The New Sudoku Players' Forum

by **David P Bird** » Thu Aug 22, 2013 9:46 pm

Split off of a new (?) view of fish (naked or hidden). JasonLion-Admin

Blue wrote: The two "r8c9" diagrams that I showed, were twists on two such fish:
r33c2477b4\r56689c399b199 and
r34c247b33\r2689c38999b59.

Agreed, very impressive, and succeeds in blowing my speculations about the relationships between NxN fish, NxM fish, and Nets to pieces.

How well does your method cope with the puzzles in the < NoFish collection > ?

David

by **blue** » Fri Aug 23, 2013 3:26 am

daj95376 wrote:
blue wrote:I have a piece of code that's telling me that the minimum rank for an Obi-Wahn fish for the r8c9 exclusion, is 4, and that the minimum number of base sectors for such a fish, is 7. (It also says that 7 is the minimum number of base sectors for an Obi-Wahn fish of any rank).

The two "r8c9" diagrams that I showed, were twists on two such fish: r33c2477b4\r56689c399b199 and r34c247b33\r2689c38999b59.

Impressive! An NxM fish for an elimination that isn't found by traditional fish patterns.

These do have interpretations as remotely finned fish -- which, I think, was the thrust of Sudtyro2's post.
I won't pretend to know whether "found by traditional methods" applies here.

For the 1st: r33c2477b4\r56689c399b199 => r8c9<>8

Code: Select all: c24b4\r589 remote fins: r2c2-r3c3=r3c9- r46c3-r3c3=r3c9- r6c23-r6c7=r79c7- (note: r6c3 is listed twice)

For the 2nd: r34c247b33\r2689c38999b59 => r8c9<>8

Code: Select all: r4c24\r89b5 remote fins: r2c2-r2c8=r3c9- r4c3-r3c3=r3c9- r4c8-r2c8=r3c9- r6c2-r6c7=r79c7-

##########

David P Bird wrote:How well does your method cope with the puzzles in the < NoFish collection > ?

For tarek's list (NoFish22-NoFish563), it says that for the specified eliminations, no Obi-Wahn fish exist.

I should note that the way I read tarek's post, the template eliminations would be apparent after placing the givens and performing the initial "line of sight" eliminations. I verified that for the first few puzzles, and then assumed it was true for the entire list.

For the earlier list (including the PMs in the post that follows), it finds Obi-Whan fish for just a few cases.

NoFish9: r2c36677b49\r11667999c248b8 => r9c4<>7 (rank 4)
NoFish21: r27c77b2\r1c36b3999 => r8c89<>2 (rank 2)

NoFish4: Obi-Whan fish for r9c8<>9, but not for r4c5<>5, r9c7<>5, or r9c3<>3.
NoFish5: Obi-Whan fish for r8c7<>9, r7c4<>9 and r7c5<>9, but not for r4c8<>5, r9c7<>5, or r9c3<>9.
Correction: NoFish5 was OK, I had used the wrong starting point.
NoFish14: Obi-Whan fish for r3c7<>8 (and r3c8<>8), but not for r3c7<>5 (or r3c8<>5).

For NoFish14, the r3c8 eliminations aren't shown in the PM diagram in the referenced post.
NoFish4 has been mentioned as having "issues", possibly, depending on which version is listed.

by **tarek** » Fri Aug 23, 2013 5:22 pm

blue wrote:For tarek's list (NoFish22-NoFish563), it says that for the specified eliminations, no Obi-Wahn fish exist.

The original list with eliminations was re-posted as it was lost with the old forum along with many posts that analysed these puzzles. Several fish have been found (Mainly large NxN fish, NxN fish with remote fins & NxM fish).

I have to say that ronk sent me -around the time it was originally posted- a list of corrections because it transpired that some of the eliminations can't be obtained by single-digit templates. The list needs reposting with the correct template eliminations before any attempts to re-analyse the puzzles are made.

