hard sudoku, "Platinum Blonde"

[David Bernier]

Rating Program: gsf's sudoku q1
Rating: 99486
Poster: coloin
Label: HardestSudokusThread-00078;Platinum_Blonde
[ from sudoku programmers pages ]


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


ref.:

http://arxiv.org/abs/1208.0370

"The Chaos Within Sudoku" at arXiv,
Maria Ercsey-Ravasz, Zoltan Toroczkai;

They say "platinum blonde" is the hardest 9x9 on their rating,
that they found on collection puzzles.
They mention 49x49 sudokus, etc.

--> Platinum Blonde (the sudoku's given name)

Note:
http://www.dailysudoku.co.uk/sudoku/play.shtml ,
when asked to solve "Platinum Bonde",
says: "Difficulty: too hard"

david bernier



--
99997066781489109195113098994290469881614963208468

[Robin Chapman]

On 10/05/2013 16:42, David Bernier wrote:
> Rating Program: gsf's sudoku q1
> Rating: 99486
> Poster: coloin
> Label: HardestSudokusThread-00078;Platinum_Blonde
> [ from sudoku programmers pages ]
>
>
> . . . | . . . | . 1 2
> . . . | . . . | . . 3
> . . 2 | 3 . . | 4 . .
> -------+-------+------
> . . 1 | 8 . . | . . 5
> . 6 . | . 7 . | 8 . .
> . . . | . . 9 | . . .
> -------+-------+------
> . . 8 | 5 . . | . . .
> 9 . . | . 4 . | 5 . .
> 4 7 . | . . 6 | . . .

839|465|712
146|782|953
752|391|486
-----------
391|824|675
564|173|829
287|659|341
-----------
628|537|194
913|248|567
475|916|238

About 7 seconds and 7000 stack pushes on my sudoku
solver (uses a depth-first search). This is a lot
more than some other vaunted "hard" sudokus.

[David Bernier]

That's interesting!

From "Champagne" at:
http://code.google.com/p/skfr-sudoku-fast-rating/downloads/list
potenitalhardest13_02.zip
release date: March 22.

98. 7.. ...
7.. ... 6..
..6 .5. ...

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

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

single line of 81 characters:
98.7.....7.....6....6.5.....4...5.3...79..5......2...1..85..9......1...4.....3.2.

+ ";11.90;11.90;11.80;GP;champagne dry;1;22;"


[rating: 11.90;11.90;11.80;GP;champagne ]



--
99997066781489109195113098994290469881614963208468

[JT]

Yeah you fucking retards go play sudoku that is all your good for.

[Robin Chapman]

On 10/05/2013 21:04, JT wrote:

> Yeah you fucking retards go play sudoku that is all your good for.

All my what?

[Robin Chapman]

981 736 245
735 492 618
426 158 397

142 685 739
367 941 582
859 327 461

218 564 973
673 219 854
594 873 126

A bit easier: 3 secs and < 3000 stack pushes.

[Richard Tobin]

In article <kmjnjr$lmq$1...@dont-email.me>,

Robin Chapman <R.J.C...@ex.ac.uk> wrote:

>>> About 7 seconds and 7000 stack pushes on my sudoku
>>> solver (uses a depth-first search).

...

>A bit easier: 3 secs and < 3000 stack pushes.

The second one takes nearly twice as long as the first on my
depth-first solver. Implementation details outweigh any notion
of difficulty!

-- Richard

[JT]

All your puny minds and ass, belongs to me.

[Robin Chapman]

A plural subject but a singular verb :-(

[JT]

Well you are the ass don't you see?
Asinus, a subgenus of Equus that includes the donkey and other asses
Donkey, Equus africanus asinus By extension, an insult meaning idiot.
Disco disco good good http://www.youtube.com/watch?v=ndN_5IrPOhc

[JT]

Extension someone was riding the idiots all the way to the market.

[Dirk Van de moortel]

David Bernier <davi...@videotron.ca> wrote:
> Rating Program: gsf's sudoku q1
> Rating: 99486
> Poster: coloin
> Label: HardestSudokusThread-00078;Platinum_Blonde
> [ from sudoku programmers pages ]
>
>
> . . . | . . . | . 1 2
> . . . | . . . | . . 3
> . . 2 | 3 . . | 4 . .
> -------+-------+------
> . . 1 | 8 . . | . . 5
> . 6 . | . 7 . | 8 . .
> . . . | . . 9 | . . .
> -------+-------+------
> . . 8 | 5 . . | . . .
> 9 . . | . 4 . | 5 . .
> 4 7 . | . . 6 | . . .

A nightmare to solve on paper.
Good grief...

