>>
>> Something you might consider is the minimal depth known when three freecells
>> are used. The minimum depth needed for two freecells would probably exceed
>> this value.
> Well, using what kind of moves? Atomic? Multi-column moves?
The same types of moves you'd allow in your distributed DFS.
Actually, my suggestion doesn't matter anymore. I found a 40-move solution to #384243 w/3 freecells. This is considerably shorter than the 80 moves generated during Pass 1 of my BFS solver when using two freecells. So, it would be difficult to estimate the maximum depth attainable during a DFS w/2 freecells.
I am about to create a version of my BFS solver that limits the number of times a "primary card" can be moved. I'm concerned about cards moving in "cycles" while being intermixed with other moves. It makes for a nightmare scenario. A scenario that, if I can block these cycles, would extend the number of moves my solver could examine. Unfortunately, it doesn't guarantee that all layouts have been examined.
A "primary card" is the highest ranking card in a move.
Received on Fri Jan 10 2014 - 10:46:55 IST