Modeling with Metaconstraints and Semantic Typing of Variables

Size: px
Start display at page:

Download "Modeling with Metaconstraints and Semantic Typing of Variables"

Transcription

1 Modelng wth Metaconstrants and Semantc Typng of Varables André Cré Unversty of Toronto John Hooker Carnege Mellon Unversty Tallys Yunes Unversty of Mam INFORMS Journal on Computng (2016) 1-13

2 Basc Problem Metaconstrants (n partcular, global constrants) n a model help convey problem structure to the solver. But they pose a fundamental problem of varable management. How to solve t? Slde 2

3 Basc Problem Metaconstrants (n partcular, global constrants) n a model help convey problem structure to the solver. But they pose a fundamental problem of varable management. How to solve t? Treat varable declaratons as database queres. In a system of semantc typng. Slde 3

4 Explotng Problem Structure You can t solve hard problems wthout explotng specal structure (No Free Lunch Theorem). Slde 4

5 Explotng Problem Structure You can t solve hard problems wthout explotng specal structure (No Free Lunch Theorem). For SAT solvers: Effcent encodng of problem n SAT form Slde 5

6 Explotng Problem Structure You can t solve hard problems wthout explotng specal structure (No Free Lunch Theorem). For SAT solvers: Effcent encodng of problem n SAT form For CP solvers: Careful choce of global constrants Redundant constrants, search strategy, etc. Slde 6

7 Explotng Problem Structure You can t solve hard problems wthout explotng specal structure (No Free Lunch Theorem). For SAT solvers: Effcent encodng of problem n SAT form For CP solvers: Careful choce of global constrants Redundant constrants, search strategy, etc. For MIP solvers: Careful choce of varables for tght formulaton Addton of vald nequaltes Slde 7

8 Conveyng structure to the solver(s) Formulate problem wth global constrants or metaconstrants to reveal structure Automatcally convert these to optmal formulaton for the solvers(s) Best choce of varables. Reformulaton of constrants. For effectve propagaton or tght relaxaton Best choce of doman flters. Generaton of vald nequaltes Slde 8

9 Conveyng structure to the solver(s) Formulate problem wth global constrants or metaconstrants to reveal structure Automatcally convert these to optmal formulaton for the solvers(s) Best choce of varables. Reformulaton of constrants. For effectve propagaton or tght relaxaton Best choce of doman flters. Generaton of vald nequaltes However, metaconstrants pose a fundamental problem of varable management Slde 9

10 Varable management problem Reformulaton typcally ntroduces new varables Dfferent metaconstrants may ntroduce varables that are functonally the same varable or related n some other way. Recognzng these relatonshps s essental to obtanng a good model (e.g., a tght contnuous relaxaton) How can the solver understand what s gong on n the model? Slde 10

11 Varable management problem Reformulaton typcally ntroduces new varables Dfferent metaconstrants may ntroduce varables that are functonally the same varable or related n some other way. Recognzng these relatonshps s essental to obtanng a good model (e.g., a tght contnuous relaxaton) How can the solver understand what s gong on n the model? Proposal: Model wth semantc typng of varables. Slde 11

12 Varable management problem Example: Let x j = worker assgned to job j c j = cost of assgnng worker to job j Fnd mn-cost assgnment: mn j c alldff ( x,, x ) 1 jx j n Slde 12

13 Varable management problem Example: Let x j = worker assgned to job j c j = cost of assgnng worker to job j Fnd mn-cost assgnment: mn Objectve functon s reformulated wth 0-1 varables: j c alldff ( x,, x ) 1 jx j n where x j yj mn cy j j j Slde 13

14 Varable management problem Example: Let x j = worker assgned to job j c j = cost of assgnng worker to job j Fnd mn-cost assgnment: mn Objectve functon s reformulated wth 0-1 varables: j c alldff ( x,, x ) 1 jx j n where x j yj mn cy j j j Alldff constrant s reformulated wth 0-1 varables: y 1, all j; y 1, all j j j Slde 14

15 Varable management problem Slde 15 Example: Let x j = worker assgned to job j c j = cost of assgnng worker to job j Fnd mn-cost assgnment: mn Objectve functon s reformulated wth 0-1 varables: Alldff constrant s reformulated wth 0-1 varables: j c alldff ( x,, x ) 1 jx j n where x j yj mn cy j j j y 1, all j; y 1, all How does the solver know that we want y j = y j, allowng the problem to be solved rapdly as a classcal assgnment problem? j j j

16 Semantc typng We assume that varables are declared. Slde 16

17 Semantc typng We assume that varables are declared. Semantc typng assgns a dfferent meanng to each varable By assocatng the varable wth a mult-place predcate and keyword. The keyword queres the relaton denoted by the predcate, as one queres a relatonal database. Slde 17

18 Semantc typng We assume that varables are declared. Semantc typng assgns a dfferent meanng to each varable By assocatng the varable wth a mult-place predcate and keyword. The keyword queres the relaton denoted by the predcate, as one queres a relatonal database. Advantage: Ths allows the solver to deduce relatonshps between varables, both orgnal or ntroduced. Can automatcally add channelng constrants. Slde 18 It s also good modelng practce.

19 How varables are ntroduced A model may nclude two formulatons of the problem that use related varables. Common n CP, because t strengthens propagaton. Slde 19

20 How varables are ntroduced A model may nclude two formulatons of the problem that use related varables. Common n CP, because t strengthens propagaton. For example, x job assgned to worker y j worker assgned to job j Solver should generate channelng constrants to relate the varables to each other: j x, y y j x Slde 20

21 How varables are ntroduced The solver may reformulate a dsjuncton of lnear systems k A x k b k usng a convex hull (or bg-m ) formulaton: k k A x b y, all k k k x x, y 1 y k k k k 0,1, all k k Other constrants may be based on same set of alternatves, and correspondng auxlary varables (y k etc.) should be equated. Slde 21

22 How varables are ntroduced A nonlnear or global solver may use McCormck factorzaton to replace nonlnear subexpressons wth auxlary varables to obtan a lnear relaxaton. Slde 22

23 How varables are ntroduced A nonlnear or global solver may use McCormck factorzaton to replace nonlnear subexpressons wth auxlary varables to obtan a lnear relaxaton. For example, blnear term xy can be lnearzed by replacng t wth new varable z and constrants L x L y L L z L x U y L U y x x y y x x y U x U y U U z U x L y U L y x x y y x x y where x Lx, Ux, y Ly, U y Factorzaton of dfferent constrants may create varables for dentcal subexpressons. They should be dentfed to get a tght relaxaton. Slde 23

24 How varables are ntroduced The solver may reformulate dfferent global constrants from CP by ntroducng varables that have the same meanng. Slde 24

25 How varables are ntroduced The solver may reformulate dfferent global constrants from CP by ntroducng varables that have the same meanng. For example, sequence constrant lmts how many jobs of a gven type can occur n gven tme nterval: and cardnalty constrant lmts how many tmes a gven job appears cardnalty x, x job n poston j Both may ntroduce varables that should be dentfed. sequence x, x job n poston j y 1 when job j occurs n poston j Slde 25

26 How varables are ntroduced The solver may ntroduce equvalent varables whle nterpretng metaconstrants desgned for classcal MIP modelng stuatons: Fxed-charge network flow Faclty locaton Lot szng Job shop schedulng Assgnment (3-dm, quadratc, etc.) Pecewse lnear Slde 26

27 Motvatng example Allocate 10 advertsng spots to 5 products x how many spots allocated to product yj 1 f j spots allocated to product A B C D E Slde 27

28 Motvatng example Allocate 10 advertsng spots to 5 products 4 spots per product x how many spots allocated to product yj 1 f j spots allocated to product A B C D E Slde 28

29 Motvatng example Allocate 10 advertsng spots to 5 products 4 spots per product Advertse 3 products x how many spots allocated to product yj 1 f j spots allocated to product A B C D E Slde 29

30 Motvatng example Allocate 10 advertsng spots to 5 products x how many spots allocated to product yj 1 f j spots allocated to product 4 spots per product Advertse 3 products 4 spots for at least one product A B C D E Slde 30

31 Motvatng example Allocate 10 advertsng spots to 5 products 4 spots per product Advertse 3 products x how many spots allocated to product yj 1 f j spots allocated to product A B C D E 4 spots for at least one product P j = proft from allocatng j spots to product Objectve: maxmze proft Slde 31

32 Motvatng example spots n {0..4} product n {A,B,C,D,E}) Index sets Slde 32

33 Motvatng example spots n {0..4} product n {A,B,C,D,E} data P{product,spots} Data nput Slde 33

34 Motvatng example spots n {0..4} product n {A,B,C,D,E} data P{product,spots} Declaraton of varable x x[] s howmany spots allocate(product ) Slde 34

