Subject Index

A Antenna, geometrical view, 671

Absolute guarantee, 710 Approximate Pareto set, 783
Absolute minimum, 108 Approximation, linear, 728
Absolute-value constraint, 37 Armijo-like procedure, 460, 576
Acceptance criterion, 722 Artificial cost function, 345, 394
Acceptance–rejection (A–R) methods, 710, 721 Artificial variables, 343
ACO. See Ant colony optimization (ACO) equality constraints, 351
Active constraint, 515 two-phase simplex method, 343
Active inequality, 152 artificial cost function, 345
Active/tight constraint, 155 degenerate basic feasible solution, 355
Adaptive numerical optimization procedure, 58 phase I algorithm, 346
Additive model, 814 infeasible problem, 346
one-way table, 814 phase II algorithm, 348
orthogonal array, 813, 815 phase I problem, definition of, 345
Advanced first order second-moment method unbounded solution, 353
(AFOSM), 841 Aspiration point, 781
Algorithm, failure of, 516 Associative law, 855
Allowable strength design (ASD) approach, Asymmetric matrix, 857
640, 647 Asymptotic error constant, 495
Allowable stress, 35 Attainability, 776
American Association of State Highway and Attainable set, 776
Transportation Officials (AASHTO), 268 Augmented Lagrangian methods, 220, 490, 492
American Institute of Steel Construction Augmented matrix, 859
(AISC), 639, 644, 652, 656
manual, 700 B
Analysis of means (ANOM), 813 Banded matrix, 857
additive model for function, 816 Bar cross-sectional shapes, 36
Analyze designs, 6 Barrier function methods, 490
Annealing process, 693 advantages and disadvantages of, 492
Answer Report from Solver, for linear programming Basic calculus concepts, review of, 113
problem, 265 Basic feasible solutions, 321, 336, 390, 412
Ant colony optimization (ACO), 755 Basic infeasible solutions, 330
algorithm, 759–760 Basis functions, 60
finding feasible solutions, 762–763 Bayesian approach, 717
pheromone deposit, 763 BBM. See Branch-and-bound methods (BBM)
pheromone evaporation, 763 Beam design problem, graphical solution, 90
problem definition, 760–761 using MATLAB, 91
algorithm for traveling salesman problem, 757 Bending stress, partial derivatives of, 612
ant behavior, 755 BFGS method. See Broyden-Fletcher-Goldfarb-Shanno
simple model/algorithm, 756 (BFGS) method
path finding capability, 756 Binary variable, 683
probabilities, 759, 763 Binomial crossover, 752
virtual ant changes, definition of, 758 Block shear, 644
Ant Colony Optimization and Swarm Intelligence Bolt insertion sequence, 748
(ANTS), 740 Boltzmann–Gibbs distribution, 694

929
930 Subject Index

Bound-constrained optimization, 572 Compression members, optimum design of

algorithm, 597 constraints, formulation of, 648
optimality conditions, 573 data/information collection, 645
projection methods, 574 design variables, definition of, 648
step-by-step algorithm, 575–576 formulation of problem, 644
step size calculation, 576 optimization criterion, 648
variable close to upper/lower bound, 577 problem formulation
Bounded objective function method, 788–789 for elastic buckling, 651
Branch-and-bound methods (BBM), 687, 711 discussion, 652
linear problems, 687 for inelastic buckling, 649
local minimizations, 690 project/problem description, 645
nonlinear continuous problems, 691 Compromise-programming methods, 787
solution of continuous subproblems, 691 Compromise solution, 779
Branching, 689 Computational algorithm, 548
Broyden-Fletcher-Goldfarb-Shanno (BFGS) method, Computational approximations, 796
482, 574 Conjugate gradient directions, 445
quasi-Newton methods, 283 Conjugate gradient method, 448
limited-memory, 576 Constrained optimizations
updating procedure, 597 algorithms, conceptual steps, 513, 514
B-splines, 630 second-order conditions, 212
Buckling constraints, 53 general constrained problems, 213
insights for, 214
C strong sufficient condition, 215
Cabinet design, 40–43 sufficient conditions, for general constrained
data and information collection, 40 problems, 214
formulation of constraints, 41 Constrained optimum design, 555
mathematical formulation, 41 numerical methods for, 511
optimization criterion, 41 algorithms for constrained problems
project/problem description, 40 implementation of iterations, 513–514
Can, design of, 28 iterative process, 512
project/problem description, 28 basic concepts related, 512
Candidate optimum points, 127 algorithm, convergence of, 516
Candidate points, 784 algorithms for constrained problems, 512
Canonical representation, 872 constraint status, at design point, 515
Cantilever beam, 20 descent function, 516
design problem, 21 constrained steepest–descent (CSD) method, 512
design variables, 23 potential constraint strategy, 556–558
formulation for optimum design, 28 Constrained problem, linearization of, 517
of hollow square cross-section, 21 linearized subproblem
notation and data for, 22 definition of, 518–519
Cantilever column, 43 notation, 517
Cantilever structures, subjected to shock Constrained problems, penalty function for, 694
input, 626 Constrained steepest descent (CSD)
Center of gravity (CG), 98 algorithm, 542, 548, 558, 560
Chain rule of differentiation, 468 geometrical interpretation, 562
Cholesky factors, 582 inexact step size, 565
Chromosome, 750, 751 with inexact step size, 565
Classical Newton method, 472 observations, 548–549
Clustering methods, 718 step size determination, 568
Coefficient matrix, 859, 871 direction, 531, 547, 591
Cofactor expansion for |A|, 861 method, 512, 547
Coil springs, 47, 48 Constraint boundary, for inequality, 73
design of, 47–49 Constraint correction (CC) algorithm, 701
Column matrix, 853 Constraint function cells, 274
 Subject Index 931
Constraint normalization, 242, 244 Derivative-based optimization method, 609
equality constraint, 244 Derivative-based search methods, 239, 424
inequality constraint, 245 Descent function, 541
Constraints, 11 calculation of, 545
inequality, 11 golden section search, 543
for LP problem, 352 second trial point, 546
sensitivity theorem, 196 value, 545
set, 107 Descent method, 427
variation sensitivity theorem, 172 Design activities, 6
Continuous functions, 14 Design change vector, 581
Continuous-variable optimization problem, 59 Designing engineering systems, 4
ContourPlot command, 79, 80 Design of experiments, for response surface
Contraction operation, 502 generation, 805
Control force constraint, 629 Design optimization
Controlled random search (CRS), 720, 731 formulated as problem, 3
basic idea of, 720 iterative process, 4
global phase, 720 overview of, 3
local phase, 721 Design problem
Conventional design method, 7 with multiple solutions, 85–86
vs. optimum design process, 6–7 with unbounded solutions, 86–87
Convex functions, 184 Design space, 773
characterization of, 181 Design under uncertainty, 833
Convex interval, 179 Design variable bounds
Convexity, 180 constraints, 623
Convex polyhedral set, 313 for global optimization problem, 729
Convex polyhedron, 322 Design variables, 22, 42, 248
Convex programming problem, 190 vector, 641
Convex sets, 184 Determinant of matrix, 860
Coordinate system set-up, 73 Determinants, 859–862
Cost coefficients, 370 leading principal minor, 862
Cost function, 37 properties of, 861–862
contours, 89 singular matrix, 862
value, 517 Deterministic methods, 709, 710
Cost space. See Criterion space DFP quasi-Newton method, 497
CPU time, 760 Diameter constraint, 49
Cramer’s rule, 860 Differential elastic line equation, 668
Criterion space, 773 Differential equation (DE), 626
two-objective optimization problem, graphical Differential evolution algorithm (DEA), 750, 752–753
representation of, 775 application, 753
Critically important, 247 crossover operation to generate the trial design, 752
Crossover operation with one-cut point, 745 donor design, generation of, 751
Cross-section, of plate girder, 270 initial population, generation of, 750
CRS. See Controlled random search (CRS) main steps, 753
CSD. See Constrained steepest descent (CSD) notation and terminology, 751
Cumulative distribution function (CDF), 834 trial design, acceptance/rejection of, 752
Curvature condition, 462 Dirac delta function, 630
Direction-finding subproblem, 589
D Directions of descent, 427
Data optimization, 646 Direct search methods, 239, 498, 739
Davidon–Fletcher–Powell (DFP) method, 479 Discontinuous functions, 14
DEA. See Differential evolution algorithm (DEA) Discrete design
Definition of minima, graphical interpretation, 109 with orthogonal arrays, 813–816
Deflection constraint, 49 variables, 684
Dependent variable cells, 274 Discrete/integer-variable optimization problem, 59
932 Subject Index