The list itself is a collection of special puzzles with the 1st elimination being a single-digit template elimination that did not have to appear -around the time of original posting- an NxN fish to explain the elimination.

by **blue** » Fri Aug 23, 2013 9:32 pm

Hi tarek,

Thanks for the heads up.
If you run across fish for any of the specified eliminations in the current list, would you post them please ?

TIA,
Blue.

by **tarek** » Sat Aug 24, 2013 8:07 am

blue wrote:Hi tarek,
Thanks for the heads up.
If you run across fish for any of the specified eliminations in the current list, would you post them please ?

No Problem,

Becuase It turned up that MY template eliminations were not absolutely correct, It will need somebody else to run through the list again and post the correct eliminations. There was a corrected version that was lost with the old forum and unfortunately I don't have a copy of. Even the original list was sent to me by ronk & ttt as I didn't keep a copy.

Unfortunately all the fish that were caught have been lost with the old forum but I'll post any that I come accross.

Tarek

by **David P Bird** » Sat Aug 24, 2013 10:42 am

Here's the copy of the NoFish collection I took but I don't know when (it's in a working spreadsheet). It's clearly from another source though as the notation order at the end of the lines is different.

Hidden Text: Show

The only solutions I have result from a search for impossible cycling fish. Here are the ones I found that make the same eliminations before I lost interest.

Hidden Text: Show

I've been too busy to spend much time on Sudoku recently but at the moment I'm thinking that with duplicated sectors in the base and cover sets there are too many options to search for NxM fish, and forcing nets would be quicker if one is prepared to use them. Once a net elimination is found, it would then rather dishonest to convert it to a fish.

As yet though, I'm still harbouring hopes that a reasonable methodology other than iterating a vast number of options may be available, even if it doesn't haven't have 100% hit rate.

by **daj95376** » Sat Aug 24, 2013 4:48 pm

tarek wrote:
blue wrote:Hi tarek,
Thanks for the heads up.
If you run across fish for any of the specified eliminations in the current list, would you post them please ?

No Problem,

Becuase It turned up that MY template eliminations were not absolutely correct, It will need somebody else to run through the list again and post the correct eliminations. There was a corrected version that was lost with the old forum and unfortunately I don't have a copy of. Even the original list was sent to me by ronk & ttt as I didn't keep a copy.

Unfortunately all the fish that were caught have been lost with the old forum but I'll post any that I come accross.

I ran my solver against Tarek's NoFish -- searching first for other techniques (including Fish) being present ... and subsequently stopping to list any Template eliminations present. Every one of my solutions had Template eliminations matching Tarek's eliminations. There were two puzzles where my solver found additional eliminations to Tarek's eliminations. The additional eliminations are (probably) insignificant.

.7....9....64...3.5....1..2......7..4..3...8...2..8.1..952.......1.9....3...5.8.. # NoFish149 r4c89,r9c9<>9
6..7......1..42.....3.8.9...4.1...3.7...3..8.........2..9...7.......5.1.8..46...3 # NoFish435 r6c8,r8c9,r9c6<>9

These two entries match results posted in DPB's list.

by **daj95376** » Sun Aug 25, 2013 3:25 am

blue wrote:For the earlier list (including the PMs in the post that follows), it finds Obi-Whan fish for just a few cases.

NoFish9: r2c36677b49\r11667999c248b8 => r9c4<>7 (rank 4)
NoFish21: r27c77b2\r1c36b3999 => r8c89<>2 (rank 2)

NoFish4: Obi-Whan fish for r9c8<>9, but not for r4c5<>5, r9c7<>5, or r9c3<>3.
NoFish5: Obi-Whan fish for r8c7<>9, r7c4<>9 and r7c5<>9, but not for r4c8<>5, r9c7<>5, or r9c3<>9.
NoFish14: Obi-Whan fish for r3c7<>8 (and r3c8<>8), but not for r3c7<>5 (or r3c8<>5).

