User talk:Heiner
From Sokoban Wiki
Hello Heiner.
The wiki contains some levels for testing the solver. You can compare your results with the one in the wiki. Maybe you can publish some of your results in the future.
- MatthiasThanks for the hint. I will publish some of my results, most probably on my user page, first, as e.g. I do not yet compute the moves, only the pushes. There is still much to do, the program is very experimental.
BTW: On the JSoko statistics pages like SolverStatistics-jsoko-DavidWSkinner-Microban,
shouldn't the unit of time be stated? I guess its milli seconds...
-- Heiner 23:02, 15 November 2010 (UTC)
Coding a solver is very time consuming. Maybe comparing your results with other solvers will help you to find out whether you are on a good way. At least in my solver I often have implemented new ideas and then recognized that the new ideas weren't that useful in most of the levels.
You are right. The time unit is missing. However, I just copied the output of the program into the wiki. The time is not important. The solver have "unlimited" time for solving levels in the comparisons.
- Matthias
Ok, yes, time is not that important. It can give an impression about how hard the job is for the solver. I'm just curious ;-)
Regarding new ideas... I suspect I will implement them anyhow. While I like to be competetive, for me the major source of fun is the programming itself. E.g. I did never before work with bipartite matchings, so I will do that.
Thanks for the results. I checked them - they are indeed all push optimal.
- Matthias
Thanks for checking... How did you do that? Did you compare with the best known solution, or do you have a solver which guarantees push optimality? --Heiner 04:13, 21 November 2010 (UTC)
I compared the results with those at sokobano.de.
However, I could have also used JSoko which includes a solver which guarantees push optimality.
--Matthias 12:26, 21 November 2010 (UTC)
Congratulations for your results in the sokhard set! I will publish new results for JSoko in some hours. I hope it can solve more levels in that set when using the "push optimal" solver instead of the "any solution" solver.
--Matthias 22:49, 27 November 2010 (UTC)