Discrete variable design, 58, 683, 813 equality constraints case, 220
Discrete variable optimum design concepts/methods, inequality constraints case, 226
683, 701 theorem, 224
adaptive numerical method, 699–701 lower bound for primal cost function, 228
continuous variable optimization, 701 problem solving, 222
basic concepts/definitions, 684 saddle points, 228
mixed variable optimum design problem theorem, 228
(MV-OPT) strong duality theorem, 227
classification of, 685 weak duality theorem, 227
definition of, 684 Dual problem, 220
branch-and-bound methods, 687 Dual tableau, 408
basic, 687–688 Dual variables, 359
general MV-OPT, 691–692 Dummy variables, 851, 852
with local minimization, 689–690 Dynamic displacement constraint, 631
dynamic rounding-off method, 696 Dynamic rounding-off algorithm, 696
algorithm, 696
integer programming, 692–693 E
linked discrete variables, 698, 699 e-Constraint, 789
neighborhood search method, 697 ε-Constraint approach, 789
selection of, 699 Eigenvalues, 882, 883
sequential linearization methods, 693 Eigenvectors, 882, 883
simulated annealing, 693–695 Elastic buckling, 647, 651
algorithm, 694 hot-rolled I–shapes, 659
Displacement constraint, 629, 634 W–shapes, 659
Displacement response Elastic line equation, 670
optimum with minimization Elements of the matrix, 852
of control effort, 636 Elimination process, 859
of error, 632 Elite points, 783
time as performance index, 637 Energy expenditure, 24
Distributive law, 855 Engineering analysis, 6
Domain elimination (DE) method, 731 Engineering design
flow diagram, 725 with analysis, 6
Domination pressure, 784 examples, 189
Double-subscript notation, for rectangular beam, 193
variables, 798 convexity, 193
D–string, 742, 744 KKT necessary conditions, 194–197
Duality, in linear programming, 399 sensitivity analysis, 197
dual LP problem, 399 vs. engineering analysis, 6
dual tableau to recover wall bracket, 189
primal solution, 407 convexity, 191
dual variables as Lagrange multipliers, 410 convex programming problem, 190
proof, 410 KKT necessary conditions, 191
equality constraints problem formulation, 191
alternate treatment, 402 sensitivity analysis, 193
treatment of, 401 Equal interval search, 438, 891
primal solution, determination of, 403 computer program, 892
standard primal LP problem, 399 Equality constraint, 58
Duality, in nonlinear programming, 220 function value, 517
equality/inequality-constrained problem, 226 gradient conditions, 150
gradient matrix of equality constraints, 221 Lagrange multiplier, 144–145
gradient of, 223 theorem, 149
Hessian of Lagrangian function, 221 problem, 143, 844
Lagrangian function, 221 Equality/inequality-constrained problem,
local duality 226, 493
 Subject Index 933
Equivalence class sharing, 785 Finite–element application, 609
Equivalent single-degree-of-freedom system Finite exact methods, 710
displacement response of, 628 Finite number, 709
Errors, meta-model, 796 First-order necessary condition, 129
Euclidean space, 884 First-order necessary conditions, 131–134
Euler stress, 647 First-order reliability method
Evtushenko’s method, 711 (FORM), 845
Exact penalty function, 539 Fitness sharing, 784
Excel, 152, 307 Flagpole, 100
Excel Solver, 237, 253, 260 Fletcher–Reeves formula, 449
for linear programming problems, 260 Flexural members, 653
Answer Report from Solver, 265 Flywheel-shaft system, 295
Sensitivity Report from Solver, 266 Free-body diagram, 34
Solver Parameters dialog box, 263 Function gradients, 520
Solver Results dialog box, 264 Function of variable, 13
worksheet, 262 Fundamental natural frequency, 53
nonlinear equation, roots of, 253
Solver Answer Report, 257 G
Solver Output, 256 Gaussian (normal) distribution, 821
Solver Parameters dialog box, 254, 255 Gaussian elimination method, 314, 858, 862, 865
Solver Results dialog box, 256 elimination process, 862
nonlinear equations, roots of set, 257 inverse, 868
Solution to KKT Cases with Solver, 260 procedure, 862–865, 867
Solver Parameters dialog box, 259 Gauss-Jordan elimination method, 317, 326, 330, 392,
worksheet, 257, 258 395, 397, 866, 869, 872
nonlinear programming general solution of linear system, 876
optimum design of springs, 266–268 inverse, 868
of plate girders, 268 in tabular form, 875
data and information collection, 270–271 General design optimization model, 144
design variables, definition of, 271 Generalized descent methods, 712, 713
formulation of constraints, 272 Generalized reduced gradient (GRG) method,
optimization criterion, 271 531, 593
project/problem description, 268–269 nonlinear, 260
solution, 274–276 General-purpose software, use of, 615
Solver Parameters Dialog Box, 273 integration of application, 616
spreadsheet layout, 272–273 software selection, 616
for unconstrained optimization problems, 260 Genetic algorithms (GAs), 778
worksheet and Solver Parameters dialog Genetic and Evolutionary Computation Conference
box, 261 (GECCO), 740
Excel worksheet, 254, 261 Geometrical representation, 10
for linear programming problem, 262 Global convergence, 428
for spring design problem, 267 properties, 494
Exhaustive search, 178 Global/local minima, definitions of, 106
Experimental errors, 796 minimum, existence of, 112
Exterior penalty methods, 491 minimum/maximum, 107, 108
Global minimum, 184, 708
F point. See Absolute minimum
Feasibility tolerance, 242 Global optimality, 178
Feasible criterion space, 774 constraint, transformation of, 187
Feasible design space, 107, 772 convex functions, 181
Feasible directions (FD) method, 531 convex programming problems, 183
Feedback loops, 5 sufficient conditions, 188
Feedback mechanism, 755 convex sets, 179
Fibonacci sequence, 437 Hessian condition, 182
934 Subject Index

