SudoCue Users Forum Index SudoCue Users
A forum for users of the SudoCue programs and the services of SudoCue.Net
 
 FAQFAQ   SearchSearch   UsergroupsUsergroups   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

SudoCue Release 3.1.0
Goto page Previous  1, 2
 
Post new topic   Reply to topic    SudoCue Users Forum Index -> Releases
View previous topic :: View next topic  
Author Message
Para
Yokozuna
Yokozuna


Joined: 08 Nov 2006
Posts: 384
Location: The Netherlands

PostPosted: Mon Aug 27, 2007 4:44 pm    Post subject: Reply with quote

Glyn wrote:
Try generating puzzles with those levels and you will see what I mean, it's much easier (ie you sometimes get one). You may also find that some of the moves required merit a high tariff in Sudocues' rating system and this may be pushing them out of your requested zone.


Hi

This is definitely true. This i mostly because any single digit technique that uses the extra constraint a lot of times becomes a color wrap (which seems to be very rare normally) instead of for example x-wing, sky-scraper, 2-string kite, etc. And as this technique is higher rated it adds to the score. The difference between easy/tough/hard, isn't that big. But i think Ruud mentioned this already before in the x-files forum, that the ratings for sudoku-X(or prob any other variant) puzzles are always (a bit) higher than normal.
I also think puzzles with extra constraints tend to have fewer givens and thus mostly need more moves to finish(at least more singles).

But i think the real problem is that sudocue just crashes when you try to create any clover/clover-X puzzles of this difficulty. It doesn't even attempt to create a puzzle at all. Normally when it can't find that rating it just goes through a lot of attempts and eventually gives up.


Thanks for the info on RN,Cn-spaces. So if i get it right in RN-space in Rows you can find hidden subsets as naked subsets and in Columns you can find X-wings, swordfish etc and in CN-space it is vice versa.

greetings

Para
Back to top
View user's profile Send private message
Ruud
Site Owner
Site Owner


Joined: 30 Dec 2005
Posts: 601

PostPosted: Mon Aug 27, 2007 9:11 pm    Post subject: Reply with quote

The new Clover-X variant is very difficult to generate. When I set the symmetry requirement to "no symmetry" and ask the program to generate an "unfair" Clover-X, it will always create one, but it often takes several minutes. Because the generator is not a background process, the program seems not to respond when it's busy. It does not crash but it keeps trying to meet your unreasonable demands Rolling Eyes

Here are a few tips to speed up the generation of Clover-X puzzles:
- add a 0 behind all limits, starting with the level that you intend to create. For example, if you want to create a tough puzzle, set the limits for "tough" and "hard" to 10 times their normal value.
- choose a symmetry with a minimum orbit size of 4. These are 90 degrees rotational, hor+ver, both diagonals and full dihedral. This lowers the chance of fully minimal puzzles which have a higher difficulty level.
- set the backtracking limit to 10000. Because there are more constraints, the average backtracking count is higher. A low setting will result in many rejected puzzles which would actually be correct.

Even with these settings, the generator may stall from time to time backtracking itself to bits on a complex puzzle. Eventually, the program will reject it and continue.

Do not forget to restore the normal settings when you want to create regular Sudokus again.

Some info:

The minimum numbers of clues I found so far is 10. I found several of them, so they must be quite common. Here's an example:
Code:
3 . 1|. . .|. . .
6 . .|. . .|. 2 .
. . .|. . .|. . 8
-----+-----+-----
. . .|. . .|. . .
. . .|. . .|. . .
. . .|. . 2|. . .
-----+-----+-----
5 . 7|. . .|. . .
. . .|. . .|. . .
. . .|. 8 .|. . 4


Here are some comparisons between the variants:

Code:
Variant    | Houses | Intersect | UR's | Rookeries
-----------+--------+-----------+------+----------
Standard   |   27   |     54    |  486 |   46656
Center Dot |   28   |     60    |  342 |   37056
Asterisk   |   28   |     60    |  318 |   33984
Sudoku-X   |   29   |     60    |  150 |   25608
Disjoint G |   36   |    108    |  162 |    8784
Windoku    |   36   |    132    |  206 |    6080
Clover     |   37   |    138    |  118 |    4420
Windoku-X  |   38   |    144    |   54 |    3448
Clover-X   |   39   |    152    |   46 |    3048

Intersections: Overlaps of 2 houses spanning 2 or more cells.
UR's: Unique Rectangles. Fewer UR's means fewer clues required.
Rookeries: Number of ways to place 9 occurrences of a digit in the grid.