Dirk Vdm

[JT]

On 10 Maj, 23:51, "Dirk Van de moortel" <dirkvandemoor...@hotspam.not>
wrote:

http://www.youtube.com/watch?v=GuVwqawKowQ

[A Nony Mouse]

In article
<d06ae04e-1b46-4cc9...@g7g2000vbv.googlegroups.com>,

JT <jonas.t...@gmail.com> wrote:

> On 10 Maj, 23:51, "Dirk Van de moortel" <dirkvandemoor...@hotspam.not>
> wrote:
> > David Bernier <david...@videotron.ca> wrote:
> > > Rating Program: gsf's sudoku q1
> > > Rating: 99486
> > > Poster: coloin
> > > Label: HardestSudokusThread-00078;Platinum_Blonde
> > > [ from sudoku programmers pages ]
> >
> > >  . . . | . . . | . 1 2
> > >  . . . | . . . | . . 3
> > >  . . 2 | 3 . . | 4 . .
> > > -------+-------+------
> > >  . . 1 | 8 . . | . . 5
> > >  . 6 . | . 7 . | 8 . .
> > >  . . . | . . 9 | . . .
> > > -------+-------+------
> > >  . . 8 | 5 . . | . . .
> > >  9 . . | . 4 . | 5 . .
> > >  4 7 . | . . 6 | . . .

`

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

4 7 5 | 9 1 6 | 2 3 8

[Robin Chapman]

On 10/05/2013 22:49, JT wrote:
>
> Extension someone was riding the idiots all the way to the market.
>

"Extension someone"?

That's a rather queer noun phrase. Any idea what it means?

[Dirk Van de moortel]

Probably something like

"Extension: someone was riding the idiots all the way to the market."

-- whatever *that* might mean.
Laziness, idiocy, authism... who knows? Who even wants to know?

Dirk Vdm

[JT]

On 12 Maj, 15:46, "Dirk Van de moortel" <dirkvandemoor...@hotspam.not>
wrote:

Well you are the donkeys after all.

[Paul]

Robin,

By counting the stack pushes, you are simply discussing the idiosyncrasies of your particular implementation rather than any objective measure. Suppose, you change your code by reordering the blanks, or by changing the order in which candidates for a square are considered, then your count will probably change wildly.

To measure this properly using a stack-based solution, you should probably randomise the order in which blanks are considered and randomise the order in which candidate digits are tried, then take an average. Of course, the randomisation shouldn't violate algorithmic constraints -- for example, perhaps you hone in on one of the blanks which has the least number of fillable digits.

Paul

[Robin Chapman]

On 20/06/2016 02:10 pm, Paul wrote:

> Robin,
>
> By counting the stack pushes, you are simply discussing the
> idiosyncrasies of your particular implementation

That's right!

> rather than any
> objective measure. Suppose, you change your code by reordering the
> blanks, or by changing the order in which candidates for a square are
> considered, then your count will probably change wildly.

Quite likely, but my programme always chooses an entry with the
minimum number of possibles for the next stack push.

> To measure this properly using a stack-based solution, you should
> probably randomise the order in which blanks are considered and
> randomise the order in which candidate digits are tried, then take an
> average.

Maybe I should, but I'm not going to bother :-)

[Paul]

Of course, your programme chooses one of the entries with the minimum number of possibilities. Otherwise it couldn't possibly work. My programme regards the coordinates of the blanks as being in lexicographic order. Presumably, you have a way of ordering the blanks (before you've considered possibilities for filling). Similarly, you probably consider trying the digits in the order 1, 2, ...9. I experimented with this sudoku: http://www.telegraph.co.uk/news/science/science-news/9359579/Worlds-hardest-sudoku-can-you-crack-it.html reversing both ordering conventions. This makes an absolutely massive difference -- With one convention the number of pops and pushes is 24174 and with another, it is 11280. The trick is to sort out the ordering conventions at the very beginning of the code so it doesn't intefere with your minimization algorithm.

Paul

[Robin Chapman]

On 20/06/2016 05:23 pm, Paul wrote:

> Of course, your programme chooses one of the entries with the
> minimum
number of possibilities. Otherwise it couldn't possibly work. My
programme regards the coordinates of the blanks as being in
lexicographic order. Presumably, you have a way of ordering the blanks
(before you've considered possibilities for filling). Similarly, you
probably consider trying the digits in the order 1, 2, ...9.

Dear, dear, nothing so crude :-)

I
experimented with this sudoku:
http://www.telegraph.co.uk/news/science/science-news/9359579/Worlds-hardest-sudoku-can-you-crack-it.html
reversing both ordering conventions. This makes an absolutely massive
difference -- With one convention the number of pops and pushes is 24174
and with another, it is 11280. The trick is to sort out the ordering
conventions at the very beginning of the code so it doesn't intefere
with your minimization algorithm.
>

My solver took 6372 pushes.

[Paul]

On Monday, June 20, 2016 at 6:46:21 PM UTC+1, Robin Chapman wrote:

> > Of course, your programme chooses one of the entries with the
> > minimum
> number of possibilities. Otherwise it couldn't possibly work. My
> programme regards the coordinates of the blanks as being in
> lexicographic order. Presumably, you have a way of ordering the blanks
> (before you've considered possibilities for filling). Similarly, you
> probably consider trying the digits in the order 1, 2, ...9.
>
> Dear, dear, nothing so crude :-)

You're absolutely right. The order in which the digits are tried needs to be dependent on their frequency in the grid. I'm looking forward to implementing this and hopefully massively improving the efficiency. I realised this problem as soon as I posted.

Paul