CODEWITHSUNDEEP
prologclpfd
user3579813
user3579813

Reputation: 21

How can we select a random data set from the list of solutions given by Prolog?

i am new to prolog and working my way through the FD solvers in it. I am exploring both SWI-Prolog and Gnu Prolog. I have a case where a constraint like below needs to be solved and one of the possible solutions needs to be picked at random based on a seed. Not sure how much of it is possible in Prolog.

Below is the equation i am trying to solve:

?- X #< 7 , X #> -10, 4*X+2*Y #< 8, Y #< 10, Y #>0, label([X]).
X = -9,
Y in 1..9 ;
X = -8,
Y in 1..9 ;
X = -7,
Y in 1..9 ;
X = -6,
Y in 1..9 ;
X = -5,
Y in 1..9 ;
X = -4,
Y in 1..9 ;
X = -3,
Y in 1..9 ;
X = -2,
Y in 1..7 ;
X = -1,
Y in 1..5 ;
X = 0,
Y in 1..3 ;
X = Y, Y = 1.

As you can see above, it gives me multiple solutions. For my application, I would be interested in choosing one of the data set from above based on a seed.

Is this possible in prolog? It would be great if you can help me with your suggestions.

Thanks, Venkat

Upvotes: 2

Views: 186

Answers (1)

Sergii Dymchenko
Sergii Dymchenko

Reputation: 7229

You can specify random choice strategy for labeling variables. That's how to do it in ECLiPSe CLP Prolog:

:- lib(ic).
one_random_solution(X, Y) :-
    X #< 7 , X #> -10, 4*X+2*Y #< 8, Y #< 10, Y #>0,
    once search([X, Y], 0, input_order, indomain_random, complete, []).

And run with:

[eclipse]: seed(10), one_random_solution(X, Y).
X = -9
Y = 5
Yes (0.00s cpu)

Unfortunately, it looks like SWI-Prolog doesn't support random choice strategy for labeling.

Upvotes: 4

Related Questions