HRoT(4, 6), (c) HT(3), and (d) GHT(3, 18, 9).

(mod 6n)) if (x1 , x2 , x3 ) = (x1 , x2 − 1, x3 ). Therefore, h(x1 , x2 , x3 ) and h(x1 , x2 , x3 ) are
adjacent.
Case 2. x3 − x3 = ±1. Because x1 + x2 + x3 = 2 and x1 + x2 + x3 = 1, (x1 , x2 , x3 ) =
(x1 , x2 , x3 − 1).
Assume that 1 ≤ x3 < n. Then h(x1 , x2 , x3 ) = (x3 , x1 − x2 + 2n) and h(x1 , x2 , x3 ) =
(x3 − 1, x1 − x2 + 2n). Assume that x3 = 0. Then h(x1 , x2 , x3 ) = (0, x1 − x2 + 2n)
and h(x1 , x2 , x3 ) = (n − 1, x1 − x2 + 5n (mod 6n)). Assume that x3 = n. Then
h(x1 , x2 , x3 ) = (0, x1 − x2 + 5n (mod 6n)) and h(x1 , x2 , x3 ) = (n − 1, x1 − x2 + 2n).
Assume that 2 − n ≤ x3 ≤ −1. Then h(x1 , x2 , x3 ) = (x3 + n, x1 − x2 + 5n (mod 6n))
ch002.tex 25/7/2008 12: 33 Page 24

24 Graph Theory and Interconnection Networks

and h(x1 , x2 , x3 ) = (x3 + n − 1, x1 − x2 + 5n (mod 6n)). Therefore, h(x1 , x2 , x3 ) and
h(x1 , x2 , x3 ) are adjacent.
Assume that e is an wraparound edge of HM(n). We have the following three
cases:
Case 3. e ∈ {((i, n − i + 1, 1 − n), (i − n, 1 − i, n)) | 1 ≤ i ≤ n}. Obviously, (x1 , x2 , x3 ) =
(i, n − i + 1, 1 − n) and (x1 , x2 , x3 ) = (i − n, 1 − i, n). Thus, h(x1 , x2 , x3 ) is (1, 4n +
2i − 1) and h(x1 , x2 , x3 ) is (0, 4n + 2i − 1). Therefore, h(x1 , x2 , x3 ) and h(x1 , x2 , x3 )
are adjacent.
Case 4. e ∈ {((1 − n, i, n − i + 1), (n, i − n, 1 − i)) | 1 ≤ i ≤ n}. Obviously, (x1 , x2 , x3 ) =
(1 − n, i, n − i + 1) and (x1 , x2 , x3 ) = (n, i − n, 1 − i). Thus, h(x1 , x2 , x3 ) is (0, 4n) if
i = 1 and (n − i + 1, n − i + 1) if 1 < i ≤ n. Again, h(x1 , x2 , x3 ) is (0, 4n − 1) if i = 1
and (n − i + 1, n − i) if 1 < i ≤ n. Therefore, h(x1 , x2 , x3 ) and h(x1 , x2 , x3 ) are adjacent.
Case 5. e ∈ {((i, 1 − n, n − i + 1), (i − n, n, 1 − i)) | 1 ≤ i ≤ n}. Obviously, (x1 , x2 , x3 ) =
(i, 1 − n, n − i + 1) and (x1 , x2 , x3 ) = (i − n, n, 1 − i). Thus, h(x1 , x2 , x3 ) is (0, 0) if
i = 1 and (n − i + 1, 3n + i − 1) if 1 < i ≤ n. Again, h(x1 , x2 , x3 ) is (0, 1) if i = 1 and
(n − i + 1, 3n + i) if 1 < i ≤ n. Hence, h(x1 , x2 , x3 ) and h(x1 , x2 , x3 ) are adjacent.
Thus, the theorem is proved.

For example, the honeycomb torus illustrated in Figure 2.1c is actually isomorphic
to the generalized honeycomb torus illustrated in Figure 2.1d.

2.2 ISOMORPHISM BETWEEN CYCLIC-CUBES

AND WRAPPED BUTTERFLY NETWORKS
The wrapped-around butterfly network WB(n, k) has n · k n vertices, and each vertex
is represented by (n + 1)-bit vectors a0 a1 . . . an−1 i where 0 ≤ i ≤ n − 1 and 1 ≤ aj ≤ k
for all 0 ≤ j ≤ n − 1. Two vertices a0 a1 . . . an−1 i and b0 b1 . . . bn−1 j are adjacent in
WB(n, k) if and only if j − i = 1 (mod n) and at = bt for all 0 ≤ t  = j ≤ n − 1. The
graph in Figure 2.2 is WB(3, 2). The WB(n, k) is widely used in communication
networks [220].

row 000
row 001
row 010
row 011
0112 row 100
row 101
row 110
row 111

FIGURE 2.2 The graph WB(3, 2).

ch002.tex 25/7/2008 12: 33 Page 25

Applications on Graph Isomorphisms 25

In 1998, Fu and Chau [118] proposed a family of graphs called cyclic-cubes,

which have even fixed degrees. They show that this family of graphs has many useful
properties, so that it can serve as an interconnection network topology. Let Gkn denote
k-ary n-dimensional cyclic-cubes. In 2000, Hung et al. [184] proved that this family
of graphs is indeed isomorphic to WB(n, k).
To define Gkn , let t1 , t2 , . . . , tn be n distinct symbols with ordering t1 > t2 · · · > tn .
Each symbol tj is assigned a rank i for 1 ≤ i ≤ k, and this ranked symbol is denoted
by tji . The graph Gkn has n · k n vertices, and each vertex of Gkn is represented by an
n-bit vector, which is a circular permutation of t1i1 t2i2 . . . tnin for 1 ≤ i1 , i2 , . . . , in ≤ k.
For example, in G24 t31 t42 t11 t22 is a vertex and t32 t21 t42 t11 is not. In other words,
    
i ij+1 ij−1 
V Gkn = tj j tj+1 . . . tnin t1i1 . . . tj−1  for 1 ≤ j ≤ n and 1 ≤ i1 , i2 , . . . , in ≤ k

To define edges in Gkn , we first define the function fs for every 1 ≤ s ≤ k, mapping
V (Gkn ) onto itself as follows:
 
i ij+1 ij−1 ij+1 ij−1 s
fs tj j tj+1 . . . tnin t1i1 . . . tj−1 = tj+1 . . . tnin t1i1 . . . tj−1 tj for any 1 ≤ s ≤ k

Note that all fj are bijective functions. Each vertex x ∈ V (Gkn ) is linked to exactly 2k
vertices fj (x) and fj−1 (x) for all 1 ≤ j ≤ k. For example, in G24 the vertex t31 t41 t11 t22 is
linked to t41 t11 t22 t31 , t41 t11 t22 t32 , t21 t31 t41 t11 , and t22 t31 t41 t11 .

THEOREM 2.2 Gkn is isomorphic to WB(n, k).

Proof: For each vertex a0 a1 . . . an−1 i in WB(n, k), we define a function π mapping
V (WB(n, k)) to V (Gkn ) as follows:
+1 ai +1 ai−2 +1
ti+1 . . . tnan−1 +1 t1a0 +1 t2a1 +1 . . . ti−1
a
π (a0 a1 . . . an−1 i) = ti i−1

For example, n = 3, k = 2, f (0000) = t11 t21 t31 , f (0001) = t21 t31 t11 , and f (0112) = t32 t11 t22 .
Obviously, the function π is bijective.
Let u = a0 a1 . . . an−1 i and v = b0 b1 . . . bn−1 j be two distinct vertices in WB(n, k).
Then π(u) and π(v) are two distinct vertices in Gkn given as follows:
+1 ai +1 ai−2 +1
ti+1 . . . tnan−1 +1 t1a0 +1 t2a1 +1 . . . ti−1
a
π(u) = ti i−1
+1 bj +1 bj−2 +1
tj+1 . . . tnbn−1 +1 t1b0 +1 t2b1 +1 . . . tj−1
b
π(v) = tj j−1

Assume that u and v are adjacent in WB(n, k). Without loss of generality, we
may assume that j = i − 1 (mod n). Thus, at = bt for all 0 ≤ t  = j ≤ n − 1; that is,
v = a0 a1 . . . ai−2 bi−1 ai . . . an−1 (i − 1). Therefore,
ai +1 i+1 +1 +1 bi−1 +1
· · · tnan−1 +1 t1a0 +1 t2a1 +1 · · · ti−1
a a
π(v) = ti+1 ti+2 i−2
ti = fbi−1 (π(u))

Thus, π(u) and π(v) are adjacent in Gkn .

ch002.tex 25/7/2008 12: 33 Page 26

26 Graph Theory and Interconnection Networks

On the other hand, suppose that π(u) and π(v) are adjacent in Gkn . Then
π(v) can be fs (π(u)) or fs−1 (π(u)) for some 1 ≤ s ≤ k. Suppose that π(v) = fs (π(u))
ai +1 ai+1 +1 a +1 ai−2 −1 s+1
for some 1 ≤ s ≤ k. Then π(v) = ti+1 ti+2 . . . tn n−1 t1a0 +1 t2a1 +1 . . . ti−1 ti
and v = a0 a1 . . . ai−1 sai . . . an−1 (i − 1). Therefore, (u, v) ∈ E(WB(n, k)). Similarly,
π(v) = fs−1 (π(u)) also implies (u, v) ∈ E(WB(n, k)).
This theorem is proved.

