Global Constraints for the Mean Aboslute Deviation and the Variance: Application to the Vertical Line Balancing

Size: px
Start display at page:

Download "Global Constraints for the Mean Aboslute Deviation and the Variance: Application to the Vertical Line Balancing"

Transcription

1 1/51 Global Constraints for the Mean Aboslute Deviation and the Variance: Application to the Vertical Line Balancing Pierre Schaus and Yves Deville, Jean-Charles Régin also collaborated Belgian Constraints UCLouvain January 16, 2008

2 2/51 Overview 1 Line Balancing Problems 2 Introduction to CP 3 Global Constraints for Variance and Mean Absolute Deviation 4 Conclusion

3 3/51 Assembly Line of the Ford-T

4 An Assembly Line is... A set of stations or working stations (WS) achieving part of assembly operations of a product. linked such that the product transit is is possible, (partial product of complete product). Taken from slides Lignes d assemblages, Conception, Organisation et gestion (Thomas L Eglise and Benoit Raucent). 4/51

5 5/51 Precedence graph (PG): DAG Taken from slides Lignes d assemblages, Conception, Organisation et gestion (Thomas L Eglise and Benoit Raucent).

6 6/51 Precedence Graph cont. Each node is a task and each task comes with a duration to achieve it.

7 7/51 Assembly Line at work Each time cycle, one product leaves the line. T c is the time between the ending of two product at the end of the line.

8 8/51 Our objective: Assign each task to a working station such that all precedences are satisfied

9 9/51 SALBP-2: Maximize The Cadence of the line The Cyle Time is fixed by the slowest WS: max(x i ) where X i is the sum of processing time of tasks in WS i. Optimization problem: Find a schedule minimizing max(x i ).

10 10/51 Vertical Line Balacing: Equalize the duration of each working stations. s = n i=1 X i is the sum of the durations of all the tasks. n is the number of working stations. Minimization of: n i=1 X i s/n n i=1 (X i s/n) 2 max{x 1...X n } max{x 1...X n } min{x 1...X n } max{ X i s/n } {X i X i s/n}...

11 11/51 Vertical Line Balacing: Equalize the duration of each working stations. In this presentation we study the 3 first: n i=1 X i s/n n i=1 (X i s/n) 2 max{x 1...X n } A 90-year-old debate:the advantages of the mean deviation by STEPHEN GORARD, University of York British Journal of Educational Studies, December 2005

12 12/51 Study of the Three Criteria on Hahn instance from Scholl Benchmark 53 Tasks, min duration of 40, max duration of 1775.

13 13/51 No criterion subsumes the others... Minimization of: Measure of: max i {X i } i (X i s/n) 2 /n i X i s/n /n max i {X i } 2336 (+3%)2400 (+4%)2418 i(x i s/n) 2 /n (+53%) (+9%)738 i X i s/n /n (+34%)298 (+0%) but i (X i s/n) 2 /n approximates very well the other two.

14 14/51 Consider now that each tasks can have a stochastic time following a gaussian distribution N d i, p d i t i d i

15 15/51 The height of each bin also follows a gaussian distribution N d tot, p d tot t 3 d 3 d tot t 2 d 2 t 1 d 1

16 The height of each bin also follows a gaussian distribution X 1 N 1 X 2 N 2 X 3 N 3 X 4 N 4 X 5 N 5 X 6 N 6 X 7 N 7 The cycle time depends on the distribution of Y = max{x 1, X 2,...} 16/51

17 17/51 Finding the distribution of Y = max{x 1, X 2,...}. Cumulative distribution: F Y (y) = P(Y y) = P(X 1 y) P(X 2 y)... P(X n y) The probability density function f Y (y) = d dy F Y (y)

18 Comparison of the 3 solutions with i : µ = d i, σ i = 0.3 d i f Y(y) max{x i } i 2e 04 4e 04 6e 04 8e 04 n (Xi s n) 2 n 1 n Xi s n n Expectation of the maximum cycle time is smaller for the solution minimizing the variance. 18/51

19 19/51 In Conclusion for the 3 criteria n i=1 (X i s/n) 2 approximates the best the other two criteria. The minimization of n i=1 (X i s/n) 2 or n i=1 X i s/n gives more robust solutions when the distribution of the tasks is stochastic (often the case when achieved by humans).

20 20/51 Introduction to Constraint Programming 1 Line Balancing Problems 2 Introduction to CP 3 Global Constraints for Variance and Mean Absolute Deviation 4 Conclusion

21 21/51 Some definitions of Constraint Programming A Constraint Satisfaction Problem (CSP) consist of: a set of variables X = {X 1,..., X n }, for each variable X i, a finite set Dom(X i ) of possible values (its domain), and a set of constraints restricting the values that the variables can simultaneously take. A solution to a CSP is an assignment of a value from its domain to every variable, in such a way that every constraint is satisfied. We may want to find: just one solution, with no preference as to which one, all solutions, an optimal, or at least a good solution, given some objective function defined in terms of some or all of the variables.

22 22/51 Constraint Each constraint is responsible to remove as much inconsistent values as possible of the domains. Example X 1 {1, 2, 5, 6}, X 2 {0, 2} a constraint C := X 1 X 2. C can remove {1, 2, 5/, 6/} {0/, 2, 4, 7} We say that a constraint filters the domains of the variables. The algorithm responsible to filter the domains for a constraint is called a propagator.

23 23/51 Searching for a solution: branching and propagation Dom(X1)={1,3} Dom(X2)={2,3,5} Dom(X3)={2,3,4} X1>X2 X2 X3 X2 X3 X1 X2 X1 X3

24 24/51 Searching for a solution: branching and propagation Dom(X1)={1,3} Dom(X2)={2,3,5} Dom(X3)={2,3,4} X1>X2 X2 X3 X2 X3 X1 X2 X1 X3

25 25/51 Searching for a solution: branching and propagation Dom(X1)=3 Dom(X2)={2,3,5} Dom(X3)={2,3,4} X1>X2 X2 X3 X2 X3 X1 X2 X1 X3

26 26/51 Searching for a solution: branching and propagation X2=2 Dom(X1)=3 Dom(X2)={2,5} Dom(X3)={2,4} X1>X2 X2 X3 X2 X3 X1 X2 X1 X3 Dom(X1)=3 Dom(X2)=2 Dom(X3)={2,4}

27 27/51 Searching for a solution: branching and propagation X2=2 Dom(X1)=3 Dom(X2)={2,5} Dom(X3)={2,4} X1>X2 X2 X3 X2 X3 X1 X2 X1 X3 Dom(X1)=3 Dom(X2)=2 Dom(X3)={2,4}

28 28/51 Searching for a solution: branching and propagation X2=2 Dom(X1)=3 Dom(X2)={2,5} Dom(X3)={2,4} X2 2 X1>X2 X2 X3 X2 X3 X1 X2 X1 X3

29 29/51 Searching for a solution: branching and propagation X2=2 Dom(X1)=3 Dom(X2)={2,5} Dom(X3)={2,4} X2 2 Dom(X1)=3 Dom(X2)=5 Dom(X3)={2,4} X1>X2 X2 X3 X2 X3 X1 X2 X1 X3

30 30/51 Different levels of consistency reached by a propagator C Dom(Y) C C Dom(X)

31 Different levels of consistency reached by a propagator: Arc-Consistency and Bound-Consistency C C Dom(Y) C C C C Dom(X) Dom(X) For some constraints, arc consistency is too costly e.g:x X n = S For others it is possible very efficiently e.g.:alldifferent(x 1,..., X n ) 31/51

32 32/51 Global Constraints and decomposition Some constraints can be decomposed into simpler constraints. e.g. AllDiff (X 1, X 2, X 3 ) is equivalent to X 1 X 2, X 1 X 3, X 2 X 3. But it is often better to consider the constraint globally to filter more values. Example Dom(X 1 ) = {1, 2}, Dom(X 2 ) = {1, 2}, Dom(X 3 ) = {1, 2, 3}. X 1 X 2, X 1 X 3, X 2 X 3 are separately arc consistent (nothing to filter). AllDiff (X 1, X 2, X 3 ) is able to filter Dom(X 3 ) = {1/, 2/, 3}

33 33/51 Modeling Line Balancing in CP Variables: ws j is the workstation of task j. b ij {0, 1}, b ij = 1 if task j scheduled at WS i. b ij = 0 otherwise. l i is the load of WS i Constraints: Reification: b ij = 1 ws j = i l i = i b ij d j where d j is the duration of task j. j b ij = 1 (each task is scheduled only once). ws j ws k if (j, k) is an edge of PG. Our three possible Objectives: min max i {l i }. min (l i ( l i )/n) 2 ). min l i ( l i )/n.

34 34/51 Improved model An easy but important redundant constraint: l i = d j i j Redundant global constraints for the bin packing: IloPack A constraint for bin packing, Paul Shaw (CP2004). IloPack+Precedences

35 35/51 Two global constraints: spread and deviation to realize our objective functions Given X = (X 1,..., X n ), one integer value s, one finite domain variable (1), (2) deviation(x, s, (1) ) holds if and only if n X i = s and (1) i=1 n n X i s. i=1 spread(x, s, (2) ) holds if and only if n n X i = s and (2) n ( Xi 2 ) s 2. i=1 i=1

36 36/51 Two global constraints: spread and deviation deviation(x, s, (1) ) holds if and only if n X i = s and (1) i=1 n n X i s. i=1 spread(x, s, (2) ) holds if and only if n n X i = s and (2) n ( Xi 2 ) s 2. i=1 It means that spread and deviation can be expressed with two simple sum constraints (decomposition). i=1 Is it really necessary to have a global constraint? Will the filtering be stronger than with the decomposition?

37 37/51 Implementation of deviation by decomposition Dom(X 1 ) = Dom(X 2 ) = [0, 10] Dom( (1) ) = [0, 8] deviation ({X 1, X 2 }, s = 10, (1) ) 10 = X 1 + X 2 and 2 X X 2 10 (1)

38 38/51 Bound-consistency on 10 = X 1 + X 2 The constraint is already bound-consistent. Nothing to prune. 10 Dom X 2 5 X 1 X 2 =10 5 Dom X 1 10

39 39/51 BC on 2 X X 2 10 (1) Dom( (1) ) = [0, 8] 10 9 Dom X 2 5 X 1 5 X Dom X 1 10

40 40/51 Now, deviation is bound-consistent 10 9 X 1 5 X X 1 X 2 =10 7 Dom X Dom X 1

41 41/51 Efficient algorithms to achieve bound-consistency deviation(x, s, (1) ) Bound consistency achieved in O(n). spread(x, s, (2) ) Bound consistency achieved in O(n log n) Time constraint, we only present the filtering of (2) for spread.

42 42/51 Filtering of Dom( (2) ) in spread(x, s, (2) ) n Q (2) = min{n (x[i] s/n) 2 x i=1 s.t. and n x[i] = s i=1 Xi min x[i] X max i i} Filtering is: Dom( (2) ) Dom( (2) ) [ Q (2), + ) Question: How to compute efficiently Q (2)?

43 43/51 Important theorem characterizing optimal solution n Q (2) = min{n (x[i] s/n) 2 x i=1 s.t. and n x[i] = s i=1 Xi min x[i] X max i Optimal solution is a ν-centered assignment [Régin et al. ] i}

44 44/51 Important theorem characterizing optimal solution Yes but the constraint n ν-centered assignment. i=1 x[i] = s must also be satisfied by the =? x[i]/ n s/n So we have to find ν such that the red line equals the green line.

45 45/51 Finding the ν-centered assignment of sum s 1 Consider a splitting of at most n 1 intervals based on the bounds of the domains. 2 For each interval, determine if there exists a ν-centered assignment of sum s with ν inside the interval. I 11 I 10 I 9 I 8 I 7 I 6 5 I 4 I 3 I 2 I 1

46 46/51 Is there a ν-centered assignment of sum s with ν I 7? ES I 7 =. I 11 I 10 I 9 I 8 I 7 I 6 5 I 4 I 3 I 2 I 1 ν must satisfy: 3 ν + ES(I 7 ) = s Answer yes if s ES(I 7) 3 I 7. Otherwise ν / I 7, try another interval.

