The New Sudoku Players' Forum

[swaatacba]

I have made a little progress with this puzzle, but I am stuck for a next step. Any suggestions?

[200e200w]

Hi, swaatacba!
First, you missed some basic moves. This is the pencilmark grid after all possible basic moves:

Code: Select all

.--------------.--------------.-------------------.
| 25   6   49  | 3     29   8 | 7     1     459   |
| 1    3   489 | 7     5    6 | 48    2     489   |
| 258  59  7   | 129   129  4 | 358   3569  35689 |
:--------------+--------------+-------------------:
| 58   1   89  | 29    249  3 | 6     7     2459  |
| 4    59  36  | 1269  8    7 | 1235  359   12359 |
| 7    2   36  | 169   149  5 | 134   8     1349  |
:--------------+--------------+-------------------:
| 9    7   15  | 4     6    2 | 1358  35    1358  |
| 36   4   25  | 8     37   1 | 9     56    2567  |
| 36   8   12  | 5     37   9 | 12    4     67    |
'--------------'--------------'-------------------'


To solve this puzzle, use this X-Wing:

Code: Select all

X-Wing: 9 r35\c28 => -9 r3c459,r5c49; stte


There is an excellent guide for fishing here - you can learn about not only X-Wings, but also all other sizes and types of fish, and the notation used. I recommend you to read the first post of it.

200e200w

[swaatacba]

Hi

Thanks for the revised version. I certainly seem to have missed some basic moves, but I cannot see how you achieved then.

Essentially you have eliminated the potential Id 3 from each of the cells 88. 89, 97 and 99. In addition, you have eliminated the Id's 1 & 2 from cell 99. Can you explain the logic for these eliminations, please?

Stephen

[Leren]

During the basic elimination phase you come to the following position.

Code: Select all

*--------------------------------------------------------*
| 25    6  2459  | 3    249   8    | 7     1     459     |
| 18    3  1489  | 7    5     6    | 48    2     489     |
| 258   59 7     | 1249 1249  149  | 3458  34569 345689  |
|----------------+-----------------+---------------------|
| 58    1  589   | 249  2489  3    | 6     7     2459    |
| 4     59 35689 | 1269 12789 1579 | 1235  359   12359   |
| 7     2  3569  | 1469 149   1459 | 1345  8     13459   |
|----------------+-----------------+---------------------|
| 9     7  15    | 14   6     2    | 13458 345   13458   |
| 12356 4  125   | 8    137   17   | 9     56-3  12567-3 |
| 1236  8  12    | 5    13479 1479 | 124-3 46-3  12467-3 |
*--------------------------------------------------------*

Note that the 3's in r7c789 are the only 3's in Row 7, so one of them must be true. Consequently you can eliminate 3 from the other cells in Box 9. This move is called locked candidates claiming.

Further on you come to the following position.

Code: Select all

*--------------------------------------------*
| 25  6   49  | 3    29  8 | 7    1    459   |
| 1   3   489 | 7    5   6 | 48   2    489   |
| 258 59  7   | 129  129 4 | 358  3569 35689 |
|-------------+------------+-----------------|
| 58  1   89  | 29   249 3 | 6    7    2459  |
| 4   59  36  | 1269 8   7 | 1235 359  12359 |
| 7   2   36  | 169  149 5 | 134  8    1349  |
|-------------+------------+-----------------|
| 9   7   15  | 4    6   2 | 1358 35   1358  |
| 36  4   25  | 8    37  1 | 9    56   2567  |
| 36  8  *12  | 5    37  9 |*12   4    67-12 |
*--------------------------------------------*

There is a naked pair (12) in r9c37 (cells marked *) so you can eliminate 1 and 2 from other cells in Row 9.

There are links to these moves here and here.

Leren

[swaatacba]

Hi Leven

"Locked candidates claiming" is not a process that I've seen previously, but it certainly looks powerful in some circs. Thank you for that.

Modelling it in Excel looks a challenge. I'll see what I can do.


Stephen

[champagne]

Hi Leven

"Locked candidates claiming" is not a process that I've seen previously, but it certainly looks powerful in some circs. Thank you for that.

Modelling it in Excel looks a challenge. I'll see what I can do.


Stephen

Forget the names, this is the simplest basic elimination with 2 cases

a) if in a box a digit is locked in a row/column, the candidate can not be in other boxes in the row/column
b) if in a row/column a candidate is locked in a box, it can not be in other rows/columns in the box

very simple and very basic

[swaatacba]

Thanks for that and I do take your point. However, it may depend on whether you are tackling SuDoku modelling as a quasi academic exercise or as a dimension conversion excise within the spreadsheet's natural 2D-grid. I have never been a SuDoku expert, but it provides an excellent model for considering more than 2 dimensions.

The task has been to convert the quasi 3 dimensional grid into a 2 dimensional surface for analysis and then back again. For the simpler processes (using the 9 x 9 grid), this has been quite straightforward. As the processes become more complex the model creator is forced to use to use a 9 x 9 x 9 grid. I confess that it is only recently that I have come to grips with this. I still have much to learn.

[Leren]

champagne wrote : Forget the names, this is the simplest basic elimination with 2 cases

Hi champagne, you are right about the names. I've always felt that this move is harder to name than to carry out !!

On some sites it's called intersection removal. For Stephen's benefit there is a link to another site here that covers both cases in a nice graphical way.

Leren

[Kozo Kataya]

swaatacba wrote : Modelling it in Excel looks a challenge. I'll see what I can do.

Would like to show one of ideas in Excel is hereunder :
Since, so-called Locked_candidates, Intersection_Removal or Pointing mighh be written as
[ A paticular mini-row in a particular box has candidates while other mini-rows in the same row are void,
candidates in other mini-row in that box can be eliminated, if there are any. ]

Pls refer attachment SD1
Steps how-to eliminate candidates in EXCEL as shown below.

step 0 range a mini-row as A1 = r1c123 and/or B2 = r2c456 etc.
step 1 count number of candidates in each mini-row
step 2 flag each mini-row, for example,
step 3 re-flag as can be eliminated = 1, else = 0
step 4 candidates elimination in a mini-row with re-flag = 1

The above steps show only mini-row looking row-direction (RR).
In addition, must apply mini-row looking column-direction (RC) too.
Also, mini-column row-direction (CR) and mini-column column-direction(CC)

Pls refer attachment SD2
For instance, given the following table with 27 of 5's candidates,
can be 9 of 5's via several steps repeated.

Pls refer attachment SD3
This shows details of re-flag from step 0 to step 1.
A combined table of [count + flag + re-flag] of RR, RC, CR and CC.

Regards

[swaatacba]

Steps how-to eliminate candidates in EXCEL as shown below.

step 0 range a mini-row as A1 = r1c123 and/or B2 = r2c456 etc.
step 1 count number of candidates in each mini-row
step 2 flag each mini-row, for example,
step 3 re-flag as can be eliminated = 1, else = 0
step 4 candidates elimination in a mini-row with re-flag = 1

The above steps show only mini-row looking row-direction (RR).
In addition, must apply mini-row looking column-direction (RC) too.
Also, mini-column row-direction (CR) and mini-column column-direction(CC)

Thank you Kozo. That looks an exceptionally powerful approach. It seems almost wholly automated.

One of curious aspect of my own approach, is to record the decisions the user/player makes in such a way that he or she IS permitted to make mistakes. Those mistakes are reflected back to the player, but are not processed. Accordingly, the player can employ a different move in the hopes of solving the puzzle.

This is all very slow and old fashioned. It is most unlikely to ever take off, but the relationship between human error and computerized systems has fascinated me for many decades.

Stephen