direct multisearch for multiobjective optimization
play

Direct multisearch for multiobjective optimization A. Ismael F. Vaz - PowerPoint PPT Presentation

Mar 03, 2023 •262 likes •1.75k views

Direct multisearch for multiobjective optimization A. Ismael F. Vaz University of Minho - Portugal aivaz@dps.uminho.pt Joint work with A. L. Custdio, J. F. A. Madeira, and L. N. Vicente Southampton October 21, 2010 A.I.F. Vaz (UMinho) DMS

  1. Direct search for single objective Outline Introduction 1 Direct search 2 Direct search for single objective 3 Direct search for multiobjective 4 Numerical results 5 Conclusions and references 6 A.I.F. Vaz (UMinho) DMS October 21, 2010 8 / 64

  2. Direct search for single objective Derivative-free optimization Problem formulation (single objective) min x ∈ Ω f ( x ) where Ω = { x ∈ R n : ℓ ≤ x ≤ u } f : R n → R ∪ { + ∞} , ℓ ∈ ( R ∪ {−∞} ) n and u ∈ ( R ∪ { + ∞} ) n We aim at solving this problem without using derivatives of f . A.I.F. Vaz (UMinho) DMS October 21, 2010 9 / 64

  3. Direct search for single objective Some definitions Positive spanning set Is a set of vectors that spans R n with nonnegative coefficients. Examples D ⊕ = { e 1 , . . . , e n , − e 1 , . . . , − e n } D ⊗ = { e 1 , . . . , e n , − e 1 , . . . , − e n , e, − e } Extreme barrier function � f ( x ) if x ∈ Ω , f Ω ( x ) = + ∞ otherwise. A.I.F. Vaz (UMinho) DMS October 21, 2010 10 / 64

  4. Direct search for single objective Some definitions Positive spanning set Is a set of vectors that spans R n with nonnegative coefficients. Examples D ⊕ = { e 1 , . . . , e n , − e 1 , . . . , − e n } D ⊗ = { e 1 , . . . , e n , − e 1 , . . . , − e n , e, − e } Extreme barrier function � f ( x ) if x ∈ Ω , f Ω ( x ) = + ∞ otherwise. A.I.F. Vaz (UMinho) DMS October 21, 2010 10 / 64

  5. Direct search for single objective Some definitions Positive spanning set Is a set of vectors that spans R n with nonnegative coefficients. Examples D ⊕ = { e 1 , . . . , e n , − e 1 , . . . , − e n } D ⊗ = { e 1 , . . . , e n , − e 1 , . . . , − e n , e, − e } Extreme barrier function � f ( x ) if x ∈ Ω , f Ω ( x ) = + ∞ otherwise. A.I.F. Vaz (UMinho) DMS October 21, 2010 10 / 64

  6. Direct search for single objective A direct-search method (0) Initialization Choose x 0 ∈ Ω , α 0 > 0 . Let D be a (possibly infinite) set of positive spanning sets. For k = 0 , 1 , 2 , . . . (1) Search step (Optional) Try to compute a point x , using a finite number of trial points, in the grid � x k + α k D k z, z ∈ N | D k | � M k = 0 with D k ⊆ D and f Ω ( x ) < f ( x k ) . If f Ω ( x ) < f ( x k ) then set x k +1 = x , declare the iteration and the search step successful, and skip the poll step. A.I.F. Vaz (UMinho) DMS October 21, 2010 11 / 64

  7. Direct search for single objective A direct-search method (0) Initialization Choose x 0 ∈ Ω , α 0 > 0 . Let D be a (possibly infinite) set of positive spanning sets. For k = 0 , 1 , 2 , . . . (1) Search step (Optional) Try to compute a point x , using a finite number of trial points, in the grid � x k + α k D k z, z ∈ N | D k | � M k = 0 with D k ⊆ D and f Ω ( x ) < f ( x k ) . If f Ω ( x ) < f ( x k ) then set x k +1 = x , declare the iteration and the search step successful, and skip the poll step. A.I.F. Vaz (UMinho) DMS October 21, 2010 11 / 64

  8. Direct search for single objective A direct-search method (0) Initialization Choose x 0 ∈ Ω , α 0 > 0 . Let D be a (possibly infinite) set of positive spanning sets. For k = 0 , 1 , 2 , . . . (1) Search step (Optional) Try to compute a point x , using a finite number of trial points, in the grid � x k + α k D k z, z ∈ N | D k | � M k = 0 with D k ⊆ D and f Ω ( x ) < f ( x k ) . If f Ω ( x ) < f ( x k ) then set x k +1 = x , declare the iteration and the search step successful, and skip the poll step. A.I.F. Vaz (UMinho) DMS October 21, 2010 11 / 64

