richardson extrapolation and higher order qmc
play

Richardson extrapolation and higher order QMC Takashi Goda School - PowerPoint PPT Presentation

Jun 22, 2023 •419 likes •1.83k views

Richardson extrapolation and higher order QMC Takashi Goda School of Engineering, University of Tokyo MCM 2019 Partly joint with Josef Dick and Takehito Yoshiki Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 1 / 61 Outline QMC

  1. Explicit construction For s = 1, C 1 can be the identity matrix, which generates the famous digital (0 , 1)-sequence called van der Corput sequence . For s = 2, C 1 and C 2 can be     1 0 · · · 0 0 · · · 0 1 0 1 · · · 0 0 · · · 1 0         C 1 =  , C 2 =  , . . . . . . ... ...     . . . . . . . . . . . .   0 0 · · · 1 1 · · · 0 0 which generates a digital (0 , m , 2)-net, known as the Hammersley point set. Not extensible in m . Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 15 / 61

  2. Explicit construction Figure: Hammersley point set for m = 6 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 16 / 61

  3. Explicit construction Figure: Hammersley point set for m = 6 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 16 / 61

  4. Explicit construction Figure: Hammersley point set for m = 6 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 16 / 61

  5. Explicit construction Figure: Hammersley point set for m = 6 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 16 / 61

  6. Explicit construction Figure: Hammersley point set for m = 6 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 16 / 61

  7. Explicit construction Figure: Hammersley point set for m = 6 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 16 / 61

  8. Explicit construction Figure: Hammersley point set for m = 6 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 16 / 61

  9. Explicit construction Figure: Hammersley point set for m = 6 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 16 / 61

  10. Explicit construction Let p 1 , p 2 , . . . ∈ F 2 [ x ] be a sequence of distinct primitive/irreducible polynomials over F 2 with e 1 ≤ e 2 ≤ · · · where e j = deg( p j ). Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 17 / 61

  11. Explicit construction Let p 1 , p 2 , . . . ∈ F 2 [ x ] be a sequence of distinct primitive/irreducible polynomials over F 2 with e 1 ≤ e 2 ≤ · · · where e j = deg( p j ). For each j , C j = ( c ( j ) k , l ) is given by the coefficients of the following Laurent series: c ( j ) c ( j ) x e j − 1 1 , 1 1 , 2 p j ( x ) = + x 2 + · · · x . . . c ( j ) c ( j ) 1 e j , 1 e j , 2 p j ( x ) = + x 2 + · · · x c ( j ) c ( j ) x e j − 1 e j +1 , 1 e j +1 , 2 ( p j ( x )) 2 = + + · · · x 2 x . . . (Sobol’ 1967; Niederreiter 1988; Tezuka 1993) Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 17 / 61

  12. Explicit construction All matrices C j ∈ F N × N are upper triangular and generate a digital 2 ( t , s )-sequence with t = ( e 1 − 1) + · · · + ( e s − 1) . (I used a C implementation for this sequence in at least more than 58636 dimensions due to Tomohiko Hironaka.) Figure: 2D projections of the first 2 6 points of the Niederreiter sequence. Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 18 / 61

  13. Polynomial lattice point sets (Niederreiter, 1992) Let p ∈ F 2 [ x ] be irreducible with deg( p ) = m Let q = ( q 1 , . . . , q s ) ∈ ( F 2 [ x ]) s with deg( q j ) < m . Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 19 / 61

  14. Polynomial lattice point sets (Niederreiter, 1992) Let p ∈ F 2 [ x ] be irreducible with deg( p ) = m Let q = ( q 1 , . . . , q s ) ∈ ( F 2 [ x ]) s with deg( q j ) < m . For each j , C j is given by the square Hankel matrix   a ( j ) a ( j ) a ( j ) · · · m 1 2 a ( j ) a ( j ) a ( j )   . . .   2 3 m +1 ∈ F m × m C j =   . . . ... 2 . . .   . . .   a ( j ) a ( j ) a ( j ) . . . m m +1 2 m − 1 where a ( j ) 1 , a ( j ) 2 , . . . are the coefficients of the Laurent series p ( x ) = a ( j ) + a ( j ) q j ( x ) 1 2 x 2 + · · · . x Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 19 / 61