2.3 1-EDGE FAULT-TOLERANT DESIGN FOR MESHES

Motivated by the study of computer and communication networks that can tolerate
failure of their components, Harary and Hayes [139] formulated the concept of edge
fault tolerance in graphs. Let G be a graph of order n and k be a positive integer. An
n-vertex graph G∗ is said to be k-edge fault-tolerant, or k-EFT, with respect to G, if
every graph obtained by removing any k-edges from G∗ contains G. For brevity, we
refer to G∗ as k-EFT(G) graph. A k-EFT(G) graph G∗ is optimal if it contains the least
number of edges among all k-EFT(G) graphs. Let eftk (G) be the difference between
the number of edges in an optimal k-EFT(G) graph and that in G.

LEMMA 2.1 Assume that G∗ is a k-EFT(G). Then δ(G∗ ) ≥ δ(G) + k.

Proof: Suppose that the lemma is false. Then there exists a k-EFT(G) graph G∗ with
δ(G∗ ) < δ(G) + k. Let v be a vertex of G∗ with degG∗ (v) = δ(G∗ ). We delete k edges
that are incident with v from G∗ to obtain a graph G. Obviously, δ(G ) < δ(G). Hence,
G is not a subgraph of G. We get a contradiction, because G∗ is a k-EFT(G). Hence,
the lemma is proved.

It is easy to see that k+1 G is a k-EFT(G) for any graph G. Moreover, Cn is an

optimal 1-EFT(Pn ). It is easy to confirm that the torus graph C(m1 , m2 , . . . , mn ) is
a 1-edge fault-tolerant graph for M(m1 , m2 , . . . , mn ) where mi ≤ 2 with 1 ≤ i ≤ n. In
Ref. 139, Harary and Hayes conjectured that C(m1 , m2 , . . . , mn ) is optimal 1-EFT
(M(m1 , m2 , . . . , mn )) for every mi ≥ 2 with 1 ≤ i ≤ n. The two-dimensional mesh
M(3, 4) is illustrated in Figure 2.3a. The graph C(3, 4) with a faulty (dashed) edge
F = (x2,3 , x2,4 ) is illustrated in Figure 2.3b. A recovery from a fault affecting the dashed
edge F = (x2,3 , x2,4 ) is shown in Figure 2.3c.
Chou and Hsu [64] disprove the conjecture by proposing the following counterex-
ample. We assume that the vertices of M(m1 , m2 , . . . , mn ) are labeled canonically.
Thus, xi1 ,i2 ,...,in is a vertex of M(m1 , m2 , . . . , mn ) if and only if 1 ≤ ij ≤ mj for 1 ≤ j ≤ n.

x31 x32 x33 x34 x31 x32 x33 x34 x34 x31 x32 x33
x21 x22 x23 x24 x21 x22 x23 x24 x24 x21 x22 x23

x11 x 12 x13 x14 x11 x12 x13 x14 x14 x11 x12 x13
(a) (b) (c)

FIGURE 2.3 (a) The graph M(3, 4), (b) the graph C(3, 4) with a faulty (dashed) edge
F = (x2,3 , x2,4 ), and (c) a recovery from a fault affecting the dashed edge F = (x2,3 , x2,4 ).
ch002.tex 25/7/2008 12: 33 Page 27

Applications on Graph Isomorphisms 27

Moreover, xi1 ,i2 ,..., in is adjacent to another vertex xj1 , j2 ,..., jn if all indices t with 1 ≤ t ≤ n
but one index k satisfy it = jt ; and the index k is such that ik  = jk satisfies |ik − jk | = 1.
Let Vp = {xi1 ,i2 ,...,in | ik = 1 or mk for some 1 ≤ k ≤ n} be the set of peripheral vertices.
Assume that xi1 ,i2 ,..., in is a vertex in Vp . The antipodal vertex of xi1 ,i2 ,...,in is xj1 , j2 ,..., jn ,
with jk = mk − ik + 1, which is another vertex in Vp . It is easy to confirm that every
vertex in Vp has exactly one antipodal. In M(m1 , m2 , . . . , mn ), we add the edges joining
each vertex in Vp to its antipodal counterpart to form a new graph P(m1 , m2 , . . . , mn ).
We call these P(m1 , m2 , . . . , mn ) projective-plane graphs because their construction
is similar to that of the projective plane when n = 2.

THEOREM 2.3 P(m1 , m2 , . . . , mn ) is 1-EFT(M(m1 , m2 , . . . , mn )) and it contains

fewer edges than that of C(m1 , m2 , . . . , mn ).
Proof: Suppose that F is the faulty edge of P(m1 , m2 , . . . , mn ) joining xi1 ,i2 ,..., in to
xi1 ,i2 ,...,in where is = is if s  = k and ik = ik + 1. Now, we reconfigure M(m1 , m2 , . . . , mn )
through the following. First we delete all the edges joining xj1 , j2 ,..., jn to xj1 , j2 ,..., jn
where js = js for every s  = k, jk = ik , and jk = ik + 1. Let A = {xj1 , j2 ,..., jn | jk ≤ ik }
and B = {xj1 , j2 ,..., jn | jk ≥ jk + 1}. In M(m1 , m2 , . . . , mn ), the sets A and B induce
two submeshes. These two submeshes are connected by edges including the set
E ∗ = {(xj1 , j2 ,..., jn , xj1 , j2 ,..., jn ) | js = ms − js + 1 for every s  = k, jk = 1, and jk = mk }.
Therefore, the subgraphs of P(m1 , m2 , . . . , mn ) generated by A, B, and E ∗ form a
mesh M  isomorphic to M(m1 , m2 , . . . , mn ). For example, the recovery of M(3, 4) for
the faulty edge F = (x2,3 , x2,4 ) in P(3, 4) is illustrated in Figure 2.4.
Obviously, |E(P(m1 , m2 , . . . , mn ))| ≤ |E(C(m1 , m2 , . . . , mn ))|. Hence, the theo-
rem is proved.

n n
COROLLARY 2.1 eft1 (M(m1 , m2 , . . . , mn )) ≤ 1/2 i=1 mi − i=1 (mi − 2) .
The projective-plane graphs are optimal for some cases, but not for all. Note
that every n-dimensional hypercube can be viewed as the mesh M(2, 2, . . . , 2). Our
P(2, 2, . . . , 2) is actually the same 1-EFT graph as that proposed in Refs 34, 139,
319, and 347. Thus, P(2, 2, . . . , 2) is an optimal 1-EFT graph. It is easy to prove
that the graph in Figure 2.5a is a 1-EFT(M(3, 2)) and the graph in Figure 2.5b is a
1-EFT(M(4, 2)). With these two examples, we know that the projective-plane graphs
may not be optimal for some cases. Furthermore, the problem of finding the optimal
1-EFT for all n-dimensional meshes remains unsolved.
In Ref. 67, Chuang et al. study the 1-EFT graphs for M(k, 2) with k ≥ 2. For
simplicity, the kth ladder graph Lk is defined to be M(k, 2). The vertices of Lk can be

x31 x32 x33 x34 x14 x31 x32 x33

x21 x22 x23 x24 x24 x21 x22 x23

x11 x12 x13 x14 x34 x11 x12 x13

FIGURE 2.4 The recovery of M(3, 4) for the faulty edge F = (x2,3 , x2,4 ) in P(3, 4).
ch002.tex 25/7/2008 12: 33 Page 28

28 Graph Theory and Interconnection Networks

x1,2 x2,2 x3,2 x1,2 x2,2 x3,2 x4,2

x1,1 x2,1 x3,1 x1,1 x2,1 x3,1 x4,1

(a) (b)

FIGURE 2.5 (a) A 1-EFT(M(3, 2)), L3∗ and (b) a 1-EFT(M(4, 2)), L4∗ .

labeled by xi, j with 1 ≤ i ≤ k and 1 ≤ j ≤ 2 canonically. The vertices x1,1 , xk,1 , x1,2 ,
and xk,2 are called the corner vertices of Lk . We have the following theorem.