Global optimization concepts/methods, 707 plotting functions, 78–79

basic solution concepts, 708 use of mathematica, 77
global minimum Graphical representation, 153, 185
characterization of, 708 Graphical solution process, 71, 72
searching for, 709 for beam design problem, 90
controlled random search for minimum-weight tubular column, 88–89
basic idea of, 720 profit maximization problem–formulation, 72–73
deterministic methods, overview of, 710 step-by-step procedure, 73–77
covering methods, 711–712 Grid points, stopping criterion, 634
generalized descent, 712–714
descents/ascents, alternation of, 713 H
Golf methods, 714 Hessian modification, 478
tunneling method, 714–715 Hessian matrix, 15, 114, 115, 139, 143, 187, 469, 472,
zooming method, 712 475, 498
deterministic/stochastic methods, 709 of Lagrange function, 579
local-global stochastic methods, 723 for quadratic form, 126
numerical performance methods, 729 second-order partial derivatives, 115
features of methods, 730 Hessian updating procedure, limited-memory, 595
stochastic zooming and domain elimination Heuristic methods, 710
methods, 731–732 Hooke–Jeeves method, 498
structural design problems, 732–733
unconstrained problems, performance, 731 I
Global optimization methods, 708 Identity matrix, 857
characteristics of, 730 IDESIGN program, 701
Global optimum design, 188 IF THEN ELSE condition, 648, 661
Global phase, 720 Implicit constraints, 59
Goal programming, 789–790 Implicit enumeration, 686
Golden section search, 439, 441, 894 Implicit functions, gradient evaluation, 609–610
subroutine GOLD, 895 Improving feasible direction, 589
Golden section search procedure, 437 Inactive constraint, 515
Goldstein test, 462, 463 Inactive inequality, 153
Golf methods, 714 Including inequality constraints, 580
Gradient-based search methods, 238, 424, 425. See also Independent variables, 316
Derivative-based search methods Independent variable transformation, 631
Gradient conditions, 150 Inelastic buckling, 647
Gradient evaluation, 534, 543 constraint, 649
implicit functions, 609–610 Inequality
of implicit functions, 609–610 constrained problem, 155
Gradient method, 442 constraint boundary, 73
Gradient projection (GP) method, 531, 591 plot, 73
Gradient vector, 114, 115, 464 constraint functions
partial derivatives of function, 114 for optimum design problem, 184
Graph-editing capability, 82 constraint function value, 517
Graphical optimization feasible/infeasible side, 74
feasible region, identification of, 80 Inexact line search, 542, 560
inequality, infeasible region basic concept of, 460
identification and shading of, 79–80 Inexact step size calculation
MATLAB uses, 81 basic concept, 560
editing of graph, 85 descent condition, 560–563
function contours, plotting of, 82 Infeasible design, 26
profit maximization problem, 83 optimization problem, 88
objective function contours, plotting of, problem, 87
80–81 Infeasible problem, 154, 248
optimum solution, identification of, 81 Infinite solutions, 316
 Subject Index 935
Inner array, 827 Lagrange multipliers, 57, 144, 145, 148, 149, 151, 155,
Insulated spherical tank design, 29 170, 172, 173, 175, 268, 359, 362, 540, 578, 581, 594,
constraints, formulation of, 30 620, 623, 635, 845
data and information collection, 30 constraint, scaling, 176, 177
optimization criterion, 30 equality constraint, 149, 844
project/problem description, 29 for lower-and upper-bound constraints, 573
Integer programming problems, 33, 43 of primal constraints, 399
Integer variable, 58, 683, 684 vectors, 221
Inter-disciplinary environment, 4 for inequality constraints, 415
Interval arithmetic, 711 Lagrangian function, 573, 588
Interval-reducing methods, 434, 539 Lagrangian methods, 708
Inverse barrier function, 491 Length of vector, 12
Inverse using cofactors, 866 Lexicographic method, 788
Irregular optimum point, graphical solution, 211 Life-cycle cooling cost, 30
Isocost curves, 139 LINDO, 307
Iterative process, 6 Linear approximations, 522, 728
Linear combination, 12
J of vectors, 876
Jacobian matrix, 593 Linear equations
m linear equations, in n unknowns, 869
K rank of matrix, 869
Karush-Kuhn-Tucker (KKT), 159, 207, 532, 708 m × n linear equations, 870–872
alternate form, 208 basic solutions, 874
cases with four inequalities, 171 set of vectors, 876
conditions, for LP problem, 411 linear independence, 876
optimality conditions, 412 vector spaces, 880
solution of, 412–414 Linear functions
first-order necessary conditions, 157, 257 constraints, 308
graphical solution, 164 cost function, 308
important observations, 160 Linear independence, 145
irregular points, 211 of vectors, 877
Lagrange function, 194 Linearized equality constraints, 517, 518
limitation of, 170 Linearized feasible region
necessary conditions, 161, 163, 207, 208, 210, graphical representation of, 521
218, 410 Linearized subproblem, 520, 535, 544
alternate form, 207–208 Linear least squares problem, 485
graphical solution, 211 Linearly independent, 877
irregular points, 210 Linear optimization problems, 851
quadratic programming (QP) problems, 415 Linear problems, BBM for, 687
second-order conditions, for constrained Linear programming (LP). See also Linear optimization
optimization, 212 problems
general constrained problems, 213 artificial variables, two-phase simplex method, 343
insights for, 214 basic concepts
strong sufficient condition, 215 boundary of feasible set, 314
sufficient conditions, for general constrained convexity of, 314
problems, 214 infinite roots, 314
solver results for, 259 related to problems, 313
worksheet and Solver Parameters dialog basic solutions, calculation of, 323
box, 258 basic solutions, 326
Kilopascal (KPa), 243 pivot step, 324
KKT. See Karush-Kuhn-Tucker (KKT) tableau, 323
basic theorems, 335
L multiple solutions, 336
Lagrange function, 228, 577, 581 graphical solution, 360
936 Subject Index

