Exotic patterns a resume

Advanced methods and approaches for solving Sudoku puzzles

Re: Exotic patterns a resume

Postby daj95376 » Sat Sep 07, 2013 3:38 pm

champagne wrote:
JC Van Hay wrote:SF(5B3C43) :=> -5r4c8 means Swordfish(5B3C43) : on 5s : r1c8=r2c9-r2c4=r6c4-r6c3=r4c3 :=> -5r4c8

OK The chain makes sense for me, but I can't see that as a swordfish.

An X-Chain for the Swordfish is typical for a manual solution by JC. It can be translated into a fish by tracking the strong links -- [b3], [c4], and [c3]. Specifically, SF(5B3C43) translates into:

Code: Select all
finned Franken Swordfish c34b3\r26 4   w/fin cell = r1c8     r4c8<>5   -or-
finned mutant  Swordfish c34b3\r26c8   w/fin cell = r4c3     r4c8<>5

You'll notice that the endpoints of his X-Chain become the (respective) fin cell for each Swordfish.
daj95376
2014 Supporter
 
Posts: 2624
Joined: 15 May 2006

Re: Exotic patterns a resume

Postby JC Van Hay » Sat Sep 07, 2013 4:43 pm

Analysis of the last puzzle in the Champagne's list of ADE :

SE 11.0 : ........1..2..3..4.3.14..5.......13....6...7..1.73...5.4..1.5..3.1..8...9....6...;686798;dob;12_12_19;11;345 ;68;2689

#1. sk-basics; UP29

Details :
r2c1,r5c6,r9c8=1 [Solved: 1]; r1c7,r5c3=3 [Unsolved: XW(3r79c49)]; UP29
LC(4r89c4) :=> -4r4c4; FXWing(5C15) :=> -5r4c23

Observations :
a. XChain[4] on 4s weakly coupled to XChain[2] on 5s in r1c3,r4c6 and to XChain[2] on 3s in r9c4.
b. No bivalue.
c. Potential exocets : (6789)r2c78,r3c13 and (2689)r12c8,r45c9

:: Chains on 4s and 5s ...
Code: Select all
+--------------------------+-----------------------+-----------------------+
| 5678(4)  56789  6789(45) | 2589     256789  2579 | 3       2689   1      |
| 1        56789  2        | 589      56789   3    | 6789    689    4      |
| 678      3      6789     | 1        4       279  | 26789   5      26789  |
+--------------------------+-----------------------+-----------------------+
| 245678   26789  46789    | 2589     2589    2459 | 1       3      2689   |
| 258(4)   2589   3        | 6        2589    1    | 289(4)  7      289    |
| 2468     1      4689     | 7        3       249  | 24689   24689  5      |
+--------------------------+-----------------------+-----------------------+
| 2678     4      678      | 239      1       279  | 5       2689   236789 |
| 3        2567   1        | 2459     2579    8    | 24679   2469   2679   |
| 9        2578   78(5)    | 23-5(4)  257     6    | 278(4)  1      2378   |
+--------------------------+-----------------------+-----------------------+
#2. Chain[4] : 4r9c4=4r9c7-4r5c7=4r5c1-4r1c1=(4-5)r1c3=5r9c3 :=> -5r9c4
Code: Select all
+--------------------------+-------------------------+-----------------------+
| 4678(5)   56789  456789  | 289(5)   256789  279(5) | 3       2689   1      |
| 1         56789  2       | 89(5)    56789   3      | 6789    689    4      |
| 678       3      6789    | 1        4       279    | 26789   5      26789  |
+--------------------------+-------------------------+-----------------------+
| 2678(45)  26789  6789(4) | 289(5)   2589    29(45) | 1       3      2689   |
| 28(45)    2589   3       | 6        2589    1      | 289(4)  7      289    |
| 2468      1      4689    | 7        3       249    | 24689   24689  5      |
+--------------------------+-------------------------+-----------------------+
| 2678      4      678     | 239      1       279    | 5       2689   236789 |
| 3         2567   1       | 29-4(5)  2579    8      | 24679   2469   2679   |
| 9         2578   578     | 23(4)    257     6      | 278(4)  1      2378   |
+--------------------------+-------------------------+-----------------------+
#3. Chain[6] : 4r9c4=5r8c4 :=> -4r8c4; 3 Singles : r9c4=4, r7c4,r9c9=3 [Solved : 3]
Code: Select all
4r8c4
4r9c4=4r9c7
======4r5c7=4r5c1
============4r4c13=4r4c6
===================5r4c6=5r1c6
============5r5c1========5r1c1==5r4c1
5r8c4====================5r12c4=5r4c4
... ?
#4. Double exocet : r12c8,r45c9 contain 2or6or8or9 -> Invalid Swordfish in rows 367 :=> -2689r3c9,r6c8; UP81