47 47/51 Complexity to filter (2) in spread(x, s, (2) ) O(n) once intervals I are known and for each ES(I ) + the number of overlapping domains. Intervals I, ES(I ) and number of overlapping variables for each I can be computed in O(n) once the bounds of variables are sorted (not so trivial). O(n log n) to sort the bound. Hence overal complexity is dominated by O(n log n) to sort the bounds. Filtering of all X i X in O(n log n) (not explained here).

48 48/51 Complexity to filter (1) in deviation(x, s, (1) ) All domains can be filtered in O(n) (not explained here).

49 49/51 Implementation and results We have implemented both propagators in Ilog Solver and we give some results. Backtracks Time(s) #Tasks #stations Max Spread Devi Max Spread Devi Buxey Hahn Wee-Mag Lutz Mulkherje

50 50/51 Conclusion We have shown that the common criterion used in line balancing that is minimization of max{x 1...X n } was not the best to equalize the loads of stations. We have show that when the tasks durations are stochastic, minimizing n i=1 (X i s/n) 2 obtain solutions with an even smaller expected cycle time than solutions minimizing max{x 1...X n }. We have introduced two global constraints and their propagators for n i=1 (X i s/n) 2 and n i=1 X i s/n respectively spread and deviation. We have implemented the propagators in Ilog and tested it on real line balancing problems.

51 51/51 Thank you for your attention!

Constraint Programming and Graph Algorithms

Constraint Programming and Graph Algorithms Constraint Programming and Graph Algorithms Kurt Mehlhorn Max-Planck-Institute for Computer Science (with significant help from Sven Thiel) I am not an expert on the subject (four publications), but I

More information

Global constraints and decompositions

Global constraints and decompositions Global constraints and decompositions Christian Bessiere Univ Montpellier, CNRS (joined work with E. Hebrard, B. Hnich, G. Katsirelos, Z. Kiziltan, N. Narodytska C.G. Quimper, T. Walsh) Outline Background

More information

Notes on Complexity of the Simple Assembly Line Balancing Problem

Notes on Complexity of the Simple Assembly Line Balancing Problem Notes on Complexity of the Simple Assembly Line Balancing Problem Lazarev A.A., Gafarov E.R. Institute of Control Sciences of the Russian Academy of Sciences, Profsoyuznaya st. 65, 117997 Moscow, Russia

More information

Projection, Inference, and Consistency

Projection, Inference, and Consistency Projection, Inference, and Consistency John Hooker Carnegie Mellon University IJCAI 2016, New York City A high-level presentation. Don t worry about the details. 2 Projection as a Unifying Concept Projection

More information

Chapter 6 Constraint Satisfaction Problems

Chapter 6 Constraint Satisfaction Problems Chapter 6 Constraint Satisfaction Problems CS5811 - Artificial Intelligence Nilufer Onder Department of Computer Science Michigan Technological University Outline CSP problem definition Backtracking search

More information

A Global Constraint for Parallelizing the Execution of Task Sets in Non-Preemptive Scheduling

A Global Constraint for Parallelizing the Execution of Task Sets in Non-Preemptive Scheduling A Global Constraint for Parallelizing the Execution of Task Sets in Non-Preemptive Scheduling 1 Introduction Michael Marte 1 Institut für Informatik, Universität München Oettingenstr. 67, 80538 München,

More information

The core of solving constraint problems using Constraint Programming (CP), with emphasis on:

The core of solving constraint problems using Constraint Programming (CP), with emphasis on: What is it about? l Theory The core of solving constraint problems using Constraint Programming (CP), with emphasis on: l Modeling l Solving: Local consistency and propagation; Backtracking search + heuristics.

More information

Alternative Methods for Obtaining. Optimization Bounds. AFOSR Program Review, April Carnegie Mellon University. Grant FA

Alternative Methods for Obtaining. Optimization Bounds. AFOSR Program Review, April Carnegie Mellon University. Grant FA Alternative Methods for Obtaining Optimization Bounds J. N. Hooker Carnegie Mellon University AFOSR Program Review, April 2012 Grant FA9550-11-1-0180 Integrating OR and CP/AI Early support by AFOSR First

More information

Announcements. Solution to Assignment 3 is posted Assignment 4 is available. c D. Poole and A. Mackworth 2017 CPSC 322 Lecture 8 1 / 25

Announcements. Solution to Assignment 3 is posted Assignment 4 is available. c D. Poole and A. Mackworth 2017 CPSC 322 Lecture 8 1 / 25 Announcements Solution to Assignment 3 is posted Assignment 4 is available c D. Poole and A. Mackworth 2017 CPSC 322 Lecture 8 1 / 25 Review: So far... Constraint satisfaction problems defined in terms

More information

Scheduling with Constraint Programming. Job Shop Cumulative Job Shop

Scheduling with Constraint Programming. Job Shop Cumulative Job Shop Scheduling with Constraint Programming Job Shop Cumulative Job Shop CP vs MIP: Task Sequencing We need to sequence a set of tasks on a machine Each task i has a specific fixed processing time p i Each

More information

Generic Adaptive Heuristics for Large Neighborhood Search

Generic Adaptive Heuristics for Large Neighborhood Search Generic Adaptive Heuristics for Large Neighborhood Search Jean-Baptiste Mairy 1, Pierre Schaus 2, and Yves Deville 1 1 Université Catholique de Louvain, Belgium 2 Dynadec Europe, Belgium Abstract. The

More information

Decision Diagrams for Discrete Optimization

Decision Diagrams for Discrete Optimization Decision Diagrams for Discrete Optimization Willem Jan van Hoeve Tepper School of Business Carnegie Mellon University www.andrew.cmu.edu/user/vanhoeve/mdd/ Acknowledgments: David Bergman, Andre Cire, Samid

More information

Introduction to Arti Intelligence

Introduction to Arti Intelligence Introduction to Arti Intelligence cial Lecture 4: Constraint satisfaction problems 1 / 48 Constraint satisfaction problems: Today Exploiting the representation of a state to accelerate search. Backtracking.