  15. Polynomial lattice point sets (Niederreiter, 1992) Let p ∈ F 2 [ x ] be irreducible with deg( p ) = m Let q = ( q 1 , . . . , q s ) ∈ ( F 2 [ x ]) s with deg( q j ) < m . For each j , C j is given by the square Hankel matrix   a ( j ) a ( j ) a ( j ) · · · m 1 2 a ( j ) a ( j ) a ( j )   . . .   2 3 m +1 ∈ F m × m C j =   . . . ... 2 . . .   . . .   a ( j ) a ( j ) a ( j ) . . . m m +1 2 m − 1 where a ( j ) 1 , a ( j ) 2 , . . . are the coefficients of the Laurent series p ( x ) = a ( j ) + a ( j ) q j ( x ) 1 2 x 2 + · · · . x The resulting digital net is called a polynomial lattice point set P ( p , q ). Usually the vector q is constructed by a (fast) computer search algorithm (Nuyens & Cools, 2006; Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 19 / 61

  16. Polynomial lattice point sets (Niederreiter, 1992) Let p ∈ F 2 [ x ] be irreducible with deg( p ) = m Let q = ( q 1 , . . . , q s ) ∈ ( F 2 [ x ]) s with deg( q j ) < m . For each j , C j is given by the square Hankel matrix   a ( j ) a ( j ) a ( j ) · · · m 1 2 a ( j ) a ( j ) a ( j )   . . .   2 3 m +1 ∈ F m × m C j =   . . . ... 2 . . .   . . .   a ( j ) a ( j ) a ( j ) . . . m m +1 2 m − 1 where a ( j ) 1 , a ( j ) 2 , . . . are the coefficients of the Laurent series p ( x ) = a ( j ) + a ( j ) q j ( x ) 1 2 x 2 + · · · . x The resulting digital net is called a polynomial lattice point set P ( p , q ). Usually the vector q is constructed by a (fast) computer search algorithm (Nuyens & Cools, 2006; P. Kritzer, this afternoon!). Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 19 / 61

  17. Outline QMC and digital nets/sequences Classical QMC Higher order QMC Richardson extrapolation and QMC Application 1: Truncation of higher order nets and sequences Application 2: Extrapolated polynomial lattice rules Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 20 / 61

  18. What if f is smooth In some applications, such as PDEs with random coefficients, f can be smooth. A proper design of QMC point sets enables higher order convergence of the integration error than O (1 / N ) as expected from the KH inequality. So far, QMC point sets achieving higher order convergence for non-periodic smooth functions are 1 Higher order digital nets/sequences (Dick, 2008; ...), and 2 Tent-transformed lattice point sets (Hickernell, 2002; ...). Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 21 / 61

  19. What if f is smooth In some applications, such as PDEs with random coefficients, f can be smooth. A proper design of QMC point sets enables higher order convergence of the integration error than O (1 / N ) as expected from the KH inequality. So far, QMC point sets achieving higher order convergence for non-periodic smooth functions are 1 Higher order digital nets/sequences (Dick, 2008; ...), and 2 Tent-transformed lattice point sets (Hickernell, 2002; ...). In this talk, I will focus on higher order digital nets/sequences . The contents of this section are mostly developed by Dick (2008). Please refer to a recent review by G. & Suzuki (arXiv:1903.12353). Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 21 / 61