THEOREM 2.4 Assume that Lk∗ is a 1-EFT(Lk ) graph. Then (i) degLk∗ (x) ≥ 3 for
any vertex x of Lk∗ and (ii) eft1 (Lk ) ≥ 2.
Proof: Assume that some vertex x with degLk∗ (x) = 2. Let e be any edge incident
with x. Then degLk∗ −e (x) = 1. Because degLk (x) ≥ 2 for any vertex x of Lk , Lk is not
a subgraph of Lk∗ − e. We find a contradiction to Lk∗ being a 1-EFT(Lk ) graph. Hence,
degLk∗ (x) ≥ 3. Because there are exactly four corner vertices in every Lk , we have
eft1 (Lk ) ≥ 2.

COROLLARY 2.2 eft1 (Lk ) > 2 if k > 4.

Proof: It is easy to check whether there are exactly three different ways of joining the
four corner vertices in Lk with two edges; namely, {(x1,1 , x1,2 ), (xk,1 , xk,2 )}, {(x1,1 , xk,1 ),
(x1,2 , xk,2 )}, and {(x1,1 , xk,2 ), (x1,2 , xk,1 )}. It is observed that none of the graphs obtained
by joining two edges to the corner vertices of Lk with k > 4 is 1-EFT(Lk ). Hence,
eft1 (Lk ) > 2 if k > 4.

Let L2∗ (L3∗ , and L4∗ , respectively) be the graph P(2, 2) (the graph in Figures 2.5a,
and 2.5b, respectively). From the previous discussion, Lk∗ is 1-EFT(Lk ) for k = 2, 3,
and 4. Note that there are exactly two edges added to Lk with k = 2, 3, and 4. By
Theorem 2.4, these graphs are optimal. By checking all three cases joining two edges
to the corner vertices of Lk , the optimal 1-EFT(Lk ) is unique for k = 2, 3, and 4. We
obtain the following theorem.

THEOREM 2.5 eft1 (Lk ) = 2 for k = 2, 3, and 4.

Let us consider the spanning supergraph L5∗ of L5 given by E(L5∗ ) = E(L5 ) ∪
{(x1,1 , x5,2 ), (x1,2 , x4,2 ), (x2,1 , x5,1 )} as shown in Figure 2.6.
Obviously, edges of L5 can be divided into the following seven classes: namely,
A = {(x1,1 , x1,2 )}, B = {(xi,1 , xi,2 ) | 2 ≤ i ≤ 4}, C = {(x5,1 , x5,2 )}, D = {(x1,1 , x2,1 ), (x1,2 ,
x2,2 )}, E = {(x2,1 , x3,1 ), (x2,2 , x3,2 )}, F = {(x3,1 , x4,1 ), (x3,2 , x4,2 )}, and G = {(x4,1 , x5,1 ),
(x4,2 , x5,2 )}. We can reconfigure L5 in L5∗ for any faulty edge e in A, B, C, D, E, F,
and G, respectively, as shown in Figures 2.7a through 2.7g, respectively. Thus, L5∗ is
1-EFT(L5 ). By Corollary 2.2, we have the following theorem.
ch002.tex 25/7/2008 12: 33 Page 29

Applications on Graph Isomorphisms 29

x1,2 x2,2 x3,2 x4,2 x5,2

x1,1 x2,1 x3,1 x4,1 x5,1

FIGURE 2.6 A 1-EFT(M(5, 2)), L5∗ .

x2,2 x3,2 x3,1 x2,1 x1,1 x2,2 x1,2 x1,1 x2,1 x3,1 x4,1 x3,1 x3,2 x4,2 x5,2 x1,2 x4,2 x4,1 x3,1 x3,2

x1,2 x4,2 x4,1 x5,1 x5,2 x3,2 x4,2 x5,2 x5,1 x4,1 x5,1 x2,1 x2,2 x1,2 x1,1 x1,1 x5,2 x5,1 x2,1 x2,2

(a) (b) (c) (d)

x3,1 x4,2 x5,1 x2,1 x2,2 x3,2 x2,2 x1,2 x4,2 x4,1 x5,2 x1,1 x1,2 x4,2 x4,1

x3,2 x4,2 x5,2 x1,1 x1,2 x3,1 x2,1 x1,1 x5,2 x5,1 x5,1 x2,1 x2,2 x3,2 x3,1
(e) (f) (g)

FIGURE 2.7 A 1-EFT(M(5, 2)), L5∗ .

THEOREM 2.6 eft1 (L5 ) = 3.

Let k be an even integer with k ≥ 4. The spanning supergraph Lk∗ of Lk is the graph
that adds E  = {(xi, j , xk−i+1, j ) | 1 ≤ i < k/2, j = 1, 2} to E(Lk ) as shown in Figure 2.8a.
The graph in Figure 2.8a is actually isomorphic to M(k/2, 2, 2), as shown in Fig-
ure 2.8b. We can reconfigure Lk in Lk∗ as shown in Figure 2.8c for any faulty edge of the
form (xi,1 , xi,2 ) or as shown in Figure 2.8d for any faulty edge of the form (xi,1 , xi+1,1 )
or (xi,2 , xi+1,2 ). Thus, M(k/2, 2, 2) is a 1-EFT(Lk ). We obtain the following theorem.

THEOREM 2.7 Assume that k is an even integer with k ≥ 4. Then eft1 (Lk ) ≤ k − 2.

Let k be an odd integer with k ≥ 7. Construct the spanning supergraph Lk∗

of Lk by adding E  = {(x1,2 , x4,2 ), (x3,2 , x6,2 ), (x2,1 , x5,1 ), (x4,1 , x7,1 ), (x1,1 , x5,2 ),
(x3,1 , x7,2 )} ∪ {(x2i, j , x2i+3, j ) | 3 ≤ i ≤ (k − 3)/2, j = 1, 2} as shown in Figure 2.9.
Obviously, edges of Lk can be divided into the following seven classes:

A = {(xi,1 , xi,2 ) | i = 1, 2} ∪ {(x2i, j , x2i+1, j ) | 4 ≤ i ≤ (k − 3)/2, j = 1, 2}

B = {(xi,1 , xi,2 ) | i = 3, 4} ∪ {(x2i−1, j , x2i, j ) | 4 ≤ i ≤ (k − 1)/2, j = 1, 2}
C = {(x5,1 , x5,2 )} ∪ {(xi,1 , xi,2 ) | 4 ≤ i ≤ (k − 1)/2}
D = {(xi,1 , xi,2 ) | i = 6, 7} ∪ {(xi,1 , xi,2 ) | i = k, k − 1}
E = {(x1, j , x2, j ) | j = 1, 2} ∪ {(x3, j , x4, j ) | j = 1, 2}
F = {(x2, j , x3, j ) | j = 1, 2} ∪ {(x5, j , x6, j ) | j = 1, 2}
G = {(x4, j , x5, j ) | j = 1, 2} ∪ {(x6, j , x7, j ) | j = 1, 2} ∪ {(xk−1, j , xk, j ) | j = 1, 2}
 xk,2 xk
1,2 xk
2,2 xk
i
1,2 xk
i,2 xk/23,2 xk/22,2 xk/21,2
30
ch002.tex

xk,1 xk
1,1 xk
2,1 xk
i
1,1 xk
i,1 xk/23,1 xk/22,1 xk/21,1

x1,2 x 2,2 x i,2 xi1,2 xk/2
1,2 xk/2,2 xk/21,2 xk/22,2 xk
i,2 xk
i1,2 xk
1,2 xk,2
x1,1 x 2,1 x i,1 xi1,1 xk/2
1,1 xk/2,1 xk/21,1 xk/22,2 xk
i,1 xk
i1,1 xk
1,1 xk,1
25/7/2008

x1,2 x2,2 x3,2 xi,2 xi1,2 xk/2
2,2 xk/2
1,2 xk/2,2

12: 33

x1,1 x2,2 x3,1 xi,1 xi1,1 xk/2
2,1 xk/2
1,1 xk/2,1

(a) (b)
Page 30

xi,2 xi,2 x
i1,2

xi,1 xi,1 xi1,1

(c) (d)

