The New Sudoku Players' Forum

[ArkieTech]

Code: Select all

 *-----------*
 |..1|9..|...|
 |.3.|24.|...|
 |86.|..5|...|
 |---+---+---|
 |6.8|...|92.|
 |91.|...|.78|
 |.52|...|4.6|
 |---+---+---|
 |...|1..|.87|
 |...|.73|.5.|
 |...|..8|3..|
 *-----------*


Play/Print this puzzle online

[Cenoman]

Code: Select all

 +----------------------+----------------------+--------------------+
 | a24-5   24    1      |  9      8      67    |  67    34    345   |
 | e57     3    e579    |  2      4      167   |  8     169   15    |
 |  8      6    e479    |  37     13     5     |  127   149   124   |
 +----------------------+----------------------+--------------------+
 |  6      47    8      |  3457   135    147   |  9     2     13    |
 |  9      1    d34     |  346    236    24    |  5     7     8     |
 |  37     5     2      |  8      139    179   |  4     13    6     |
 +----------------------+----------------------+--------------------+
 | b2345   249  c3456   |  1      2569   249   |  26    8     7     |
 |  12     8     46     |  46     7      3     |  12    5     9     |
 |  1257   279   567    |  56     2569   8     |  3     146   124   |
 +----------------------+----------------------+--------------------+

(4)r1c1 = (4-3)r7c1 = r7c3 - (3=4)r5c3 - (479=5)b1p469 => -5r1c1; ste

Note the subchain: (4)r1c1 = (4-3)r7c1 = r7c3 - (3=4)r5c3 => -4 r3c3; lclste
Or: (573=4)r26c1,r5c3 - (479=5)b1p469 => -5r1c1; ste

[SteveG48]

Code: Select all

 *-----------------------------------------------------------*
 | 245   24    1     | 9     8    e67    | 67   d34   d345   |
 | 57    3     579   | 2     4    e167   | 8     169  d15    |
 | 8     6    b479   | 37    13    5     | 127  c149  c124   |
 *-------------------+-------------------+-------------------|
 | 6     47    8     | 3457  135   147   | 9     2     13    |
 | 9     1    a34    | 346   236   24    | 5     7     8     |
 | 7-3   5     2     | 8    f139  f179   | 4    f13    6     |
 *-------------------+-------------------+-------------------|
 | 2345  249   3456  | 1     2569  249   | 26    8     7     |
 | 12    8     46    | 46    7     3     | 12    5     9     |
 | 1257  279   567   | 56    2569  8     | 3     146   124   |
 *-----------------------------------------------------------*


(3=4)r5c3 - r3c3 = r3c89 - (4=135)b3p236 - (1=67)r12c6 - (7=139)r6c568 => -3 r6c1 ; stte

[SpAce]

Code: Select all

.-------------------.---------------------.--------------------.
|  245   24    1    |  9     8      6-7   | d6(7)   34     345 |
| a5[7]  3    b59#7 |  2     4     a16[7] |  8     d1(6)9  15  |
|  8     6     479  |  37    13     5     |  127    149    124 |
:-------------------+---------------------+--------------------:
|  6     47    8    |  3457  135    147   |  9      2      13  |
|  9     1     34   |  346   236    24    |  5      7      8   |
| a3[7]  5     2    |  8     139   a19[7] |  4      13     6   |
:-------------------+---------------------+--------------------:
|  2345  249   3456 |  1     2569   249   |  26     8      7   |
|  12    8     46   |  46    7      3     |  12     5      9   |
|  1257  279  c567  | c56    2569   8     |  3     c146    124 |
'-------------------'---------------------'--------------------'

Kraken X-Wing:

(7)r26\c16 = r2c3 - (756)r9c348 = (67)b3p51 => -7 r1c6; stte

5x5 TM: Show

Code: Select all

 7r6c6 7r6c1
 7r2c6 7r2c1 7r2c3
             7r9c3 56r9c34
                    6r9c8  6r2c8
 7r1c7                     6r1c7
--------------------------------
-7r1c6

Edit. Lol! That's not even close to stte (or btte for that matter)!! Why didn't anyone say anything? Well, I know what happened, and it's Hodoku's fault! When I find an elimination that seems potential (i.e. produces immediate placements) I click "Solve up to" ("progress" set to singles) and see if it solves the puzzle. However, for some stupid reason it doesn't work similarly in "Learning" mode which I'd forgot to turn off after testing something. Instead it solves the puzzle all the way regardless of the progress setting. That has happened at least once before, but then I noticed it almost immediately. Frustrating. The one time I actually spotted an almost-X-Wing!!

[Ngisa]

Code: Select all

+----------------------+----------------------+--------------------+
|  245     24     1    | 9       8       d67  |d67      34     345 |
| a5-7*    3      579  | 2       4        167 | 8      e169    15  |
|  8       6     g479  | 37      13       5   | 127    f149*   124 |
+----------------------+----------------------+--------------------+
|  6       47     8    | 3457    135      147 | 9       2      13  |
|  9       1     h34   | 346     236      24  | 5       7      8   |
|ib37      5      2    | 8       139     c179 | 4       13     6   |
+----------------------+----------------------+--------------------+
|  2345    249    3456 | 1       2569     249 | 26      8      7   |
|  12      8      46   | 46      7        3   | 12      5      9   |
|  1257    279    567  | 56      2569     8   | 3       146    124 |
+----------------------+----------------------+--------------------+


(7*)r2c1 - r6c1 = r6c6 - (76)r1c67 = (6-9)r2c8 = (9*)r3c8 - (7*9*=4)r3c3 - (4=3)r5c3 - (3=7)r6c1 => - 7r2c1; stte

Clement

[SpAce]

Hi Cenoman,

You said here that you're "pleased to make some ads for full tagging," and on my part I was pleased to get your initial input on the matter. However, the best way to advertise it would be to answer my follow-up question about its real-world applicability for manual solving. How do you actually do it for anything but trivial cases? I've asked that before but didn't get any answer then either. Based on my recent couple of experiments with the little I understand of it, I can't say I'm yet convinced that it's a viable tool for everyday solving. I find the concept interesting and elegant, though, so please feel free to convince me!

Anyway, here's my attempt for today's puzzle, tagged just enough to prove everyone's solutions so far (I think):

partial tagging: Show

Code: Select all

.-------------------------------------.---------------------------.--------------------------------.
| [4D]f.5E.2   [4]f.2    1            |  9      8      [6a.7A]G   | 6A.7a  [4]F.(3)f  [4]F.(5e.3)f |
| (7c.[5C]e)f   3      ([7]C.[5]e.9)f |  2      4      [6A.7]G.1g | 8       6a.19      1e.5E       |
|  8            6      ([7]C.9)f.4F   |  3G.7g  1G.3g   5         | 7A.12  [4]f.19    [4]f.12      |
:-------------------------------------+---------------------------+--------------------------------:
|  6            4b.7B    8            |  3457   135     147       | 9       2          13          |
|  9            1        3b.4B        |  346    236     24        | 5       7          8           |
|  3B.7b        5        2            |  8     (139)b   7B.(19)b  | 4      (13)b       6           |
:-------------------------------------+---------------------------+--------------------------------:
|  3b.4d.25     249      3B.456       |  1      2569    249       | 2A.6a   8          7           |
|  12           8        46           |  46     7       3         | 12      5          9           |
|  1257         7b.29    7c.(56)C     | (56)C   2569    8         | 3       6A.14      124         |
'-------------------------------------'---------------------------'--------------------------------'



a=A -7c6- B=b -7c1- c=C -56r9- A=a => +a (proves my solution)

E -1n1- D=d -7n1- b=B -4c3- F=f -579b1- E => -E (proves Cenoman's first stte solution)

F -4b1- D=d -7n1- b=B -4c3- F => -F (proves Cenoman's btte solution)

E -5b1- C=c -7c1- b=B -4c3- F=f -579b1- E => -E (proves Cenoman's second stte)

b=B -4c3- F=f -35b3- E=e -1r2- g=G -7c7- B=b => +b (proves Steve's solution, also indirectly Clement's since b-c)

That was the first time I tagged anything but basic bilocal/bivalue links, and despite being still pretty simple and only partially tagged, it's already quite unreadable (not to mention tedious and error-prone to produce). When groups and ALS nodes are tagged each candidate can have multiple tags, which complicates things a lot. In this case I solved that problem with brackets ([] for groups, () for ALSs), but I don't see it as a viable option for more complicated cases which would require deeper nesting of tags. How do you deal with it?

So, I reiterate my question: can you share your technique of how you actually use full tagging manually? Or am I close to it already? As far as I see, there's no way to do it with actual coloring tools (of Hodoku, for example), except for very simple cases that use only basic links and have no more than a few tag pairs. That's a big drawback for a visual solver, even if we discount the hard labor of working with text-based pm grids. GEM doesn't suffer from that problem because it has a static cap on the number of colors it needs, so it's easy and quick to use with Hodoku as well as on paper (in which case I use symbols instead of colors). It has its own drawbacks, and less elegance perhaps, but for most practical purposes it still seems much more user-friendly and efficient. I'm open to be convinced otherwise, though. Please do!

Added. To underscore my point, here's a bit more complete tagging (as I understand it):

a bit fuller tagging: Show

Code: Select all

.-----------------------------------------------------.---------------------------------------------------------------.-------------------------------------------.
|  2d.[4h]S.[5k]U       2D.[4d]S       1              |  9                     8                     6m.[7M]A         |  6M.7m        3C.[4c]Js      3c.[4]js.5K  |
| [5r]KU.[7R]W          3             [5u]K.[7]rW.9o  |  2                     4                     1a.6M.[7w]A      |  8           [1]A.6m.9O     [1K]A.5k      |
|  8                    6              4s.[7w]r.9O    |  3A.7a                 1A.3a                 5                | [1b]a.2e.7M  [1]a.[4]JS.9o  [1]a.2E.[4]jS |
:-----------------------------------------------------+---------------------------------------------------------------+-------------------------------------------:
|  6                    4g.7G          8              | [3]ac.[4]Gi.5l.[7A]g  [1]aC.[3]Ac.5L        [1]AC.[4]GI.[7]ag |  9            2              1c.3C        |
|  9                    1              3g.4G          | [3]aG.[4]gi.6n         2f.[3]G.6N            2F.[4f]gI        |  5            7              8            |
|  3G.7g                5              2              |  8                    [1]ac.[3]A.9p         [1]Ac.[7G]a.9P    |  4            1C.3c          6            |
:-----------------------------------------------------+---------------------------------------------------------------+-------------------------------------------:
| [2]D.3g.[4H]It.[5]u  [2]d.[4]It.9q   3G.[4]I.[5]U.6 |  1                    [2]F.[5]l.[6]n.[9]PQ   2f.4i.[9p]Q      | [2M]E.6m      8              7            |
|  1b.[2B]D             8             [4i]T.6I        |  4I.[6i]N              7                     3                |  1B.[2b]E     5              9            |
|  1B.[2]D.[5]uv.7     [2]d.7g.9Q     [5]Uv.6.7r      | [5L]V.[6l]N           [2]F.[5]lV.[6]n.[9q]P  8                |  3           [1]b.4j.6M     [1]b.2e.4J    |
'-----------------------------------------------------'---------------------------------------------------------------'-------------------------------------------'


As far as I know, all basic and group links are now fully tagged (and dots added to improve readability a bit). It's missing all ALS tags and other advanced links, so it's still very far from "full" tagging -- but I don't think any more information is possible to fit into one grid (nor am I willing to add it). I was willing to do this exercise exactly this once, but probably never again. I think it's pretty clear that it's not a human solving technique if taken to this level and beyond, no matter how powerful and elegant it might be. At the very least the tags must be generated by a computer, but even then I think the end result is better to be left analyzed by a computer too, as there's so much information overload in a very unreadable and unintuitive format. I have no idea how the missing ALS tags etc would be added anyway, except external to the grid.

So... I can easily see how it could be used to implement a powerful and generic software solver, but it's almost impossible to imagine how it could work as an enjoyable human technique. Or is there something I'm still missing?

(Edit: The conversation has continued in PM.)

[eleven]

Code: Select all

 *------------------------------------------------------------------*
 |  24-5   24    1      |  9      8      67    |  67    34    345   |
 | *57     3    *579    |  2      4      167   |  8     169   1-5   |
 |  8      6    *479    |  37     13     5     |  127   149   124   |
 |----------------------+----------------------+--------------------|
 |  6      47    8      |  3457   135    147   |  9     2     13    |
 |  9      1    *34     |  346    236    24    |  5     7     8     |
 | *37     5     2      |  8      139    179   |  4     13    6     |
 |----------------------+----------------------+--------------------|
 |  2345   249   3456   |  1      2569   249   |  26    8     7     |
 |  12     8     46     |  46     7      3     |  12    5     9     |
 |  1257   279   567    |  56     2569   8     |  3     146   124   |
 *------------------------------------------------------------------*

5 cells, 5 digits 34579, only the 7 can be twice:
7r6c1 -> 5r2c1 => 5r2c13, -5r1c1,r2c9; stte

[SteveG48]

Edit. Lol! That's not even close to stte (or btte for that matter)!! Why didn't anyone say anything?

I don't know about everyone else, but I'll generally look at the chains to see if I understand, but I won't check to see if the elimination does what is claimed unless something suggests that it might not.

Well, I know what happened, and it's Hodoku's fault! When I find an elimination that seems potential (i.e. produces immediate placements) I click "Solve up to" ("progress" set to singles) and see if it solves the puzzle. However, for some stupid reason it doesn't work similarly in "Learning" mode which I'd forgot to turn off after testing something. Instead it solves the puzzle all the way regardless of the progress setting. That has happened at least once before, but then I noticed it almost immediately. Frustrating. The one time I actually spotted an almost-X-Wing!!

Interesting. I use Hodoku to do the basics. Then, if I think an elimination is potential, I just eliminate it and see what the progress indicator says. Whether it does the job or not, then I hit the back arrow and go about my business.

[skysea575]

010020300004005060070000008006900070000100002030048000500006040000800106008000000
.1..2.3..
..4..5.6.
.7......8
..69...7.
...1....2
.3..48...
5....6.4.
...8..1.6
..8......
1stDifficulty is 1.2



010070300004005060070000008006900070000100002030048000500006040000800106002000000
.1..7.3..
..4..5.6.
.7......8
..69...7.
...1....2
.3..48...
5....6.4.
...8..1.6
..2......
2nd question difficulty 7.2

The difference of the two Numbers bigger difference of difficulty.

[SpAce]

I don't know about everyone else, but I'll generally look at the chains to see if I understand, but I won't check to see if the elimination does what is claimed unless something suggests that it might not.

The same here, so it's not really surprising no one noticed. The only times I remember checking that in someone's solution have been when an stte-marked elimination has seemed too easy for the puzzle (and has in fact been btte).

Interesting. I use Hodoku to do the basics.

Me too, usually. Sometimes -- like this time -- I like to do basics too just to maintain routine (I rarely do other solving these days than these dailies). That's why I didn't catch the error immediately because I didn't click the progress button at the beginning as usual.

Then, if I think an elimination is potential, I just eliminate it and see what the progress indicator says.

Lol again! Now I really feel stupid! Thanks for that! I don't think I've ever really looked at the progress indicator except to see if a manually entered puzzle is valid. Yep, here it says it clearly:

In HoDoKu's default configuration the only steps thought of as easy are singles. That means, that once the indicator's bullet becomes white (default color for "Easy"), the sudoku can be solved with Singles only.

Why did I never think of that??? That's definitely a smarter way to do it! Now I can actually set the progress button to clear all basics instead of just singles. Does the green bullet mean btte?

[SteveG48]

Does the green bullet mean btte?

Yes, if you're using the default colors.