Linear programming (LP). See also Linear optimization “≤ type” constraints, treatment of, 310
problems (cont.) “≥ type” constraints, treatment of, 311
optimum solution, 356 unrestricted variables, 311
number of basic solutions, 322 Linear programming methods, for optimum design,
to problems, 321 307, 389
postoptimality analysis, 356 alternate simplex method, 397
coefficient matrix, changes, 372 simplex algorithm, 396–397
constraint limits, changes, 358 simplex method, derivation, 389
cost function, change, 359 artificial cost function, 394–395
Lagrange multipliers, recovery of, 358 basic variable, 393
Lagrange multiplier values, 358 canonical form, 389–390
limits on changes in resources, 365 nonbasic variable, 391, 392
new values of basic variables, 366 cost function, 391
ranging cost coefficients, 369 optimum cost function, 392
basic variables, 370 pivot step, 395–396
nonbasic variables, 369 reduced cost coefficients, 392
ranging right-side parameters, 365 unbounded problem, 392
problems, 26, 33, 57, 307 Linear simultaneous equations, 457
Answer Report from Solver, 265 Linear systems, 858–859
Excel solver, 260 Line search function, 430
with multiple solutions, 340 Line search problem, 429
Sensitivity Report from Solver, 266 Line search termination criterion, 431, 456
Solver Parameters dialog box, 263 Linked discrete variable, 683
Solver Results dialog box, 264 Lipschitz condition, 548
solving, 307 Lipschitz constant, 710, 711
worksheet, 262 Load and resistance factors design (LRFD)
simplex method, 329 approach, 640
basic idea, 330 Local duality theorem, 220, 224, 225
basic steps, 330 Local-global stochastic methods, 723
cost function in terms of nonbasic conceptual, algorithm, 723, 724
variables, 330 domain elimination method, 724–726
2D/3D space, 329 stopping criteria, 726
KKT conditions, problem, 411 operations analysis of methods, 727
standard problem, definition of, 308 checking, proximity of point, 727
expanded form, 309 design variable constraints, 729
matrix form, 309 point and trajectory distance, 728
summation form, 309 trajectory approximation, 728
terminology, 319 stochastic zooming method, 726
basic feasible solution, 319 Local minimum point. See Relative minimum
basic solution, 319 Local optimality conditions, 128
basic variables, 319 Log barrier function, 491
basis, 319 LP. See Linear programming (LP)
convex polyhedron, 319
degenerate basic solution, 319
feasible solution, 319 M
feasible solution, 319 Marquardt’s algorithm, 478
nonbasic variables, 319 Marquardt’s method, 478
optimum basic solution, 319 Mathematica, 152
optimum solution, 319 Mathematical approximations, 796
vertex/extreme point, 319 Mathematica Optimization Tool Box, 237
transcription to standard, 310 Mating strings, 743
maximization of function, 311 MATLAB, 152, 165, 279, 307
nonnegative constraint limits, 310 constrained optimum design problems, 285
treatment of inequalities, 310 file, profit maximization problem, 83
 Subject Index 937
graphical solution method and basic optimization Mean value first-order second-moment method
concepts, 279 (MVFOSM), 841
optimization toolbox, 279–281 Megapascal (MPa), 243
output from, explanation of, 281 Members for flexure, optimum design of, 652
scalar/array/matrix operations, 280 constraints, formulation of, 660
variables/expressions, 279 data/information collection, 652
optimum design examples, 288 deflection requirement, 660
column design for minimum mass, 290 design of flexural members, 653
data/information collection, 291 data for optimizing, 653
design variables, definition of, 292 moment strength requirement, 656
formulation of constraints, 292 nominal bending strength
optimization criterion, 292 of compact shapes, 657
project/problem statement, 290 of noncompact shapes, 658
flywheel design for minimum mass, 293 project/problem description, 652
constraints, formulation of, 297 shear strength requirement, 659
data/information collection, 296 Member stresses, 623
design variables, definition of, 297 Metaheuristics methods, 764
optimization criterion, 297 Meta-models
project/problem statement, 293 for design optimization, 771, 773
maximum shear stress location, 288–289 mathematical model, 772
constraints, formulation of, 289 response surface method (RSM), 773
criterion, 289 normalization of variables, 776, 778
data and information collection, 288 normalization of variables, procedure, 777–778
design variables, definition of, 288 quadratic response surface generation,
program, 81, 82 773–777
unconstrained optimum design errors, 796
problems, 282 examples, 796
MATLAB Optimization Tool Box, 237 m-File, for objective function, 283
Matrices Minimum-area beam design problem, 91
addition of matrices, 853 Minimum-cost cylindrical tank design, 46–47
condition number, 885 Minimum-weight system, 700
definition of, 851 Minimum-weight tubular column design, 43, 89
elementary row–column operations, 856 formulation 1 for, 44–45
equivalence of matrices, 856 formulation 2 for, 45–46
inverse of matrix graphical solution, 88–89
Gaussian elimination, 867 Mixed variable optimum design problem (MV-OPT)
Gauss–Jordan elimination, 866 classification of, 685
inverse by cofactors, 866 combinatorial problems, 685–686
multiplication of matrices, 853 design variables, to other parameters, 685
norm of, 884 discrete variables, nondiscrete values, 685
condition number, 884 functions continuous/differentiable, 685
vectors, 884 functions nondifferentiable, 685
null matrix, 853 solution concepts, overview of, 686–687
partitioning of matrices, 857–858 definition of, 684
scalar product/dot product of vectors, 856 Modified Newton method, 473–477, 713, 897
square matrices, 857 computer program, 904
transpose of a matrix, 855 Monte Carlo simulation (MCS), 845
types of, 853 Most probable failure point (MPFP), 842
vector, 853 reliability index, 845, 846
Matrix of order, 852 geometric representation, 843
Maximin fitness function, 783 Most probable point (MPP), 842
Maximum constraint violation, 540 Move limits, 525
Maximum point, definition of, 111 Multilayered graphical representation
m-Digit binary string, 742 of discrete variable problem, 762
938 Subject Index

