The New Sudoku Players' Forum

[mw210461]

From a book "Meister Sudokus" ("Master Sudokus" for you), Naumann und Goebel, Cologne, Germany, puzzle 40 of 126):

.6.|..4|...
...|.2.|...
3.5|8..|..7
-----------
65.|..7|.9.
.4.|...|.2.
.7.|1..|.36
-----------
4..|..6|9.1
...|.8.|...
...|5..|.8.


Reduced by logic to:

.6.|..4|.1.
..4|.2.|.6.
395|861|247
-----------
65.|..7|19.
.4.|6..|72.
.7.|1..|.36
-----------
4..|..6|951
5..|48.|.7.
...|51.|.8.

The final solution is known, albeit not the reasonable way to continue.
All I am asking for is the next ONE to TWO logical steps, i.e. the reason for these choices. Don't need the name of the rule, just the RATIONALE for continuing.
I solve Sudokus for a while now, am decent (at least I thought so until I bought this book, 35 of these 126 I could NOT solve so far, some needed 45+ min). Newspaper puzzles I can typically solve in 3-10minutes, but these here got me stomped. Have more examples of 'bad' ones if needed.
The book claims that they all have unique choices, probably true.

Apparently I am missing one or two solution paths that I am not aware of so far, or I am overlooking the obvious, or I am simply dumb (see Subject).

Thanks!
michael

[Glyn]

Michael
My guess is that you have found all the single candidates which brings you to the grid you posted with 41 cells known. At this stage you have probably got these possibilities left.

Code: Select all

.------------------.------------------.------------------.
| 278   6     278  | 379   3579  4    | 358   1     3589 |
| 178   18    4    | 379   2     359  | 358   6     3589 |
| 3     9     5    | 8     6     1    | 2     4     7    |
:------------------+------------------+------------------:
| 6     5     238  | 23    34    7    | 1     9     48   |
| 189   4     1389 | 6     359   3589 | 7     2     58   |
| 289   7     289  | 1     459   2589 | 458   3     6    |
:------------------+------------------+------------------:
| 4     238   2378 | 237   37    6    | 9     5     1    |
| 5     123   12369| 4     8     239  | 36    7     23   |
| 279   23    23679| 5     1     239  | 346   8     234  |
'------------------'------------------'------------------'

You now need some new moves (not difficult though).
All these are forms of locked candidates.

In Box 4 in which column must the 3's go?
In Box 8 on what row must the 7's go?
In Column 2 in which box must the 2's go?

What is the consequence on Row 7 Column 3?

It should be plain sailing from there. This will get you to the next level and some of those ones that have stumped you should be easier.

[mw210461]

Thanks for the quick reply!
It was the exact answer I wanted and needed, and of course it makes sense.
Sort of sad as I DID know the three answers by reason, yet failed to see their immediate impact on the cell scan for Row 7 Col 3.
It also shows a weakness of my current strategy for finding candidates by cell scanning, an extra ELIMINATION step is needed to purge out impossibles. Or better phrased: when scanning a cell within a box one should always include numbers as 'given', if their location is unknown within the box, but has to be clearly outside the currently scanned cell!

Having solved many games I developed the habit to only write down candidates if their location is limited to TWO possibilities, trying to avoid overload in cells.
I found this to be very helpful for filling the gaps quickly later when the 'front clears', and most published puzzles do not need more than that, even if marked 'hard'. Newspapers and mags typically do not over-challenge their readers, they don't want to aggravate or disillusion them.
Apparently this method is not sufficient for "master" levels as used in this book.

As a 'non' chess player I constantly grapple about my obvious inability to foresee the implication of my doings - I seem to be unable to predict the consequences of testing choices, that may lead to conflicts at some time down the road. Good chess player 'plan' and 'see' their figures steps ahead, I wonder how they'd score on a Sudoku grid?
But maybe this method is not even the goal of Sudoku, and Kakuro as well.

I will keep trying, our local supermarket will offer another "Expert" book tomorrow, cheap and lasting entertainment for 2Euros.

Thanks again!
michael

[Pat]

hey mw210461,

if you enjoy this level of puzzles -- as i do! -- take a look at Puzzles needing lots of locked candidates

[mw210461]

Not so sure if I really enjoy them, I seem to suck when it comes to what I consider "hard" ones. With concentration and care I could finish another couple of the difficult ones in the book, still about 28/126 to go..

About 1 year ago I wrote my own Sudoku application (Windows, VB) and added features for logging traces, cell scans, immediate error trapping, step by step help hints with rules, auto solving with rule explanation etc. It will solve any reasonable puzzle with brute force as well.
It also designs new puzzles, but these random ones usually suck due to ambiguity and weird constellations. Found the same issue with electronic pocket Sudoku games - their "games" lack logic and reason in many cases.

Funny that my own program could actually help me - the AUTHOR! - in several cases when I got stuck myself. Mostly due to more accuracy and care of course, as it only "knows" what I taught it to know.
The puzzles of this book are out of my league, and my little proggie bends backwards as well. It will solve all by guessing, but it can no longer explain the reasons, that's what brought me here...
Adding more code for "locked candidates" is an option but I need to digest these first to come up with reasonable code.

Is the following partly solved puzzle another example of "locked candidates"? I don't see any.
It started as a brutal "24" (57 missing) and I nailed it down to 43 empty cells.
I DO know that Col 5 Rows 1 and 2 appear to be "key" - ie. guessing 9 or 7 will roll it all up. But HOW do I KNOW this? Are there other key cells that are more obvious?

Code start:

..3|2..|.6.
5..|3..|..1
.47|...|...
-----------
...|.81|..2
.6.|...|.9.
1..|43.|...
-----------
...|...|57.
7..|..8|..6
.9.|..7|4..

Got to:

.13|2..|.6.
5..|3..|..1
.47|81.|...
-----------
...|681|..2
.6.|752|19.
1..|439|6..
-----------
...|..3|57.
7..|..8|..6
.9.|.67|4..

Would you judge puzzles of this level 'hard'?
I never found any like these in regular newspapers or standard magazines, not even in mixed Sudoku booklets. They seem to be addressed to "Profis" ("pros") as the book subtitle suggests, and that I am clearly not


Thanks again
michael

[Glyn]

Michael
Naked Pairs and Triplets or Hidden pairs and locked candidates should be all that's needed.
Lots of locked candidates though.
As for the difficulty, it is at the top end of the range for a newspaper, an end of the week offering, due to the number of non-trivial moves. It is possible to find harder moves than these in newspapers but the number required would be limited.
Without a progress grid showing the other candidates left, it is difficult to tell how close to the end you are. Sounds promising though.