More information

Decomposition of the NVALUE constraint

Decomposition of the NVALUE constraint Decomposition of the NVALUE constraint Christian Bessiere 1, George Katsirelos 2, Nina Narodytska 3, Claude-Guy Quimper 4, and Toby Walsh 3 1 LIRMM, CNRS, Montpellier, email: bessiere@lirmm.fr 2 CRIL-CNRS,

More information

Mathematics for Decision Making: An Introduction. Lecture 13

Mathematics for Decision Making: An Introduction. Lecture 13 Mathematics for Decision Making: An Introduction Lecture 13 Matthias Köppe UC Davis, Mathematics February 17, 2009 13 1 Reminder: Flows in networks General structure: Flows in networks In general, consider

More information

Stochastic Decision Diagrams

Stochastic Decision Diagrams Stochastic Decision Diagrams John Hooker CORS/INFORMS Montréal June 2015 Objective Relaxed decision diagrams provide an generalpurpose method for discrete optimization. When the problem has a dynamic programming

More information

Boosting Set Constraint Propagation for Network Design

Boosting Set Constraint Propagation for Network Design Boosting Set Constraint Propagation for Network Design Justin Yip 1, Pascal Van Hentenryck 1, and Carmen Gervet 2 1 Brown University, Box 1910, Providence, RI 02912, USA 2 German University in Cairo, New

More information

Symmetry Breaking Ordering Constraints

Symmetry Breaking Ordering Constraints Symmetry Breaking Ordering Constraints PhD dissertation Uppsala University, Sweden, March 2004 Zeynep KIZILTAN Università di Bologna, Italy URL: http://lia.deis.unibo.it/~zk/ Overview Brief Background

More information

Algorithms. NP -Complete Problems. Dong Kyue Kim Hanyang University

Algorithms. NP -Complete Problems. Dong Kyue Kim Hanyang University Algorithms NP -Complete Problems Dong Kyue Kim Hanyang University dqkim@hanyang.ac.kr The Class P Definition 13.2 Polynomially bounded An algorithm is said to be polynomially bounded if its worst-case

More information

arxiv:cs/ v2 [cs.dm] 21 Aug 2001

arxiv:cs/ v2 [cs.dm] 21 Aug 2001 Solving Assembly Line Balancing Problems by Combining IP and CP Alexander Bockmayr and Nicolai Pisaruk arxiv:cs/0106002v2 [cs.dm] 21 Aug 2001 Université Henri Poincaré, LORIA B.P. 239, F-54506 Vandœuvre-lès-Nancy,

More information

Probabilistic Graphical Models

Probabilistic Graphical Models Probabilistic Graphical Models David Sontag New York University Lecture 6, March 7, 2013 David Sontag (NYU) Graphical Models Lecture 6, March 7, 2013 1 / 25 Today s lecture 1 Dual decomposition 2 MAP inference

More information

SOLVING FINITE-DOMAIN LINEAR CONSTRAINTS IN PRESENCE OF THE ALLDIFFERENT

SOLVING FINITE-DOMAIN LINEAR CONSTRAINTS IN PRESENCE OF THE ALLDIFFERENT Logical Methods in Computer Science Vol. 12(3:5)2016, pp. 1 28 www.lmcs-online.org Submitted Jul. 31, 2015 Published Sep. 5, 2016 SOLVING FINITE-DOMAIN LINEAR CONSTRAINTS IN PRESENCE OF THE ALLDIFFERENT

More information

Decision Diagrams for Sequencing and Scheduling

Decision Diagrams for Sequencing and Scheduling Decision Diagrams for Sequencing and Scheduling Willem-Jan van Hoeve Tepper School of Business Carnegie Mellon University www.andrew.cmu.edu/user/vanhoeve/mdd/ Plan What can MDDs do for Combinatorial Optimization?

More information

Introduction to Bin Packing Problems

Introduction to Bin Packing Problems Introduction to Bin Packing Problems Fabio Furini March 13, 2015 Outline Origins and applications Applications: Definition: Bin Packing Problem (BPP) Solution techniques for the BPP Heuristic Algorithms

More information

Decision Procedures An Algorithmic Point of View

Decision Procedures An Algorithmic Point of View An Algorithmic Point of View ILP References: Integer Programming / Laurence Wolsey Deciding ILPs with Branch & Bound Intro. To mathematical programming / Hillier, Lieberman Daniel Kroening and Ofer Strichman

More information

Consistency as Projection

Consistency as Projection Consistency as Projection John Hooker Carnegie Mellon University INFORMS 2015, Philadelphia USA Consistency as Projection Reconceive consistency in constraint programming as a form of projection. For eample,

More information

Projection in Logic, CP, and Optimization

Projection in Logic, CP, and Optimization Projection in Logic, CP, and Optimization John Hooker Carnegie Mellon University Workshop on Logic and Search Melbourne, 2017 Projection as a Unifying Concept Projection is a fundamental concept in logic,

More information

A CP Approach to the Balanced Academic Curriculum Problem

A CP Approach to the Balanced Academic Curriculum Problem A CP Approach to the Balanced Academic Curriculum Problem Jean-Noël Monette, Pierre Schaus, Stéphane Zampelli, Yves Deville and Pierre Dupont Department of Computing Sciences and Engineering Université

More information

Mathematical formulation for a mixed-model assembly line balancing problem with stochastic processing times

Mathematical formulation for a mixed-model assembly line balancing problem with stochastic processing times Fourth LACCEI International Latin American and Caribbean Conference for Engineering and Technology (LACCET 2006) Breaing Frontiers and Barriers in Engineering: Education, Research and Practice,2-23 June

More information

Interleaved Alldifferent Constraints: CSP vs. SAT Approaches

Interleaved Alldifferent Constraints: CSP vs. SAT Approaches Interleaved Alldifferent Constraints: CSP vs. SAT Approaches Frédéric Lardeux 3, Eric Monfroy 1,2, and Frédéric Saubion 3 1 Universidad Técnica Federico Santa María, Valparaíso, Chile 2 LINA, Université