FIGURE 2.8 (a) a 1-EFT(Lk ) where k is even and k ≥ 4, (b) the three-dimensional mesh M(k/2, 2, 2), (c) reconfigured Lk for any faulty edge of the
Lk∗ ,
form (xi,1 , xi,2 ), and (d) reconfigured Lk for any faulty edge of the form (xi,1 , xi+1,1 ) or (xi,2 , xi+1,2 ).
Graph Theory and Interconnection Networks
Another Random Scribd Document
with Unrelated Content
but the endeavour in calling that New Parliament had been, not to
choose any but such as had hope and faith in Christ. "The Lord," he
observed, "shakes the hills and mountains, and they reel; and God
hath a hill too, an high hill as the hill of Bashan, and the chariots of
God are twenty thousand, even thousands of angels, and God will
dwell upon this hill for ever." Such a speech had never before been
delivered at the opening of a Parliament.
When it was finished, a formal instrument devolving authority on
the members was placed on the table, with the General's signature
and seal, after which he left the Chamber,—politically "nothing more
than the brewer's son of Huntingdon."[47]
Francis Rouse, provost of Eton, was elected
speaker—before which proceeding the members Little Parliament.
prayed without a chaplain—"Eight or ten speaking
in prayer to God, and some briefly from the Word." "Much of the
presence of Christ and His Spirit," says a person who was present,
"appearing that day to the great gladdening of the hearts of many—
some affirming they never enjoyed so much of the spirit and
presence of Christ, in any of the meetings and exercises of religion,
in all their lives, as they did that day." On the day following they
again spent some time in prayer, and prayer, it is recorded, "was
daily performed by one member or other, as they were found free to
perform it." Presently again they devoted a day to prayer, which
"was done by the members, principally by such as had not done
service before, when also the Lord General was present, and it was a
very comfortable day."[48] The Lord General had been specially
invited to join the Assembly, together with Harrison, Desborough,
Lambert, and Tomlinson.
Major-General Harrison—a rather noble though not refined-
looking man, with flowing locks, irregular features, aquiline nose,
and black eyes, often flashing with enthusiasm—had distinguished
himself at Basinghouse, Preston, and elsewhere, as a stern soldier,
and had been heard by Baxter breaking out into a rapture at
Somerton. He was now getting deep in the study of prophecy, and
was expecting the reign of the saints to succeed the four great
monarchies described by the prophet Daniel. Desborough—a ruder
and coarser man than Harrison, as his face, eyebrows, form, and
gait all betokened—who had shewn himself a gallant soldier in the
storming of Bristol, and had become brother-in-law to Oliver
Cromwell—was an Independent, and, after the Restoration, became
a member of Dr. Owen's church, in London. Major General Lambert—
who had done good service in the wars, had routed the Scots at
Linlithgow, and achieved daring feats at Worcester—was Cromwell's
particular friend, and adhered to him throughout his career, when
others turned their backs; a shrewd, clever, practical sort of person.
Tomlinson, Colonel of the Guard at King Charles' execution, was not,
beyond that circumstance, at all a noteworthy individual.[49]
The Little Parliament altered the marriage law,
[50] 1653, July.
which, owing to recent confusion, and
consequent irregularities in domestic life, needed amendment.
Matrimony was considered by these new legislators in its relation to
the State, and was treated by them simply in the character of a civil
contract; possibly, in part at least, with the view of diminishing
clerical influence, but also with so remarkable an insight into what is
just and wise, as to anticipate modern legislation in this respect both
in England and on the Continent. Parties were to obtain a certificate
from the registrar of the parish, and then solemnly before a justice
of the peace to take each other for husband and wife. The religious
sanctions of the wedding bond were left entirely to the will of the
parties united, and these sometimes were so connected with the
secular part of the ceremony, that the service altogether resembled
the solemnization of matrimony at the present day in a
Nonconformist church in the presence of a registrar.[51]
The Parliament had scarcely commenced its
sittings when it entered upon the consideration of Question of
Tithes.
the important subject of tithes; and on the 15th of
July, it was determined by a majority of twenty-five—in a House
consisting of one hundred and eleven members—that the
maintenance of ministers by tithes should not be continued after the
3rd day of November following. On the 19th of July, upon a renewal
of the debate relative to this subject, the question whether
incumbents possessed a propriety in this kind of income, was
referred to the consideration of a committee specially appointed for
that purpose. And to this same committee a different business was
committed on the 26th of the next month. On that day were
presented petitions from several churches in Devonshire and
Gloucestershire, seeking the further reformation of religion, in
connexion with which it was resolved: "That a committee be
appointed to consider of some way to be propounded to the House,
how ignorant, profane, and scandalous ministers may be rejected;
That it be referred to the same committee to consider of some way
to be propounded to the House, for the encouragement of such
godly and able persons as shall preach the Gospel; and That it be
referred to the committee for tithes."[52] The consideration of the
question as to whether incumbents had any property in tithes having
been previously entrusted to this committee, that question now
became mixed up with the other relating to the character of the
clergy; and what the Little Parliament ultimately did respecting the
one has been supposed to have been a legislative decision of the
other, which is not the fact.
Two parties were in contention, one disposed to
retain the old tithe system, the other bent upon 1653.
supporting ministers in some other way. Harrison
led the latter division, and occupied the extreme left in relation to
the moderates, who were swayed by Cromwell, and who, as to
tithes, agreed with both the Presbyterian and Episcopal parties. The
House generally concurred in the opinion that the collection of tithes
by the clergy was a grievance; yet perhaps only a few members
were prepared to vote for putting an end to that method of support
without the provision of some legal substitute. These few, following
Harrison, were intent on having the impost repealed at once, leaving
only such other provision "as God should direct." A distinction was
admitted throughout the debate between the claims of impropriators
and the claims of incumbents. The whole House was willing to
compensate impropriators in case of the forfeiture of their rights. It
was the case of incumbents alone which came under the
consideration of the committee. The utmost measure of change
formally proposed was, to put an end to the payment of the clergy in
the old way, and to equalize benefices by reducing those of £200 per
annum and upwards, and by increasing smaller incomes. It was also
suggested that a provision should be made to meet the wants of
ministers' widows and children.[53]
The larger question respecting ignorant,
profane, and scandalous ministers, involved also Question of
Tithes.
the minor one touching presentation to benefices;
the predominant feeling of the members favoured the right of
congregations to choose their own pastors. The Puritan current had
long been setting in that direction, and arguments were now urged
to the effect that it was unreasonable for people not to be allowed to
select their own spiritual guides; much, in short, being advanced
upon the subject, of the same kind as is common in the present day.
The parties having a pecuniary interest in the maintenance of things
as they were, did all they could to keep on their side the men in
power. Such people sought protection from the Lord General against
what they deemed Parliament-robbery. However, on the 17th of
November, the right of presentation to benefices by patrons was
condemned by a resolution, and a bill in accordance with it was
ordered to be brought in. The effect of this decision was to place the
election of the clergy in the hands of parishioners.
On the 2nd of December came the long-
expected and much-dreaded report of the Tithe 1653, December.
Committee. It first recommended the sending forth
of certain authorized commissioners to enquire into ministerial
character, and next it gave a deliverance with regard to ministerial
maintenance. As this portion of the report is so imperfectly explained
by historians, we will present it to the reader as it is printed in the
journals:—
 "Resolved—That it be presented to the Parliament: That all such
as are, or shall be approved for public preachers of the Gospel in the
public meeting-places, shall have and enjoy the maintenance already
settled by law, and such other encouragement as the Parliament
already hath appointed, or hereafter shall appoint; and that where
any scruple payment of tithes, the three next justices of the peace,
or two of them, shall, upon complaint, call the parties concerned
before them, and by the oaths of lawful witnesses shall duly
apportion the value of the said tithes, to be paid either in money or
land, by them to be set out according to the said value, to be held
and enjoyed by him that was to have the said tithes; and, in case
such apportioned value be not duly paid or enjoyed, according to the
order of the said justices, the tithes shall be paid in kind, and shall
be recovered in any court of record. Upon hearing and considering
what hath been offered to this Committee touching propriety in
tithes of incumbents, rectors, possessors of donatives, or propriate
tithes, it is the opinion of this Committee, and resolved to be so
reported to the Parliament, that the said persons have a legal
propriety in tithes."[54]
An earnest debate ensued, "managed day by day," says one who
was present, "with very great seriousness, many arguments and
Scriptures being alleged," and "very little of heat or passion being
shewed all that time, only one gentleman or two that were for the
report, seeing themselves and their party so engaged, flew out a
little, complaining of the expense of time, to have given a check to
the going on of the debate."[55] The first part of the report, relating
to the method of removing scandalous ministers, upon being put to
the vote, was rejected by a majority of two, fifty-six voting against
fifty-four. The second part, relating to the mode of supporting
ministers, and the rights of property possessed by incumbents, was
not put to the vote at all. The probable rejection of it might be
inferred, but no formal rejection was expressed. What is sometimes
represented as the decided abolition of tithes amounted to no more
than the rejection of a Committee of Triers. That Committee was a
favourite scheme of Cromwell's, and was afterwards by him
practically carried out; he also favoured the continuance of the old
method of supporting ministers. But the Little Parliament indicated a