  20. Quality measure: high order t -value Let α ∈ N . Let t be an integer such that, for any choice 1 ≤ d j , v j < · · · < d j , 1 ≤ α m , 0 ≤ v j ≤ α m , 1 ≤ j ≤ s , with min( v j ,α ) s ∑ ∑ d j , i = α m − t , j =1 i =1 the d 1 , v 1 , . . . , d 1 , 1 -th rows of C 1 . . . the d s , v s , . . . , d s , 1 -th rows of C s are linearly independent over F 2 . P is called an order α digital ( t , m , s ) -net . Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 22 / 61

  21. Quality measure: high order t -value Order α digital ( t , m , s )-nets hold equi-distribution properties: union of dyadic elementary boxes contains the fair number of points (shown later visually). Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 23 / 61

  22. Quality measure: high order t -value Order α digital ( t , m , s )-nets hold equi-distribution properties: union of dyadic elementary boxes contains the fair number of points (shown later visually). Let t be an integer such that, for any α m ≥ t , the first 2 m points of S are an order α digital ( t , m , s )-net. S is called an order α digital ( t , s ) -sequence . Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 23 / 61

  23. Explicit construction Define D α : [0 , 1) α → [0 , 1) by  x 1 = (0 .ξ (1) 1 ξ (1) 2 ξ (1) . . . ) 2  3   x 2 = (0 .ξ (2) 1 ξ (2) 2 ξ (2)  . . . ) 2  3 �→ (0 . ξ (1) 1 ξ (2) . . . ξ ( α ) ξ (1) 2 ξ (2) . . . ξ ( α ) . . . ) 2 . . 1 1 2 2 . .    � �� � � �� �  x α = (0 .ξ ( α ) ξ ( α ) ξ ( α )  α α . . . ) 2 1 2 3 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 24 / 61

  24. Explicit construction Define D α : [0 , 1) α → [0 , 1) by  x 1 = (0 .ξ (1) 1 ξ (1) 2 ξ (1) . . . ) 2  3   x 2 = (0 .ξ (2) 1 ξ (2) 2 ξ (2)  . . . ) 2  3 �→ (0 . ξ (1) 1 ξ (2) . . . ξ ( α ) ξ (1) 2 ξ (2) . . . ξ ( α ) . . . ) 2 . . 1 1 2 2 . .    � �� � � �� �  x α = (0 .ξ ( α ) ξ ( α ) ξ ( α )  α α . . . ) 2 1 2 3 For x ∈ [0 , 1) α s , let D α ( x ) = ( D α ( x 1 , . . . , x α ) , D α ( x α +1 , . . . , x 2 α ) , . . . , D α ( x α ( s − 1)+1 , . . . , x α s )) ∈ [0 , 1) s . Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 24 / 61

  25. Explicit construction Define D α : [0 , 1) α → [0 , 1) by  x 1 = (0 .ξ (1) 1 ξ (1) 2 ξ (1) . . . ) 2  3   x 2 = (0 .ξ (2) 1 ξ (2) 2 ξ (2)  . . . ) 2  3 �→ (0 . ξ (1) 1 ξ (2) . . . ξ ( α ) ξ (1) 2 ξ (2) . . . ξ ( α ) . . . ) 2 . . 1 1 2 2 . .    � �� � � �� �  x α = (0 .ξ ( α ) ξ ( α ) ξ ( α )  α α . . . ) 2 1 2 3 For x ∈ [0 , 1) α s , let D α ( x ) = ( D α ( x 1 , . . . , x α ) , D α ( x α +1 , . . . , x 2 α ) , . . . , D α ( x α ( s − 1)+1 , . . . , x α s )) ∈ [0 , 1) s . For a digital ( t , m , α s )-net P , we write D α ( P ) = {D α ( x ) | x ∈ P } . Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 24 / 61

  26. Another look at construction Let P be a digital ( t , m , α s )-net with C 1 , . . . , C α s ∈ F m × m . We write 2       c (1) c (2) c ( α ) 1 1 1 . . .  .   .   .  C 1 =  , C 2 =  , . . . , C α =  , . . . . . . .    c (1) c (2) c ( α ) m m m Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 25 / 61

