Авторизация
Поиск по указателям
Griewank A. — Evaluating derivatives: principles and techniques of algorithmic differentiation
Обсудите книгу на научном форуме
Нашли опечатку? Выделите ее мышкой и нажмите Ctrl+Enter
Название: Evaluating derivatives: principles and techniques of algorithmic differentiation
Автор: Griewank A.
Аннотация: Algorithmic, or automatic, differentiation (AD) is concerned with the accurate and efficient evaluation of derivatives for functions defined by computer programs. No truncation errors are incurred, and the resulting numerical derivative values can be used for all scientific computations that are based on linear, quadratic, or even higher order approximations to nonlinear scalar or vector functions. In particular, AD has been applied to optimization, parameter identification, equation solving, the numerical integration of differential equations, and combinations thereof. Apart from quantifying sensitivities numerically, AD techniques can also provide structural information, e.g., sparsity pattern and generic rank of Jacobian matrices.
This first comprehensive treatment of AD describes all chainrule-based techniques for evaluating derivatives of composite functions with particular emphasis on the reverse, or adjoint, mode. The corresponding complexity analysis shows that gradients are always relatively cheap, while the cost of evaluating Jacobian and Hessian matrices is found to be strongly dependent on problem structure and its efficient exploitation. Attempts to minimize operations count and/or memory requirement lead to hard combinatorial optimization problems in the case of Jacobians and a well-defined trade-off curve between spatial and temporal complexity for gradient evaluations.
The book is divided into three parts: a stand-alone introduction to the fundamentals of AD and its software, a thorough treatment of methods for sparse problems, and final chapters on higher derivatives, nonsmooth problems, and program reversal schedules. Each of the chapters concludes with examples and exercises suitable for students with a basic understanding of differential calculus, procedural programming, and numerical linear algebra.
Язык:
Рубрика: Математика /
Статус предметного указателя: Готов указатель с номерами страниц
ed2k: ed2k stats
Год издания: 1987
Количество страниц: 394
Добавлена в каталог: 03.03.2014
Операции: Положить на полку |
Скопировать ссылку для форума | Скопировать ID
Предметный указатель
Activity analysis 111
Additive tasks 29
Addressing scheme & 26
ADIFOR 40 45 86 113 128 142 159 185 189 211
Adjoint as Lagrange multiplier 52
Adjoint complexity 84
Adjoint consistency 74
Adjoint gradient complexity 57
Adjoint higher order 238
Adjoint mode 8
Adjoint of adjoint 69 79
Adjoint of tangent 69 82
Adjoint operation 10 53
Adjoint pairs 55
Adjoint procedure 49
Adjoint second-order 83 107
Adjoint sensitivity equation 284
Adjoint Taylor exponential 233
Adjoint value 51
Adjoint variable 8
Adjoint, implicit 284
Adjoint, incremental 48
Adjoint, nonincremental 48 58
ADOL-C 112 225 233 312
Advancing 313
Algorithmic differentiation (AD) 1
Alias-safe 40 53
Aliasing 74
allocation 26
Approximants of Intermediates 221
Assignment, incremental 48 73
Assignment, iterative 71
Assignment, operator 100
Assignment, single 26 305
Average domain size 125
Back-elimination 170
Bough 313
Call tree 306
Chain rule 24
Chance constraint 253
Cheap gradient principle 60
Checkpointing 319
Chromatic number 140
Code preparation 210
Coleman — Verma partition 149
Coloring 140
Compatibility requirements 30
complexity 29—34
composition 223
Compressed tensor 152
Compression, column 139
Compression, combined 148
Compression, CPR 146
Compression, NR 146
Compression, row 139
Compression, simultaneous 152
Compression, two-sided 151
Consistency, adjoint 75
Consistency, derivative 28
Consistency, pattern 145
Contractive Preconditioning 289
Convergence rate 288
Copy-on-write 54
Cotangent 46
Cross-country 163
Curtis-Powell-Reid (CPR) seeding 140
Deactivation 101 291
Deinitializing 99
Derivative, addressing 41
Derivative, aliasing 41
Derivative, Cartesian 37
Derivative, directional 42 97 137
Derivative, discontinuous 252
Derivative, recurrence 291
Derivative, second 151
Difference quotient 137
Differentiable piecewise 261
Differential algebraic equation 247
Differential equation ordinary 243
Differentiation, algorithmic or automatic or computational 1
Differentiation, backward 46
Differentiation, generalized 264
Differentiation, reverse 46
Divided difference see "Difference quotient"
Doublet 96
Dynamic programming 322
Edge-elimination 160
Edge-elimination rule 169
Elemental Differentiability 24
Elemental function 21
Elemental partials 22
Elemental Task Boundedness 33
Elimination, cross-country 167
Elimination, Gaussian 166
