Re: how to value selection in tree search?

Chuang Liu wrote:
> hi there:
> I am using library fd to implement a tree search
> method. I like to control the value selection for
> variables during the search process. I have a variable
> whose domain is [1, 2, 3, 4, 5], and I associate a
> weight to every value. I want the search process to
> try value with higher weight first.
> 
> I tried to sort the value domain based on the weight
> of values like
> V :: [2, 1, 3, 5, 4],
> indomain(V) 
> 
> But indomain always pick the 1 first. 


If all your variable domains are subsets of one "superdomain"
which you can order, then you can define a variable-independent
value-selection predicate and use that instead of indomain.
E.g. for your example:

% enumerate values 1..9 in some heuristic order
domain_value(7).
domain_value(2).
domain_value(6).
domain_value(1).
domain_value(3).
domain_value(9).
domain_value(5).
domain_value(4).
domain_value(8).


?- V :: 1..5, domain_value(V).
V = 2
Yes (0.00s cpu, solution 1, maybe more) ? ;
V = 1
Yes (0.00s cpu, solution 2, maybe more) ? ;
V = 3
Yes (0.00s cpu, solution 3, maybe more) ? ;
V = 5
Yes (0.00s cpu, solution 4, maybe more) ? ;
V = 4
Yes (0.00s cpu, solution 5, maybe more) ? ;
No (0.00s cpu)

?- V :: [4,6,7], domain_value(V).
V = 7
Yes (0.00s cpu, solution 1, maybe more) ? ;
V = 6
Yes (0.00s cpu, solution 2, maybe more) ? ;
V = 4
Yes (0.00s cpu, solution 3, maybe more) ? ;
No (0.00s cpu)


-- 
  Joachim Schimpf              /             phone: +44 20 7594 8187
  IC-Parc                     /      mailto:J.Schimpf@imperial.ac.uk
  Imperial College London    /    http://www.icparc.ic.ac.uk/eclipse