35 Motvatng example spots n {0..4} product n {A,B,C,D,E} data P{product,spots} Declaraton of varable x x[] s howmany spots allocate(product ) Ths makes t a varable declaraton Slde 35

36 Motvatng example spots n {0..4} product n {A,B,C,D,E} data P{product,spots} Declaraton of varable x x[] s howmany spots allocate(product ) Ths s the semantc type Slde 36

37 Motvatng example spots n {0..4} product n {A,B,C,D,E} data P{product,spots} Declaraton of varable x x[] s howmany spots allocate(product ) Slde 37 Indcates an nteger quantty Other keywords: howmuch whether

38 Motvatng example spots n {0..4} product n {A,B,C,D,E} data P{product,spots} Declaraton of varable x x[] s howmany spots allocate(product ) How many of what? Slde 38

39 Motvatng example spots n {0..4} product n {A,B,C,D,E} data P{product,spots} Declaraton of varable x x[] s howmany spots allocate(product ) 2-place predcate assocated wth varable x Slde 39 Every varable s assocated wth a predcate that gves t meanng

40 Motvatng example spots n {0..4} product n {A,B,C,D,E} data P{product,spots} Declaraton of varable x x[] s howmany spots allocate(product ) Other term of the predcate Slde 40

41 Motvatng example spots n {0..4} product n {A,B,C,D,E} data P{product,spots} Declaraton of varable x x[] s howmany spots allocate(product ) Assocates ndex of x[] wth ndex set product Slde 41

42 Motvatng example max x 10, y 2, y y 1, x jy, all spots n {0..4} j j product n {A,B,C,D,E} j j data P{product,spots} x[] s howmany spots allocate(product ) maxmze sum{product } P[,x[]] Objectve functon P x Slde 42

43 Motvatng example max x 10, y 2, y y 1, x jy, all spots n {0..4} product n {A,B,C,D,E} data P{product,spots} j j x[] s howmany spots allocate(product ) maxmze sum{product } P[,x[]] sum{product } x[] <= spots avalable P x j j Slde 43

44 Motvatng example max x 10, y 2, y y 1, x jy, all spots n {0..4} j j product n {A,B,C,D,E} j j data P{product,spots} x[] s howmany spots allocate(product ) maxmze sum{product } P[I,x[]] sum{product } x[] <= 10 y[,j] s whether allocate(product, spots j) Declare y j P x Indcates 0-1 varable Slde 44

45 Motvatng example max x 10, y 2, y y 1, x jy, all spots n {0..4} j j product n {A,B,C,D,E} j j data P{product,spots} x[] s howmany spots allocate(product ) maxmze sum{product } P[,x[]] sum{product } x[] <= 10 y[,j] s whether allocate(product, spots j) Declare y j P x Assocated wth same predcate as x[] Slde 45

46 Motvatng example spots n {0..4} product n {A,B,C,D,E} data P{product,spots} max x 10, y 2, y y 1, x jy, all x[] s howmany spots allocate(product ) maxmze sum{product } P[,x[]] sum{product } x[] <= 10 y[,j] s whether allocate(product, spots j) sum{product } y[,0] >= 2 At least 2 products not advertsed j P x j j j Slde 46

47 Motvatng example spots n {0..4} product n {A,B,C,D,E} data P{product,spots} x[] s howmany spots allocate(product ) maxmze sum{product } P[,x[]] sum{product } x[] <= 10 y[,j] s whether allocate(product, spots j) sum{product } y[,0] >= 2 sum{product } y[,4] >= 1 max x x 10, y 2, y j P y 1, x jy, all j j j At least 1 product gets 4 spots Slde 47

48 Motvatng example max x 10, y 2, y y 1, x jy, all spots n {0..4} j j product n {A,B,C,D,E} j j data P{product,spots} x[] s howmany spots allocate(product ) maxmze sum{product } P[,x[]] sum{product } x[] <= 10 y[,j] s whether allocate(product, spots j) sum{product } y[,0] >= 2 sum{product } y[,4] >= 1 {product } sum{spots j} y[,j] = 1 {product } x[] = sum{spots j} j*y[,j] Solver generates lnkng constrants because x[] and y[,j] are assocated wth the same predcate. P x Slde 48

49 Motvatng example max spots n {0..4} product n {A,B,C,D,E} data P{product,spots} j j x[] s howmany spots allocate (product ) maxmze sum{product } P[,x[]] Ths constrant must be lnearzed. Solver generates x 10, y 2, y 1 y 1, x jy, all y [,j] s whether allocate(product, spots j) z j j z P y, y 1, x jy, all j j j j j 0 j 0 j 0 Slde 49

50 Motvatng example max spots n {0..4} product n {A,B,C,D,E} data P{product,spots} j j x[] s howmany spots allocate (product ) maxmze sum{product } P[,x[]] Ths constrant must be lnearzed. Solver generates x 10, y 2, y 1 y 1, x jy, all y [,j] s whether allocate(product, spots j) y and y are dentfed because they have the same type: y[,j] s whether allocate(product, spots j) z j j z P y, y 1, x jy, all j j j j j 0 j 0 j 0 Slde 50

51 Predcates and relatons Predcate allocate denotes 2-place relaton (set of tuples). Schematcally ndcated by: 1 2 product spots x Slde 51

52 Predcates and relatons Predcate allocate denotes 2-place relaton (set of tuples). Schematcally ndcated by: 1 2 product spots x Column correspondng to a varable must be a functon of other columns. Slde 52

53 Predcates and relatons Predcate allocate denotes 2-place relaton (set of tuples). Schematcally ndcated by: 1 2 product spots x Declaraton of x[] as howmany spots allocate (product ) and y[,j] as whether allocate (product, spots j) query the relaton for how many and whether. Slde 53

54 Predcates and relatons Predcate allocate denotes 2-place relaton (set of tuples). Schematcally ndcated by: 1 2 product spots x Declaraton of x[] as howmany spots allocate (product ) and y[,j] as whether allocate (product, spots j) query the relaton for how many and whether. In general, keywords are queres (analogous to relatonal database) Slde 54

55 Predcates and relatons Relaton table reveals channelng constrants. For example, x[] s whch job assgn(worker ) y[j] s whch worker assgn(job ) 1 2 job worker j, x, y j Slde 55 We can read off the channelng constrants j x x y y j y x

56 Predcates and relatons If several jobs can be assgned to a worker, we declare z[] s whchset job assgn(worker ) The channelng constrants are j z y Slde 56

57 Prevous work Model management uses semantc typng to help combne models and use nhertance. Orgnally nspred by object-orented programmng Bradley & Clemence (1988) Quddty: a rgorous attempt to analyze condtons for varable dentfcaton Bhargava, Kmbrough & Krshnan (1991) SML uses typng n a structured modelng framework Geoffron (1992) Ascend uses strongly-typed, object-orented modelng Bhargava, Krshnan & Pela (1998) Slde 57

58 Prevous work Our semantc typng dffers: Less ambtous because t doesn t attempt model management. There s only one model. More ambtous because we recognze relatonshps other than equvalence. We manage varables ntroduced by solver. Slde 58

59 Prevous work Modelng systems that convey some structure to solver: CP modelng systems use global constrants. AIMMS uses typed ndex sets. MnZnc reformulates metaconstrants for specfc solvers. Savle Row uses common subexpresson elmnaton. OPL, Xpress-Kals, Comet, etc., use nterval varables. SAT solver SymChaff uses hgh-level AI plannng language PDDL. SIMPL has full metaconstrant capablty. Slde 59

60 Prevous work However, none of these systems deals systematcally wth the varable management problem. We address t wth semantc typng of varables. Slde 60

61 Assgnment problem mn worker n {1..m} job n {1..n} data C{worker,job} x[] s whch job assgn(worker ) mnmze sum{worker } C[,x[]] alldff{x[*]} alldff x,, c x 1 x n Slde 61

62 Assgnment problem mn worker n {1..m} job n {1..n} data C{worker,job} x[] s whch job assgn(worker ) mnmze sum{worker } C[,x[]] alldff{x[*]} alldff x,, c x 1 x n Objectve functon s formulated max c y, x y, all j j j j y[,j] s whether assgn(worker, job j) Slde 62

63 Assgnment problem mn worker n {1..m} job n {1..n} data C{worker,job} x[] s whch job assgn(worker ) mnmze sum{worker } C[,x[]] alldff{x[*]} alldff x,, c x 1 x n Objectve functon s formulated max c y, x y, all j j j j y[,j] s whether assgn(worker, job j) Alldff s formulated Slde 63 y 1, all, y 1, all j, x jy, all j j j j j y [,j] s whether assgn(worker, job j) Solver dentfes y and y to create classcal AP.