  27. Another look at construction Let P be a digital ( t , m , α s )-net with C 1 , . . . , C α s ∈ F m × m . We write 2       c (1) c (2) c ( α ) 1 1 1 . . .  .   .   .  C 1 =  , C 2 =  , . . . , C α =  , . . . . . . .    c (1) c (2) c ( α ) m m m D α ( P ) is a digital net with D 1 , . . . , D s ∈ F α m × m where 2     c (1) c ( α +1) 1 1 c (2) c ( α +2)     1 .  1     .    . .     . .         c ( α ) c (2 α )     1 1 .     .  .   .  D 1 = , D 2 = , . . . . . .         c (1) c ( α +1)     m m     c (2) c ( α +2)         m m     . . . .     . .     c ( α ) c (2 α ) m m Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 25 / 61

  28. High order t -value For a digital ( t , m , α s )-net P , D α ( P ) is an order α digital ( t ′ , m , s )-net with { ⌊ s ( α − 1) ⌋} t ′ ≤ α min m , t + . 2 For a digital ( t , α s )-sequences S , D α ( S ) is an order α digital ( t ′ , s )-sequences with t ′ ≤ α t + s α ( α − 1) . 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 26 / 61

  29. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  30. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  31. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  32. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  33. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  34. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  35. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  36. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  37. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  38. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  39. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  40. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  41. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  42. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  43. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  44. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  45. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  46. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  47. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  48. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  49. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  50. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  51. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  52. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  53. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  54. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  55. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  56. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  57. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  58. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  59. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  60. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  61. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  62. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  63. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  64. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  65. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  66. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  67. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  68. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  69. Interlaced Sobol’ point sets with α = 2 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 27 / 61

  70. Interlaced polynomial lattice point sets Let p ∈ F 2 [ x ] be irreducible with deg( p ) = m , and let q = ( q 1 , . . . , q α s ) ∈ ( F 2 [ x ]) α s with deg( q j ) < m . An interlaced polynomial lattice point set is just D α ( P ( p , q )) . Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 28 / 61

  71. Interlaced polynomial lattice point sets Let p ∈ F 2 [ x ] be irreducible with deg( p ) = m , and let q = ( q 1 , . . . , q α s ) ∈ ( F 2 [ x ]) α s with deg( q j ) < m . An interlaced polynomial lattice point set is just D α ( P ( p , q )) . The vector q can be constructed component by component. In the simplest case, the necessary construction cost is of O ( α sN log N ) with O ( N ) memory (G. & Dick, 2015; G., 2015). In applications to PDEs with random coefficients, the criterion sometimes becomes a bit complicated, requiring O ( α sN log N + α 2 s 2 N ) construction cost with O ( α sN ) memory (Dick, Kuo, Le Gia, Nuyens & Schwab, 2014). Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 28 / 61

  72. Outline QMC and digital nets/sequences Classical QMC Higher order QMC Richardson extrapolation and QMC Application 1: Truncation of higher order nets and sequences Application 2: Extrapolated polynomial lattice rules Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 29 / 61

  73. Walsh functions For k = ( . . . κ 1 κ 0 ) 2 ∈ N 0 , the k -th Walsh function is defined by wal k ( x ) = ( − 1) κ 0 ξ 1 + κ 1 ξ 2 + ··· , where x = (0 .ξ 1 ξ 2 . . . ) 2 ∈ [0 , 1), unique in the sense that infinitely many ξ i are equal to 0. Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 30 / 61

  74. Walsh functions For k = ( . . . κ 1 κ 0 ) 2 ∈ N 0 , the k -th Walsh function is defined by wal k ( x ) = ( − 1) κ 0 ξ 1 + κ 1 ξ 2 + ··· , where x = (0 .ξ 1 ξ 2 . . . ) 2 ∈ [0 , 1), unique in the sense that infinitely many ξ i are equal to 0. For s ≥ 1 and k = ( k 1 , . . . , k s ) ∈ N s 0 , the k -th Walsh function is defined by s ∏ wal k ( x ) = wal k j ( x j ) . j =1 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 30 / 61

