1 / 52

The Rules of Modelling Automatic Generation of Constraint Programs through Refinement

The Rules of Modelling Automatic Generation of Constraint Programs through Refinement. Artificial Intelligence Group Dept of Computer Science University of York. School of Computer Science University of St Andrews. Alan M. Frisch Ian Miguel

mercer
Download Presentation

The Rules of Modelling Automatic Generation of Constraint Programs through Refinement

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. The Rules of ModellingAutomatic Generation of Constraint Programs through Refinement Artificial Intelligence Group Dept of Computer Science University of York School of Computer Science University of St Andrews Alan M. Frisch Ian Miguel Joint work with Matt Grum, Chris Jefferson & Bernadette Martinez Hernandez

  2. Allocating a Divisible Resource:An Application to Grid Computing Alan Frisch Artificial Intelligence Group Dept. Computer Science University of York 4pm, CS103

  3. Modelling Bottleneck • Useful for solving a wide range of important, complex problems including scheduling, allocation, layout, configuration,… • Modelling a problem as a constraint program requires moderate/great expertise. • Major barrier to widespread use of CP.

  4. What are CP researchers doing? • Model Generation: • Done in private. • Model Selection: • Case studies • Some generalisations from analysis & experiments

  5. Reducing the Modelling Bottleneck:Automated Modelling • 2 to 3 years: We have been generating models more and more systematically. • Embed systematic viewpoint in automated system (CONJURE). • Takes specification in high-level language (ESSENCE). • Reformulates specification into a set of correct constraint programs. • Ongoing: add heuristics for model selection.

  6. Outline • Reduced constraint programming. • ESSENCE. • CONJURE. • The big picture.

  7. Part IReduced Constraint Programming

  8. What is Constraint Programming? • Broad view: programming in which constraints play a central role. • Narrow view: solve combinatorial (optimisation) problem by • mapping it to a constraint satisfaction problem • solving the constraint satisfaction problem • mapping solution back to original problem

  9. What is the (finite domain) CSP? An instance comprises: • Finite set of variables • Each associated with a finite domain • Finite set of constraints on the values taken by the variables • (Objective function) Solution is assignment of values to variables that satisfies the constraints (and optimises objective function)

  10. Modelling: Just Do It! Find 3 distinct non-zero digits that sum to 9.

  11. Just Do It! Find 3 distinct non-zero digits that sum to 9. • Variables: X, Y, Z

  12. Just Do It! Find 3 distinct non-zero digits that sum to 9. • Variables/Domains: X, Y, Z::{1,..9}

  13. Just Do It! Find 3 distinct non-zero digits that sum to 9. • Variables/Domains: X, Y, Z::{1,..9} • Constraints: X  Y, Y Z, X Z, X+Y+Z=9

  14. Just Do It! Find 3 distinct non-zero digits that sum to 9. • Variables/Domains: X, Y, Z :: {1..9} • Constraints: X  Y, Y Z, X Z, X+Y+Z=9 X  Y  Z

  15. Just Do It! Find 3 non-zero digits that sum to 9. • Variables/Domains: X, Y, Z :: {1..9} • Constraints: X  Y, Y Z, X Z, X+Y+Z=9 X  Y  ZX < Y < Z

  16. Just Do It for a Problem Given n and s, find n distinct non-zero digits that sum to s.

  17. Model: Explicit Representation Given n and s, find n distinct non-zero digits that sum to s. … 1 2 3 n X … 1..9 1..9 1..9 1..9 n:nat, s:natX: matrix (indexed by 1.. n) of 1..9 given find such that AllDiff(X)Si1..n X[i] = s ?

  18. Model: Occurrence Representation Given n and s, find n distinct non-zero digits that sum to s. … 1 2 3 9 … D 0/1 0/1 0/1 0/1 n:nat, s:natD: matrix (indexed by 1..9) of 0..1? given find such that D[1] + 2D[2] + … + 9D[9] = sD[1] + D[2] + … + D[9] = n

  19. The SONET Problem Specification • Given nrings rings, nnodes nodes, a set of pairs of nodes (communication demand) and an integer capacity (of each ring). Install nodes on rings satisfying demand and capacity constraints. Minimise installations. Instance • nrings=2, nnodes=5, capacity = 4 • demand: n1 & n3, n1 & n4, n2 & n3, n2 & n4, n3 & n5 Solution n3 n1 n2 n5 n4 n3

  20. A Model of SONET Problem rings-nodes Rings 0..1 0..1 0..1 0..1 0..1 0..1 0..1 0..1 Nodes rings-nodes[r,n] = 1 iff node n is installed on ring r. 0..1 0..1 0..1 0..1 0..1 0..1 0..1 0..1 0..1 0..1 0..1 0..1 0..1 0..1 0..1 0..1

  21. A Model of SONET Problem rings-nodes Rings 0..1 0..1 0..1 0..1 0..1 0..1 0..1 0..1 Nodes ScalarProduct > 0 0..1 0..1 0..1 0..1 0..1 0..1 0..1 0..1 Minimise: SS rings-nodes[r,n] 0..1 0..1 0..1 0..1 0..1 0..1 0..1 0..1 n  Nodes r  Rings … S c S c

  22. Part II Abstract Problem Specifications The ESSENCE Language

  23. ESSENCE • The language enables problems to be specified at a level of abstraction above that at which modelling decisions are made. • This requires features not found in current constraint programming languages.

