1 / 13

A new approach to SMTI (in CP)

A new approach to SMTI (in CP). By Chris Unsworth. Constraint satisfaction. Constraint satisfaction problem (CSP) is a triple <V,D,C> where: V is set of variables Each X in V has set of values, D_X Usually assume finite domain {true,false}, {red,blue,green}, [0,10], …

jeri
Télécharger la présentation

A new approach to SMTI (in CP)

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. A new approach to SMTI (in CP) By Chris Unsworth

  2. Constraint satisfaction • Constraint satisfaction problem (CSP) is a triple <V,D,C> where: • V is set of variables • Each X in V has set of values, D_X • Usually assume finite domain • {true,false}, {red,blue,green}, [0,10], … • C is set of constraints Goal: find assignment of values to variables to satisfy all the constraints

  3. Overview • Stable Marriage with Ties and Incomplete Preference lists (SMTI) • The Problem • Current CP solutions cant handle “real world” instances • The New idea • What's good about it • What's not so good about it

  4. The Stable Marriage Problem We have n men and n women Each man ranks the n women And each woman ranks the n men Men Women Jan Liz Sue : Ian Jon Bob : Jon Ian Bob : Bob Jon Ian Bob Ian Jon : Sue Jan Liz : Liz Jan Sue : Jan Sue Liz Objective : To find a matching of men to women Such that the matching is Stable

  5. The Stable Marriage Problem A Matching But not a stable one  Bob and Sue would rather be matched to each other than to there assigned partners Men Women Jan Liz Sue : Ian Jon Bob : Jon Ian Bob : Bob Jon Ian Bob Ian Jon : Sue Jan Liz : Liz Jan Sue : Jan Sue Liz In this matching Bob and Sue form a Blocking pair A matching is only stable iff it contains no Blocking pairs

  6. Ties and incomplete Preference lists Men Women Ann Joe Liz Zoe Jes : Tom Alf Bob Ian : Ian (Alf Jim) : (Alf Ian) Tom Bob : Tom (Jim Ian Bob) Alf : Ian Jim (Tom Bob) Alf Bob Tom Ian Jim : Zoe (Ann Liz) Joe : Liz Jes (Ann Zoe) : (Ann Jes Liz Zoe) : Ann Jes Liz Zoe Joe : Joe Zoe Jes Ties indicate indifference (shown in brackets) Incomplete list indicate some potential partners are unacceptable

  7. Previous CP solutions w1 w2 w3 w4 w5 : m3 m1 m2 m4 : m4 (m1 m5) : (m1 m4) m3 m2 : m3 (m5 m4 m2) m1 : m4 m5 (m3 m2) m1 : w4 (w1 w3) w2 m2 : w3 w5 (w1 w4) m3 : (w1 w5 w3 w4) m4 : w1 w5 w3 w4 w2 m5 : w2 w4 w5 : 1 2 3 4 6 : 1 2 3 6 : 1 2 3 4 6 : 1 2 3 4 5 6 : 1 2 3 4 6 x1 x2 x3 x4 x5 : 1 2 3 4 6 : 1 2 3 4 6 : 1 2 3 4 6 : 1 2 3 4 5 6 : 1 2 3 6 y1 y2 y3 y4 y5 xPy1: {2,2,3}{4,4,4}{2,3,3}{1,1,1}{6,6,6} xPy2: {3,3,4}{6,6,6}{1,1,1}{3,4,4}{2,2,2} xPy3: {1,1,4}{6,6,6}{1,3,4}{1,4,4}{1,2,4} xPy4: {1,1,1}{5,5,5}{3,3,3}{4,4,4}{2,2,2} xPy5: {6,6,6}{1,1,1}{6,6,6}{2,2,2}{3,3,3}

  8. The Problem • Real world Instances of HRTI • 781 Residents, list length 6 with no ties • 53 Hospitals, have many ties • Latest Data set not yet returned a single matching in over a week! • Similar results seen with random datasets

  9. The new idea • All matchings in an SM instance have the same people matched (are the same size) • A SMTI instance can be made into an SM instance by arbitrarily breaking the ties • Instead of searching for the best matching……. • Find the best way to break the ties

  10. Modelling preference lists • m1 : w4 (w1 w3) w2 • Modeled as L constrained integer variables (where L = length of preference list) • mpl1,1 = {4} • mpl1,2 = {1,3} • mpl1,3 = {1,3} • mpl1,4 = {2} • Use these as search variables

  11. What’s good about it? • Solutions guaranteed at every leaf in the search tree • Reduce the number of possible matchings searched • Problem becomes combinatorial, loads of research in other areas for ideas. • E.g. Job shop scheduling etc

  12. What’s not so good about it? • Many ways of breaking ties may lead to the same matching • Naive analysis shows O(tL!tn) leaf nodes (where tL = length of ties and tn = number of ties) • compared to O(n!) for the previous solution • Larger model • O(n3) instead of the optimal O(n2)

  13. Conclusions & future work • Very much “work in progress” • Has some very nice properties • Valid solution at every node • Has some very nasty potential pitfalls • Loads of solutions

More Related