  75. Walsh functions Figure: The k -th Walsh functions for k = 0 , 1 , 2 , 3 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 31 / 61

  76. Walsh functions Figure: The k -th Walsh functions for k = 4 , 5 , 6 , 7 Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 32 / 61

  77. Walsh functions Every Walsh function is a piecewise constant function. Takashi Goda (U. Tokyo) Extrapolation and HOQMC MCM 2019 33 / 61

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Numerical differentiation &amp; How truncation error and Richardson extrapolation floating

Numerical differentiation &amp; How truncation error and Richardson extrapolation floating

Numerical and Scientific Computing with Applications David F . Gleich CS 314, Purdue November 2, 2016 In this class you should learn: Numerical differentiation &amp; How truncation error and Richardson extrapolation floating point

97 views • 5 slides
York University www.cs.york.ac.uk/~ndm First order vs Higher order Higher order:

York University www.cs.york.ac.uk/~ndm First order vs Higher order Higher order:

First Order Haskell Neil Mitchell York University www.cs.york.ac.uk/~ndm First order vs Higher order Higher order: functions are values Can be passed around Stored in data structure Harder for reasoning and analysis More

483 views • 15 slides
Why higher-order logic? Higher-Order logic Expressive Mathematics Verification

Why higher-order logic? Higher-Order logic Expressive Mathematics Verification

Extending SMT solvers to higher-order logic Haniel Barbosa Andrew Reynolds Daniel El Ouraoui Clark Barrett Cesare Tinelli SMT 2019 20190707, Lisbon, PT To be published in the proceedings of CADE 2019 Why higher-order logic?

569 views • 25 slides
Higher order complexity Hugo Fre Mathieu Hoyrup CCA 2013 Hugo Fre Higher order

Higher order complexity Hugo Fre Mathieu Hoyrup CCA 2013 Hugo Fre Higher order

Background &amp; motivation Higher order strategies Higher order Turing machines Computable analysis Conclusion Higher order complexity Hugo Fre Mathieu Hoyrup CCA 2013 Hugo Fre Higher order complexity 1/12 Background &amp;

383 views • 15 slides
Higher-order Interpretations for Higher-order Complexity Emmanuel Hainry &amp; Romain Pchoux

Higher-order Interpretations for Higher-order Complexity Emmanuel Hainry &amp; Romain Pchoux

Higher-order Interpretations for Higher-order Complexity Emmanuel Hainry &amp; Romain Pchoux DICE-FOPARA, 22-23 april 2017 CNRS, INRIA, Universit de Lorraine &amp; LORIA, Nancy, France 1 Introduction First-order computability and

801 views • 46 slides
Higher-order Interpretations for Higher-order Complexity Emmanuel Hainry &amp; Romain Pchoux

Higher-order Interpretations for Higher-order Complexity Emmanuel Hainry &amp; Romain Pchoux

Higher-order Interpretations for Higher-order Complexity Emmanuel Hainry &amp; Romain Pchoux LPAR, 11 may 2017 CNRS, INRIA, Universit de Lorraine &amp; LORIA, Nancy, France 1 Introduction First-order computability and complexity

665 views • 39 slides
Outline Higher Order Statistics First, second and higher-order statistics Matthias Hennig

Outline Higher Order Statistics First, second and higher-order statistics Matthias Hennig

Outline Higher Order Statistics First, second and higher-order statistics Matthias Hennig Generative models, recognition models Neural Information Processing Sparse Coding School of Informatics, University of Edinburgh Independent Components

358 views • 9 slides
Outline Higher order is commonly used on convergence and on derivatives in opti- Trust Region with