Multi-objective genetic algorithms (GAs), 781 differential evolution algorithm, 750, 752–753
elitist strategy, 783 crossover operation to generate the trial
niche techniques, 784–785 design, 752
Pareto fitness function, 783 donor design, generation of, 751
Pareto-set filter, 783 initial population, generation of, 750
ranking, 782–783 notation and terminology, 751
Tournament selection, 784 trial design, acceptance/rejection of, 752
vector-evaluated, 782 genetic algorithms (GA), 747
Multi-objective optimization methods, 780, 790, 791 applications of, 749
Multi-objective optimization problems, 771 basic concepts/definitions, 741
Multi-objective optimum design concepts/methods, design representation, 741–742
771, 773 fitness function, 743
criterion space/design space, 773–776 initial generation/starting design set, 742–743
objective functions, normalization of, 780–781 crossover
optimization engine, 781 number of, 746
pareto optimal set, generation of, 780 fundamentals of, 743
preferences/utility functions, 779–780 crossover, 744
problem definition, 771–772 mutation, 745
selection of methods, 790 population, leader of, 746
solution concepts, 776 reproduction procedure, 744
compromise solution, 779 stopping criteria, 747
efficiency/dominance, 778 mutation, number of, 745
pareto optimality, 777 for optimum design, 741
utopia point, 778 immigration, 747
weak pareto optimality, 777 multiple runs, for problem, 747
vector methods/scalarization methods, 780 particle swarm optimization (PSO), 764
Multi-objective problems, 241 algorithm, 765–766
Multiple optimum designs, 86 swarm behavior/terminology, 764
Multiple performance requirements, three-bar sequencing-type problems, genetic algorithm, 748
structure, 619 relocation, 749
asymmetric three–bar structure, 621–625 NBR-6123 code, 666
comparison of solutions, 625 n-Dimensional column vector, 859
symmetric three–bar structure, 619–621 Necessary conditions, 136
Multiplier methods, 490 concepts of, 127
Multivariable unconstrained minimization, 283 for constrained problem, 152
MV-OPT problems, 685, 687 gradient condition, geometrical meaning of, 158
Karush–Kuhn–Tucker necessary conditions, 154
N KKT conditions, 159
Natural frequency, 51 first-order necessary conditions, 170
Nature-inspired algorithms, 739 important observations, 160
Nature-inspired methods, 739 limitation of, 170
Nature-inspired search methods, 739 role of inequalities, 152
algorithms, 739 switching conditions, 158
drawbacks of, 740 Nelder–Mead algorithm, 503–505, 721
ant colony optimization, 755 Nelder–Mead simplex method, 498
algorithm, 759–760 Newton methods, 428, 455, 473, 496
finding feasible solutions, 762–763 Newton–Raphson iterative procedure, 579
pheromone deposit, 763 Newton–Raphson method, 142, 578, 579, 592
pheromone evaporation, 763 Newton search direction, 474
problem definition, 760–761 Niche techniques, 784
algorithm for traveling salesman problem, 757 Noise coefficient, 722
ant behavior, 755 Noncompact shape, 664
simple model/algorithm, 756 Nondifferentiable problems, 694
virtual ant changes, definition of, 758 Nondominated check, 783
 Subject Index 939
Nongradient-based method, 283 Numerical solution process, for optimum design, 250
Nonhomogeneous system, 859 algorithm, 252
Nonlinear discrete problems, 692 feasible points, 251–252
Nonlinear equations, 489 general purpose software, integration of application,
Excel solver 250–251
roots of set, 257
Excel solver roots, 253 O
roots of set Off-diagonal elements, 121, 123, 857
Solution to KKT Cases with Solver, 260 One-cut-point, 744
Solver Parameters dialog box, 259 One-dimensional minimization, 459
worksheet, 257, 258 Optimal Bayesian estimate, 717
Solver Answer Report, 257 Optimal control problems, 625, 627
Solver Output, 256 Optimality conditions, 128
Solver Parameters dialog box, 254, 255 basic concept, 128
Solver Results dialog box, 256 functions of several variables, 135
Nonlinear optimization problems, 519 for functions of single variable, 129
Nonlinear programming algorithms, 537 for unconstrained variable problems, 131
Nonlinear programming, Excel solver Optimality criteria methods, 105
optimum design of springs, 266–268 Optimization engine, 781
Nonlinear programming, optimal control of systems, 625 Optimization methods, 4
minimum time control problem, 637–639 classification of, 106
prototype optimal control problem, 625–629 Optimization problem, 4
state variable, minimization of error, 629–635 Optimization toolbox functions, 281
minimum control effort problem, 635–637 Optimization variable, 23
numerical results, 631 Optimum control forces
numerical solution, formulation for, 630–631 to minimize control effort, 636
problem normalization, 631–634 to minimize error, 633
results discussion, 634–635 to minimize time, 638
system motion, formulations, 639 Optimum design process
Nonlinear programming problem (NLP) methods, 26, formulation, of complex engineering systems, 603
54, 307, 423, 577, 688 general mathematical model
Nonquadratic case, 497 active/inactive/violated constraints, 57
Nonunimodal function, 433 application to different engineering fields, 56
∞-Norm, 885 discrete/integer design variables, 58
Normalization procedure, 803, 805 feasible set, 57
Normalized shear stress, 290 “greater than type” constraints, treatment of, 56
Norm of matrices, 884 maximization problem treatment, 55
Norm of vectors, 12, 884 standard design optimization model, 54–55
Notebook, 78 standard model, important observations, 56–57
Null/zero matrix, 853 types of problems, 59
Numerical algorithms, 556 continuous/discrete-variable, 59
Numerical aspects, of problem formulation. See design variables as functions, 60
Problem formulation, numerical aspects of dynamic-response, 60
Numerical optimization methods implicit constraints, 59
feasible directions, method of, 588–590 network optimization problems, 59
generalized reduced gradient method, 592–593 smooth/nonsmooth, 59
gradient projection method, 591–592 problem formulation, numerical aspects of, 241
Numerical performance methods, 729 vs. conventional, 6–7
Numerical search methods, 238, 432, 512 vs. optimal control, 8
derivative-based methods, 238–239 Optimum Lagrange multipliers, 174
derivative-free methods, 240 Optimum points, representation of, 107
direct search methods, 239–240 Orthogonal arrays method, 807
nature-inspired search methods, 240 Orthogonal steepest–descent paths, 468
selection of method, 241 Outer array, 827
940 Subject Index

