The New Sudoku Players' Forum

[Hud]

11-05-05 puzzle in "The Arizona Republic". They put their hardest rating on this one and I could only finally solve it using "swordfish". Thanks to Angus for his tutorial. Did anyone else work this one?

5 . . | . .4 | . . 2
3 . . | . . . | . 6 .
. . 8 | 5 . . |4 . .
-------+-------+------
8 . . | . 9. | 6 . 4
. . . | 1 8 6| . . .
1 . 3 | . 2 . | . . 5
-------+-------+------
. . 4 |. . .9| 2 . .
. 1 . | . . . | . . 6
6 . . | 2 . .| . . 7

[emm]

Hud, I couldn’t find the swordfish, Pappacom labelled the puzzle VHard but valid, Susser solved it with chains and Simple Sudoku solved it by colouring. If this is the grid at the hard point, colouring seems more straightforward.

Code: Select all

+-------------------+-------------------+-------------------+
| 5     79    6     | 89    13    4     | 138   378    2    |
| 3     4     1     | 89    7     2     | 5     6      89   |
| 2     79    8     | 5     6     13    | 4     37     19   |
+-------------------+-------------------+-------------------+
| 8     2     7     | 3     9     5     | 6     1       4   |
| 4     5     9     | 1     8     6     | 7     2       3   |
| 1     6     3     | 4     2     7     | 89    89      5   |
+-------------------+-------------------+-------------------+
| 7     38    4     | 6     135   9     | 2     358    18   |
| 9     1     2     | 7     45    38    | 38    45      6   |
| 6     38    5     | 2     134   138   | 1389  3489    7   |
+-------------------+-------------------+-------------------+

For colouring you need to find conjugate pairs – just 2 of that number in any row, column or box. Then you label the numbers alternately – say red and blue. Because they’re pairs, either the reds or the blues will be correct - you don’t know which yet. The rule is that if you find two of the same colour in the same group then they are the incorrect ones.

Code: Select all

red          blue       red        blue         red
r1c5   ->    r3c6  ->   r9c6
r1c5   ->    r1c7  ->   r3c9  ->   r7c9   ->    r9c7
                                   r7c9   ->    r7c5

Code: Select all

 .   .   .   |  .  1r  .  |  1b   .   . 
  .   .   .  |  .  .   .  |  .   .   . 
  .   .   .  |  .  .  1b  |  .   .   1r 
-------------+------------+------------
  .   .    . |  .   .  .  |  .   .   . 
  .   .   .  |  .   .  .  |  .   .   . 
  .   .   .  |  .   .  .  |  .   .   . 
-------------+------------+------------
  .   .   .  | .   1r     |  .   .   1b 
  .  .   .   |  .  .   .  |  .   .   . 
  .   .   .  |  .  1   1r |  1r   .   .

Since two reds end up in same group - both box 8 and row 9 - they must be false and you can eliminate all the red 1s.

[Pat]

11-05-05 puzzle in "The Arizona Republic":

Code: Select all

 5 . . | . . 4 | . . 2
 3 . . | . . . | . 6 .
 . . 8 | 5 . . | 4 . .
-------+-------+------
 8 . . | . 9 . | 6 . 4
 . . . | 1 8 6 | . . .
 1 . 3 | . 2 . | . . 5
-------+-------+------
 . . 4 | . . 9 | 2 . .
 . 1 . | . . . | . . 6
 6 . . | 2 . . | . . 7

They put their hardest rating on this one,
and I could only solve it using Swordfish.

Pappacom labelled the puzzle VHard but valid.

Susser solved it with chains,
and Simple Sudoku solved it by colouring.

quite amazing, if it does require something beyond X-wing to solve - while Pappocom rates it Very Hard!

or did i miss an important announcement??
- Pat

[emm]

I wondered about this too, Pat – I expected Pappocom to call it invalid. Is it new that it will accept a puzzle it wouldn’t create itself and calls arguably unfair – or have there always been some it would accept?

[Hud]

I'm still a bit fuzzy when it comes to "swordfish" I tried what I thought was a swordfish and it solved it but I may have just lucked out. I read Angus's blurb on it but I still have some questions about it. Please forgive any terminology mistakes but:

after finding the 3 columns or rows having 3 or less candidates in each (I also assume they have to be in separate groups of 3 blocks) does there then have to be 3 intersecting perpendicular columns or rows in separate groups of 3 blocks in order to rule out candidates?

Angus's example shows it that way but I don't know if it's mandatory or not.

Also, can you do multiple swordfish on a puzzle?


Can you do more than one swordfish with the same candidate?

Sorry about all the questions but this is driving me nutty.

By the way, how do you guys get the sample grid to post so nicely? Mine always end up crappy looking.
Tom

[Hud]

I just did a search and found Nick67's explanation of swordfish. I believe that answers some of my questions but not all of them.

[emm]

I’m not sure what you mean by separate groups of blocks – the three arms of the swordfish will always each be in a different box but don’t have to be spread across all the boxes. Here the swordfish is going down the columns and is all in boxes 1,4,6.

