HoDoKu

Programs which generate, solve, and analyze Sudoku puzzles

Postby hobiwan » Fri Feb 06, 2009 10:50 pm

I only consider ERs valid, if they contain at least three candidates. With two candidates the ER degenerates into an X-Chain (or a Franken X-Wing). I am not entirely sure whether I should change that.

Whats your view?

[edit: corrected wrong answer]
hobiwan
2012 Supporter
 
Posts: 321
Joined: 16 January 2008
Location: Klagenfurt

Postby ronk » Fri Feb 06, 2009 11:05 pm

hobiwan wrote:I only consider ERs valid, if they contain at least three candidates.
Ditto for me.
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

Postby wintder » Sat Feb 07, 2009 12:59 am

hobiwan wrote:I only consider ERs valid, if they contain at least three candidates. With two candidates the ER degenerates into an X-Chain (or a Franken X-Wing). I am not entirely sure whether I should change that.

Whats your view?

[edit: corrected wrong answer]


You name and mark 2-string kite and skyscraper, two of the three turbots.
The missing turbot is this "ER" mentioned above.

I'd like to see it listed as a Turbot.

I would prefer the 3 Turbots to be shown distinct from ER, in fact I would really like that.
wintder
 
Posts: 297
Joined: 24 April 2007

Postby hobiwan » Sat Feb 07, 2009 3:35 am

wintder wrote:I'd like to see it listed as a Turbot.

I would prefer the 3 Turbots to be shown distinct from ER, in fact I would really like that.

I can implement Turbot as seperate technique (disabled by default), then you can configure the solver as you like.

I have three or four techniques I want to add/overhaul first, I'll tackle turbot after that.
hobiwan
2012 Supporter
 
Posts: 321
Joined: 16 January 2008
Location: Klagenfurt

Postby wintder » Sat Feb 07, 2009 4:30 am

hobiwan wrote:
wintder wrote:I'd like to see it listed as a Turbot.

I would prefer the 3 Turbots to be shown distinct from ER, in fact I would really like that.

I can implement Turbot as seperate technique (disabled by default), then you can configure the solver as you like.

I have three or four techniques I want to add/overhaul first, I'll tackle turbot after that.




As far as ER being restricted to +2 candidates.

ER is venerable. It does not have such a restriction. It is less old than turbot.

Ronk has strong preferences. They are his own.
wintder
 
Posts: 297
Joined: 24 April 2007

Postby ronk » Sat Feb 07, 2009 4:56 am

wintder wrote:As far as ER being restricted to +2 candidates.

ER is venerable. It does not have such a restriction. It is less old than turbot.

Ronk has strong preferences. They are his own.

To be consistent with that philosophy, we should list ALS xz-rule as an alternate term for naked pair.:)
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

Postby wintder » Sat Feb 07, 2009 9:46 pm

I have seen this solver list a naked pair elimination as a skyscraper.
It is all a matter of perspective, I suppose.:)

It did list the naked pair well before the other.
wintder
 
Posts: 297
Joined: 24 April 2007

Postby hobiwan » Sat Feb 07, 2009 11:08 pm

I will make ERs with only two candidates optional in the next release but the option will be disabled by default.
hobiwan
2012 Supporter
 
Posts: 321
Joined: 16 January 2008
Location: Klagenfurt

re: degenerate cases

Postby Pat » Sat Feb 07, 2009 11:46 pm

wintder wrote:
hobiwan wrote:
wintder wrote:In this markup,
the ER isn't listed as a "possible" --

Code: Select all
.--------------.---------------.--------------.
| 7   28   89  | 236   5    4  | 39   1    68 |
| 19  6    3   | 8     7    19 | 4    2    5  |
| 5   128  4   | 1236  136  69 | 7    39   68 |
:--------------+---------------+--------------:
| 2   7    56  | 4     36   58 | 389  369  1  |
| 4   13   156 | 9     2    58 | 38   36   7  |
|#69  38   89  | 13-6  136  7  | 5    4    2  |
:--------------+---------------+--------------:
| 8   5    2   |*16    4    3  | 16   7    9  |
| 3   9    16  | 7     8    2  | 16   5    4  |
|*16  4    7   | 5     9   *16 | 2    8    3  |
'--------------'---------------'--------------'



I only consider an ER valid, if it contains at least 3 candidates.
With only 2 candidates, the ER degenerates into an X-Chain (or a Franken X-Wing).
I am not entirely sure whether I should change that.
What's your view?