[edit]

And I also found a 9 clue minimum Clover-X Very Happy
Code:
. . .|. . .|. . .
. . .|. . .|6 . .
. . .|. . .|. . .
-----+-----+-----
. . .|. . .|. . .
. . .|. . .|. . .
. . 5|4 . .|. . .
-----+-----+-----
. . .|. . .|. . 1
4 . .|. . 2|. . .
. 3 .|. . .|7 . 8

Ruud
_________________
“If the human brain were so simple that we could understand it, we would be so simple that we couldn't.” - Emerson M Pugh
Back to top
View user's profile Send private message Send e-mail
Brian
Regular
Regular


Joined: 30 Jan 2007
Posts: 11
Location: Stockholm, Sweden.

PostPosted: Wed Aug 29, 2007 1:43 pm    Post subject: Bug in the new service release Reply with quote

There seems to be a bug in the routines to generate the RN and CN views. I seem to have far more candidates in RN space than in RC. Could it be that the program generates the candidates in RN and CN space from the digits already in place but ignores candidates that I have eliminated in RC space?

I was about halfway through today’s Nightmare when I noticed the problem.
If you paste this code into the latest service release of Sudocue you can recreate the problem:
u7 {249} {259} u8 {34} 1 6 {349} {345}
6 {149} 3 {259} {47} {25} {1579} 8 {457}
8 {14} {59} {59} {3467} {36} {137} {1347} 2
u4 u3 {168} 7 {125} {68} {58} {26} 9
2 5 {68} u4 {36} 9 {378} {367} 1
9 7 {168} {25} {125} {368} 4 {236} {356}
1 u8 4 3 {25} {25} {79} {679} {67}
u3 u6 u7 u1 u9 u4 2 5 8
u5 {29} {29} 6 8 u7 {13} {134} {34}

Go to RN space and into ultra colouring mode. Colour the 5 in row 1 column 5 pink and the 3 in row 3 column 1 blue. Switch to CN space and see the corresponding pink 1 in r5, c5 and blue 3 in r1, c3. Switch to RC space and you do not see a pink 5 in r1, c5 or a blue 1 in r3, c3.

Best regards,

Brian.
Back to top
View user's profile Send private message
Ruud
Site Owner
Site Owner


Joined: 30 Dec 2005
Posts: 601

PostPosted: Thu Aug 30, 2007 6:13 pm    Post subject: Reply with quote

Thanks Brian,

The program seems to mess up the synchronization between the candidates in RC/RN/CN space when you do manual eliminations. I'll try to have a fix for this early next week.

cheers,
Ruud
Back to top
View user's profile Send private message Send e-mail
nj3h
Gold Member
Gold Member


Joined: 10 Jul 2006
Posts: 111
Location: Virginia / USA

PostPosted: Fri Aug 31, 2007 12:58 am    Post subject: Reply with quote

Hi Ruud,

A few posts ago you provided some guidance about settings when generating Clover-X puzzles. Perhaps SudoCue could automatically use values that the player sets on a new settings area on the screen with the other difficulty factors. A separate set of difficulty level settings could be entered on the same screen for the Clover-X variant. When other variants are chosen then the program would automatically revert back to the standard settings.

Regards,
George
Back to top
View user's profile Send private message
Ron Moore
Addict
Addict


Joined: 13 Aug 2006
Posts: 72
Location: New Mexico

PostPosted: Thu Jan 03, 2008 7:09 pm    Post subject: Sue de Coq, v 3.1.0.1 Reply with quote

Ruud,

Sudocue version 3.1.0.1 seems to detect most Sue de Coq patterns (it's found several which I've missed), but I've found a few which it doesn't. I've slowly began undertaking some of the so-called "Unsolvable" puzzles from sudoku.org.uk (here). This is the position in Unsolvable #12 after initial basic eliminations.

Start position, Unsolvable #12:

..73..4......4..9..5.....6......76..5...3...1..39.8....2.....7..1..8......46.92..

Code:

·--------------------·------------------------·----------------------·
| 12689  689    7    |  3        12569  1256  | 4      1258   258    |
| 12368  368    1268 |  1578-2   4      1256  | 1578   9      2578   |
| 4      5      1289 |  178-2    1279   12    | 1378   6      2378   |
·--------------------+------------------------+----------------------·
| 1289   489    1289 |  15-24    125    7     | 6      23458  234589 |
| 5      46789  2689 | B24       3      246   | 789    248    1      |
| 1267   467    3    |  9        1256   8     | 57     245    2457   |
·--------------------+------------------------+----------------------·
| 3689   2      5689 | C145     A15     34-15 | 13589  7      345689 |
| 3679   1      569  | C2457     8      234-5 | 359    345    34569  |
| 378    378    4    |  6       A157    9     | 2      1358   358    |
·--------------------·------------------------·----------------------·


Here we have a "classic" Sue de Coq pattern. I term it "classic" because it satisfies a constraint given in the original Sue de Coq post (here), that all candidate digits in the pattern be found in the line/box intersection set (set C in my notation):
    set C (in box 8/column 4 intersection) = r78c4, digits 12457
    set A (in box 8) = r79c5, digits 157
    set B (in column 4) = r5c4, digits 24
The pattern results in the eliminations shown in box 8 and column 4. I have Sue de Coq set at fairly high priority (before the ALS XZ rule, and finned swordfish). Sudocue 3.1.0.1 does find the finned swordfish elimination of (1)r3c5 but does not find the Sue de Coq.