  9. Direct search for single objective A direct-search method (0) Initialization Choose x 0 ∈ Ω , α 0 > 0 . Let D be a (possibly infinite) set of positive spanning sets. For k = 0 , 1 , 2 , . . . (1) Search step (Optional) Try to compute a point x , using a finite number of trial points, in the grid � x k + α k D k z, z ∈ N | D k | � M k = 0 with D k ⊆ D and f Ω ( x ) < f ( x k ) . If f Ω ( x ) < f ( x k ) then set x k +1 = x , declare the iteration and the search step successful, and skip the poll step. A.I.F. Vaz (UMinho) DMS October 21, 2010 11 / 64

  10. Direct search for single objective A direct-search method (0) Initialization Choose x 0 ∈ Ω , α 0 > 0 . Let D be a (possibly infinite) set of positive spanning sets. For k = 0 , 1 , 2 , . . . (1) Search step (Optional) Try to compute a point x , using a finite number of trial points, in the grid � x k + α k D k z, z ∈ N | D k | � M k = 0 with D k ⊆ D and f Ω ( x ) < f ( x k ) . If f Ω ( x ) < f ( x k ) then set x k +1 = x , declare the iteration and the search step successful, and skip the poll step. A.I.F. Vaz (UMinho) DMS October 21, 2010 11 / 64

  11. Direct search for single objective A direct-search method (2) Poll step Optionally order the poll set P k = { x k + α k d, d ∈ D k } with D k ⊆ D . If a poll point x k + α k d k is found such that f Ω ( x k + α k d k ) < f ( x k ) then stop polling, set x k +1 = x k + α k d k , and declare the iteration and the poll step successful. Otherwise declare the iteration (and the poll step) unsuccessful and set x k +1 = x k . A.I.F. Vaz (UMinho) DMS October 21, 2010 12 / 64

  12. Direct search for single objective A direct-search method (2) Poll step Optionally order the poll set P k = { x k + α k d, d ∈ D k } with D k ⊆ D . If a poll point x k + α k d k is found such that f Ω ( x k + α k d k ) < f ( x k ) then stop polling, set x k +1 = x k + α k d k , and declare the iteration and the poll step successful. Otherwise declare the iteration (and the poll step) unsuccessful and set x k +1 = x k . A.I.F. Vaz (UMinho) DMS October 21, 2010 12 / 64

  13. Direct search for single objective A direct-search method (2) Poll step Optionally order the poll set P k = { x k + α k d, d ∈ D k } with D k ⊆ D . If a poll point x k + α k d k is found such that f Ω ( x k + α k d k ) < f ( x k ) then stop polling, set x k +1 = x k + α k d k , and declare the iteration and the poll step successful. Otherwise declare the iteration (and the poll step) unsuccessful and set x k +1 = x k . A.I.F. Vaz (UMinho) DMS October 21, 2010 12 / 64

  14. Direct search for single objective A direct-search method (3) Step size update: If the iteration was successful then maintain the step size parameter ( α k +1 = α k ) or double it ( α k +1 = 2 α k ) after two consecutive poll successes along the same direction. If the iteration was unsuccessful, halve the step size parameter ( α k +1 = α k / 2 ). A.I.F. Vaz (UMinho) DMS October 21, 2010 13 / 64

  15. Direct search for single objective A direct-search method (3) Step size update: If the iteration was successful then maintain the step size parameter ( α k +1 = α k ) or double it ( α k +1 = 2 α k ) after two consecutive poll successes along the same direction. If the iteration was unsuccessful, halve the step size parameter ( α k +1 = α k / 2 ). A.I.F. Vaz (UMinho) DMS October 21, 2010 13 / 64

  16. Direct search for single objective Some comments We could present the previous algorithm in a different form, namely by fixing the set D k ( D k = D, ∀ k ) not to change with the iteration number (problem with only bound constraints). allowing the set D k to be computed in a way to conform with possible linear constraints. to use a forcing function ρ ( · ) ( e.g. , ρ ( t ) = t 2 ) instead of a integer lattice (the mesh M k ). A forcing function ρ ( · ) is continuous, positive, and satisfies lim t − → 0 + ρ ( t ) /t = 0 and ρ ( t 1 ) ≤ ρ ( t 2 ) if t 1 < t 2 . A point x is accepted (successful) in the search step if f Ω ( x ) < f ( x k ) − ρ ( α k ) . A.I.F. Vaz (UMinho) DMS October 21, 2010 14 / 64

  17. Direct search for single objective Some comments We could present the previous algorithm in a different form, namely by fixing the set D k ( D k = D, ∀ k ) not to change with the iteration number (problem with only bound constraints). allowing the set D k to be computed in a way to conform with possible linear constraints. to use a forcing function ρ ( · ) ( e.g. , ρ ( t ) = t 2 ) instead of a integer lattice (the mesh M k ). A forcing function ρ ( · ) is continuous, positive, and satisfies lim t − → 0 + ρ ( t ) /t = 0 and ρ ( t 1 ) ≤ ρ ( t 2 ) if t 1 < t 2 . A point x is accepted (successful) in the search step if f Ω ( x ) < f ( x k ) − ρ ( α k ) . A.I.F. Vaz (UMinho) DMS October 21, 2010 14 / 64