Elimination, rule 171
Elimination, sequence 180
Evaluation, graph 16
Evaluation, procedure 16
Evaluation, procedure, three-part 18
Evaluation, program 16
Evaluation, trace 5 16
Evolution 72 180 319
Extended Jacobian 22
Extended system 21
External memory 304
Feasible partition 150
Finite selection 261
Finite Slopes 281
Finite termination 172
Fixed point 298
Fortran calling sequence 113
Forward 37 161
forward compatibility 27
Forward incremental 173
Forward mode 6 7
Forward motion or sweep 49
Forward nonincremental 173
Forward overwriting 41
Forward propagation 127
Forward sparse 130
Front-elimination 171
Function addressing 71
Function allocation 26
Function composite 15 19
Function elemental 18 20
Function Heaviside 266
Function intermediate 20
Function kink 265
Function pole 267
Function reciprocal 23
Function rewriting 211
Function root 266
Function step 266
Funnel 188
Gaussian elimination 166
Generic rank 187
Gradient see "Adjoint"
Gradient complexity 56
Gradient, higher order 233
Gradient, principle 60
Gradient, propagation 45
Gradient, reduced 52
Graph, bipartite 168
Graph, computational 333
Graph, quotient 184
hessian 121
Hessian cost ratio 155
Hessian graph 190
Hessian Lagrangian 133
Hessian, reduced width 155
IEEE arithmetic 266
Implementation, forward 98
Implementation, reverse 103
Implicit, adjoint 284
Implicit, tangent 284
Incremental, forward 173
Incremental, recursion 48
Incremental, reverse 173
Independent variable see "Variable"
Index domain 123
Index range 124
Initialization function 21
Interaction binary 142
Interface contraction 185
intermediate see "Variable"
Isolated Criticalities 267
Jacobian 121 161 201
Jacobian Regularity 283
Jacobian, extended 163
Jacobian, generalized 280
Jacobian, global 187
Jacobian, higher order 233
Jacobian, local 187
Jacobian, transpose 64
Jacobian-vector 39
Joint allocation xviii
Karush — Kuhn — Tucker 52 284
Kink function 265
Laurent model 276
Laurent number 270
Levi-Civita number 256
Lifespan 56
Lighthouse example 253
Limited-Memory BFGS 89
Link, strong or weak 312
Local Jacobians and procedures 187
Locations 26
LU factorization 88
Markowitz heuristic 179
Matrix chaining 159 184
Matrix compression 135 139
Matrix reconstruction 153
Matrix seed 137
Matrix sparse 121
Matrix width 125
Matrix, Vandermonde 143
Memory location 26
Mode, forward 6 7
Mode, joint or split 312
Mode, reverse 8 103 104
Mode, selection 211
Mode, vector 42 55 109
Motion 309
Motion, cross-country 309
Multistep contractivity 292 296
Newsam — Ramsdell (NR) principle 139
Newsam — Ramsdell (NR) seeding 143
Newton method 256
Newton scenario 286
Newton step 195 286
Newton truncated 108
Nonincremental adjoint 58
Nonincremental adjoint, evaluation 48
Nonincremental form 80
Nonincremental forward 173
Nonincremental recursion 48
Nonincremental reverse 173
Nonlinear width or height 209
Odyssee 40 79 86 113 305 312 317
Overwrite 5 27 56
Pairs, compatible 308
Parallel chain 331
Parallel program 333
Partial separability 206
Partials, elemental 39
Partials, mixed 219
Partition Coleman — Verma 149
Path connectedness 124
Pathlength and pathvalue 168 184
Piecewise differentiability 261
Piggyback approach 286 300
Pole function 267
Polynomial core 23
Pre-value 53 76
Preaccumulation 159 185
Preconditioning 63 298
Preelimination 178
Preferred direction 251
Preprocessor 95 113
Problem preparation 209
Program, branch 266
Program, overloading 93
Program, transformation 91
Program, variable 5
Quotient convergence factor 289
Quotient graph 316
Randomly accessed memory 41
Range size 126
Rank-one example 199
Recalculation 304
Reciprocal 23
Recomputation 78
recording 313 315
Recursion, incremental or nonincremental 48
Reduced width 154
Reflexive 80
Regular Arc and Determinacy 279
Related task 29
Return motion or sweep 304
returning 312 313 315
Reversal 304 313
Reversal chain 320
Reversal joint 308 311
Reversal schedule 303—313 315—334
Reversal schedule, chain 328
Reversal schedule, parallel 330
Reversal Schedules 321
Reversal split 308 311
reverse 37 161
Reverse, compatibility 27
Reverse, differentiation 46
Reverse, implementation 103
Reverse, incremental 173
Reverse, mode 8 10
Reverse, motion or sweep 49
Reverse, nonincremental 173
Reverse, propagation 129
Reverse, sparse 132
Reverse, statement-level 188
Reverse, sweep 49
reversing 310
Root convergence factor 292
Root function 266
Runtime functional 31
Schedule, optimal 326
Schedule, reversal 315 328 330
Second derivatives 130 132
Seed, directions 37
Seed, matrix 137
Seeding, Curtis — Powell — Reid (CPR) 140
Seeding, Newsam — Ramsdell (NR) 155
Sensitivity, perturbation 51
Separability, argument 126 207
Separability, partial 131 202
Separability, value 206
Simplified recurrence 291
Source transformation 111
Sparsity 137
Sparsity, compression 137
Sparsity, dynamic 122
Реклама