Outline Higher order is commonly used on convergence and on derivatives in opti- Trust Region with

Workshop AD Higher Order Workshop AD Higher Order Outline Higher order is commonly used on convergence and on derivatives in opti- Trust Region with a Cubic Model mization. First order methods are gradient based and have

468 views • 7 slides
On Higher-Order Program Verification and Two Notions of Higher-Order Model Checking Naoki

On Higher-Order Program Verification and Two Notions of Higher-Order Model Checking Naoki

On Higher-Order Program Verification and Two Notions of Higher-Order Model Checking Naoki Kobayashi University of Tokyo Summaries of papers from POPL09, POPL17 (joint work with Etienne Lozes, Florian Bruse), and more recent work (joint work

963 views • 65 slides
A modeling and simulation perspective on extrapolation EMA Workshop on extrapolation of efficacy

A modeling and simulation perspective on extrapolation EMA Workshop on extrapolation of efficacy

A modeling and simulation perspective on extrapolation EMA Workshop on extrapolation of efficacy and safety in medicine development across age groups, 17 18 May 2016, European Medicines Agency, London Ine Skottheim Rusten on behalf of the

742 views • 32 slides
v1 v2 vn my-map : (define (my-map f xs) F F F (F v2) (F vn) (F v1) (if (null? xs) null

v1 v2 vn my-map : (define (my-map f xs) F F F (F v2) (F vn) (F v1) (if (null? xs) null

Higher-order List Func2ons Higher-Order List Functions A function is higher-order if it takes another in Racket function as an input and/or returns another function as a result. E.g. app-3-5 , make-linear-function , flip2 from the previous

519 views • 6 slides
v1 v2 vn (if (null? xs) null F F F (cons ( f (first xs)) (F vn) (F v2) (F v1) (my-map f

v1 v2 vn (if (null? xs) null F F F (cons ( f (first xs)) (F vn) (F v2) (F v1) (my-map f

Higher-order List Func2ons Higher-Order List Functions A function is higher-order if it takes another in Racket function as an input and/or returns another function as a result. E.g. app-3-5 , make-linear-function , flip2 . We will now study

300 views • 6 slides
More JavaScript! Higher-Order Functions, Callbacks, and Array Methods Higher-Order Functions

More JavaScript! Higher-Order Functions, Callbacks, and Array Methods Higher-Order Functions

More JavaScript! Higher-Order Functions, Callbacks, and Array Methods Higher-Order Functions A Higher-Order Function is any function that operates on any other function, either by taking them in as arguments or returning them. In

308 views • 9 slides
Higher Order Proof Engineering Robert White ILLC/INRIA Cool Logic, ILLC 1/23 Higher Order

Higher Order Proof Engineering Robert White ILLC/INRIA Cool Logic, ILLC 1/23 Higher Order

Higher Order Proof Engineering Higher Order Proof Engineering Robert White ILLC/INRIA Cool Logic, ILLC 1/23 Higher Order Proof Engineering Outline Introduction HOL Light and HOL Proof Checking OpenTheory Holide and Dedukti ProofCloud

767 views • 23 slides
Extrapolation framework Status quo and issues to be resolved EMA extrapolation workshop 2015-09

Extrapolation framework Status quo and issues to be resolved EMA extrapolation workshop 2015-09

Extrapolation framework Status quo and issues to be resolved EMA extrapolation workshop 2015-09 Christoph Male Austrian alternate PDCO delegate Medical University of Vienna, Department of Paediatrics An agency of the European Union Objectives

346 views • 22 slides
EMA extrapolation fram ew ork Extrapolation w orkshop 2 0 1 6 -0 5 Christoph Male Austrian PDCO

EMA extrapolation fram ew ork Extrapolation w orkshop 2 0 1 6 -0 5 Christoph Male Austrian PDCO