64 Latn squares j Numbers n every row and column are dstnct. We wll use three formulatons to mprove propagaton. Slde 64

65 Latn squares j Numbers n every row and column are dstnct. We wll use three formulatons to mprove propagaton. alldff x, x, all 1 alldff x, x, all j 1 j 1 1k n nj alldff y, y, all alldff n y y, all k j1 1k nk alldff z, x, all j alldff z, x, all k jn nk row, col, num n {1..n} x[,j] s whch num assgn(row, col j) y[,k] s whch col assgn(row, num k) z[j,k] s whch row assgn(col j, num k) Slde 65

66 Latn squares j Numbers n every row and column are dstnct. We wll use three formulatons to mprove propagaton. alldff x, x, all 1 alldff x, x, all j 1 j 1 1k n nj alldff y, y, all alldff n y y, all k j1 1k nk alldff z, x, all j alldff z, x, all k jn nk row, col, num n {1..n} x[,j] s whch num assgn(row, col j) y[,k] s whch col assgn(row, num k) z[j,k] s whch row assgn(col j, num k) {row } alldff{x[,*]); {col j} alldff{x[*,j]) {row } alldff{y[,*]); {num k} alldff{y[*,j]) {col j} alldff{z[j,*]); {num k} alldff{z[*,k]) Slde 66

67 Latn squares The predcate assgn denotes the 3-place relaton num col row k, x j j, y k, z jk row, col, num n {1..n} x[,j] s whch num assgn(row, col j) y[,k] s whch col assgn(row, num k) z[j,k] s whch row assgn(col j, num k) {row } alldff{x[,*]); {col j} alldff{x[*,j]) {row } alldff{y[,*]); {num k} alldff{y[*,j]) {col j} alldff{z[j,*]); {num k} alldff{z[*,k]) Slde 67

68 Latn squares The predcate assgn denotes the 3-place relaton num col row k, x j j, y k, z jk We can read off the channelng constrants: k x, j y, z, all, j, k z y z x y x jk k jk j k kj whch can be propagated. Slde 68

69 Latn squares {row } alldff{x[,*]); {col j} alldff{x[*,j]) {row } alldff{y[,*]); {num k} alldff{y[*,j]) {col j} alldff{z[j,*]); {num k} alldff{z[*,k]) The 3 formulatons generate 3 dentcal MIP models: x x x x x k ; 1, all, j; 1, all, k; 1, all j, k j jk jk jk jk k k j y y y y y j, 1, all, k; 1, all, j; 1, all j, k k jk jk jk jk j j k z z z z z jk jk, jk 1, all j, k; jk 1 j jk, all, ; 1, all, k j k Slde 69

70 Latn squares {row } alldff{x[,*]); {col j} alldff{x[*,j]) {row } alldff{y[,*]); {num k} alldff{y[*,j]) {col j} alldff{z[j,*]); {num k} alldff{z[*,k]) The 3 formulatons generate 3 dentcal MIP models: x x x x x k ; 1, all, j; 1, all, k; 1, all j, k y y y y y j, 1, all, k; 1, all, j; 1, all j, k z z z z z jk jk, jk 1, all j, k; jk 1 j jk Slde 70 j jk jk jk jk k k j k jk jk jk jk j j k The solver declares, all, ; 1, all, x y z,, jk jk jk k whether assgn(row, col j, num k) So t treats them as the same varable and generates only 1 MIP model. j k

71 Multple whch varables In general, an n-place predcate that denotes the relaton 1 k k + 1 n term 1 term k term k+1 term n 1 1, x (1) x k 1, k k ( k ) n for whch varables, where j ( ) 1 j1 j1 n generates the channelng constrants j x, all,,, j 1,, k j 1 j1 j1 k (1) ( j1) ( j1) ( k ) k 1 n x x x x 1 n Slde 71

72 Multple whether varables whether keywords serve as projecton operators on the relaton. y[,j,d] s whether assgn(worker, job j, day d) Project out d : y1[,j] s whether assgn(worker, job j) Project out j and d : y2[] s whether assgn(worker ) Slde 72

73 Short forms Declare x to be cost of actvty : x[] s howmuch cost(actvty ) whch s short for the formal declaraton x[] s howmuch cost cost(actvty ) n whch a new term cost s generated Declare x to be cost: x s howmuch cost whch s short for x s howmuch cost cost() Slde 73

74 Pecewse lnear Pecewse lnear functon z = f(x) Breakponts n A, ordnates n C f(x) x s howmuch output ndex n {1..n} data A,C{ndex} A z s howmuch cost pecewse(x,z,a,c) ths metaconstrant defnes z = f(x) C x Slde 74

75 Pecewse lnear Pecewse lnear functon z = f(x) Breakponts n A, ordnates n C x s howmuch output ndex n {1..n} data A,C{ndex} z s howmuch cost pecewse(x,z,a,c) Solver generates the locally deal model n1 n1 c 1 c x a1 x, z c1 x a a C f(x) ( a a ) x ( a a ), 0,1, 1,, n 1 A x We need to declare auxlary varables, x Slde 75

76 Pecewse lnear Pecewse lnear functon z = f(x) Breakponts n A, ordnates n C x s howmuch output ndex n {1..n} data A,C{ndex} z s howmuch cost pecewse(x,z,a,c) pecewse constrant nduces solver to declare a new ndex set that assocates ndex wth A, and use t to declare, x xbar[] s howmuch output.a(ndex ) delta[] s whether lastpostve output.a(ndex ) C f(x) A x Both declaratons create predcates nherted from output and A Slde 76

77 Pecewse lnear f(x) Suppose there s another pecewse functon on the same break ponts f(x) x s howmuch output ndex n {1..n} x data A,C{ndex} A z s howmuch cost pecewse(x,z,a,c) data C {ndex} z s howmuch proft pecewse(x,z,a,c ) x [] s howmuch cost output.a(ndex ) delta [] s whether lastpostve output.a(ndex) Slde 77

78 Pecewse lnear f(x) Suppose there s another pecewse functon on the same break ponts x s howmuch output ndex n {1..n} data A,C{ndex} z s howmuch cost pecewse(x,z,a,c) data C {ndex} z s howmuch proft pecewse(x,z,a,c ) x [] s howmuch cost output.a(ndex ) delta [] s whether lastpostve output.a(ndex) A f(x) Because new pecewse constrant s assocated wth the same x and A, solver agan creates output.a. x The solver creates varables and x wth same types as and x and so dentfes them. Slde 78

79 Interval varables cumulatve x, D, R, L Each job j runs for a tme nterval x j. We wsh to schedule jobs so that total resource consumpton never exceeds L. job n {1..n} tme n {t..t} data W,D,R{job} wndow, duraton, resource runnng n [tme,tme] makes runnng an nterval varable x[j] s when runnng sched(job j) subset W[j] cumulatve(x,d,r,l) L x W, all j j j Slde 79

80 Interval varables Each job j runs for a tme nterval x j. We wsh to schedule jobs so that total resource consumpton never exceeds L. job n {1..n} tme n {t..t} data W,D,R{job} wndow, duraton, resource Solver generates the model delta[j,t] s whether runnng.start sched(job j, tme t) ph[j,t] s whether runnng sched(job j, tme t) Slde 80 t 1, all j; R L, all t jt j jt j, all t, t wth 0 t t D, all j jt jt j cumulatve x, D, R, L x W, all j runnng n [tme,tme] makes runnng an nterval varable x[j] s when runnng sched(job j) subset W[j] cumulatve(x,d,r,l) j j

81 Interval varables Suppose we want fnsh tmes to be separated by at least T 0 cumulatve x, D, R, L job n {1..n} tme n {t..t} data W,D,R{job} runnng n [tme,tme] x[j] s when runnng sched(job j) subset W[j] cumulatve(x,d,r,l) {job j, job k j<>k} x[j].end x[k].end >= T0 end j end k x W, all j j j x x T, all j, k, j k delta[j,t] s whether runnng.start sched(job j, tme t) ph[j,t] s whether runnng sched(job j, tme t) 0 Slde 81

82 Interval varables Suppose we want fnsh tmes j j to be separated by at least T 0 end end x j xk T0, all j, k, j k job n {1..n} tme n {t..t} data W,D,R{job} runnng n [tme,tme] x[j] s when runnng sched(job j) subset W[j] cumulatve(x,d,r,l) {job j, job k j<>k} x[j].end x[k].end >= T0 delta[j,t] s whether runnng.start sched(job j, tme t) ph[j,t] s whether runnng sched(job j, tme t) Solver generates cumulatve x, D, R, L jt kt 1, all t, t wth 0 t t T0, all j, t wth j k epslon[j,t] s whether runnng.end sched(job j, tme t) x W, all j Slde 82

83 Interval varables Varables jt and jt are related by an offset. Solver assocates runnng.end n declaraton of jt wth runnng.start n declaraton of jt and deduces e, jt, all j, t j t D j delta[j,t] s whether runnng.start sched(job j, tme t) ph[j,t] s whether runnng sched(job j, tme t) epslon[j,t] s whether runnng.end sched(job j, tme t) Slde 83

