Discrete Calculus - Applied Analysis on Graphs for Computational Science

von: Leo J. Grady, Jonathan Polimeni

Springer-Verlag, 2010

ISBN: 9781849962902 , 366 Seiten

Format: PDF, OL

Kopierschutz: Wasserzeichen

Windows PC,Mac OSX geeignet für alle DRM-fähigen eReader Apple iPad, Android Tablet PC's Online-Lesen für: Windows PC,Mac OSX,Linux

Preis: 181,89 EUR

Mehr zum Inhalt

Discrete Calculus - Applied Analysis on Graphs for Computational Science


 

Preface

6

Contents

7

Acronyms

12

Discrete Calculus: History and Future

14

Discrete Calculus

14

Origins of Vector Calculus

15

Origins of Discrete Calculus

16

Discrete vs. Discretized

17

Complex Networks

19

Content Extraction

20

Organization of the Book

21

Intended Audience

21

A Brief Review of Discrete Calculus

23

Introduction to Discrete Calculus

24

Topology and the Fundamental Theorem of Calculus

25

Differential Forms

27

Exterior Algebra and Antisymmetric Tensors

28

The Vector Spaces of p-Vectors and p-Forms

28

Manifolds, Tangent Spaces, and Cotangent Spaces

31

The Metric Tensor: Mapping p-Forms to p-Vectors

34

Differentiation and Integration of Forms

37

The Exterior Derivative

37

The Poincaré Lemma

41

The Hodge Star Operator

43

The Codifferential Operator and the Laplace-de Rham Operator

47

Differential Forms and Linear Pairings

48

Discrete Calculus

49

Discrete Domains

50

Orientation

53

The Incidence Matrix

55

Chains

56

The Discrete Boundary Operator

57

Discrete Forms and the Coboundary Operator

59

Primal and Dual Complexes

61

The Role of a Metric: the Metric Tensor, the Discrete Hodge Star Operator, and Weighted Complexes

