My solution "only" enumerates W^N states, not N^W. TLC finds the N=4, W=40 solution on my laptop. It chokes on N=5, obviously.
MC
On Wednesday, February 11, 2015 at 3:03:05 PM UTC-5, Leslie Lamport wrote:
Michel's solution requires TLC to enumerate N^W states and then throw away the ones that don't satisfy the entire Init predicate. Hence, it can't handle the original problem with N=4, W=40. My solution requires TLC to construct a much smaller set, so it handles that model. However, it still can't handle much bigger models. Since ProB does not does not require enumerating all cases, it should be able to handle very big models.
On Wed, Feb 11, 2015 at 11:33 AM, Michel Charpentier
<michel...@xxxxxxx> wrote:
Another solution, which I find easier to read:
http://cs.unh.edu/~charpov/stone.tla
It probably requires more work from TLC but less thinking on my part!
MC
On Wednesday, February 4, 2015 at 7:56:27 PM UTC-5, Leslie Lamport wrote:I just posted in the CodePlex repository a TLA+/TLC solution to this problem for arbitrary weight of stone and number of pieces. The repository is at
Click on Source Code > examples > Stones. However, I recommend trying to write such a solution yourself before looking at mine.
On Wed, Feb 4, 2015 at 1:48 PM, TLA Plus <tlap...@xxxxxxxxx> wrote:
Thanks for the puzzle. It took me just a little more time to write a spec that allows TLC to solve the problem than to solve it myself. (However, I had heard it before and so recreating the solution was easy.) TLC found the solution in a few seconds. I will leave this as a puzzle for readers to solve, and I hope others will do the same. I therefore rejected an email someone just sent that reveals the solution, although not the TLA+ spec that allows TLC to solve it. After TLC shows you the answer, the reasoning behind it becomes fairly obvious.