For NoFish14, the r3c8 eliminations aren't shown in the PM diagram in the referenced post.
NoFish4 has been mentioned as having "issues", possibly, depending on which version is listed.

FWIW:

Code: Select all: Obi-Wahn NoFish #21: solvable .......39.5....6....68...4129.5.3.7...5..74....1.......7.16..9.6...85...1........ +-----------------------+ | . . . | . . . | . 3 9 | | . 5 . | . . . | 6 . . | | . . 6 | 8 . . | . 4 1 | |-------+-------+-------| | 2 9 . | 5 . 3 | . 7 . | | . . 5 | . . 7 | 4 . . | | . . 1 | . . . | . . . | |-------+-------+-------| | . 7 . | 1 6 . | . 9 . | | 6 . . | . 8 5 | . . . | | 1 . . | . . . | . . . | +-----------------------+ r3 Naked Triple <> 239 r3c57 r2 b2 Locked Candidate 1 <> 3 r2c13 c9b6 Locked Candidate 1 <> 3 r78c9 c3b7 Locked Candidate 1 <> 9 r2c3 c3b7 Locked Candidate 2 <> 3 r89c2 c1b1 Locked Candidate 2 <> 4 r2c3 c6b8 Locked Candidate 2 <> 4 r8c4,r9c45 r2 Naked Triple <> 278 r2c145 c18 X-Wing f/s <> 8 r1c7,r2c3 all extraneous r14 X-Wing f/s <> 8 r2c3 r24 X-Wing f/s <> 8 r1c7 r47 X-Wing f/s <> 8 r9c7 r49 X-Wing f/s <> 8 r7c3 5-Fish r27c7b26\r1c36 9+r8|b9 Bm 212\230 <> 2 r8c8 extraneous 5-Fish r27c7b26\r1c368 +r8|b9 Bm 212\230 <> 2 r8c9 5-Fish r27c7b26\r1c3689 Bm 212\140 <> 2 r8c89 Solution: 817456239452391687936872541298543176365917428741628953573164892629785314184239765

Code: Select all: Obi-Wahn NoFish #14: solvable 5...4.2...1..2.7....21......7..1...62...3...4..6....7...4.7.69.3.........9.3.1..2 +-----------------------+ | 5 . . | . 4 . | 2 . . | | . 1 . | . 2 . | 7 . . | | . . 2 | 1 . . | . . . | |-------+-------+-------| | . 7 . | . 1 . | . . 6 | | 2 . . | . 3 . | . . 4 | | . . 6 | . . . | . 7 . | |-------+-------+-------| | . . 4 | . 7 . | 6 9 . | | 3 . . | . . . | . . . | | . 9 . | 3 . 1 | . . 2 | +-----------------------+ r5 Naked Pair <> 58 r5c467 c7b6 Locked Candidate 1 <> 3 r3c7 r8 b8 Locked Candidate 1 <> 4 r8c78 r8 Naked Triple <> 158 r8c2456 4-Fish r57c59\r6c2b8+r3|c8|b3 fm 220\211 <> 5 r3c8 extraneous 4-Fish r57c59\r6c2b8+r3|c8|b3 fm 220\211 <> 8 r3c8 extraneous 5-Fish r57c59b9\r6c28b8+r3|b3 fm 221\221 <> 5 r3c7 (+r3c8) 5-Fish r57c59b9\r6c28b8+r3|b3 fm 221\221 <> 8 r3c7 (+r3c8) Solution: 537649218419823765682157439873914526251736984946285371124578693368492157795361842

I'm not sure at what point you are solving NoFish4 and NoFish5.