PS

Interpretation of #3. : Kraken 5C146 : InvalidSwordfish(5r12c146)=5r5c1=528c4 -> ...
5r8c4=*[4r4c13=(4-5)r4c6=5r1c6-5r2c4=*5r5c1]-4r5c1=4r5c7-4r9c7=4r9c4 :=> -4r8c4(r9c4=4)

while, after some optimization, pjb's approach below may be identically interpreted as
Code: Select all
+----------------------------+--------------------------+-----------------------+
| 45678     56789    456789  | 2589     256789   279(5) | 3       2689   1      |
| 1         6789(5)  2       | 89(5)    6789(5)  3      | 6789    689    4      |
| 678       3        6789    | 1        4        279    | 26789   5      26789  |
+----------------------------+--------------------------+-----------------------+
| 25678(4)  26789    6789(4) | 2589     2589     29(45) | 1       3      2689   |
| 28(45)    289(5)   3       | 6        289(5)   1      | 289(4)  7      289    |
| 2468      1        4689    | 7        3        249    | 24689   24689  5      |
+----------------------------+--------------------------+-----------------------+
| 2678      4        678     | 239      1        279    | 5       2689   236789 |
| 3         267(5)   1       | 29-4(5)  279(5)   8      | 24679   2469   2679   |
| 9         2578     578     | 235(4)   257      6      | 278(4)  1      2378   |
+----------------------------+--------------------------+-----------------------+
Chain[7] : Kraken 5R258 : InvalidSwordfish(5r258c25)=5r5c1=5r28c4 -> ...
5r8c4=*[4r4c13=(4-5)r4c6=5r1c6-5r2c4=*5r5c1]-4r5c1=4r5c7-4r9c7=4r9c4 :=> -4r8c4(r9c4=4)
Last edited by JC Van Hay on Sun Sep 08, 2013 9:05 am, edited 1 time in total.
JC Van Hay
 
Posts: 719
Joined: 22 May 2010

Re: Exotic patterns a resume

Postby David P Bird » Sat Sep 07, 2013 11:07 pm

Here's my approach to JC's puzzle above

If the (2689)DJE is true, (4)r6c8 is true
Hence (4)r6c8 = (29)spoiler:r7c4 - (3)r7c4 = (3-4)r9c4 = (4)r8c4 => r8c8 <> 4
The eliminations that follow include the spoiler digits so the DJE is proved true.

This is interesting as a single spoiler cell contains two digits that don't comply with the partial fish requirements.