  24. SONET Specification nrings:nat, nnodes:nat, capacity:nat Nodes be 1..nnodes demand: set of set (size 2) of Nodes rings: mset (size nrings) of set (maxsize capacity) of Nodes Sr  rings .|r| pair  demand . r  rings . pair  r given letting given find minimising such that

  25. ESSENCE Provides Abstract TypesSets, Multisets, Partitions, Range Types, Unnamed TypesFunctions, Relations, Enumerated Types, Sequences… nrings:nat, nnodes:nat, capacity:nat Nodes be 1..nnodes demand: set of set (size 2) of Nodes rings: mset (size nrings) of set (maxsize capacity) of Nodes Sr  rings .|r| pair  demand . r  rings . pair  r given letting given find minimising such that Abstract types

  26. ESSENCE SupportsArbitrarily-Nested Types Arbitrary nesting of types nrings:nat, nnodes:nat, capacity:nat Nodes be 1..nnodes demand: set of set (size 2) of Nodes rings: mset (size nrings) of set (maxsize capacity) of Nodes Sr  rings .|r| pair  demand . r  rings . pair  r given letting given find minimising such that

  27. ESSENCE Supports Quantification over Decision Variables nrings:nat, nnodes:nat, capacity:nat Nodes be 1..nnodes demand: set of set (size 2) of Nodes rings: mset (size nrings) of set (maxsize capacity) of Nodes Sr  rings .|r| pair  demand . r  rings . pair  r given letting given find minimising such that Quantifying Over Decision Variables

  28. How Usable is ESSENCE? • Specifications of ~50 problems found in the CSP literature written by an undergraduate with no background in constraint programming. • URL: http://www.cs.york.ac.uk/aig/constraints/

  29. ESSENCE’ Abstract types • ESSENCE’ has a similar level of abstraction to existing constraint languages. Arbitrary nesting of types ESSENCE’ = ESSENCE - Quantifying Over Decision Variables

  30. Model Generation: Informal Given n and s, find n distinct non-zero digits that sum to s. ESSENCE’ n:nat, s:natX: matrix (indexed by 1.. n) of 1..9 given find such that AllDiff(X)Si1..n X[i] = s ESSENCE’ n:nat, s:natD: matrix (indexed by 1..9) of 0..1 given find such that D[1] + 2D[2] + … + 9D[9] = sD[1] + D[2] + … + D[9] = n

  31. Formalisation of Model Generation n:nat, s:intX: set (size n) of 1..9 given find S x = s such that x  X ESSENCE’ n:nat, s:natX: matrix (indexed by 1.. n) of 1..9 given find such that AllDiff(X)Si1..n X[i] = s ESSENCE’ n:nat, s:natD: matrix (indexed by 1..9) of 0..1 given find such that D[1] + 2D[2] + … + 9D[9] = sD[1] + D[2] + … + D[9] = n

  32. Part III Generating Models: The CONJURE System

  33. A Theory of Model Generation Must Account for • Representing complex decision variables. • SONET involves finding a multiset of sets. • Exploiting channelling. • Generate, maintain multiple representations. • Identifying and breaking symmetries. • X+Y+Z = c  X <= Y <= Z • Performing transformations to improve efficiency. • X<=Y<=Z, AllDiff(X,Y,Z)  X < Y < Z

  34. CONJURE • CONJURE automatically generates a set of alternative ESSENCE’ models from ESSENCE specifications. • Core: set of refinement rules. • Refine an ESSENCE expression into a set of ESSENCE’ expressions. • Produce a set because often many alternative refinements of an ESSENCE expression.

  35. Rules Must be Compositional • Compositional refinement: • Refinement of an expression is composed of the refinements of its sub-expressions. • Necessary to handle: • Unbounded nesting of constraints (usual) • |A  (B  C) | < X • Unbounded nesting of types (unique) • set, set of sets, set of set of sets, …

  36. Challenge: Compositionality is Difficult • Complex variables: can have multiple refinements. Not all support all operations. • Nested variables: refinement of outer type must be determined without looking arbitrarily deep into the nesting. • Operators: refinement depends on refinement of operands. • Operators: recursion needed to handle arbitrary nesting of operands, e.g. 

