Thursday, 28th March 2024
Puzzles Solved Yesterday: 124
Forum Index
 
Page 2 of 2<12
Loopy Format and user puzzles
Tilps
Kwon-Tom Obsessive
Puzzles: 6482
Best Total: 20m 22s
Posted - 2008.03.02 19:08:57
Ctrl+Y is redo, not Ctrl+R

Ctrl+R only works in combination with Ctrl+M, it only ever goes backwards to the last time you pressed ctrl+M - and if you've never pressed ctrl+M it goes all the way back to empty board.

Additionally I have some focusing issues which mean that sometimes the program ignores the key presses, I thought I had fixed them, but maybe not.  Using the menu should always work...

Edit:
Beta 35 includes a change which might help with the focusing issues.
Last edited by Tilps - 2008.03.02 21:08:38
Jankonyex
Kwon-Tom Obsessive
Puzzles: 5661
Best Total: 9m 35s
Posted - 2008.03.03 08:56:16
Quote:
Originally Posted by tilps
C....
I mean...
sometimes Ctrl+Z or Ctrl+Y is not work.
e.g. If the last time 'Actions' showed that 'Undo' is not allowed, Ctrl+Z is never work until I open 'Actions' again and showed that 'Undo' is allowed.
Tilps
Kwon-Tom Obsessive
Puzzles: 6482
Best Total: 20m 22s
Posted - 2008.03.03 11:05:48
Aha! - thankyou for that piece of insight, it suddenly makes sense why it hasn't been working.  It will take more then a couple of minutes to fix though so not tonight.

Edit:
Fixed in beta 36 - I think.
Last edited by Tilps - 2008.03.03 20:05:51
Naivoj
Kwon-Tom Addict
Puzzles: 314
Best Total: 33m 50s
Posted - 2008.03.04 09:22:53
Quote:
Originally Posted by tilps
If you want to time based on leaving the program running and using the CPU time, don't click solve - just open the puzzle.  But first change your settings to remove all entries from the 'rating types to show'.
I suppose that means to turn off everything in Setting/Play? The only option that was checked there was "Let AutoMove avoid closing loop too early", which I unchecked.
Quote:
Originally Posted by tilps
That way the only rating it will perform is the full solve to validate that the puzzle actually has a single solution.
Additionally this way you don't need to change it from a single cpu to dual cpu mode.  The only downside is that when its done you wont be able to visually inspect that the solution is correct, you'll just have to trust me .
I then loaded #18 and observed that LoopDeLoop3B36 was only using one of my 2 processors as you predicted: the CPU time stopped at 1:01:39, but this is more than solving as you said that it also verifying there is only one solution. So afterward I pressed "Solve" and observed again that only one processor was used: the puzzle was solved at CPU time 1:52:44; My conclusions are LoopDeLoop3B36
  - solved #18 on my computer in 51 min 5 sec (1:52:44 - 1:01:39)
       - without checking if there is only one solution.
  - only needed an extra 10 min 34 sec to verify there is only one solution
       - which is impressive!
       - Unique test on our program take an eternity. Currently running: 3 hours and counting...
       - But my friend is planning to improve this in the near future.
Tilps
Kwon-Tom Obsessive
Puzzles: 6482
Best Total: 20m 22s
Posted - 2008.03.04 09:54:01
oh - my solver has no concept of 'cheating'.  The full solver never does anything other than solve complete with verify of uniqueness.  The extra 10 minutes would be the non-full ratings. (or some kind of background variation... which there shouldn't be, the solver is deterministic.)  The solve run by clicking the button is the same as the solve run by the ratings thread first.

The rating thread runs the full solver, which will find/verify a solution (it may take forever, but it will get there eventually...).  Once that has been done and the existance of a single solution has been verified it runs the 'rating types to show' which is the text box I asked you to empty - by default that text box contains G,SI,S - G means current settings (why I choose G i have no idea...), SI means simple solver with interactions, and S means simple solver.  For each of these (assuming your current settings isn't the full solver) it runs a binary search to work out the minimum lookahead required to solve the puzzle, but in this case none of them would work even with maximum lookahead, so that would be reasonably fast.  With a puzzle this big and based on how long it takes your computer to run the full solve - I would say 10 minutes to run those 3 would be reasonable.

My non-full solvers, if they find a solution gaurantee it is the only solution, but are not gauranteed to find a solution.  They usually run quite quickly too, but your puzzles use advanced logic either in loop closing or something else I am missing, so the simple solvers fail to complete the job.

All of the above assumes a lack of bugs in my solvers - bugs will cause it to fail to find solutions, or in much much rarer scenarios find solutions which don't exist.
Last edited by Tilps - 2008.03.04 11:47:26
Page 2 of 2<12

Forum Index