More information

CMSC 722, AI Planning. Planning and Scheduling

CMSC 722, AI Planning. Planning and Scheduling CMSC 722, AI Planning Planning and Scheduling Dana S. Nau University of Maryland 1:26 PM April 24, 2012 1 Scheduling Given: actions to perform set of resources to use time constraints» e.g., the ones computed

More information

BATCH PROCESSING WITH SEQUENCE DEPENDENT SETUP TIMES: NEW RESULTS

BATCH PROCESSING WITH SEQUENCE DEPENDENT SETUP TIMES: NEW RESULTS BATCH PROCESSING WITH SEQUENCE DEPENDENT SETUP TIMES: NEW RESULTS PETR VILÍM Charles University, Faculty of Mathematics and Physics, Malostranské náměstí 2/25, Praha 1, Czech Republic, e-mail: vilim@kti.mff.cuni.cz

More information

Continuously Degrading Resource and Interval Dependent Activity Durations in Nuclear Medicine Patient Scheduling

Continuously Degrading Resource and Interval Dependent Activity Durations in Nuclear Medicine Patient Scheduling Continuously Degrading Resource and Interval Dependent Activity Durations in Nuclear Medicine Patient Scheduling Cyrille Dejemeppe and Yves Deville ICTEAM, Université Catholique de Louvain (UCLouvain),

More information

Constraint Technology for Solving Combinatorial Problems

Constraint Technology for Solving Combinatorial Problems First Exam for Course 1DL023: Constraint Technology for Solving Combinatorial Problems Autumn 2009, Uppsala University, Sweden Prepared by Pierre Flener Wednesday 21 October 2009, from 08:00 to 13:00,

More information

Discrete (and Continuous) Optimization WI4 131

Discrete (and Continuous) Optimization WI4 131 Discrete (and Continuous) Optimization WI4 131 Kees Roos Technische Universiteit Delft Faculteit Electrotechniek, Wiskunde en Informatica Afdeling Informatie, Systemen en Algoritmiek e-mail: C.Roos@ewi.tudelft.nl

More information

n! (k 1)!(n k)! = F (X) U(0, 1). (x, y) = n(n 1) ( F (y) F (x) ) n 2

n! (k 1)!(n k)! = F (X) U(0, 1). (x, y) = n(n 1) ( F (y) F (x) ) n 2 Order statistics Ex. 4. (*. Let independent variables X,..., X n have U(0, distribution. Show that for every x (0,, we have P ( X ( < x and P ( X (n > x as n. Ex. 4.2 (**. By using induction or otherwise,

More information

Hierarchical Boosting and Filter Generation

Hierarchical Boosting and Filter Generation January 29, 2007 Plan Combining Classifiers Boosting Neural Network Structure of AdaBoost Image processing Hierarchical Boosting Hierarchical Structure Filters Combining Classifiers Combining Classifiers

More information

CPSC 540: Machine Learning

CPSC 540: Machine Learning CPSC 540: Machine Learning Undirected Graphical Models Mark Schmidt University of British Columbia Winter 2016 Admin Assignment 3: 2 late days to hand it in today, Thursday is final day. Assignment 4:

More information

Combining Symmetry Breaking with Other Constraints: lexicographic ordering with sums

Combining Symmetry Breaking with Other Constraints: lexicographic ordering with sums Combining Symmetry Breaking with Other Constraints: lexicographic ordering with sums Brahim Hnich 1, Zeynep Kiziltan 2, and Toby Walsh 1 1 Cork Constraint Computation Center, University College Cork, Ireland.

More information

Projection, Consistency, and George Boole

Projection, Consistency, and George Boole Projection, Consistency, and George Boole John Hooker Carnegie Mellon University CP 2015, Cork, Ireland Projection as a Unifying Concept Projection underlies both optimization and logical inference. Optimization

More information

Structure Learning: the good, the bad, the ugly

Structure Learning: the good, the bad, the ugly Readings: K&F: 15.1, 15.2, 15.3, 15.4, 15.5 Structure Learning: the good, the bad, the ugly Graphical Models 10708 Carlos Guestrin Carnegie Mellon University September 29 th, 2006 1 Understanding the uniform

More information

How to Relax. CP 2008 Slide 1. John Hooker Carnegie Mellon University September 2008

How to Relax. CP 2008 Slide 1. John Hooker Carnegie Mellon University September 2008 How to Relax Slide 1 John Hooker Carnegie Mellon University September 2008 Two ways to relax Relax your mind and body. Relax your problem formulations. Slide 2 Relaxing a problem Feasible set of original

More information

The Balance Constraint Family

The Balance Constraint Family The Balance Constraint Family Christian Bessiere 1, Emmanuel Hebrard 2, George Katsirelos 3, Zeynep Kiziltan 4, Émilie Picard-Cantin 5, Claude-Guy Quimper 5, and Toby Walsh 6 1 CNRS, University of Montpellier,

More information

Machine Learning for Software Engineering

Machine Learning for Software Engineering Machine Learning for Software Engineering Dimensionality Reduction Prof. Dr.-Ing. Norbert Siegmund Intelligent Software Systems 1 2 Exam Info Scheduled for Tuesday 25 th of July 11-13h (same time as the

More information

PART 4 INTEGER PROGRAMMING

PART 4 INTEGER PROGRAMMING PART 4 INTEGER PROGRAMMING 102 Read Chapters 11 and 12 in textbook 103 A capital budgeting problem We want to invest $19 000 Four investment opportunities which cannot be split (take it or leave it) 1.

More information

Integer Programming, Constraint Programming, and their Combination

Integer Programming, Constraint Programming, and their Combination Integer Programming, Constraint Programming, and their Combination Alexander Bockmayr Freie Universität Berlin & DFG Research Center Matheon Eindhoven, 27 January 2006 Discrete Optimization General framework

More information

What we will try to do...