Out-of-plane loads, two-member frame, 617 Potential constraint strategy, 548, 556, 615
Output from optimization function, 281 Potential cost functions, 603
Potential energy function, 487
P Potentially active, 556
Pareto optimality, 777 Practical applications
Pareto optimal set, 777, 785 development of problem formulation, 60–61
illustration of, 775 of optimization, 601
Pareto optimal solution, 787 Practical design optimization, issues, 614
Pareto-set filter approach, 783 algorithm, selection of, 614
Partial derivatives, 114 potential constraint strategy, 614
of functions, 14 robustness, 614
of vector functions, 15 good optimization algorithm, attributes
Partial pivoting, 865 of, 614–615
Particle swarm optimization (PSO), 764 Practical design optimization problems, formulation
algorithm, 765–766 of, 602
swarm behavior/terminology, 764 example of, 603–604
Pattern search methods, 499 general guidelines, 602–603
PDF. See Probability density function (PDF) Predictive dynamics, 675
Penalty function method, 490 Preliminary design, 5
Penalty methods, 713 Primal cost function, 228
Performance requirements, 25 Primal problem, 408
Pheromone density, 760 Principle of stationary potential energy, 486
Pheromone values, 757 Probability density function (PDF), 819,
Physical programming, 779 833, 834
Pitting, 288 limit state, 839
Plate girders Probability mass function, 833
design problem, spreadsheet for, 274 Problem formulation process, 20
Excel solver, 268 cantilever beam, optimization criterion, 25
data and information collection, 270–271 constraints, formulation of
design variables, definition of, 271 equality and inequality, 26
formulation of constraints, 272 feasible design, 26
optimization criterion, 271 linear/nonlinear, 26
project/problem description, 268–269 restrictions, 25
solution, 274–276 data/information collection, 21
Solver Parameters Dialog Box, 273 design variables
spreadsheet layout, 272–273 for cantilever beam, 23
Point maximizing, 55 design variables, definition of, 22
Points, 8 numerical aspects of, 241
Poisson’s ratio, 288 general guidelines, 241–242
Polak–Ribiére formulas, 449 iterative process for development, 248–250
Pole structure, section, 668 scaling of constraints, 242–243
Polyhedron, 322 scaling of design variables, 246–247
Positive definite quadratic function, 497 optimization criterion, 24
Postoptimality analysis, 171, 356 project goals, 20
changing constraint limits, 171–172 Problem parameter vector, 820
constraint, effect of scaling, 176 Profit maximization problem, 83, 315, 328
constraint variation sensitivity result, generalization ABCDE, feasible region, 75
of, 177 graphical representation, 85
cost function scaling, effect of, 175 graphical solution, 77
first-order changes, in cost function, 172 objective function contour, 76
Lagrange multipliers, 171 Projection operator, 574
nonnegativity of, 173 Pshenichny’s descent function, 539, 540
practical use of, 174 PSO. See Particle swarm optimization (PSO)
Potential constraint index set, 556 Push-off factor, 590
 Subject Index 941
Q covariance, 837
QP. See Quadratic programming (QP) cumulative distribution function (CDF), 834
Quadratic approximation, 457, 800 expected value, 835–836
Quadratic form, 122 Gaussian (normal) distribution, 837
Quadratic interpolation, 456 inverse, 838
Quadratic loss function, 491, 825 mean/variance, 836
Quadratic programming (QP), 531 probability density function (PDF), 833
definition of, 414 probability of failure, 835
KKT conditions reliability index, 837
necessary, 415 standard deviation, 836
transformation of, 415–416 Reliability index, geometric representation, 843
problems, 414, 555 Reliable algorithms, 516
methods to solve, 555 Resource limits, 308, 358
simplex method for problem solving, 416–417 Response surface generation (RSG)
subproblem, 533, 537, 538, 594 design of experiments, 805–807
graphical representation, 533 Response surface method (RSM), 773,
Quadratic programming subproblem, 536 797, 805
direct solution, of QP subproblem, 596–597 approximate a function, 782
KKT necessary conditions, 594–596 bending stress constraint, 772
solution, 593 least squares method, 797
Quadratic response surface, generation of, mean values, one-way table
773–777, 799 graphical representation of, 815
Quasi-Newton methods, 428, 479, 497, 580, 587 normalization of variables, 776, 778
procedure, 777–778
R optimization, 810
Random tunneling, 722 optimum values, 812
Random variable, with mean value, 819 quadratic response surface generation,
Realistic practical bounds, 247 773–777
Real telecommunication steel pole, 666 shear stress constraint, 773
Recall, step size calculation problem, 456 Response surface methods, 240
Rectangular beam design problem, second-order Robust design approach
conditions, 218 defined, 817
Hessian of Lagrangian, 219 effect of uncertainties in problem parameters, 818
Hessians of cost function, 219 robust optimization, 818
KKT necessary conditions, 218, 220 mean, 818
Rectangular matrix, 857 probability density function, 819
Rectangular m × n system, 858 problem definition, 820–823
Reduced cost coefficients, 395 standard deviation, 819
Reduced gradient, 592 variance, 819
Reduced sample points, 718 Taguchi method, 825–827
Regression analysis, 485 Robust design method, 818
Relative minimum, 108 Robust design optimization, 818
Reliability-based design optimization (RBDO), design Robustness, 817
under uncertainty, 795, 833 index, 823
formulation of, 848 Robust optimization solution, 821, 824
reliability index, 838 Root-finding process, 488
advanced first order second-moment method Roulette wheel process, 744
(AFOSM), 841–845 Row matrix, 853
limit state equation, 838–839 RSM. See Response surface method (RSM)
linear limit state equation, 840 Rupture limit state constraint, 642
nonlinear limit state equation, 840
review of background material, 833 S
coefficient of variation, 837 Saddle point theorem, 228
correlation coefficient, 837 Sample computer programs, 891
942 Subject Index