David
David P Bird
2010 Supporter
 
Posts: 1043
Joined: 16 September 2008
Location: Middle England

Re: Exotic patterns a resume

Postby pjb » Sun Sep 08, 2013 4:25 am

Another approach: State after basics and finned X-Wing of 5s at r14c16, r5c1

Code: Select all
45678  56789  456789 | 2589   256789 2579   | 3      2689   1     
1      56789  2      | 589    56789  3      | 6789   689    4     
678    3      6789   | 1      4      279    | 26789  5      26789 
---------------------+----------------------+---------------------
245678 26789  46789  | 2589   2589   2459   | 1      3      2689   
2458   2589   3      | 6      2589   1      | 2489   7      289   
2468   1      4689   | 7      3      249    | 24689  24689  5     
---------------------+----------------------+---------------------
2678   4      678    | 239    1      279    | 5      2689   236789
3      2567   1      | 2459   2579   8      | 24679  2469   2679   
9      2578   578    | 2345   257    6      | 2478   1      2378   

Then whether 4 is true in r5c1 or r5c7, the 4 at r9c4 is true.
A:
(4) r5c7 - r9c7 = r9c4;
B:
1. (4) r5c1 - r1c1 = (4-5) r1c3 = r9c3 => 5 false at r9c45
2. (4) r5c1 - r4c13 = (4-5) r4c6 = r1c6 - r1c123 = r2c2 - r5c2 = r5c5 => 5 at r8c5 is false
3. Since 5 is false at r9c45, r8c5, 5 is true at r8c4
4. (4) r8c4 = r9c4

Following this there is a double Exocet of 2689 which cracks the puzzle.
Phil
pjb
2014 Supporter
 
Posts: 2672
Joined: 11 September 2011
Location: Sydney, Australia

Re: Exotic patterns a resume

Postby champagne » Sun Sep 08, 2013 9:39 am

JC Van Hay wrote:Analysis of the last puzzle in the Champagne's list of ADE :

SE 11.0 : ........1..2..3..4.3.14..5.......13....6...7..1.73...5.4..1.5..3.1..8...9....6...;686798;dob;12_12_19;11;345 ;68;2689

4r9c4=4r9c7
======4r5c7=4r5c1
============4r4c13=4r4c6
===================5r4c6=5r1c6
============5r5c1========5r1c1==5r4c1
5r8c4====================5r12c4=5r4c4

#4. Double exocet : r12c8,r45c9 ...

Hi JC,

All that is very interesting seen from my side.

I am modifying the search of JE's and other exotic patterns to include your findings.

In the entire file, I got 13 new double Jexocets and 2 new single JE if the search is launched after XY chains.

In that example, the search has to be done after what is called "dynamic expansion" in serate mode. I'll add that step ASAP.

We are here in the low part of the ratings and the JE (likely the same would be seen with other patterns) appears with more cells solved. For anybody having studied the rating process in serate, this is not surprising. The lowest is the number of candidates, the lowest is the chance to reach a total number of candidates involved in any elimination.


Another interesting point is that, following David's and JPB' attacks, it's seems possible to have specific paths, but this would require much more work to be applied by the computer.
champagne
2017 Supporter
 
Posts: 7465
Joined: 02 August 2007
Location: France Brittany

Re: Exotic patterns a resume

Postby champagne » Sun Sep 08, 2013 1:31 pm

I added solving techniques till dynamic plus (subject to more control) and now 55 puzzles with an exocet pattern have been seen

all have a double exocet except these 3

........1.....2.....3.4..5.....562....52...63.76.3......4..3..63...2.8..9..4..1..;352623;dob;12_12_03;r4c4 r6c4 r3c6 r7c5 1789
........1.....2.....3.4..5.....2.6....4..3..7.8.4..1....52...733...572..9.7.3....;434949;dob;12_12_03;r8c4 r9c4 r3c6 r5c5 1689
...........1..2..3.4..5..16.....7.5.....1..343....58....45..9...56......1...6..4.;798639;DOB;13_01;r4c3 r6c3 r3c1 r9c2 2789


I'll split the file keeping these 55 puzzles in a separate file

EDIT : the third in that list is very easy with a specific position
champagne
2017 Supporter
 
Posts: 7465
Joined: 02 August 2007
Location: France Brittany

Re: Exotic patterns a resume

Postby David P Bird » Mon Sep 09, 2013 11:01 am

Here are the 'exemplars' for the Almost Double JExocet pattern with 4 base digits and 4 target cells when a single base digit can occur in 3 parallel lines in the partial fish cells.
The pattern used is (1234)DoubleExocet[a:r1c12,r2c7,r3c5][b:r3c89,r1c5,r2c3]
The partial fish columns are 3,5,& 7 where (1) occurs 3 times in the other two bands and (234) each occur twice.