  37. Refinement via r Operator • Function mapping ESSENCE expression to a set of ESSENCE’ expressions. • Defined by set of equalities. • One per syntactic construct. • For easy presentation, each equation broken into clauses: • RuleName1r(e)  e1’ • RuleName2r(e)  e2’ • … • RuleNamenr(e)  en’

  38. Refining a Specification • Refine constraints & objective function in turn. • Compose results to form final model. • Multiple refinements composed = multiple models. MicroSonet: nnodes:nat, capacity:nat given letting find such that Nodes be 1..nnodes ring:set (size capacity) of Nodes 4  ring 4  ring:set (maxsize capacity) of Nodes

  39. Refining MicroSonet nnodes:nat, capacity:nat given letting find such that Nodes be 1..nnodes • Refinement begins with r(4  ring): • Relevant clause of Element rule refines both sides, composes the results. ring:set (size capacity) of Nodes 4 ring  Compositionalityat work r(4) = {4} r(ring)

  40. { S’ S’=genSym(S,matrix(indxd by lb..ub)of bool) such that sum(S’)=n | } true • Generates a 1-d matrix, S’ Indexed by lb..ub S= n … 0/1 0/1 0/1 0/1 0/1 FixedSizeSet1: Occurrence Rep FixedSizeSet1r(S: set (size n) of lb..ub )  • Associate an axiom schema with this rule: • i t. iS S’ [i]

  41. ESSENCE’Occ Model: MicroSonet nnodes:nat, capacity:nat given letting find such that Nodes be 1..nnodes ESSENCE ring:set (size capacity) of Nodes 4 ring nnodes:nat, capacity:nat given letting find such that Nodes be 1..nnodes ESSENCE’ ring’ : matrix (indexed by Nodes) of bool sum(ring’ )=capacity ring’ [4]=1 1 2 3 nnodes ring’ 0/1 0/1 0/1 … 0/1

  42. { S’’ S’= genSym(S, matrix (indexed by 1..n) of t) such that f | S’’ : matrix (indexed by 1..n) of _ fr(AllDiff(S’ )) } S’’r(S’ ) • Generates intermediate 1-d matrix, S’ : … 1 2 3 n S’ t t t t … AllDiff(S’ ) • S’ refined to give fully refined matrix S’’ • All-different constraint refined, added to local model. FixedSizeSet2: Explicit Rep FixedSizeSet2r (S: set (size n) of t)  NB t could be arbitrarily complex • Axiom schema: i t. i Sj  1..n (S’ [j]=i)

  43. ESSENCE’ Explicit Model: MicroSonet nnodes:nat, capacity:nat given letting find such that Nodes be 1..nnodes ESSENCE ring:set (size capacity) of Nodes 4 ring nnodes:nat, capacity:nat given letting find such that Nodes be 1..nnodes ring’:matrix (indexed by 1..capacity) of Nodes ESSENCE’ AllDifferent(ring’ ) ring’ [1] = 4 OR ring’ [2] = 4 OR… capacity 2 1 ring’ … Nodes Nodes Nodes

  44. Channelling Models • Good constraint models often contain multiple representations of same abstract variable. • Different constraints can often be stated most effectively on different representations. • We use channelling constraints to maintain consistency between the representations. • CONJURE can generate these constraints automatically.

  45. (3) (3) i t (S1’ [i] j  1..n (S2’ [j]=i)) Automatically GeneratingChannelling Models S:set (size n) of t (1) (2) FixedSizeSet2 FixedSizeSet1 S1’ :matrix (indxd by t) of bool S2’ :matrix (indxd by 1..n) of t • i t (iS S1’ [i]) • i t (iSj  1..n (S2’ [j]=i))

  46. (Some) Symmetry Detection is Free! • Some refinements, whether manual or automatic, introduce symmetry. • Some rule clauses introduce symmetry every time they are used. • The clause can annotate the refined expression with a description of the symmetry. • E.g. explicit representation of a set of 3 integers drawn from 1..n. 1 2 3 1..n 1..n 1..n • Have named the indices, where the set did not. • Introduced index symmetry.

  47. Part IV Conclusion: The Big Picture

  48. What we Hope to Achieve • Much of modelling can be formalised as refining ESSENCE to ESSENCE’ • Reduce modelling bottleneck with CONJURE. • Ongoing work: • Develop heuristics to select among competing models.

  49. E E1 E2 M1 M2 M3 M4 Why Might we Fail to Build a Competent Automatic Modeller? • Can’t get compositionality to work. • Some/much of modelling is involved in writing ESSENCE specs.

  50. Why Might we Fail to Build a Competent Automatic Modeller? • Can’t get compositionality to work. • Some/much of modelling is involved in writing ESSENCE specs. • Failure to reach closure: • Endless list of new modelling steps to encapsulate in rules. • General mathematical reasoning needed to perform transformations: • Adding implied constraints. • Obtaining rule applicability. • Obtaining more effective models.

More Related