Sawmill operation, 32 Sign support column, 100–102

data and information collection, 32 Simplest stochastic method
definition of design variables, 32 for global optimization, 717
formulation of constraints, 33 Simplex method, 338, 389, 417
mathematical formulation, 33 for LP problem, 411
optimization criterion, 33 Simulated annealing (SA) method, 687, 693
project/problem description, 32 Single-objective optimization problem, 772
Scalar function, 884 graphical representation of, 773
Scalarization, 780 Singular matrix, 866
Scalar matrix, 857 Slack variables, 154, 191, 310, 367
Scalar multiplication, 881 Slenderness ratios, 647
Scalar quantity, 12 constraint, 642
Scaling procedure, 247 SLP. See Sequential linear programming (SLP)
Schaffer’s method, 782 Small feasible region, 108
Schema, 741 Smooth optimization problems, 238
Search direction definition, 575 Solve constrained design problems, 489
Search methods, 106 Solver Answer Report, for spring design problem, 269
Second-order conditions, for constrained Solver Parameters dialog box, 255, 260
optimization, 212 for linear programming problem, 263
general constrained problems, 213 for plate girder design problem, 275
insights for, 214 Solver Results dialog box, 256
strong sufficient condition, 215 for linear programming problem, 264
sufficient conditions, for general constrained Solving n linear equations
problems, 214 in n unknowns, 858
Second-order necessary conditions, 130, 136, 213 Spherical tank, with intermediate variables, 31
Selection process, 782 Spring constant, 628
Selection step, 752 Spring design problem
Self-explanatory flowchart, 6 with design variables, 49
Sensitivity analysis, 171 material data, 697
Sensitivity Report from Solver SQP. See Sequential quadratic programming (SQP)
for linear programming problem, 266 Square matrix, 852, 857
Sequential linear programming (SLP) Standard design optimization model, 54
algorithm, 524–526, 535 defined, 707
method, 530 Standard LP problem, 308
move limits, 524 Stationary points, first-order necessary condition, 129
observations, 530–531 Steepest ascent, 465
positive/negative design changes, 526 Steepest–descent algorithm, 443
selection of proper move limits, 525 Steepest-descent direction, 158, 531
Sequential quadratic programming (SQP), 531, 547, 617 Steepest descent method, 442, 444, 469, 470, 479
algorithm, 582, 587 computer program, 898
descent functions, 587–588 gradient-based methods for unconstrained
methods, 577, 582, 631, 731 optimization, 897
QP subproblem, 578 Steepest–descent steps, 719
observations, 587 Step-by-step algorithm, 499
quadratic programming subproblem Step size calculation subproblem
derivation of, 578–580 descent function, 539–541
quasi-Newton Hessian approximation, 580–582 line search, 542
subproblem, search direction calculation, 532 Step size determination, 429, 562
definition of, 532 Stochastic integration, 722–723
solving, 537 Stochastic methods, overview of, 710, 716
Serviceability requirement, 660 clustering methods, 718, 719
Sets, 9 density clustering, 719
Shear-stress constraint, 49 mode analysis clustering, 720
Show Formulas command, 262 reduced sample points, 718
 Subject Index 943
single linkage clustering, 719 constraints, formulation of, 642–643
vector quantization, 720 data and information collection, 640
multistart method, 717 design variables, definition of, 641
stopping criterion, 717 discussion, 644
pure random search method, 717 optimization criterion, 642
Stochastic zooming method (ZOOM), 724, 731 optimum design of, 639
Stress constraints, 52 project/problem description, 640
Strong duality theorem, 227 Terminal displacement constraint, 629
Strong Wolfe conditions, 462 Terminal velocity constraint, 629
Structural optimization problems, alternative Three-bar truss, 51
formulations, 672 Three elementary row–column operations, 862
two–member frame design, alternate Time-dependent optimization problems, 674
formulation, 673 Time-dependent problems
Submatrices, 857 alternative formulations, 674
Subprogram calls, 616 digital human modeling, 675
Subroutine EQUAL, 891 mechanical/structural design problems, 674
Subroutine FUNCT, 891, 894 Total pivoting procedure, 865
Subroutine SYSEQ, 897 Trade-off, 789
Subvectors, 857 Trajectory methods, 712
Sufficient condition, 130 Transformation function, 490
concepts of, 127 Transformation of variables, 801
Sufficient-decrease condition, 461 Transportation problems, 33
Summation notation, 11, 118 Traveling salesman (TS), 758
Superscript notation, 11 problem, 748, 758
Surplus variable, 310 Tripod, 103
Swarm intelligence methods, 764 Tubular column, 44
Symmetric matrix, 857 Tunneling method, 714
Symmetric three–bar structure, 621 basic concept of, 715
Symmetric three-bar truss, minimum-weight global descent property, 715
design of, 50–54 Twice-continuous differentiability, 13
System evolution model, 5 Two-bar bracket design, 33, 34
data and information collection, 34
T definition of design variables, 35
Tableau form, 324 formulation of constraints, 37
Taguchi method, 825, 827, 828, 830 optimization criterion, 37
Tangential vector, 464 project/problem description, 33
Taylor expansion, of constraint, 844 Two-bar bracket problem, with intermediate
Taylor series, 840 variables, 38
Taylor series expansion, 822 Two-bar truss, 486
Taylor’s expansion, 117–119, 129, 464 minimization of total potential energy, 487
function of two variables, 119 Two-cut point
of Lagrange function, 213 crossover operation with, 745
linear function, 120 methods, 744
quadratic form, differentiation of, 126 Two–member frame design, 605
quadratic forms/definite matrices, 120 Two-objective optimization problem, 772
Telecommunication poles, 667
optimum design of, 664 U
constraints, formulation of, 670 Unconstrained methods, engineering applications, 484
data/information collection, 665 data interpolation, 485
design variables, definition of, 669 nonlinear equations, solution of, 488–489
optimization criterion, 669 total potential energy, minimization of, 486
project/problem description, 664 Unconstrained minimization, 489
Ten-bar cantilever truss, 700, 702 problem, 455. See also Unconstrained optimum
Tension members, 641 design, numerical methods
944 Subject Index