Code: Select all
*-------------*-------------*-------------*     *-------------*-------------*-------------*   
| 12  12  5   | 347 34  347 | 689 689 689 |     | 12  12  5   | 347 34  347 | 689 689 689 |
| 347 347 34  | 689 689 689 | 12  125 125 |     | 346 346 34  | 589 589 589 | 12  127 127 |
| 689 689 689 | 125 12  125 | 7   34  34  |     | 789 789 789 | 126 12  126 | 5   34  34  |
*-------------*-------------*-------------*     *-------------*-------------*-------------*
     Both JEs True (Repeating Triples /)             Both JEs True (Repeating Pairs /)   

*-------------*-------------*-------------*     *-------------*-------------*-------------*   
| 12  12  5   | 347 34  347 | 689 689 689 |     | 12  12  5   | 347 34  347 | 689 689 689 |
| 347 347 34  | 689 689 689 | 25  15  15  |     | 346 346 34  | 589 589 589 | 27  17  17  |
| 689 689 689 | 15  25  15  | 7   34  34  |     | 789 789 789 | 16  26  16  | 5   34  34  |
*-------------*-------------*-------------*     *-------------*-------------*-------------*
    (34)JE True (Repeating Triples /)                (34)JE True (Repeating Pairs /)   
    (12)JE false (1 in neither target)               (12)JE False (1 in neither target)         

*-------------*-------------*-------------*     *-------------*-------------*-------------*   
| 12  12  5   | 479 479 479 | 368 368 368 |     | 12  12  5   | 489 489 489 | 367 367 367 |
| 478 478 478 | 13  6   13  | 256 256 256 |     | 467 467 467 | 13  5   13  | 289 289 289 |
| 369 369 369 | 258 258 258 | 7   14  14  |     | 389 389 389 | 267 267 267 | 5   14  14  |
*-------------*-------------*-------------*     *-------------*-------------*-------------*
     Both False (Repeating Pairs /)                  Both False (Repeating Pairs \)
     (1 in both base cells)                          (1 in both base cells)

In 4 of the 6 cases the digits repeat in the mini-rows as pairs or triples in the same diagonal direction.

Observations:

1. Commonly the companion cells, r1c3, r2c5, and r3c7, will be singles and if they hold two instances of one digit and one of another, at least one of the component patterns must be true (which will occur reasonably frequently)

2. One pattern must also be true when r1c7 and r3c3 cannot hold a base digit between them. In the configuration shown, when the pattern is false, (3) must be true in c3b58, so with only one of columns 3 & 7 capable of being satisfied in the partial fish, at least one of these cells would have to be true. It also follows that whether the DJE is true or false, any base digit eliminated from both these cells can be eliminated completely from mini-lines r1b3,r2b2,r3b1.

3. Similarly whether the DJE is true or false, digits 2,3,4 (the unspoilt base digits) always travel in the / diagonal direction between mini-rows in the band. Hence if one of them has been eliminated from one mini-line, it can also be eliminated from two others.

4. For the double pattern to be false. the spoiler digit, (1) must be true 3 times in the partial fish as it can't be true in c357 in the DJE band.

5. [Added] When more than two digits can simultaneously occupy 3 parallel lines in the partial fish cells, it's possible for both JEs to be false, with these digits absent from the target cells. In this case it isn't necessary for one of the base digits to occupy both sets of base cells, although this is still a possibility.

With no solver program, I can't check how useful these deductions will prove to be in practice, so would appreciate any feedback.