What we will try to do... Constraint Satisfaction Problems (CSP) (Where we postpone making difficult decisions until they become easy to make) R&N: Chap. 5 Slides from Jean-Claude Latombe at Stanford University (used with permission)

More information

Section 8.1. Vector Notation

Section 8.1. Vector Notation Section 8.1 Vector Notation Definition 8.1 Random Vector A random vector is a column vector X = [ X 1 ]. X n Each Xi is a random variable. Definition 8.2 Vector Sample Value A sample value of a random

More information

Empirical Risk Minimization is an incomplete inductive principle Thomas P. Minka

Empirical Risk Minimization is an incomplete inductive principle Thomas P. Minka Empirical Risk Minimization is an incomplete inductive principle Thomas P. Minka February 20, 2001 Abstract Empirical Risk Minimization (ERM) only utilizes the loss function defined for the task and is

More information

Integer Programming Methods LNMB

Integer Programming Methods LNMB Integer Programming Methods LNMB 2017 2018 Dion Gijswijt homepage.tudelft.nl/64a8q/intpm/ Dion Gijswijt Intro IntPM 2017-2018 1 / 24 Organisation Webpage: homepage.tudelft.nl/64a8q/intpm/ Book: Integer

More information

Search and Lookahead. Bernhard Nebel, Julien Hué, and Stefan Wölfl. June 4/6, 2012

Search and Lookahead. Bernhard Nebel, Julien Hué, and Stefan Wölfl. June 4/6, 2012 Search and Lookahead Bernhard Nebel, Julien Hué, and Stefan Wölfl Albert-Ludwigs-Universität Freiburg June 4/6, 2012 Search and Lookahead Enforcing consistency is one way of solving constraint networks:

More information

Revisiting the Limits of MAP Inference by MWSS on Perfect Graphs

Revisiting the Limits of MAP Inference by MWSS on Perfect Graphs Revisiting the Limits of MAP Inference by MWSS on Perfect Graphs Adrian Weller University of Cambridge CP 2015 Cork, Ireland Slides and full paper at http://mlg.eng.cam.ac.uk/adrian/ 1 / 21 Motivation:

More information

Message-Passing Algorithms for GMRFs and Non-Linear Optimization

Message-Passing Algorithms for GMRFs and Non-Linear Optimization Message-Passing Algorithms for GMRFs and Non-Linear Optimization Jason Johnson Joint Work with Dmitry Malioutov, Venkat Chandrasekaran and Alan Willsky Stochastic Systems Group, MIT NIPS Workshop: Approximate

More information

Inference and Representation

Inference and Representation Inference and Representation David Sontag New York University Lecture 5, Sept. 30, 2014 David Sontag (NYU) Inference and Representation Lecture 5, Sept. 30, 2014 1 / 16 Today s lecture 1 Running-time of

More information

Integer Linear Programming

Integer Linear Programming Integer Linear Programming Solution : cutting planes and Branch and Bound Hugues Talbot Laboratoire CVN April 13, 2018 IP Resolution Gomory s cutting planes Solution branch-and-bound General method Resolution

More information

Stochastic Constraint Programming: a seamless modeling framework for decision making under uncertainty

Stochastic Constraint Programming: a seamless modeling framework for decision making under uncertainty Stochastic Constraint Programming: a seamless modeling framework for decision making under uncertainty Dr Roberto Rossi 1 1 LDI, Wageningen UR, the Netherlands Mansholt Lecture Research A Complete Overview

More information

11/8/2018. Overview. PERT / CPM Part 2

11/8/2018. Overview. PERT / CPM Part 2 /8/08 PERT / CPM Part BSAD 0 Dave Novak Fall 08 Source: Anderson et al., 0 Quantitative Methods for Business th edition some slides are directly from J. Loucks 0 Cengage Learning Overview Last class introduce

More information

A conjecture on the alphabet size needed to produce all correlation classes of pairs of words

A conjecture on the alphabet size needed to produce all correlation classes of pairs of words A conjecture on the alphabet size needed to produce all correlation classes of pairs of words Paul Leopardi Thanks: Jörg Arndt, Michael Barnsley, Richard Brent, Sylvain Forêt, Judy-anne Osborn. Mathematical

More information

ahmaxsat: Description and Evaluation of a Branch and Bound Max-SAT Solver

ahmaxsat: Description and Evaluation of a Branch and Bound Max-SAT Solver Journal on Satisfiability, Boolean Modeling and Computation 9 (2015) 89-128 ahmaxsat: Description and Evaluation of a Branch and Bound Max-SAT Solver André Abramé Djamal Habet Aix Marseille Université,

More information

The two-dimensional bin-packing problem is the problem of orthogonally packing a given set of rectangles

The two-dimensional bin-packing problem is the problem of orthogonally packing a given set of rectangles INFORMS Journal on Computing Vol. 19, No. 1, Winter 2007, pp. 36 51 issn 1091-9856 eissn 1526-5528 07 1901 0036 informs doi 10.1287/ijoc.1060.0181 2007 INFORMS Using Decomposition Techniques and Constraint

More information

Learning Gaussian Graphical Models with Unknown Group Sparsity

Learning Gaussian Graphical Models with Unknown Group Sparsity Learning Gaussian Graphical Models with Unknown Group Sparsity Kevin Murphy Ben Marlin Depts. of Statistics & Computer Science Univ. British Columbia Canada Connections Graphical models Density estimation

More information

Backtracking search: look-back. ICS 275 Spring 2014

Backtracking search: look-back. ICS 275 Spring 2014 Backtracking search: look-back ICS 275 Look-back: Backjumping / Learning Backjumping: In deadends, go back to the most recent culprit. Learning: constraint-recording, nogood recording. good-recording Backjumping

More information

Image Segmentation Using Minimum st Cut

Image Segmentation Using Minimum st Cut Image Segmentation Using Minimum st Cut Tomas Werner Center for Machine Perception Czech Technical University Prague Minimum st Cut 2 Undirected graph (V, ) with nodes v V and edges vv ( ) V 2 very edge

More information

Introduction. In this talk

Introduction. In this talk Introduction Scarab Constraint Model Advanced Solving System Architecture and Implementation of a Prototyping Tool for SAT-based Constraint Programming Systems Takehide Soh 1, Naoyuki Tamura 1, Mutsunori

More information

n! (k 1)!(n k)! = F (X) U(0, 1). (x, y) = n(n 1) ( F (y) F (x) ) n 2

n! (k 1)!(n k)! = F (X) U(0, 1). (x, y) = n(n 1) ( F (y) F (x) ) n 2 Order statistics Ex. 4.1 (*. Let independent variables X 1,..., X n have U(0, 1 distribution. Show that for every x (0, 1, we have P ( X (1 < x 1 and P ( X (n > x 1 as n. Ex. 4.2 (**. By using induction