84 Interval varables Varables jt and jt are related by an offset. Solver assocates runnng.end n declaraton of jt wth runnng.start n declaraton of jt and deduces e, jt, all j, t j t D j Solver also assocates runnng.end n declaraton of jt wth runnng n declaraton of jt and deduces the redundant constrants, all t, twth 0 t t D, all j jt jt j delta[j,t] s whether runnng.start sched(job j, tme t) ph[j,t] s whether runnng sched(job j, tme t) epslon[j,t] s whether runnng.end sched(job j, tme t) Slde 84

85 TSP wth Sde Constrants Travelng salesman problem wth mssng arcs and precedence constrants. mn alldff x, crcut s x x, all, j wth prec 1 Succ cty, poston n {1..n} data D{cty, cty} Dstances data Prec{cty, cty} Prec[,j]=1 f must precede j data Succ{cty} Succ[j] = set of successors of cty j s D s j j Slde 85

86 TSP wth Sde Constrants Travelng salesman problem wth mssng arcs and precedence constrants. mn alldff x, crcut cty, poston n {1..n} data D{cty, cty} Dstances data Prec{cty, cty} Prec[,j]=1 f must precede j data Succ{cty} Succ[j] = set of successors of cty j Two varable systems: x[] s whch poston orderng(cty ) s[] s successor cty orderng(cty ) subset Succ[] D s x x, all, j wth prec 1 s Succ s j j Slde 86

87 TSP wth Sde Constrants Travelng salesman problem wth mssng arcs and precedence constrants. mn alldff x, crcut cty, poston n {1..n} data D{cty, cty} Dstances data Prec{cty, cty} Prec[,j]=1 f must precede j data Succ{cty} Succ[j] = set of successors of cty j Two varable systems: x[] s whch poston orderng(cty ) s[] s successor cty orderng(cty ) subset Succ[] s x x, all, j wth prec 1 Succ Precedence constrants requre x varables prec{cty, cty j Prec[,j] = 1}: x[] < x[j] Mssng arc constrants (mplct n data Succ) requre s varables s D s j j Slde 87

88 TSP wth Sde Constrants Travelng salesman problem wth mssng arcs and precedence constrants. cty, poston n {1..n} data D{cty, cty} Dstances data Prec{cty, cty} Prec[,j]=1 f must precede j data Succ{cty} Succ[j] = set of successors of cty j Two varable systems: x[] s whch poston orderng(cty ) s[] s successor cty orderng(cty ) subset Succ[] Precedence constrants requre x varables prec{cty, cty j Prec[,j] = 1}: x[] < x[j] Mssng arc constrants (mplct n data Succ) requre s varables mn sum {cty } D[,s[]] Objectve functon Slde 88 mn alldff D x, crcut s x x, all, j wth prec 1 s Succ s j j

89 TSP wth Sde Constrants The solver can gve alldff(x) a conventonal assgnment model usng z k = whether cty s n poston k. z[,k] s whether orderng(cty, poston k) Slde 89

90 TSP wth Sde Constrants The solver can gve alldff(x) a conventonal assgnment model usng z k = whether cty s n poston k. z[,k] s whether orderng(cty, poston k) For crcut(s), the solver can ntroduce w j = whether cty mmedately precedes cty j. w[,j] s whether successor orderng(cty, cty j) Slde 90

91 TSP wth Sde Constrants The solver can gve alldff(x) a conventonal assgnment model usng z k = whether cty s n poston k. z[,k] s whether orderng(cty, poston k) For crcut(s), the solver can ntroduce w j = whether cty mmedately precedes cty j. w[,j] s whether successor orderng(cty, cty j) Declaraton of z tells solver that predcate s orderng(cty,poston), not orderng(cty,cty). Slde 91

92 TSP wth Sde Constrants The solver can gve alldff(x) a conventonal assgnment model usng z k = whether cty s n poston k. z[,k] s whether orderng(cty, poston k) For crcut(s), the solver can ntroduce w j = whether cty mmedately precedes cty j. w[,j] s whether successor orderng(cty, cty j) Declaraton of z tells solver that predcate s orderng(cty,poston), not orderng(cty,cty). Solver generates cuttng planes n w-space and s-space. Slde 92

93 TSP wth Sde Constrants The solver can gve alldff(x) a conventonal assgnment model usng z k = whether cty s n poston k. z[,k] s whether orderng(cty, poston k) For crcut(s), the solver can ntroduce w j = whether cty mmedately precedes cty j. w[,j] s whether successor orderng(cty, cty j) Declaraton of z tells solver that predcate s orderng(cty,poston), not orderng(cty,cty). Solver generates cuttng planes n w-space and s-space. The successor keyword tells solver how z and w relate. Slde 93, all t, twth 0 t t D, all j jt jt j

94 TSP wth Sde Constrants Suppose we also have constrants on whch cty s n poston k. Smply declare y[k] = whch cty orderng(poston k) The solver generates the channelng constrants between y[k] and x[] = whch poston s cty Slde 94

95 TSP wth Sde Constrants Suppose we also have constrants on whch cty s n poston k. Smply declare y[k] = whch cty orderng(poston k) The solver generates the channelng constrants between y[k] and x[] = whch poston s cty The solver can also ntroduce a second (equvalent) objectve functon mn sum{poston k} D[y[k],y[k+1]] whch may mprove boundng. MIP Slde 95

96 Pros and Cons of Semantc Typng Pros Conveys problem structure to the solver(s) by allowng use of metaconstants Incorporates state of the art n formulaton, vald nequaltes Allows solver to expand repertory of technques Doman flterng, propagaton, cuttng plane algorthms Good modelng practce Self-documentng Bug detecton Slde 96

97 Pros and Cons of Semantc Typng Cons Modeler must be famlar wth a large collecton of metaconstrants Rather than few prmtve constrants Slde 97

98 Pros and Cons of Semantc Typng Cons Modeler must be famlar wth a large collecton of metaconstrants Rather than few prmtve constrants Response Modeler must be famlar wth the underlyng concepts anyway Modelng system can offer sophstcated help, mprove modelng Slde 98

99 Pros and Cons of Semantc Typng Cons Modeler must be famlar wth a large collecton of metaconstrants Rather than few prmtve constrants Response Modeler must be famlar wth the underlyng concepts anyway Modelng system can offer sophstcated help, mprove modelng OR, SAT communty s not accustomed to hgh-level modelng Typed languages lke Ascend never really caught on, resstance to CP. Slde 99

100 Pros and Cons of Semantc Typng Cons Modeler must be famlar wth a large collecton of metaconstrants Rather than few prmtve constrants Slde 100 Response Modeler must be famlar wth the underlyng concepts anyway Modelng system can offer sophstcated help, mprove modelng OR, SAT communty s not accustomed to hgh-level modelng Typed languages lke Ascend never really caught on, resstance to CP. Response Tran the next generaton!

Single-Facility Scheduling over Long Time Horizons by Logic-based Benders Decomposition

Single-Facility Scheduling over Long Time Horizons by Logic-based Benders Decomposition Sngle-Faclty Schedulng over Long Tme Horzons by Logc-based Benders Decomposton Elvn Coban and J. N. Hooker Tepper School of Busness, Carnege Mellon Unversty ecoban@andrew.cmu.edu, john@hooker.tepper.cmu.edu

More information

An Integrated OR/CP Method for Planning and Scheduling

An Integrated OR/CP Method for Planning and Scheduling An Integrated OR/CP Method for Plannng and Schedulng John Hooer Carnege Mellon Unversty IT Unversty of Copenhagen June 2005 The Problem Allocate tass to facltes. Schedule tass assgned to each faclty. Subect

More information

Global Optimization of Truss. Structure Design INFORMS J. N. Hooker. Tallys Yunes. Slide 1

Global Optimization of Truss. Structure Design INFORMS J. N. Hooker. Tallys Yunes. Slide 1 Slde 1 Global Optmzaton of Truss Structure Desgn J. N. Hooker Tallys Yunes INFORMS 2010 Truss Structure Desgn Select sze of each bar (possbly zero) to support the load whle mnmzng weght. Bar szes are dscrete.