[Edit 1] Missing cases added for when one JE can be true and the other false.
[Edit 2] Observation 5 added.
Last edited by David P Bird on Wed Sep 25, 2013 10:19 pm, edited 3 times in total.
David P Bird
2010 Supporter
 
Posts: 1043
Joined: 16 September 2008
Location: Middle England

Re: Exotic patterns a resume

Postby David P Bird » Mon Sep 09, 2013 11:02 am

Champagne, as a follow-on to my previous post:

I believe at one time you were checking for JExocet patterns before conducting a complete Exocet analysis, but I'm not sure if you are still doing this. If you aren't, then those exemplar deductions will have limited value for you. What's the situation?

David
David P Bird
2010 Supporter
 
Posts: 1043
Joined: 16 September 2008
Location: Middle England

Re: Exotic patterns a resume

Postby champagne » Mon Sep 09, 2013 3:18 pm

David P Bird wrote:Champagne, as a follow-on to my previous post:

I believe at one time you were checking for JExocet patterns before conducting a complete Exocet analysis, but I'm not sure if you are still doing this. If you aren't, then those exemplar deductions will have limited value for you. What's the situation?

David


The JE search remains the first task when I am looking for exotic pattern. It is also currently the only exotic pattern active in the solver part of my program.

Handling partial exocets (I don't like that much ADE because the situation is very different whether you have 3 digits or less filling the property) remains very difficult and I have a limited use of such patterns in my old solver. In the new one, I just see them, but I am not yet to the point to use them.

I had a look to your tables. IMO, you have only a small part of the possible patterns with partial exocets. In the first table for example, nothing force 1 to be in the targets r2c7 r3c5 if it is not a JE digit.

On one side, I am convinced that you can extract some efficient associated properties for a manual solver, on the other side, I see it of little interest in the solver due to the quantity of code to add.
champagne
2017 Supporter
 
Posts: 7465
Joined: 02 August 2007
Location: France Brittany

Re: Exotic patterns a resume

Postby David P Bird » Mon Sep 09, 2013 5:17 pm

Champagne wrote:I had a look to your tables. IMO, you have only a small part of the possible patterns with partial exocets. In the first table for example, nothing force 1 to be in the targets r2c7 r3c5 if it is not a JE digit.

Thanks. Reviewing this, I do think I've missed a case but not the one you describe.
If the (34)JE is true, then (1) needn't occupy either of the targets for the potential (12)JE and (2) could occupy one or two of the them.
I've edited my post to include this. I can't see your case at all though, perhaps you could give me an example.

Champagne wrote: On one side, I am convinced that you can extract some efficient associated properties for a manual solver, on the other side, I see it of little interest in the solver due to the quantity of code to add.

In that case, it's much as I thought, whatever I find won't be of any interest to you (well at least currently).
David P Bird
2010 Supporter
 
Posts: 1043
Joined: 16 September 2008
Location: Middle England

Re: Exotic patterns a resume

Postby David P Bird » Mon Sep 16, 2013 9:57 am

Here's an interesting Almost Double JExocet from Champagne's list with one digit providing two spoilers:

........1.....2.....3.4..5....6..7....8.3..4..15.8.9.6..94.8..353..9....8......9.;521159;dob;12_12_03;152;1267 ;167
Code: Select all
 *-------------------------*-------------------------*-------------------------*
 | 24679   2456789 2467#   | 35789   567 #   35679   | 23468   2678#   <1>     | 267
 | 14679   456789  1467#   | 135789  1567#   <2>     | 3468    678 #   4789    | 167
 | 12679   26789   <3>     | 1789    <4>     1679    | 268     <5>     2789    |
 *-------------------------*-------------------------*-------------------------*
 | 2349    249     24  #   | <6>     125 #   1459    | <7>     1238#   258     | 12
 | 2679    2679    <8>     | 12579   <3>     1579    | 125     <4>     25      |
 | 2347    <1>     <5>     | 27      <8>     47      | <9>     23  #   <6>     | 2
 *-------------------------*-------------------------*-------------------------*
 | 1267a   267a    <9>     | <4>     12567   <8>     | 1256    1267b   <3>     |
 | <5>     <3>     12467   | 127b    <9>     167b    | 12468   12678a  2478    |
 | <8>     2467    12467b  | 12357   12567a  13567   | 12456   <9>     2457    |
 *-------------------------*-------------------------*-------------------------*
                    S                 S                         S