As far as ER being restricted to 2+ candidates -- ER does not have such a restriction.


i go along with wintder's view,
as i generally prefer the inclusion of degenerate cases
User avatar
Pat
 
Posts: 3440
Joined: 18 July 2005

Re: re: degenerate cases

Postby hobiwan » Sun Feb 08, 2009 12:05 am

Pat, thanks for the links (I have always troubles finding something). As to the ER, I will stick with the option (mostly because I have already done it).

Pat wrote:this finned Mutant X-wing
    c1b8\r9c4 with fin=r6c1
    ==> exclude 6 at r6c4

I prefer
Finned Franken X-Wing: 6 c1b8 r69 fr7c4 => r6c4<>6
but my solver will find both.
hobiwan
2012 Supporter
 
Posts: 321
Joined: 16 January 2008
Location: Klagenfurt

Postby wintder » Wed Feb 11, 2009 9:34 pm

Hi again!

Code: Select all
.-----------------.----------------.-----------------.
| 1235  1235  9   | 7   6     135  | 13    8     4   |
| 135   6     15  | 4   8     1359 | 139   2     7   |
| 4     7     8   | 23  29    139  | 5     6     139 |
:-----------------+----------------+-----------------:
| 8     25    57  | 1   27    39   | 4     39    6   |
| 139   134   6   | 8   5     47   | 2     1379  139 |
| 1239  123   147 | 23  2479  6    | 1379  5     8   |
:-----------------+----------------+-----------------:
| 15    145   2   | 6   3     47   | 8     179   159 |
| 7     8     35  | 9   1     2    | 6     4     35  |
| 6     9     134 | 5   47    8    | 137   13    2   |
'-----------------'----------------'-----------------'


In this markup I don't see this xy-chain listed except as ALS variants.

Code: Select all
.-----------------.----------------.-----------------.
| 1235  1235  9   | 7   6     135  | 13    8     4   |
| 135   6    #15  | 4   8     1359 | 139   2     7   |
| 4     7     8   | 23  29    139  | 5     6     139 |
:-----------------+----------------+-----------------:
| 8     25   #57  | 1  #27   #39   | 4    #39    6   |
| 139   134   6   | 8   5     47   | 2     1379  139 |
| 1239  123   147 |#23  2479  6    | 1379  5     8   |
:-----------------+----------------+-----------------:
| 15    145   2   | 6   3     47   | 8     179   159 |
| 7     8     35  | 9   1     2    | 6     4     35  |
| 6     9     34-1| 5   47    8    | 137  #13    2   |
'-----------------'----------------'-----------------'



Almost Locked Set XY-Wing: A=r248c3 - {1357}, B=r49c8 - {139}, C=r4c56,r6c4 - {2379}, X,Y=7,9, Z=1 => r9c3<>1

Thanks.
wintder
 
Posts: 297
Joined: 24 April 2007

Postby hobiwan » Wed Feb 11, 2009 11:47 pm

wintder wrote:In this markup I don't see this xy-chain listed except as ALS variants.

Could you please check again, because I do get it (last entry before the Discontinuous Nice Loops):
Code: Select all
.------------------.-------------------.------------------.
| 1235  1235   9   |  7    6      135  | 13     8     4   |
| 135   6     *15  |  4    8      1359 | 139    2     7   |
| 4     7      8   |  23   29     139  | 5      6     139 |
:------------------+-------------------+------------------:
| 8     25    *57  |  1   *27    *39   | 4     *39    6   |
| 139   134    6   |  8    5      47   | 2      1379  139 |
| 1239  123    147 | *23   2479   6    | 1379   5     8   |
:------------------+-------------------+------------------:
| 15    145    2   |  6    3      47   | 8      179   159 |
| 7     8      35  |  9    1      2    | 6      4     35  |
| 6     9     -134 |  5    47     8    | 137   *13    2   |
'------------------'-------------------'------------------'