More information

Combining Constraint Programming and Integer Programming

Combining Constraint Programming and Integer Programming Combnng Constrant Programmng and Integer Programmng GLOBAL CONSTRAINT OPTIMIZATION COMPONENT Specal Purpose Algorthm mn c T x +(x- 0 ) x( + ()) =1 x( - ()) =1 FILTERING ALGORITHM COST-BASED FILTERING ALGORITHM

More information

EEL 6266 Power System Operation and Control. Chapter 3 Economic Dispatch Using Dynamic Programming

EEL 6266 Power System Operation and Control. Chapter 3 Economic Dispatch Using Dynamic Programming EEL 6266 Power System Operaton and Control Chapter 3 Economc Dspatch Usng Dynamc Programmng Pecewse Lnear Cost Functons Common practce many utltes prefer to represent ther generator cost functons as sngle-

More information

Amiri s Supply Chain Model. System Engineering b Department of Mathematics and Statistics c Odette School of Business

Amiri s Supply Chain Model. System Engineering b Department of Mathematics and Statistics c Odette School of Business Amr s Supply Chan Model by S. Ashtab a,, R.J. Caron b E. Selvarajah c a Department of Industral Manufacturng System Engneerng b Department of Mathematcs Statstcs c Odette School of Busness Unversty of

More information

Planning and Scheduling to Minimize Makespan & Tardiness. John Hooker Carnegie Mellon University September 2006

Planning and Scheduling to Minimize Makespan & Tardiness. John Hooker Carnegie Mellon University September 2006 Plannng and Schedulng to Mnmze Makespan & ardness John Hooker Carnege Mellon Unversty September 2006 he Problem Gven a set of tasks, each wth a deadlne 2 he Problem Gven a set of tasks, each wth a deadlne

More information

Kernel Methods and SVMs Extension

Kernel Methods and SVMs Extension Kernel Methods and SVMs Extenson The purpose of ths document s to revew materal covered n Machne Learnng 1 Supervsed Learnng regardng support vector machnes (SVMs). Ths document also provdes a general

More information

College of Computer & Information Science Fall 2009 Northeastern University 20 October 2009

College of Computer & Information Science Fall 2009 Northeastern University 20 October 2009 College of Computer & Informaton Scence Fall 2009 Northeastern Unversty 20 October 2009 CS7880: Algorthmc Power Tools Scrbe: Jan Wen and Laura Poplawsk Lecture Outlne: Prmal-dual schema Network Desgn:

More information

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017 U.C. Berkeley CS94: Beyond Worst-Case Analyss Handout 4s Luca Trevsan September 5, 07 Summary of Lecture 4 In whch we ntroduce semdefnte programmng and apply t to Max Cut. Semdefnte Programmng Recall that

More information

Yong Joon Ryang. 1. Introduction Consider the multicommodity transportation problem with convex quadratic cost function. 1 2 (x x0 ) T Q(x x 0 )

Yong Joon Ryang. 1. Introduction Consider the multicommodity transportation problem with convex quadratic cost function. 1 2 (x x0 ) T Q(x x 0 ) Kangweon-Kyungk Math. Jour. 4 1996), No. 1, pp. 7 16 AN ITERATIVE ROW-ACTION METHOD FOR MULTICOMMODITY TRANSPORTATION PROBLEMS Yong Joon Ryang Abstract. The optmzaton problems wth quadratc constrants often

More information

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers Psychology 282 Lecture #24 Outlne Regresson Dagnostcs: Outlers In an earler lecture we studed the statstcal assumptons underlyng the regresson model, ncludng the followng ponts: Formal statement of assumptons.

More information

Optimization Methods for Engineering Design. Logic-Based. John Hooker. Turkish Operational Research Society. Carnegie Mellon University

Optimization Methods for Engineering Design. Logic-Based. John Hooker. Turkish Operational Research Society. Carnegie Mellon University Logc-Based Optmzaton Methods for Engneerng Desgn John Hooker Carnege Mellon Unerst Turksh Operatonal Research Socet Ankara June 1999 Jont work wth: Srnas Bollapragada General Electrc R&D Omar Ghattas Cl

More information

Lecture 10 Support Vector Machines II

Lecture 10 Support Vector Machines II Lecture 10 Support Vector Machnes II 22 February 2016 Taylor B. Arnold Yale Statstcs STAT 365/665 1/28 Notes: Problem 3 s posted and due ths upcomng Frday There was an early bug n the fake-test data; fxed

More information

Computing Correlated Equilibria in Multi-Player Games

Computing Correlated Equilibria in Multi-Player Games Computng Correlated Equlbra n Mult-Player Games Chrstos H. Papadmtrou Presented by Zhanxang Huang December 7th, 2005 1 The Author Dr. Chrstos H. Papadmtrou CS professor at UC Berkley (taught at Harvard,

More information

Simultaneous Optimization of Berth Allocation, Quay Crane Assignment and Quay Crane Scheduling Problems in Container Terminals

Simultaneous Optimization of Berth Allocation, Quay Crane Assignment and Quay Crane Scheduling Problems in Container Terminals Smultaneous Optmzaton of Berth Allocaton, Quay Crane Assgnment and Quay Crane Schedulng Problems n Contaner Termnals Necat Aras, Yavuz Türkoğulları, Z. Caner Taşkın, Kuban Altınel Abstract In ths work,

More information

Support Vector Machines. Vibhav Gogate The University of Texas at dallas

Support Vector Machines. Vibhav Gogate The University of Texas at dallas Support Vector Machnes Vbhav Gogate he Unversty of exas at dallas What We have Learned So Far? 1. Decson rees. Naïve Bayes 3. Lnear Regresson 4. Logstc Regresson 5. Perceptron 6. Neural networks 7. K-Nearest

More information

Feature Selection: Part 1

Feature Selection: Part 1 CSE 546: Machne Learnng Lecture 5 Feature Selecton: Part 1 Instructor: Sham Kakade 1 Regresson n the hgh dmensonal settng How do we learn when the number of features d s greater than the sample sze n?

More information

Winter 2008 CS567 Stochastic Linear/Integer Programming Guest Lecturer: Xu, Huan

Winter 2008 CS567 Stochastic Linear/Integer Programming Guest Lecturer: Xu, Huan Wnter 2008 CS567 Stochastc Lnear/Integer Programmng Guest Lecturer: Xu, Huan Class 2: More Modelng Examples 1 Capacty Expanson Capacty expanson models optmal choces of the tmng and levels of nvestments

More information

Solutions to exam in SF1811 Optimization, Jan 14, 2015

Solutions to exam in SF1811 Optimization, Jan 14, 2015 Solutons to exam n SF8 Optmzaton, Jan 4, 25 3 3 O------O -4 \ / \ / The network: \/ where all lnks go from left to rght. /\ / \ / \ 6 O------O -5 2 4.(a) Let x = ( x 3, x 4, x 23, x 24 ) T, where the varable

More information

The Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction

The Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction ECONOMICS 5* -- NOTE (Summary) ECON 5* -- NOTE The Multple Classcal Lnear Regresson Model (CLRM): Specfcaton and Assumptons. Introducton CLRM stands for the Classcal Lnear Regresson Model. The CLRM s also

More information

A Search-Infer-and-Relax Framework for. Integrating Solution Methods. Carnegie Mellon University CPAIOR, May John Hooker

A Search-Infer-and-Relax Framework for. Integrating Solution Methods. Carnegie Mellon University CPAIOR, May John Hooker A Search-Infer-and-Rela Framework for Integratng Soluton Methods John Hooker Carnege Mellon Unversty CPAIOR, May 005 CPAIOR 005 Why ntegrate soluton methods? One-stop shoppng. One solver does t all. CPAIOR

More information

Modeling curves. Graphs: y = ax+b, y = sin(x) Implicit ax + by + c = 0, x 2 +y 2 =r 2 Parametric:

Modeling curves. Graphs: y = ax+b, y = sin(x) Implicit ax + by + c = 0, x 2 +y 2 =r 2 Parametric: Modelng curves Types of Curves Graphs: y = ax+b, y = sn(x) Implct ax + by + c = 0, x 2 +y 2 =r 2 Parametrc: x = ax + bxt x = cos t y = ay + byt y = snt Parametrc are the most common mplct are also used,

More information

A Modeling System to Combine Optimization and Constraint. Programming. INFORMS, November Carnegie Mellon University.

A Modeling System to Combine Optimization and Constraint. Programming. INFORMS, November Carnegie Mellon University. A Modelng Sstem to Combne Optmzaton and Constrant Programmng John Hooker Carnege Mellon Unverst INFORMS November 000 Based on ont work wth Ignaco Grossmann Hak-Jn Km Mara Axlo Osoro Greger Ottosson Erlendr

More information

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix Lectures - Week 4 Matrx norms, Condtonng, Vector Spaces, Lnear Independence, Spannng sets and Bass, Null space and Range of a Matrx Matrx Norms Now we turn to assocatng a number to each matrx. We could

More information

Lecture 10 Support Vector Machines. Oct

Lecture 10 Support Vector Machines. Oct Lecture 10 Support Vector Machnes Oct - 20-2008 Lnear Separators Whch of the lnear separators s optmal? Concept of Margn Recall that n Perceptron, we learned that the convergence rate of the Perceptron

More information

VQ widely used in coding speech, image, and video

VQ widely used in coding speech, image, and video at Scalar quantzers are specal cases of vector quantzers (VQ): they are constraned to look at one sample at a tme (memoryless) VQ does not have such constrant better RD perfomance expected Source codng

More information

1 Convex Optimization

1 Convex Optimization Convex Optmzaton We wll consder convex optmzaton problems. Namely, mnmzaton problems where the objectve s convex (we assume no constrants for now). Such problems often arse n machne learnng. For example,

More information

Assortment Optimization under MNL

Assortment Optimization under MNL Assortment Optmzaton under MNL Haotan Song Aprl 30, 2017 1 Introducton The assortment optmzaton problem ams to fnd the revenue-maxmzng assortment of products to offer when the prces of products are fxed.