Unconstrained minimum, representation golden section search, 437

of, 110, 111 algorithm for 1D search by golden sections, 440
Unconstrained optimization methods, 488 initial bracketing of minimum-phase I, 437
algorithms, rate of convergence, 494 initial bracketing of minimum point, 438
convergence ratio, 495 interval of uncertainty-phase II, 439
definitions, 494 initial bracketing of minimum-phase I, 434
linear convergence, 495 process, 435
order of convergence, 494 reducing interval of uncertainty-phase II, 435
quadratic convergence, 495 Excel solver, 260
superlinear convergence, 495 worksheet and Solver Parameters dialog box, 261
conjugate Gradient method, 497 general algorithm, 426
constrained problem using, solution of, 489 general iterative algorithm, 425
direct search methods, 498 numerical methods, 423
contraction, 502 to compute step size, 432
expansion, 502 1D search methods, 433
Hooke–Jeeves algorithm, 499–500 general concepts, 432
Hooke–Jeeves methods, 499 interval-reducing methods, 434
exploratory search, 499 unimodal functions, 432
pattern search, 499 search direction determination, 445
initial simplex, 503 conjugate gradient algorithm, 446
Nelder–Mead algorithm, 503–504 convergence of, 447
Nelder–Mead simplex method, 500–501 steepest–descent method, 442–443
reflection, 501 step size determination, basic ideas, 429
shrinking operation, 503 analytical method to compute step size, 431
termination criterion, 503 1D minimization problem, 430
univariate search, 498 reduction to function, 429
Newton method, 496 subproblem, definition of, 429
quasi-Newton methods, 497 Unconstrained optimum design, numerical
BFGS quasi-newton method, 498 methods, 455
DFP method, nonquadratic case, 497 design variables, scaling of, 469
DFP method, quadratic case, 497 Newton method, search direction determination, 472
sequential unconstrained minimization classical Newton method, 472–473
techniques, 490 Marquardt modification, 478–479
augmented Lagrangian algorithm, modified Newton method, 473–477
493–494 polynomial interpolation, 456
augmented Lagrangian (multiplier) methods, alternate quadratic interpolation, 459
492–494 Armijo’s rule, inexact line search, 460–461
barrier function methods, 491–492 Goldstein test, inexact line search, 462–463
penalty function method, 491 quadratic curve fitting, 456–458
steepest-descent method, 495 Wolfe conditions, inexact line search, 462
nonquadratic case, 496 quasi-Newton methods, search direction
quadratic function, 495 determination, 479
Unconstrained optimization problems, 813 BFGS method, direct Hessian updating, 482–483
algorithms, descent direction/convergence DFP method, inverse Hessian updating, 479–481
of, 426 steepest-descent directions, orthogonality of, 468
convergence of, 428 steepest–descent method, 463
descent direction/descent step, gradient vector, properties of, 463–465
427–428 step size determination, 456
rate of convergence, 428 Unconstrained points, 109
classification of, 424 Unconstrained problem, 128
conjugate gradient methods, 448 Uniform prior distribution, 717
optimum solution, 448 Upper triangular matrix, 857
equal–interval search US–British, SI Units, 15, 16
alternate equal–interval search, 436 conversion factors, 16
 Subject Index 945
User-defined constant, 821 Weakly Pareto optimal points, 792
Utility function, 780 Weierstrass theorem, 112, 113
Utopia point, 787 Weighted global criterion method,
786–788
V Weighted min–max method, 785
Variable-interval search method, 437 additional constraints, 786
Variance represents dispersion, 819 advantages of, 786
Vector-evaluated genetic algorithm (VEGA), 782 disadvantages of, 786
Vector functions, partial derivatives of, 15 Weighted sum method, 785
Vector optimization methods, 720, 771, 780 multi-objective optimization, 785
Vector quantization multistart, 719 Weighted Tchebycheff method, 785
Vector representation, 9 Welded plate girders, 268
Vectors, 8 Wiener process, 722
norm of, 884 Wolfe conditions, 462
Vector space, 880 W–shape
Velocity responses for the flexural member, 653
at optimum with minimization of time, 638 for member, 640
Vertical column, with eccentric load, 291
Violated constraint, 515 Y
V–strings, 746 Young’s modulus, 44

W Z
Water tower support column, 98, 99 Zero reduced cost coefficient, 334
Weak duality theorem, 227 Zooming method, 712