  18. Direct search for single objective Some comments We could present the previous algorithm in a different form, namely by fixing the set D k ( D k = D, ∀ k ) not to change with the iteration number (problem with only bound constraints). allowing the set D k to be computed in a way to conform with possible linear constraints. to use a forcing function ρ ( · ) ( e.g. , ρ ( t ) = t 2 ) instead of a integer lattice (the mesh M k ). A forcing function ρ ( · ) is continuous, positive, and satisfies lim t − → 0 + ρ ( t ) /t = 0 and ρ ( t 1 ) ≤ ρ ( t 2 ) if t 1 < t 2 . A point x is accepted (successful) in the search step if f Ω ( x ) < f ( x k ) − ρ ( α k ) . A.I.F. Vaz (UMinho) DMS October 21, 2010 14 / 64

  19. Direct search for single objective Some comments We could present the previous algorithm in a different form, namely by fixing the set D k ( D k = D, ∀ k ) not to change with the iteration number (problem with only bound constraints). allowing the set D k to be computed in a way to conform with possible linear constraints. to use a forcing function ρ ( · ) ( e.g. , ρ ( t ) = t 2 ) instead of a integer lattice (the mesh M k ). A forcing function ρ ( · ) is continuous, positive, and satisfies lim t − → 0 + ρ ( t ) /t = 0 and ρ ( t 1 ) ≤ ρ ( t 2 ) if t 1 < t 2 . A point x is accepted (successful) in the search step if f Ω ( x ) < f ( x k ) − ρ ( α k ) . The presented algorithm just suits for the multiobjective version to be described. A.I.F. Vaz (UMinho) DMS October 21, 2010 14 / 64

  20. Direct search for multiobjective Outline Introduction 1 Direct search 2 Direct search for single objective 3 Direct search for multiobjective 4 Numerical results 5 Conclusions and references 6 A.I.F. Vaz (UMinho) DMS October 21, 2010 15 / 64

  21. Direct search for multiobjective Derivative-free multiobjective optimization MOO problem x ∈ Ω F ( x ) ≡ ( f 1 ( x ) , f 2 ( x ) , . . . , f m ( x )) ⊤ min where Ω = { x ∈ R n : ℓ ≤ x ≤ u } f j : R n → R ∪ { + ∞} , j = 1 , . . . , m , ℓ ∈ ( R ∪ {−∞} ) n and u ∈ ( R ∪ { + ∞} ) n Several objectives, often conflicting. Functions with unknown derivatives. Expensive function evaluations, possibly subject to noise. Impractical to compute approximations to derivatives. A.I.F. Vaz (UMinho) DMS October 21, 2010 16 / 64

  22. Direct search for multiobjective Derivative-free multiobjective optimization MOO problem x ∈ Ω F ( x ) ≡ ( f 1 ( x ) , f 2 ( x ) , . . . , f m ( x )) ⊤ min where Ω = { x ∈ R n : ℓ ≤ x ≤ u } f j : R n → R ∪ { + ∞} , j = 1 , . . . , m , ℓ ∈ ( R ∪ {−∞} ) n and u ∈ ( R ∪ { + ∞} ) n Several objectives, often conflicting. Functions with unknown derivatives. Expensive function evaluations, possibly subject to noise. Impractical to compute approximations to derivatives. A.I.F. Vaz (UMinho) DMS October 21, 2010 16 / 64

  23. Direct search for multiobjective Derivative-free multiobjective optimization MOO problem x ∈ Ω F ( x ) ≡ ( f 1 ( x ) , f 2 ( x ) , . . . , f m ( x )) ⊤ min where Ω = { x ∈ R n : ℓ ≤ x ≤ u } f j : R n → R ∪ { + ∞} , j = 1 , . . . , m , ℓ ∈ ( R ∪ {−∞} ) n and u ∈ ( R ∪ { + ∞} ) n Several objectives, often conflicting. Functions with unknown derivatives. Expensive function evaluations, possibly subject to noise. Impractical to compute approximations to derivatives. A.I.F. Vaz (UMinho) DMS October 21, 2010 16 / 64

  24. Direct search for multiobjective Derivative-free multiobjective optimization MOO problem x ∈ Ω F ( x ) ≡ ( f 1 ( x ) , f 2 ( x ) , . . . , f m ( x )) ⊤ min where Ω = { x ∈ R n : ℓ ≤ x ≤ u } f j : R n → R ∪ { + ∞} , j = 1 , . . . , m , ℓ ∈ ( R ∪ {−∞} ) n and u ∈ ( R ∪ { + ∞} ) n Several objectives, often conflicting. Functions with unknown derivatives. Expensive function evaluations, possibly subject to noise. Impractical to compute approximations to derivatives. A.I.F. Vaz (UMinho) DMS October 21, 2010 16 / 64

  25. Direct search for multiobjective DMS algorithmic main lines Does not aggregate any of the objective functions. Generalizes ALL direct-search methods of directional type to MOO. Makes use of search/poll paradigm. Implements an optional search step (only to disseminate the search). Tries to capture the whole Pareto front from the polling procedure. A.I.F. Vaz (UMinho) DMS October 21, 2010 17 / 64