(1267)Almost Double JE:[a=r7c12,r8c8,r9c5],[b=r8c46,r7c8,r9c3]

The spoilers are A, (2)r4c5, & B, (2)r6c8, as if either of them were false, the partial fish cells could only accommodate two (2) truths restricted to either r14 or c38, and the DJE would be made.

This chain then proves the DJE is good as one or other of these spoilers must be false:
DJE = (2)spoilerA:r4c5 – (2)r4c123 = (267-3)r5c12,r6c1 = (3-2)spoilerB:r6c8 = DJE
David P Bird
2010 Supporter
 
Posts: 1043
Joined: 16 September 2008
Location: Middle England

Re: Exotic patterns a resume

Postby sultan vinegar » Mon Sep 16, 2013 12:47 pm

I agree that 2r6c8 is a spoiler, but not 2r4c5.
sultan vinegar
 
Posts: 81
Joined: 27 August 2013

Re: Exotic patterns a resume

Postby David P Bird » Mon Sep 16, 2013 2:02 pm

sultan vinegar wrote:I agree that 2r6c8 is a spoiler, but not 2r4c5.

See requirement 3 in the < JExocet Pattern Definition >. In the partial fish cells for (2), if (2)r6c8 is false, only 2 rows can hold truths and if (2)r4c5 is false it becomes only 2 columns.

DPB
David P Bird
2010 Supporter
 
Posts: 1043
Joined: 16 September 2008
Location: Middle England

Re: Exotic patterns a resume

Postby champagne » Mon Sep 16, 2013 2:41 pm

David P Bird wrote:
sultan vinegar wrote:I agree that 2r6c8 is a spoiler, but not 2r4c5.

See requirement 3 in the < JExocet Pattern Definition >. In the partial fish cells for (2), if (2)r6c8 is false, only 2 rows can hold truths and if (2)r4c5 is false it becomes only 2 columns.

DPB



As i am not looking for JE's in that situation, I see it as "sultan vinegar".

The situation is clear in columns 3 5 8 if one of the digits 1;6;7 occupies both bases.

If the digit 2 occupies both, we have still possible in column 8 2r6c8

but 2r4c5 => 7r6c4 => Hidden pair 67 r4c12 => 2r6c1 => <2>r6c8

I did not check whether we go to the end, but then r7c5=5 and r8c3=4
champagne
2017 Supporter
 
Posts: 7465
Joined: 02 August 2007
Location: France Brittany

Re: Exotic patterns a resume

Postby David P Bird » Mon Sep 16, 2013 3:57 pm

Champagne, you post made me think I'd slipped up for a moment! However my deduction is good because:

For a single JE we must show that the partial fish cells can't hold 3 truths for any of the base digits. But this also proves a digit can''t occupy both sets of base cells in a double JEs with 4 target cells. Here if a digit is repeated in the base cells it can't be true in the JE band in any of the cross lines. In two boxes it is eliminated by the base cells and in the third box, the only cell not seen by a base pair, is a companion cell that can't contain a given.

The chain I gave showed that (2)r4c5 and (2)r6c8 couldn't both be true, and so the partial fish cells for (2) would either be reduced to 2 rows or 2 columns and so couldn't hold 3 truths.
David P Bird
2010 Supporter
 
Posts: 1043
Joined: 16 September 2008
Location: Middle England

PreviousNext

Return to Advanced solving techniques