Hi Lorenzo, the problem with your code is that findall/3 generates copies of the domain variables. So you are constraining these copies instead of the original variables. For example: test(X,Y) :- X #:: 0..1, Y #:: 0..1, findall(Z, member(Z, [X,Y]), L), L = [A,B], A #> B, writeln(L). gives this: ?- test(X, Y). [1, 0] X = X{[0, 1]} Y = Y{[0, 1]} Yes (0.00s cpu) As you can see, A and B (the copies generated by findall/3) have been constrained (and instantiated), whereas X and Y (the original variables) have been left unconstrained. You would need to update your program to avoid using the findall/3 call to determine the OtherDomains list for each domain variable. However, the best would be to use the cumulative/4 scheduling constraint provided by ECLiPSe's edgefinder library: :- lib(ic). :- lib(ic_edge_finder). schedule(Domains):- getTasks(Tasks), ( foreach(task(_,Dur), Tasks), foreach(Dur, Durations), foreach(1, Resources), foreach(Dom, Domains) do Dom #:: 0..59 ), cumulative(Domains, Durations, Resources, 1), labeling(Domains). % the rest of the program stays unchanged Cheers, Thorsten Am 23.05.2012 18:03, schrieb Lorenzo Stoakes: > Hi all, > > I am trying to leverage CLP in order to solve a rather simple > scheduling problem. the problem consists of a series of tasks, each of > which with a specified length. > > I want to determine all the permutations of start times, with the > constraint that they cannot run at the same time. > > I define these tasks as facts in the knowledge base, e.g.:- > > task(a, 3), > task(b, 2). > > So far, I have been resorting to writing very imperative code littered > with for's and foreach's to constrain things. I write code to use > findall/3 to obtain all N tasks, then I generate N domain variables, > setting them to 0..59 (minutes in one hour). I then use (perhaps > rather unwisely), remove/3 to get each element in the list along with > every other element, to which I apply the constraint Domain #\= > OtherDomain + I for I = 0 to Length-1. > > I have two problems here. 1, this code is rather awful and really, a > twisty turny version of imperative code bending over backwards to try > to work in prolog, and 2 - it doesn't work. I am getting the following > response:- > > [eclipse 41]: top. > [0, 0] > > Delayed goals: > -(_285{0 .. 59}) + _267{0 .. 59} #\= 0 > -(_285{0 .. 59}) + _267{0 .. 59} #\= 1 > -(_285{0 .. 59}) + _267{0 .. 59} #\= 2 > -(_321{0 .. 59}) + _303{0 .. 59} #\= 0 > -(_321{0 .. 59}) + _303{0 .. 59} #\= 1 > Yes (0.00s cpu, solution 1, maybe more) ? > > Which means that variables are somehow getting reassigned, which > rather confuses me. > > I know this might be a rather 'send me the codez' sort of question, > but I am rather stuck and I am sure one of you CLP experts must be > able to come up with a simpler, more declarative and all-round more > CLP-y way of doing things, and also perhaps something that works :) > > I would appreciate any help anybody could give, massively. > > Cheers, > Lorenzo > > Exhibit A, The Code:- > > The code:- > > :- lib(ic). > > task(a, 3). > task(b, 2). > > % We want to schedule these tasks to take up 60 minutes. So > essentially permutations. > > remove(X, [X|Rest], Rest). > remove(X, [Y|Rest], [Y|RestAnswer]):- > remove(X, Rest, RestAnswer). > > getTasks(Tasks):- > findall(task(Name, Length), task(Name, Length), Tasks). > > schedule(Domains):- > getTasks(Tasks), > length(Tasks, N), > length(Domains, N), > Domains :: 0..59, > findall([Domain,OtherDomains], remove(Domain, Domains, > OtherDomains), SplitDomains), > (foreach(task(_, Length), Tasks), foreach([Domain, > OtherDomains], SplitDomains) do > (foreach(OtherDomain, OtherDomains), param(Domain, Length) do > (for(I,0,Length-1), param(Domain, OtherDomain) do > ( > Domain #\= OtherDomain+I > ) > ) > ) > ), > labeling(Domains). > > top:- > schedule(Domains), > writeln(Domains). > > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > > > _______________________________________________ > ECLiPSe-CLP-Users mailing list > ECLiPSe-CLP-Users_at_lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/eclipse-clp-usersReceived on Fri May 25 2012 - 08:04:06 CEST
This archive was generated by hypermail 2.2.0 : Sat Jun 02 2012 - 06:14:11 CEST