  26. Direct search for multiobjective DMS algorithmic main lines Does not aggregate any of the objective functions. Generalizes ALL direct-search methods of directional type to MOO. Makes use of search/poll paradigm. Implements an optional search step (only to disseminate the search). Tries to capture the whole Pareto front from the polling procedure. A.I.F. Vaz (UMinho) DMS October 21, 2010 17 / 64

  27. Direct search for multiobjective DMS algorithmic main lines Does not aggregate any of the objective functions. Generalizes ALL direct-search methods of directional type to MOO. Makes use of search/poll paradigm. Implements an optional search step (only to disseminate the search). Tries to capture the whole Pareto front from the polling procedure. A.I.F. Vaz (UMinho) DMS October 21, 2010 17 / 64

  28. Direct search for multiobjective DMS algorithmic main lines Does not aggregate any of the objective functions. Generalizes ALL direct-search methods of directional type to MOO. Makes use of search/poll paradigm. Implements an optional search step (only to disseminate the search). Tries to capture the whole Pareto front from the polling procedure. A.I.F. Vaz (UMinho) DMS October 21, 2010 17 / 64

  29. Direct search for multiobjective DMS algorithmic main lines Does not aggregate any of the objective functions. Generalizes ALL direct-search methods of directional type to MOO. Makes use of search/poll paradigm. Implements an optional search step (only to disseminate the search). Tries to capture the whole Pareto front from the polling procedure. A.I.F. Vaz (UMinho) DMS October 21, 2010 17 / 64

  30. Direct search for multiobjective DMS algorithmic main lines Keeps a list of feasible nondominated points. Poll centers are chosen from the list. Successful iterations correspond to list changes. A.I.F. Vaz (UMinho) DMS October 21, 2010 18 / 64

  31. Direct search for multiobjective DMS algorithmic main lines Keeps a list of feasible nondominated points. Poll centers are chosen from the list. Successful iterations correspond to list changes. A.I.F. Vaz (UMinho) DMS October 21, 2010 18 / 64

  32. Direct search for multiobjective DMS algorithmic main lines Keeps a list of feasible nondominated points. Poll centers are chosen from the list. Successful iterations correspond to list changes. A.I.F. Vaz (UMinho) DMS October 21, 2010 18 / 64

  33. Direct search for multiobjective DMS example A.I.F. Vaz (UMinho) DMS October 21, 2010 19 / 64

  34. Direct search for multiobjective DMS example A.I.F. Vaz (UMinho) DMS October 21, 2010 20 / 64

  35. Direct search for multiobjective DMS example A.I.F. Vaz (UMinho) DMS October 21, 2010 21 / 64

  36. Direct search for multiobjective DMS example A.I.F. Vaz (UMinho) DMS October 21, 2010 22 / 64

  37. Direct search for multiobjective DMS example A.I.F. Vaz (UMinho) DMS October 21, 2010 23 / 64

  38. Direct search for multiobjective DMS example A.I.F. Vaz (UMinho) DMS October 21, 2010 24 / 64

  39. Direct search for multiobjective DMS example A.I.F. Vaz (UMinho) DMS October 21, 2010 25 / 64

  40. Direct search for multiobjective DMS example A.I.F. Vaz (UMinho) DMS October 21, 2010 26 / 64

  41. Direct search for multiobjective DMS example A.I.F. Vaz (UMinho) DMS October 21, 2010 27 / 64

  42. Direct search for multiobjective DMS search & poll steps Evaluate a finite set of feasible points ֒ → L add . Remove dominated points from L k ∪ L add ֒ → L filtered . Select list of feasible nondominated points ֒ → L trial . Compare L trial to L k (success if L trial � = L k , unsuccess otherwise). A.I.F. Vaz (UMinho) DMS October 21, 2010 28 / 64

  43. Direct search for multiobjective DMS search & poll steps Evaluate a finite set of feasible points ֒ → L add . Remove dominated points from L k ∪ L add ֒ → L filtered . Select list of feasible nondominated points ֒ → L trial . Compare L trial to L k (success if L trial � = L k , unsuccess otherwise). A.I.F. Vaz (UMinho) DMS October 21, 2010 28 / 64

  44. Direct search for multiobjective DMS search & poll steps Evaluate a finite set of feasible points ֒ → L add . Remove dominated points from L k ∪ L add ֒ → L filtered . Select list of feasible nondominated points ֒ → L trial . Compare L trial to L k (success if L trial � = L k , unsuccess otherwise). A.I.F. Vaz (UMinho) DMS October 21, 2010 28 / 64

  45. Direct search for multiobjective DMS search & poll steps Evaluate a finite set of feasible points ֒ → L add . Remove dominated points from L k ∪ L add ֒ → L filtered . Select list of feasible nondominated points ֒ → L trial . Compare L trial to L k (success if L trial � = L k , unsuccess otherwise). A.I.F. Vaz (UMinho) DMS October 21, 2010 28 / 64