More information

Linear Feature Engineering 11

Linear Feature Engineering 11 Lnear Feature Engneerng 11 2 Least-Squares 2.1 Smple least-squares Consder the followng dataset. We have a bunch of nputs x and correspondng outputs y. The partcular values n ths dataset are x y 0.23 0.19

More information

Least squares cubic splines without B-splines S.K. Lucas

Least squares cubic splines without B-splines S.K. Lucas Least squares cubc splnes wthout B-splnes S.K. Lucas School of Mathematcs and Statstcs, Unversty of South Australa, Mawson Lakes SA 595 e-mal: stephen.lucas@unsa.edu.au Submtted to the Gazette of the Australan

More information

10.34 Numerical Methods Applied to Chemical Engineering Fall Homework #3: Systems of Nonlinear Equations and Optimization

10.34 Numerical Methods Applied to Chemical Engineering Fall Homework #3: Systems of Nonlinear Equations and Optimization 10.34 Numercal Methods Appled to Chemcal Engneerng Fall 2015 Homework #3: Systems of Nonlnear Equatons and Optmzaton Problem 1 (30 ponts). A (homogeneous) azeotrope s a composton of a multcomponent mxture

More information

For now, let us focus on a specific model of neurons. These are simplified from reality but can achieve remarkable results.

For now, let us focus on a specific model of neurons. These are simplified from reality but can achieve remarkable results. Neural Networks : Dervaton compled by Alvn Wan from Professor Jtendra Malk s lecture Ths type of computaton s called deep learnng and s the most popular method for many problems, such as computer vson

More information

Outline and Reading. Dynamic Programming. Dynamic Programming revealed. Computing Fibonacci. The General Dynamic Programming Technique

Outline and Reading. Dynamic Programming. Dynamic Programming revealed. Computing Fibonacci. The General Dynamic Programming Technique Outlne and Readng Dynamc Programmng The General Technque ( 5.3.2) -1 Knapsac Problem ( 5.3.3) Matrx Chan-Product ( 5.3.1) Dynamc Programmng verson 1.4 1 Dynamc Programmng verson 1.4 2 Dynamc Programmng

More information

Interactive Bi-Level Multi-Objective Integer. Non-linear Programming Problem

Interactive Bi-Level Multi-Objective Integer. Non-linear Programming Problem Appled Mathematcal Scences Vol 5 0 no 65 3 33 Interactve B-Level Mult-Objectve Integer Non-lnear Programmng Problem O E Emam Department of Informaton Systems aculty of Computer Scence and nformaton Helwan

More information

The Minimum Universal Cost Flow in an Infeasible Flow Network

The Minimum Universal Cost Flow in an Infeasible Flow Network Journal of Scences, Islamc Republc of Iran 17(2): 175-180 (2006) Unversty of Tehran, ISSN 1016-1104 http://jscencesutacr The Mnmum Unversal Cost Flow n an Infeasble Flow Network H Saleh Fathabad * M Bagheran

More information

Notes on Frequency Estimation in Data Streams

Notes on Frequency Estimation in Data Streams Notes on Frequency Estmaton n Data Streams In (one of) the data streamng model(s), the data s a sequence of arrvals a 1, a 2,..., a m of the form a j = (, v) where s the dentty of the tem and belongs to

More information

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE Analytcal soluton s usually not possble when exctaton vares arbtrarly wth tme or f the system s nonlnear. Such problems can be solved by numercal tmesteppng

More information

Transfer Functions. Convenient representation of a linear, dynamic model. A transfer function (TF) relates one input and one output: ( ) system

Transfer Functions. Convenient representation of a linear, dynamic model. A transfer function (TF) relates one input and one output: ( ) system Transfer Functons Convenent representaton of a lnear, dynamc model. A transfer functon (TF) relates one nput and one output: x t X s y t system Y s The followng termnology s used: x y nput output forcng

More information

ECE559VV Project Report

ECE559VV Project Report ECE559VV Project Report (Supplementary Notes Loc Xuan Bu I. MAX SUM-RATE SCHEDULING: THE UPLINK CASE We have seen (n the presentaton that, for downlnk (broadcast channels, the strategy maxmzng the sum-rate

More information

Recent Developments in Disjunctive Programming

Recent Developments in Disjunctive Programming Recent Developments n Dsjunctve Programmng Aldo Vecchett (*) and Ignaco E. Grossmann (**) (*) INGAR Insttuto de Desarrollo Dseño Unversdad Tecnologca Naconal Santa Fe Argentna e-mal: aldovec@alpha.arcrde.edu.ar

More information

APPENDIX A Some Linear Algebra

APPENDIX A Some Linear Algebra APPENDIX A Some Lnear Algebra The collecton of m, n matrces A.1 Matrces a 1,1,..., a 1,n A = a m,1,..., a m,n wth real elements a,j s denoted by R m,n. If n = 1 then A s called a column vector. Smlarly,

More information

p 1 c 2 + p 2 c 2 + p 3 c p m c 2

p 1 c 2 + p 2 c 2 + p 3 c p m c 2 Where to put a faclty? Gven locatons p 1,..., p m n R n of m houses, want to choose a locaton c n R n for the fre staton. Want c to be as close as possble to all the house. We know how to measure dstance

More information

NP-Completeness : Proofs

NP-Completeness : Proofs NP-Completeness : Proofs Proof Methods A method to show a decson problem Π NP-complete s as follows. (1) Show Π NP. (2) Choose an NP-complete problem Π. (3) Show Π Π. A method to show an optmzaton problem

More information

Economics 101. Lecture 4 - Equilibrium and Efficiency

Economics 101. Lecture 4 - Equilibrium and Efficiency Economcs 0 Lecture 4 - Equlbrum and Effcency Intro As dscussed n the prevous lecture, we wll now move from an envronment where we looed at consumers mang decsons n solaton to analyzng economes full of

More information

COS 521: Advanced Algorithms Game Theory and Linear Programming

COS 521: Advanced Algorithms Game Theory and Linear Programming COS 521: Advanced Algorthms Game Theory and Lnear Programmng Moses Charkar February 27, 2013 In these notes, we ntroduce some basc concepts n game theory and lnear programmng (LP). We show a connecton

More information

Resource Allocation with a Budget Constraint for Computing Independent Tasks in the Cloud

Resource Allocation with a Budget Constraint for Computing Independent Tasks in the Cloud Resource Allocaton wth a Budget Constrant for Computng Independent Tasks n the Cloud Wemng Sh and Bo Hong School of Electrcal and Computer Engneerng Georga Insttute of Technology, USA 2nd IEEE Internatonal

More information

Solutions HW #2. minimize. Ax = b. Give the dual problem, and make the implicit equality constraints explicit. Solution.

Solutions HW #2. minimize. Ax = b. Give the dual problem, and make the implicit equality constraints explicit. Solution. Solutons HW #2 Dual of general LP. Fnd the dual functon of the LP mnmze subject to c T x Gx h Ax = b. Gve the dual problem, and make the mplct equalty constrants explct. Soluton. 1. The Lagrangan s L(x,

More information

More metrics on cartesian products

More metrics on cartesian products More metrcs on cartesan products If (X, d ) are metrc spaces for 1 n, then n Secton II4 of the lecture notes we defned three metrcs on X whose underlyng topologes are the product topology The purpose of

More information

An Integrated Asset Allocation and Path Planning Method to to Search for a Moving Target in in a Dynamic Environment

An Integrated Asset Allocation and Path Planning Method to to Search for a Moving Target in in a Dynamic Environment An Integrated Asset Allocaton and Path Plannng Method to to Search for a Movng Target n n a Dynamc Envronment Woosun An Mansha Mshra Chulwoo Park Prof. Krshna R. Pattpat Dept. of Electrcal and Computer

More information

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography CSc 6974 and ECSE 6966 Math. Tech. for Vson, Graphcs and Robotcs Lecture 21, Aprl 17, 2006 Estmatng A Plane Homography Overvew We contnue wth a dscusson of the major ssues, usng estmaton of plane projectve

More information

Incremental and Encoding Formulations for Mixed Integer Programming

Incremental and Encoding Formulations for Mixed Integer Programming Incremental and Encodng Formulatons for Mxed Integer Programmng Sercan Yıldız a, Juan Pablo Velma b,c, a Tepper School of Busness, Carnege Mellon Unversty, 5000 Forbes Ave., Pttsburgh, PA 15213, Unted