wish to change that method; yet what they decided now, and even
their distinct votes against tithes in the preceding July, did not
necessarily imply that they intended to terminate altogether the
state maintenance of religious worship.[56]
It is desirable here once more to pause, and to
consider the opinions of Fifth Monarchy men, who Fifth Monarchy
Men.
were at this time becoming very numerous and
very active.
They may be divided into three classes. The first was composed
of mere millenarians, who entertained views not essentially different
from those which had been held in ancient times with regard to the
reign of Christ upon the earth; and whatever may be thought of
these persons in some other respects, their opinions cannot be
regarded as involving anything discreditable to their reputation,
inasmuch as in substance those opinions had received the sanction
of a great scholar, Henry Mead, and of a distinguished philosopher,
Henry Moore.
The second class consisted of theoretical
theocrats, people who talked in an extravagant 1653.
manner respecting Divine dominion, and generally
opposed the authority of Oliver Cromwell; yet they appear to have
been inoffensive persons, not at all disposed to attempt any violent
measures for the realization of their wild and mystical dreams. John
Tillinghurst, the now forgotten author of several publications on
prophecy, which were popular at the time of which we are speaking,
belonged to this order of anti-Cromwell millenarians, who were very
bold and busy, but for the most part very harmless. Some of them
wrote and preached with great confidence upon the prophecies of
Daniel and the Revelations of St. John, only to find, in a short time,
their theories exploded by the criticism of facts.
But the third class included a number of practical theocrats who
blended republican ideas with their theological speculations, and
who were quite disposed, whenever the opportunity offered, to open
by force a path by which the saints might advance to the
government of the world. Venner before and after the Restoration
went even this length; and Harrison, although a different man from
Venner, was also a military theologian, ready with carnal weapons to
cast down strongholds. Certain Anabaptists also belonged to the
same division.
The fifth monarchy fever, in its fanatical
symptoms, threatened much mischief to English Fifth Monarchy
Men.
society. As in the tenth century the notion then so
common throughout Europe—that the end of the world had come,
that the whole social fabric had crumbled away, and was to give
place to an order of things just on the point of descending from
heaven—proceeded from the existing confusion of the age; so the
prevalence of fifth monarchy dreams, in the earlier part of the
Commonwealth, was aided, no doubt, by the convulsions which had
so rudely shaken the whole of the British Empire. The actual
overthrow of the hierarchy, the peerage, and the throne, with the
desolations of civil war, and the disturbance of the whole country,
could not but throw the minds of many into a maddening whirl, and
dash them completely off their balance. Fifth monarchism arose, in a
great measure, out of the ruins of English monarchism. The
enthusiastic visions which absorbed certain minds were, to a large
extent, the effect of prevalent disorder. Men's brains were crazed by
what they had recently witnessed, and their insanity created omens
of other things, brighter or more terrible, which were yet to come.
During the sittings of the Little Parliament, the
fifth monarchy delusion reached its height; and to 1653.
that delusion, though not to that alone, may be
attributed, as to their cause, certain incidents which, although
exaggerated by both Clarendon and Baxter, were not by any means
imaginary. These writers speak of this period, as if the very
existence of the Christian ministry, of rational religion, of the two
Universities, and of Christian learning in general, were then on the
point of complete extinction.[57] Things certainly were not in that
condition. What was actually done in reference to the support of the
clergy has been already described, and it can by no means be made
to sustain the sweeping assertions of those authors; yet,
notwithstanding, the interests of the great seats of education were
placed in fearful jeopardy, as will appear when we have to notice the
history of the University of Oxford at that time; nor can it be denied
that the demolition of some noble ecclesiastical edifices had been
contemplated even before the breaking up of the Long Parliament.
The Norwich Corporation, so early as 1650, debated what should be
done with the cathedral of that city; and in the same year, Yarmouth
was seeking to share in the spoil.[58] Further still, on the 9th of July,
1652, it was referred to a committee by the House of Commons, to
consider this question: "What cathedrals are fit to stand, and what
to be pulled down." Such intended destruction betokened other
spoliations, and the Long Parliament having set such an example,
the Little Parliament had encouragement to proceed in a similar
path.
The proceedings of Harrison and his party could not fail to alarm
the Presbyterians; even the Independents, with Cromwell at their
head, in spite of their broad views of ecclesiastical questions, were
also convinced that nothing but confusion could result from the wild
schemes of republican visionaries and Fifth Monarchists. The country
could not feel confidence in those who formed the slight majority of
the House, and sober-minded men apprehended nothing but ruin
from the continuance of their power. The Parliament itself, with such
an even balance of parties, and amidst so much distraction, had a
consciousness of its own incapacity, which led the members speedily
to resign their powers into the Lord General's hands.[59]
This resignation, Cromwell's acceptance of it,
and his consequent assumption of supreme Feake at
Blackfriars.
authority, drove the millenarian democrats into a
still more violent expression of extravagant views, and into still more
decidedly energetic opposition to the Lord General. Preachers of that
day, in close alliance with Harrison, advocated in the pulpit the cause
which he and his party upheld in the senate. A House at Blackfriars
is repeatedly mentioned in contemporary letters as the head
quarters of this menacing agitation. Feake—a well-known Anabaptist
and Fifth Monarchy man—there held forth in a strain of rude
eloquence, and greatly distinguished himself as leader of a large
band of sympathizing disciples. To letters written by contemporaries
we are chiefly indebted for what we know of the proceedings of
these enthusiasts.
"I know not," says the writer of an intercepted
epistle, "whether you have formerly heard of the 1653.
Monday's lecture at Blackfriars, where three or four
of the Anabaptistical ministers preach constantly, with very great
bitterness, against the present Government, but especially against
his Excellency, calling him 'the man of sin,' 'the old dragon,' and
many other scripture ill names; the chief of them is one Feake, a
bold and crafty orator, and of high reputation amongst them. It has
been wondered the General has so patiently permitted them; but
yesterday I heard the true reason of it, which is, that he cannot help
it, for they preach by an Act of the late Parliament, which the council
of state cannot over-rule, and this Parliament will not abolish it; but
on Tuesday last, as I take it, they were called before a private
committee, where your General was present, who told them that the
ill odour they had cast upon the Government has given confidence to
our enemies abroad and at home, (meaning the Scots,) and would
bring the Parliament into contempt; and that whatsoever ill effect
followed, they must be accountable for it. Feake replied that he
desired that what the General said and what he answered might be
recorded in heaven; and that it was his tampering with the king, and
his assuming an exorbitant power, which made these disorders; and
so held forth the Fifth Monarchy. The General answered, that when
he heard him begin with a record in heaven, he did not expect that
he would have told such a lie upon earth; but assured him that
whensoever they should be harder pressed by the enemy than they
yet had been, it would be necessary to begin first with them; and so
dismissed them. I forgot to tell you that the General had brought
Sterry,[60] and two or three more of his ministers, to oppose spirit to
spirit, and to advise Feake and the rest to obedience, as the most
necessary way to bring in the kingdom of Christ. But it is believed
we shall have very much trouble from the Anabaptists, yet it is
thought their power is nothing so great in the army as in the House;
they have none above a captain of their party besides Harrison, who,
it is thought, will betray all the rest: but whether the General will
ease himself of those in the House by the old way of purging, or the
new one of dissolving, rests in his own and his officers' breasts."[61]
The district of Blackfriars claimed to be
independent of the municipal authorities of 1653, December.
London. The inhabitants asserted an inheritance of
the privileges of sanctuary, formerly pertaining to that famous
monastery which had given its name to the neighbourhood. Hence,
to find shelter and protection within the precincts of the ancient
foundation, players, who had been driven out of the city, here
erected a theatre; and Papists, who were proscribed by law, here
assembled for worship. And it is not a little curious that Puritans also
were somewhat numerous in the same locality; a fact which is
indicated by their presenting what seems to have been an influential
petition to the Lords of the Privy Council against the continuance of
stage-plays by their dramatic neighbours.[62] Blackfriars, as we have
seen, is also mentioned among the places in which certain
Nonconformists were wont to meet in the first quarter of the
seventeenth century; and in this same place we now meet with an
Anabaptist assembly listening to the popular preachers of
millenarianism.
A letter from an eye-witness communicates
additional information respecting these meetings. Feake and Powell.
The writer states that he had been to one of them,
and had heard Feake preach upon the subject of the little horn
described in the book of Daniel; and he states that in the course of
the sermon the preacher exclaimed, "I know some would have the
late King Charles to be meant by this little horn; but as I said at first,
I'll name nobody. God will make it clear shortly to His people who is
meant here." When Feake had concluded his portion of the service,
Vavasour Powell continued to discourse on the same subject, in a
similar strain of interpretation—still more explicitly reflecting on
public men and measures than his predecessor had done—
interpreting the king of the north to signify the late monarch, and
inveighing bitterly against the military commanders of the day, as
the sole cause of the pressure of taxation. The leading points of the
sermon were, that Christ was setting up a fifth monarchy in the
world; that a spirit of prophecy had been communicated to the
saints, whereby they were enabled to describe future events; and