  46. Direct search for multiobjective Direct MultiSearch for MOO (0) Initialization Choose x 0 ∈ Ω with F ( x 0 ) < + ∞ , α 0 > 0 . Set L 0 = { ( x 0 ; α 0 ) } . Let D be a (possibly infinite) set of positive spanning sets. For k = 0 , 1 , 2 , . . . (1) Selection of iterate point Order L k and select ( x k ; α k ) ∈ L k . A.I.F. Vaz (UMinho) DMS October 21, 2010 29 / 64

  47. Direct search for multiobjective Direct MultiSearch for MOO (0) Initialization Choose x 0 ∈ Ω with F ( x 0 ) < + ∞ , α 0 > 0 . Set L 0 = { ( x 0 ; α 0 ) } . Let D be a (possibly infinite) set of positive spanning sets. For k = 0 , 1 , 2 , . . . (1) Selection of iterate point Order L k and select ( x k ; α k ) ∈ L k . A.I.F. Vaz (UMinho) DMS October 21, 2010 29 / 64

  48. Direct search for multiobjective Direct MultiSearch for MOO (0) Initialization Choose x 0 ∈ Ω with F ( x 0 ) < + ∞ , α 0 > 0 . Set L 0 = { ( x 0 ; α 0 ) } . Let D be a (possibly infinite) set of positive spanning sets. For k = 0 , 1 , 2 , . . . (1) Selection of iterate point Order L k and select ( x k ; α k ) ∈ L k . A.I.F. Vaz (UMinho) DMS October 21, 2010 29 / 64

  49. Direct search for multiobjective Direct MultiSearch for MOO (0) Initialization Choose x 0 ∈ Ω with F ( x 0 ) < + ∞ , α 0 > 0 . Set L 0 = { ( x 0 ; α 0 ) } . Let D be a (possibly infinite) set of positive spanning sets. For k = 0 , 1 , 2 , . . . (1) Selection of iterate point Order L k and select ( x k ; α k ) ∈ L k . A.I.F. Vaz (UMinho) DMS October 21, 2010 29 / 64

  50. Direct search for multiobjective Direct MultiSearch for MOO (2) Search step (Optional) Evaluate a finite set of points L add = { ( z s ; α k ) } s ∈ S (in the mesh or using a forcing function). ( L k ; L add ) ֒ → L filtered ֒ → L trial If success is achieved then set L k +1 = L trial , declare the iteration and the search step successful, and skip the poll step. A.I.F. Vaz (UMinho) DMS October 21, 2010 30 / 64

  51. Direct search for multiobjective Direct MultiSearch for MOO (2) Search step (Optional) Evaluate a finite set of points L add = { ( z s ; α k ) } s ∈ S (in the mesh or using a forcing function). ( L k ; L add ) ֒ → L filtered ֒ → L trial If success is achieved then set L k +1 = L trial , declare the iteration and the search step successful, and skip the poll step. A.I.F. Vaz (UMinho) DMS October 21, 2010 30 / 64

  52. Direct search for multiobjective Direct MultiSearch for MOO (3) Poll step Evaluate L add = { ( x k + α k d ; α k ) , d ∈ D k }, with D k ⊆ D ( L k ; L add ) ֒ → L filtered ֒ → L trial If success is achieved then set L k +1 = L trial , declare the iteration and the poll step successful Otherwise declare the iteration (and the poll step) unsuccessful and set L k +1 = L trial A.I.F. Vaz (UMinho) DMS October 21, 2010 31 / 64

  53. Direct search for multiobjective Direct MultiSearch for MOO (3) Poll step Evaluate L add = { ( x k + α k d ; α k ) , d ∈ D k }, with D k ⊆ D ( L k ; L add ) ֒ → L filtered ֒ → L trial If success is achieved then set L k +1 = L trial , declare the iteration and the poll step successful Otherwise declare the iteration (and the poll step) unsuccessful and set L k +1 = L trial A.I.F. Vaz (UMinho) DMS October 21, 2010 31 / 64

  54. Direct search for multiobjective Direct MultiSearch for MOO (3) Poll step Evaluate L add = { ( x k + α k d ; α k ) , d ∈ D k }, with D k ⊆ D ( L k ; L add ) ֒ → L filtered ֒ → L trial If success is achieved then set L k +1 = L trial , declare the iteration and the poll step successful Otherwise declare the iteration (and the poll step) unsuccessful and set L k +1 = L trial A.I.F. Vaz (UMinho) DMS October 21, 2010 31 / 64

  55. Direct search for multiobjective Direct MultiSearch for MOO (4) Step size update: If the iteration was successful then maintain the step size parameter ( α k +1 = α k ) or double it ( α k +1 = 2 α k ) after two consecutive poll successes along the same direction. If the iteration was unsuccessful, halve the step size parameter ( α k +1 = α k / 2 ). A.I.F. Vaz (UMinho) DMS October 21, 2010 32 / 64

