5/13/07 Nightmare

Discuss the <a href="http://www.sudocue.net/daily.php">Daily Sudoku Nightmare</a> here
Post Reply
Sudtyro
Hooked
Hooked
Posts: 49
Joined: Tue Jan 16, 2007 8:17 pm

5/13/07 Nightmare

Post by Sudtyro »

After running through the basics:

Code: Select all

----------------------------------------------------
47    269   147  | 3    17   8    | 269   5    26     
239   23689 1368 | 12   69   5    | 4     2369 7      
23579 23569 367  | 27   4    69   | 8     1    236    
-----------------+----------------+-----------------
1     356   36   | 57   2    4    | 367   8    9      
459   569   2    | 8    357  37   | 1     467  46     
8     7     34   | 6    19   19   | 23    234  5      
-----------------+----------------+-----------------
237   4     9    | 57-1 8    1367 | 23567 2367 1236   
6     238   5    | 9    137  137  | 237   2347 12348  
37    1     378  | 4    3567 2    | 35679 3679 368    
----------------------------------------------------
To advance beyond this point, there are numerous grouped AICs and ALSs available. However, one very productive ALS-XZ rule (from Andrew Stuart’s solver) is:

(1=2378)r8c2567 - (8=23691)r2c12458 => r7c4 <> 1

This particular ALS application was hard to spot manually, so I’m curious to know if there are more obvious alternative solution strategies that provide the same elimination?
Jean-Christophe
Addict
Addict
Posts: 92
Joined: Mon Apr 23, 2007 12:50 pm
Location: Belgium

Post by Jean-Christophe »

Here is an (non grouped) AIC which yields the same elimination:

(1)R2C4=(1)R2C3-(8)R2C3=(8)R2C2-(8)R8C2=(8)R8C9-(1)R8C9=(1)R7C9 -> R7C4<>1

Not sure it's easier to spot, thought
Sudtyro
Hooked
Hooked
Posts: 49
Joined: Tue Jan 16, 2007 8:17 pm

And another question...

Post by Sudtyro »

Thanks, J-C...very nice chain and maybe easier to find, at least for me!

This brings up another general question:
For every grouped AIC, including ALS-rule applications, does there always exist an equivalent non-grouped AIC for the same elimination?
This particular case seems to say yes, but what about, say, a simple XYZ-Wing? Are there always other cells available that allow for a non-grouped AIC?
Jean-Christophe
Addict
Addict
Posts: 92
Joined: Mon Apr 23, 2007 12:50 pm
Location: Belgium

Re: And another question...

Post by Jean-Christophe »

Sudtyro wrote:For every grouped AIC, including ALS-rule applications, does there always exist an equivalent non-grouped AIC for the same elimination?
This particular case seems to say yes, but what about, say, a simple XYZ-Wing? Are there always other cells available that allow for a non-grouped AIC?
No, sometimes one may find other ways or techniques to eliminate the same candidate, but not always.
Ron Moore
Addict
Addict
Posts: 72
Joined: Sun Aug 13, 2006 3:34 am
Location: New Mexico

Uniqueness Argument

Post by Ron Moore »

Sudtyro,

You may not consider this easy to see, but I've found that uniqueness based arguments are often productive, so I tend to give priority to looking for and using such arguments.

Code: Select all

---------------------------------------------------- 
47    269   147  | 3    17   8    | 269   5    26      
239   23689 1368 | 12   69   5    | 4     2369 7      
23579 23569 367  |*2-7  4    69   | 8     1    236    
-----------------+----------------+----------------- 
1     356   36   |#57   2    4    | 367   8    9      
459   569   2    | 8   ^357 ^37   | 1     467  46      
8     7     34   | 6    19   19   | 23    234  5      
-----------------+----------------+----------------- 
237   4     9    |#157  8    1367 | 23567 2367 1236    
6     238   5    | 9   ^137 ^137  | 237   2347 12348  
37    1     378  | 4    3567 2    | 35679 3679 368    
----------------------------------------------------
There is a potential deadly pattern in r58c56 (marked with "^"in the diagram), based on the digits 3,7. To avoid the deadly pattern, we must have a 5 in r5c5 or a 1 in one of r8c56. We can use this fact in an AIC, as follows:
  • (7=5)r4c4 - (5=37)r5c56 - UR - (37=1)r8c56 - (1=57)r47c4 => r3c4 <> 7
Of course, this quickly leads to (or follows from) the elimination you give.
Sudtyro
Hooked
Hooked
Posts: 49
Joined: Tue Jan 16, 2007 8:17 pm

Re: Uniqueness Argument

Post by Sudtyro »

Thanks, RM...that’s a clever and interesting AIC with its use of the potential UR.