More information

Primer on High-Order Moment Estimators

Primer on High-Order Moment Estimators Prmer on Hgh-Order Moment Estmators Ton M. Whted July 2007 The Errors-n-Varables Model We wll start wth the classcal EIV for one msmeasured regressor. The general case s n Erckson and Whted Econometrc

More information

FUZZY GOAL PROGRAMMING VS ORDINARY FUZZY PROGRAMMING APPROACH FOR MULTI OBJECTIVE PROGRAMMING PROBLEM

FUZZY GOAL PROGRAMMING VS ORDINARY FUZZY PROGRAMMING APPROACH FOR MULTI OBJECTIVE PROGRAMMING PROBLEM Internatonal Conference on Ceramcs, Bkaner, Inda Internatonal Journal of Modern Physcs: Conference Seres Vol. 22 (2013) 757 761 World Scentfc Publshng Company DOI: 10.1142/S2010194513010982 FUZZY GOAL

More information

On a direct solver for linear least squares problems

On a direct solver for linear least squares problems ISSN 2066-6594 Ann. Acad. Rom. Sc. Ser. Math. Appl. Vol. 8, No. 2/2016 On a drect solver for lnear least squares problems Constantn Popa Abstract The Null Space (NS) algorthm s a drect solver for lnear

More information

On the Multicriteria Integer Network Flow Problem

On the Multicriteria Integer Network Flow Problem BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 5, No 2 Sofa 2005 On the Multcrtera Integer Network Flow Problem Vassl Vasslev, Marana Nkolova, Maryana Vassleva Insttute of

More information

An Admission Control Algorithm in Cloud Computing Systems

An Admission Control Algorithm in Cloud Computing Systems An Admsson Control Algorthm n Cloud Computng Systems Authors: Frank Yeong-Sung Ln Department of Informaton Management Natonal Tawan Unversty Tape, Tawan, R.O.C. ysln@m.ntu.edu.tw Yngje Lan Management Scence

More information

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems Numercal Analyss by Dr. Anta Pal Assstant Professor Department of Mathematcs Natonal Insttute of Technology Durgapur Durgapur-713209 emal: anta.bue@gmal.com 1 . Chapter 5 Soluton of System of Lnear Equatons

More information

Société de Calcul Mathématique SA

Société de Calcul Mathématique SA Socété de Calcul Mathématque SA Outls d'ade à la décson Tools for decson help Probablstc Studes: Normalzng the Hstograms Bernard Beauzamy December, 202 I. General constructon of the hstogram Any probablstc

More information

Module 2. Random Processes. Version 2 ECE IIT, Kharagpur

Module 2. Random Processes. Version 2 ECE IIT, Kharagpur Module Random Processes Lesson 6 Functons of Random Varables After readng ths lesson, ou wll learn about cdf of functon of a random varable. Formula for determnng the pdf of a random varable. Let, X be

More information

PHYS 450 Spring semester Lecture 02: Dealing with Experimental Uncertainties. Ron Reifenberger Birck Nanotechnology Center Purdue University

PHYS 450 Spring semester Lecture 02: Dealing with Experimental Uncertainties. Ron Reifenberger Birck Nanotechnology Center Purdue University PHYS 45 Sprng semester 7 Lecture : Dealng wth Expermental Uncertantes Ron Refenberger Brck anotechnology Center Purdue Unversty Lecture Introductory Comments Expermental errors (really expermental uncertantes)

More information

Lecture 12: Classification

Lecture 12: Classification Lecture : Classfcaton g Dscrmnant functons g The optmal Bayes classfer g Quadratc classfers g Eucldean and Mahalanobs metrcs g K Nearest Neghbor Classfers Intellgent Sensor Systems Rcardo Guterrez-Osuna

More information

Linear Approximation with Regularization and Moving Least Squares

Linear Approximation with Regularization and Moving Least Squares Lnear Approxmaton wth Regularzaton and Movng Least Squares Igor Grešovn May 007 Revson 4.6 (Revson : March 004). 5 4 3 0.5 3 3.5 4 Contents: Lnear Fttng...4. Weghted Least Squares n Functon Approxmaton...

More information

Report on Image warping

Report on Image warping Report on Image warpng Xuan Ne, Dec. 20, 2004 Ths document summarzed the algorthms of our mage warpng soluton for further study, and there s a detaled descrpton about the mplementaton of these algorthms.

More information

Exercises. 18 Algorithms

Exercises. 18 Algorithms 18 Algorthms Exercses 0.1. In each of the followng stuatons, ndcate whether f = O(g), or f = Ω(g), or both (n whch case f = Θ(g)). f(n) g(n) (a) n 100 n 200 (b) n 1/2 n 2/3 (c) 100n + log n n + (log n)

More information

Which Separator? Spring 1

Which Separator? Spring 1 Whch Separator? 6.034 - Sprng 1 Whch Separator? Mamze the margn to closest ponts 6.034 - Sprng Whch Separator? Mamze the margn to closest ponts 6.034 - Sprng 3 Margn of a pont " # y (w $ + b) proportonal

More information

Canonical transformations

Canonical transformations Canoncal transformatons November 23, 2014 Recall that we have defned a symplectc transformaton to be any lnear transformaton M A B leavng the symplectc form nvarant, Ω AB M A CM B DΩ CD Coordnate transformatons,

More information

An Interactive Optimisation Tool for Allocation Problems

An Interactive Optimisation Tool for Allocation Problems An Interactve Optmsaton ool for Allocaton Problems Fredr Bonäs, Joam Westerlund and apo Westerlund Process Desgn Laboratory, Faculty of echnology, Åbo Aadem Unversty, uru 20500, Fnland hs paper presents

More information

CS 3710: Visual Recognition Classification and Detection. Adriana Kovashka Department of Computer Science January 13, 2015

CS 3710: Visual Recognition Classification and Detection. Adriana Kovashka Department of Computer Science January 13, 2015 CS 3710: Vsual Recognton Classfcaton and Detecton Adrana Kovashka Department of Computer Scence January 13, 2015 Plan for Today Vsual recognton bascs part 2: Classfcaton and detecton Adrana s research

More information

Math Review. CptS 223 Advanced Data Structures. Larry Holder School of Electrical Engineering and Computer Science Washington State University

Math Review. CptS 223 Advanced Data Structures. Larry Holder School of Electrical Engineering and Computer Science Washington State University Math Revew CptS 223 dvanced Data Structures Larry Holder School of Electrcal Engneerng and Computer Scence Washngton State Unversty 1 Why do we need math n a data structures course? nalyzng data structures

More information

Lecture 5 Decoding Binary BCH Codes

Lecture 5 Decoding Binary BCH Codes Lecture 5 Decodng Bnary BCH Codes In ths class, we wll ntroduce dfferent methods for decodng BCH codes 51 Decodng the [15, 7, 5] 2 -BCH Code Consder the [15, 7, 5] 2 -code C we ntroduced n the last lecture

More information

Image classification. Given the bag-of-features representations of images from different classes, how do we learn a model for distinguishing i them?

Image classification. Given the bag-of-features representations of images from different classes, how do we learn a model for distinguishing i them? Image classfcaton Gven te bag-of-features representatons of mages from dfferent classes ow do we learn a model for dstngusng tem? Classfers Learn a decson rule assgnng bag-offeatures representatons of

More information

Nice plotting of proteins II

Nice plotting of proteins II Nce plottng of protens II Fnal remark regardng effcency: It s possble to wrte the Newton representaton n a way that can be computed effcently, usng smlar bracketng that we made for the frst representaton

More information

Problem Set 9 Solutions

Problem Set 9 Solutions Desgn and Analyss of Algorthms May 4, 2015 Massachusetts Insttute of Technology 6.046J/18.410J Profs. Erk Demane, Srn Devadas, and Nancy Lynch Problem Set 9 Solutons Problem Set 9 Solutons Ths problem

More information

Errors for Linear Systems

Errors for Linear Systems Errors for Lnear Systems When we solve a lnear system Ax b we often do not know A and b exactly, but have only approxmatons  and ˆb avalable. Then the best thng we can do s to solve ˆx ˆb exactly whch

More information

Lecture 21: Numerical methods for pricing American type derivatives

Lecture 21: Numerical methods for pricing American type derivatives Lecture 21: Numercal methods for prcng Amercan type dervatves Xaoguang Wang STAT 598W Aprl 10th, 2014 (STAT 598W) Lecture 21 1 / 26 Outlne 1 Fnte Dfference Method Explct Method Penalty Method (STAT 598W)

More information

CS 331 DESIGN AND ANALYSIS OF ALGORITHMS DYNAMIC PROGRAMMING. Dr. Daisy Tang

CS 331 DESIGN AND ANALYSIS OF ALGORITHMS DYNAMIC PROGRAMMING. Dr. Daisy Tang CS DESIGN ND NLYSIS OF LGORITHMS DYNMIC PROGRMMING Dr. Dasy Tang Dynamc Programmng Idea: Problems can be dvded nto stages Soluton s a sequence o decsons and the decson at the current stage s based on the

More information

Week 2. This week, we covered operations on sets and cardinality.