Code: Select all: Obi-Wahn NoFish #4: ..72....62...4..758.....1..1.4.6.8...2.8....3.....9....3...4..14..6..2.....7..... +-----------------------+ | . . 7 | 2 . . | . . 6 | | 2 . . | . 4 . | . 7 5 | | 8 . . | . . . | 1 . . | |-------+-------+-------| | 1 . 4 | . 6 . | 8 . . | | . 2 . | 8 . . | . . 3 | | . . . | . . 9 | . . . | |-------+-------+-------| | . 3 . | . . 4 | . . 1 | | 4 . . | 6 . . | 2 . . | | . . . | 7 . . | . . . | +-----------------------+ r7 Naked Triple <> 569 r7c35 r6 b4 Locked Candidate 1 <> 3 r6c5 c7b6 Locked Candidate 1 <> 6 r9c7 2-Fish r1b8\c15 fF 101\020 <> 9 r7c1 4-Fish r1c249\r349b1 fm 130\301 <> 9 r9c5 4-Fish r1c29b2\r349b1 Bm 121\301 <> 9 r9c5 4-Fish r1c249\r34c5b1 fm 130\211 <> 9 r9c5 4-Fish r1c249\r34b18 fm 130\202 <> 9 r9c5 +--------------------------------------------------------------------------------+ | 359 1 7 | 2 359 35 | 4 8 6 | | 2 69 369 | 1 4 8 | 39 7 5 | | 8 4 359 | 359 7 6 | 1 239 29 | |--------------------------+--------------------------+--------------------------| | 1 59 4 | 35 6 2357 | 8 259 279 | | 5679 2 569 | 8 15 157 | 569 4 3 | | 3567 568 3568 | 4 25 9 | 56 1 27 | |--------------------------+--------------------------+--------------------------| | 56 3 28 | 59 28 4 | 7 569 1 | | 4 7 159 | 6 1359 135 | 2 359 8 | | 569 5689 125689 | 7 12358 1235 | 359 3569 4 | +--------------------------------------------------------------------------------+ Templates (A: 1) <> 5 r4c8,r9c7 Templates (A: 1) <> 9 r9c38 *** not necessary

Code: Select all: Obi-Wahn NoFish #5: ..4..8......2..1.....5.134.9.....8.64.8...71.751.....383...4........6..75.7...... +-----------------------+ | . . 4 | . . 8 | . . . | | . . . | 2 . . | 1 . . | | . . . | 5 . 1 | 3 4 . | |-------+-------+-------| | 9 . . | . . . | 8 . 6 | | 4 . 8 | . . . | 7 1 . | | 7 5 1 | . . . | . . 3 | |-------+-------+-------| | 8 3 . | . . 4 | . . . | | . . . | . . 6 | . . 7 | | 5 . 7 | . . . | . . . | +-----------------------+ c8 Naked Pair <> 29 r1289c8 r6 Naked Pair <> 29 r6c45 b9 Naked Triple <> 259 r9c7 c8b9 Naked Pair <> 38 r2c8 r8 b7 Locked Candidate 1 <> 2 r8c57 r7 b9 Locked Candidate 1 <> 2 r7c5 5-Fish c368b37\r2369b9 BF 032\401 <> 9 r8c7 5-Fish r157b37\c24579 AF 302\050 <> 9 r8c7 5-Fish r157b37\c2459b9 BF 302\041 <> 9 r8c7 +--------------------------------------------------------------+ | 1236 179 4 | 369 3679 8 | 29 67 5 | | 36 789 5 | 2 4 39 | 1 67 89 | | 26 789 29 | 5 679 1 | 3 4 289 | |--------------------+--------------------+--------------------| | 9 2 3 | 4 1 7 | 8 5 6 | | 4 6 8 | 39 239 5 | 7 1 29 | | 7 5 1 | 68 68 29 | 4 29 3 | |--------------------+--------------------+--------------------| | 8 3 6 | 7 5 4 | 29 29 1 | | 12 4 29 | 1389 389 6 | 5 38 7 | | 5 19 7 | 1389 2389 239 | 6 38 4 | +--------------------------------------------------------------+ Templates (A: 1) <> 9 r3c3,r8c45,r9c2

by **blue** » Sun Aug 25, 2013 6:36 am

Hi Danny,

daj95376 wrote:I'm not sure at what point you are solving NoFish4 and NoFish5.

