Constraint Satisfaction

Problems

Chapter 5
Section 1 – 3

4 Feb 2004 CS 3243 - Constraint Satisfaction 1

 Outline
 Constraint Satisfaction Problems (CSP)
 Backtracking search for CSPs
 Local search for CSPs

4 Feb 2004 CS 3243 - Constraint Satisfaction 2

 Constraint satisfaction problems (CSPs)

 Standard search problem:


 state is a "black box“ – any data structure that supports successor
function, heuristic function, and goal test


 CSP:

 state is defined by variables Xi with values from domain Di


 goal test is a set of constraints specifying allowable combinations of

values for subsets of variables


 Simple example of a formal representation language

4 Feb 2004 CS 3243 - Constraint Satisfaction 3
 Example: Map-Coloring

 Variables WA, NT, Q, NSW, V, SA, T

 Domains Di = {red,green,blue}
 Constraints: adjacent regions must have different colors


 e.g., WA ≠ NT, or (WA,NT) in {(red,green),(red,blue),(green,red),

(green,blue),(blue,red),(blue,green)}


4 Feb 2004 CS 3243 - Constraint Satisfaction 4

 Example: Map-Coloring

 Solutions are complete and consistent assignments,

e.g., WA = red, NT = green,Q = red,NSW =
green,V = red,SA = blue,T = green

4 Feb 2004 CS 3243 - Constraint Satisfaction 5
 Constraint graph
 Binary CSP: each constraint relates two variables


 Constraint graph: nodes are variables, arcs are constraints



4 Feb 2004 CS 3243 - Constraint Satisfaction 6

 Varieties of CSPs
 Discrete variables

 finite domains:
 n variables, domain size d  O(dn) complete assignments
 e.g., Boolean CSPs, incl.~Boolean satisfiability (NP-complete)
 infinite domains:
 integers, strings, etc.
 e.g., job scheduling, variables are start/end days for each job
 need a constraint language, e.g., StartJob1 + 5 ≤ StartJob3

 Continuous variables

 e.g., start/end times for Hubble Space Telescope observations
 linear constraints solvable in polynomial time by linear programming
4 Feb 2004 CS 3243 - Constraint Satisfaction 7
 Varieties of constraints
 Unary constraints involve a single variable,
 e.g., SA ≠ green


 Binary constraints involve pairs of variables,

 e.g., SA ≠ WA


 Higher-order constraints involve 3 or more

variables,
 e.g., cryptarithmetic
4 Feb 2004
column constraints
CS 3243 - Constraint Satisfaction 8
 Example: Cryptarithmetic

 Variables: F T U W
R O X 1 X2 X 3
 Domains: {0,1,2,3,4,5,6,7,8,9}
 Constraints: Alldiff (F,T,U,W,R,O)

 O + O = R + 10 · X1


 X1 + W + W = U + 10 · X2


X2
4 Feb 2004 + T + T = O + CS
103243
· X3- Constraint Satisfaction 9
 Real-world CSPs
 Assignment problems
 e.g., who teaches what class


 Timetabling problems

 e.g., which class is offered when and where?


 Transportation scheduling

 Factory scheduling


Notice that many

4 Feb 2004
 real-world problems involve real-10
CS 3243 - Constraint Satisfaction
 Standard search formulation (incremental)

Let's start with the straightforward approach, then fix it

States are defined by the values assigned so far

 Initial state: the empty assignment { }

 Successor function: assign a value to an unassigned variable that does
not conflict with current assignment
 fail if no legal assignments

 Goal test: the current assignment is complete

1. This is the same for all CSPs

2. Every solution appears at depth n with n variables
 use depth-first search
3. Path is irrelevant, so can also use complete-state formulation
4 Feb 2004 CS 3243 - Constraint Satisfaction 11
4. b = (n - l )d at depth l, hence n! · dn leaves
 Backtracking search
 Variable assignments are commutative}, i.e.,
[ WA = red then NT = green ] same as [ NT = green then WA = red ]

 Only need to consider assignments to a single variable at each node

 b = d and there are $d^n$ leaves

 Depth-first search for CSPs with single-variable assignments is called

backtracking search


 Backtracking search is the basic uninformed algorithm for CSPs



 Can solve n-queens for n ≈ 25


4 Feb 2004 CS 3243 - Constraint Satisfaction 12
 Backtracking search

4 Feb 2004 CS 3243 - Constraint Satisfaction 13

 Backtracking example

4 Feb 2004 CS 3243 - Constraint Satisfaction 14

 Backtracking example

4 Feb 2004 CS 3243 - Constraint Satisfaction 15

 Backtracking example

4 Feb 2004 CS 3243 - Constraint Satisfaction 16

 Backtracking example

4 Feb 2004 CS 3243 - Constraint Satisfaction 17

 Improving backtracking efficiency

 General-purpose methods can give huge

gains in speed:


 Which variable should be assigned next?