Code: Select all

 1 1 1 | . . . | 1 1 1
 . . . | . . . | . . .
 . . . | . . . | . . .
-------+-------+------
 1 1 . | . . . | 1 1 1
 . . . | . . . | . . .
 . . . | . . . | . . .
-------+-------+------
 1 . 1 | . . . | 1 1 1
 . . . | . . . | . . .
 . . . | . . . | . . .

You can exclude other candidate 1s across rows 1,4,7

Pappocom puzzles don’t have swordfish so you don’t see many here. Other programs create puzzles with swordfish so you could try those out for practice. Theoretically I don’t see why you couldn’t have a puzzle with more than one swordfish and more than one for the same number. Look at this site - Sadman - for some examples.

This is how I get the perfect grid - I copied someone else’s and saved it in Word. I insert my numbers as needed, paste it back into my reply, highlight it and click code. It works like magic!

[CathyW]

This is how I get the perfect grid - I copied someone else’s and saved it in Word. I insert my numbers as needed, paste it back into my reply, highlight it and click code. It works like magic!

Amazing That's exactly what I did too.

[Hud]

em, I didn't realize that swordfish weren't required on Pappocom puzzles. I'm working on one of its puzzles that I was sure needed it. I'll check it out again to see if I can solve it using the usual methods.

In your example, I'm confused about how the boxes are numbered. Is the upper left box no. 1 and the lower right one box 9? It does simplify matters that the candidate columns don't have to each occupy their own individual third (chute?) of the 9 cells though.

When I tried to post a grid, I cut and pasted a previous nice looking grid onto my post but it got skewed in translation. I'll check out using "word" if I can

[emm]

In your example, I'm confused about how the boxes are numbered. Is the upper left box no. 1 and the lower right one box 9?

Yes. Apologies for my typo - the swordfish is in box 1,4,7

It does simplify matters that the candidate columns don't have to each occupy their own individual third (chute?) of the 9 cells though.

Vertical chute = stack. Here’s Pappocom’s definition of Basic Terms.

When I tried to post a grid … it got skewed in translation.

Yeah, sometimes it does that. Make sure you highlight the grid and click code. Also click the preview button before you submit - then you can adjust if needed.

[Hud]

em, thanx for clarifying. I redid the Pappocom puzzle I mentioned and as advertised, it didn't require swordfish. I am getting a hankering to go tuna fishing but that's on a different forum. The puzzle did require even a quint naked group. First time I've encountered that.

[Pat]

11-05-05 puzzle in "The Arizona Republic":

Code: Select all

 5 . . | . . 4 | . . 2
 3 . . | . . . | . 6 .
 . . 8 | 5 . . | 4 . .
-------+-------+------
 8 . . | . 9 . | 6 . 4
 . . . | 1 8 6 | . . .
 1 . 3 | . 2 . | . . 5
-------+-------+------
 . . 4 | . . 9 | 2 . .
 . 1 . | . . . | . . 6
 6 . . | 2 . . | . . 7

They put their hardest rating on this one,
and I could only solve it using Swordfish.

Pappacom labelled the puzzle VHard but valid.

Susser solved it with chains,
and Simple Sudoku solved it by colouring.

quite amazing, if it does require something beyond X-wing to solve - while Pappocom rates it Very Hard!

or did i miss an important announcement??

I wondered about this too – I expected Pappocom to call it invalid.

Is it new that it will accept a puzzle it wouldn’t create itself and calls arguably unfair
– or have there always been some it would accept?

i'd consider this very odd!

is Pappocom too busy to comment??

- Pat

[emm]

Must have had lots of replies to the For Sale sign.

[emm]

Pat - I came across this in an earlier thread

In the Help files for my Sudoku program, I talk of a "blurred area" between logic and T&E. I don't say so in the Help files, but I am referring to 3 or 4 techniques, one of which is the "Nishio"…

If a puzzle which requires a Nishio for its solution is Dubbed into my program, my program can solve it. To that extent, I recognize the puzzle as valid.

However, the user is told that the puzzle is valid but arguably unfair. However, my program will not itself create a puzzle which requires a Nishio for solution. That is because I consider that the Nishio technique is nearer to T&E than to logic.

[Pat]

Pat - I came across this in an earlier thread

In the Help files for my SuDoku program, I talk of a "blurred area" between logic and T&E. I don't say so in the Help files, but I am referring to 3 or 4 techniques, one of which is the "Nishio"---

If a puzzle which requires a Nishio for its solution is Dubbed into my program, my program can solve it. To that extent, I recognize the puzzle as valid. However, the user is told that the puzzle is valid but arguably unfair.

However, my program will not itself create a puzzle which requires a Nishio for solution. That is because I consider that the Nishio technique is nearer to T&E than to logic.

hey, em, sorry for the delay, i only now saw this---

and i'm not sure what you're trying to say here:
regarding this specific puzzle we were discussing,
does the Pappocom software rate it Very Hard -
or valid but arguably unfair??

- Pat