View previous topic :: View next topic |
Author |
Message |
| ArtoI
| Joined: 20 Sep 2006 | Posts: 3 | : | | Items |
|
Posted: Fri Sep 22, 2006 1:34 pm Post subject: AIroot - universal solving technique |
|
|
Hello,
I have just invented universal technique AIroot, which solves all puzzles (at least have solved so far). It is based on the fact that, there can arrive three types on conflicts, when solving.
1. basic conflicts - there are only N-1 different candidates in N cell in the area
2. fish conflicts - when eliminating number from N rows/columns, it will disappear also from N+1 columns/rows.
3. unique conflicts - this pattern means multiple solutions
When all candidates, which cause conflict immediately, have been eliminated (AIroot link 1), it is time to start linking. Linking take into account all candidates, which cause any type of conflict immediately in the new situation. If any of the conflicts appear in second link, the number can be eliminated by some AIroot link 2 technique. So, the AIroot method is some kind of enlargement of the Trebor's Tabling, because it takes into account all conflict types.
All other techniques can be also be converted to the AIroot. For example: single = AIroot link 1 technique. In normal single all candidates, which can be seen by single cell, can be eliminated. In the AIroot the way of thinking is opposite, illegal candidate cause basic conflict (empty cell), so it can be eliminated. Respectively all fish like and UR techniques are AIroot link 1 techniques.
XY-wings and Bowman Bingo are AIroot link 2 techniques and so on.
Regards,
ArtoI |
|
Back to top |
|
|
| Ruud Site Admin
| Joined: 17 Sep 2005 | Posts: 708 | : | Location: Netherlands | Items |
|
Posted: Fri Sep 22, 2006 3:59 pm Post subject: |
|
|
It is not clear from your description how this works.
Can you explain it with an example?
Ruud _________________ Meet me at sudocue.net |
|
Back to top |
|
|
| ArtoI
| Joined: 20 Sep 2006 | Posts: 3 | : | | Items |
|
Posted: Thu Oct 05, 2006 1:34 pm Post subject: |
|
|
Procedure of using AIroot is quite simple.
1. put candidate to its' cell
2. check if any of the conflict appear
3. eliminate all candidates, which cause conflict in the new situation
4. repeat 2 and 3 until conflict appear of eliminations are not possible
5. if you want to find out the minimum number of links needed to solve puzzle, eliminate only these candidates, which can be eliminated with the minimum steps in each situation |
|
Back to top |
|
|
| gsf
| Joined: 18 Aug 2005 | Posts: 411 | : | Location: NJ USA | Items |
|
Posted: Thu Oct 05, 2006 4:51 pm Post subject: |
|
|
ArtoI wrote: | Procedure of using AIroot is quite simple.
1. put candidate to its' cell
2. check if any of the conflict appear
3. eliminate all candidates, which cause conflict in the new situation
4. repeat 2 and 3 until conflict appear of eliminations are not possible
|
... until conflict appear ...
and then either give up or backtrack
an ArtoI solver is a backtrack solver
nothing new unless you have insights into selecting the best cells and candidate values to guess on |
|
Back to top |
|
|
| ArtoI
| Joined: 20 Sep 2006 | Posts: 3 | : | | Items |
|
Posted: Wed Oct 11, 2006 1:38 pm Post subject: |
|
|
Correct conclusion. The benefit of my sover is, that it finds out the shortest number of links needed to make next elimination. This feature is useful, when comparing the hardest sudokus, because it is possible to make scoring, which is independent from the subjective scoring of different solving techniques and also independent from application order of the different solving techniques.
All backtrack solvers I have seen before use only limited set of conflicts. AIroot method includes them all.
So, improvements are small but usefull. |
|
Back to top |
|
|
|