 In what order should its values be tried?



 Can we detect inevitable failure early?



4 Feb 2004 CS 3243 - Constraint Satisfaction 18

 Most constrained variable
 Most constrained variable:
choose the variable with the fewest legal values

 a.k.a. minimum remaining values (MRV)

heuristic


4 Feb 2004 CS 3243 - Constraint Satisfaction 19

 Most constraining variable
 Tie-breaker among most constrained
variables
 Most constraining variable:


 choose the variable with the most constraints on

remaining variables


4 Feb 2004 CS 3243 - Constraint Satisfaction 20

 Least constraining value
 Given a variable, choose the least
constraining value:


 the one that rules out the fewest values in the

remaining variables


 Combining these
4 Feb 2004 CS 3243 heuristics makes 1000
- Constraint Satisfaction 21
 Forward checking
 Idea:
 Keep track of remaining legal values for unassigned variables
 Terminate search when any variable has no legal values


4 Feb 2004 CS 3243 - Constraint Satisfaction 22

 Forward checking
 Idea:
 Keep track of remaining legal values for unassigned variables
 Terminate search when any variable has no legal values


4 Feb 2004 CS 3243 - Constraint Satisfaction 23

 Forward checking
 Idea:
 Keep track of remaining legal values for unassigned variables
 Terminate search when any variable has no legal values


4 Feb 2004 CS 3243 - Constraint Satisfaction 24

 Forward checking
 Idea:
 Keep track of remaining legal values for unassigned variables
 Terminate search when any variable has no legal values


4 Feb 2004 CS 3243 - Constraint Satisfaction 25

 Constraint propagation
 Forward checking propagates information from assigned to
unassigned variables, but doesn't provide early detection for
all failures:


 NT and SA cannot both be blue!



 Constraint propagation
4 Feb 2004 CS 3243 repeatedly enforces constraints
- Constraint Satisfaction 26
 Arc consistency
 Simplest form of propagation makes each arc consistent
 X Y is consistent iff


for every value x of X there is some allowed y

4 Feb 2004 CS 3243 - Constraint Satisfaction 27

 Arc consistency
 Simplest form of propagation makes each arc consistent
 X Y is consistent iff


for every value x of X there is some allowed y

4 Feb 2004 CS 3243 - Constraint Satisfaction 28

 Arc consistency
 Simplest form of propagation makes each arc consistent
 X Y is consistent iff


for every value x of X there is some allowed y

 If X loses a value, neighbors of X need to be rechecked



4 Feb 2004 CS 3243 - Constraint Satisfaction 29

 Arc consistency
 Simplest form of propagation makes each arc consistent
 X Y is consistent iff


for every value x of X there is some allowed y

 If X loses a value, neighbors of X need to be rechecked

 Arc consistency detects failure earlier than forward checking

4 Feb 2004 CS 3243 - Constraint Satisfaction 30
 Arc consistency algorithm AC-3

 Time complexity: O(n2d3)


4 Feb 2004 CS 3243 - Constraint Satisfaction 31
 Local search for CSPs
 Hill-climbing, simulated annealing typically work with
"complete" states, i.e., all variables assigned


 To apply to CSPs:

 allow states with unsatisfied constraints


 operators reassign variable values



 Variable selection: randomly select any conflicted variable



 Value selection by CS
4 Feb 2004 min-conflicts heuristic:
3243 - Constraint Satisfaction 32
 Example: 4-Queens
 States: 4 queens in 4 columns (44 = 256 states)


 Actions: move queen in column


 Goal test: no attacks


 Evaluation: h(n) = number of attacks



 Given random initial

4 Feb 2004 state,
CS 3243 - Constraint solve n-queens in almost 33
can Satisfaction
constant time for arbitrary n with high probability (e.g., n =
 Summary
 CSPs are a special kind of problem:

 states defined by values of a fixed set of variables


 goal test defined by constraints on variable values



 Backtracking = depth-first search with one variable assigned per node



 Variable ordering and value selection heuristics help significantly



 Forward checking prevents assignments that guarantee later failure



 Constraint propagation (e.g., arc consistency) does additional work to

constrain values and detect inconsistencies
4

Feb 2004 CS 3243 - Constraint Satisfaction 34