that the design of Christ was to destroy all antichristian forms,
including established churches together with their clergy. Upon this
third particular, the reporter states that Powell was somewhat
copious, and said "they must down, though they were never so
strongly protected, for Christ is none of their Lord Protectors, though
the army-men protect them." "Yes," said he, "and rather than those
shall down, they will pull Parliaments in pieces, and this made them
break the last Parliament; for on Saturday, the 10th of December,
the House refused to settle a commission of ministers to ride in
circuits, as the judges did, and judge who were fit to be continued
or put out of their livings, and so to maintain them upon the old
corrupt foundation still. And when the House would not yield that
these antichristian clergymen and tithes should be upheld, then, on
Monday following, in the morning, they were thrust out (I mean the
few honest men of them that were present) by violence; and the
rest (as they had agreed beforehand) went and subscribed their
names to a paper giving up their authority in the name of the whole;
whereas none of the honest men would subscribe or surrender, save
only some three or four, who have since professed their hearty
sorrow to me for it. This is true, and we must speak it out, for our
mouths shall not be stopped with paper-proclamations." ... Further,
in relation to the Parliament, he remarked, "they were broken by
force, and it was a business plotted by the great army-men,
clergymen, and their party together." ... Powell afterwards "flew into
many strange ejaculations, 'Lord! what have our army-men all
apostatized from their principles? What is become of all their
declarations, protestations, and professions? Are they choked with
lands, and parks, and manors? Let us go home, and pray, and say,
'Lord, wilt Thou have Oliver Cromwell to reign over us, or Jesus
Christ to reign over us.'" "I know," he proceeded, "there are many
gracious souls in the army, and of good principles, but the greater
they grow, the more they are corrupted with lands and honours. I'll
tell you, it was a common proverb that we had among us of the
General, that in the field he was the graciousest and most gallant
man in the world; but out of the field, and when he came home to
government, the worst." This strange preacher told his congregation
that "snares were laid for them, and spies set over them, and that
they might be deprived of the benefit of meeting in that place. But
then (said he) we will meet at another, and if we be driven thence,
we will meet at private houses, and if we cannot have liberty there,
we will into the fields, and if we be driven thence, we will into
corners, for we will never give over, and God will not permit this
spirit to go down. He will be the support of the spirits of His people.
He complained also of the faltering of divers who had formerly been
very forward at this meeting, but now drew back, and therefore he
prayed that the Lord would hold up the meeting."[63]
Powell having concluded, somebody seated in
one corner of the gallery began to speak, and 1653, December.
would have replied to the preacher; but, though he
strained his voice with the utmost violence to overcome the outcries
of the congregation, he was compelled, after half-an-hour's tumult,
to hold his peace. A Mr. Colaine, amidst the confusion, ascended the
pulpit, and afterwards expounded the fifth chapter of Hosea,
representing the state of things in England as parallel to that which
the prophet portrayed, and inveighing strongly against the national
clergy of Antichrist, and the parochial priests of Baal.
According to another letter, personal allusions to Cromwell even
yet more violent occurred in the discourses of these misguided men.
Powell and Feake called him "the dissemblingist perjured villain in
the world," and desired any friends of his, who might be present, to
go and report this to him, adding, that the Protector's reign would
be short, and "that he should be served worse than that great tyrant
the last Lord Protector was, he being altogether as bad, if not worse
than he."[64]
These fanatics threw themselves with
earnestness into the Dutch war. That conflict, Feake and Powell.
looking at the political and religious character of
the combatants, strikes us as very strange, both parties being
republicans, and both being defenders of religious liberty: but it had
arisen from commercial and maritime rivalries, into which additional
bitterness had been shed by the natural sympathies of the Prince of
Orange with the Stuart family. A confederation of the two
commonwealths, for the promotion of civil freedom and the interests
of Protestantism throughout Europe, formed an English dream at the
end of the civil wars; and what had at first been contemplated as a
subject for peaceful negotiation was afterwards absurdly sought to
be accomplished by naval battles. The republican zeal and Protestant
fervour of Feake and his friends enlisted them on the side of a
thorough union between the two states, and they stipulated for it as
an indispensable condition of peace. That England should persevere
till Holland could be yoked to her in humble submission for the
attainment of these civil and religious ends, constituted a staple
theme in the harangues at Blackfriars. Conciliation and compromise
were condemned. The preachers denounced in the wildest way the
statesman-like views of Cromwell, who felt anxious to put an end to
the deadly struggle of two countries, between which policy as well
as justice dictated alliance with mutual independence. His opponents
did all they could to stir up the people of England against the
Netherlanders, and one of the Dutch deputies, who went to hear
them, wrote home, declaring that their sermons were "most horrid
trumpets of fire, murder, and flame."[65] Millenarianism thus became
mixed up with political schemes; and these Commonwealth
visionaries believed that God had given Holland to the English as a
"landing place of the saints, whence they should proceed to pluck
the whore of Babylon from her chair, and to establish the kingdom of
Christ on the Continent."[66]
Between the resignation of the Little Parliament on the 12th of
October, and the date of the last of these letters, a great change had
come over the government of England. Cromwell and his council of
officers, "after several days seeking of God," had determined
formally to avow the perpetuation of what was already a fact—that
supreme authority should rest in a single person, even in Oliver
himself. His title was to be "Lord Protector," and with him was to be
associated "a council of godly, able, and discreet persons," consisting
of not more than twenty-one.
On Friday, the 16th of December, about three
o'clock in the afternoon, his Highness went in Cromwell made
Protector.
procession from Whitehall to the Court of Chancery
in Westminster. Commissioners of the Great Seal, scarlet-robed
Judges and Barons, and the Lord Mayor and Corporation of London,
with the usual accompaniments of civic splendour, were in
attendance on the occasion. There on a dais stood a chair of state,
near to a table on which lay a roll of parchment, containing, in a
summary of forty-two articles, the fundamental principles of the
Protectorate government.[67] His Highness having subscribed the
document, and having sworn to maintain the constitution which it
prescribed, sat down on his throne, and then received into his hands
the great seal of the realm, the Lord Mayor's sword, and the cap of
maintenance. His portrait at that moment has been sketched in the
following graphic words: "Fifty-four years old gone April last; brown
hair and moustache are getting gray;" "massive stature; big massive
head, of somewhat leonine aspect—wart above the right eye-brow;
nose of considerable blunt aquiline proportions; strict yet copious
lips, full of all tremulous sensibilities, and also, if need were, of all
fierceness and rigours; deep, loving eyes, call them grave, call them
stern, looking from under those craggy brows as if in life-long
sorrow."[68]
 As the Protector returned to Whitehall, the Lord Mayor,
uncovered, carried the sword before him; and in the banqueting
house, Mr. Lockier, the Protector's chaplain, delivered an exhortation;
after which the Lord Mayor, the Aldermen, and the Judges all
departed home.[69]
A very wonderful day's work was that of the
State of Affairs at
16th of December.[70] It was an act of usurpation the Time.
beyond all doubt, yet one which it would be
pedantic to criticise according to constitutional rules. For when a
house is catching fire, or a vessel is on the edge of a rock, people in
their senses do not expect that a strong man, who can extinguish
the flame, or steer the ship into open waters, should stand on
ceremony and wait for red-tape formalities. A crisis had come, such
as must come in the wake of great revolutions, when, however
firmly we may maintain Locke's principles of the origin of
government, we find people in such a state of perfect helplessness,
and former rulers so utterly destitute of power to rule, so incapable,
so inevitably rushing to destruction, dragging the whole country
along with them—that in mercy to mankind we feel all theories
woven of the wisest webs must be laid aside, and a competently
skilful hand be allowed to gather up the scattered threads in such
fashion as it may be able. Public affairs had now reached such a
crisis that the alternative was a Protector or destruction, Cromwell or
chaos. The fire long kindling now at length burst into a blaze, and
there was only one man who could put an end to the conflagration.
The ship was within an inch of foundering, and there remained but
one pilot who had the power of steering her off the rocks.
The Church, it should be recollected, had not
ceased to be a State Church. The constant 1653.
discussion of its affairs by Parliament and the
Council of State implied its continued subjection to secular control.
No scheme of severance had been propounded, though certain
proposals had been made which might seem to involve such a result.
However desirable it may appear to some of our readers that the
civil and ecclesiastical powers in a country should stand each on its
own basis, and not interfere with one another's action; whatever
anticipations may be formed of a new and better era of Christian
civilization, to be inaugurated when a separation of the Church from
the State shall take place, subject to the authority of New Testament
teaching, aided by the lights of experience through the exercise of
political sagacity, and under the inspiration of Christian
disinterestedness; yet every one must see that the time had not
then come for such a revolution. Such a revolution involved not only
the settlement of questions touching ecclesiastical property and
revenue, but also the determination of two other points, namely,
that religion should be left to its own unfettered exercise, and that
no man should be disqualified by his theological opinions for the
discharge of the offices and the enjoyment of the honours of the
State. Now, without doing injustice to the character of the Little
Parliament, we certainly may go so far as to say that it never
indicated the possession of that clearsightedness, that deep wisdom,
