The New Sudoku Players' Forum

[Red Ed]

As a benchmark, how about this commercial one from Gakken Mook Pocket Puzzle 100 Series 20: Pocket Number Place 8 ...

Code: Select all

+---+---+---+
|...|916|...|
|..3|...|1..|
|.4.|...|.9.|
+---+---+---+
|1..|.8.|..7|
|5..|3.9|..1|
|6..|.2.|..5|
+---+---+---+
|.9.|...|.4.|
|..8|...|2..|
|...|458|...|
+---+---+---+

Obviously I don't claim this as my own! But I'd like to know how it rates, to give me and maybe others something to try to beat in future.

[vidarino]

But I'd like to know how it rates, to give me and maybe others something to try to beat in future.

It's well under par for this thread, I'm afraid. It solves with nothing but singles and locked candidates. A "superior" puzzle needs at least some pairs and/or triples and/or X-Wings, as far as I know.

That doesn't have to mean that your puzzle is easy, though. The ones in the "inferior" thread are singles only, but they are by far the hardest singles-only puzzles I have ever seen.

Vidar

[tarek]

Nice one henk, I actually in one step could spot 6 x-wings (although probably only 2 were needed).......

Having said that, There are no doubles or triples needed to solve...

Tarek

there are if you remove R1C1 and R9C9

No there still isn't , those x-wings are too powerful.

I can make an improvement on Vidarino's latest qualifying offering too:

It is an improvemrnt, one of the naked doubles is swapped with a naked triple, as it is an improvement ... It will go on the list.

Tarek

[Red Ed]

OK, I get it.

How about this. Not so pretty at first sight until you realise that each pair of diametrically opposite digits always adds to 10 (in the solution as well as in the puzzle itself).

Code: Select all

. . 1 . 9 . . . 2
2 . . 3 . . 1 . .
. . . . . 4 3 . .
. . . . . . 5 4 6
. 7 . . . . . 3 .
4 6 5 . . . . . .
. . 7 6 . . . . .
. . 9 . . 7 . . 8
8 . . . 1 . 9 . .

I think it needs an X-wing ...

[ab]

I think it needs an X-wing ...

Yes it does, and pairs and locked candidates. Should make the grade

[ab]

Having said that, There are no doubles or triples needed to solve...

Tarek

there are if you remove R1C1 and R9C9

No there still isn't , those x-wings are too powerful.


Tarek

ah yeah you can use locked candidates instead!

[JPF]

Would this one go on the list ?

Code: Select all

Puzzle #04

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


JPF

[tarek]

Red Ed,
Good one, definitely a superior

Would this one go on the list ?

Unfortunately not, it is harder than a superior. it had the required number of doubles, but it needed something more than the hidden triples to solve.

All rules of posting the Superior puzzles are found with the updated Superior puzzle list at the start of this thread (you follow this link:http://forum.enjoysudoku.com/viewtopic.php?p=23814#p23814)

There is also my personal rankings of the 1st 62 puzzles here

Tarek

[Red Ed]

Here's another sum-to-10 grid, this time with rotational symmetry and needing an x-wing to solve:

Code: Select all

 . . 1 2 . . . . .
 . . . . . 3 4 . .
 . 2 9 . . . 3 . 7
 . 9 . . 3 . . . 4
 . . . 9 . 1 . . .
 6 . . . 7 . . 1 .
 3 . 7 . . . 1 8 .
 . . 6 7 . . . . .
 . . . . . 8 9 . .

This slightly less pretty sum-to-10 needs a hidden triple:

Code: Select all

 . 1 . . . . . 2 .
 . . . . . 8 3 . .
 . . 4 5 7 . . . .
 6 3 5 . . . . . 9
 . . . . . . . . .
 1 . . . . . 5 7 4
 . . . . 3 5 6 . .
 . . 7 2 . . . . .
 . 8 . . . . . 9 .

[tarek]

Wow Red Ed,

That first puzzle, I was able to construct 8 x-wings in the same step, for that it will be a superior. Would not score highly on the difficulty or variability rankings though.

The second puzzle needs a hidden triple & (a naked triple or a hidden double), again a superior........

Tarek

[Ruud]

I was admiring these sum-to-10 submissions by Red Ed, when I suddenly realized that they are not minimal.

Take this beauty, for instance:

Code: Select all

 . . 1 2 . . . . .
 . . . . . 3 4 . .
 . 2 9 . . . 3 . 7
 . 9 . . 3 . . . 4
 . . . 9 . 1 . . .
 6 . . . 7 . . 1 .
 3 . 7 . . . 1 8 .
 . . 6 7 . . . . .
 . . . . . 8 9 . .

Because I know it is a sum-to-10, it can easily be minimized to:

Code: Select all

 . . 1 2 . . . . .
 . . . . . 3 4 . .
 . 2 9 . . . 3 . 7
 . 9 . . 3 . . . 4
 . . . 9 . . . . .
 . . . . . . . . .
 . . . . . . . . .
 . . . . . . . . .
 . . . . . . . . .

I'm not sure if there is more that can be eliminated from these puzzles.

Another thought: Of course you lose symmetry in the clues, but the first step in solving it restores the symmetry. In fact, these puzzles can be solved without ever losing symmetry along the way.

Ruud

[vidarino]

Hehe, quite correctly observed, Ruud.

This type of symmetry popped up in another thread, actually.

I still haven't implemented the idea i aired there (symmetry similar to N vs 10-N, but with the pairings unknown at start time), but I might get around to it soon. Unless someone beats me to it.

[tarek]

if the sum to (N) was on a 90 degrees rotational symmetry...... Then you would have an extra cnstraint because you are looking at 4 cells (A killer taste) [i.e. highlight all symmeterical clue cells with a colour, all cells in that particular colour should sum up to 10], we know that we still can't use the central cell, but it would be interesting to see the cells in box 5

Tarek

[tarek]

I'm going to post the next rankings as soon as we hit the 75 mark, the general formula would depend on variability primarily.........
if there are 8 x-wings that can be deduced by looking at a given grid candidates The score would be 1 x-wing, however 2 x-wings that can't be deduced from looking at one grid would score 2 x-wings, this would go for every technique....

then the weighting would be bised towards Hidden triples mostly & x-wings.

The formula would then be:

Code: Select all

Hidden Triples*5+XWings*2+Naked triples+Hidden doubles+Naked doubles*0.5

What would be your suggestions ???

Tarek

[ab]

So you don't rate locked candidates at all Tarek? I think they're fun.