The New Sudoku Players' Forum

[David P Bird]

For each puzzle, two additional exclusions r24c7<>3 are valid.
These are relatively simple 0-rank multi-fish, at least for a software program.

Thanks. Thinking that there wouldn't be any rank 0 patterns, I stopped looking when I found the Almost Multi-fish. Checking today, by changing c7 to c4 in the cover set I get your simple MF.

For both puzzles this adds your eliminations in c7, but loses mine in c4.
For the second puzzle (41106) follow-on locked sets get these c4 eliminations, but in the first (34189) no locked sets are formed and they don't immediately become apparent.

I don't know if Xsudo finds them anyway, but in my case I need to explore this accidental discovery further.

[pjb]

For virtually every puzzle containing an SK-loop that I've encountered, I've also found a simple row-column rank zero multifish, except for the following one:

Code: Select all

98.7.....76....8....5......6...5..4...89..7.......3..2..96..1......2...3.....4.5.;539;GP;H89;G13


Am I missing something, (most likely), or is there something odd about it?

pjb

[champagne]

For virtually every puzzle containing an SK-loop that I've encountered, I've also found a simple row-column rank zero multifish, except for the following one:

Code: Select all

98.7.....76....8....5......6...5..4...89..7.......3..2..96..1......2...3.....4.5.;539;GP;H89;G13


Am I missing something, (most likely), or is there something odd about it?

pjb

in my statistic file, that one has a row and a mixed (2 rows 2 columns) multi_fish.

I have only one puzzle tagged as having an SK loop and no multi fish
it is number 7765

EDIT : for puzzle 7765, the box cover on a row based multi_fish is not studied by my solver to-day.
I'll flag manually the puzzle in the data base.

For puzzle 539, my statistics files seems out of date. I have three rank_0 logic, not 2

Hidden Text: Show

16 Truths = {2R1257 3R1257 4R1257 5R1257 }
16 Links = {2c68 3c58 4c59 5c69 1n37 2n34 5n12 7n12 }


19 Truths = {2C12347 3C12347 4C12347 5C1247 }
19 Links = {2r349 3r349 4r368 5r68 1n37 2n34 5n12 7n12 }


16 Truths = {2R57 2C47 3R57 3C47 4R57 4C47 5R57 5C47 }
16 Links = {2b59 3b68 4b59 5b68 1n7 2n4 3n47 5n12 7n12 }

[ronk]

I have only one puzzle tagged as having an SK loop and no multi fish; it is number 7765

There is another 0-rank pattern that requires box covers (links).

Code: Select all

987......65.9..4....4.......4.3..8......6..2......1..5..97..3......2..6......5..1 # 7765;GP;H1604

Spoiler alert: Hides a 0-rank multi-fish: Show

___

Code: Select all

     18 Truths = {12R12347 56R1347}
     18 Links = {15c58 26c69 47n1 7n2 4n3 13n47 12b1}
     7 Eliminations --> r3c7<>79, r7c1<>48, r1c4<>4, r3c4<>8, r4c1<>7

[pjb]

Having thought I had developed techniques with the help of you all to produce a rank zero multifish for virtually all puzzles with SK-loops, I came across these:

Code: Select all

..9.....6.1...8.3.2.....7.........1.....73..5.4.8.5.....6.....9.5.3...4.7.....2..;9698;TkP;3952
..1.....8.3...7.4.6.....9.........72....15....4.2.3.....9...6...2.3...5.8.......1;9681;TkP;3642
..3.....1.2...4.7.6.....8.......7.9....26.....5.3....7..1.....6.9.5...4.8.....3..;9650;TkP;5008


They all share the same pattern. I expect I'm missing something, but can't balance truths and links.
Thank you all again
pjb

[champagne]

here after the results of my solver

regards

champagne

Hidden Text: Show

Code: Select all

..9.....6.1...8.3.2.....7.........1.....73..5.4.8.5.....6.....9.5.3...4.7.....2..;9698;TkP;3952

X     7+    9g    |27+   2+    27+   |X     2+    6g   
6+    X     7+    |2679+ 269+  X     |9+    X     2+   
2g    6+    X     |69+   69+   69+   |7g    9+    X     

69+   2679+ 27+   |269+  269+  269+  |69+   X     27+   
69+   269+  2+    |269+  7g    X     |69+   269+  X     
69+   X     27+   |X     269+  X     |69+   2679  27+   

X     2+    6g    |27+   2+    27+   |X     7+    9g   
9+    X     2+    |X     269+  2679+ |6+    X     7+   
7g    9+    X     |69+   69+   69+   |2g    6+    X     


SLG rank 0
16 Truths = {2R28 2C28 6R28 6C28 7R28 7C28 9R28 9C28 }
16 Links = {2b37 6b19 7b19 9b37 2n45 4n2 5n28 6n8 8n56 }
elims 45r2c4 45r2c5 38r4c2 8r5c2 8r5c8 18r8c5 1r8c6
11 elims done
..1.....8.3...7.4.6.....9.........72....15....4.2.3.....9...6...2.3...5.8.......1;9681;TkP;3642


floors PM 1689
9+    9+    1g    |69+   69+   69+   |X     6+    8g   
9+    X     8+    |1689+ 689+  X     |1+    X     6+   
6g    8+    8+    |18+   8+    18+   |9g    1+    X     