I also want to mention another Sue de Coq pattern in this same position, but this one is not classic.

Code:

·---------------------·------------------------·----------------------·
| 12689  689     7    |  3       269-15   1256 | 4      1258   258    |
| 12368  368     1268 |  12578   4        1256 | 1578   9      2578   |
| 4      5       1289 |  1278    279-1    12   | 1378   6      2378   |
·---------------------+------------------------+----------------------·
| 1289   489     1289 |  15-24  C125      7    | 6      23458  234589 |
| 5      6789-4  2689 | A24      3       A246  | 789    28-4   1      |
| 1267   467     3    |  9      C1256     8    | 57     245    2457   |
·---------------------+------------------------+----------------------·
| 3689   2       5689 |  145    B15       1345 | 13589  7      345689 |
| 3679   1       569  |  2457    8        2345 | 359    345    34569  |
| 378    378     4    |  6       7-15     9    | 2      1358   358    |
·---------------------·------------------------·----------------------·


Here we have
    set C (in box 5/column 5 intersection) = r46c5, digits 1256
    set A (in box 5) = r5c46, digits 246
    set B (in column 5) = r7c5, digits 15
This is not a classic position since digit 4 in set A does not appear in set C. However, it's not difficult to see that only two of the basic Sue de Coq constraints, that sets A and B have no common candidate digits, and that the total cell count = total distinct candidate digit count, are enough to make the usual subset counting argument behind Sue de Coq eliminations. In this case, with respect to the set of five cells in the pattern (A union B union C), each of the digits 1 and 5 have max multiplicity 1 since these candidates all lie in column 5; the remaining digits (2, 4, and 6) in the pattern all lie in box 5 so each of these also has max multiplicity of 1. Since we have only 5 digits which can be used to fill the 5 cells of the pattern, and each of these digits has max multiplicity of 1, in order to fill all cells each digit must appear (exactly once) in the pattern. This gives the usual eliminations in the primary line and box of the Sue de Coq pattern; also, in this case, since the digit 4 candidates in the pattern lie in row 5 only (r5c46), digit 4 can be eliminated in other cells of row 5, as shown.

Sudocue version 3.1.0.1 does not find this, but interestingly enough it does find the ALS XZ rule elimination of (2)r4c4 using the sets A and {B union C}.

So, Ruud, if you decide to investigate the logic for detecting classic Sue de Coq patterns, you might give consideration to extending the logic to find non-classic patterns.

Here's another classic position which Sudocue 3.1.0.1 does not find. It's a six cell pattern which arises in Unsolvable #16 after initial basic eliminations and two naked quads.

Start position, Unsolvable #16:

.9.3.......7...6......24.3.91......8.........4....5.27.5.87..6...1...5.....5.6.9.

Code:

·--------------------------·--------------------·--------------------·
| C12568    9       24-568 | 3     1568   178   | 12478  1478  1245  |
| C12358    234-8   7      | 19    1589   189   | 6      148   12459 |
| C1568    A68     A568    | 1679  2      4     | 178    3     159   |
·--------------------------+--------------------+--------------------·
|  9        1       2356   | 2467  346    237   | 34     45    8     |
|  578-23   2378    2358   | 1249  13489  12389 | 1349   145   6     |
|  4        368     368    | 169   13689  5     | 139    2     7     |
·--------------------------+--------------------+--------------------·
| B23       5       9      | 8     7      123   | 124    6     1234  |
|  678      678     1      | 249   349    239   | 5      78    23    |
|  78-23    23478   2348   | 5     13     6     | 78     9     123   |
·--------------------------·--------------------·--------------------·

    set C (in box 1/column 1 intersection) = r123c1, digits 123568
    set A (in box 1) = r3c23, digits 568
    set B (in column 1) = r7c1, digits 23
This gives the eliminations shown in box 1 and column 1.
Back to top
View user's profile Send private message
Display posts from previous:   
Post new topic   Reply to topic    SudoCue Users Forum Index -> Releases All times are GMT
Goto page Previous  1, 2
Page 2 of 2

 
Jump to:  
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