More information

Algorithms for propagating resource constraints in AI planning and scheduling: Existing approaches and new results

Algorithms for propagating resource constraints in AI planning and scheduling: Existing approaches and new results Artificial Intelligence 143 (2003) 151 188 www.elsevier.com/locate/artint Algorithms for propagating resource constraints in AI planning and scheduling: Existing approaches and new results Philippe Laborie

More information

Random Variables and Their Distributions

Random Variables and Their Distributions Chapter 3 Random Variables and Their Distributions A random variable (r.v.) is a function that assigns one and only one numerical value to each simple event in an experiment. We will denote r.vs by capital

More information

Lecture 11 October 11, Information Dissemination through Social Networks

Lecture 11 October 11, Information Dissemination through Social Networks CS 284r: Incentives and Information in Networks Fall 2013 Prof. Yaron Singer Lecture 11 October 11, 2013 Scribe: Michael Tingley, K. Nathaniel Tucker 1 Overview In today s lecture we will start the second

More information

is called an integer programming (IP) problem. model is called a mixed integer programming (MIP)

is called an integer programming (IP) problem. model is called a mixed integer programming (MIP) INTEGER PROGRAMMING Integer Programming g In many problems the decision variables must have integer values. Example: assign people, machines, and vehicles to activities in integer quantities. If this is

More information

Information Theory, Statistics, and Decision Trees

Information Theory, Statistics, and Decision Trees Information Theory, Statistics, and Decision Trees Léon Bottou COS 424 4/6/2010 Summary 1. Basic information theory. 2. Decision trees. 3. Information theory and statistics. Léon Bottou 2/31 COS 424 4/6/2010

More information

Introduction to Natural Computation. Lecture 9. Multilayer Perceptrons and Backpropagation. Peter Lewis

Introduction to Natural Computation. Lecture 9. Multilayer Perceptrons and Backpropagation. Peter Lewis Introduction to Natural Computation Lecture 9 Multilayer Perceptrons and Backpropagation Peter Lewis 1 / 25 Overview of the Lecture Why multilayer perceptrons? Some applications of multilayer perceptrons.

More information

Integer vs. constraint programming. IP vs. CP: Language

Integer vs. constraint programming. IP vs. CP: Language Discrete Math for Bioinformatics WS 0/, by A. Bockmayr/K. Reinert,. Januar 0, 0:6 00 Integer vs. constraint programming Practical Problem Solving Model building: Language Model solving: Algorithms IP vs.

More information

Basics of Multivariate Modelling and Data Analysis

Basics of Multivariate Modelling and Data Analysis Basics of Multivariate Modelling and Data Analysis Kurt-Erik Häggblom 6. Principal component analysis (PCA) 6.1 Overview 6.2 Essentials of PCA 6.3 Numerical calculation of PCs 6.4 Effects of data preprocessing

More information

Machine Learning. Part 1. Linear Regression. Machine Learning: Regression Case. .. Dennis Sun DATA 401 Data Science Alex Dekhtyar..

Machine Learning. Part 1. Linear Regression. Machine Learning: Regression Case. .. Dennis Sun DATA 401 Data Science Alex Dekhtyar.. .. Dennis Sun DATA 401 Data Science Alex Dekhtyar.. Machine Learning. Part 1. Linear Regression Machine Learning: Regression Case. Dataset. Consider a collection of features X = {X 1,...,X n }, such that

More information

A Quadratic Propagator for the Inter-Distance Constraint

A Quadratic Propagator for the Inter-Distance Constraint Constraint Programming Letters 3 (2008) 21 35 Submitted 11/2006; Published 2/2008 A Quadratic Propagator for the Inter-Distance Constraint Claude-Guy Quimper QUIMPER@ALUMNI.UWATERLOO.CA University of Waterloo

More information

Lecture 7: Edge Detection

Lecture 7: Edge Detection #1 Lecture 7: Edge Detection Saad J Bedros sbedros@umn.edu Review From Last Lecture Definition of an Edge First Order Derivative Approximation as Edge Detector #2 This Lecture Examples of Edge Detection

More information

Adding Flexibility to Russian Doll Search

Adding Flexibility to Russian Doll Search Adding Flexibility to Russian Doll Search Margarita Razgon and Gregory M. Provan Department of Computer Science, University College Cork, Ireland {m.razgon g.provan}@cs.ucc.ie Abstract The Weighted Constraint

More information

Optimization Bounds from Binary Decision Diagrams

Optimization Bounds from Binary Decision Diagrams Optimization Bounds from Binary Decision Diagrams J. N. Hooker Joint work with David Bergman, André Ciré, Willem van Hoeve Carnegie Mellon University ICS 203 Binary Decision Diagrams BDDs historically

More information

Completion Time of Fuzzy GERT-type Networks with Loops

Completion Time of Fuzzy GERT-type Networks with Loops Completion Time of Fuzzy GERT-type Networks with Loops Sina Ghaffari 1, Seyed Saeid Hashemin 2 1 Department of Industrial Engineering, Ardabil Branch, Islamic Azad university, Ardabil, Iran 2 Department

More information

Lecture 7 Random Signal Analysis

Lecture 7 Random Signal Analysis Lecture 7 Random Signal Analysis 7. Introduction to Probability 7. Amplitude Distributions 7.3 Uniform, Gaussian, and Other Distributions 7.4 Power and Power Density Spectra 7.5 Properties of the Power

More information

Logic, Optimization and Data Analytics

Logic, Optimization and Data Analytics Logic, Optimization and Data Analytics John Hooker Carnegie Mellon University United Technologies Research Center, Cork, Ireland August 2015 Thesis Logic and optimization have an underlying unity. Ideas

More information

Modelling for Constraint Programming

Modelling for Constraint Programming Modelling for Constraint Programming Barbara Smith 3. Symmetry, Viewpoints Symmetry in CSPs A symmetry transforms any solution into another Sometimes symmetry is inherent in the problem (e.g. chessboard

More information

IEOR E4570: Machine Learning for OR&FE Spring 2015 c 2015 by Martin Haugh. The EM Algorithm

IEOR E4570: Machine Learning for OR&FE Spring 2015 c 2015 by Martin Haugh. The EM Algorithm IEOR E4570: Machine Learning for OR&FE Spring 205 c 205 by Martin Haugh The EM Algorithm The EM algorithm is used for obtaining maximum likelihood estimates of parameters when some of the data is missing.