and that broad sympathy which are essential to the satisfactory
solution of the practical problems included in the change just
indicated. The members had neither the intellectual nor the moral
qualifications requisite for the task. It was in those days more
difficult than it is at present to draw the line where free religious
action comes to an end, and something else quite different from it
begins; for millenarian opinions ran over into Fifth Monarchy
schemes, and the Dutch wars had become mixed up in men's brains
with the dominion of the saints. The Little Parliament lacked the
mental power necessary in those days for carrying out the doctrines
of voluntaryism, even had they understood them. But they did not
understand them. Had they done so, they would not have clung to
the idea in any form, of the State supporting the Christian ministry,
nor would they have cherished the conviction that certain theological
qualifications are indispensable for the discharge of political trusts.
And further, if the Little Parliament had been
composed of the wisest of mortals, and had plainly State of Affairs at
the Time.
and skilfully propounded a system of pure
voluntaryism, such as is ably and successfully advocated in our own
time, still, with the Presbyterians all against them; with many of the
Independents against them; and with the Episcopalians also against
them; in short, with the bulk of the wealth, of the intelligence, and
of the power of the country against them—how useless would have
been their attempts to work out the measure. Common sense
teaches, and voluntaryism in its very nature implies, that before it
can be established as the exclusive method of dealing with spiritual
interests, a very large number of those who have to adopt it must be
convinced of its wisdom. And as to the alternative of revising the
Establishment, and placing it upon grounds adapted to the needs of
existing society, that also was an undertaking which, it is needless to
repeat, the Little Parliament did not accomplish, and one, too, with
which the whole history of that Assembly proved that it was utterly
incompetent to deal. The whole web of ecclesiastical affairs had
raveled out, and it devolved on a more than ordinarily skilful hand to
gather up the threads and arrange them in some sort of order.
Cromwell ever shewed himself to be a practical
man, by no means wedded to any fine-spun theory. 1653.
No ideal republic, such as was conceived by Plato
or by Harrington, floated before his imagination. In this respect a
marked distinction existed between him and his contemporaries of
the philosophical schools which were led by Sir Harry Vane and
Algernon Sidney; and, as in pure politics, so in ecclesiastical politics,
he aimed simply at accomplishing what he saw to be practicable. His
strong religious feelings, the mystic cast of his piety, his enthusiastic
faith in prayer and providence, never turned him aside from plain
paths of human action, where he could get common people to walk
and work beside him. Whatever idea he might have had as to what
was best in itself, and under other circumstances than those of
England in his own day, then rocking with the throes of revolution,
certainly the plan which he adopted was not that of attempting the
exclusive establishment of a voluntary system of supporting religion.
He saw that to alienate church property from sacred uses—had he
wished to do so—would arouse against him at once all the
Presbyterians of the country, and would give them a rallying point
and a battle cry quite sufficient to render them irresistible. He knew,
that supported in this respect by Episcopalians, and not without
sympathy amongst Independents, the Presbyterians would have
protested against spoliation, and would have contended for the
inviolateness of tithe property with a temper too fierce and with an
amount of influence too strong for any government to resist with
success. He perceived the wisdom of conciliating the Presbyterian
party, and even on that ground he would shrink from provoking
them by the confiscation of all church revenues. His keen eye also
discerned such a spirit in some of the sects, such violent anti-social
principles abroad, such elements seething in the cauldron of
religious excitement, that he felt it would not be safe to leave all
theological teachers at that time to do and say just what they liked
without any sort of legal restraint. The liberty which he believed it
just and right to concede in reference to the discussion of simple
questions of divinity, he did not consider it just and right to afford to
all sorts of semi-political agitations; which, under the cover of
prophetical study and of transcendental schemes of society, directly
tended to overthrow all law and order, and with law and order, the
very liberty which such enthusiasts themselves really desired to
enthrone.
What, then, was the kind of National Church
which Cromwell's practical sagacity led him to Cromwell's Policy.
establish? Though he might not work according to
any definite theory, and was mainly prompted by the quick insight of
his own genius, yet there could not but be some principles lying at
the basis of his operations. Three politico-ecclesiastical theories of
union may be entertained: that of the Church's mistress-ship over
the State, that of the Church's servitude to the State, and that of the
Church's marriage with the State. What the Lord Protector aimed at
accomplishing appears far removed from the first of these. He would
not allow Presbyters, or Pastors, or Preachers of any kind, any more
than Anglican Priests, to lord it over the people. He would carry the
staff in his own hands. At the same time, he did not put the Church
in perfect servitude. Though Erastian in one way, his method of
ecclesiastical government does not appear to have been so in
another. Whilst the appointment and recognition of ministers
receiving State pay were placed under the authority of persons who
owed their official position to State appointment, yet the inner
working, the worship, and the discipline of Churches continued to be
left free to a very large extent. Perhaps, on the whole, Cromwell's
Broad Church embodied more of the idea of the marriage of the
Church with the State than any other Establishment which ever
existed.[71]
His ecclesiastical policy rested on five principles:
—State recognition, State control, State support, 1653.
State protection, and State penalties. How those
principles were developed in Cromwell's administration will be seen
in the next chapter.
 CHAPTER IV.
To prevent confusion, let it be distinctly stated at once, that in
tracing the form which the new ecclesiastical establishment assumed
under the impress of Cromwell's genius, we confine ourselves in this
chapter to the legislation of nine months; consisting of those
ordinances which were issued between the end of the Little
Parliament, in December, 1653, and the opening of the first
Protectorate Parliament, in September, 1654. During this period, the
foundations of the Protector's ecclesiastical policy were laid.
I. State Recognition.—The articles of government—the conception
and inspiration of which must be regarded as proceeding from
Cromwell—distinctly declared "that the Christian religion, as
contained in the Scriptures, be held forth and recommended as the
public profession of these nations."[72] Christianity being thus
recognized as part and parcel of the law of the land, the sanctions of
religion were introduced at the inauguration of the Protector; the
solemnities of worship and of preaching were connected with all
special public acts; and the exercises of devotion constantly
accompanied the ordinary business of Parliament. The State
continued to recognize religion by the appointment of fast days,
which were of frequent occurrence; whilst the Scotch brethren
objected to this exercise of civil authority as an Erastian intrusion
into the spiritual realm.[73] Preachers, both Presbyterian and
Independent, were appointed on these occasions; and a fast day
sometimes was solemnized by a service which lasted from nine
o'clock in the morning until four in the afternoon.
By an express article, all who professed the
Roman Catholic religion were disabled from voting, 1654.
as well as from being elected; and as the Act which
had been passed against execrable opinions, treated as culprits and
subjected to penalties those who opposed Christianity, it virtually
deprived all such persons of the electoral franchise. Infidels and
heretics, also, who attacked or undermined the foundations of the
Christian faith, forfeited the rights of denizenship. But these laws did
not affect the social position of any individuals who professed
Protestantism in any of its usually-recognized forms of orthodoxy. All
the "sects" were accepted as citizens. So were the Presbyterians.
And, although Prelacy was forbidden, there was nothing which could
legally prevent an Episcopalian from going to the poll to give or
receive the vote of a freeman. Still, we must not forget that, since
the Common Prayer Book had been prohibited, any one who
persisted in using its formularies might have both his franchise and
his freedom brought into peril.[74] From these facts, it is evident that
England under the Protectorate was, in theory, a religious
Commonwealth; that the State possessed a spiritual as well as a
secular character; that Christianity was considered essential to the
welfare of society; and that an irreligious man was not regarded as a
faithful subject. But this theory of the Commonwealth as a Christian
State must not be confounded with the theory of the National
Church as connected with the Commonwealth. The lines of limitation
in the two cases were not the same. Considerable differences
existed between the Christianity which entitled all its disciples to the
franchise of the citizen and the Christianity which entitled its
ministerial advocates to the support of the State. What those
differences were will be indicated as we proceed.
II. State Control.—The laws made certain
distinctions between what was civil and what was State Control.
sacred. They followed the early legislation of the
Long Parliament by withdrawing all secular matters from
ecclesiastical authority. Wills received careful attention from the Little
Parliament in 1653, when commissioners were appointed to
superintend that business, and to grant administrations "in the late
provinces of Canterbury and York." Their powers were defined, and
the probate fees to be taken by registrars were, after the payment
of expenses, to be appropriated to the support of the navy. The Act
of 1653 was revived in 1654, and more commissioners were added
to the existing number.[75]
The main control over the Church consisted, not in any Act of
Uniformity—nor in the establishment of a particular creed—nor in
the maintenance of a simple mode of worship, but in the
appointment of a spiritual tribunal, invested with the power of
determining who were fitting persons to fulfil the Christian ministry.
In the month of March, 1653-4, an ordinance appeared,[76] reciting
that there had been no certain method adopted for supplying
vacancies with able ministers, in consequence of which the rights of