For NoFish4, at the point that you show.
Here's one of the Obi-Wahn fish for the r9c8 elimination.

Hidden Text: Show

For NoFish5, I was using the wrong starting point -- the point just before the r8c7 elimination(s) you showed.
When I use your starting point, I don't find anything. I'll make a note in the earlier post.

For NoFish14, I must have goten mixed up somehow -- read the wrong line in the output, or something.
I do find the eliminations for all 4 candidates (58r3c78), just as you did.

BTW: I saw your post about the template eliminations for NoFish22-NoFish563.
I got the same results, and no Obi-Whan fish for the few eliminations that were missing.

Blue.

by **tarek** » Sun Aug 25, 2013 11:07 am

Thanks to Danny, blue, David & others for verifying the eliminations. This should allow targeting the correct eliminations in the future. Let the fishing begin

by **daj95376** » Sun Aug 25, 2013 6:21 pm

From my modest perspective, Fish logic uses Base/Cover sets and (often) fin cells to determine eliminations. NoFish contain invalid X-Cycles/Loops. However, dealing with an invalid X-Cycle seems beyond the reach of Fish logic. What's now needed is a way to detect these invalid X-Cycles and determine if an elimination exists.

Back when Tarek originally posted his NoFish, I tested a few of his initial offerings and noticed that the systematic placement of givens forced the invalid X-Cycle to exist right from the start.

Tarek #1 (NoFish #22): 12/25/09 (mm/dd/yy)

Code: Select all: +-----------------------+ | . . . | . . . | . . 9 | | . . 5 | . . 1 | . 3 . | | . 7 . | . 6 . | 4 . . | |-------+-------+-------| | 6 . . | . . 3 | . . 2 | | . . 4 | . 8 . | . 1 . | | . 3 . | . . 7 | 5 . . | |-------+-------+-------| | . . 9 | . . 5 | . 6 . | | . 1 . | . 3 . | 7 . . | | . . . | 2 . . | . . 8 | +-----------------------+ +-----------------------------------------------------------------------+ | 12348 2468 12368 | 34578 2457 248 | 1268 2578 9 | | 2489 24689 5 | 4789 2479 1 | 268 3 67 | | 12389 7 1238 | 3589 6 289 | 4 258 15 | |-----------------------+-----------------------+-----------------------| | 6 589 178 | 1459 1459 3 | 89 4789 2 | | 2579 259 4 | 569 8 269 | 369 1 367 | | 1289 3 128 | 1469 1249 7 | 5 489 46 | |-----------------------+-----------------------+-----------------------| | 23478 248 9 | 1478 147 5 | 123 6 134 | | 2458 1 268 | 4689 3 4689 | 7 2459 45 | | 3457 456 367 | 2 1479 469 | 139 459 8 | +-----------------------------------------------------------------------+ # 144 eliminations remain (*) invalid X-Cycle (g) guardian cells to prevent (*): r1c235678,r235678c1,r2c2,r3c3 r1c1=2 eliminates all guardian cells and forces presence of invalid X-Cycle +-----------------------------------+ | -2 g2 g2 | . g2 g2 | g2 g2 . | | g2 g2 . | . *2 . | *2 . . | | g2 . g2 | . . *2 | . *2 . | |-----------+-----------+-----------| | . . . | . . . | . . 2 | | g2 *2 . | . . *2 | . . . | | g2 . *2 | . *2 . | . . . | |-----------+-----------+-----------| | g2 *2 . | . . . | *2 . . | | g2 . *2 | . . . | . *2 . | | . . . | 2 . . | . . . | +-----------------------------------+

Yes, the X-Cycle can be split into two smaller oddagons, but I like the pattern afforded by taking the longer route.

by **David P Bird** » Mon Aug 26, 2013 8:36 am

DAJ, at the risk of teaching grandma to suck eggs, you don't need to specify the full set of guardian cells, just enough to ensure that each of the invalid pattern cells sees a member of the set you use. For NoFish #22 the options are:
r234578c1,r2c2,r3c3 (by rows)
r1c235678,r2c2,r3c3 (by columns)
r1c5678,r5678c1 (by boxes)

There is similarity between these sets for the invalid pattern and the sets of fin and PE cells for a fish pattern, which is why I consider the invalid cell set to form the body cells of an impossible fish.