XY-Chain: 1 r2c3 -5- r4c3 -7- r4c5 -2- r6c4 -3- r4c6 -9- r4c8 -3- r9c8 => r9c3<>1

Other variant listed (besides the ALS you mentioned):
Code: Select all
.-------------------.-------------------.------------------.
| 1235  1235   9    |  7    6      135  | 13     8     4   |
| 135   6      15   |  4    8      1359 | 139    2     7   |
| 4     7      8    |  23   29     139  | 5      6     139 |
:-------------------+-------------------+------------------:
| 8     25    *57   |  1   *27    *39   | 4     *39    6   |
| 139   134    6    |  8    5      47   | 2      1379  139 |
| 1239  123   *147  | *23   2479   6    | 1379   5     8   |
:-------------------+-------------------+------------------:
| 15    145    2    |  6    3      47   | 8      179   159 |
| 7     8      35   |  9    1      2    | 6      4     35  |
| 6     9     *-134 |  5    47     8    | 137   *13    2   |
'-------------------'-------------------'------------------'

Discontinuous Nice Loop r9c3 -1- r9c8 -3- r4c8 =3= r4c6 -3- r6c4 -2- r4c5 -7- r4c3 =7= r6c3 =4= r9c3 => r9c3<>1

I have thought about implementing different sort orders in the "All Steps" dialog, but that would mean some major changes to the code, so dont expect that anytime soon.
hobiwan
2012 Supporter
 
Posts: 321
Joined: 16 January 2008
Location: Klagenfurt

Postby wintder » Thu Feb 12, 2009 6:00 am

hobiwan wrote:Could you please check again, because I do get it (last entry before the Discontinuous Nice Loops):


Not there for me running winxp and version 1.0 of the solver.

Edited to add:

I reset to defaults and the chain appeared. I'll tinker around Friday to see if I can puzzle out what stopped it from appearing.:)

It was that I set the chain length to 12 from the default 20.
I'd have thought 12 was enough.:)
wintder
 
Posts: 297
Joined: 24 April 2007

Postby hobiwan » Mon Mar 02, 2009 11:00 pm

Version 1.1 is now out!

New features:
    Kraken fish
    Turbot fish
    ERs with only two candidates in the ER
    Doubly linked ALS-XZ
    SDC with the same additional candidate in both flanking sets
    "Find all steps" is now freely configurable

For Turbot Fish the default search order has been changed a bit. It is now:
    Skycraper
    2-String Kite
    Turbot Fish
    Empty Rectangle
    W-Wing
If Skyscraper and/or 2-String Kite is disabled, the eliminations are reported as Turbot Fish. If Turbot fish is disabled as well, they will be reported as X-Chains.

Chain ordering has been changed (chains without ALS are now treated as beeing shorter than chains with less nodes but large ALS). Plus the AIC algorithm didnt catch some possible eliminations, so solutions to hard puzzles will be different from v1.0.
hobiwan
2012 Supporter
 
Posts: 321
Joined: 16 January 2008
Location: Klagenfurt

Postby wintder » Tue Mar 03, 2009 1:04 pm

Hi!

This is a puzzle from the patterns game posted by Tarek.
Code: Select all
. . .|1 . 7|. . .
. . .|9 . 5|. . .
. . 3|. 8 .|5 . .
-----+-----+-----
. 9 .|. . .|. 7 .
. . 5|2 . 6|4 . .
1 . .|. . .|. . 8
-----+-----+-----
7 . .|. . .|. . 9
. . 6|. 4 .|3 . .
. 2 .|. . .|. 1 .


The sashimi swordfish have only half the available eliminations listed.
I tried versions 1.0 and 1.1.

These sashimi swords should be eliminating like a skyscraper, in fact I mentally call this pattern a sky-sword.

Thanks!
wintder
 
Posts: 297
Joined: 24 April 2007

PreviousNext

Return to Software