The search uses IDA* algorithm with walking distance heuristic - you have to run the following command at the start
create_wd_database.to generate all 24964 possible boards for this heuristic (otherwise it won't work) - it shouldn't take a bit more than a minute on a decent computer.
You can also use a much weaker heuristic (which doesn't have to generate anything) MAX(inversion distance, manhattan) but you have to change the heuristic in the code.
% Use this heuristic if you don't want to generate boards for walking distance heuristic
%heuristic(Current, _, Res) :-
% manhattan(Current, M),
% inversion_distance(Current, I),
% Res is max(M, I).
heuristic(Current, _, Res) :-
get_walking_distance(Current, Res).The puzzle board is represented as one list of numbers 0-15 left to right, top to bottom, eg. the final state is represented as [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,0].
Then you can simply run (for example)
solve([1,10,15,4,0,13,6,8,2,9,3,7,14,5,12,11], M).(I recommend to do it like this though - so you can see the time and length of the solution)
time((solve([1,10,15,4,0,13,6,8,2,9,3,7,14,5,12,11], M), write(M), length(M,L))).The solvabilty of the particular configuration is tested at the start of the solve.
You can also try to run prepared tests:
test0- unsolvabletest1- 35 movestest2- 36 movestest3- 37 movestest4- 40 movestest5- 41 movestest6- 46 moves
Each test shouldn't take more than a couple of minutes at max