Help With Sudoku Explainer

Interactive on-site game threads go here

Re: original Sudoku explainer mirror site

Postby tarek » Wed Jan 01, 2020 6:45 pm

I harvested what I assume are all the pages from https://diuf.unifr.ch/pai/people/juillera/Sudoku/ (dead link) after searching through webarchive

I mirrored it locally and imo has some useful info. These can be hosted on this forum (as the programmers forum). I also would like to added it to the Sukaku Explainer GitHub wiki pages but I don't know hoe to imbed .htm files in them!

tarek
User avatar
tarek
 
Posts: 3537
Joined: 05 January 2006

Re: Help With Sudoku Explainer

Postby 1to9only » Wed Jan 01, 2020 7:33 pm

I think there are only 3 files that may be of interest: FAQ.html, InterestingSudokus.html and SudokuEquiv.html - unless you've collected more!
The text can be extracted, and formatted using the markdown notation favored by the git readme (.md) and wikis.
1to9only
 
Posts: 1655
Joined: 04 April 2018

Re: original Sudoku explainer mirror site

Postby tarek » Wed Jan 01, 2020 11:45 pm

I used an html to markdown online converter. Only minor text editing was required afterwards

Uploaded 4 pages to the project's wiki pages

The 1st page (with links to the rest) is at https://github.com/SudokuMonster/SukakuExplainer/wiki/SE121---Sudoku-Explainer
User avatar
tarek
 
Posts: 3537
Joined: 05 January 2006

Re: Help With Sudoku Explainer

Postby RW » Fri May 15, 2020 6:24 pm

Hi tarek,

I'm looking at this Disjoint Group + Diagonal puzzle in sukaku explainer 1.15.10:

000000000007000000000604003000000000510090034000000000600701000000000200000000000

After a few trivial moves we get here:

Code: Select all
+--------+--------------------------------------------------------------+--------+
| 12348  | 6        234589   12359    12358    35789    14589    124589 | 2578   |
+--------+--------+                                            +--------+--------+
| 1349   | 2458   | 7        139      258      3589     6      | 2458   | 1589   |
|        +--------+--------+                          +--------+--------+        |
| 1289     2589   | 1258   | 6        12578    4      | 158    | 125789   3      |
|                 +--------+--------+        +--------+--------+                 |
| 23489    234789   23489  | 1345   | 13457  | 6      | 1589     12589    2589   |
|                          +--------+--------+--------+                          |
| 5        1        6        8      | 9      | 2        7        3        4      |
|                          +--------+--------+--------+                          |
| 234789   23489    289    | 1345   | 1345   | 57     | 1589     125689   125689 |
|                 +--------+--------+        +--------+--------+                 |
| 6        234589 | 23458  | 7        23458    1      | 3458   | 4589     589    |
|        +--------+--------+                          +--------+--------+        |
| 1349   | 4578   | 13589    349      4568     3589     2      | 45678  | 1589   |
+--------+--------+                                            +--------+--------+
| 123478 | 234589   12589    23459    234568   589      134589   145689 | 15678  |
+--------+--------------------------------------------------------------+--------+

At this point it suggests a skyscraper and 'get all hints' only shows a bunch of turbot fishes. However, we have the following moves:

-Locked candidates 2 on diagonal (eliminate from rest of box 1)
-Locked candidates 6 on diagonal (eliminate from rest of box 9)
-Locked candidates 3 on diagonal (eliminate from rest of top left disjoint group)
-Locked candidates 1 on anti-diagonal (eliminate from rest of bottom left disjoint group)
-Locked candidates 7 in lower right disjoint group (eliminate from rest of diagonal)

How come these moves are not shown?

RW

Edit: seeing similar misses in other puzzles as well. It appears diagonals are not properly accounted for when checking locked candidates.
Edit 2: looking at the code, it seems it doesn't check diagonals for locked candidates at all...
RW
2010 Supporter
 
Posts: 1010
Joined: 16 March 2006

Re: Help With Sudoku Explainer

Postby tarek » Sat May 16, 2020 10:39 am

Try this version 1.17.5

https://www.transfernow.net/VqOkWK052020

It uses generalized intersections amongst other techniques

tarek
User avatar
tarek
 
Posts: 3537
Joined: 05 January 2006

Re: Help With Sudoku Explainer

Postby RW » Sat May 16, 2020 11:11 am

Thanks! That version correctly downgrades the puzzle from ER 4.3 to 2.7.

Btw. In the code I noticed it searches for locked candidates whenever potential positions >= 2. Should be enough to check only when cardinality is 2 or 3 (add 4 if window region involved). Maybe you already changed this in the new version.

RW
RW
2010 Supporter
 
Posts: 1010
Joined: 16 March 2006

Re: Help With Sudoku Explainer

Postby tarek » Sun May 17, 2020 7:52 am

Thanks RW. I'll look into that when I start working on that project again.

tarek
User avatar
tarek
 
Posts: 3537
Joined: 05 January 2006

Re: Help With Sudoku Explainer

Postby tarek » Tue May 19, 2020 11:56 am

RW wrote:In the code I noticed it searches for locked candidates whenever potential positions >= 2. Should be enough to check only when cardinality is 2 or 3 (add 4 if window region involved). Maybe you already changed this in the new version.

Looking at some recent examples … The genralized intersections (or generalized locked candidates if you prefer) may eliminate a candidate from a target cell(s) outside that region that sees all the locked candidates. I found examples of 7 cells from a single region that see the same cell (e.g. Disjoint Groups Asterisk X Sudoku where r8c8 for instance can see 7 of the asterisk group cells)

I'm not sure if I kept that as the upper limit or not. This upper limit is different from one variant too another and indeed on the current resolution state of the puzzle … I didn't put any dynamic limits and most likely it is either no upper limit or a fixed upper limit at 7 candidates

tarek
User avatar
tarek
 
Posts: 3537
Joined: 05 January 2006

Previous

Return to Interactive games