|
View previous topic :: View next topic |
Author |
Message |
| coloin
| Joined: 05 May 2005 | Posts: 97 | : | | Items |
|
Posted: Sat Apr 12, 2008 8:50 am Post subject: |
|
|
JPF wrote: | there are 18 NAR |
Repeating my analysis
I generated 46656 random NAR templates.
There were 19 of these which were rainbow, this confrims the random grid probably was representative. Making a few more confirmed the ceiling of essentially different NAR templates is 439.
The generation of the rainbow grids with 9 clues in any position but with a clue in every box [9!][362880] is slightly trickier [for me]. I am still thinking there wont be many essentially different - there will be 46656 equivalent NAR templates straight off. Possibly not all possibilities will occur in our 17-puzzles.
Approx 50% of 17-puzzles have 9 clue values
Approx 80% of 17-puzzles have a clue in all 9 boxes
Approx 50% of 17-puzzles have 8 clue values
Most 17-puzzles have a clue in 8 of the boxes
This is another way of catagorizing "equivalent" clues, the value of which is not yet established.
Having said all this we actually do have a way of reducing and minlexing the 9 clue sub-puzzles from all our 17-puzzles. Clue and box distribution clearly possible. Displaying and searching for puzzles in the "gaps" could be done.
Having said all this we actually do have a way of reducing and minlexing the 12 clue sub-puzzles from all our 17-puzzles as well.
However, it is not possible to test each of the many duplicate 12-combinations with a {-0+5} even if we are able to get assistance from the fixed clues in the [2] minlex combinations.
Or is it ?
Just how remote are the recently found 17-puzzles ?
Last edited by coloin on Tue May 06, 2008 4:58 pm; edited 1 time in total |
|
Back to top |
|
|
| gsf
| Joined: 18 Aug 2005 | Posts: 411 | : | Location: NJ USA | Items |
|
Posted: Sat Apr 12, 2008 1:43 pm Post subject: |
|
|
coloin wrote: |
Having said all this we actually do have a way of reducing and minlexing the 12 clue sub-puzzles from all our 17-puzzles as well.
However, it is not possible to test each of the many duplicate 12-combinations with a {-0+5} even if we are able to get assistance from the fixed clues in the [2] minlex combinations.
|
there are 220,940,238 essentially different 12 clue subgrids in |G|=47733
I'll see about squeezing some distributions early next week |
|
Back to top |
|
|
| coloin
| Joined: 05 May 2005 | Posts: 97 | : | | Items |
|
Posted: Tue Apr 15, 2008 7:27 pm Post subject: |
|
|
A little thought on what we are doing here.......
I dont think 221M {-0+5} is going to be possible !
However, possibly, an improved method, is to compare the common clues in the puzzles from the minlex grid solutions. I think I half-heartedly tried it a while back.
I was able to do a cumbersome and incomplete {-0+4} on a random selection of 1000 x 13 clue subpuzzles. The subpuzzles were from the puzzles in the minlex representation of the grid. I got the original 17s but not surprizingly not new ones. I enlisted the help of the fixed clues in the minlex grid solutions. I am sure it could be optimized.
Before we consider this on a big scale, we should ask how remote are recently found 17-puzzles are ?
If we take the puzzles which solve to the minlex grid, how many of the 12-subpuzzles from the new 17-puzzle correalate directly with any from the set of all 12 puzzles from the total 17-puzzles. I did a small scale version of this, comparing the 12s from a remote puzzle with 12s from 200 other 17 puzzles [all puzzles give minlex soutions] the maximum overlap of clues that I could find was 4 clues out of the 12.[disappointing] Grid solutions with more than one 17-puzzle would taint this complete analysis.
This shows that using fixed clues probably aint going to work.
The reason why every puzzle has a neighbour with 12 clues the same is due to the large number of isomorphic variations of each individual puzzle, and, of course, the large number of 17-puzzles.
C |
|
Back to top |
|
|
| gsf
| Joined: 18 Aug 2005 | Posts: 411 | : | Location: NJ USA | Items |
|
Posted: Thu Apr 17, 2008 3:11 pm Post subject: |
|
|
ChPicard has just recorded 2 new 17s
I'm currently at the random luck stage with my current strategy and it looks like luck is running out
did you apply some new strategy or old luck? |
|
Back to top |
|
|
| coloin
| Joined: 05 May 2005 | Posts: 97 | : | | Items |
|
Posted: Fri May 09, 2008 7:00 pm Post subject: |
|
|
JPF wrote: | a rough estimate of the number non equivalent patterns is 81!/(16!x 65!)/(2 x 6^8)=10 000 527 188 |
Heres an idea for the 16 proof !
This pattern apparently doest have any valid puzzles. Code: | +---+---+---+
|***|***|***|
|***|***|***|
|...|***|...|
+---+---+---+
|...|*.*|...|
|...|*.*|...|
|...|*.*|...|
+---+---+---+
|...|***|...|
|***|***|***|
|***|***|***|
+---+---+---+ |
we also know that there needs two be a clue in 1 of the 18/27 cells in a horizontal band or vertical shute.
Code: | +---+---+---+
|...|...|...|
|...|...|...|
|***|***|***|
+---+---+---+
|***|***|***|
|***|***|***|
|***|***|***|
+---+---+---+
|***|***|***|
|***|***|***|
|***|***|***|
+---+---+---+ |
we can therefore remove all the 16 patterns which have only these clues
Code: |
+---+---+---+ +---+---+---+
|*..|*.*|..*| |*..|*.*|..*|
|...|*.*|...| |...|*.*|...|
|...|*.*|...| |...|*.*|...|
+---+---+---+ +---+---+---+
|***|***|***| |***|***|***|
|...|***|...| |*..|***|...|
|***|***|***| |***|***|***|
+---+---+---+ +---+---+---+
|...|*.*|...| |...|*.*|...|
|...|*.*|...| |...|*.*|...|
|*..|*.*|..*| |*..|*.*|..*|
+---+---+---+ +---+---+---+ also dont have puzzles I believe. |
Just keep whittling away !
It might be possible to demonstrate that 12 or more clues cant possibly hit all the recipricals of these patterns simultaneously.
We know 17 clues can
C |
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
Powered by phpBB © 2001, 2005 phpBB Group
Igloo Theme Version 1.0 :: Created By: Andrew Charron
|