65

The Metric Tensor and the Associated Inner Product

65

The Discrete Hodge Star Operator

66

Weights

68

The Volume Cochain

68

The Dual Coboundary Operator

70

The Discrete Laplace-de Rham Operator

71

Structure of Discrete Physical Laws

74

Examples of Discrete Calculus

74

Fundamental Theorem of Calculus and the Generalized Stokes' Theorem

76

Generalized Stokes' Theorem on a 1-Complex

76

Comparison with Finite Differences Operators

80

Generalized Stokes' Theorem on a 2-Complex

82

Generalized Stokes' Theorem on a 3-Complex

83

The Helmholtz Decomposition

84

Algorithm for Computing a Helmholtz Decomposition of a Flow Field

87

Matrix Representation of Discrete Calculus Identities

88

Integration by Parts

88

Other Identities

89

Elliptic Equations

90

Variational Principles

93

Diffusion

93

Advection

97

Concluding Remarks

99

Circuit Theory and Other Discrete Physical Models

101

Circuit Laws

103

Steady-State Solutions

104

Dependent Sources

107

Energy Minimization

109

Power Minimization with Nonlinear Resistors

111

AC Circuits

111

Connections Between Circuit Theory and Other Discrete Domains

114

Spring Networks

114

Random Walks

116

Gaussian Markov Random Fields

120

Tree Counting

122

Linear Algebra Applied to Circuit Analysis

127

The Delta-Wye and Star-Mesh Transforms

127

Minimum-Degree Orderings

129

Conclusion

131

Applications of Discrete Calculus

133

Building a Weighted Complex from Data

134

Determining Edges and Cycles

135

Defining an Edge Set

135

Edges from an Ambient Metric

136

Edges by k-Nearest Neighbors

137

Edges from a Delaunay Triangulation

137

Adding Edges via the Watts-Strogatz Model

137

Defining a Cycle Set

138

Defining Cycles Geometrically: Cycles from an Embedding

139

Defining Cycles Algebraically

140

Cycle Sets and Duality

143

Deriving Edge Weights

144

Edge Weights to Reflect Geometry

144

Edge Weights to Penalize Data Outliers

145

Univariate Data

145

Computing Weights from Multivariate Data

150

Edge Weights to Cause Repulsion

152

Edge Weights to Represent Joint Statistics

153

Deducing Edge Weights from Observations

153

The Underdetermined Case

154

The Overdetermined/Inconsistent Case

155

Obtaining Higher-Order Weights to Penalize Outliers

156

Weights Beyond Flows

158

Metrics Defined on a Complex

159

Conclusion

162

Filtering on Graphs

164

Fourier and Spectral Filtering on a Graph

165

Graphs that Are Not Shift-Invariant

168

The Origins of High Frequency Noise

171

Energy Minimization Methods for Filtering

172

The Basic Energy Minimization Model

172

Iterative Minimization

173

Extended Basic Energy Model

176

The Total Variation Model

177

Filtering with Implicit Discontinuities

179

Filtering with Explicit, but Unknown, Discontinuities

181

Filtering by Gradient Manipulation

183

Nonlocal Filtering

183

Filtering Vectors and Flows

184

Translating Scalar Filtering to Flow Filtering

185

Filtering Higher-Order Cochains

188

Applications

189

Image Processing

189

Regular Graphs and Space-Invariant Processing

189

Space-Variant Imaging

191

Three-Dimensional Mesh Filtering

194

Mesh Fairing

194

Filtering Data on a Surface

196

Geospatial Data

201

Filtering Flow Data-Brain Connectivity

202

Conclusion

206

Clustering and Segmentation

207

Targeted Clustering

208

Primal Targeted Clustering

209

Probabilities Assigned to a Subset

210

Known Labels for a Subset of Nodes

213

Negative Weights

217

Dual Targeted Clustering

218

Dual Algorithms in Three Dimensions

221

Untargeted Clustering

223

Primal Untargeted Clustering

224

Dual Untargeted Clustering

228

Semi-targeted Clustering

229

The k-Means Model

229

Clustering Higher-Order Cells

235

Clustering Edges

235

Targeted Edge Clustering

235

Untargeted Edge Clustering

236

Applications

237

Image Segmentation

237

Social Networks

243

Machine Learning and Classification

244

Gene Expression

248

Conclusion

250

Manifold Learning and Ranking

251

Manifold Learning

252

Multidimensional Scaling and Isomap

253

Laplacian Eigenmaps and Spectral Coordinates

255

Locality Preserving Projections

257

Relationship to Clustering

259

Manifold Learning on Edge Data

259

Ranking

261

PageRank

261

PageRank as Advection

263

HITS

264

Applications

265

Shape Characterization

265

Point Correspondence

268

Web Search

270

Judicial Citation

272

Conclusion

274

Measuring Networks

275

Measures of Graph Connectedness

276

Graph Distance

276

Node Centrality

277

Distance-Based Properties of a Graph

278

Measures of Graph Separability

282

Clustering Measures

282

Small-World Graphs

285

Topological Measures

287

Geometric Measures

289

Discrete Gaussian Curvature

290

Discrete Mean Curvature

291

Applications

293

Social Networks

293

Chemical Graph Theory

294

Conclusion

296

Appendix A Representation and Storage of a Graph and Complex

298

General Representations for Complexes

298

Cells List Representation

298

Operator Representation

299

Representation of 1-Complexes

300

Neighbor List Representation

300

Appendix B Optimization

302

Real-Valued Optimization

303

Unconstrained Direct Solutions

304

Constrained Direct Solutions

305

Optimization with Boundary Conditions

305

Optimization with Linear Equality Constraints

308

Optimization with Linear Inequality Constraints

310

Ratio Optimization

312

Descent Methods

315

Gradient Descent

315

Newton's Method

315

Descent Methods for Constrained Optimization

318

Nonconvex Energy Optimization over Real Variables

319

Integer-Valued Optimization

319

Linear Objective Functions

319

Quadratic Objective Functions

321

Pure Quadratic

321

General Pairwise Terms

323

Higher-Order Terms

325

General Integer Programming Problems

325

Appendix C The Hodge Theorem: A Generalization of the Helmholtz Decomposition

326

The Helmholtz Theorem

326

The Hodge Decomposition

333

Summary of Notation

338

References

341

Index

359

Color Plates

366