More information

CS360 Homework 12 Solution

CS360 Homework 12 Solution CS360 Homework 12 Solution Constraint Satisfaction 1) Consider the following constraint satisfaction problem with variables x, y and z, each with domain {1, 2, 3}, and constraints C 1 and C 2, defined

More information

Energy-efficient scheduling

Energy-efficient scheduling Energy-efficient scheduling Guillaume Aupy 1, Anne Benoit 1,2, Paul Renaud-Goud 1 and Yves Robert 1,2,3 1. Ecole Normale Supérieure de Lyon, France 2. Institut Universitaire de France 3. University of

More information

Chris Bishop s PRML Ch. 8: Graphical Models

Chris Bishop s PRML Ch. 8: Graphical Models Chris Bishop s PRML Ch. 8: Graphical Models January 24, 2008 Introduction Visualize the structure of a probabilistic model Design and motivate new models Insights into the model s properties, in particular

More information

Math 5490 Network Flows

Math 5490 Network Flows Math 90 Network Flows Lecture 8: Flow Decomposition Algorithm Stephen Billups University of Colorado at Denver Math 90Network Flows p./6 Flow Decomposition Algorithms Two approaches to modeling network

More information

Functional Dependencies and Normalization

Functional Dependencies and Normalization Functional Dependencies and Normalization There are many forms of constraints on relational database schemata other than key dependencies. Undoubtedly most important is the functional dependency. A functional

More information

MVE165/MMG630, Applied Optimization Lecture 6 Integer linear programming: models and applications; complexity. Ann-Brith Strömberg

MVE165/MMG630, Applied Optimization Lecture 6 Integer linear programming: models and applications; complexity. Ann-Brith Strömberg MVE165/MMG630, Integer linear programming: models and applications; complexity Ann-Brith Strömberg 2011 04 01 Modelling with integer variables (Ch. 13.1) Variables Linear programming (LP) uses continuous

More information

UCLA STAT 233 Statistical Methods in Biomedical Imaging

UCLA STAT 233 Statistical Methods in Biomedical Imaging UCLA STAT 233 Statistical Methods in Biomedical Imaging Instructor: Ivo Dinov, Asst. Prof. In Statistics and Neurology University of California, Los Angeles, Spring 2004 http://www.stat.ucla.edu/~dinov/

More information

CPSC 540: Machine Learning

CPSC 540: Machine Learning CPSC 540: Machine Learning Mark Schmidt University of British Columbia Winter 2019 Last Time: Monte Carlo Methods If we want to approximate expectations of random functions, E[g(x)] = g(x)p(x) or E[g(x)]

More information

Recoverable Robustness in Scheduling Problems

Recoverable Robustness in Scheduling Problems Master Thesis Computing Science Recoverable Robustness in Scheduling Problems Author: J.M.J. Stoef (3470997) J.M.J.Stoef@uu.nl Supervisors: dr. J.A. Hoogeveen J.A.Hoogeveen@uu.nl dr. ir. J.M. van den Akker

More information

Simple Linear Regression: The Model

Simple Linear Regression: The Model Simple Linear Regression: The Model task: quantifying the effect of change X in X on Y, with some constant β 1 : Y = β 1 X, linear relationship between X and Y, however, relationship subject to a random

More information

Probabilistic Graphical Networks: Definitions and Basic Results

Probabilistic Graphical Networks: Definitions and Basic Results This document gives a cursory overview of Probabilistic Graphical Networks. The material has been gleaned from different sources. I make no claim to original authorship of this material. Bayesian Graphical

More information

A Framework for Integrating Optimization and Constraint Programming

A Framework for Integrating Optimization and Constraint Programming A Framework for Integrating Optimization and Constraint Programming John Hooker Carnegie Mellon University SARA 2007 SARA 07 Slide Underlying theme Model vs. solution method. How can we match the model

More information

Mock Exam Künstliche Intelligenz-1. Different problems test different skills and knowledge, so do not get stuck on one problem.

Mock Exam Künstliche Intelligenz-1. Different problems test different skills and knowledge, so do not get stuck on one problem. Name: Matriculation Number: Mock Exam Künstliche Intelligenz-1 January 9., 2017 You have one hour(sharp) for the test; Write the solutions to the sheet. The estimated time for solving this exam is 53 minutes,

More information

Chapter 4 Deliberation with Temporal Domain Models

Chapter 4 Deliberation with Temporal Domain Models Last update: April 10, 2018 Chapter 4 Deliberation with Temporal Domain Models Section 4.4: Constraint Management Section 4.5: Acting with Temporal Models Automated Planning and Acting Malik Ghallab, Dana

More information

Module 7-2 Decomposition Approach

Module 7-2 Decomposition Approach Module 7-2 Decomposition Approach Chanan Singh Texas A&M University Decomposition Approach l Now we will describe a method of decomposing the state space into subsets for the purpose of calculating the

More information

An Energy Cost Aware Cumulative

An Energy Cost Aware Cumulative An Energy Cost Aware Cumulative Helmut Simonis and Tarik Hadzic Cork Constraint Computation Centre Computer Science Department University College Cork Ireland Modref 200, St. Andrews Helmut Simonis and

More information

Readings: K&F: 16.3, 16.4, Graphical Models Carlos Guestrin Carnegie Mellon University October 6 th, 2008

Readings: K&F: 16.3, 16.4, Graphical Models Carlos Guestrin Carnegie Mellon University October 6 th, 2008 Readings: K&F: 16.3, 16.4, 17.3 Bayesian Param. Learning Bayesian Structure Learning Graphical Models 10708 Carlos Guestrin Carnegie Mellon University October 6 th, 2008 10-708 Carlos Guestrin 2006-2008

More information