[Edit for late addendum]: Perhaps one could even get the original elimination directly from
(5)r7c4 = (5)r4c4 - (5=37)r5c56 - UR - (37=1)r8c56 => r7c4 <> 1
Last edited by Sudtyro on Thu Jun 14, 2007 6:08 pm, edited 1 time in total.
Sudtyro
Hooked
Hooked
Posts: 49
Joined: Tue Jan 16, 2007 8:17 pm

5/13/07 Nightmare - revisited

Post by Sudtyro »

To continue a bit on this thread...
After follow-up and one Turbot Fish, (7): r5c8 = r4c6 - r4c4 = r7c4 => r7c8 <> 7, the unresolved 7’s grid becomes:

Code: Select all

. . . | . . . | . . .
. . . | . . . | . . .
7 . 7 | . . . | . . .
------+-------+------
. . . | 7 . . | 7 . .
. . . | . . 7 | . 7 .
. . . | . . . | . . .
------+-------+------
7 . . | 7 . 7 | 7 . .
. . . | . . 7 | 7 7 .
7 . 7 | . . . | . 7 .
I’m no fish expert, but based on Ron Moore’s recent Nightmare posting (1 June 2007 – “Almost” Swordfish), the above grid appears to host an “almost” X-wing with two possible “spoilers.”

r39c13 would be a column (or row) X-Wing except for the spoiler in r7c1 (or r9c8). Following Ron’s lead, an AIC for the column X-wing plus spoiler is
[X-wing r39c13] = r7c1 – r7c4 = r4c4 - r4c7 = r5c8 => r9c8 <> 7.
But, note that this elimination now makes the “almost” column X-wing a true row X-wing that then forces r7c1 <> 7.

The double elimination suggests that other techniques/patterns are available. E.g., the grouped AIC,
r9c13 = r7c1 - r7c4 = r4c4 - r4c7 = r5c8 => r9c8 <> 7,
yields the same elimination without specifically using r3c13. Still, it seems much easier to initially spot that “almost” X-wing than to pick out the grouped AIC (or a more complex single-digit pattern).
Para
Yokozuna
Yokozuna
Posts: 384
Joined: Wed Nov 08, 2006 7:42 pm
Location: The Netherlands

Post by Para »

Hi

Thought you might be interested to know that there is a Finned X-wing
There is a Finned X-wing on R58 with R8C7 being the fin in the grid before the turbot fish that eliminates 7 from both R79C8.

Code: Select all

. . . | . . . | . . . 
. . . | . . . | . . . 
7 . 7 | . . . | . . . 
------+-------+------ 
. . . | 7 . . | 7 . . 
. . . | . . 7*| . 7*. 
. . . | . . . | . . . 
------+-------+------ 
7 . . | 7 . 7 | 7 7-. 
. . . | . . 7*| 7#7*. 
7 . 7 | . . . | . 7-.
greetings

Para
Sudtyro
Hooked
Hooked
Posts: 49
Joined: Tue Jan 16, 2007 8:17 pm

Post by Sudtyro »

Thanks, Para...you are "da Master!"
Memo to self: Submit all future posts before happy hour!
Myth Jellies
Hooked
Hooked
Posts: 42
Joined: Tue Apr 04, 2006 7:07 am

Post by Myth Jellies »

Having additional ways to see easily overlooked reductions is no bad thing. :)
Sudtyro
Hooked
Hooked
Posts: 49
Joined: Tue Jan 16, 2007 8:17 pm

Post by Sudtyro »

MJ,

Regarding your comment, I just recently ran across arcilla’s “a new(?) view of fish (naked or hidden)” 11/3/2006 thread in the other forum. It appears that his technique would very neatly find both the finned and the “almost” X-Wings in the 7’s grid.
Arcilla’s lists for Para's grid would appear as:

Row numbers, per col: (379)(-)(39)(47)(-)(578)(478)(5789)(-)
Col numbers, per row: (-)(-)(13)(47)(68)(-)(14678)(678)(138)

The row numbers reveal c1’s (379) and c3’s (39), for one of the “almost” X-Wings, r7c1 being the spoiler.
The col numbers reveal r3’s (13) and r9’s (138), for the other “almost” X-Wing, r9c8 being the spoiler.
The col numbers also reveal r5’s (68) and r8’s (678), for the finned X-Wing, r8c7 being the fin.
[Edit]: Just noticed that the row numbers also reveal c4's (47) and c7's (478), for yet another finned X-Wing, r8c7 again being the fin.

That seems pretty amazing!
Last edited by Sudtyro on Wed Jul 04, 2007 10:41 am, edited 1 time in total.
Myth Jellies
Hooked
Hooked
Posts: 42
Joined: Tue Apr 04, 2006 7:07 am

Post by Myth Jellies »

Yes, Arcilla's trick converts fish into either hidden or naked sets. In Arcilla's representation, Almost Locked Sets and Almost Hidden Sets will correspond to finned fish, although the fin may not occupy the same box as any of the fish vertices (which is hard to see in Arcilla's representation) and so might not be immediately useful. Also note that some finned fish can have 2 to 4 candidates in the "fin" and still be immediately useful. These will translate into something more complicated than an ALS, etc.
Novajlija
Rookie
Rookie
Posts: 5
Joined: Sat Nov 17, 2007 1:40 pm

Post by Novajlija »

Number 4 is LOCKED in R1C1 and R1C3. If there is naked pair 4-7 then we have R1C5=1, R3C4=7, R4C7=7 and R6C3=4. So we ken eliminate number 7 from R1C1 or R1C1=4.

There is few more this kind of elimination...
Post Reply