The New Sudoku Players' Forum

[eleven]

My formulation was not correct there. The link 5r6c7|r7c5=r7c7 is wrong, should be r6c7|r7c5=r7c78.
And it is artificial, not in a way easy to follow.
Yes, as a reader i would prefer the way you did it.

[totuan]

Hi All,
I’m a NEWBIE here with second post…

An example from old post (for very long time then I could not remember where it was…).
The combination between Almost XY-wing & Almost Kite that is presented by NET (or as Diagram)
Original Puzzle:

Code: Select all

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

From original puzzle and after SSTS (not included X-wing on 3’s):

Code: Select all

 *-----------------------------------------------------------------------------*
 | 1269    39      1236    | 4       8       259     | 235679  3679    25679   |
 | 4       5       236     | 1       29      7       | 2369    8       269     |
 | 29      8       7       | 6       2359    2359    | 1       4       259     |
 |-------------------------+-------------------------+-------------------------|
 | 3       126     9       | 8       16      25      | 2567    167     4       |
 | 8       126     4       | 2359    7       2359    | 2569    169     12569   |
 | 7       126     5       | 29      16      4       | 8       169     3       |
 |-------------------------+-------------------------+-------------------------|
 | 129     7       8       | 239     239     6       | 4       5       19      |
 | 569     4       36      | 579     59      1       | 3679    2       8       |
 | 12569   39      1236    | 2579    4       8       | 3679    13679   1679    |
 *-----------------------------------------------------------------------------*

1- Look at 9’s on Row 2 and Box 9: if r2c7 & r9c789<>9 => Kite 9’s : r2c5=r2c9-r7c9=r8c7 => r8c5<>9
2- Look at Almost XY-wing (259)r28c5/r1c6: if r1c6<>9 => XY-wing (259)r28c5/r1c6 => r3c5<>5 => r8c5=5 => r8c5<>9

Present as diagram: => r8c5<>9, stte.

Code: Select all

(9)r2c7-(3)r2c7=(3)r2c3-(3=9)r1c2--(9)r1c6=(XY-wing:259)r28c5/r1c6*
 ||                              |
(9)r9c789-(9)r9c2=(9)r1c2--------
 ||
{Kite 9’s: r2c5=r2c9-r7c9=r8c7)*

By thinking on Net you can combine many parterns or techniques together like A*ALS, A*AUR, A*XYZ-wing, A*Fish…

P/s:
- Nice to see you all still online here.
- Hi champagne – my Big Brother: yes, it’s me (totuan – ttt)
- I have to wait long time after submit my post to see it active, d'ont know why...?

HAPPY NEW YEAR – 2018 to All – SudokuLand.

totuan

[SpAce]

My formulation was not correct there. The link 5r6c7|r7c5=r7c7 is wrong, should be r6c7|r7c5=r7c78.

I guess you missed my earlier note on that:

http://forum.enjoysudoku.com/line-check-for-sudocue-nightmare-sun-dec-9-2007-t34296-60.html#p261986

Should be r78c7, right? [[5r4c9 -> -5r6c7 -> [5r7c7 | 5r8c7]] | [5r4c5 -> -5r7c5 -> 5c7c7]] => [[5r7c7 | 5r8c7] | 5r7c7] => [5r7c7 | 5r8c7] <=> 5r78c7. Or am I missing something?

And it is artificial, not in a way easy to follow.
Yes, as a reader i would prefer the way you did it.

Well, as a learner I actually prefer the way you did it. It may not be the easiest to follow or the simplest possible way to do it, but that's exactly why it helps me learn better. Different perspectives and sometimes even obscurities force one to think more, which leads to deeper learning. For example, it's unnatural for me to use ALS nodes to make jumps in chains (unless it's actually necessary), and I don't think that style is very readable in general, but I still like seeing experienced solvers use it because it helps me learn to spot and use ALSs better.

Also, the readability problem with that ALS jumping style is not in itself or its users but more of a flaw of Eureka. It hides the digit placements of ALSs, which makes it harder to follow such links without seeing the grid. For instance, this part of your chain is quite hard to follow without seeing the grid: (5=1)r36c7-(1=7)r2c4,r3c5, even though those are very small and simple ALSs. Written differently it's longer but very easy to follow: (5=4)r6c7-(4=1)r3c7-(1=4)r3c5-(4=7)r2c4. Yet I don't mind seeing the former style because it teaches me more. Easy things aren't the best teachers.

[SpAce]

I’m a NEWBIE here with second post…

Welcome, totuan.

The combination between Almost XY-wing & Almost Kite that is presented by NET (or as Diagram)

Thanks for the interesting example. I'm basically a newbie too, so it took a little bit of deciphering. In such a relatively complex net example I'd personally find it clearer if you presented all the cases first that you're proving to lead to the same conclusion. That way it would be easier to see that at least one of the cases, and hence the conclusion, must be true. Yes, the information was there but it wasn't obvious at first. I guess the two cases that cover all bases here are:

((r2c7 | r9c789) = 9) | ((r2c7 | r9c789) <> 9)

Or three cases in practice:

9r2c7 | 9r9c789 | (-9r2c7 & -9r9c789)

Your proofs for those three cases seem fine to me. As a minor detail, I don't think it's a Kite but an Empty Rectangle (anyone disagree?)

[Added: Could it be notated like this:

ER[(9)r8c7=r7c9-r2c9=(9)r2c5] = [[(9-3)r2c7 = (3)r2c3] | [(9)r9c789 - (9=3)r9c2]] - (3=9)r1c2 - (9)r1c6 = XY-Wing[(5=2)r1c6-(2=9)r2c5-(9=5)r8c5] - (5)r3c5 = (5)r8c5

Endpoints: [(9)r8c7 | (9)r2c5 | (5)r8c5] => -9 r8c5

??]

