So I've finally done my program. It gave me a lot of work, but it ends up working just fine!
Just for the ones who wanna know how I've done it:
So it starts by creating a vector with the boxes. Than the program looks for the closest box that has a possible knock (most of the boards has a start position in one of those). While it looks for that box it saves the path and when it finds the box he creates a tree node...
Tsuk
Last Activity: Dec 28 '10, 07:16 PM
Joined: May 19 '10
Location:
-
I know that the binary heap does that, my doubt is how to use it for this particular problem.
At the moment, my strategy is to find all the possible paths till the program hits the final box, using a kind of tree(every node has 4 son nodes, and is double linked to be possible to reach the father using the son). Than I use Breadth Search First to look for the shortest path.
But I still think that are another more efficient way to do...Leave a comment:
-
Exactly, it only allows 2 "childs", but I'll need 4 "childs", one for each direction. Or its possible to apply the heap condition with 4 "childs"?
Theoretically I think is possible, I'm just not seeing how to implement it....Leave a comment:
-
I'm thinking about do it that way, but I'm not seeing how I'm going to create a the tree. If I want to use BFS, the tree will be "Breadth created" or it only mathers after create the tree?
I was thinking about another way to do it. It was a depth search with the help of a counter that saves the number of levels. For example: I start filling the tree till it gets the end. The counter saves the number of levels. Then it goes to another...Leave a comment:
-
I'd think about it, but it can't be binary because there are 4 possible paths for each node... right?
Thanks for you'r help....Leave a comment:
-
The fastest algorithm for TipOver Solver using tree
I am making a tipover solver in C.
I'm trying to solve it by creating a tree with all the possible paths and then process all the tree with a level search algorithm, looking for the shortest path. This implementation is pretty easy to do but takes more processing time than I was expecting. So now I'm thinking about do it with a variation of the A* algorithm. Do you think it will run faster? Or the problem is in looking to all the possible...
No activity results to display
Leave a comment: