Oddities of Sudoku Explainer

Programs which generate, solve, and analyze Sudoku puzzles

Oddities of Sudoku Explainer

Postby ronk » Sun Apr 19, 2009 3:58 pm

I think many are aware that Sudoku Explainer (SE) does not group cells. This includes single-digit grouping of the Empty Rectangle (ER) pattern, and the multi-digit grouping of almost-locked-sets (ALS).

But I was surprised by its apparent mis-handling of a single, a naked single. In the following pencilmarks ...
Code: Select all
4...6...16..2.3.54..2......7..52...9.4.78.2.392..36.7586.3.219729.....3..35.97.42

47..653216..2.3.54352......7.352...954678921392..36.75864352197297...53.135.97.42

 *-----------------------------------------------------------*
 | 4     7     89    | 89    6     5     | 3     2     1     |
 | 6     18    189   | 2     17    3     | 789   5     4     |
 | 3     5     2     | 1489  147   148   | 6789  68    68    |
 |-------------------+-------------------+-------------------|
 | 7     18    3     | 5     2     14    | 468   68    9     |
 | 5     4     6     | 7     8     9     | 2     1     3     |
 | 9     2     18    | 14    3     6     | 48    7     5     |
 |-------------------+-------------------+-------------------|
 | 8     6     4     | 3     5     2     | 1     9     7     |
 | 2     9     7     | 1468  14    148   | 5     3     68    |
 | 1     3     5     | 68    9     7     | 68    4     2     |
 *-----------------------------------------------------------*

... SE v1.2.1 first finds locked candidates (SE's pointing) for r2c5<>1, and then mysteriously misses the naked single r2c5=7. Instead it single hints with the Direct Hidden Pair <14> in r38c5 to make the placement. Using [Get all hints] doesn't find the single either. What am I doing wrong?

FWIW the SE rating for a (naked) single is 1.0; for a direct hidden pair it's 2.0.
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

Postby 999_Springs » Sun Apr 19, 2009 4:09 pm

A direct hidden pair is rated 2.0 but a naked single is rated 2.3 which is, in my opinion, just ridiculous. This is why direct hidden pairs are ranked before naked singles.

The "get all hints" button only gets hints that don't have an outcome similar to an elimination given earlier, so r2c5=7 isn't found twice.

Another oddity: try loading up SE, manually remove all candidates 1 from r2378c2378, press "get next hint" three times. It should find something. But it doesn't. It just lags.

And speaking of lags, here's Eioru's "Sudoku Explainer lag puzzle":
Code: Select all
 . 4 5 . 6 7 . 8 9
 6 . 7 8 . 9 4 . 5
 8 9 . 4 5 . 6 7 .
 4 . . 5 7 8 . 9 6
 5 . 6 . 9 4 7 . 8
 7 8 9 2 . 6 5 4 .
 9 6 8 7 . . . 5 4
 . 5 4 6 8 . 9 . 7
 1 7 . 9 4 5 8 6 . 

Put it into SE, hit F9, and time it... (it took 322 seconds on my computer...)
Hidden Text: Show
Once upon a time I was a teenager who was active on here 2007-2011
999_Springs
 
Posts: 367
Joined: 27 January 2007
Location: In the toilet, flushing down springs, one by one.

Postby ronk » Sun Apr 19, 2009 4:41 pm

999_Springs wrote:A direct hidden pair is rated 2.0 but a naked single is rated 2.3 which is, in my opinion, just ridiculous. This is why direct hidden pairs are ranked before naked singles.

[Get next hint][Apply hint] stepping thru the original puzzle gives "Hint rating: 1.0" for the single r5c8=1. So you're suggesting that 1.0 is incorrect?

999_Springs wrote:Another oddity: try loading up SE, manually remove all candidates 1 from r2378c2378, press "get next hint" three times. It should find something. But it doesn't. It just lags.

Thanks, I didn't know that.
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

Postby 999_Springs » Sun Apr 19, 2009 5:48 pm

ronk wrote:
999_Springs wrote:A direct hidden pair is rated 2.0 but a naked single is rated 2.3 which is, in my opinion, just ridiculous. This is why direct hidden pairs are ranked before naked singles.

[Get next hint][Apply hint] stepping thru the original puzzle gives "Hint rating: 1.0" for the single r5c8=1. So you're suggesting that 1.0 is incorrect?

A naked single is rated 1.0 if it's the last remaining empty cell in a sector and 2.3 if it isn't.
Hidden Text: Show
Once upon a time I was a teenager who was active on here 2007-2011
999_Springs
 
Posts: 367
Joined: 27 January 2007
Location: In the toilet, flushing down springs, one by one.

Postby udosuk » Mon Apr 20, 2009 6:31 am

999_Springs wrote:A naked single is rated 1.0 if it's the last remaining empty cell in a sector and 2.3 if it isn't.

Wouldn't that makes it a hidden single too, thus the 1.0 rating?

I guess SE adopts that 2.3 rating because it assumes most players solves without pencilmarks, which is a really unrealistic assumption IMO.

When solving without pencilmarks naked singles are probably harder to spot than hidden pairs (and perhaps even hidden triples).:idea:
udosuk
 
Posts: 2698
Joined: 17 July 2005

Postby ronk » Mon Apr 20, 2009 1:42 pm

udosuk wrote:
999_Springs wrote:A naked single is rated 1.0 if it's the last remaining empty cell in a sector and 2.3 if it isn't.

Wouldn't that makes it a hidden single too, thus the 1.0 rating?

Ruud (whom I miss) called that a Full House. If anyone else used the term, I'm not aware of it.

udosuk wrote:When solving without pencilmarks naked singles are probably harder to spot than hidden pairs (and perhaps even hidden triples).

Rhetorically, do the terms naked single and hidden single even make sense when solving without pencilmarks?:)

::: Off-topic :::
To get some idea of the birth pains of sudoku terminology, take a look at the onetime alternative names for hidden single in the "External links" section here.
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

Postby udosuk » Mon Apr 20, 2009 3:52 pm

ronk wrote:Rhetorically, do the terms naked single and hidden single even make sense when solving without pencilmarks?:)

Actually, when I solve without pencilmarks, I do (consciously) search for both.

To search for hidden singles, I use "scanning" and "cross-hatching" as described by the sudopedia link on "hidden single" you quoted above.

As for naked singles, I just inspect the 20 "buddies" of a certain cell (8 of same row, 8 of same column, 4 others of same box) to see if they account for 8 of the 9 digits. If this is the case, then I know the cell must hold the remaining digit.

Note this approach doesn't necessarily find all naked singles as in solving with pencilmarks, because there might be naked subsets which together account for a set of several digits, which would be hard to see just by inspecting blank cells. I guess with a good memory I could memorise some naked subsets without pencilmarking, but that would be "mental pencilmarking" which sort of makes the whole "pencilmarkless solving" practice meaningless.:)
udosuk
 
Posts: 2698
Joined: 17 July 2005

Postby daj95376 » Mon Apr 20, 2009 6:05 pm

ronk wrote:
udosuk wrote:
999_Springs wrote:A naked single is rated 1.0 if it's the last remaining empty cell in a sector and 2.3 if it isn't.

Wouldn't that makes it a hidden single too, thus the 1.0 rating?

Ruud (whom I miss) called that a Full House. If anyone else used the term, I'm not aware of it.

HoDoKu uses Full House in its output. It was a shock to see it the first time!
daj95376
2014 Supporter
 
Posts: 2624
Joined: 15 May 2006

Postby ronk » Tue Apr 21, 2009 2:47 am

udosuk wrote:
ronk wrote:Rhetorically, do the terms naked single and hidden single even make sense when solving without pencilmarks?:)

Actually, when I solve without pencilmarks, I do (consciously) search for both.

I think the 'naked' is due to there being no candidates other than the unplaced single ... and the 'hidden' because there are other candidates. These 'other candidates' are a product of pencilmarking.
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

Postby udosuk » Tue Apr 21, 2009 6:12 am

ronk wrote:I think the 'naked' is due to there being no candidates other than the unplaced single ... and the 'hidden' because there are other candidates. These 'other candidates' are a product of pencilmarking.

I'm afraid I have to disagree here. Without consideration of different candidates there wouldn't be much "Sudoku solving", pencilmarking or not.

For me the definition of "pencilmarking" is using some methods to record the possible contents of unsolved cells.:idea:
udosuk
 
Posts: 2698
Joined: 17 July 2005

Postby ronk » Tue Apr 21, 2009 1:14 pm

udosuk wrote:
ronk wrote:I think the 'naked' is due to there being no candidates other than the unplaced single ... and the 'hidden' because there are other candidates. These 'other candidates' are a product of pencilmarking.

I'm afraid I have to disagree here. Without consideration of different candidates there wouldn't be much "Sudoku solving", pencilmarking or not.

For me the definition of "pencilmarking" is using some methods to record the possible contents of unsolved cells.:idea:

Within the context of this discussion, it should have been clear that I was writing about "recorded" candidates.

Moreover, here Ruud wrote:Hidden Single ... Without pencilmarks, the term “hidden” is meaningless.
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

Postby udosuk » Tue Apr 21, 2009 6:08 pm

I see, so we're back to the same old argument on word usage ("pedantry" as called by some).

The most important issue for me is there exist these 2 techniques which I'd employ when solving without pencilmarks:

Technique 1: scanning a row/column/box for a certain digit to see if there is only 1 possible cell for that digit.

Technique 2: inspecting a cell and its "buddies" to see if there is only 1 possible digit for that cell.

I can swear I can do both without any "pencilmarking" (i.e. "marked/recorded candidates"). Whether you believe it or not isn't my business.

The remaining (unimportant) issue is how we name these techniques. Since in both techniques the target is to spot a "single possibility", I have no objection to use "single" to name them.

As for the words "hidden" and "naked":

Technique 1: I'm scanning a room (row/column/box) to look for something (a single possible cell to hold a digit), i.e. that something is "hidden" from me. Therefore I can associate the word "hidden" for this very technique.

Technique 2: I'm inspecting a cell and its buddies for possible content (digits) for the cell. Consider each "false digit" as a piece of "cover" for a cell. When a buddy cell contains a certain digit, it's "removing a cover" from the cell. When there is only 1 possible digit left in the cell, it must be the "true digit". In that case all the "false digits", i.e. "covers", have been removed from the cell. And what do we call something/somebody with "all covers removed"? "Naked" sounds a pretty appropriate word.

(Note in here "possible digits" can be called "candidates", but by no means are they related to "pencilmarks" because they are never marked/recorded.:idea: )

See? I've just justified the names "hidden single" and "naked single" for these 2 techniques without associating to any context of "pencilmarks" or "recorded candidates" etc. So these names are perfectly valid in pencilmark or pencilmarkless solving. End of argument.:!:



PS: While I respect Ruud as a great (retired) puzzle master and fully admire his contribution on the Sudopedia and such, I never treat his words like holy bible verses.:idea:
Last edited by udosuk on Tue Apr 21, 2009 2:17 pm, edited 2 times in total.
udosuk
 
Posts: 2698
Joined: 17 July 2005

Postby StrmCkr » Tue Apr 21, 2009 6:11 pm

depends how you are candidate listing.

you can mark
what isnt given. {naked: last candidate}

and
what digits are given in the opposit rows/columns + line of sight.

{which is what digits can only go here?)

[hidden in opposit rows/columns. ]

i do agree that from some points of view hidden sets are actually easier to find then naked.

naked require manual tracking of digits per cell + cells.

can swear I can do both without any "pencilmarking" (i.e. "marked/recorded candidates"). Whether you believe it or not isn't my business.


i do the same thing
:)
Some do, some teach, the rest look it up.
User avatar
StrmCkr
 
Posts: 679
Joined: 05 September 2006

Postby ronk » Tue Apr 21, 2009 6:36 pm

udosuk wrote:The most important issue for me is there exist these 2 techniques which I'd employ when solving without pencilmarks:

Technique 1: scanning a row/column/box for a certain digit to see if there is only 1 possible cell for that digit.

Technique 2: inspecting a cell and its "buddies" to see if there is only 1 possible digit for that cell.

I can swear I can do both without any "pencilmarking" (i.e. "marked/recorded candidates"). Whether you believe it or not isn't my business.:

I do the same, so I believe you. But I can swear I at one time used both techniques without even knowing they were called hidden and naked singles, respectively. You should be able to guess when I learned the terms.

P.S. Sure seems like the bad economy has put lots of people in the U.S.A. in a foul mood. The economy in Australia must be bad too.:)
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

Postby udosuk » Wed Apr 22, 2009 10:18 am

ronk wrote:P.S. Sure seems like the bad economy has put lots of people in the U.S.A. in a foul mood. The economy in Australia must be bad too.:)

Sigh...

When you get involved in an argument (heated or not), there are many reasons. Being in a foul/bad mood is one. Another one is that you think the other person is not being correct and you are not the type that keeps silent about it.

For the record, I was in a pretty nice mood when I decided to get involve in this argument. And the economy here isn't any worse compared to the global status. The other day I was just chatting with Luke451 (another USA resident) about basketball betting with his extra pocket-money. Also, I'm not the type that would let the economy status affect my mood.:idea:
udosuk
 
Posts: 2698
Joined: 17 July 2005

Next

Return to Software