I have to wait long time after submit my post to see it active, d'ont know why...?

I've understood that it's a spam/troll filter. It happened to me too, so I guess anyone's first few posts go through a moderator. After that you can post without a delay. Be patient.

[eleven]

Oh, a One-stepper.

This is an alternative with the almost xy-wing, only using small nets.

Code: Select all

     *-----------------------------------------------------------------------------*
     | 1269   d39      1236    | 4       8      c25+9    | 235679  3679    25679   |
     | 4       5      d236     | 1     dc29      7       | 2369    8      e269     |
     | 29      8       7       | 6      b2359    2359    | 1       4       259     |
     |-------------------------+-------------------------+-------------------------|
     | 3       126     9       | 8       16      25      | 2567    167     4       |
     | 8       126     4       | 2359    7       2359    | 2569    169     12569   |
     | 7       126     5       | 29      16      4       | 8       169     3       |
     |-------------------------+-------------------------+-------------------------|
     |f129     7       8       |g239    g239     6       | 4       5      f19      |
     | 569     4       36      | 579   ca59      1       | 3679    2       8       |
     | 12569  e39      1236    | 2579    4       8       | 3679    13679   1679    |
     *-----------------------------------------------------------------------------*

5r8c5=r3c5-xy-wing259r1c6,r28c5=9r1c6-(9=236)r2c35,r1c2-(2|3|6=9)r2c9,r9c2-r7c19=r7c45 => -9r8c5

[SpAce]

This is an alternative with the almost xy-wing, only using small nets.
5r8c5=r3c5-xy-wing259r1c6,r28c5=9r1c6-(9=236)r2c35,r1c2-(2|3|6=9)r2c9,r9c2-r7c19=r7c45 => -9r8c5

Nice. (But how about some white space to make it more readable?) I see it like this backwards:

(9)r7c45 = [[r7c1 - (9=3)r9c2 - (3=9)r2c2] | [r7c9 - r2c9 = WXYZ-Wing:2369:r1c2,r2c359]] - (9)r1c6 = XY-Wing:259:r1c6,r28c5 - (5)r3c5 = (5)r8c5

I also tried to write it without complex nodes, but then I had to use a memory marker and lost bidirectionality:

(5)r8c5 = [[(9)r8c5 - (9=2*)r2c5] & [(5)r3c5]] - (2|5=9)r1c6 - (9=3)r1c2 - [[(3|2*=6)r2c3 - (6|2*=9)r2c9 - r7c9] | [(3=9)r9c2 - r7c1]] = (9)r7c45

Can anyone come up with a way to make that style work?

[JC Van Hay]

Here is the Forbidding Matrix equivalent to the ttt's diagram if the FM is read from the rows 7 & 8 :

Code: Select all

+--------------------+--------------------+-------------------------+
| 1269   (39)  1236  | 4     8      (259) | 235679  3679     25679  |
| 4      5     26(3) | 1     (29)   7     | 26(39)  8        26(9)  |
| 29     8     7     | 6     239-5  2359  | 1       4        259    |
+--------------------+--------------------+-------------------------+
| 3      126   9     | 8     16     25    | 2567    167      4      |
| 8      126   4     | 2359  7      2359  | 2569    169      12569  |
| 7      126   5     | 29    16     4     | 8       169      3      |
+--------------------+--------------------+-------------------------+
| 129    7     8     | 239   239    6     | 4       5        1(9)   |
| 569    4     36    | 579   (59)   1     | 367(9)  2        8      |
| 12569  3(9)  1236  | 2579  4      8     | 367(9)  1367(9)  167(9) |
+--------------------+--------------------+-------------------------+
|5r8c5 9r8c5      |
|      9r2c5 2r2c5|
|5r1c6       2r1c6|9r1c6
                   9r1c2 3r1c2
                         3r2c3 3r2c7
                   9r1c2             9r9c2
                                     9r9c789|9r8c7 9r7c9|
                               9r2c7        |9r2c5 9r2c9|
 5r8c5                                       9r8c5
   |
   V
-5r3c5; stte

[JC Van Hay]

Strange that there is no feedback to my previous post. Further more, one is trying to write down an AIC in an extremely complicated way with "logical operators" while totuan ttt's diagram is written as a bidirectional AIC. To wit :

(5=9)r8c5-TurboFish(9r2c59, 9r7c9.9r8c7)=*L2-Wing[3r2c3=3r2c7-9r2c7=*9r9c789-9r9c2=9r1c2]-(3=9)r1c2-9r1c6=XY-Wing(259)r1c6,r28c5 -> DC={5r8c5, 5r1c6} => -5r3c5; stte

[totuan]

Hi JC and All,
In fact, I only intend to show you another view on finding eliminations that are quite useful for puzzles with ERs around 9.0 or higher. I do not want to make more complicated presentation by Eureka notation so I present it by simple Eureka notation with branches or as diagram.

Thanks to All and have nice weekend!
totuan

[eleven]

Strange that there is no feedback to my previous post.

Maybe we are just lacking a good explanation of the forbidden matrix. All i found out was that the rows have (here) exactly one truth and the columns at least one. But how to find one manually ? I don't want to write down hundreds of lines and then try to combine them.

In the AIC chains i try to reflect, how i can continue with a link (e.g. here the 5r8c5=9r1c6 from the almost xy-wing) to get an elimination, without having to remember past steps (just the starting point).

@ttt, i do know your nets. Great stuff.