EMA extrapolation fram ew ork Extrapolation w orkshop 2 0 1 6 -0 5 Christoph Male Austrian PDCO delegate (alternate) Medical University of Vienna, Department of Paediatrics An agency of the European Union Evidence base for medicine use in

398 views • 15 slides
Usefulness of a Carbon target in DUNE ND : first thoughts DUNE ND meeting 15 May 2019

Usefulness of a Carbon target in DUNE ND : first thoughts DUNE ND meeting 15 May 2019

Usefulness of a Carbon target in DUNE ND : first thoughts DUNE ND meeting 15 May 2019 S.Bolognesi IRFU/CEA Why neutrino-nucleus interactions are important Modeling of neutrino-nucleus interactions is needed for ND FD extrapolation

688 views • 13 slides
Extracting neutrino oscillation parameters using a simultaneous fit of the e appearance and

Extracting neutrino oscillation parameters using a simultaneous fit of the e appearance and

Extracting neutrino oscillation parameters using a simultaneous fit of the e appearance and disappearance data in the NOvA experiment Prabhjot Singh , University of Delhi, India On Behalf of the NOvA Collaboration DPF 2017 Fermilab

617 views • 20 slides
Extrapolation of operator moments, with problems C. Brezinski, applications to linear algebra

Extrapolation of operator moments, with problems C. Brezinski, applications to linear algebra

Extrapolation of operator moments, with applications to linear algebra Extrapolation of operator moments, with problems C. Brezinski, applications to linear algebra problems P. Fika, + and M. Mitrouli + Introduction C. Brezinski,

517 views • 28 slides
Nonparametric estimation of extreme risks from heavy-tailed distributions St ephane GIRARD

Nonparametric estimation of extreme risks from heavy-tailed distributions St ephane GIRARD

Nonparametric estimation of extreme risks from heavy-tailed distributions St ephane GIRARD joint work with Jonathan EL METHNI &amp; Laurent GARDES September 2013 Outline Extreme risk measures Estimators and asymptotic results Extrapolation

562 views • 34 slides
Richardsons Extrapolation Suppose h = 0 we have a formula N 1 ( h ) that approximates an

Richardsons Extrapolation Suppose h = 0 we have a formula N 1 ( h ) that approximates an

Richardsons Extrapolation Suppose h = 0 we have a formula N 1 ( h ) that approximates an unknown value M M N 1 ( h ) = K 1 h + K 2 h 2 + K 3 h 3 + , (7) for some unknown constants K 1 , K 2 , K 3 , . . . . If K 1 = 0,

473 views • 8 slides
Performance evaluation of an efficient double-double BLAS1 function with error-free

Performance evaluation of an efficient double-double BLAS1 function with error-free

Performance evaluation of an efficient double-double BLAS1 function with error-free transformation and its application to explicit extrapolation methods Tomonori Kouya Shizuoka Institute of Science and Technology kouya.tomonori@sist.ac.jp

606 views • 24 slides
4. Numerical Quadrature: Where analytical abilities end . . . 4. Numerical Quadrature: Where

4. Numerical Quadrature: Where analytical abilities end . . . 4. Numerical Quadrature: Where

Preliminary Remarks Simple and Composite Rules Extrapolation Other Approaches Applications of Quadrature 4. Numerical Quadrature: Where analytical abilities end . . . 4. Numerical Quadrature: Where analytical abilities end . . . Numerical

463 views • 32 slides
Channels &amp; Keyframes CSE169: Computer Animation Instructor: Steve Rotenberg UCSD, Spring

Channels &amp; Keyframes CSE169: Computer Animation Instructor: Steve Rotenberg UCSD, Spring

Channels &amp; Keyframes CSE169: Computer Animation Instructor: Steve Rotenberg UCSD, Spring 2016 Animation Rigging and Animation Animation System ... Pose 1 2 N Rigging System Triangles Renderer

938 views • 62 slides

More recommend