19+   1689+ 68+   |689+  689+  689+  |18+   X     X     
9+    689+  68+   |689+  1g    X     |8+    689+  69+   
19+   X     68+   |X     689+  X     |18+   1689  69+   

1+    1+    9g    |18+   8+    18+   |6g    8+    X     
1+    X     6+    |X     689+  1689+ |8+    X     9+   
8g    6+    6+    |69+   69+   69+   |X     9+    1g   


SLG rank 0
16 Truths = {1R28 1C28 6R28 6C28 8R28 8C28 9R28 9C28 }
16 Links = {1b37 6b37 8b19 9b19 2n45 4n2 5n28 6n8 8n56 }


..3.....1.2...4.7.6.....8.......7.9....26.....5.3....7..1.....6.9.5...4.8.....3..;9650;TkP;5008


X     8+    3g    |68+   8+    68+   |6+    6+    1g   
1+    X     8+    |168+  138+  X     |6+    X     3+   
6g    1+    X     |1+    13+   13+   |8g    3+    3+   

13+   1368+ 68+   |18+   18+   X     |16+   X     38+   
13+   138+  8+    |X     6g    18+   |1+    138+  38+   
1+    X     68+   |3g    18+   18+   |16+   168+  X     

3+    3+    1g    |8+    38+   38+   |X     8+    6g   
3+    X     6+    |X     138+  1368+ |1+    X     8+   
8g    6+    6+    |16+   1+    16+   |3g    1+    X     


SLG rank 0
16 Truths = {1R28 1C28 3R28 3C28 6R28 6C28 8R28 8C28 }
16 Links = {1b19 3b37 6b37 8b19 2n45 4n2 5n28 6n8 8n56 }


[pjb]

Thank you champagne

In all the other SK-loop puzzles, I've managed to find a multifish with all the truths in the rows alone and all the links in columns, nodes and occasionally boxes. This makes the coding quite easy. However, your solutions require truths to be in both rows and columns, so puzzles with the pattern above must be different in some way. My challenge now is to work out how to code for solutions such as you provided here

Best wishes
Phil

[ronk]

Having thought I had developed techniques with the help of you all to produce a rank zero multifish for virtually all puzzles with SK-loops, I came across these:

Code: Select all

..9.....6.1...8.3.2.....7.........1.....73..5.4.8.5.....6.....9.5.3...4.7.....2..;9698;TkP;3952
..1.....8.3...7.4.6.....9.........72....15....4.2.3.....9...6...2.3...5.8.......1;9681;TkP;3642
..3.....1.2...4.7.6.....8.......7.9....26.....5.3....7..1.....6.9.5...4.8.....3..;9650;TkP;5008


Similar to 7765;GP;H1604 above, box links are required in the "central box", which is b5 for these three puzzles. To give others the chance to find these multi-fish on their own, I'll wait a few days before posting the details.

champagne's, I didn't think your program handled box covers (links) yet, but row-multifish are indicated in the "04c multi_fish rank 0.txt" file.

[champagne]

I didn't think champagne's program handled box covers (links) yet, but multi-fish in rows are indicated in his "04c multi_fish rank 0.txt" file.

The pattern with 2 rows 2 columns as truths belongs to the first group of patterns coded in my solver. That pattern has always a box cover where rows and columns intersect (AFAIK).

But my solver still does not consider box cover with a pure rows or pure columns base.

champagne

[ronk]

I didn't think champagne's program handled box covers (links) yet, but multi-fish in rows are indicated in his "04c multi_fish rank 0.txt" file.

But my solver still does not consider box cover with a pure rows or pure columns base.

So why are these three puzzles listed in the two "multifish" files? Are you counting sk-loops as multi-fish too?

[champagne]

But my solver still does not consider box cover with a pure rows or pure columns base.

So why are these three puzzles listed in the two "multifish" files? Are you counting sk-loops as multi-fish too?

I don't understand your point.
These are not pure rows or pure columns base sets (truths), but a mix of 2 rows and 2 columns.

That pattern was already recognised by my old solver, untouched for now more than one year, and likely 2 years.

In that pattern, the links are done of box covers in the 4 boxes containing the crossings and from cells covers.
That pattern was named G3 in the old solver

so far, my new solver looks for the following patterns

a) row based (+ optional cells) and columns + cells cover
b) columns based (+ optional cells) and rows + cells cover
c) 2 row + 2 cells and box cover + cells cover
d) a pattern describing one special solution you gave in that thread, but I have only one example of that pattern, so I forgot it exists.

I confirm that I don't study boxes in a pure row base or a pure column base

[ronk]

But my solver still does not consider box cover with a pure rows or pure columns base.

So why are these three puzzles listed in the two "multifish" files?

I don't understand your point.

I don't know how to make my question any simpler.

[champagne]

why are these three puzzles listed in the two "multifish" files?

I don't understand your point.

I don't know how to make my question any simpler.

My first reaction would be to answer

They are in the files "multifish" because a "multifish" has been identified by the solver, but it can not be the right answer, so I wait for more information

[ronk]

They are in the files "multifish" because a "multifish" has been identified by the solver ...

Posting the multifish "identified by the solver" (for one of the three puzzles) would be helpful.

[champagne]

They are in the files "multifish" because a "multifish" has been identified by the solver ...

Posting the multifish "identified by the solver" (for one of the three puzzles) would be helpful.

All three are in the answer to pjb (who clearly has seen them) but in hidden mode,
sorry, but I try to keep posts as small as possible to make easier a quick search.
May be this is not the right way