Because the pattern must be false, each one of these sets must contain a truth, and if no cell is seen by all members of a set to provide an elimination, it can be split into two strongly linked subsets which may provide an elimination chain.

A quick way to check for an impossible fish is to check the box diagonals the pattern cells occupy. Here, in 5 out of the 6 containing boxes they are in the the \ direction, and in the other it is the / direction. As these counts are both odd, the pattern can't be simply coloured and so is impossible.

That's a lot quicker than surveying the guardians for individual oddagons.

PS Why can't you over-eat and fall asleep in front of a re-run on the telly like everyone else?

by **sultan vinegar** » Tue Aug 27, 2013 3:01 am

Hi everyone, this is my first post after rapidly getting up to speed on advanced techniques. For this elimination:

Code: Select all: . . . | . . 8 | . . . 8 8 . | . . . | . 8 . . . 8 | . . . | . . 8 ---------+----------+--------- . . 8 | . 8 . | . 8 . 8 . . | 8 . . | . 8 . . 8 8 | . 8 . | 8 . . ---------+----------+--------- F . 8 | . 8 . | 8 8 8 . G . | . 8 . | . 8 T 8 G . | 8 . . | 8 . .

I see two overlapping fish (both for digit 8), at least one of which is true, and both fish eliminate the target. Bear with me.

F is weakly linked to G, so at least one of F,G is false.

If F is false, then the 5 truth, 7 link fish c1247b3\r25689c9b9 is true. The target T is in 0 base sets and 3 link sets, so is eliminated automatically under rank 2 logic.

If G is false, then the 3 truth, 4 link fish (grouped x-chain) c27b3\r26c9b9 is true. The target T is in 0 base sets and 2 link sets, so is eliminated automatically under rank 1 logic.

Either way, the target T is eliminated.

by **tarek** » Wed Aug 28, 2013 9:46 pm

I've updated my NoFish22-NoFish563 List to accommodate the corrected eliminations. Thanks to Danny & blue.

by **daj95376** » Mon Sep 02, 2013 5:13 am

blue wrote:For NoFish4, at the point that you show.
Here's one of the Obi-Wahn fish for the r9c8 elimination.
r347c15b77\r18999c3489b4

It's like the 5x5 fish, r347c15\r1c489b4, with r9c1 as a normal fin, and r3c3 and r8c5 as remote fins.
[ (r3c3-r8c3=r9c123-) and (r8c5-r8c3=r9c123-) ]

I've been discussing the elimination r9c8<>9 with StrmCkr and an idea he's been investigating. IMO, his results include a single remote fin cell. Then, while staring at the candidate grid for <9>, it hit me that a simple pattern with two remote fin cells existed.

Code: Select all: Kraken Empty Rectangle r3b7\c38 r3c4 - r7c4 = r7c8 - r9c8 r3c9 - r2c7 = r2c23 - r1c1; X-Wing r59c17; r9c8<>9

Note: I did not research to see if anyone else has already posted the same results. I just thought that it'd be appropriate to this thread.

The New Sudoku Players' Forum

The NoFish List

The NoFish List

Re: a new (?) view of fish (naked or hidden)

Re: NoFish22-NoFish563 list

Re: a new (?) view of fish (naked or hidden)

Re: a new (?) view of fish (naked or hidden)

Re: a new (?) view of fish (naked or hidden)

Re: NoFish22-NoFish563 list

Re: a new (?) view of fish (naked or hidden)

Re: a new (?) view of fish (naked or hidden)

Re: The NoFish List

NoFish #22

Re: The NoFish List

Re: The NoFish List

Re: The NoFish List

Re: NoFish4