Week 2. This week, we covered operations on sets and cardinality. Week 2 Ths week, we covered operatons on sets and cardnalty. Defnton 0.1 (Correspondence). A correspondence between two sets A and B s a set S contaned n A B = {(a, b) a A, b B}. A correspondence from

More information

Lecture 20: Lift and Project, SDP Duality. Today we will study the Lift and Project method. Then we will prove the SDP duality theorem.

Lecture 20: Lift and Project, SDP Duality. Today we will study the Lift and Project method. Then we will prove the SDP duality theorem. prnceton u. sp 02 cos 598B: algorthms and complexty Lecture 20: Lft and Project, SDP Dualty Lecturer: Sanjeev Arora Scrbe:Yury Makarychev Today we wll study the Lft and Project method. Then we wll prove

More information

Structure and Drive Paul A. Jensen Copyright July 20, 2003

Structure and Drive Paul A. Jensen Copyright July 20, 2003 Structure and Drve Paul A. Jensen Copyrght July 20, 2003 A system s made up of several operatons wth flow passng between them. The structure of the system descrbes the flow paths from nputs to outputs.

More information

ExxonMobil. Juan Pablo Ruiz Ignacio E. Grossmann. Department of Chemical Engineering Center for Advanced Process Decision-making. Pittsburgh, PA 15213

ExxonMobil. Juan Pablo Ruiz Ignacio E. Grossmann. Department of Chemical Engineering Center for Advanced Process Decision-making. Pittsburgh, PA 15213 ExxonMobl Multperod Blend Schedulng Problem Juan Pablo Ruz Ignaco E. Grossmann Department of Chemcal Engneerng Center for Advanced Process Decson-makng Unversty Pttsburgh, PA 15213 1 Motvaton - Large cost

More information

Lecture 17: Lee-Sidford Barrier

Lecture 17: Lee-Sidford Barrier CSE 599: Interplay between Convex Optmzaton and Geometry Wnter 2018 Lecturer: Yn Tat Lee Lecture 17: Lee-Sdford Barrer Dsclamer: Please tell me any mstake you notced. In ths lecture, we talk about the

More information

4DVAR, according to the name, is a four-dimensional variational method.

4DVAR, according to the name, is a four-dimensional variational method. 4D-Varatonal Data Assmlaton (4D-Var) 4DVAR, accordng to the name, s a four-dmensonal varatonal method. 4D-Var s actually a drect generalzaton of 3D-Var to handle observatons that are dstrbuted n tme. The

More information

The Geometry of Logit and Probit

The Geometry of Logit and Probit The Geometry of Logt and Probt Ths short note s meant as a supplement to Chapters and 3 of Spatal Models of Parlamentary Votng and the notaton and reference to fgures n the text below s to those two chapters.

More information

princeton univ. F 17 cos 521: Advanced Algorithm Design Lecture 7: LP Duality Lecturer: Matt Weinberg

princeton univ. F 17 cos 521: Advanced Algorithm Design Lecture 7: LP Duality Lecturer: Matt Weinberg prnceton unv. F 17 cos 521: Advanced Algorthm Desgn Lecture 7: LP Dualty Lecturer: Matt Wenberg Scrbe: LP Dualty s an extremely useful tool for analyzng structural propertes of lnear programs. Whle there

More information

Calculation of time complexity (3%)

Calculation of time complexity (3%) Problem 1. (30%) Calculaton of tme complexty (3%) Gven n ctes, usng exhaust search to see every result takes O(n!). Calculaton of tme needed to solve the problem (2%) 40 ctes:40! dfferent tours 40 add

More information

Linear Classification, SVMs and Nearest Neighbors

Linear Classification, SVMs and Nearest Neighbors 1 CSE 473 Lecture 25 (Chapter 18) Lnear Classfcaton, SVMs and Nearest Neghbors CSE AI faculty + Chrs Bshop, Dan Klen, Stuart Russell, Andrew Moore Motvaton: Face Detecton How do we buld a classfer to dstngush

More information

PHYS 705: Classical Mechanics. Calculus of Variations II

PHYS 705: Classical Mechanics. Calculus of Variations II 1 PHYS 705: Classcal Mechancs Calculus of Varatons II 2 Calculus of Varatons: Generalzaton (no constrant yet) Suppose now that F depends on several dependent varables : We need to fnd such that has a statonary

More information

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity LINEAR REGRESSION ANALYSIS MODULE IX Lecture - 30 Multcollnearty Dr. Shalabh Department of Mathematcs and Statstcs Indan Insttute of Technology Kanpur 2 Remedes for multcollnearty Varous technques have

More information

Robust observed-state feedback design. for discrete-time systems rational in the uncertainties

Robust observed-state feedback design. for discrete-time systems rational in the uncertainties Robust observed-state feedback desgn for dscrete-tme systems ratonal n the uncertantes Dmtr Peaucelle Yosho Ebhara & Yohe Hosoe Semnar at Kolloquum Technsche Kybernetk, May 10, 016 Unversty of Stuttgart

More information

Line Drawing and Clipping Week 1, Lecture 2

Line Drawing and Clipping Week 1, Lecture 2 CS 43 Computer Graphcs I Lne Drawng and Clppng Week, Lecture 2 Davd Breen, Wllam Regl and Maxm Peysakhov Geometrc and Intellgent Computng Laboratory Department of Computer Scence Drexel Unversty http://gcl.mcs.drexel.edu

More information

Chapter 8 Indicator Variables

Chapter 8 Indicator Variables Chapter 8 Indcator Varables In general, e explanatory varables n any regresson analyss are assumed to be quanttatve n nature. For example, e varables lke temperature, dstance, age etc. are quanttatve n

More information

An efficient algorithm for multivariate Maclaurin Newton transformation

An efficient algorithm for multivariate Maclaurin Newton transformation Annales UMCS Informatca AI VIII, 2 2008) 5 14 DOI: 10.2478/v10065-008-0020-6 An effcent algorthm for multvarate Maclaurn Newton transformaton Joanna Kapusta Insttute of Mathematcs and Computer Scence,

More information

Stanford University CS359G: Graph Partitioning and Expanders Handout 4 Luca Trevisan January 13, 2011

Stanford University CS359G: Graph Partitioning and Expanders Handout 4 Luca Trevisan January 13, 2011 Stanford Unversty CS359G: Graph Parttonng and Expanders Handout 4 Luca Trevsan January 3, 0 Lecture 4 In whch we prove the dffcult drecton of Cheeger s nequalty. As n the past lectures, consder an undrected

More information

Portfolios with Trading Constraints and Payout Restrictions

Portfolios with Trading Constraints and Payout Restrictions Portfolos wth Tradng Constrants and Payout Restrctons John R. Brge Northwestern Unversty (ont wor wth Chrs Donohue Xaodong Xu and Gongyun Zhao) 1 General Problem (Very) long-term nvestor (eample: unversty

More information

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity LINEAR REGRESSION ANALYSIS MODULE IX Lecture - 31 Multcollnearty Dr. Shalabh Department of Mathematcs and Statstcs Indan Insttute of Technology Kanpur 6. Rdge regresson The OLSE s the best lnear unbased

More information

Finite Mixture Models and Expectation Maximization. Most slides are from: Dr. Mario Figueiredo, Dr. Anil Jain and Dr. Rong Jin

Finite Mixture Models and Expectation Maximization. Most slides are from: Dr. Mario Figueiredo, Dr. Anil Jain and Dr. Rong Jin Fnte Mxture Models and Expectaton Maxmzaton Most sldes are from: Dr. Maro Fgueredo, Dr. Anl Jan and Dr. Rong Jn Recall: The Supervsed Learnng Problem Gven a set of n samples X {(x, y )},,,n Chapter 3 of

More information

Chapter 4 The Wave Equation

Chapter 4 The Wave Equation Chapter 4 The Wave Equaton Another classcal example of a hyperbolc PDE s a wave equaton. The wave equaton s a second-order lnear hyperbolc PDE that descrbes the propagaton of a varety of waves, such as

More information

Integrated approach in solving parallel machine scheduling and location (ScheLoc) problem

Integrated approach in solving parallel machine scheduling and location (ScheLoc) problem Internatonal Journal of Industral Engneerng Computatons 7 (2016) 573 584 Contents lsts avalable at GrowngScence Internatonal Journal of Industral Engneerng Computatons homepage: www.growngscence.com/ec

More information

One Dimension Again. Chapter Fourteen

One Dimension Again. Chapter Fourteen hapter Fourteen One Dmenson Agan 4 Scalar Lne Integrals Now we agan consder the dea of the ntegral n one dmenson When we were ntroduced to the ntegral back n elementary school, we consdered only functons

More information