  56. Direct search for multiobjective Direct MultiSearch for MOO (4) Step size update: If the iteration was successful then maintain the step size parameter ( α k +1 = α k ) or double it ( α k +1 = 2 α k ) after two consecutive poll successes along the same direction. If the iteration was unsuccessful, halve the step size parameter ( α k +1 = α k / 2 ). A.I.F. Vaz (UMinho) DMS October 21, 2010 32 / 64

  57. Direct search for multiobjective Numerical Example — Problem SP1 [Huband et al.] � Evaluated points since beginning. � Current iterate list. A.I.F. Vaz (UMinho) DMS October 21, 2010 33 / 64

  58. Direct search for multiobjective Numerical example — problem SP1 [Huband et al.] � Evaluated poll points. � Evaluated points since beginning. A.I.F. Vaz (UMinho) DMS October 21, 2010 34 / 64

  59. Direct search for multiobjective Numerical example — problem SP1 [Huband et al.] • Nondominated evaluated poll points. A.I.F. Vaz (UMinho) DMS October 21, 2010 35 / 64

  60. Direct search for multiobjective Numerical example — problem SP1 [Huband et al.] � Evaluated poll points. � Evaluated points since beginning. � Current iterate list. A.I.F. Vaz (UMinho) DMS October 21, 2010 36 / 64

  61. Direct search for multiobjective Numerical example — problem SP1 [Huband et al.] � Evaluated poll points. � Evaluated points since beginning. A.I.F. Vaz (UMinho) DMS October 21, 2010 37 / 64

  62. Direct search for multiobjective Numerical example — problem SP1 [Huband et al.] • Nondominated evaluated poll points. A.I.F. Vaz (UMinho) DMS October 21, 2010 38 / 64

  63. Direct search for multiobjective Numerical example — problem SP1 [Huband et al.] � Evaluated poll points. � Evaluated points since beginning. � Current iterate list. A.I.F. Vaz (UMinho) DMS October 21, 2010 39 / 64

  64. Direct search for multiobjective Numerical example — problem SP1 [Huband et al.] � Evaluated poll points. � Evaluated points since beginning. � Current iterate list. A.I.F. Vaz (UMinho) DMS October 21, 2010 40 / 64

  65. Direct search for multiobjective Numerical example — problem SP1 [Huband et al.] � Evaluated poll points. � Evaluated points since beginning. � Current iterate list. A.I.F. Vaz (UMinho) DMS October 21, 2010 41 / 64

  66. Direct search for multiobjective Numerical example — problem SP1 [Huband et al.] � Evaluated poll points. � Evaluated points since beginning. � Current iterate list. A.I.F. Vaz (UMinho) DMS October 21, 2010 42 / 64

  67. Direct search for multiobjective Numerical example — problem SP1 [Huband et al.] � Evaluated poll points. � Evaluated points since beginning. � Current iterate list. A.I.F. Vaz (UMinho) DMS October 21, 2010 43 / 64

Recommend

Direct Multisearch for Multiobjective Optimization Ana Lusa Custdio 1 Jos F. Aguilar Madeira

Direct Multisearch for Multiobjective Optimization Ana Lusa Custdio 1 Jos F. Aguilar Madeira

Direct Multisearch for Multiobjective Optimization Ana Lusa Custdio 1 Jos F. Aguilar Madeira 2 A. Ismael F. Vaz 3 Lus Nunes Vicente 4 2 IDMEC-IST, ISEL 1 Universidade Nova de Lisboa 3 Universidade do Minho 4 Universidade de Coimbra

966 views • 93 slides
Heuristic Algorithms for Multiobjective Combinatorial Optimization Adapted from a tutorial by

Heuristic Algorithms for Multiobjective Combinatorial Optimization Adapted from a tutorial by

[center] HEURISTIC OPTIMIZATION Heuristic Algorithms for Multiobjective Combinatorial Optimization Adapted from a tutorial by Lu s Paquete 1 / 36 Introduction Multiobjective Combinatorial Optimization Problems (MCOPs) I Many real-life

314 views • 18 slides
Multiobjective Optimization Performance Measures in Many-Objective Optimization Facult: Eduardo

Multiobjective Optimization Performance Measures in Many-Objective Optimization Facult: Eduardo

Multiobjective Optimization Performance Measures in Many-Objective Optimization Facult: Eduardo G. Carrano Frederico G. Guimares Lucas S. Batista {egcarrano,fredericoguimaraes,lusoba}@ufmg.br www.ppgee.ufmg.br/ lusoba Universidade

229 views • 12 slides
Reduced-Order Approaches for PDE-Constrained Multiobjective Optimization joint work with U

Reduced-Order Approaches for PDE-Constrained Multiobjective Optimization joint work with U

Reduced-Order Approaches for PDE-Constrained Multiobjective Optimization joint work with U Dubrovnik / TU Eindhoven / U Erlangen / TU Mnchen / U Paderborn Stefan Volkwein University of Konstanz, Chair Numerical Optimization New Trends in

284 views • 12 slides
Multiobjective Multiobjective Genetic Algorithms for Genetic Algorithms for Multiscaling

Multiobjective Multiobjective Genetic Algorithms for Genetic Algorithms for Multiscaling

Multiobjective Multiobjective Genetic Algorithms for Genetic Algorithms for Multiscaling Multiscaling Excited-State Direct Excited-State Direct Dynamics in Photochemistry Dynamics in Photochemistry Kumara Sastry 1 , D.D. Johnson 2 , A. L.