patrons had been prejudiced, and "weak, scandalous, popish, and ill-
affected persons had intruded themselves, or been brought in, to
the great grief and trouble of the good people of this nation." As a
remedy, it was ordained that every person presented to a benefice,
or appointed to a lecture, should be approved by certain
Commissioners who were named for that purpose. No mention is
made of any standard of faith, of any mode of worship, or of any
scheme of polity. Episcopacy, Presbyterianism, Independency, anti-
Pædobaptism—in short, particular forms of Christianity are entirely
unnamed and unnoticed. In general terms, power was vested in the
Commissioners:—they were to grant admission to the ministry; their
certificate being a sufficient induction; but a vote of exclusion did
not acquire validity unless nine members were present at the time
when the vote was passed. Appointments made by these
Commissioners did not interfere with the rights of patronage. They
had no authority to dispose of Church benefices, or to elect
lecturers; but only to determine upon the qualifications of those
whom the patrons presented or the people chose. Nor did the law
construe the decision of these judges "to be any solemn or sacred
setting apart of a person to any particular office in the ministry." In
short, the Commissioners formed a board, and nothing more, for the
examination of persons who presented themselves for the ministerial
office. So far, it bore a likeness to the Assembly of Divines, for they
had exercised similar functions in the examination of clergymen; but
then they had been more numerous, and had been wont to consult
Church standards and formularies for the guidance of their
judgment. Nothing of the sort limited the power of the new
Commissioners, and, moreover, their unfettered power was lodged in
comparatively few hands. Some creed, statute, canon, or established
usage, had in all similar cases been recognized as a rule of action;
but in this instance everything was left for determination by the
wisdom or the will of irresponsible functionaries.[77] No distinct
articles of faith were prescribed. No subscription whatever was
enforced.[78] The only way to form an idea of the character of a
Church so circumstanced is to infer what it must have been from the
known opinions and characters of such powerful officers. The
Commission was composed of men of very different characters.
Some had much prejudice and party spirit, with little judgment, and
less charity. No confidence could be placed in Hugh Peters, and in
others of a similar stamp; but there were amongst the members
individuals of great wisdom and large benevolence—such as Manton,
Goodwin, Owen, and many more. Presbyterians, Independents, and
Baptists were of the number; and, so far, the constitution of the
tribunal permitted access to benefices by ministers belonging to all
three denominations. The proportion of different religious parties in
the commission suggests what was likely to be the proportion of
those admitted to preferment. Nothing hindered the admission of
the members of any sect whatever, not even Episcopalians, provided
they did not use the Book of Common Prayer; and such persons
actually were admitted; but it is not probable that many would be
included in the establishment who occupied a position far beyond
the circle of the Commissioners' opinions.
Another tribunal appeared in August, 1654, for
ejecting "scandalous, ignorant, and insufficient State Control.
ministers and schoolmasters."[79] Unlike the
former, this Commission branched out into manifold divisions, which,
in fact, formed ecclesiastical courts of assize, spreading over the
whole country. Long lists of distinguished laymen are contained in
the ordinance—including the names of the Lords Wharton, Fairfax,
Lisle, Say and Sele, Sir A. Haselrig, Sir Anthony A. Cooper, and
Nathaniel Fiennes. They were to bring before them all clergymen
and teachers who were punishable by the Act against blasphemous
opinions, or who were guilty of profanity, perjury, popish opinions,
adultery, fornication, drunkenness, haunting taverns, quarrelling,
fighting, playing at cards or dice, or profaning the Sabbath day. So
ran the enumeration of the first class of scandals, cognizable by
these judges. Then came words pointing to such as had publicly and
frequently used the Common Prayer Book since the first of January
last, or should at any time afterwards do so; such as publicly and
profanely scoffed at or reviled the strict profession or professors of
religion or godliness, or encouraged, or countenanced, "by word or
practice, any Whitsun ales, wakes, Morris-dances, May-poles, stage-
plays," or similar licentious practices; and such as declared their
disaffection to the present government. As to "negligent" ministers,
they are defined to be "non-residents," and such as omitted the
public exercise of preaching and praying on the Lord's Day: but "the
ignorant and insufficient" are not defined at all, they are left to be
declared and judged by the Commissioners in every county, or by
any five of them, together with five of the ministers mentioned in
the ordinance. To them, therefore, in this respect, remained a wide
margin of discretion, and individuals guiltless of the scandals and
offences before enumerated—yet being charged in general terms by
their parishioners with ministerial incompetency—were left to the
mercy and the conscience of these lay and clerical assessors. Their
character was the only guarantee that justice would be
administered; and sometimes proofs appeared shewing how perilous
a thing it was to the interests of the parties arraigned that even to
men of established integrity there should be entrusted such large
powers, especially at a time when party spirit on all sides ran so
high.
III. State Support.—The articles of government
declared, that as soon as might be, a provision less 1654.
subject to contention and more certain than the
one existing should be made for the maintenance of "able and
painful" teachers, for instructing the people, and for the discovery
and confutation of error, heresy, and whatever is contrary to sound
doctrine. But until such provision could be devised, the existing
maintenance was not to be taken away or impeached. Also the
ordinances of 1647, as to tithes, were in 1654 declared to remain in
full force; and further still, for the more efficient support of the
ministry, an ordinance of the 2nd of September, 1654, directed that
there should be a union of small parishes and a division of large
ones—authority for that purpose being vested in a Commission,
according to a common plan then adopted in all business of that
description.[80]
The Long Parliament, in the month of February,
1648, had commanded churchwardens and State Protection.
overseers of the poor to assess every inhabitant of
a parish, in such sums as those officers should think proper; no
mention being made of holding any vestry meetings whatever for
that purpose. The law declared that such rates should be
appropriated for repairing the fabric of the church, and keeping in
order the churchyard and walls; for providing books to be used in
Divine worship; and for the bread and wine required in the
administration of the Lord's supper. When the rate had been
confirmed by two justices of the peace, the churchwardens were
authorized and required to levy payment and to recover by "distress"
where payment was refused. The justices, "in default of such
distress," might commit the defaulter to the common gaol.[81] This
church-rate law remained unrepealed, and therefore was available
for the support of worship by all those who were now incorporated
in the Establishment. In the ordinance of 1654, for uniting and for
severing parishes, reference is made to rates, taxes, parochial rights,
charges and duties, as acknowledged sources of revenue.[82]
IV. State Protection.—The Articles of
Government extended protection, within certain 1654.
limits, to professing Christians who did not share in
the resources and immunities of the State Church. Religious
compulsion was forbidden, religious persuasion was recommended;
and it was expressly declared, "That such as profess faith in God by
Jesus Christ (though differing in judgment from the doctrine,
worship, or discipline publicly held forth), shall not be restrained
from, but shall be protected in, the profession of the faith and
exercise of their religion, so as they abuse not this liberty to the civil
injury of others, and to the actual disturbance of the public peace on
their parts, provided this liberty be not extended to Popery nor
Prelacy, nor to such as under the profession of Christ hold forth and
practise licentiousness."[83] The shield of the law was thus placed
over all Protestant sects whose liberty in no way threatened the
security of the Government.
V. State Penalties.—First, the Papists were
deprived of all religious freedom and of all political State Penalties.
rights, and this act of injustice was perpetrated as
a retaliation which their own habitual intolerance had provoked; and
as a precaution which the tendency of their system and their
Jesuitical and treasonable practices had rendered expedient. The
circumstances in which Prelatists were placed by the legislation of
the Long Parliament have been explained. These circumstances
remained unaltered; and Prelacy was now conjoined with Popery in
the prohibition expressed by the articles. The supporters of Prelacy
were known to be disaffected to the Government, and whenever
that disaffection manifested itself in overt acts, the magistrates were
justified in punishing the offenders; but to inflict penalties for using
the Prayer Book was an unrighteous proceeding, no more to be
excused than was the persecution of Nonconformists for their
worship, after the Restoration. Sometimes that persecution has been
defended or its guilt has been extenuated on the ground that the
very religion of the Separatists made them disloyal; persons who
condemn that plea as being insult added to injury must not set up a
similar one on behalf of the rulers of the Commonwealth. After the
mention of Prelacy in the articles comes a denial of freedom to such
as maintained tenets inimical to the principles of public morality and
order;[84] Fifth Monarchists, therefore, preaching after the fashion of
Welcome to Our Bookstore - The Ultimate Destination for Book Lovers
Are you passionate about books and eager to explore new worlds of
knowledge? At our website, we offer a vast collection of books that
cater to every interest and age group. From classic literature to
specialized publications, self-help books, and children’s stories, we
have it all! Each book is a gateway to new adventures, helping you
expand your knowledge and nourish your soul
Experience Convenient and Enjoyable Book Shopping Our website is more
than just an online bookstore—it’s a bridge connecting readers to the
timeless values of culture and wisdom. With a sleek and user-friendly
interface and a smart search system, you can find your favorite books
quickly and easily. Enjoy special promotions, fast home delivery, and
a seamless shopping experience that saves you time and enhances your
love for reading.
Let us accompany you on the journey of exploring knowledge and
personal growth!

ebookgate.com