[eclipse-clp-users] Question about using bb_min/minimize

From: Seppe vanden Broucke <macuyiko_at_gmail.com>
Date: Fri, 09 Oct 2009 17:58:55 +0200

I've been trying to solve a certain problem with ECLiPSe. I'm almost 
finished writing the source, but I'm having a problem when calling minimize.

The program looks like this:

%================= snip ==============
% :-lib(ic_global).

cost(1, 3, 1, 3.82).
cost(1, 3, 2, 0.45).
cost(1, 3, 4, 0.45).
% ...

solve(TripletNR, Triplets, Cost) :-
    dim(Triplets, [TripletNR,3]),
    SatNR is TripletNR*2, 
    Triplets[1..TripletNR,1..3] :: 1..SatNR,

    % a list of constraints, which are working...
    % I now want to calculate the cost
    ( for(K,1,TripletNR), fromto(0,In,Out,Cost), param(Triplets)
        My1 is Triplets[K,1],
        My2 is Triplets[K,2],
        My3 is Triplets[K,3],
        cost(My1, My2, My3, Mycost),
        Out is In+Mycost

    % minimize:
    minimize(labeling(Triplets), Cost).
%================= snip ==============

I've been using this as an example: 

When I run my program with solve(2, Triplets, Cost)., the output is:
?- solve(2, Triplets, Cost).
Triplets = []([](1, 3, 1), [](2, 4, 2))
Cost = 3.8699999999999997
Yes (0.02s cpu, solution 1, maybe more)
Found a solution with cost 3.8699999999999997

Which is just the first labelled solution, pressing more (or using 
findall) just iterates all the possible solutions:
Found a solution with cost 3.8699999999999997
Found a solution with cost 4.13
Found a solution with cost 0.56
Found a solution with cost 1.5899999999999999

Running the example gives the minimum solution and immediately converges 
using branch and bound.

What am I doing wrong? I believe the problem to be in my fromto-loop. I 
use structures the calculate my total cost, while the example uses 
matrices... I tried using a three-dimensional array []([]([](...))), but 
that gave "instantiation fault in arg" errors when using Costarray[My1, 
My2, My3].

Thanks in advance!

- S. vanden Broucke
Received on Fri Oct 09 2009 - 15:59:10 CEST

This archive was generated by hypermail 2.2.0 : Thu Feb 02 2012 - 02:31:58 CET