235 views • 11 slides
Multiobjective Parameter Optimization of Elliptic PDEs using the Reduced Basis Method S.

Multiobjective Parameter Optimization of Elliptic PDEs using the Reduced Basis Method S.

Multiobjective Parameter Optimization of Elliptic PDEs using the Reduced Basis Method S. Banholzer, B. Gebken, M. Dellnitz, S. Peitz, S. Volkwein University of Konstanz / Paderborn University Workshop: New trends in PDE constrained optimization

652 views • 61 slides
Budgeted Bayesian Multiobjective Optimization David Gaudrie 1 , Rodolphe Le Riche 2 , Victor

Budgeted Bayesian Multiobjective Optimization David Gaudrie 1 , Rodolphe Le Riche 2 , Victor

Introduction Preferences Budgeted optimization Batch criteria Budgeted Bayesian Multiobjective Optimization David Gaudrie 1 , Rodolphe Le Riche 2 , Victor Picheny 3 , t Enaux 1 , Vincent Herbert 1 Beno 1 Groupe PSA 2 CNRS LIMOS,

963 views • 84 slides
Unit 2: Problem Classification and Difficulty in Optimization Learning goals Unit 2 I. What

Unit 2: Problem Classification and Difficulty in Optimization Learning goals Unit 2 I. What

Unit 2: Problem Classification and Difficulty in Optimization Learning goals Unit 2 I. What is the subject area of multiobjective decision analysis and multiobjective optimization; How does it relate to the more general field of systems

606 views • 46 slides
A Short Tutorial on Evolutionary Multiobjective Optimization Carlos A. Coello Coello

A Short Tutorial on Evolutionary Multiobjective Optimization Carlos A. Coello Coello

A Short Tutorial on Evolutionary Multiobjective Optimization Carlos A. Coello Coello CINVESTAV-IPN Depto. de Ingenier a El ectrica Secci on de Computaci on Av. Instituto Polit ecnico Nacional No. 2508 Col. San Pedro Zacatenco

1.15k views • 77 slides
Evolutionary Multiobjective Optimization: Current and Future Challenges Carlos A. Coello Coello

Evolutionary Multiobjective Optimization: Current and Future Challenges Carlos A. Coello Coello

Evolutionary Multiobjective Optimization: Current and Future Challenges Carlos A. Coello Coello CINVESTAV-IPN Depto. de Ingenier a El ectrica Secci on de Computaci on Av. Instituto Polit ecnico Nacional No. 2508 Col. San

642 views • 35 slides
Event-Triggered Interactive Gradient Descent for Real-Time Multiobjective Optimization Pio Ong and

Event-Triggered Interactive Gradient Descent for Real-Time Multiobjective Optimization Pio Ong and

Event-Triggered Interactive Gradient Descent for Real-Time Multiobjective Optimization Pio Ong and Jorge Cort es Mechanical and Aerospace Engineering University of California, San Diego http://carmenere.ucsd.edu/jorge 56th IEEE Conference on

732 views • 40 slides
MULTIOBJECTIVE OPTIMIZATION OF AUTOMOTIVE VEHICLE GAGE PANEL Anatolijs Melnikovs*

MULTIOBJECTIVE OPTIMIZATION OF AUTOMOTIVE VEHICLE GAGE PANEL Anatolijs Melnikovs*

International Conference on Vibration Problems 2011, September 5-8, 2011, Prague, Czech Republic MULTIOBJECTIVE OPTIMIZATION OF AUTOMOTIVE VEHICLE GAGE PANEL Anatolijs Melnikovs* anatolijs.melnikovs@inbox.lv Alexander Janushevskis* Janis

576 views • 28 slides
fadsdfasadfs Ordered Weighted Average Optimization in Multiobjective Spanning Tree Problems andez

fadsdfasadfs Ordered Weighted Average Optimization in Multiobjective Spanning Tree Problems andez

fadsdfasadfs Ordered Weighted Average Optimization in Multiobjective Spanning Tree Problems andez 1 Elena Fern Miguel Pozo 2 Justo Puerto 2 1 Universitat Polit` ecnica de Catalunya - BarcelonaTech 2 Universidad de Sevilla The Ordered Weighted

630 views • 45 slides
Direct Optimization CSC2547 Adamo Young, Dami Choi, Sepehr Abbasi Zadeh Direct Optimization

Direct Optimization CSC2547 Adamo Young, Dami Choi, Sepehr Abbasi Zadeh Direct Optimization

Direct Optimization CSC2547 Adamo Young, Dami Choi, Sepehr Abbasi Zadeh Direct Optimization A way to obtain gradient estimates that directly optimizes a non-differentiable objective. It has first appeared in structured prediction

1.2k views • 106 slides
Recent Results and Open Problems in Evolutionary Multiobjective Optimization Carlos A. Coello

Recent Results and Open Problems in Evolutionary Multiobjective Optimization Carlos A. Coello

Contents Introduction Recent Results and Open Problems The Challenges of this Century Conclusions Recent Results and Open Problems in Evolutionary Multiobjective Optimization Carlos A. Coello Coello CINVESTAV-IPN Evolutionary Computation

704 views • 67 slides
Balancing Performance and Efficiency in a Robotic Fish with Evolutionary Multiobjective

Balancing Performance and Efficiency in a Robotic Fish with Evolutionary Multiobjective

Balancing Performance and Efficiency in a Robotic Fish with Evolutionary Multiobjective Optimization Anthony J. Clark, Jianxun Wang, Xiaobo Tan, and Philip K. McKinley Michigan State University, East Lansing, MI, USA Anthony Clark -- IEEE ICES

709 views • 23 slides
Getting Lost or Getting Trapped: On the Effect of Moves to Incomparable Points in Multiobjective

Getting Lost or Getting Trapped: On the Effect of Moves to Incomparable Points in Multiobjective

Introduction Analysis and Tempering Summary/Conclusions/What next? Getting Lost or Getting Trapped: On the Effect of Moves to Incomparable Points in Multiobjective Hillclimbing Workshop on Theoretical Aspects of Evolutionary Multiobjective

661 views • 35 slides
Optimization-Based Control: Direct Collocation Methods for Trajectory and Policy Optimization CS

Optimization-Based Control: Direct Collocation Methods for Trajectory and Policy Optimization CS

Optimization-Based Control: Direct Collocation Methods for Trajectory and Policy Optimization CS 287: Advanced Robotics, Fall 2019 Guest Lecture Igor Mordatch Overview Previously: Locally optimal control (shooting vs. collocation)

1.54k views • 138 slides
4. Fitness landscape analysis for multi-objective optimization problems Fitness landscape

4. Fitness landscape analysis for multi-objective optimization problems Fitness landscape

Multiobjective Optimization Optimization algorithms Features in MO Prediction Set-based FL 4. Fitness landscape analysis for multi-objective optimization problems Fitness landscape analysis for understanding and designing local search

713 views • 58 slides
Metaheuristics 4.3 Main design issues of multiobjective metaheuristics 4.4 Fitness assignment

Metaheuristics 4.3 Main design issues of multiobjective metaheuristics 4.4 Fitness assignment

Metaheuristics 4.3 Main design issues of multiobjective metaheuristics 4.4 Fitness assignment strategies Adrian Horga 4.3 Main design issues of multiobjective metaheuristics 4.3 Issues Algorithms for solving MOPs Exact Useful for

458 views • 21 slides
A Novel Multiobjective Framework for Cell Switch-Off in Dense Cellular Networks alez G 1 , Halim

A Novel Multiobjective Framework for Cell Switch-Off in Dense Cellular Networks alez G 1 , Halim

A Novel Multiobjective Framework for Cell Switch-Off in Dense Cellular Networks alez G 1 , Halim Yanikomeroglu 2 , David Gonz a-Lozano 1 , Silvia Ruiz Boqu e 1 Mario Garc 1 Departament of Signal Theory and Communications Universitat

511 views • 37 slides
Multiobjective planning for farms, using the Dominance-based Rough Set Approach Luisa Paolotti

Multiobjective planning for farms, using the Dominance-based Rough Set Approach Luisa Paolotti

Multiobjective planning for farms, using the Dominance-based Rough Set Approach Luisa Paolotti PhD in Agri-food Economics and Politics Faculty of Agriculture, University of Perugia - Italy 2010 COST IC0602 International Doctoral School

536 views • 29 slides
Otimizao Multiobjective Evolutionary Algorithm based on Multiobjetivo Decomposition

Otimizao Multiobjective Evolutionary Algorithm based on Multiobjetivo Decomposition

Prof. Lucas de Souza Batista - DEE/EE/UFMG Otimizao Multiobjective Evolutionary Algorithm based on Multiobjetivo Decomposition (MOEA/D) Introduo O MOEA/D decompe um problema de otimizao multiobjetivo em N subproblemas

673 views • 20 slides
Population-Based Incremental Learning for Multiobjective Optimisation Sujin Bureerat and Krit

Population-Based Incremental Learning for Multiobjective Optimisation Sujin Bureerat and Krit

Population-Based Incremental Learning for Multiobjective Optimisation Sujin Bureerat and Krit Sriworamas Department of Mechanical Engineering, Faculty of Engineering, Khon Kaen University, Thailand, 40002 Sujbur@kku.ac.th Outlines

562 views • 38 slides

More recommend